diff.con commit Merge branch 'ew/send-email-readable-message-id' into maint (ffaa7c5)
   1/*
   2 * Copyright (C) 2005 Junio C Hamano
   3 */
   4#include "cache.h"
   5#include "tempfile.h"
   6#include "quote.h"
   7#include "diff.h"
   8#include "diffcore.h"
   9#include "delta.h"
  10#include "xdiff-interface.h"
  11#include "color.h"
  12#include "attr.h"
  13#include "run-command.h"
  14#include "utf8.h"
  15#include "userdiff.h"
  16#include "submodule-config.h"
  17#include "submodule.h"
  18#include "ll-merge.h"
  19#include "string-list.h"
  20#include "argv-array.h"
  21
  22#ifdef NO_FAST_WORKING_DIRECTORY
  23#define FAST_WORKING_DIRECTORY 0
  24#else
  25#define FAST_WORKING_DIRECTORY 1
  26#endif
  27
  28static int diff_detect_rename_default;
  29static int diff_rename_limit_default = 400;
  30static int diff_suppress_blank_empty;
  31static int diff_use_color_default = -1;
  32static int diff_context_default = 3;
  33static const char *diff_word_regex_cfg;
  34static const char *external_diff_cmd_cfg;
  35static const char *diff_order_file_cfg;
  36int diff_auto_refresh_index = 1;
  37static int diff_mnemonic_prefix;
  38static int diff_no_prefix;
  39static int diff_stat_graph_width;
  40static int diff_dirstat_permille_default = 30;
  41static struct diff_options default_diff_options;
  42static long diff_algorithm;
  43
  44static char diff_colors[][COLOR_MAXLEN] = {
  45        GIT_COLOR_RESET,
  46        GIT_COLOR_NORMAL,       /* CONTEXT */
  47        GIT_COLOR_BOLD,         /* METAINFO */
  48        GIT_COLOR_CYAN,         /* FRAGINFO */
  49        GIT_COLOR_RED,          /* OLD */
  50        GIT_COLOR_GREEN,        /* NEW */
  51        GIT_COLOR_YELLOW,       /* COMMIT */
  52        GIT_COLOR_BG_RED,       /* WHITESPACE */
  53        GIT_COLOR_NORMAL,       /* FUNCINFO */
  54};
  55
  56static int parse_diff_color_slot(const char *var)
  57{
  58        if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  59                return DIFF_CONTEXT;
  60        if (!strcasecmp(var, "meta"))
  61                return DIFF_METAINFO;
  62        if (!strcasecmp(var, "frag"))
  63                return DIFF_FRAGINFO;
  64        if (!strcasecmp(var, "old"))
  65                return DIFF_FILE_OLD;
  66        if (!strcasecmp(var, "new"))
  67                return DIFF_FILE_NEW;
  68        if (!strcasecmp(var, "commit"))
  69                return DIFF_COMMIT;
  70        if (!strcasecmp(var, "whitespace"))
  71                return DIFF_WHITESPACE;
  72        if (!strcasecmp(var, "func"))
  73                return DIFF_FUNCINFO;
  74        return -1;
  75}
  76
  77static int parse_dirstat_params(struct diff_options *options, const char *params_string,
  78                                struct strbuf *errmsg)
  79{
  80        char *params_copy = xstrdup(params_string);
  81        struct string_list params = STRING_LIST_INIT_NODUP;
  82        int ret = 0;
  83        int i;
  84
  85        if (*params_copy)
  86                string_list_split_in_place(&params, params_copy, ',', -1);
  87        for (i = 0; i < params.nr; i++) {
  88                const char *p = params.items[i].string;
  89                if (!strcmp(p, "changes")) {
  90                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  91                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  92                } else if (!strcmp(p, "lines")) {
  93                        DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  94                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  95                } else if (!strcmp(p, "files")) {
  96                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  97                        DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  98                } else if (!strcmp(p, "noncumulative")) {
  99                        DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
 100                } else if (!strcmp(p, "cumulative")) {
 101                        DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 102                } else if (isdigit(*p)) {
 103                        char *end;
 104                        int permille = strtoul(p, &end, 10) * 10;
 105                        if (*end == '.' && isdigit(*++end)) {
 106                                /* only use first digit */
 107                                permille += *end - '0';
 108                                /* .. and ignore any further digits */
 109                                while (isdigit(*++end))
 110                                        ; /* nothing */
 111                        }
 112                        if (!*end)
 113                                options->dirstat_permille = permille;
 114                        else {
 115                                strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 116                                            p);
 117                                ret++;
 118                        }
 119                } else {
 120                        strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 121                        ret++;
 122                }
 123
 124        }
 125        string_list_clear(&params, 0);
 126        free(params_copy);
 127        return ret;
 128}
 129
 130static int parse_submodule_params(struct diff_options *options, const char *value)
 131{
 132        if (!strcmp(value, "log"))
 133                DIFF_OPT_SET(options, SUBMODULE_LOG);
 134        else if (!strcmp(value, "short"))
 135                DIFF_OPT_CLR(options, SUBMODULE_LOG);
 136        else
 137                return -1;
 138        return 0;
 139}
 140
 141static int git_config_rename(const char *var, const char *value)
 142{
 143        if (!value)
 144                return DIFF_DETECT_RENAME;
 145        if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 146                return  DIFF_DETECT_COPY;
 147        return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 148}
 149
 150long parse_algorithm_value(const char *value)
 151{
 152        if (!value)
 153                return -1;
 154        else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 155                return 0;
 156        else if (!strcasecmp(value, "minimal"))
 157                return XDF_NEED_MINIMAL;
 158        else if (!strcasecmp(value, "patience"))
 159                return XDF_PATIENCE_DIFF;
 160        else if (!strcasecmp(value, "histogram"))
 161                return XDF_HISTOGRAM_DIFF;
 162        return -1;
 163}
 164
 165/*
 166 * These are to give UI layer defaults.
 167 * The core-level commands such as git-diff-files should
 168 * never be affected by the setting of diff.renames
 169 * the user happens to have in the configuration file.
 170 */
 171int git_diff_ui_config(const char *var, const char *value, void *cb)
 172{
 173        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 174                diff_use_color_default = git_config_colorbool(var, value);
 175                return 0;
 176        }
 177        if (!strcmp(var, "diff.context")) {
 178                diff_context_default = git_config_int(var, value);
 179                if (diff_context_default < 0)
 180                        return -1;
 181                return 0;
 182        }
 183        if (!strcmp(var, "diff.renames")) {
 184                diff_detect_rename_default = git_config_rename(var, value);
 185                return 0;
 186        }
 187        if (!strcmp(var, "diff.autorefreshindex")) {
 188                diff_auto_refresh_index = git_config_bool(var, value);
 189                return 0;
 190        }
 191        if (!strcmp(var, "diff.mnemonicprefix")) {
 192                diff_mnemonic_prefix = git_config_bool(var, value);
 193                return 0;
 194        }
 195        if (!strcmp(var, "diff.noprefix")) {
 196                diff_no_prefix = git_config_bool(var, value);
 197                return 0;
 198        }
 199        if (!strcmp(var, "diff.statgraphwidth")) {
 200                diff_stat_graph_width = git_config_int(var, value);
 201                return 0;
 202        }
 203        if (!strcmp(var, "diff.external"))
 204                return git_config_string(&external_diff_cmd_cfg, var, value);
 205        if (!strcmp(var, "diff.wordregex"))
 206                return git_config_string(&diff_word_regex_cfg, var, value);
 207        if (!strcmp(var, "diff.orderfile"))
 208                return git_config_pathname(&diff_order_file_cfg, var, value);
 209
 210        if (!strcmp(var, "diff.ignoresubmodules"))
 211                handle_ignore_submodules_arg(&default_diff_options, value);
 212
 213        if (!strcmp(var, "diff.submodule")) {
 214                if (parse_submodule_params(&default_diff_options, value))
 215                        warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 216                                value);
 217                return 0;
 218        }
 219
 220        if (!strcmp(var, "diff.algorithm")) {
 221                diff_algorithm = parse_algorithm_value(value);
 222                if (diff_algorithm < 0)
 223                        return -1;
 224                return 0;
 225        }
 226
 227        if (git_color_config(var, value, cb) < 0)
 228                return -1;
 229
 230        return git_diff_basic_config(var, value, cb);
 231}
 232
 233int git_diff_basic_config(const char *var, const char *value, void *cb)
 234{
 235        const char *name;
 236
 237        if (!strcmp(var, "diff.renamelimit")) {
 238                diff_rename_limit_default = git_config_int(var, value);
 239                return 0;
 240        }
 241
 242        if (userdiff_config(var, value) < 0)
 243                return -1;
 244
 245        if (skip_prefix(var, "diff.color.", &name) ||
 246            skip_prefix(var, "color.diff.", &name)) {
 247                int slot = parse_diff_color_slot(name);
 248                if (slot < 0)
 249                        return 0;
 250                if (!value)
 251                        return config_error_nonbool(var);
 252                return color_parse(value, diff_colors[slot]);
 253        }
 254
 255        /* like GNU diff's --suppress-blank-empty option  */
 256        if (!strcmp(var, "diff.suppressblankempty") ||
 257                        /* for backwards compatibility */
 258                        !strcmp(var, "diff.suppress-blank-empty")) {
 259                diff_suppress_blank_empty = git_config_bool(var, value);
 260                return 0;
 261        }
 262
 263        if (!strcmp(var, "diff.dirstat")) {
 264                struct strbuf errmsg = STRBUF_INIT;
 265                default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 266                if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 267                        warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 268                                errmsg.buf);
 269                strbuf_release(&errmsg);
 270                diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 271                return 0;
 272        }
 273
 274        if (starts_with(var, "submodule."))
 275                return parse_submodule_config_option(var, value);
 276
 277        return git_default_config(var, value, cb);
 278}
 279
 280static char *quote_two(const char *one, const char *two)
 281{
 282        int need_one = quote_c_style(one, NULL, NULL, 1);
 283        int need_two = quote_c_style(two, NULL, NULL, 1);
 284        struct strbuf res = STRBUF_INIT;
 285
 286        if (need_one + need_two) {
 287                strbuf_addch(&res, '"');
 288                quote_c_style(one, &res, NULL, 1);
 289                quote_c_style(two, &res, NULL, 1);
 290                strbuf_addch(&res, '"');
 291        } else {
 292                strbuf_addstr(&res, one);
 293                strbuf_addstr(&res, two);
 294        }
 295        return strbuf_detach(&res, NULL);
 296}
 297
 298static const char *external_diff(void)
 299{
 300        static const char *external_diff_cmd = NULL;
 301        static int done_preparing = 0;
 302
 303        if (done_preparing)
 304                return external_diff_cmd;
 305        external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 306        if (!external_diff_cmd)
 307                external_diff_cmd = external_diff_cmd_cfg;
 308        done_preparing = 1;
 309        return external_diff_cmd;
 310}
 311
 312/*
 313 * Keep track of files used for diffing. Sometimes such an entry
 314 * refers to a temporary file, sometimes to an existing file, and
 315 * sometimes to "/dev/null".
 316 */
 317static struct diff_tempfile {
 318        /*
 319         * filename external diff should read from, or NULL if this
 320         * entry is currently not in use:
 321         */
 322        const char *name;
 323
 324        char hex[GIT_SHA1_HEXSZ + 1];
 325        char mode[10];
 326
 327        /*
 328         * If this diff_tempfile instance refers to a temporary file,
 329         * this tempfile object is used to manage its lifetime.
 330         */
 331        struct tempfile tempfile;
 332} diff_temp[2];
 333
 334typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 335
 336struct emit_callback {
 337        int color_diff;
 338        unsigned ws_rule;
 339        int blank_at_eof_in_preimage;
 340        int blank_at_eof_in_postimage;
 341        int lno_in_preimage;
 342        int lno_in_postimage;
 343        sane_truncate_fn truncate;
 344        const char **label_path;
 345        struct diff_words_data *diff_words;
 346        struct diff_options *opt;
 347        int *found_changesp;
 348        struct strbuf *header;
 349};
 350
 351static int count_lines(const char *data, int size)
 352{
 353        int count, ch, completely_empty = 1, nl_just_seen = 0;
 354        count = 0;
 355        while (0 < size--) {
 356                ch = *data++;
 357                if (ch == '\n') {
 358                        count++;
 359                        nl_just_seen = 1;
 360                        completely_empty = 0;
 361                }
 362                else {
 363                        nl_just_seen = 0;
 364                        completely_empty = 0;
 365                }
 366        }
 367        if (completely_empty)
 368                return 0;
 369        if (!nl_just_seen)
 370                count++; /* no trailing newline */
 371        return count;
 372}
 373
 374static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 375{
 376        if (!DIFF_FILE_VALID(one)) {
 377                mf->ptr = (char *)""; /* does not matter */
 378                mf->size = 0;
 379                return 0;
 380        }
 381        else if (diff_populate_filespec(one, 0))
 382                return -1;
 383
 384        mf->ptr = one->data;
 385        mf->size = one->size;
 386        return 0;
 387}
 388
 389/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 390static unsigned long diff_filespec_size(struct diff_filespec *one)
 391{
 392        if (!DIFF_FILE_VALID(one))
 393                return 0;
 394        diff_populate_filespec(one, CHECK_SIZE_ONLY);
 395        return one->size;
 396}
 397
 398static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 399{
 400        char *ptr = mf->ptr;
 401        long size = mf->size;
 402        int cnt = 0;
 403
 404        if (!size)
 405                return cnt;
 406        ptr += size - 1; /* pointing at the very end */
 407        if (*ptr != '\n')
 408                ; /* incomplete line */
 409        else
 410                ptr--; /* skip the last LF */
 411        while (mf->ptr < ptr) {
 412                char *prev_eol;
 413                for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 414                        if (*prev_eol == '\n')
 415                                break;
 416                if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 417                        break;
 418                cnt++;
 419                ptr = prev_eol - 1;
 420        }
 421        return cnt;
 422}
 423
 424static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 425                               struct emit_callback *ecbdata)
 426{
 427        int l1, l2, at;
 428        unsigned ws_rule = ecbdata->ws_rule;
 429        l1 = count_trailing_blank(mf1, ws_rule);
 430        l2 = count_trailing_blank(mf2, ws_rule);
 431        if (l2 <= l1) {
 432                ecbdata->blank_at_eof_in_preimage = 0;
 433                ecbdata->blank_at_eof_in_postimage = 0;
 434                return;
 435        }
 436        at = count_lines(mf1->ptr, mf1->size);
 437        ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 438
 439        at = count_lines(mf2->ptr, mf2->size);
 440        ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 441}
 442
 443static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 444                        int first, const char *line, int len)
 445{
 446        int has_trailing_newline, has_trailing_carriage_return;
 447        int nofirst;
 448        FILE *file = o->file;
 449
 450        fputs(diff_line_prefix(o), file);
 451
 452        if (len == 0) {
 453                has_trailing_newline = (first == '\n');
 454                has_trailing_carriage_return = (!has_trailing_newline &&
 455                                                (first == '\r'));
 456                nofirst = has_trailing_newline || has_trailing_carriage_return;
 457        } else {
 458                has_trailing_newline = (len > 0 && line[len-1] == '\n');
 459                if (has_trailing_newline)
 460                        len--;
 461                has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 462                if (has_trailing_carriage_return)
 463                        len--;
 464                nofirst = 0;
 465        }
 466
 467        if (len || !nofirst) {
 468                fputs(set, file);
 469                if (!nofirst)
 470                        fputc(first, file);
 471                fwrite(line, len, 1, file);
 472                fputs(reset, file);
 473        }
 474        if (has_trailing_carriage_return)
 475                fputc('\r', file);
 476        if (has_trailing_newline)
 477                fputc('\n', file);
 478}
 479
 480static void emit_line(struct diff_options *o, const char *set, const char *reset,
 481                      const char *line, int len)
 482{
 483        emit_line_0(o, set, reset, line[0], line+1, len-1);
 484}
 485
 486static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 487{
 488        if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 489              ecbdata->blank_at_eof_in_preimage &&
 490              ecbdata->blank_at_eof_in_postimage &&
 491              ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 492              ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 493                return 0;
 494        return ws_blank_line(line, len, ecbdata->ws_rule);
 495}
 496
 497static void emit_line_checked(const char *reset,
 498                              struct emit_callback *ecbdata,
 499                              const char *line, int len,
 500                              enum color_diff color,
 501                              unsigned ws_error_highlight,
 502                              char sign)
 503{
 504        const char *set = diff_get_color(ecbdata->color_diff, color);
 505        const char *ws = NULL;
 506
 507        if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
 508                ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 509                if (!*ws)
 510                        ws = NULL;
 511        }
 512
 513        if (!ws)
 514                emit_line_0(ecbdata->opt, set, reset, sign, line, len);
 515        else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
 516                /* Blank line at EOF - paint '+' as well */
 517                emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
 518        else {
 519                /* Emit just the prefix, then the rest. */
 520                emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
 521                ws_check_emit(line, len, ecbdata->ws_rule,
 522                              ecbdata->opt->file, set, reset, ws);
 523        }
 524}
 525
 526static void emit_add_line(const char *reset,
 527                          struct emit_callback *ecbdata,
 528                          const char *line, int len)
 529{
 530        emit_line_checked(reset, ecbdata, line, len,
 531                          DIFF_FILE_NEW, WSEH_NEW, '+');
 532}
 533
 534static void emit_del_line(const char *reset,
 535                          struct emit_callback *ecbdata,
 536                          const char *line, int len)
 537{
 538        emit_line_checked(reset, ecbdata, line, len,
 539                          DIFF_FILE_OLD, WSEH_OLD, '-');
 540}
 541
 542static void emit_context_line(const char *reset,
 543                              struct emit_callback *ecbdata,
 544                              const char *line, int len)
 545{
 546        emit_line_checked(reset, ecbdata, line, len,
 547                          DIFF_CONTEXT, WSEH_CONTEXT, ' ');
 548}
 549
 550static void emit_hunk_header(struct emit_callback *ecbdata,
 551                             const char *line, int len)
 552{
 553        const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
 554        const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 555        const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 556        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 557        static const char atat[2] = { '@', '@' };
 558        const char *cp, *ep;
 559        struct strbuf msgbuf = STRBUF_INIT;
 560        int org_len = len;
 561        int i = 1;
 562
 563        /*
 564         * As a hunk header must begin with "@@ -<old>, +<new> @@",
 565         * it always is at least 10 bytes long.
 566         */
 567        if (len < 10 ||
 568            memcmp(line, atat, 2) ||
 569            !(ep = memmem(line + 2, len - 2, atat, 2))) {
 570                emit_line(ecbdata->opt, context, reset, line, len);
 571                return;
 572        }
 573        ep += 2; /* skip over @@ */
 574
 575        /* The hunk header in fraginfo color */
 576        strbuf_addstr(&msgbuf, frag);
 577        strbuf_add(&msgbuf, line, ep - line);
 578        strbuf_addstr(&msgbuf, reset);
 579
 580        /*
 581         * trailing "\r\n"
 582         */
 583        for ( ; i < 3; i++)
 584                if (line[len - i] == '\r' || line[len - i] == '\n')
 585                        len--;
 586
 587        /* blank before the func header */
 588        for (cp = ep; ep - line < len; ep++)
 589                if (*ep != ' ' && *ep != '\t')
 590                        break;
 591        if (ep != cp) {
 592                strbuf_addstr(&msgbuf, context);
 593                strbuf_add(&msgbuf, cp, ep - cp);
 594                strbuf_addstr(&msgbuf, reset);
 595        }
 596
 597        if (ep < line + len) {
 598                strbuf_addstr(&msgbuf, func);
 599                strbuf_add(&msgbuf, ep, line + len - ep);
 600                strbuf_addstr(&msgbuf, reset);
 601        }
 602
 603        strbuf_add(&msgbuf, line + len, org_len - len);
 604        emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 605        strbuf_release(&msgbuf);
 606}
 607
 608static struct diff_tempfile *claim_diff_tempfile(void) {
 609        int i;
 610        for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 611                if (!diff_temp[i].name)
 612                        return diff_temp + i;
 613        die("BUG: diff is failing to clean up its tempfiles");
 614}
 615
 616static void remove_tempfile(void)
 617{
 618        int i;
 619        for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 620                if (is_tempfile_active(&diff_temp[i].tempfile))
 621                        delete_tempfile(&diff_temp[i].tempfile);
 622                diff_temp[i].name = NULL;
 623        }
 624}
 625
 626static void print_line_count(FILE *file, int count)
 627{
 628        switch (count) {
 629        case 0:
 630                fprintf(file, "0,0");
 631                break;
 632        case 1:
 633                fprintf(file, "1");
 634                break;
 635        default:
 636                fprintf(file, "1,%d", count);
 637                break;
 638        }
 639}
 640
 641static void emit_rewrite_lines(struct emit_callback *ecb,
 642                               int prefix, const char *data, int size)
 643{
 644        const char *endp = NULL;
 645        static const char *nneof = " No newline at end of file\n";
 646        const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 647
 648        while (0 < size) {
 649                int len;
 650
 651                endp = memchr(data, '\n', size);
 652                len = endp ? (endp - data + 1) : size;
 653                if (prefix != '+') {
 654                        ecb->lno_in_preimage++;
 655                        emit_del_line(reset, ecb, data, len);
 656                } else {
 657                        ecb->lno_in_postimage++;
 658                        emit_add_line(reset, ecb, data, len);
 659                }
 660                size -= len;
 661                data += len;
 662        }
 663        if (!endp) {
 664                const char *context = diff_get_color(ecb->color_diff,
 665                                                     DIFF_CONTEXT);
 666                putc('\n', ecb->opt->file);
 667                emit_line_0(ecb->opt, context, reset, '\\',
 668                            nneof, strlen(nneof));
 669        }
 670}
 671
 672static void emit_rewrite_diff(const char *name_a,
 673                              const char *name_b,
 674                              struct diff_filespec *one,
 675                              struct diff_filespec *two,
 676                              struct userdiff_driver *textconv_one,
 677                              struct userdiff_driver *textconv_two,
 678                              struct diff_options *o)
 679{
 680        int lc_a, lc_b;
 681        const char *name_a_tab, *name_b_tab;
 682        const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 683        const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 684        const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 685        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 686        const char *a_prefix, *b_prefix;
 687        char *data_one, *data_two;
 688        size_t size_one, size_two;
 689        struct emit_callback ecbdata;
 690        const char *line_prefix = diff_line_prefix(o);
 691
 692        if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 693                a_prefix = o->b_prefix;
 694                b_prefix = o->a_prefix;
 695        } else {
 696                a_prefix = o->a_prefix;
 697                b_prefix = o->b_prefix;
 698        }
 699
 700        name_a += (*name_a == '/');
 701        name_b += (*name_b == '/');
 702        name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 703        name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 704
 705        strbuf_reset(&a_name);
 706        strbuf_reset(&b_name);
 707        quote_two_c_style(&a_name, a_prefix, name_a, 0);
 708        quote_two_c_style(&b_name, b_prefix, name_b, 0);
 709
 710        size_one = fill_textconv(textconv_one, one, &data_one);
 711        size_two = fill_textconv(textconv_two, two, &data_two);
 712
 713        memset(&ecbdata, 0, sizeof(ecbdata));
 714        ecbdata.color_diff = want_color(o->use_color);
 715        ecbdata.found_changesp = &o->found_changes;
 716        ecbdata.ws_rule = whitespace_rule(name_b);
 717        ecbdata.opt = o;
 718        if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 719                mmfile_t mf1, mf2;
 720                mf1.ptr = (char *)data_one;
 721                mf2.ptr = (char *)data_two;
 722                mf1.size = size_one;
 723                mf2.size = size_two;
 724                check_blank_at_eof(&mf1, &mf2, &ecbdata);
 725        }
 726        ecbdata.lno_in_preimage = 1;
 727        ecbdata.lno_in_postimage = 1;
 728
 729        lc_a = count_lines(data_one, size_one);
 730        lc_b = count_lines(data_two, size_two);
 731        fprintf(o->file,
 732                "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 733                line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 734                line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 735                line_prefix, fraginfo);
 736        if (!o->irreversible_delete)
 737                print_line_count(o->file, lc_a);
 738        else
 739                fprintf(o->file, "?,?");
 740        fprintf(o->file, " +");
 741        print_line_count(o->file, lc_b);
 742        fprintf(o->file, " @@%s\n", reset);
 743        if (lc_a && !o->irreversible_delete)
 744                emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 745        if (lc_b)
 746                emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 747        if (textconv_one)
 748                free((char *)data_one);
 749        if (textconv_two)
 750                free((char *)data_two);
 751}
 752
 753struct diff_words_buffer {
 754        mmfile_t text;
 755        long alloc;
 756        struct diff_words_orig {
 757                const char *begin, *end;
 758        } *orig;
 759        int orig_nr, orig_alloc;
 760};
 761
 762static void diff_words_append(char *line, unsigned long len,
 763                struct diff_words_buffer *buffer)
 764{
 765        ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 766        line++;
 767        len--;
 768        memcpy(buffer->text.ptr + buffer->text.size, line, len);
 769        buffer->text.size += len;
 770        buffer->text.ptr[buffer->text.size] = '\0';
 771}
 772
 773struct diff_words_style_elem {
 774        const char *prefix;
 775        const char *suffix;
 776        const char *color; /* NULL; filled in by the setup code if
 777                            * color is enabled */
 778};
 779
 780struct diff_words_style {
 781        enum diff_words_type type;
 782        struct diff_words_style_elem new, old, ctx;
 783        const char *newline;
 784};
 785
 786static struct diff_words_style diff_words_styles[] = {
 787        { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 788        { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 789        { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 790};
 791
 792struct diff_words_data {
 793        struct diff_words_buffer minus, plus;
 794        const char *current_plus;
 795        int last_minus;
 796        struct diff_options *opt;
 797        regex_t *word_regex;
 798        enum diff_words_type type;
 799        struct diff_words_style *style;
 800};
 801
 802static int fn_out_diff_words_write_helper(FILE *fp,
 803                                          struct diff_words_style_elem *st_el,
 804                                          const char *newline,
 805                                          size_t count, const char *buf,
 806                                          const char *line_prefix)
 807{
 808        int print = 0;
 809
 810        while (count) {
 811                char *p = memchr(buf, '\n', count);
 812                if (print)
 813                        fputs(line_prefix, fp);
 814                if (p != buf) {
 815                        if (st_el->color && fputs(st_el->color, fp) < 0)
 816                                return -1;
 817                        if (fputs(st_el->prefix, fp) < 0 ||
 818                            fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 819                            fputs(st_el->suffix, fp) < 0)
 820                                return -1;
 821                        if (st_el->color && *st_el->color
 822                            && fputs(GIT_COLOR_RESET, fp) < 0)
 823                                return -1;
 824                }
 825                if (!p)
 826                        return 0;
 827                if (fputs(newline, fp) < 0)
 828                        return -1;
 829                count -= p + 1 - buf;
 830                buf = p + 1;
 831                print = 1;
 832        }
 833        return 0;
 834}
 835
 836/*
 837 * '--color-words' algorithm can be described as:
 838 *
 839 *   1. collect a the minus/plus lines of a diff hunk, divided into
 840 *      minus-lines and plus-lines;
 841 *
 842 *   2. break both minus-lines and plus-lines into words and
 843 *      place them into two mmfile_t with one word for each line;
 844 *
 845 *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 846 *
 847 * And for the common parts of the both file, we output the plus side text.
 848 * diff_words->current_plus is used to trace the current position of the plus file
 849 * which printed. diff_words->last_minus is used to trace the last minus word
 850 * printed.
 851 *
 852 * For '--graph' to work with '--color-words', we need to output the graph prefix
 853 * on each line of color words output. Generally, there are two conditions on
 854 * which we should output the prefix.
 855 *
 856 *   1. diff_words->last_minus == 0 &&
 857 *      diff_words->current_plus == diff_words->plus.text.ptr
 858 *
 859 *      that is: the plus text must start as a new line, and if there is no minus
 860 *      word printed, a graph prefix must be printed.
 861 *
 862 *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 863 *      *(diff_words->current_plus - 1) == '\n'
 864 *
 865 *      that is: a graph prefix must be printed following a '\n'
 866 */
 867static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 868{
 869        if ((diff_words->last_minus == 0 &&
 870                diff_words->current_plus == diff_words->plus.text.ptr) ||
 871                (diff_words->current_plus > diff_words->plus.text.ptr &&
 872                *(diff_words->current_plus - 1) == '\n')) {
 873                return 1;
 874        } else {
 875                return 0;
 876        }
 877}
 878
 879static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 880{
 881        struct diff_words_data *diff_words = priv;
 882        struct diff_words_style *style = diff_words->style;
 883        int minus_first, minus_len, plus_first, plus_len;
 884        const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 885        struct diff_options *opt = diff_words->opt;
 886        const char *line_prefix;
 887
 888        if (line[0] != '@' || parse_hunk_header(line, len,
 889                        &minus_first, &minus_len, &plus_first, &plus_len))
 890                return;
 891
 892        assert(opt);
 893        line_prefix = diff_line_prefix(opt);
 894
 895        /* POSIX requires that first be decremented by one if len == 0... */
 896        if (minus_len) {
 897                minus_begin = diff_words->minus.orig[minus_first].begin;
 898                minus_end =
 899                        diff_words->minus.orig[minus_first + minus_len - 1].end;
 900        } else
 901                minus_begin = minus_end =
 902                        diff_words->minus.orig[minus_first].end;
 903
 904        if (plus_len) {
 905                plus_begin = diff_words->plus.orig[plus_first].begin;
 906                plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 907        } else
 908                plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 909
 910        if (color_words_output_graph_prefix(diff_words)) {
 911                fputs(line_prefix, diff_words->opt->file);
 912        }
 913        if (diff_words->current_plus != plus_begin) {
 914                fn_out_diff_words_write_helper(diff_words->opt->file,
 915                                &style->ctx, style->newline,
 916                                plus_begin - diff_words->current_plus,
 917                                diff_words->current_plus, line_prefix);
 918                if (*(plus_begin - 1) == '\n')
 919                        fputs(line_prefix, diff_words->opt->file);
 920        }
 921        if (minus_begin != minus_end) {
 922                fn_out_diff_words_write_helper(diff_words->opt->file,
 923                                &style->old, style->newline,
 924                                minus_end - minus_begin, minus_begin,
 925                                line_prefix);
 926        }
 927        if (plus_begin != plus_end) {
 928                fn_out_diff_words_write_helper(diff_words->opt->file,
 929                                &style->new, style->newline,
 930                                plus_end - plus_begin, plus_begin,
 931                                line_prefix);
 932        }
 933
 934        diff_words->current_plus = plus_end;
 935        diff_words->last_minus = minus_first;
 936}
 937
 938/* This function starts looking at *begin, and returns 0 iff a word was found. */
 939static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 940                int *begin, int *end)
 941{
 942        if (word_regex && *begin < buffer->size) {
 943                regmatch_t match[1];
 944                if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 945                        char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 946                                        '\n', match[0].rm_eo - match[0].rm_so);
 947                        *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 948                        *begin += match[0].rm_so;
 949                        return *begin >= *end;
 950                }
 951                return -1;
 952        }
 953
 954        /* find the next word */
 955        while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 956                (*begin)++;
 957        if (*begin >= buffer->size)
 958                return -1;
 959
 960        /* find the end of the word */
 961        *end = *begin + 1;
 962        while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 963                (*end)++;
 964
 965        return 0;
 966}
 967
 968/*
 969 * This function splits the words in buffer->text, stores the list with
 970 * newline separator into out, and saves the offsets of the original words
 971 * in buffer->orig.
 972 */
 973static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 974                regex_t *word_regex)
 975{
 976        int i, j;
 977        long alloc = 0;
 978
 979        out->size = 0;
 980        out->ptr = NULL;
 981
 982        /* fake an empty "0th" word */
 983        ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 984        buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 985        buffer->orig_nr = 1;
 986
 987        for (i = 0; i < buffer->text.size; i++) {
 988                if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 989                        return;
 990
 991                /* store original boundaries */
 992                ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 993                                buffer->orig_alloc);
 994                buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 995                buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 996                buffer->orig_nr++;
 997
 998                /* store one word */
 999                ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1000                memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1001                out->ptr[out->size + j - i] = '\n';
1002                out->size += j - i + 1;
1003
1004                i = j - 1;
1005        }
1006}
1007
1008/* this executes the word diff on the accumulated buffers */
1009static void diff_words_show(struct diff_words_data *diff_words)
1010{
1011        xpparam_t xpp;
1012        xdemitconf_t xecfg;
1013        mmfile_t minus, plus;
1014        struct diff_words_style *style = diff_words->style;
1015
1016        struct diff_options *opt = diff_words->opt;
1017        const char *line_prefix;
1018
1019        assert(opt);
1020        line_prefix = diff_line_prefix(opt);
1021
1022        /* special case: only removal */
1023        if (!diff_words->plus.text.size) {
1024                fputs(line_prefix, diff_words->opt->file);
1025                fn_out_diff_words_write_helper(diff_words->opt->file,
1026                        &style->old, style->newline,
1027                        diff_words->minus.text.size,
1028                        diff_words->minus.text.ptr, line_prefix);
1029                diff_words->minus.text.size = 0;
1030                return;
1031        }
1032
1033        diff_words->current_plus = diff_words->plus.text.ptr;
1034        diff_words->last_minus = 0;
1035
1036        memset(&xpp, 0, sizeof(xpp));
1037        memset(&xecfg, 0, sizeof(xecfg));
1038        diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1039        diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1040        xpp.flags = 0;
1041        /* as only the hunk header will be parsed, we need a 0-context */
1042        xecfg.ctxlen = 0;
1043        if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1044                          &xpp, &xecfg))
1045                die("unable to generate word diff");
1046        free(minus.ptr);
1047        free(plus.ptr);
1048        if (diff_words->current_plus != diff_words->plus.text.ptr +
1049                        diff_words->plus.text.size) {
1050                if (color_words_output_graph_prefix(diff_words))
1051                        fputs(line_prefix, diff_words->opt->file);
1052                fn_out_diff_words_write_helper(diff_words->opt->file,
1053                        &style->ctx, style->newline,
1054                        diff_words->plus.text.ptr + diff_words->plus.text.size
1055                        - diff_words->current_plus, diff_words->current_plus,
1056                        line_prefix);
1057        }
1058        diff_words->minus.text.size = diff_words->plus.text.size = 0;
1059}
1060
1061/* In "color-words" mode, show word-diff of words accumulated in the buffer */
1062static void diff_words_flush(struct emit_callback *ecbdata)
1063{
1064        if (ecbdata->diff_words->minus.text.size ||
1065            ecbdata->diff_words->plus.text.size)
1066                diff_words_show(ecbdata->diff_words);
1067}
1068
1069static void diff_filespec_load_driver(struct diff_filespec *one)
1070{
1071        /* Use already-loaded driver */
1072        if (one->driver)
1073                return;
1074
1075        if (S_ISREG(one->mode))
1076                one->driver = userdiff_find_by_path(one->path);
1077
1078        /* Fallback to default settings */
1079        if (!one->driver)
1080                one->driver = userdiff_find_by_name("default");
1081}
1082
1083static const char *userdiff_word_regex(struct diff_filespec *one)
1084{
1085        diff_filespec_load_driver(one);
1086        return one->driver->word_regex;
1087}
1088
1089static void init_diff_words_data(struct emit_callback *ecbdata,
1090                                 struct diff_options *orig_opts,
1091                                 struct diff_filespec *one,
1092                                 struct diff_filespec *two)
1093{
1094        int i;
1095        struct diff_options *o = xmalloc(sizeof(struct diff_options));
1096        memcpy(o, orig_opts, sizeof(struct diff_options));
1097
1098        ecbdata->diff_words =
1099                xcalloc(1, sizeof(struct diff_words_data));
1100        ecbdata->diff_words->type = o->word_diff;
1101        ecbdata->diff_words->opt = o;
1102        if (!o->word_regex)
1103                o->word_regex = userdiff_word_regex(one);
1104        if (!o->word_regex)
1105                o->word_regex = userdiff_word_regex(two);
1106        if (!o->word_regex)
1107                o->word_regex = diff_word_regex_cfg;
1108        if (o->word_regex) {
1109                ecbdata->diff_words->word_regex = (regex_t *)
1110                        xmalloc(sizeof(regex_t));
1111                if (regcomp(ecbdata->diff_words->word_regex,
1112                            o->word_regex,
1113                            REG_EXTENDED | REG_NEWLINE))
1114                        die ("Invalid regular expression: %s",
1115                             o->word_regex);
1116        }
1117        for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1118                if (o->word_diff == diff_words_styles[i].type) {
1119                        ecbdata->diff_words->style =
1120                                &diff_words_styles[i];
1121                        break;
1122                }
1123        }
1124        if (want_color(o->use_color)) {
1125                struct diff_words_style *st = ecbdata->diff_words->style;
1126                st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1127                st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1128                st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1129        }
1130}
1131
1132static void free_diff_words_data(struct emit_callback *ecbdata)
1133{
1134        if (ecbdata->diff_words) {
1135                diff_words_flush(ecbdata);
1136                free (ecbdata->diff_words->opt);
1137                free (ecbdata->diff_words->minus.text.ptr);
1138                free (ecbdata->diff_words->minus.orig);
1139                free (ecbdata->diff_words->plus.text.ptr);
1140                free (ecbdata->diff_words->plus.orig);
1141                if (ecbdata->diff_words->word_regex) {
1142                        regfree(ecbdata->diff_words->word_regex);
1143                        free(ecbdata->diff_words->word_regex);
1144                }
1145                free(ecbdata->diff_words);
1146                ecbdata->diff_words = NULL;
1147        }
1148}
1149
1150const char *diff_get_color(int diff_use_color, enum color_diff ix)
1151{
1152        if (want_color(diff_use_color))
1153                return diff_colors[ix];
1154        return "";
1155}
1156
1157const char *diff_line_prefix(struct diff_options *opt)
1158{
1159        struct strbuf *msgbuf;
1160        if (!opt->output_prefix)
1161                return "";
1162
1163        msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1164        return msgbuf->buf;
1165}
1166
1167static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1168{
1169        const char *cp;
1170        unsigned long allot;
1171        size_t l = len;
1172
1173        if (ecb->truncate)
1174                return ecb->truncate(line, len);
1175        cp = line;
1176        allot = l;
1177        while (0 < l) {
1178                (void) utf8_width(&cp, &l);
1179                if (!cp)
1180                        break; /* truncated in the middle? */
1181        }
1182        return allot - l;
1183}
1184
1185static void find_lno(const char *line, struct emit_callback *ecbdata)
1186{
1187        const char *p;
1188        ecbdata->lno_in_preimage = 0;
1189        ecbdata->lno_in_postimage = 0;
1190        p = strchr(line, '-');
1191        if (!p)
1192                return; /* cannot happen */
1193        ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1194        p = strchr(p, '+');
1195        if (!p)
1196                return; /* cannot happen */
1197        ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1198}
1199
1200static void fn_out_consume(void *priv, char *line, unsigned long len)
1201{
1202        struct emit_callback *ecbdata = priv;
1203        const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1204        const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1205        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1206        struct diff_options *o = ecbdata->opt;
1207        const char *line_prefix = diff_line_prefix(o);
1208
1209        if (ecbdata->header) {
1210                fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1211                strbuf_reset(ecbdata->header);
1212                ecbdata->header = NULL;
1213        }
1214        *(ecbdata->found_changesp) = 1;
1215
1216        if (ecbdata->label_path[0]) {
1217                const char *name_a_tab, *name_b_tab;
1218
1219                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1220                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1221
1222                fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1223                        line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1224                fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1225                        line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1226                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1227        }
1228
1229        if (diff_suppress_blank_empty
1230            && len == 2 && line[0] == ' ' && line[1] == '\n') {
1231                line[0] = '\n';
1232                len = 1;
1233        }
1234
1235        if (line[0] == '@') {
1236                if (ecbdata->diff_words)
1237                        diff_words_flush(ecbdata);
1238                len = sane_truncate_line(ecbdata, line, len);
1239                find_lno(line, ecbdata);
1240                emit_hunk_header(ecbdata, line, len);
1241                if (line[len-1] != '\n')
1242                        putc('\n', ecbdata->opt->file);
1243                return;
1244        }
1245
1246        if (len < 1) {
1247                emit_line(ecbdata->opt, reset, reset, line, len);
1248                if (ecbdata->diff_words
1249                    && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1250                        fputs("~\n", ecbdata->opt->file);
1251                return;
1252        }
1253
1254        if (ecbdata->diff_words) {
1255                if (line[0] == '-') {
1256                        diff_words_append(line, len,
1257                                          &ecbdata->diff_words->minus);
1258                        return;
1259                } else if (line[0] == '+') {
1260                        diff_words_append(line, len,
1261                                          &ecbdata->diff_words->plus);
1262                        return;
1263                } else if (starts_with(line, "\\ ")) {
1264                        /*
1265                         * Eat the "no newline at eof" marker as if we
1266                         * saw a "+" or "-" line with nothing on it,
1267                         * and return without diff_words_flush() to
1268                         * defer processing. If this is the end of
1269                         * preimage, more "+" lines may come after it.
1270                         */
1271                        return;
1272                }
1273                diff_words_flush(ecbdata);
1274                if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1275                        emit_line(ecbdata->opt, context, reset, line, len);
1276                        fputs("~\n", ecbdata->opt->file);
1277                } else {
1278                        /*
1279                         * Skip the prefix character, if any.  With
1280                         * diff_suppress_blank_empty, there may be
1281                         * none.
1282                         */
1283                        if (line[0] != '\n') {
1284                              line++;
1285                              len--;
1286                        }
1287                        emit_line(ecbdata->opt, context, reset, line, len);
1288                }
1289                return;
1290        }
1291
1292        switch (line[0]) {
1293        case '+':
1294                ecbdata->lno_in_postimage++;
1295                emit_add_line(reset, ecbdata, line + 1, len - 1);
1296                break;
1297        case '-':
1298                ecbdata->lno_in_preimage++;
1299                emit_del_line(reset, ecbdata, line + 1, len - 1);
1300                break;
1301        case ' ':
1302                ecbdata->lno_in_postimage++;
1303                ecbdata->lno_in_preimage++;
1304                emit_context_line(reset, ecbdata, line + 1, len - 1);
1305                break;
1306        default:
1307                /* incomplete line at the end */
1308                ecbdata->lno_in_preimage++;
1309                emit_line(ecbdata->opt,
1310                          diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1311                          reset, line, len);
1312                break;
1313        }
1314}
1315
1316static char *pprint_rename(const char *a, const char *b)
1317{
1318        const char *old = a;
1319        const char *new = b;
1320        struct strbuf name = STRBUF_INIT;
1321        int pfx_length, sfx_length;
1322        int pfx_adjust_for_slash;
1323        int len_a = strlen(a);
1324        int len_b = strlen(b);
1325        int a_midlen, b_midlen;
1326        int qlen_a = quote_c_style(a, NULL, NULL, 0);
1327        int qlen_b = quote_c_style(b, NULL, NULL, 0);
1328
1329        if (qlen_a || qlen_b) {
1330                quote_c_style(a, &name, NULL, 0);
1331                strbuf_addstr(&name, " => ");
1332                quote_c_style(b, &name, NULL, 0);
1333                return strbuf_detach(&name, NULL);
1334        }
1335
1336        /* Find common prefix */
1337        pfx_length = 0;
1338        while (*old && *new && *old == *new) {
1339                if (*old == '/')
1340                        pfx_length = old - a + 1;
1341                old++;
1342                new++;
1343        }
1344
1345        /* Find common suffix */
1346        old = a + len_a;
1347        new = b + len_b;
1348        sfx_length = 0;
1349        /*
1350         * If there is a common prefix, it must end in a slash.  In
1351         * that case we let this loop run 1 into the prefix to see the
1352         * same slash.
1353         *
1354         * If there is no common prefix, we cannot do this as it would
1355         * underrun the input strings.
1356         */
1357        pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1358        while (a + pfx_length - pfx_adjust_for_slash <= old &&
1359               b + pfx_length - pfx_adjust_for_slash <= new &&
1360               *old == *new) {
1361                if (*old == '/')
1362                        sfx_length = len_a - (old - a);
1363                old--;
1364                new--;
1365        }
1366
1367        /*
1368         * pfx{mid-a => mid-b}sfx
1369         * {pfx-a => pfx-b}sfx
1370         * pfx{sfx-a => sfx-b}
1371         * name-a => name-b
1372         */
1373        a_midlen = len_a - pfx_length - sfx_length;
1374        b_midlen = len_b - pfx_length - sfx_length;
1375        if (a_midlen < 0)
1376                a_midlen = 0;
1377        if (b_midlen < 0)
1378                b_midlen = 0;
1379
1380        strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1381        if (pfx_length + sfx_length) {
1382                strbuf_add(&name, a, pfx_length);
1383                strbuf_addch(&name, '{');
1384        }
1385        strbuf_add(&name, a + pfx_length, a_midlen);
1386        strbuf_addstr(&name, " => ");
1387        strbuf_add(&name, b + pfx_length, b_midlen);
1388        if (pfx_length + sfx_length) {
1389                strbuf_addch(&name, '}');
1390                strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1391        }
1392        return strbuf_detach(&name, NULL);
1393}
1394
1395struct diffstat_t {
1396        int nr;
1397        int alloc;
1398        struct diffstat_file {
1399                char *from_name;
1400                char *name;
1401                char *print_name;
1402                unsigned is_unmerged:1;
1403                unsigned is_binary:1;
1404                unsigned is_renamed:1;
1405                unsigned is_interesting:1;
1406                uintmax_t added, deleted;
1407        } **files;
1408};
1409
1410static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1411                                          const char *name_a,
1412                                          const char *name_b)
1413{
1414        struct diffstat_file *x;
1415        x = xcalloc(1, sizeof(*x));
1416        ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1417        diffstat->files[diffstat->nr++] = x;
1418        if (name_b) {
1419                x->from_name = xstrdup(name_a);
1420                x->name = xstrdup(name_b);
1421                x->is_renamed = 1;
1422        }
1423        else {
1424                x->from_name = NULL;
1425                x->name = xstrdup(name_a);
1426        }
1427        return x;
1428}
1429
1430static void diffstat_consume(void *priv, char *line, unsigned long len)
1431{
1432        struct diffstat_t *diffstat = priv;
1433        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1434
1435        if (line[0] == '+')
1436                x->added++;
1437        else if (line[0] == '-')
1438                x->deleted++;
1439}
1440
1441const char mime_boundary_leader[] = "------------";
1442
1443static int scale_linear(int it, int width, int max_change)
1444{
1445        if (!it)
1446                return 0;
1447        /*
1448         * make sure that at least one '-' or '+' is printed if
1449         * there is any change to this path. The easiest way is to
1450         * scale linearly as if the alloted width is one column shorter
1451         * than it is, and then add 1 to the result.
1452         */
1453        return 1 + (it * (width - 1) / max_change);
1454}
1455
1456static void show_name(FILE *file,
1457                      const char *prefix, const char *name, int len)
1458{
1459        fprintf(file, " %s%-*s |", prefix, len, name);
1460}
1461
1462static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1463{
1464        if (cnt <= 0)
1465                return;
1466        fprintf(file, "%s", set);
1467        while (cnt--)
1468                putc(ch, file);
1469        fprintf(file, "%s", reset);
1470}
1471
1472static void fill_print_name(struct diffstat_file *file)
1473{
1474        char *pname;
1475
1476        if (file->print_name)
1477                return;
1478
1479        if (!file->is_renamed) {
1480                struct strbuf buf = STRBUF_INIT;
1481                if (quote_c_style(file->name, &buf, NULL, 0)) {
1482                        pname = strbuf_detach(&buf, NULL);
1483                } else {
1484                        pname = file->name;
1485                        strbuf_release(&buf);
1486                }
1487        } else {
1488                pname = pprint_rename(file->from_name, file->name);
1489        }
1490        file->print_name = pname;
1491}
1492
1493int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1494{
1495        struct strbuf sb = STRBUF_INIT;
1496        int ret;
1497
1498        if (!files) {
1499                assert(insertions == 0 && deletions == 0);
1500                return fprintf(fp, "%s\n", " 0 files changed");
1501        }
1502
1503        strbuf_addf(&sb,
1504                    (files == 1) ? " %d file changed" : " %d files changed",
1505                    files);
1506
1507        /*
1508         * For binary diff, the caller may want to print "x files
1509         * changed" with insertions == 0 && deletions == 0.
1510         *
1511         * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1512         * is probably less confusing (i.e skip over "2 files changed
1513         * but nothing about added/removed lines? Is this a bug in Git?").
1514         */
1515        if (insertions || deletions == 0) {
1516                strbuf_addf(&sb,
1517                            (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1518                            insertions);
1519        }
1520
1521        if (deletions || insertions == 0) {
1522                strbuf_addf(&sb,
1523                            (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1524                            deletions);
1525        }
1526        strbuf_addch(&sb, '\n');
1527        ret = fputs(sb.buf, fp);
1528        strbuf_release(&sb);
1529        return ret;
1530}
1531
1532static void show_stats(struct diffstat_t *data, struct diff_options *options)
1533{
1534        int i, len, add, del, adds = 0, dels = 0;
1535        uintmax_t max_change = 0, max_len = 0;
1536        int total_files = data->nr, count;
1537        int width, name_width, graph_width, number_width = 0, bin_width = 0;
1538        const char *reset, *add_c, *del_c;
1539        const char *line_prefix = "";
1540        int extra_shown = 0;
1541
1542        if (data->nr == 0)
1543                return;
1544
1545        line_prefix = diff_line_prefix(options);
1546        count = options->stat_count ? options->stat_count : data->nr;
1547
1548        reset = diff_get_color_opt(options, DIFF_RESET);
1549        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1550        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1551
1552        /*
1553         * Find the longest filename and max number of changes
1554         */
1555        for (i = 0; (i < count) && (i < data->nr); i++) {
1556                struct diffstat_file *file = data->files[i];
1557                uintmax_t change = file->added + file->deleted;
1558
1559                if (!file->is_interesting && (change == 0)) {
1560                        count++; /* not shown == room for one more */
1561                        continue;
1562                }
1563                fill_print_name(file);
1564                len = strlen(file->print_name);
1565                if (max_len < len)
1566                        max_len = len;
1567
1568                if (file->is_unmerged) {
1569                        /* "Unmerged" is 8 characters */
1570                        bin_width = bin_width < 8 ? 8 : bin_width;
1571                        continue;
1572                }
1573                if (file->is_binary) {
1574                        /* "Bin XXX -> YYY bytes" */
1575                        int w = 14 + decimal_width(file->added)
1576                                + decimal_width(file->deleted);
1577                        bin_width = bin_width < w ? w : bin_width;
1578                        /* Display change counts aligned with "Bin" */
1579                        number_width = 3;
1580                        continue;
1581                }
1582
1583                if (max_change < change)
1584                        max_change = change;
1585        }
1586        count = i; /* where we can stop scanning in data->files[] */
1587
1588        /*
1589         * We have width = stat_width or term_columns() columns total.
1590         * We want a maximum of min(max_len, stat_name_width) for the name part.
1591         * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1592         * We also need 1 for " " and 4 + decimal_width(max_change)
1593         * for " | NNNN " and one the empty column at the end, altogether
1594         * 6 + decimal_width(max_change).
1595         *
1596         * If there's not enough space, we will use the smaller of
1597         * stat_name_width (if set) and 5/8*width for the filename,
1598         * and the rest for constant elements + graph part, but no more
1599         * than stat_graph_width for the graph part.
1600         * (5/8 gives 50 for filename and 30 for the constant parts + graph
1601         * for the standard terminal size).
1602         *
1603         * In other words: stat_width limits the maximum width, and
1604         * stat_name_width fixes the maximum width of the filename,
1605         * and is also used to divide available columns if there
1606         * aren't enough.
1607         *
1608         * Binary files are displayed with "Bin XXX -> YYY bytes"
1609         * instead of the change count and graph. This part is treated
1610         * similarly to the graph part, except that it is not
1611         * "scaled". If total width is too small to accommodate the
1612         * guaranteed minimum width of the filename part and the
1613         * separators and this message, this message will "overflow"
1614         * making the line longer than the maximum width.
1615         */
1616
1617        if (options->stat_width == -1)
1618                width = term_columns() - options->output_prefix_length;
1619        else
1620                width = options->stat_width ? options->stat_width : 80;
1621        number_width = decimal_width(max_change) > number_width ?
1622                decimal_width(max_change) : number_width;
1623
1624        if (options->stat_graph_width == -1)
1625                options->stat_graph_width = diff_stat_graph_width;
1626
1627        /*
1628         * Guarantee 3/8*16==6 for the graph part
1629         * and 5/8*16==10 for the filename part
1630         */
1631        if (width < 16 + 6 + number_width)
1632                width = 16 + 6 + number_width;
1633
1634        /*
1635         * First assign sizes that are wanted, ignoring available width.
1636         * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1637         * starting from "XXX" should fit in graph_width.
1638         */
1639        graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1640        if (options->stat_graph_width &&
1641            options->stat_graph_width < graph_width)
1642                graph_width = options->stat_graph_width;
1643
1644        name_width = (options->stat_name_width > 0 &&
1645                      options->stat_name_width < max_len) ?
1646                options->stat_name_width : max_len;
1647
1648        /*
1649         * Adjust adjustable widths not to exceed maximum width
1650         */
1651        if (name_width + number_width + 6 + graph_width > width) {
1652                if (graph_width > width * 3/8 - number_width - 6) {
1653                        graph_width = width * 3/8 - number_width - 6;
1654                        if (graph_width < 6)
1655                                graph_width = 6;
1656                }
1657
1658                if (options->stat_graph_width &&
1659                    graph_width > options->stat_graph_width)
1660                        graph_width = options->stat_graph_width;
1661                if (name_width > width - number_width - 6 - graph_width)
1662                        name_width = width - number_width - 6 - graph_width;
1663                else
1664                        graph_width = width - number_width - 6 - name_width;
1665        }
1666
1667        /*
1668         * From here name_width is the width of the name area,
1669         * and graph_width is the width of the graph area.
1670         * max_change is used to scale graph properly.
1671         */
1672        for (i = 0; i < count; i++) {
1673                const char *prefix = "";
1674                struct diffstat_file *file = data->files[i];
1675                char *name = file->print_name;
1676                uintmax_t added = file->added;
1677                uintmax_t deleted = file->deleted;
1678                int name_len;
1679
1680                if (!file->is_interesting && (added + deleted == 0))
1681                        continue;
1682
1683                /*
1684                 * "scale" the filename
1685                 */
1686                len = name_width;
1687                name_len = strlen(name);
1688                if (name_width < name_len) {
1689                        char *slash;
1690                        prefix = "...";
1691                        len -= 3;
1692                        name += name_len - len;
1693                        slash = strchr(name, '/');
1694                        if (slash)
1695                                name = slash;
1696                }
1697
1698                if (file->is_binary) {
1699                        fprintf(options->file, "%s", line_prefix);
1700                        show_name(options->file, prefix, name, len);
1701                        fprintf(options->file, " %*s", number_width, "Bin");
1702                        if (!added && !deleted) {
1703                                putc('\n', options->file);
1704                                continue;
1705                        }
1706                        fprintf(options->file, " %s%"PRIuMAX"%s",
1707                                del_c, deleted, reset);
1708                        fprintf(options->file, " -> ");
1709                        fprintf(options->file, "%s%"PRIuMAX"%s",
1710                                add_c, added, reset);
1711                        fprintf(options->file, " bytes");
1712                        fprintf(options->file, "\n");
1713                        continue;
1714                }
1715                else if (file->is_unmerged) {
1716                        fprintf(options->file, "%s", line_prefix);
1717                        show_name(options->file, prefix, name, len);
1718                        fprintf(options->file, " Unmerged\n");
1719                        continue;
1720                }
1721
1722                /*
1723                 * scale the add/delete
1724                 */
1725                add = added;
1726                del = deleted;
1727
1728                if (graph_width <= max_change) {
1729                        int total = scale_linear(add + del, graph_width, max_change);
1730                        if (total < 2 && add && del)
1731                                /* width >= 2 due to the sanity check */
1732                                total = 2;
1733                        if (add < del) {
1734                                add = scale_linear(add, graph_width, max_change);
1735                                del = total - add;
1736                        } else {
1737                                del = scale_linear(del, graph_width, max_change);
1738                                add = total - del;
1739                        }
1740                }
1741                fprintf(options->file, "%s", line_prefix);
1742                show_name(options->file, prefix, name, len);
1743                fprintf(options->file, " %*"PRIuMAX"%s",
1744                        number_width, added + deleted,
1745                        added + deleted ? " " : "");
1746                show_graph(options->file, '+', add, add_c, reset);
1747                show_graph(options->file, '-', del, del_c, reset);
1748                fprintf(options->file, "\n");
1749        }
1750
1751        for (i = 0; i < data->nr; i++) {
1752                struct diffstat_file *file = data->files[i];
1753                uintmax_t added = file->added;
1754                uintmax_t deleted = file->deleted;
1755
1756                if (file->is_unmerged ||
1757                    (!file->is_interesting && (added + deleted == 0))) {
1758                        total_files--;
1759                        continue;
1760                }
1761
1762                if (!file->is_binary) {
1763                        adds += added;
1764                        dels += deleted;
1765                }
1766                if (i < count)
1767                        continue;
1768                if (!extra_shown)
1769                        fprintf(options->file, "%s ...\n", line_prefix);
1770                extra_shown = 1;
1771        }
1772        fprintf(options->file, "%s", line_prefix);
1773        print_stat_summary(options->file, total_files, adds, dels);
1774}
1775
1776static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1777{
1778        int i, adds = 0, dels = 0, total_files = data->nr;
1779
1780        if (data->nr == 0)
1781                return;
1782
1783        for (i = 0; i < data->nr; i++) {
1784                int added = data->files[i]->added;
1785                int deleted= data->files[i]->deleted;
1786
1787                if (data->files[i]->is_unmerged ||
1788                    (!data->files[i]->is_interesting && (added + deleted == 0))) {
1789                        total_files--;
1790                } else if (!data->files[i]->is_binary) { /* don't count bytes */
1791                        adds += added;
1792                        dels += deleted;
1793                }
1794        }
1795        fprintf(options->file, "%s", diff_line_prefix(options));
1796        print_stat_summary(options->file, total_files, adds, dels);
1797}
1798
1799static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1800{
1801        int i;
1802
1803        if (data->nr == 0)
1804                return;
1805
1806        for (i = 0; i < data->nr; i++) {
1807                struct diffstat_file *file = data->files[i];
1808
1809                fprintf(options->file, "%s", diff_line_prefix(options));
1810
1811                if (file->is_binary)
1812                        fprintf(options->file, "-\t-\t");
1813                else
1814                        fprintf(options->file,
1815                                "%"PRIuMAX"\t%"PRIuMAX"\t",
1816                                file->added, file->deleted);
1817                if (options->line_termination) {
1818                        fill_print_name(file);
1819                        if (!file->is_renamed)
1820                                write_name_quoted(file->name, options->file,
1821                                                  options->line_termination);
1822                        else {
1823                                fputs(file->print_name, options->file);
1824                                putc(options->line_termination, options->file);
1825                        }
1826                } else {
1827                        if (file->is_renamed) {
1828                                putc('\0', options->file);
1829                                write_name_quoted(file->from_name, options->file, '\0');
1830                        }
1831                        write_name_quoted(file->name, options->file, '\0');
1832                }
1833        }
1834}
1835
1836struct dirstat_file {
1837        const char *name;
1838        unsigned long changed;
1839};
1840
1841struct dirstat_dir {
1842        struct dirstat_file *files;
1843        int alloc, nr, permille, cumulative;
1844};
1845
1846static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1847                unsigned long changed, const char *base, int baselen)
1848{
1849        unsigned long this_dir = 0;
1850        unsigned int sources = 0;
1851        const char *line_prefix = diff_line_prefix(opt);
1852
1853        while (dir->nr) {
1854                struct dirstat_file *f = dir->files;
1855                int namelen = strlen(f->name);
1856                unsigned long this;
1857                char *slash;
1858
1859                if (namelen < baselen)
1860                        break;
1861                if (memcmp(f->name, base, baselen))
1862                        break;
1863                slash = strchr(f->name + baselen, '/');
1864                if (slash) {
1865                        int newbaselen = slash + 1 - f->name;
1866                        this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1867                        sources++;
1868                } else {
1869                        this = f->changed;
1870                        dir->files++;
1871                        dir->nr--;
1872                        sources += 2;
1873                }
1874                this_dir += this;
1875        }
1876
1877        /*
1878         * We don't report dirstat's for
1879         *  - the top level
1880         *  - or cases where everything came from a single directory
1881         *    under this directory (sources == 1).
1882         */
1883        if (baselen && sources != 1) {
1884                if (this_dir) {
1885                        int permille = this_dir * 1000 / changed;
1886                        if (permille >= dir->permille) {
1887                                fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1888                                        permille / 10, permille % 10, baselen, base);
1889                                if (!dir->cumulative)
1890                                        return 0;
1891                        }
1892                }
1893        }
1894        return this_dir;
1895}
1896
1897static int dirstat_compare(const void *_a, const void *_b)
1898{
1899        const struct dirstat_file *a = _a;
1900        const struct dirstat_file *b = _b;
1901        return strcmp(a->name, b->name);
1902}
1903
1904static void show_dirstat(struct diff_options *options)
1905{
1906        int i;
1907        unsigned long changed;
1908        struct dirstat_dir dir;
1909        struct diff_queue_struct *q = &diff_queued_diff;
1910
1911        dir.files = NULL;
1912        dir.alloc = 0;
1913        dir.nr = 0;
1914        dir.permille = options->dirstat_permille;
1915        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1916
1917        changed = 0;
1918        for (i = 0; i < q->nr; i++) {
1919                struct diff_filepair *p = q->queue[i];
1920                const char *name;
1921                unsigned long copied, added, damage;
1922                int content_changed;
1923
1924                name = p->two->path ? p->two->path : p->one->path;
1925
1926                if (p->one->sha1_valid && p->two->sha1_valid)
1927                        content_changed = hashcmp(p->one->sha1, p->two->sha1);
1928                else
1929                        content_changed = 1;
1930
1931                if (!content_changed) {
1932                        /*
1933                         * The SHA1 has not changed, so pre-/post-content is
1934                         * identical. We can therefore skip looking at the
1935                         * file contents altogether.
1936                         */
1937                        damage = 0;
1938                        goto found_damage;
1939                }
1940
1941                if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1942                        /*
1943                         * In --dirstat-by-file mode, we don't really need to
1944                         * look at the actual file contents at all.
1945                         * The fact that the SHA1 changed is enough for us to
1946                         * add this file to the list of results
1947                         * (with each file contributing equal damage).
1948                         */
1949                        damage = 1;
1950                        goto found_damage;
1951                }
1952
1953                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1954                        diff_populate_filespec(p->one, 0);
1955                        diff_populate_filespec(p->two, 0);
1956                        diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1957                                               &copied, &added);
1958                        diff_free_filespec_data(p->one);
1959                        diff_free_filespec_data(p->two);
1960                } else if (DIFF_FILE_VALID(p->one)) {
1961                        diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1962                        copied = added = 0;
1963                        diff_free_filespec_data(p->one);
1964                } else if (DIFF_FILE_VALID(p->two)) {
1965                        diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1966                        copied = 0;
1967                        added = p->two->size;
1968                        diff_free_filespec_data(p->two);
1969                } else
1970                        continue;
1971
1972                /*
1973                 * Original minus copied is the removed material,
1974                 * added is the new material.  They are both damages
1975                 * made to the preimage.
1976                 * If the resulting damage is zero, we know that
1977                 * diffcore_count_changes() considers the two entries to
1978                 * be identical, but since content_changed is true, we
1979                 * know that there must have been _some_ kind of change,
1980                 * so we force all entries to have damage > 0.
1981                 */
1982                damage = (p->one->size - copied) + added;
1983                if (!damage)
1984                        damage = 1;
1985
1986found_damage:
1987                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1988                dir.files[dir.nr].name = name;
1989                dir.files[dir.nr].changed = damage;
1990                changed += damage;
1991                dir.nr++;
1992        }
1993
1994        /* This can happen even with many files, if everything was renames */
1995        if (!changed)
1996                return;
1997
1998        /* Show all directories with more than x% of the changes */
1999        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2000        gather_dirstat(options, &dir, changed, "", 0);
2001}
2002
2003static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2004{
2005        int i;
2006        unsigned long changed;
2007        struct dirstat_dir dir;
2008
2009        if (data->nr == 0)
2010                return;
2011
2012        dir.files = NULL;
2013        dir.alloc = 0;
2014        dir.nr = 0;
2015        dir.permille = options->dirstat_permille;
2016        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2017
2018        changed = 0;
2019        for (i = 0; i < data->nr; i++) {
2020                struct diffstat_file *file = data->files[i];
2021                unsigned long damage = file->added + file->deleted;
2022                if (file->is_binary)
2023                        /*
2024                         * binary files counts bytes, not lines. Must find some
2025                         * way to normalize binary bytes vs. textual lines.
2026                         * The following heuristic assumes that there are 64
2027                         * bytes per "line".
2028                         * This is stupid and ugly, but very cheap...
2029                         */
2030                        damage = (damage + 63) / 64;
2031                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2032                dir.files[dir.nr].name = file->name;
2033                dir.files[dir.nr].changed = damage;
2034                changed += damage;
2035                dir.nr++;
2036        }
2037
2038        /* This can happen even with many files, if everything was renames */
2039        if (!changed)
2040                return;
2041
2042        /* Show all directories with more than x% of the changes */
2043        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2044        gather_dirstat(options, &dir, changed, "", 0);
2045}
2046
2047static void free_diffstat_info(struct diffstat_t *diffstat)
2048{
2049        int i;
2050        for (i = 0; i < diffstat->nr; i++) {
2051                struct diffstat_file *f = diffstat->files[i];
2052                if (f->name != f->print_name)
2053                        free(f->print_name);
2054                free(f->name);
2055                free(f->from_name);
2056                free(f);
2057        }
2058        free(diffstat->files);
2059}
2060
2061struct checkdiff_t {
2062        const char *filename;
2063        int lineno;
2064        int conflict_marker_size;
2065        struct diff_options *o;
2066        unsigned ws_rule;
2067        unsigned status;
2068};
2069
2070static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2071{
2072        char firstchar;
2073        int cnt;
2074
2075        if (len < marker_size + 1)
2076                return 0;
2077        firstchar = line[0];
2078        switch (firstchar) {
2079        case '=': case '>': case '<': case '|':
2080                break;
2081        default:
2082                return 0;
2083        }
2084        for (cnt = 1; cnt < marker_size; cnt++)
2085                if (line[cnt] != firstchar)
2086                        return 0;
2087        /* line[1] thru line[marker_size-1] are same as firstchar */
2088        if (len < marker_size + 1 || !isspace(line[marker_size]))
2089                return 0;
2090        return 1;
2091}
2092
2093static void checkdiff_consume(void *priv, char *line, unsigned long len)
2094{
2095        struct checkdiff_t *data = priv;
2096        int marker_size = data->conflict_marker_size;
2097        const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2098        const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2099        const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2100        char *err;
2101        const char *line_prefix;
2102
2103        assert(data->o);
2104        line_prefix = diff_line_prefix(data->o);
2105
2106        if (line[0] == '+') {
2107                unsigned bad;
2108                data->lineno++;
2109                if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2110                        data->status |= 1;
2111                        fprintf(data->o->file,
2112                                "%s%s:%d: leftover conflict marker\n",
2113                                line_prefix, data->filename, data->lineno);
2114                }
2115                bad = ws_check(line + 1, len - 1, data->ws_rule);
2116                if (!bad)
2117                        return;
2118                data->status |= bad;
2119                err = whitespace_error_string(bad);
2120                fprintf(data->o->file, "%s%s:%d: %s.\n",
2121                        line_prefix, data->filename, data->lineno, err);
2122                free(err);
2123                emit_line(data->o, set, reset, line, 1);
2124                ws_check_emit(line + 1, len - 1, data->ws_rule,
2125                              data->o->file, set, reset, ws);
2126        } else if (line[0] == ' ') {
2127                data->lineno++;
2128        } else if (line[0] == '@') {
2129                char *plus = strchr(line, '+');
2130                if (plus)
2131                        data->lineno = strtol(plus, NULL, 10) - 1;
2132                else
2133                        die("invalid diff");
2134        }
2135}
2136
2137static unsigned char *deflate_it(char *data,
2138                                 unsigned long size,
2139                                 unsigned long *result_size)
2140{
2141        int bound;
2142        unsigned char *deflated;
2143        git_zstream stream;
2144
2145        git_deflate_init(&stream, zlib_compression_level);
2146        bound = git_deflate_bound(&stream, size);
2147        deflated = xmalloc(bound);
2148        stream.next_out = deflated;
2149        stream.avail_out = bound;
2150
2151        stream.next_in = (unsigned char *)data;
2152        stream.avail_in = size;
2153        while (git_deflate(&stream, Z_FINISH) == Z_OK)
2154                ; /* nothing */
2155        git_deflate_end(&stream);
2156        *result_size = stream.total_out;
2157        return deflated;
2158}
2159
2160static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2161                                  const char *prefix)
2162{
2163        void *cp;
2164        void *delta;
2165        void *deflated;
2166        void *data;
2167        unsigned long orig_size;
2168        unsigned long delta_size;
2169        unsigned long deflate_size;
2170        unsigned long data_size;
2171
2172        /* We could do deflated delta, or we could do just deflated two,
2173         * whichever is smaller.
2174         */
2175        delta = NULL;
2176        deflated = deflate_it(two->ptr, two->size, &deflate_size);
2177        if (one->size && two->size) {
2178                delta = diff_delta(one->ptr, one->size,
2179                                   two->ptr, two->size,
2180                                   &delta_size, deflate_size);
2181                if (delta) {
2182                        void *to_free = delta;
2183                        orig_size = delta_size;
2184                        delta = deflate_it(delta, delta_size, &delta_size);
2185                        free(to_free);
2186                }
2187        }
2188
2189        if (delta && delta_size < deflate_size) {
2190                fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2191                free(deflated);
2192                data = delta;
2193                data_size = delta_size;
2194        }
2195        else {
2196                fprintf(file, "%sliteral %lu\n", prefix, two->size);
2197                free(delta);
2198                data = deflated;
2199                data_size = deflate_size;
2200        }
2201
2202        /* emit data encoded in base85 */
2203        cp = data;
2204        while (data_size) {
2205                int bytes = (52 < data_size) ? 52 : data_size;
2206                char line[70];
2207                data_size -= bytes;
2208                if (bytes <= 26)
2209                        line[0] = bytes + 'A' - 1;
2210                else
2211                        line[0] = bytes - 26 + 'a' - 1;
2212                encode_85(line + 1, cp, bytes);
2213                cp = (char *) cp + bytes;
2214                fprintf(file, "%s", prefix);
2215                fputs(line, file);
2216                fputc('\n', file);
2217        }
2218        fprintf(file, "%s\n", prefix);
2219        free(data);
2220}
2221
2222static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2223                             const char *prefix)
2224{
2225        fprintf(file, "%sGIT binary patch\n", prefix);
2226        emit_binary_diff_body(file, one, two, prefix);
2227        emit_binary_diff_body(file, two, one, prefix);
2228}
2229
2230int diff_filespec_is_binary(struct diff_filespec *one)
2231{
2232        if (one->is_binary == -1) {
2233                diff_filespec_load_driver(one);
2234                if (one->driver->binary != -1)
2235                        one->is_binary = one->driver->binary;
2236                else {
2237                        if (!one->data && DIFF_FILE_VALID(one))
2238                                diff_populate_filespec(one, CHECK_BINARY);
2239                        if (one->is_binary == -1 && one->data)
2240                                one->is_binary = buffer_is_binary(one->data,
2241                                                one->size);
2242                        if (one->is_binary == -1)
2243                                one->is_binary = 0;
2244                }
2245        }
2246        return one->is_binary;
2247}
2248
2249static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2250{
2251        diff_filespec_load_driver(one);
2252        return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2253}
2254
2255void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2256{
2257        if (!options->a_prefix)
2258                options->a_prefix = a;
2259        if (!options->b_prefix)
2260                options->b_prefix = b;
2261}
2262
2263struct userdiff_driver *get_textconv(struct diff_filespec *one)
2264{
2265        if (!DIFF_FILE_VALID(one))
2266                return NULL;
2267
2268        diff_filespec_load_driver(one);
2269        return userdiff_get_textconv(one->driver);
2270}
2271
2272static void builtin_diff(const char *name_a,
2273                         const char *name_b,
2274                         struct diff_filespec *one,
2275                         struct diff_filespec *two,
2276                         const char *xfrm_msg,
2277                         int must_show_header,
2278                         struct diff_options *o,
2279                         int complete_rewrite)
2280{
2281        mmfile_t mf1, mf2;
2282        const char *lbl[2];
2283        char *a_one, *b_two;
2284        const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2285        const char *reset = diff_get_color_opt(o, DIFF_RESET);
2286        const char *a_prefix, *b_prefix;
2287        struct userdiff_driver *textconv_one = NULL;
2288        struct userdiff_driver *textconv_two = NULL;
2289        struct strbuf header = STRBUF_INIT;
2290        const char *line_prefix = diff_line_prefix(o);
2291
2292        if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2293                        (!one->mode || S_ISGITLINK(one->mode)) &&
2294                        (!two->mode || S_ISGITLINK(two->mode))) {
2295                const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2296                const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2297                show_submodule_summary(o->file, one->path ? one->path : two->path,
2298                                line_prefix,
2299                                one->sha1, two->sha1, two->dirty_submodule,
2300                                meta, del, add, reset);
2301                return;
2302        }
2303
2304        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2305                textconv_one = get_textconv(one);
2306                textconv_two = get_textconv(two);
2307        }
2308
2309        diff_set_mnemonic_prefix(o, "a/", "b/");
2310        if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2311                a_prefix = o->b_prefix;
2312                b_prefix = o->a_prefix;
2313        } else {
2314                a_prefix = o->a_prefix;
2315                b_prefix = o->b_prefix;
2316        }
2317
2318        /* Never use a non-valid filename anywhere if at all possible */
2319        name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2320        name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2321
2322        a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2323        b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2324        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2325        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2326        strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2327        if (lbl[0][0] == '/') {
2328                /* /dev/null */
2329                strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2330                if (xfrm_msg)
2331                        strbuf_addstr(&header, xfrm_msg);
2332                must_show_header = 1;
2333        }
2334        else if (lbl[1][0] == '/') {
2335                strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2336                if (xfrm_msg)
2337                        strbuf_addstr(&header, xfrm_msg);
2338                must_show_header = 1;
2339        }
2340        else {
2341                if (one->mode != two->mode) {
2342                        strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2343                        strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2344                        must_show_header = 1;
2345                }
2346                if (xfrm_msg)
2347                        strbuf_addstr(&header, xfrm_msg);
2348
2349                /*
2350                 * we do not run diff between different kind
2351                 * of objects.
2352                 */
2353                if ((one->mode ^ two->mode) & S_IFMT)
2354                        goto free_ab_and_return;
2355                if (complete_rewrite &&
2356                    (textconv_one || !diff_filespec_is_binary(one)) &&
2357                    (textconv_two || !diff_filespec_is_binary(two))) {
2358                        fprintf(o->file, "%s", header.buf);
2359                        strbuf_reset(&header);
2360                        emit_rewrite_diff(name_a, name_b, one, two,
2361                                                textconv_one, textconv_two, o);
2362                        o->found_changes = 1;
2363                        goto free_ab_and_return;
2364                }
2365        }
2366
2367        if (o->irreversible_delete && lbl[1][0] == '/') {
2368                fprintf(o->file, "%s", header.buf);
2369                strbuf_reset(&header);
2370                goto free_ab_and_return;
2371        } else if (!DIFF_OPT_TST(o, TEXT) &&
2372            ( (!textconv_one && diff_filespec_is_binary(one)) ||
2373              (!textconv_two && diff_filespec_is_binary(two)) )) {
2374                if (!one->data && !two->data &&
2375                    S_ISREG(one->mode) && S_ISREG(two->mode) &&
2376                    !DIFF_OPT_TST(o, BINARY)) {
2377                        if (!hashcmp(one->sha1, two->sha1)) {
2378                                if (must_show_header)
2379                                        fprintf(o->file, "%s", header.buf);
2380                                goto free_ab_and_return;
2381                        }
2382                        fprintf(o->file, "%s", header.buf);
2383                        fprintf(o->file, "%sBinary files %s and %s differ\n",
2384                                line_prefix, lbl[0], lbl[1]);
2385                        goto free_ab_and_return;
2386                }
2387                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2388                        die("unable to read files to diff");
2389                /* Quite common confusing case */
2390                if (mf1.size == mf2.size &&
2391                    !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2392                        if (must_show_header)
2393                                fprintf(o->file, "%s", header.buf);
2394                        goto free_ab_and_return;
2395                }
2396                fprintf(o->file, "%s", header.buf);
2397                strbuf_reset(&header);
2398                if (DIFF_OPT_TST(o, BINARY))
2399                        emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2400                else
2401                        fprintf(o->file, "%sBinary files %s and %s differ\n",
2402                                line_prefix, lbl[0], lbl[1]);
2403                o->found_changes = 1;
2404        } else {
2405                /* Crazy xdl interfaces.. */
2406                const char *diffopts = getenv("GIT_DIFF_OPTS");
2407                const char *v;
2408                xpparam_t xpp;
2409                xdemitconf_t xecfg;
2410                struct emit_callback ecbdata;
2411                const struct userdiff_funcname *pe;
2412
2413                if (must_show_header) {
2414                        fprintf(o->file, "%s", header.buf);
2415                        strbuf_reset(&header);
2416                }
2417
2418                mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2419                mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2420
2421                pe = diff_funcname_pattern(one);
2422                if (!pe)
2423                        pe = diff_funcname_pattern(two);
2424
2425                memset(&xpp, 0, sizeof(xpp));
2426                memset(&xecfg, 0, sizeof(xecfg));
2427                memset(&ecbdata, 0, sizeof(ecbdata));
2428                ecbdata.label_path = lbl;
2429                ecbdata.color_diff = want_color(o->use_color);
2430                ecbdata.found_changesp = &o->found_changes;
2431                ecbdata.ws_rule = whitespace_rule(name_b);
2432                if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2433                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2434                ecbdata.opt = o;
2435                ecbdata.header = header.len ? &header : NULL;
2436                xpp.flags = o->xdl_opts;
2437                xecfg.ctxlen = o->context;
2438                xecfg.interhunkctxlen = o->interhunkcontext;
2439                xecfg.flags = XDL_EMIT_FUNCNAMES;
2440                if (DIFF_OPT_TST(o, FUNCCONTEXT))
2441                        xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2442                if (pe)
2443                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2444                if (!diffopts)
2445                        ;
2446                else if (skip_prefix(diffopts, "--unified=", &v))
2447                        xecfg.ctxlen = strtoul(v, NULL, 10);
2448                else if (skip_prefix(diffopts, "-u", &v))
2449                        xecfg.ctxlen = strtoul(v, NULL, 10);
2450                if (o->word_diff)
2451                        init_diff_words_data(&ecbdata, o, one, two);
2452                if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2453                                  &xpp, &xecfg))
2454                        die("unable to generate diff for %s", one->path);
2455                if (o->word_diff)
2456                        free_diff_words_data(&ecbdata);
2457                if (textconv_one)
2458                        free(mf1.ptr);
2459                if (textconv_two)
2460                        free(mf2.ptr);
2461                xdiff_clear_find_func(&xecfg);
2462        }
2463
2464 free_ab_and_return:
2465        strbuf_release(&header);
2466        diff_free_filespec_data(one);
2467        diff_free_filespec_data(two);
2468        free(a_one);
2469        free(b_two);
2470        return;
2471}
2472
2473static void builtin_diffstat(const char *name_a, const char *name_b,
2474                             struct diff_filespec *one,
2475                             struct diff_filespec *two,
2476                             struct diffstat_t *diffstat,
2477                             struct diff_options *o,
2478                             struct diff_filepair *p)
2479{
2480        mmfile_t mf1, mf2;
2481        struct diffstat_file *data;
2482        int same_contents;
2483        int complete_rewrite = 0;
2484
2485        if (!DIFF_PAIR_UNMERGED(p)) {
2486                if (p->status == DIFF_STATUS_MODIFIED && p->score)
2487                        complete_rewrite = 1;
2488        }
2489
2490        data = diffstat_add(diffstat, name_a, name_b);
2491        data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2492
2493        if (!one || !two) {
2494                data->is_unmerged = 1;
2495                return;
2496        }
2497
2498        same_contents = !hashcmp(one->sha1, two->sha1);
2499
2500        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2501                data->is_binary = 1;
2502                if (same_contents) {
2503                        data->added = 0;
2504                        data->deleted = 0;
2505                } else {
2506                        data->added = diff_filespec_size(two);
2507                        data->deleted = diff_filespec_size(one);
2508                }
2509        }
2510
2511        else if (complete_rewrite) {
2512                diff_populate_filespec(one, 0);
2513                diff_populate_filespec(two, 0);
2514                data->deleted = count_lines(one->data, one->size);
2515                data->added = count_lines(two->data, two->size);
2516        }
2517
2518        else if (!same_contents) {
2519                /* Crazy xdl interfaces.. */
2520                xpparam_t xpp;
2521                xdemitconf_t xecfg;
2522
2523                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2524                        die("unable to read files to diff");
2525
2526                memset(&xpp, 0, sizeof(xpp));
2527                memset(&xecfg, 0, sizeof(xecfg));
2528                xpp.flags = o->xdl_opts;
2529                xecfg.ctxlen = o->context;
2530                xecfg.interhunkctxlen = o->interhunkcontext;
2531                if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2532                                  &xpp, &xecfg))
2533                        die("unable to generate diffstat for %s", one->path);
2534        }
2535
2536        diff_free_filespec_data(one);
2537        diff_free_filespec_data(two);
2538}
2539
2540static void builtin_checkdiff(const char *name_a, const char *name_b,
2541                              const char *attr_path,
2542                              struct diff_filespec *one,
2543                              struct diff_filespec *two,
2544                              struct diff_options *o)
2545{
2546        mmfile_t mf1, mf2;
2547        struct checkdiff_t data;
2548
2549        if (!two)
2550                return;
2551
2552        memset(&data, 0, sizeof(data));
2553        data.filename = name_b ? name_b : name_a;
2554        data.lineno = 0;
2555        data.o = o;
2556        data.ws_rule = whitespace_rule(attr_path);
2557        data.conflict_marker_size = ll_merge_marker_size(attr_path);
2558
2559        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2560                die("unable to read files to diff");
2561
2562        /*
2563         * All the other codepaths check both sides, but not checking
2564         * the "old" side here is deliberate.  We are checking the newly
2565         * introduced changes, and as long as the "new" side is text, we
2566         * can and should check what it introduces.
2567         */
2568        if (diff_filespec_is_binary(two))
2569                goto free_and_return;
2570        else {
2571                /* Crazy xdl interfaces.. */
2572                xpparam_t xpp;
2573                xdemitconf_t xecfg;
2574
2575                memset(&xpp, 0, sizeof(xpp));
2576                memset(&xecfg, 0, sizeof(xecfg));
2577                xecfg.ctxlen = 1; /* at least one context line */
2578                xpp.flags = 0;
2579                if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2580                                  &xpp, &xecfg))
2581                        die("unable to generate checkdiff for %s", one->path);
2582
2583                if (data.ws_rule & WS_BLANK_AT_EOF) {
2584                        struct emit_callback ecbdata;
2585                        int blank_at_eof;
2586
2587                        ecbdata.ws_rule = data.ws_rule;
2588                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2589                        blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2590
2591                        if (blank_at_eof) {
2592                                static char *err;
2593                                if (!err)
2594                                        err = whitespace_error_string(WS_BLANK_AT_EOF);
2595                                fprintf(o->file, "%s:%d: %s.\n",
2596                                        data.filename, blank_at_eof, err);
2597                                data.status = 1; /* report errors */
2598                        }
2599                }
2600        }
2601 free_and_return:
2602        diff_free_filespec_data(one);
2603        diff_free_filespec_data(two);
2604        if (data.status)
2605                DIFF_OPT_SET(o, CHECK_FAILED);
2606}
2607
2608struct diff_filespec *alloc_filespec(const char *path)
2609{
2610        struct diff_filespec *spec;
2611
2612        FLEXPTR_ALLOC_STR(spec, path, path);
2613        spec->count = 1;
2614        spec->is_binary = -1;
2615        return spec;
2616}
2617
2618void free_filespec(struct diff_filespec *spec)
2619{
2620        if (!--spec->count) {
2621                diff_free_filespec_data(spec);
2622                free(spec);
2623        }
2624}
2625
2626void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2627                   int sha1_valid, unsigned short mode)
2628{
2629        if (mode) {
2630                spec->mode = canon_mode(mode);
2631                hashcpy(spec->sha1, sha1);
2632                spec->sha1_valid = sha1_valid;
2633        }
2634}
2635
2636/*
2637 * Given a name and sha1 pair, if the index tells us the file in
2638 * the work tree has that object contents, return true, so that
2639 * prepare_temp_file() does not have to inflate and extract.
2640 */
2641static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2642{
2643        const struct cache_entry *ce;
2644        struct stat st;
2645        int pos, len;
2646
2647        /*
2648         * We do not read the cache ourselves here, because the
2649         * benchmark with my previous version that always reads cache
2650         * shows that it makes things worse for diff-tree comparing
2651         * two linux-2.6 kernel trees in an already checked out work
2652         * tree.  This is because most diff-tree comparisons deal with
2653         * only a small number of files, while reading the cache is
2654         * expensive for a large project, and its cost outweighs the
2655         * savings we get by not inflating the object to a temporary
2656         * file.  Practically, this code only helps when we are used
2657         * by diff-cache --cached, which does read the cache before
2658         * calling us.
2659         */
2660        if (!active_cache)
2661                return 0;
2662
2663        /* We want to avoid the working directory if our caller
2664         * doesn't need the data in a normal file, this system
2665         * is rather slow with its stat/open/mmap/close syscalls,
2666         * and the object is contained in a pack file.  The pack
2667         * is probably already open and will be faster to obtain
2668         * the data through than the working directory.  Loose
2669         * objects however would tend to be slower as they need
2670         * to be individually opened and inflated.
2671         */
2672        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2673                return 0;
2674
2675        len = strlen(name);
2676        pos = cache_name_pos(name, len);
2677        if (pos < 0)
2678                return 0;
2679        ce = active_cache[pos];
2680
2681        /*
2682         * This is not the sha1 we are looking for, or
2683         * unreusable because it is not a regular file.
2684         */
2685        if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2686                return 0;
2687
2688        /*
2689         * If ce is marked as "assume unchanged", there is no
2690         * guarantee that work tree matches what we are looking for.
2691         */
2692        if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2693                return 0;
2694
2695        /*
2696         * If ce matches the file in the work tree, we can reuse it.
2697         */
2698        if (ce_uptodate(ce) ||
2699            (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2700                return 1;
2701
2702        return 0;
2703}
2704
2705static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2706{
2707        struct strbuf buf = STRBUF_INIT;
2708        char *dirty = "";
2709
2710        /* Are we looking at the work tree? */
2711        if (s->dirty_submodule)
2712                dirty = "-dirty";
2713
2714        strbuf_addf(&buf, "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2715        s->size = buf.len;
2716        if (size_only) {
2717                s->data = NULL;
2718                strbuf_release(&buf);
2719        } else {
2720                s->data = strbuf_detach(&buf, NULL);
2721                s->should_free = 1;
2722        }
2723        return 0;
2724}
2725
2726/*
2727 * While doing rename detection and pickaxe operation, we may need to
2728 * grab the data for the blob (or file) for our own in-core comparison.
2729 * diff_filespec has data and size fields for this purpose.
2730 */
2731int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2732{
2733        int size_only = flags & CHECK_SIZE_ONLY;
2734        int err = 0;
2735        /*
2736         * demote FAIL to WARN to allow inspecting the situation
2737         * instead of refusing.
2738         */
2739        enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2740                                    ? SAFE_CRLF_WARN
2741                                    : safe_crlf);
2742
2743        if (!DIFF_FILE_VALID(s))
2744                die("internal error: asking to populate invalid file.");
2745        if (S_ISDIR(s->mode))
2746                return -1;
2747
2748        if (s->data)
2749                return 0;
2750
2751        if (size_only && 0 < s->size)
2752                return 0;
2753
2754        if (S_ISGITLINK(s->mode))
2755                return diff_populate_gitlink(s, size_only);
2756
2757        if (!s->sha1_valid ||
2758            reuse_worktree_file(s->path, s->sha1, 0)) {
2759                struct strbuf buf = STRBUF_INIT;
2760                struct stat st;
2761                int fd;
2762
2763                if (lstat(s->path, &st) < 0) {
2764                        if (errno == ENOENT) {
2765                        err_empty:
2766                                err = -1;
2767                        empty:
2768                                s->data = (char *)"";
2769                                s->size = 0;
2770                                return err;
2771                        }
2772                }
2773                s->size = xsize_t(st.st_size);
2774                if (!s->size)
2775                        goto empty;
2776                if (S_ISLNK(st.st_mode)) {
2777                        struct strbuf sb = STRBUF_INIT;
2778
2779                        if (strbuf_readlink(&sb, s->path, s->size))
2780                                goto err_empty;
2781                        s->size = sb.len;
2782                        s->data = strbuf_detach(&sb, NULL);
2783                        s->should_free = 1;
2784                        return 0;
2785                }
2786                if (size_only)
2787                        return 0;
2788                if ((flags & CHECK_BINARY) &&
2789                    s->size > big_file_threshold && s->is_binary == -1) {
2790                        s->is_binary = 1;
2791                        return 0;
2792                }
2793                fd = open(s->path, O_RDONLY);
2794                if (fd < 0)
2795                        goto err_empty;
2796                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2797                close(fd);
2798                s->should_munmap = 1;
2799
2800                /*
2801                 * Convert from working tree format to canonical git format
2802                 */
2803                if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2804                        size_t size = 0;
2805                        munmap(s->data, s->size);
2806                        s->should_munmap = 0;
2807                        s->data = strbuf_detach(&buf, &size);
2808                        s->size = size;
2809                        s->should_free = 1;
2810                }
2811        }
2812        else {
2813                enum object_type type;
2814                if (size_only || (flags & CHECK_BINARY)) {
2815                        type = sha1_object_info(s->sha1, &s->size);
2816                        if (type < 0)
2817                                die("unable to read %s", sha1_to_hex(s->sha1));
2818                        if (size_only)
2819                                return 0;
2820                        if (s->size > big_file_threshold && s->is_binary == -1) {
2821                                s->is_binary = 1;
2822                                return 0;
2823                        }
2824                }
2825                s->data = read_sha1_file(s->sha1, &type, &s->size);
2826                if (!s->data)
2827                        die("unable to read %s", sha1_to_hex(s->sha1));
2828                s->should_free = 1;
2829        }
2830        return 0;
2831}
2832
2833void diff_free_filespec_blob(struct diff_filespec *s)
2834{
2835        if (s->should_free)
2836                free(s->data);
2837        else if (s->should_munmap)
2838                munmap(s->data, s->size);
2839
2840        if (s->should_free || s->should_munmap) {
2841                s->should_free = s->should_munmap = 0;
2842                s->data = NULL;
2843        }
2844}
2845
2846void diff_free_filespec_data(struct diff_filespec *s)
2847{
2848        diff_free_filespec_blob(s);
2849        free(s->cnt_data);
2850        s->cnt_data = NULL;
2851}
2852
2853static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2854                           void *blob,
2855                           unsigned long size,
2856                           const unsigned char *sha1,
2857                           int mode)
2858{
2859        int fd;
2860        struct strbuf buf = STRBUF_INIT;
2861        struct strbuf template = STRBUF_INIT;
2862        char *path_dup = xstrdup(path);
2863        const char *base = basename(path_dup);
2864
2865        /* Generate "XXXXXX_basename.ext" */
2866        strbuf_addstr(&template, "XXXXXX_");
2867        strbuf_addstr(&template, base);
2868
2869        fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2870        if (fd < 0)
2871                die_errno("unable to create temp-file");
2872        if (convert_to_working_tree(path,
2873                        (const char *)blob, (size_t)size, &buf)) {
2874                blob = buf.buf;
2875                size = buf.len;
2876        }
2877        if (write_in_full(fd, blob, size) != size)
2878                die_errno("unable to write temp-file");
2879        close_tempfile(&temp->tempfile);
2880        temp->name = get_tempfile_path(&temp->tempfile);
2881        sha1_to_hex_r(temp->hex, sha1);
2882        xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
2883        strbuf_release(&buf);
2884        strbuf_release(&template);
2885        free(path_dup);
2886}
2887
2888static struct diff_tempfile *prepare_temp_file(const char *name,
2889                struct diff_filespec *one)
2890{
2891        struct diff_tempfile *temp = claim_diff_tempfile();
2892
2893        if (!DIFF_FILE_VALID(one)) {
2894        not_a_valid_file:
2895                /* A '-' entry produces this for file-2, and
2896                 * a '+' entry produces this for file-1.
2897                 */
2898                temp->name = "/dev/null";
2899                xsnprintf(temp->hex, sizeof(temp->hex), ".");
2900                xsnprintf(temp->mode, sizeof(temp->mode), ".");
2901                return temp;
2902        }
2903
2904        if (!S_ISGITLINK(one->mode) &&
2905            (!one->sha1_valid ||
2906             reuse_worktree_file(name, one->sha1, 1))) {
2907                struct stat st;
2908                if (lstat(name, &st) < 0) {
2909                        if (errno == ENOENT)
2910                                goto not_a_valid_file;
2911                        die_errno("stat(%s)", name);
2912                }
2913                if (S_ISLNK(st.st_mode)) {
2914                        struct strbuf sb = STRBUF_INIT;
2915                        if (strbuf_readlink(&sb, name, st.st_size) < 0)
2916                                die_errno("readlink(%s)", name);
2917                        prep_temp_blob(name, temp, sb.buf, sb.len,
2918                                       (one->sha1_valid ?
2919                                        one->sha1 : null_sha1),
2920                                       (one->sha1_valid ?
2921                                        one->mode : S_IFLNK));
2922                        strbuf_release(&sb);
2923                }
2924                else {
2925                        /* we can borrow from the file in the work tree */
2926                        temp->name = name;
2927                        if (!one->sha1_valid)
2928                                sha1_to_hex_r(temp->hex, null_sha1);
2929                        else
2930                                sha1_to_hex_r(temp->hex, one->sha1);
2931                        /* Even though we may sometimes borrow the
2932                         * contents from the work tree, we always want
2933                         * one->mode.  mode is trustworthy even when
2934                         * !(one->sha1_valid), as long as
2935                         * DIFF_FILE_VALID(one).
2936                         */
2937                        xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
2938                }
2939                return temp;
2940        }
2941        else {
2942                if (diff_populate_filespec(one, 0))
2943                        die("cannot read data blob for %s", one->path);
2944                prep_temp_blob(name, temp, one->data, one->size,
2945                               one->sha1, one->mode);
2946        }
2947        return temp;
2948}
2949
2950static void add_external_diff_name(struct argv_array *argv,
2951                                   const char *name,
2952                                   struct diff_filespec *df)
2953{
2954        struct diff_tempfile *temp = prepare_temp_file(name, df);
2955        argv_array_push(argv, temp->name);
2956        argv_array_push(argv, temp->hex);
2957        argv_array_push(argv, temp->mode);
2958}
2959
2960/* An external diff command takes:
2961 *
2962 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2963 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2964 *
2965 */
2966static void run_external_diff(const char *pgm,
2967                              const char *name,
2968                              const char *other,
2969                              struct diff_filespec *one,
2970                              struct diff_filespec *two,
2971                              const char *xfrm_msg,
2972                              int complete_rewrite,
2973                              struct diff_options *o)
2974{
2975        struct argv_array argv = ARGV_ARRAY_INIT;
2976        struct argv_array env = ARGV_ARRAY_INIT;
2977        struct diff_queue_struct *q = &diff_queued_diff;
2978
2979        argv_array_push(&argv, pgm);
2980        argv_array_push(&argv, name);
2981
2982        if (one && two) {
2983                add_external_diff_name(&argv, name, one);
2984                if (!other)
2985                        add_external_diff_name(&argv, name, two);
2986                else {
2987                        add_external_diff_name(&argv, other, two);
2988                        argv_array_push(&argv, other);
2989                        argv_array_push(&argv, xfrm_msg);
2990                }
2991        }
2992
2993        argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2994        argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2995
2996        if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
2997                die(_("external diff died, stopping at %s"), name);
2998
2999        remove_tempfile();
3000        argv_array_clear(&argv);
3001        argv_array_clear(&env);
3002}
3003
3004static int similarity_index(struct diff_filepair *p)
3005{
3006        return p->score * 100 / MAX_SCORE;
3007}
3008
3009static void fill_metainfo(struct strbuf *msg,
3010                          const char *name,
3011                          const char *other,
3012                          struct diff_filespec *one,
3013                          struct diff_filespec *two,
3014                          struct diff_options *o,
3015                          struct diff_filepair *p,
3016                          int *must_show_header,
3017                          int use_color)
3018{
3019        const char *set = diff_get_color(use_color, DIFF_METAINFO);
3020        const char *reset = diff_get_color(use_color, DIFF_RESET);
3021        const char *line_prefix = diff_line_prefix(o);
3022
3023        *must_show_header = 1;
3024        strbuf_init(msg, PATH_MAX * 2 + 300);
3025        switch (p->status) {
3026        case DIFF_STATUS_COPIED:
3027                strbuf_addf(msg, "%s%ssimilarity index %d%%",
3028                            line_prefix, set, similarity_index(p));
3029                strbuf_addf(msg, "%s\n%s%scopy from ",
3030                            reset,  line_prefix, set);
3031                quote_c_style(name, msg, NULL, 0);
3032                strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3033                quote_c_style(other, msg, NULL, 0);
3034                strbuf_addf(msg, "%s\n", reset);
3035                break;
3036        case DIFF_STATUS_RENAMED:
3037                strbuf_addf(msg, "%s%ssimilarity index %d%%",
3038                            line_prefix, set, similarity_index(p));
3039                strbuf_addf(msg, "%s\n%s%srename from ",
3040                            reset, line_prefix, set);
3041                quote_c_style(name, msg, NULL, 0);
3042                strbuf_addf(msg, "%s\n%s%srename to ",
3043                            reset, line_prefix, set);
3044                quote_c_style(other, msg, NULL, 0);
3045                strbuf_addf(msg, "%s\n", reset);
3046                break;
3047        case DIFF_STATUS_MODIFIED:
3048                if (p->score) {
3049                        strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3050                                    line_prefix,
3051                                    set, similarity_index(p), reset);
3052                        break;
3053                }
3054                /* fallthru */
3055        default:
3056                *must_show_header = 0;
3057        }
3058        if (one && two && hashcmp(one->sha1, two->sha1)) {
3059                int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3060
3061                if (DIFF_OPT_TST(o, BINARY)) {
3062                        mmfile_t mf;
3063                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3064                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3065                                abbrev = 40;
3066                }
3067                strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3068                            find_unique_abbrev(one->sha1, abbrev));
3069                strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3070                if (one->mode == two->mode)
3071                        strbuf_addf(msg, " %06o", one->mode);
3072                strbuf_addf(msg, "%s\n", reset);
3073        }
3074}
3075
3076static void run_diff_cmd(const char *pgm,
3077                         const char *name,
3078                         const char *other,
3079                         const char *attr_path,
3080                         struct diff_filespec *one,
3081                         struct diff_filespec *two,
3082                         struct strbuf *msg,
3083                         struct diff_options *o,
3084                         struct diff_filepair *p)
3085{
3086        const char *xfrm_msg = NULL;
3087        int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3088        int must_show_header = 0;
3089
3090
3091        if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3092                struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3093                if (drv && drv->external)
3094                        pgm = drv->external;
3095        }
3096
3097        if (msg) {
3098                /*
3099                 * don't use colors when the header is intended for an
3100                 * external diff driver
3101                 */
3102                fill_metainfo(msg, name, other, one, two, o, p,
3103                              &must_show_header,
3104                              want_color(o->use_color) && !pgm);
3105                xfrm_msg = msg->len ? msg->buf : NULL;
3106        }
3107
3108        if (pgm) {
3109                run_external_diff(pgm, name, other, one, two, xfrm_msg,
3110                                  complete_rewrite, o);
3111                return;
3112        }
3113        if (one && two)
3114                builtin_diff(name, other ? other : name,
3115                             one, two, xfrm_msg, must_show_header,
3116                             o, complete_rewrite);
3117        else
3118                fprintf(o->file, "* Unmerged path %s\n", name);
3119}
3120
3121static void diff_fill_sha1_info(struct diff_filespec *one)
3122{
3123        if (DIFF_FILE_VALID(one)) {
3124                if (!one->sha1_valid) {
3125                        struct stat st;
3126                        if (one->is_stdin) {
3127                                hashcpy(one->sha1, null_sha1);
3128                                return;
3129                        }
3130                        if (lstat(one->path, &st) < 0)
3131                                die_errno("stat '%s'", one->path);
3132                        if (index_path(one->sha1, one->path, &st, 0))
3133                                die("cannot hash %s", one->path);
3134                }
3135        }
3136        else
3137                hashclr(one->sha1);
3138}
3139
3140static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3141{
3142        /* Strip the prefix but do not molest /dev/null and absolute paths */
3143        if (*namep && **namep != '/') {
3144                *namep += prefix_length;
3145                if (**namep == '/')
3146                        ++*namep;
3147        }
3148        if (*otherp && **otherp != '/') {
3149                *otherp += prefix_length;
3150                if (**otherp == '/')
3151                        ++*otherp;
3152        }
3153}
3154
3155static void run_diff(struct diff_filepair *p, struct diff_options *o)
3156{
3157        const char *pgm = external_diff();
3158        struct strbuf msg;
3159        struct diff_filespec *one = p->one;
3160        struct diff_filespec *two = p->two;
3161        const char *name;
3162        const char *other;
3163        const char *attr_path;
3164
3165        name  = p->one->path;
3166        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3167        attr_path = name;
3168        if (o->prefix_length)
3169                strip_prefix(o->prefix_length, &name, &other);
3170
3171        if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3172                pgm = NULL;
3173
3174        if (DIFF_PAIR_UNMERGED(p)) {
3175                run_diff_cmd(pgm, name, NULL, attr_path,
3176                             NULL, NULL, NULL, o, p);
3177                return;
3178        }
3179
3180        diff_fill_sha1_info(one);
3181        diff_fill_sha1_info(two);
3182
3183        if (!pgm &&
3184            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3185            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3186                /*
3187                 * a filepair that changes between file and symlink
3188                 * needs to be split into deletion and creation.
3189                 */
3190                struct diff_filespec *null = alloc_filespec(two->path);
3191                run_diff_cmd(NULL, name, other, attr_path,
3192                             one, null, &msg, o, p);
3193                free(null);
3194                strbuf_release(&msg);
3195
3196                null = alloc_filespec(one->path);
3197                run_diff_cmd(NULL, name, other, attr_path,
3198                             null, two, &msg, o, p);
3199                free(null);
3200        }
3201        else
3202                run_diff_cmd(pgm, name, other, attr_path,
3203                             one, two, &msg, o, p);
3204
3205        strbuf_release(&msg);
3206}
3207
3208static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3209                         struct diffstat_t *diffstat)
3210{
3211        const char *name;
3212        const char *other;
3213
3214        if (DIFF_PAIR_UNMERGED(p)) {
3215                /* unmerged */
3216                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3217                return;
3218        }
3219
3220        name = p->one->path;
3221        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3222
3223        if (o->prefix_length)
3224                strip_prefix(o->prefix_length, &name, &other);
3225
3226        diff_fill_sha1_info(p->one);
3227        diff_fill_sha1_info(p->two);
3228
3229        builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3230}
3231
3232static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3233{
3234        const char *name;
3235        const char *other;
3236        const char *attr_path;
3237
3238        if (DIFF_PAIR_UNMERGED(p)) {
3239                /* unmerged */
3240                return;
3241        }
3242
3243        name = p->one->path;
3244        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3245        attr_path = other ? other : name;
3246
3247        if (o->prefix_length)
3248                strip_prefix(o->prefix_length, &name, &other);
3249
3250        diff_fill_sha1_info(p->one);
3251        diff_fill_sha1_info(p->two);
3252
3253        builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3254}
3255
3256void diff_setup(struct diff_options *options)
3257{
3258        memcpy(options, &default_diff_options, sizeof(*options));
3259
3260        options->file = stdout;
3261
3262        options->line_termination = '\n';
3263        options->break_opt = -1;
3264        options->rename_limit = -1;
3265        options->dirstat_permille = diff_dirstat_permille_default;
3266        options->context = diff_context_default;
3267        options->ws_error_highlight = WSEH_NEW;
3268        DIFF_OPT_SET(options, RENAME_EMPTY);
3269
3270        /* pathchange left =NULL by default */
3271        options->change = diff_change;
3272        options->add_remove = diff_addremove;
3273        options->use_color = diff_use_color_default;
3274        options->detect_rename = diff_detect_rename_default;
3275        options->xdl_opts |= diff_algorithm;
3276
3277        options->orderfile = diff_order_file_cfg;
3278
3279        if (diff_no_prefix) {
3280                options->a_prefix = options->b_prefix = "";
3281        } else if (!diff_mnemonic_prefix) {
3282                options->a_prefix = "a/";
3283                options->b_prefix = "b/";
3284        }
3285}
3286
3287void diff_setup_done(struct diff_options *options)
3288{
3289        int count = 0;
3290
3291        if (options->set_default)
3292                options->set_default(options);
3293
3294        if (options->output_format & DIFF_FORMAT_NAME)
3295                count++;
3296        if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3297                count++;
3298        if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3299                count++;
3300        if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3301                count++;
3302        if (count > 1)
3303                die("--name-only, --name-status, --check and -s are mutually exclusive");
3304
3305        /*
3306         * Most of the time we can say "there are changes"
3307         * only by checking if there are changed paths, but
3308         * --ignore-whitespace* options force us to look
3309         * inside contents.
3310         */
3311
3312        if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3313            DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3314            DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3315                DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3316        else
3317                DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3318
3319        if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3320                options->detect_rename = DIFF_DETECT_COPY;
3321
3322        if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3323                options->prefix = NULL;
3324        if (options->prefix)
3325                options->prefix_length = strlen(options->prefix);
3326        else
3327                options->prefix_length = 0;
3328
3329        if (options->output_format & (DIFF_FORMAT_NAME |
3330                                      DIFF_FORMAT_NAME_STATUS |
3331                                      DIFF_FORMAT_CHECKDIFF |
3332                                      DIFF_FORMAT_NO_OUTPUT))
3333                options->output_format &= ~(DIFF_FORMAT_RAW |
3334                                            DIFF_FORMAT_NUMSTAT |
3335                                            DIFF_FORMAT_DIFFSTAT |
3336                                            DIFF_FORMAT_SHORTSTAT |
3337                                            DIFF_FORMAT_DIRSTAT |
3338                                            DIFF_FORMAT_SUMMARY |
3339                                            DIFF_FORMAT_PATCH);
3340
3341        /*
3342         * These cases always need recursive; we do not drop caller-supplied
3343         * recursive bits for other formats here.
3344         */
3345        if (options->output_format & (DIFF_FORMAT_PATCH |
3346                                      DIFF_FORMAT_NUMSTAT |
3347                                      DIFF_FORMAT_DIFFSTAT |
3348                                      DIFF_FORMAT_SHORTSTAT |
3349                                      DIFF_FORMAT_DIRSTAT |
3350                                      DIFF_FORMAT_SUMMARY |
3351                                      DIFF_FORMAT_CHECKDIFF))
3352                DIFF_OPT_SET(options, RECURSIVE);
3353        /*
3354         * Also pickaxe would not work very well if you do not say recursive
3355         */
3356        if (options->pickaxe)
3357                DIFF_OPT_SET(options, RECURSIVE);
3358        /*
3359         * When patches are generated, submodules diffed against the work tree
3360         * must be checked for dirtiness too so it can be shown in the output
3361         */
3362        if (options->output_format & DIFF_FORMAT_PATCH)
3363                DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3364
3365        if (options->detect_rename && options->rename_limit < 0)
3366                options->rename_limit = diff_rename_limit_default;
3367        if (options->setup & DIFF_SETUP_USE_CACHE) {
3368                if (!active_cache)
3369                        /* read-cache does not die even when it fails
3370                         * so it is safe for us to do this here.  Also
3371                         * it does not smudge active_cache or active_nr
3372                         * when it fails, so we do not have to worry about
3373                         * cleaning it up ourselves either.
3374                         */
3375                        read_cache();
3376        }
3377        if (options->abbrev <= 0 || 40 < options->abbrev)
3378                options->abbrev = 40; /* full */
3379
3380        /*
3381         * It does not make sense to show the first hit we happened
3382         * to have found.  It does not make sense not to return with
3383         * exit code in such a case either.
3384         */
3385        if (DIFF_OPT_TST(options, QUICK)) {
3386                options->output_format = DIFF_FORMAT_NO_OUTPUT;
3387                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3388        }
3389
3390        options->diff_path_counter = 0;
3391
3392        if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3393                die(_("--follow requires exactly one pathspec"));
3394}
3395
3396static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3397{
3398        char c, *eq;
3399        int len;
3400
3401        if (*arg != '-')
3402                return 0;
3403        c = *++arg;
3404        if (!c)
3405                return 0;
3406        if (c == arg_short) {
3407                c = *++arg;
3408                if (!c)
3409                        return 1;
3410                if (val && isdigit(c)) {
3411                        char *end;
3412                        int n = strtoul(arg, &end, 10);
3413                        if (*end)
3414                                return 0;
3415                        *val = n;
3416                        return 1;
3417                }
3418                return 0;
3419        }
3420        if (c != '-')
3421                return 0;
3422        arg++;
3423        eq = strchrnul(arg, '=');
3424        len = eq - arg;
3425        if (!len || strncmp(arg, arg_long, len))
3426                return 0;
3427        if (*eq) {
3428                int n;
3429                char *end;
3430                if (!isdigit(*++eq))
3431                        return 0;
3432                n = strtoul(eq, &end, 10);
3433                if (*end)
3434                        return 0;
3435                *val = n;
3436        }
3437        return 1;
3438}
3439
3440static int diff_scoreopt_parse(const char *opt);
3441
3442static inline int short_opt(char opt, const char **argv,
3443                            const char **optarg)
3444{
3445        const char *arg = argv[0];
3446        if (arg[0] != '-' || arg[1] != opt)
3447                return 0;
3448        if (arg[2] != '\0') {
3449                *optarg = arg + 2;
3450                return 1;
3451        }
3452        if (!argv[1])
3453                die("Option '%c' requires a value", opt);
3454        *optarg = argv[1];
3455        return 2;
3456}
3457
3458int parse_long_opt(const char *opt, const char **argv,
3459                   const char **optarg)
3460{
3461        const char *arg = argv[0];
3462        if (!skip_prefix(arg, "--", &arg))
3463                return 0;
3464        if (!skip_prefix(arg, opt, &arg))
3465                return 0;
3466        if (*arg == '=') { /* stuck form: --option=value */
3467                *optarg = arg + 1;
3468                return 1;
3469        }
3470        if (*arg != '\0')
3471                return 0;
3472        /* separate form: --option value */
3473        if (!argv[1])
3474                die("Option '--%s' requires a value", opt);
3475        *optarg = argv[1];
3476        return 2;
3477}
3478
3479static int stat_opt(struct diff_options *options, const char **av)
3480{
3481        const char *arg = av[0];
3482        char *end;
3483        int width = options->stat_width;
3484        int name_width = options->stat_name_width;
3485        int graph_width = options->stat_graph_width;
3486        int count = options->stat_count;
3487        int argcount = 1;
3488
3489        if (!skip_prefix(arg, "--stat", &arg))
3490                die("BUG: stat option does not begin with --stat: %s", arg);
3491        end = (char *)arg;
3492
3493        switch (*arg) {
3494        case '-':
3495                if (skip_prefix(arg, "-width", &arg)) {
3496                        if (*arg == '=')
3497                                width = strtoul(arg + 1, &end, 10);
3498                        else if (!*arg && !av[1])
3499                                die("Option '--stat-width' requires a value");
3500                        else if (!*arg) {
3501                                width = strtoul(av[1], &end, 10);
3502                                argcount = 2;
3503                        }
3504                } else if (skip_prefix(arg, "-name-width", &arg)) {
3505                        if (*arg == '=')
3506                                name_width = strtoul(arg + 1, &end, 10);
3507                        else if (!*arg && !av[1])
3508                                die("Option '--stat-name-width' requires a value");
3509                        else if (!*arg) {
3510                                name_width = strtoul(av[1], &end, 10);
3511                                argcount = 2;
3512                        }
3513                } else if (skip_prefix(arg, "-graph-width", &arg)) {
3514                        if (*arg == '=')
3515                                graph_width = strtoul(arg + 1, &end, 10);
3516                        else if (!*arg && !av[1])
3517                                die("Option '--stat-graph-width' requires a value");
3518                        else if (!*arg) {
3519                                graph_width = strtoul(av[1], &end, 10);
3520                                argcount = 2;
3521                        }
3522                } else if (skip_prefix(arg, "-count", &arg)) {
3523                        if (*arg == '=')
3524                                count = strtoul(arg + 1, &end, 10);
3525                        else if (!*arg && !av[1])
3526                                die("Option '--stat-count' requires a value");
3527                        else if (!*arg) {
3528                                count = strtoul(av[1], &end, 10);
3529                                argcount = 2;
3530                        }
3531                }
3532                break;
3533        case '=':
3534                width = strtoul(arg+1, &end, 10);
3535                if (*end == ',')
3536                        name_width = strtoul(end+1, &end, 10);
3537                if (*end == ',')
3538                        count = strtoul(end+1, &end, 10);
3539        }
3540
3541        /* Important! This checks all the error cases! */
3542        if (*end)
3543                return 0;
3544        options->output_format |= DIFF_FORMAT_DIFFSTAT;
3545        options->stat_name_width = name_width;
3546        options->stat_graph_width = graph_width;
3547        options->stat_width = width;
3548        options->stat_count = count;
3549        return argcount;
3550}
3551
3552static int parse_dirstat_opt(struct diff_options *options, const char *params)
3553{
3554        struct strbuf errmsg = STRBUF_INIT;
3555        if (parse_dirstat_params(options, params, &errmsg))
3556                die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3557                    errmsg.buf);
3558        strbuf_release(&errmsg);
3559        /*
3560         * The caller knows a dirstat-related option is given from the command
3561         * line; allow it to say "return this_function();"
3562         */
3563        options->output_format |= DIFF_FORMAT_DIRSTAT;
3564        return 1;
3565}
3566
3567static int parse_submodule_opt(struct diff_options *options, const char *value)
3568{
3569        if (parse_submodule_params(options, value))
3570                die(_("Failed to parse --submodule option parameter: '%s'"),
3571                        value);
3572        return 1;
3573}
3574
3575static const char diff_status_letters[] = {
3576        DIFF_STATUS_ADDED,
3577        DIFF_STATUS_COPIED,
3578        DIFF_STATUS_DELETED,
3579        DIFF_STATUS_MODIFIED,
3580        DIFF_STATUS_RENAMED,
3581        DIFF_STATUS_TYPE_CHANGED,
3582        DIFF_STATUS_UNKNOWN,
3583        DIFF_STATUS_UNMERGED,
3584        DIFF_STATUS_FILTER_AON,
3585        DIFF_STATUS_FILTER_BROKEN,
3586        '\0',
3587};
3588
3589static unsigned int filter_bit['Z' + 1];
3590
3591static void prepare_filter_bits(void)
3592{
3593        int i;
3594
3595        if (!filter_bit[DIFF_STATUS_ADDED]) {
3596                for (i = 0; diff_status_letters[i]; i++)
3597                        filter_bit[(int) diff_status_letters[i]] = (1 << i);
3598        }
3599}
3600
3601static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3602{
3603        return opt->filter & filter_bit[(int) status];
3604}
3605
3606static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3607{
3608        int i, optch;
3609
3610        prepare_filter_bits();
3611
3612        /*
3613         * If there is a negation e.g. 'd' in the input, and we haven't
3614         * initialized the filter field with another --diff-filter, start
3615         * from full set of bits, except for AON.
3616         */
3617        if (!opt->filter) {
3618                for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3619                        if (optch < 'a' || 'z' < optch)
3620                                continue;
3621                        opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3622                        opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3623                        break;
3624                }
3625        }
3626
3627        for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3628                unsigned int bit;
3629                int negate;
3630
3631                if ('a' <= optch && optch <= 'z') {
3632                        negate = 1;
3633                        optch = toupper(optch);
3634                } else {
3635                        negate = 0;
3636                }
3637
3638                bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3639                if (!bit)
3640                        return optarg[i];
3641                if (negate)
3642                        opt->filter &= ~bit;
3643                else
3644                        opt->filter |= bit;
3645        }
3646        return 0;
3647}
3648
3649static void enable_patch_output(int *fmt) {
3650        *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3651        *fmt |= DIFF_FORMAT_PATCH;
3652}
3653
3654static int parse_one_token(const char **arg, const char *token)
3655{
3656        const char *rest;
3657        if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
3658                *arg = rest;
3659                return 1;
3660        }
3661        return 0;
3662}
3663
3664static int parse_ws_error_highlight(struct diff_options *opt, const char *arg)
3665{
3666        const char *orig_arg = arg;
3667        unsigned val = 0;
3668        while (*arg) {
3669                if (parse_one_token(&arg, "none"))
3670                        val = 0;
3671                else if (parse_one_token(&arg, "default"))
3672                        val = WSEH_NEW;
3673                else if (parse_one_token(&arg, "all"))
3674                        val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
3675                else if (parse_one_token(&arg, "new"))
3676                        val |= WSEH_NEW;
3677                else if (parse_one_token(&arg, "old"))
3678                        val |= WSEH_OLD;
3679                else if (parse_one_token(&arg, "context"))
3680                        val |= WSEH_CONTEXT;
3681                else {
3682                        error("unknown value after ws-error-highlight=%.*s",
3683                              (int)(arg - orig_arg), orig_arg);
3684                        return 0;
3685                }
3686                if (*arg)
3687                        arg++;
3688        }
3689        opt->ws_error_highlight = val;
3690        return 1;
3691}
3692
3693int diff_opt_parse(struct diff_options *options,
3694                   const char **av, int ac, const char *prefix)
3695{
3696        const char *arg = av[0];
3697        const char *optarg;
3698        int argcount;
3699
3700        if (!prefix)
3701                prefix = "";
3702
3703        /* Output format options */
3704        if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3705            || opt_arg(arg, 'U', "unified", &options->context))
3706                enable_patch_output(&options->output_format);
3707        else if (!strcmp(arg, "--raw"))
3708                options->output_format |= DIFF_FORMAT_RAW;
3709        else if (!strcmp(arg, "--patch-with-raw")) {
3710                enable_patch_output(&options->output_format);
3711                options->output_format |= DIFF_FORMAT_RAW;
3712        } else if (!strcmp(arg, "--numstat"))
3713                options->output_format |= DIFF_FORMAT_NUMSTAT;
3714        else if (!strcmp(arg, "--shortstat"))
3715                options->output_format |= DIFF_FORMAT_SHORTSTAT;
3716        else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3717                return parse_dirstat_opt(options, "");
3718        else if (skip_prefix(arg, "-X", &arg))
3719                return parse_dirstat_opt(options, arg);
3720        else if (skip_prefix(arg, "--dirstat=", &arg))
3721                return parse_dirstat_opt(options, arg);
3722        else if (!strcmp(arg, "--cumulative"))
3723                return parse_dirstat_opt(options, "cumulative");
3724        else if (!strcmp(arg, "--dirstat-by-file"))
3725                return parse_dirstat_opt(options, "files");
3726        else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3727                parse_dirstat_opt(options, "files");
3728                return parse_dirstat_opt(options, arg);
3729        }
3730        else if (!strcmp(arg, "--check"))
3731                options->output_format |= DIFF_FORMAT_CHECKDIFF;
3732        else if (!strcmp(arg, "--summary"))
3733                options->output_format |= DIFF_FORMAT_SUMMARY;
3734        else if (!strcmp(arg, "--patch-with-stat")) {
3735                enable_patch_output(&options->output_format);
3736                options->output_format |= DIFF_FORMAT_DIFFSTAT;
3737        } else if (!strcmp(arg, "--name-only"))
3738                options->output_format |= DIFF_FORMAT_NAME;
3739        else if (!strcmp(arg, "--name-status"))
3740                options->output_format |= DIFF_FORMAT_NAME_STATUS;
3741        else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3742                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3743        else if (starts_with(arg, "--stat"))
3744                /* --stat, --stat-width, --stat-name-width, or --stat-count */
3745                return stat_opt(options, av);
3746
3747        /* renames options */
3748        else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3749                 !strcmp(arg, "--break-rewrites")) {
3750                if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3751                        return error("invalid argument to -B: %s", arg+2);
3752        }
3753        else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3754                 !strcmp(arg, "--find-renames")) {
3755                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3756                        return error("invalid argument to -M: %s", arg+2);
3757                options->detect_rename = DIFF_DETECT_RENAME;
3758        }
3759        else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3760                options->irreversible_delete = 1;
3761        }
3762        else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3763                 !strcmp(arg, "--find-copies")) {
3764                if (options->detect_rename == DIFF_DETECT_COPY)
3765                        DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3766                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3767                        return error("invalid argument to -C: %s", arg+2);
3768                options->detect_rename = DIFF_DETECT_COPY;
3769        }
3770        else if (!strcmp(arg, "--no-renames"))
3771                options->detect_rename = 0;
3772        else if (!strcmp(arg, "--rename-empty"))
3773                DIFF_OPT_SET(options, RENAME_EMPTY);
3774        else if (!strcmp(arg, "--no-rename-empty"))
3775                DIFF_OPT_CLR(options, RENAME_EMPTY);
3776        else if (!strcmp(arg, "--relative"))
3777                DIFF_OPT_SET(options, RELATIVE_NAME);
3778        else if (skip_prefix(arg, "--relative=", &arg)) {
3779                DIFF_OPT_SET(options, RELATIVE_NAME);
3780                options->prefix = arg;
3781        }
3782
3783        /* xdiff options */
3784        else if (!strcmp(arg, "--minimal"))
3785                DIFF_XDL_SET(options, NEED_MINIMAL);
3786        else if (!strcmp(arg, "--no-minimal"))
3787                DIFF_XDL_CLR(options, NEED_MINIMAL);
3788        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3789                DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3790        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3791                DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3792        else if (!strcmp(arg, "--ignore-space-at-eol"))
3793                DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3794        else if (!strcmp(arg, "--ignore-blank-lines"))
3795                DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3796        else if (!strcmp(arg, "--patience"))
3797                options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3798        else if (!strcmp(arg, "--histogram"))
3799                options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3800        else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3801                long value = parse_algorithm_value(optarg);
3802                if (value < 0)
3803                        return error("option diff-algorithm accepts \"myers\", "
3804                                     "\"minimal\", \"patience\" and \"histogram\"");
3805                /* clear out previous settings */
3806                DIFF_XDL_CLR(options, NEED_MINIMAL);
3807                options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3808                options->xdl_opts |= value;
3809                return argcount;
3810        }
3811
3812        /* flags options */
3813        else if (!strcmp(arg, "--binary")) {
3814                enable_patch_output(&options->output_format);
3815                DIFF_OPT_SET(options, BINARY);
3816        }
3817        else if (!strcmp(arg, "--full-index"))
3818                DIFF_OPT_SET(options, FULL_INDEX);
3819        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3820                DIFF_OPT_SET(options, TEXT);
3821        else if (!strcmp(arg, "-R"))
3822                DIFF_OPT_SET(options, REVERSE_DIFF);
3823        else if (!strcmp(arg, "--find-copies-harder"))
3824                DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3825        else if (!strcmp(arg, "--follow"))
3826                DIFF_OPT_SET(options, FOLLOW_RENAMES);
3827        else if (!strcmp(arg, "--no-follow")) {
3828                DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3829                DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3830        } else if (!strcmp(arg, "--color"))
3831                options->use_color = 1;
3832        else if (skip_prefix(arg, "--color=", &arg)) {
3833                int value = git_config_colorbool(NULL, arg);
3834                if (value < 0)
3835                        return error("option `color' expects \"always\", \"auto\", or \"never\"");
3836                options->use_color = value;
3837        }
3838        else if (!strcmp(arg, "--no-color"))
3839                options->use_color = 0;
3840        else if (!strcmp(arg, "--color-words")) {
3841                options->use_color = 1;
3842                options->word_diff = DIFF_WORDS_COLOR;
3843        }
3844        else if (skip_prefix(arg, "--color-words=", &arg)) {
3845                options->use_color = 1;
3846                options->word_diff = DIFF_WORDS_COLOR;
3847                options->word_regex = arg;
3848        }
3849        else if (!strcmp(arg, "--word-diff")) {
3850                if (options->word_diff == DIFF_WORDS_NONE)
3851                        options->word_diff = DIFF_WORDS_PLAIN;
3852        }
3853        else if (skip_prefix(arg, "--word-diff=", &arg)) {
3854                if (!strcmp(arg, "plain"))
3855                        options->word_diff = DIFF_WORDS_PLAIN;
3856                else if (!strcmp(arg, "color")) {
3857                        options->use_color = 1;
3858                        options->word_diff = DIFF_WORDS_COLOR;
3859                }
3860                else if (!strcmp(arg, "porcelain"))
3861                        options->word_diff = DIFF_WORDS_PORCELAIN;
3862                else if (!strcmp(arg, "none"))
3863                        options->word_diff = DIFF_WORDS_NONE;
3864                else
3865                        die("bad --word-diff argument: %s", arg);
3866        }
3867        else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3868                if (options->word_diff == DIFF_WORDS_NONE)
3869                        options->word_diff = DIFF_WORDS_PLAIN;
3870                options->word_regex = optarg;
3871                return argcount;
3872        }
3873        else if (!strcmp(arg, "--exit-code"))
3874                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3875        else if (!strcmp(arg, "--quiet"))
3876                DIFF_OPT_SET(options, QUICK);
3877        else if (!strcmp(arg, "--ext-diff"))
3878                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3879        else if (!strcmp(arg, "--no-ext-diff"))
3880                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3881        else if (!strcmp(arg, "--textconv"))
3882                DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3883        else if (!strcmp(arg, "--no-textconv"))
3884                DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3885        else if (!strcmp(arg, "--ignore-submodules")) {
3886                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3887                handle_ignore_submodules_arg(options, "all");
3888        } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3889                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3890                handle_ignore_submodules_arg(options, arg);
3891        } else if (!strcmp(arg, "--submodule"))
3892                DIFF_OPT_SET(options, SUBMODULE_LOG);
3893        else if (skip_prefix(arg, "--submodule=", &arg))
3894                return parse_submodule_opt(options, arg);
3895        else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3896                return parse_ws_error_highlight(options, arg);
3897
3898        /* misc options */
3899        else if (!strcmp(arg, "-z"))
3900                options->line_termination = 0;
3901        else if ((argcount = short_opt('l', av, &optarg))) {
3902                options->rename_limit = strtoul(optarg, NULL, 10);
3903                return argcount;
3904        }
3905        else if ((argcount = short_opt('S', av, &optarg))) {
3906                options->pickaxe = optarg;
3907                options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3908                return argcount;
3909        } else if ((argcount = short_opt('G', av, &optarg))) {
3910                options->pickaxe = optarg;
3911                options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3912                return argcount;
3913        }
3914        else if (!strcmp(arg, "--pickaxe-all"))
3915                options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3916        else if (!strcmp(arg, "--pickaxe-regex"))
3917                options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3918        else if ((argcount = short_opt('O', av, &optarg))) {
3919                const char *path = prefix_filename(prefix, strlen(prefix), optarg);
3920                options->orderfile = xstrdup(path);
3921                return argcount;
3922        }
3923        else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3924                int offending = parse_diff_filter_opt(optarg, options);
3925                if (offending)
3926                        die("unknown change class '%c' in --diff-filter=%s",
3927                            offending, optarg);
3928                return argcount;
3929        }
3930        else if (!strcmp(arg, "--abbrev"))
3931                options->abbrev = DEFAULT_ABBREV;
3932        else if (skip_prefix(arg, "--abbrev=", &arg)) {
3933                options->abbrev = strtoul(arg, NULL, 10);
3934                if (options->abbrev < MINIMUM_ABBREV)
3935                        options->abbrev = MINIMUM_ABBREV;
3936                else if (40 < options->abbrev)
3937                        options->abbrev = 40;
3938        }
3939        else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3940                options->a_prefix = optarg;
3941                return argcount;
3942        }
3943        else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3944                options->b_prefix = optarg;
3945                return argcount;
3946        }
3947        else if (!strcmp(arg, "--no-prefix"))
3948                options->a_prefix = options->b_prefix = "";
3949        else if (opt_arg(arg, '\0', "inter-hunk-context",
3950                         &options->interhunkcontext))
3951                ;
3952        else if (!strcmp(arg, "-W"))
3953                DIFF_OPT_SET(options, FUNCCONTEXT);
3954        else if (!strcmp(arg, "--function-context"))
3955                DIFF_OPT_SET(options, FUNCCONTEXT);
3956        else if (!strcmp(arg, "--no-function-context"))
3957                DIFF_OPT_CLR(options, FUNCCONTEXT);
3958        else if ((argcount = parse_long_opt("output", av, &optarg))) {
3959                const char *path = prefix_filename(prefix, strlen(prefix), optarg);
3960                options->file = fopen(path, "w");
3961                if (!options->file)
3962                        die_errno("Could not open '%s'", path);
3963                options->close_file = 1;
3964                return argcount;
3965        } else
3966                return 0;
3967        return 1;
3968}
3969
3970int parse_rename_score(const char **cp_p)
3971{
3972        unsigned long num, scale;
3973        int ch, dot;
3974        const char *cp = *cp_p;
3975
3976        num = 0;
3977        scale = 1;
3978        dot = 0;
3979        for (;;) {
3980                ch = *cp;
3981                if ( !dot && ch == '.' ) {
3982                        scale = 1;
3983                        dot = 1;
3984                } else if ( ch == '%' ) {
3985                        scale = dot ? scale*100 : 100;
3986                        cp++;   /* % is always at the end */
3987                        break;
3988                } else if ( ch >= '0' && ch <= '9' ) {
3989                        if ( scale < 100000 ) {
3990                                scale *= 10;
3991                                num = (num*10) + (ch-'0');
3992                        }
3993                } else {
3994                        break;
3995                }
3996                cp++;
3997        }
3998        *cp_p = cp;
3999
4000        /* user says num divided by scale and we say internally that
4001         * is MAX_SCORE * num / scale.
4002         */
4003        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4004}
4005
4006static int diff_scoreopt_parse(const char *opt)
4007{
4008        int opt1, opt2, cmd;
4009
4010        if (*opt++ != '-')
4011                return -1;
4012        cmd = *opt++;
4013        if (cmd == '-') {
4014                /* convert the long-form arguments into short-form versions */
4015                if (skip_prefix(opt, "break-rewrites", &opt)) {
4016                        if (*opt == 0 || *opt++ == '=')
4017                                cmd = 'B';
4018                } else if (skip_prefix(opt, "find-copies", &opt)) {
4019                        if (*opt == 0 || *opt++ == '=')
4020                                cmd = 'C';
4021                } else if (skip_prefix(opt, "find-renames", &opt)) {
4022                        if (*opt == 0 || *opt++ == '=')
4023                                cmd = 'M';
4024                }
4025        }
4026        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4027                return -1; /* that is not a -M, -C, or -B option */
4028
4029        opt1 = parse_rename_score(&opt);
4030        if (cmd != 'B')
4031                opt2 = 0;
4032        else {
4033                if (*opt == 0)
4034                        opt2 = 0;
4035                else if (*opt != '/')
4036                        return -1; /* we expect -B80/99 or -B80 */
4037                else {
4038                        opt++;
4039                        opt2 = parse_rename_score(&opt);
4040                }
4041        }
4042        if (*opt != 0)
4043                return -1;
4044        return opt1 | (opt2 << 16);
4045}
4046
4047struct diff_queue_struct diff_queued_diff;
4048
4049void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4050{
4051        ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4052        queue->queue[queue->nr++] = dp;
4053}
4054
4055struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4056                                 struct diff_filespec *one,
4057                                 struct diff_filespec *two)
4058{
4059        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4060        dp->one = one;
4061        dp->two = two;
4062        if (queue)
4063                diff_q(queue, dp);
4064        return dp;
4065}
4066
4067void diff_free_filepair(struct diff_filepair *p)
4068{
4069        free_filespec(p->one);
4070        free_filespec(p->two);
4071        free(p);
4072}
4073
4074/* This is different from find_unique_abbrev() in that
4075 * it stuffs the result with dots for alignment.
4076 */
4077const char *diff_unique_abbrev(const unsigned char *sha1, int len)
4078{
4079        int abblen;
4080        const char *abbrev;
4081        if (len == 40)
4082                return sha1_to_hex(sha1);
4083
4084        abbrev = find_unique_abbrev(sha1, len);
4085        abblen = strlen(abbrev);
4086        if (abblen < 37) {
4087                static char hex[41];
4088                if (len < abblen && abblen <= len + 2)
4089                        xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4090                else
4091                        xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4092                return hex;
4093        }
4094        return sha1_to_hex(sha1);
4095}
4096
4097static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4098{
4099        int line_termination = opt->line_termination;
4100        int inter_name_termination = line_termination ? '\t' : '\0';
4101
4102        fprintf(opt->file, "%s", diff_line_prefix(opt));
4103        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4104                fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4105                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
4106                fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4107        }
4108        if (p->score) {
4109                fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4110                        inter_name_termination);
4111        } else {
4112                fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4113        }
4114
4115        if (p->status == DIFF_STATUS_COPIED ||
4116            p->status == DIFF_STATUS_RENAMED) {
4117                const char *name_a, *name_b;
4118                name_a = p->one->path;
4119                name_b = p->two->path;
4120                strip_prefix(opt->prefix_length, &name_a, &name_b);
4121                write_name_quoted(name_a, opt->file, inter_name_termination);
4122                write_name_quoted(name_b, opt->file, line_termination);
4123        } else {
4124                const char *name_a, *name_b;
4125                name_a = p->one->mode ? p->one->path : p->two->path;
4126                name_b = NULL;
4127                strip_prefix(opt->prefix_length, &name_a, &name_b);
4128                write_name_quoted(name_a, opt->file, line_termination);
4129        }
4130}
4131
4132int diff_unmodified_pair(struct diff_filepair *p)
4133{
4134        /* This function is written stricter than necessary to support
4135         * the currently implemented transformers, but the idea is to
4136         * let transformers to produce diff_filepairs any way they want,
4137         * and filter and clean them up here before producing the output.
4138         */
4139        struct diff_filespec *one = p->one, *two = p->two;
4140
4141        if (DIFF_PAIR_UNMERGED(p))
4142                return 0; /* unmerged is interesting */
4143
4144        /* deletion, addition, mode or type change
4145         * and rename are all interesting.
4146         */
4147        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4148            DIFF_PAIR_MODE_CHANGED(p) ||
4149            strcmp(one->path, two->path))
4150                return 0;
4151
4152        /* both are valid and point at the same path.  that is, we are
4153         * dealing with a change.
4154         */
4155        if (one->sha1_valid && two->sha1_valid &&
4156            !hashcmp(one->sha1, two->sha1) &&
4157            !one->dirty_submodule && !two->dirty_submodule)
4158                return 1; /* no change */
4159        if (!one->sha1_valid && !two->sha1_valid)
4160                return 1; /* both look at the same file on the filesystem. */
4161        return 0;
4162}
4163
4164static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4165{
4166        if (diff_unmodified_pair(p))
4167                return;
4168
4169        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4170            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4171                return; /* no tree diffs in patch format */
4172
4173        run_diff(p, o);
4174}
4175
4176static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4177                            struct diffstat_t *diffstat)
4178{
4179        if (diff_unmodified_pair(p))
4180                return;
4181
4182        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4183            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4184                return; /* no useful stat for tree diffs */
4185
4186        run_diffstat(p, o, diffstat);
4187}
4188
4189static void diff_flush_checkdiff(struct diff_filepair *p,
4190                struct diff_options *o)
4191{
4192        if (diff_unmodified_pair(p))
4193                return;
4194
4195        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4196            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4197                return; /* nothing to check in tree diffs */
4198
4199        run_checkdiff(p, o);
4200}
4201
4202int diff_queue_is_empty(void)
4203{
4204        struct diff_queue_struct *q = &diff_queued_diff;
4205        int i;
4206        for (i = 0; i < q->nr; i++)
4207                if (!diff_unmodified_pair(q->queue[i]))
4208                        return 0;
4209        return 1;
4210}
4211
4212#if DIFF_DEBUG
4213void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4214{
4215        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4216                x, one ? one : "",
4217                s->path,
4218                DIFF_FILE_VALID(s) ? "valid" : "invalid",
4219                s->mode,
4220                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4221        fprintf(stderr, "queue[%d] %s size %lu\n",
4222                x, one ? one : "",
4223                s->size);
4224}
4225
4226void diff_debug_filepair(const struct diff_filepair *p, int i)
4227{
4228        diff_debug_filespec(p->one, i, "one");
4229        diff_debug_filespec(p->two, i, "two");
4230        fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4231                p->score, p->status ? p->status : '?',
4232                p->one->rename_used, p->broken_pair);
4233}
4234
4235void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4236{
4237        int i;
4238        if (msg)
4239                fprintf(stderr, "%s\n", msg);
4240        fprintf(stderr, "q->nr = %d\n", q->nr);
4241        for (i = 0; i < q->nr; i++) {
4242                struct diff_filepair *p = q->queue[i];
4243                diff_debug_filepair(p, i);
4244        }
4245}
4246#endif
4247
4248static void diff_resolve_rename_copy(void)
4249{
4250        int i;
4251        struct diff_filepair *p;
4252        struct diff_queue_struct *q = &diff_queued_diff;
4253
4254        diff_debug_queue("resolve-rename-copy", q);
4255
4256        for (i = 0; i < q->nr; i++) {
4257                p = q->queue[i];
4258                p->status = 0; /* undecided */
4259                if (DIFF_PAIR_UNMERGED(p))
4260                        p->status = DIFF_STATUS_UNMERGED;
4261                else if (!DIFF_FILE_VALID(p->one))
4262                        p->status = DIFF_STATUS_ADDED;
4263                else if (!DIFF_FILE_VALID(p->two))
4264                        p->status = DIFF_STATUS_DELETED;
4265                else if (DIFF_PAIR_TYPE_CHANGED(p))
4266                        p->status = DIFF_STATUS_TYPE_CHANGED;
4267
4268                /* from this point on, we are dealing with a pair
4269                 * whose both sides are valid and of the same type, i.e.
4270                 * either in-place edit or rename/copy edit.
4271                 */
4272                else if (DIFF_PAIR_RENAME(p)) {
4273                        /*
4274                         * A rename might have re-connected a broken
4275                         * pair up, causing the pathnames to be the
4276                         * same again. If so, that's not a rename at
4277                         * all, just a modification..
4278                         *
4279                         * Otherwise, see if this source was used for
4280                         * multiple renames, in which case we decrement
4281                         * the count, and call it a copy.
4282                         */
4283                        if (!strcmp(p->one->path, p->two->path))
4284                                p->status = DIFF_STATUS_MODIFIED;
4285                        else if (--p->one->rename_used > 0)
4286                                p->status = DIFF_STATUS_COPIED;
4287                        else
4288                                p->status = DIFF_STATUS_RENAMED;
4289                }
4290                else if (hashcmp(p->one->sha1, p->two->sha1) ||
4291                         p->one->mode != p->two->mode ||
4292                         p->one->dirty_submodule ||
4293                         p->two->dirty_submodule ||
4294                         is_null_sha1(p->one->sha1))
4295                        p->status = DIFF_STATUS_MODIFIED;
4296                else {
4297                        /* This is a "no-change" entry and should not
4298                         * happen anymore, but prepare for broken callers.
4299                         */
4300                        error("feeding unmodified %s to diffcore",
4301                              p->one->path);
4302                        p->status = DIFF_STATUS_UNKNOWN;
4303                }
4304        }
4305        diff_debug_queue("resolve-rename-copy done", q);
4306}
4307
4308static int check_pair_status(struct diff_filepair *p)
4309{
4310        switch (p->status) {
4311        case DIFF_STATUS_UNKNOWN:
4312                return 0;
4313        case 0:
4314                die("internal error in diff-resolve-rename-copy");
4315        default:
4316                return 1;
4317        }
4318}
4319
4320static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4321{
4322        int fmt = opt->output_format;
4323
4324        if (fmt & DIFF_FORMAT_CHECKDIFF)
4325                diff_flush_checkdiff(p, opt);
4326        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4327                diff_flush_raw(p, opt);
4328        else if (fmt & DIFF_FORMAT_NAME) {
4329                const char *name_a, *name_b;
4330                name_a = p->two->path;
4331                name_b = NULL;
4332                strip_prefix(opt->prefix_length, &name_a, &name_b);
4333                write_name_quoted(name_a, opt->file, opt->line_termination);
4334        }
4335}
4336
4337static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4338{
4339        if (fs->mode)
4340                fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4341        else
4342                fprintf(file, " %s ", newdelete);
4343        write_name_quoted(fs->path, file, '\n');
4344}
4345
4346
4347static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4348                const char *line_prefix)
4349{
4350        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4351                fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4352                        p->two->mode, show_name ? ' ' : '\n');
4353                if (show_name) {
4354                        write_name_quoted(p->two->path, file, '\n');
4355                }
4356        }
4357}
4358
4359static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4360                        const char *line_prefix)
4361{
4362        char *names = pprint_rename(p->one->path, p->two->path);
4363
4364        fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4365        free(names);
4366        show_mode_change(file, p, 0, line_prefix);
4367}
4368
4369static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4370{
4371        FILE *file = opt->file;
4372        const char *line_prefix = diff_line_prefix(opt);
4373
4374        switch(p->status) {
4375        case DIFF_STATUS_DELETED:
4376                fputs(line_prefix, file);
4377                show_file_mode_name(file, "delete", p->one);
4378                break;
4379        case DIFF_STATUS_ADDED:
4380                fputs(line_prefix, file);
4381                show_file_mode_name(file, "create", p->two);
4382                break;
4383        case DIFF_STATUS_COPIED:
4384                fputs(line_prefix, file);
4385                show_rename_copy(file, "copy", p, line_prefix);
4386                break;
4387        case DIFF_STATUS_RENAMED:
4388                fputs(line_prefix, file);
4389                show_rename_copy(file, "rename", p, line_prefix);
4390                break;
4391        default:
4392                if (p->score) {
4393                        fprintf(file, "%s rewrite ", line_prefix);
4394                        write_name_quoted(p->two->path, file, ' ');
4395                        fprintf(file, "(%d%%)\n", similarity_index(p));
4396                }
4397                show_mode_change(file, p, !p->score, line_prefix);
4398                break;
4399        }
4400}
4401
4402struct patch_id_t {
4403        git_SHA_CTX *ctx;
4404        int patchlen;
4405};
4406
4407static int remove_space(char *line, int len)
4408{
4409        int i;
4410        char *dst = line;
4411        unsigned char c;
4412
4413        for (i = 0; i < len; i++)
4414                if (!isspace((c = line[i])))
4415                        *dst++ = c;
4416
4417        return dst - line;
4418}
4419
4420static void patch_id_consume(void *priv, char *line, unsigned long len)
4421{
4422        struct patch_id_t *data = priv;
4423        int new_len;
4424
4425        /* Ignore line numbers when computing the SHA1 of the patch */
4426        if (starts_with(line, "@@ -"))
4427                return;
4428
4429        new_len = remove_space(line, len);
4430
4431        git_SHA1_Update(data->ctx, line, new_len);
4432        data->patchlen += new_len;
4433}
4434
4435/* returns 0 upon success, and writes result into sha1 */
4436static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4437{
4438        struct diff_queue_struct *q = &diff_queued_diff;
4439        int i;
4440        git_SHA_CTX ctx;
4441        struct patch_id_t data;
4442        char buffer[PATH_MAX * 4 + 20];
4443
4444        git_SHA1_Init(&ctx);
4445        memset(&data, 0, sizeof(struct patch_id_t));
4446        data.ctx = &ctx;
4447
4448        for (i = 0; i < q->nr; i++) {
4449                xpparam_t xpp;
4450                xdemitconf_t xecfg;
4451                mmfile_t mf1, mf2;
4452                struct diff_filepair *p = q->queue[i];
4453                int len1, len2;
4454
4455                memset(&xpp, 0, sizeof(xpp));
4456                memset(&xecfg, 0, sizeof(xecfg));
4457                if (p->status == 0)
4458                        return error("internal diff status error");
4459                if (p->status == DIFF_STATUS_UNKNOWN)
4460                        continue;
4461                if (diff_unmodified_pair(p))
4462                        continue;
4463                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4464                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4465                        continue;
4466                if (DIFF_PAIR_UNMERGED(p))
4467                        continue;
4468
4469                diff_fill_sha1_info(p->one);
4470                diff_fill_sha1_info(p->two);
4471                if (fill_mmfile(&mf1, p->one) < 0 ||
4472                                fill_mmfile(&mf2, p->two) < 0)
4473                        return error("unable to read files to diff");
4474
4475                len1 = remove_space(p->one->path, strlen(p->one->path));
4476                len2 = remove_space(p->two->path, strlen(p->two->path));
4477                if (p->one->mode == 0)
4478                        len1 = snprintf(buffer, sizeof(buffer),
4479                                        "diff--gita/%.*sb/%.*s"
4480                                        "newfilemode%06o"
4481                                        "---/dev/null"
4482                                        "+++b/%.*s",
4483                                        len1, p->one->path,
4484                                        len2, p->two->path,
4485                                        p->two->mode,
4486                                        len2, p->two->path);
4487                else if (p->two->mode == 0)
4488                        len1 = snprintf(buffer, sizeof(buffer),
4489                                        "diff--gita/%.*sb/%.*s"
4490                                        "deletedfilemode%06o"
4491                                        "---a/%.*s"
4492                                        "+++/dev/null",
4493                                        len1, p->one->path,
4494                                        len2, p->two->path,
4495                                        p->one->mode,
4496                                        len1, p->one->path);
4497                else
4498                        len1 = snprintf(buffer, sizeof(buffer),
4499                                        "diff--gita/%.*sb/%.*s"
4500                                        "---a/%.*s"
4501                                        "+++b/%.*s",
4502                                        len1, p->one->path,
4503                                        len2, p->two->path,
4504                                        len1, p->one->path,
4505                                        len2, p->two->path);
4506                git_SHA1_Update(&ctx, buffer, len1);
4507
4508                if (diff_filespec_is_binary(p->one) ||
4509                    diff_filespec_is_binary(p->two)) {
4510                        git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4511                        git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4512                        continue;
4513                }
4514
4515                xpp.flags = 0;
4516                xecfg.ctxlen = 3;
4517                xecfg.flags = 0;
4518                if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4519                                  &xpp, &xecfg))
4520                        return error("unable to generate patch-id diff for %s",
4521                                     p->one->path);
4522        }
4523
4524        git_SHA1_Final(sha1, &ctx);
4525        return 0;
4526}
4527
4528int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4529{
4530        struct diff_queue_struct *q = &diff_queued_diff;
4531        int i;
4532        int result = diff_get_patch_id(options, sha1);
4533
4534        for (i = 0; i < q->nr; i++)
4535                diff_free_filepair(q->queue[i]);
4536
4537        free(q->queue);
4538        DIFF_QUEUE_CLEAR(q);
4539
4540        return result;
4541}
4542
4543static int is_summary_empty(const struct diff_queue_struct *q)
4544{
4545        int i;
4546
4547        for (i = 0; i < q->nr; i++) {
4548                const struct diff_filepair *p = q->queue[i];
4549
4550                switch (p->status) {
4551                case DIFF_STATUS_DELETED:
4552                case DIFF_STATUS_ADDED:
4553                case DIFF_STATUS_COPIED:
4554                case DIFF_STATUS_RENAMED:
4555                        return 0;
4556                default:
4557                        if (p->score)
4558                                return 0;
4559                        if (p->one->mode && p->two->mode &&
4560                            p->one->mode != p->two->mode)
4561                                return 0;
4562                        break;
4563                }
4564        }
4565        return 1;
4566}
4567
4568static const char rename_limit_warning[] =
4569"inexact rename detection was skipped due to too many files.";
4570
4571static const char degrade_cc_to_c_warning[] =
4572"only found copies from modified paths due to too many files.";
4573
4574static const char rename_limit_advice[] =
4575"you may want to set your %s variable to at least "
4576"%d and retry the command.";
4577
4578void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4579{
4580        if (degraded_cc)
4581                warning(degrade_cc_to_c_warning);
4582        else if (needed)
4583                warning(rename_limit_warning);
4584        else
4585                return;
4586        if (0 < needed && needed < 32767)
4587                warning(rename_limit_advice, varname, needed);
4588}
4589
4590void diff_flush(struct diff_options *options)
4591{
4592        struct diff_queue_struct *q = &diff_queued_diff;
4593        int i, output_format = options->output_format;
4594        int separator = 0;
4595        int dirstat_by_line = 0;
4596
4597        /*
4598         * Order: raw, stat, summary, patch
4599         * or:    name/name-status/checkdiff (other bits clear)
4600         */
4601        if (!q->nr)
4602                goto free_queue;
4603
4604        if (output_format & (DIFF_FORMAT_RAW |
4605                             DIFF_FORMAT_NAME |
4606                             DIFF_FORMAT_NAME_STATUS |
4607                             DIFF_FORMAT_CHECKDIFF)) {
4608                for (i = 0; i < q->nr; i++) {
4609                        struct diff_filepair *p = q->queue[i];
4610                        if (check_pair_status(p))
4611                                flush_one_pair(p, options);
4612                }
4613                separator++;
4614        }
4615
4616        if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4617                dirstat_by_line = 1;
4618
4619        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4620            dirstat_by_line) {
4621                struct diffstat_t diffstat;
4622
4623                memset(&diffstat, 0, sizeof(struct diffstat_t));
4624                for (i = 0; i < q->nr; i++) {
4625                        struct diff_filepair *p = q->queue[i];
4626                        if (check_pair_status(p))
4627                                diff_flush_stat(p, options, &diffstat);
4628                }
4629                if (output_format & DIFF_FORMAT_NUMSTAT)
4630                        show_numstat(&diffstat, options);
4631                if (output_format & DIFF_FORMAT_DIFFSTAT)
4632                        show_stats(&diffstat, options);
4633                if (output_format & DIFF_FORMAT_SHORTSTAT)
4634                        show_shortstats(&diffstat, options);
4635                if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4636                        show_dirstat_by_line(&diffstat, options);
4637                free_diffstat_info(&diffstat);
4638                separator++;
4639        }
4640        if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4641                show_dirstat(options);
4642
4643        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4644                for (i = 0; i < q->nr; i++) {
4645                        diff_summary(options, q->queue[i]);
4646                }
4647                separator++;
4648        }
4649
4650        if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4651            DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4652            DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4653                /*
4654                 * run diff_flush_patch for the exit status. setting
4655                 * options->file to /dev/null should be safe, because we
4656                 * aren't supposed to produce any output anyway.
4657                 */
4658                if (options->close_file)
4659                        fclose(options->file);
4660                options->file = fopen("/dev/null", "w");
4661                if (!options->file)
4662                        die_errno("Could not open /dev/null");
4663                options->close_file = 1;
4664                for (i = 0; i < q->nr; i++) {
4665                        struct diff_filepair *p = q->queue[i];
4666                        if (check_pair_status(p))
4667                                diff_flush_patch(p, options);
4668                        if (options->found_changes)
4669                                break;
4670                }
4671        }
4672
4673        if (output_format & DIFF_FORMAT_PATCH) {
4674                if (separator) {
4675                        fprintf(options->file, "%s%c",
4676                                diff_line_prefix(options),
4677                                options->line_termination);
4678                        if (options->stat_sep) {
4679                                /* attach patch instead of inline */
4680                                fputs(options->stat_sep, options->file);
4681                        }
4682                }
4683
4684                for (i = 0; i < q->nr; i++) {
4685                        struct diff_filepair *p = q->queue[i];
4686                        if (check_pair_status(p))
4687                                diff_flush_patch(p, options);
4688                }
4689        }
4690
4691        if (output_format & DIFF_FORMAT_CALLBACK)
4692                options->format_callback(q, options, options->format_callback_data);
4693
4694        for (i = 0; i < q->nr; i++)
4695                diff_free_filepair(q->queue[i]);
4696free_queue:
4697        free(q->queue);
4698        DIFF_QUEUE_CLEAR(q);
4699        if (options->close_file)
4700                fclose(options->file);
4701
4702        /*
4703         * Report the content-level differences with HAS_CHANGES;
4704         * diff_addremove/diff_change does not set the bit when
4705         * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4706         */
4707        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4708                if (options->found_changes)
4709                        DIFF_OPT_SET(options, HAS_CHANGES);
4710                else
4711                        DIFF_OPT_CLR(options, HAS_CHANGES);
4712        }
4713}
4714
4715static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4716{
4717        return (((p->status == DIFF_STATUS_MODIFIED) &&
4718                 ((p->score &&
4719                   filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4720                  (!p->score &&
4721                   filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4722                ((p->status != DIFF_STATUS_MODIFIED) &&
4723                 filter_bit_tst(p->status, options)));
4724}
4725
4726static void diffcore_apply_filter(struct diff_options *options)
4727{
4728        int i;
4729        struct diff_queue_struct *q = &diff_queued_diff;
4730        struct diff_queue_struct outq;
4731
4732        DIFF_QUEUE_CLEAR(&outq);
4733
4734        if (!options->filter)
4735                return;
4736
4737        if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4738                int found;
4739                for (i = found = 0; !found && i < q->nr; i++) {
4740                        if (match_filter(options, q->queue[i]))
4741                                found++;
4742                }
4743                if (found)
4744                        return;
4745
4746                /* otherwise we will clear the whole queue
4747                 * by copying the empty outq at the end of this
4748                 * function, but first clear the current entries
4749                 * in the queue.
4750                 */
4751                for (i = 0; i < q->nr; i++)
4752                        diff_free_filepair(q->queue[i]);
4753        }
4754        else {
4755                /* Only the matching ones */
4756                for (i = 0; i < q->nr; i++) {
4757                        struct diff_filepair *p = q->queue[i];
4758                        if (match_filter(options, p))
4759                                diff_q(&outq, p);
4760                        else
4761                                diff_free_filepair(p);
4762                }
4763        }
4764        free(q->queue);
4765        *q = outq;
4766}
4767
4768/* Check whether two filespecs with the same mode and size are identical */
4769static int diff_filespec_is_identical(struct diff_filespec *one,
4770                                      struct diff_filespec *two)
4771{
4772        if (S_ISGITLINK(one->mode))
4773                return 0;
4774        if (diff_populate_filespec(one, 0))
4775                return 0;
4776        if (diff_populate_filespec(two, 0))
4777                return 0;
4778        return !memcmp(one->data, two->data, one->size);
4779}
4780
4781static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4782{
4783        if (p->done_skip_stat_unmatch)
4784                return p->skip_stat_unmatch_result;
4785
4786        p->done_skip_stat_unmatch = 1;
4787        p->skip_stat_unmatch_result = 0;
4788        /*
4789         * 1. Entries that come from stat info dirtiness
4790         *    always have both sides (iow, not create/delete),
4791         *    one side of the object name is unknown, with
4792         *    the same mode and size.  Keep the ones that
4793         *    do not match these criteria.  They have real
4794         *    differences.
4795         *
4796         * 2. At this point, the file is known to be modified,
4797         *    with the same mode and size, and the object
4798         *    name of one side is unknown.  Need to inspect
4799         *    the identical contents.
4800         */
4801        if (!DIFF_FILE_VALID(p->one) || /* (1) */
4802            !DIFF_FILE_VALID(p->two) ||
4803            (p->one->sha1_valid && p->two->sha1_valid) ||
4804            (p->one->mode != p->two->mode) ||
4805            diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4806            diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4807            (p->one->size != p->two->size) ||
4808            !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4809                p->skip_stat_unmatch_result = 1;
4810        return p->skip_stat_unmatch_result;
4811}
4812
4813static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4814{
4815        int i;
4816        struct diff_queue_struct *q = &diff_queued_diff;
4817        struct diff_queue_struct outq;
4818        DIFF_QUEUE_CLEAR(&outq);
4819
4820        for (i = 0; i < q->nr; i++) {
4821                struct diff_filepair *p = q->queue[i];
4822
4823                if (diff_filespec_check_stat_unmatch(p))
4824                        diff_q(&outq, p);
4825                else {
4826                        /*
4827                         * The caller can subtract 1 from skip_stat_unmatch
4828                         * to determine how many paths were dirty only
4829                         * due to stat info mismatch.
4830                         */
4831                        if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4832                                diffopt->skip_stat_unmatch++;
4833                        diff_free_filepair(p);
4834                }
4835        }
4836        free(q->queue);
4837        *q = outq;
4838}
4839
4840static int diffnamecmp(const void *a_, const void *b_)
4841{
4842        const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4843        const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4844        const char *name_a, *name_b;
4845
4846        name_a = a->one ? a->one->path : a->two->path;
4847        name_b = b->one ? b->one->path : b->two->path;
4848        return strcmp(name_a, name_b);
4849}
4850
4851void diffcore_fix_diff_index(struct diff_options *options)
4852{
4853        struct diff_queue_struct *q = &diff_queued_diff;
4854        qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4855}
4856
4857void diffcore_std(struct diff_options *options)
4858{
4859        /* NOTE please keep the following in sync with diff_tree_combined() */
4860        if (options->skip_stat_unmatch)
4861                diffcore_skip_stat_unmatch(options);
4862        if (!options->found_follow) {
4863                /* See try_to_follow_renames() in tree-diff.c */
4864                if (options->break_opt != -1)
4865                        diffcore_break(options->break_opt);
4866                if (options->detect_rename)
4867                        diffcore_rename(options);
4868                if (options->break_opt != -1)
4869                        diffcore_merge_broken();
4870        }
4871        if (options->pickaxe)
4872                diffcore_pickaxe(options);
4873        if (options->orderfile)
4874                diffcore_order(options->orderfile);
4875        if (!options->found_follow)
4876                /* See try_to_follow_renames() in tree-diff.c */
4877                diff_resolve_rename_copy();
4878        diffcore_apply_filter(options);
4879
4880        if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4881                DIFF_OPT_SET(options, HAS_CHANGES);
4882        else
4883                DIFF_OPT_CLR(options, HAS_CHANGES);
4884
4885        options->found_follow = 0;
4886}
4887
4888int diff_result_code(struct diff_options *opt, int status)
4889{
4890        int result = 0;
4891
4892        diff_warn_rename_limit("diff.renameLimit",
4893                               opt->needed_rename_limit,
4894                               opt->degraded_cc_to_c);
4895        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4896            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4897                return status;
4898        if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4899            DIFF_OPT_TST(opt, HAS_CHANGES))
4900                result |= 01;
4901        if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4902            DIFF_OPT_TST(opt, CHECK_FAILED))
4903                result |= 02;
4904        return result;
4905}
4906
4907int diff_can_quit_early(struct diff_options *opt)
4908{
4909        return (DIFF_OPT_TST(opt, QUICK) &&
4910                !opt->filter &&
4911                DIFF_OPT_TST(opt, HAS_CHANGES));
4912}
4913
4914/*
4915 * Shall changes to this submodule be ignored?
4916 *
4917 * Submodule changes can be configured to be ignored separately for each path,
4918 * but that configuration can be overridden from the command line.
4919 */
4920static int is_submodule_ignored(const char *path, struct diff_options *options)
4921{
4922        int ignored = 0;
4923        unsigned orig_flags = options->flags;
4924        if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4925                set_diffopt_flags_from_submodule_config(options, path);
4926        if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4927                ignored = 1;
4928        options->flags = orig_flags;
4929        return ignored;
4930}
4931
4932void diff_addremove(struct diff_options *options,
4933                    int addremove, unsigned mode,
4934                    const unsigned char *sha1,
4935                    int sha1_valid,
4936                    const char *concatpath, unsigned dirty_submodule)
4937{
4938        struct diff_filespec *one, *two;
4939
4940        if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4941                return;
4942
4943        /* This may look odd, but it is a preparation for
4944         * feeding "there are unchanged files which should
4945         * not produce diffs, but when you are doing copy
4946         * detection you would need them, so here they are"
4947         * entries to the diff-core.  They will be prefixed
4948         * with something like '=' or '*' (I haven't decided
4949         * which but should not make any difference).
4950         * Feeding the same new and old to diff_change()
4951         * also has the same effect.
4952         * Before the final output happens, they are pruned after
4953         * merged into rename/copy pairs as appropriate.
4954         */
4955        if (DIFF_OPT_TST(options, REVERSE_DIFF))
4956                addremove = (addremove == '+' ? '-' :
4957                             addremove == '-' ? '+' : addremove);
4958
4959        if (options->prefix &&
4960            strncmp(concatpath, options->prefix, options->prefix_length))
4961                return;
4962
4963        one = alloc_filespec(concatpath);
4964        two = alloc_filespec(concatpath);
4965
4966        if (addremove != '+')
4967                fill_filespec(one, sha1, sha1_valid, mode);
4968        if (addremove != '-') {
4969                fill_filespec(two, sha1, sha1_valid, mode);
4970                two->dirty_submodule = dirty_submodule;
4971        }
4972
4973        diff_queue(&diff_queued_diff, one, two);
4974        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4975                DIFF_OPT_SET(options, HAS_CHANGES);
4976}
4977
4978void diff_change(struct diff_options *options,
4979                 unsigned old_mode, unsigned new_mode,
4980                 const unsigned char *old_sha1,
4981                 const unsigned char *new_sha1,
4982                 int old_sha1_valid, int new_sha1_valid,
4983                 const char *concatpath,
4984                 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4985{
4986        struct diff_filespec *one, *two;
4987        struct diff_filepair *p;
4988
4989        if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4990            is_submodule_ignored(concatpath, options))
4991                return;
4992
4993        if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4994                unsigned tmp;
4995                const unsigned char *tmp_c;
4996                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4997                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4998                tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4999                        new_sha1_valid = tmp;
5000                tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
5001                        new_dirty_submodule = tmp;
5002        }
5003
5004        if (options->prefix &&
5005            strncmp(concatpath, options->prefix, options->prefix_length))
5006                return;
5007
5008        one = alloc_filespec(concatpath);
5009        two = alloc_filespec(concatpath);
5010        fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5011        fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5012        one->dirty_submodule = old_dirty_submodule;
5013        two->dirty_submodule = new_dirty_submodule;
5014        p = diff_queue(&diff_queued_diff, one, two);
5015
5016        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5017                return;
5018
5019        if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5020            !diff_filespec_check_stat_unmatch(p))
5021                return;
5022
5023        DIFF_OPT_SET(options, HAS_CHANGES);
5024}
5025
5026struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5027{
5028        struct diff_filepair *pair;
5029        struct diff_filespec *one, *two;
5030
5031        if (options->prefix &&
5032            strncmp(path, options->prefix, options->prefix_length))
5033                return NULL;
5034
5035        one = alloc_filespec(path);
5036        two = alloc_filespec(path);
5037        pair = diff_queue(&diff_queued_diff, one, two);
5038        pair->is_unmerged = 1;
5039        return pair;
5040}
5041
5042static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5043                size_t *outsize)
5044{
5045        struct diff_tempfile *temp;
5046        const char *argv[3];
5047        const char **arg = argv;
5048        struct child_process child = CHILD_PROCESS_INIT;
5049        struct strbuf buf = STRBUF_INIT;
5050        int err = 0;
5051
5052        temp = prepare_temp_file(spec->path, spec);
5053        *arg++ = pgm;
5054        *arg++ = temp->name;
5055        *arg = NULL;
5056
5057        child.use_shell = 1;
5058        child.argv = argv;
5059        child.out = -1;
5060        if (start_command(&child)) {
5061                remove_tempfile();
5062                return NULL;
5063        }
5064
5065        if (strbuf_read(&buf, child.out, 0) < 0)
5066                err = error("error reading from textconv command '%s'", pgm);
5067        close(child.out);
5068
5069        if (finish_command(&child) || err) {
5070                strbuf_release(&buf);
5071                remove_tempfile();
5072                return NULL;
5073        }
5074        remove_tempfile();
5075
5076        return strbuf_detach(&buf, outsize);
5077}
5078
5079size_t fill_textconv(struct userdiff_driver *driver,
5080                     struct diff_filespec *df,
5081                     char **outbuf)
5082{
5083        size_t size;
5084
5085        if (!driver) {
5086                if (!DIFF_FILE_VALID(df)) {
5087                        *outbuf = "";
5088                        return 0;
5089                }
5090                if (diff_populate_filespec(df, 0))
5091                        die("unable to read files to diff");
5092                *outbuf = df->data;
5093                return df->size;
5094        }
5095
5096        if (!driver->textconv)
5097                die("BUG: fill_textconv called with non-textconv driver");
5098
5099        if (driver->textconv_cache && df->sha1_valid) {
5100                *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5101                                          &size);
5102                if (*outbuf)
5103                        return size;
5104        }
5105
5106        *outbuf = run_textconv(driver->textconv, df, &size);
5107        if (!*outbuf)
5108                die("unable to read files to diff");
5109
5110        if (driver->textconv_cache && df->sha1_valid) {
5111                /* ignore errors, as we might be in a readonly repository */
5112                notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5113                                size);
5114                /*
5115                 * we could save up changes and flush them all at the end,
5116                 * but we would need an extra call after all diffing is done.
5117                 * Since generating a cache entry is the slow path anyway,
5118                 * this extra overhead probably isn't a big deal.
5119                 */
5120                notes_cache_write(driver->textconv_cache);
5121        }
5122
5123        return size;
5124}
5125
5126void setup_diff_pager(struct diff_options *opt)
5127{
5128        /*
5129         * If the user asked for our exit code, then either they want --quiet
5130         * or --exit-code. We should definitely not bother with a pager in the
5131         * former case, as we will generate no output. Since we still properly
5132         * report our exit code even when a pager is run, we _could_ run a
5133         * pager with --exit-code. But since we have not done so historically,
5134         * and because it is easy to find people oneline advising "git diff
5135         * --exit-code" in hooks and other scripts, we do not do so.
5136         */
5137        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5138            check_pager_config("diff") != 0)
5139                setup_pager();
5140}