a360ab5063e6025b909d6f11a9b09ebbd6347e0d
   1/*
   2 * Copyright (C) 2005 Junio C Hamano
   3 */
   4#include "cache.h"
   5#include "quote.h"
   6#include "diff.h"
   7#include "diffcore.h"
   8#include "delta.h"
   9#include "xdiff-interface.h"
  10#include "color.h"
  11#include "attr.h"
  12#include "run-command.h"
  13#include "utf8.h"
  14#include "userdiff.h"
  15#include "sigchain.h"
  16#include "submodule.h"
  17#include "ll-merge.h"
  18#include "string-list.h"
  19#include "argv-array.h"
  20
  21#ifdef NO_FAST_WORKING_DIRECTORY
  22#define FAST_WORKING_DIRECTORY 0
  23#else
  24#define FAST_WORKING_DIRECTORY 1
  25#endif
  26
  27static int diff_detect_rename_default;
  28static int diff_rename_limit_default = 400;
  29static int diff_suppress_blank_empty;
  30static int diff_use_color_default = -1;
  31static int diff_context_default = 3;
  32static const char *diff_word_regex_cfg;
  33static const char *external_diff_cmd_cfg;
  34static const char *diff_order_file_cfg;
  35int diff_auto_refresh_index = 1;
  36static int diff_mnemonic_prefix;
  37static int diff_no_prefix;
  38static int diff_stat_graph_width;
  39static int diff_dirstat_permille_default = 30;
  40static struct diff_options default_diff_options;
  41static long diff_algorithm;
  42
  43static char diff_colors[][COLOR_MAXLEN] = {
  44        GIT_COLOR_RESET,
  45        GIT_COLOR_NORMAL,       /* PLAIN */
  46        GIT_COLOR_BOLD,         /* METAINFO */
  47        GIT_COLOR_CYAN,         /* FRAGINFO */
  48        GIT_COLOR_RED,          /* OLD */
  49        GIT_COLOR_GREEN,        /* NEW */
  50        GIT_COLOR_YELLOW,       /* COMMIT */
  51        GIT_COLOR_BG_RED,       /* WHITESPACE */
  52        GIT_COLOR_NORMAL,       /* FUNCINFO */
  53};
  54
  55static int parse_diff_color_slot(const char *var, int ofs)
  56{
  57        if (!strcasecmp(var+ofs, "plain"))
  58                return DIFF_PLAIN;
  59        if (!strcasecmp(var+ofs, "meta"))
  60                return DIFF_METAINFO;
  61        if (!strcasecmp(var+ofs, "frag"))
  62                return DIFF_FRAGINFO;
  63        if (!strcasecmp(var+ofs, "old"))
  64                return DIFF_FILE_OLD;
  65        if (!strcasecmp(var+ofs, "new"))
  66                return DIFF_FILE_NEW;
  67        if (!strcasecmp(var+ofs, "commit"))
  68                return DIFF_COMMIT;
  69        if (!strcasecmp(var+ofs, "whitespace"))
  70                return DIFF_WHITESPACE;
  71        if (!strcasecmp(var+ofs, "func"))
  72                return DIFF_FUNCINFO;
  73        return -1;
  74}
  75
  76static int parse_dirstat_params(struct diff_options *options, const char *params_string,
  77                                struct strbuf *errmsg)
  78{
  79        char *params_copy = xstrdup(params_string);
  80        struct string_list params = STRING_LIST_INIT_NODUP;
  81        int ret = 0;
  82        int i;
  83
  84        if (*params_copy)
  85                string_list_split_in_place(&params, params_copy, ',', -1);
  86        for (i = 0; i < params.nr; i++) {
  87                const char *p = params.items[i].string;
  88                if (!strcmp(p, "changes")) {
  89                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  90                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  91                } else if (!strcmp(p, "lines")) {
  92                        DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  93                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  94                } else if (!strcmp(p, "files")) {
  95                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  96                        DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  97                } else if (!strcmp(p, "noncumulative")) {
  98                        DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
  99                } else if (!strcmp(p, "cumulative")) {
 100                        DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 101                } else if (isdigit(*p)) {
 102                        char *end;
 103                        int permille = strtoul(p, &end, 10) * 10;
 104                        if (*end == '.' && isdigit(*++end)) {
 105                                /* only use first digit */
 106                                permille += *end - '0';
 107                                /* .. and ignore any further digits */
 108                                while (isdigit(*++end))
 109                                        ; /* nothing */
 110                        }
 111                        if (!*end)
 112                                options->dirstat_permille = permille;
 113                        else {
 114                                strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 115                                            p);
 116                                ret++;
 117                        }
 118                } else {
 119                        strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 120                        ret++;
 121                }
 122
 123        }
 124        string_list_clear(&params, 0);
 125        free(params_copy);
 126        return ret;
 127}
 128
 129static int parse_submodule_params(struct diff_options *options, const char *value)
 130{
 131        if (!strcmp(value, "log"))
 132                DIFF_OPT_SET(options, SUBMODULE_LOG);
 133        else if (!strcmp(value, "short"))
 134                DIFF_OPT_CLR(options, SUBMODULE_LOG);
 135        else
 136                return -1;
 137        return 0;
 138}
 139
 140static int git_config_rename(const char *var, const char *value)
 141{
 142        if (!value)
 143                return DIFF_DETECT_RENAME;
 144        if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 145                return  DIFF_DETECT_COPY;
 146        return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 147}
 148
 149long parse_algorithm_value(const char *value)
 150{
 151        if (!value)
 152                return -1;
 153        else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 154                return 0;
 155        else if (!strcasecmp(value, "minimal"))
 156                return XDF_NEED_MINIMAL;
 157        else if (!strcasecmp(value, "patience"))
 158                return XDF_PATIENCE_DIFF;
 159        else if (!strcasecmp(value, "histogram"))
 160                return XDF_HISTOGRAM_DIFF;
 161        return -1;
 162}
 163
 164/*
 165 * These are to give UI layer defaults.
 166 * The core-level commands such as git-diff-files should
 167 * never be affected by the setting of diff.renames
 168 * the user happens to have in the configuration file.
 169 */
 170int git_diff_ui_config(const char *var, const char *value, void *cb)
 171{
 172        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 173                diff_use_color_default = git_config_colorbool(var, value);
 174                return 0;
 175        }
 176        if (!strcmp(var, "diff.context")) {
 177                diff_context_default = git_config_int(var, value);
 178                if (diff_context_default < 0)
 179                        return -1;
 180                return 0;
 181        }
 182        if (!strcmp(var, "diff.renames")) {
 183                diff_detect_rename_default = git_config_rename(var, value);
 184                return 0;
 185        }
 186        if (!strcmp(var, "diff.autorefreshindex")) {
 187                diff_auto_refresh_index = git_config_bool(var, value);
 188                return 0;
 189        }
 190        if (!strcmp(var, "diff.mnemonicprefix")) {
 191                diff_mnemonic_prefix = git_config_bool(var, value);
 192                return 0;
 193        }
 194        if (!strcmp(var, "diff.noprefix")) {
 195                diff_no_prefix = git_config_bool(var, value);
 196                return 0;
 197        }
 198        if (!strcmp(var, "diff.statgraphwidth")) {
 199                diff_stat_graph_width = git_config_int(var, value);
 200                return 0;
 201        }
 202        if (!strcmp(var, "diff.external"))
 203                return git_config_string(&external_diff_cmd_cfg, var, value);
 204        if (!strcmp(var, "diff.wordregex"))
 205                return git_config_string(&diff_word_regex_cfg, var, value);
 206        if (!strcmp(var, "diff.orderfile"))
 207                return git_config_pathname(&diff_order_file_cfg, var, value);
 208
 209        if (!strcmp(var, "diff.ignoresubmodules"))
 210                handle_ignore_submodules_arg(&default_diff_options, value);
 211
 212        if (!strcmp(var, "diff.submodule")) {
 213                if (parse_submodule_params(&default_diff_options, value))
 214                        warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 215                                value);
 216                return 0;
 217        }
 218
 219        if (!strcmp(var, "diff.algorithm")) {
 220                diff_algorithm = parse_algorithm_value(value);
 221                if (diff_algorithm < 0)
 222                        return -1;
 223                return 0;
 224        }
 225
 226        if (git_color_config(var, value, cb) < 0)
 227                return -1;
 228
 229        return git_diff_basic_config(var, value, cb);
 230}
 231
 232int git_diff_basic_config(const char *var, const char *value, void *cb)
 233{
 234        if (!strcmp(var, "diff.renamelimit")) {
 235                diff_rename_limit_default = git_config_int(var, value);
 236                return 0;
 237        }
 238
 239        if (userdiff_config(var, value) < 0)
 240                return -1;
 241
 242        if (starts_with(var, "diff.color.") || starts_with(var, "color.diff.")) {
 243                int slot = parse_diff_color_slot(var, 11);
 244                if (slot < 0)
 245                        return 0;
 246                if (!value)
 247                        return config_error_nonbool(var);
 248                color_parse(value, var, diff_colors[slot]);
 249                return 0;
 250        }
 251
 252        /* like GNU diff's --suppress-blank-empty option  */
 253        if (!strcmp(var, "diff.suppressblankempty") ||
 254                        /* for backwards compatibility */
 255                        !strcmp(var, "diff.suppress-blank-empty")) {
 256                diff_suppress_blank_empty = git_config_bool(var, value);
 257                return 0;
 258        }
 259
 260        if (!strcmp(var, "diff.dirstat")) {
 261                struct strbuf errmsg = STRBUF_INIT;
 262                default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 263                if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 264                        warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 265                                errmsg.buf);
 266                strbuf_release(&errmsg);
 267                diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 268                return 0;
 269        }
 270
 271        if (starts_with(var, "submodule."))
 272                return parse_submodule_config_option(var, value);
 273
 274        return git_default_config(var, value, cb);
 275}
 276
 277static char *quote_two(const char *one, const char *two)
 278{
 279        int need_one = quote_c_style(one, NULL, NULL, 1);
 280        int need_two = quote_c_style(two, NULL, NULL, 1);
 281        struct strbuf res = STRBUF_INIT;
 282
 283        if (need_one + need_two) {
 284                strbuf_addch(&res, '"');
 285                quote_c_style(one, &res, NULL, 1);
 286                quote_c_style(two, &res, NULL, 1);
 287                strbuf_addch(&res, '"');
 288        } else {
 289                strbuf_addstr(&res, one);
 290                strbuf_addstr(&res, two);
 291        }
 292        return strbuf_detach(&res, NULL);
 293}
 294
 295static const char *external_diff(void)
 296{
 297        static const char *external_diff_cmd = NULL;
 298        static int done_preparing = 0;
 299
 300        if (done_preparing)
 301                return external_diff_cmd;
 302        external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 303        if (!external_diff_cmd)
 304                external_diff_cmd = external_diff_cmd_cfg;
 305        done_preparing = 1;
 306        return external_diff_cmd;
 307}
 308
 309static struct diff_tempfile {
 310        const char *name; /* filename external diff should read from */
 311        char hex[41];
 312        char mode[10];
 313        char tmp_path[PATH_MAX];
 314} diff_temp[2];
 315
 316typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 317
 318struct emit_callback {
 319        int color_diff;
 320        unsigned ws_rule;
 321        int blank_at_eof_in_preimage;
 322        int blank_at_eof_in_postimage;
 323        int lno_in_preimage;
 324        int lno_in_postimage;
 325        sane_truncate_fn truncate;
 326        const char **label_path;
 327        struct diff_words_data *diff_words;
 328        struct diff_options *opt;
 329        int *found_changesp;
 330        struct strbuf *header;
 331};
 332
 333static int count_lines(const char *data, int size)
 334{
 335        int count, ch, completely_empty = 1, nl_just_seen = 0;
 336        count = 0;
 337        while (0 < size--) {
 338                ch = *data++;
 339                if (ch == '\n') {
 340                        count++;
 341                        nl_just_seen = 1;
 342                        completely_empty = 0;
 343                }
 344                else {
 345                        nl_just_seen = 0;
 346                        completely_empty = 0;
 347                }
 348        }
 349        if (completely_empty)
 350                return 0;
 351        if (!nl_just_seen)
 352                count++; /* no trailing newline */
 353        return count;
 354}
 355
 356static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 357{
 358        if (!DIFF_FILE_VALID(one)) {
 359                mf->ptr = (char *)""; /* does not matter */
 360                mf->size = 0;
 361                return 0;
 362        }
 363        else if (diff_populate_filespec(one, 0))
 364                return -1;
 365
 366        mf->ptr = one->data;
 367        mf->size = one->size;
 368        return 0;
 369}
 370
 371/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 372static unsigned long diff_filespec_size(struct diff_filespec *one)
 373{
 374        if (!DIFF_FILE_VALID(one))
 375                return 0;
 376        diff_populate_filespec(one, 1);
 377        return one->size;
 378}
 379
 380static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 381{
 382        char *ptr = mf->ptr;
 383        long size = mf->size;
 384        int cnt = 0;
 385
 386        if (!size)
 387                return cnt;
 388        ptr += size - 1; /* pointing at the very end */
 389        if (*ptr != '\n')
 390                ; /* incomplete line */
 391        else
 392                ptr--; /* skip the last LF */
 393        while (mf->ptr < ptr) {
 394                char *prev_eol;
 395                for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 396                        if (*prev_eol == '\n')
 397                                break;
 398                if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 399                        break;
 400                cnt++;
 401                ptr = prev_eol - 1;
 402        }
 403        return cnt;
 404}
 405
 406static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 407                               struct emit_callback *ecbdata)
 408{
 409        int l1, l2, at;
 410        unsigned ws_rule = ecbdata->ws_rule;
 411        l1 = count_trailing_blank(mf1, ws_rule);
 412        l2 = count_trailing_blank(mf2, ws_rule);
 413        if (l2 <= l1) {
 414                ecbdata->blank_at_eof_in_preimage = 0;
 415                ecbdata->blank_at_eof_in_postimage = 0;
 416                return;
 417        }
 418        at = count_lines(mf1->ptr, mf1->size);
 419        ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 420
 421        at = count_lines(mf2->ptr, mf2->size);
 422        ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 423}
 424
 425static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 426                        int first, const char *line, int len)
 427{
 428        int has_trailing_newline, has_trailing_carriage_return;
 429        int nofirst;
 430        FILE *file = o->file;
 431
 432        fputs(diff_line_prefix(o), file);
 433
 434        if (len == 0) {
 435                has_trailing_newline = (first == '\n');
 436                has_trailing_carriage_return = (!has_trailing_newline &&
 437                                                (first == '\r'));
 438                nofirst = has_trailing_newline || has_trailing_carriage_return;
 439        } else {
 440                has_trailing_newline = (len > 0 && line[len-1] == '\n');
 441                if (has_trailing_newline)
 442                        len--;
 443                has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 444                if (has_trailing_carriage_return)
 445                        len--;
 446                nofirst = 0;
 447        }
 448
 449        if (len || !nofirst) {
 450                fputs(set, file);
 451                if (!nofirst)
 452                        fputc(first, file);
 453                fwrite(line, len, 1, file);
 454                fputs(reset, file);
 455        }
 456        if (has_trailing_carriage_return)
 457                fputc('\r', file);
 458        if (has_trailing_newline)
 459                fputc('\n', file);
 460}
 461
 462static void emit_line(struct diff_options *o, const char *set, const char *reset,
 463                      const char *line, int len)
 464{
 465        emit_line_0(o, set, reset, line[0], line+1, len-1);
 466}
 467
 468static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 469{
 470        if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 471              ecbdata->blank_at_eof_in_preimage &&
 472              ecbdata->blank_at_eof_in_postimage &&
 473              ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 474              ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 475                return 0;
 476        return ws_blank_line(line, len, ecbdata->ws_rule);
 477}
 478
 479static void emit_add_line(const char *reset,
 480                          struct emit_callback *ecbdata,
 481                          const char *line, int len)
 482{
 483        const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 484        const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 485
 486        if (!*ws)
 487                emit_line_0(ecbdata->opt, set, reset, '+', line, len);
 488        else if (new_blank_line_at_eof(ecbdata, line, len))
 489                /* Blank line at EOF - paint '+' as well */
 490                emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
 491        else {
 492                /* Emit just the prefix, then the rest. */
 493                emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
 494                ws_check_emit(line, len, ecbdata->ws_rule,
 495                              ecbdata->opt->file, set, reset, ws);
 496        }
 497}
 498
 499static void emit_hunk_header(struct emit_callback *ecbdata,
 500                             const char *line, int len)
 501{
 502        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 503        const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 504        const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 505        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 506        static const char atat[2] = { '@', '@' };
 507        const char *cp, *ep;
 508        struct strbuf msgbuf = STRBUF_INIT;
 509        int org_len = len;
 510        int i = 1;
 511
 512        /*
 513         * As a hunk header must begin with "@@ -<old>, +<new> @@",
 514         * it always is at least 10 bytes long.
 515         */
 516        if (len < 10 ||
 517            memcmp(line, atat, 2) ||
 518            !(ep = memmem(line + 2, len - 2, atat, 2))) {
 519                emit_line(ecbdata->opt, plain, reset, line, len);
 520                return;
 521        }
 522        ep += 2; /* skip over @@ */
 523
 524        /* The hunk header in fraginfo color */
 525        strbuf_add(&msgbuf, frag, strlen(frag));
 526        strbuf_add(&msgbuf, line, ep - line);
 527        strbuf_add(&msgbuf, reset, strlen(reset));
 528
 529        /*
 530         * trailing "\r\n"
 531         */
 532        for ( ; i < 3; i++)
 533                if (line[len - i] == '\r' || line[len - i] == '\n')
 534                        len--;
 535
 536        /* blank before the func header */
 537        for (cp = ep; ep - line < len; ep++)
 538                if (*ep != ' ' && *ep != '\t')
 539                        break;
 540        if (ep != cp) {
 541                strbuf_add(&msgbuf, plain, strlen(plain));
 542                strbuf_add(&msgbuf, cp, ep - cp);
 543                strbuf_add(&msgbuf, reset, strlen(reset));
 544        }
 545
 546        if (ep < line + len) {
 547                strbuf_add(&msgbuf, func, strlen(func));
 548                strbuf_add(&msgbuf, ep, line + len - ep);
 549                strbuf_add(&msgbuf, reset, strlen(reset));
 550        }
 551
 552        strbuf_add(&msgbuf, line + len, org_len - len);
 553        emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 554        strbuf_release(&msgbuf);
 555}
 556
 557static struct diff_tempfile *claim_diff_tempfile(void) {
 558        int i;
 559        for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 560                if (!diff_temp[i].name)
 561                        return diff_temp + i;
 562        die("BUG: diff is failing to clean up its tempfiles");
 563}
 564
 565static int remove_tempfile_installed;
 566
 567static void remove_tempfile(void)
 568{
 569        int i;
 570        for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 571                if (diff_temp[i].name == diff_temp[i].tmp_path)
 572                        unlink_or_warn(diff_temp[i].name);
 573                diff_temp[i].name = NULL;
 574        }
 575}
 576
 577static void remove_tempfile_on_signal(int signo)
 578{
 579        remove_tempfile();
 580        sigchain_pop(signo);
 581        raise(signo);
 582}
 583
 584static void print_line_count(FILE *file, int count)
 585{
 586        switch (count) {
 587        case 0:
 588                fprintf(file, "0,0");
 589                break;
 590        case 1:
 591                fprintf(file, "1");
 592                break;
 593        default:
 594                fprintf(file, "1,%d", count);
 595                break;
 596        }
 597}
 598
 599static void emit_rewrite_lines(struct emit_callback *ecb,
 600                               int prefix, const char *data, int size)
 601{
 602        const char *endp = NULL;
 603        static const char *nneof = " No newline at end of file\n";
 604        const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 605        const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 606
 607        while (0 < size) {
 608                int len;
 609
 610                endp = memchr(data, '\n', size);
 611                len = endp ? (endp - data + 1) : size;
 612                if (prefix != '+') {
 613                        ecb->lno_in_preimage++;
 614                        emit_line_0(ecb->opt, old, reset, '-',
 615                                    data, len);
 616                } else {
 617                        ecb->lno_in_postimage++;
 618                        emit_add_line(reset, ecb, data, len);
 619                }
 620                size -= len;
 621                data += len;
 622        }
 623        if (!endp) {
 624                const char *plain = diff_get_color(ecb->color_diff,
 625                                                   DIFF_PLAIN);
 626                putc('\n', ecb->opt->file);
 627                emit_line_0(ecb->opt, plain, reset, '\\',
 628                            nneof, strlen(nneof));
 629        }
 630}
 631
 632static void emit_rewrite_diff(const char *name_a,
 633                              const char *name_b,
 634                              struct diff_filespec *one,
 635                              struct diff_filespec *two,
 636                              struct userdiff_driver *textconv_one,
 637                              struct userdiff_driver *textconv_two,
 638                              struct diff_options *o)
 639{
 640        int lc_a, lc_b;
 641        const char *name_a_tab, *name_b_tab;
 642        const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 643        const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 644        const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 645        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 646        const char *a_prefix, *b_prefix;
 647        char *data_one, *data_two;
 648        size_t size_one, size_two;
 649        struct emit_callback ecbdata;
 650        const char *line_prefix = diff_line_prefix(o);
 651
 652        if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 653                a_prefix = o->b_prefix;
 654                b_prefix = o->a_prefix;
 655        } else {
 656                a_prefix = o->a_prefix;
 657                b_prefix = o->b_prefix;
 658        }
 659
 660        name_a += (*name_a == '/');
 661        name_b += (*name_b == '/');
 662        name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 663        name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 664
 665        strbuf_reset(&a_name);
 666        strbuf_reset(&b_name);
 667        quote_two_c_style(&a_name, a_prefix, name_a, 0);
 668        quote_two_c_style(&b_name, b_prefix, name_b, 0);
 669
 670        size_one = fill_textconv(textconv_one, one, &data_one);
 671        size_two = fill_textconv(textconv_two, two, &data_two);
 672
 673        memset(&ecbdata, 0, sizeof(ecbdata));
 674        ecbdata.color_diff = want_color(o->use_color);
 675        ecbdata.found_changesp = &o->found_changes;
 676        ecbdata.ws_rule = whitespace_rule(name_b);
 677        ecbdata.opt = o;
 678        if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 679                mmfile_t mf1, mf2;
 680                mf1.ptr = (char *)data_one;
 681                mf2.ptr = (char *)data_two;
 682                mf1.size = size_one;
 683                mf2.size = size_two;
 684                check_blank_at_eof(&mf1, &mf2, &ecbdata);
 685        }
 686        ecbdata.lno_in_preimage = 1;
 687        ecbdata.lno_in_postimage = 1;
 688
 689        lc_a = count_lines(data_one, size_one);
 690        lc_b = count_lines(data_two, size_two);
 691        fprintf(o->file,
 692                "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 693                line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 694                line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 695                line_prefix, fraginfo);
 696        if (!o->irreversible_delete)
 697                print_line_count(o->file, lc_a);
 698        else
 699                fprintf(o->file, "?,?");
 700        fprintf(o->file, " +");
 701        print_line_count(o->file, lc_b);
 702        fprintf(o->file, " @@%s\n", reset);
 703        if (lc_a && !o->irreversible_delete)
 704                emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 705        if (lc_b)
 706                emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 707        if (textconv_one)
 708                free((char *)data_one);
 709        if (textconv_two)
 710                free((char *)data_two);
 711}
 712
 713struct diff_words_buffer {
 714        mmfile_t text;
 715        long alloc;
 716        struct diff_words_orig {
 717                const char *begin, *end;
 718        } *orig;
 719        int orig_nr, orig_alloc;
 720};
 721
 722static void diff_words_append(char *line, unsigned long len,
 723                struct diff_words_buffer *buffer)
 724{
 725        ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 726        line++;
 727        len--;
 728        memcpy(buffer->text.ptr + buffer->text.size, line, len);
 729        buffer->text.size += len;
 730        buffer->text.ptr[buffer->text.size] = '\0';
 731}
 732
 733struct diff_words_style_elem {
 734        const char *prefix;
 735        const char *suffix;
 736        const char *color; /* NULL; filled in by the setup code if
 737                            * color is enabled */
 738};
 739
 740struct diff_words_style {
 741        enum diff_words_type type;
 742        struct diff_words_style_elem new, old, ctx;
 743        const char *newline;
 744};
 745
 746static struct diff_words_style diff_words_styles[] = {
 747        { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 748        { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 749        { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 750};
 751
 752struct diff_words_data {
 753        struct diff_words_buffer minus, plus;
 754        const char *current_plus;
 755        int last_minus;
 756        struct diff_options *opt;
 757        regex_t *word_regex;
 758        enum diff_words_type type;
 759        struct diff_words_style *style;
 760};
 761
 762static int fn_out_diff_words_write_helper(FILE *fp,
 763                                          struct diff_words_style_elem *st_el,
 764                                          const char *newline,
 765                                          size_t count, const char *buf,
 766                                          const char *line_prefix)
 767{
 768        int print = 0;
 769
 770        while (count) {
 771                char *p = memchr(buf, '\n', count);
 772                if (print)
 773                        fputs(line_prefix, fp);
 774                if (p != buf) {
 775                        if (st_el->color && fputs(st_el->color, fp) < 0)
 776                                return -1;
 777                        if (fputs(st_el->prefix, fp) < 0 ||
 778                            fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 779                            fputs(st_el->suffix, fp) < 0)
 780                                return -1;
 781                        if (st_el->color && *st_el->color
 782                            && fputs(GIT_COLOR_RESET, fp) < 0)
 783                                return -1;
 784                }
 785                if (!p)
 786                        return 0;
 787                if (fputs(newline, fp) < 0)
 788                        return -1;
 789                count -= p + 1 - buf;
 790                buf = p + 1;
 791                print = 1;
 792        }
 793        return 0;
 794}
 795
 796/*
 797 * '--color-words' algorithm can be described as:
 798 *
 799 *   1. collect a the minus/plus lines of a diff hunk, divided into
 800 *      minus-lines and plus-lines;
 801 *
 802 *   2. break both minus-lines and plus-lines into words and
 803 *      place them into two mmfile_t with one word for each line;
 804 *
 805 *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 806 *
 807 * And for the common parts of the both file, we output the plus side text.
 808 * diff_words->current_plus is used to trace the current position of the plus file
 809 * which printed. diff_words->last_minus is used to trace the last minus word
 810 * printed.
 811 *
 812 * For '--graph' to work with '--color-words', we need to output the graph prefix
 813 * on each line of color words output. Generally, there are two conditions on
 814 * which we should output the prefix.
 815 *
 816 *   1. diff_words->last_minus == 0 &&
 817 *      diff_words->current_plus == diff_words->plus.text.ptr
 818 *
 819 *      that is: the plus text must start as a new line, and if there is no minus
 820 *      word printed, a graph prefix must be printed.
 821 *
 822 *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 823 *      *(diff_words->current_plus - 1) == '\n'
 824 *
 825 *      that is: a graph prefix must be printed following a '\n'
 826 */
 827static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 828{
 829        if ((diff_words->last_minus == 0 &&
 830                diff_words->current_plus == diff_words->plus.text.ptr) ||
 831                (diff_words->current_plus > diff_words->plus.text.ptr &&
 832                *(diff_words->current_plus - 1) == '\n')) {
 833                return 1;
 834        } else {
 835                return 0;
 836        }
 837}
 838
 839static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 840{
 841        struct diff_words_data *diff_words = priv;
 842        struct diff_words_style *style = diff_words->style;
 843        int minus_first, minus_len, plus_first, plus_len;
 844        const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 845        struct diff_options *opt = diff_words->opt;
 846        const char *line_prefix;
 847
 848        if (line[0] != '@' || parse_hunk_header(line, len,
 849                        &minus_first, &minus_len, &plus_first, &plus_len))
 850                return;
 851
 852        assert(opt);
 853        line_prefix = diff_line_prefix(opt);
 854
 855        /* POSIX requires that first be decremented by one if len == 0... */
 856        if (minus_len) {
 857                minus_begin = diff_words->minus.orig[minus_first].begin;
 858                minus_end =
 859                        diff_words->minus.orig[minus_first + minus_len - 1].end;
 860        } else
 861                minus_begin = minus_end =
 862                        diff_words->minus.orig[minus_first].end;
 863
 864        if (plus_len) {
 865                plus_begin = diff_words->plus.orig[plus_first].begin;
 866                plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 867        } else
 868                plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 869
 870        if (color_words_output_graph_prefix(diff_words)) {
 871                fputs(line_prefix, diff_words->opt->file);
 872        }
 873        if (diff_words->current_plus != plus_begin) {
 874                fn_out_diff_words_write_helper(diff_words->opt->file,
 875                                &style->ctx, style->newline,
 876                                plus_begin - diff_words->current_plus,
 877                                diff_words->current_plus, line_prefix);
 878                if (*(plus_begin - 1) == '\n')
 879                        fputs(line_prefix, diff_words->opt->file);
 880        }
 881        if (minus_begin != minus_end) {
 882                fn_out_diff_words_write_helper(diff_words->opt->file,
 883                                &style->old, style->newline,
 884                                minus_end - minus_begin, minus_begin,
 885                                line_prefix);
 886        }
 887        if (plus_begin != plus_end) {
 888                fn_out_diff_words_write_helper(diff_words->opt->file,
 889                                &style->new, style->newline,
 890                                plus_end - plus_begin, plus_begin,
 891                                line_prefix);
 892        }
 893
 894        diff_words->current_plus = plus_end;
 895        diff_words->last_minus = minus_first;
 896}
 897
 898/* This function starts looking at *begin, and returns 0 iff a word was found. */
 899static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 900                int *begin, int *end)
 901{
 902        if (word_regex && *begin < buffer->size) {
 903                regmatch_t match[1];
 904                if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 905                        char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 906                                        '\n', match[0].rm_eo - match[0].rm_so);
 907                        *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 908                        *begin += match[0].rm_so;
 909                        return *begin >= *end;
 910                }
 911                return -1;
 912        }
 913
 914        /* find the next word */
 915        while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 916                (*begin)++;
 917        if (*begin >= buffer->size)
 918                return -1;
 919
 920        /* find the end of the word */
 921        *end = *begin + 1;
 922        while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 923                (*end)++;
 924
 925        return 0;
 926}
 927
 928/*
 929 * This function splits the words in buffer->text, stores the list with
 930 * newline separator into out, and saves the offsets of the original words
 931 * in buffer->orig.
 932 */
 933static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 934                regex_t *word_regex)
 935{
 936        int i, j;
 937        long alloc = 0;
 938
 939        out->size = 0;
 940        out->ptr = NULL;
 941
 942        /* fake an empty "0th" word */
 943        ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 944        buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 945        buffer->orig_nr = 1;
 946
 947        for (i = 0; i < buffer->text.size; i++) {
 948                if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 949                        return;
 950
 951                /* store original boundaries */
 952                ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 953                                buffer->orig_alloc);
 954                buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 955                buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 956                buffer->orig_nr++;
 957
 958                /* store one word */
 959                ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 960                memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 961                out->ptr[out->size + j - i] = '\n';
 962                out->size += j - i + 1;
 963
 964                i = j - 1;
 965        }
 966}
 967
 968/* this executes the word diff on the accumulated buffers */
 969static void diff_words_show(struct diff_words_data *diff_words)
 970{
 971        xpparam_t xpp;
 972        xdemitconf_t xecfg;
 973        mmfile_t minus, plus;
 974        struct diff_words_style *style = diff_words->style;
 975
 976        struct diff_options *opt = diff_words->opt;
 977        const char *line_prefix;
 978
 979        assert(opt);
 980        line_prefix = diff_line_prefix(opt);
 981
 982        /* special case: only removal */
 983        if (!diff_words->plus.text.size) {
 984                fputs(line_prefix, diff_words->opt->file);
 985                fn_out_diff_words_write_helper(diff_words->opt->file,
 986                        &style->old, style->newline,
 987                        diff_words->minus.text.size,
 988                        diff_words->minus.text.ptr, line_prefix);
 989                diff_words->minus.text.size = 0;
 990                return;
 991        }
 992
 993        diff_words->current_plus = diff_words->plus.text.ptr;
 994        diff_words->last_minus = 0;
 995
 996        memset(&xpp, 0, sizeof(xpp));
 997        memset(&xecfg, 0, sizeof(xecfg));
 998        diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 999        diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1000        xpp.flags = 0;
1001        /* as only the hunk header will be parsed, we need a 0-context */
1002        xecfg.ctxlen = 0;
1003        xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1004                      &xpp, &xecfg);
1005        free(minus.ptr);
1006        free(plus.ptr);
1007        if (diff_words->current_plus != diff_words->plus.text.ptr +
1008                        diff_words->plus.text.size) {
1009                if (color_words_output_graph_prefix(diff_words))
1010                        fputs(line_prefix, diff_words->opt->file);
1011                fn_out_diff_words_write_helper(diff_words->opt->file,
1012                        &style->ctx, style->newline,
1013                        diff_words->plus.text.ptr + diff_words->plus.text.size
1014                        - diff_words->current_plus, diff_words->current_plus,
1015                        line_prefix);
1016        }
1017        diff_words->minus.text.size = diff_words->plus.text.size = 0;
1018}
1019
1020/* In "color-words" mode, show word-diff of words accumulated in the buffer */
1021static void diff_words_flush(struct emit_callback *ecbdata)
1022{
1023        if (ecbdata->diff_words->minus.text.size ||
1024            ecbdata->diff_words->plus.text.size)
1025                diff_words_show(ecbdata->diff_words);
1026}
1027
1028static void diff_filespec_load_driver(struct diff_filespec *one)
1029{
1030        /* Use already-loaded driver */
1031        if (one->driver)
1032                return;
1033
1034        if (S_ISREG(one->mode))
1035                one->driver = userdiff_find_by_path(one->path);
1036
1037        /* Fallback to default settings */
1038        if (!one->driver)
1039                one->driver = userdiff_find_by_name("default");
1040}
1041
1042static const char *userdiff_word_regex(struct diff_filespec *one)
1043{
1044        diff_filespec_load_driver(one);
1045        return one->driver->word_regex;
1046}
1047
1048static void init_diff_words_data(struct emit_callback *ecbdata,
1049                                 struct diff_options *orig_opts,
1050                                 struct diff_filespec *one,
1051                                 struct diff_filespec *two)
1052{
1053        int i;
1054        struct diff_options *o = xmalloc(sizeof(struct diff_options));
1055        memcpy(o, orig_opts, sizeof(struct diff_options));
1056
1057        ecbdata->diff_words =
1058                xcalloc(1, sizeof(struct diff_words_data));
1059        ecbdata->diff_words->type = o->word_diff;
1060        ecbdata->diff_words->opt = o;
1061        if (!o->word_regex)
1062                o->word_regex = userdiff_word_regex(one);
1063        if (!o->word_regex)
1064                o->word_regex = userdiff_word_regex(two);
1065        if (!o->word_regex)
1066                o->word_regex = diff_word_regex_cfg;
1067        if (o->word_regex) {
1068                ecbdata->diff_words->word_regex = (regex_t *)
1069                        xmalloc(sizeof(regex_t));
1070                if (regcomp(ecbdata->diff_words->word_regex,
1071                            o->word_regex,
1072                            REG_EXTENDED | REG_NEWLINE))
1073                        die ("Invalid regular expression: %s",
1074                             o->word_regex);
1075        }
1076        for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1077                if (o->word_diff == diff_words_styles[i].type) {
1078                        ecbdata->diff_words->style =
1079                                &diff_words_styles[i];
1080                        break;
1081                }
1082        }
1083        if (want_color(o->use_color)) {
1084                struct diff_words_style *st = ecbdata->diff_words->style;
1085                st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1086                st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1087                st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1088        }
1089}
1090
1091static void free_diff_words_data(struct emit_callback *ecbdata)
1092{
1093        if (ecbdata->diff_words) {
1094                diff_words_flush(ecbdata);
1095                free (ecbdata->diff_words->opt);
1096                free (ecbdata->diff_words->minus.text.ptr);
1097                free (ecbdata->diff_words->minus.orig);
1098                free (ecbdata->diff_words->plus.text.ptr);
1099                free (ecbdata->diff_words->plus.orig);
1100                if (ecbdata->diff_words->word_regex) {
1101                        regfree(ecbdata->diff_words->word_regex);
1102                        free(ecbdata->diff_words->word_regex);
1103                }
1104                free(ecbdata->diff_words);
1105                ecbdata->diff_words = NULL;
1106        }
1107}
1108
1109const char *diff_get_color(int diff_use_color, enum color_diff ix)
1110{
1111        if (want_color(diff_use_color))
1112                return diff_colors[ix];
1113        return "";
1114}
1115
1116const char *diff_line_prefix(struct diff_options *opt)
1117{
1118        struct strbuf *msgbuf;
1119        if (!opt->output_prefix)
1120                return "";
1121
1122        msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1123        return msgbuf->buf;
1124}
1125
1126static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1127{
1128        const char *cp;
1129        unsigned long allot;
1130        size_t l = len;
1131
1132        if (ecb->truncate)
1133                return ecb->truncate(line, len);
1134        cp = line;
1135        allot = l;
1136        while (0 < l) {
1137                (void) utf8_width(&cp, &l);
1138                if (!cp)
1139                        break; /* truncated in the middle? */
1140        }
1141        return allot - l;
1142}
1143
1144static void find_lno(const char *line, struct emit_callback *ecbdata)
1145{
1146        const char *p;
1147        ecbdata->lno_in_preimage = 0;
1148        ecbdata->lno_in_postimage = 0;
1149        p = strchr(line, '-');
1150        if (!p)
1151                return; /* cannot happen */
1152        ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1153        p = strchr(p, '+');
1154        if (!p)
1155                return; /* cannot happen */
1156        ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1157}
1158
1159static void fn_out_consume(void *priv, char *line, unsigned long len)
1160{
1161        struct emit_callback *ecbdata = priv;
1162        const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1163        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1164        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1165        struct diff_options *o = ecbdata->opt;
1166        const char *line_prefix = diff_line_prefix(o);
1167
1168        if (ecbdata->header) {
1169                fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1170                strbuf_reset(ecbdata->header);
1171                ecbdata->header = NULL;
1172        }
1173        *(ecbdata->found_changesp) = 1;
1174
1175        if (ecbdata->label_path[0]) {
1176                const char *name_a_tab, *name_b_tab;
1177
1178                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1179                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1180
1181                fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1182                        line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1183                fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1184                        line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1185                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1186        }
1187
1188        if (diff_suppress_blank_empty
1189            && len == 2 && line[0] == ' ' && line[1] == '\n') {
1190                line[0] = '\n';
1191                len = 1;
1192        }
1193
1194        if (line[0] == '@') {
1195                if (ecbdata->diff_words)
1196                        diff_words_flush(ecbdata);
1197                len = sane_truncate_line(ecbdata, line, len);
1198                find_lno(line, ecbdata);
1199                emit_hunk_header(ecbdata, line, len);
1200                if (line[len-1] != '\n')
1201                        putc('\n', ecbdata->opt->file);
1202                return;
1203        }
1204
1205        if (len < 1) {
1206                emit_line(ecbdata->opt, reset, reset, line, len);
1207                if (ecbdata->diff_words
1208                    && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1209                        fputs("~\n", ecbdata->opt->file);
1210                return;
1211        }
1212
1213        if (ecbdata->diff_words) {
1214                if (line[0] == '-') {
1215                        diff_words_append(line, len,
1216                                          &ecbdata->diff_words->minus);
1217                        return;
1218                } else if (line[0] == '+') {
1219                        diff_words_append(line, len,
1220                                          &ecbdata->diff_words->plus);
1221                        return;
1222                } else if (starts_with(line, "\\ ")) {
1223                        /*
1224                         * Eat the "no newline at eof" marker as if we
1225                         * saw a "+" or "-" line with nothing on it,
1226                         * and return without diff_words_flush() to
1227                         * defer processing. If this is the end of
1228                         * preimage, more "+" lines may come after it.
1229                         */
1230                        return;
1231                }
1232                diff_words_flush(ecbdata);
1233                if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1234                        emit_line(ecbdata->opt, plain, reset, line, len);
1235                        fputs("~\n", ecbdata->opt->file);
1236                } else {
1237                        /*
1238                         * Skip the prefix character, if any.  With
1239                         * diff_suppress_blank_empty, there may be
1240                         * none.
1241                         */
1242                        if (line[0] != '\n') {
1243                              line++;
1244                              len--;
1245                        }
1246                        emit_line(ecbdata->opt, plain, reset, line, len);
1247                }
1248                return;
1249        }
1250
1251        if (line[0] != '+') {
1252                const char *color =
1253                        diff_get_color(ecbdata->color_diff,
1254                                       line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1255                ecbdata->lno_in_preimage++;
1256                if (line[0] == ' ')
1257                        ecbdata->lno_in_postimage++;
1258                emit_line(ecbdata->opt, color, reset, line, len);
1259        } else {
1260                ecbdata->lno_in_postimage++;
1261                emit_add_line(reset, ecbdata, line + 1, len - 1);
1262        }
1263}
1264
1265static char *pprint_rename(const char *a, const char *b)
1266{
1267        const char *old = a;
1268        const char *new = b;
1269        struct strbuf name = STRBUF_INIT;
1270        int pfx_length, sfx_length;
1271        int pfx_adjust_for_slash;
1272        int len_a = strlen(a);
1273        int len_b = strlen(b);
1274        int a_midlen, b_midlen;
1275        int qlen_a = quote_c_style(a, NULL, NULL, 0);
1276        int qlen_b = quote_c_style(b, NULL, NULL, 0);
1277
1278        if (qlen_a || qlen_b) {
1279                quote_c_style(a, &name, NULL, 0);
1280                strbuf_addstr(&name, " => ");
1281                quote_c_style(b, &name, NULL, 0);
1282                return strbuf_detach(&name, NULL);
1283        }
1284
1285        /* Find common prefix */
1286        pfx_length = 0;
1287        while (*old && *new && *old == *new) {
1288                if (*old == '/')
1289                        pfx_length = old - a + 1;
1290                old++;
1291                new++;
1292        }
1293
1294        /* Find common suffix */
1295        old = a + len_a;
1296        new = b + len_b;
1297        sfx_length = 0;
1298        /*
1299         * If there is a common prefix, it must end in a slash.  In
1300         * that case we let this loop run 1 into the prefix to see the
1301         * same slash.
1302         *
1303         * If there is no common prefix, we cannot do this as it would
1304         * underrun the input strings.
1305         */
1306        pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1307        while (a + pfx_length - pfx_adjust_for_slash <= old &&
1308               b + pfx_length - pfx_adjust_for_slash <= new &&
1309               *old == *new) {
1310                if (*old == '/')
1311                        sfx_length = len_a - (old - a);
1312                old--;
1313                new--;
1314        }
1315
1316        /*
1317         * pfx{mid-a => mid-b}sfx
1318         * {pfx-a => pfx-b}sfx
1319         * pfx{sfx-a => sfx-b}
1320         * name-a => name-b
1321         */
1322        a_midlen = len_a - pfx_length - sfx_length;
1323        b_midlen = len_b - pfx_length - sfx_length;
1324        if (a_midlen < 0)
1325                a_midlen = 0;
1326        if (b_midlen < 0)
1327                b_midlen = 0;
1328
1329        strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1330        if (pfx_length + sfx_length) {
1331                strbuf_add(&name, a, pfx_length);
1332                strbuf_addch(&name, '{');
1333        }
1334        strbuf_add(&name, a + pfx_length, a_midlen);
1335        strbuf_addstr(&name, " => ");
1336        strbuf_add(&name, b + pfx_length, b_midlen);
1337        if (pfx_length + sfx_length) {
1338                strbuf_addch(&name, '}');
1339                strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1340        }
1341        return strbuf_detach(&name, NULL);
1342}
1343
1344struct diffstat_t {
1345        int nr;
1346        int alloc;
1347        struct diffstat_file {
1348                char *from_name;
1349                char *name;
1350                char *print_name;
1351                unsigned is_unmerged:1;
1352                unsigned is_binary:1;
1353                unsigned is_renamed:1;
1354                unsigned is_interesting:1;
1355                uintmax_t added, deleted;
1356        } **files;
1357};
1358
1359static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1360                                          const char *name_a,
1361                                          const char *name_b)
1362{
1363        struct diffstat_file *x;
1364        x = xcalloc(sizeof (*x), 1);
1365        if (diffstat->nr == diffstat->alloc) {
1366                diffstat->alloc = alloc_nr(diffstat->alloc);
1367                diffstat->files = xrealloc(diffstat->files,
1368                                diffstat->alloc * sizeof(x));
1369        }
1370        diffstat->files[diffstat->nr++] = x;
1371        if (name_b) {
1372                x->from_name = xstrdup(name_a);
1373                x->name = xstrdup(name_b);
1374                x->is_renamed = 1;
1375        }
1376        else {
1377                x->from_name = NULL;
1378                x->name = xstrdup(name_a);
1379        }
1380        return x;
1381}
1382
1383static void diffstat_consume(void *priv, char *line, unsigned long len)
1384{
1385        struct diffstat_t *diffstat = priv;
1386        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1387
1388        if (line[0] == '+')
1389                x->added++;
1390        else if (line[0] == '-')
1391                x->deleted++;
1392}
1393
1394const char mime_boundary_leader[] = "------------";
1395
1396static int scale_linear(int it, int width, int max_change)
1397{
1398        if (!it)
1399                return 0;
1400        /*
1401         * make sure that at least one '-' or '+' is printed if
1402         * there is any change to this path. The easiest way is to
1403         * scale linearly as if the alloted width is one column shorter
1404         * than it is, and then add 1 to the result.
1405         */
1406        return 1 + (it * (width - 1) / max_change);
1407}
1408
1409static void show_name(FILE *file,
1410                      const char *prefix, const char *name, int len)
1411{
1412        fprintf(file, " %s%-*s |", prefix, len, name);
1413}
1414
1415static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1416{
1417        if (cnt <= 0)
1418                return;
1419        fprintf(file, "%s", set);
1420        while (cnt--)
1421                putc(ch, file);
1422        fprintf(file, "%s", reset);
1423}
1424
1425static void fill_print_name(struct diffstat_file *file)
1426{
1427        char *pname;
1428
1429        if (file->print_name)
1430                return;
1431
1432        if (!file->is_renamed) {
1433                struct strbuf buf = STRBUF_INIT;
1434                if (quote_c_style(file->name, &buf, NULL, 0)) {
1435                        pname = strbuf_detach(&buf, NULL);
1436                } else {
1437                        pname = file->name;
1438                        strbuf_release(&buf);
1439                }
1440        } else {
1441                pname = pprint_rename(file->from_name, file->name);
1442        }
1443        file->print_name = pname;
1444}
1445
1446int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1447{
1448        struct strbuf sb = STRBUF_INIT;
1449        int ret;
1450
1451        if (!files) {
1452                assert(insertions == 0 && deletions == 0);
1453                return fprintf(fp, "%s\n", " 0 files changed");
1454        }
1455
1456        strbuf_addf(&sb,
1457                    (files == 1) ? " %d file changed" : " %d files changed",
1458                    files);
1459
1460        /*
1461         * For binary diff, the caller may want to print "x files
1462         * changed" with insertions == 0 && deletions == 0.
1463         *
1464         * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1465         * is probably less confusing (i.e skip over "2 files changed
1466         * but nothing about added/removed lines? Is this a bug in Git?").
1467         */
1468        if (insertions || deletions == 0) {
1469                /*
1470                 * TRANSLATORS: "+" in (+) is a line addition marker;
1471                 * do not translate it.
1472                 */
1473                strbuf_addf(&sb,
1474                            (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1475                            insertions);
1476        }
1477
1478        if (deletions || insertions == 0) {
1479                /*
1480                 * TRANSLATORS: "-" in (-) is a line removal marker;
1481                 * do not translate it.
1482                 */
1483                strbuf_addf(&sb,
1484                            (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1485                            deletions);
1486        }
1487        strbuf_addch(&sb, '\n');
1488        ret = fputs(sb.buf, fp);
1489        strbuf_release(&sb);
1490        return ret;
1491}
1492
1493static void show_stats(struct diffstat_t *data, struct diff_options *options)
1494{
1495        int i, len, add, del, adds = 0, dels = 0;
1496        uintmax_t max_change = 0, max_len = 0;
1497        int total_files = data->nr, count;
1498        int width, name_width, graph_width, number_width = 0, bin_width = 0;
1499        const char *reset, *add_c, *del_c;
1500        const char *line_prefix = "";
1501        int extra_shown = 0;
1502
1503        if (data->nr == 0)
1504                return;
1505
1506        line_prefix = diff_line_prefix(options);
1507        count = options->stat_count ? options->stat_count : data->nr;
1508
1509        reset = diff_get_color_opt(options, DIFF_RESET);
1510        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1511        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1512
1513        /*
1514         * Find the longest filename and max number of changes
1515         */
1516        for (i = 0; (i < count) && (i < data->nr); i++) {
1517                struct diffstat_file *file = data->files[i];
1518                uintmax_t change = file->added + file->deleted;
1519
1520                if (!file->is_interesting && (change == 0)) {
1521                        count++; /* not shown == room for one more */
1522                        continue;
1523                }
1524                fill_print_name(file);
1525                len = strlen(file->print_name);
1526                if (max_len < len)
1527                        max_len = len;
1528
1529                if (file->is_unmerged) {
1530                        /* "Unmerged" is 8 characters */
1531                        bin_width = bin_width < 8 ? 8 : bin_width;
1532                        continue;
1533                }
1534                if (file->is_binary) {
1535                        /* "Bin XXX -> YYY bytes" */
1536                        int w = 14 + decimal_width(file->added)
1537                                + decimal_width(file->deleted);
1538                        bin_width = bin_width < w ? w : bin_width;
1539                        /* Display change counts aligned with "Bin" */
1540                        number_width = 3;
1541                        continue;
1542                }
1543
1544                if (max_change < change)
1545                        max_change = change;
1546        }
1547        count = i; /* where we can stop scanning in data->files[] */
1548
1549        /*
1550         * We have width = stat_width or term_columns() columns total.
1551         * We want a maximum of min(max_len, stat_name_width) for the name part.
1552         * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1553         * We also need 1 for " " and 4 + decimal_width(max_change)
1554         * for " | NNNN " and one the empty column at the end, altogether
1555         * 6 + decimal_width(max_change).
1556         *
1557         * If there's not enough space, we will use the smaller of
1558         * stat_name_width (if set) and 5/8*width for the filename,
1559         * and the rest for constant elements + graph part, but no more
1560         * than stat_graph_width for the graph part.
1561         * (5/8 gives 50 for filename and 30 for the constant parts + graph
1562         * for the standard terminal size).
1563         *
1564         * In other words: stat_width limits the maximum width, and
1565         * stat_name_width fixes the maximum width of the filename,
1566         * and is also used to divide available columns if there
1567         * aren't enough.
1568         *
1569         * Binary files are displayed with "Bin XXX -> YYY bytes"
1570         * instead of the change count and graph. This part is treated
1571         * similarly to the graph part, except that it is not
1572         * "scaled". If total width is too small to accommodate the
1573         * guaranteed minimum width of the filename part and the
1574         * separators and this message, this message will "overflow"
1575         * making the line longer than the maximum width.
1576         */
1577
1578        if (options->stat_width == -1)
1579                width = term_columns() - options->output_prefix_length;
1580        else
1581                width = options->stat_width ? options->stat_width : 80;
1582        number_width = decimal_width(max_change) > number_width ?
1583                decimal_width(max_change) : number_width;
1584
1585        if (options->stat_graph_width == -1)
1586                options->stat_graph_width = diff_stat_graph_width;
1587
1588        /*
1589         * Guarantee 3/8*16==6 for the graph part
1590         * and 5/8*16==10 for the filename part
1591         */
1592        if (width < 16 + 6 + number_width)
1593                width = 16 + 6 + number_width;
1594
1595        /*
1596         * First assign sizes that are wanted, ignoring available width.
1597         * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1598         * starting from "XXX" should fit in graph_width.
1599         */
1600        graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1601        if (options->stat_graph_width &&
1602            options->stat_graph_width < graph_width)
1603                graph_width = options->stat_graph_width;
1604
1605        name_width = (options->stat_name_width > 0 &&
1606                      options->stat_name_width < max_len) ?
1607                options->stat_name_width : max_len;
1608
1609        /*
1610         * Adjust adjustable widths not to exceed maximum width
1611         */
1612        if (name_width + number_width + 6 + graph_width > width) {
1613                if (graph_width > width * 3/8 - number_width - 6) {
1614                        graph_width = width * 3/8 - number_width - 6;
1615                        if (graph_width < 6)
1616                                graph_width = 6;
1617                }
1618
1619                if (options->stat_graph_width &&
1620                    graph_width > options->stat_graph_width)
1621                        graph_width = options->stat_graph_width;
1622                if (name_width > width - number_width - 6 - graph_width)
1623                        name_width = width - number_width - 6 - graph_width;
1624                else
1625                        graph_width = width - number_width - 6 - name_width;
1626        }
1627
1628        /*
1629         * From here name_width is the width of the name area,
1630         * and graph_width is the width of the graph area.
1631         * max_change is used to scale graph properly.
1632         */
1633        for (i = 0; i < count; i++) {
1634                const char *prefix = "";
1635                struct diffstat_file *file = data->files[i];
1636                char *name = file->print_name;
1637                uintmax_t added = file->added;
1638                uintmax_t deleted = file->deleted;
1639                int name_len;
1640
1641                if (!file->is_interesting && (added + deleted == 0))
1642                        continue;
1643
1644                /*
1645                 * "scale" the filename
1646                 */
1647                len = name_width;
1648                name_len = strlen(name);
1649                if (name_width < name_len) {
1650                        char *slash;
1651                        prefix = "...";
1652                        len -= 3;
1653                        name += name_len - len;
1654                        slash = strchr(name, '/');
1655                        if (slash)
1656                                name = slash;
1657                }
1658
1659                if (file->is_binary) {
1660                        fprintf(options->file, "%s", line_prefix);
1661                        show_name(options->file, prefix, name, len);
1662                        fprintf(options->file, " %*s", number_width, "Bin");
1663                        if (!added && !deleted) {
1664                                putc('\n', options->file);
1665                                continue;
1666                        }
1667                        fprintf(options->file, " %s%"PRIuMAX"%s",
1668                                del_c, deleted, reset);
1669                        fprintf(options->file, " -> ");
1670                        fprintf(options->file, "%s%"PRIuMAX"%s",
1671                                add_c, added, reset);
1672                        fprintf(options->file, " bytes");
1673                        fprintf(options->file, "\n");
1674                        continue;
1675                }
1676                else if (file->is_unmerged) {
1677                        fprintf(options->file, "%s", line_prefix);
1678                        show_name(options->file, prefix, name, len);
1679                        fprintf(options->file, " Unmerged\n");
1680                        continue;
1681                }
1682
1683                /*
1684                 * scale the add/delete
1685                 */
1686                add = added;
1687                del = deleted;
1688
1689                if (graph_width <= max_change) {
1690                        int total = scale_linear(add + del, graph_width, max_change);
1691                        if (total < 2 && add && del)
1692                                /* width >= 2 due to the sanity check */
1693                                total = 2;
1694                        if (add < del) {
1695                                add = scale_linear(add, graph_width, max_change);
1696                                del = total - add;
1697                        } else {
1698                                del = scale_linear(del, graph_width, max_change);
1699                                add = total - del;
1700                        }
1701                }
1702                fprintf(options->file, "%s", line_prefix);
1703                show_name(options->file, prefix, name, len);
1704                fprintf(options->file, " %*"PRIuMAX"%s",
1705                        number_width, added + deleted,
1706                        added + deleted ? " " : "");
1707                show_graph(options->file, '+', add, add_c, reset);
1708                show_graph(options->file, '-', del, del_c, reset);
1709                fprintf(options->file, "\n");
1710        }
1711
1712        for (i = 0; i < data->nr; i++) {
1713                struct diffstat_file *file = data->files[i];
1714                uintmax_t added = file->added;
1715                uintmax_t deleted = file->deleted;
1716
1717                if (file->is_unmerged ||
1718                    (!file->is_interesting && (added + deleted == 0))) {
1719                        total_files--;
1720                        continue;
1721                }
1722
1723                if (!file->is_binary) {
1724                        adds += added;
1725                        dels += deleted;
1726                }
1727                if (i < count)
1728                        continue;
1729                if (!extra_shown)
1730                        fprintf(options->file, "%s ...\n", line_prefix);
1731                extra_shown = 1;
1732        }
1733        fprintf(options->file, "%s", line_prefix);
1734        print_stat_summary(options->file, total_files, adds, dels);
1735}
1736
1737static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1738{
1739        int i, adds = 0, dels = 0, total_files = data->nr;
1740
1741        if (data->nr == 0)
1742                return;
1743
1744        for (i = 0; i < data->nr; i++) {
1745                int added = data->files[i]->added;
1746                int deleted= data->files[i]->deleted;
1747
1748                if (data->files[i]->is_unmerged ||
1749                    (!data->files[i]->is_interesting && (added + deleted == 0))) {
1750                        total_files--;
1751                } else if (!data->files[i]->is_binary) { /* don't count bytes */
1752                        adds += added;
1753                        dels += deleted;
1754                }
1755        }
1756        fprintf(options->file, "%s", diff_line_prefix(options));
1757        print_stat_summary(options->file, total_files, adds, dels);
1758}
1759
1760static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1761{
1762        int i;
1763
1764        if (data->nr == 0)
1765                return;
1766
1767        for (i = 0; i < data->nr; i++) {
1768                struct diffstat_file *file = data->files[i];
1769
1770                fprintf(options->file, "%s", diff_line_prefix(options));
1771
1772                if (file->is_binary)
1773                        fprintf(options->file, "-\t-\t");
1774                else
1775                        fprintf(options->file,
1776                                "%"PRIuMAX"\t%"PRIuMAX"\t",
1777                                file->added, file->deleted);
1778                if (options->line_termination) {
1779                        fill_print_name(file);
1780                        if (!file->is_renamed)
1781                                write_name_quoted(file->name, options->file,
1782                                                  options->line_termination);
1783                        else {
1784                                fputs(file->print_name, options->file);
1785                                putc(options->line_termination, options->file);
1786                        }
1787                } else {
1788                        if (file->is_renamed) {
1789                                putc('\0', options->file);
1790                                write_name_quoted(file->from_name, options->file, '\0');
1791                        }
1792                        write_name_quoted(file->name, options->file, '\0');
1793                }
1794        }
1795}
1796
1797struct dirstat_file {
1798        const char *name;
1799        unsigned long changed;
1800};
1801
1802struct dirstat_dir {
1803        struct dirstat_file *files;
1804        int alloc, nr, permille, cumulative;
1805};
1806
1807static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1808                unsigned long changed, const char *base, int baselen)
1809{
1810        unsigned long this_dir = 0;
1811        unsigned int sources = 0;
1812        const char *line_prefix = diff_line_prefix(opt);
1813
1814        while (dir->nr) {
1815                struct dirstat_file *f = dir->files;
1816                int namelen = strlen(f->name);
1817                unsigned long this;
1818                char *slash;
1819
1820                if (namelen < baselen)
1821                        break;
1822                if (memcmp(f->name, base, baselen))
1823                        break;
1824                slash = strchr(f->name + baselen, '/');
1825                if (slash) {
1826                        int newbaselen = slash + 1 - f->name;
1827                        this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1828                        sources++;
1829                } else {
1830                        this = f->changed;
1831                        dir->files++;
1832                        dir->nr--;
1833                        sources += 2;
1834                }
1835                this_dir += this;
1836        }
1837
1838        /*
1839         * We don't report dirstat's for
1840         *  - the top level
1841         *  - or cases where everything came from a single directory
1842         *    under this directory (sources == 1).
1843         */
1844        if (baselen && sources != 1) {
1845                if (this_dir) {
1846                        int permille = this_dir * 1000 / changed;
1847                        if (permille >= dir->permille) {
1848                                fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1849                                        permille / 10, permille % 10, baselen, base);
1850                                if (!dir->cumulative)
1851                                        return 0;
1852                        }
1853                }
1854        }
1855        return this_dir;
1856}
1857
1858static int dirstat_compare(const void *_a, const void *_b)
1859{
1860        const struct dirstat_file *a = _a;
1861        const struct dirstat_file *b = _b;
1862        return strcmp(a->name, b->name);
1863}
1864
1865static void show_dirstat(struct diff_options *options)
1866{
1867        int i;
1868        unsigned long changed;
1869        struct dirstat_dir dir;
1870        struct diff_queue_struct *q = &diff_queued_diff;
1871
1872        dir.files = NULL;
1873        dir.alloc = 0;
1874        dir.nr = 0;
1875        dir.permille = options->dirstat_permille;
1876        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1877
1878        changed = 0;
1879        for (i = 0; i < q->nr; i++) {
1880                struct diff_filepair *p = q->queue[i];
1881                const char *name;
1882                unsigned long copied, added, damage;
1883                int content_changed;
1884
1885                name = p->two->path ? p->two->path : p->one->path;
1886
1887                if (p->one->sha1_valid && p->two->sha1_valid)
1888                        content_changed = hashcmp(p->one->sha1, p->two->sha1);
1889                else
1890                        content_changed = 1;
1891
1892                if (!content_changed) {
1893                        /*
1894                         * The SHA1 has not changed, so pre-/post-content is
1895                         * identical. We can therefore skip looking at the
1896                         * file contents altogether.
1897                         */
1898                        damage = 0;
1899                        goto found_damage;
1900                }
1901
1902                if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1903                        /*
1904                         * In --dirstat-by-file mode, we don't really need to
1905                         * look at the actual file contents at all.
1906                         * The fact that the SHA1 changed is enough for us to
1907                         * add this file to the list of results
1908                         * (with each file contributing equal damage).
1909                         */
1910                        damage = 1;
1911                        goto found_damage;
1912                }
1913
1914                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1915                        diff_populate_filespec(p->one, 0);
1916                        diff_populate_filespec(p->two, 0);
1917                        diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1918                                               &copied, &added);
1919                        diff_free_filespec_data(p->one);
1920                        diff_free_filespec_data(p->two);
1921                } else if (DIFF_FILE_VALID(p->one)) {
1922                        diff_populate_filespec(p->one, 1);
1923                        copied = added = 0;
1924                        diff_free_filespec_data(p->one);
1925                } else if (DIFF_FILE_VALID(p->two)) {
1926                        diff_populate_filespec(p->two, 1);
1927                        copied = 0;
1928                        added = p->two->size;
1929                        diff_free_filespec_data(p->two);
1930                } else
1931                        continue;
1932
1933                /*
1934                 * Original minus copied is the removed material,
1935                 * added is the new material.  They are both damages
1936                 * made to the preimage.
1937                 * If the resulting damage is zero, we know that
1938                 * diffcore_count_changes() considers the two entries to
1939                 * be identical, but since content_changed is true, we
1940                 * know that there must have been _some_ kind of change,
1941                 * so we force all entries to have damage > 0.
1942                 */
1943                damage = (p->one->size - copied) + added;
1944                if (!damage)
1945                        damage = 1;
1946
1947found_damage:
1948                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1949                dir.files[dir.nr].name = name;
1950                dir.files[dir.nr].changed = damage;
1951                changed += damage;
1952                dir.nr++;
1953        }
1954
1955        /* This can happen even with many files, if everything was renames */
1956        if (!changed)
1957                return;
1958
1959        /* Show all directories with more than x% of the changes */
1960        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1961        gather_dirstat(options, &dir, changed, "", 0);
1962}
1963
1964static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1965{
1966        int i;
1967        unsigned long changed;
1968        struct dirstat_dir dir;
1969
1970        if (data->nr == 0)
1971                return;
1972
1973        dir.files = NULL;
1974        dir.alloc = 0;
1975        dir.nr = 0;
1976        dir.permille = options->dirstat_permille;
1977        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1978
1979        changed = 0;
1980        for (i = 0; i < data->nr; i++) {
1981                struct diffstat_file *file = data->files[i];
1982                unsigned long damage = file->added + file->deleted;
1983                if (file->is_binary)
1984                        /*
1985                         * binary files counts bytes, not lines. Must find some
1986                         * way to normalize binary bytes vs. textual lines.
1987                         * The following heuristic assumes that there are 64
1988                         * bytes per "line".
1989                         * This is stupid and ugly, but very cheap...
1990                         */
1991                        damage = (damage + 63) / 64;
1992                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1993                dir.files[dir.nr].name = file->name;
1994                dir.files[dir.nr].changed = damage;
1995                changed += damage;
1996                dir.nr++;
1997        }
1998
1999        /* This can happen even with many files, if everything was renames */
2000        if (!changed)
2001                return;
2002
2003        /* Show all directories with more than x% of the changes */
2004        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2005        gather_dirstat(options, &dir, changed, "", 0);
2006}
2007
2008static void free_diffstat_info(struct diffstat_t *diffstat)
2009{
2010        int i;
2011        for (i = 0; i < diffstat->nr; i++) {
2012                struct diffstat_file *f = diffstat->files[i];
2013                if (f->name != f->print_name)
2014                        free(f->print_name);
2015                free(f->name);
2016                free(f->from_name);
2017                free(f);
2018        }
2019        free(diffstat->files);
2020}
2021
2022struct checkdiff_t {
2023        const char *filename;
2024        int lineno;
2025        int conflict_marker_size;
2026        struct diff_options *o;
2027        unsigned ws_rule;
2028        unsigned status;
2029};
2030
2031static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2032{
2033        char firstchar;
2034        int cnt;
2035
2036        if (len < marker_size + 1)
2037                return 0;
2038        firstchar = line[0];
2039        switch (firstchar) {
2040        case '=': case '>': case '<': case '|':
2041                break;
2042        default:
2043                return 0;
2044        }
2045        for (cnt = 1; cnt < marker_size; cnt++)
2046                if (line[cnt] != firstchar)
2047                        return 0;
2048        /* line[1] thru line[marker_size-1] are same as firstchar */
2049        if (len < marker_size + 1 || !isspace(line[marker_size]))
2050                return 0;
2051        return 1;
2052}
2053
2054static void checkdiff_consume(void *priv, char *line, unsigned long len)
2055{
2056        struct checkdiff_t *data = priv;
2057        int marker_size = data->conflict_marker_size;
2058        const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2059        const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2060        const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2061        char *err;
2062        const char *line_prefix;
2063
2064        assert(data->o);
2065        line_prefix = diff_line_prefix(data->o);
2066
2067        if (line[0] == '+') {
2068                unsigned bad;
2069                data->lineno++;
2070                if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2071                        data->status |= 1;
2072                        fprintf(data->o->file,
2073                                "%s%s:%d: leftover conflict marker\n",
2074                                line_prefix, data->filename, data->lineno);
2075                }
2076                bad = ws_check(line + 1, len - 1, data->ws_rule);
2077                if (!bad)
2078                        return;
2079                data->status |= bad;
2080                err = whitespace_error_string(bad);
2081                fprintf(data->o->file, "%s%s:%d: %s.\n",
2082                        line_prefix, data->filename, data->lineno, err);
2083                free(err);
2084                emit_line(data->o, set, reset, line, 1);
2085                ws_check_emit(line + 1, len - 1, data->ws_rule,
2086                              data->o->file, set, reset, ws);
2087        } else if (line[0] == ' ') {
2088                data->lineno++;
2089        } else if (line[0] == '@') {
2090                char *plus = strchr(line, '+');
2091                if (plus)
2092                        data->lineno = strtol(plus, NULL, 10) - 1;
2093                else
2094                        die("invalid diff");
2095        }
2096}
2097
2098static unsigned char *deflate_it(char *data,
2099                                 unsigned long size,
2100                                 unsigned long *result_size)
2101{
2102        int bound;
2103        unsigned char *deflated;
2104        git_zstream stream;
2105
2106        memset(&stream, 0, sizeof(stream));
2107        git_deflate_init(&stream, zlib_compression_level);
2108        bound = git_deflate_bound(&stream, size);
2109        deflated = xmalloc(bound);
2110        stream.next_out = deflated;
2111        stream.avail_out = bound;
2112
2113        stream.next_in = (unsigned char *)data;
2114        stream.avail_in = size;
2115        while (git_deflate(&stream, Z_FINISH) == Z_OK)
2116                ; /* nothing */
2117        git_deflate_end(&stream);
2118        *result_size = stream.total_out;
2119        return deflated;
2120}
2121
2122static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2123                                  const char *prefix)
2124{
2125        void *cp;
2126        void *delta;
2127        void *deflated;
2128        void *data;
2129        unsigned long orig_size;
2130        unsigned long delta_size;
2131        unsigned long deflate_size;
2132        unsigned long data_size;
2133
2134        /* We could do deflated delta, or we could do just deflated two,
2135         * whichever is smaller.
2136         */
2137        delta = NULL;
2138        deflated = deflate_it(two->ptr, two->size, &deflate_size);
2139        if (one->size && two->size) {
2140                delta = diff_delta(one->ptr, one->size,
2141                                   two->ptr, two->size,
2142                                   &delta_size, deflate_size);
2143                if (delta) {
2144                        void *to_free = delta;
2145                        orig_size = delta_size;
2146                        delta = deflate_it(delta, delta_size, &delta_size);
2147                        free(to_free);
2148                }
2149        }
2150
2151        if (delta && delta_size < deflate_size) {
2152                fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2153                free(deflated);
2154                data = delta;
2155                data_size = delta_size;
2156        }
2157        else {
2158                fprintf(file, "%sliteral %lu\n", prefix, two->size);
2159                free(delta);
2160                data = deflated;
2161                data_size = deflate_size;
2162        }
2163
2164        /* emit data encoded in base85 */
2165        cp = data;
2166        while (data_size) {
2167                int bytes = (52 < data_size) ? 52 : data_size;
2168                char line[70];
2169                data_size -= bytes;
2170                if (bytes <= 26)
2171                        line[0] = bytes + 'A' - 1;
2172                else
2173                        line[0] = bytes - 26 + 'a' - 1;
2174                encode_85(line + 1, cp, bytes);
2175                cp = (char *) cp + bytes;
2176                fprintf(file, "%s", prefix);
2177                fputs(line, file);
2178                fputc('\n', file);
2179        }
2180        fprintf(file, "%s\n", prefix);
2181        free(data);
2182}
2183
2184static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2185                             const char *prefix)
2186{
2187        fprintf(file, "%sGIT binary patch\n", prefix);
2188        emit_binary_diff_body(file, one, two, prefix);
2189        emit_binary_diff_body(file, two, one, prefix);
2190}
2191
2192int diff_filespec_is_binary(struct diff_filespec *one)
2193{
2194        if (one->is_binary == -1) {
2195                diff_filespec_load_driver(one);
2196                if (one->driver->binary != -1)
2197                        one->is_binary = one->driver->binary;
2198                else {
2199                        if (!one->data && DIFF_FILE_VALID(one))
2200                                diff_populate_filespec(one, 0);
2201                        if (one->data)
2202                                one->is_binary = buffer_is_binary(one->data,
2203                                                one->size);
2204                        if (one->is_binary == -1)
2205                                one->is_binary = 0;
2206                }
2207        }
2208        return one->is_binary;
2209}
2210
2211static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2212{
2213        diff_filespec_load_driver(one);
2214        return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2215}
2216
2217void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2218{
2219        if (!options->a_prefix)
2220                options->a_prefix = a;
2221        if (!options->b_prefix)
2222                options->b_prefix = b;
2223}
2224
2225struct userdiff_driver *get_textconv(struct diff_filespec *one)
2226{
2227        if (!DIFF_FILE_VALID(one))
2228                return NULL;
2229
2230        diff_filespec_load_driver(one);
2231        return userdiff_get_textconv(one->driver);
2232}
2233
2234static void builtin_diff(const char *name_a,
2235                         const char *name_b,
2236                         struct diff_filespec *one,
2237                         struct diff_filespec *two,
2238                         const char *xfrm_msg,
2239                         int must_show_header,
2240                         struct diff_options *o,
2241                         int complete_rewrite)
2242{
2243        mmfile_t mf1, mf2;
2244        const char *lbl[2];
2245        char *a_one, *b_two;
2246        const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2247        const char *reset = diff_get_color_opt(o, DIFF_RESET);
2248        const char *a_prefix, *b_prefix;
2249        struct userdiff_driver *textconv_one = NULL;
2250        struct userdiff_driver *textconv_two = NULL;
2251        struct strbuf header = STRBUF_INIT;
2252        const char *line_prefix = diff_line_prefix(o);
2253
2254        if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2255                        (!one->mode || S_ISGITLINK(one->mode)) &&
2256                        (!two->mode || S_ISGITLINK(two->mode))) {
2257                const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2258                const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2259                show_submodule_summary(o->file, one->path ? one->path : two->path,
2260                                line_prefix,
2261                                one->sha1, two->sha1, two->dirty_submodule,
2262                                meta, del, add, reset);
2263                return;
2264        }
2265
2266        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2267                textconv_one = get_textconv(one);
2268                textconv_two = get_textconv(two);
2269        }
2270
2271        diff_set_mnemonic_prefix(o, "a/", "b/");
2272        if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2273                a_prefix = o->b_prefix;
2274                b_prefix = o->a_prefix;
2275        } else {
2276                a_prefix = o->a_prefix;
2277                b_prefix = o->b_prefix;
2278        }
2279
2280        /* Never use a non-valid filename anywhere if at all possible */
2281        name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2282        name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2283
2284        a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2285        b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2286        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2287        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2288        strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2289        if (lbl[0][0] == '/') {
2290                /* /dev/null */
2291                strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2292                if (xfrm_msg)
2293                        strbuf_addstr(&header, xfrm_msg);
2294                must_show_header = 1;
2295        }
2296        else if (lbl[1][0] == '/') {
2297                strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2298                if (xfrm_msg)
2299                        strbuf_addstr(&header, xfrm_msg);
2300                must_show_header = 1;
2301        }
2302        else {
2303                if (one->mode != two->mode) {
2304                        strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2305                        strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2306                        must_show_header = 1;
2307                }
2308                if (xfrm_msg)
2309                        strbuf_addstr(&header, xfrm_msg);
2310
2311                /*
2312                 * we do not run diff between different kind
2313                 * of objects.
2314                 */
2315                if ((one->mode ^ two->mode) & S_IFMT)
2316                        goto free_ab_and_return;
2317                if (complete_rewrite &&
2318                    (textconv_one || !diff_filespec_is_binary(one)) &&
2319                    (textconv_two || !diff_filespec_is_binary(two))) {
2320                        fprintf(o->file, "%s", header.buf);
2321                        strbuf_reset(&header);
2322                        emit_rewrite_diff(name_a, name_b, one, two,
2323                                                textconv_one, textconv_two, o);
2324                        o->found_changes = 1;
2325                        goto free_ab_and_return;
2326                }
2327        }
2328
2329        if (o->irreversible_delete && lbl[1][0] == '/') {
2330                fprintf(o->file, "%s", header.buf);
2331                strbuf_reset(&header);
2332                goto free_ab_and_return;
2333        } else if (!DIFF_OPT_TST(o, TEXT) &&
2334            ( (!textconv_one && diff_filespec_is_binary(one)) ||
2335              (!textconv_two && diff_filespec_is_binary(two)) )) {
2336                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2337                        die("unable to read files to diff");
2338                /* Quite common confusing case */
2339                if (mf1.size == mf2.size &&
2340                    !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2341                        if (must_show_header)
2342                                fprintf(o->file, "%s", header.buf);
2343                        goto free_ab_and_return;
2344                }
2345                fprintf(o->file, "%s", header.buf);
2346                strbuf_reset(&header);
2347                if (DIFF_OPT_TST(o, BINARY))
2348                        emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2349                else
2350                        fprintf(o->file, "%sBinary files %s and %s differ\n",
2351                                line_prefix, lbl[0], lbl[1]);
2352                o->found_changes = 1;
2353        } else {
2354                /* Crazy xdl interfaces.. */
2355                const char *diffopts = getenv("GIT_DIFF_OPTS");
2356                xpparam_t xpp;
2357                xdemitconf_t xecfg;
2358                struct emit_callback ecbdata;
2359                const struct userdiff_funcname *pe;
2360
2361                if (must_show_header) {
2362                        fprintf(o->file, "%s", header.buf);
2363                        strbuf_reset(&header);
2364                }
2365
2366                mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2367                mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2368
2369                pe = diff_funcname_pattern(one);
2370                if (!pe)
2371                        pe = diff_funcname_pattern(two);
2372
2373                memset(&xpp, 0, sizeof(xpp));
2374                memset(&xecfg, 0, sizeof(xecfg));
2375                memset(&ecbdata, 0, sizeof(ecbdata));
2376                ecbdata.label_path = lbl;
2377                ecbdata.color_diff = want_color(o->use_color);
2378                ecbdata.found_changesp = &o->found_changes;
2379                ecbdata.ws_rule = whitespace_rule(name_b);
2380                if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2381                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2382                ecbdata.opt = o;
2383                ecbdata.header = header.len ? &header : NULL;
2384                xpp.flags = o->xdl_opts;
2385                xecfg.ctxlen = o->context;
2386                xecfg.interhunkctxlen = o->interhunkcontext;
2387                xecfg.flags = XDL_EMIT_FUNCNAMES;
2388                if (DIFF_OPT_TST(o, FUNCCONTEXT))
2389                        xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2390                if (pe)
2391                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2392                if (!diffopts)
2393                        ;
2394                else if (starts_with(diffopts, "--unified="))
2395                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2396                else if (starts_with(diffopts, "-u"))
2397                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2398                if (o->word_diff)
2399                        init_diff_words_data(&ecbdata, o, one, two);
2400                xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2401                              &xpp, &xecfg);
2402                if (o->word_diff)
2403                        free_diff_words_data(&ecbdata);
2404                if (textconv_one)
2405                        free(mf1.ptr);
2406                if (textconv_two)
2407                        free(mf2.ptr);
2408                xdiff_clear_find_func(&xecfg);
2409        }
2410
2411 free_ab_and_return:
2412        strbuf_release(&header);
2413        diff_free_filespec_data(one);
2414        diff_free_filespec_data(two);
2415        free(a_one);
2416        free(b_two);
2417        return;
2418}
2419
2420static void builtin_diffstat(const char *name_a, const char *name_b,
2421                             struct diff_filespec *one,
2422                             struct diff_filespec *two,
2423                             struct diffstat_t *diffstat,
2424                             struct diff_options *o,
2425                             struct diff_filepair *p)
2426{
2427        mmfile_t mf1, mf2;
2428        struct diffstat_file *data;
2429        int same_contents;
2430        int complete_rewrite = 0;
2431
2432        if (!DIFF_PAIR_UNMERGED(p)) {
2433                if (p->status == DIFF_STATUS_MODIFIED && p->score)
2434                        complete_rewrite = 1;
2435        }
2436
2437        data = diffstat_add(diffstat, name_a, name_b);
2438        data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2439
2440        if (!one || !two) {
2441                data->is_unmerged = 1;
2442                return;
2443        }
2444
2445        same_contents = !hashcmp(one->sha1, two->sha1);
2446
2447        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2448                data->is_binary = 1;
2449                if (same_contents) {
2450                        data->added = 0;
2451                        data->deleted = 0;
2452                } else {
2453                        data->added = diff_filespec_size(two);
2454                        data->deleted = diff_filespec_size(one);
2455                }
2456        }
2457
2458        else if (complete_rewrite) {
2459                diff_populate_filespec(one, 0);
2460                diff_populate_filespec(two, 0);
2461                data->deleted = count_lines(one->data, one->size);
2462                data->added = count_lines(two->data, two->size);
2463        }
2464
2465        else if (!same_contents) {
2466                /* Crazy xdl interfaces.. */
2467                xpparam_t xpp;
2468                xdemitconf_t xecfg;
2469
2470                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2471                        die("unable to read files to diff");
2472
2473                memset(&xpp, 0, sizeof(xpp));
2474                memset(&xecfg, 0, sizeof(xecfg));
2475                xpp.flags = o->xdl_opts;
2476                xecfg.ctxlen = o->context;
2477                xecfg.interhunkctxlen = o->interhunkcontext;
2478                xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2479                              &xpp, &xecfg);
2480        }
2481
2482        diff_free_filespec_data(one);
2483        diff_free_filespec_data(two);
2484}
2485
2486static void builtin_checkdiff(const char *name_a, const char *name_b,
2487                              const char *attr_path,
2488                              struct diff_filespec *one,
2489                              struct diff_filespec *two,
2490                              struct diff_options *o)
2491{
2492        mmfile_t mf1, mf2;
2493        struct checkdiff_t data;
2494
2495        if (!two)
2496                return;
2497
2498        memset(&data, 0, sizeof(data));
2499        data.filename = name_b ? name_b : name_a;
2500        data.lineno = 0;
2501        data.o = o;
2502        data.ws_rule = whitespace_rule(attr_path);
2503        data.conflict_marker_size = ll_merge_marker_size(attr_path);
2504
2505        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2506                die("unable to read files to diff");
2507
2508        /*
2509         * All the other codepaths check both sides, but not checking
2510         * the "old" side here is deliberate.  We are checking the newly
2511         * introduced changes, and as long as the "new" side is text, we
2512         * can and should check what it introduces.
2513         */
2514        if (diff_filespec_is_binary(two))
2515                goto free_and_return;
2516        else {
2517                /* Crazy xdl interfaces.. */
2518                xpparam_t xpp;
2519                xdemitconf_t xecfg;
2520
2521                memset(&xpp, 0, sizeof(xpp));
2522                memset(&xecfg, 0, sizeof(xecfg));
2523                xecfg.ctxlen = 1; /* at least one context line */
2524                xpp.flags = 0;
2525                xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2526                              &xpp, &xecfg);
2527
2528                if (data.ws_rule & WS_BLANK_AT_EOF) {
2529                        struct emit_callback ecbdata;
2530                        int blank_at_eof;
2531
2532                        ecbdata.ws_rule = data.ws_rule;
2533                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2534                        blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2535
2536                        if (blank_at_eof) {
2537                                static char *err;
2538                                if (!err)
2539                                        err = whitespace_error_string(WS_BLANK_AT_EOF);
2540                                fprintf(o->file, "%s:%d: %s.\n",
2541                                        data.filename, blank_at_eof, err);
2542                                data.status = 1; /* report errors */
2543                        }
2544                }
2545        }
2546 free_and_return:
2547        diff_free_filespec_data(one);
2548        diff_free_filespec_data(two);
2549        if (data.status)
2550                DIFF_OPT_SET(o, CHECK_FAILED);
2551}
2552
2553struct diff_filespec *alloc_filespec(const char *path)
2554{
2555        int namelen = strlen(path);
2556        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2557
2558        memset(spec, 0, sizeof(*spec));
2559        spec->path = (char *)(spec + 1);
2560        memcpy(spec->path, path, namelen+1);
2561        spec->count = 1;
2562        spec->is_binary = -1;
2563        return spec;
2564}
2565
2566void free_filespec(struct diff_filespec *spec)
2567{
2568        if (!--spec->count) {
2569                diff_free_filespec_data(spec);
2570                free(spec);
2571        }
2572}
2573
2574void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2575                   int sha1_valid, unsigned short mode)
2576{
2577        if (mode) {
2578                spec->mode = canon_mode(mode);
2579                hashcpy(spec->sha1, sha1);
2580                spec->sha1_valid = sha1_valid;
2581        }
2582}
2583
2584/*
2585 * Given a name and sha1 pair, if the index tells us the file in
2586 * the work tree has that object contents, return true, so that
2587 * prepare_temp_file() does not have to inflate and extract.
2588 */
2589static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2590{
2591        const struct cache_entry *ce;
2592        struct stat st;
2593        int pos, len;
2594
2595        /*
2596         * We do not read the cache ourselves here, because the
2597         * benchmark with my previous version that always reads cache
2598         * shows that it makes things worse for diff-tree comparing
2599         * two linux-2.6 kernel trees in an already checked out work
2600         * tree.  This is because most diff-tree comparisons deal with
2601         * only a small number of files, while reading the cache is
2602         * expensive for a large project, and its cost outweighs the
2603         * savings we get by not inflating the object to a temporary
2604         * file.  Practically, this code only helps when we are used
2605         * by diff-cache --cached, which does read the cache before
2606         * calling us.
2607         */
2608        if (!active_cache)
2609                return 0;
2610
2611        /* We want to avoid the working directory if our caller
2612         * doesn't need the data in a normal file, this system
2613         * is rather slow with its stat/open/mmap/close syscalls,
2614         * and the object is contained in a pack file.  The pack
2615         * is probably already open and will be faster to obtain
2616         * the data through than the working directory.  Loose
2617         * objects however would tend to be slower as they need
2618         * to be individually opened and inflated.
2619         */
2620        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2621                return 0;
2622
2623        len = strlen(name);
2624        pos = cache_name_pos(name, len);
2625        if (pos < 0)
2626                return 0;
2627        ce = active_cache[pos];
2628
2629        /*
2630         * This is not the sha1 we are looking for, or
2631         * unreusable because it is not a regular file.
2632         */
2633        if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2634                return 0;
2635
2636        /*
2637         * If ce is marked as "assume unchanged", there is no
2638         * guarantee that work tree matches what we are looking for.
2639         */
2640        if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2641                return 0;
2642
2643        /*
2644         * If ce matches the file in the work tree, we can reuse it.
2645         */
2646        if (ce_uptodate(ce) ||
2647            (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2648                return 1;
2649
2650        return 0;
2651}
2652
2653static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2654{
2655        int len;
2656        char *data = xmalloc(100), *dirty = "";
2657
2658        /* Are we looking at the work tree? */
2659        if (s->dirty_submodule)
2660                dirty = "-dirty";
2661
2662        len = snprintf(data, 100,
2663                       "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2664        s->data = data;
2665        s->size = len;
2666        s->should_free = 1;
2667        if (size_only) {
2668                s->data = NULL;
2669                free(data);
2670        }
2671        return 0;
2672}
2673
2674/*
2675 * While doing rename detection and pickaxe operation, we may need to
2676 * grab the data for the blob (or file) for our own in-core comparison.
2677 * diff_filespec has data and size fields for this purpose.
2678 */
2679int diff_populate_filespec(struct diff_filespec *s, int size_only)
2680{
2681        int err = 0;
2682        /*
2683         * demote FAIL to WARN to allow inspecting the situation
2684         * instead of refusing.
2685         */
2686        enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2687                                    ? SAFE_CRLF_WARN
2688                                    : safe_crlf);
2689
2690        if (!DIFF_FILE_VALID(s))
2691                die("internal error: asking to populate invalid file.");
2692        if (S_ISDIR(s->mode))
2693                return -1;
2694
2695        if (s->data)
2696                return 0;
2697
2698        if (size_only && 0 < s->size)
2699                return 0;
2700
2701        if (S_ISGITLINK(s->mode))
2702                return diff_populate_gitlink(s, size_only);
2703
2704        if (!s->sha1_valid ||
2705            reuse_worktree_file(s->path, s->sha1, 0)) {
2706                struct strbuf buf = STRBUF_INIT;
2707                struct stat st;
2708                int fd;
2709
2710                if (lstat(s->path, &st) < 0) {
2711                        if (errno == ENOENT) {
2712                        err_empty:
2713                                err = -1;
2714                        empty:
2715                                s->data = (char *)"";
2716                                s->size = 0;
2717                                return err;
2718                        }
2719                }
2720                s->size = xsize_t(st.st_size);
2721                if (!s->size)
2722                        goto empty;
2723                if (S_ISLNK(st.st_mode)) {
2724                        struct strbuf sb = STRBUF_INIT;
2725
2726                        if (strbuf_readlink(&sb, s->path, s->size))
2727                                goto err_empty;
2728                        s->size = sb.len;
2729                        s->data = strbuf_detach(&sb, NULL);
2730                        s->should_free = 1;
2731                        return 0;
2732                }
2733                if (size_only)
2734                        return 0;
2735                fd = open(s->path, O_RDONLY);
2736                if (fd < 0)
2737                        goto err_empty;
2738                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2739                close(fd);
2740                s->should_munmap = 1;
2741
2742                /*
2743                 * Convert from working tree format to canonical git format
2744                 */
2745                if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2746                        size_t size = 0;
2747                        munmap(s->data, s->size);
2748                        s->should_munmap = 0;
2749                        s->data = strbuf_detach(&buf, &size);
2750                        s->size = size;
2751                        s->should_free = 1;
2752                }
2753        }
2754        else {
2755                enum object_type type;
2756                if (size_only) {
2757                        type = sha1_object_info(s->sha1, &s->size);
2758                        if (type < 0)
2759                                die("unable to read %s", sha1_to_hex(s->sha1));
2760                } else {
2761                        s->data = read_sha1_file(s->sha1, &type, &s->size);
2762                        if (!s->data)
2763                                die("unable to read %s", sha1_to_hex(s->sha1));
2764                        s->should_free = 1;
2765                }
2766        }
2767        return 0;
2768}
2769
2770void diff_free_filespec_blob(struct diff_filespec *s)
2771{
2772        if (s->should_free)
2773                free(s->data);
2774        else if (s->should_munmap)
2775                munmap(s->data, s->size);
2776
2777        if (s->should_free || s->should_munmap) {
2778                s->should_free = s->should_munmap = 0;
2779                s->data = NULL;
2780        }
2781}
2782
2783void diff_free_filespec_data(struct diff_filespec *s)
2784{
2785        diff_free_filespec_blob(s);
2786        free(s->cnt_data);
2787        s->cnt_data = NULL;
2788}
2789
2790static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2791                           void *blob,
2792                           unsigned long size,
2793                           const unsigned char *sha1,
2794                           int mode)
2795{
2796        int fd;
2797        struct strbuf buf = STRBUF_INIT;
2798        struct strbuf template = STRBUF_INIT;
2799        char *path_dup = xstrdup(path);
2800        const char *base = basename(path_dup);
2801
2802        /* Generate "XXXXXX_basename.ext" */
2803        strbuf_addstr(&template, "XXXXXX_");
2804        strbuf_addstr(&template, base);
2805
2806        fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2807                        strlen(base) + 1);
2808        if (fd < 0)
2809                die_errno("unable to create temp-file");
2810        if (convert_to_working_tree(path,
2811                        (const char *)blob, (size_t)size, &buf)) {
2812                blob = buf.buf;
2813                size = buf.len;
2814        }
2815        if (write_in_full(fd, blob, size) != size)
2816                die_errno("unable to write temp-file");
2817        close(fd);
2818        temp->name = temp->tmp_path;
2819        strcpy(temp->hex, sha1_to_hex(sha1));
2820        temp->hex[40] = 0;
2821        sprintf(temp->mode, "%06o", mode);
2822        strbuf_release(&buf);
2823        strbuf_release(&template);
2824        free(path_dup);
2825}
2826
2827static struct diff_tempfile *prepare_temp_file(const char *name,
2828                struct diff_filespec *one)
2829{
2830        struct diff_tempfile *temp = claim_diff_tempfile();
2831
2832        if (!DIFF_FILE_VALID(one)) {
2833        not_a_valid_file:
2834                /* A '-' entry produces this for file-2, and
2835                 * a '+' entry produces this for file-1.
2836                 */
2837                temp->name = "/dev/null";
2838                strcpy(temp->hex, ".");
2839                strcpy(temp->mode, ".");
2840                return temp;
2841        }
2842
2843        if (!remove_tempfile_installed) {
2844                atexit(remove_tempfile);
2845                sigchain_push_common(remove_tempfile_on_signal);
2846                remove_tempfile_installed = 1;
2847        }
2848
2849        if (!S_ISGITLINK(one->mode) &&
2850            (!one->sha1_valid ||
2851             reuse_worktree_file(name, one->sha1, 1))) {
2852                struct stat st;
2853                if (lstat(name, &st) < 0) {
2854                        if (errno == ENOENT)
2855                                goto not_a_valid_file;
2856                        die_errno("stat(%s)", name);
2857                }
2858                if (S_ISLNK(st.st_mode)) {
2859                        struct strbuf sb = STRBUF_INIT;
2860                        if (strbuf_readlink(&sb, name, st.st_size) < 0)
2861                                die_errno("readlink(%s)", name);
2862                        prep_temp_blob(name, temp, sb.buf, sb.len,
2863                                       (one->sha1_valid ?
2864                                        one->sha1 : null_sha1),
2865                                       (one->sha1_valid ?
2866                                        one->mode : S_IFLNK));
2867                        strbuf_release(&sb);
2868                }
2869                else {
2870                        /* we can borrow from the file in the work tree */
2871                        temp->name = name;
2872                        if (!one->sha1_valid)
2873                                strcpy(temp->hex, sha1_to_hex(null_sha1));
2874                        else
2875                                strcpy(temp->hex, sha1_to_hex(one->sha1));
2876                        /* Even though we may sometimes borrow the
2877                         * contents from the work tree, we always want
2878                         * one->mode.  mode is trustworthy even when
2879                         * !(one->sha1_valid), as long as
2880                         * DIFF_FILE_VALID(one).
2881                         */
2882                        sprintf(temp->mode, "%06o", one->mode);
2883                }
2884                return temp;
2885        }
2886        else {
2887                if (diff_populate_filespec(one, 0))
2888                        die("cannot read data blob for %s", one->path);
2889                prep_temp_blob(name, temp, one->data, one->size,
2890                               one->sha1, one->mode);
2891        }
2892        return temp;
2893}
2894
2895/* An external diff command takes:
2896 *
2897 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2898 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2899 *
2900 */
2901static void run_external_diff(const char *pgm,
2902                              const char *name,
2903                              const char *other,
2904                              struct diff_filespec *one,
2905                              struct diff_filespec *two,
2906                              const char *xfrm_msg,
2907                              int complete_rewrite,
2908                              struct diff_options *o)
2909{
2910        struct argv_array argv = ARGV_ARRAY_INIT;
2911        struct argv_array env = ARGV_ARRAY_INIT;
2912        struct diff_queue_struct *q = &diff_queued_diff;
2913
2914        argv_array_push(&argv, pgm);
2915        argv_array_push(&argv, name);
2916
2917        if (one && two) {
2918                struct diff_tempfile *temp_one, *temp_two;
2919                const char *othername = (other ? other : name);
2920                temp_one = prepare_temp_file(name, one);
2921                temp_two = prepare_temp_file(othername, two);
2922                argv_array_push(&argv, temp_one->name);
2923                argv_array_push(&argv, temp_one->hex);
2924                argv_array_push(&argv, temp_one->mode);
2925                argv_array_push(&argv, temp_two->name);
2926                argv_array_push(&argv, temp_two->hex);
2927                argv_array_push(&argv, temp_two->mode);
2928                if (other) {
2929                        argv_array_push(&argv, other);
2930                        argv_array_push(&argv, xfrm_msg);
2931                }
2932        }
2933
2934        argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2935        argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2936
2937        if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
2938                die(_("external diff died, stopping at %s"), name);
2939
2940        remove_tempfile();
2941        argv_array_clear(&argv);
2942        argv_array_clear(&env);
2943}
2944
2945static int similarity_index(struct diff_filepair *p)
2946{
2947        return p->score * 100 / MAX_SCORE;
2948}
2949
2950static void fill_metainfo(struct strbuf *msg,
2951                          const char *name,
2952                          const char *other,
2953                          struct diff_filespec *one,
2954                          struct diff_filespec *two,
2955                          struct diff_options *o,
2956                          struct diff_filepair *p,
2957                          int *must_show_header,
2958                          int use_color)
2959{
2960        const char *set = diff_get_color(use_color, DIFF_METAINFO);
2961        const char *reset = diff_get_color(use_color, DIFF_RESET);
2962        const char *line_prefix = diff_line_prefix(o);
2963
2964        *must_show_header = 1;
2965        strbuf_init(msg, PATH_MAX * 2 + 300);
2966        switch (p->status) {
2967        case DIFF_STATUS_COPIED:
2968                strbuf_addf(msg, "%s%ssimilarity index %d%%",
2969                            line_prefix, set, similarity_index(p));
2970                strbuf_addf(msg, "%s\n%s%scopy from ",
2971                            reset,  line_prefix, set);
2972                quote_c_style(name, msg, NULL, 0);
2973                strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2974                quote_c_style(other, msg, NULL, 0);
2975                strbuf_addf(msg, "%s\n", reset);
2976                break;
2977        case DIFF_STATUS_RENAMED:
2978                strbuf_addf(msg, "%s%ssimilarity index %d%%",
2979                            line_prefix, set, similarity_index(p));
2980                strbuf_addf(msg, "%s\n%s%srename from ",
2981                            reset, line_prefix, set);
2982                quote_c_style(name, msg, NULL, 0);
2983                strbuf_addf(msg, "%s\n%s%srename to ",
2984                            reset, line_prefix, set);
2985                quote_c_style(other, msg, NULL, 0);
2986                strbuf_addf(msg, "%s\n", reset);
2987                break;
2988        case DIFF_STATUS_MODIFIED:
2989                if (p->score) {
2990                        strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2991                                    line_prefix,
2992                                    set, similarity_index(p), reset);
2993                        break;
2994                }
2995                /* fallthru */
2996        default:
2997                *must_show_header = 0;
2998        }
2999        if (one && two && hashcmp(one->sha1, two->sha1)) {
3000                int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3001
3002                if (DIFF_OPT_TST(o, BINARY)) {
3003                        mmfile_t mf;
3004                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3005                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3006                                abbrev = 40;
3007                }
3008                strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3009                            find_unique_abbrev(one->sha1, abbrev));
3010                strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3011                if (one->mode == two->mode)
3012                        strbuf_addf(msg, " %06o", one->mode);
3013                strbuf_addf(msg, "%s\n", reset);
3014        }
3015}
3016
3017static void run_diff_cmd(const char *pgm,
3018                         const char *name,
3019                         const char *other,
3020                         const char *attr_path,
3021                         struct diff_filespec *one,
3022                         struct diff_filespec *two,
3023                         struct strbuf *msg,
3024                         struct diff_options *o,
3025                         struct diff_filepair *p)
3026{
3027        const char *xfrm_msg = NULL;
3028        int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3029        int must_show_header = 0;
3030
3031
3032        if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3033                struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3034                if (drv && drv->external)
3035                        pgm = drv->external;
3036        }
3037
3038        if (msg) {
3039                /*
3040                 * don't use colors when the header is intended for an
3041                 * external diff driver
3042                 */
3043                fill_metainfo(msg, name, other, one, two, o, p,
3044                              &must_show_header,
3045                              want_color(o->use_color) && !pgm);
3046                xfrm_msg = msg->len ? msg->buf : NULL;
3047        }
3048
3049        if (pgm) {
3050                run_external_diff(pgm, name, other, one, two, xfrm_msg,
3051                                  complete_rewrite, o);
3052                return;
3053        }
3054        if (one && two)
3055                builtin_diff(name, other ? other : name,
3056                             one, two, xfrm_msg, must_show_header,
3057                             o, complete_rewrite);
3058        else
3059                fprintf(o->file, "* Unmerged path %s\n", name);
3060}
3061
3062static void diff_fill_sha1_info(struct diff_filespec *one)
3063{
3064        if (DIFF_FILE_VALID(one)) {
3065                if (!one->sha1_valid) {
3066                        struct stat st;
3067                        if (one->is_stdin) {
3068                                hashcpy(one->sha1, null_sha1);
3069                                return;
3070                        }
3071                        if (lstat(one->path, &st) < 0)
3072                                die_errno("stat '%s'", one->path);
3073                        if (index_path(one->sha1, one->path, &st, 0))
3074                                die("cannot hash %s", one->path);
3075                }
3076        }
3077        else
3078                hashclr(one->sha1);
3079}
3080
3081static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3082{
3083        /* Strip the prefix but do not molest /dev/null and absolute paths */
3084        if (*namep && **namep != '/') {
3085                *namep += prefix_length;
3086                if (**namep == '/')
3087                        ++*namep;
3088        }
3089        if (*otherp && **otherp != '/') {
3090                *otherp += prefix_length;
3091                if (**otherp == '/')
3092                        ++*otherp;
3093        }
3094}
3095
3096static void run_diff(struct diff_filepair *p, struct diff_options *o)
3097{
3098        const char *pgm = external_diff();
3099        struct strbuf msg;
3100        struct diff_filespec *one = p->one;
3101        struct diff_filespec *two = p->two;
3102        const char *name;
3103        const char *other;
3104        const char *attr_path;
3105
3106        name  = p->one->path;
3107        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3108        attr_path = name;
3109        if (o->prefix_length)
3110                strip_prefix(o->prefix_length, &name, &other);
3111
3112        if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3113                pgm = NULL;
3114
3115        if (DIFF_PAIR_UNMERGED(p)) {
3116                run_diff_cmd(pgm, name, NULL, attr_path,
3117                             NULL, NULL, NULL, o, p);
3118                return;
3119        }
3120
3121        diff_fill_sha1_info(one);
3122        diff_fill_sha1_info(two);
3123
3124        if (!pgm &&
3125            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3126            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3127                /*
3128                 * a filepair that changes between file and symlink
3129                 * needs to be split into deletion and creation.
3130                 */
3131                struct diff_filespec *null = alloc_filespec(two->path);
3132                run_diff_cmd(NULL, name, other, attr_path,
3133                             one, null, &msg, o, p);
3134                free(null);
3135                strbuf_release(&msg);
3136
3137                null = alloc_filespec(one->path);
3138                run_diff_cmd(NULL, name, other, attr_path,
3139                             null, two, &msg, o, p);
3140                free(null);
3141        }
3142        else
3143                run_diff_cmd(pgm, name, other, attr_path,
3144                             one, two, &msg, o, p);
3145
3146        strbuf_release(&msg);
3147}
3148
3149static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3150                         struct diffstat_t *diffstat)
3151{
3152        const char *name;
3153        const char *other;
3154
3155        if (DIFF_PAIR_UNMERGED(p)) {
3156                /* unmerged */
3157                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3158                return;
3159        }
3160
3161        name = p->one->path;
3162        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3163
3164        if (o->prefix_length)
3165                strip_prefix(o->prefix_length, &name, &other);
3166
3167        diff_fill_sha1_info(p->one);
3168        diff_fill_sha1_info(p->two);
3169
3170        builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3171}
3172
3173static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3174{
3175        const char *name;
3176        const char *other;
3177        const char *attr_path;
3178
3179        if (DIFF_PAIR_UNMERGED(p)) {
3180                /* unmerged */
3181                return;
3182        }
3183
3184        name = p->one->path;
3185        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3186        attr_path = other ? other : name;
3187
3188        if (o->prefix_length)
3189                strip_prefix(o->prefix_length, &name, &other);
3190
3191        diff_fill_sha1_info(p->one);
3192        diff_fill_sha1_info(p->two);
3193
3194        builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3195}
3196
3197void diff_setup(struct diff_options *options)
3198{
3199        memcpy(options, &default_diff_options, sizeof(*options));
3200
3201        options->file = stdout;
3202
3203        options->line_termination = '\n';
3204        options->break_opt = -1;
3205        options->rename_limit = -1;
3206        options->dirstat_permille = diff_dirstat_permille_default;
3207        options->context = diff_context_default;
3208        DIFF_OPT_SET(options, RENAME_EMPTY);
3209
3210        options->change = diff_change;
3211        options->add_remove = diff_addremove;
3212        options->use_color = diff_use_color_default;
3213        options->detect_rename = diff_detect_rename_default;
3214        options->xdl_opts |= diff_algorithm;
3215
3216        options->orderfile = diff_order_file_cfg;
3217
3218        if (diff_no_prefix) {
3219                options->a_prefix = options->b_prefix = "";
3220        } else if (!diff_mnemonic_prefix) {
3221                options->a_prefix = "a/";
3222                options->b_prefix = "b/";
3223        }
3224}
3225
3226void diff_setup_done(struct diff_options *options)
3227{
3228        int count = 0;
3229
3230        if (options->set_default)
3231                options->set_default(options);
3232
3233        if (options->output_format & DIFF_FORMAT_NAME)
3234                count++;
3235        if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3236                count++;
3237        if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3238                count++;
3239        if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3240                count++;
3241        if (count > 1)
3242                die("--name-only, --name-status, --check and -s are mutually exclusive");
3243
3244        /*
3245         * Most of the time we can say "there are changes"
3246         * only by checking if there are changed paths, but
3247         * --ignore-whitespace* options force us to look
3248         * inside contents.
3249         */
3250
3251        if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3252            DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3253            DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3254                DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3255        else
3256                DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3257
3258        if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3259                options->detect_rename = DIFF_DETECT_COPY;
3260
3261        if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3262                options->prefix = NULL;
3263        if (options->prefix)
3264                options->prefix_length = strlen(options->prefix);
3265        else
3266                options->prefix_length = 0;
3267
3268        if (options->output_format & (DIFF_FORMAT_NAME |
3269                                      DIFF_FORMAT_NAME_STATUS |
3270                                      DIFF_FORMAT_CHECKDIFF |
3271                                      DIFF_FORMAT_NO_OUTPUT))
3272                options->output_format &= ~(DIFF_FORMAT_RAW |
3273                                            DIFF_FORMAT_NUMSTAT |
3274                                            DIFF_FORMAT_DIFFSTAT |
3275                                            DIFF_FORMAT_SHORTSTAT |
3276                                            DIFF_FORMAT_DIRSTAT |
3277                                            DIFF_FORMAT_SUMMARY |
3278                                            DIFF_FORMAT_PATCH);
3279
3280        /*
3281         * These cases always need recursive; we do not drop caller-supplied
3282         * recursive bits for other formats here.
3283         */
3284        if (options->output_format & (DIFF_FORMAT_PATCH |
3285                                      DIFF_FORMAT_NUMSTAT |
3286                                      DIFF_FORMAT_DIFFSTAT |
3287                                      DIFF_FORMAT_SHORTSTAT |
3288                                      DIFF_FORMAT_DIRSTAT |
3289                                      DIFF_FORMAT_SUMMARY |
3290                                      DIFF_FORMAT_CHECKDIFF))
3291                DIFF_OPT_SET(options, RECURSIVE);
3292        /*
3293         * Also pickaxe would not work very well if you do not say recursive
3294         */
3295        if (options->pickaxe)
3296                DIFF_OPT_SET(options, RECURSIVE);
3297        /*
3298         * When patches are generated, submodules diffed against the work tree
3299         * must be checked for dirtiness too so it can be shown in the output
3300         */
3301        if (options->output_format & DIFF_FORMAT_PATCH)
3302                DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3303
3304        if (options->detect_rename && options->rename_limit < 0)
3305                options->rename_limit = diff_rename_limit_default;
3306        if (options->setup & DIFF_SETUP_USE_CACHE) {
3307                if (!active_cache)
3308                        /* read-cache does not die even when it fails
3309                         * so it is safe for us to do this here.  Also
3310                         * it does not smudge active_cache or active_nr
3311                         * when it fails, so we do not have to worry about
3312                         * cleaning it up ourselves either.
3313                         */
3314                        read_cache();
3315        }
3316        if (options->abbrev <= 0 || 40 < options->abbrev)
3317                options->abbrev = 40; /* full */
3318
3319        /*
3320         * It does not make sense to show the first hit we happened
3321         * to have found.  It does not make sense not to return with
3322         * exit code in such a case either.
3323         */
3324        if (DIFF_OPT_TST(options, QUICK)) {
3325                options->output_format = DIFF_FORMAT_NO_OUTPUT;
3326                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3327        }
3328
3329        options->diff_path_counter = 0;
3330}
3331
3332static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3333{
3334        char c, *eq;
3335        int len;
3336
3337        if (*arg != '-')
3338                return 0;
3339        c = *++arg;
3340        if (!c)
3341                return 0;
3342        if (c == arg_short) {
3343                c = *++arg;
3344                if (!c)
3345                        return 1;
3346                if (val && isdigit(c)) {
3347                        char *end;
3348                        int n = strtoul(arg, &end, 10);
3349                        if (*end)
3350                                return 0;
3351                        *val = n;
3352                        return 1;
3353                }
3354                return 0;
3355        }
3356        if (c != '-')
3357                return 0;
3358        arg++;
3359        eq = strchr(arg, '=');
3360        if (eq)
3361                len = eq - arg;
3362        else
3363                len = strlen(arg);
3364        if (!len || strncmp(arg, arg_long, len))
3365                return 0;
3366        if (eq) {
3367                int n;
3368                char *end;
3369                if (!isdigit(*++eq))
3370                        return 0;
3371                n = strtoul(eq, &end, 10);
3372                if (*end)
3373                        return 0;
3374                *val = n;
3375        }
3376        return 1;
3377}
3378
3379static int diff_scoreopt_parse(const char *opt);
3380
3381static inline int short_opt(char opt, const char **argv,
3382                            const char **optarg)
3383{
3384        const char *arg = argv[0];
3385        if (arg[0] != '-' || arg[1] != opt)
3386                return 0;
3387        if (arg[2] != '\0') {
3388                *optarg = arg + 2;
3389                return 1;
3390        }
3391        if (!argv[1])
3392                die("Option '%c' requires a value", opt);
3393        *optarg = argv[1];
3394        return 2;
3395}
3396
3397int parse_long_opt(const char *opt, const char **argv,
3398                   const char **optarg)
3399{
3400        const char *arg = argv[0];
3401        if (arg[0] != '-' || arg[1] != '-')
3402                return 0;
3403        arg += strlen("--");
3404        if (!starts_with(arg, opt))
3405                return 0;
3406        arg += strlen(opt);
3407        if (*arg == '=') { /* stuck form: --option=value */
3408                *optarg = arg + 1;
3409                return 1;
3410        }
3411        if (*arg != '\0')
3412                return 0;
3413        /* separate form: --option value */
3414        if (!argv[1])
3415                die("Option '--%s' requires a value", opt);
3416        *optarg = argv[1];
3417        return 2;
3418}
3419
3420static int stat_opt(struct diff_options *options, const char **av)
3421{
3422        const char *arg = av[0];
3423        char *end;
3424        int width = options->stat_width;
3425        int name_width = options->stat_name_width;
3426        int graph_width = options->stat_graph_width;
3427        int count = options->stat_count;
3428        int argcount = 1;
3429
3430        arg += strlen("--stat");
3431        end = (char *)arg;
3432
3433        switch (*arg) {
3434        case '-':
3435                if (starts_with(arg, "-width")) {
3436                        arg += strlen("-width");
3437                        if (*arg == '=')
3438                                width = strtoul(arg + 1, &end, 10);
3439                        else if (!*arg && !av[1])
3440                                die("Option '--stat-width' requires a value");
3441                        else if (!*arg) {
3442                                width = strtoul(av[1], &end, 10);
3443                                argcount = 2;
3444                        }
3445                } else if (starts_with(arg, "-name-width")) {
3446                        arg += strlen("-name-width");
3447                        if (*arg == '=')
3448                                name_width = strtoul(arg + 1, &end, 10);
3449                        else if (!*arg && !av[1])
3450                                die("Option '--stat-name-width' requires a value");
3451                        else if (!*arg) {
3452                                name_width = strtoul(av[1], &end, 10);
3453                                argcount = 2;
3454                        }
3455                } else if (starts_with(arg, "-graph-width")) {
3456                        arg += strlen("-graph-width");
3457                        if (*arg == '=')
3458                                graph_width = strtoul(arg + 1, &end, 10);
3459                        else if (!*arg && !av[1])
3460                                die("Option '--stat-graph-width' requires a value");
3461                        else if (!*arg) {
3462                                graph_width = strtoul(av[1], &end, 10);
3463                                argcount = 2;
3464                        }
3465                } else if (starts_with(arg, "-count")) {
3466                        arg += strlen("-count");
3467                        if (*arg == '=')
3468                                count = strtoul(arg + 1, &end, 10);
3469                        else if (!*arg && !av[1])
3470                                die("Option '--stat-count' requires a value");
3471                        else if (!*arg) {
3472                                count = strtoul(av[1], &end, 10);
3473                                argcount = 2;
3474                        }
3475                }
3476                break;
3477        case '=':
3478                width = strtoul(arg+1, &end, 10);
3479                if (*end == ',')
3480                        name_width = strtoul(end+1, &end, 10);
3481                if (*end == ',')
3482                        count = strtoul(end+1, &end, 10);
3483        }
3484
3485        /* Important! This checks all the error cases! */
3486        if (*end)
3487                return 0;
3488        options->output_format |= DIFF_FORMAT_DIFFSTAT;
3489        options->stat_name_width = name_width;
3490        options->stat_graph_width = graph_width;
3491        options->stat_width = width;
3492        options->stat_count = count;
3493        return argcount;
3494}
3495
3496static int parse_dirstat_opt(struct diff_options *options, const char *params)
3497{
3498        struct strbuf errmsg = STRBUF_INIT;
3499        if (parse_dirstat_params(options, params, &errmsg))
3500                die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3501                    errmsg.buf);
3502        strbuf_release(&errmsg);
3503        /*
3504         * The caller knows a dirstat-related option is given from the command
3505         * line; allow it to say "return this_function();"
3506         */
3507        options->output_format |= DIFF_FORMAT_DIRSTAT;
3508        return 1;
3509}
3510
3511static int parse_submodule_opt(struct diff_options *options, const char *value)
3512{
3513        if (parse_submodule_params(options, value))
3514                die(_("Failed to parse --submodule option parameter: '%s'"),
3515                        value);
3516        return 1;
3517}
3518
3519static const char diff_status_letters[] = {
3520        DIFF_STATUS_ADDED,
3521        DIFF_STATUS_COPIED,
3522        DIFF_STATUS_DELETED,
3523        DIFF_STATUS_MODIFIED,
3524        DIFF_STATUS_RENAMED,
3525        DIFF_STATUS_TYPE_CHANGED,
3526        DIFF_STATUS_UNKNOWN,
3527        DIFF_STATUS_UNMERGED,
3528        DIFF_STATUS_FILTER_AON,
3529        DIFF_STATUS_FILTER_BROKEN,
3530        '\0',
3531};
3532
3533static unsigned int filter_bit['Z' + 1];
3534
3535static void prepare_filter_bits(void)
3536{
3537        int i;
3538
3539        if (!filter_bit[DIFF_STATUS_ADDED]) {
3540                for (i = 0; diff_status_letters[i]; i++)
3541                        filter_bit[(int) diff_status_letters[i]] = (1 << i);
3542        }
3543}
3544
3545static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3546{
3547        return opt->filter & filter_bit[(int) status];
3548}
3549
3550static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3551{
3552        int i, optch;
3553
3554        prepare_filter_bits();
3555
3556        /*
3557         * If there is a negation e.g. 'd' in the input, and we haven't
3558         * initialized the filter field with another --diff-filter, start
3559         * from full set of bits, except for AON.
3560         */
3561        if (!opt->filter) {
3562                for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3563                        if (optch < 'a' || 'z' < optch)
3564                                continue;
3565                        opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3566                        opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3567                        break;
3568                }
3569        }
3570
3571        for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3572                unsigned int bit;
3573                int negate;
3574
3575                if ('a' <= optch && optch <= 'z') {
3576                        negate = 1;
3577                        optch = toupper(optch);
3578                } else {
3579                        negate = 0;
3580                }
3581
3582                bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3583                if (!bit)
3584                        return optarg[i];
3585                if (negate)
3586                        opt->filter &= ~bit;
3587                else
3588                        opt->filter |= bit;
3589        }
3590        return 0;
3591}
3592
3593/* Used only by "diff-files" and "diff --no-index" */
3594void handle_deprecated_show_diff_q(struct diff_options *opt)
3595{
3596        warning("'diff -q' and 'diff-files -q' are deprecated.");
3597        warning("Use 'diff --diff-filter=d' instead to ignore deleted filepairs.");
3598        parse_diff_filter_opt("d", opt);
3599}
3600
3601static void enable_patch_output(int *fmt) {
3602        *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3603        *fmt |= DIFF_FORMAT_PATCH;
3604}
3605
3606int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3607{
3608        const char *arg = av[0];
3609        const char *optarg;
3610        int argcount;
3611
3612        /* Output format options */
3613        if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3614            || opt_arg(arg, 'U', "unified", &options->context))
3615                enable_patch_output(&options->output_format);
3616        else if (!strcmp(arg, "--raw"))
3617                options->output_format |= DIFF_FORMAT_RAW;
3618        else if (!strcmp(arg, "--patch-with-raw")) {
3619                enable_patch_output(&options->output_format);
3620                options->output_format |= DIFF_FORMAT_RAW;
3621        } else if (!strcmp(arg, "--numstat"))
3622                options->output_format |= DIFF_FORMAT_NUMSTAT;
3623        else if (!strcmp(arg, "--shortstat"))
3624                options->output_format |= DIFF_FORMAT_SHORTSTAT;
3625        else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3626                return parse_dirstat_opt(options, "");
3627        else if (starts_with(arg, "-X"))
3628                return parse_dirstat_opt(options, arg + 2);
3629        else if (starts_with(arg, "--dirstat="))
3630                return parse_dirstat_opt(options, arg + 10);
3631        else if (!strcmp(arg, "--cumulative"))
3632                return parse_dirstat_opt(options, "cumulative");
3633        else if (!strcmp(arg, "--dirstat-by-file"))
3634                return parse_dirstat_opt(options, "files");
3635        else if (starts_with(arg, "--dirstat-by-file=")) {
3636                parse_dirstat_opt(options, "files");
3637                return parse_dirstat_opt(options, arg + 18);
3638        }
3639        else if (!strcmp(arg, "--check"))
3640                options->output_format |= DIFF_FORMAT_CHECKDIFF;
3641        else if (!strcmp(arg, "--summary"))
3642                options->output_format |= DIFF_FORMAT_SUMMARY;
3643        else if (!strcmp(arg, "--patch-with-stat")) {
3644                enable_patch_output(&options->output_format);
3645                options->output_format |= DIFF_FORMAT_DIFFSTAT;
3646        } else if (!strcmp(arg, "--name-only"))
3647                options->output_format |= DIFF_FORMAT_NAME;
3648        else if (!strcmp(arg, "--name-status"))
3649                options->output_format |= DIFF_FORMAT_NAME_STATUS;
3650        else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3651                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3652        else if (starts_with(arg, "--stat"))
3653                /* --stat, --stat-width, --stat-name-width, or --stat-count */
3654                return stat_opt(options, av);
3655
3656        /* renames options */
3657        else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3658                 !strcmp(arg, "--break-rewrites")) {
3659                if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3660                        return error("invalid argument to -B: %s", arg+2);
3661        }
3662        else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3663                 !strcmp(arg, "--find-renames")) {
3664                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3665                        return error("invalid argument to -M: %s", arg+2);
3666                options->detect_rename = DIFF_DETECT_RENAME;
3667        }
3668        else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3669                options->irreversible_delete = 1;
3670        }
3671        else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3672                 !strcmp(arg, "--find-copies")) {
3673                if (options->detect_rename == DIFF_DETECT_COPY)
3674                        DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3675                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3676                        return error("invalid argument to -C: %s", arg+2);
3677                options->detect_rename = DIFF_DETECT_COPY;
3678        }
3679        else if (!strcmp(arg, "--no-renames"))
3680                options->detect_rename = 0;
3681        else if (!strcmp(arg, "--rename-empty"))
3682                DIFF_OPT_SET(options, RENAME_EMPTY);
3683        else if (!strcmp(arg, "--no-rename-empty"))
3684                DIFF_OPT_CLR(options, RENAME_EMPTY);
3685        else if (!strcmp(arg, "--relative"))
3686                DIFF_OPT_SET(options, RELATIVE_NAME);
3687        else if (starts_with(arg, "--relative=")) {
3688                DIFF_OPT_SET(options, RELATIVE_NAME);
3689                options->prefix = arg + 11;
3690        }
3691
3692        /* xdiff options */
3693        else if (!strcmp(arg, "--minimal"))
3694                DIFF_XDL_SET(options, NEED_MINIMAL);
3695        else if (!strcmp(arg, "--no-minimal"))
3696                DIFF_XDL_CLR(options, NEED_MINIMAL);
3697        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3698                DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3699        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3700                DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3701        else if (!strcmp(arg, "--ignore-space-at-eol"))
3702                DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3703        else if (!strcmp(arg, "--ignore-blank-lines"))
3704                DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3705        else if (!strcmp(arg, "--patience"))
3706                options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3707        else if (!strcmp(arg, "--histogram"))
3708                options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3709        else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3710                long value = parse_algorithm_value(optarg);
3711                if (value < 0)
3712                        return error("option diff-algorithm accepts \"myers\", "
3713                                     "\"minimal\", \"patience\" and \"histogram\"");
3714                /* clear out previous settings */
3715                DIFF_XDL_CLR(options, NEED_MINIMAL);
3716                options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3717                options->xdl_opts |= value;
3718                return argcount;
3719        }
3720
3721        /* flags options */
3722        else if (!strcmp(arg, "--binary")) {
3723                enable_patch_output(&options->output_format);
3724                DIFF_OPT_SET(options, BINARY);
3725        }
3726        else if (!strcmp(arg, "--full-index"))
3727                DIFF_OPT_SET(options, FULL_INDEX);
3728        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3729                DIFF_OPT_SET(options, TEXT);
3730        else if (!strcmp(arg, "-R"))
3731                DIFF_OPT_SET(options, REVERSE_DIFF);
3732        else if (!strcmp(arg, "--find-copies-harder"))
3733                DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3734        else if (!strcmp(arg, "--follow"))
3735                DIFF_OPT_SET(options, FOLLOW_RENAMES);
3736        else if (!strcmp(arg, "--no-follow"))
3737                DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3738        else if (!strcmp(arg, "--color"))
3739                options->use_color = 1;
3740        else if (starts_with(arg, "--color=")) {
3741                int value = git_config_colorbool(NULL, arg+8);
3742                if (value < 0)
3743                        return error("option `color' expects \"always\", \"auto\", or \"never\"");
3744                options->use_color = value;
3745        }
3746        else if (!strcmp(arg, "--no-color"))
3747                options->use_color = 0;
3748        else if (!strcmp(arg, "--color-words")) {
3749                options->use_color = 1;
3750                options->word_diff = DIFF_WORDS_COLOR;
3751        }
3752        else if (starts_with(arg, "--color-words=")) {
3753                options->use_color = 1;
3754                options->word_diff = DIFF_WORDS_COLOR;
3755                options->word_regex = arg + 14;
3756        }
3757        else if (!strcmp(arg, "--word-diff")) {
3758                if (options->word_diff == DIFF_WORDS_NONE)
3759                        options->word_diff = DIFF_WORDS_PLAIN;
3760        }
3761        else if (starts_with(arg, "--word-diff=")) {
3762                const char *type = arg + 12;
3763                if (!strcmp(type, "plain"))
3764                        options->word_diff = DIFF_WORDS_PLAIN;
3765                else if (!strcmp(type, "color")) {
3766                        options->use_color = 1;
3767                        options->word_diff = DIFF_WORDS_COLOR;
3768                }
3769                else if (!strcmp(type, "porcelain"))
3770                        options->word_diff = DIFF_WORDS_PORCELAIN;
3771                else if (!strcmp(type, "none"))
3772                        options->word_diff = DIFF_WORDS_NONE;
3773                else
3774                        die("bad --word-diff argument: %s", type);
3775        }
3776        else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3777                if (options->word_diff == DIFF_WORDS_NONE)
3778                        options->word_diff = DIFF_WORDS_PLAIN;
3779                options->word_regex = optarg;
3780                return argcount;
3781        }
3782        else if (!strcmp(arg, "--exit-code"))
3783                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3784        else if (!strcmp(arg, "--quiet"))
3785                DIFF_OPT_SET(options, QUICK);
3786        else if (!strcmp(arg, "--ext-diff"))
3787                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3788        else if (!strcmp(arg, "--no-ext-diff"))
3789                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3790        else if (!strcmp(arg, "--textconv"))
3791                DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3792        else if (!strcmp(arg, "--no-textconv"))
3793                DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3794        else if (!strcmp(arg, "--ignore-submodules")) {
3795                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3796                handle_ignore_submodules_arg(options, "all");
3797        } else if (starts_with(arg, "--ignore-submodules=")) {
3798                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3799                handle_ignore_submodules_arg(options, arg + 20);
3800        } else if (!strcmp(arg, "--submodule"))
3801                DIFF_OPT_SET(options, SUBMODULE_LOG);
3802        else if (starts_with(arg, "--submodule="))
3803                return parse_submodule_opt(options, arg + 12);
3804
3805        /* misc options */
3806        else if (!strcmp(arg, "-z"))
3807                options->line_termination = 0;
3808        else if ((argcount = short_opt('l', av, &optarg))) {
3809                options->rename_limit = strtoul(optarg, NULL, 10);
3810                return argcount;
3811        }
3812        else if ((argcount = short_opt('S', av, &optarg))) {
3813                options->pickaxe = optarg;
3814                options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3815                return argcount;
3816        } else if ((argcount = short_opt('G', av, &optarg))) {
3817                options->pickaxe = optarg;
3818                options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3819                return argcount;
3820        }
3821        else if (!strcmp(arg, "--pickaxe-all"))
3822                options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3823        else if (!strcmp(arg, "--pickaxe-regex"))
3824                options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3825        else if ((argcount = short_opt('O', av, &optarg))) {
3826                options->orderfile = optarg;
3827                return argcount;
3828        }
3829        else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3830                int offending = parse_diff_filter_opt(optarg, options);
3831                if (offending)
3832                        die("unknown change class '%c' in --diff-filter=%s",
3833                            offending, optarg);
3834                return argcount;
3835        }
3836        else if (!strcmp(arg, "--abbrev"))
3837                options->abbrev = DEFAULT_ABBREV;
3838        else if (starts_with(arg, "--abbrev=")) {
3839                options->abbrev = strtoul(arg + 9, NULL, 10);
3840                if (options->abbrev < MINIMUM_ABBREV)
3841                        options->abbrev = MINIMUM_ABBREV;
3842                else if (40 < options->abbrev)
3843                        options->abbrev = 40;
3844        }
3845        else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3846                options->a_prefix = optarg;
3847                return argcount;
3848        }
3849        else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3850                options->b_prefix = optarg;
3851                return argcount;
3852        }
3853        else if (!strcmp(arg, "--no-prefix"))
3854                options->a_prefix = options->b_prefix = "";
3855        else if (opt_arg(arg, '\0', "inter-hunk-context",
3856                         &options->interhunkcontext))
3857                ;
3858        else if (!strcmp(arg, "-W"))
3859                DIFF_OPT_SET(options, FUNCCONTEXT);
3860        else if (!strcmp(arg, "--function-context"))
3861                DIFF_OPT_SET(options, FUNCCONTEXT);
3862        else if (!strcmp(arg, "--no-function-context"))
3863                DIFF_OPT_CLR(options, FUNCCONTEXT);
3864        else if ((argcount = parse_long_opt("output", av, &optarg))) {
3865                options->file = fopen(optarg, "w");
3866                if (!options->file)
3867                        die_errno("Could not open '%s'", optarg);
3868                options->close_file = 1;
3869                return argcount;
3870        } else
3871                return 0;
3872        return 1;
3873}
3874
3875int parse_rename_score(const char **cp_p)
3876{
3877        unsigned long num, scale;
3878        int ch, dot;
3879        const char *cp = *cp_p;
3880
3881        num = 0;
3882        scale = 1;
3883        dot = 0;
3884        for (;;) {
3885                ch = *cp;
3886                if ( !dot && ch == '.' ) {
3887                        scale = 1;
3888                        dot = 1;
3889                } else if ( ch == '%' ) {
3890                        scale = dot ? scale*100 : 100;
3891                        cp++;   /* % is always at the end */
3892                        break;
3893                } else if ( ch >= '0' && ch <= '9' ) {
3894                        if ( scale < 100000 ) {
3895                                scale *= 10;
3896                                num = (num*10) + (ch-'0');
3897                        }
3898                } else {
3899                        break;
3900                }
3901                cp++;
3902        }
3903        *cp_p = cp;
3904
3905        /* user says num divided by scale and we say internally that
3906         * is MAX_SCORE * num / scale.
3907         */
3908        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3909}
3910
3911static int diff_scoreopt_parse(const char *opt)
3912{
3913        int opt1, opt2, cmd;
3914
3915        if (*opt++ != '-')
3916                return -1;
3917        cmd = *opt++;
3918        if (cmd == '-') {
3919                /* convert the long-form arguments into short-form versions */
3920                if (starts_with(opt, "break-rewrites")) {
3921                        opt += strlen("break-rewrites");
3922                        if (*opt == 0 || *opt++ == '=')
3923                                cmd = 'B';
3924                } else if (starts_with(opt, "find-copies")) {
3925                        opt += strlen("find-copies");
3926                        if (*opt == 0 || *opt++ == '=')
3927                                cmd = 'C';
3928                } else if (starts_with(opt, "find-renames")) {
3929                        opt += strlen("find-renames");
3930                        if (*opt == 0 || *opt++ == '=')
3931                                cmd = 'M';
3932                }
3933        }
3934        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3935                return -1; /* that is not a -M, -C, or -B option */
3936
3937        opt1 = parse_rename_score(&opt);
3938        if (cmd != 'B')
3939                opt2 = 0;
3940        else {
3941                if (*opt == 0)
3942                        opt2 = 0;
3943                else if (*opt != '/')
3944                        return -1; /* we expect -B80/99 or -B80 */
3945                else {
3946                        opt++;
3947                        opt2 = parse_rename_score(&opt);
3948                }
3949        }
3950        if (*opt != 0)
3951                return -1;
3952        return opt1 | (opt2 << 16);
3953}
3954
3955struct diff_queue_struct diff_queued_diff;
3956
3957void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3958{
3959        if (queue->alloc <= queue->nr) {
3960                queue->alloc = alloc_nr(queue->alloc);
3961                queue->queue = xrealloc(queue->queue,
3962                                        sizeof(dp) * queue->alloc);
3963        }
3964        queue->queue[queue->nr++] = dp;
3965}
3966
3967struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3968                                 struct diff_filespec *one,
3969                                 struct diff_filespec *two)
3970{
3971        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3972        dp->one = one;
3973        dp->two = two;
3974        if (queue)
3975                diff_q(queue, dp);
3976        return dp;
3977}
3978
3979void diff_free_filepair(struct diff_filepair *p)
3980{
3981        free_filespec(p->one);
3982        free_filespec(p->two);
3983        free(p);
3984}
3985
3986/* This is different from find_unique_abbrev() in that
3987 * it stuffs the result with dots for alignment.
3988 */
3989const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3990{
3991        int abblen;
3992        const char *abbrev;
3993        if (len == 40)
3994                return sha1_to_hex(sha1);
3995
3996        abbrev = find_unique_abbrev(sha1, len);
3997        abblen = strlen(abbrev);
3998        if (abblen < 37) {
3999                static char hex[41];
4000                if (len < abblen && abblen <= len + 2)
4001                        sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
4002                else
4003                        sprintf(hex, "%s...", abbrev);
4004                return hex;
4005        }
4006        return sha1_to_hex(sha1);
4007}
4008
4009static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4010{
4011        int line_termination = opt->line_termination;
4012        int inter_name_termination = line_termination ? '\t' : '\0';
4013
4014        fprintf(opt->file, "%s", diff_line_prefix(opt));
4015        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4016                fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4017                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
4018                fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4019        }
4020        if (p->score) {
4021                fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4022                        inter_name_termination);
4023        } else {
4024                fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4025        }
4026
4027        if (p->status == DIFF_STATUS_COPIED ||
4028            p->status == DIFF_STATUS_RENAMED) {
4029                const char *name_a, *name_b;
4030                name_a = p->one->path;
4031                name_b = p->two->path;
4032                strip_prefix(opt->prefix_length, &name_a, &name_b);
4033                write_name_quoted(name_a, opt->file, inter_name_termination);
4034                write_name_quoted(name_b, opt->file, line_termination);
4035        } else {
4036                const char *name_a, *name_b;
4037                name_a = p->one->mode ? p->one->path : p->two->path;
4038                name_b = NULL;
4039                strip_prefix(opt->prefix_length, &name_a, &name_b);
4040                write_name_quoted(name_a, opt->file, line_termination);
4041        }
4042}
4043
4044int diff_unmodified_pair(struct diff_filepair *p)
4045{
4046        /* This function is written stricter than necessary to support
4047         * the currently implemented transformers, but the idea is to
4048         * let transformers to produce diff_filepairs any way they want,
4049         * and filter and clean them up here before producing the output.
4050         */
4051        struct diff_filespec *one = p->one, *two = p->two;
4052
4053        if (DIFF_PAIR_UNMERGED(p))
4054                return 0; /* unmerged is interesting */
4055
4056        /* deletion, addition, mode or type change
4057         * and rename are all interesting.
4058         */
4059        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4060            DIFF_PAIR_MODE_CHANGED(p) ||
4061            strcmp(one->path, two->path))
4062                return 0;
4063
4064        /* both are valid and point at the same path.  that is, we are
4065         * dealing with a change.
4066         */
4067        if (one->sha1_valid && two->sha1_valid &&
4068            !hashcmp(one->sha1, two->sha1) &&
4069            !one->dirty_submodule && !two->dirty_submodule)
4070                return 1; /* no change */
4071        if (!one->sha1_valid && !two->sha1_valid)
4072                return 1; /* both look at the same file on the filesystem. */
4073        return 0;
4074}
4075
4076static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4077{
4078        if (diff_unmodified_pair(p))
4079                return;
4080
4081        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4082            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4083                return; /* no tree diffs in patch format */
4084
4085        run_diff(p, o);
4086}
4087
4088static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4089                            struct diffstat_t *diffstat)
4090{
4091        if (diff_unmodified_pair(p))
4092                return;
4093
4094        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4095            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4096                return; /* no useful stat for tree diffs */
4097
4098        run_diffstat(p, o, diffstat);
4099}
4100
4101static void diff_flush_checkdiff(struct diff_filepair *p,
4102                struct diff_options *o)
4103{
4104        if (diff_unmodified_pair(p))
4105                return;
4106
4107        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4108            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4109                return; /* nothing to check in tree diffs */
4110
4111        run_checkdiff(p, o);
4112}
4113
4114int diff_queue_is_empty(void)
4115{
4116        struct diff_queue_struct *q = &diff_queued_diff;
4117        int i;
4118        for (i = 0; i < q->nr; i++)
4119                if (!diff_unmodified_pair(q->queue[i]))
4120                        return 0;
4121        return 1;
4122}
4123
4124#if DIFF_DEBUG
4125void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4126{
4127        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4128                x, one ? one : "",
4129                s->path,
4130                DIFF_FILE_VALID(s) ? "valid" : "invalid",
4131                s->mode,
4132                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4133        fprintf(stderr, "queue[%d] %s size %lu\n",
4134                x, one ? one : "",
4135                s->size);
4136}
4137
4138void diff_debug_filepair(const struct diff_filepair *p, int i)
4139{
4140        diff_debug_filespec(p->one, i, "one");
4141        diff_debug_filespec(p->two, i, "two");
4142        fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4143                p->score, p->status ? p->status : '?',
4144                p->one->rename_used, p->broken_pair);
4145}
4146
4147void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4148{
4149        int i;
4150        if (msg)
4151                fprintf(stderr, "%s\n", msg);
4152        fprintf(stderr, "q->nr = %d\n", q->nr);
4153        for (i = 0; i < q->nr; i++) {
4154                struct diff_filepair *p = q->queue[i];
4155                diff_debug_filepair(p, i);
4156        }
4157}
4158#endif
4159
4160static void diff_resolve_rename_copy(void)
4161{
4162        int i;
4163        struct diff_filepair *p;
4164        struct diff_queue_struct *q = &diff_queued_diff;
4165
4166        diff_debug_queue("resolve-rename-copy", q);
4167
4168        for (i = 0; i < q->nr; i++) {
4169                p = q->queue[i];
4170                p->status = 0; /* undecided */
4171                if (DIFF_PAIR_UNMERGED(p))
4172                        p->status = DIFF_STATUS_UNMERGED;
4173                else if (!DIFF_FILE_VALID(p->one))
4174                        p->status = DIFF_STATUS_ADDED;
4175                else if (!DIFF_FILE_VALID(p->two))
4176                        p->status = DIFF_STATUS_DELETED;
4177                else if (DIFF_PAIR_TYPE_CHANGED(p))
4178                        p->status = DIFF_STATUS_TYPE_CHANGED;
4179
4180                /* from this point on, we are dealing with a pair
4181                 * whose both sides are valid and of the same type, i.e.
4182                 * either in-place edit or rename/copy edit.
4183                 */
4184                else if (DIFF_PAIR_RENAME(p)) {
4185                        /*
4186                         * A rename might have re-connected a broken
4187                         * pair up, causing the pathnames to be the
4188                         * same again. If so, that's not a rename at
4189                         * all, just a modification..
4190                         *
4191                         * Otherwise, see if this source was used for
4192                         * multiple renames, in which case we decrement
4193                         * the count, and call it a copy.
4194                         */
4195                        if (!strcmp(p->one->path, p->two->path))
4196                                p->status = DIFF_STATUS_MODIFIED;
4197                        else if (--p->one->rename_used > 0)
4198                                p->status = DIFF_STATUS_COPIED;
4199                        else
4200                                p->status = DIFF_STATUS_RENAMED;
4201                }
4202                else if (hashcmp(p->one->sha1, p->two->sha1) ||
4203                         p->one->mode != p->two->mode ||
4204                         p->one->dirty_submodule ||
4205                         p->two->dirty_submodule ||
4206                         is_null_sha1(p->one->sha1))
4207                        p->status = DIFF_STATUS_MODIFIED;
4208                else {
4209                        /* This is a "no-change" entry and should not
4210                         * happen anymore, but prepare for broken callers.
4211                         */
4212                        error("feeding unmodified %s to diffcore",
4213                              p->one->path);
4214                        p->status = DIFF_STATUS_UNKNOWN;
4215                }
4216        }
4217        diff_debug_queue("resolve-rename-copy done", q);
4218}
4219
4220static int check_pair_status(struct diff_filepair *p)
4221{
4222        switch (p->status) {
4223        case DIFF_STATUS_UNKNOWN:
4224                return 0;
4225        case 0:
4226                die("internal error in diff-resolve-rename-copy");
4227        default:
4228                return 1;
4229        }
4230}
4231
4232static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4233{
4234        int fmt = opt->output_format;
4235
4236        if (fmt & DIFF_FORMAT_CHECKDIFF)
4237                diff_flush_checkdiff(p, opt);
4238        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4239                diff_flush_raw(p, opt);
4240        else if (fmt & DIFF_FORMAT_NAME) {
4241                const char *name_a, *name_b;
4242                name_a = p->two->path;
4243                name_b = NULL;
4244                strip_prefix(opt->prefix_length, &name_a, &name_b);
4245                write_name_quoted(name_a, opt->file, opt->line_termination);
4246        }
4247}
4248
4249static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4250{
4251        if (fs->mode)
4252                fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4253        else
4254                fprintf(file, " %s ", newdelete);
4255        write_name_quoted(fs->path, file, '\n');
4256}
4257
4258
4259static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4260                const char *line_prefix)
4261{
4262        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4263                fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4264                        p->two->mode, show_name ? ' ' : '\n');
4265                if (show_name) {
4266                        write_name_quoted(p->two->path, file, '\n');
4267                }
4268        }
4269}
4270
4271static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4272                        const char *line_prefix)
4273{
4274        char *names = pprint_rename(p->one->path, p->two->path);
4275
4276        fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4277        free(names);
4278        show_mode_change(file, p, 0, line_prefix);
4279}
4280
4281static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4282{
4283        FILE *file = opt->file;
4284        const char *line_prefix = diff_line_prefix(opt);
4285
4286        switch(p->status) {
4287        case DIFF_STATUS_DELETED:
4288                fputs(line_prefix, file);
4289                show_file_mode_name(file, "delete", p->one);
4290                break;
4291        case DIFF_STATUS_ADDED:
4292                fputs(line_prefix, file);
4293                show_file_mode_name(file, "create", p->two);
4294                break;
4295        case DIFF_STATUS_COPIED:
4296                fputs(line_prefix, file);
4297                show_rename_copy(file, "copy", p, line_prefix);
4298                break;
4299        case DIFF_STATUS_RENAMED:
4300                fputs(line_prefix, file);
4301                show_rename_copy(file, "rename", p, line_prefix);
4302                break;
4303        default:
4304                if (p->score) {
4305                        fprintf(file, "%s rewrite ", line_prefix);
4306                        write_name_quoted(p->two->path, file, ' ');
4307                        fprintf(file, "(%d%%)\n", similarity_index(p));
4308                }
4309                show_mode_change(file, p, !p->score, line_prefix);
4310                break;
4311        }
4312}
4313
4314struct patch_id_t {
4315        git_SHA_CTX *ctx;
4316        int patchlen;
4317};
4318
4319static int remove_space(char *line, int len)
4320{
4321        int i;
4322        char *dst = line;
4323        unsigned char c;
4324
4325        for (i = 0; i < len; i++)
4326                if (!isspace((c = line[i])))
4327                        *dst++ = c;
4328
4329        return dst - line;
4330}
4331
4332static void patch_id_consume(void *priv, char *line, unsigned long len)
4333{
4334        struct patch_id_t *data = priv;
4335        int new_len;
4336
4337        /* Ignore line numbers when computing the SHA1 of the patch */
4338        if (starts_with(line, "@@ -"))
4339                return;
4340
4341        new_len = remove_space(line, len);
4342
4343        git_SHA1_Update(data->ctx, line, new_len);
4344        data->patchlen += new_len;
4345}
4346
4347/* returns 0 upon success, and writes result into sha1 */
4348static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4349{
4350        struct diff_queue_struct *q = &diff_queued_diff;
4351        int i;
4352        git_SHA_CTX ctx;
4353        struct patch_id_t data;
4354        char buffer[PATH_MAX * 4 + 20];
4355
4356        git_SHA1_Init(&ctx);
4357        memset(&data, 0, sizeof(struct patch_id_t));
4358        data.ctx = &ctx;
4359
4360        for (i = 0; i < q->nr; i++) {
4361                xpparam_t xpp;
4362                xdemitconf_t xecfg;
4363                mmfile_t mf1, mf2;
4364                struct diff_filepair *p = q->queue[i];
4365                int len1, len2;
4366
4367                memset(&xpp, 0, sizeof(xpp));
4368                memset(&xecfg, 0, sizeof(xecfg));
4369                if (p->status == 0)
4370                        return error("internal diff status error");
4371                if (p->status == DIFF_STATUS_UNKNOWN)
4372                        continue;
4373                if (diff_unmodified_pair(p))
4374                        continue;
4375                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4376                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4377                        continue;
4378                if (DIFF_PAIR_UNMERGED(p))
4379                        continue;
4380
4381                diff_fill_sha1_info(p->one);
4382                diff_fill_sha1_info(p->two);
4383                if (fill_mmfile(&mf1, p->one) < 0 ||
4384                                fill_mmfile(&mf2, p->two) < 0)
4385                        return error("unable to read files to diff");
4386
4387                len1 = remove_space(p->one->path, strlen(p->one->path));
4388                len2 = remove_space(p->two->path, strlen(p->two->path));
4389                if (p->one->mode == 0)
4390                        len1 = snprintf(buffer, sizeof(buffer),
4391                                        "diff--gita/%.*sb/%.*s"
4392                                        "newfilemode%06o"
4393                                        "---/dev/null"
4394                                        "+++b/%.*s",
4395                                        len1, p->one->path,
4396                                        len2, p->two->path,
4397                                        p->two->mode,
4398                                        len2, p->two->path);
4399                else if (p->two->mode == 0)
4400                        len1 = snprintf(buffer, sizeof(buffer),
4401                                        "diff--gita/%.*sb/%.*s"
4402                                        "deletedfilemode%06o"
4403                                        "---a/%.*s"
4404                                        "+++/dev/null",
4405                                        len1, p->one->path,
4406                                        len2, p->two->path,
4407                                        p->one->mode,
4408                                        len1, p->one->path);
4409                else
4410                        len1 = snprintf(buffer, sizeof(buffer),
4411                                        "diff--gita/%.*sb/%.*s"
4412                                        "---a/%.*s"
4413                                        "+++b/%.*s",
4414                                        len1, p->one->path,
4415                                        len2, p->two->path,
4416                                        len1, p->one->path,
4417                                        len2, p->two->path);
4418                git_SHA1_Update(&ctx, buffer, len1);
4419
4420                if (diff_filespec_is_binary(p->one) ||
4421                    diff_filespec_is_binary(p->two)) {
4422                        git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4423                        git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4424                        continue;
4425                }
4426
4427                xpp.flags = 0;
4428                xecfg.ctxlen = 3;
4429                xecfg.flags = 0;
4430                xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4431                              &xpp, &xecfg);
4432        }
4433
4434        git_SHA1_Final(sha1, &ctx);
4435        return 0;
4436}
4437
4438int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4439{
4440        struct diff_queue_struct *q = &diff_queued_diff;
4441        int i;
4442        int result = diff_get_patch_id(options, sha1);
4443
4444        for (i = 0; i < q->nr; i++)
4445                diff_free_filepair(q->queue[i]);
4446
4447        free(q->queue);
4448        DIFF_QUEUE_CLEAR(q);
4449
4450        return result;
4451}
4452
4453static int is_summary_empty(const struct diff_queue_struct *q)
4454{
4455        int i;
4456
4457        for (i = 0; i < q->nr; i++) {
4458                const struct diff_filepair *p = q->queue[i];
4459
4460                switch (p->status) {
4461                case DIFF_STATUS_DELETED:
4462                case DIFF_STATUS_ADDED:
4463                case DIFF_STATUS_COPIED:
4464                case DIFF_STATUS_RENAMED:
4465                        return 0;
4466                default:
4467                        if (p->score)
4468                                return 0;
4469                        if (p->one->mode && p->two->mode &&
4470                            p->one->mode != p->two->mode)
4471                                return 0;
4472                        break;
4473                }
4474        }
4475        return 1;
4476}
4477
4478static const char rename_limit_warning[] =
4479"inexact rename detection was skipped due to too many files.";
4480
4481static const char degrade_cc_to_c_warning[] =
4482"only found copies from modified paths due to too many files.";
4483
4484static const char rename_limit_advice[] =
4485"you may want to set your %s variable to at least "
4486"%d and retry the command.";
4487
4488void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4489{
4490        if (degraded_cc)
4491                warning(degrade_cc_to_c_warning);
4492        else if (needed)
4493                warning(rename_limit_warning);
4494        else
4495                return;
4496        if (0 < needed && needed < 32767)
4497                warning(rename_limit_advice, varname, needed);
4498}
4499
4500void diff_flush(struct diff_options *options)
4501{
4502        struct diff_queue_struct *q = &diff_queued_diff;
4503        int i, output_format = options->output_format;
4504        int separator = 0;
4505        int dirstat_by_line = 0;
4506
4507        /*
4508         * Order: raw, stat, summary, patch
4509         * or:    name/name-status/checkdiff (other bits clear)
4510         */
4511        if (!q->nr)
4512                goto free_queue;
4513
4514        if (output_format & (DIFF_FORMAT_RAW |
4515                             DIFF_FORMAT_NAME |
4516                             DIFF_FORMAT_NAME_STATUS |
4517                             DIFF_FORMAT_CHECKDIFF)) {
4518                for (i = 0; i < q->nr; i++) {
4519                        struct diff_filepair *p = q->queue[i];
4520                        if (check_pair_status(p))
4521                                flush_one_pair(p, options);
4522                }
4523                separator++;
4524        }
4525
4526        if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4527                dirstat_by_line = 1;
4528
4529        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4530            dirstat_by_line) {
4531                struct diffstat_t diffstat;
4532
4533                memset(&diffstat, 0, sizeof(struct diffstat_t));
4534                for (i = 0; i < q->nr; i++) {
4535                        struct diff_filepair *p = q->queue[i];
4536                        if (check_pair_status(p))
4537                                diff_flush_stat(p, options, &diffstat);
4538                }
4539                if (output_format & DIFF_FORMAT_NUMSTAT)
4540                        show_numstat(&diffstat, options);
4541                if (output_format & DIFF_FORMAT_DIFFSTAT)
4542                        show_stats(&diffstat, options);
4543                if (output_format & DIFF_FORMAT_SHORTSTAT)
4544                        show_shortstats(&diffstat, options);
4545                if (output_format & DIFF_FORMAT_DIRSTAT)
4546                        show_dirstat_by_line(&diffstat, options);
4547                free_diffstat_info(&diffstat);
4548                separator++;
4549        }
4550        if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4551                show_dirstat(options);
4552
4553        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4554                for (i = 0; i < q->nr; i++) {
4555                        diff_summary(options, q->queue[i]);
4556                }
4557                separator++;
4558        }
4559
4560        if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4561            DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4562            DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4563                /*
4564                 * run diff_flush_patch for the exit status. setting
4565                 * options->file to /dev/null should be safe, because we
4566                 * aren't supposed to produce any output anyway.
4567                 */
4568                if (options->close_file)
4569                        fclose(options->file);
4570                options->file = fopen("/dev/null", "w");
4571                if (!options->file)
4572                        die_errno("Could not open /dev/null");
4573                options->close_file = 1;
4574                for (i = 0; i < q->nr; i++) {
4575                        struct diff_filepair *p = q->queue[i];
4576                        if (check_pair_status(p))
4577                                diff_flush_patch(p, options);
4578                        if (options->found_changes)
4579                                break;
4580                }
4581        }
4582
4583        if (output_format & DIFF_FORMAT_PATCH) {
4584                if (separator) {
4585                        fprintf(options->file, "%s%c",
4586                                diff_line_prefix(options),
4587                                options->line_termination);
4588                        if (options->stat_sep) {
4589                                /* attach patch instead of inline */
4590                                fputs(options->stat_sep, options->file);
4591                        }
4592                }
4593
4594                for (i = 0; i < q->nr; i++) {
4595                        struct diff_filepair *p = q->queue[i];
4596                        if (check_pair_status(p))
4597                                diff_flush_patch(p, options);
4598                }
4599        }
4600
4601        if (output_format & DIFF_FORMAT_CALLBACK)
4602                options->format_callback(q, options, options->format_callback_data);
4603
4604        for (i = 0; i < q->nr; i++)
4605                diff_free_filepair(q->queue[i]);
4606free_queue:
4607        free(q->queue);
4608        DIFF_QUEUE_CLEAR(q);
4609        if (options->close_file)
4610                fclose(options->file);
4611
4612        /*
4613         * Report the content-level differences with HAS_CHANGES;
4614         * diff_addremove/diff_change does not set the bit when
4615         * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4616         */
4617        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4618                if (options->found_changes)
4619                        DIFF_OPT_SET(options, HAS_CHANGES);
4620                else
4621                        DIFF_OPT_CLR(options, HAS_CHANGES);
4622        }
4623}
4624
4625static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4626{
4627        return (((p->status == DIFF_STATUS_MODIFIED) &&
4628                 ((p->score &&
4629                   filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4630                  (!p->score &&
4631                   filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4632                ((p->status != DIFF_STATUS_MODIFIED) &&
4633                 filter_bit_tst(p->status, options)));
4634}
4635
4636static void diffcore_apply_filter(struct diff_options *options)
4637{
4638        int i;
4639        struct diff_queue_struct *q = &diff_queued_diff;
4640        struct diff_queue_struct outq;
4641
4642        DIFF_QUEUE_CLEAR(&outq);
4643
4644        if (!options->filter)
4645                return;
4646
4647        if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4648                int found;
4649                for (i = found = 0; !found && i < q->nr; i++) {
4650                        if (match_filter(options, q->queue[i]))
4651                                found++;
4652                }
4653                if (found)
4654                        return;
4655
4656                /* otherwise we will clear the whole queue
4657                 * by copying the empty outq at the end of this
4658                 * function, but first clear the current entries
4659                 * in the queue.
4660                 */
4661                for (i = 0; i < q->nr; i++)
4662                        diff_free_filepair(q->queue[i]);
4663        }
4664        else {
4665                /* Only the matching ones */
4666                for (i = 0; i < q->nr; i++) {
4667                        struct diff_filepair *p = q->queue[i];
4668                        if (match_filter(options, p))
4669                                diff_q(&outq, p);
4670                        else
4671                                diff_free_filepair(p);
4672                }
4673        }
4674        free(q->queue);
4675        *q = outq;
4676}
4677
4678/* Check whether two filespecs with the same mode and size are identical */
4679static int diff_filespec_is_identical(struct diff_filespec *one,
4680                                      struct diff_filespec *two)
4681{
4682        if (S_ISGITLINK(one->mode))
4683                return 0;
4684        if (diff_populate_filespec(one, 0))
4685                return 0;
4686        if (diff_populate_filespec(two, 0))
4687                return 0;
4688        return !memcmp(one->data, two->data, one->size);
4689}
4690
4691static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4692{
4693        if (p->done_skip_stat_unmatch)
4694                return p->skip_stat_unmatch_result;
4695
4696        p->done_skip_stat_unmatch = 1;
4697        p->skip_stat_unmatch_result = 0;
4698        /*
4699         * 1. Entries that come from stat info dirtiness
4700         *    always have both sides (iow, not create/delete),
4701         *    one side of the object name is unknown, with
4702         *    the same mode and size.  Keep the ones that
4703         *    do not match these criteria.  They have real
4704         *    differences.
4705         *
4706         * 2. At this point, the file is known to be modified,
4707         *    with the same mode and size, and the object
4708         *    name of one side is unknown.  Need to inspect
4709         *    the identical contents.
4710         */
4711        if (!DIFF_FILE_VALID(p->one) || /* (1) */
4712            !DIFF_FILE_VALID(p->two) ||
4713            (p->one->sha1_valid && p->two->sha1_valid) ||
4714            (p->one->mode != p->two->mode) ||
4715            diff_populate_filespec(p->one, 1) ||
4716            diff_populate_filespec(p->two, 1) ||
4717            (p->one->size != p->two->size) ||
4718            !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4719                p->skip_stat_unmatch_result = 1;
4720        return p->skip_stat_unmatch_result;
4721}
4722
4723static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4724{
4725        int i;
4726        struct diff_queue_struct *q = &diff_queued_diff;
4727        struct diff_queue_struct outq;
4728        DIFF_QUEUE_CLEAR(&outq);
4729
4730        for (i = 0; i < q->nr; i++) {
4731                struct diff_filepair *p = q->queue[i];
4732
4733                if (diff_filespec_check_stat_unmatch(p))
4734                        diff_q(&outq, p);
4735                else {
4736                        /*
4737                         * The caller can subtract 1 from skip_stat_unmatch
4738                         * to determine how many paths were dirty only
4739                         * due to stat info mismatch.
4740                         */
4741                        if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4742                                diffopt->skip_stat_unmatch++;
4743                        diff_free_filepair(p);
4744                }
4745        }
4746        free(q->queue);
4747        *q = outq;
4748}
4749
4750static int diffnamecmp(const void *a_, const void *b_)
4751{
4752        const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4753        const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4754        const char *name_a, *name_b;
4755
4756        name_a = a->one ? a->one->path : a->two->path;
4757        name_b = b->one ? b->one->path : b->two->path;
4758        return strcmp(name_a, name_b);
4759}
4760
4761void diffcore_fix_diff_index(struct diff_options *options)
4762{
4763        struct diff_queue_struct *q = &diff_queued_diff;
4764        qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4765}
4766
4767void diffcore_std(struct diff_options *options)
4768{
4769        if (options->skip_stat_unmatch)
4770                diffcore_skip_stat_unmatch(options);
4771        if (!options->found_follow) {
4772                /* See try_to_follow_renames() in tree-diff.c */
4773                if (options->break_opt != -1)
4774                        diffcore_break(options->break_opt);
4775                if (options->detect_rename)
4776                        diffcore_rename(options);
4777                if (options->break_opt != -1)
4778                        diffcore_merge_broken();
4779        }
4780        if (options->pickaxe)
4781                diffcore_pickaxe(options);
4782        if (options->orderfile)
4783                diffcore_order(options->orderfile);
4784        if (!options->found_follow)
4785                /* See try_to_follow_renames() in tree-diff.c */
4786                diff_resolve_rename_copy();
4787        diffcore_apply_filter(options);
4788
4789        if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4790                DIFF_OPT_SET(options, HAS_CHANGES);
4791        else
4792                DIFF_OPT_CLR(options, HAS_CHANGES);
4793
4794        options->found_follow = 0;
4795}
4796
4797int diff_result_code(struct diff_options *opt, int status)
4798{
4799        int result = 0;
4800
4801        diff_warn_rename_limit("diff.renameLimit",
4802                               opt->needed_rename_limit,
4803                               opt->degraded_cc_to_c);
4804        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4805            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4806                return status;
4807        if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4808            DIFF_OPT_TST(opt, HAS_CHANGES))
4809                result |= 01;
4810        if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4811            DIFF_OPT_TST(opt, CHECK_FAILED))
4812                result |= 02;
4813        return result;
4814}
4815
4816int diff_can_quit_early(struct diff_options *opt)
4817{
4818        return (DIFF_OPT_TST(opt, QUICK) &&
4819                !opt->filter &&
4820                DIFF_OPT_TST(opt, HAS_CHANGES));
4821}
4822
4823/*
4824 * Shall changes to this submodule be ignored?
4825 *
4826 * Submodule changes can be configured to be ignored separately for each path,
4827 * but that configuration can be overridden from the command line.
4828 */
4829static int is_submodule_ignored(const char *path, struct diff_options *options)
4830{
4831        int ignored = 0;
4832        unsigned orig_flags = options->flags;
4833        if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4834                set_diffopt_flags_from_submodule_config(options, path);
4835        if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4836                ignored = 1;
4837        options->flags = orig_flags;
4838        return ignored;
4839}
4840
4841void diff_addremove(struct diff_options *options,
4842                    int addremove, unsigned mode,
4843                    const unsigned char *sha1,
4844                    int sha1_valid,
4845                    const char *concatpath, unsigned dirty_submodule)
4846{
4847        struct diff_filespec *one, *two;
4848
4849        if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4850                return;
4851
4852        /* This may look odd, but it is a preparation for
4853         * feeding "there are unchanged files which should
4854         * not produce diffs, but when you are doing copy
4855         * detection you would need them, so here they are"
4856         * entries to the diff-core.  They will be prefixed
4857         * with something like '=' or '*' (I haven't decided
4858         * which but should not make any difference).
4859         * Feeding the same new and old to diff_change()
4860         * also has the same effect.
4861         * Before the final output happens, they are pruned after
4862         * merged into rename/copy pairs as appropriate.
4863         */
4864        if (DIFF_OPT_TST(options, REVERSE_DIFF))
4865                addremove = (addremove == '+' ? '-' :
4866                             addremove == '-' ? '+' : addremove);
4867
4868        if (options->prefix &&
4869            strncmp(concatpath, options->prefix, options->prefix_length))
4870                return;
4871
4872        one = alloc_filespec(concatpath);
4873        two = alloc_filespec(concatpath);
4874
4875        if (addremove != '+')
4876                fill_filespec(one, sha1, sha1_valid, mode);
4877        if (addremove != '-') {
4878                fill_filespec(two, sha1, sha1_valid, mode);
4879                two->dirty_submodule = dirty_submodule;
4880        }
4881
4882        diff_queue(&diff_queued_diff, one, two);
4883        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4884                DIFF_OPT_SET(options, HAS_CHANGES);
4885}
4886
4887void diff_change(struct diff_options *options,
4888                 unsigned old_mode, unsigned new_mode,
4889                 const unsigned char *old_sha1,
4890                 const unsigned char *new_sha1,
4891                 int old_sha1_valid, int new_sha1_valid,
4892                 const char *concatpath,
4893                 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4894{
4895        struct diff_filespec *one, *two;
4896        struct diff_filepair *p;
4897
4898        if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4899            is_submodule_ignored(concatpath, options))
4900                return;
4901
4902        if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4903                unsigned tmp;
4904                const unsigned char *tmp_c;
4905                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4906                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4907                tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4908                        new_sha1_valid = tmp;
4909                tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4910                        new_dirty_submodule = tmp;
4911        }
4912
4913        if (options->prefix &&
4914            strncmp(concatpath, options->prefix, options->prefix_length))
4915                return;
4916
4917        one = alloc_filespec(concatpath);
4918        two = alloc_filespec(concatpath);
4919        fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4920        fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4921        one->dirty_submodule = old_dirty_submodule;
4922        two->dirty_submodule = new_dirty_submodule;
4923        p = diff_queue(&diff_queued_diff, one, two);
4924
4925        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4926                return;
4927
4928        if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
4929            !diff_filespec_check_stat_unmatch(p))
4930                return;
4931
4932        DIFF_OPT_SET(options, HAS_CHANGES);
4933}
4934
4935struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4936{
4937        struct diff_filepair *pair;
4938        struct diff_filespec *one, *two;
4939
4940        if (options->prefix &&
4941            strncmp(path, options->prefix, options->prefix_length))
4942                return NULL;
4943
4944        one = alloc_filespec(path);
4945        two = alloc_filespec(path);
4946        pair = diff_queue(&diff_queued_diff, one, two);
4947        pair->is_unmerged = 1;
4948        return pair;
4949}
4950
4951static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4952                size_t *outsize)
4953{
4954        struct diff_tempfile *temp;
4955        const char *argv[3];
4956        const char **arg = argv;
4957        struct child_process child;
4958        struct strbuf buf = STRBUF_INIT;
4959        int err = 0;
4960
4961        temp = prepare_temp_file(spec->path, spec);
4962        *arg++ = pgm;
4963        *arg++ = temp->name;
4964        *arg = NULL;
4965
4966        memset(&child, 0, sizeof(child));
4967        child.use_shell = 1;
4968        child.argv = argv;
4969        child.out = -1;
4970        if (start_command(&child)) {
4971                remove_tempfile();
4972                return NULL;
4973        }
4974
4975        if (strbuf_read(&buf, child.out, 0) < 0)
4976                err = error("error reading from textconv command '%s'", pgm);
4977        close(child.out);
4978
4979        if (finish_command(&child) || err) {
4980                strbuf_release(&buf);
4981                remove_tempfile();
4982                return NULL;
4983        }
4984        remove_tempfile();
4985
4986        return strbuf_detach(&buf, outsize);
4987}
4988
4989size_t fill_textconv(struct userdiff_driver *driver,
4990                     struct diff_filespec *df,
4991                     char **outbuf)
4992{
4993        size_t size;
4994
4995        if (!driver || !driver->textconv) {
4996                if (!DIFF_FILE_VALID(df)) {
4997                        *outbuf = "";
4998                        return 0;
4999                }
5000                if (diff_populate_filespec(df, 0))
5001                        die("unable to read files to diff");
5002                *outbuf = df->data;
5003                return df->size;
5004        }
5005
5006        if (driver->textconv_cache && df->sha1_valid) {
5007                *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5008                                          &size);
5009                if (*outbuf)
5010                        return size;
5011        }
5012
5013        *outbuf = run_textconv(driver->textconv, df, &size);
5014        if (!*outbuf)
5015                die("unable to read files to diff");
5016
5017        if (driver->textconv_cache && df->sha1_valid) {
5018                /* ignore errors, as we might be in a readonly repository */
5019                notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5020                                size);
5021                /*
5022                 * we could save up changes and flush them all at the end,
5023                 * but we would need an extra call after all diffing is done.
5024                 * Since generating a cache entry is the slow path anyway,
5025                 * this extra overhead probably isn't a big deal.
5026                 */
5027                notes_cache_write(driver->textconv_cache);
5028        }
5029
5030        return size;
5031}
5032
5033void setup_diff_pager(struct diff_options *opt)
5034{
5035        /*
5036         * If the user asked for our exit code, then either they want --quiet
5037         * or --exit-code. We should definitely not bother with a pager in the
5038         * former case, as we will generate no output. Since we still properly
5039         * report our exit code even when a pager is run, we _could_ run a
5040         * pager with --exit-code. But since we have not done so historically,
5041         * and because it is easy to find people oneline advising "git diff
5042         * --exit-code" in hooks and other scripts, we do not do so.
5043         */
5044        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5045            check_pager_config("diff") != 0)
5046                setup_pager();
5047}