4105260122878a81994077e1552dcc305f501037
   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
  19#ifdef NO_FAST_WORKING_DIRECTORY
  20#define FAST_WORKING_DIRECTORY 0
  21#else
  22#define FAST_WORKING_DIRECTORY 1
  23#endif
  24
  25static int diff_detect_rename_default;
  26static int diff_rename_limit_default = 400;
  27static int diff_suppress_blank_empty;
  28static int diff_use_color_default = -1;
  29static const char *diff_word_regex_cfg;
  30static const char *external_diff_cmd_cfg;
  31int diff_auto_refresh_index = 1;
  32static int diff_mnemonic_prefix;
  33static int diff_no_prefix;
  34static int diff_stat_graph_width;
  35static int diff_dirstat_permille_default = 30;
  36static struct diff_options default_diff_options;
  37
  38static char diff_colors[][COLOR_MAXLEN] = {
  39        GIT_COLOR_RESET,
  40        GIT_COLOR_NORMAL,       /* PLAIN */
  41        GIT_COLOR_BOLD,         /* METAINFO */
  42        GIT_COLOR_CYAN,         /* FRAGINFO */
  43        GIT_COLOR_RED,          /* OLD */
  44        GIT_COLOR_GREEN,        /* NEW */
  45        GIT_COLOR_YELLOW,       /* COMMIT */
  46        GIT_COLOR_BG_RED,       /* WHITESPACE */
  47        GIT_COLOR_NORMAL,       /* FUNCINFO */
  48};
  49
  50static int parse_diff_color_slot(const char *var, int ofs)
  51{
  52        if (!strcasecmp(var+ofs, "plain"))
  53                return DIFF_PLAIN;
  54        if (!strcasecmp(var+ofs, "meta"))
  55                return DIFF_METAINFO;
  56        if (!strcasecmp(var+ofs, "frag"))
  57                return DIFF_FRAGINFO;
  58        if (!strcasecmp(var+ofs, "old"))
  59                return DIFF_FILE_OLD;
  60        if (!strcasecmp(var+ofs, "new"))
  61                return DIFF_FILE_NEW;
  62        if (!strcasecmp(var+ofs, "commit"))
  63                return DIFF_COMMIT;
  64        if (!strcasecmp(var+ofs, "whitespace"))
  65                return DIFF_WHITESPACE;
  66        if (!strcasecmp(var+ofs, "func"))
  67                return DIFF_FUNCINFO;
  68        return -1;
  69}
  70
  71static int parse_dirstat_params(struct diff_options *options, const char *params,
  72                                struct strbuf *errmsg)
  73{
  74        const char *p = params;
  75        int p_len, ret = 0;
  76
  77        while (*p) {
  78                p_len = strchrnul(p, ',') - p;
  79                if (!memcmp(p, "changes", p_len)) {
  80                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  81                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  82                } else if (!memcmp(p, "lines", p_len)) {
  83                        DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  84                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  85                } else if (!memcmp(p, "files", p_len)) {
  86                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  87                        DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  88                } else if (!memcmp(p, "noncumulative", p_len)) {
  89                        DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
  90                } else if (!memcmp(p, "cumulative", p_len)) {
  91                        DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
  92                } else if (isdigit(*p)) {
  93                        char *end;
  94                        int permille = strtoul(p, &end, 10) * 10;
  95                        if (*end == '.' && isdigit(*++end)) {
  96                                /* only use first digit */
  97                                permille += *end - '0';
  98                                /* .. and ignore any further digits */
  99                                while (isdigit(*++end))
 100                                        ; /* nothing */
 101                        }
 102                        if (end - p == p_len)
 103                                options->dirstat_permille = permille;
 104                        else {
 105                                strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%.*s'\n"),
 106                                            p_len, p);
 107                                ret++;
 108                        }
 109                } else {
 110                        strbuf_addf(errmsg, _("  Unknown dirstat parameter '%.*s'\n"),
 111                                    p_len, p);
 112                        ret++;
 113                }
 114
 115                p += p_len;
 116
 117                if (*p)
 118                        p++; /* more parameters, swallow separator */
 119        }
 120        return ret;
 121}
 122
 123static int git_config_rename(const char *var, const char *value)
 124{
 125        if (!value)
 126                return DIFF_DETECT_RENAME;
 127        if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 128                return  DIFF_DETECT_COPY;
 129        return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 130}
 131
 132/*
 133 * These are to give UI layer defaults.
 134 * The core-level commands such as git-diff-files should
 135 * never be affected by the setting of diff.renames
 136 * the user happens to have in the configuration file.
 137 */
 138int git_diff_ui_config(const char *var, const char *value, void *cb)
 139{
 140        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 141                diff_use_color_default = git_config_colorbool(var, value);
 142                return 0;
 143        }
 144        if (!strcmp(var, "diff.renames")) {
 145                diff_detect_rename_default = git_config_rename(var, value);
 146                return 0;
 147        }
 148        if (!strcmp(var, "diff.autorefreshindex")) {
 149                diff_auto_refresh_index = git_config_bool(var, value);
 150                return 0;
 151        }
 152        if (!strcmp(var, "diff.mnemonicprefix")) {
 153                diff_mnemonic_prefix = git_config_bool(var, value);
 154                return 0;
 155        }
 156        if (!strcmp(var, "diff.noprefix")) {
 157                diff_no_prefix = git_config_bool(var, value);
 158                return 0;
 159        }
 160        if (!strcmp(var, "diff.statgraphwidth")) {
 161                diff_stat_graph_width = git_config_int(var, value);
 162                return 0;
 163        }
 164        if (!strcmp(var, "diff.external"))
 165                return git_config_string(&external_diff_cmd_cfg, var, value);
 166        if (!strcmp(var, "diff.wordregex"))
 167                return git_config_string(&diff_word_regex_cfg, var, value);
 168
 169        if (!strcmp(var, "diff.ignoresubmodules"))
 170                handle_ignore_submodules_arg(&default_diff_options, value);
 171
 172        if (git_color_config(var, value, cb) < 0)
 173                return -1;
 174
 175        return git_diff_basic_config(var, value, cb);
 176}
 177
 178int git_diff_basic_config(const char *var, const char *value, void *cb)
 179{
 180        if (!strcmp(var, "diff.renamelimit")) {
 181                diff_rename_limit_default = git_config_int(var, value);
 182                return 0;
 183        }
 184
 185        if (userdiff_config(var, value) < 0)
 186                return -1;
 187
 188        if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 189                int slot = parse_diff_color_slot(var, 11);
 190                if (slot < 0)
 191                        return 0;
 192                if (!value)
 193                        return config_error_nonbool(var);
 194                color_parse(value, var, diff_colors[slot]);
 195                return 0;
 196        }
 197
 198        /* like GNU diff's --suppress-blank-empty option  */
 199        if (!strcmp(var, "diff.suppressblankempty") ||
 200                        /* for backwards compatibility */
 201                        !strcmp(var, "diff.suppress-blank-empty")) {
 202                diff_suppress_blank_empty = git_config_bool(var, value);
 203                return 0;
 204        }
 205
 206        if (!strcmp(var, "diff.dirstat")) {
 207                struct strbuf errmsg = STRBUF_INIT;
 208                default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 209                if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 210                        warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 211                                errmsg.buf);
 212                strbuf_release(&errmsg);
 213                diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 214                return 0;
 215        }
 216
 217        if (!prefixcmp(var, "submodule."))
 218                return parse_submodule_config_option(var, value);
 219
 220        return git_default_config(var, value, cb);
 221}
 222
 223static char *quote_two(const char *one, const char *two)
 224{
 225        int need_one = quote_c_style(one, NULL, NULL, 1);
 226        int need_two = quote_c_style(two, NULL, NULL, 1);
 227        struct strbuf res = STRBUF_INIT;
 228
 229        if (need_one + need_two) {
 230                strbuf_addch(&res, '"');
 231                quote_c_style(one, &res, NULL, 1);
 232                quote_c_style(two, &res, NULL, 1);
 233                strbuf_addch(&res, '"');
 234        } else {
 235                strbuf_addstr(&res, one);
 236                strbuf_addstr(&res, two);
 237        }
 238        return strbuf_detach(&res, NULL);
 239}
 240
 241static const char *external_diff(void)
 242{
 243        static const char *external_diff_cmd = NULL;
 244        static int done_preparing = 0;
 245
 246        if (done_preparing)
 247                return external_diff_cmd;
 248        external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 249        if (!external_diff_cmd)
 250                external_diff_cmd = external_diff_cmd_cfg;
 251        done_preparing = 1;
 252        return external_diff_cmd;
 253}
 254
 255static struct diff_tempfile {
 256        const char *name; /* filename external diff should read from */
 257        char hex[41];
 258        char mode[10];
 259        char tmp_path[PATH_MAX];
 260} diff_temp[2];
 261
 262typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 263
 264struct emit_callback {
 265        int color_diff;
 266        unsigned ws_rule;
 267        int blank_at_eof_in_preimage;
 268        int blank_at_eof_in_postimage;
 269        int lno_in_preimage;
 270        int lno_in_postimage;
 271        sane_truncate_fn truncate;
 272        const char **label_path;
 273        struct diff_words_data *diff_words;
 274        struct diff_options *opt;
 275        int *found_changesp;
 276        struct strbuf *header;
 277};
 278
 279static int count_lines(const char *data, int size)
 280{
 281        int count, ch, completely_empty = 1, nl_just_seen = 0;
 282        count = 0;
 283        while (0 < size--) {
 284                ch = *data++;
 285                if (ch == '\n') {
 286                        count++;
 287                        nl_just_seen = 1;
 288                        completely_empty = 0;
 289                }
 290                else {
 291                        nl_just_seen = 0;
 292                        completely_empty = 0;
 293                }
 294        }
 295        if (completely_empty)
 296                return 0;
 297        if (!nl_just_seen)
 298                count++; /* no trailing newline */
 299        return count;
 300}
 301
 302static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 303{
 304        if (!DIFF_FILE_VALID(one)) {
 305                mf->ptr = (char *)""; /* does not matter */
 306                mf->size = 0;
 307                return 0;
 308        }
 309        else if (diff_populate_filespec(one, 0))
 310                return -1;
 311
 312        mf->ptr = one->data;
 313        mf->size = one->size;
 314        return 0;
 315}
 316
 317/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 318static unsigned long diff_filespec_size(struct diff_filespec *one)
 319{
 320        if (!DIFF_FILE_VALID(one))
 321                return 0;
 322        diff_populate_filespec(one, 1);
 323        return one->size;
 324}
 325
 326static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 327{
 328        char *ptr = mf->ptr;
 329        long size = mf->size;
 330        int cnt = 0;
 331
 332        if (!size)
 333                return cnt;
 334        ptr += size - 1; /* pointing at the very end */
 335        if (*ptr != '\n')
 336                ; /* incomplete line */
 337        else
 338                ptr--; /* skip the last LF */
 339        while (mf->ptr < ptr) {
 340                char *prev_eol;
 341                for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 342                        if (*prev_eol == '\n')
 343                                break;
 344                if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 345                        break;
 346                cnt++;
 347                ptr = prev_eol - 1;
 348        }
 349        return cnt;
 350}
 351
 352static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 353                               struct emit_callback *ecbdata)
 354{
 355        int l1, l2, at;
 356        unsigned ws_rule = ecbdata->ws_rule;
 357        l1 = count_trailing_blank(mf1, ws_rule);
 358        l2 = count_trailing_blank(mf2, ws_rule);
 359        if (l2 <= l1) {
 360                ecbdata->blank_at_eof_in_preimage = 0;
 361                ecbdata->blank_at_eof_in_postimage = 0;
 362                return;
 363        }
 364        at = count_lines(mf1->ptr, mf1->size);
 365        ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 366
 367        at = count_lines(mf2->ptr, mf2->size);
 368        ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 369}
 370
 371static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 372                        int first, const char *line, int len)
 373{
 374        int has_trailing_newline, has_trailing_carriage_return;
 375        int nofirst;
 376        FILE *file = o->file;
 377
 378        if (o->output_prefix) {
 379                struct strbuf *msg = NULL;
 380                msg = o->output_prefix(o, o->output_prefix_data);
 381                assert(msg);
 382                fwrite(msg->buf, msg->len, 1, file);
 383        }
 384
 385        if (len == 0) {
 386                has_trailing_newline = (first == '\n');
 387                has_trailing_carriage_return = (!has_trailing_newline &&
 388                                                (first == '\r'));
 389                nofirst = has_trailing_newline || has_trailing_carriage_return;
 390        } else {
 391                has_trailing_newline = (len > 0 && line[len-1] == '\n');
 392                if (has_trailing_newline)
 393                        len--;
 394                has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 395                if (has_trailing_carriage_return)
 396                        len--;
 397                nofirst = 0;
 398        }
 399
 400        if (len || !nofirst) {
 401                fputs(set, file);
 402                if (!nofirst)
 403                        fputc(first, file);
 404                fwrite(line, len, 1, file);
 405                fputs(reset, file);
 406        }
 407        if (has_trailing_carriage_return)
 408                fputc('\r', file);
 409        if (has_trailing_newline)
 410                fputc('\n', file);
 411}
 412
 413static void emit_line(struct diff_options *o, const char *set, const char *reset,
 414                      const char *line, int len)
 415{
 416        emit_line_0(o, set, reset, line[0], line+1, len-1);
 417}
 418
 419static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 420{
 421        if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 422              ecbdata->blank_at_eof_in_preimage &&
 423              ecbdata->blank_at_eof_in_postimage &&
 424              ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 425              ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 426                return 0;
 427        return ws_blank_line(line, len, ecbdata->ws_rule);
 428}
 429
 430static void emit_add_line(const char *reset,
 431                          struct emit_callback *ecbdata,
 432                          const char *line, int len)
 433{
 434        const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 435        const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 436
 437        if (!*ws)
 438                emit_line_0(ecbdata->opt, set, reset, '+', line, len);
 439        else if (new_blank_line_at_eof(ecbdata, line, len))
 440                /* Blank line at EOF - paint '+' as well */
 441                emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
 442        else {
 443                /* Emit just the prefix, then the rest. */
 444                emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
 445                ws_check_emit(line, len, ecbdata->ws_rule,
 446                              ecbdata->opt->file, set, reset, ws);
 447        }
 448}
 449
 450static void emit_hunk_header(struct emit_callback *ecbdata,
 451                             const char *line, int len)
 452{
 453        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 454        const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 455        const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 456        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 457        static const char atat[2] = { '@', '@' };
 458        const char *cp, *ep;
 459        struct strbuf msgbuf = STRBUF_INIT;
 460        int org_len = len;
 461        int i = 1;
 462
 463        /*
 464         * As a hunk header must begin with "@@ -<old>, +<new> @@",
 465         * it always is at least 10 bytes long.
 466         */
 467        if (len < 10 ||
 468            memcmp(line, atat, 2) ||
 469            !(ep = memmem(line + 2, len - 2, atat, 2))) {
 470                emit_line(ecbdata->opt, plain, reset, line, len);
 471                return;
 472        }
 473        ep += 2; /* skip over @@ */
 474
 475        /* The hunk header in fraginfo color */
 476        strbuf_add(&msgbuf, frag, strlen(frag));
 477        strbuf_add(&msgbuf, line, ep - line);
 478        strbuf_add(&msgbuf, reset, strlen(reset));
 479
 480        /*
 481         * trailing "\r\n"
 482         */
 483        for ( ; i < 3; i++)
 484                if (line[len - i] == '\r' || line[len - i] == '\n')
 485                        len--;
 486
 487        /* blank before the func header */
 488        for (cp = ep; ep - line < len; ep++)
 489                if (*ep != ' ' && *ep != '\t')
 490                        break;
 491        if (ep != cp) {
 492                strbuf_add(&msgbuf, plain, strlen(plain));
 493                strbuf_add(&msgbuf, cp, ep - cp);
 494                strbuf_add(&msgbuf, reset, strlen(reset));
 495        }
 496
 497        if (ep < line + len) {
 498                strbuf_add(&msgbuf, func, strlen(func));
 499                strbuf_add(&msgbuf, ep, line + len - ep);
 500                strbuf_add(&msgbuf, reset, strlen(reset));
 501        }
 502
 503        strbuf_add(&msgbuf, line + len, org_len - len);
 504        emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 505        strbuf_release(&msgbuf);
 506}
 507
 508static struct diff_tempfile *claim_diff_tempfile(void) {
 509        int i;
 510        for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 511                if (!diff_temp[i].name)
 512                        return diff_temp + i;
 513        die("BUG: diff is failing to clean up its tempfiles");
 514}
 515
 516static int remove_tempfile_installed;
 517
 518static void remove_tempfile(void)
 519{
 520        int i;
 521        for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 522                if (diff_temp[i].name == diff_temp[i].tmp_path)
 523                        unlink_or_warn(diff_temp[i].name);
 524                diff_temp[i].name = NULL;
 525        }
 526}
 527
 528static void remove_tempfile_on_signal(int signo)
 529{
 530        remove_tempfile();
 531        sigchain_pop(signo);
 532        raise(signo);
 533}
 534
 535static void print_line_count(FILE *file, int count)
 536{
 537        switch (count) {
 538        case 0:
 539                fprintf(file, "0,0");
 540                break;
 541        case 1:
 542                fprintf(file, "1");
 543                break;
 544        default:
 545                fprintf(file, "1,%d", count);
 546                break;
 547        }
 548}
 549
 550static void emit_rewrite_lines(struct emit_callback *ecb,
 551                               int prefix, const char *data, int size)
 552{
 553        const char *endp = NULL;
 554        static const char *nneof = " No newline at end of file\n";
 555        const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 556        const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 557
 558        while (0 < size) {
 559                int len;
 560
 561                endp = memchr(data, '\n', size);
 562                len = endp ? (endp - data + 1) : size;
 563                if (prefix != '+') {
 564                        ecb->lno_in_preimage++;
 565                        emit_line_0(ecb->opt, old, reset, '-',
 566                                    data, len);
 567                } else {
 568                        ecb->lno_in_postimage++;
 569                        emit_add_line(reset, ecb, data, len);
 570                }
 571                size -= len;
 572                data += len;
 573        }
 574        if (!endp) {
 575                const char *plain = diff_get_color(ecb->color_diff,
 576                                                   DIFF_PLAIN);
 577                putc('\n', ecb->opt->file);
 578                emit_line_0(ecb->opt, plain, reset, '\\',
 579                            nneof, strlen(nneof));
 580        }
 581}
 582
 583static void emit_rewrite_diff(const char *name_a,
 584                              const char *name_b,
 585                              struct diff_filespec *one,
 586                              struct diff_filespec *two,
 587                              struct userdiff_driver *textconv_one,
 588                              struct userdiff_driver *textconv_two,
 589                              struct diff_options *o)
 590{
 591        int lc_a, lc_b;
 592        const char *name_a_tab, *name_b_tab;
 593        const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 594        const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 595        const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 596        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 597        const char *a_prefix, *b_prefix;
 598        char *data_one, *data_two;
 599        size_t size_one, size_two;
 600        struct emit_callback ecbdata;
 601        char *line_prefix = "";
 602        struct strbuf *msgbuf;
 603
 604        if (o && o->output_prefix) {
 605                msgbuf = o->output_prefix(o, o->output_prefix_data);
 606                line_prefix = msgbuf->buf;
 607        }
 608
 609        if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 610                a_prefix = o->b_prefix;
 611                b_prefix = o->a_prefix;
 612        } else {
 613                a_prefix = o->a_prefix;
 614                b_prefix = o->b_prefix;
 615        }
 616
 617        name_a += (*name_a == '/');
 618        name_b += (*name_b == '/');
 619        name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 620        name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 621
 622        strbuf_reset(&a_name);
 623        strbuf_reset(&b_name);
 624        quote_two_c_style(&a_name, a_prefix, name_a, 0);
 625        quote_two_c_style(&b_name, b_prefix, name_b, 0);
 626
 627        size_one = fill_textconv(textconv_one, one, &data_one);
 628        size_two = fill_textconv(textconv_two, two, &data_two);
 629
 630        memset(&ecbdata, 0, sizeof(ecbdata));
 631        ecbdata.color_diff = want_color(o->use_color);
 632        ecbdata.found_changesp = &o->found_changes;
 633        ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 634        ecbdata.opt = o;
 635        if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 636                mmfile_t mf1, mf2;
 637                mf1.ptr = (char *)data_one;
 638                mf2.ptr = (char *)data_two;
 639                mf1.size = size_one;
 640                mf2.size = size_two;
 641                check_blank_at_eof(&mf1, &mf2, &ecbdata);
 642        }
 643        ecbdata.lno_in_preimage = 1;
 644        ecbdata.lno_in_postimage = 1;
 645
 646        lc_a = count_lines(data_one, size_one);
 647        lc_b = count_lines(data_two, size_two);
 648        fprintf(o->file,
 649                "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 650                line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 651                line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 652                line_prefix, fraginfo);
 653        if (!o->irreversible_delete)
 654                print_line_count(o->file, lc_a);
 655        else
 656                fprintf(o->file, "?,?");
 657        fprintf(o->file, " +");
 658        print_line_count(o->file, lc_b);
 659        fprintf(o->file, " @@%s\n", reset);
 660        if (lc_a && !o->irreversible_delete)
 661                emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 662        if (lc_b)
 663                emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 664        if (textconv_one)
 665                free((char *)data_one);
 666        if (textconv_two)
 667                free((char *)data_two);
 668}
 669
 670struct diff_words_buffer {
 671        mmfile_t text;
 672        long alloc;
 673        struct diff_words_orig {
 674                const char *begin, *end;
 675        } *orig;
 676        int orig_nr, orig_alloc;
 677};
 678
 679static void diff_words_append(char *line, unsigned long len,
 680                struct diff_words_buffer *buffer)
 681{
 682        ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 683        line++;
 684        len--;
 685        memcpy(buffer->text.ptr + buffer->text.size, line, len);
 686        buffer->text.size += len;
 687        buffer->text.ptr[buffer->text.size] = '\0';
 688}
 689
 690struct diff_words_style_elem {
 691        const char *prefix;
 692        const char *suffix;
 693        const char *color; /* NULL; filled in by the setup code if
 694                            * color is enabled */
 695};
 696
 697struct diff_words_style {
 698        enum diff_words_type type;
 699        struct diff_words_style_elem new, old, ctx;
 700        const char *newline;
 701};
 702
 703static struct diff_words_style diff_words_styles[] = {
 704        { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 705        { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 706        { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 707};
 708
 709struct diff_words_data {
 710        struct diff_words_buffer minus, plus;
 711        const char *current_plus;
 712        int last_minus;
 713        struct diff_options *opt;
 714        regex_t *word_regex;
 715        enum diff_words_type type;
 716        struct diff_words_style *style;
 717};
 718
 719static int fn_out_diff_words_write_helper(FILE *fp,
 720                                          struct diff_words_style_elem *st_el,
 721                                          const char *newline,
 722                                          size_t count, const char *buf,
 723                                          const char *line_prefix)
 724{
 725        int print = 0;
 726
 727        while (count) {
 728                char *p = memchr(buf, '\n', count);
 729                if (print)
 730                        fputs(line_prefix, fp);
 731                if (p != buf) {
 732                        if (st_el->color && fputs(st_el->color, fp) < 0)
 733                                return -1;
 734                        if (fputs(st_el->prefix, fp) < 0 ||
 735                            fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 736                            fputs(st_el->suffix, fp) < 0)
 737                                return -1;
 738                        if (st_el->color && *st_el->color
 739                            && fputs(GIT_COLOR_RESET, fp) < 0)
 740                                return -1;
 741                }
 742                if (!p)
 743                        return 0;
 744                if (fputs(newline, fp) < 0)
 745                        return -1;
 746                count -= p + 1 - buf;
 747                buf = p + 1;
 748                print = 1;
 749        }
 750        return 0;
 751}
 752
 753/*
 754 * '--color-words' algorithm can be described as:
 755 *
 756 *   1. collect a the minus/plus lines of a diff hunk, divided into
 757 *      minus-lines and plus-lines;
 758 *
 759 *   2. break both minus-lines and plus-lines into words and
 760 *      place them into two mmfile_t with one word for each line;
 761 *
 762 *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 763 *
 764 * And for the common parts of the both file, we output the plus side text.
 765 * diff_words->current_plus is used to trace the current position of the plus file
 766 * which printed. diff_words->last_minus is used to trace the last minus word
 767 * printed.
 768 *
 769 * For '--graph' to work with '--color-words', we need to output the graph prefix
 770 * on each line of color words output. Generally, there are two conditions on
 771 * which we should output the prefix.
 772 *
 773 *   1. diff_words->last_minus == 0 &&
 774 *      diff_words->current_plus == diff_words->plus.text.ptr
 775 *
 776 *      that is: the plus text must start as a new line, and if there is no minus
 777 *      word printed, a graph prefix must be printed.
 778 *
 779 *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 780 *      *(diff_words->current_plus - 1) == '\n'
 781 *
 782 *      that is: a graph prefix must be printed following a '\n'
 783 */
 784static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 785{
 786        if ((diff_words->last_minus == 0 &&
 787                diff_words->current_plus == diff_words->plus.text.ptr) ||
 788                (diff_words->current_plus > diff_words->plus.text.ptr &&
 789                *(diff_words->current_plus - 1) == '\n')) {
 790                return 1;
 791        } else {
 792                return 0;
 793        }
 794}
 795
 796static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 797{
 798        struct diff_words_data *diff_words = priv;
 799        struct diff_words_style *style = diff_words->style;
 800        int minus_first, minus_len, plus_first, plus_len;
 801        const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 802        struct diff_options *opt = diff_words->opt;
 803        struct strbuf *msgbuf;
 804        char *line_prefix = "";
 805
 806        if (line[0] != '@' || parse_hunk_header(line, len,
 807                        &minus_first, &minus_len, &plus_first, &plus_len))
 808                return;
 809
 810        assert(opt);
 811        if (opt->output_prefix) {
 812                msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 813                line_prefix = msgbuf->buf;
 814        }
 815
 816        /* POSIX requires that first be decremented by one if len == 0... */
 817        if (minus_len) {
 818                minus_begin = diff_words->minus.orig[minus_first].begin;
 819                minus_end =
 820                        diff_words->minus.orig[minus_first + minus_len - 1].end;
 821        } else
 822                minus_begin = minus_end =
 823                        diff_words->minus.orig[minus_first].end;
 824
 825        if (plus_len) {
 826                plus_begin = diff_words->plus.orig[plus_first].begin;
 827                plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 828        } else
 829                plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 830
 831        if (color_words_output_graph_prefix(diff_words)) {
 832                fputs(line_prefix, diff_words->opt->file);
 833        }
 834        if (diff_words->current_plus != plus_begin) {
 835                fn_out_diff_words_write_helper(diff_words->opt->file,
 836                                &style->ctx, style->newline,
 837                                plus_begin - diff_words->current_plus,
 838                                diff_words->current_plus, line_prefix);
 839                if (*(plus_begin - 1) == '\n')
 840                        fputs(line_prefix, diff_words->opt->file);
 841        }
 842        if (minus_begin != minus_end) {
 843                fn_out_diff_words_write_helper(diff_words->opt->file,
 844                                &style->old, style->newline,
 845                                minus_end - minus_begin, minus_begin,
 846                                line_prefix);
 847        }
 848        if (plus_begin != plus_end) {
 849                fn_out_diff_words_write_helper(diff_words->opt->file,
 850                                &style->new, style->newline,
 851                                plus_end - plus_begin, plus_begin,
 852                                line_prefix);
 853        }
 854
 855        diff_words->current_plus = plus_end;
 856        diff_words->last_minus = minus_first;
 857}
 858
 859/* This function starts looking at *begin, and returns 0 iff a word was found. */
 860static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 861                int *begin, int *end)
 862{
 863        if (word_regex && *begin < buffer->size) {
 864                regmatch_t match[1];
 865                if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 866                        char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 867                                        '\n', match[0].rm_eo - match[0].rm_so);
 868                        *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 869                        *begin += match[0].rm_so;
 870                        return *begin >= *end;
 871                }
 872                return -1;
 873        }
 874
 875        /* find the next word */
 876        while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 877                (*begin)++;
 878        if (*begin >= buffer->size)
 879                return -1;
 880
 881        /* find the end of the word */
 882        *end = *begin + 1;
 883        while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 884                (*end)++;
 885
 886        return 0;
 887}
 888
 889/*
 890 * This function splits the words in buffer->text, stores the list with
 891 * newline separator into out, and saves the offsets of the original words
 892 * in buffer->orig.
 893 */
 894static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 895                regex_t *word_regex)
 896{
 897        int i, j;
 898        long alloc = 0;
 899
 900        out->size = 0;
 901        out->ptr = NULL;
 902
 903        /* fake an empty "0th" word */
 904        ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 905        buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 906        buffer->orig_nr = 1;
 907
 908        for (i = 0; i < buffer->text.size; i++) {
 909                if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 910                        return;
 911
 912                /* store original boundaries */
 913                ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 914                                buffer->orig_alloc);
 915                buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 916                buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 917                buffer->orig_nr++;
 918
 919                /* store one word */
 920                ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 921                memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 922                out->ptr[out->size + j - i] = '\n';
 923                out->size += j - i + 1;
 924
 925                i = j - 1;
 926        }
 927}
 928
 929/* this executes the word diff on the accumulated buffers */
 930static void diff_words_show(struct diff_words_data *diff_words)
 931{
 932        xpparam_t xpp;
 933        xdemitconf_t xecfg;
 934        mmfile_t minus, plus;
 935        struct diff_words_style *style = diff_words->style;
 936
 937        struct diff_options *opt = diff_words->opt;
 938        struct strbuf *msgbuf;
 939        char *line_prefix = "";
 940
 941        assert(opt);
 942        if (opt->output_prefix) {
 943                msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 944                line_prefix = msgbuf->buf;
 945        }
 946
 947        /* special case: only removal */
 948        if (!diff_words->plus.text.size) {
 949                fputs(line_prefix, diff_words->opt->file);
 950                fn_out_diff_words_write_helper(diff_words->opt->file,
 951                        &style->old, style->newline,
 952                        diff_words->minus.text.size,
 953                        diff_words->minus.text.ptr, line_prefix);
 954                diff_words->minus.text.size = 0;
 955                return;
 956        }
 957
 958        diff_words->current_plus = diff_words->plus.text.ptr;
 959        diff_words->last_minus = 0;
 960
 961        memset(&xpp, 0, sizeof(xpp));
 962        memset(&xecfg, 0, sizeof(xecfg));
 963        diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 964        diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 965        xpp.flags = 0;
 966        /* as only the hunk header will be parsed, we need a 0-context */
 967        xecfg.ctxlen = 0;
 968        xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 969                      &xpp, &xecfg);
 970        free(minus.ptr);
 971        free(plus.ptr);
 972        if (diff_words->current_plus != diff_words->plus.text.ptr +
 973                        diff_words->plus.text.size) {
 974                if (color_words_output_graph_prefix(diff_words))
 975                        fputs(line_prefix, diff_words->opt->file);
 976                fn_out_diff_words_write_helper(diff_words->opt->file,
 977                        &style->ctx, style->newline,
 978                        diff_words->plus.text.ptr + diff_words->plus.text.size
 979                        - diff_words->current_plus, diff_words->current_plus,
 980                        line_prefix);
 981        }
 982        diff_words->minus.text.size = diff_words->plus.text.size = 0;
 983}
 984
 985/* In "color-words" mode, show word-diff of words accumulated in the buffer */
 986static void diff_words_flush(struct emit_callback *ecbdata)
 987{
 988        if (ecbdata->diff_words->minus.text.size ||
 989            ecbdata->diff_words->plus.text.size)
 990                diff_words_show(ecbdata->diff_words);
 991}
 992
 993static void diff_filespec_load_driver(struct diff_filespec *one)
 994{
 995        /* Use already-loaded driver */
 996        if (one->driver)
 997                return;
 998
 999        if (S_ISREG(one->mode))
1000                one->driver = userdiff_find_by_path(one->path);
1001
1002        /* Fallback to default settings */
1003        if (!one->driver)
1004                one->driver = userdiff_find_by_name("default");
1005}
1006
1007static const char *userdiff_word_regex(struct diff_filespec *one)
1008{
1009        diff_filespec_load_driver(one);
1010        return one->driver->word_regex;
1011}
1012
1013static void init_diff_words_data(struct emit_callback *ecbdata,
1014                                 struct diff_options *orig_opts,
1015                                 struct diff_filespec *one,
1016                                 struct diff_filespec *two)
1017{
1018        int i;
1019        struct diff_options *o = xmalloc(sizeof(struct diff_options));
1020        memcpy(o, orig_opts, sizeof(struct diff_options));
1021
1022        ecbdata->diff_words =
1023                xcalloc(1, sizeof(struct diff_words_data));
1024        ecbdata->diff_words->type = o->word_diff;
1025        ecbdata->diff_words->opt = o;
1026        if (!o->word_regex)
1027                o->word_regex = userdiff_word_regex(one);
1028        if (!o->word_regex)
1029                o->word_regex = userdiff_word_regex(two);
1030        if (!o->word_regex)
1031                o->word_regex = diff_word_regex_cfg;
1032        if (o->word_regex) {
1033                ecbdata->diff_words->word_regex = (regex_t *)
1034                        xmalloc(sizeof(regex_t));
1035                if (regcomp(ecbdata->diff_words->word_regex,
1036                            o->word_regex,
1037                            REG_EXTENDED | REG_NEWLINE))
1038                        die ("Invalid regular expression: %s",
1039                             o->word_regex);
1040        }
1041        for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1042                if (o->word_diff == diff_words_styles[i].type) {
1043                        ecbdata->diff_words->style =
1044                                &diff_words_styles[i];
1045                        break;
1046                }
1047        }
1048        if (want_color(o->use_color)) {
1049                struct diff_words_style *st = ecbdata->diff_words->style;
1050                st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1051                st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1052                st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1053        }
1054}
1055
1056static void free_diff_words_data(struct emit_callback *ecbdata)
1057{
1058        if (ecbdata->diff_words) {
1059                diff_words_flush(ecbdata);
1060                free (ecbdata->diff_words->opt);
1061                free (ecbdata->diff_words->minus.text.ptr);
1062                free (ecbdata->diff_words->minus.orig);
1063                free (ecbdata->diff_words->plus.text.ptr);
1064                free (ecbdata->diff_words->plus.orig);
1065                if (ecbdata->diff_words->word_regex) {
1066                        regfree(ecbdata->diff_words->word_regex);
1067                        free(ecbdata->diff_words->word_regex);
1068                }
1069                free(ecbdata->diff_words);
1070                ecbdata->diff_words = NULL;
1071        }
1072}
1073
1074const char *diff_get_color(int diff_use_color, enum color_diff ix)
1075{
1076        if (want_color(diff_use_color))
1077                return diff_colors[ix];
1078        return "";
1079}
1080
1081static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1082{
1083        const char *cp;
1084        unsigned long allot;
1085        size_t l = len;
1086
1087        if (ecb->truncate)
1088                return ecb->truncate(line, len);
1089        cp = line;
1090        allot = l;
1091        while (0 < l) {
1092                (void) utf8_width(&cp, &l);
1093                if (!cp)
1094                        break; /* truncated in the middle? */
1095        }
1096        return allot - l;
1097}
1098
1099static void find_lno(const char *line, struct emit_callback *ecbdata)
1100{
1101        const char *p;
1102        ecbdata->lno_in_preimage = 0;
1103        ecbdata->lno_in_postimage = 0;
1104        p = strchr(line, '-');
1105        if (!p)
1106                return; /* cannot happen */
1107        ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1108        p = strchr(p, '+');
1109        if (!p)
1110                return; /* cannot happen */
1111        ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1112}
1113
1114static void fn_out_consume(void *priv, char *line, unsigned long len)
1115{
1116        struct emit_callback *ecbdata = priv;
1117        const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1118        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1119        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1120        struct diff_options *o = ecbdata->opt;
1121        char *line_prefix = "";
1122        struct strbuf *msgbuf;
1123
1124        if (o && o->output_prefix) {
1125                msgbuf = o->output_prefix(o, o->output_prefix_data);
1126                line_prefix = msgbuf->buf;
1127        }
1128
1129        if (ecbdata->header) {
1130                fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1131                strbuf_reset(ecbdata->header);
1132                ecbdata->header = NULL;
1133        }
1134        *(ecbdata->found_changesp) = 1;
1135
1136        if (ecbdata->label_path[0]) {
1137                const char *name_a_tab, *name_b_tab;
1138
1139                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1140                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1141
1142                fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1143                        line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1144                fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1145                        line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1146                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1147        }
1148
1149        if (diff_suppress_blank_empty
1150            && len == 2 && line[0] == ' ' && line[1] == '\n') {
1151                line[0] = '\n';
1152                len = 1;
1153        }
1154
1155        if (line[0] == '@') {
1156                if (ecbdata->diff_words)
1157                        diff_words_flush(ecbdata);
1158                len = sane_truncate_line(ecbdata, line, len);
1159                find_lno(line, ecbdata);
1160                emit_hunk_header(ecbdata, line, len);
1161                if (line[len-1] != '\n')
1162                        putc('\n', ecbdata->opt->file);
1163                return;
1164        }
1165
1166        if (len < 1) {
1167                emit_line(ecbdata->opt, reset, reset, line, len);
1168                if (ecbdata->diff_words
1169                    && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1170                        fputs("~\n", ecbdata->opt->file);
1171                return;
1172        }
1173
1174        if (ecbdata->diff_words) {
1175                if (line[0] == '-') {
1176                        diff_words_append(line, len,
1177                                          &ecbdata->diff_words->minus);
1178                        return;
1179                } else if (line[0] == '+') {
1180                        diff_words_append(line, len,
1181                                          &ecbdata->diff_words->plus);
1182                        return;
1183                } else if (!prefixcmp(line, "\\ ")) {
1184                        /*
1185                         * Eat the "no newline at eof" marker as if we
1186                         * saw a "+" or "-" line with nothing on it,
1187                         * and return without diff_words_flush() to
1188                         * defer processing. If this is the end of
1189                         * preimage, more "+" lines may come after it.
1190                         */
1191                        return;
1192                }
1193                diff_words_flush(ecbdata);
1194                if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1195                        emit_line(ecbdata->opt, plain, reset, line, len);
1196                        fputs("~\n", ecbdata->opt->file);
1197                } else {
1198                        /*
1199                         * Skip the prefix character, if any.  With
1200                         * diff_suppress_blank_empty, there may be
1201                         * none.
1202                         */
1203                        if (line[0] != '\n') {
1204                              line++;
1205                              len--;
1206                        }
1207                        emit_line(ecbdata->opt, plain, reset, line, len);
1208                }
1209                return;
1210        }
1211
1212        if (line[0] != '+') {
1213                const char *color =
1214                        diff_get_color(ecbdata->color_diff,
1215                                       line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1216                ecbdata->lno_in_preimage++;
1217                if (line[0] == ' ')
1218                        ecbdata->lno_in_postimage++;
1219                emit_line(ecbdata->opt, color, reset, line, len);
1220        } else {
1221                ecbdata->lno_in_postimage++;
1222                emit_add_line(reset, ecbdata, line + 1, len - 1);
1223        }
1224}
1225
1226static char *pprint_rename(const char *a, const char *b)
1227{
1228        const char *old = a;
1229        const char *new = b;
1230        struct strbuf name = STRBUF_INIT;
1231        int pfx_length, sfx_length;
1232        int len_a = strlen(a);
1233        int len_b = strlen(b);
1234        int a_midlen, b_midlen;
1235        int qlen_a = quote_c_style(a, NULL, NULL, 0);
1236        int qlen_b = quote_c_style(b, NULL, NULL, 0);
1237
1238        if (qlen_a || qlen_b) {
1239                quote_c_style(a, &name, NULL, 0);
1240                strbuf_addstr(&name, " => ");
1241                quote_c_style(b, &name, NULL, 0);
1242                return strbuf_detach(&name, NULL);
1243        }
1244
1245        /* Find common prefix */
1246        pfx_length = 0;
1247        while (*old && *new && *old == *new) {
1248                if (*old == '/')
1249                        pfx_length = old - a + 1;
1250                old++;
1251                new++;
1252        }
1253
1254        /* Find common suffix */
1255        old = a + len_a;
1256        new = b + len_b;
1257        sfx_length = 0;
1258        while (a <= old && b <= new && *old == *new) {
1259                if (*old == '/')
1260                        sfx_length = len_a - (old - a);
1261                old--;
1262                new--;
1263        }
1264
1265        /*
1266         * pfx{mid-a => mid-b}sfx
1267         * {pfx-a => pfx-b}sfx
1268         * pfx{sfx-a => sfx-b}
1269         * name-a => name-b
1270         */
1271        a_midlen = len_a - pfx_length - sfx_length;
1272        b_midlen = len_b - pfx_length - sfx_length;
1273        if (a_midlen < 0)
1274                a_midlen = 0;
1275        if (b_midlen < 0)
1276                b_midlen = 0;
1277
1278        strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1279        if (pfx_length + sfx_length) {
1280                strbuf_add(&name, a, pfx_length);
1281                strbuf_addch(&name, '{');
1282        }
1283        strbuf_add(&name, a + pfx_length, a_midlen);
1284        strbuf_addstr(&name, " => ");
1285        strbuf_add(&name, b + pfx_length, b_midlen);
1286        if (pfx_length + sfx_length) {
1287                strbuf_addch(&name, '}');
1288                strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1289        }
1290        return strbuf_detach(&name, NULL);
1291}
1292
1293struct diffstat_t {
1294        int nr;
1295        int alloc;
1296        struct diffstat_file {
1297                char *from_name;
1298                char *name;
1299                char *print_name;
1300                unsigned is_unmerged:1;
1301                unsigned is_binary:1;
1302                unsigned is_renamed:1;
1303                unsigned is_interesting:1;
1304                uintmax_t added, deleted;
1305        } **files;
1306};
1307
1308static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1309                                          const char *name_a,
1310                                          const char *name_b)
1311{
1312        struct diffstat_file *x;
1313        x = xcalloc(sizeof (*x), 1);
1314        if (diffstat->nr == diffstat->alloc) {
1315                diffstat->alloc = alloc_nr(diffstat->alloc);
1316                diffstat->files = xrealloc(diffstat->files,
1317                                diffstat->alloc * sizeof(x));
1318        }
1319        diffstat->files[diffstat->nr++] = x;
1320        if (name_b) {
1321                x->from_name = xstrdup(name_a);
1322                x->name = xstrdup(name_b);
1323                x->is_renamed = 1;
1324        }
1325        else {
1326                x->from_name = NULL;
1327                x->name = xstrdup(name_a);
1328        }
1329        return x;
1330}
1331
1332static void diffstat_consume(void *priv, char *line, unsigned long len)
1333{
1334        struct diffstat_t *diffstat = priv;
1335        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1336
1337        if (line[0] == '+')
1338                x->added++;
1339        else if (line[0] == '-')
1340                x->deleted++;
1341}
1342
1343const char mime_boundary_leader[] = "------------";
1344
1345static int scale_linear(int it, int width, int max_change)
1346{
1347        if (!it)
1348                return 0;
1349        /*
1350         * make sure that at least one '-' or '+' is printed if
1351         * there is any change to this path. The easiest way is to
1352         * scale linearly as if the alloted width is one column shorter
1353         * than it is, and then add 1 to the result.
1354         */
1355        return 1 + (it * (width - 1) / max_change);
1356}
1357
1358static void show_name(FILE *file,
1359                      const char *prefix, const char *name, int len)
1360{
1361        fprintf(file, " %s%-*s |", prefix, len, name);
1362}
1363
1364static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1365{
1366        if (cnt <= 0)
1367                return;
1368        fprintf(file, "%s", set);
1369        while (cnt--)
1370                putc(ch, file);
1371        fprintf(file, "%s", reset);
1372}
1373
1374static void fill_print_name(struct diffstat_file *file)
1375{
1376        char *pname;
1377
1378        if (file->print_name)
1379                return;
1380
1381        if (!file->is_renamed) {
1382                struct strbuf buf = STRBUF_INIT;
1383                if (quote_c_style(file->name, &buf, NULL, 0)) {
1384                        pname = strbuf_detach(&buf, NULL);
1385                } else {
1386                        pname = file->name;
1387                        strbuf_release(&buf);
1388                }
1389        } else {
1390                pname = pprint_rename(file->from_name, file->name);
1391        }
1392        file->print_name = pname;
1393}
1394
1395int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1396{
1397        struct strbuf sb = STRBUF_INIT;
1398        int ret;
1399
1400        if (!files) {
1401                assert(insertions == 0 && deletions == 0);
1402                return fprintf(fp, "%s\n", " 0 files changed");
1403        }
1404
1405        strbuf_addf(&sb,
1406                    (files == 1) ? " %d file changed" : " %d files changed",
1407                    files);
1408
1409        /*
1410         * For binary diff, the caller may want to print "x files
1411         * changed" with insertions == 0 && deletions == 0.
1412         *
1413         * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1414         * is probably less confusing (i.e skip over "2 files changed
1415         * but nothing about added/removed lines? Is this a bug in Git?").
1416         */
1417        if (insertions || deletions == 0) {
1418                /*
1419                 * TRANSLATORS: "+" in (+) is a line addition marker;
1420                 * do not translate it.
1421                 */
1422                strbuf_addf(&sb,
1423                            (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1424                            insertions);
1425        }
1426
1427        if (deletions || insertions == 0) {
1428                /*
1429                 * TRANSLATORS: "-" in (-) is a line removal marker;
1430                 * do not translate it.
1431                 */
1432                strbuf_addf(&sb,
1433                            (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1434                            deletions);
1435        }
1436        strbuf_addch(&sb, '\n');
1437        ret = fputs(sb.buf, fp);
1438        strbuf_release(&sb);
1439        return ret;
1440}
1441
1442static void show_stats(struct diffstat_t *data, struct diff_options *options)
1443{
1444        int i, len, add, del, adds = 0, dels = 0;
1445        uintmax_t max_change = 0, max_len = 0;
1446        int total_files = data->nr, count;
1447        int width, name_width, graph_width, number_width = 0, bin_width = 0;
1448        const char *reset, *add_c, *del_c;
1449        const char *line_prefix = "";
1450        int extra_shown = 0;
1451        struct strbuf *msg = NULL;
1452
1453        if (data->nr == 0)
1454                return;
1455
1456        if (options->output_prefix) {
1457                msg = options->output_prefix(options, options->output_prefix_data);
1458                line_prefix = msg->buf;
1459        }
1460
1461        count = options->stat_count ? options->stat_count : data->nr;
1462
1463        reset = diff_get_color_opt(options, DIFF_RESET);
1464        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1465        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1466
1467        /*
1468         * Find the longest filename and max number of changes
1469         */
1470        for (i = 0; (i < count) && (i < data->nr); i++) {
1471                struct diffstat_file *file = data->files[i];
1472                uintmax_t change = file->added + file->deleted;
1473
1474                if (!file->is_interesting && (change == 0)) {
1475                        count++; /* not shown == room for one more */
1476                        continue;
1477                }
1478                fill_print_name(file);
1479                len = strlen(file->print_name);
1480                if (max_len < len)
1481                        max_len = len;
1482
1483                if (file->is_unmerged) {
1484                        /* "Unmerged" is 8 characters */
1485                        bin_width = bin_width < 8 ? 8 : bin_width;
1486                        continue;
1487                }
1488                if (file->is_binary) {
1489                        /* "Bin XXX -> YYY bytes" */
1490                        int w = 14 + decimal_width(file->added)
1491                                + decimal_width(file->deleted);
1492                        bin_width = bin_width < w ? w : bin_width;
1493                        /* Display change counts aligned with "Bin" */
1494                        number_width = 3;
1495                        continue;
1496                }
1497
1498                if (max_change < change)
1499                        max_change = change;
1500        }
1501        count = i; /* where we can stop scanning in data->files[] */
1502
1503        /*
1504         * We have width = stat_width or term_columns() columns total.
1505         * We want a maximum of min(max_len, stat_name_width) for the name part.
1506         * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1507         * We also need 1 for " " and 4 + decimal_width(max_change)
1508         * for " | NNNN " and one the empty column at the end, altogether
1509         * 6 + decimal_width(max_change).
1510         *
1511         * If there's not enough space, we will use the smaller of
1512         * stat_name_width (if set) and 5/8*width for the filename,
1513         * and the rest for constant elements + graph part, but no more
1514         * than stat_graph_width for the graph part.
1515         * (5/8 gives 50 for filename and 30 for the constant parts + graph
1516         * for the standard terminal size).
1517         *
1518         * In other words: stat_width limits the maximum width, and
1519         * stat_name_width fixes the maximum width of the filename,
1520         * and is also used to divide available columns if there
1521         * aren't enough.
1522         *
1523         * Binary files are displayed with "Bin XXX -> YYY bytes"
1524         * instead of the change count and graph. This part is treated
1525         * similarly to the graph part, except that it is not
1526         * "scaled". If total width is too small to accomodate the
1527         * guaranteed minimum width of the filename part and the
1528         * separators and this message, this message will "overflow"
1529         * making the line longer than the maximum width.
1530         */
1531
1532        if (options->stat_width == -1)
1533                width = term_columns() - options->output_prefix_length;
1534        else
1535                width = options->stat_width ? options->stat_width : 80;
1536        number_width = decimal_width(max_change) > number_width ?
1537                decimal_width(max_change) : number_width;
1538
1539        if (options->stat_graph_width == -1)
1540                options->stat_graph_width = diff_stat_graph_width;
1541
1542        /*
1543         * Guarantee 3/8*16==6 for the graph part
1544         * and 5/8*16==10 for the filename part
1545         */
1546        if (width < 16 + 6 + number_width)
1547                width = 16 + 6 + number_width;
1548
1549        /*
1550         * First assign sizes that are wanted, ignoring available width.
1551         * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1552         * starting from "XXX" should fit in graph_width.
1553         */
1554        graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1555        if (options->stat_graph_width &&
1556            options->stat_graph_width < graph_width)
1557                graph_width = options->stat_graph_width;
1558
1559        name_width = (options->stat_name_width > 0 &&
1560                      options->stat_name_width < max_len) ?
1561                options->stat_name_width : max_len;
1562
1563        /*
1564         * Adjust adjustable widths not to exceed maximum width
1565         */
1566        if (name_width + number_width + 6 + graph_width > width) {
1567                if (graph_width > width * 3/8 - number_width - 6) {
1568                        graph_width = width * 3/8 - number_width - 6;
1569                        if (graph_width < 6)
1570                                graph_width = 6;
1571                }
1572
1573                if (options->stat_graph_width &&
1574                    graph_width > options->stat_graph_width)
1575                        graph_width = options->stat_graph_width;
1576                if (name_width > width - number_width - 6 - graph_width)
1577                        name_width = width - number_width - 6 - graph_width;
1578                else
1579                        graph_width = width - number_width - 6 - name_width;
1580        }
1581
1582        /*
1583         * From here name_width is the width of the name area,
1584         * and graph_width is the width of the graph area.
1585         * max_change is used to scale graph properly.
1586         */
1587        for (i = 0; i < count; i++) {
1588                const char *prefix = "";
1589                struct diffstat_file *file = data->files[i];
1590                char *name = file->print_name;
1591                uintmax_t added = file->added;
1592                uintmax_t deleted = file->deleted;
1593                int name_len;
1594
1595                if (!file->is_interesting && (added + deleted == 0))
1596                        continue;
1597
1598                /*
1599                 * "scale" the filename
1600                 */
1601                len = name_width;
1602                name_len = strlen(name);
1603                if (name_width < name_len) {
1604                        char *slash;
1605                        prefix = "...";
1606                        len -= 3;
1607                        name += name_len - len;
1608                        slash = strchr(name, '/');
1609                        if (slash)
1610                                name = slash;
1611                }
1612
1613                if (file->is_binary) {
1614                        fprintf(options->file, "%s", line_prefix);
1615                        show_name(options->file, prefix, name, len);
1616                        fprintf(options->file, " %*s", number_width, "Bin");
1617                        if (!added && !deleted) {
1618                                putc('\n', options->file);
1619                                continue;
1620                        }
1621                        fprintf(options->file, " %s%"PRIuMAX"%s",
1622                                del_c, deleted, reset);
1623                        fprintf(options->file, " -> ");
1624                        fprintf(options->file, "%s%"PRIuMAX"%s",
1625                                add_c, added, reset);
1626                        fprintf(options->file, " bytes");
1627                        fprintf(options->file, "\n");
1628                        continue;
1629                }
1630                else if (file->is_unmerged) {
1631                        fprintf(options->file, "%s", line_prefix);
1632                        show_name(options->file, prefix, name, len);
1633                        fprintf(options->file, " Unmerged\n");
1634                        continue;
1635                }
1636
1637                /*
1638                 * scale the add/delete
1639                 */
1640                add = added;
1641                del = deleted;
1642
1643                if (graph_width <= max_change) {
1644                        int total = add + del;
1645
1646                        total = scale_linear(add + del, graph_width, max_change);
1647                        if (total < 2 && add && del)
1648                                /* width >= 2 due to the sanity check */
1649                                total = 2;
1650                        if (add < del) {
1651                                add = scale_linear(add, graph_width, max_change);
1652                                del = total - add;
1653                        } else {
1654                                del = scale_linear(del, graph_width, max_change);
1655                                add = total - del;
1656                        }
1657                }
1658                fprintf(options->file, "%s", line_prefix);
1659                show_name(options->file, prefix, name, len);
1660                fprintf(options->file, " %*"PRIuMAX"%s",
1661                        number_width, added + deleted,
1662                        added + deleted ? " " : "");
1663                show_graph(options->file, '+', add, add_c, reset);
1664                show_graph(options->file, '-', del, del_c, reset);
1665                fprintf(options->file, "\n");
1666        }
1667
1668        for (i = 0; i < data->nr; i++) {
1669                struct diffstat_file *file = data->files[i];
1670                uintmax_t added = file->added;
1671                uintmax_t deleted = file->deleted;
1672                if (!file->is_interesting && (added + deleted == 0)) {
1673                        total_files--;
1674                        continue;
1675                }
1676
1677                if (!file->is_binary && !file->is_unmerged) {
1678                        adds += added;
1679                        dels += deleted;
1680                }
1681                if (i < count)
1682                        continue;
1683                if (!extra_shown)
1684                        fprintf(options->file, "%s ...\n", line_prefix);
1685                extra_shown = 1;
1686        }
1687        fprintf(options->file, "%s", line_prefix);
1688        print_stat_summary(options->file, total_files, adds, dels);
1689}
1690
1691static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1692{
1693        int i, adds = 0, dels = 0, total_files = data->nr;
1694
1695        if (data->nr == 0)
1696                return;
1697
1698        for (i = 0; i < data->nr; i++) {
1699                int added = data->files[i]->added;
1700                int deleted= data->files[i]->deleted;
1701
1702                if (data->files[i]->is_unmerged)
1703                        continue;
1704                if (!data->files[i]->is_interesting && (added + deleted == 0)) {
1705                        total_files--;
1706                } else if (!data->files[i]->is_binary) { /* don't count bytes */
1707                        adds += added;
1708                        dels += deleted;
1709                }
1710        }
1711        if (options->output_prefix) {
1712                struct strbuf *msg = NULL;
1713                msg = options->output_prefix(options,
1714                                options->output_prefix_data);
1715                fprintf(options->file, "%s", msg->buf);
1716        }
1717        print_stat_summary(options->file, total_files, adds, dels);
1718}
1719
1720static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1721{
1722        int i;
1723
1724        if (data->nr == 0)
1725                return;
1726
1727        for (i = 0; i < data->nr; i++) {
1728                struct diffstat_file *file = data->files[i];
1729
1730                if (options->output_prefix) {
1731                        struct strbuf *msg = NULL;
1732                        msg = options->output_prefix(options,
1733                                        options->output_prefix_data);
1734                        fprintf(options->file, "%s", msg->buf);
1735                }
1736
1737                if (file->is_binary)
1738                        fprintf(options->file, "-\t-\t");
1739                else
1740                        fprintf(options->file,
1741                                "%"PRIuMAX"\t%"PRIuMAX"\t",
1742                                file->added, file->deleted);
1743                if (options->line_termination) {
1744                        fill_print_name(file);
1745                        if (!file->is_renamed)
1746                                write_name_quoted(file->name, options->file,
1747                                                  options->line_termination);
1748                        else {
1749                                fputs(file->print_name, options->file);
1750                                putc(options->line_termination, options->file);
1751                        }
1752                } else {
1753                        if (file->is_renamed) {
1754                                putc('\0', options->file);
1755                                write_name_quoted(file->from_name, options->file, '\0');
1756                        }
1757                        write_name_quoted(file->name, options->file, '\0');
1758                }
1759        }
1760}
1761
1762struct dirstat_file {
1763        const char *name;
1764        unsigned long changed;
1765};
1766
1767struct dirstat_dir {
1768        struct dirstat_file *files;
1769        int alloc, nr, permille, cumulative;
1770};
1771
1772static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1773                unsigned long changed, const char *base, int baselen)
1774{
1775        unsigned long this_dir = 0;
1776        unsigned int sources = 0;
1777        const char *line_prefix = "";
1778        struct strbuf *msg = NULL;
1779
1780        if (opt->output_prefix) {
1781                msg = opt->output_prefix(opt, opt->output_prefix_data);
1782                line_prefix = msg->buf;
1783        }
1784
1785        while (dir->nr) {
1786                struct dirstat_file *f = dir->files;
1787                int namelen = strlen(f->name);
1788                unsigned long this;
1789                char *slash;
1790
1791                if (namelen < baselen)
1792                        break;
1793                if (memcmp(f->name, base, baselen))
1794                        break;
1795                slash = strchr(f->name + baselen, '/');
1796                if (slash) {
1797                        int newbaselen = slash + 1 - f->name;
1798                        this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1799                        sources++;
1800                } else {
1801                        this = f->changed;
1802                        dir->files++;
1803                        dir->nr--;
1804                        sources += 2;
1805                }
1806                this_dir += this;
1807        }
1808
1809        /*
1810         * We don't report dirstat's for
1811         *  - the top level
1812         *  - or cases where everything came from a single directory
1813         *    under this directory (sources == 1).
1814         */
1815        if (baselen && sources != 1) {
1816                if (this_dir) {
1817                        int permille = this_dir * 1000 / changed;
1818                        if (permille >= dir->permille) {
1819                                fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1820                                        permille / 10, permille % 10, baselen, base);
1821                                if (!dir->cumulative)
1822                                        return 0;
1823                        }
1824                }
1825        }
1826        return this_dir;
1827}
1828
1829static int dirstat_compare(const void *_a, const void *_b)
1830{
1831        const struct dirstat_file *a = _a;
1832        const struct dirstat_file *b = _b;
1833        return strcmp(a->name, b->name);
1834}
1835
1836static void show_dirstat(struct diff_options *options)
1837{
1838        int i;
1839        unsigned long changed;
1840        struct dirstat_dir dir;
1841        struct diff_queue_struct *q = &diff_queued_diff;
1842
1843        dir.files = NULL;
1844        dir.alloc = 0;
1845        dir.nr = 0;
1846        dir.permille = options->dirstat_permille;
1847        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1848
1849        changed = 0;
1850        for (i = 0; i < q->nr; i++) {
1851                struct diff_filepair *p = q->queue[i];
1852                const char *name;
1853                unsigned long copied, added, damage;
1854                int content_changed;
1855
1856                name = p->two->path ? p->two->path : p->one->path;
1857
1858                if (p->one->sha1_valid && p->two->sha1_valid)
1859                        content_changed = hashcmp(p->one->sha1, p->two->sha1);
1860                else
1861                        content_changed = 1;
1862
1863                if (!content_changed) {
1864                        /*
1865                         * The SHA1 has not changed, so pre-/post-content is
1866                         * identical. We can therefore skip looking at the
1867                         * file contents altogether.
1868                         */
1869                        damage = 0;
1870                        goto found_damage;
1871                }
1872
1873                if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1874                        /*
1875                         * In --dirstat-by-file mode, we don't really need to
1876                         * look at the actual file contents at all.
1877                         * The fact that the SHA1 changed is enough for us to
1878                         * add this file to the list of results
1879                         * (with each file contributing equal damage).
1880                         */
1881                        damage = 1;
1882                        goto found_damage;
1883                }
1884
1885                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1886                        diff_populate_filespec(p->one, 0);
1887                        diff_populate_filespec(p->two, 0);
1888                        diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1889                                               &copied, &added);
1890                        diff_free_filespec_data(p->one);
1891                        diff_free_filespec_data(p->two);
1892                } else if (DIFF_FILE_VALID(p->one)) {
1893                        diff_populate_filespec(p->one, 1);
1894                        copied = added = 0;
1895                        diff_free_filespec_data(p->one);
1896                } else if (DIFF_FILE_VALID(p->two)) {
1897                        diff_populate_filespec(p->two, 1);
1898                        copied = 0;
1899                        added = p->two->size;
1900                        diff_free_filespec_data(p->two);
1901                } else
1902                        continue;
1903
1904                /*
1905                 * Original minus copied is the removed material,
1906                 * added is the new material.  They are both damages
1907                 * made to the preimage.
1908                 * If the resulting damage is zero, we know that
1909                 * diffcore_count_changes() considers the two entries to
1910                 * be identical, but since content_changed is true, we
1911                 * know that there must have been _some_ kind of change,
1912                 * so we force all entries to have damage > 0.
1913                 */
1914                damage = (p->one->size - copied) + added;
1915                if (!damage)
1916                        damage = 1;
1917
1918found_damage:
1919                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1920                dir.files[dir.nr].name = name;
1921                dir.files[dir.nr].changed = damage;
1922                changed += damage;
1923                dir.nr++;
1924        }
1925
1926        /* This can happen even with many files, if everything was renames */
1927        if (!changed)
1928                return;
1929
1930        /* Show all directories with more than x% of the changes */
1931        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1932        gather_dirstat(options, &dir, changed, "", 0);
1933}
1934
1935static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1936{
1937        int i;
1938        unsigned long changed;
1939        struct dirstat_dir dir;
1940
1941        if (data->nr == 0)
1942                return;
1943
1944        dir.files = NULL;
1945        dir.alloc = 0;
1946        dir.nr = 0;
1947        dir.permille = options->dirstat_permille;
1948        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1949
1950        changed = 0;
1951        for (i = 0; i < data->nr; i++) {
1952                struct diffstat_file *file = data->files[i];
1953                unsigned long damage = file->added + file->deleted;
1954                if (file->is_binary)
1955                        /*
1956                         * binary files counts bytes, not lines. Must find some
1957                         * way to normalize binary bytes vs. textual lines.
1958                         * The following heuristic assumes that there are 64
1959                         * bytes per "line".
1960                         * This is stupid and ugly, but very cheap...
1961                         */
1962                        damage = (damage + 63) / 64;
1963                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1964                dir.files[dir.nr].name = file->name;
1965                dir.files[dir.nr].changed = damage;
1966                changed += damage;
1967                dir.nr++;
1968        }
1969
1970        /* This can happen even with many files, if everything was renames */
1971        if (!changed)
1972                return;
1973
1974        /* Show all directories with more than x% of the changes */
1975        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1976        gather_dirstat(options, &dir, changed, "", 0);
1977}
1978
1979static void free_diffstat_info(struct diffstat_t *diffstat)
1980{
1981        int i;
1982        for (i = 0; i < diffstat->nr; i++) {
1983                struct diffstat_file *f = diffstat->files[i];
1984                if (f->name != f->print_name)
1985                        free(f->print_name);
1986                free(f->name);
1987                free(f->from_name);
1988                free(f);
1989        }
1990        free(diffstat->files);
1991}
1992
1993struct checkdiff_t {
1994        const char *filename;
1995        int lineno;
1996        int conflict_marker_size;
1997        struct diff_options *o;
1998        unsigned ws_rule;
1999        unsigned status;
2000};
2001
2002static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2003{
2004        char firstchar;
2005        int cnt;
2006
2007        if (len < marker_size + 1)
2008                return 0;
2009        firstchar = line[0];
2010        switch (firstchar) {
2011        case '=': case '>': case '<': case '|':
2012                break;
2013        default:
2014                return 0;
2015        }
2016        for (cnt = 1; cnt < marker_size; cnt++)
2017                if (line[cnt] != firstchar)
2018                        return 0;
2019        /* line[1] thru line[marker_size-1] are same as firstchar */
2020        if (len < marker_size + 1 || !isspace(line[marker_size]))
2021                return 0;
2022        return 1;
2023}
2024
2025static void checkdiff_consume(void *priv, char *line, unsigned long len)
2026{
2027        struct checkdiff_t *data = priv;
2028        int marker_size = data->conflict_marker_size;
2029        const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2030        const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2031        const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2032        char *err;
2033        char *line_prefix = "";
2034        struct strbuf *msgbuf;
2035
2036        assert(data->o);
2037        if (data->o->output_prefix) {
2038                msgbuf = data->o->output_prefix(data->o,
2039                        data->o->output_prefix_data);
2040                line_prefix = msgbuf->buf;
2041        }
2042
2043        if (line[0] == '+') {
2044                unsigned bad;
2045                data->lineno++;
2046                if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2047                        data->status |= 1;
2048                        fprintf(data->o->file,
2049                                "%s%s:%d: leftover conflict marker\n",
2050                                line_prefix, data->filename, data->lineno);
2051                }
2052                bad = ws_check(line + 1, len - 1, data->ws_rule);
2053                if (!bad)
2054                        return;
2055                data->status |= bad;
2056                err = whitespace_error_string(bad);
2057                fprintf(data->o->file, "%s%s:%d: %s.\n",
2058                        line_prefix, data->filename, data->lineno, err);
2059                free(err);
2060                emit_line(data->o, set, reset, line, 1);
2061                ws_check_emit(line + 1, len - 1, data->ws_rule,
2062                              data->o->file, set, reset, ws);
2063        } else if (line[0] == ' ') {
2064                data->lineno++;
2065        } else if (line[0] == '@') {
2066                char *plus = strchr(line, '+');
2067                if (plus)
2068                        data->lineno = strtol(plus, NULL, 10) - 1;
2069                else
2070                        die("invalid diff");
2071        }
2072}
2073
2074static unsigned char *deflate_it(char *data,
2075                                 unsigned long size,
2076                                 unsigned long *result_size)
2077{
2078        int bound;
2079        unsigned char *deflated;
2080        git_zstream stream;
2081
2082        memset(&stream, 0, sizeof(stream));
2083        git_deflate_init(&stream, zlib_compression_level);
2084        bound = git_deflate_bound(&stream, size);
2085        deflated = xmalloc(bound);
2086        stream.next_out = deflated;
2087        stream.avail_out = bound;
2088
2089        stream.next_in = (unsigned char *)data;
2090        stream.avail_in = size;
2091        while (git_deflate(&stream, Z_FINISH) == Z_OK)
2092                ; /* nothing */
2093        git_deflate_end(&stream);
2094        *result_size = stream.total_out;
2095        return deflated;
2096}
2097
2098static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2099{
2100        void *cp;
2101        void *delta;
2102        void *deflated;
2103        void *data;
2104        unsigned long orig_size;
2105        unsigned long delta_size;
2106        unsigned long deflate_size;
2107        unsigned long data_size;
2108
2109        /* We could do deflated delta, or we could do just deflated two,
2110         * whichever is smaller.
2111         */
2112        delta = NULL;
2113        deflated = deflate_it(two->ptr, two->size, &deflate_size);
2114        if (one->size && two->size) {
2115                delta = diff_delta(one->ptr, one->size,
2116                                   two->ptr, two->size,
2117                                   &delta_size, deflate_size);
2118                if (delta) {
2119                        void *to_free = delta;
2120                        orig_size = delta_size;
2121                        delta = deflate_it(delta, delta_size, &delta_size);
2122                        free(to_free);
2123                }
2124        }
2125
2126        if (delta && delta_size < deflate_size) {
2127                fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2128                free(deflated);
2129                data = delta;
2130                data_size = delta_size;
2131        }
2132        else {
2133                fprintf(file, "%sliteral %lu\n", prefix, two->size);
2134                free(delta);
2135                data = deflated;
2136                data_size = deflate_size;
2137        }
2138
2139        /* emit data encoded in base85 */
2140        cp = data;
2141        while (data_size) {
2142                int bytes = (52 < data_size) ? 52 : data_size;
2143                char line[70];
2144                data_size -= bytes;
2145                if (bytes <= 26)
2146                        line[0] = bytes + 'A' - 1;
2147                else
2148                        line[0] = bytes - 26 + 'a' - 1;
2149                encode_85(line + 1, cp, bytes);
2150                cp = (char *) cp + bytes;
2151                fprintf(file, "%s", prefix);
2152                fputs(line, file);
2153                fputc('\n', file);
2154        }
2155        fprintf(file, "%s\n", prefix);
2156        free(data);
2157}
2158
2159static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2160{
2161        fprintf(file, "%sGIT binary patch\n", prefix);
2162        emit_binary_diff_body(file, one, two, prefix);
2163        emit_binary_diff_body(file, two, one, prefix);
2164}
2165
2166int diff_filespec_is_binary(struct diff_filespec *one)
2167{
2168        if (one->is_binary == -1) {
2169                diff_filespec_load_driver(one);
2170                if (one->driver->binary != -1)
2171                        one->is_binary = one->driver->binary;
2172                else {
2173                        if (!one->data && DIFF_FILE_VALID(one))
2174                                diff_populate_filespec(one, 0);
2175                        if (one->data)
2176                                one->is_binary = buffer_is_binary(one->data,
2177                                                one->size);
2178                        if (one->is_binary == -1)
2179                                one->is_binary = 0;
2180                }
2181        }
2182        return one->is_binary;
2183}
2184
2185static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2186{
2187        diff_filespec_load_driver(one);
2188        return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2189}
2190
2191void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2192{
2193        if (!options->a_prefix)
2194                options->a_prefix = a;
2195        if (!options->b_prefix)
2196                options->b_prefix = b;
2197}
2198
2199struct userdiff_driver *get_textconv(struct diff_filespec *one)
2200{
2201        if (!DIFF_FILE_VALID(one))
2202                return NULL;
2203
2204        diff_filespec_load_driver(one);
2205        return userdiff_get_textconv(one->driver);
2206}
2207
2208static void builtin_diff(const char *name_a,
2209                         const char *name_b,
2210                         struct diff_filespec *one,
2211                         struct diff_filespec *two,
2212                         const char *xfrm_msg,
2213                         int must_show_header,
2214                         struct diff_options *o,
2215                         int complete_rewrite)
2216{
2217        mmfile_t mf1, mf2;
2218        const char *lbl[2];
2219        char *a_one, *b_two;
2220        const char *set = diff_get_color_opt(o, DIFF_METAINFO);
2221        const char *reset = diff_get_color_opt(o, DIFF_RESET);
2222        const char *a_prefix, *b_prefix;
2223        struct userdiff_driver *textconv_one = NULL;
2224        struct userdiff_driver *textconv_two = NULL;
2225        struct strbuf header = STRBUF_INIT;
2226        struct strbuf *msgbuf;
2227        char *line_prefix = "";
2228
2229        if (o->output_prefix) {
2230                msgbuf = o->output_prefix(o, o->output_prefix_data);
2231                line_prefix = msgbuf->buf;
2232        }
2233
2234        if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2235                        (!one->mode || S_ISGITLINK(one->mode)) &&
2236                        (!two->mode || S_ISGITLINK(two->mode))) {
2237                const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2238                const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2239                show_submodule_summary(o->file, one ? one->path : two->path,
2240                                one->sha1, two->sha1, two->dirty_submodule,
2241                                del, add, reset);
2242                return;
2243        }
2244
2245        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2246                textconv_one = get_textconv(one);
2247                textconv_two = get_textconv(two);
2248        }
2249
2250        diff_set_mnemonic_prefix(o, "a/", "b/");
2251        if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2252                a_prefix = o->b_prefix;
2253                b_prefix = o->a_prefix;
2254        } else {
2255                a_prefix = o->a_prefix;
2256                b_prefix = o->b_prefix;
2257        }
2258
2259        /* Never use a non-valid filename anywhere if at all possible */
2260        name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2261        name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2262
2263        a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2264        b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2265        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2266        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2267        strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
2268        if (lbl[0][0] == '/') {
2269                /* /dev/null */
2270                strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
2271                if (xfrm_msg)
2272                        strbuf_addstr(&header, xfrm_msg);
2273                must_show_header = 1;
2274        }
2275        else if (lbl[1][0] == '/') {
2276                strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
2277                if (xfrm_msg)
2278                        strbuf_addstr(&header, xfrm_msg);
2279                must_show_header = 1;
2280        }
2281        else {
2282                if (one->mode != two->mode) {
2283                        strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
2284                        strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
2285                        must_show_header = 1;
2286                }
2287                if (xfrm_msg)
2288                        strbuf_addstr(&header, xfrm_msg);
2289
2290                /*
2291                 * we do not run diff between different kind
2292                 * of objects.
2293                 */
2294                if ((one->mode ^ two->mode) & S_IFMT)
2295                        goto free_ab_and_return;
2296                if (complete_rewrite &&
2297                    (textconv_one || !diff_filespec_is_binary(one)) &&
2298                    (textconv_two || !diff_filespec_is_binary(two))) {
2299                        fprintf(o->file, "%s", header.buf);
2300                        strbuf_reset(&header);
2301                        emit_rewrite_diff(name_a, name_b, one, two,
2302                                                textconv_one, textconv_two, o);
2303                        o->found_changes = 1;
2304                        goto free_ab_and_return;
2305                }
2306        }
2307
2308        if (o->irreversible_delete && lbl[1][0] == '/') {
2309                fprintf(o->file, "%s", header.buf);
2310                strbuf_reset(&header);
2311                goto free_ab_and_return;
2312        } else if (!DIFF_OPT_TST(o, TEXT) &&
2313            ( (!textconv_one && diff_filespec_is_binary(one)) ||
2314              (!textconv_two && diff_filespec_is_binary(two)) )) {
2315                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2316                        die("unable to read files to diff");
2317                /* Quite common confusing case */
2318                if (mf1.size == mf2.size &&
2319                    !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2320                        if (must_show_header)
2321                                fprintf(o->file, "%s", header.buf);
2322                        goto free_ab_and_return;
2323                }
2324                fprintf(o->file, "%s", header.buf);
2325                strbuf_reset(&header);
2326                if (DIFF_OPT_TST(o, BINARY))
2327                        emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2328                else
2329                        fprintf(o->file, "%sBinary files %s and %s differ\n",
2330                                line_prefix, lbl[0], lbl[1]);
2331                o->found_changes = 1;
2332        } else {
2333                /* Crazy xdl interfaces.. */
2334                const char *diffopts = getenv("GIT_DIFF_OPTS");
2335                xpparam_t xpp;
2336                xdemitconf_t xecfg;
2337                struct emit_callback ecbdata;
2338                const struct userdiff_funcname *pe;
2339
2340                if (must_show_header) {
2341                        fprintf(o->file, "%s", header.buf);
2342                        strbuf_reset(&header);
2343                }
2344
2345                mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2346                mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2347
2348                pe = diff_funcname_pattern(one);
2349                if (!pe)
2350                        pe = diff_funcname_pattern(two);
2351
2352                memset(&xpp, 0, sizeof(xpp));
2353                memset(&xecfg, 0, sizeof(xecfg));
2354                memset(&ecbdata, 0, sizeof(ecbdata));
2355                ecbdata.label_path = lbl;
2356                ecbdata.color_diff = want_color(o->use_color);
2357                ecbdata.found_changesp = &o->found_changes;
2358                ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2359                if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2360                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2361                ecbdata.opt = o;
2362                ecbdata.header = header.len ? &header : NULL;
2363                xpp.flags = o->xdl_opts;
2364                xecfg.ctxlen = o->context;
2365                xecfg.interhunkctxlen = o->interhunkcontext;
2366                xecfg.flags = XDL_EMIT_FUNCNAMES;
2367                if (DIFF_OPT_TST(o, FUNCCONTEXT))
2368                        xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2369                if (pe)
2370                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2371                if (!diffopts)
2372                        ;
2373                else if (!prefixcmp(diffopts, "--unified="))
2374                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2375                else if (!prefixcmp(diffopts, "-u"))
2376                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2377                if (o->word_diff)
2378                        init_diff_words_data(&ecbdata, o, one, two);
2379                xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2380                              &xpp, &xecfg);
2381                if (o->word_diff)
2382                        free_diff_words_data(&ecbdata);
2383                if (textconv_one)
2384                        free(mf1.ptr);
2385                if (textconv_two)
2386                        free(mf2.ptr);
2387                xdiff_clear_find_func(&xecfg);
2388        }
2389
2390 free_ab_and_return:
2391        strbuf_release(&header);
2392        diff_free_filespec_data(one);
2393        diff_free_filespec_data(two);
2394        free(a_one);
2395        free(b_two);
2396        return;
2397}
2398
2399static void builtin_diffstat(const char *name_a, const char *name_b,
2400                             struct diff_filespec *one,
2401                             struct diff_filespec *two,
2402                             struct diffstat_t *diffstat,
2403                             struct diff_options *o,
2404                             struct diff_filepair *p)
2405{
2406        mmfile_t mf1, mf2;
2407        struct diffstat_file *data;
2408        int same_contents;
2409        int complete_rewrite = 0;
2410
2411        if (!DIFF_PAIR_UNMERGED(p)) {
2412                if (p->status == DIFF_STATUS_MODIFIED && p->score)
2413                        complete_rewrite = 1;
2414        }
2415
2416        data = diffstat_add(diffstat, name_a, name_b);
2417        data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2418
2419        if (!one || !two) {
2420                data->is_unmerged = 1;
2421                return;
2422        }
2423
2424        same_contents = !hashcmp(one->sha1, two->sha1);
2425
2426        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2427                data->is_binary = 1;
2428                if (same_contents) {
2429                        data->added = 0;
2430                        data->deleted = 0;
2431                } else {
2432                        data->added = diff_filespec_size(two);
2433                        data->deleted = diff_filespec_size(one);
2434                }
2435        }
2436
2437        else if (complete_rewrite) {
2438                diff_populate_filespec(one, 0);
2439                diff_populate_filespec(two, 0);
2440                data->deleted = count_lines(one->data, one->size);
2441                data->added = count_lines(two->data, two->size);
2442        }
2443
2444        else if (!same_contents) {
2445                /* Crazy xdl interfaces.. */
2446                xpparam_t xpp;
2447                xdemitconf_t xecfg;
2448
2449                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2450                        die("unable to read files to diff");
2451
2452                memset(&xpp, 0, sizeof(xpp));
2453                memset(&xecfg, 0, sizeof(xecfg));
2454                xpp.flags = o->xdl_opts;
2455                xecfg.ctxlen = o->context;
2456                xecfg.interhunkctxlen = o->interhunkcontext;
2457                xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2458                              &xpp, &xecfg);
2459        }
2460
2461        diff_free_filespec_data(one);
2462        diff_free_filespec_data(two);
2463}
2464
2465static void builtin_checkdiff(const char *name_a, const char *name_b,
2466                              const char *attr_path,
2467                              struct diff_filespec *one,
2468                              struct diff_filespec *two,
2469                              struct diff_options *o)
2470{
2471        mmfile_t mf1, mf2;
2472        struct checkdiff_t data;
2473
2474        if (!two)
2475                return;
2476
2477        memset(&data, 0, sizeof(data));
2478        data.filename = name_b ? name_b : name_a;
2479        data.lineno = 0;
2480        data.o = o;
2481        data.ws_rule = whitespace_rule(attr_path);
2482        data.conflict_marker_size = ll_merge_marker_size(attr_path);
2483
2484        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2485                die("unable to read files to diff");
2486
2487        /*
2488         * All the other codepaths check both sides, but not checking
2489         * the "old" side here is deliberate.  We are checking the newly
2490         * introduced changes, and as long as the "new" side is text, we
2491         * can and should check what it introduces.
2492         */
2493        if (diff_filespec_is_binary(two))
2494                goto free_and_return;
2495        else {
2496                /* Crazy xdl interfaces.. */
2497                xpparam_t xpp;
2498                xdemitconf_t xecfg;
2499
2500                memset(&xpp, 0, sizeof(xpp));
2501                memset(&xecfg, 0, sizeof(xecfg));
2502                xecfg.ctxlen = 1; /* at least one context line */
2503                xpp.flags = 0;
2504                xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2505                              &xpp, &xecfg);
2506
2507                if (data.ws_rule & WS_BLANK_AT_EOF) {
2508                        struct emit_callback ecbdata;
2509                        int blank_at_eof;
2510
2511                        ecbdata.ws_rule = data.ws_rule;
2512                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2513                        blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2514
2515                        if (blank_at_eof) {
2516                                static char *err;
2517                                if (!err)
2518                                        err = whitespace_error_string(WS_BLANK_AT_EOF);
2519                                fprintf(o->file, "%s:%d: %s.\n",
2520                                        data.filename, blank_at_eof, err);
2521                                data.status = 1; /* report errors */
2522                        }
2523                }
2524        }
2525 free_and_return:
2526        diff_free_filespec_data(one);
2527        diff_free_filespec_data(two);
2528        if (data.status)
2529                DIFF_OPT_SET(o, CHECK_FAILED);
2530}
2531
2532struct diff_filespec *alloc_filespec(const char *path)
2533{
2534        int namelen = strlen(path);
2535        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2536
2537        memset(spec, 0, sizeof(*spec));
2538        spec->path = (char *)(spec + 1);
2539        memcpy(spec->path, path, namelen+1);
2540        spec->count = 1;
2541        spec->is_binary = -1;
2542        return spec;
2543}
2544
2545void free_filespec(struct diff_filespec *spec)
2546{
2547        if (!--spec->count) {
2548                diff_free_filespec_data(spec);
2549                free(spec);
2550        }
2551}
2552
2553void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2554                   int sha1_valid, unsigned short mode)
2555{
2556        if (mode) {
2557                spec->mode = canon_mode(mode);
2558                hashcpy(spec->sha1, sha1);
2559                spec->sha1_valid = sha1_valid;
2560        }
2561}
2562
2563/*
2564 * Given a name and sha1 pair, if the index tells us the file in
2565 * the work tree has that object contents, return true, so that
2566 * prepare_temp_file() does not have to inflate and extract.
2567 */
2568static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2569{
2570        struct cache_entry *ce;
2571        struct stat st;
2572        int pos, len;
2573
2574        /*
2575         * We do not read the cache ourselves here, because the
2576         * benchmark with my previous version that always reads cache
2577         * shows that it makes things worse for diff-tree comparing
2578         * two linux-2.6 kernel trees in an already checked out work
2579         * tree.  This is because most diff-tree comparisons deal with
2580         * only a small number of files, while reading the cache is
2581         * expensive for a large project, and its cost outweighs the
2582         * savings we get by not inflating the object to a temporary
2583         * file.  Practically, this code only helps when we are used
2584         * by diff-cache --cached, which does read the cache before
2585         * calling us.
2586         */
2587        if (!active_cache)
2588                return 0;
2589
2590        /* We want to avoid the working directory if our caller
2591         * doesn't need the data in a normal file, this system
2592         * is rather slow with its stat/open/mmap/close syscalls,
2593         * and the object is contained in a pack file.  The pack
2594         * is probably already open and will be faster to obtain
2595         * the data through than the working directory.  Loose
2596         * objects however would tend to be slower as they need
2597         * to be individually opened and inflated.
2598         */
2599        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2600                return 0;
2601
2602        len = strlen(name);
2603        pos = cache_name_pos(name, len);
2604        if (pos < 0)
2605                return 0;
2606        ce = active_cache[pos];
2607
2608        /*
2609         * This is not the sha1 we are looking for, or
2610         * unreusable because it is not a regular file.
2611         */
2612        if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2613                return 0;
2614
2615        /*
2616         * If ce is marked as "assume unchanged", there is no
2617         * guarantee that work tree matches what we are looking for.
2618         */
2619        if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2620                return 0;
2621
2622        /*
2623         * If ce matches the file in the work tree, we can reuse it.
2624         */
2625        if (ce_uptodate(ce) ||
2626            (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2627                return 1;
2628
2629        return 0;
2630}
2631
2632static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2633{
2634        int len;
2635        char *data = xmalloc(100), *dirty = "";
2636
2637        /* Are we looking at the work tree? */
2638        if (s->dirty_submodule)
2639                dirty = "-dirty";
2640
2641        len = snprintf(data, 100,
2642                       "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2643        s->data = data;
2644        s->size = len;
2645        s->should_free = 1;
2646        if (size_only) {
2647                s->data = NULL;
2648                free(data);
2649        }
2650        return 0;
2651}
2652
2653/*
2654 * While doing rename detection and pickaxe operation, we may need to
2655 * grab the data for the blob (or file) for our own in-core comparison.
2656 * diff_filespec has data and size fields for this purpose.
2657 */
2658int diff_populate_filespec(struct diff_filespec *s, int size_only)
2659{
2660        int err = 0;
2661        if (!DIFF_FILE_VALID(s))
2662                die("internal error: asking to populate invalid file.");
2663        if (S_ISDIR(s->mode))
2664                return -1;
2665
2666        if (s->data)
2667                return 0;
2668
2669        if (size_only && 0 < s->size)
2670                return 0;
2671
2672        if (S_ISGITLINK(s->mode))
2673                return diff_populate_gitlink(s, size_only);
2674
2675        if (!s->sha1_valid ||
2676            reuse_worktree_file(s->path, s->sha1, 0)) {
2677                struct strbuf buf = STRBUF_INIT;
2678                struct stat st;
2679                int fd;
2680
2681                if (lstat(s->path, &st) < 0) {
2682                        if (errno == ENOENT) {
2683                        err_empty:
2684                                err = -1;
2685                        empty:
2686                                s->data = (char *)"";
2687                                s->size = 0;
2688                                return err;
2689                        }
2690                }
2691                s->size = xsize_t(st.st_size);
2692                if (!s->size)
2693                        goto empty;
2694                if (S_ISLNK(st.st_mode)) {
2695                        struct strbuf sb = STRBUF_INIT;
2696
2697                        if (strbuf_readlink(&sb, s->path, s->size))
2698                                goto err_empty;
2699                        s->size = sb.len;
2700                        s->data = strbuf_detach(&sb, NULL);
2701                        s->should_free = 1;
2702                        return 0;
2703                }
2704                if (size_only)
2705                        return 0;
2706                fd = open(s->path, O_RDONLY);
2707                if (fd < 0)
2708                        goto err_empty;
2709                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2710                close(fd);
2711                s->should_munmap = 1;
2712
2713                /*
2714                 * Convert from working tree format to canonical git format
2715                 */
2716                if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2717                        size_t size = 0;
2718                        munmap(s->data, s->size);
2719                        s->should_munmap = 0;
2720                        s->data = strbuf_detach(&buf, &size);
2721                        s->size = size;
2722                        s->should_free = 1;
2723                }
2724        }
2725        else {
2726                enum object_type type;
2727                if (size_only) {
2728                        type = sha1_object_info(s->sha1, &s->size);
2729                        if (type < 0)
2730                                die("unable to read %s", sha1_to_hex(s->sha1));
2731                } else {
2732                        s->data = read_sha1_file(s->sha1, &type, &s->size);
2733                        if (!s->data)
2734                                die("unable to read %s", sha1_to_hex(s->sha1));
2735                        s->should_free = 1;
2736                }
2737        }
2738        return 0;
2739}
2740
2741void diff_free_filespec_blob(struct diff_filespec *s)
2742{
2743        if (s->should_free)
2744                free(s->data);
2745        else if (s->should_munmap)
2746                munmap(s->data, s->size);
2747
2748        if (s->should_free || s->should_munmap) {
2749                s->should_free = s->should_munmap = 0;
2750                s->data = NULL;
2751        }
2752}
2753
2754void diff_free_filespec_data(struct diff_filespec *s)
2755{
2756        diff_free_filespec_blob(s);
2757        free(s->cnt_data);
2758        s->cnt_data = NULL;
2759}
2760
2761static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2762                           void *blob,
2763                           unsigned long size,
2764                           const unsigned char *sha1,
2765                           int mode)
2766{
2767        int fd;
2768        struct strbuf buf = STRBUF_INIT;
2769        struct strbuf template = STRBUF_INIT;
2770        char *path_dup = xstrdup(path);
2771        const char *base = basename(path_dup);
2772
2773        /* Generate "XXXXXX_basename.ext" */
2774        strbuf_addstr(&template, "XXXXXX_");
2775        strbuf_addstr(&template, base);
2776
2777        fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2778                        strlen(base) + 1);
2779        if (fd < 0)
2780                die_errno("unable to create temp-file");
2781        if (convert_to_working_tree(path,
2782                        (const char *)blob, (size_t)size, &buf)) {
2783                blob = buf.buf;
2784                size = buf.len;
2785        }
2786        if (write_in_full(fd, blob, size) != size)
2787                die_errno("unable to write temp-file");
2788        close(fd);
2789        temp->name = temp->tmp_path;
2790        strcpy(temp->hex, sha1_to_hex(sha1));
2791        temp->hex[40] = 0;
2792        sprintf(temp->mode, "%06o", mode);
2793        strbuf_release(&buf);
2794        strbuf_release(&template);
2795        free(path_dup);
2796}
2797
2798static struct diff_tempfile *prepare_temp_file(const char *name,
2799                struct diff_filespec *one)
2800{
2801        struct diff_tempfile *temp = claim_diff_tempfile();
2802
2803        if (!DIFF_FILE_VALID(one)) {
2804        not_a_valid_file:
2805                /* A '-' entry produces this for file-2, and
2806                 * a '+' entry produces this for file-1.
2807                 */
2808                temp->name = "/dev/null";
2809                strcpy(temp->hex, ".");
2810                strcpy(temp->mode, ".");
2811                return temp;
2812        }
2813
2814        if (!remove_tempfile_installed) {
2815                atexit(remove_tempfile);
2816                sigchain_push_common(remove_tempfile_on_signal);
2817                remove_tempfile_installed = 1;
2818        }
2819
2820        if (!one->sha1_valid ||
2821            reuse_worktree_file(name, one->sha1, 1)) {
2822                struct stat st;
2823                if (lstat(name, &st) < 0) {
2824                        if (errno == ENOENT)
2825                                goto not_a_valid_file;
2826                        die_errno("stat(%s)", name);
2827                }
2828                if (S_ISLNK(st.st_mode)) {
2829                        struct strbuf sb = STRBUF_INIT;
2830                        if (strbuf_readlink(&sb, name, st.st_size) < 0)
2831                                die_errno("readlink(%s)", name);
2832                        prep_temp_blob(name, temp, sb.buf, sb.len,
2833                                       (one->sha1_valid ?
2834                                        one->sha1 : null_sha1),
2835                                       (one->sha1_valid ?
2836                                        one->mode : S_IFLNK));
2837                        strbuf_release(&sb);
2838                }
2839                else {
2840                        /* we can borrow from the file in the work tree */
2841                        temp->name = name;
2842                        if (!one->sha1_valid)
2843                                strcpy(temp->hex, sha1_to_hex(null_sha1));
2844                        else
2845                                strcpy(temp->hex, sha1_to_hex(one->sha1));
2846                        /* Even though we may sometimes borrow the
2847                         * contents from the work tree, we always want
2848                         * one->mode.  mode is trustworthy even when
2849                         * !(one->sha1_valid), as long as
2850                         * DIFF_FILE_VALID(one).
2851                         */
2852                        sprintf(temp->mode, "%06o", one->mode);
2853                }
2854                return temp;
2855        }
2856        else {
2857                if (diff_populate_filespec(one, 0))
2858                        die("cannot read data blob for %s", one->path);
2859                prep_temp_blob(name, temp, one->data, one->size,
2860                               one->sha1, one->mode);
2861        }
2862        return temp;
2863}
2864
2865/* An external diff command takes:
2866 *
2867 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2868 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2869 *
2870 */
2871static void run_external_diff(const char *pgm,
2872                              const char *name,
2873                              const char *other,
2874                              struct diff_filespec *one,
2875                              struct diff_filespec *two,
2876                              const char *xfrm_msg,
2877                              int complete_rewrite)
2878{
2879        const char *spawn_arg[10];
2880        int retval;
2881        const char **arg = &spawn_arg[0];
2882
2883        if (one && two) {
2884                struct diff_tempfile *temp_one, *temp_two;
2885                const char *othername = (other ? other : name);
2886                temp_one = prepare_temp_file(name, one);
2887                temp_two = prepare_temp_file(othername, two);
2888                *arg++ = pgm;
2889                *arg++ = name;
2890                *arg++ = temp_one->name;
2891                *arg++ = temp_one->hex;
2892                *arg++ = temp_one->mode;
2893                *arg++ = temp_two->name;
2894                *arg++ = temp_two->hex;
2895                *arg++ = temp_two->mode;
2896                if (other) {
2897                        *arg++ = other;
2898                        *arg++ = xfrm_msg;
2899                }
2900        } else {
2901                *arg++ = pgm;
2902                *arg++ = name;
2903        }
2904        *arg = NULL;
2905        fflush(NULL);
2906        retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2907        remove_tempfile();
2908        if (retval) {
2909                fprintf(stderr, "external diff died, stopping at %s.\n", name);
2910                exit(1);
2911        }
2912}
2913
2914static int similarity_index(struct diff_filepair *p)
2915{
2916        return p->score * 100 / MAX_SCORE;
2917}
2918
2919static void fill_metainfo(struct strbuf *msg,
2920                          const char *name,
2921                          const char *other,
2922                          struct diff_filespec *one,
2923                          struct diff_filespec *two,
2924                          struct diff_options *o,
2925                          struct diff_filepair *p,
2926                          int *must_show_header,
2927                          int use_color)
2928{
2929        const char *set = diff_get_color(use_color, DIFF_METAINFO);
2930        const char *reset = diff_get_color(use_color, DIFF_RESET);
2931        struct strbuf *msgbuf;
2932        char *line_prefix = "";
2933
2934        *must_show_header = 1;
2935        if (o->output_prefix) {
2936                msgbuf = o->output_prefix(o, o->output_prefix_data);
2937                line_prefix = msgbuf->buf;
2938        }
2939        strbuf_init(msg, PATH_MAX * 2 + 300);
2940        switch (p->status) {
2941        case DIFF_STATUS_COPIED:
2942                strbuf_addf(msg, "%s%ssimilarity index %d%%",
2943                            line_prefix, set, similarity_index(p));
2944                strbuf_addf(msg, "%s\n%s%scopy from ",
2945                            reset,  line_prefix, set);
2946                quote_c_style(name, msg, NULL, 0);
2947                strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2948                quote_c_style(other, msg, NULL, 0);
2949                strbuf_addf(msg, "%s\n", reset);
2950                break;
2951        case DIFF_STATUS_RENAMED:
2952                strbuf_addf(msg, "%s%ssimilarity index %d%%",
2953                            line_prefix, set, similarity_index(p));
2954                strbuf_addf(msg, "%s\n%s%srename from ",
2955                            reset, line_prefix, set);
2956                quote_c_style(name, msg, NULL, 0);
2957                strbuf_addf(msg, "%s\n%s%srename to ",
2958                            reset, line_prefix, set);
2959                quote_c_style(other, msg, NULL, 0);
2960                strbuf_addf(msg, "%s\n", reset);
2961                break;
2962        case DIFF_STATUS_MODIFIED:
2963                if (p->score) {
2964                        strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2965                                    line_prefix,
2966                                    set, similarity_index(p), reset);
2967                        break;
2968                }
2969                /* fallthru */
2970        default:
2971                *must_show_header = 0;
2972        }
2973        if (one && two && hashcmp(one->sha1, two->sha1)) {
2974                int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2975
2976                if (DIFF_OPT_TST(o, BINARY)) {
2977                        mmfile_t mf;
2978                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2979                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2980                                abbrev = 40;
2981                }
2982                strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2983                            find_unique_abbrev(one->sha1, abbrev));
2984                strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2985                if (one->mode == two->mode)
2986                        strbuf_addf(msg, " %06o", one->mode);
2987                strbuf_addf(msg, "%s\n", reset);
2988        }
2989}
2990
2991static void run_diff_cmd(const char *pgm,
2992                         const char *name,
2993                         const char *other,
2994                         const char *attr_path,
2995                         struct diff_filespec *one,
2996                         struct diff_filespec *two,
2997                         struct strbuf *msg,
2998                         struct diff_options *o,
2999                         struct diff_filepair *p)
3000{
3001        const char *xfrm_msg = NULL;
3002        int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3003        int must_show_header = 0;
3004
3005
3006        if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3007                struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3008                if (drv && drv->external)
3009                        pgm = drv->external;
3010        }
3011
3012        if (msg) {
3013                /*
3014                 * don't use colors when the header is intended for an
3015                 * external diff driver
3016                 */
3017                fill_metainfo(msg, name, other, one, two, o, p,
3018                              &must_show_header,
3019                              want_color(o->use_color) && !pgm);
3020                xfrm_msg = msg->len ? msg->buf : NULL;
3021        }
3022
3023        if (pgm) {
3024                run_external_diff(pgm, name, other, one, two, xfrm_msg,
3025                                  complete_rewrite);
3026                return;
3027        }
3028        if (one && two)
3029                builtin_diff(name, other ? other : name,
3030                             one, two, xfrm_msg, must_show_header,
3031                             o, complete_rewrite);
3032        else
3033                fprintf(o->file, "* Unmerged path %s\n", name);
3034}
3035
3036static void diff_fill_sha1_info(struct diff_filespec *one)
3037{
3038        if (DIFF_FILE_VALID(one)) {
3039                if (!one->sha1_valid) {
3040                        struct stat st;
3041                        if (one->is_stdin) {
3042                                hashcpy(one->sha1, null_sha1);
3043                                return;
3044                        }
3045                        if (lstat(one->path, &st) < 0)
3046                                die_errno("stat '%s'", one->path);
3047                        if (index_path(one->sha1, one->path, &st, 0))
3048                                die("cannot hash %s", one->path);
3049                }
3050        }
3051        else
3052                hashclr(one->sha1);
3053}
3054
3055static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3056{
3057        /* Strip the prefix but do not molest /dev/null and absolute paths */
3058        if (*namep && **namep != '/') {
3059                *namep += prefix_length;
3060                if (**namep == '/')
3061                        ++*namep;
3062        }
3063        if (*otherp && **otherp != '/') {
3064                *otherp += prefix_length;
3065                if (**otherp == '/')
3066                        ++*otherp;
3067        }
3068}
3069
3070static void run_diff(struct diff_filepair *p, struct diff_options *o)
3071{
3072        const char *pgm = external_diff();
3073        struct strbuf msg;
3074        struct diff_filespec *one = p->one;
3075        struct diff_filespec *two = p->two;
3076        const char *name;
3077        const char *other;
3078        const char *attr_path;
3079
3080        name  = p->one->path;
3081        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3082        attr_path = name;
3083        if (o->prefix_length)
3084                strip_prefix(o->prefix_length, &name, &other);
3085
3086        if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3087                pgm = NULL;
3088
3089        if (DIFF_PAIR_UNMERGED(p)) {
3090                run_diff_cmd(pgm, name, NULL, attr_path,
3091                             NULL, NULL, NULL, o, p);
3092                return;
3093        }
3094
3095        diff_fill_sha1_info(one);
3096        diff_fill_sha1_info(two);
3097
3098        if (!pgm &&
3099            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3100            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3101                /*
3102                 * a filepair that changes between file and symlink
3103                 * needs to be split into deletion and creation.
3104                 */
3105                struct diff_filespec *null = alloc_filespec(two->path);
3106                run_diff_cmd(NULL, name, other, attr_path,
3107                             one, null, &msg, o, p);
3108                free(null);
3109                strbuf_release(&msg);
3110
3111                null = alloc_filespec(one->path);
3112                run_diff_cmd(NULL, name, other, attr_path,
3113                             null, two, &msg, o, p);
3114                free(null);
3115        }
3116        else
3117                run_diff_cmd(pgm, name, other, attr_path,
3118                             one, two, &msg, o, p);
3119
3120        strbuf_release(&msg);
3121}
3122
3123static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3124                         struct diffstat_t *diffstat)
3125{
3126        const char *name;
3127        const char *other;
3128
3129        if (DIFF_PAIR_UNMERGED(p)) {
3130                /* unmerged */
3131                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3132                return;
3133        }
3134
3135        name = p->one->path;
3136        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3137
3138        if (o->prefix_length)
3139                strip_prefix(o->prefix_length, &name, &other);
3140
3141        diff_fill_sha1_info(p->one);
3142        diff_fill_sha1_info(p->two);
3143
3144        builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3145}
3146
3147static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3148{
3149        const char *name;
3150        const char *other;
3151        const char *attr_path;
3152
3153        if (DIFF_PAIR_UNMERGED(p)) {
3154                /* unmerged */
3155                return;
3156        }
3157
3158        name = p->one->path;
3159        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3160        attr_path = other ? other : name;
3161
3162        if (o->prefix_length)
3163                strip_prefix(o->prefix_length, &name, &other);
3164
3165        diff_fill_sha1_info(p->one);
3166        diff_fill_sha1_info(p->two);
3167
3168        builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3169}
3170
3171void diff_setup(struct diff_options *options)
3172{
3173        memcpy(options, &default_diff_options, sizeof(*options));
3174
3175        options->file = stdout;
3176
3177        options->line_termination = '\n';
3178        options->break_opt = -1;
3179        options->rename_limit = -1;
3180        options->dirstat_permille = diff_dirstat_permille_default;
3181        options->context = 3;
3182        DIFF_OPT_SET(options, RENAME_EMPTY);
3183
3184        options->change = diff_change;
3185        options->add_remove = diff_addremove;
3186        options->use_color = diff_use_color_default;
3187        options->detect_rename = diff_detect_rename_default;
3188
3189        if (diff_no_prefix) {
3190                options->a_prefix = options->b_prefix = "";
3191        } else if (!diff_mnemonic_prefix) {
3192                options->a_prefix = "a/";
3193                options->b_prefix = "b/";
3194        }
3195}
3196
3197void diff_setup_done(struct diff_options *options)
3198{
3199        int count = 0;
3200
3201        if (options->output_format & DIFF_FORMAT_NAME)
3202                count++;
3203        if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3204                count++;
3205        if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3206                count++;
3207        if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3208                count++;
3209        if (count > 1)
3210                die("--name-only, --name-status, --check and -s are mutually exclusive");
3211
3212        /*
3213         * Most of the time we can say "there are changes"
3214         * only by checking if there are changed paths, but
3215         * --ignore-whitespace* options force us to look
3216         * inside contents.
3217         */
3218
3219        if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3220            DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3221            DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3222                DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3223        else
3224                DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3225
3226        if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3227                options->detect_rename = DIFF_DETECT_COPY;
3228
3229        if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3230                options->prefix = NULL;
3231        if (options->prefix)
3232                options->prefix_length = strlen(options->prefix);
3233        else
3234                options->prefix_length = 0;
3235
3236        if (options->output_format & (DIFF_FORMAT_NAME |
3237                                      DIFF_FORMAT_NAME_STATUS |
3238                                      DIFF_FORMAT_CHECKDIFF |
3239                                      DIFF_FORMAT_NO_OUTPUT))
3240                options->output_format &= ~(DIFF_FORMAT_RAW |
3241                                            DIFF_FORMAT_NUMSTAT |
3242                                            DIFF_FORMAT_DIFFSTAT |
3243                                            DIFF_FORMAT_SHORTSTAT |
3244                                            DIFF_FORMAT_DIRSTAT |
3245                                            DIFF_FORMAT_SUMMARY |
3246                                            DIFF_FORMAT_PATCH);
3247
3248        /*
3249         * These cases always need recursive; we do not drop caller-supplied
3250         * recursive bits for other formats here.
3251         */
3252        if (options->output_format & (DIFF_FORMAT_PATCH |
3253                                      DIFF_FORMAT_NUMSTAT |
3254                                      DIFF_FORMAT_DIFFSTAT |
3255                                      DIFF_FORMAT_SHORTSTAT |
3256                                      DIFF_FORMAT_DIRSTAT |
3257                                      DIFF_FORMAT_SUMMARY |
3258                                      DIFF_FORMAT_CHECKDIFF))
3259                DIFF_OPT_SET(options, RECURSIVE);
3260        /*
3261         * Also pickaxe would not work very well if you do not say recursive
3262         */
3263        if (options->pickaxe)
3264                DIFF_OPT_SET(options, RECURSIVE);
3265        /*
3266         * When patches are generated, submodules diffed against the work tree
3267         * must be checked for dirtiness too so it can be shown in the output
3268         */
3269        if (options->output_format & DIFF_FORMAT_PATCH)
3270                DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3271
3272        if (options->detect_rename && options->rename_limit < 0)
3273                options->rename_limit = diff_rename_limit_default;
3274        if (options->setup & DIFF_SETUP_USE_CACHE) {
3275                if (!active_cache)
3276                        /* read-cache does not die even when it fails
3277                         * so it is safe for us to do this here.  Also
3278                         * it does not smudge active_cache or active_nr
3279                         * when it fails, so we do not have to worry about
3280                         * cleaning it up ourselves either.
3281                         */
3282                        read_cache();
3283        }
3284        if (options->abbrev <= 0 || 40 < options->abbrev)
3285                options->abbrev = 40; /* full */
3286
3287        /*
3288         * It does not make sense to show the first hit we happened
3289         * to have found.  It does not make sense not to return with
3290         * exit code in such a case either.
3291         */
3292        if (DIFF_OPT_TST(options, QUICK)) {
3293                options->output_format = DIFF_FORMAT_NO_OUTPUT;
3294                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3295        }
3296}
3297
3298static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3299{
3300        char c, *eq;
3301        int len;
3302
3303        if (*arg != '-')
3304                return 0;
3305        c = *++arg;
3306        if (!c)
3307                return 0;
3308        if (c == arg_short) {
3309                c = *++arg;
3310                if (!c)
3311                        return 1;
3312                if (val && isdigit(c)) {
3313                        char *end;
3314                        int n = strtoul(arg, &end, 10);
3315                        if (*end)
3316                                return 0;
3317                        *val = n;
3318                        return 1;
3319                }
3320                return 0;
3321        }
3322        if (c != '-')
3323                return 0;
3324        arg++;
3325        eq = strchr(arg, '=');
3326        if (eq)
3327                len = eq - arg;
3328        else
3329                len = strlen(arg);
3330        if (!len || strncmp(arg, arg_long, len))
3331                return 0;
3332        if (eq) {
3333                int n;
3334                char *end;
3335                if (!isdigit(*++eq))
3336                        return 0;
3337                n = strtoul(eq, &end, 10);
3338                if (*end)
3339                        return 0;
3340                *val = n;
3341        }
3342        return 1;
3343}
3344
3345static int diff_scoreopt_parse(const char *opt);
3346
3347static inline int short_opt(char opt, const char **argv,
3348                            const char **optarg)
3349{
3350        const char *arg = argv[0];
3351        if (arg[0] != '-' || arg[1] != opt)
3352                return 0;
3353        if (arg[2] != '\0') {
3354                *optarg = arg + 2;
3355                return 1;
3356        }
3357        if (!argv[1])
3358                die("Option '%c' requires a value", opt);
3359        *optarg = argv[1];
3360        return 2;
3361}
3362
3363int parse_long_opt(const char *opt, const char **argv,
3364                   const char **optarg)
3365{
3366        const char *arg = argv[0];
3367        if (arg[0] != '-' || arg[1] != '-')
3368                return 0;
3369        arg += strlen("--");
3370        if (prefixcmp(arg, opt))
3371                return 0;
3372        arg += strlen(opt);
3373        if (*arg == '=') { /* sticked form: --option=value */
3374                *optarg = arg + 1;
3375                return 1;
3376        }
3377        if (*arg != '\0')
3378                return 0;
3379        /* separate form: --option value */
3380        if (!argv[1])
3381                die("Option '--%s' requires a value", opt);
3382        *optarg = argv[1];
3383        return 2;
3384}
3385
3386static int stat_opt(struct diff_options *options, const char **av)
3387{
3388        const char *arg = av[0];
3389        char *end;
3390        int width = options->stat_width;
3391        int name_width = options->stat_name_width;
3392        int graph_width = options->stat_graph_width;
3393        int count = options->stat_count;
3394        int argcount = 1;
3395
3396        arg += strlen("--stat");
3397        end = (char *)arg;
3398
3399        switch (*arg) {
3400        case '-':
3401                if (!prefixcmp(arg, "-width")) {
3402                        arg += strlen("-width");
3403                        if (*arg == '=')
3404                                width = strtoul(arg + 1, &end, 10);
3405                        else if (!*arg && !av[1])
3406                                die("Option '--stat-width' requires a value");
3407                        else if (!*arg) {
3408                                width = strtoul(av[1], &end, 10);
3409                                argcount = 2;
3410                        }
3411                } else if (!prefixcmp(arg, "-name-width")) {
3412                        arg += strlen("-name-width");
3413                        if (*arg == '=')
3414                                name_width = strtoul(arg + 1, &end, 10);
3415                        else if (!*arg && !av[1])
3416                                die("Option '--stat-name-width' requires a value");
3417                        else if (!*arg) {
3418                                name_width = strtoul(av[1], &end, 10);
3419                                argcount = 2;
3420                        }
3421                } else if (!prefixcmp(arg, "-graph-width")) {
3422                        arg += strlen("-graph-width");
3423                        if (*arg == '=')
3424                                graph_width = strtoul(arg + 1, &end, 10);
3425                        else if (!*arg && !av[1])
3426                                die("Option '--stat-graph-width' requires a value");
3427                        else if (!*arg) {
3428                                graph_width = strtoul(av[1], &end, 10);
3429                                argcount = 2;
3430                        }
3431                } else if (!prefixcmp(arg, "-count")) {
3432                        arg += strlen("-count");
3433                        if (*arg == '=')
3434                                count = strtoul(arg + 1, &end, 10);
3435                        else if (!*arg && !av[1])
3436                                die("Option '--stat-count' requires a value");
3437                        else if (!*arg) {
3438                                count = strtoul(av[1], &end, 10);
3439                                argcount = 2;
3440                        }
3441                }
3442                break;
3443        case '=':
3444                width = strtoul(arg+1, &end, 10);
3445                if (*end == ',')
3446                        name_width = strtoul(end+1, &end, 10);
3447                if (*end == ',')
3448                        count = strtoul(end+1, &end, 10);
3449        }
3450
3451        /* Important! This checks all the error cases! */
3452        if (*end)
3453                return 0;
3454        options->output_format |= DIFF_FORMAT_DIFFSTAT;
3455        options->stat_name_width = name_width;
3456        options->stat_graph_width = graph_width;
3457        options->stat_width = width;
3458        options->stat_count = count;
3459        return argcount;
3460}
3461
3462static int parse_dirstat_opt(struct diff_options *options, const char *params)
3463{
3464        struct strbuf errmsg = STRBUF_INIT;
3465        if (parse_dirstat_params(options, params, &errmsg))
3466                die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3467                    errmsg.buf);
3468        strbuf_release(&errmsg);
3469        /*
3470         * The caller knows a dirstat-related option is given from the command
3471         * line; allow it to say "return this_function();"
3472         */
3473        options->output_format |= DIFF_FORMAT_DIRSTAT;
3474        return 1;
3475}
3476
3477int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3478{
3479        const char *arg = av[0];
3480        const char *optarg;
3481        int argcount;
3482
3483        /* Output format options */
3484        if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3485                options->output_format |= DIFF_FORMAT_PATCH;
3486        else if (opt_arg(arg, 'U', "unified", &options->context))
3487                options->output_format |= DIFF_FORMAT_PATCH;
3488        else if (!strcmp(arg, "--raw"))
3489                options->output_format |= DIFF_FORMAT_RAW;
3490        else if (!strcmp(arg, "--patch-with-raw"))
3491                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3492        else if (!strcmp(arg, "--numstat"))
3493                options->output_format |= DIFF_FORMAT_NUMSTAT;
3494        else if (!strcmp(arg, "--shortstat"))
3495                options->output_format |= DIFF_FORMAT_SHORTSTAT;
3496        else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3497                return parse_dirstat_opt(options, "");
3498        else if (!prefixcmp(arg, "-X"))
3499                return parse_dirstat_opt(options, arg + 2);
3500        else if (!prefixcmp(arg, "--dirstat="))
3501                return parse_dirstat_opt(options, arg + 10);
3502        else if (!strcmp(arg, "--cumulative"))
3503                return parse_dirstat_opt(options, "cumulative");
3504        else if (!strcmp(arg, "--dirstat-by-file"))
3505                return parse_dirstat_opt(options, "files");
3506        else if (!prefixcmp(arg, "--dirstat-by-file=")) {
3507                parse_dirstat_opt(options, "files");
3508                return parse_dirstat_opt(options, arg + 18);
3509        }
3510        else if (!strcmp(arg, "--check"))
3511                options->output_format |= DIFF_FORMAT_CHECKDIFF;
3512        else if (!strcmp(arg, "--summary"))
3513                options->output_format |= DIFF_FORMAT_SUMMARY;
3514        else if (!strcmp(arg, "--patch-with-stat"))
3515                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3516        else if (!strcmp(arg, "--name-only"))
3517                options->output_format |= DIFF_FORMAT_NAME;
3518        else if (!strcmp(arg, "--name-status"))
3519                options->output_format |= DIFF_FORMAT_NAME_STATUS;
3520        else if (!strcmp(arg, "-s"))
3521                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3522        else if (!prefixcmp(arg, "--stat"))
3523                /* --stat, --stat-width, --stat-name-width, or --stat-count */
3524                return stat_opt(options, av);
3525
3526        /* renames options */
3527        else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3528                 !strcmp(arg, "--break-rewrites")) {
3529                if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3530                        return error("invalid argument to -B: %s", arg+2);
3531        }
3532        else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3533                 !strcmp(arg, "--find-renames")) {
3534                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3535                        return error("invalid argument to -M: %s", arg+2);
3536                options->detect_rename = DIFF_DETECT_RENAME;
3537        }
3538        else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3539                options->irreversible_delete = 1;
3540        }
3541        else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3542                 !strcmp(arg, "--find-copies")) {
3543                if (options->detect_rename == DIFF_DETECT_COPY)
3544                        DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3545                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3546                        return error("invalid argument to -C: %s", arg+2);
3547                options->detect_rename = DIFF_DETECT_COPY;
3548        }
3549        else if (!strcmp(arg, "--no-renames"))
3550                options->detect_rename = 0;
3551        else if (!strcmp(arg, "--rename-empty"))
3552                DIFF_OPT_SET(options, RENAME_EMPTY);
3553        else if (!strcmp(arg, "--no-rename-empty"))
3554                DIFF_OPT_CLR(options, RENAME_EMPTY);
3555        else if (!strcmp(arg, "--relative"))
3556                DIFF_OPT_SET(options, RELATIVE_NAME);
3557        else if (!prefixcmp(arg, "--relative=")) {
3558                DIFF_OPT_SET(options, RELATIVE_NAME);
3559                options->prefix = arg + 11;
3560        }
3561
3562        /* xdiff options */
3563        else if (!strcmp(arg, "--minimal"))
3564                DIFF_XDL_SET(options, NEED_MINIMAL);
3565        else if (!strcmp(arg, "--no-minimal"))
3566                DIFF_XDL_CLR(options, NEED_MINIMAL);
3567        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3568                DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3569        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3570                DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3571        else if (!strcmp(arg, "--ignore-space-at-eol"))
3572                DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3573        else if (!strcmp(arg, "--patience"))
3574                options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3575        else if (!strcmp(arg, "--histogram"))
3576                options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3577
3578        /* flags options */
3579        else if (!strcmp(arg, "--binary")) {
3580                options->output_format |= DIFF_FORMAT_PATCH;
3581                DIFF_OPT_SET(options, BINARY);
3582        }
3583        else if (!strcmp(arg, "--full-index"))
3584                DIFF_OPT_SET(options, FULL_INDEX);
3585        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3586                DIFF_OPT_SET(options, TEXT);
3587        else if (!strcmp(arg, "-R"))
3588                DIFF_OPT_SET(options, REVERSE_DIFF);
3589        else if (!strcmp(arg, "--find-copies-harder"))
3590                DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3591        else if (!strcmp(arg, "--follow"))
3592                DIFF_OPT_SET(options, FOLLOW_RENAMES);
3593        else if (!strcmp(arg, "--color"))
3594                options->use_color = 1;
3595        else if (!prefixcmp(arg, "--color=")) {
3596                int value = git_config_colorbool(NULL, arg+8);
3597                if (value < 0)
3598                        return error("option `color' expects \"always\", \"auto\", or \"never\"");
3599                options->use_color = value;
3600        }
3601        else if (!strcmp(arg, "--no-color"))
3602                options->use_color = 0;
3603        else if (!strcmp(arg, "--color-words")) {
3604                options->use_color = 1;
3605                options->word_diff = DIFF_WORDS_COLOR;
3606        }
3607        else if (!prefixcmp(arg, "--color-words=")) {
3608                options->use_color = 1;
3609                options->word_diff = DIFF_WORDS_COLOR;
3610                options->word_regex = arg + 14;
3611        }
3612        else if (!strcmp(arg, "--word-diff")) {
3613                if (options->word_diff == DIFF_WORDS_NONE)
3614                        options->word_diff = DIFF_WORDS_PLAIN;
3615        }
3616        else if (!prefixcmp(arg, "--word-diff=")) {
3617                const char *type = arg + 12;
3618                if (!strcmp(type, "plain"))
3619                        options->word_diff = DIFF_WORDS_PLAIN;
3620                else if (!strcmp(type, "color")) {
3621                        options->use_color = 1;
3622                        options->word_diff = DIFF_WORDS_COLOR;
3623                }
3624                else if (!strcmp(type, "porcelain"))
3625                        options->word_diff = DIFF_WORDS_PORCELAIN;
3626                else if (!strcmp(type, "none"))
3627                        options->word_diff = DIFF_WORDS_NONE;
3628                else
3629                        die("bad --word-diff argument: %s", type);
3630        }
3631        else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3632                if (options->word_diff == DIFF_WORDS_NONE)
3633                        options->word_diff = DIFF_WORDS_PLAIN;
3634                options->word_regex = optarg;
3635                return argcount;
3636        }
3637        else if (!strcmp(arg, "--exit-code"))
3638                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3639        else if (!strcmp(arg, "--quiet"))
3640                DIFF_OPT_SET(options, QUICK);
3641        else if (!strcmp(arg, "--ext-diff"))
3642                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3643        else if (!strcmp(arg, "--no-ext-diff"))
3644                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3645        else if (!strcmp(arg, "--textconv"))
3646                DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3647        else if (!strcmp(arg, "--no-textconv"))
3648                DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3649        else if (!strcmp(arg, "--ignore-submodules")) {
3650                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3651                handle_ignore_submodules_arg(options, "all");
3652        } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3653                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3654                handle_ignore_submodules_arg(options, arg + 20);
3655        } else if (!strcmp(arg, "--submodule"))
3656                DIFF_OPT_SET(options, SUBMODULE_LOG);
3657        else if (!prefixcmp(arg, "--submodule=")) {
3658                if (!strcmp(arg + 12, "log"))
3659                        DIFF_OPT_SET(options, SUBMODULE_LOG);
3660        }
3661
3662        /* misc options */
3663        else if (!strcmp(arg, "-z"))
3664                options->line_termination = 0;
3665        else if ((argcount = short_opt('l', av, &optarg))) {
3666                options->rename_limit = strtoul(optarg, NULL, 10);
3667                return argcount;
3668        }
3669        else if ((argcount = short_opt('S', av, &optarg))) {
3670                options->pickaxe = optarg;
3671                options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3672                return argcount;
3673        } else if ((argcount = short_opt('G', av, &optarg))) {
3674                options->pickaxe = optarg;
3675                options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3676                return argcount;
3677        }
3678        else if (!strcmp(arg, "--pickaxe-all"))
3679                options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3680        else if (!strcmp(arg, "--pickaxe-regex"))
3681                options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3682        else if ((argcount = short_opt('O', av, &optarg))) {
3683                options->orderfile = optarg;
3684                return argcount;
3685        }
3686        else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3687                options->filter = optarg;
3688                return argcount;
3689        }
3690        else if (!strcmp(arg, "--abbrev"))
3691                options->abbrev = DEFAULT_ABBREV;
3692        else if (!prefixcmp(arg, "--abbrev=")) {
3693                options->abbrev = strtoul(arg + 9, NULL, 10);
3694                if (options->abbrev < MINIMUM_ABBREV)
3695                        options->abbrev = MINIMUM_ABBREV;
3696                else if (40 < options->abbrev)
3697                        options->abbrev = 40;
3698        }
3699        else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3700                options->a_prefix = optarg;
3701                return argcount;
3702        }
3703        else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3704                options->b_prefix = optarg;
3705                return argcount;
3706        }
3707        else if (!strcmp(arg, "--no-prefix"))
3708                options->a_prefix = options->b_prefix = "";
3709        else if (opt_arg(arg, '\0', "inter-hunk-context",
3710                         &options->interhunkcontext))
3711                ;
3712        else if (!strcmp(arg, "-W"))
3713                DIFF_OPT_SET(options, FUNCCONTEXT);
3714        else if (!strcmp(arg, "--function-context"))
3715                DIFF_OPT_SET(options, FUNCCONTEXT);
3716        else if (!strcmp(arg, "--no-function-context"))
3717                DIFF_OPT_CLR(options, FUNCCONTEXT);
3718        else if ((argcount = parse_long_opt("output", av, &optarg))) {
3719                options->file = fopen(optarg, "w");
3720                if (!options->file)
3721                        die_errno("Could not open '%s'", optarg);
3722                options->close_file = 1;
3723                return argcount;
3724        } else
3725                return 0;
3726        return 1;
3727}
3728
3729int parse_rename_score(const char **cp_p)
3730{
3731        unsigned long num, scale;
3732        int ch, dot;
3733        const char *cp = *cp_p;
3734
3735        num = 0;
3736        scale = 1;
3737        dot = 0;
3738        for (;;) {
3739                ch = *cp;
3740                if ( !dot && ch == '.' ) {
3741                        scale = 1;
3742                        dot = 1;
3743                } else if ( ch == '%' ) {
3744                        scale = dot ? scale*100 : 100;
3745                        cp++;   /* % is always at the end */
3746                        break;
3747                } else if ( ch >= '0' && ch <= '9' ) {
3748                        if ( scale < 100000 ) {
3749                                scale *= 10;
3750                                num = (num*10) + (ch-'0');
3751                        }
3752                } else {
3753                        break;
3754                }
3755                cp++;
3756        }
3757        *cp_p = cp;
3758
3759        /* user says num divided by scale and we say internally that
3760         * is MAX_SCORE * num / scale.
3761         */
3762        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3763}
3764
3765static int diff_scoreopt_parse(const char *opt)
3766{
3767        int opt1, opt2, cmd;
3768
3769        if (*opt++ != '-')
3770                return -1;
3771        cmd = *opt++;
3772        if (cmd == '-') {
3773                /* convert the long-form arguments into short-form versions */
3774                if (!prefixcmp(opt, "break-rewrites")) {
3775                        opt += strlen("break-rewrites");
3776                        if (*opt == 0 || *opt++ == '=')
3777                                cmd = 'B';
3778                } else if (!prefixcmp(opt, "find-copies")) {
3779                        opt += strlen("find-copies");
3780                        if (*opt == 0 || *opt++ == '=')
3781                                cmd = 'C';
3782                } else if (!prefixcmp(opt, "find-renames")) {
3783                        opt += strlen("find-renames");
3784                        if (*opt == 0 || *opt++ == '=')
3785                                cmd = 'M';
3786                }
3787        }
3788        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3789                return -1; /* that is not a -M, -C nor -B option */
3790
3791        opt1 = parse_rename_score(&opt);
3792        if (cmd != 'B')
3793                opt2 = 0;
3794        else {
3795                if (*opt == 0)
3796                        opt2 = 0;
3797                else if (*opt != '/')
3798                        return -1; /* we expect -B80/99 or -B80 */
3799                else {
3800                        opt++;
3801                        opt2 = parse_rename_score(&opt);
3802                }
3803        }
3804        if (*opt != 0)
3805                return -1;
3806        return opt1 | (opt2 << 16);
3807}
3808
3809struct diff_queue_struct diff_queued_diff;
3810
3811void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3812{
3813        if (queue->alloc <= queue->nr) {
3814                queue->alloc = alloc_nr(queue->alloc);
3815                queue->queue = xrealloc(queue->queue,
3816                                        sizeof(dp) * queue->alloc);
3817        }
3818        queue->queue[queue->nr++] = dp;
3819}
3820
3821struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3822                                 struct diff_filespec *one,
3823                                 struct diff_filespec *two)
3824{
3825        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3826        dp->one = one;
3827        dp->two = two;
3828        if (queue)
3829                diff_q(queue, dp);
3830        return dp;
3831}
3832
3833void diff_free_filepair(struct diff_filepair *p)
3834{
3835        free_filespec(p->one);
3836        free_filespec(p->two);
3837        free(p);
3838}
3839
3840/* This is different from find_unique_abbrev() in that
3841 * it stuffs the result with dots for alignment.
3842 */
3843const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3844{
3845        int abblen;
3846        const char *abbrev;
3847        if (len == 40)
3848                return sha1_to_hex(sha1);
3849
3850        abbrev = find_unique_abbrev(sha1, len);
3851        abblen = strlen(abbrev);
3852        if (abblen < 37) {
3853                static char hex[41];
3854                if (len < abblen && abblen <= len + 2)
3855                        sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3856                else
3857                        sprintf(hex, "%s...", abbrev);
3858                return hex;
3859        }
3860        return sha1_to_hex(sha1);
3861}
3862
3863static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3864{
3865        int line_termination = opt->line_termination;
3866        int inter_name_termination = line_termination ? '\t' : '\0';
3867        if (opt->output_prefix) {
3868                struct strbuf *msg = NULL;
3869                msg = opt->output_prefix(opt, opt->output_prefix_data);
3870                fprintf(opt->file, "%s", msg->buf);
3871        }
3872
3873        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3874                fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3875                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
3876                fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3877        }
3878        if (p->score) {
3879                fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3880                        inter_name_termination);
3881        } else {
3882                fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3883        }
3884
3885        if (p->status == DIFF_STATUS_COPIED ||
3886            p->status == DIFF_STATUS_RENAMED) {
3887                const char *name_a, *name_b;
3888                name_a = p->one->path;
3889                name_b = p->two->path;
3890                strip_prefix(opt->prefix_length, &name_a, &name_b);
3891                write_name_quoted(name_a, opt->file, inter_name_termination);
3892                write_name_quoted(name_b, opt->file, line_termination);
3893        } else {
3894                const char *name_a, *name_b;
3895                name_a = p->one->mode ? p->one->path : p->two->path;
3896                name_b = NULL;
3897                strip_prefix(opt->prefix_length, &name_a, &name_b);
3898                write_name_quoted(name_a, opt->file, line_termination);
3899        }
3900}
3901
3902int diff_unmodified_pair(struct diff_filepair *p)
3903{
3904        /* This function is written stricter than necessary to support
3905         * the currently implemented transformers, but the idea is to
3906         * let transformers to produce diff_filepairs any way they want,
3907         * and filter and clean them up here before producing the output.
3908         */
3909        struct diff_filespec *one = p->one, *two = p->two;
3910
3911        if (DIFF_PAIR_UNMERGED(p))
3912                return 0; /* unmerged is interesting */
3913
3914        /* deletion, addition, mode or type change
3915         * and rename are all interesting.
3916         */
3917        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3918            DIFF_PAIR_MODE_CHANGED(p) ||
3919            strcmp(one->path, two->path))
3920                return 0;
3921
3922        /* both are valid and point at the same path.  that is, we are
3923         * dealing with a change.
3924         */
3925        if (one->sha1_valid && two->sha1_valid &&
3926            !hashcmp(one->sha1, two->sha1) &&
3927            !one->dirty_submodule && !two->dirty_submodule)
3928                return 1; /* no change */
3929        if (!one->sha1_valid && !two->sha1_valid)
3930                return 1; /* both look at the same file on the filesystem. */
3931        return 0;
3932}
3933
3934static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3935{
3936        if (diff_unmodified_pair(p))
3937                return;
3938
3939        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3940            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3941                return; /* no tree diffs in patch format */
3942
3943        run_diff(p, o);
3944}
3945
3946static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3947                            struct diffstat_t *diffstat)
3948{
3949        if (diff_unmodified_pair(p))
3950                return;
3951
3952        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3953            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3954                return; /* no useful stat for tree diffs */
3955
3956        run_diffstat(p, o, diffstat);
3957}
3958
3959static void diff_flush_checkdiff(struct diff_filepair *p,
3960                struct diff_options *o)
3961{
3962        if (diff_unmodified_pair(p))
3963                return;
3964
3965        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3966            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3967                return; /* nothing to check in tree diffs */
3968
3969        run_checkdiff(p, o);
3970}
3971
3972int diff_queue_is_empty(void)
3973{
3974        struct diff_queue_struct *q = &diff_queued_diff;
3975        int i;
3976        for (i = 0; i < q->nr; i++)
3977                if (!diff_unmodified_pair(q->queue[i]))
3978                        return 0;
3979        return 1;
3980}
3981
3982#if DIFF_DEBUG
3983void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3984{
3985        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3986                x, one ? one : "",
3987                s->path,
3988                DIFF_FILE_VALID(s) ? "valid" : "invalid",
3989                s->mode,
3990                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3991        fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3992                x, one ? one : "",
3993                s->size, s->xfrm_flags);
3994}
3995
3996void diff_debug_filepair(const struct diff_filepair *p, int i)
3997{
3998        diff_debug_filespec(p->one, i, "one");
3999        diff_debug_filespec(p->two, i, "two");
4000        fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4001                p->score, p->status ? p->status : '?',
4002                p->one->rename_used, p->broken_pair);
4003}
4004
4005void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4006{
4007        int i;
4008        if (msg)
4009                fprintf(stderr, "%s\n", msg);
4010        fprintf(stderr, "q->nr = %d\n", q->nr);
4011        for (i = 0; i < q->nr; i++) {
4012                struct diff_filepair *p = q->queue[i];
4013                diff_debug_filepair(p, i);
4014        }
4015}
4016#endif
4017
4018static void diff_resolve_rename_copy(void)
4019{
4020        int i;
4021        struct diff_filepair *p;
4022        struct diff_queue_struct *q = &diff_queued_diff;
4023
4024        diff_debug_queue("resolve-rename-copy", q);
4025
4026        for (i = 0; i < q->nr; i++) {
4027                p = q->queue[i];
4028                p->status = 0; /* undecided */
4029                if (DIFF_PAIR_UNMERGED(p))
4030                        p->status = DIFF_STATUS_UNMERGED;
4031                else if (!DIFF_FILE_VALID(p->one))
4032                        p->status = DIFF_STATUS_ADDED;
4033                else if (!DIFF_FILE_VALID(p->two))
4034                        p->status = DIFF_STATUS_DELETED;
4035                else if (DIFF_PAIR_TYPE_CHANGED(p))
4036                        p->status = DIFF_STATUS_TYPE_CHANGED;
4037
4038                /* from this point on, we are dealing with a pair
4039                 * whose both sides are valid and of the same type, i.e.
4040                 * either in-place edit or rename/copy edit.
4041                 */
4042                else if (DIFF_PAIR_RENAME(p)) {
4043                        /*
4044                         * A rename might have re-connected a broken
4045                         * pair up, causing the pathnames to be the
4046                         * same again. If so, that's not a rename at
4047                         * all, just a modification..
4048                         *
4049                         * Otherwise, see if this source was used for
4050                         * multiple renames, in which case we decrement
4051                         * the count, and call it a copy.
4052                         */
4053                        if (!strcmp(p->one->path, p->two->path))
4054                                p->status = DIFF_STATUS_MODIFIED;
4055                        else if (--p->one->rename_used > 0)
4056                                p->status = DIFF_STATUS_COPIED;
4057                        else
4058                                p->status = DIFF_STATUS_RENAMED;
4059                }
4060                else if (hashcmp(p->one->sha1, p->two->sha1) ||
4061                         p->one->mode != p->two->mode ||
4062                         p->one->dirty_submodule ||
4063                         p->two->dirty_submodule ||
4064                         is_null_sha1(p->one->sha1))
4065                        p->status = DIFF_STATUS_MODIFIED;
4066                else {
4067                        /* This is a "no-change" entry and should not
4068                         * happen anymore, but prepare for broken callers.
4069                         */
4070                        error("feeding unmodified %s to diffcore",
4071                              p->one->path);
4072                        p->status = DIFF_STATUS_UNKNOWN;
4073                }
4074        }
4075        diff_debug_queue("resolve-rename-copy done", q);
4076}
4077
4078static int check_pair_status(struct diff_filepair *p)
4079{
4080        switch (p->status) {
4081        case DIFF_STATUS_UNKNOWN:
4082                return 0;
4083        case 0:
4084                die("internal error in diff-resolve-rename-copy");
4085        default:
4086                return 1;
4087        }
4088}
4089
4090static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4091{
4092        int fmt = opt->output_format;
4093
4094        if (fmt & DIFF_FORMAT_CHECKDIFF)
4095                diff_flush_checkdiff(p, opt);
4096        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4097                diff_flush_raw(p, opt);
4098        else if (fmt & DIFF_FORMAT_NAME) {
4099                const char *name_a, *name_b;
4100                name_a = p->two->path;
4101                name_b = NULL;
4102                strip_prefix(opt->prefix_length, &name_a, &name_b);
4103                write_name_quoted(name_a, opt->file, opt->line_termination);
4104        }
4105}
4106
4107static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4108{
4109        if (fs->mode)
4110                fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4111        else
4112                fprintf(file, " %s ", newdelete);
4113        write_name_quoted(fs->path, file, '\n');
4114}
4115
4116
4117static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4118                const char *line_prefix)
4119{
4120        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4121                fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4122                        p->two->mode, show_name ? ' ' : '\n');
4123                if (show_name) {
4124                        write_name_quoted(p->two->path, file, '\n');
4125                }
4126        }
4127}
4128
4129static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4130                        const char *line_prefix)
4131{
4132        char *names = pprint_rename(p->one->path, p->two->path);
4133
4134        fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4135        free(names);
4136        show_mode_change(file, p, 0, line_prefix);
4137}
4138
4139static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4140{
4141        FILE *file = opt->file;
4142        char *line_prefix = "";
4143
4144        if (opt->output_prefix) {
4145                struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
4146                line_prefix = buf->buf;
4147        }
4148
4149        switch(p->status) {
4150        case DIFF_STATUS_DELETED:
4151                fputs(line_prefix, file);
4152                show_file_mode_name(file, "delete", p->one);
4153                break;
4154        case DIFF_STATUS_ADDED:
4155                fputs(line_prefix, file);
4156                show_file_mode_name(file, "create", p->two);
4157                break;
4158        case DIFF_STATUS_COPIED:
4159                fputs(line_prefix, file);
4160                show_rename_copy(file, "copy", p, line_prefix);
4161                break;
4162        case DIFF_STATUS_RENAMED:
4163                fputs(line_prefix, file);
4164                show_rename_copy(file, "rename", p, line_prefix);
4165                break;
4166        default:
4167                if (p->score) {
4168                        fprintf(file, "%s rewrite ", line_prefix);
4169                        write_name_quoted(p->two->path, file, ' ');
4170                        fprintf(file, "(%d%%)\n", similarity_index(p));
4171                }
4172                show_mode_change(file, p, !p->score, line_prefix);
4173                break;
4174        }
4175}
4176
4177struct patch_id_t {
4178        git_SHA_CTX *ctx;
4179        int patchlen;
4180};
4181
4182static int remove_space(char *line, int len)
4183{
4184        int i;
4185        char *dst = line;
4186        unsigned char c;
4187
4188        for (i = 0; i < len; i++)
4189                if (!isspace((c = line[i])))
4190                        *dst++ = c;
4191
4192        return dst - line;
4193}
4194
4195static void patch_id_consume(void *priv, char *line, unsigned long len)
4196{
4197        struct patch_id_t *data = priv;
4198        int new_len;
4199
4200        /* Ignore line numbers when computing the SHA1 of the patch */
4201        if (!prefixcmp(line, "@@ -"))
4202                return;
4203
4204        new_len = remove_space(line, len);
4205
4206        git_SHA1_Update(data->ctx, line, new_len);
4207        data->patchlen += new_len;
4208}
4209
4210/* returns 0 upon success, and writes result into sha1 */
4211static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4212{
4213        struct diff_queue_struct *q = &diff_queued_diff;
4214        int i;
4215        git_SHA_CTX ctx;
4216        struct patch_id_t data;
4217        char buffer[PATH_MAX * 4 + 20];
4218
4219        git_SHA1_Init(&ctx);
4220        memset(&data, 0, sizeof(struct patch_id_t));
4221        data.ctx = &ctx;
4222
4223        for (i = 0; i < q->nr; i++) {
4224                xpparam_t xpp;
4225                xdemitconf_t xecfg;
4226                mmfile_t mf1, mf2;
4227                struct diff_filepair *p = q->queue[i];
4228                int len1, len2;
4229
4230                memset(&xpp, 0, sizeof(xpp));
4231                memset(&xecfg, 0, sizeof(xecfg));
4232                if (p->status == 0)
4233                        return error("internal diff status error");
4234                if (p->status == DIFF_STATUS_UNKNOWN)
4235                        continue;
4236                if (diff_unmodified_pair(p))
4237                        continue;
4238                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4239                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4240                        continue;
4241                if (DIFF_PAIR_UNMERGED(p))
4242                        continue;
4243
4244                diff_fill_sha1_info(p->one);
4245                diff_fill_sha1_info(p->two);
4246                if (fill_mmfile(&mf1, p->one) < 0 ||
4247                                fill_mmfile(&mf2, p->two) < 0)
4248                        return error("unable to read files to diff");
4249
4250                len1 = remove_space(p->one->path, strlen(p->one->path));
4251                len2 = remove_space(p->two->path, strlen(p->two->path));
4252                if (p->one->mode == 0)
4253                        len1 = snprintf(buffer, sizeof(buffer),
4254                                        "diff--gita/%.*sb/%.*s"
4255                                        "newfilemode%06o"
4256                                        "---/dev/null"
4257                                        "+++b/%.*s",
4258                                        len1, p->one->path,
4259                                        len2, p->two->path,
4260                                        p->two->mode,
4261                                        len2, p->two->path);
4262                else if (p->two->mode == 0)
4263                        len1 = snprintf(buffer, sizeof(buffer),
4264                                        "diff--gita/%.*sb/%.*s"
4265                                        "deletedfilemode%06o"
4266                                        "---a/%.*s"
4267                                        "+++/dev/null",
4268                                        len1, p->one->path,
4269                                        len2, p->two->path,
4270                                        p->one->mode,
4271                                        len1, p->one->path);
4272                else
4273                        len1 = snprintf(buffer, sizeof(buffer),
4274                                        "diff--gita/%.*sb/%.*s"
4275                                        "---a/%.*s"
4276                                        "+++b/%.*s",
4277                                        len1, p->one->path,
4278                                        len2, p->two->path,
4279                                        len1, p->one->path,
4280                                        len2, p->two->path);
4281                git_SHA1_Update(&ctx, buffer, len1);
4282
4283                if (diff_filespec_is_binary(p->one) ||
4284                    diff_filespec_is_binary(p->two)) {
4285                        git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4286                        git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4287                        continue;
4288                }
4289
4290                xpp.flags = 0;
4291                xecfg.ctxlen = 3;
4292                xecfg.flags = 0;
4293                xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4294                              &xpp, &xecfg);
4295        }
4296
4297        git_SHA1_Final(sha1, &ctx);
4298        return 0;
4299}
4300
4301int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4302{
4303        struct diff_queue_struct *q = &diff_queued_diff;
4304        int i;
4305        int result = diff_get_patch_id(options, sha1);
4306
4307        for (i = 0; i < q->nr; i++)
4308                diff_free_filepair(q->queue[i]);
4309
4310        free(q->queue);
4311        DIFF_QUEUE_CLEAR(q);
4312
4313        return result;
4314}
4315
4316static int is_summary_empty(const struct diff_queue_struct *q)
4317{
4318        int i;
4319
4320        for (i = 0; i < q->nr; i++) {
4321                const struct diff_filepair *p = q->queue[i];
4322
4323                switch (p->status) {
4324                case DIFF_STATUS_DELETED:
4325                case DIFF_STATUS_ADDED:
4326                case DIFF_STATUS_COPIED:
4327                case DIFF_STATUS_RENAMED:
4328                        return 0;
4329                default:
4330                        if (p->score)
4331                                return 0;
4332                        if (p->one->mode && p->two->mode &&
4333                            p->one->mode != p->two->mode)
4334                                return 0;
4335                        break;
4336                }
4337        }
4338        return 1;
4339}
4340
4341static const char rename_limit_warning[] =
4342"inexact rename detection was skipped due to too many files.";
4343
4344static const char degrade_cc_to_c_warning[] =
4345"only found copies from modified paths due to too many files.";
4346
4347static const char rename_limit_advice[] =
4348"you may want to set your %s variable to at least "
4349"%d and retry the command.";
4350
4351void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4352{
4353        if (degraded_cc)
4354                warning(degrade_cc_to_c_warning);
4355        else if (needed)
4356                warning(rename_limit_warning);
4357        else
4358                return;
4359        if (0 < needed && needed < 32767)
4360                warning(rename_limit_advice, varname, needed);
4361}
4362
4363void diff_flush(struct diff_options *options)
4364{
4365        struct diff_queue_struct *q = &diff_queued_diff;
4366        int i, output_format = options->output_format;
4367        int separator = 0;
4368        int dirstat_by_line = 0;
4369
4370        /*
4371         * Order: raw, stat, summary, patch
4372         * or:    name/name-status/checkdiff (other bits clear)
4373         */
4374        if (!q->nr)
4375                goto free_queue;
4376
4377        if (output_format & (DIFF_FORMAT_RAW |
4378                             DIFF_FORMAT_NAME |
4379                             DIFF_FORMAT_NAME_STATUS |
4380                             DIFF_FORMAT_CHECKDIFF)) {
4381                for (i = 0; i < q->nr; i++) {
4382                        struct diff_filepair *p = q->queue[i];
4383                        if (check_pair_status(p))
4384                                flush_one_pair(p, options);
4385                }
4386                separator++;
4387        }
4388
4389        if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4390                dirstat_by_line = 1;
4391
4392        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4393            dirstat_by_line) {
4394                struct diffstat_t diffstat;
4395
4396                memset(&diffstat, 0, sizeof(struct diffstat_t));
4397                for (i = 0; i < q->nr; i++) {
4398                        struct diff_filepair *p = q->queue[i];
4399                        if (check_pair_status(p))
4400                                diff_flush_stat(p, options, &diffstat);
4401                }
4402                if (output_format & DIFF_FORMAT_NUMSTAT)
4403                        show_numstat(&diffstat, options);
4404                if (output_format & DIFF_FORMAT_DIFFSTAT)
4405                        show_stats(&diffstat, options);
4406                if (output_format & DIFF_FORMAT_SHORTSTAT)
4407                        show_shortstats(&diffstat, options);
4408                if (output_format & DIFF_FORMAT_DIRSTAT)
4409                        show_dirstat_by_line(&diffstat, options);
4410                free_diffstat_info(&diffstat);
4411                separator++;
4412        }
4413        if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4414                show_dirstat(options);
4415
4416        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4417                for (i = 0; i < q->nr; i++) {
4418                        diff_summary(options, q->queue[i]);
4419                }
4420                separator++;
4421        }
4422
4423        if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4424            DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4425            DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4426                /*
4427                 * run diff_flush_patch for the exit status. setting
4428                 * options->file to /dev/null should be safe, becaue we
4429                 * aren't supposed to produce any output anyway.
4430                 */
4431                if (options->close_file)
4432                        fclose(options->file);
4433                options->file = fopen("/dev/null", "w");
4434                if (!options->file)
4435                        die_errno("Could not open /dev/null");
4436                options->close_file = 1;
4437                for (i = 0; i < q->nr; i++) {
4438                        struct diff_filepair *p = q->queue[i];
4439                        if (check_pair_status(p))
4440                                diff_flush_patch(p, options);
4441                        if (options->found_changes)
4442                                break;
4443                }
4444        }
4445
4446        if (output_format & DIFF_FORMAT_PATCH) {
4447                if (separator) {
4448                        if (options->output_prefix) {
4449                                struct strbuf *msg = NULL;
4450                                msg = options->output_prefix(options,
4451                                        options->output_prefix_data);
4452                                fwrite(msg->buf, msg->len, 1, stdout);
4453                        }
4454                        putc(options->line_termination, options->file);
4455                        if (options->stat_sep) {
4456                                /* attach patch instead of inline */
4457                                fputs(options->stat_sep, options->file);
4458                        }
4459                }
4460
4461                for (i = 0; i < q->nr; i++) {
4462                        struct diff_filepair *p = q->queue[i];
4463                        if (check_pair_status(p))
4464                                diff_flush_patch(p, options);
4465                }
4466        }
4467
4468        if (output_format & DIFF_FORMAT_CALLBACK)
4469                options->format_callback(q, options, options->format_callback_data);
4470
4471        for (i = 0; i < q->nr; i++)
4472                diff_free_filepair(q->queue[i]);
4473free_queue:
4474        free(q->queue);
4475        DIFF_QUEUE_CLEAR(q);
4476        if (options->close_file)
4477                fclose(options->file);
4478
4479        /*
4480         * Report the content-level differences with HAS_CHANGES;
4481         * diff_addremove/diff_change does not set the bit when
4482         * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4483         */
4484        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4485                if (options->found_changes)
4486                        DIFF_OPT_SET(options, HAS_CHANGES);
4487                else
4488                        DIFF_OPT_CLR(options, HAS_CHANGES);
4489        }
4490}
4491
4492static void diffcore_apply_filter(const char *filter)
4493{
4494        int i;
4495        struct diff_queue_struct *q = &diff_queued_diff;
4496        struct diff_queue_struct outq;
4497        DIFF_QUEUE_CLEAR(&outq);
4498
4499        if (!filter)
4500                return;
4501
4502        if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4503                int found;
4504                for (i = found = 0; !found && i < q->nr; i++) {
4505                        struct diff_filepair *p = q->queue[i];
4506                        if (((p->status == DIFF_STATUS_MODIFIED) &&
4507                             ((p->score &&
4508                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4509                              (!p->score &&
4510                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4511                            ((p->status != DIFF_STATUS_MODIFIED) &&
4512                             strchr(filter, p->status)))
4513                                found++;
4514                }
4515                if (found)
4516                        return;
4517
4518                /* otherwise we will clear the whole queue
4519                 * by copying the empty outq at the end of this
4520                 * function, but first clear the current entries
4521                 * in the queue.
4522                 */
4523                for (i = 0; i < q->nr; i++)
4524                        diff_free_filepair(q->queue[i]);
4525        }
4526        else {
4527                /* Only the matching ones */
4528                for (i = 0; i < q->nr; i++) {
4529                        struct diff_filepair *p = q->queue[i];
4530
4531                        if (((p->status == DIFF_STATUS_MODIFIED) &&
4532                             ((p->score &&
4533                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4534                              (!p->score &&
4535                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4536                            ((p->status != DIFF_STATUS_MODIFIED) &&
4537                             strchr(filter, p->status)))
4538                                diff_q(&outq, p);
4539                        else
4540                                diff_free_filepair(p);
4541                }
4542        }
4543        free(q->queue);
4544        *q = outq;
4545}
4546
4547/* Check whether two filespecs with the same mode and size are identical */
4548static int diff_filespec_is_identical(struct diff_filespec *one,
4549                                      struct diff_filespec *two)
4550{
4551        if (S_ISGITLINK(one->mode))
4552                return 0;
4553        if (diff_populate_filespec(one, 0))
4554                return 0;
4555        if (diff_populate_filespec(two, 0))
4556                return 0;
4557        return !memcmp(one->data, two->data, one->size);
4558}
4559
4560static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4561{
4562        int i;
4563        struct diff_queue_struct *q = &diff_queued_diff;
4564        struct diff_queue_struct outq;
4565        DIFF_QUEUE_CLEAR(&outq);
4566
4567        for (i = 0; i < q->nr; i++) {
4568                struct diff_filepair *p = q->queue[i];
4569
4570                /*
4571                 * 1. Entries that come from stat info dirtiness
4572                 *    always have both sides (iow, not create/delete),
4573                 *    one side of the object name is unknown, with
4574                 *    the same mode and size.  Keep the ones that
4575                 *    do not match these criteria.  They have real
4576                 *    differences.
4577                 *
4578                 * 2. At this point, the file is known to be modified,
4579                 *    with the same mode and size, and the object
4580                 *    name of one side is unknown.  Need to inspect
4581                 *    the identical contents.
4582                 */
4583                if (!DIFF_FILE_VALID(p->one) || /* (1) */
4584                    !DIFF_FILE_VALID(p->two) ||
4585                    (p->one->sha1_valid && p->two->sha1_valid) ||
4586                    (p->one->mode != p->two->mode) ||
4587                    diff_populate_filespec(p->one, 1) ||
4588                    diff_populate_filespec(p->two, 1) ||
4589                    (p->one->size != p->two->size) ||
4590                    !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4591                        diff_q(&outq, p);
4592                else {
4593                        /*
4594                         * The caller can subtract 1 from skip_stat_unmatch
4595                         * to determine how many paths were dirty only
4596                         * due to stat info mismatch.
4597                         */
4598                        if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4599                                diffopt->skip_stat_unmatch++;
4600                        diff_free_filepair(p);
4601                }
4602        }
4603        free(q->queue);
4604        *q = outq;
4605}
4606
4607static int diffnamecmp(const void *a_, const void *b_)
4608{
4609        const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4610        const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4611        const char *name_a, *name_b;
4612
4613        name_a = a->one ? a->one->path : a->two->path;
4614        name_b = b->one ? b->one->path : b->two->path;
4615        return strcmp(name_a, name_b);
4616}
4617
4618void diffcore_fix_diff_index(struct diff_options *options)
4619{
4620        struct diff_queue_struct *q = &diff_queued_diff;
4621        qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4622}
4623
4624void diffcore_std(struct diff_options *options)
4625{
4626        if (options->skip_stat_unmatch)
4627                diffcore_skip_stat_unmatch(options);
4628        if (!options->found_follow) {
4629                /* See try_to_follow_renames() in tree-diff.c */
4630                if (options->break_opt != -1)
4631                        diffcore_break(options->break_opt);
4632                if (options->detect_rename)
4633                        diffcore_rename(options);
4634                if (options->break_opt != -1)
4635                        diffcore_merge_broken();
4636        }
4637        if (options->pickaxe)
4638                diffcore_pickaxe(options);
4639        if (options->orderfile)
4640                diffcore_order(options->orderfile);
4641        if (!options->found_follow)
4642                /* See try_to_follow_renames() in tree-diff.c */
4643                diff_resolve_rename_copy();
4644        diffcore_apply_filter(options->filter);
4645
4646        if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4647                DIFF_OPT_SET(options, HAS_CHANGES);
4648        else
4649                DIFF_OPT_CLR(options, HAS_CHANGES);
4650
4651        options->found_follow = 0;
4652}
4653
4654int diff_result_code(struct diff_options *opt, int status)
4655{
4656        int result = 0;
4657
4658        diff_warn_rename_limit("diff.renamelimit",
4659                               opt->needed_rename_limit,
4660                               opt->degraded_cc_to_c);
4661        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4662            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4663                return status;
4664        if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4665            DIFF_OPT_TST(opt, HAS_CHANGES))
4666                result |= 01;
4667        if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4668            DIFF_OPT_TST(opt, CHECK_FAILED))
4669                result |= 02;
4670        return result;
4671}
4672
4673int diff_can_quit_early(struct diff_options *opt)
4674{
4675        return (DIFF_OPT_TST(opt, QUICK) &&
4676                !opt->filter &&
4677                DIFF_OPT_TST(opt, HAS_CHANGES));
4678}
4679
4680/*
4681 * Shall changes to this submodule be ignored?
4682 *
4683 * Submodule changes can be configured to be ignored separately for each path,
4684 * but that configuration can be overridden from the command line.
4685 */
4686static int is_submodule_ignored(const char *path, struct diff_options *options)
4687{
4688        int ignored = 0;
4689        unsigned orig_flags = options->flags;
4690        if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4691                set_diffopt_flags_from_submodule_config(options, path);
4692        if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4693                ignored = 1;
4694        options->flags = orig_flags;
4695        return ignored;
4696}
4697
4698void diff_addremove(struct diff_options *options,
4699                    int addremove, unsigned mode,
4700                    const unsigned char *sha1,
4701                    int sha1_valid,
4702                    const char *concatpath, unsigned dirty_submodule)
4703{
4704        struct diff_filespec *one, *two;
4705
4706        if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4707                return;
4708
4709        /* This may look odd, but it is a preparation for
4710         * feeding "there are unchanged files which should
4711         * not produce diffs, but when you are doing copy
4712         * detection you would need them, so here they are"
4713         * entries to the diff-core.  They will be prefixed
4714         * with something like '=' or '*' (I haven't decided
4715         * which but should not make any difference).
4716         * Feeding the same new and old to diff_change()
4717         * also has the same effect.
4718         * Before the final output happens, they are pruned after
4719         * merged into rename/copy pairs as appropriate.
4720         */
4721        if (DIFF_OPT_TST(options, REVERSE_DIFF))
4722                addremove = (addremove == '+' ? '-' :
4723                             addremove == '-' ? '+' : addremove);
4724
4725        if (options->prefix &&
4726            strncmp(concatpath, options->prefix, options->prefix_length))
4727                return;
4728
4729        one = alloc_filespec(concatpath);
4730        two = alloc_filespec(concatpath);
4731
4732        if (addremove != '+')
4733                fill_filespec(one, sha1, sha1_valid, mode);
4734        if (addremove != '-') {
4735                fill_filespec(two, sha1, sha1_valid, mode);
4736                two->dirty_submodule = dirty_submodule;
4737        }
4738
4739        diff_queue(&diff_queued_diff, one, two);
4740        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4741                DIFF_OPT_SET(options, HAS_CHANGES);
4742}
4743
4744void diff_change(struct diff_options *options,
4745                 unsigned old_mode, unsigned new_mode,
4746                 const unsigned char *old_sha1,
4747                 const unsigned char *new_sha1,
4748                 int old_sha1_valid, int new_sha1_valid,
4749                 const char *concatpath,
4750                 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4751{
4752        struct diff_filespec *one, *two;
4753
4754        if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4755            is_submodule_ignored(concatpath, options))
4756                return;
4757
4758        if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4759                unsigned tmp;
4760                const unsigned char *tmp_c;
4761                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4762                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4763                tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4764                        new_sha1_valid = tmp;
4765                tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4766                        new_dirty_submodule = tmp;
4767        }
4768
4769        if (options->prefix &&
4770            strncmp(concatpath, options->prefix, options->prefix_length))
4771                return;
4772
4773        one = alloc_filespec(concatpath);
4774        two = alloc_filespec(concatpath);
4775        fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4776        fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4777        one->dirty_submodule = old_dirty_submodule;
4778        two->dirty_submodule = new_dirty_submodule;
4779
4780        diff_queue(&diff_queued_diff, one, two);
4781        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4782                DIFF_OPT_SET(options, HAS_CHANGES);
4783}
4784
4785struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4786{
4787        struct diff_filepair *pair;
4788        struct diff_filespec *one, *two;
4789
4790        if (options->prefix &&
4791            strncmp(path, options->prefix, options->prefix_length))
4792                return NULL;
4793
4794        one = alloc_filespec(path);
4795        two = alloc_filespec(path);
4796        pair = diff_queue(&diff_queued_diff, one, two);
4797        pair->is_unmerged = 1;
4798        return pair;
4799}
4800
4801static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4802                size_t *outsize)
4803{
4804        struct diff_tempfile *temp;
4805        const char *argv[3];
4806        const char **arg = argv;
4807        struct child_process child;
4808        struct strbuf buf = STRBUF_INIT;
4809        int err = 0;
4810
4811        temp = prepare_temp_file(spec->path, spec);
4812        *arg++ = pgm;
4813        *arg++ = temp->name;
4814        *arg = NULL;
4815
4816        memset(&child, 0, sizeof(child));
4817        child.use_shell = 1;
4818        child.argv = argv;
4819        child.out = -1;
4820        if (start_command(&child)) {
4821                remove_tempfile();
4822                return NULL;
4823        }
4824
4825        if (strbuf_read(&buf, child.out, 0) < 0)
4826                err = error("error reading from textconv command '%s'", pgm);
4827        close(child.out);
4828
4829        if (finish_command(&child) || err) {
4830                strbuf_release(&buf);
4831                remove_tempfile();
4832                return NULL;
4833        }
4834        remove_tempfile();
4835
4836        return strbuf_detach(&buf, outsize);
4837}
4838
4839size_t fill_textconv(struct userdiff_driver *driver,
4840                     struct diff_filespec *df,
4841                     char **outbuf)
4842{
4843        size_t size;
4844
4845        if (!driver || !driver->textconv) {
4846                if (!DIFF_FILE_VALID(df)) {
4847                        *outbuf = "";
4848                        return 0;
4849                }
4850                if (diff_populate_filespec(df, 0))
4851                        die("unable to read files to diff");
4852                *outbuf = df->data;
4853                return df->size;
4854        }
4855
4856        if (driver->textconv_cache && df->sha1_valid) {
4857                *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4858                                          &size);
4859                if (*outbuf)
4860                        return size;
4861        }
4862
4863        *outbuf = run_textconv(driver->textconv, df, &size);
4864        if (!*outbuf)
4865                die("unable to read files to diff");
4866
4867        if (driver->textconv_cache && df->sha1_valid) {
4868                /* ignore errors, as we might be in a readonly repository */
4869                notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4870                                size);
4871                /*
4872                 * we could save up changes and flush them all at the end,
4873                 * but we would need an extra call after all diffing is done.
4874                 * Since generating a cache entry is the slow path anyway,
4875                 * this extra overhead probably isn't a big deal.
4876                 */
4877                notes_cache_write(driver->textconv_cache);
4878        }
4879
4880        return size;
4881}