95c51c0b7df0fe68743dd571143196d5760d4abe
   1/*
   2 * Copyright (C) 2005 Junio C Hamano
   3 */
   4#include "cache.h"
   5#include "config.h"
   6#include "tempfile.h"
   7#include "quote.h"
   8#include "diff.h"
   9#include "diffcore.h"
  10#include "delta.h"
  11#include "xdiff-interface.h"
  12#include "color.h"
  13#include "attr.h"
  14#include "run-command.h"
  15#include "utf8.h"
  16#include "userdiff.h"
  17#include "submodule-config.h"
  18#include "submodule.h"
  19#include "hashmap.h"
  20#include "ll-merge.h"
  21#include "string-list.h"
  22#include "argv-array.h"
  23#include "graph.h"
  24#include "packfile.h"
  25
  26#ifdef NO_FAST_WORKING_DIRECTORY
  27#define FAST_WORKING_DIRECTORY 0
  28#else
  29#define FAST_WORKING_DIRECTORY 1
  30#endif
  31
  32static int diff_detect_rename_default;
  33static int diff_indent_heuristic = 1;
  34static int diff_rename_limit_default = 400;
  35static int diff_suppress_blank_empty;
  36static int diff_use_color_default = -1;
  37static int diff_color_moved_default;
  38static int diff_context_default = 3;
  39static int diff_interhunk_context_default;
  40static const char *diff_word_regex_cfg;
  41static const char *external_diff_cmd_cfg;
  42static const char *diff_order_file_cfg;
  43int diff_auto_refresh_index = 1;
  44static int diff_mnemonic_prefix;
  45static int diff_no_prefix;
  46static int diff_stat_graph_width;
  47static int diff_dirstat_permille_default = 30;
  48static struct diff_options default_diff_options;
  49static long diff_algorithm;
  50static unsigned ws_error_highlight_default = WSEH_NEW;
  51
  52static char diff_colors[][COLOR_MAXLEN] = {
  53        GIT_COLOR_RESET,
  54        GIT_COLOR_NORMAL,       /* CONTEXT */
  55        GIT_COLOR_BOLD,         /* METAINFO */
  56        GIT_COLOR_CYAN,         /* FRAGINFO */
  57        GIT_COLOR_RED,          /* OLD */
  58        GIT_COLOR_GREEN,        /* NEW */
  59        GIT_COLOR_YELLOW,       /* COMMIT */
  60        GIT_COLOR_BG_RED,       /* WHITESPACE */
  61        GIT_COLOR_NORMAL,       /* FUNCINFO */
  62        GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  63        GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  64        GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  65        GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  66        GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  67        GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  68        GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  69        GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  70};
  71
  72static NORETURN void die_want_option(const char *option_name)
  73{
  74        die(_("option '%s' requires a value"), option_name);
  75}
  76
  77static int parse_diff_color_slot(const char *var)
  78{
  79        if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  80                return DIFF_CONTEXT;
  81        if (!strcasecmp(var, "meta"))
  82                return DIFF_METAINFO;
  83        if (!strcasecmp(var, "frag"))
  84                return DIFF_FRAGINFO;
  85        if (!strcasecmp(var, "old"))
  86                return DIFF_FILE_OLD;
  87        if (!strcasecmp(var, "new"))
  88                return DIFF_FILE_NEW;
  89        if (!strcasecmp(var, "commit"))
  90                return DIFF_COMMIT;
  91        if (!strcasecmp(var, "whitespace"))
  92                return DIFF_WHITESPACE;
  93        if (!strcasecmp(var, "func"))
  94                return DIFF_FUNCINFO;
  95        if (!strcasecmp(var, "oldmoved"))
  96                return DIFF_FILE_OLD_MOVED;
  97        if (!strcasecmp(var, "oldmovedalternative"))
  98                return DIFF_FILE_OLD_MOVED_ALT;
  99        if (!strcasecmp(var, "oldmoveddimmed"))
 100                return DIFF_FILE_OLD_MOVED_DIM;
 101        if (!strcasecmp(var, "oldmovedalternativedimmed"))
 102                return DIFF_FILE_OLD_MOVED_ALT_DIM;
 103        if (!strcasecmp(var, "newmoved"))
 104                return DIFF_FILE_NEW_MOVED;
 105        if (!strcasecmp(var, "newmovedalternative"))
 106                return DIFF_FILE_NEW_MOVED_ALT;
 107        if (!strcasecmp(var, "newmoveddimmed"))
 108                return DIFF_FILE_NEW_MOVED_DIM;
 109        if (!strcasecmp(var, "newmovedalternativedimmed"))
 110                return DIFF_FILE_NEW_MOVED_ALT_DIM;
 111        return -1;
 112}
 113
 114static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 115                                struct strbuf *errmsg)
 116{
 117        char *params_copy = xstrdup(params_string);
 118        struct string_list params = STRING_LIST_INIT_NODUP;
 119        int ret = 0;
 120        int i;
 121
 122        if (*params_copy)
 123                string_list_split_in_place(&params, params_copy, ',', -1);
 124        for (i = 0; i < params.nr; i++) {
 125                const char *p = params.items[i].string;
 126                if (!strcmp(p, "changes")) {
 127                        options->flags.dirstat_by_line = 0;
 128                        options->flags.dirstat_by_file = 0;
 129                } else if (!strcmp(p, "lines")) {
 130                        options->flags.dirstat_by_line = 1;
 131                        options->flags.dirstat_by_file = 0;
 132                } else if (!strcmp(p, "files")) {
 133                        options->flags.dirstat_by_line = 0;
 134                        options->flags.dirstat_by_file = 1;
 135                } else if (!strcmp(p, "noncumulative")) {
 136                        options->flags.dirstat_cumulative = 0;
 137                } else if (!strcmp(p, "cumulative")) {
 138                        options->flags.dirstat_cumulative = 1;
 139                } else if (isdigit(*p)) {
 140                        char *end;
 141                        int permille = strtoul(p, &end, 10) * 10;
 142                        if (*end == '.' && isdigit(*++end)) {
 143                                /* only use first digit */
 144                                permille += *end - '0';
 145                                /* .. and ignore any further digits */
 146                                while (isdigit(*++end))
 147                                        ; /* nothing */
 148                        }
 149                        if (!*end)
 150                                options->dirstat_permille = permille;
 151                        else {
 152                                strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 153                                            p);
 154                                ret++;
 155                        }
 156                } else {
 157                        strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 158                        ret++;
 159                }
 160
 161        }
 162        string_list_clear(&params, 0);
 163        free(params_copy);
 164        return ret;
 165}
 166
 167static int parse_submodule_params(struct diff_options *options, const char *value)
 168{
 169        if (!strcmp(value, "log"))
 170                options->submodule_format = DIFF_SUBMODULE_LOG;
 171        else if (!strcmp(value, "short"))
 172                options->submodule_format = DIFF_SUBMODULE_SHORT;
 173        else if (!strcmp(value, "diff"))
 174                options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 175        else
 176                return -1;
 177        return 0;
 178}
 179
 180static int git_config_rename(const char *var, const char *value)
 181{
 182        if (!value)
 183                return DIFF_DETECT_RENAME;
 184        if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 185                return  DIFF_DETECT_COPY;
 186        return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 187}
 188
 189long parse_algorithm_value(const char *value)
 190{
 191        if (!value)
 192                return -1;
 193        else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 194                return 0;
 195        else if (!strcasecmp(value, "minimal"))
 196                return XDF_NEED_MINIMAL;
 197        else if (!strcasecmp(value, "patience"))
 198                return XDF_PATIENCE_DIFF;
 199        else if (!strcasecmp(value, "histogram"))
 200                return XDF_HISTOGRAM_DIFF;
 201        return -1;
 202}
 203
 204static int parse_one_token(const char **arg, const char *token)
 205{
 206        const char *rest;
 207        if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 208                *arg = rest;
 209                return 1;
 210        }
 211        return 0;
 212}
 213
 214static int parse_ws_error_highlight(const char *arg)
 215{
 216        const char *orig_arg = arg;
 217        unsigned val = 0;
 218
 219        while (*arg) {
 220                if (parse_one_token(&arg, "none"))
 221                        val = 0;
 222                else if (parse_one_token(&arg, "default"))
 223                        val = WSEH_NEW;
 224                else if (parse_one_token(&arg, "all"))
 225                        val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 226                else if (parse_one_token(&arg, "new"))
 227                        val |= WSEH_NEW;
 228                else if (parse_one_token(&arg, "old"))
 229                        val |= WSEH_OLD;
 230                else if (parse_one_token(&arg, "context"))
 231                        val |= WSEH_CONTEXT;
 232                else {
 233                        return -1 - (int)(arg - orig_arg);
 234                }
 235                if (*arg)
 236                        arg++;
 237        }
 238        return val;
 239}
 240
 241/*
 242 * These are to give UI layer defaults.
 243 * The core-level commands such as git-diff-files should
 244 * never be affected by the setting of diff.renames
 245 * the user happens to have in the configuration file.
 246 */
 247void init_diff_ui_defaults(void)
 248{
 249        diff_detect_rename_default = DIFF_DETECT_RENAME;
 250}
 251
 252int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 253{
 254        if (!strcmp(var, "diff.indentheuristic"))
 255                diff_indent_heuristic = git_config_bool(var, value);
 256        return 0;
 257}
 258
 259static int parse_color_moved(const char *arg)
 260{
 261        switch (git_parse_maybe_bool(arg)) {
 262        case 0:
 263                return COLOR_MOVED_NO;
 264        case 1:
 265                return COLOR_MOVED_DEFAULT;
 266        default:
 267                break;
 268        }
 269
 270        if (!strcmp(arg, "no"))
 271                return COLOR_MOVED_NO;
 272        else if (!strcmp(arg, "plain"))
 273                return COLOR_MOVED_PLAIN;
 274        else if (!strcmp(arg, "blocks"))
 275                return COLOR_MOVED_BLOCKS;
 276        else if (!strcmp(arg, "zebra"))
 277                return COLOR_MOVED_ZEBRA;
 278        else if (!strcmp(arg, "default"))
 279                return COLOR_MOVED_DEFAULT;
 280        else if (!strcmp(arg, "dimmed_zebra"))
 281                return COLOR_MOVED_ZEBRA_DIM;
 282        else
 283                return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed_zebra', 'plain'"));
 284}
 285
 286int git_diff_ui_config(const char *var, const char *value, void *cb)
 287{
 288        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 289                diff_use_color_default = git_config_colorbool(var, value);
 290                return 0;
 291        }
 292        if (!strcmp(var, "diff.colormoved")) {
 293                int cm = parse_color_moved(value);
 294                if (cm < 0)
 295                        return -1;
 296                diff_color_moved_default = cm;
 297                return 0;
 298        }
 299        if (!strcmp(var, "diff.context")) {
 300                diff_context_default = git_config_int(var, value);
 301                if (diff_context_default < 0)
 302                        return -1;
 303                return 0;
 304        }
 305        if (!strcmp(var, "diff.interhunkcontext")) {
 306                diff_interhunk_context_default = git_config_int(var, value);
 307                if (diff_interhunk_context_default < 0)
 308                        return -1;
 309                return 0;
 310        }
 311        if (!strcmp(var, "diff.renames")) {
 312                diff_detect_rename_default = git_config_rename(var, value);
 313                return 0;
 314        }
 315        if (!strcmp(var, "diff.autorefreshindex")) {
 316                diff_auto_refresh_index = git_config_bool(var, value);
 317                return 0;
 318        }
 319        if (!strcmp(var, "diff.mnemonicprefix")) {
 320                diff_mnemonic_prefix = git_config_bool(var, value);
 321                return 0;
 322        }
 323        if (!strcmp(var, "diff.noprefix")) {
 324                diff_no_prefix = git_config_bool(var, value);
 325                return 0;
 326        }
 327        if (!strcmp(var, "diff.statgraphwidth")) {
 328                diff_stat_graph_width = git_config_int(var, value);
 329                return 0;
 330        }
 331        if (!strcmp(var, "diff.external"))
 332                return git_config_string(&external_diff_cmd_cfg, var, value);
 333        if (!strcmp(var, "diff.wordregex"))
 334                return git_config_string(&diff_word_regex_cfg, var, value);
 335        if (!strcmp(var, "diff.orderfile"))
 336                return git_config_pathname(&diff_order_file_cfg, var, value);
 337
 338        if (!strcmp(var, "diff.ignoresubmodules"))
 339                handle_ignore_submodules_arg(&default_diff_options, value);
 340
 341        if (!strcmp(var, "diff.submodule")) {
 342                if (parse_submodule_params(&default_diff_options, value))
 343                        warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 344                                value);
 345                return 0;
 346        }
 347
 348        if (!strcmp(var, "diff.algorithm")) {
 349                diff_algorithm = parse_algorithm_value(value);
 350                if (diff_algorithm < 0)
 351                        return -1;
 352                return 0;
 353        }
 354
 355        if (!strcmp(var, "diff.wserrorhighlight")) {
 356                int val = parse_ws_error_highlight(value);
 357                if (val < 0)
 358                        return -1;
 359                ws_error_highlight_default = val;
 360                return 0;
 361        }
 362
 363        if (git_color_config(var, value, cb) < 0)
 364                return -1;
 365
 366        return git_diff_basic_config(var, value, cb);
 367}
 368
 369int git_diff_basic_config(const char *var, const char *value, void *cb)
 370{
 371        const char *name;
 372
 373        if (!strcmp(var, "diff.renamelimit")) {
 374                diff_rename_limit_default = git_config_int(var, value);
 375                return 0;
 376        }
 377
 378        if (userdiff_config(var, value) < 0)
 379                return -1;
 380
 381        if (skip_prefix(var, "diff.color.", &name) ||
 382            skip_prefix(var, "color.diff.", &name)) {
 383                int slot = parse_diff_color_slot(name);
 384                if (slot < 0)
 385                        return 0;
 386                if (!value)
 387                        return config_error_nonbool(var);
 388                return color_parse(value, diff_colors[slot]);
 389        }
 390
 391        /* like GNU diff's --suppress-blank-empty option  */
 392        if (!strcmp(var, "diff.suppressblankempty") ||
 393                        /* for backwards compatibility */
 394                        !strcmp(var, "diff.suppress-blank-empty")) {
 395                diff_suppress_blank_empty = git_config_bool(var, value);
 396                return 0;
 397        }
 398
 399        if (!strcmp(var, "diff.dirstat")) {
 400                struct strbuf errmsg = STRBUF_INIT;
 401                default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 402                if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 403                        warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 404                                errmsg.buf);
 405                strbuf_release(&errmsg);
 406                diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 407                return 0;
 408        }
 409
 410        if (git_diff_heuristic_config(var, value, cb) < 0)
 411                return -1;
 412
 413        return git_default_config(var, value, cb);
 414}
 415
 416static char *quote_two(const char *one, const char *two)
 417{
 418        int need_one = quote_c_style(one, NULL, NULL, 1);
 419        int need_two = quote_c_style(two, NULL, NULL, 1);
 420        struct strbuf res = STRBUF_INIT;
 421
 422        if (need_one + need_two) {
 423                strbuf_addch(&res, '"');
 424                quote_c_style(one, &res, NULL, 1);
 425                quote_c_style(two, &res, NULL, 1);
 426                strbuf_addch(&res, '"');
 427        } else {
 428                strbuf_addstr(&res, one);
 429                strbuf_addstr(&res, two);
 430        }
 431        return strbuf_detach(&res, NULL);
 432}
 433
 434static const char *external_diff(void)
 435{
 436        static const char *external_diff_cmd = NULL;
 437        static int done_preparing = 0;
 438
 439        if (done_preparing)
 440                return external_diff_cmd;
 441        external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 442        if (!external_diff_cmd)
 443                external_diff_cmd = external_diff_cmd_cfg;
 444        done_preparing = 1;
 445        return external_diff_cmd;
 446}
 447
 448/*
 449 * Keep track of files used for diffing. Sometimes such an entry
 450 * refers to a temporary file, sometimes to an existing file, and
 451 * sometimes to "/dev/null".
 452 */
 453static struct diff_tempfile {
 454        /*
 455         * filename external diff should read from, or NULL if this
 456         * entry is currently not in use:
 457         */
 458        const char *name;
 459
 460        char hex[GIT_MAX_HEXSZ + 1];
 461        char mode[10];
 462
 463        /*
 464         * If this diff_tempfile instance refers to a temporary file,
 465         * this tempfile object is used to manage its lifetime.
 466         */
 467        struct tempfile *tempfile;
 468} diff_temp[2];
 469
 470struct emit_callback {
 471        int color_diff;
 472        unsigned ws_rule;
 473        int blank_at_eof_in_preimage;
 474        int blank_at_eof_in_postimage;
 475        int lno_in_preimage;
 476        int lno_in_postimage;
 477        const char **label_path;
 478        struct diff_words_data *diff_words;
 479        struct diff_options *opt;
 480        struct strbuf *header;
 481};
 482
 483static int count_lines(const char *data, int size)
 484{
 485        int count, ch, completely_empty = 1, nl_just_seen = 0;
 486        count = 0;
 487        while (0 < size--) {
 488                ch = *data++;
 489                if (ch == '\n') {
 490                        count++;
 491                        nl_just_seen = 1;
 492                        completely_empty = 0;
 493                }
 494                else {
 495                        nl_just_seen = 0;
 496                        completely_empty = 0;
 497                }
 498        }
 499        if (completely_empty)
 500                return 0;
 501        if (!nl_just_seen)
 502                count++; /* no trailing newline */
 503        return count;
 504}
 505
 506static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 507{
 508        if (!DIFF_FILE_VALID(one)) {
 509                mf->ptr = (char *)""; /* does not matter */
 510                mf->size = 0;
 511                return 0;
 512        }
 513        else if (diff_populate_filespec(one, 0))
 514                return -1;
 515
 516        mf->ptr = one->data;
 517        mf->size = one->size;
 518        return 0;
 519}
 520
 521/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 522static unsigned long diff_filespec_size(struct diff_filespec *one)
 523{
 524        if (!DIFF_FILE_VALID(one))
 525                return 0;
 526        diff_populate_filespec(one, CHECK_SIZE_ONLY);
 527        return one->size;
 528}
 529
 530static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 531{
 532        char *ptr = mf->ptr;
 533        long size = mf->size;
 534        int cnt = 0;
 535
 536        if (!size)
 537                return cnt;
 538        ptr += size - 1; /* pointing at the very end */
 539        if (*ptr != '\n')
 540                ; /* incomplete line */
 541        else
 542                ptr--; /* skip the last LF */
 543        while (mf->ptr < ptr) {
 544                char *prev_eol;
 545                for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 546                        if (*prev_eol == '\n')
 547                                break;
 548                if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 549                        break;
 550                cnt++;
 551                ptr = prev_eol - 1;
 552        }
 553        return cnt;
 554}
 555
 556static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 557                               struct emit_callback *ecbdata)
 558{
 559        int l1, l2, at;
 560        unsigned ws_rule = ecbdata->ws_rule;
 561        l1 = count_trailing_blank(mf1, ws_rule);
 562        l2 = count_trailing_blank(mf2, ws_rule);
 563        if (l2 <= l1) {
 564                ecbdata->blank_at_eof_in_preimage = 0;
 565                ecbdata->blank_at_eof_in_postimage = 0;
 566                return;
 567        }
 568        at = count_lines(mf1->ptr, mf1->size);
 569        ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 570
 571        at = count_lines(mf2->ptr, mf2->size);
 572        ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 573}
 574
 575static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 576                        int first, const char *line, int len)
 577{
 578        int has_trailing_newline, has_trailing_carriage_return;
 579        int nofirst;
 580        FILE *file = o->file;
 581
 582        fputs(diff_line_prefix(o), file);
 583
 584        if (len == 0) {
 585                has_trailing_newline = (first == '\n');
 586                has_trailing_carriage_return = (!has_trailing_newline &&
 587                                                (first == '\r'));
 588                nofirst = has_trailing_newline || has_trailing_carriage_return;
 589        } else {
 590                has_trailing_newline = (len > 0 && line[len-1] == '\n');
 591                if (has_trailing_newline)
 592                        len--;
 593                has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 594                if (has_trailing_carriage_return)
 595                        len--;
 596                nofirst = 0;
 597        }
 598
 599        if (len || !nofirst) {
 600                fputs(set, file);
 601                if (!nofirst)
 602                        fputc(first, file);
 603                fwrite(line, len, 1, file);
 604                fputs(reset, file);
 605        }
 606        if (has_trailing_carriage_return)
 607                fputc('\r', file);
 608        if (has_trailing_newline)
 609                fputc('\n', file);
 610}
 611
 612static void emit_line(struct diff_options *o, const char *set, const char *reset,
 613                      const char *line, int len)
 614{
 615        emit_line_0(o, set, reset, line[0], line+1, len-1);
 616}
 617
 618enum diff_symbol {
 619        DIFF_SYMBOL_BINARY_DIFF_HEADER,
 620        DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 621        DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 622        DIFF_SYMBOL_BINARY_DIFF_BODY,
 623        DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 624        DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 625        DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 626        DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 627        DIFF_SYMBOL_STATS_LINE,
 628        DIFF_SYMBOL_WORD_DIFF,
 629        DIFF_SYMBOL_STAT_SEP,
 630        DIFF_SYMBOL_SUMMARY,
 631        DIFF_SYMBOL_SUBMODULE_ADD,
 632        DIFF_SYMBOL_SUBMODULE_DEL,
 633        DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 634        DIFF_SYMBOL_SUBMODULE_MODIFIED,
 635        DIFF_SYMBOL_SUBMODULE_HEADER,
 636        DIFF_SYMBOL_SUBMODULE_ERROR,
 637        DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 638        DIFF_SYMBOL_REWRITE_DIFF,
 639        DIFF_SYMBOL_BINARY_FILES,
 640        DIFF_SYMBOL_HEADER,
 641        DIFF_SYMBOL_FILEPAIR_PLUS,
 642        DIFF_SYMBOL_FILEPAIR_MINUS,
 643        DIFF_SYMBOL_WORDS_PORCELAIN,
 644        DIFF_SYMBOL_WORDS,
 645        DIFF_SYMBOL_CONTEXT,
 646        DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 647        DIFF_SYMBOL_PLUS,
 648        DIFF_SYMBOL_MINUS,
 649        DIFF_SYMBOL_NO_LF_EOF,
 650        DIFF_SYMBOL_CONTEXT_FRAGINFO,
 651        DIFF_SYMBOL_CONTEXT_MARKER,
 652        DIFF_SYMBOL_SEPARATOR
 653};
 654/*
 655 * Flags for content lines:
 656 * 0..12 are whitespace rules
 657 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 658 * 16 is marking if the line is blank at EOF
 659 */
 660#define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 661#define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 662#define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 663#define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 664#define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 665
 666/*
 667 * This struct is used when we need to buffer the output of the diff output.
 668 *
 669 * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 670 * into the pre/post image file. This pointer could be a union with the
 671 * line pointer. By storing an offset into the file instead of the literal line,
 672 * we can decrease the memory footprint for the buffered output. At first we
 673 * may want to only have indirection for the content lines, but we could also
 674 * enhance the state for emitting prefabricated lines, e.g. the similarity
 675 * score line or hunk/file headers would only need to store a number or path
 676 * and then the output can be constructed later on depending on state.
 677 */
 678struct emitted_diff_symbol {
 679        const char *line;
 680        int len;
 681        int flags;
 682        enum diff_symbol s;
 683};
 684#define EMITTED_DIFF_SYMBOL_INIT {NULL}
 685
 686struct emitted_diff_symbols {
 687        struct emitted_diff_symbol *buf;
 688        int nr, alloc;
 689};
 690#define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 691
 692static void append_emitted_diff_symbol(struct diff_options *o,
 693                                       struct emitted_diff_symbol *e)
 694{
 695        struct emitted_diff_symbol *f;
 696
 697        ALLOC_GROW(o->emitted_symbols->buf,
 698                   o->emitted_symbols->nr + 1,
 699                   o->emitted_symbols->alloc);
 700        f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 701
 702        memcpy(f, e, sizeof(struct emitted_diff_symbol));
 703        f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 704}
 705
 706struct moved_entry {
 707        struct hashmap_entry ent;
 708        const struct emitted_diff_symbol *es;
 709        struct moved_entry *next_line;
 710};
 711
 712static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 713                           const void *entry,
 714                           const void *entry_or_key,
 715                           const void *keydata)
 716{
 717        const struct diff_options *diffopt = hashmap_cmp_fn_data;
 718        const struct moved_entry *a = entry;
 719        const struct moved_entry *b = entry_or_key;
 720
 721        return !xdiff_compare_lines(a->es->line, a->es->len,
 722                                    b->es->line, b->es->len,
 723                                    diffopt->xdl_opts);
 724}
 725
 726static struct moved_entry *prepare_entry(struct diff_options *o,
 727                                         int line_no)
 728{
 729        struct moved_entry *ret = xmalloc(sizeof(*ret));
 730        struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 731
 732        ret->ent.hash = xdiff_hash_string(l->line, l->len, o->xdl_opts);
 733        ret->es = l;
 734        ret->next_line = NULL;
 735
 736        return ret;
 737}
 738
 739static void add_lines_to_move_detection(struct diff_options *o,
 740                                        struct hashmap *add_lines,
 741                                        struct hashmap *del_lines)
 742{
 743        struct moved_entry *prev_line = NULL;
 744
 745        int n;
 746        for (n = 0; n < o->emitted_symbols->nr; n++) {
 747                struct hashmap *hm;
 748                struct moved_entry *key;
 749
 750                switch (o->emitted_symbols->buf[n].s) {
 751                case DIFF_SYMBOL_PLUS:
 752                        hm = add_lines;
 753                        break;
 754                case DIFF_SYMBOL_MINUS:
 755                        hm = del_lines;
 756                        break;
 757                default:
 758                        prev_line = NULL;
 759                        continue;
 760                }
 761
 762                key = prepare_entry(o, n);
 763                if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 764                        prev_line->next_line = key;
 765
 766                hashmap_add(hm, key);
 767                prev_line = key;
 768        }
 769}
 770
 771static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 772                                         int pmb_nr)
 773{
 774        int lp, rp;
 775
 776        /* Shrink the set of potential block to the remaining running */
 777        for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 778                while (lp < pmb_nr && pmb[lp])
 779                        lp++;
 780                /* lp points at the first NULL now */
 781
 782                while (rp > -1 && !pmb[rp])
 783                        rp--;
 784                /* rp points at the last non-NULL */
 785
 786                if (lp < pmb_nr && rp > -1 && lp < rp) {
 787                        pmb[lp] = pmb[rp];
 788                        pmb[rp] = NULL;
 789                        rp--;
 790                        lp++;
 791                }
 792        }
 793
 794        /* Remember the number of running sets */
 795        return rp + 1;
 796}
 797
 798/*
 799 * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
 800 *
 801 * Otherwise, if the last block has fewer alphanumeric characters than
 802 * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
 803 * that block.
 804 *
 805 * The last block consists of the (n - block_length)'th line up to but not
 806 * including the nth line.
 807 *
 808 * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
 809 * Think of a way to unify them.
 810 */
 811static void adjust_last_block(struct diff_options *o, int n, int block_length)
 812{
 813        int i, alnum_count = 0;
 814        if (o->color_moved == COLOR_MOVED_PLAIN)
 815                return;
 816        for (i = 1; i < block_length + 1; i++) {
 817                const char *c = o->emitted_symbols->buf[n - i].line;
 818                for (; *c; c++) {
 819                        if (!isalnum(*c))
 820                                continue;
 821                        alnum_count++;
 822                        if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
 823                                return;
 824                }
 825        }
 826        for (i = 1; i < block_length + 1; i++)
 827                o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
 828}
 829
 830/* Find blocks of moved code, delegate actual coloring decision to helper */
 831static void mark_color_as_moved(struct diff_options *o,
 832                                struct hashmap *add_lines,
 833                                struct hashmap *del_lines)
 834{
 835        struct moved_entry **pmb = NULL; /* potentially moved blocks */
 836        int pmb_nr = 0, pmb_alloc = 0;
 837        int n, flipped_block = 1, block_length = 0;
 838
 839
 840        for (n = 0; n < o->emitted_symbols->nr; n++) {
 841                struct hashmap *hm = NULL;
 842                struct moved_entry *key;
 843                struct moved_entry *match = NULL;
 844                struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 845                int i;
 846
 847                switch (l->s) {
 848                case DIFF_SYMBOL_PLUS:
 849                        hm = del_lines;
 850                        key = prepare_entry(o, n);
 851                        match = hashmap_get(hm, key, NULL);
 852                        free(key);
 853                        break;
 854                case DIFF_SYMBOL_MINUS:
 855                        hm = add_lines;
 856                        key = prepare_entry(o, n);
 857                        match = hashmap_get(hm, key, NULL);
 858                        free(key);
 859                        break;
 860                default:
 861                        flipped_block = 1;
 862                }
 863
 864                if (!match) {
 865                        adjust_last_block(o, n, block_length);
 866                        pmb_nr = 0;
 867                        block_length = 0;
 868                        continue;
 869                }
 870
 871                l->flags |= DIFF_SYMBOL_MOVED_LINE;
 872
 873                if (o->color_moved == COLOR_MOVED_PLAIN)
 874                        continue;
 875
 876                /* Check any potential block runs, advance each or nullify */
 877                for (i = 0; i < pmb_nr; i++) {
 878                        struct moved_entry *p = pmb[i];
 879                        struct moved_entry *pnext = (p && p->next_line) ?
 880                                        p->next_line : NULL;
 881                        if (pnext && !hm->cmpfn(o, pnext, match, NULL)) {
 882                                pmb[i] = p->next_line;
 883                        } else {
 884                                pmb[i] = NULL;
 885                        }
 886                }
 887
 888                pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
 889
 890                if (pmb_nr == 0) {
 891                        /*
 892                         * The current line is the start of a new block.
 893                         * Setup the set of potential blocks.
 894                         */
 895                        for (; match; match = hashmap_get_next(hm, match)) {
 896                                ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
 897                                pmb[pmb_nr++] = match;
 898                        }
 899
 900                        flipped_block = (flipped_block + 1) % 2;
 901
 902                        adjust_last_block(o, n, block_length);
 903                        block_length = 0;
 904                }
 905
 906                block_length++;
 907
 908                if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
 909                        l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
 910        }
 911        adjust_last_block(o, n, block_length);
 912
 913        free(pmb);
 914}
 915
 916#define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
 917  (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
 918static void dim_moved_lines(struct diff_options *o)
 919{
 920        int n;
 921        for (n = 0; n < o->emitted_symbols->nr; n++) {
 922                struct emitted_diff_symbol *prev = (n != 0) ?
 923                                &o->emitted_symbols->buf[n - 1] : NULL;
 924                struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 925                struct emitted_diff_symbol *next =
 926                                (n < o->emitted_symbols->nr - 1) ?
 927                                &o->emitted_symbols->buf[n + 1] : NULL;
 928
 929                /* Not a plus or minus line? */
 930                if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
 931                        continue;
 932
 933                /* Not a moved line? */
 934                if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
 935                        continue;
 936
 937                /*
 938                 * If prev or next are not a plus or minus line,
 939                 * pretend they don't exist
 940                 */
 941                if (prev && prev->s != DIFF_SYMBOL_PLUS &&
 942                            prev->s != DIFF_SYMBOL_MINUS)
 943                        prev = NULL;
 944                if (next && next->s != DIFF_SYMBOL_PLUS &&
 945                            next->s != DIFF_SYMBOL_MINUS)
 946                        next = NULL;
 947
 948                /* Inside a block? */
 949                if ((prev &&
 950                    (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
 951                    (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
 952                    (next &&
 953                    (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
 954                    (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
 955                        l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
 956                        continue;
 957                }
 958
 959                /* Check if we are at an interesting bound: */
 960                if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
 961                    (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
 962                       (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
 963                        continue;
 964                if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
 965                    (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
 966                       (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
 967                        continue;
 968
 969                /*
 970                 * The boundary to prev and next are not interesting,
 971                 * so this line is not interesting as a whole
 972                 */
 973                l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
 974        }
 975}
 976
 977static void emit_line_ws_markup(struct diff_options *o,
 978                                const char *set, const char *reset,
 979                                const char *line, int len, char sign,
 980                                unsigned ws_rule, int blank_at_eof)
 981{
 982        const char *ws = NULL;
 983
 984        if (o->ws_error_highlight & ws_rule) {
 985                ws = diff_get_color_opt(o, DIFF_WHITESPACE);
 986                if (!*ws)
 987                        ws = NULL;
 988        }
 989
 990        if (!ws)
 991                emit_line_0(o, set, reset, sign, line, len);
 992        else if (blank_at_eof)
 993                /* Blank line at EOF - paint '+' as well */
 994                emit_line_0(o, ws, reset, sign, line, len);
 995        else {
 996                /* Emit just the prefix, then the rest. */
 997                emit_line_0(o, set, reset, sign, "", 0);
 998                ws_check_emit(line, len, ws_rule,
 999                              o->file, set, reset, ws);
1000        }
1001}
1002
1003static void emit_diff_symbol_from_struct(struct diff_options *o,
1004                                         struct emitted_diff_symbol *eds)
1005{
1006        static const char *nneof = " No newline at end of file\n";
1007        const char *context, *reset, *set, *meta, *fraginfo;
1008        struct strbuf sb = STRBUF_INIT;
1009
1010        enum diff_symbol s = eds->s;
1011        const char *line = eds->line;
1012        int len = eds->len;
1013        unsigned flags = eds->flags;
1014
1015        switch (s) {
1016        case DIFF_SYMBOL_NO_LF_EOF:
1017                context = diff_get_color_opt(o, DIFF_CONTEXT);
1018                reset = diff_get_color_opt(o, DIFF_RESET);
1019                putc('\n', o->file);
1020                emit_line_0(o, context, reset, '\\',
1021                            nneof, strlen(nneof));
1022                break;
1023        case DIFF_SYMBOL_SUBMODULE_HEADER:
1024        case DIFF_SYMBOL_SUBMODULE_ERROR:
1025        case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1026        case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1027        case DIFF_SYMBOL_SUMMARY:
1028        case DIFF_SYMBOL_STATS_LINE:
1029        case DIFF_SYMBOL_BINARY_DIFF_BODY:
1030        case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1031                emit_line(o, "", "", line, len);
1032                break;
1033        case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1034        case DIFF_SYMBOL_CONTEXT_MARKER:
1035                context = diff_get_color_opt(o, DIFF_CONTEXT);
1036                reset = diff_get_color_opt(o, DIFF_RESET);
1037                emit_line(o, context, reset, line, len);
1038                break;
1039        case DIFF_SYMBOL_SEPARATOR:
1040                fprintf(o->file, "%s%c",
1041                        diff_line_prefix(o),
1042                        o->line_termination);
1043                break;
1044        case DIFF_SYMBOL_CONTEXT:
1045                set = diff_get_color_opt(o, DIFF_CONTEXT);
1046                reset = diff_get_color_opt(o, DIFF_RESET);
1047                emit_line_ws_markup(o, set, reset, line, len, ' ',
1048                                    flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1049                break;
1050        case DIFF_SYMBOL_PLUS:
1051                switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1052                                 DIFF_SYMBOL_MOVED_LINE_ALT |
1053                                 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1054                case DIFF_SYMBOL_MOVED_LINE |
1055                     DIFF_SYMBOL_MOVED_LINE_ALT |
1056                     DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1057                        set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1058                        break;
1059                case DIFF_SYMBOL_MOVED_LINE |
1060                     DIFF_SYMBOL_MOVED_LINE_ALT:
1061                        set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1062                        break;
1063                case DIFF_SYMBOL_MOVED_LINE |
1064                     DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1065                        set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1066                        break;
1067                case DIFF_SYMBOL_MOVED_LINE:
1068                        set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1069                        break;
1070                default:
1071                        set = diff_get_color_opt(o, DIFF_FILE_NEW);
1072                }
1073                reset = diff_get_color_opt(o, DIFF_RESET);
1074                emit_line_ws_markup(o, set, reset, line, len, '+',
1075                                    flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1076                                    flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1077                break;
1078        case DIFF_SYMBOL_MINUS:
1079                switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1080                                 DIFF_SYMBOL_MOVED_LINE_ALT |
1081                                 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1082                case DIFF_SYMBOL_MOVED_LINE |
1083                     DIFF_SYMBOL_MOVED_LINE_ALT |
1084                     DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1085                        set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1086                        break;
1087                case DIFF_SYMBOL_MOVED_LINE |
1088                     DIFF_SYMBOL_MOVED_LINE_ALT:
1089                        set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1090                        break;
1091                case DIFF_SYMBOL_MOVED_LINE |
1092                     DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1093                        set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1094                        break;
1095                case DIFF_SYMBOL_MOVED_LINE:
1096                        set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1097                        break;
1098                default:
1099                        set = diff_get_color_opt(o, DIFF_FILE_OLD);
1100                }
1101                reset = diff_get_color_opt(o, DIFF_RESET);
1102                emit_line_ws_markup(o, set, reset, line, len, '-',
1103                                    flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1104                break;
1105        case DIFF_SYMBOL_WORDS_PORCELAIN:
1106                context = diff_get_color_opt(o, DIFF_CONTEXT);
1107                reset = diff_get_color_opt(o, DIFF_RESET);
1108                emit_line(o, context, reset, line, len);
1109                fputs("~\n", o->file);
1110                break;
1111        case DIFF_SYMBOL_WORDS:
1112                context = diff_get_color_opt(o, DIFF_CONTEXT);
1113                reset = diff_get_color_opt(o, DIFF_RESET);
1114                /*
1115                 * Skip the prefix character, if any.  With
1116                 * diff_suppress_blank_empty, there may be
1117                 * none.
1118                 */
1119                if (line[0] != '\n') {
1120                        line++;
1121                        len--;
1122                }
1123                emit_line(o, context, reset, line, len);
1124                break;
1125        case DIFF_SYMBOL_FILEPAIR_PLUS:
1126                meta = diff_get_color_opt(o, DIFF_METAINFO);
1127                reset = diff_get_color_opt(o, DIFF_RESET);
1128                fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1129                        line, reset,
1130                        strchr(line, ' ') ? "\t" : "");
1131                break;
1132        case DIFF_SYMBOL_FILEPAIR_MINUS:
1133                meta = diff_get_color_opt(o, DIFF_METAINFO);
1134                reset = diff_get_color_opt(o, DIFF_RESET);
1135                fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1136                        line, reset,
1137                        strchr(line, ' ') ? "\t" : "");
1138                break;
1139        case DIFF_SYMBOL_BINARY_FILES:
1140        case DIFF_SYMBOL_HEADER:
1141                fprintf(o->file, "%s", line);
1142                break;
1143        case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1144                fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1145                break;
1146        case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1147                fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1148                break;
1149        case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1150                fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1151                break;
1152        case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1153                fputs(diff_line_prefix(o), o->file);
1154                fputc('\n', o->file);
1155                break;
1156        case DIFF_SYMBOL_REWRITE_DIFF:
1157                fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1158                reset = diff_get_color_opt(o, DIFF_RESET);
1159                emit_line(o, fraginfo, reset, line, len);
1160                break;
1161        case DIFF_SYMBOL_SUBMODULE_ADD:
1162                set = diff_get_color_opt(o, DIFF_FILE_NEW);
1163                reset = diff_get_color_opt(o, DIFF_RESET);
1164                emit_line(o, set, reset, line, len);
1165                break;
1166        case DIFF_SYMBOL_SUBMODULE_DEL:
1167                set = diff_get_color_opt(o, DIFF_FILE_OLD);
1168                reset = diff_get_color_opt(o, DIFF_RESET);
1169                emit_line(o, set, reset, line, len);
1170                break;
1171        case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1172                fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1173                        diff_line_prefix(o), line);
1174                break;
1175        case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1176                fprintf(o->file, "%sSubmodule %s contains modified content\n",
1177                        diff_line_prefix(o), line);
1178                break;
1179        case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1180                emit_line(o, "", "", " 0 files changed\n",
1181                          strlen(" 0 files changed\n"));
1182                break;
1183        case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1184                emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1185                break;
1186        case DIFF_SYMBOL_WORD_DIFF:
1187                fprintf(o->file, "%.*s", len, line);
1188                break;
1189        case DIFF_SYMBOL_STAT_SEP:
1190                fputs(o->stat_sep, o->file);
1191                break;
1192        default:
1193                die("BUG: unknown diff symbol");
1194        }
1195        strbuf_release(&sb);
1196}
1197
1198static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1199                             const char *line, int len, unsigned flags)
1200{
1201        struct emitted_diff_symbol e = {line, len, flags, s};
1202
1203        if (o->emitted_symbols)
1204                append_emitted_diff_symbol(o, &e);
1205        else
1206                emit_diff_symbol_from_struct(o, &e);
1207}
1208
1209void diff_emit_submodule_del(struct diff_options *o, const char *line)
1210{
1211        emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1212}
1213
1214void diff_emit_submodule_add(struct diff_options *o, const char *line)
1215{
1216        emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1217}
1218
1219void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1220{
1221        emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1222                         path, strlen(path), 0);
1223}
1224
1225void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1226{
1227        emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1228                         path, strlen(path), 0);
1229}
1230
1231void diff_emit_submodule_header(struct diff_options *o, const char *header)
1232{
1233        emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1234                         header, strlen(header), 0);
1235}
1236
1237void diff_emit_submodule_error(struct diff_options *o, const char *err)
1238{
1239        emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1240}
1241
1242void diff_emit_submodule_pipethrough(struct diff_options *o,
1243                                     const char *line, int len)
1244{
1245        emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1246}
1247
1248static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1249{
1250        if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1251              ecbdata->blank_at_eof_in_preimage &&
1252              ecbdata->blank_at_eof_in_postimage &&
1253              ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1254              ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1255                return 0;
1256        return ws_blank_line(line, len, ecbdata->ws_rule);
1257}
1258
1259static void emit_add_line(const char *reset,
1260                          struct emit_callback *ecbdata,
1261                          const char *line, int len)
1262{
1263        unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1264        if (new_blank_line_at_eof(ecbdata, line, len))
1265                flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1266
1267        emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1268}
1269
1270static void emit_del_line(const char *reset,
1271                          struct emit_callback *ecbdata,
1272                          const char *line, int len)
1273{
1274        unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1275        emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1276}
1277
1278static void emit_context_line(const char *reset,
1279                              struct emit_callback *ecbdata,
1280                              const char *line, int len)
1281{
1282        unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1283        emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1284}
1285
1286static void emit_hunk_header(struct emit_callback *ecbdata,
1287                             const char *line, int len)
1288{
1289        const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1290        const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1291        const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1292        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1293        static const char atat[2] = { '@', '@' };
1294        const char *cp, *ep;
1295        struct strbuf msgbuf = STRBUF_INIT;
1296        int org_len = len;
1297        int i = 1;
1298
1299        /*
1300         * As a hunk header must begin with "@@ -<old>, +<new> @@",
1301         * it always is at least 10 bytes long.
1302         */
1303        if (len < 10 ||
1304            memcmp(line, atat, 2) ||
1305            !(ep = memmem(line + 2, len - 2, atat, 2))) {
1306                emit_diff_symbol(ecbdata->opt,
1307                                 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1308                return;
1309        }
1310        ep += 2; /* skip over @@ */
1311
1312        /* The hunk header in fraginfo color */
1313        strbuf_addstr(&msgbuf, frag);
1314        strbuf_add(&msgbuf, line, ep - line);
1315        strbuf_addstr(&msgbuf, reset);
1316
1317        /*
1318         * trailing "\r\n"
1319         */
1320        for ( ; i < 3; i++)
1321                if (line[len - i] == '\r' || line[len - i] == '\n')
1322                        len--;
1323
1324        /* blank before the func header */
1325        for (cp = ep; ep - line < len; ep++)
1326                if (*ep != ' ' && *ep != '\t')
1327                        break;
1328        if (ep != cp) {
1329                strbuf_addstr(&msgbuf, context);
1330                strbuf_add(&msgbuf, cp, ep - cp);
1331                strbuf_addstr(&msgbuf, reset);
1332        }
1333
1334        if (ep < line + len) {
1335                strbuf_addstr(&msgbuf, func);
1336                strbuf_add(&msgbuf, ep, line + len - ep);
1337                strbuf_addstr(&msgbuf, reset);
1338        }
1339
1340        strbuf_add(&msgbuf, line + len, org_len - len);
1341        strbuf_complete_line(&msgbuf);
1342        emit_diff_symbol(ecbdata->opt,
1343                         DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1344        strbuf_release(&msgbuf);
1345}
1346
1347static struct diff_tempfile *claim_diff_tempfile(void) {
1348        int i;
1349        for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1350                if (!diff_temp[i].name)
1351                        return diff_temp + i;
1352        die("BUG: diff is failing to clean up its tempfiles");
1353}
1354
1355static void remove_tempfile(void)
1356{
1357        int i;
1358        for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1359                if (is_tempfile_active(diff_temp[i].tempfile))
1360                        delete_tempfile(&diff_temp[i].tempfile);
1361                diff_temp[i].name = NULL;
1362        }
1363}
1364
1365static void add_line_count(struct strbuf *out, int count)
1366{
1367        switch (count) {
1368        case 0:
1369                strbuf_addstr(out, "0,0");
1370                break;
1371        case 1:
1372                strbuf_addstr(out, "1");
1373                break;
1374        default:
1375                strbuf_addf(out, "1,%d", count);
1376                break;
1377        }
1378}
1379
1380static void emit_rewrite_lines(struct emit_callback *ecb,
1381                               int prefix, const char *data, int size)
1382{
1383        const char *endp = NULL;
1384        const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1385
1386        while (0 < size) {
1387                int len;
1388
1389                endp = memchr(data, '\n', size);
1390                len = endp ? (endp - data + 1) : size;
1391                if (prefix != '+') {
1392                        ecb->lno_in_preimage++;
1393                        emit_del_line(reset, ecb, data, len);
1394                } else {
1395                        ecb->lno_in_postimage++;
1396                        emit_add_line(reset, ecb, data, len);
1397                }
1398                size -= len;
1399                data += len;
1400        }
1401        if (!endp)
1402                emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1403}
1404
1405static void emit_rewrite_diff(const char *name_a,
1406                              const char *name_b,
1407                              struct diff_filespec *one,
1408                              struct diff_filespec *two,
1409                              struct userdiff_driver *textconv_one,
1410                              struct userdiff_driver *textconv_two,
1411                              struct diff_options *o)
1412{
1413        int lc_a, lc_b;
1414        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1415        const char *a_prefix, *b_prefix;
1416        char *data_one, *data_two;
1417        size_t size_one, size_two;
1418        struct emit_callback ecbdata;
1419        struct strbuf out = STRBUF_INIT;
1420
1421        if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1422                a_prefix = o->b_prefix;
1423                b_prefix = o->a_prefix;
1424        } else {
1425                a_prefix = o->a_prefix;
1426                b_prefix = o->b_prefix;
1427        }
1428
1429        name_a += (*name_a == '/');
1430        name_b += (*name_b == '/');
1431
1432        strbuf_reset(&a_name);
1433        strbuf_reset(&b_name);
1434        quote_two_c_style(&a_name, a_prefix, name_a, 0);
1435        quote_two_c_style(&b_name, b_prefix, name_b, 0);
1436
1437        size_one = fill_textconv(textconv_one, one, &data_one);
1438        size_two = fill_textconv(textconv_two, two, &data_two);
1439
1440        memset(&ecbdata, 0, sizeof(ecbdata));
1441        ecbdata.color_diff = want_color(o->use_color);
1442        ecbdata.ws_rule = whitespace_rule(name_b);
1443        ecbdata.opt = o;
1444        if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1445                mmfile_t mf1, mf2;
1446                mf1.ptr = (char *)data_one;
1447                mf2.ptr = (char *)data_two;
1448                mf1.size = size_one;
1449                mf2.size = size_two;
1450                check_blank_at_eof(&mf1, &mf2, &ecbdata);
1451        }
1452        ecbdata.lno_in_preimage = 1;
1453        ecbdata.lno_in_postimage = 1;
1454
1455        lc_a = count_lines(data_one, size_one);
1456        lc_b = count_lines(data_two, size_two);
1457
1458        emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1459                         a_name.buf, a_name.len, 0);
1460        emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1461                         b_name.buf, b_name.len, 0);
1462
1463        strbuf_addstr(&out, "@@ -");
1464        if (!o->irreversible_delete)
1465                add_line_count(&out, lc_a);
1466        else
1467                strbuf_addstr(&out, "?,?");
1468        strbuf_addstr(&out, " +");
1469        add_line_count(&out, lc_b);
1470        strbuf_addstr(&out, " @@\n");
1471        emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1472        strbuf_release(&out);
1473
1474        if (lc_a && !o->irreversible_delete)
1475                emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1476        if (lc_b)
1477                emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1478        if (textconv_one)
1479                free((char *)data_one);
1480        if (textconv_two)
1481                free((char *)data_two);
1482}
1483
1484struct diff_words_buffer {
1485        mmfile_t text;
1486        unsigned long alloc;
1487        struct diff_words_orig {
1488                const char *begin, *end;
1489        } *orig;
1490        int orig_nr, orig_alloc;
1491};
1492
1493static void diff_words_append(char *line, unsigned long len,
1494                struct diff_words_buffer *buffer)
1495{
1496        ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1497        line++;
1498        len--;
1499        memcpy(buffer->text.ptr + buffer->text.size, line, len);
1500        buffer->text.size += len;
1501        buffer->text.ptr[buffer->text.size] = '\0';
1502}
1503
1504struct diff_words_style_elem {
1505        const char *prefix;
1506        const char *suffix;
1507        const char *color; /* NULL; filled in by the setup code if
1508                            * color is enabled */
1509};
1510
1511struct diff_words_style {
1512        enum diff_words_type type;
1513        struct diff_words_style_elem new_word, old_word, ctx;
1514        const char *newline;
1515};
1516
1517static struct diff_words_style diff_words_styles[] = {
1518        { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1519        { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1520        { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1521};
1522
1523struct diff_words_data {
1524        struct diff_words_buffer minus, plus;
1525        const char *current_plus;
1526        int last_minus;
1527        struct diff_options *opt;
1528        regex_t *word_regex;
1529        enum diff_words_type type;
1530        struct diff_words_style *style;
1531};
1532
1533static int fn_out_diff_words_write_helper(struct diff_options *o,
1534                                          struct diff_words_style_elem *st_el,
1535                                          const char *newline,
1536                                          size_t count, const char *buf)
1537{
1538        int print = 0;
1539        struct strbuf sb = STRBUF_INIT;
1540
1541        while (count) {
1542                char *p = memchr(buf, '\n', count);
1543                if (print)
1544                        strbuf_addstr(&sb, diff_line_prefix(o));
1545
1546                if (p != buf) {
1547                        const char *reset = st_el->color && *st_el->color ?
1548                                            GIT_COLOR_RESET : NULL;
1549                        if (st_el->color && *st_el->color)
1550                                strbuf_addstr(&sb, st_el->color);
1551                        strbuf_addstr(&sb, st_el->prefix);
1552                        strbuf_add(&sb, buf, p ? p - buf : count);
1553                        strbuf_addstr(&sb, st_el->suffix);
1554                        if (reset)
1555                                strbuf_addstr(&sb, reset);
1556                }
1557                if (!p)
1558                        goto out;
1559
1560                strbuf_addstr(&sb, newline);
1561                count -= p + 1 - buf;
1562                buf = p + 1;
1563                print = 1;
1564                if (count) {
1565                        emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1566                                         sb.buf, sb.len, 0);
1567                        strbuf_reset(&sb);
1568                }
1569        }
1570
1571out:
1572        if (sb.len)
1573                emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1574                                 sb.buf, sb.len, 0);
1575        strbuf_release(&sb);
1576        return 0;
1577}
1578
1579/*
1580 * '--color-words' algorithm can be described as:
1581 *
1582 *   1. collect the minus/plus lines of a diff hunk, divided into
1583 *      minus-lines and plus-lines;
1584 *
1585 *   2. break both minus-lines and plus-lines into words and
1586 *      place them into two mmfile_t with one word for each line;
1587 *
1588 *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1589 *
1590 * And for the common parts of the both file, we output the plus side text.
1591 * diff_words->current_plus is used to trace the current position of the plus file
1592 * which printed. diff_words->last_minus is used to trace the last minus word
1593 * printed.
1594 *
1595 * For '--graph' to work with '--color-words', we need to output the graph prefix
1596 * on each line of color words output. Generally, there are two conditions on
1597 * which we should output the prefix.
1598 *
1599 *   1. diff_words->last_minus == 0 &&
1600 *      diff_words->current_plus == diff_words->plus.text.ptr
1601 *
1602 *      that is: the plus text must start as a new line, and if there is no minus
1603 *      word printed, a graph prefix must be printed.
1604 *
1605 *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1606 *      *(diff_words->current_plus - 1) == '\n'
1607 *
1608 *      that is: a graph prefix must be printed following a '\n'
1609 */
1610static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1611{
1612        if ((diff_words->last_minus == 0 &&
1613                diff_words->current_plus == diff_words->plus.text.ptr) ||
1614                (diff_words->current_plus > diff_words->plus.text.ptr &&
1615                *(diff_words->current_plus - 1) == '\n')) {
1616                return 1;
1617        } else {
1618                return 0;
1619        }
1620}
1621
1622static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1623{
1624        struct diff_words_data *diff_words = priv;
1625        struct diff_words_style *style = diff_words->style;
1626        int minus_first, minus_len, plus_first, plus_len;
1627        const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1628        struct diff_options *opt = diff_words->opt;
1629        const char *line_prefix;
1630
1631        if (line[0] != '@' || parse_hunk_header(line, len,
1632                        &minus_first, &minus_len, &plus_first, &plus_len))
1633                return;
1634
1635        assert(opt);
1636        line_prefix = diff_line_prefix(opt);
1637
1638        /* POSIX requires that first be decremented by one if len == 0... */
1639        if (minus_len) {
1640                minus_begin = diff_words->minus.orig[minus_first].begin;
1641                minus_end =
1642                        diff_words->minus.orig[minus_first + minus_len - 1].end;
1643        } else
1644                minus_begin = minus_end =
1645                        diff_words->minus.orig[minus_first].end;
1646
1647        if (plus_len) {
1648                plus_begin = diff_words->plus.orig[plus_first].begin;
1649                plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1650        } else
1651                plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1652
1653        if (color_words_output_graph_prefix(diff_words)) {
1654                fputs(line_prefix, diff_words->opt->file);
1655        }
1656        if (diff_words->current_plus != plus_begin) {
1657                fn_out_diff_words_write_helper(diff_words->opt,
1658                                &style->ctx, style->newline,
1659                                plus_begin - diff_words->current_plus,
1660                                diff_words->current_plus);
1661        }
1662        if (minus_begin != minus_end) {
1663                fn_out_diff_words_write_helper(diff_words->opt,
1664                                &style->old_word, style->newline,
1665                                minus_end - minus_begin, minus_begin);
1666        }
1667        if (plus_begin != plus_end) {
1668                fn_out_diff_words_write_helper(diff_words->opt,
1669                                &style->new_word, style->newline,
1670                                plus_end - plus_begin, plus_begin);
1671        }
1672
1673        diff_words->current_plus = plus_end;
1674        diff_words->last_minus = minus_first;
1675}
1676
1677/* This function starts looking at *begin, and returns 0 iff a word was found. */
1678static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1679                int *begin, int *end)
1680{
1681        if (word_regex && *begin < buffer->size) {
1682                regmatch_t match[1];
1683                if (!regexec_buf(word_regex, buffer->ptr + *begin,
1684                                 buffer->size - *begin, 1, match, 0)) {
1685                        char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1686                                        '\n', match[0].rm_eo - match[0].rm_so);
1687                        *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1688                        *begin += match[0].rm_so;
1689                        return *begin >= *end;
1690                }
1691                return -1;
1692        }
1693
1694        /* find the next word */
1695        while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1696                (*begin)++;
1697        if (*begin >= buffer->size)
1698                return -1;
1699
1700        /* find the end of the word */
1701        *end = *begin + 1;
1702        while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1703                (*end)++;
1704
1705        return 0;
1706}
1707
1708/*
1709 * This function splits the words in buffer->text, stores the list with
1710 * newline separator into out, and saves the offsets of the original words
1711 * in buffer->orig.
1712 */
1713static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1714                regex_t *word_regex)
1715{
1716        int i, j;
1717        long alloc = 0;
1718
1719        out->size = 0;
1720        out->ptr = NULL;
1721
1722        /* fake an empty "0th" word */
1723        ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1724        buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1725        buffer->orig_nr = 1;
1726
1727        for (i = 0; i < buffer->text.size; i++) {
1728                if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1729                        return;
1730
1731                /* store original boundaries */
1732                ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1733                                buffer->orig_alloc);
1734                buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1735                buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1736                buffer->orig_nr++;
1737
1738                /* store one word */
1739                ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1740                memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1741                out->ptr[out->size + j - i] = '\n';
1742                out->size += j - i + 1;
1743
1744                i = j - 1;
1745        }
1746}
1747
1748/* this executes the word diff on the accumulated buffers */
1749static void diff_words_show(struct diff_words_data *diff_words)
1750{
1751        xpparam_t xpp;
1752        xdemitconf_t xecfg;
1753        mmfile_t minus, plus;
1754        struct diff_words_style *style = diff_words->style;
1755
1756        struct diff_options *opt = diff_words->opt;
1757        const char *line_prefix;
1758
1759        assert(opt);
1760        line_prefix = diff_line_prefix(opt);
1761
1762        /* special case: only removal */
1763        if (!diff_words->plus.text.size) {
1764                emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1765                                 line_prefix, strlen(line_prefix), 0);
1766                fn_out_diff_words_write_helper(diff_words->opt,
1767                        &style->old_word, style->newline,
1768                        diff_words->minus.text.size,
1769                        diff_words->minus.text.ptr);
1770                diff_words->minus.text.size = 0;
1771                return;
1772        }
1773
1774        diff_words->current_plus = diff_words->plus.text.ptr;
1775        diff_words->last_minus = 0;
1776
1777        memset(&xpp, 0, sizeof(xpp));
1778        memset(&xecfg, 0, sizeof(xecfg));
1779        diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1780        diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1781        xpp.flags = 0;
1782        /* as only the hunk header will be parsed, we need a 0-context */
1783        xecfg.ctxlen = 0;
1784        if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1785                          &xpp, &xecfg))
1786                die("unable to generate word diff");
1787        free(minus.ptr);
1788        free(plus.ptr);
1789        if (diff_words->current_plus != diff_words->plus.text.ptr +
1790                        diff_words->plus.text.size) {
1791                if (color_words_output_graph_prefix(diff_words))
1792                        emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1793                                         line_prefix, strlen(line_prefix), 0);
1794                fn_out_diff_words_write_helper(diff_words->opt,
1795                        &style->ctx, style->newline,
1796                        diff_words->plus.text.ptr + diff_words->plus.text.size
1797                        - diff_words->current_plus, diff_words->current_plus);
1798        }
1799        diff_words->minus.text.size = diff_words->plus.text.size = 0;
1800}
1801
1802/* In "color-words" mode, show word-diff of words accumulated in the buffer */
1803static void diff_words_flush(struct emit_callback *ecbdata)
1804{
1805        struct diff_options *wo = ecbdata->diff_words->opt;
1806
1807        if (ecbdata->diff_words->minus.text.size ||
1808            ecbdata->diff_words->plus.text.size)
1809                diff_words_show(ecbdata->diff_words);
1810
1811        if (wo->emitted_symbols) {
1812                struct diff_options *o = ecbdata->opt;
1813                struct emitted_diff_symbols *wol = wo->emitted_symbols;
1814                int i;
1815
1816                /*
1817                 * NEEDSWORK:
1818                 * Instead of appending each, concat all words to a line?
1819                 */
1820                for (i = 0; i < wol->nr; i++)
1821                        append_emitted_diff_symbol(o, &wol->buf[i]);
1822
1823                for (i = 0; i < wol->nr; i++)
1824                        free((void *)wol->buf[i].line);
1825
1826                wol->nr = 0;
1827        }
1828}
1829
1830static void diff_filespec_load_driver(struct diff_filespec *one)
1831{
1832        /* Use already-loaded driver */
1833        if (one->driver)
1834                return;
1835
1836        if (S_ISREG(one->mode))
1837                one->driver = userdiff_find_by_path(one->path);
1838
1839        /* Fallback to default settings */
1840        if (!one->driver)
1841                one->driver = userdiff_find_by_name("default");
1842}
1843
1844static const char *userdiff_word_regex(struct diff_filespec *one)
1845{
1846        diff_filespec_load_driver(one);
1847        return one->driver->word_regex;
1848}
1849
1850static void init_diff_words_data(struct emit_callback *ecbdata,
1851                                 struct diff_options *orig_opts,
1852                                 struct diff_filespec *one,
1853                                 struct diff_filespec *two)
1854{
1855        int i;
1856        struct diff_options *o = xmalloc(sizeof(struct diff_options));
1857        memcpy(o, orig_opts, sizeof(struct diff_options));
1858
1859        ecbdata->diff_words =
1860                xcalloc(1, sizeof(struct diff_words_data));
1861        ecbdata->diff_words->type = o->word_diff;
1862        ecbdata->diff_words->opt = o;
1863
1864        if (orig_opts->emitted_symbols)
1865                o->emitted_symbols =
1866                        xcalloc(1, sizeof(struct emitted_diff_symbols));
1867
1868        if (!o->word_regex)
1869                o->word_regex = userdiff_word_regex(one);
1870        if (!o->word_regex)
1871                o->word_regex = userdiff_word_regex(two);
1872        if (!o->word_regex)
1873                o->word_regex = diff_word_regex_cfg;
1874        if (o->word_regex) {
1875                ecbdata->diff_words->word_regex = (regex_t *)
1876                        xmalloc(sizeof(regex_t));
1877                if (regcomp(ecbdata->diff_words->word_regex,
1878                            o->word_regex,
1879                            REG_EXTENDED | REG_NEWLINE))
1880                        die ("Invalid regular expression: %s",
1881                             o->word_regex);
1882        }
1883        for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1884                if (o->word_diff == diff_words_styles[i].type) {
1885                        ecbdata->diff_words->style =
1886                                &diff_words_styles[i];
1887                        break;
1888                }
1889        }
1890        if (want_color(o->use_color)) {
1891                struct diff_words_style *st = ecbdata->diff_words->style;
1892                st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1893                st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1894                st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1895        }
1896}
1897
1898static void free_diff_words_data(struct emit_callback *ecbdata)
1899{
1900        if (ecbdata->diff_words) {
1901                diff_words_flush(ecbdata);
1902                free (ecbdata->diff_words->opt->emitted_symbols);
1903                free (ecbdata->diff_words->opt);
1904                free (ecbdata->diff_words->minus.text.ptr);
1905                free (ecbdata->diff_words->minus.orig);
1906                free (ecbdata->diff_words->plus.text.ptr);
1907                free (ecbdata->diff_words->plus.orig);
1908                if (ecbdata->diff_words->word_regex) {
1909                        regfree(ecbdata->diff_words->word_regex);
1910                        free(ecbdata->diff_words->word_regex);
1911                }
1912                FREE_AND_NULL(ecbdata->diff_words);
1913        }
1914}
1915
1916const char *diff_get_color(int diff_use_color, enum color_diff ix)
1917{
1918        if (want_color(diff_use_color))
1919                return diff_colors[ix];
1920        return "";
1921}
1922
1923const char *diff_line_prefix(struct diff_options *opt)
1924{
1925        struct strbuf *msgbuf;
1926        if (!opt->output_prefix)
1927                return "";
1928
1929        msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1930        return msgbuf->buf;
1931}
1932
1933static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1934{
1935        const char *cp;
1936        unsigned long allot;
1937        size_t l = len;
1938
1939        cp = line;
1940        allot = l;
1941        while (0 < l) {
1942                (void) utf8_width(&cp, &l);
1943                if (!cp)
1944                        break; /* truncated in the middle? */
1945        }
1946        return allot - l;
1947}
1948
1949static void find_lno(const char *line, struct emit_callback *ecbdata)
1950{
1951        const char *p;
1952        ecbdata->lno_in_preimage = 0;
1953        ecbdata->lno_in_postimage = 0;
1954        p = strchr(line, '-');
1955        if (!p)
1956                return; /* cannot happen */
1957        ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1958        p = strchr(p, '+');
1959        if (!p)
1960                return; /* cannot happen */
1961        ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1962}
1963
1964static void fn_out_consume(void *priv, char *line, unsigned long len)
1965{
1966        struct emit_callback *ecbdata = priv;
1967        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1968        struct diff_options *o = ecbdata->opt;
1969
1970        o->found_changes = 1;
1971
1972        if (ecbdata->header) {
1973                emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
1974                                 ecbdata->header->buf, ecbdata->header->len, 0);
1975                strbuf_reset(ecbdata->header);
1976                ecbdata->header = NULL;
1977        }
1978
1979        if (ecbdata->label_path[0]) {
1980                emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1981                                 ecbdata->label_path[0],
1982                                 strlen(ecbdata->label_path[0]), 0);
1983                emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1984                                 ecbdata->label_path[1],
1985                                 strlen(ecbdata->label_path[1]), 0);
1986                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1987        }
1988
1989        if (diff_suppress_blank_empty
1990            && len == 2 && line[0] == ' ' && line[1] == '\n') {
1991                line[0] = '\n';
1992                len = 1;
1993        }
1994
1995        if (line[0] == '@') {
1996                if (ecbdata->diff_words)
1997                        diff_words_flush(ecbdata);
1998                len = sane_truncate_line(ecbdata, line, len);
1999                find_lno(line, ecbdata);
2000                emit_hunk_header(ecbdata, line, len);
2001                return;
2002        }
2003
2004        if (ecbdata->diff_words) {
2005                enum diff_symbol s =
2006                        ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2007                        DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2008                if (line[0] == '-') {
2009                        diff_words_append(line, len,
2010                                          &ecbdata->diff_words->minus);
2011                        return;
2012                } else if (line[0] == '+') {
2013                        diff_words_append(line, len,
2014                                          &ecbdata->diff_words->plus);
2015                        return;
2016                } else if (starts_with(line, "\\ ")) {
2017                        /*
2018                         * Eat the "no newline at eof" marker as if we
2019                         * saw a "+" or "-" line with nothing on it,
2020                         * and return without diff_words_flush() to
2021                         * defer processing. If this is the end of
2022                         * preimage, more "+" lines may come after it.
2023                         */
2024                        return;
2025                }
2026                diff_words_flush(ecbdata);
2027                emit_diff_symbol(o, s, line, len, 0);
2028                return;
2029        }
2030
2031        switch (line[0]) {
2032        case '+':
2033                ecbdata->lno_in_postimage++;
2034                emit_add_line(reset, ecbdata, line + 1, len - 1);
2035                break;
2036        case '-':
2037                ecbdata->lno_in_preimage++;
2038                emit_del_line(reset, ecbdata, line + 1, len - 1);
2039                break;
2040        case ' ':
2041                ecbdata->lno_in_postimage++;
2042                ecbdata->lno_in_preimage++;
2043                emit_context_line(reset, ecbdata, line + 1, len - 1);
2044                break;
2045        default:
2046                /* incomplete line at the end */
2047                ecbdata->lno_in_preimage++;
2048                emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2049                                 line, len, 0);
2050                break;
2051        }
2052}
2053
2054static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2055{
2056        const char *old_name = a;
2057        const char *new_name = b;
2058        int pfx_length, sfx_length;
2059        int pfx_adjust_for_slash;
2060        int len_a = strlen(a);
2061        int len_b = strlen(b);
2062        int a_midlen, b_midlen;
2063        int qlen_a = quote_c_style(a, NULL, NULL, 0);
2064        int qlen_b = quote_c_style(b, NULL, NULL, 0);
2065
2066        if (qlen_a || qlen_b) {
2067                quote_c_style(a, name, NULL, 0);
2068                strbuf_addstr(name, " => ");
2069                quote_c_style(b, name, NULL, 0);
2070                return;
2071        }
2072
2073        /* Find common prefix */
2074        pfx_length = 0;
2075        while (*old_name && *new_name && *old_name == *new_name) {
2076                if (*old_name == '/')
2077                        pfx_length = old_name - a + 1;
2078                old_name++;
2079                new_name++;
2080        }
2081
2082        /* Find common suffix */
2083        old_name = a + len_a;
2084        new_name = b + len_b;
2085        sfx_length = 0;
2086        /*
2087         * If there is a common prefix, it must end in a slash.  In
2088         * that case we let this loop run 1 into the prefix to see the
2089         * same slash.
2090         *
2091         * If there is no common prefix, we cannot do this as it would
2092         * underrun the input strings.
2093         */
2094        pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2095        while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2096               b + pfx_length - pfx_adjust_for_slash <= new_name &&
2097               *old_name == *new_name) {
2098                if (*old_name == '/')
2099                        sfx_length = len_a - (old_name - a);
2100                old_name--;
2101                new_name--;
2102        }
2103
2104        /*
2105         * pfx{mid-a => mid-b}sfx
2106         * {pfx-a => pfx-b}sfx
2107         * pfx{sfx-a => sfx-b}
2108         * name-a => name-b
2109         */
2110        a_midlen = len_a - pfx_length - sfx_length;
2111        b_midlen = len_b - pfx_length - sfx_length;
2112        if (a_midlen < 0)
2113                a_midlen = 0;
2114        if (b_midlen < 0)
2115                b_midlen = 0;
2116
2117        strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2118        if (pfx_length + sfx_length) {
2119                strbuf_add(name, a, pfx_length);
2120                strbuf_addch(name, '{');
2121        }
2122        strbuf_add(name, a + pfx_length, a_midlen);
2123        strbuf_addstr(name, " => ");
2124        strbuf_add(name, b + pfx_length, b_midlen);
2125        if (pfx_length + sfx_length) {
2126                strbuf_addch(name, '}');
2127                strbuf_add(name, a + len_a - sfx_length, sfx_length);
2128        }
2129}
2130
2131struct diffstat_t {
2132        int nr;
2133        int alloc;
2134        struct diffstat_file {
2135                char *from_name;
2136                char *name;
2137                char *print_name;
2138                const char *comments;
2139                unsigned is_unmerged:1;
2140                unsigned is_binary:1;
2141                unsigned is_renamed:1;
2142                unsigned is_interesting:1;
2143                uintmax_t added, deleted;
2144        } **files;
2145};
2146
2147static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2148                                          const char *name_a,
2149                                          const char *name_b)
2150{
2151        struct diffstat_file *x;
2152        x = xcalloc(1, sizeof(*x));
2153        ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2154        diffstat->files[diffstat->nr++] = x;
2155        if (name_b) {
2156                x->from_name = xstrdup(name_a);
2157                x->name = xstrdup(name_b);
2158                x->is_renamed = 1;
2159        }
2160        else {
2161                x->from_name = NULL;
2162                x->name = xstrdup(name_a);
2163        }
2164        return x;
2165}
2166
2167static void diffstat_consume(void *priv, char *line, unsigned long len)
2168{
2169        struct diffstat_t *diffstat = priv;
2170        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2171
2172        if (line[0] == '+')
2173                x->added++;
2174        else if (line[0] == '-')
2175                x->deleted++;
2176}
2177
2178const char mime_boundary_leader[] = "------------";
2179
2180static int scale_linear(int it, int width, int max_change)
2181{
2182        if (!it)
2183                return 0;
2184        /*
2185         * make sure that at least one '-' or '+' is printed if
2186         * there is any change to this path. The easiest way is to
2187         * scale linearly as if the alloted width is one column shorter
2188         * than it is, and then add 1 to the result.
2189         */
2190        return 1 + (it * (width - 1) / max_change);
2191}
2192
2193static void show_graph(struct strbuf *out, char ch, int cnt,
2194                       const char *set, const char *reset)
2195{
2196        if (cnt <= 0)
2197                return;
2198        strbuf_addstr(out, set);
2199        strbuf_addchars(out, ch, cnt);
2200        strbuf_addstr(out, reset);
2201}
2202
2203static void fill_print_name(struct diffstat_file *file)
2204{
2205        struct strbuf pname = STRBUF_INIT;
2206
2207        if (file->print_name)
2208                return;
2209
2210        if (file->is_renamed)
2211                pprint_rename(&pname, file->from_name, file->name);
2212        else
2213                quote_c_style(file->name, &pname, NULL, 0);
2214
2215        if (file->comments)
2216                strbuf_addf(&pname, " (%s)", file->comments);
2217
2218        file->print_name = strbuf_detach(&pname, NULL);
2219}
2220
2221static void print_stat_summary_inserts_deletes(struct diff_options *options,
2222                int files, int insertions, int deletions)
2223{
2224        struct strbuf sb = STRBUF_INIT;
2225
2226        if (!files) {
2227                assert(insertions == 0 && deletions == 0);
2228                emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2229                                 NULL, 0, 0);
2230                return;
2231        }
2232
2233        strbuf_addf(&sb,
2234                    (files == 1) ? " %d file changed" : " %d files changed",
2235                    files);
2236
2237        /*
2238         * For binary diff, the caller may want to print "x files
2239         * changed" with insertions == 0 && deletions == 0.
2240         *
2241         * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2242         * is probably less confusing (i.e skip over "2 files changed
2243         * but nothing about added/removed lines? Is this a bug in Git?").
2244         */
2245        if (insertions || deletions == 0) {
2246                strbuf_addf(&sb,
2247                            (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2248                            insertions);
2249        }
2250
2251        if (deletions || insertions == 0) {
2252                strbuf_addf(&sb,
2253                            (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2254                            deletions);
2255        }
2256        strbuf_addch(&sb, '\n');
2257        emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2258                         sb.buf, sb.len, 0);
2259        strbuf_release(&sb);
2260}
2261
2262void print_stat_summary(FILE *fp, int files,
2263                        int insertions, int deletions)
2264{
2265        struct diff_options o;
2266        memset(&o, 0, sizeof(o));
2267        o.file = fp;
2268
2269        print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2270}
2271
2272static void show_stats(struct diffstat_t *data, struct diff_options *options)
2273{
2274        int i, len, add, del, adds = 0, dels = 0;
2275        uintmax_t max_change = 0, max_len = 0;
2276        int total_files = data->nr, count;
2277        int width, name_width, graph_width, number_width = 0, bin_width = 0;
2278        const char *reset, *add_c, *del_c;
2279        int extra_shown = 0;
2280        const char *line_prefix = diff_line_prefix(options);
2281        struct strbuf out = STRBUF_INIT;
2282
2283        if (data->nr == 0)
2284                return;
2285
2286        count = options->stat_count ? options->stat_count : data->nr;
2287
2288        reset = diff_get_color_opt(options, DIFF_RESET);
2289        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2290        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2291
2292        /*
2293         * Find the longest filename and max number of changes
2294         */
2295        for (i = 0; (i < count) && (i < data->nr); i++) {
2296                struct diffstat_file *file = data->files[i];
2297                uintmax_t change = file->added + file->deleted;
2298
2299                if (!file->is_interesting && (change == 0)) {
2300                        count++; /* not shown == room for one more */
2301                        continue;
2302                }
2303                fill_print_name(file);
2304                len = strlen(file->print_name);
2305                if (max_len < len)
2306                        max_len = len;
2307
2308                if (file->is_unmerged) {
2309                        /* "Unmerged" is 8 characters */
2310                        bin_width = bin_width < 8 ? 8 : bin_width;
2311                        continue;
2312                }
2313                if (file->is_binary) {
2314                        /* "Bin XXX -> YYY bytes" */
2315                        int w = 14 + decimal_width(file->added)
2316                                + decimal_width(file->deleted);
2317                        bin_width = bin_width < w ? w : bin_width;
2318                        /* Display change counts aligned with "Bin" */
2319                        number_width = 3;
2320                        continue;
2321                }
2322
2323                if (max_change < change)
2324                        max_change = change;
2325        }
2326        count = i; /* where we can stop scanning in data->files[] */
2327
2328        /*
2329         * We have width = stat_width or term_columns() columns total.
2330         * We want a maximum of min(max_len, stat_name_width) for the name part.
2331         * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2332         * We also need 1 for " " and 4 + decimal_width(max_change)
2333         * for " | NNNN " and one the empty column at the end, altogether
2334         * 6 + decimal_width(max_change).
2335         *
2336         * If there's not enough space, we will use the smaller of
2337         * stat_name_width (if set) and 5/8*width for the filename,
2338         * and the rest for constant elements + graph part, but no more
2339         * than stat_graph_width for the graph part.
2340         * (5/8 gives 50 for filename and 30 for the constant parts + graph
2341         * for the standard terminal size).
2342         *
2343         * In other words: stat_width limits the maximum width, and
2344         * stat_name_width fixes the maximum width of the filename,
2345         * and is also used to divide available columns if there
2346         * aren't enough.
2347         *
2348         * Binary files are displayed with "Bin XXX -> YYY bytes"
2349         * instead of the change count and graph. This part is treated
2350         * similarly to the graph part, except that it is not
2351         * "scaled". If total width is too small to accommodate the
2352         * guaranteed minimum width of the filename part and the
2353         * separators and this message, this message will "overflow"
2354         * making the line longer than the maximum width.
2355         */
2356
2357        if (options->stat_width == -1)
2358                width = term_columns() - strlen(line_prefix);
2359        else
2360                width = options->stat_width ? options->stat_width : 80;
2361        number_width = decimal_width(max_change) > number_width ?
2362                decimal_width(max_change) : number_width;
2363
2364        if (options->stat_graph_width == -1)
2365                options->stat_graph_width = diff_stat_graph_width;
2366
2367        /*
2368         * Guarantee 3/8*16==6 for the graph part
2369         * and 5/8*16==10 for the filename part
2370         */
2371        if (width < 16 + 6 + number_width)
2372                width = 16 + 6 + number_width;
2373
2374        /*
2375         * First assign sizes that are wanted, ignoring available width.
2376         * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2377         * starting from "XXX" should fit in graph_width.
2378         */
2379        graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2380        if (options->stat_graph_width &&
2381            options->stat_graph_width < graph_width)
2382                graph_width = options->stat_graph_width;
2383
2384        name_width = (options->stat_name_width > 0 &&
2385                      options->stat_name_width < max_len) ?
2386                options->stat_name_width : max_len;
2387
2388        /*
2389         * Adjust adjustable widths not to exceed maximum width
2390         */
2391        if (name_width + number_width + 6 + graph_width > width) {
2392                if (graph_width > width * 3/8 - number_width - 6) {
2393                        graph_width = width * 3/8 - number_width - 6;
2394                        if (graph_width < 6)
2395                                graph_width = 6;
2396                }
2397
2398                if (options->stat_graph_width &&
2399                    graph_width > options->stat_graph_width)
2400                        graph_width = options->stat_graph_width;
2401                if (name_width > width - number_width - 6 - graph_width)
2402                        name_width = width - number_width - 6 - graph_width;
2403                else
2404                        graph_width = width - number_width - 6 - name_width;
2405        }
2406
2407        /*
2408         * From here name_width is the width of the name area,
2409         * and graph_width is the width of the graph area.
2410         * max_change is used to scale graph properly.
2411         */
2412        for (i = 0; i < count; i++) {
2413                const char *prefix = "";
2414                struct diffstat_file *file = data->files[i];
2415                char *name = file->print_name;
2416                uintmax_t added = file->added;
2417                uintmax_t deleted = file->deleted;
2418                int name_len;
2419
2420                if (!file->is_interesting && (added + deleted == 0))
2421                        continue;
2422
2423                /*
2424                 * "scale" the filename
2425                 */
2426                len = name_width;
2427                name_len = strlen(name);
2428                if (name_width < name_len) {
2429                        char *slash;
2430                        prefix = "...";
2431                        len -= 3;
2432                        name += name_len - len;
2433                        slash = strchr(name, '/');
2434                        if (slash)
2435                                name = slash;
2436                }
2437
2438                if (file->is_binary) {
2439                        strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2440                        strbuf_addf(&out, " %*s", number_width, "Bin");
2441                        if (!added && !deleted) {
2442                                strbuf_addch(&out, '\n');
2443                                emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2444                                                 out.buf, out.len, 0);
2445                                strbuf_reset(&out);
2446                                continue;
2447                        }
2448                        strbuf_addf(&out, " %s%"PRIuMAX"%s",
2449                                del_c, deleted, reset);
2450                        strbuf_addstr(&out, " -> ");
2451                        strbuf_addf(&out, "%s%"PRIuMAX"%s",
2452                                add_c, added, reset);
2453                        strbuf_addstr(&out, " bytes\n");
2454                        emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2455                                         out.buf, out.len, 0);
2456                        strbuf_reset(&out);
2457                        continue;
2458                }
2459                else if (file->is_unmerged) {
2460                        strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2461                        strbuf_addstr(&out, " Unmerged\n");
2462                        emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2463                                         out.buf, out.len, 0);
2464                        strbuf_reset(&out);
2465                        continue;
2466                }
2467
2468                /*
2469                 * scale the add/delete
2470                 */
2471                add = added;
2472                del = deleted;
2473
2474                if (graph_width <= max_change) {
2475                        int total = scale_linear(add + del, graph_width, max_change);
2476                        if (total < 2 && add && del)
2477                                /* width >= 2 due to the sanity check */
2478                                total = 2;
2479                        if (add < del) {
2480                                add = scale_linear(add, graph_width, max_change);
2481                                del = total - add;
2482                        } else {
2483                                del = scale_linear(del, graph_width, max_change);
2484                                add = total - del;
2485                        }
2486                }
2487                strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2488                strbuf_addf(&out, " %*"PRIuMAX"%s",
2489                        number_width, added + deleted,
2490                        added + deleted ? " " : "");
2491                show_graph(&out, '+', add, add_c, reset);
2492                show_graph(&out, '-', del, del_c, reset);
2493                strbuf_addch(&out, '\n');
2494                emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2495                                 out.buf, out.len, 0);
2496                strbuf_reset(&out);
2497        }
2498
2499        for (i = 0; i < data->nr; i++) {
2500                struct diffstat_file *file = data->files[i];
2501                uintmax_t added = file->added;
2502                uintmax_t deleted = file->deleted;
2503
2504                if (file->is_unmerged ||
2505                    (!file->is_interesting && (added + deleted == 0))) {
2506                        total_files--;
2507                        continue;
2508                }
2509
2510                if (!file->is_binary) {
2511                        adds += added;
2512                        dels += deleted;
2513                }
2514                if (i < count)
2515                        continue;
2516                if (!extra_shown)
2517                        emit_diff_symbol(options,
2518                                         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2519                                         NULL, 0, 0);
2520                extra_shown = 1;
2521        }
2522
2523        print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2524        strbuf_release(&out);
2525}
2526
2527static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2528{
2529        int i, adds = 0, dels = 0, total_files = data->nr;
2530
2531        if (data->nr == 0)
2532                return;
2533
2534        for (i = 0; i < data->nr; i++) {
2535                int added = data->files[i]->added;
2536                int deleted = data->files[i]->deleted;
2537
2538                if (data->files[i]->is_unmerged ||
2539                    (!data->files[i]->is_interesting && (added + deleted == 0))) {
2540                        total_files--;
2541                } else if (!data->files[i]->is_binary) { /* don't count bytes */
2542                        adds += added;
2543                        dels += deleted;
2544                }
2545        }
2546        print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2547}
2548
2549static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2550{
2551        int i;
2552
2553        if (data->nr == 0)
2554                return;
2555
2556        for (i = 0; i < data->nr; i++) {
2557                struct diffstat_file *file = data->files[i];
2558
2559                fprintf(options->file, "%s", diff_line_prefix(options));
2560
2561                if (file->is_binary)
2562                        fprintf(options->file, "-\t-\t");
2563                else
2564                        fprintf(options->file,
2565                                "%"PRIuMAX"\t%"PRIuMAX"\t",
2566                                file->added, file->deleted);
2567                if (options->line_termination) {
2568                        fill_print_name(file);
2569                        if (!file->is_renamed)
2570                                write_name_quoted(file->name, options->file,
2571                                                  options->line_termination);
2572                        else {
2573                                fputs(file->print_name, options->file);
2574                                putc(options->line_termination, options->file);
2575                        }
2576                } else {
2577                        if (file->is_renamed) {
2578                                putc('\0', options->file);
2579                                write_name_quoted(file->from_name, options->file, '\0');
2580                        }
2581                        write_name_quoted(file->name, options->file, '\0');
2582                }
2583        }
2584}
2585
2586struct dirstat_file {
2587        const char *name;
2588        unsigned long changed;
2589};
2590
2591struct dirstat_dir {
2592        struct dirstat_file *files;
2593        int alloc, nr, permille, cumulative;
2594};
2595
2596static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2597                unsigned long changed, const char *base, int baselen)
2598{
2599        unsigned long sum_changes = 0;
2600        unsigned int sources = 0;
2601        const char *line_prefix = diff_line_prefix(opt);
2602
2603        while (dir->nr) {
2604                struct dirstat_file *f = dir->files;
2605                int namelen = strlen(f->name);
2606                unsigned long changes;
2607                char *slash;
2608
2609                if (namelen < baselen)
2610                        break;
2611                if (memcmp(f->name, base, baselen))
2612                        break;
2613                slash = strchr(f->name + baselen, '/');
2614                if (slash) {
2615                        int newbaselen = slash + 1 - f->name;
2616                        changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2617                        sources++;
2618                } else {
2619                        changes = f->changed;
2620                        dir->files++;
2621                        dir->nr--;
2622                        sources += 2;
2623                }
2624                sum_changes += changes;
2625        }
2626
2627        /*
2628         * We don't report dirstat's for
2629         *  - the top level
2630         *  - or cases where everything came from a single directory
2631         *    under this directory (sources == 1).
2632         */
2633        if (baselen && sources != 1) {
2634                if (sum_changes) {
2635                        int permille = sum_changes * 1000 / changed;
2636                        if (permille >= dir->permille) {
2637                                fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2638                                        permille / 10, permille % 10, baselen, base);
2639                                if (!dir->cumulative)
2640                                        return 0;
2641                        }
2642                }
2643        }
2644        return sum_changes;
2645}
2646
2647static int dirstat_compare(const void *_a, const void *_b)
2648{
2649        const struct dirstat_file *a = _a;
2650        const struct dirstat_file *b = _b;
2651        return strcmp(a->name, b->name);
2652}
2653
2654static void show_dirstat(struct diff_options *options)
2655{
2656        int i;
2657        unsigned long changed;
2658        struct dirstat_dir dir;
2659        struct diff_queue_struct *q = &diff_queued_diff;
2660
2661        dir.files = NULL;
2662        dir.alloc = 0;
2663        dir.nr = 0;
2664        dir.permille = options->dirstat_permille;
2665        dir.cumulative = options->flags.dirstat_cumulative;
2666
2667        changed = 0;
2668        for (i = 0; i < q->nr; i++) {
2669                struct diff_filepair *p = q->queue[i];
2670                const char *name;
2671                unsigned long copied, added, damage;
2672                int content_changed;
2673
2674                name = p->two->path ? p->two->path : p->one->path;
2675
2676                if (p->one->oid_valid && p->two->oid_valid)
2677                        content_changed = oidcmp(&p->one->oid, &p->two->oid);
2678                else
2679                        content_changed = 1;
2680
2681                if (!content_changed) {
2682                        /*
2683                         * The SHA1 has not changed, so pre-/post-content is
2684                         * identical. We can therefore skip looking at the
2685                         * file contents altogether.
2686                         */
2687                        damage = 0;
2688                        goto found_damage;
2689                }
2690
2691                if (options->flags.dirstat_by_file) {
2692                        /*
2693                         * In --dirstat-by-file mode, we don't really need to
2694                         * look at the actual file contents at all.
2695                         * The fact that the SHA1 changed is enough for us to
2696                         * add this file to the list of results
2697                         * (with each file contributing equal damage).
2698                         */
2699                        damage = 1;
2700                        goto found_damage;
2701                }
2702
2703                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2704                        diff_populate_filespec(p->one, 0);
2705                        diff_populate_filespec(p->two, 0);
2706                        diffcore_count_changes(p->one, p->two, NULL, NULL,
2707                                               &copied, &added);
2708                        diff_free_filespec_data(p->one);
2709                        diff_free_filespec_data(p->two);
2710                } else if (DIFF_FILE_VALID(p->one)) {
2711                        diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2712                        copied = added = 0;
2713                        diff_free_filespec_data(p->one);
2714                } else if (DIFF_FILE_VALID(p->two)) {
2715                        diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2716                        copied = 0;
2717                        added = p->two->size;
2718                        diff_free_filespec_data(p->two);
2719                } else
2720                        continue;
2721
2722                /*
2723                 * Original minus copied is the removed material,
2724                 * added is the new material.  They are both damages
2725                 * made to the preimage.
2726                 * If the resulting damage is zero, we know that
2727                 * diffcore_count_changes() considers the two entries to
2728                 * be identical, but since content_changed is true, we
2729                 * know that there must have been _some_ kind of change,
2730                 * so we force all entries to have damage > 0.
2731                 */
2732                damage = (p->one->size - copied) + added;
2733                if (!damage)
2734                        damage = 1;
2735
2736found_damage:
2737                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2738                dir.files[dir.nr].name = name;
2739                dir.files[dir.nr].changed = damage;
2740                changed += damage;
2741                dir.nr++;
2742        }
2743
2744        /* This can happen even with many files, if everything was renames */
2745        if (!changed)
2746                return;
2747
2748        /* Show all directories with more than x% of the changes */
2749        QSORT(dir.files, dir.nr, dirstat_compare);
2750        gather_dirstat(options, &dir, changed, "", 0);
2751}
2752
2753static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2754{
2755        int i;
2756        unsigned long changed;
2757        struct dirstat_dir dir;
2758
2759        if (data->nr == 0)
2760                return;
2761
2762        dir.files = NULL;
2763        dir.alloc = 0;
2764        dir.nr = 0;
2765        dir.permille = options->dirstat_permille;
2766        dir.cumulative = options->flags.dirstat_cumulative;
2767
2768        changed = 0;
2769        for (i = 0; i < data->nr; i++) {
2770                struct diffstat_file *file = data->files[i];
2771                unsigned long damage = file->added + file->deleted;
2772                if (file->is_binary)
2773                        /*
2774                         * binary files counts bytes, not lines. Must find some
2775                         * way to normalize binary bytes vs. textual lines.
2776                         * The following heuristic assumes that there are 64
2777                         * bytes per "line".
2778                         * This is stupid and ugly, but very cheap...
2779                         */
2780                        damage = DIV_ROUND_UP(damage, 64);
2781                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2782                dir.files[dir.nr].name = file->name;
2783                dir.files[dir.nr].changed = damage;
2784                changed += damage;
2785                dir.nr++;
2786        }
2787
2788        /* This can happen even with many files, if everything was renames */
2789        if (!changed)
2790                return;
2791
2792        /* Show all directories with more than x% of the changes */
2793        QSORT(dir.files, dir.nr, dirstat_compare);
2794        gather_dirstat(options, &dir, changed, "", 0);
2795}
2796
2797static void free_diffstat_info(struct diffstat_t *diffstat)
2798{
2799        int i;
2800        for (i = 0; i < diffstat->nr; i++) {
2801                struct diffstat_file *f = diffstat->files[i];
2802                free(f->print_name);
2803                free(f->name);
2804                free(f->from_name);
2805                free(f);
2806        }
2807        free(diffstat->files);
2808}
2809
2810struct checkdiff_t {
2811        const char *filename;
2812        int lineno;
2813        int conflict_marker_size;
2814        struct diff_options *o;
2815        unsigned ws_rule;
2816        unsigned status;
2817};
2818
2819static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2820{
2821        char firstchar;
2822        int cnt;
2823
2824        if (len < marker_size + 1)
2825                return 0;
2826        firstchar = line[0];
2827        switch (firstchar) {
2828        case '=': case '>': case '<': case '|':
2829                break;
2830        default:
2831                return 0;
2832        }
2833        for (cnt = 1; cnt < marker_size; cnt++)
2834                if (line[cnt] != firstchar)
2835                        return 0;
2836        /* line[1] thru line[marker_size-1] are same as firstchar */
2837        if (len < marker_size + 1 || !isspace(line[marker_size]))
2838                return 0;
2839        return 1;
2840}
2841
2842static void checkdiff_consume(void *priv, char *line, unsigned long len)
2843{
2844        struct checkdiff_t *data = priv;
2845        int marker_size = data->conflict_marker_size;
2846        const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2847        const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2848        const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2849        char *err;
2850        const char *line_prefix;
2851
2852        assert(data->o);
2853        line_prefix = diff_line_prefix(data->o);
2854
2855        if (line[0] == '+') {
2856                unsigned bad;
2857                data->lineno++;
2858                if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2859                        data->status |= 1;
2860                        fprintf(data->o->file,
2861                                "%s%s:%d: leftover conflict marker\n",
2862                                line_prefix, data->filename, data->lineno);
2863                }
2864                bad = ws_check(line + 1, len - 1, data->ws_rule);
2865                if (!bad)
2866                        return;
2867                data->status |= bad;
2868                err = whitespace_error_string(bad);
2869                fprintf(data->o->file, "%s%s:%d: %s.\n",
2870                        line_prefix, data->filename, data->lineno, err);
2871                free(err);
2872                emit_line(data->o, set, reset, line, 1);
2873                ws_check_emit(line + 1, len - 1, data->ws_rule,
2874                              data->o->file, set, reset, ws);
2875        } else if (line[0] == ' ') {
2876                data->lineno++;
2877        } else if (line[0] == '@') {
2878                char *plus = strchr(line, '+');
2879                if (plus)
2880                        data->lineno = strtol(plus, NULL, 10) - 1;
2881                else
2882                        die("invalid diff");
2883        }
2884}
2885
2886static unsigned char *deflate_it(char *data,
2887                                 unsigned long size,
2888                                 unsigned long *result_size)
2889{
2890        int bound;
2891        unsigned char *deflated;
2892        git_zstream stream;
2893
2894        git_deflate_init(&stream, zlib_compression_level);
2895        bound = git_deflate_bound(&stream, size);
2896        deflated = xmalloc(bound);
2897        stream.next_out = deflated;
2898        stream.avail_out = bound;
2899
2900        stream.next_in = (unsigned char *)data;
2901        stream.avail_in = size;
2902        while (git_deflate(&stream, Z_FINISH) == Z_OK)
2903                ; /* nothing */
2904        git_deflate_end(&stream);
2905        *result_size = stream.total_out;
2906        return deflated;
2907}
2908
2909static void emit_binary_diff_body(struct diff_options *o,
2910                                  mmfile_t *one, mmfile_t *two)
2911{
2912        void *cp;
2913        void *delta;
2914        void *deflated;
2915        void *data;
2916        unsigned long orig_size;
2917        unsigned long delta_size;
2918        unsigned long deflate_size;
2919        unsigned long data_size;
2920
2921        /* We could do deflated delta, or we could do just deflated two,
2922         * whichever is smaller.
2923         */
2924        delta = NULL;
2925        deflated = deflate_it(two->ptr, two->size, &deflate_size);
2926        if (one->size && two->size) {
2927                delta = diff_delta(one->ptr, one->size,
2928                                   two->ptr, two->size,
2929                                   &delta_size, deflate_size);
2930                if (delta) {
2931                        void *to_free = delta;
2932                        orig_size = delta_size;
2933                        delta = deflate_it(delta, delta_size, &delta_size);
2934                        free(to_free);
2935                }
2936        }
2937
2938        if (delta && delta_size < deflate_size) {
2939                char *s = xstrfmt("%lu", orig_size);
2940                emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
2941                                 s, strlen(s), 0);
2942                free(s);
2943                free(deflated);
2944                data = delta;
2945                data_size = delta_size;
2946        } else {
2947                char *s = xstrfmt("%lu", two->size);
2948                emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
2949                                 s, strlen(s), 0);
2950                free(s);
2951                free(delta);
2952                data = deflated;
2953                data_size = deflate_size;
2954        }
2955
2956        /* emit data encoded in base85 */
2957        cp = data;
2958        while (data_size) {
2959                int len;
2960                int bytes = (52 < data_size) ? 52 : data_size;
2961                char line[71];
2962                data_size -= bytes;
2963                if (bytes <= 26)
2964                        line[0] = bytes + 'A' - 1;
2965                else
2966                        line[0] = bytes - 26 + 'a' - 1;
2967                encode_85(line + 1, cp, bytes);
2968                cp = (char *) cp + bytes;
2969
2970                len = strlen(line);
2971                line[len++] = '\n';
2972                line[len] = '\0';
2973
2974                emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
2975                                 line, len, 0);
2976        }
2977        emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
2978        free(data);
2979}
2980
2981static void emit_binary_diff(struct diff_options *o,
2982                             mmfile_t *one, mmfile_t *two)
2983{
2984        emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
2985        emit_binary_diff_body(o, one, two);
2986        emit_binary_diff_body(o, two, one);
2987}
2988
2989int diff_filespec_is_binary(struct diff_filespec *one)
2990{
2991        if (one->is_binary == -1) {
2992                diff_filespec_load_driver(one);
2993                if (one->driver->binary != -1)
2994                        one->is_binary = one->driver->binary;
2995                else {
2996                        if (!one->data && DIFF_FILE_VALID(one))
2997                                diff_populate_filespec(one, CHECK_BINARY);
2998                        if (one->is_binary == -1 && one->data)
2999                                one->is_binary = buffer_is_binary(one->data,
3000                                                one->size);
3001                        if (one->is_binary == -1)
3002                                one->is_binary = 0;
3003                }
3004        }
3005        return one->is_binary;
3006}
3007
3008static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3009{
3010        diff_filespec_load_driver(one);
3011        return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3012}
3013
3014void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3015{
3016        if (!options->a_prefix)
3017                options->a_prefix = a;
3018        if (!options->b_prefix)
3019                options->b_prefix = b;
3020}
3021
3022struct userdiff_driver *get_textconv(struct diff_filespec *one)
3023{
3024        if (!DIFF_FILE_VALID(one))
3025                return NULL;
3026
3027        diff_filespec_load_driver(one);
3028        return userdiff_get_textconv(one->driver);
3029}
3030
3031static void builtin_diff(const char *name_a,
3032                         const char *name_b,
3033                         struct diff_filespec *one,
3034                         struct diff_filespec *two,
3035                         const char *xfrm_msg,
3036                         int must_show_header,
3037                         struct diff_options *o,
3038                         int complete_rewrite)
3039{
3040        mmfile_t mf1, mf2;
3041        const char *lbl[2];
3042        char *a_one, *b_two;
3043        const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3044        const char *reset = diff_get_color_opt(o, DIFF_RESET);
3045        const char *a_prefix, *b_prefix;
3046        struct userdiff_driver *textconv_one = NULL;
3047        struct userdiff_driver *textconv_two = NULL;
3048        struct strbuf header = STRBUF_INIT;
3049        const char *line_prefix = diff_line_prefix(o);
3050
3051        diff_set_mnemonic_prefix(o, "a/", "b/");
3052        if (o->flags.reverse_diff) {
3053                a_prefix = o->b_prefix;
3054                b_prefix = o->a_prefix;
3055        } else {
3056                a_prefix = o->a_prefix;
3057                b_prefix = o->b_prefix;
3058        }
3059
3060        if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3061            (!one->mode || S_ISGITLINK(one->mode)) &&
3062            (!two->mode || S_ISGITLINK(two->mode))) {
3063                show_submodule_summary(o, one->path ? one->path : two->path,
3064                                &one->oid, &two->oid,
3065                                two->dirty_submodule);
3066                return;
3067        } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3068                   (!one->mode || S_ISGITLINK(one->mode)) &&
3069                   (!two->mode || S_ISGITLINK(two->mode))) {
3070                show_submodule_inline_diff(o, one->path ? one->path : two->path,
3071                                &one->oid, &two->oid,
3072                                two->dirty_submodule);
3073                return;
3074        }
3075
3076        if (o->flags.allow_textconv) {
3077                textconv_one = get_textconv(one);
3078                textconv_two = get_textconv(two);
3079        }
3080
3081        /* Never use a non-valid filename anywhere if at all possible */
3082        name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3083        name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3084
3085        a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3086        b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3087        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3088        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3089        strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3090        if (lbl[0][0] == '/') {
3091                /* /dev/null */
3092                strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3093                if (xfrm_msg)
3094                        strbuf_addstr(&header, xfrm_msg);
3095                must_show_header = 1;
3096        }
3097        else if (lbl[1][0] == '/') {
3098                strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3099                if (xfrm_msg)
3100                        strbuf_addstr(&header, xfrm_msg);
3101                must_show_header = 1;
3102        }
3103        else {
3104                if (one->mode != two->mode) {
3105                        strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3106                        strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3107                        must_show_header = 1;
3108                }
3109                if (xfrm_msg)
3110                        strbuf_addstr(&header, xfrm_msg);
3111
3112                /*
3113                 * we do not run diff between different kind
3114                 * of objects.
3115                 */
3116                if ((one->mode ^ two->mode) & S_IFMT)
3117                        goto free_ab_and_return;
3118                if (complete_rewrite &&
3119                    (textconv_one || !diff_filespec_is_binary(one)) &&
3120                    (textconv_two || !diff_filespec_is_binary(two))) {
3121                        emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3122                                         header.buf, header.len, 0);
3123                        strbuf_reset(&header);
3124                        emit_rewrite_diff(name_a, name_b, one, two,
3125                                                textconv_one, textconv_two, o);
3126                        o->found_changes = 1;
3127                        goto free_ab_and_return;
3128                }
3129        }
3130
3131        if (o->irreversible_delete && lbl[1][0] == '/') {
3132                emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3133                                 header.len, 0);
3134                strbuf_reset(&header);
3135                goto free_ab_and_return;
3136        } else if (!o->flags.text &&
3137            ( (!textconv_one && diff_filespec_is_binary(one)) ||
3138              (!textconv_two && diff_filespec_is_binary(two)) )) {
3139                struct strbuf sb = STRBUF_INIT;
3140                if (!one->data && !two->data &&
3141                    S_ISREG(one->mode) && S_ISREG(two->mode) &&
3142                    !o->flags.binary) {
3143                        if (!oidcmp(&one->oid, &two->oid)) {
3144                                if (must_show_header)
3145                                        emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3146                                                         header.buf, header.len,
3147                                                         0);
3148                                goto free_ab_and_return;
3149                        }
3150                        emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3151                                         header.buf, header.len, 0);
3152                        strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3153                                    diff_line_prefix(o), lbl[0], lbl[1]);
3154                        emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3155                                         sb.buf, sb.len, 0);
3156                        strbuf_release(&sb);
3157                        goto free_ab_and_return;
3158                }
3159                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3160                        die("unable to read files to diff");
3161                /* Quite common confusing case */
3162                if (mf1.size == mf2.size &&
3163                    !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3164                        if (must_show_header)
3165                                emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3166                                                 header.buf, header.len, 0);
3167                        goto free_ab_and_return;
3168                }
3169                emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3170                strbuf_reset(&header);
3171                if (o->flags.binary)
3172                        emit_binary_diff(o, &mf1, &mf2);
3173                else {
3174                        strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3175                                    diff_line_prefix(o), lbl[0], lbl[1]);
3176                        emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3177                                         sb.buf, sb.len, 0);
3178                        strbuf_release(&sb);
3179                }
3180                o->found_changes = 1;
3181        } else {
3182                /* Crazy xdl interfaces.. */
3183                const char *diffopts = getenv("GIT_DIFF_OPTS");
3184                const char *v;
3185                xpparam_t xpp;
3186                xdemitconf_t xecfg;
3187                struct emit_callback ecbdata;
3188                const struct userdiff_funcname *pe;
3189
3190                if (must_show_header) {
3191                        emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3192                                         header.buf, header.len, 0);
3193                        strbuf_reset(&header);
3194                }
3195
3196                mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3197                mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3198
3199                pe = diff_funcname_pattern(one);
3200                if (!pe)
3201                        pe = diff_funcname_pattern(two);
3202
3203                memset(&xpp, 0, sizeof(xpp));
3204                memset(&xecfg, 0, sizeof(xecfg));
3205                memset(&ecbdata, 0, sizeof(ecbdata));
3206                ecbdata.label_path = lbl;
3207                ecbdata.color_diff = want_color(o->use_color);
3208                ecbdata.ws_rule = whitespace_rule(name_b);
3209                if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3210                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
3211                ecbdata.opt = o;
3212                ecbdata.header = header.len ? &header : NULL;
3213                xpp.flags = o->xdl_opts;
3214                xpp.anchors = o->anchors;
3215                xpp.anchors_nr = o->anchors_nr;
3216                xecfg.ctxlen = o->context;
3217                xecfg.interhunkctxlen = o->interhunkcontext;
3218                xecfg.flags = XDL_EMIT_FUNCNAMES;
3219                if (o->flags.funccontext)
3220                        xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3221                if (pe)
3222                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3223                if (!diffopts)
3224                        ;
3225                else if (skip_prefix(diffopts, "--unified=", &v))
3226                        xecfg.ctxlen = strtoul(v, NULL, 10);
3227                else if (skip_prefix(diffopts, "-u", &v))
3228                        xecfg.ctxlen = strtoul(v, NULL, 10);
3229                if (o->word_diff)
3230                        init_diff_words_data(&ecbdata, o, one, two);
3231                if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3232                                  &xpp, &xecfg))
3233                        die("unable to generate diff for %s", one->path);
3234                if (o->word_diff)
3235                        free_diff_words_data(&ecbdata);
3236                if (textconv_one)
3237                        free(mf1.ptr);
3238                if (textconv_two)
3239                        free(mf2.ptr);
3240                xdiff_clear_find_func(&xecfg);
3241        }
3242
3243 free_ab_and_return:
3244        strbuf_release(&header);
3245        diff_free_filespec_data(one);
3246        diff_free_filespec_data(two);
3247        free(a_one);
3248        free(b_two);
3249        return;
3250}
3251
3252static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3253{
3254        if (!is_renamed) {
3255                if (p->status == DIFF_STATUS_ADDED) {
3256                        if (S_ISLNK(p->two->mode))
3257                                return "new +l";
3258                        else if ((p->two->mode & 0777) == 0755)
3259                                return "new +x";
3260                        else
3261                                return "new";
3262                } else if (p->status == DIFF_STATUS_DELETED)
3263                        return "gone";
3264        }
3265        if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3266                return "mode -l";
3267        else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3268                return "mode +l";
3269        else if ((p->one->mode & 0777) == 0644 &&
3270                 (p->two->mode & 0777) == 0755)
3271                return "mode +x";
3272        else if ((p->one->mode & 0777) == 0755 &&
3273                 (p->two->mode & 0777) == 0644)
3274                return "mode -x";
3275        return NULL;
3276}
3277
3278static void builtin_diffstat(const char *name_a, const char *name_b,
3279                             struct diff_filespec *one,
3280                             struct diff_filespec *two,
3281                             struct diffstat_t *diffstat,
3282                             struct diff_options *o,
3283                             struct diff_filepair *p)
3284{
3285        mmfile_t mf1, mf2;
3286        struct diffstat_file *data;
3287        int same_contents;
3288        int complete_rewrite = 0;
3289
3290        if (!DIFF_PAIR_UNMERGED(p)) {
3291                if (p->status == DIFF_STATUS_MODIFIED && p->score)
3292                        complete_rewrite = 1;
3293        }
3294
3295        data = diffstat_add(diffstat, name_a, name_b);
3296        data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3297        if (o->flags.stat_with_summary)
3298                data->comments = get_compact_summary(p, data->is_renamed);
3299
3300        if (!one || !two) {
3301                data->is_unmerged = 1;
3302                return;
3303        }
3304
3305        same_contents = !oidcmp(&one->oid, &two->oid);
3306
3307        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3308                data->is_binary = 1;
3309                if (same_contents) {
3310                        data->added = 0;
3311                        data->deleted = 0;
3312                } else {
3313                        data->added = diff_filespec_size(two);
3314                        data->deleted = diff_filespec_size(one);
3315                }
3316        }
3317
3318        else if (complete_rewrite) {
3319                diff_populate_filespec(one, 0);
3320                diff_populate_filespec(two, 0);
3321                data->deleted = count_lines(one->data, one->size);
3322                data->added = count_lines(two->data, two->size);
3323        }
3324
3325        else if (!same_contents) {
3326                /* Crazy xdl interfaces.. */
3327                xpparam_t xpp;
3328                xdemitconf_t xecfg;
3329
3330                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3331                        die("unable to read files to diff");
3332
3333                memset(&xpp, 0, sizeof(xpp));
3334                memset(&xecfg, 0, sizeof(xecfg));
3335                xpp.flags = o->xdl_opts;
3336                xpp.anchors = o->anchors;
3337                xpp.anchors_nr = o->anchors_nr;
3338                xecfg.ctxlen = o->context;
3339                xecfg.interhunkctxlen = o->interhunkcontext;
3340                if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3341                                  &xpp, &xecfg))
3342                        die("unable to generate diffstat for %s", one->path);
3343        }
3344
3345        diff_free_filespec_data(one);
3346        diff_free_filespec_data(two);
3347}
3348
3349static void builtin_checkdiff(const char *name_a, const char *name_b,
3350                              const char *attr_path,
3351                              struct diff_filespec *one,
3352                              struct diff_filespec *two,
3353                              struct diff_options *o)
3354{
3355        mmfile_t mf1, mf2;
3356        struct checkdiff_t data;
3357
3358        if (!two)
3359                return;
3360
3361        memset(&data, 0, sizeof(data));
3362        data.filename = name_b ? name_b : name_a;
3363        data.lineno = 0;
3364        data.o = o;
3365        data.ws_rule = whitespace_rule(attr_path);
3366        data.conflict_marker_size = ll_merge_marker_size(attr_path);
3367
3368        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3369                die("unable to read files to diff");
3370
3371        /*
3372         * All the other codepaths check both sides, but not checking
3373         * the "old" side here is deliberate.  We are checking the newly
3374         * introduced changes, and as long as the "new" side is text, we
3375         * can and should check what it introduces.
3376         */
3377        if (diff_filespec_is_binary(two))
3378                goto free_and_return;
3379        else {
3380                /* Crazy xdl interfaces.. */
3381                xpparam_t xpp;
3382                xdemitconf_t xecfg;
3383
3384                memset(&xpp, 0, sizeof(xpp));
3385                memset(&xecfg, 0, sizeof(xecfg));
3386                xecfg.ctxlen = 1; /* at least one context line */
3387                xpp.flags = 0;
3388                if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3389                                  &xpp, &xecfg))
3390                        die("unable to generate checkdiff for %s", one->path);
3391
3392                if (data.ws_rule & WS_BLANK_AT_EOF) {
3393                        struct emit_callback ecbdata;
3394                        int blank_at_eof;
3395
3396                        ecbdata.ws_rule = data.ws_rule;
3397                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
3398                        blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3399
3400                        if (blank_at_eof) {
3401                                static char *err;
3402                                if (!err)
3403                                        err = whitespace_error_string(WS_BLANK_AT_EOF);
3404                                fprintf(o->file, "%s:%d: %s.\n",
3405                                        data.filename, blank_at_eof, err);
3406                                data.status = 1; /* report errors */
3407                        }
3408                }
3409        }
3410 free_and_return:
3411        diff_free_filespec_data(one);
3412        diff_free_filespec_data(two);
3413        if (data.status)
3414                o->flags.check_failed = 1;
3415}
3416
3417struct diff_filespec *alloc_filespec(const char *path)
3418{
3419        struct diff_filespec *spec;
3420
3421        FLEXPTR_ALLOC_STR(spec, path, path);
3422        spec->count = 1;
3423        spec->is_binary = -1;
3424        return spec;
3425}
3426
3427void free_filespec(struct diff_filespec *spec)
3428{
3429        if (!--spec->count) {
3430                diff_free_filespec_data(spec);
3431                free(spec);
3432        }
3433}
3434
3435void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3436                   int oid_valid, unsigned short mode)
3437{
3438        if (mode) {
3439                spec->mode = canon_mode(mode);
3440                oidcpy(&spec->oid, oid);
3441                spec->oid_valid = oid_valid;
3442        }
3443}
3444
3445/*
3446 * Given a name and sha1 pair, if the index tells us the file in
3447 * the work tree has that object contents, return true, so that
3448 * prepare_temp_file() does not have to inflate and extract.
3449 */
3450static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3451{
3452        const struct cache_entry *ce;
3453        struct stat st;
3454        int pos, len;
3455
3456        /*
3457         * We do not read the cache ourselves here, because the
3458         * benchmark with my previous version that always reads cache
3459         * shows that it makes things worse for diff-tree comparing
3460         * two linux-2.6 kernel trees in an already checked out work
3461         * tree.  This is because most diff-tree comparisons deal with
3462         * only a small number of files, while reading the cache is
3463         * expensive for a large project, and its cost outweighs the
3464         * savings we get by not inflating the object to a temporary
3465         * file.  Practically, this code only helps when we are used
3466         * by diff-cache --cached, which does read the cache before
3467         * calling us.
3468         */
3469        if (!active_cache)
3470                return 0;
3471
3472        /* We want to avoid the working directory if our caller
3473         * doesn't need the data in a normal file, this system
3474         * is rather slow with its stat/open/mmap/close syscalls,
3475         * and the object is contained in a pack file.  The pack
3476         * is probably already open and will be faster to obtain
3477         * the data through than the working directory.  Loose
3478         * objects however would tend to be slower as they need
3479         * to be individually opened and inflated.
3480         */
3481        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
3482                return 0;
3483
3484        /*
3485         * Similarly, if we'd have to convert the file contents anyway, that
3486         * makes the optimization not worthwhile.
3487         */
3488        if (!want_file && would_convert_to_git(&the_index, name))
3489                return 0;
3490
3491        len = strlen(name);
3492        pos = cache_name_pos(name, len);
3493        if (pos < 0)
3494                return 0;
3495        ce = active_cache[pos];
3496
3497        /*
3498         * This is not the sha1 we are looking for, or
3499         * unreusable because it is not a regular file.
3500         */
3501        if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3502                return 0;
3503
3504        /*
3505         * If ce is marked as "assume unchanged", there is no
3506         * guarantee that work tree matches what we are looking for.
3507         */
3508        if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3509                return 0;
3510
3511        /*
3512         * If ce matches the file in the work tree, we can reuse it.
3513         */
3514        if (ce_uptodate(ce) ||
3515            (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3516                return 1;
3517
3518        return 0;
3519}
3520
3521static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3522{
3523        struct strbuf buf = STRBUF_INIT;
3524        char *dirty = "";
3525
3526        /* Are we looking at the work tree? */
3527        if (s->dirty_submodule)
3528                dirty = "-dirty";
3529
3530        strbuf_addf(&buf, "Subproject commit %s%s\n",
3531                    oid_to_hex(&s->oid), dirty);
3532        s->size = buf.len;
3533        if (size_only) {
3534                s->data = NULL;
3535                strbuf_release(&buf);
3536        } else {
3537                s->data = strbuf_detach(&buf, NULL);
3538                s->should_free = 1;
3539        }
3540        return 0;
3541}
3542
3543/*
3544 * While doing rename detection and pickaxe operation, we may need to
3545 * grab the data for the blob (or file) for our own in-core comparison.
3546 * diff_filespec has data and size fields for this purpose.
3547 */
3548int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3549{
3550        int size_only = flags & CHECK_SIZE_ONLY;
3551        int err = 0;
3552        int conv_flags = global_conv_flags_eol;
3553        /*
3554         * demote FAIL to WARN to allow inspecting the situation
3555         * instead of refusing.
3556         */
3557        if (conv_flags & CONV_EOL_RNDTRP_DIE)
3558                conv_flags = CONV_EOL_RNDTRP_WARN;
3559
3560        if (!DIFF_FILE_VALID(s))
3561                die("internal error: asking to populate invalid file.");
3562        if (S_ISDIR(s->mode))
3563                return -1;
3564
3565        if (s->data)
3566                return 0;
3567
3568        if (size_only && 0 < s->size)
3569                return 0;
3570
3571        if (S_ISGITLINK(s->mode))
3572                return diff_populate_gitlink(s, size_only);
3573
3574        if (!s->oid_valid ||
3575            reuse_worktree_file(s->path, &s->oid, 0)) {
3576                struct strbuf buf = STRBUF_INIT;
3577                struct stat st;
3578                int fd;
3579
3580                if (lstat(s->path, &st) < 0) {
3581                err_empty:
3582                        err = -1;
3583                empty:
3584                        s->data = (char *)"";
3585                        s->size = 0;
3586                        return err;
3587                }
3588                s->size = xsize_t(st.st_size);
3589                if (!s->size)
3590                        goto empty;
3591                if (S_ISLNK(st.st_mode)) {
3592                        struct strbuf sb = STRBUF_INIT;
3593
3594                        if (strbuf_readlink(&sb, s->path, s->size))
3595                                goto err_empty;
3596                        s->size = sb.len;
3597                        s->data = strbuf_detach(&sb, NULL);
3598                        s->should_free = 1;
3599                        return 0;
3600                }
3601
3602                /*
3603                 * Even if the caller would be happy with getting
3604                 * only the size, we cannot return early at this
3605                 * point if the path requires us to run the content
3606                 * conversion.
3607                 */
3608                if (size_only && !would_convert_to_git(&the_index, s->path))
3609                        return 0;
3610
3611                /*
3612                 * Note: this check uses xsize_t(st.st_size) that may
3613                 * not be the true size of the blob after it goes
3614                 * through convert_to_git().  This may not strictly be
3615                 * correct, but the whole point of big_file_threshold
3616                 * and is_binary check being that we want to avoid
3617                 * opening the file and inspecting the contents, this
3618                 * is probably fine.
3619                 */
3620                if ((flags & CHECK_BINARY) &&
3621                    s->size > big_file_threshold && s->is_binary == -1) {
3622                        s->is_binary = 1;
3623                        return 0;
3624                }
3625                fd = open(s->path, O_RDONLY);
3626                if (fd < 0)
3627                        goto err_empty;
3628                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3629                close(fd);
3630                s->should_munmap = 1;
3631
3632                /*
3633                 * Convert from working tree format to canonical git format
3634                 */
3635                if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, conv_flags)) {
3636                        size_t size = 0;
3637                        munmap(s->data, s->size);
3638                        s->should_munmap = 0;
3639                        s->data = strbuf_detach(&buf, &size);
3640                        s->size = size;
3641                        s->should_free = 1;
3642                }
3643        }
3644        else {
3645                enum object_type type;
3646                if (size_only || (flags & CHECK_BINARY)) {
3647                        type = oid_object_info(&s->oid, &s->size);
3648                        if (type < 0)
3649                                die("unable to read %s",
3650                                    oid_to_hex(&s->oid));
3651                        if (size_only)
3652                                return 0;
3653                        if (s->size > big_file_threshold && s->is_binary == -1) {
3654                                s->is_binary = 1;
3655                                return 0;
3656                        }
3657                }
3658                s->data = read_object_file(&s->oid, &type, &s->size);
3659                if (!s->data)
3660                        die("unable to read %s", oid_to_hex(&s->oid));
3661                s->should_free = 1;
3662        }
3663        return 0;
3664}
3665
3666void diff_free_filespec_blob(struct diff_filespec *s)
3667{
3668        if (s->should_free)
3669                free(s->data);
3670        else if (s->should_munmap)
3671                munmap(s->data, s->size);
3672
3673        if (s->should_free || s->should_munmap) {
3674                s->should_free = s->should_munmap = 0;
3675                s->data = NULL;
3676        }
3677}
3678
3679void diff_free_filespec_data(struct diff_filespec *s)
3680{
3681        diff_free_filespec_blob(s);
3682        FREE_AND_NULL(s->cnt_data);
3683}
3684
3685static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3686                           void *blob,
3687                           unsigned long size,
3688                           const struct object_id *oid,
3689                           int mode)
3690{
3691        struct strbuf buf = STRBUF_INIT;
3692        struct strbuf tempfile = STRBUF_INIT;
3693        char *path_dup = xstrdup(path);
3694        const char *base = basename(path_dup);
3695
3696        /* Generate "XXXXXX_basename.ext" */
3697        strbuf_addstr(&tempfile, "XXXXXX_");
3698        strbuf_addstr(&tempfile, base);
3699
3700        temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
3701        if (!temp->tempfile)
3702                die_errno("unable to create temp-file");
3703        if (convert_to_working_tree(path,
3704                        (const char *)blob, (size_t)size, &buf)) {
3705                blob = buf.buf;
3706                size = buf.len;
3707        }
3708        if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3709            close_tempfile_gently(temp->tempfile))
3710                die_errno("unable to write temp-file");
3711        temp->name = get_tempfile_path(temp->tempfile);
3712        oid_to_hex_r(temp->hex, oid);
3713        xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3714        strbuf_release(&buf);
3715        strbuf_release(&tempfile);
3716        free(path_dup);
3717}
3718
3719static struct diff_tempfile *prepare_temp_file(const char *name,
3720                struct diff_filespec *one)
3721{
3722        struct diff_tempfile *temp = claim_diff_tempfile();
3723
3724        if (!DIFF_FILE_VALID(one)) {
3725        not_a_valid_file:
3726                /* A '-' entry produces this for file-2, and
3727                 * a '+' entry produces this for file-1.
3728                 */
3729                temp->name = "/dev/null";
3730                xsnprintf(temp->hex, sizeof(temp->hex), ".");
3731                xsnprintf(temp->mode, sizeof(temp->mode), ".");
3732                return temp;
3733        }
3734
3735        if (!S_ISGITLINK(one->mode) &&
3736            (!one->oid_valid ||
3737             reuse_worktree_file(name, &one->oid, 1))) {
3738                struct stat st;
3739                if (lstat(name, &st) < 0) {
3740                        if (errno == ENOENT)
3741                                goto not_a_valid_file;
3742                        die_errno("stat(%s)", name);
3743                }
3744                if (S_ISLNK(st.st_mode)) {
3745                        struct strbuf sb = STRBUF_INIT;
3746                        if (strbuf_readlink(&sb, name, st.st_size) < 0)
3747                                die_errno("readlink(%s)", name);
3748                        prep_temp_blob(name, temp, sb.buf, sb.len,
3749                                       (one->oid_valid ?
3750                                        &one->oid : &null_oid),
3751                                       (one->oid_valid ?
3752                                        one->mode : S_IFLNK));
3753                        strbuf_release(&sb);
3754                }
3755                else {
3756                        /* we can borrow from the file in the work tree */
3757                        temp->name = name;
3758                        if (!one->oid_valid)
3759                                oid_to_hex_r(temp->hex, &null_oid);
3760                        else
3761                                oid_to_hex_r(temp->hex, &one->oid);
3762                        /* Even though we may sometimes borrow the
3763                         * contents from the work tree, we always want
3764                         * one->mode.  mode is trustworthy even when
3765                         * !(one->oid_valid), as long as
3766                         * DIFF_FILE_VALID(one).
3767                         */
3768                        xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3769                }
3770                return temp;
3771        }
3772        else {
3773                if (diff_populate_filespec(one, 0))
3774                        die("cannot read data blob for %s", one->path);
3775                prep_temp_blob(name, temp, one->data, one->size,
3776                               &one->oid, one->mode);
3777        }
3778        return temp;
3779}
3780
3781static void add_external_diff_name(struct argv_array *argv,
3782                                   const char *name,
3783                                   struct diff_filespec *df)
3784{
3785        struct diff_tempfile *temp = prepare_temp_file(name, df);
3786        argv_array_push(argv, temp->name);
3787        argv_array_push(argv, temp->hex);
3788        argv_array_push(argv, temp->mode);
3789}
3790
3791/* An external diff command takes:
3792 *
3793 * diff-cmd name infile1 infile1-sha1 infile1-mode \
3794 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3795 *
3796 */
3797static void run_external_diff(const char *pgm,
3798                              const char *name,
3799                              const char *other,
3800                              struct diff_filespec *one,
3801                              struct diff_filespec *two,
3802                              const char *xfrm_msg,
3803                              int complete_rewrite,
3804                              struct diff_options *o)
3805{
3806        struct argv_array argv = ARGV_ARRAY_INIT;
3807        struct argv_array env = ARGV_ARRAY_INIT;
3808        struct diff_queue_struct *q = &diff_queued_diff;
3809
3810        argv_array_push(&argv, pgm);
3811        argv_array_push(&argv, name);
3812
3813        if (one && two) {
3814                add_external_diff_name(&argv, name, one);
3815                if (!other)
3816                        add_external_diff_name(&argv, name, two);
3817                else {
3818                        add_external_diff_name(&argv, other, two);
3819                        argv_array_push(&argv, other);
3820                        argv_array_push(&argv, xfrm_msg);
3821                }
3822        }
3823
3824        argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3825        argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3826
3827        if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3828                die(_("external diff died, stopping at %s"), name);
3829
3830        remove_tempfile();
3831        argv_array_clear(&argv);
3832        argv_array_clear(&env);
3833}
3834
3835static int similarity_index(struct diff_filepair *p)
3836{
3837        return p->score * 100 / MAX_SCORE;
3838}
3839
3840static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3841{
3842        if (startup_info->have_repository)
3843                return find_unique_abbrev(oid, abbrev);
3844        else {
3845                char *hex = oid_to_hex(oid);
3846                if (abbrev < 0)
3847                        abbrev = FALLBACK_DEFAULT_ABBREV;
3848                if (abbrev > GIT_SHA1_HEXSZ)
3849                        die("BUG: oid abbreviation out of range: %d", abbrev);
3850                if (abbrev)
3851                        hex[abbrev] = '\0';
3852                return hex;
3853        }
3854}
3855
3856static void fill_metainfo(struct strbuf *msg,
3857                          const char *name,
3858                          const char *other,
3859                          struct diff_filespec *one,
3860                          struct diff_filespec *two,
3861                          struct diff_options *o,
3862                          struct diff_filepair *p,
3863                          int *must_show_header,
3864                          int use_color)
3865{
3866        const char *set = diff_get_color(use_color, DIFF_METAINFO);
3867        const char *reset = diff_get_color(use_color, DIFF_RESET);
3868        const char *line_prefix = diff_line_prefix(o);
3869
3870        *must_show_header = 1;
3871        strbuf_init(msg, PATH_MAX * 2 + 300);
3872        switch (p->status) {
3873        case DIFF_STATUS_COPIED:
3874                strbuf_addf(msg, "%s%ssimilarity index %d%%",
3875                            line_prefix, set, similarity_index(p));
3876                strbuf_addf(msg, "%s\n%s%scopy from ",
3877                            reset,  line_prefix, set);
3878                quote_c_style(name, msg, NULL, 0);
3879                strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3880                quote_c_style(other, msg, NULL, 0);
3881                strbuf_addf(msg, "%s\n", reset);
3882                break;
3883        case DIFF_STATUS_RENAMED:
3884                strbuf_addf(msg, "%s%ssimilarity index %d%%",
3885                            line_prefix, set, similarity_index(p));
3886                strbuf_addf(msg, "%s\n%s%srename from ",
3887                            reset, line_prefix, set);
3888                quote_c_style(name, msg, NULL, 0);
3889                strbuf_addf(msg, "%s\n%s%srename to ",
3890                            reset, line_prefix, set);
3891                quote_c_style(other, msg, NULL, 0);
3892                strbuf_addf(msg, "%s\n", reset);
3893                break;
3894        case DIFF_STATUS_MODIFIED:
3895                if (p->score) {
3896                        strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3897                                    line_prefix,
3898                                    set, similarity_index(p), reset);
3899                        break;
3900                }
3901                /* fallthru */
3902        default:
3903                *must_show_header = 0;
3904        }
3905        if (one && two && oidcmp(&one->oid, &two->oid)) {
3906                int abbrev = o->flags.full_index ? 40 : DEFAULT_ABBREV;
3907
3908                if (o->flags.binary) {
3909                        mmfile_t mf;
3910                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3911                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3912                                abbrev = 40;
3913                }
3914                strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3915                            diff_abbrev_oid(&one->oid, abbrev),
3916                            diff_abbrev_oid(&two->oid, abbrev));
3917                if (one->mode == two->mode)
3918                        strbuf_addf(msg, " %06o", one->mode);
3919                strbuf_addf(msg, "%s\n", reset);
3920        }
3921}
3922
3923static void run_diff_cmd(const char *pgm,
3924                         const char *name,
3925                         const char *other,
3926                         const char *attr_path,
3927                         struct diff_filespec *one,
3928                         struct diff_filespec *two,
3929                         struct strbuf *msg,
3930                         struct diff_options *o,
3931                         struct diff_filepair *p)
3932{
3933        const char *xfrm_msg = NULL;
3934        int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3935        int must_show_header = 0;
3936
3937
3938        if (o->flags.allow_external) {
3939                struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3940                if (drv && drv->external)
3941                        pgm = drv->external;
3942        }
3943
3944        if (msg) {
3945                /*
3946                 * don't use colors when the header is intended for an
3947                 * external diff driver
3948                 */
3949                fill_metainfo(msg, name, other, one, two, o, p,
3950                              &must_show_header,
3951                              want_color(o->use_color) && !pgm);
3952                xfrm_msg = msg->len ? msg->buf : NULL;
3953        }
3954
3955        if (pgm) {
3956                run_external_diff(pgm, name, other, one, two, xfrm_msg,
3957                                  complete_rewrite, o);
3958                return;
3959        }
3960        if (one && two)
3961                builtin_diff(name, other ? other : name,
3962                             one, two, xfrm_msg, must_show_header,
3963                             o, complete_rewrite);
3964        else
3965                fprintf(o->file, "* Unmerged path %s\n", name);
3966}
3967
3968static void diff_fill_oid_info(struct diff_filespec *one)
3969{
3970        if (DIFF_FILE_VALID(one)) {
3971                if (!one->oid_valid) {
3972                        struct stat st;
3973                        if (one->is_stdin) {
3974                                oidclr(&one->oid);
3975                                return;
3976                        }
3977                        if (lstat(one->path, &st) < 0)
3978                                die_errno("stat '%s'", one->path);
3979                        if (index_path(&one->oid, one->path, &st, 0))
3980                                die("cannot hash %s", one->path);
3981                }
3982        }
3983        else
3984                oidclr(&one->oid);
3985}
3986
3987static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3988{
3989        /* Strip the prefix but do not molest /dev/null and absolute paths */
3990        if (*namep && **namep != '/') {
3991                *namep += prefix_length;
3992                if (**namep == '/')
3993                        ++*namep;
3994        }
3995        if (*otherp && **otherp != '/') {
3996                *otherp += prefix_length;
3997                if (**otherp == '/')
3998                        ++*otherp;
3999        }
4000}
4001
4002static void run_diff(struct diff_filepair *p, struct diff_options *o)
4003{
4004        const char *pgm = external_diff();
4005        struct strbuf msg;
4006        struct diff_filespec *one = p->one;
4007        struct diff_filespec *two = p->two;
4008        const char *name;
4009        const char *other;
4010        const char *attr_path;
4011
4012        name  = one->path;
4013        other = (strcmp(name, two->path) ? two->path : NULL);
4014        attr_path = name;
4015        if (o->prefix_length)
4016                strip_prefix(o->prefix_length, &name, &other);
4017
4018        if (!o->flags.allow_external)
4019                pgm = NULL;
4020
4021        if (DIFF_PAIR_UNMERGED(p)) {
4022                run_diff_cmd(pgm, name, NULL, attr_path,
4023                             NULL, NULL, NULL, o, p);
4024                return;
4025        }
4026
4027        diff_fill_oid_info(one);
4028        diff_fill_oid_info(two);
4029
4030        if (!pgm &&
4031            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4032            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4033                /*
4034                 * a filepair that changes between file and symlink
4035                 * needs to be split into deletion and creation.
4036                 */
4037                struct diff_filespec *null = alloc_filespec(two->path);
4038                run_diff_cmd(NULL, name, other, attr_path,
4039                             one, null, &msg, o, p);
4040                free(null);
4041                strbuf_release(&msg);
4042
4043                null = alloc_filespec(one->path);
4044                run_diff_cmd(NULL, name, other, attr_path,
4045                             null, two, &msg, o, p);
4046                free(null);
4047        }
4048        else
4049                run_diff_cmd(pgm, name, other, attr_path,
4050                             one, two, &msg, o, p);
4051
4052        strbuf_release(&msg);
4053}
4054
4055static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4056                         struct diffstat_t *diffstat)
4057{
4058        const char *name;
4059        const char *other;
4060
4061        if (DIFF_PAIR_UNMERGED(p)) {
4062                /* unmerged */
4063                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4064                return;
4065        }
4066
4067        name = p->one->path;
4068        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4069
4070        if (o->prefix_length)
4071                strip_prefix(o->prefix_length, &name, &other);
4072
4073        diff_fill_oid_info(p->one);
4074        diff_fill_oid_info(p->two);
4075
4076        builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4077}
4078
4079static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4080{
4081        const char *name;
4082        const char *other;
4083        const char *attr_path;
4084
4085        if (DIFF_PAIR_UNMERGED(p)) {
4086                /* unmerged */
4087                return;
4088        }
4089
4090        name = p->one->path;
4091        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4092        attr_path = other ? other : name;
4093
4094        if (o->prefix_length)
4095                strip_prefix(o->prefix_length, &name, &other);
4096
4097        diff_fill_oid_info(p->one);
4098        diff_fill_oid_info(p->two);
4099
4100        builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4101}
4102
4103void diff_setup(struct diff_options *options)
4104{
4105        memcpy(options, &default_diff_options, sizeof(*options));
4106
4107        options->file = stdout;
4108
4109        options->abbrev = DEFAULT_ABBREV;
4110        options->line_termination = '\n';
4111        options->break_opt = -1;
4112        options->rename_limit = -1;
4113        options->dirstat_permille = diff_dirstat_permille_default;
4114        options->context = diff_context_default;
4115        options->interhunkcontext = diff_interhunk_context_default;
4116        options->ws_error_highlight = ws_error_highlight_default;
4117        options->flags.rename_empty = 1;
4118        options->objfind = NULL;
4119
4120        /* pathchange left =NULL by default */
4121        options->change = diff_change;
4122        options->add_remove = diff_addremove;
4123        options->use_color = diff_use_color_default;
4124        options->detect_rename = diff_detect_rename_default;
4125        options->xdl_opts |= diff_algorithm;
4126        if (diff_indent_heuristic)
4127                DIFF_XDL_SET(options, INDENT_HEURISTIC);
4128
4129        options->orderfile = diff_order_file_cfg;
4130
4131        if (diff_no_prefix) {
4132                options->a_prefix = options->b_prefix = "";
4133        } else if (!diff_mnemonic_prefix) {
4134                options->a_prefix = "a/";
4135                options->b_prefix = "b/";
4136        }
4137
4138        options->color_moved = diff_color_moved_default;
4139}
4140
4141void diff_setup_done(struct diff_options *options)
4142{
4143        unsigned check_mask = DIFF_FORMAT_NAME |
4144                              DIFF_FORMAT_NAME_STATUS |
4145                              DIFF_FORMAT_CHECKDIFF |
4146                              DIFF_FORMAT_NO_OUTPUT;
4147
4148        if (options->set_default)
4149                options->set_default(options);
4150
4151        if (HAS_MULTI_BITS(options->output_format & check_mask))
4152                die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4153
4154        if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4155                die(_("-G, -S and --find-object are mutually exclusive"));
4156
4157        /*
4158         * Most of the time we can say "there are changes"
4159         * only by checking if there are changed paths, but
4160         * --ignore-whitespace* options force us to look
4161         * inside contents.
4162         */
4163
4164        if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4165                options->flags.diff_from_contents = 1;
4166        else
4167                options->flags.diff_from_contents = 0;
4168
4169        if (options->flags.find_copies_harder)
4170                options->detect_rename = DIFF_DETECT_COPY;
4171
4172        if (!options->flags.relative_name)
4173                options->prefix = NULL;
4174        if (options->prefix)
4175                options->prefix_length = strlen(options->prefix);
4176        else
4177                options->prefix_length = 0;
4178
4179        if (options->output_format & (DIFF_FORMAT_NAME |
4180                                      DIFF_FORMAT_NAME_STATUS |
4181                                      DIFF_FORMAT_CHECKDIFF |
4182                                      DIFF_FORMAT_NO_OUTPUT))
4183                options->output_format &= ~(DIFF_FORMAT_RAW |
4184                                            DIFF_FORMAT_NUMSTAT |
4185                                            DIFF_FORMAT_DIFFSTAT |
4186                                            DIFF_FORMAT_SHORTSTAT |
4187                                            DIFF_FORMAT_DIRSTAT |
4188                                            DIFF_FORMAT_SUMMARY |
4189                                            DIFF_FORMAT_PATCH);
4190
4191        /*
4192         * These cases always need recursive; we do not drop caller-supplied
4193         * recursive bits for other formats here.
4194         */
4195        if (options->output_format & (DIFF_FORMAT_PATCH |
4196                                      DIFF_FORMAT_NUMSTAT |
4197                                      DIFF_FORMAT_DIFFSTAT |
4198                                      DIFF_FORMAT_SHORTSTAT |
4199                                      DIFF_FORMAT_DIRSTAT |
4200                                      DIFF_FORMAT_SUMMARY |
4201                                      DIFF_FORMAT_CHECKDIFF))
4202                options->flags.recursive = 1;
4203        /*
4204         * Also pickaxe would not work very well if you do not say recursive
4205         */
4206        if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4207                options->flags.recursive = 1;
4208        /*
4209         * When patches are generated, submodules diffed against the work tree
4210         * must be checked for dirtiness too so it can be shown in the output
4211         */
4212        if (options->output_format & DIFF_FORMAT_PATCH)
4213                options->flags.dirty_submodules = 1;
4214
4215        if (options->detect_rename && options->rename_limit < 0)
4216                options->rename_limit = diff_rename_limit_default;
4217        if (options->setup & DIFF_SETUP_USE_CACHE) {
4218                if (!active_cache)
4219                        /* read-cache does not die even when it fails
4220                         * so it is safe for us to do this here.  Also
4221                         * it does not smudge active_cache or active_nr
4222                         * when it fails, so we do not have to worry about
4223                         * cleaning it up ourselves either.
4224                         */
4225                        read_cache();
4226        }
4227        if (40 < options->abbrev)
4228                options->abbrev = 40; /* full */
4229
4230        /*
4231         * It does not make sense to show the first hit we happened
4232         * to have found.  It does not make sense not to return with
4233         * exit code in such a case either.
4234         */
4235        if (options->flags.quick) {
4236                options->output_format = DIFF_FORMAT_NO_OUTPUT;
4237                options->flags.exit_with_status = 1;
4238        }
4239
4240        options->diff_path_counter = 0;
4241
4242        if (options->flags.follow_renames && options->pathspec.nr != 1)
4243                die(_("--follow requires exactly one pathspec"));
4244
4245        if (!options->use_color || external_diff())
4246                options->color_moved = 0;
4247}
4248
4249static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4250{
4251        char c, *eq;
4252        int len;
4253
4254        if (*arg != '-')
4255                return 0;
4256        c = *++arg;
4257        if (!c)
4258                return 0;
4259        if (c == arg_short) {
4260                c = *++arg;
4261                if (!c)
4262                        return 1;
4263                if (val && isdigit(c)) {
4264                        char *end;
4265                        int n = strtoul(arg, &end, 10);
4266                        if (*end)
4267                                return 0;
4268                        *val = n;
4269                        return 1;
4270                }
4271                return 0;
4272        }
4273        if (c != '-')
4274                return 0;
4275        arg++;
4276        eq = strchrnul(arg, '=');
4277        len = eq - arg;
4278        if (!len || strncmp(arg, arg_long, len))
4279                return 0;
4280        if (*eq) {
4281                int n;
4282                char *end;
4283                if (!isdigit(*++eq))
4284                        return 0;
4285                n = strtoul(eq, &end, 10);
4286                if (*end)
4287                        return 0;
4288                *val = n;
4289        }
4290        return 1;
4291}
4292
4293static int diff_scoreopt_parse(const char *opt);
4294
4295static inline int short_opt(char opt, const char **argv,
4296                            const char **optarg)
4297{
4298        const char *arg = argv[0];
4299        if (arg[0] != '-' || arg[1] != opt)
4300                return 0;
4301        if (arg[2] != '\0') {
4302                *optarg = arg + 2;
4303                return 1;
4304        }
4305        if (!argv[1])
4306                die("Option '%c' requires a value", opt);
4307        *optarg = argv[1];
4308        return 2;
4309}
4310
4311int parse_long_opt(const char *opt, const char **argv,
4312                   const char **optarg)
4313{
4314        const char *arg = argv[0];
4315        if (!skip_prefix(arg, "--", &arg))
4316                return 0;
4317        if (!skip_prefix(arg, opt, &arg))
4318                return 0;
4319        if (*arg == '=') { /* stuck form: --option=value */
4320                *optarg = arg + 1;
4321                return 1;
4322        }
4323        if (*arg != '\0')
4324                return 0;
4325        /* separate form: --option value */
4326        if (!argv[1])
4327                die("Option '--%s' requires a value", opt);
4328        *optarg = argv[1];
4329        return 2;
4330}
4331
4332static int stat_opt(struct diff_options *options, const char **av)
4333{
4334        const char *arg = av[0];
4335        char *end;
4336        int width = options->stat_width;
4337        int name_width = options->stat_name_width;
4338        int graph_width = options->stat_graph_width;
4339        int count = options->stat_count;
4340        int argcount = 1;
4341
4342        if (!skip_prefix(arg, "--stat", &arg))
4343                die("BUG: stat option does not begin with --stat: %s", arg);
4344        end = (char *)arg;
4345
4346        switch (*arg) {
4347        case '-':
4348                if (skip_prefix(arg, "-width", &arg)) {
4349                        if (*arg == '=')
4350                                width = strtoul(arg + 1, &end, 10);
4351                        else if (!*arg && !av[1])
4352                                die_want_option("--stat-width");
4353                        else if (!*arg) {
4354                                width = strtoul(av[1], &end, 10);
4355                                argcount = 2;
4356                        }
4357                } else if (skip_prefix(arg, "-name-width", &arg)) {
4358                        if (*arg == '=')
4359                                name_width = strtoul(arg + 1, &end, 10);
4360                        else if (!*arg && !av[1])
4361                                die_want_option("--stat-name-width");
4362                        else if (!*arg) {
4363                                name_width = strtoul(av[1], &end, 10);
4364                                argcount = 2;
4365                        }
4366                } else if (skip_prefix(arg, "-graph-width", &arg)) {
4367                        if (*arg == '=')
4368                                graph_width = strtoul(arg + 1, &end, 10);
4369                        else if (!*arg && !av[1])
4370                                die_want_option("--stat-graph-width");
4371                        else if (!*arg) {
4372                                graph_width = strtoul(av[1], &end, 10);
4373                                argcount = 2;
4374                        }
4375                } else if (skip_prefix(arg, "-count", &arg)) {
4376                        if (*arg == '=')
4377                                count = strtoul(arg + 1, &end, 10);
4378                        else if (!*arg && !av[1])
4379                                die_want_option("--stat-count");
4380                        else if (!*arg) {
4381                                count = strtoul(av[1], &end, 10);
4382                                argcount = 2;
4383                        }
4384                }
4385                break;
4386        case '=':
4387                width = strtoul(arg+1, &end, 10);
4388                if (*end == ',')
4389                        name_width = strtoul(end+1, &end, 10);
4390                if (*end == ',')
4391                        count = strtoul(end+1, &end, 10);
4392        }
4393
4394        /* Important! This checks all the error cases! */
4395        if (*end)
4396                return 0;
4397        options->output_format |= DIFF_FORMAT_DIFFSTAT;
4398        options->stat_name_width = name_width;
4399        options->stat_graph_width = graph_width;
4400        options->stat_width = width;
4401        options->stat_count = count;
4402        return argcount;
4403}
4404
4405static int parse_dirstat_opt(struct diff_options *options, const char *params)
4406{
4407        struct strbuf errmsg = STRBUF_INIT;
4408        if (parse_dirstat_params(options, params, &errmsg))
4409                die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4410                    errmsg.buf);
4411        strbuf_release(&errmsg);
4412        /*
4413         * The caller knows a dirstat-related option is given from the command
4414         * line; allow it to say "return this_function();"
4415         */
4416        options->output_format |= DIFF_FORMAT_DIRSTAT;
4417        return 1;
4418}
4419
4420static int parse_submodule_opt(struct diff_options *options, const char *value)
4421{
4422        if (parse_submodule_params(options, value))
4423                die(_("Failed to parse --submodule option parameter: '%s'"),
4424                        value);
4425        return 1;
4426}
4427
4428static const char diff_status_letters[] = {
4429        DIFF_STATUS_ADDED,
4430        DIFF_STATUS_COPIED,
4431        DIFF_STATUS_DELETED,
4432        DIFF_STATUS_MODIFIED,
4433        DIFF_STATUS_RENAMED,
4434        DIFF_STATUS_TYPE_CHANGED,
4435        DIFF_STATUS_UNKNOWN,
4436        DIFF_STATUS_UNMERGED,
4437        DIFF_STATUS_FILTER_AON,
4438        DIFF_STATUS_FILTER_BROKEN,
4439        '\0',
4440};
4441
4442static unsigned int filter_bit['Z' + 1];
4443
4444static void prepare_filter_bits(void)
4445{
4446        int i;
4447
4448        if (!filter_bit[DIFF_STATUS_ADDED]) {
4449                for (i = 0; diff_status_letters[i]; i++)
4450                        filter_bit[(int) diff_status_letters[i]] = (1 << i);
4451        }
4452}
4453
4454static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4455{
4456        return opt->filter & filter_bit[(int) status];
4457}
4458
4459static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4460{
4461        int i, optch;
4462
4463        prepare_filter_bits();
4464
4465        /*
4466         * If there is a negation e.g. 'd' in the input, and we haven't
4467         * initialized the filter field with another --diff-filter, start
4468         * from full set of bits, except for AON.
4469         */
4470        if (!opt->filter) {
4471                for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4472                        if (optch < 'a' || 'z' < optch)
4473                                continue;
4474                        opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4475                        opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4476                        break;
4477                }
4478        }
4479
4480        for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4481                unsigned int bit;
4482                int negate;
4483
4484                if ('a' <= optch && optch <= 'z') {
4485                        negate = 1;
4486                        optch = toupper(optch);
4487                } else {
4488                        negate = 0;
4489                }
4490
4491                bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4492                if (!bit)
4493                        return optarg[i];
4494                if (negate)
4495                        opt->filter &= ~bit;
4496                else
4497                        opt->filter |= bit;
4498        }
4499        return 0;
4500}
4501
4502static void enable_patch_output(int *fmt) {
4503        *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4504        *fmt |= DIFF_FORMAT_PATCH;
4505}
4506
4507static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4508{
4509        int val = parse_ws_error_highlight(arg);
4510
4511        if (val < 0) {
4512                error("unknown value after ws-error-highlight=%.*s",
4513                      -1 - val, arg);
4514                return 0;
4515        }
4516        opt->ws_error_highlight = val;
4517        return 1;
4518}
4519
4520static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4521{
4522        struct object_id oid;
4523
4524        if (get_oid(arg, &oid))
4525                return error("unable to resolve '%s'", arg);
4526
4527        if (!opt->objfind)
4528                opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4529
4530        opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4531        opt->flags.recursive = 1;
4532        opt->flags.tree_in_recursive = 1;
4533        oidset_insert(opt->objfind, &oid);
4534        return 1;
4535}
4536
4537int diff_opt_parse(struct diff_options *options,
4538                   const char **av, int ac, const char *prefix)
4539{
4540        const char *arg = av[0];
4541        const char *optarg;
4542        int argcount;
4543
4544        if (!prefix)
4545                prefix = "";
4546
4547        /* Output format options */
4548        if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4549            || opt_arg(arg, 'U', "unified", &options->context))
4550                enable_patch_output(&options->output_format);
4551        else if (!strcmp(arg, "--raw"))
4552                options->output_format |= DIFF_FORMAT_RAW;
4553        else if (!strcmp(arg, "--patch-with-raw")) {
4554                enable_patch_output(&options->output_format);
4555                options->output_format |= DIFF_FORMAT_RAW;
4556        } else if (!strcmp(arg, "--numstat"))
4557                options->output_format |= DIFF_FORMAT_NUMSTAT;
4558        else if (!strcmp(arg, "--shortstat"))
4559                options->output_format |= DIFF_FORMAT_SHORTSTAT;
4560        else if (skip_prefix(arg, "-X", &arg) ||
4561                 skip_to_optional_arg(arg, "--dirstat", &arg))
4562                return parse_dirstat_opt(options, arg);
4563        else if (!strcmp(arg, "--cumulative"))
4564                return parse_dirstat_opt(options, "cumulative");
4565        else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4566                parse_dirstat_opt(options, "files");
4567                return parse_dirstat_opt(options, arg);
4568        }
4569        else if (!strcmp(arg, "--check"))
4570                options->output_format |= DIFF_FORMAT_CHECKDIFF;
4571        else if (!strcmp(arg, "--summary"))
4572                options->output_format |= DIFF_FORMAT_SUMMARY;
4573        else if (!strcmp(arg, "--patch-with-stat")) {
4574                enable_patch_output(&options->output_format);
4575                options->output_format |= DIFF_FORMAT_DIFFSTAT;
4576        } else if (!strcmp(arg, "--name-only"))
4577                options->output_format |= DIFF_FORMAT_NAME;
4578        else if (!strcmp(arg, "--name-status"))
4579                options->output_format |= DIFF_FORMAT_NAME_STATUS;
4580        else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4581                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4582        else if (starts_with(arg, "--stat"))
4583                /* --stat, --stat-width, --stat-name-width, or --stat-count */
4584                return stat_opt(options, av);
4585        else if (!strcmp(arg, "--compact-summary")) {
4586                 options->flags.stat_with_summary = 1;
4587                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4588        } else if (!strcmp(arg, "--no-compact-summary"))
4589                 options->flags.stat_with_summary = 0;
4590
4591        /* renames options */
4592        else if (starts_with(arg, "-B") ||
4593                 skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4594                if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4595                        return error("invalid argument to -B: %s", arg+2);
4596        }
4597        else if (starts_with(arg, "-M") ||
4598                 skip_to_optional_arg(arg, "--find-renames", NULL)) {
4599                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4600                        return error("invalid argument to -M: %s", arg+2);
4601                options->detect_rename = DIFF_DETECT_RENAME;
4602        }
4603        else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4604                options->irreversible_delete = 1;
4605        }
4606        else if (starts_with(arg, "-C") ||
4607                 skip_to_optional_arg(arg, "--find-copies", NULL)) {
4608                if (options->detect_rename == DIFF_DETECT_COPY)
4609                        options->flags.find_copies_harder = 1;
4610                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4611                        return error("invalid argument to -C: %s", arg+2);
4612                options->detect_rename = DIFF_DETECT_COPY;
4613        }
4614        else if (!strcmp(arg, "--no-renames"))
4615                options->detect_rename = 0;
4616        else if (!strcmp(arg, "--rename-empty"))
4617                options->flags.rename_empty = 1;
4618        else if (!strcmp(arg, "--no-rename-empty"))
4619                options->flags.rename_empty = 0;
4620        else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4621                options->flags.relative_name = 1;
4622                if (arg)
4623                        options->prefix = arg;
4624        }
4625
4626        /* xdiff options */
4627        else if (!strcmp(arg, "--minimal"))
4628                DIFF_XDL_SET(options, NEED_MINIMAL);
4629        else if (!strcmp(arg, "--no-minimal"))
4630                DIFF_XDL_CLR(options, NEED_MINIMAL);
4631        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4632                DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4633        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4634                DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4635        else if (!strcmp(arg, "--ignore-space-at-eol"))
4636                DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4637        else if (!strcmp(arg, "--ignore-cr-at-eol"))
4638                DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4639        else if (!strcmp(arg, "--ignore-blank-lines"))
4640                DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4641        else if (!strcmp(arg, "--indent-heuristic"))
4642                DIFF_XDL_SET(options, INDENT_HEURISTIC);
4643        else if (!strcmp(arg, "--no-indent-heuristic"))
4644                DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4645        else if (!strcmp(arg, "--patience")) {
4646                int i;
4647                options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4648                /*
4649                 * Both --patience and --anchored use PATIENCE_DIFF
4650                 * internally, so remove any anchors previously
4651                 * specified.
4652                 */
4653                for (i = 0; i < options->anchors_nr; i++)
4654                        free(options->anchors[i]);
4655                options->anchors_nr = 0;
4656        } else if (!strcmp(arg, "--histogram"))
4657                options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4658        else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4659                long value = parse_algorithm_value(optarg);
4660                if (value < 0)
4661                        return error("option diff-algorithm accepts \"myers\", "
4662                                     "\"minimal\", \"patience\" and \"histogram\"");
4663                /* clear out previous settings */
4664                DIFF_XDL_CLR(options, NEED_MINIMAL);
4665                options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4666                options->xdl_opts |= value;
4667                return argcount;
4668        } else if (skip_prefix(arg, "--anchored=", &arg)) {
4669                options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4670                ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4671                           options->anchors_alloc);
4672                options->anchors[options->anchors_nr++] = xstrdup(arg);
4673        }
4674
4675        /* flags options */
4676        else if (!strcmp(arg, "--binary")) {
4677                enable_patch_output(&options->output_format);
4678                options->flags.binary = 1;
4679        }
4680        else if (!strcmp(arg, "--full-index"))
4681                options->flags.full_index = 1;
4682        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4683                options->flags.text = 1;
4684        else if (!strcmp(arg, "-R"))
4685                options->flags.reverse_diff = 1;
4686        else if (!strcmp(arg, "--find-copies-harder"))
4687                options->flags.find_copies_harder = 1;
4688        else if (!strcmp(arg, "--follow"))
4689                options->flags.follow_renames = 1;
4690        else if (!strcmp(arg, "--no-follow")) {
4691                options->flags.follow_renames = 0;
4692                options->flags.default_follow_renames = 0;
4693        } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
4694                int value = git_config_colorbool(NULL, arg);
4695                if (value < 0)
4696                        return error("option `color' expects \"always\", \"auto\", or \"never\"");
4697                options->use_color = value;
4698        }
4699        else if (!strcmp(arg, "--no-color"))
4700                options->use_color = 0;
4701        else if (!strcmp(arg, "--color-moved")) {
4702                if (diff_color_moved_default)
4703                        options->color_moved = diff_color_moved_default;
4704                if (options->color_moved == COLOR_MOVED_NO)
4705                        options->color_moved = COLOR_MOVED_DEFAULT;
4706        } else if (!strcmp(arg, "--no-color-moved"))
4707                options->color_moved = COLOR_MOVED_NO;
4708        else if (skip_prefix(arg, "--color-moved=", &arg)) {
4709                int cm = parse_color_moved(arg);
4710                if (cm < 0)
4711                        die("bad --color-moved argument: %s", arg);
4712                options->color_moved = cm;
4713        } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
4714                options->use_color = 1;
4715                options->word_diff = DIFF_WORDS_COLOR;
4716        }
4717        else if (!strcmp(arg, "--word-diff")) {
4718                if (options->word_diff == DIFF_WORDS_NONE)
4719                        options->word_diff = DIFF_WORDS_PLAIN;
4720        }
4721        else if (skip_prefix(arg, "--word-diff=", &arg)) {
4722                if (!strcmp(arg, "plain"))
4723                        options->word_diff = DIFF_WORDS_PLAIN;
4724                else if (!strcmp(arg, "color")) {
4725                        options->use_color = 1;
4726                        options->word_diff = DIFF_WORDS_COLOR;
4727                }
4728                else if (!strcmp(arg, "porcelain"))
4729                        options->word_diff = DIFF_WORDS_PORCELAIN;
4730                else if (!strcmp(arg, "none"))
4731                        options->word_diff = DIFF_WORDS_NONE;
4732                else
4733                        die("bad --word-diff argument: %s", arg);
4734        }
4735        else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4736                if (options->word_diff == DIFF_WORDS_NONE)
4737                        options->word_diff = DIFF_WORDS_PLAIN;
4738                options->word_regex = optarg;
4739                return argcount;
4740        }
4741        else if (!strcmp(arg, "--exit-code"))
4742                options->flags.exit_with_status = 1;
4743        else if (!strcmp(arg, "--quiet"))
4744                options->flags.quick = 1;
4745        else if (!strcmp(arg, "--ext-diff"))
4746                options->flags.allow_external = 1;
4747        else if (!strcmp(arg, "--no-ext-diff"))
4748                options->flags.allow_external = 0;
4749        else if (!strcmp(arg, "--textconv")) {
4750                options->flags.allow_textconv = 1;
4751                options->flags.textconv_set_via_cmdline = 1;
4752        } else if (!strcmp(arg, "--no-textconv"))
4753                options->flags.allow_textconv = 0;
4754        else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
4755                options->flags.override_submodule_config = 1;
4756                handle_ignore_submodules_arg(options, arg);
4757        } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
4758                return parse_submodule_opt(options, arg);
4759        else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4760                return parse_ws_error_highlight_opt(options, arg);
4761        else if (!strcmp(arg, "--ita-invisible-in-index"))
4762                options->ita_invisible_in_index = 1;
4763        else if (!strcmp(arg, "--ita-visible-in-index"))
4764                options->ita_invisible_in_index = 0;
4765
4766        /* misc options */
4767        else if (!strcmp(arg, "-z"))
4768                options->line_termination = 0;
4769        else if ((argcount = short_opt('l', av, &optarg))) {
4770                options->rename_limit = strtoul(optarg, NULL, 10);
4771                return argcount;
4772        }
4773        else if ((argcount = short_opt('S', av, &optarg))) {
4774                options->pickaxe = optarg;
4775                options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4776                return argcount;
4777        } else if ((argcount = short_opt('G', av, &optarg))) {
4778                options->pickaxe = optarg;
4779                options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4780                return argcount;
4781        }
4782        else if (!strcmp(arg, "--pickaxe-all"))
4783                options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4784        else if (!strcmp(arg, "--pickaxe-regex"))
4785                options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4786        else if ((argcount = short_opt('O', av, &optarg))) {
4787                options->orderfile = prefix_filename(prefix, optarg);
4788                return argcount;
4789        } else if (skip_prefix(arg, "--find-object=", &arg))
4790                return parse_objfind_opt(options, arg);
4791        else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4792                int offending = parse_diff_filter_opt(optarg, options);
4793                if (offending)
4794                        die("unknown change class '%c' in --diff-filter=%s",
4795                            offending, optarg);
4796                return argcount;
4797        }
4798        else if (!strcmp(arg, "--no-abbrev"))
4799                options->abbrev = 0;
4800        else if (!strcmp(arg, "--abbrev"))
4801                options->abbrev = DEFAULT_ABBREV;
4802        else if (skip_prefix(arg, "--abbrev=", &arg)) {
4803                options->abbrev = strtoul(arg, NULL, 10);
4804                if (options->abbrev < MINIMUM_ABBREV)
4805                        options->abbrev = MINIMUM_ABBREV;
4806                else if (40 < options->abbrev)
4807                        options->abbrev = 40;
4808        }
4809        else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4810                options->a_prefix = optarg;
4811                return argcount;
4812        }
4813        else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4814                options->line_prefix = optarg;
4815                options->line_prefix_length = strlen(options->line_prefix);
4816                graph_setup_line_prefix(options);
4817                return argcount;
4818        }
4819        else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4820                options->b_prefix = optarg;
4821                return argcount;
4822        }
4823        else if (!strcmp(arg, "--no-prefix"))
4824                options->a_prefix = options->b_prefix = "";
4825        else if (opt_arg(arg, '\0', "inter-hunk-context",
4826                         &options->interhunkcontext))
4827                ;
4828        else if (!strcmp(arg, "-W"))
4829                options->flags.funccontext = 1;
4830        else if (!strcmp(arg, "--function-context"))
4831                options->flags.funccontext = 1;
4832        else if (!strcmp(arg, "--no-function-context"))
4833                options->flags.funccontext = 0;
4834        else if ((argcount = parse_long_opt("output", av, &optarg))) {
4835                char *path = prefix_filename(prefix, optarg);
4836                options->file = xfopen(path, "w");
4837                options->close_file = 1;
4838                if (options->use_color != GIT_COLOR_ALWAYS)
4839                        options->use_color = GIT_COLOR_NEVER;
4840                free(path);
4841                return argcount;
4842        } else
4843                return 0;
4844        return 1;
4845}
4846
4847int parse_rename_score(const char **cp_p)
4848{
4849        unsigned long num, scale;
4850        int ch, dot;
4851        const char *cp = *cp_p;
4852
4853        num = 0;
4854        scale = 1;
4855        dot = 0;
4856        for (;;) {
4857                ch = *cp;
4858                if ( !dot && ch == '.' ) {
4859                        scale = 1;
4860                        dot = 1;
4861                } else if ( ch == '%' ) {
4862                        scale = dot ? scale*100 : 100;
4863                        cp++;   /* % is always at the end */
4864                        break;
4865                } else if ( ch >= '0' && ch <= '9' ) {
4866                        if ( scale < 100000 ) {
4867                                scale *= 10;
4868                                num = (num*10) + (ch-'0');
4869                        }
4870                } else {
4871                        break;
4872                }
4873                cp++;
4874        }
4875        *cp_p = cp;
4876
4877        /* user says num divided by scale and we say internally that
4878         * is MAX_SCORE * num / scale.
4879         */
4880        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4881}
4882
4883static int diff_scoreopt_parse(const char *opt)
4884{
4885        int opt1, opt2, cmd;
4886
4887        if (*opt++ != '-')
4888                return -1;
4889        cmd = *opt++;
4890        if (cmd == '-') {
4891                /* convert the long-form arguments into short-form versions */
4892                if (skip_prefix(opt, "break-rewrites", &opt)) {
4893                        if (*opt == 0 || *opt++ == '=')
4894                                cmd = 'B';
4895                } else if (skip_prefix(opt, "find-copies", &opt)) {
4896                        if (*opt == 0 || *opt++ == '=')
4897                                cmd = 'C';
4898                } else if (skip_prefix(opt, "find-renames", &opt)) {
4899                        if (*opt == 0 || *opt++ == '=')
4900                                cmd = 'M';
4901                }
4902        }
4903        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4904                return -1; /* that is not a -M, -C, or -B option */
4905
4906        opt1 = parse_rename_score(&opt);
4907        if (cmd != 'B')
4908                opt2 = 0;
4909        else {
4910                if (*opt == 0)
4911                        opt2 = 0;
4912                else if (*opt != '/')
4913                        return -1; /* we expect -B80/99 or -B80 */
4914                else {
4915                        opt++;
4916                        opt2 = parse_rename_score(&opt);
4917                }
4918        }
4919        if (*opt != 0)
4920                return -1;
4921        return opt1 | (opt2 << 16);
4922}
4923
4924struct diff_queue_struct diff_queued_diff;
4925
4926void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4927{
4928        ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4929        queue->queue[queue->nr++] = dp;
4930}
4931
4932struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4933                                 struct diff_filespec *one,
4934                                 struct diff_filespec *two)
4935{
4936        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4937        dp->one = one;
4938        dp->two = two;
4939        if (queue)
4940                diff_q(queue, dp);
4941        return dp;
4942}
4943
4944void diff_free_filepair(struct diff_filepair *p)
4945{
4946        free_filespec(p->one);
4947        free_filespec(p->two);
4948        free(p);
4949}
4950
4951const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4952{
4953        int abblen;
4954        const char *abbrev;
4955
4956        /* Do we want all 40 hex characters? */
4957        if (len == GIT_SHA1_HEXSZ)
4958                return oid_to_hex(oid);
4959
4960        /* An abbreviated value is fine, possibly followed by an ellipsis. */
4961        abbrev = diff_abbrev_oid(oid, len);
4962
4963        if (!print_sha1_ellipsis())
4964                return abbrev;
4965
4966        abblen = strlen(abbrev);
4967
4968        /*
4969         * In well-behaved cases, where the abbreviated result is the
4970         * same as the requested length, append three dots after the
4971         * abbreviation (hence the whole logic is limited to the case
4972         * where abblen < 37); when the actual abbreviated result is a
4973         * bit longer than the requested length, we reduce the number
4974         * of dots so that they match the well-behaved ones.  However,
4975         * if the actual abbreviation is longer than the requested
4976         * length by more than three, we give up on aligning, and add
4977         * three dots anyway, to indicate that the output is not the
4978         * full object name.  Yes, this may be suboptimal, but this
4979         * appears only in "diff --raw --abbrev" output and it is not
4980         * worth the effort to change it now.  Note that this would
4981         * likely to work fine when the automatic sizing of default
4982         * abbreviation length is used--we would be fed -1 in "len" in
4983         * that case, and will end up always appending three-dots, but
4984         * the automatic sizing is supposed to give abblen that ensures
4985         * uniqueness across all objects (statistically speaking).
4986         */
4987        if (abblen < GIT_SHA1_HEXSZ - 3) {
4988                static char hex[GIT_MAX_HEXSZ + 1];
4989                if (len < abblen && abblen <= len + 2)
4990                        xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4991                else
4992                        xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4993                return hex;
4994        }
4995
4996        return oid_to_hex(oid);
4997}
4998
4999static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5000{
5001        int line_termination = opt->line_termination;
5002        int inter_name_termination = line_termination ? '\t' : '\0';
5003
5004        fprintf(opt->file, "%s", diff_line_prefix(opt));
5005        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5006                fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5007                        diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5008                fprintf(opt->file, "%s ",
5009                        diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5010        }
5011        if (p->score) {
5012                fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5013                        inter_name_termination);
5014        } else {
5015                fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5016        }
5017
5018        if (p->status == DIFF_STATUS_COPIED ||
5019            p->status == DIFF_STATUS_RENAMED) {
5020                const char *name_a, *name_b;
5021                name_a = p->one->path;
5022                name_b = p->two->path;
5023                strip_prefix(opt->prefix_length, &name_a, &name_b);
5024                write_name_quoted(name_a, opt->file, inter_name_termination);
5025                write_name_quoted(name_b, opt->file, line_termination);
5026        } else {
5027                const char *name_a, *name_b;
5028                name_a = p->one->mode ? p->one->path : p->two->path;
5029                name_b = NULL;
5030                strip_prefix(opt->prefix_length, &name_a, &name_b);
5031                write_name_quoted(name_a, opt->file, line_termination);
5032        }
5033}
5034
5035int diff_unmodified_pair(struct diff_filepair *p)
5036{
5037        /* This function is written stricter than necessary to support
5038         * the currently implemented transformers, but the idea is to
5039         * let transformers to produce diff_filepairs any way they want,
5040         * and filter and clean them up here before producing the output.
5041         */
5042        struct diff_filespec *one = p->one, *two = p->two;
5043
5044        if (DIFF_PAIR_UNMERGED(p))
5045                return 0; /* unmerged is interesting */
5046
5047        /* deletion, addition, mode or type change
5048         * and rename are all interesting.
5049         */
5050        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5051            DIFF_PAIR_MODE_CHANGED(p) ||
5052            strcmp(one->path, two->path))
5053                return 0;
5054
5055        /* both are valid and point at the same path.  that is, we are
5056         * dealing with a change.
5057         */
5058        if (one->oid_valid && two->oid_valid &&
5059            !oidcmp(&one->oid, &two->oid) &&
5060            !one->dirty_submodule && !two->dirty_submodule)
5061                return 1; /* no change */
5062        if (!one->oid_valid && !two->oid_valid)
5063                return 1; /* both look at the same file on the filesystem. */
5064        return 0;
5065}
5066
5067static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5068{
5069        if (diff_unmodified_pair(p))
5070                return;
5071
5072        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5073            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5074                return; /* no tree diffs in patch format */
5075
5076        run_diff(p, o);
5077}
5078
5079static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5080                            struct diffstat_t *diffstat)
5081{
5082        if (diff_unmodified_pair(p))
5083                return;
5084
5085        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5086            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5087                return; /* no useful stat for tree diffs */
5088
5089        run_diffstat(p, o, diffstat);
5090}
5091
5092static void diff_flush_checkdiff(struct diff_filepair *p,
5093                struct diff_options *o)
5094{
5095        if (diff_unmodified_pair(p))
5096                return;
5097
5098        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5099            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5100                return; /* nothing to check in tree diffs */
5101
5102        run_checkdiff(p, o);
5103}
5104
5105int diff_queue_is_empty(void)
5106{
5107        struct diff_queue_struct *q = &diff_queued_diff;
5108        int i;
5109        for (i = 0; i < q->nr; i++)
5110                if (!diff_unmodified_pair(q->queue[i]))
5111                        return 0;
5112        return 1;
5113}
5114
5115#if DIFF_DEBUG
5116void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5117{
5118        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5119                x, one ? one : "",
5120                s->path,
5121                DIFF_FILE_VALID(s) ? "valid" : "invalid",
5122                s->mode,
5123                s->oid_valid ? oid_to_hex(&s->oid) : "");
5124        fprintf(stderr, "queue[%d] %s size %lu\n",
5125                x, one ? one : "",
5126                s->size);
5127}
5128
5129void diff_debug_filepair(const struct diff_filepair *p, int i)
5130{
5131        diff_debug_filespec(p->one, i, "one");
5132        diff_debug_filespec(p->two, i, "two");
5133        fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5134                p->score, p->status ? p->status : '?',
5135                p->one->rename_used, p->broken_pair);
5136}
5137
5138void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5139{
5140        int i;
5141        if (msg)
5142                fprintf(stderr, "%s\n", msg);
5143        fprintf(stderr, "q->nr = %d\n", q->nr);
5144        for (i = 0; i < q->nr; i++) {
5145                struct diff_filepair *p = q->queue[i];
5146                diff_debug_filepair(p, i);
5147        }
5148}
5149#endif
5150
5151static void diff_resolve_rename_copy(void)
5152{
5153        int i;
5154        struct diff_filepair *p;
5155        struct diff_queue_struct *q = &diff_queued_diff;
5156
5157        diff_debug_queue("resolve-rename-copy", q);
5158
5159        for (i = 0; i < q->nr; i++) {
5160                p = q->queue[i];
5161                p->status = 0; /* undecided */
5162                if (DIFF_PAIR_UNMERGED(p))
5163                        p->status = DIFF_STATUS_UNMERGED;
5164                else if (!DIFF_FILE_VALID(p->one))
5165                        p->status = DIFF_STATUS_ADDED;
5166                else if (!DIFF_FILE_VALID(p->two))
5167                        p->status = DIFF_STATUS_DELETED;
5168                else if (DIFF_PAIR_TYPE_CHANGED(p))
5169                        p->status = DIFF_STATUS_TYPE_CHANGED;
5170
5171                /* from this point on, we are dealing with a pair
5172                 * whose both sides are valid and of the same type, i.e.
5173                 * either in-place edit or rename/copy edit.
5174                 */
5175                else if (DIFF_PAIR_RENAME(p)) {
5176                        /*
5177                         * A rename might have re-connected a broken
5178                         * pair up, causing the pathnames to be the
5179                         * same again. If so, that's not a rename at
5180                         * all, just a modification..
5181                         *
5182                         * Otherwise, see if this source was used for
5183                         * multiple renames, in which case we decrement
5184                         * the count, and call it a copy.
5185                         */
5186                        if (!strcmp(p->one->path, p->two->path))
5187                                p->status = DIFF_STATUS_MODIFIED;
5188                        else if (--p->one->rename_used > 0)
5189                                p->status = DIFF_STATUS_COPIED;
5190                        else
5191                                p->status = DIFF_STATUS_RENAMED;
5192                }
5193                else if (oidcmp(&p->one->oid, &p->two->oid) ||
5194                         p->one->mode != p->two->mode ||
5195                         p->one->dirty_submodule ||
5196                         p->two->dirty_submodule ||
5197                         is_null_oid(&p->one->oid))
5198                        p->status = DIFF_STATUS_MODIFIED;
5199                else {
5200                        /* This is a "no-change" entry and should not
5201                         * happen anymore, but prepare for broken callers.
5202                         */
5203                        error("feeding unmodified %s to diffcore",
5204                              p->one->path);
5205                        p->status = DIFF_STATUS_UNKNOWN;
5206                }
5207        }
5208        diff_debug_queue("resolve-rename-copy done", q);
5209}
5210
5211static int check_pair_status(struct diff_filepair *p)
5212{
5213        switch (p->status) {
5214        case DIFF_STATUS_UNKNOWN:
5215                return 0;
5216        case 0:
5217                die("internal error in diff-resolve-rename-copy");
5218        default:
5219                return 1;
5220        }
5221}
5222
5223static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5224{
5225        int fmt = opt->output_format;
5226
5227        if (fmt & DIFF_FORMAT_CHECKDIFF)
5228                diff_flush_checkdiff(p, opt);
5229        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5230                diff_flush_raw(p, opt);
5231        else if (fmt & DIFF_FORMAT_NAME) {
5232                const char *name_a, *name_b;
5233                name_a = p->two->path;
5234                name_b = NULL;
5235                strip_prefix(opt->prefix_length, &name_a, &name_b);
5236                fprintf(opt->file, "%s", diff_line_prefix(opt));
5237                write_name_quoted(name_a, opt->file, opt->line_termination);
5238        }
5239}
5240
5241static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5242{
5243        struct strbuf sb = STRBUF_INIT;
5244        if (fs->mode)
5245                strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5246        else
5247                strbuf_addf(&sb, " %s ", newdelete);
5248
5249        quote_c_style(fs->path, &sb, NULL, 0);
5250        strbuf_addch(&sb, '\n');
5251        emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5252                         sb.buf, sb.len, 0);
5253        strbuf_release(&sb);
5254}
5255
5256static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5257                int show_name)
5258{
5259        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5260                struct strbuf sb = STRBUF_INIT;
5261                strbuf_addf(&sb, " mode change %06o => %06o",
5262                            p->one->mode, p->two->mode);
5263                if (show_name) {
5264                        strbuf_addch(&sb, ' ');
5265                        quote_c_style(p->two->path, &sb, NULL, 0);
5266                }
5267                strbuf_addch(&sb, '\n');
5268                emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5269                                 sb.buf, sb.len, 0);
5270                strbuf_release(&sb);
5271        }
5272}
5273
5274static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5275                struct diff_filepair *p)
5276{
5277        struct strbuf sb = STRBUF_INIT;
5278        struct strbuf names = STRBUF_INIT;
5279
5280        pprint_rename(&names, p->one->path, p->two->path);
5281        strbuf_addf(&sb, " %s %s (%d%%)\n",
5282                    renamecopy, names.buf, similarity_index(p));
5283        strbuf_release(&names);
5284        emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5285                                 sb.buf, sb.len, 0);
5286        show_mode_change(opt, p, 0);
5287        strbuf_release(&sb);
5288}
5289
5290static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5291{
5292        switch(p->status) {
5293        case DIFF_STATUS_DELETED:
5294                show_file_mode_name(opt, "delete", p->one);
5295                break;
5296        case DIFF_STATUS_ADDED:
5297                show_file_mode_name(opt, "create", p->two);
5298                break;
5299        case DIFF_STATUS_COPIED:
5300                show_rename_copy(opt, "copy", p);
5301                break;
5302        case DIFF_STATUS_RENAMED:
5303                show_rename_copy(opt, "rename", p);
5304                break;
5305        default:
5306                if (p->score) {
5307                        struct strbuf sb = STRBUF_INIT;
5308                        strbuf_addstr(&sb, " rewrite ");
5309                        quote_c_style(p->two->path, &sb, NULL, 0);
5310                        strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5311                        emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5312                                         sb.buf, sb.len, 0);
5313                        strbuf_release(&sb);
5314                }
5315                show_mode_change(opt, p, !p->score);
5316                break;
5317        }
5318}
5319
5320struct patch_id_t {
5321        git_SHA_CTX *ctx;
5322        int patchlen;
5323};
5324
5325static int remove_space(char *line, int len)
5326{
5327        int i;
5328        char *dst = line;
5329        unsigned char c;
5330
5331        for (i = 0; i < len; i++)
5332                if (!isspace((c = line[i])))
5333                        *dst++ = c;
5334
5335        return dst - line;
5336}
5337
5338static void patch_id_consume(void *priv, char *line, unsigned long len)
5339{
5340        struct patch_id_t *data = priv;
5341        int new_len;
5342
5343        /* Ignore line numbers when computing the SHA1 of the patch */
5344        if (starts_with(line, "@@ -"))
5345                return;
5346
5347        new_len = remove_space(line, len);
5348
5349        git_SHA1_Update(data->ctx, line, new_len);
5350        data->patchlen += new_len;
5351}
5352
5353static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5354{
5355        git_SHA1_Update(ctx, str, strlen(str));
5356}
5357
5358static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5359{
5360        /* large enough for 2^32 in octal */
5361        char buf[12];
5362        int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5363        git_SHA1_Update(ctx, buf, len);
5364}
5365
5366/* returns 0 upon success, and writes result into sha1 */
5367static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5368{
5369        struct diff_queue_struct *q = &diff_queued_diff;
5370        int i;
5371        git_SHA_CTX ctx;
5372        struct patch_id_t data;
5373
5374        git_SHA1_Init(&ctx);
5375        memset(&data, 0, sizeof(struct patch_id_t));
5376        data.ctx = &ctx;
5377
5378        for (i = 0; i < q->nr; i++) {
5379                xpparam_t xpp;
5380                xdemitconf_t xecfg;
5381                mmfile_t mf1, mf2;
5382                struct diff_filepair *p = q->queue[i];
5383                int len1, len2;
5384
5385                memset(&xpp, 0, sizeof(xpp));
5386                memset(&xecfg, 0, sizeof(xecfg));
5387                if (p->status == 0)
5388                        return error("internal diff status error");
5389                if (p->status == DIFF_STATUS_UNKNOWN)
5390                        continue;
5391                if (diff_unmodified_pair(p))
5392                        continue;
5393                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5394                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5395                        continue;
5396                if (DIFF_PAIR_UNMERGED(p))
5397                        continue;
5398
5399                diff_fill_oid_info(p->one);
5400                diff_fill_oid_info(p->two);
5401
5402                len1 = remove_space(p->one->path, strlen(p->one->path));
5403                len2 = remove_space(p->two->path, strlen(p->two->path));
5404                patch_id_add_string(&ctx, "diff--git");
5405                patch_id_add_string(&ctx, "a/");
5406                git_SHA1_Update(&ctx, p->one->path, len1);
5407                patch_id_add_string(&ctx, "b/");
5408                git_SHA1_Update(&ctx, p->two->path, len2);
5409
5410                if (p->one->mode == 0) {
5411                        patch_id_add_string(&ctx, "newfilemode");
5412                        patch_id_add_mode(&ctx, p->two->mode);
5413                        patch_id_add_string(&ctx, "---/dev/null");
5414                        patch_id_add_string(&ctx, "+++b/");
5415                        git_SHA1_Update(&ctx, p->two->path, len2);
5416                } else if (p->two->mode == 0) {
5417                        patch_id_add_string(&ctx, "deletedfilemode");
5418                        patch_id_add_mode(&ctx, p->one->mode);
5419                        patch_id_add_string(&ctx, "---a/");
5420                        git_SHA1_Update(&ctx, p->one->path, len1);
5421                        patch_id_add_string(&ctx, "+++/dev/null");
5422                } else {
5423                        patch_id_add_string(&ctx, "---a/");
5424                        git_SHA1_Update(&ctx, p->one->path, len1);
5425                        patch_id_add_string(&ctx, "+++b/");
5426                        git_SHA1_Update(&ctx, p->two->path, len2);
5427                }
5428
5429                if (diff_header_only)
5430                        continue;
5431
5432                if (fill_mmfile(&mf1, p->one) < 0 ||
5433                    fill_mmfile(&mf2, p->two) < 0)
5434                        return error("unable to read files to diff");
5435
5436                if (diff_filespec_is_binary(p->one) ||
5437                    diff_filespec_is_binary(p->two)) {
5438                        git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5439                                        GIT_SHA1_HEXSZ);
5440                        git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5441                                        GIT_SHA1_HEXSZ);
5442                        continue;
5443                }
5444
5445                xpp.flags = 0;
5446                xecfg.ctxlen = 3;
5447                xecfg.flags = 0;
5448                if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5449                                  &xpp, &xecfg))
5450                        return error("unable to generate patch-id diff for %s",
5451                                     p->one->path);
5452        }
5453
5454        git_SHA1_Final(oid->hash, &ctx);
5455        return 0;
5456}
5457
5458int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5459{
5460        struct diff_queue_struct *q = &diff_queued_diff;
5461        int i;
5462        int result = diff_get_patch_id(options, oid, diff_header_only);
5463
5464        for (i = 0; i < q->nr; i++)
5465                diff_free_filepair(q->queue[i]);
5466
5467        free(q->queue);
5468        DIFF_QUEUE_CLEAR(q);
5469
5470        return result;
5471}
5472
5473static int is_summary_empty(const struct diff_queue_struct *q)
5474{
5475        int i;
5476
5477        for (i = 0; i < q->nr; i++) {
5478                const struct diff_filepair *p = q->queue[i];
5479
5480                switch (p->status) {
5481                case DIFF_STATUS_DELETED:
5482                case DIFF_STATUS_ADDED:
5483                case DIFF_STATUS_COPIED:
5484                case DIFF_STATUS_RENAMED:
5485                        return 0;
5486                default:
5487                        if (p->score)
5488                                return 0;
5489                        if (p->one->mode && p->two->mode &&
5490                            p->one->mode != p->two->mode)
5491                                return 0;
5492                        break;
5493                }
5494        }
5495        return 1;
5496}
5497
5498static const char rename_limit_warning[] =
5499N_("inexact rename detection was skipped due to too many files.");
5500
5501static const char degrade_cc_to_c_warning[] =
5502N_("only found copies from modified paths due to too many files.");
5503
5504static const char rename_limit_advice[] =
5505N_("you may want to set your %s variable to at least "
5506   "%d and retry the command.");
5507
5508void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5509{
5510        fflush(stdout);
5511        if (degraded_cc)
5512                warning(_(degrade_cc_to_c_warning));
5513        else if (needed)
5514                warning(_(rename_limit_warning));
5515        else
5516                return;
5517        if (0 < needed)
5518                warning(_(rename_limit_advice), varname, needed);
5519}
5520
5521static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5522{
5523        int i;
5524        static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5525        struct diff_queue_struct *q = &diff_queued_diff;
5526
5527        if (WSEH_NEW & WS_RULE_MASK)
5528                die("BUG: WS rules bit mask overlaps with diff symbol flags");
5529
5530        if (o->color_moved)
5531                o->emitted_symbols = &esm;
5532
5533        for (i = 0; i < q->nr; i++) {
5534                struct diff_filepair *p = q->queue[i];
5535                if (check_pair_status(p))
5536                        diff_flush_patch(p, o);
5537        }
5538
5539        if (o->emitted_symbols) {
5540                if (o->color_moved) {
5541                        struct hashmap add_lines, del_lines;
5542
5543                        hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5544                        hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5545
5546                        add_lines_to_move_detection(o, &add_lines, &del_lines);
5547                        mark_color_as_moved(o, &add_lines, &del_lines);
5548                        if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5549                                dim_moved_lines(o);
5550
5551                        hashmap_free(&add_lines, 0);
5552                        hashmap_free(&del_lines, 0);
5553                }
5554
5555                for (i = 0; i < esm.nr; i++)
5556                        emit_diff_symbol_from_struct(o, &esm.buf[i]);
5557
5558                for (i = 0; i < esm.nr; i++)
5559                        free((void *)esm.buf[i].line);
5560        }
5561        esm.nr = 0;
5562}
5563
5564void diff_flush(struct diff_options *options)
5565{
5566        struct diff_queue_struct *q = &diff_queued_diff;
5567        int i, output_format = options->output_format;
5568        int separator = 0;
5569        int dirstat_by_line = 0;
5570
5571        /*
5572         * Order: raw, stat, summary, patch
5573         * or:    name/name-status/checkdiff (other bits clear)
5574         */
5575        if (!q->nr)
5576                goto free_queue;
5577
5578        if (output_format & (DIFF_FORMAT_RAW |
5579                             DIFF_FORMAT_NAME |
5580                             DIFF_FORMAT_NAME_STATUS |
5581                             DIFF_FORMAT_CHECKDIFF)) {
5582                for (i = 0; i < q->nr; i++) {
5583                        struct diff_filepair *p = q->queue[i];
5584                        if (check_pair_status(p))
5585                                flush_one_pair(p, options);
5586                }
5587                separator++;
5588        }
5589
5590        if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5591                dirstat_by_line = 1;
5592
5593        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5594            dirstat_by_line) {
5595                struct diffstat_t diffstat;
5596
5597                memset(&diffstat, 0, sizeof(struct diffstat_t));
5598                for (i = 0; i < q->nr; i++) {
5599                        struct diff_filepair *p = q->queue[i];
5600                        if (check_pair_status(p))
5601                                diff_flush_stat(p, options, &diffstat);
5602                }
5603                if (output_format & DIFF_FORMAT_NUMSTAT)
5604                        show_numstat(&diffstat, options);
5605                if (output_format & DIFF_FORMAT_DIFFSTAT)
5606                        show_stats(&diffstat, options);
5607                if (output_format & DIFF_FORMAT_SHORTSTAT)
5608                        show_shortstats(&diffstat, options);
5609                if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5610                        show_dirstat_by_line(&diffstat, options);
5611                free_diffstat_info(&diffstat);
5612                separator++;
5613        }
5614        if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5615                show_dirstat(options);
5616
5617        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5618                for (i = 0; i < q->nr; i++) {
5619                        diff_summary(options, q->queue[i]);
5620                }
5621                separator++;
5622        }
5623
5624        if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5625            options->flags.exit_with_status &&
5626            options->flags.diff_from_contents) {
5627                /*
5628                 * run diff_flush_patch for the exit status. setting
5629                 * options->file to /dev/null should be safe, because we
5630                 * aren't supposed to produce any output anyway.
5631                 */
5632                if (options->close_file)
5633                        fclose(options->file);
5634                options->file = xfopen("/dev/null", "w");
5635                options->close_file = 1;
5636                options->color_moved = 0;
5637                for (i = 0; i < q->nr; i++) {
5638                        struct diff_filepair *p = q->queue[i];
5639                        if (check_pair_status(p))
5640                                diff_flush_patch(p, options);
5641                        if (options->found_changes)
5642                                break;
5643                }
5644        }
5645
5646        if (output_format & DIFF_FORMAT_PATCH) {
5647                if (separator) {
5648                        emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5649                        if (options->stat_sep)
5650                                /* attach patch instead of inline */
5651                                emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5652                                                 NULL, 0, 0);
5653                }
5654
5655                diff_flush_patch_all_file_pairs(options);
5656        }
5657
5658        if (output_format & DIFF_FORMAT_CALLBACK)
5659                options->format_callback(q, options, options->format_callback_data);
5660
5661        for (i = 0; i < q->nr; i++)
5662                diff_free_filepair(q->queue[i]);
5663free_queue:
5664        free(q->queue);
5665        DIFF_QUEUE_CLEAR(q);
5666        if (options->close_file)
5667                fclose(options->file);
5668
5669        /*
5670         * Report the content-level differences with HAS_CHANGES;
5671         * diff_addremove/diff_change does not set the bit when
5672         * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5673         */
5674        if (options->flags.diff_from_contents) {
5675                if (options->found_changes)
5676                        options->flags.has_changes = 1;
5677                else
5678                        options->flags.has_changes = 0;
5679        }
5680}
5681
5682static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5683{
5684        return (((p->status == DIFF_STATUS_MODIFIED) &&
5685                 ((p->score &&
5686                   filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5687                  (!p->score &&
5688                   filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5689                ((p->status != DIFF_STATUS_MODIFIED) &&
5690                 filter_bit_tst(p->status, options)));
5691}
5692
5693static void diffcore_apply_filter(struct diff_options *options)
5694{
5695        int i;
5696        struct diff_queue_struct *q = &diff_queued_diff;
5697        struct diff_queue_struct outq;
5698
5699        DIFF_QUEUE_CLEAR(&outq);
5700
5701        if (!options->filter)
5702                return;
5703
5704        if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5705                int found;
5706                for (i = found = 0; !found && i < q->nr; i++) {
5707                        if (match_filter(options, q->queue[i]))
5708                                found++;
5709                }
5710                if (found)
5711                        return;
5712
5713                /* otherwise we will clear the whole queue
5714                 * by copying the empty outq at the end of this
5715                 * function, but first clear the current entries
5716                 * in the queue.
5717                 */
5718                for (i = 0; i < q->nr; i++)
5719                        diff_free_filepair(q->queue[i]);
5720        }
5721        else {
5722                /* Only the matching ones */
5723                for (i = 0; i < q->nr; i++) {
5724                        struct diff_filepair *p = q->queue[i];
5725                        if (match_filter(options, p))
5726                                diff_q(&outq, p);
5727                        else
5728                                diff_free_filepair(p);
5729                }
5730        }
5731        free(q->queue);
5732        *q = outq;
5733}
5734
5735/* Check whether two filespecs with the same mode and size are identical */
5736static int diff_filespec_is_identical(struct diff_filespec *one,
5737                                      struct diff_filespec *two)
5738{
5739        if (S_ISGITLINK(one->mode))
5740                return 0;
5741        if (diff_populate_filespec(one, 0))
5742                return 0;
5743        if (diff_populate_filespec(two, 0))
5744                return 0;
5745        return !memcmp(one->data, two->data, one->size);
5746}
5747
5748static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5749{
5750        if (p->done_skip_stat_unmatch)
5751                return p->skip_stat_unmatch_result;
5752
5753        p->done_skip_stat_unmatch = 1;
5754        p->skip_stat_unmatch_result = 0;
5755        /*
5756         * 1. Entries that come from stat info dirtiness
5757         *    always have both sides (iow, not create/delete),
5758         *    one side of the object name is unknown, with
5759         *    the same mode and size.  Keep the ones that
5760         *    do not match these criteria.  They have real
5761         *    differences.
5762         *
5763         * 2. At this point, the file is known to be modified,
5764         *    with the same mode and size, and the object
5765         *    name of one side is unknown.  Need to inspect
5766         *    the identical contents.
5767         */
5768        if (!DIFF_FILE_VALID(p->one) || /* (1) */
5769            !DIFF_FILE_VALID(p->two) ||
5770            (p->one->oid_valid && p->two->oid_valid) ||
5771            (p->one->mode != p->two->mode) ||
5772            diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5773            diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5774            (p->one->size != p->two->size) ||
5775            !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5776                p->skip_stat_unmatch_result = 1;
5777        return p->skip_stat_unmatch_result;
5778}
5779
5780static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5781{
5782        int i;
5783        struct diff_queue_struct *q = &diff_queued_diff;
5784        struct diff_queue_struct outq;
5785        DIFF_QUEUE_CLEAR(&outq);
5786
5787        for (i = 0; i < q->nr; i++) {
5788                struct diff_filepair *p = q->queue[i];
5789
5790                if (diff_filespec_check_stat_unmatch(p))
5791                        diff_q(&outq, p);
5792                else {
5793                        /*
5794                         * The caller can subtract 1 from skip_stat_unmatch
5795                         * to determine how many paths were dirty only
5796                         * due to stat info mismatch.
5797                         */
5798                        if (!diffopt->flags.no_index)
5799                                diffopt->skip_stat_unmatch++;
5800                        diff_free_filepair(p);
5801                }
5802        }
5803        free(q->queue);
5804        *q = outq;
5805}
5806
5807static int diffnamecmp(const void *a_, const void *b_)
5808{
5809        const struct diff_filepair *a = *((const struct diff_filepair **)a_);
5810        const struct diff_filepair *b = *((const struct diff_filepair **)b_);
5811        const char *name_a, *name_b;
5812
5813        name_a = a->one ? a->one->path : a->two->path;
5814        name_b = b->one ? b->one->path : b->two->path;
5815        return strcmp(name_a, name_b);
5816}
5817
5818void diffcore_fix_diff_index(struct diff_options *options)
5819{
5820        struct diff_queue_struct *q = &diff_queued_diff;
5821        QSORT(q->queue, q->nr, diffnamecmp);
5822}
5823
5824void diffcore_std(struct diff_options *options)
5825{
5826        /* NOTE please keep the following in sync with diff_tree_combined() */
5827        if (options->skip_stat_unmatch)
5828                diffcore_skip_stat_unmatch(options);
5829        if (!options->found_follow) {
5830                /* See try_to_follow_renames() in tree-diff.c */
5831                if (options->break_opt != -1)
5832                        diffcore_break(options->break_opt);
5833                if (options->detect_rename)
5834                        diffcore_rename(options);
5835                if (options->break_opt != -1)
5836                        diffcore_merge_broken();
5837        }
5838        if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
5839                diffcore_pickaxe(options);
5840        if (options->orderfile)
5841                diffcore_order(options->orderfile);
5842        if (!options->found_follow)
5843                /* See try_to_follow_renames() in tree-diff.c */
5844                diff_resolve_rename_copy();
5845        diffcore_apply_filter(options);
5846
5847        if (diff_queued_diff.nr && !options->flags.diff_from_contents)
5848                options->flags.has_changes = 1;
5849        else
5850                options->flags.has_changes = 0;
5851
5852        options->found_follow = 0;
5853}
5854
5855int diff_result_code(struct diff_options *opt, int status)
5856{
5857        int result = 0;
5858
5859        diff_warn_rename_limit("diff.renameLimit",
5860                               opt->needed_rename_limit,
5861                               opt->degraded_cc_to_c);
5862        if (!opt->flags.exit_with_status &&
5863            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5864                return status;
5865        if (opt->flags.exit_with_status &&
5866            opt->flags.has_changes)
5867                result |= 01;
5868        if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5869            opt->flags.check_failed)
5870                result |= 02;
5871        return result;
5872}
5873
5874int diff_can_quit_early(struct diff_options *opt)
5875{
5876        return (opt->flags.quick &&
5877                !opt->filter &&
5878                opt->flags.has_changes);
5879}
5880
5881/*
5882 * Shall changes to this submodule be ignored?
5883 *
5884 * Submodule changes can be configured to be ignored separately for each path,
5885 * but that configuration can be overridden from the command line.
5886 */
5887static int is_submodule_ignored(const char *path, struct diff_options *options)
5888{
5889        int ignored = 0;
5890        struct diff_flags orig_flags = options->flags;
5891        if (!options->flags.override_submodule_config)
5892                set_diffopt_flags_from_submodule_config(options, path);
5893        if (options->flags.ignore_submodules)
5894                ignored = 1;
5895        options->flags = orig_flags;
5896        return ignored;
5897}
5898
5899void diff_addremove(struct diff_options *options,
5900                    int addremove, unsigned mode,
5901                    const struct object_id *oid,
5902                    int oid_valid,
5903                    const char *concatpath, unsigned dirty_submodule)
5904{
5905        struct diff_filespec *one, *two;
5906
5907        if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5908                return;
5909
5910        /* This may look odd, but it is a preparation for
5911         * feeding "there are unchanged files which should
5912         * not produce diffs, but when you are doing copy
5913         * detection you would need them, so here they are"
5914         * entries to the diff-core.  They will be prefixed
5915         * with something like '=' or '*' (I haven't decided
5916         * which but should not make any difference).
5917         * Feeding the same new and old to diff_change()
5918         * also has the same effect.
5919         * Before the final output happens, they are pruned after
5920         * merged into rename/copy pairs as appropriate.
5921         */
5922        if (options->flags.reverse_diff)
5923                addremove = (addremove == '+' ? '-' :
5924                             addremove == '-' ? '+' : addremove);
5925
5926        if (options->prefix &&
5927            strncmp(concatpath, options->prefix, options->prefix_length))
5928                return;
5929
5930        one = alloc_filespec(concatpath);
5931        two = alloc_filespec(concatpath);
5932
5933        if (addremove != '+')
5934                fill_filespec(one, oid, oid_valid, mode);
5935        if (addremove != '-') {
5936                fill_filespec(two, oid, oid_valid, mode);
5937                two->dirty_submodule = dirty_submodule;
5938        }
5939
5940        diff_queue(&diff_queued_diff, one, two);
5941        if (!options->flags.diff_from_contents)
5942                options->flags.has_changes = 1;
5943}
5944
5945void diff_change(struct diff_options *options,
5946                 unsigned old_mode, unsigned new_mode,
5947                 const struct object_id *old_oid,
5948                 const struct object_id *new_oid,
5949                 int old_oid_valid, int new_oid_valid,
5950                 const char *concatpath,
5951                 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5952{
5953        struct diff_filespec *one, *two;
5954        struct diff_filepair *p;
5955
5956        if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5957            is_submodule_ignored(concatpath, options))
5958                return;
5959
5960        if (options->flags.reverse_diff) {
5961                SWAP(old_mode, new_mode);
5962                SWAP(old_oid, new_oid);
5963                SWAP(old_oid_valid, new_oid_valid);
5964                SWAP(old_dirty_submodule, new_dirty_submodule);
5965        }
5966
5967        if (options->prefix &&
5968            strncmp(concatpath, options->prefix, options->prefix_length))
5969                return;
5970
5971        one = alloc_filespec(concatpath);
5972        two = alloc_filespec(concatpath);
5973        fill_filespec(one, old_oid, old_oid_valid, old_mode);
5974        fill_filespec(two, new_oid, new_oid_valid, new_mode);
5975        one->dirty_submodule = old_dirty_submodule;
5976        two->dirty_submodule = new_dirty_submodule;
5977        p = diff_queue(&diff_queued_diff, one, two);
5978
5979        if (options->flags.diff_from_contents)
5980                return;
5981
5982        if (options->flags.quick && options->skip_stat_unmatch &&
5983            !diff_filespec_check_stat_unmatch(p))
5984                return;
5985
5986        options->flags.has_changes = 1;
5987}
5988
5989struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5990{
5991        struct diff_filepair *pair;
5992        struct diff_filespec *one, *two;
5993
5994        if (options->prefix &&
5995            strncmp(path, options->prefix, options->prefix_length))
5996                return NULL;
5997
5998        one = alloc_filespec(path);
5999        two = alloc_filespec(path);
6000        pair = diff_queue(&diff_queued_diff, one, two);
6001        pair->is_unmerged = 1;
6002        return pair;
6003}
6004
6005static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6006                size_t *outsize)
6007{
6008        struct diff_tempfile *temp;
6009        const char *argv[3];
6010        const char **arg = argv;
6011        struct child_process child = CHILD_PROCESS_INIT;
6012        struct strbuf buf = STRBUF_INIT;
6013        int err = 0;
6014
6015        temp = prepare_temp_file(spec->path, spec);
6016        *arg++ = pgm;
6017        *arg++ = temp->name;
6018        *arg = NULL;
6019
6020        child.use_shell = 1;
6021        child.argv = argv;
6022        child.out = -1;
6023        if (start_command(&child)) {
6024                remove_tempfile();
6025                return NULL;
6026        }
6027
6028        if (strbuf_read(&buf, child.out, 0) < 0)
6029                err = error("error reading from textconv command '%s'", pgm);
6030        close(child.out);
6031
6032        if (finish_command(&child) || err) {
6033                strbuf_release(&buf);
6034                remove_tempfile();
6035                return NULL;
6036        }
6037        remove_tempfile();
6038
6039        return strbuf_detach(&buf, outsize);
6040}
6041
6042size_t fill_textconv(struct userdiff_driver *driver,
6043                     struct diff_filespec *df,
6044                     char **outbuf)
6045{
6046        size_t size;
6047
6048        if (!driver) {
6049                if (!DIFF_FILE_VALID(df)) {
6050                        *outbuf = "";
6051                        return 0;
6052                }
6053                if (diff_populate_filespec(df, 0))
6054                        die("unable to read files to diff");
6055                *outbuf = df->data;
6056                return df->size;
6057        }
6058
6059        if (!driver->textconv)
6060                die("BUG: fill_textconv called with non-textconv driver");
6061
6062        if (driver->textconv_cache && df->oid_valid) {
6063                *outbuf = notes_cache_get(driver->textconv_cache,
6064                                          &df->oid,
6065                                          &size);
6066                if (*outbuf)
6067                        return size;
6068        }
6069
6070        *outbuf = run_textconv(driver->textconv, df, &size);
6071        if (!*outbuf)
6072                die("unable to read files to diff");
6073
6074        if (driver->textconv_cache && df->oid_valid) {
6075                /* ignore errors, as we might be in a readonly repository */
6076                notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6077                                size);
6078                /*
6079                 * we could save up changes and flush them all at the end,
6080                 * but we would need an extra call after all diffing is done.
6081                 * Since generating a cache entry is the slow path anyway,
6082                 * this extra overhead probably isn't a big deal.
6083                 */
6084                notes_cache_write(driver->textconv_cache);
6085        }
6086
6087        return size;
6088}
6089
6090int textconv_object(const char *path,
6091                    unsigned mode,
6092                    const struct object_id *oid,
6093                    int oid_valid,
6094                    char **buf,
6095                    unsigned long *buf_size)
6096{
6097        struct diff_filespec *df;
6098        struct userdiff_driver *textconv;
6099
6100        df = alloc_filespec(path);
6101        fill_filespec(df, oid, oid_valid, mode);
6102        textconv = get_textconv(df);
6103        if (!textconv) {
6104                free_filespec(df);
6105                return 0;
6106        }
6107
6108        *buf_size = fill_textconv(textconv, df, buf);
6109        free_filespec(df);
6110        return 1;
6111}
6112
6113void setup_diff_pager(struct diff_options *opt)
6114{
6115        /*
6116         * If the user asked for our exit code, then either they want --quiet
6117         * or --exit-code. We should definitely not bother with a pager in the
6118         * former case, as we will generate no output. Since we still properly
6119         * report our exit code even when a pager is run, we _could_ run a
6120         * pager with --exit-code. But since we have not done so historically,
6121         * and because it is easy to find people oneline advising "git diff
6122         * --exit-code" in hooks and other scripts, we do not do so.
6123         */
6124        if (!opt->flags.exit_with_status &&
6125            check_pager_config("diff") != 0)
6126                setup_pager();
6127}