diff.c

   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
  16 #ifdef NO_FAST_WORKING_DIRECTORY
  17 #define FAST_WORKING_DIRECTORY 0
  18 #else
  19 #define FAST_WORKING_DIRECTORY 1
  20 #endif
  21
  22 static int diff_detect_rename_default;
  23 static int diff_rename_limit_default = 200;
  24 static int diff_suppress_blank_empty;
  25 int diff_use_color_default = -1;
  26 static const char *external_diff_cmd_cfg;
  27 int diff_auto_refresh_index = 1;
  28 static int diff_mnemonic_prefix;
  29
  30 static char diff_colors[][COLOR_MAXLEN] = {
  31         "\033[m",       /* reset */
  32         "",             /* PLAIN (normal) */
  33         "\033[1m",      /* METAINFO (bold) */
  34         "\033[36m",     /* FRAGINFO (cyan) */
  35         "\033[31m",     /* OLD (red) */
  36         "\033[32m",     /* NEW (green) */
  37         "\033[33m",     /* COMMIT (yellow) */
  38         "\033[41m",     /* WHITESPACE (red background) */
  39 };
  40
  41 static void diff_filespec_load_driver(struct diff_filespec *one);
  42 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
  43
  44 static int parse_diff_color_slot(const char *var, int ofs)
  45 {
  46         if (!strcasecmp(var+ofs, "plain"))
  47                 return DIFF_PLAIN;
  48         if (!strcasecmp(var+ofs, "meta"))
  49                 return DIFF_METAINFO;
  50         if (!strcasecmp(var+ofs, "frag"))
  51                 return DIFF_FRAGINFO;
  52         if (!strcasecmp(var+ofs, "old"))
  53                 return DIFF_FILE_OLD;
  54         if (!strcasecmp(var+ofs, "new"))
  55                 return DIFF_FILE_NEW;
  56         if (!strcasecmp(var+ofs, "commit"))
  57                 return DIFF_COMMIT;
  58         if (!strcasecmp(var+ofs, "whitespace"))
  59                 return DIFF_WHITESPACE;
  60         die("bad config variable '%s'", var);
  61 }
  62
  63 /*
  64  * These are to give UI layer defaults.
  65  * The core-level commands such as git-diff-files should
  66  * never be affected by the setting of diff.renames
  67  * the user happens to have in the configuration file.
  68  */
  69 int git_diff_ui_config(const char *var, const char *value, void *cb)
  70 {
  71         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
  72                 diff_use_color_default = git_config_colorbool(var, value, -1);
  73                 return 0;
  74         }
  75         if (!strcmp(var, "diff.renames")) {
  76                 if (!value)
  77                         diff_detect_rename_default = DIFF_DETECT_RENAME;
  78                 else if (!strcasecmp(value, "copies") ||
  79                          !strcasecmp(value, "copy"))
  80                         diff_detect_rename_default = DIFF_DETECT_COPY;
  81                 else if (git_config_bool(var,value))
  82                         diff_detect_rename_default = DIFF_DETECT_RENAME;
  83                 return 0;
  84         }
  85         if (!strcmp(var, "diff.autorefreshindex")) {
  86                 diff_auto_refresh_index = git_config_bool(var, value);
  87                 return 0;
  88         }
  89         if (!strcmp(var, "diff.mnemonicprefix")) {
  90                 diff_mnemonic_prefix = git_config_bool(var, value);
  91                 return 0;
  92         }
  93         if (!strcmp(var, "diff.external"))
  94                 return git_config_string(&external_diff_cmd_cfg, var, value);
  95
  96         switch (userdiff_config_porcelain(var, value)) {
  97                 case 0: break;
  98                 case -1: return -1;
  99                 default: return 0;
 100         }
 101
 102         return git_diff_basic_config(var, value, cb);
 103 }
 104
 105 int git_diff_basic_config(const char *var, const char *value, void *cb)
 106 {
 107         if (!strcmp(var, "diff.renamelimit")) {
 108                 diff_rename_limit_default = git_config_int(var, value);
 109                 return 0;
 110         }
 111
 112         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 113                 int slot = parse_diff_color_slot(var, 11);
 114                 if (!value)
 115                         return config_error_nonbool(var);
 116                 color_parse(value, var, diff_colors[slot]);
 117                 return 0;
 118         }
 119
 120         /* like GNU diff's --suppress-blank-empty option  */
 121         if (!strcmp(var, "diff.suppress-blank-empty")) {
 122                 diff_suppress_blank_empty = git_config_bool(var, value);
 123                 return 0;
 124         }
 125
 126         switch (userdiff_config_basic(var, value)) {
 127                 case 0: break;
 128                 case -1: return -1;
 129                 default: return 0;
 130         }
 131
 132         return git_color_default_config(var, value, cb);
 133 }
 134
 135 static char *quote_two(const char *one, const char *two)
 136 {
 137         int need_one = quote_c_style(one, NULL, NULL, 1);
 138         int need_two = quote_c_style(two, NULL, NULL, 1);
 139         struct strbuf res = STRBUF_INIT;
 140
 141         if (need_one + need_two) {
 142                 strbuf_addch(&res, '"');
 143                 quote_c_style(one, &res, NULL, 1);
 144                 quote_c_style(two, &res, NULL, 1);
 145                 strbuf_addch(&res, '"');
 146         } else {
 147                 strbuf_addstr(&res, one);
 148                 strbuf_addstr(&res, two);
 149         }
 150         return strbuf_detach(&res, NULL);
 151 }
 152
 153 static const char *external_diff(void)
 154 {
 155         static const char *external_diff_cmd = NULL;
 156         static int done_preparing = 0;
 157
 158         if (done_preparing)
 159                 return external_diff_cmd;
 160         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 161         if (!external_diff_cmd)
 162                 external_diff_cmd = external_diff_cmd_cfg;
 163         done_preparing = 1;
 164         return external_diff_cmd;
 165 }
 166
 167 static struct diff_tempfile {
 168         const char *name; /* filename external diff should read from */
 169         char hex[41];
 170         char mode[10];
 171         char tmp_path[PATH_MAX];
 172 } diff_temp[2];
 173
 174 static int count_lines(const char *data, int size)
 175 {
 176         int count, ch, completely_empty = 1, nl_just_seen = 0;
 177         count = 0;
 178         while (0 < size--) {
 179                 ch = *data++;
 180                 if (ch == '\n') {
 181                         count++;
 182                         nl_just_seen = 1;
 183                         completely_empty = 0;
 184                 }
 185                 else {
 186                         nl_just_seen = 0;
 187                         completely_empty = 0;
 188                 }
 189         }
 190         if (completely_empty)
 191                 return 0;
 192         if (!nl_just_seen)
 193                 count++; /* no trailing newline */
 194         return count;
 195 }
 196
 197 static void print_line_count(FILE *file, int count)
 198 {
 199         switch (count) {
 200         case 0:
 201                 fprintf(file, "0,0");
 202                 break;
 203         case 1:
 204                 fprintf(file, "1");
 205                 break;
 206         default:
 207                 fprintf(file, "1,%d", count);
 208                 break;
 209         }
 210 }
 211
 212 static void copy_file_with_prefix(FILE *file,
 213                                   int prefix, const char *data, int size,
 214                                   const char *set, const char *reset)
 215 {
 216         int ch, nl_just_seen = 1;
 217         while (0 < size--) {
 218                 ch = *data++;
 219                 if (nl_just_seen) {
 220                         fputs(set, file);
 221                         putc(prefix, file);
 222                 }
 223                 if (ch == '\n') {
 224                         nl_just_seen = 1;
 225                         fputs(reset, file);
 226                 } else
 227                         nl_just_seen = 0;
 228                 putc(ch, file);
 229         }
 230         if (!nl_just_seen)
 231                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 232 }
 233
 234 static void emit_rewrite_diff(const char *name_a,
 235                               const char *name_b,
 236                               struct diff_filespec *one,
 237                               struct diff_filespec *two,
 238                               struct diff_options *o)
 239 {
 240         int lc_a, lc_b;
 241         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 242         const char *name_a_tab, *name_b_tab;
 243         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 244         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 245         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 246         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 247         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 248         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 249         const char *a_prefix, *b_prefix;
 250
 251         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 252                 a_prefix = o->b_prefix;
 253                 b_prefix = o->a_prefix;
 254         } else {
 255                 a_prefix = o->a_prefix;
 256                 b_prefix = o->b_prefix;
 257         }
 258
 259         name_a += (*name_a == '/');
 260         name_b += (*name_b == '/');
 261         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 262         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 263
 264         strbuf_reset(&a_name);
 265         strbuf_reset(&b_name);
 266         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 267         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 268
 269         diff_populate_filespec(one, 0);
 270         diff_populate_filespec(two, 0);
 271         lc_a = count_lines(one->data, one->size);
 272         lc_b = count_lines(two->data, two->size);
 273         fprintf(o->file,
 274                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 275                 metainfo, a_name.buf, name_a_tab, reset,
 276                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 277         print_line_count(o->file, lc_a);
 278         fprintf(o->file, " +");
 279         print_line_count(o->file, lc_b);
 280         fprintf(o->file, " @@%s\n", reset);
 281         if (lc_a)
 282                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
 283         if (lc_b)
 284                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
 285 }
 286
 287 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 288 {
 289         if (!DIFF_FILE_VALID(one)) {
 290                 mf->ptr = (char *)""; /* does not matter */
 291                 mf->size = 0;
 292                 return 0;
 293         }
 294         else if (diff_populate_filespec(one, 0))
 295                 return -1;
 296
 297         mf->ptr = one->data;
 298         mf->size = one->size;
 299         return 0;
 300 }
 301
 302 struct diff_words_buffer {
 303         mmfile_t text;
 304         long alloc;
 305         long current; /* output pointer */
 306         int suppressed_newline;
 307 };
 308
 309 static void diff_words_append(char *line, unsigned long len,
 310                 struct diff_words_buffer *buffer)
 311 {
 312         if (buffer->text.size + len > buffer->alloc) {
 313                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 314                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 315         }
 316         line++;
 317         len--;
 318         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 319         buffer->text.size += len;
 320 }
 321
 322 struct diff_words_data {
 323         struct diff_words_buffer minus, plus;
 324         FILE *file;
 325 };
 326
 327 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 328                 int suppress_newline)
 329 {
 330         const char *ptr;
 331         int eol = 0;
 332
 333         if (len == 0)
 334                 return;
 335
 336         ptr  = buffer->text.ptr + buffer->current;
 337         buffer->current += len;
 338
 339         if (ptr[len - 1] == '\n') {
 340                 eol = 1;
 341                 len--;
 342         }
 343
 344         fputs(diff_get_color(1, color), file);
 345         fwrite(ptr, len, 1, file);
 346         fputs(diff_get_color(1, DIFF_RESET), file);
 347
 348         if (eol) {
 349                 if (suppress_newline)
 350                         buffer->suppressed_newline = 1;
 351                 else
 352                         putc('\n', file);
 353         }
 354 }
 355
 356 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 357 {
 358         struct diff_words_data *diff_words = priv;
 359
 360         if (diff_words->minus.suppressed_newline) {
 361                 if (line[0] != '+')
 362                         putc('\n', diff_words->file);
 363                 diff_words->minus.suppressed_newline = 0;
 364         }
 365
 366         len--;
 367         switch (line[0]) {
 368                 case '-':
 369                         print_word(diff_words->file,
 370                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 371                         break;
 372                 case '+':
 373                         print_word(diff_words->file,
 374                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 375                         break;
 376                 case ' ':
 377                         print_word(diff_words->file,
 378                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 379                         diff_words->minus.current += len;
 380                         break;
 381         }
 382 }
 383
 384 /* this executes the word diff on the accumulated buffers */
 385 static void diff_words_show(struct diff_words_data *diff_words)
 386 {
 387         xpparam_t xpp;
 388         xdemitconf_t xecfg;
 389         xdemitcb_t ecb;
 390         mmfile_t minus, plus;
 391         int i;
 392
 393         memset(&xecfg, 0, sizeof(xecfg));
 394         minus.size = diff_words->minus.text.size;
 395         minus.ptr = xmalloc(minus.size);
 396         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 397         for (i = 0; i < minus.size; i++)
 398                 if (isspace(minus.ptr[i]))
 399                         minus.ptr[i] = '\n';
 400         diff_words->minus.current = 0;
 401
 402         plus.size = diff_words->plus.text.size;
 403         plus.ptr = xmalloc(plus.size);
 404         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 405         for (i = 0; i < plus.size; i++)
 406                 if (isspace(plus.ptr[i]))
 407                         plus.ptr[i] = '\n';
 408         diff_words->plus.current = 0;
 409
 410         xpp.flags = XDF_NEED_MINIMAL;
 411         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 412         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 413                       &xpp, &xecfg, &ecb);
 414         free(minus.ptr);
 415         free(plus.ptr);
 416         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 417
 418         if (diff_words->minus.suppressed_newline) {
 419                 putc('\n', diff_words->file);
 420                 diff_words->minus.suppressed_newline = 0;
 421         }
 422 }
 423
 424 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 425
 426 struct emit_callback {
 427         int nparents, color_diff;
 428         unsigned ws_rule;
 429         sane_truncate_fn truncate;
 430         const char **label_path;
 431         struct diff_words_data *diff_words;
 432         int *found_changesp;
 433         FILE *file;
 434 };
 435
 436 static void free_diff_words_data(struct emit_callback *ecbdata)
 437 {
 438         if (ecbdata->diff_words) {
 439                 /* flush buffers */
 440                 if (ecbdata->diff_words->minus.text.size ||
 441                                 ecbdata->diff_words->plus.text.size)
 442                         diff_words_show(ecbdata->diff_words);
 443
 444                 free (ecbdata->diff_words->minus.text.ptr);
 445                 free (ecbdata->diff_words->plus.text.ptr);
 446                 free(ecbdata->diff_words);
 447                 ecbdata->diff_words = NULL;
 448         }
 449 }
 450
 451 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 452 {
 453         if (diff_use_color)
 454                 return diff_colors[ix];
 455         return "";
 456 }
 457
 458 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 459 {
 460         int has_trailing_newline, has_trailing_carriage_return;
 461
 462         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 463         if (has_trailing_newline)
 464                 len--;
 465         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 466         if (has_trailing_carriage_return)
 467                 len--;
 468
 469         fputs(set, file);
 470         fwrite(line, len, 1, file);
 471         fputs(reset, file);
 472         if (has_trailing_carriage_return)
 473                 fputc('\r', file);
 474         if (has_trailing_newline)
 475                 fputc('\n', file);
 476 }
 477
 478 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 479 {
 480         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 481         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 482
 483         if (!*ws)
 484                 emit_line(ecbdata->file, set, reset, line, len);
 485         else {
 486                 /* Emit just the prefix, then the rest. */
 487                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 488                 ws_check_emit(line + ecbdata->nparents,
 489                               len - ecbdata->nparents, ecbdata->ws_rule,
 490                               ecbdata->file, set, reset, ws);
 491         }
 492 }
 493
 494 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 495 {
 496         const char *cp;
 497         unsigned long allot;
 498         size_t l = len;
 499
 500         if (ecb->truncate)
 501                 return ecb->truncate(line, len);
 502         cp = line;
 503         allot = l;
 504         while (0 < l) {
 505                 (void) utf8_width(&cp, &l);
 506                 if (!cp)
 507                         break; /* truncated in the middle? */
 508         }
 509         return allot - l;
 510 }
 511
 512 static void fn_out_consume(void *priv, char *line, unsigned long len)
 513 {
 514         int i;
 515         int color;
 516         struct emit_callback *ecbdata = priv;
 517         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 518         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 519         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 520
 521         *(ecbdata->found_changesp) = 1;
 522
 523         if (ecbdata->label_path[0]) {
 524                 const char *name_a_tab, *name_b_tab;
 525
 526                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 527                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 528
 529                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 530                         meta, ecbdata->label_path[0], reset, name_a_tab);
 531                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 532                         meta, ecbdata->label_path[1], reset, name_b_tab);
 533                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 534         }
 535
 536         if (diff_suppress_blank_empty
 537             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 538                 line[0] = '\n';
 539                 len = 1;
 540         }
 541
 542         /* This is not really necessary for now because
 543          * this codepath only deals with two-way diffs.
 544          */
 545         for (i = 0; i < len && line[i] == '@'; i++)
 546                 ;
 547         if (2 <= i && i < len && line[i] == ' ') {
 548                 ecbdata->nparents = i - 1;
 549                 len = sane_truncate_line(ecbdata, line, len);
 550                 emit_line(ecbdata->file,
 551                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 552                           reset, line, len);
 553                 if (line[len-1] != '\n')
 554                         putc('\n', ecbdata->file);
 555                 return;
 556         }
 557
 558         if (len < ecbdata->nparents) {
 559                 emit_line(ecbdata->file, reset, reset, line, len);
 560                 return;
 561         }
 562
 563         color = DIFF_PLAIN;
 564         if (ecbdata->diff_words && ecbdata->nparents != 1)
 565                 /* fall back to normal diff */
 566                 free_diff_words_data(ecbdata);
 567         if (ecbdata->diff_words) {
 568                 if (line[0] == '-') {
 569                         diff_words_append(line, len,
 570                                           &ecbdata->diff_words->minus);
 571                         return;
 572                 } else if (line[0] == '+') {
 573                         diff_words_append(line, len,
 574                                           &ecbdata->diff_words->plus);
 575                         return;
 576                 }
 577                 if (ecbdata->diff_words->minus.text.size ||
 578                     ecbdata->diff_words->plus.text.size)
 579                         diff_words_show(ecbdata->diff_words);
 580                 line++;
 581                 len--;
 582                 emit_line(ecbdata->file, plain, reset, line, len);
 583                 return;
 584         }
 585         for (i = 0; i < ecbdata->nparents && len; i++) {
 586                 if (line[i] == '-')
 587                         color = DIFF_FILE_OLD;
 588                 else if (line[i] == '+')
 589                         color = DIFF_FILE_NEW;
 590         }
 591
 592         if (color != DIFF_FILE_NEW) {
 593                 emit_line(ecbdata->file,
 594                           diff_get_color(ecbdata->color_diff, color),
 595                           reset, line, len);
 596                 return;
 597         }
 598         emit_add_line(reset, ecbdata, line, len);
 599 }
 600
 601 static char *pprint_rename(const char *a, const char *b)
 602 {
 603         const char *old = a;
 604         const char *new = b;
 605         struct strbuf name = STRBUF_INIT;
 606         int pfx_length, sfx_length;
 607         int len_a = strlen(a);
 608         int len_b = strlen(b);
 609         int a_midlen, b_midlen;
 610         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 611         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 612
 613         if (qlen_a || qlen_b) {
 614                 quote_c_style(a, &name, NULL, 0);
 615                 strbuf_addstr(&name, " => ");
 616                 quote_c_style(b, &name, NULL, 0);
 617                 return strbuf_detach(&name, NULL);
 618         }
 619
 620         /* Find common prefix */
 621         pfx_length = 0;
 622         while (*old && *new && *old == *new) {
 623                 if (*old == '/')
 624                         pfx_length = old - a + 1;
 625                 old++;
 626                 new++;
 627         }
 628
 629         /* Find common suffix */
 630         old = a + len_a;
 631         new = b + len_b;
 632         sfx_length = 0;
 633         while (a <= old && b <= new && *old == *new) {
 634                 if (*old == '/')
 635                         sfx_length = len_a - (old - a);
 636                 old--;
 637                 new--;
 638         }
 639
 640         /*
 641          * pfx{mid-a => mid-b}sfx
 642          * {pfx-a => pfx-b}sfx
 643          * pfx{sfx-a => sfx-b}
 644          * name-a => name-b
 645          */
 646         a_midlen = len_a - pfx_length - sfx_length;
 647         b_midlen = len_b - pfx_length - sfx_length;
 648         if (a_midlen < 0)
 649                 a_midlen = 0;
 650         if (b_midlen < 0)
 651                 b_midlen = 0;
 652
 653         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 654         if (pfx_length + sfx_length) {
 655                 strbuf_add(&name, a, pfx_length);
 656                 strbuf_addch(&name, '{');
 657         }
 658         strbuf_add(&name, a + pfx_length, a_midlen);
 659         strbuf_addstr(&name, " => ");
 660         strbuf_add(&name, b + pfx_length, b_midlen);
 661         if (pfx_length + sfx_length) {
 662                 strbuf_addch(&name, '}');
 663                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 664         }
 665         return strbuf_detach(&name, NULL);
 666 }
 667
 668 struct diffstat_t {
 669         int nr;
 670         int alloc;
 671         struct diffstat_file {
 672                 char *from_name;
 673                 char *name;
 674                 char *print_name;
 675                 unsigned is_unmerged:1;
 676                 unsigned is_binary:1;
 677                 unsigned is_renamed:1;
 678                 unsigned int added, deleted;
 679         } **files;
 680 };
 681
 682 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 683                                           const char *name_a,
 684                                           const char *name_b)
 685 {
 686         struct diffstat_file *x;
 687         x = xcalloc(sizeof (*x), 1);
 688         if (diffstat->nr == diffstat->alloc) {
 689                 diffstat->alloc = alloc_nr(diffstat->alloc);
 690                 diffstat->files = xrealloc(diffstat->files,
 691                                 diffstat->alloc * sizeof(x));
 692         }
 693         diffstat->files[diffstat->nr++] = x;
 694         if (name_b) {
 695                 x->from_name = xstrdup(name_a);
 696                 x->name = xstrdup(name_b);
 697                 x->is_renamed = 1;
 698         }
 699         else {
 700                 x->from_name = NULL;
 701                 x->name = xstrdup(name_a);
 702         }
 703         return x;
 704 }
 705
 706 static void diffstat_consume(void *priv, char *line, unsigned long len)
 707 {
 708         struct diffstat_t *diffstat = priv;
 709         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 710
 711         if (line[0] == '+')
 712                 x->added++;
 713         else if (line[0] == '-')
 714                 x->deleted++;
 715 }
 716
 717 const char mime_boundary_leader[] = "------------";
 718
 719 static int scale_linear(int it, int width, int max_change)
 720 {
 721         /*
 722          * make sure that at least one '-' is printed if there were deletions,
 723          * and likewise for '+'.
 724          */
 725         if (max_change < 2)
 726                 return it;
 727         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 728 }
 729
 730 static void show_name(FILE *file,
 731                       const char *prefix, const char *name, int len,
 732                       const char *reset, const char *set)
 733 {
 734         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 735 }
 736
 737 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 738 {
 739         if (cnt <= 0)
 740                 return;
 741         fprintf(file, "%s", set);
 742         while (cnt--)
 743                 putc(ch, file);
 744         fprintf(file, "%s", reset);
 745 }
 746
 747 static void fill_print_name(struct diffstat_file *file)
 748 {
 749         char *pname;
 750
 751         if (file->print_name)
 752                 return;
 753
 754         if (!file->is_renamed) {
 755                 struct strbuf buf = STRBUF_INIT;
 756                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 757                         pname = strbuf_detach(&buf, NULL);
 758                 } else {
 759                         pname = file->name;
 760                         strbuf_release(&buf);
 761                 }
 762         } else {
 763                 pname = pprint_rename(file->from_name, file->name);
 764         }
 765         file->print_name = pname;
 766 }
 767
 768 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 769 {
 770         int i, len, add, del, total, adds = 0, dels = 0;
 771         int max_change = 0, max_len = 0;
 772         int total_files = data->nr;
 773         int width, name_width;
 774         const char *reset, *set, *add_c, *del_c;
 775
 776         if (data->nr == 0)
 777                 return;
 778
 779         width = options->stat_width ? options->stat_width : 80;
 780         name_width = options->stat_name_width ? options->stat_name_width : 50;
 781
 782         /* Sanity: give at least 5 columns to the graph,
 783          * but leave at least 10 columns for the name.
 784          */
 785         if (width < 25)
 786                 width = 25;
 787         if (name_width < 10)
 788                 name_width = 10;
 789         else if (width < name_width + 15)
 790                 name_width = width - 15;
 791
 792         /* Find the longest filename and max number of changes */
 793         reset = diff_get_color_opt(options, DIFF_RESET);
 794         set   = diff_get_color_opt(options, DIFF_PLAIN);
 795         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 796         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 797
 798         for (i = 0; i < data->nr; i++) {
 799                 struct diffstat_file *file = data->files[i];
 800                 int change = file->added + file->deleted;
 801                 fill_print_name(file);
 802                 len = strlen(file->print_name);
 803                 if (max_len < len)
 804                         max_len = len;
 805
 806                 if (file->is_binary || file->is_unmerged)
 807                         continue;
 808                 if (max_change < change)
 809                         max_change = change;
 810         }
 811
 812         /* Compute the width of the graph part;
 813          * 10 is for one blank at the beginning of the line plus
 814          * " | count " between the name and the graph.
 815          *
 816          * From here on, name_width is the width of the name area,
 817          * and width is the width of the graph area.
 818          */
 819         name_width = (name_width < max_len) ? name_width : max_len;
 820         if (width < (name_width + 10) + max_change)
 821                 width = width - (name_width + 10);
 822         else
 823                 width = max_change;
 824
 825         for (i = 0; i < data->nr; i++) {
 826                 const char *prefix = "";
 827                 char *name = data->files[i]->print_name;
 828                 int added = data->files[i]->added;
 829                 int deleted = data->files[i]->deleted;
 830                 int name_len;
 831
 832                 /*
 833                  * "scale" the filename
 834                  */
 835                 len = name_width;
 836                 name_len = strlen(name);
 837                 if (name_width < name_len) {
 838                         char *slash;
 839                         prefix = "...";
 840                         len -= 3;
 841                         name += name_len - len;
 842                         slash = strchr(name, '/');
 843                         if (slash)
 844                                 name = slash;
 845                 }
 846
 847                 if (data->files[i]->is_binary) {
 848                         show_name(options->file, prefix, name, len, reset, set);
 849                         fprintf(options->file, "  Bin ");
 850                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 851                         fprintf(options->file, " -> ");
 852                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 853                         fprintf(options->file, " bytes");
 854                         fprintf(options->file, "\n");
 855                         continue;
 856                 }
 857                 else if (data->files[i]->is_unmerged) {
 858                         show_name(options->file, prefix, name, len, reset, set);
 859                         fprintf(options->file, "  Unmerged\n");
 860                         continue;
 861                 }
 862                 else if (!data->files[i]->is_renamed &&
 863                          (added + deleted == 0)) {
 864                         total_files--;
 865                         continue;
 866                 }
 867
 868                 /*
 869                  * scale the add/delete
 870                  */
 871                 add = added;
 872                 del = deleted;
 873                 total = add + del;
 874                 adds += add;
 875                 dels += del;
 876
 877                 if (width <= max_change) {
 878                         add = scale_linear(add, width, max_change);
 879                         del = scale_linear(del, width, max_change);
 880                         total = add + del;
 881                 }
 882                 show_name(options->file, prefix, name, len, reset, set);
 883                 fprintf(options->file, "%5d%s", added + deleted,
 884                                 added + deleted ? " " : "");
 885                 show_graph(options->file, '+', add, add_c, reset);
 886                 show_graph(options->file, '-', del, del_c, reset);
 887                 fprintf(options->file, "\n");
 888         }
 889         fprintf(options->file,
 890                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 891                set, total_files, adds, dels, reset);
 892 }
 893
 894 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 895 {
 896         int i, adds = 0, dels = 0, total_files = data->nr;
 897
 898         if (data->nr == 0)
 899                 return;
 900
 901         for (i = 0; i < data->nr; i++) {
 902                 if (!data->files[i]->is_binary &&
 903                     !data->files[i]->is_unmerged) {
 904                         int added = data->files[i]->added;
 905                         int deleted= data->files[i]->deleted;
 906                         if (!data->files[i]->is_renamed &&
 907                             (added + deleted == 0)) {
 908                                 total_files--;
 909                         } else {
 910                                 adds += added;
 911                                 dels += deleted;
 912                         }
 913                 }
 914         }
 915         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 916                total_files, adds, dels);
 917 }
 918
 919 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 920 {
 921         int i;
 922
 923         if (data->nr == 0)
 924                 return;
 925
 926         for (i = 0; i < data->nr; i++) {
 927                 struct diffstat_file *file = data->files[i];
 928
 929                 if (file->is_binary)
 930                         fprintf(options->file, "-\t-\t");
 931                 else
 932                         fprintf(options->file,
 933                                 "%d\t%d\t", file->added, file->deleted);
 934                 if (options->line_termination) {
 935                         fill_print_name(file);
 936                         if (!file->is_renamed)
 937                                 write_name_quoted(file->name, options->file,
 938                                                   options->line_termination);
 939                         else {
 940                                 fputs(file->print_name, options->file);
 941                                 putc(options->line_termination, options->file);
 942                         }
 943                 } else {
 944                         if (file->is_renamed) {
 945                                 putc('\0', options->file);
 946                                 write_name_quoted(file->from_name, options->file, '\0');
 947                         }
 948                         write_name_quoted(file->name, options->file, '\0');
 949                 }
 950         }
 951 }
 952
 953 struct dirstat_file {
 954         const char *name;
 955         unsigned long changed;
 956 };
 957
 958 struct dirstat_dir {
 959         struct dirstat_file *files;
 960         int alloc, nr, percent, cumulative;
 961 };
 962
 963 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
 964 {
 965         unsigned long this_dir = 0;
 966         unsigned int sources = 0;
 967
 968         while (dir->nr) {
 969                 struct dirstat_file *f = dir->files;
 970                 int namelen = strlen(f->name);
 971                 unsigned long this;
 972                 char *slash;
 973
 974                 if (namelen < baselen)
 975                         break;
 976                 if (memcmp(f->name, base, baselen))
 977                         break;
 978                 slash = strchr(f->name + baselen, '/');
 979                 if (slash) {
 980                         int newbaselen = slash + 1 - f->name;
 981                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
 982                         sources++;
 983                 } else {
 984                         this = f->changed;
 985                         dir->files++;
 986                         dir->nr--;
 987                         sources += 2;
 988                 }
 989                 this_dir += this;
 990         }
 991
 992         /*
 993          * We don't report dirstat's for
 994          *  - the top level
 995          *  - or cases where everything came from a single directory
 996          *    under this directory (sources == 1).
 997          */
 998         if (baselen && sources != 1) {
 999                 int permille = this_dir * 1000 / changed;
1000                 if (permille) {
1001                         int percent = permille / 10;
1002                         if (percent >= dir->percent) {
1003                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1004                                 if (!dir->cumulative)
1005                                         return 0;
1006                         }
1007                 }
1008         }
1009         return this_dir;
1010 }
1011
1012 static int dirstat_compare(const void *_a, const void *_b)
1013 {
1014         const struct dirstat_file *a = _a;
1015         const struct dirstat_file *b = _b;
1016         return strcmp(a->name, b->name);
1017 }
1018
1019 static void show_dirstat(struct diff_options *options)
1020 {
1021         int i;
1022         unsigned long changed;
1023         struct dirstat_dir dir;
1024         struct diff_queue_struct *q = &diff_queued_diff;
1025
1026         dir.files = NULL;
1027         dir.alloc = 0;
1028         dir.nr = 0;
1029         dir.percent = options->dirstat_percent;
1030         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1031
1032         changed = 0;
1033         for (i = 0; i < q->nr; i++) {
1034                 struct diff_filepair *p = q->queue[i];
1035                 const char *name;
1036                 unsigned long copied, added, damage;
1037
1038                 name = p->one->path ? p->one->path : p->two->path;
1039
1040                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1041                         diff_populate_filespec(p->one, 0);
1042                         diff_populate_filespec(p->two, 0);
1043                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1044                                                &copied, &added);
1045                         diff_free_filespec_data(p->one);
1046                         diff_free_filespec_data(p->two);
1047                 } else if (DIFF_FILE_VALID(p->one)) {
1048                         diff_populate_filespec(p->one, 1);
1049                         copied = added = 0;
1050                         diff_free_filespec_data(p->one);
1051                 } else if (DIFF_FILE_VALID(p->two)) {
1052                         diff_populate_filespec(p->two, 1);
1053                         copied = 0;
1054                         added = p->two->size;
1055                         diff_free_filespec_data(p->two);
1056                 } else
1057                         continue;
1058
1059                 /*
1060                  * Original minus copied is the removed material,
1061                  * added is the new material.  They are both damages
1062                  * made to the preimage. In --dirstat-by-file mode, count
1063                  * damaged files, not damaged lines. This is done by
1064                  * counting only a single damaged line per file.
1065                  */
1066                 damage = (p->one->size - copied) + added;
1067                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1068                         damage = 1;
1069
1070                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1071                 dir.files[dir.nr].name = name;
1072                 dir.files[dir.nr].changed = damage;
1073                 changed += damage;
1074                 dir.nr++;
1075         }
1076
1077         /* This can happen even with many files, if everything was renames */
1078         if (!changed)
1079                 return;
1080
1081         /* Show all directories with more than x% of the changes */
1082         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1083         gather_dirstat(options->file, &dir, changed, "", 0);
1084 }
1085
1086 static void free_diffstat_info(struct diffstat_t *diffstat)
1087 {
1088         int i;
1089         for (i = 0; i < diffstat->nr; i++) {
1090                 struct diffstat_file *f = diffstat->files[i];
1091                 if (f->name != f->print_name)
1092                         free(f->print_name);
1093                 free(f->name);
1094                 free(f->from_name);
1095                 free(f);
1096         }
1097         free(diffstat->files);
1098 }
1099
1100 struct checkdiff_t {
1101         const char *filename;
1102         int lineno;
1103         struct diff_options *o;
1104         unsigned ws_rule;
1105         unsigned status;
1106         int trailing_blanks_start;
1107 };
1108
1109 static int is_conflict_marker(const char *line, unsigned long len)
1110 {
1111         char firstchar;
1112         int cnt;
1113
1114         if (len < 8)
1115                 return 0;
1116         firstchar = line[0];
1117         switch (firstchar) {
1118         case '=': case '>': case '<':
1119                 break;
1120         default:
1121                 return 0;
1122         }
1123         for (cnt = 1; cnt < 7; cnt++)
1124                 if (line[cnt] != firstchar)
1125                         return 0;
1126         /* line[0] thru line[6] are same as firstchar */
1127         if (firstchar == '=') {
1128                 /* divider between ours and theirs? */
1129                 if (len != 8 || line[7] != '\n')
1130                         return 0;
1131         } else if (len < 8 || !isspace(line[7])) {
1132                 /* not divider before ours nor after theirs */
1133                 return 0;
1134         }
1135         return 1;
1136 }
1137
1138 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1139 {
1140         struct checkdiff_t *data = priv;
1141         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1142         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1143         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1144         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1145         char *err;
1146
1147         if (line[0] == '+') {
1148                 unsigned bad;
1149                 data->lineno++;
1150                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1151                         data->trailing_blanks_start = 0;
1152                 else if (!data->trailing_blanks_start)
1153                         data->trailing_blanks_start = data->lineno;
1154                 if (is_conflict_marker(line + 1, len - 1)) {
1155                         data->status |= 1;
1156                         fprintf(data->o->file,
1157                                 "%s:%d: leftover conflict marker\n",
1158                                 data->filename, data->lineno);
1159                 }
1160                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1161                 if (!bad)
1162                         return;
1163                 data->status |= bad;
1164                 err = whitespace_error_string(bad);
1165                 fprintf(data->o->file, "%s:%d: %s.\n",
1166                         data->filename, data->lineno, err);
1167                 free(err);
1168                 emit_line(data->o->file, set, reset, line, 1);
1169                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1170                               data->o->file, set, reset, ws);
1171         } else if (line[0] == ' ') {
1172                 data->lineno++;
1173                 data->trailing_blanks_start = 0;
1174         } else if (line[0] == '@') {
1175                 char *plus = strchr(line, '+');
1176                 if (plus)
1177                         data->lineno = strtol(plus, NULL, 10) - 1;
1178                 else
1179                         die("invalid diff");
1180                 data->trailing_blanks_start = 0;
1181         }
1182 }
1183
1184 static unsigned char *deflate_it(char *data,
1185                                  unsigned long size,
1186                                  unsigned long *result_size)
1187 {
1188         int bound;
1189         unsigned char *deflated;
1190         z_stream stream;
1191
1192         memset(&stream, 0, sizeof(stream));
1193         deflateInit(&stream, zlib_compression_level);
1194         bound = deflateBound(&stream, size);
1195         deflated = xmalloc(bound);
1196         stream.next_out = deflated;
1197         stream.avail_out = bound;
1198
1199         stream.next_in = (unsigned char *)data;
1200         stream.avail_in = size;
1201         while (deflate(&stream, Z_FINISH) == Z_OK)
1202                 ; /* nothing */
1203         deflateEnd(&stream);
1204         *result_size = stream.total_out;
1205         return deflated;
1206 }
1207
1208 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1209 {
1210         void *cp;
1211         void *delta;
1212         void *deflated;
1213         void *data;
1214         unsigned long orig_size;
1215         unsigned long delta_size;
1216         unsigned long deflate_size;
1217         unsigned long data_size;
1218
1219         /* We could do deflated delta, or we could do just deflated two,
1220          * whichever is smaller.
1221          */
1222         delta = NULL;
1223         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1224         if (one->size && two->size) {
1225                 delta = diff_delta(one->ptr, one->size,
1226                                    two->ptr, two->size,
1227                                    &delta_size, deflate_size);
1228                 if (delta) {
1229                         void *to_free = delta;
1230                         orig_size = delta_size;
1231                         delta = deflate_it(delta, delta_size, &delta_size);
1232                         free(to_free);
1233                 }
1234         }
1235
1236         if (delta && delta_size < deflate_size) {
1237                 fprintf(file, "delta %lu\n", orig_size);
1238                 free(deflated);
1239                 data = delta;
1240                 data_size = delta_size;
1241         }
1242         else {
1243                 fprintf(file, "literal %lu\n", two->size);
1244                 free(delta);
1245                 data = deflated;
1246                 data_size = deflate_size;
1247         }
1248
1249         /* emit data encoded in base85 */
1250         cp = data;
1251         while (data_size) {
1252                 int bytes = (52 < data_size) ? 52 : data_size;
1253                 char line[70];
1254                 data_size -= bytes;
1255                 if (bytes <= 26)
1256                         line[0] = bytes + 'A' - 1;
1257                 else
1258                         line[0] = bytes - 26 + 'a' - 1;
1259                 encode_85(line + 1, cp, bytes);
1260                 cp = (char *) cp + bytes;
1261                 fputs(line, file);
1262                 fputc('\n', file);
1263         }
1264         fprintf(file, "\n");
1265         free(data);
1266 }
1267
1268 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1269 {
1270         fprintf(file, "GIT binary patch\n");
1271         emit_binary_diff_body(file, one, two);
1272         emit_binary_diff_body(file, two, one);
1273 }
1274
1275 static void diff_filespec_load_driver(struct diff_filespec *one)
1276 {
1277         if (!one->driver)
1278                 one->driver = userdiff_find_by_path(one->path);
1279         if (!one->driver)
1280                 one->driver = userdiff_find_by_name("default");
1281 }
1282
1283 int diff_filespec_is_binary(struct diff_filespec *one)
1284 {
1285         if (one->is_binary == -1) {
1286                 diff_filespec_load_driver(one);
1287                 if (one->driver->binary != -1)
1288                         one->is_binary = one->driver->binary;
1289                 else {
1290                         if (!one->data && DIFF_FILE_VALID(one))
1291                                 diff_populate_filespec(one, 0);
1292                         if (one->data)
1293                                 one->is_binary = buffer_is_binary(one->data,
1294                                                 one->size);
1295                         if (one->is_binary == -1)
1296                                 one->is_binary = 0;
1297                 }
1298         }
1299         return one->is_binary;
1300 }
1301
1302 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1303 {
1304         diff_filespec_load_driver(one);
1305         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1306 }
1307
1308 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1309 {
1310         if (!options->a_prefix)
1311                 options->a_prefix = a;
1312         if (!options->b_prefix)
1313                 options->b_prefix = b;
1314 }
1315
1316 static const char *get_textconv(struct diff_filespec *one)
1317 {
1318         if (!DIFF_FILE_VALID(one))
1319                 return NULL;
1320         diff_filespec_load_driver(one);
1321         return one->driver->textconv;
1322 }
1323
1324 static void builtin_diff(const char *name_a,
1325                          const char *name_b,
1326                          struct diff_filespec *one,
1327                          struct diff_filespec *two,
1328                          const char *xfrm_msg,
1329                          struct diff_options *o,
1330                          int complete_rewrite)
1331 {
1332         mmfile_t mf1, mf2;
1333         const char *lbl[2];
1334         char *a_one, *b_two;
1335         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1336         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1337         const char *a_prefix, *b_prefix;
1338         const char *textconv_one, *textconv_two;
1339
1340         diff_set_mnemonic_prefix(o, "a/", "b/");
1341         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1342                 a_prefix = o->b_prefix;
1343                 b_prefix = o->a_prefix;
1344         } else {
1345                 a_prefix = o->a_prefix;
1346                 b_prefix = o->b_prefix;
1347         }
1348
1349         /* Never use a non-valid filename anywhere if at all possible */
1350         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1351         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1352
1353         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1354         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1355         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1356         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1357         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1358         if (lbl[0][0] == '/') {
1359                 /* /dev/null */
1360                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1361                 if (xfrm_msg && xfrm_msg[0])
1362                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1363         }
1364         else if (lbl[1][0] == '/') {
1365                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1366                 if (xfrm_msg && xfrm_msg[0])
1367                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1368         }
1369         else {
1370                 if (one->mode != two->mode) {
1371                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1372                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1373                 }
1374                 if (xfrm_msg && xfrm_msg[0])
1375                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1376                 /*
1377                  * we do not run diff between different kind
1378                  * of objects.
1379                  */
1380                 if ((one->mode ^ two->mode) & S_IFMT)
1381                         goto free_ab_and_return;
1382                 if (complete_rewrite) {
1383                         emit_rewrite_diff(name_a, name_b, one, two, o);
1384                         o->found_changes = 1;
1385                         goto free_ab_and_return;
1386                 }
1387         }
1388
1389         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1390                 die("unable to read files to diff");
1391
1392         textconv_one = get_textconv(one);
1393         textconv_two = get_textconv(two);
1394
1395         if (!DIFF_OPT_TST(o, TEXT) &&
1396             ( (diff_filespec_is_binary(one) && !textconv_one) ||
1397               (diff_filespec_is_binary(two) && !textconv_two) )) {
1398                 /* Quite common confusing case */
1399                 if (mf1.size == mf2.size &&
1400                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1401                         goto free_ab_and_return;
1402                 if (DIFF_OPT_TST(o, BINARY))
1403                         emit_binary_diff(o->file, &mf1, &mf2);
1404                 else
1405                         fprintf(o->file, "Binary files %s and %s differ\n",
1406                                 lbl[0], lbl[1]);
1407                 o->found_changes = 1;
1408         }
1409         else {
1410                 /* Crazy xdl interfaces.. */
1411                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1412                 xpparam_t xpp;
1413                 xdemitconf_t xecfg;
1414                 xdemitcb_t ecb;
1415                 struct emit_callback ecbdata;
1416                 const struct userdiff_funcname *pe;
1417
1418                 if (textconv_one) {
1419                         size_t size;
1420                         mf1.ptr = run_textconv(textconv_one, one, &size);
1421                         if (!mf1.ptr)
1422                                 die("unable to read files to diff");
1423                         mf1.size = size;
1424                 }
1425                 if (textconv_two) {
1426                         size_t size;
1427                         mf2.ptr = run_textconv(textconv_two, two, &size);
1428                         if (!mf2.ptr)
1429                                 die("unable to read files to diff");
1430                         mf2.size = size;
1431                 }
1432
1433                 pe = diff_funcname_pattern(one);
1434                 if (!pe)
1435                         pe = diff_funcname_pattern(two);
1436
1437                 memset(&xecfg, 0, sizeof(xecfg));
1438                 memset(&ecbdata, 0, sizeof(ecbdata));
1439                 ecbdata.label_path = lbl;
1440                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1441                 ecbdata.found_changesp = &o->found_changes;
1442                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1443                 ecbdata.file = o->file;
1444                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1445                 xecfg.ctxlen = o->context;
1446                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1447                 if (pe)
1448                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1449                 if (!diffopts)
1450                         ;
1451                 else if (!prefixcmp(diffopts, "--unified="))
1452                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1453                 else if (!prefixcmp(diffopts, "-u"))
1454                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1455                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1456                         ecbdata.diff_words =
1457                                 xcalloc(1, sizeof(struct diff_words_data));
1458                         ecbdata.diff_words->file = o->file;
1459                 }
1460                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1461                               &xpp, &xecfg, &ecb);
1462                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1463                         free_diff_words_data(&ecbdata);
1464                 if (textconv_one)
1465                         free(mf1.ptr);
1466                 if (textconv_two)
1467                         free(mf2.ptr);
1468         }
1469
1470  free_ab_and_return:
1471         diff_free_filespec_data(one);
1472         diff_free_filespec_data(two);
1473         free(a_one);
1474         free(b_two);
1475         return;
1476 }
1477
1478 static void builtin_diffstat(const char *name_a, const char *name_b,
1479                              struct diff_filespec *one,
1480                              struct diff_filespec *two,
1481                              struct diffstat_t *diffstat,
1482                              struct diff_options *o,
1483                              int complete_rewrite)
1484 {
1485         mmfile_t mf1, mf2;
1486         struct diffstat_file *data;
1487
1488         data = diffstat_add(diffstat, name_a, name_b);
1489
1490         if (!one || !two) {
1491                 data->is_unmerged = 1;
1492                 return;
1493         }
1494         if (complete_rewrite) {
1495                 diff_populate_filespec(one, 0);
1496                 diff_populate_filespec(two, 0);
1497                 data->deleted = count_lines(one->data, one->size);
1498                 data->added = count_lines(two->data, two->size);
1499                 goto free_and_return;
1500         }
1501         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1502                 die("unable to read files to diff");
1503
1504         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1505                 data->is_binary = 1;
1506                 data->added = mf2.size;
1507                 data->deleted = mf1.size;
1508         } else {
1509                 /* Crazy xdl interfaces.. */
1510                 xpparam_t xpp;
1511                 xdemitconf_t xecfg;
1512                 xdemitcb_t ecb;
1513
1514                 memset(&xecfg, 0, sizeof(xecfg));
1515                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1516                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1517                               &xpp, &xecfg, &ecb);
1518         }
1519
1520  free_and_return:
1521         diff_free_filespec_data(one);
1522         diff_free_filespec_data(two);
1523 }
1524
1525 static void builtin_checkdiff(const char *name_a, const char *name_b,
1526                               const char *attr_path,
1527                               struct diff_filespec *one,
1528                               struct diff_filespec *two,
1529                               struct diff_options *o)
1530 {
1531         mmfile_t mf1, mf2;
1532         struct checkdiff_t data;
1533
1534         if (!two)
1535                 return;
1536
1537         memset(&data, 0, sizeof(data));
1538         data.filename = name_b ? name_b : name_a;
1539         data.lineno = 0;
1540         data.o = o;
1541         data.ws_rule = whitespace_rule(attr_path);
1542
1543         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1544                 die("unable to read files to diff");
1545
1546         /*
1547          * All the other codepaths check both sides, but not checking
1548          * the "old" side here is deliberate.  We are checking the newly
1549          * introduced changes, and as long as the "new" side is text, we
1550          * can and should check what it introduces.
1551          */
1552         if (diff_filespec_is_binary(two))
1553                 goto free_and_return;
1554         else {
1555                 /* Crazy xdl interfaces.. */
1556                 xpparam_t xpp;
1557                 xdemitconf_t xecfg;
1558                 xdemitcb_t ecb;
1559
1560                 memset(&xecfg, 0, sizeof(xecfg));
1561                 xecfg.ctxlen = 1; /* at least one context line */
1562                 xpp.flags = XDF_NEED_MINIMAL;
1563                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1564                               &xpp, &xecfg, &ecb);
1565
1566                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1567                     data.trailing_blanks_start) {
1568                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1569                                 data.filename, data.trailing_blanks_start);
1570                         data.status = 1; /* report errors */
1571                 }
1572         }
1573  free_and_return:
1574         diff_free_filespec_data(one);
1575         diff_free_filespec_data(two);
1576         if (data.status)
1577                 DIFF_OPT_SET(o, CHECK_FAILED);
1578 }
1579
1580 struct diff_filespec *alloc_filespec(const char *path)
1581 {
1582         int namelen = strlen(path);
1583         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1584
1585         memset(spec, 0, sizeof(*spec));
1586         spec->path = (char *)(spec + 1);
1587         memcpy(spec->path, path, namelen+1);
1588         spec->count = 1;
1589         spec->is_binary = -1;
1590         return spec;
1591 }
1592
1593 void free_filespec(struct diff_filespec *spec)
1594 {
1595         if (!--spec->count) {
1596                 diff_free_filespec_data(spec);
1597                 free(spec);
1598         }
1599 }
1600
1601 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1602                    unsigned short mode)
1603 {
1604         if (mode) {
1605                 spec->mode = canon_mode(mode);
1606                 hashcpy(spec->sha1, sha1);
1607                 spec->sha1_valid = !is_null_sha1(sha1);
1608         }
1609 }
1610
1611 /*
1612  * Given a name and sha1 pair, if the index tells us the file in
1613  * the work tree has that object contents, return true, so that
1614  * prepare_temp_file() does not have to inflate and extract.
1615  */
1616 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1617 {
1618         struct cache_entry *ce;
1619         struct stat st;
1620         int pos, len;
1621
1622         /* We do not read the cache ourselves here, because the
1623          * benchmark with my previous version that always reads cache
1624          * shows that it makes things worse for diff-tree comparing
1625          * two linux-2.6 kernel trees in an already checked out work
1626          * tree.  This is because most diff-tree comparisons deal with
1627          * only a small number of files, while reading the cache is
1628          * expensive for a large project, and its cost outweighs the
1629          * savings we get by not inflating the object to a temporary
1630          * file.  Practically, this code only helps when we are used
1631          * by diff-cache --cached, which does read the cache before
1632          * calling us.
1633          */
1634         if (!active_cache)
1635                 return 0;
1636
1637         /* We want to avoid the working directory if our caller
1638          * doesn't need the data in a normal file, this system
1639          * is rather slow with its stat/open/mmap/close syscalls,
1640          * and the object is contained in a pack file.  The pack
1641          * is probably already open and will be faster to obtain
1642          * the data through than the working directory.  Loose
1643          * objects however would tend to be slower as they need
1644          * to be individually opened and inflated.
1645          */
1646         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1647                 return 0;
1648
1649         len = strlen(name);
1650         pos = cache_name_pos(name, len);
1651         if (pos < 0)
1652                 return 0;
1653         ce = active_cache[pos];
1654
1655         /*
1656          * This is not the sha1 we are looking for, or
1657          * unreusable because it is not a regular file.
1658          */
1659         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1660                 return 0;
1661
1662         /*
1663          * If ce matches the file in the work tree, we can reuse it.
1664          */
1665         if (ce_uptodate(ce) ||
1666             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1667                 return 1;
1668
1669         return 0;
1670 }
1671
1672 static int populate_from_stdin(struct diff_filespec *s)
1673 {
1674         struct strbuf buf = STRBUF_INIT;
1675         size_t size = 0;
1676
1677         if (strbuf_read(&buf, 0, 0) < 0)
1678                 return error("error while reading from stdin %s",
1679                                      strerror(errno));
1680
1681         s->should_munmap = 0;
1682         s->data = strbuf_detach(&buf, &size);
1683         s->size = size;
1684         s->should_free = 1;
1685         return 0;
1686 }
1687
1688 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1689 {
1690         int len;
1691         char *data = xmalloc(100);
1692         len = snprintf(data, 100,
1693                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1694         s->data = data;
1695         s->size = len;
1696         s->should_free = 1;
1697         if (size_only) {
1698                 s->data = NULL;
1699                 free(data);
1700         }
1701         return 0;
1702 }
1703
1704 /*
1705  * While doing rename detection and pickaxe operation, we may need to
1706  * grab the data for the blob (or file) for our own in-core comparison.
1707  * diff_filespec has data and size fields for this purpose.
1708  */
1709 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1710 {
1711         int err = 0;
1712         if (!DIFF_FILE_VALID(s))
1713                 die("internal error: asking to populate invalid file.");
1714         if (S_ISDIR(s->mode))
1715                 return -1;
1716
1717         if (s->data)
1718                 return 0;
1719
1720         if (size_only && 0 < s->size)
1721                 return 0;
1722
1723         if (S_ISGITLINK(s->mode))
1724                 return diff_populate_gitlink(s, size_only);
1725
1726         if (!s->sha1_valid ||
1727             reuse_worktree_file(s->path, s->sha1, 0)) {
1728                 struct strbuf buf = STRBUF_INIT;
1729                 struct stat st;
1730                 int fd;
1731
1732                 if (!strcmp(s->path, "-"))
1733                         return populate_from_stdin(s);
1734
1735                 if (lstat(s->path, &st) < 0) {
1736                         if (errno == ENOENT) {
1737                         err_empty:
1738                                 err = -1;
1739                         empty:
1740                                 s->data = (char *)"";
1741                                 s->size = 0;
1742                                 return err;
1743                         }
1744                 }
1745                 s->size = xsize_t(st.st_size);
1746                 if (!s->size)
1747                         goto empty;
1748                 if (size_only)
1749                         return 0;
1750                 if (S_ISLNK(st.st_mode)) {
1751                         int ret;
1752                         s->data = xmalloc(s->size);
1753                         s->should_free = 1;
1754                         ret = readlink(s->path, s->data, s->size);
1755                         if (ret < 0) {
1756                                 free(s->data);
1757                                 goto err_empty;
1758                         }
1759                         return 0;
1760                 }
1761                 fd = open(s->path, O_RDONLY);
1762                 if (fd < 0)
1763                         goto err_empty;
1764                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1765                 close(fd);
1766                 s->should_munmap = 1;
1767
1768                 /*
1769                  * Convert from working tree format to canonical git format
1770                  */
1771                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1772                         size_t size = 0;
1773                         munmap(s->data, s->size);
1774                         s->should_munmap = 0;
1775                         s->data = strbuf_detach(&buf, &size);
1776                         s->size = size;
1777                         s->should_free = 1;
1778                 }
1779         }
1780         else {
1781                 enum object_type type;
1782                 if (size_only)
1783                         type = sha1_object_info(s->sha1, &s->size);
1784                 else {
1785                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1786                         s->should_free = 1;
1787                 }
1788         }
1789         return 0;
1790 }
1791
1792 void diff_free_filespec_blob(struct diff_filespec *s)
1793 {
1794         if (s->should_free)
1795                 free(s->data);
1796         else if (s->should_munmap)
1797                 munmap(s->data, s->size);
1798
1799         if (s->should_free || s->should_munmap) {
1800                 s->should_free = s->should_munmap = 0;
1801                 s->data = NULL;
1802         }
1803 }
1804
1805 void diff_free_filespec_data(struct diff_filespec *s)
1806 {
1807         diff_free_filespec_blob(s);
1808         free(s->cnt_data);
1809         s->cnt_data = NULL;
1810 }
1811
1812 static void prep_temp_blob(struct diff_tempfile *temp,
1813                            void *blob,
1814                            unsigned long size,
1815                            const unsigned char *sha1,
1816                            int mode)
1817 {
1818         int fd;
1819
1820         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1821         if (fd < 0)
1822                 die("unable to create temp-file: %s", strerror(errno));
1823         if (write_in_full(fd, blob, size) != size)
1824                 die("unable to write temp-file");
1825         close(fd);
1826         temp->name = temp->tmp_path;
1827         strcpy(temp->hex, sha1_to_hex(sha1));
1828         temp->hex[40] = 0;
1829         sprintf(temp->mode, "%06o", mode);
1830 }
1831
1832 static void prepare_temp_file(const char *name,
1833                               struct diff_tempfile *temp,
1834                               struct diff_filespec *one)
1835 {
1836         if (!DIFF_FILE_VALID(one)) {
1837         not_a_valid_file:
1838                 /* A '-' entry produces this for file-2, and
1839                  * a '+' entry produces this for file-1.
1840                  */
1841                 temp->name = "/dev/null";
1842                 strcpy(temp->hex, ".");
1843                 strcpy(temp->mode, ".");
1844                 return;
1845         }
1846
1847         if (!one->sha1_valid ||
1848             reuse_worktree_file(name, one->sha1, 1)) {
1849                 struct stat st;
1850                 if (lstat(name, &st) < 0) {
1851                         if (errno == ENOENT)
1852                                 goto not_a_valid_file;
1853                         die("stat(%s): %s", name, strerror(errno));
1854                 }
1855                 if (S_ISLNK(st.st_mode)) {
1856                         int ret;
1857                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1858                         size_t sz = xsize_t(st.st_size);
1859                         if (sizeof(buf) <= st.st_size)
1860                                 die("symlink too long: %s", name);
1861                         ret = readlink(name, buf, sz);
1862                         if (ret < 0)
1863                                 die("readlink(%s)", name);
1864                         prep_temp_blob(temp, buf, sz,
1865                                        (one->sha1_valid ?
1866                                         one->sha1 : null_sha1),
1867                                        (one->sha1_valid ?
1868                                         one->mode : S_IFLNK));
1869                 }
1870                 else {
1871                         /* we can borrow from the file in the work tree */
1872                         temp->name = name;
1873                         if (!one->sha1_valid)
1874                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1875                         else
1876                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1877                         /* Even though we may sometimes borrow the
1878                          * contents from the work tree, we always want
1879                          * one->mode.  mode is trustworthy even when
1880                          * !(one->sha1_valid), as long as
1881                          * DIFF_FILE_VALID(one).
1882                          */
1883                         sprintf(temp->mode, "%06o", one->mode);
1884                 }
1885                 return;
1886         }
1887         else {
1888                 if (diff_populate_filespec(one, 0))
1889                         die("cannot read data blob for %s", one->path);
1890                 prep_temp_blob(temp, one->data, one->size,
1891                                one->sha1, one->mode);
1892         }
1893 }
1894
1895 static void remove_tempfile(void)
1896 {
1897         int i;
1898
1899         for (i = 0; i < 2; i++)
1900                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1901                         unlink(diff_temp[i].name);
1902                         diff_temp[i].name = NULL;
1903                 }
1904 }
1905
1906 static void remove_tempfile_on_signal(int signo)
1907 {
1908         remove_tempfile();
1909         signal(SIGINT, SIG_DFL);
1910         raise(signo);
1911 }
1912
1913 /* An external diff command takes:
1914  *
1915  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1916  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1917  *
1918  */
1919 static void run_external_diff(const char *pgm,
1920                               const char *name,
1921                               const char *other,
1922                               struct diff_filespec *one,
1923                               struct diff_filespec *two,
1924                               const char *xfrm_msg,
1925                               int complete_rewrite)
1926 {
1927         const char *spawn_arg[10];
1928         struct diff_tempfile *temp = diff_temp;
1929         int retval;
1930         static int atexit_asked = 0;
1931         const char *othername;
1932         const char **arg = &spawn_arg[0];
1933
1934         othername = (other? other : name);
1935         if (one && two) {
1936                 prepare_temp_file(name, &temp[0], one);
1937                 prepare_temp_file(othername, &temp[1], two);
1938                 if (! atexit_asked &&
1939                     (temp[0].name == temp[0].tmp_path ||
1940                      temp[1].name == temp[1].tmp_path)) {
1941                         atexit_asked = 1;
1942                         atexit(remove_tempfile);
1943                 }
1944                 signal(SIGINT, remove_tempfile_on_signal);
1945         }
1946
1947         if (one && two) {
1948                 *arg++ = pgm;
1949                 *arg++ = name;
1950                 *arg++ = temp[0].name;
1951                 *arg++ = temp[0].hex;
1952                 *arg++ = temp[0].mode;
1953                 *arg++ = temp[1].name;
1954                 *arg++ = temp[1].hex;
1955                 *arg++ = temp[1].mode;
1956                 if (other) {
1957                         *arg++ = other;
1958                         *arg++ = xfrm_msg;
1959                 }
1960         } else {
1961                 *arg++ = pgm;
1962                 *arg++ = name;
1963         }
1964         *arg = NULL;
1965         fflush(NULL);
1966         retval = run_command_v_opt(spawn_arg, 0);
1967         remove_tempfile();
1968         if (retval) {
1969                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1970                 exit(1);
1971         }
1972 }
1973
1974 static void run_diff_cmd(const char *pgm,
1975                          const char *name,
1976                          const char *other,
1977                          const char *attr_path,
1978                          struct diff_filespec *one,
1979                          struct diff_filespec *two,
1980                          const char *xfrm_msg,
1981                          struct diff_options *o,
1982                          int complete_rewrite)
1983 {
1984         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
1985                 pgm = NULL;
1986         else {
1987                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
1988                 if (drv && drv->external)
1989                         pgm = drv->external;
1990         }
1991
1992         if (pgm) {
1993                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1994                                   complete_rewrite);
1995                 return;
1996         }
1997         if (one && two)
1998                 builtin_diff(name, other ? other : name,
1999                              one, two, xfrm_msg, o, complete_rewrite);
2000         else
2001                 fprintf(o->file, "* Unmerged path %s\n", name);
2002 }
2003
2004 static void diff_fill_sha1_info(struct diff_filespec *one)
2005 {
2006         if (DIFF_FILE_VALID(one)) {
2007                 if (!one->sha1_valid) {
2008                         struct stat st;
2009                         if (!strcmp(one->path, "-")) {
2010                                 hashcpy(one->sha1, null_sha1);
2011                                 return;
2012                         }
2013                         if (lstat(one->path, &st) < 0)
2014                                 die("stat %s", one->path);
2015                         if (index_path(one->sha1, one->path, &st, 0))
2016                                 die("cannot hash %s\n", one->path);
2017                 }
2018         }
2019         else
2020                 hashclr(one->sha1);
2021 }
2022
2023 static int similarity_index(struct diff_filepair *p)
2024 {
2025         return p->score * 100 / MAX_SCORE;
2026 }
2027
2028 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2029 {
2030         /* Strip the prefix but do not molest /dev/null and absolute paths */
2031         if (*namep && **namep != '/')
2032                 *namep += prefix_length;
2033         if (*otherp && **otherp != '/')
2034                 *otherp += prefix_length;
2035 }
2036
2037 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2038 {
2039         const char *pgm = external_diff();
2040         struct strbuf msg;
2041         char *xfrm_msg;
2042         struct diff_filespec *one = p->one;
2043         struct diff_filespec *two = p->two;
2044         const char *name;
2045         const char *other;
2046         const char *attr_path;
2047         int complete_rewrite = 0;
2048
2049         name  = p->one->path;
2050         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2051         attr_path = name;
2052         if (o->prefix_length)
2053                 strip_prefix(o->prefix_length, &name, &other);
2054
2055         if (DIFF_PAIR_UNMERGED(p)) {
2056                 run_diff_cmd(pgm, name, NULL, attr_path,
2057                              NULL, NULL, NULL, o, 0);
2058                 return;
2059         }
2060
2061         diff_fill_sha1_info(one);
2062         diff_fill_sha1_info(two);
2063
2064         strbuf_init(&msg, PATH_MAX * 2 + 300);
2065         switch (p->status) {
2066         case DIFF_STATUS_COPIED:
2067                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2068                 strbuf_addstr(&msg, "\ncopy from ");
2069                 quote_c_style(name, &msg, NULL, 0);
2070                 strbuf_addstr(&msg, "\ncopy to ");
2071                 quote_c_style(other, &msg, NULL, 0);
2072                 strbuf_addch(&msg, '\n');
2073                 break;
2074         case DIFF_STATUS_RENAMED:
2075                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2076                 strbuf_addstr(&msg, "\nrename from ");
2077                 quote_c_style(name, &msg, NULL, 0);
2078                 strbuf_addstr(&msg, "\nrename to ");
2079                 quote_c_style(other, &msg, NULL, 0);
2080                 strbuf_addch(&msg, '\n');
2081                 break;
2082         case DIFF_STATUS_MODIFIED:
2083                 if (p->score) {
2084                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2085                                         similarity_index(p));
2086                         complete_rewrite = 1;
2087                         break;
2088                 }
2089                 /* fallthru */
2090         default:
2091                 /* nothing */
2092                 ;
2093         }
2094
2095         if (hashcmp(one->sha1, two->sha1)) {
2096                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2097
2098                 if (DIFF_OPT_TST(o, BINARY)) {
2099                         mmfile_t mf;
2100                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2101                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2102                                 abbrev = 40;
2103                 }
2104                 strbuf_addf(&msg, "index %.*s..%.*s",
2105                                 abbrev, sha1_to_hex(one->sha1),
2106                                 abbrev, sha1_to_hex(two->sha1));
2107                 if (one->mode == two->mode)
2108                         strbuf_addf(&msg, " %06o", one->mode);
2109                 strbuf_addch(&msg, '\n');
2110         }
2111
2112         if (msg.len)
2113                 strbuf_setlen(&msg, msg.len - 1);
2114         xfrm_msg = msg.len ? msg.buf : NULL;
2115
2116         if (!pgm &&
2117             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2118             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2119                 /* a filepair that changes between file and symlink
2120                  * needs to be split into deletion and creation.
2121                  */
2122                 struct diff_filespec *null = alloc_filespec(two->path);
2123                 run_diff_cmd(NULL, name, other, attr_path,
2124                              one, null, xfrm_msg, o, 0);
2125                 free(null);
2126                 null = alloc_filespec(one->path);
2127                 run_diff_cmd(NULL, name, other, attr_path,
2128                              null, two, xfrm_msg, o, 0);
2129                 free(null);
2130         }
2131         else
2132                 run_diff_cmd(pgm, name, other, attr_path,
2133                              one, two, xfrm_msg, o, complete_rewrite);
2134
2135         strbuf_release(&msg);
2136 }
2137
2138 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2139                          struct diffstat_t *diffstat)
2140 {
2141         const char *name;
2142         const char *other;
2143         int complete_rewrite = 0;
2144
2145         if (DIFF_PAIR_UNMERGED(p)) {
2146                 /* unmerged */
2147                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2148                 return;
2149         }
2150
2151         name = p->one->path;
2152         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2153
2154         if (o->prefix_length)
2155                 strip_prefix(o->prefix_length, &name, &other);
2156
2157         diff_fill_sha1_info(p->one);
2158         diff_fill_sha1_info(p->two);
2159
2160         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2161                 complete_rewrite = 1;
2162         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2163 }
2164
2165 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2166 {
2167         const char *name;
2168         const char *other;
2169         const char *attr_path;
2170
2171         if (DIFF_PAIR_UNMERGED(p)) {
2172                 /* unmerged */
2173                 return;
2174         }
2175
2176         name = p->one->path;
2177         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2178         attr_path = other ? other : name;
2179
2180         if (o->prefix_length)
2181                 strip_prefix(o->prefix_length, &name, &other);
2182
2183         diff_fill_sha1_info(p->one);
2184         diff_fill_sha1_info(p->two);
2185
2186         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2187 }
2188
2189 void diff_setup(struct diff_options *options)
2190 {
2191         memset(options, 0, sizeof(*options));
2192
2193         options->file = stdout;
2194
2195         options->line_termination = '\n';
2196         options->break_opt = -1;
2197         options->rename_limit = -1;
2198         options->dirstat_percent = 3;
2199         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2200         options->context = 3;
2201
2202         options->change = diff_change;
2203         options->add_remove = diff_addremove;
2204         if (diff_use_color_default > 0)
2205                 DIFF_OPT_SET(options, COLOR_DIFF);
2206         else
2207                 DIFF_OPT_CLR(options, COLOR_DIFF);
2208         options->detect_rename = diff_detect_rename_default;
2209
2210         if (!diff_mnemonic_prefix) {
2211                 options->a_prefix = "a/";
2212                 options->b_prefix = "b/";
2213         }
2214 }
2215
2216 int diff_setup_done(struct diff_options *options)
2217 {
2218         int count = 0;
2219
2220         if (options->output_format & DIFF_FORMAT_NAME)
2221                 count++;
2222         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2223                 count++;
2224         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2225                 count++;
2226         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2227                 count++;
2228         if (count > 1)
2229                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2230
2231         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2232                 options->detect_rename = DIFF_DETECT_COPY;
2233
2234         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2235                 options->prefix = NULL;
2236         if (options->prefix)
2237                 options->prefix_length = strlen(options->prefix);
2238         else
2239                 options->prefix_length = 0;
2240
2241         if (options->output_format & (DIFF_FORMAT_NAME |
2242                                       DIFF_FORMAT_NAME_STATUS |
2243                                       DIFF_FORMAT_CHECKDIFF |
2244                                       DIFF_FORMAT_NO_OUTPUT))
2245                 options->output_format &= ~(DIFF_FORMAT_RAW |
2246                                             DIFF_FORMAT_NUMSTAT |
2247                                             DIFF_FORMAT_DIFFSTAT |
2248                                             DIFF_FORMAT_SHORTSTAT |
2249                                             DIFF_FORMAT_DIRSTAT |
2250                                             DIFF_FORMAT_SUMMARY |
2251                                             DIFF_FORMAT_PATCH);
2252
2253         /*
2254          * These cases always need recursive; we do not drop caller-supplied
2255          * recursive bits for other formats here.
2256          */
2257         if (options->output_format & (DIFF_FORMAT_PATCH |
2258                                       DIFF_FORMAT_NUMSTAT |
2259                                       DIFF_FORMAT_DIFFSTAT |
2260                                       DIFF_FORMAT_SHORTSTAT |
2261                                       DIFF_FORMAT_DIRSTAT |
2262                                       DIFF_FORMAT_SUMMARY |
2263                                       DIFF_FORMAT_CHECKDIFF))
2264                 DIFF_OPT_SET(options, RECURSIVE);
2265         /*
2266          * Also pickaxe would not work very well if you do not say recursive
2267          */
2268         if (options->pickaxe)
2269                 DIFF_OPT_SET(options, RECURSIVE);
2270
2271         if (options->detect_rename && options->rename_limit < 0)
2272                 options->rename_limit = diff_rename_limit_default;
2273         if (options->setup & DIFF_SETUP_USE_CACHE) {
2274                 if (!active_cache)
2275                         /* read-cache does not die even when it fails
2276                          * so it is safe for us to do this here.  Also
2277                          * it does not smudge active_cache or active_nr
2278                          * when it fails, so we do not have to worry about
2279                          * cleaning it up ourselves either.
2280                          */
2281                         read_cache();
2282         }
2283         if (options->abbrev <= 0 || 40 < options->abbrev)
2284                 options->abbrev = 40; /* full */
2285
2286         /*
2287          * It does not make sense to show the first hit we happened
2288          * to have found.  It does not make sense not to return with
2289          * exit code in such a case either.
2290          */
2291         if (DIFF_OPT_TST(options, QUIET)) {
2292                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2293                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2294         }
2295
2296         return 0;
2297 }
2298
2299 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2300 {
2301         char c, *eq;
2302         int len;
2303
2304         if (*arg != '-')
2305                 return 0;
2306         c = *++arg;
2307         if (!c)
2308                 return 0;
2309         if (c == arg_short) {
2310                 c = *++arg;
2311                 if (!c)
2312                         return 1;
2313                 if (val && isdigit(c)) {
2314                         char *end;
2315                         int n = strtoul(arg, &end, 10);
2316                         if (*end)
2317                                 return 0;
2318                         *val = n;
2319                         return 1;
2320                 }
2321                 return 0;
2322         }
2323         if (c != '-')
2324                 return 0;
2325         arg++;
2326         eq = strchr(arg, '=');
2327         if (eq)
2328                 len = eq - arg;
2329         else
2330                 len = strlen(arg);
2331         if (!len || strncmp(arg, arg_long, len))
2332                 return 0;
2333         if (eq) {
2334                 int n;
2335                 char *end;
2336                 if (!isdigit(*++eq))
2337                         return 0;
2338                 n = strtoul(eq, &end, 10);
2339                 if (*end)
2340                         return 0;
2341                 *val = n;
2342         }
2343         return 1;
2344 }
2345
2346 static int diff_scoreopt_parse(const char *opt);
2347
2348 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2349 {
2350         const char *arg = av[0];
2351
2352         /* Output format options */
2353         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2354                 options->output_format |= DIFF_FORMAT_PATCH;
2355         else if (opt_arg(arg, 'U', "unified", &options->context))
2356                 options->output_format |= DIFF_FORMAT_PATCH;
2357         else if (!strcmp(arg, "--raw"))
2358                 options->output_format |= DIFF_FORMAT_RAW;
2359         else if (!strcmp(arg, "--patch-with-raw"))
2360                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2361         else if (!strcmp(arg, "--numstat"))
2362                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2363         else if (!strcmp(arg, "--shortstat"))
2364                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2365         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2366                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2367         else if (!strcmp(arg, "--cumulative")) {
2368                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2369                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2370         } else if (opt_arg(arg, 0, "dirstat-by-file",
2371                            &options->dirstat_percent)) {
2372                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2373                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2374         }
2375         else if (!strcmp(arg, "--check"))
2376                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2377         else if (!strcmp(arg, "--summary"))
2378                 options->output_format |= DIFF_FORMAT_SUMMARY;
2379         else if (!strcmp(arg, "--patch-with-stat"))
2380                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2381         else if (!strcmp(arg, "--name-only"))
2382                 options->output_format |= DIFF_FORMAT_NAME;
2383         else if (!strcmp(arg, "--name-status"))
2384                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2385         else if (!strcmp(arg, "-s"))
2386                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2387         else if (!prefixcmp(arg, "--stat")) {
2388                 char *end;
2389                 int width = options->stat_width;
2390                 int name_width = options->stat_name_width;
2391                 arg += 6;
2392                 end = (char *)arg;
2393
2394                 switch (*arg) {
2395                 case '-':
2396                         if (!prefixcmp(arg, "-width="))
2397                                 width = strtoul(arg + 7, &end, 10);
2398                         else if (!prefixcmp(arg, "-name-width="))
2399                                 name_width = strtoul(arg + 12, &end, 10);
2400                         break;
2401                 case '=':
2402                         width = strtoul(arg+1, &end, 10);
2403                         if (*end == ',')
2404                                 name_width = strtoul(end+1, &end, 10);
2405                 }
2406
2407                 /* Important! This checks all the error cases! */
2408                 if (*end)
2409                         return 0;
2410                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2411                 options->stat_name_width = name_width;
2412                 options->stat_width = width;
2413         }
2414
2415         /* renames options */
2416         else if (!prefixcmp(arg, "-B")) {
2417                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2418                         return -1;
2419         }
2420         else if (!prefixcmp(arg, "-M")) {
2421                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2422                         return -1;
2423                 options->detect_rename = DIFF_DETECT_RENAME;
2424         }
2425         else if (!prefixcmp(arg, "-C")) {
2426                 if (options->detect_rename == DIFF_DETECT_COPY)
2427                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2428                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2429                         return -1;
2430                 options->detect_rename = DIFF_DETECT_COPY;
2431         }
2432         else if (!strcmp(arg, "--no-renames"))
2433                 options->detect_rename = 0;
2434         else if (!strcmp(arg, "--relative"))
2435                 DIFF_OPT_SET(options, RELATIVE_NAME);
2436         else if (!prefixcmp(arg, "--relative=")) {
2437                 DIFF_OPT_SET(options, RELATIVE_NAME);
2438                 options->prefix = arg + 11;
2439         }
2440
2441         /* xdiff options */
2442         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2443                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2444         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2445                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2446         else if (!strcmp(arg, "--ignore-space-at-eol"))
2447                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2448
2449         /* flags options */
2450         else if (!strcmp(arg, "--binary")) {
2451                 options->output_format |= DIFF_FORMAT_PATCH;
2452                 DIFF_OPT_SET(options, BINARY);
2453         }
2454         else if (!strcmp(arg, "--full-index"))
2455                 DIFF_OPT_SET(options, FULL_INDEX);
2456         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2457                 DIFF_OPT_SET(options, TEXT);
2458         else if (!strcmp(arg, "-R"))
2459                 DIFF_OPT_SET(options, REVERSE_DIFF);
2460         else if (!strcmp(arg, "--find-copies-harder"))
2461                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2462         else if (!strcmp(arg, "--follow"))
2463                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2464         else if (!strcmp(arg, "--color"))
2465                 DIFF_OPT_SET(options, COLOR_DIFF);
2466         else if (!strcmp(arg, "--no-color"))
2467                 DIFF_OPT_CLR(options, COLOR_DIFF);
2468         else if (!strcmp(arg, "--color-words"))
2469                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2470         else if (!strcmp(arg, "--exit-code"))
2471                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2472         else if (!strcmp(arg, "--quiet"))
2473                 DIFF_OPT_SET(options, QUIET);
2474         else if (!strcmp(arg, "--ext-diff"))
2475                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2476         else if (!strcmp(arg, "--no-ext-diff"))
2477                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2478         else if (!strcmp(arg, "--ignore-submodules"))
2479                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2480
2481         /* misc options */
2482         else if (!strcmp(arg, "-z"))
2483                 options->line_termination = 0;
2484         else if (!prefixcmp(arg, "-l"))
2485                 options->rename_limit = strtoul(arg+2, NULL, 10);
2486         else if (!prefixcmp(arg, "-S"))
2487                 options->pickaxe = arg + 2;
2488         else if (!strcmp(arg, "--pickaxe-all"))
2489                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2490         else if (!strcmp(arg, "--pickaxe-regex"))
2491                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2492         else if (!prefixcmp(arg, "-O"))
2493                 options->orderfile = arg + 2;
2494         else if (!prefixcmp(arg, "--diff-filter="))
2495                 options->filter = arg + 14;
2496         else if (!strcmp(arg, "--abbrev"))
2497                 options->abbrev = DEFAULT_ABBREV;
2498         else if (!prefixcmp(arg, "--abbrev=")) {
2499                 options->abbrev = strtoul(arg + 9, NULL, 10);
2500                 if (options->abbrev < MINIMUM_ABBREV)
2501                         options->abbrev = MINIMUM_ABBREV;
2502                 else if (40 < options->abbrev)
2503                         options->abbrev = 40;
2504         }
2505         else if (!prefixcmp(arg, "--src-prefix="))
2506                 options->a_prefix = arg + 13;
2507         else if (!prefixcmp(arg, "--dst-prefix="))
2508                 options->b_prefix = arg + 13;
2509         else if (!strcmp(arg, "--no-prefix"))
2510                 options->a_prefix = options->b_prefix = "";
2511         else if (!prefixcmp(arg, "--output=")) {
2512                 options->file = fopen(arg + strlen("--output="), "w");
2513                 options->close_file = 1;
2514         } else
2515                 return 0;
2516         return 1;
2517 }
2518
2519 static int parse_num(const char **cp_p)
2520 {
2521         unsigned long num, scale;
2522         int ch, dot;
2523         const char *cp = *cp_p;
2524
2525         num = 0;
2526         scale = 1;
2527         dot = 0;
2528         for(;;) {
2529                 ch = *cp;
2530                 if ( !dot && ch == '.' ) {
2531                         scale = 1;
2532                         dot = 1;
2533                 } else if ( ch == '%' ) {
2534                         scale = dot ? scale*100 : 100;
2535                         cp++;   /* % is always at the end */
2536                         break;
2537                 } else if ( ch >= '0' && ch <= '9' ) {
2538                         if ( scale < 100000 ) {
2539                                 scale *= 10;
2540                                 num = (num*10) + (ch-'0');
2541                         }
2542                 } else {
2543                         break;
2544                 }
2545                 cp++;
2546         }
2547         *cp_p = cp;
2548
2549         /* user says num divided by scale and we say internally that
2550          * is MAX_SCORE * num / scale.
2551          */
2552         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2553 }
2554
2555 static int diff_scoreopt_parse(const char *opt)
2556 {
2557         int opt1, opt2, cmd;
2558
2559         if (*opt++ != '-')
2560                 return -1;
2561         cmd = *opt++;
2562         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2563                 return -1; /* that is not a -M, -C nor -B option */
2564
2565         opt1 = parse_num(&opt);
2566         if (cmd != 'B')
2567                 opt2 = 0;
2568         else {
2569                 if (*opt == 0)
2570                         opt2 = 0;
2571                 else if (*opt != '/')
2572                         return -1; /* we expect -B80/99 or -B80 */
2573                 else {
2574                         opt++;
2575                         opt2 = parse_num(&opt);
2576                 }
2577         }
2578         if (*opt != 0)
2579                 return -1;
2580         return opt1 | (opt2 << 16);
2581 }
2582
2583 struct diff_queue_struct diff_queued_diff;
2584
2585 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2586 {
2587         if (queue->alloc <= queue->nr) {
2588                 queue->alloc = alloc_nr(queue->alloc);
2589                 queue->queue = xrealloc(queue->queue,
2590                                         sizeof(dp) * queue->alloc);
2591         }
2592         queue->queue[queue->nr++] = dp;
2593 }
2594
2595 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2596                                  struct diff_filespec *one,
2597                                  struct diff_filespec *two)
2598 {
2599         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2600         dp->one = one;
2601         dp->two = two;
2602         if (queue)
2603                 diff_q(queue, dp);
2604         return dp;
2605 }
2606
2607 void diff_free_filepair(struct diff_filepair *p)
2608 {
2609         free_filespec(p->one);
2610         free_filespec(p->two);
2611         free(p);
2612 }
2613
2614 /* This is different from find_unique_abbrev() in that
2615  * it stuffs the result with dots for alignment.
2616  */
2617 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2618 {
2619         int abblen;
2620         const char *abbrev;
2621         if (len == 40)
2622                 return sha1_to_hex(sha1);
2623
2624         abbrev = find_unique_abbrev(sha1, len);
2625         abblen = strlen(abbrev);
2626         if (abblen < 37) {
2627                 static char hex[41];
2628                 if (len < abblen && abblen <= len + 2)
2629                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2630                 else
2631                         sprintf(hex, "%s...", abbrev);
2632                 return hex;
2633         }
2634         return sha1_to_hex(sha1);
2635 }
2636
2637 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2638 {
2639         int line_termination = opt->line_termination;
2640         int inter_name_termination = line_termination ? '\t' : '\0';
2641
2642         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2643                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2644                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2645                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2646         }
2647         if (p->score) {
2648                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2649                         inter_name_termination);
2650         } else {
2651                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2652         }
2653
2654         if (p->status == DIFF_STATUS_COPIED ||
2655             p->status == DIFF_STATUS_RENAMED) {
2656                 const char *name_a, *name_b;
2657                 name_a = p->one->path;
2658                 name_b = p->two->path;
2659                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2660                 write_name_quoted(name_a, opt->file, inter_name_termination);
2661                 write_name_quoted(name_b, opt->file, line_termination);
2662         } else {
2663                 const char *name_a, *name_b;
2664                 name_a = p->one->mode ? p->one->path : p->two->path;
2665                 name_b = NULL;
2666                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2667                 write_name_quoted(name_a, opt->file, line_termination);
2668         }
2669 }
2670
2671 int diff_unmodified_pair(struct diff_filepair *p)
2672 {
2673         /* This function is written stricter than necessary to support
2674          * the currently implemented transformers, but the idea is to
2675          * let transformers to produce diff_filepairs any way they want,
2676          * and filter and clean them up here before producing the output.
2677          */
2678         struct diff_filespec *one = p->one, *two = p->two;
2679
2680         if (DIFF_PAIR_UNMERGED(p))
2681                 return 0; /* unmerged is interesting */
2682
2683         /* deletion, addition, mode or type change
2684          * and rename are all interesting.
2685          */
2686         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2687             DIFF_PAIR_MODE_CHANGED(p) ||
2688             strcmp(one->path, two->path))
2689                 return 0;
2690
2691         /* both are valid and point at the same path.  that is, we are
2692          * dealing with a change.
2693          */
2694         if (one->sha1_valid && two->sha1_valid &&
2695             !hashcmp(one->sha1, two->sha1))
2696                 return 1; /* no change */
2697         if (!one->sha1_valid && !two->sha1_valid)
2698                 return 1; /* both look at the same file on the filesystem. */
2699         return 0;
2700 }
2701
2702 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2703 {
2704         if (diff_unmodified_pair(p))
2705                 return;
2706
2707         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2708             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2709                 return; /* no tree diffs in patch format */
2710
2711         run_diff(p, o);
2712 }
2713
2714 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2715                             struct diffstat_t *diffstat)
2716 {
2717         if (diff_unmodified_pair(p))
2718                 return;
2719
2720         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2721             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2722                 return; /* no tree diffs in patch format */
2723
2724         run_diffstat(p, o, diffstat);
2725 }
2726
2727 static void diff_flush_checkdiff(struct diff_filepair *p,
2728                 struct diff_options *o)
2729 {
2730         if (diff_unmodified_pair(p))
2731                 return;
2732
2733         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2734             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2735                 return; /* no tree diffs in patch format */
2736
2737         run_checkdiff(p, o);
2738 }
2739
2740 int diff_queue_is_empty(void)
2741 {
2742         struct diff_queue_struct *q = &diff_queued_diff;
2743         int i;
2744         for (i = 0; i < q->nr; i++)
2745                 if (!diff_unmodified_pair(q->queue[i]))
2746                         return 0;
2747         return 1;
2748 }
2749
2750 #if DIFF_DEBUG
2751 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2752 {
2753         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2754                 x, one ? one : "",
2755                 s->path,
2756                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2757                 s->mode,
2758                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2759         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2760                 x, one ? one : "",
2761                 s->size, s->xfrm_flags);
2762 }
2763
2764 void diff_debug_filepair(const struct diff_filepair *p, int i)
2765 {
2766         diff_debug_filespec(p->one, i, "one");
2767         diff_debug_filespec(p->two, i, "two");
2768         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2769                 p->score, p->status ? p->status : '?',
2770                 p->one->rename_used, p->broken_pair);
2771 }
2772
2773 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2774 {
2775         int i;
2776         if (msg)
2777                 fprintf(stderr, "%s\n", msg);
2778         fprintf(stderr, "q->nr = %d\n", q->nr);
2779         for (i = 0; i < q->nr; i++) {
2780                 struct diff_filepair *p = q->queue[i];
2781                 diff_debug_filepair(p, i);
2782         }
2783 }
2784 #endif
2785
2786 static void diff_resolve_rename_copy(void)
2787 {
2788         int i;
2789         struct diff_filepair *p;
2790         struct diff_queue_struct *q = &diff_queued_diff;
2791
2792         diff_debug_queue("resolve-rename-copy", q);
2793
2794         for (i = 0; i < q->nr; i++) {
2795                 p = q->queue[i];
2796                 p->status = 0; /* undecided */
2797                 if (DIFF_PAIR_UNMERGED(p))
2798                         p->status = DIFF_STATUS_UNMERGED;
2799                 else if (!DIFF_FILE_VALID(p->one))
2800                         p->status = DIFF_STATUS_ADDED;
2801                 else if (!DIFF_FILE_VALID(p->two))
2802                         p->status = DIFF_STATUS_DELETED;
2803                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2804                         p->status = DIFF_STATUS_TYPE_CHANGED;
2805
2806                 /* from this point on, we are dealing with a pair
2807                  * whose both sides are valid and of the same type, i.e.
2808                  * either in-place edit or rename/copy edit.
2809                  */
2810                 else if (DIFF_PAIR_RENAME(p)) {
2811                         /*
2812                          * A rename might have re-connected a broken
2813                          * pair up, causing the pathnames to be the
2814                          * same again. If so, that's not a rename at
2815                          * all, just a modification..
2816                          *
2817                          * Otherwise, see if this source was used for
2818                          * multiple renames, in which case we decrement
2819                          * the count, and call it a copy.
2820                          */
2821                         if (!strcmp(p->one->path, p->two->path))
2822                                 p->status = DIFF_STATUS_MODIFIED;
2823                         else if (--p->one->rename_used > 0)
2824                                 p->status = DIFF_STATUS_COPIED;
2825                         else
2826                                 p->status = DIFF_STATUS_RENAMED;
2827                 }
2828                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2829                          p->one->mode != p->two->mode ||
2830                          is_null_sha1(p->one->sha1))
2831                         p->status = DIFF_STATUS_MODIFIED;
2832                 else {
2833                         /* This is a "no-change" entry and should not
2834                          * happen anymore, but prepare for broken callers.
2835                          */
2836                         error("feeding unmodified %s to diffcore",
2837                               p->one->path);
2838                         p->status = DIFF_STATUS_UNKNOWN;
2839                 }
2840         }
2841         diff_debug_queue("resolve-rename-copy done", q);
2842 }
2843
2844 static int check_pair_status(struct diff_filepair *p)
2845 {
2846         switch (p->status) {
2847         case DIFF_STATUS_UNKNOWN:
2848                 return 0;
2849         case 0:
2850                 die("internal error in diff-resolve-rename-copy");
2851         default:
2852                 return 1;
2853         }
2854 }
2855
2856 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2857 {
2858         int fmt = opt->output_format;
2859
2860         if (fmt & DIFF_FORMAT_CHECKDIFF)
2861                 diff_flush_checkdiff(p, opt);
2862         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2863                 diff_flush_raw(p, opt);
2864         else if (fmt & DIFF_FORMAT_NAME) {
2865                 const char *name_a, *name_b;
2866                 name_a = p->two->path;
2867                 name_b = NULL;
2868                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2869                 write_name_quoted(name_a, opt->file, opt->line_termination);
2870         }
2871 }
2872
2873 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2874 {
2875         if (fs->mode)
2876                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2877         else
2878                 fprintf(file, " %s ", newdelete);
2879         write_name_quoted(fs->path, file, '\n');
2880 }
2881
2882
2883 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2884 {
2885         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2886                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2887                         show_name ? ' ' : '\n');
2888                 if (show_name) {
2889                         write_name_quoted(p->two->path, file, '\n');
2890                 }
2891         }
2892 }
2893
2894 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2895 {
2896         char *names = pprint_rename(p->one->path, p->two->path);
2897
2898         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2899         free(names);
2900         show_mode_change(file, p, 0);
2901 }
2902
2903 static void diff_summary(FILE *file, struct diff_filepair *p)
2904 {
2905         switch(p->status) {
2906         case DIFF_STATUS_DELETED:
2907                 show_file_mode_name(file, "delete", p->one);
2908                 break;
2909         case DIFF_STATUS_ADDED:
2910                 show_file_mode_name(file, "create", p->two);
2911                 break;
2912         case DIFF_STATUS_COPIED:
2913                 show_rename_copy(file, "copy", p);
2914                 break;
2915         case DIFF_STATUS_RENAMED:
2916                 show_rename_copy(file, "rename", p);
2917                 break;
2918         default:
2919                 if (p->score) {
2920                         fputs(" rewrite ", file);
2921                         write_name_quoted(p->two->path, file, ' ');
2922                         fprintf(file, "(%d%%)\n", similarity_index(p));
2923                 }
2924                 show_mode_change(file, p, !p->score);
2925                 break;
2926         }
2927 }
2928
2929 struct patch_id_t {
2930         git_SHA_CTX *ctx;
2931         int patchlen;
2932 };
2933
2934 static int remove_space(char *line, int len)
2935 {
2936         int i;
2937         char *dst = line;
2938         unsigned char c;
2939
2940         for (i = 0; i < len; i++)
2941                 if (!isspace((c = line[i])))
2942                         *dst++ = c;
2943
2944         return dst - line;
2945 }
2946
2947 static void patch_id_consume(void *priv, char *line, unsigned long len)
2948 {
2949         struct patch_id_t *data = priv;
2950         int new_len;
2951
2952         /* Ignore line numbers when computing the SHA1 of the patch */
2953         if (!prefixcmp(line, "@@ -"))
2954                 return;
2955
2956         new_len = remove_space(line, len);
2957
2958         git_SHA1_Update(data->ctx, line, new_len);
2959         data->patchlen += new_len;
2960 }
2961
2962 /* returns 0 upon success, and writes result into sha1 */
2963 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2964 {
2965         struct diff_queue_struct *q = &diff_queued_diff;
2966         int i;
2967         git_SHA_CTX ctx;
2968         struct patch_id_t data;
2969         char buffer[PATH_MAX * 4 + 20];
2970
2971         git_SHA1_Init(&ctx);
2972         memset(&data, 0, sizeof(struct patch_id_t));
2973         data.ctx = &ctx;
2974
2975         for (i = 0; i < q->nr; i++) {
2976                 xpparam_t xpp;
2977                 xdemitconf_t xecfg;
2978                 xdemitcb_t ecb;
2979                 mmfile_t mf1, mf2;
2980                 struct diff_filepair *p = q->queue[i];
2981                 int len1, len2;
2982
2983                 memset(&xecfg, 0, sizeof(xecfg));
2984                 if (p->status == 0)
2985                         return error("internal diff status error");
2986                 if (p->status == DIFF_STATUS_UNKNOWN)
2987                         continue;
2988                 if (diff_unmodified_pair(p))
2989                         continue;
2990                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2991                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2992                         continue;
2993                 if (DIFF_PAIR_UNMERGED(p))
2994                         continue;
2995
2996                 diff_fill_sha1_info(p->one);
2997                 diff_fill_sha1_info(p->two);
2998                 if (fill_mmfile(&mf1, p->one) < 0 ||
2999                                 fill_mmfile(&mf2, p->two) < 0)
3000                         return error("unable to read files to diff");
3001
3002                 len1 = remove_space(p->one->path, strlen(p->one->path));
3003                 len2 = remove_space(p->two->path, strlen(p->two->path));
3004                 if (p->one->mode == 0)
3005                         len1 = snprintf(buffer, sizeof(buffer),
3006                                         "diff--gita/%.*sb/%.*s"
3007                                         "newfilemode%06o"
3008                                         "---/dev/null"
3009                                         "+++b/%.*s",
3010                                         len1, p->one->path,
3011                                         len2, p->two->path,
3012                                         p->two->mode,
3013                                         len2, p->two->path);
3014                 else if (p->two->mode == 0)
3015                         len1 = snprintf(buffer, sizeof(buffer),
3016                                         "diff--gita/%.*sb/%.*s"
3017                                         "deletedfilemode%06o"
3018                                         "---a/%.*s"
3019                                         "+++/dev/null",
3020                                         len1, p->one->path,
3021                                         len2, p->two->path,
3022                                         p->one->mode,
3023                                         len1, p->one->path);
3024                 else
3025                         len1 = snprintf(buffer, sizeof(buffer),
3026                                         "diff--gita/%.*sb/%.*s"
3027                                         "---a/%.*s"
3028                                         "+++b/%.*s",
3029                                         len1, p->one->path,
3030                                         len2, p->two->path,
3031                                         len1, p->one->path,
3032                                         len2, p->two->path);
3033                 git_SHA1_Update(&ctx, buffer, len1);
3034
3035                 xpp.flags = XDF_NEED_MINIMAL;
3036                 xecfg.ctxlen = 3;
3037                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3038                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3039                               &xpp, &xecfg, &ecb);
3040         }
3041
3042         git_SHA1_Final(sha1, &ctx);
3043         return 0;
3044 }
3045
3046 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3047 {
3048         struct diff_queue_struct *q = &diff_queued_diff;
3049         int i;
3050         int result = diff_get_patch_id(options, sha1);
3051
3052         for (i = 0; i < q->nr; i++)
3053                 diff_free_filepair(q->queue[i]);
3054
3055         free(q->queue);
3056         q->queue = NULL;
3057         q->nr = q->alloc = 0;
3058
3059         return result;
3060 }
3061
3062 static int is_summary_empty(const struct diff_queue_struct *q)
3063 {
3064         int i;
3065
3066         for (i = 0; i < q->nr; i++) {
3067                 const struct diff_filepair *p = q->queue[i];
3068
3069                 switch (p->status) {
3070                 case DIFF_STATUS_DELETED:
3071                 case DIFF_STATUS_ADDED:
3072                 case DIFF_STATUS_COPIED:
3073                 case DIFF_STATUS_RENAMED:
3074                         return 0;
3075                 default:
3076                         if (p->score)
3077                                 return 0;
3078                         if (p->one->mode && p->two->mode &&
3079                             p->one->mode != p->two->mode)
3080                                 return 0;
3081                         break;
3082                 }
3083         }
3084         return 1;
3085 }
3086
3087 void diff_flush(struct diff_options *options)
3088 {
3089         struct diff_queue_struct *q = &diff_queued_diff;
3090         int i, output_format = options->output_format;
3091         int separator = 0;
3092
3093         /*
3094          * Order: raw, stat, summary, patch
3095          * or:    name/name-status/checkdiff (other bits clear)
3096          */
3097         if (!q->nr)
3098                 goto free_queue;
3099
3100         if (output_format & (DIFF_FORMAT_RAW |
3101                              DIFF_FORMAT_NAME |
3102                              DIFF_FORMAT_NAME_STATUS |
3103                              DIFF_FORMAT_CHECKDIFF)) {
3104                 for (i = 0; i < q->nr; i++) {
3105                         struct diff_filepair *p = q->queue[i];
3106                         if (check_pair_status(p))
3107                                 flush_one_pair(p, options);
3108                 }
3109                 separator++;
3110         }
3111
3112         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3113                 struct diffstat_t diffstat;
3114
3115                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3116                 for (i = 0; i < q->nr; i++) {
3117                         struct diff_filepair *p = q->queue[i];
3118                         if (check_pair_status(p))
3119                                 diff_flush_stat(p, options, &diffstat);
3120                 }
3121                 if (output_format & DIFF_FORMAT_NUMSTAT)
3122                         show_numstat(&diffstat, options);
3123                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3124                         show_stats(&diffstat, options);
3125                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3126                         show_shortstats(&diffstat, options);
3127                 free_diffstat_info(&diffstat);
3128                 separator++;
3129         }
3130         if (output_format & DIFF_FORMAT_DIRSTAT)
3131                 show_dirstat(options);
3132
3133         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3134                 for (i = 0; i < q->nr; i++)
3135                         diff_summary(options->file, q->queue[i]);
3136                 separator++;
3137         }
3138
3139         if (output_format & DIFF_FORMAT_PATCH) {
3140                 if (separator) {
3141                         putc(options->line_termination, options->file);
3142                         if (options->stat_sep) {
3143                                 /* attach patch instead of inline */
3144                                 fputs(options->stat_sep, options->file);
3145                         }
3146                 }
3147
3148                 for (i = 0; i < q->nr; i++) {
3149                         struct diff_filepair *p = q->queue[i];
3150                         if (check_pair_status(p))
3151                                 diff_flush_patch(p, options);
3152                 }
3153         }
3154
3155         if (output_format & DIFF_FORMAT_CALLBACK)
3156                 options->format_callback(q, options, options->format_callback_data);
3157
3158         for (i = 0; i < q->nr; i++)
3159                 diff_free_filepair(q->queue[i]);
3160 free_queue:
3161         free(q->queue);
3162         q->queue = NULL;
3163         q->nr = q->alloc = 0;
3164         if (options->close_file)
3165                 fclose(options->file);
3166 }
3167
3168 static void diffcore_apply_filter(const char *filter)
3169 {
3170         int i;
3171         struct diff_queue_struct *q = &diff_queued_diff;
3172         struct diff_queue_struct outq;
3173         outq.queue = NULL;
3174         outq.nr = outq.alloc = 0;
3175
3176         if (!filter)
3177                 return;
3178
3179         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3180                 int found;
3181                 for (i = found = 0; !found && i < q->nr; i++) {
3182                         struct diff_filepair *p = q->queue[i];
3183                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3184                              ((p->score &&
3185                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3186                               (!p->score &&
3187                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3188                             ((p->status != DIFF_STATUS_MODIFIED) &&
3189                              strchr(filter, p->status)))
3190                                 found++;
3191                 }
3192                 if (found)
3193                         return;
3194
3195                 /* otherwise we will clear the whole queue
3196                  * by copying the empty outq at the end of this
3197                  * function, but first clear the current entries
3198                  * in the queue.
3199                  */
3200                 for (i = 0; i < q->nr; i++)
3201                         diff_free_filepair(q->queue[i]);
3202         }
3203         else {
3204                 /* Only the matching ones */
3205                 for (i = 0; i < q->nr; i++) {
3206                         struct diff_filepair *p = q->queue[i];
3207
3208                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3209                              ((p->score &&
3210                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3211                               (!p->score &&
3212                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3213                             ((p->status != DIFF_STATUS_MODIFIED) &&
3214                              strchr(filter, p->status)))
3215                                 diff_q(&outq, p);
3216                         else
3217                                 diff_free_filepair(p);
3218                 }
3219         }
3220         free(q->queue);
3221         *q = outq;
3222 }
3223
3224 /* Check whether two filespecs with the same mode and size are identical */
3225 static int diff_filespec_is_identical(struct diff_filespec *one,
3226                                       struct diff_filespec *two)
3227 {
3228         if (S_ISGITLINK(one->mode))
3229                 return 0;
3230         if (diff_populate_filespec(one, 0))
3231                 return 0;
3232         if (diff_populate_filespec(two, 0))
3233                 return 0;
3234         return !memcmp(one->data, two->data, one->size);
3235 }
3236
3237 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3238 {
3239         int i;
3240         struct diff_queue_struct *q = &diff_queued_diff;
3241         struct diff_queue_struct outq;
3242         outq.queue = NULL;
3243         outq.nr = outq.alloc = 0;
3244
3245         for (i = 0; i < q->nr; i++) {
3246                 struct diff_filepair *p = q->queue[i];
3247
3248                 /*
3249                  * 1. Entries that come from stat info dirtyness
3250                  *    always have both sides (iow, not create/delete),
3251                  *    one side of the object name is unknown, with
3252                  *    the same mode and size.  Keep the ones that
3253                  *    do not match these criteria.  They have real
3254                  *    differences.
3255                  *
3256                  * 2. At this point, the file is known to be modified,
3257                  *    with the same mode and size, and the object
3258                  *    name of one side is unknown.  Need to inspect
3259                  *    the identical contents.
3260                  */
3261                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3262                     !DIFF_FILE_VALID(p->two) ||
3263                     (p->one->sha1_valid && p->two->sha1_valid) ||
3264                     (p->one->mode != p->two->mode) ||
3265                     diff_populate_filespec(p->one, 1) ||
3266                     diff_populate_filespec(p->two, 1) ||
3267                     (p->one->size != p->two->size) ||
3268                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3269                         diff_q(&outq, p);
3270                 else {
3271                         /*
3272                          * The caller can subtract 1 from skip_stat_unmatch
3273                          * to determine how many paths were dirty only
3274                          * due to stat info mismatch.
3275                          */
3276                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3277                                 diffopt->skip_stat_unmatch++;
3278                         diff_free_filepair(p);
3279                 }
3280         }
3281         free(q->queue);
3282         *q = outq;
3283 }
3284
3285 void diffcore_std(struct diff_options *options)
3286 {
3287         if (options->skip_stat_unmatch)
3288                 diffcore_skip_stat_unmatch(options);
3289         if (options->break_opt != -1)
3290                 diffcore_break(options->break_opt);
3291         if (options->detect_rename)
3292                 diffcore_rename(options);
3293         if (options->break_opt != -1)
3294                 diffcore_merge_broken();
3295         if (options->pickaxe)
3296                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3297         if (options->orderfile)
3298                 diffcore_order(options->orderfile);
3299         diff_resolve_rename_copy();
3300         diffcore_apply_filter(options->filter);
3301
3302         if (diff_queued_diff.nr)
3303                 DIFF_OPT_SET(options, HAS_CHANGES);
3304         else
3305                 DIFF_OPT_CLR(options, HAS_CHANGES);
3306 }
3307
3308 int diff_result_code(struct diff_options *opt, int status)
3309 {
3310         int result = 0;
3311         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3312             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3313                 return status;
3314         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3315             DIFF_OPT_TST(opt, HAS_CHANGES))
3316                 result |= 01;
3317         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3318             DIFF_OPT_TST(opt, CHECK_FAILED))
3319                 result |= 02;
3320         return result;
3321 }
3322
3323 void diff_addremove(struct diff_options *options,
3324                     int addremove, unsigned mode,
3325                     const unsigned char *sha1,
3326                     const char *concatpath)
3327 {
3328         struct diff_filespec *one, *two;
3329
3330         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3331                 return;
3332
3333         /* This may look odd, but it is a preparation for
3334          * feeding "there are unchanged files which should
3335          * not produce diffs, but when you are doing copy
3336          * detection you would need them, so here they are"
3337          * entries to the diff-core.  They will be prefixed
3338          * with something like '=' or '*' (I haven't decided
3339          * which but should not make any difference).
3340          * Feeding the same new and old to diff_change()
3341          * also has the same effect.
3342          * Before the final output happens, they are pruned after
3343          * merged into rename/copy pairs as appropriate.
3344          */
3345         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3346                 addremove = (addremove == '+' ? '-' :
3347                              addremove == '-' ? '+' : addremove);
3348
3349         if (options->prefix &&
3350             strncmp(concatpath, options->prefix, options->prefix_length))
3351                 return;
3352
3353         one = alloc_filespec(concatpath);
3354         two = alloc_filespec(concatpath);
3355
3356         if (addremove != '+')
3357                 fill_filespec(one, sha1, mode);
3358         if (addremove != '-')
3359                 fill_filespec(two, sha1, mode);
3360
3361         diff_queue(&diff_queued_diff, one, two);
3362         DIFF_OPT_SET(options, HAS_CHANGES);
3363 }
3364
3365 void diff_change(struct diff_options *options,
3366                  unsigned old_mode, unsigned new_mode,
3367                  const unsigned char *old_sha1,
3368                  const unsigned char *new_sha1,
3369                  const char *concatpath)
3370 {
3371         struct diff_filespec *one, *two;
3372
3373         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3374                         && S_ISGITLINK(new_mode))
3375                 return;
3376
3377         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3378                 unsigned tmp;
3379                 const unsigned char *tmp_c;
3380                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3381                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3382         }
3383
3384         if (options->prefix &&
3385             strncmp(concatpath, options->prefix, options->prefix_length))
3386                 return;
3387
3388         one = alloc_filespec(concatpath);
3389         two = alloc_filespec(concatpath);
3390         fill_filespec(one, old_sha1, old_mode);
3391         fill_filespec(two, new_sha1, new_mode);
3392
3393         diff_queue(&diff_queued_diff, one, two);
3394         DIFF_OPT_SET(options, HAS_CHANGES);
3395 }
3396
3397 void diff_unmerge(struct diff_options *options,
3398                   const char *path,
3399                   unsigned mode, const unsigned char *sha1)
3400 {
3401         struct diff_filespec *one, *two;
3402
3403         if (options->prefix &&
3404             strncmp(path, options->prefix, options->prefix_length))
3405                 return;
3406
3407         one = alloc_filespec(path);
3408         two = alloc_filespec(path);
3409         fill_filespec(one, sha1, mode);
3410         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3411 }
3412
3413 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3414                 size_t *outsize)
3415 {
3416         struct diff_tempfile temp;
3417         const char *argv[3];
3418         const char **arg = argv;
3419         struct child_process child;
3420         struct strbuf buf = STRBUF_INIT;
3421
3422         prepare_temp_file(spec->path, &temp, spec);
3423         *arg++ = pgm;
3424         *arg++ = temp.name;
3425         *arg = NULL;
3426
3427         memset(&child, 0, sizeof(child));
3428         child.argv = argv;
3429         child.out = -1;
3430         if (start_command(&child) != 0 ||
3431             strbuf_read(&buf, child.out, 0) < 0 ||
3432             finish_command(&child) != 0) {
3433                 if (temp.name == temp.tmp_path)
3434                         unlink(temp.name);
3435                 error("error running textconv command '%s'", pgm);
3436                 return NULL;
3437         }
3438         if (temp.name == temp.tmp_path)
3439                 unlink(temp.name);
3440
3441         return strbuf_detach(&buf, outsize);
3442 }