git.oblomov.eu Git - git/blob - 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         return git_diff_basic_config(var, value, cb);
  97 }
  98
  99 int git_diff_basic_config(const char *var, const char *value, void *cb)
 100 {
 101         if (!strcmp(var, "diff.renamelimit")) {
 102                 diff_rename_limit_default = git_config_int(var, value);
 103                 return 0;
 104         }
 105
 106         switch (userdiff_config(var, value)) {
 107                 case 0: break;
 108                 case -1: return -1;
 109                 default: 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.suppressblankempty") ||
 122                         /* for backwards compatibility */
 123                         !strcmp(var, "diff.suppress-blank-empty")) {
 124                 diff_suppress_blank_empty = git_config_bool(var, value);
 125                 return 0;
 126         }
 127
 128         return git_color_default_config(var, value, cb);
 129 }
 130
 131 static char *quote_two(const char *one, const char *two)
 132 {
 133         int need_one = quote_c_style(one, NULL, NULL, 1);
 134         int need_two = quote_c_style(two, NULL, NULL, 1);
 135         struct strbuf res = STRBUF_INIT;
 136
 137         if (need_one + need_two) {
 138                 strbuf_addch(&res, '"');
 139                 quote_c_style(one, &res, NULL, 1);
 140                 quote_c_style(two, &res, NULL, 1);
 141                 strbuf_addch(&res, '"');
 142         } else {
 143                 strbuf_addstr(&res, one);
 144                 strbuf_addstr(&res, two);
 145         }
 146         return strbuf_detach(&res, NULL);
 147 }
 148
 149 static const char *external_diff(void)
 150 {
 151         static const char *external_diff_cmd = NULL;
 152         static int done_preparing = 0;
 153
 154         if (done_preparing)
 155                 return external_diff_cmd;
 156         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 157         if (!external_diff_cmd)
 158                 external_diff_cmd = external_diff_cmd_cfg;
 159         done_preparing = 1;
 160         return external_diff_cmd;
 161 }
 162
 163 static struct diff_tempfile {
 164         const char *name; /* filename external diff should read from */
 165         char hex[41];
 166         char mode[10];
 167         char tmp_path[PATH_MAX];
 168 } diff_temp[2];
 169
 170 static int count_lines(const char *data, int size)
 171 {
 172         int count, ch, completely_empty = 1, nl_just_seen = 0;
 173         count = 0;
 174         while (0 < size--) {
 175                 ch = *data++;
 176                 if (ch == '\n') {
 177                         count++;
 178                         nl_just_seen = 1;
 179                         completely_empty = 0;
 180                 }
 181                 else {
 182                         nl_just_seen = 0;
 183                         completely_empty = 0;
 184                 }
 185         }
 186         if (completely_empty)
 187                 return 0;
 188         if (!nl_just_seen)
 189                 count++; /* no trailing newline */
 190         return count;
 191 }
 192
 193 static void print_line_count(FILE *file, int count)
 194 {
 195         switch (count) {
 196         case 0:
 197                 fprintf(file, "0,0");
 198                 break;
 199         case 1:
 200                 fprintf(file, "1");
 201                 break;
 202         default:
 203                 fprintf(file, "1,%d", count);
 204                 break;
 205         }
 206 }
 207
 208 static void copy_file_with_prefix(FILE *file,
 209                                   int prefix, const char *data, int size,
 210                                   const char *set, const char *reset)
 211 {
 212         int ch, nl_just_seen = 1;
 213         while (0 < size--) {
 214                 ch = *data++;
 215                 if (nl_just_seen) {
 216                         fputs(set, file);
 217                         putc(prefix, file);
 218                 }
 219                 if (ch == '\n') {
 220                         nl_just_seen = 1;
 221                         fputs(reset, file);
 222                 } else
 223                         nl_just_seen = 0;
 224                 putc(ch, file);
 225         }
 226         if (!nl_just_seen)
 227                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 228 }
 229
 230 static void emit_rewrite_diff(const char *name_a,
 231                               const char *name_b,
 232                               struct diff_filespec *one,
 233                               struct diff_filespec *two,
 234                               const char *textconv_one,
 235                               const char *textconv_two,
 236                               struct diff_options *o)
 237 {
 238         int lc_a, lc_b;
 239         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 240         const char *name_a_tab, *name_b_tab;
 241         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 242         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 243         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 244         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 245         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 246         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 247         const char *a_prefix, *b_prefix;
 248         const char *data_one, *data_two;
 249         size_t size_one, size_two;
 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         if (textconv_one) {
 272                 data_one = run_textconv(textconv_one, one, &size_one);
 273                 if (!data_one)
 274                         die("unable to read files to diff");
 275         }
 276         else {
 277                 data_one = one->data;
 278                 size_one = one->size;
 279         }
 280         if (textconv_two) {
 281                 data_two = run_textconv(textconv_two, two, &size_two);
 282                 if (!data_two)
 283                         die("unable to read files to diff");
 284         }
 285         else {
 286                 data_two = two->data;
 287                 size_two = two->size;
 288         }
 289
 290         lc_a = count_lines(data_one, size_one);
 291         lc_b = count_lines(data_two, size_two);
 292         fprintf(o->file,
 293                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 294                 metainfo, a_name.buf, name_a_tab, reset,
 295                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 296         print_line_count(o->file, lc_a);
 297         fprintf(o->file, " +");
 298         print_line_count(o->file, lc_b);
 299         fprintf(o->file, " @@%s\n", reset);
 300         if (lc_a)
 301                 copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
 302         if (lc_b)
 303                 copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
 304 }
 305
 306 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 307 {
 308         if (!DIFF_FILE_VALID(one)) {
 309                 mf->ptr = (char *)""; /* does not matter */
 310                 mf->size = 0;
 311                 return 0;
 312         }
 313         else if (diff_populate_filespec(one, 0))
 314                 return -1;
 315
 316         mf->ptr = one->data;
 317         mf->size = one->size;
 318         return 0;
 319 }
 320
 321 struct diff_words_buffer {
 322         mmfile_t text;
 323         long alloc;
 324         long current; /* output pointer */
 325         int suppressed_newline;
 326 };
 327
 328 static void diff_words_append(char *line, unsigned long len,
 329                 struct diff_words_buffer *buffer)
 330 {
 331         if (buffer->text.size + len > buffer->alloc) {
 332                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 333                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 334         }
 335         line++;
 336         len--;
 337         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 338         buffer->text.size += len;
 339 }
 340
 341 struct diff_words_data {
 342         struct diff_words_buffer minus, plus;
 343         FILE *file;
 344 };
 345
 346 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 347                 int suppress_newline)
 348 {
 349         const char *ptr;
 350         int eol = 0;
 351
 352         if (len == 0)
 353                 return;
 354
 355         ptr  = buffer->text.ptr + buffer->current;
 356         buffer->current += len;
 357
 358         if (ptr[len - 1] == '\n') {
 359                 eol = 1;
 360                 len--;
 361         }
 362
 363         fputs(diff_get_color(1, color), file);
 364         fwrite(ptr, len, 1, file);
 365         fputs(diff_get_color(1, DIFF_RESET), file);
 366
 367         if (eol) {
 368                 if (suppress_newline)
 369                         buffer->suppressed_newline = 1;
 370                 else
 371                         putc('\n', file);
 372         }
 373 }
 374
 375 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 376 {
 377         struct diff_words_data *diff_words = priv;
 378
 379         if (diff_words->minus.suppressed_newline) {
 380                 if (line[0] != '+')
 381                         putc('\n', diff_words->file);
 382                 diff_words->minus.suppressed_newline = 0;
 383         }
 384
 385         len--;
 386         switch (line[0]) {
 387                 case '-':
 388                         print_word(diff_words->file,
 389                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 390                         break;
 391                 case '+':
 392                         print_word(diff_words->file,
 393                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 394                         break;
 395                 case ' ':
 396                         print_word(diff_words->file,
 397                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 398                         diff_words->minus.current += len;
 399                         break;
 400         }
 401 }
 402
 403 /* this executes the word diff on the accumulated buffers */
 404 static void diff_words_show(struct diff_words_data *diff_words)
 405 {
 406         xpparam_t xpp;
 407         xdemitconf_t xecfg;
 408         xdemitcb_t ecb;
 409         mmfile_t minus, plus;
 410         int i;
 411
 412         memset(&xpp, 0, sizeof(xpp));
 413         memset(&xecfg, 0, sizeof(xecfg));
 414         minus.size = diff_words->minus.text.size;
 415         minus.ptr = xmalloc(minus.size);
 416         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 417         for (i = 0; i < minus.size; i++)
 418                 if (isspace(minus.ptr[i]))
 419                         minus.ptr[i] = '\n';
 420         diff_words->minus.current = 0;
 421
 422         plus.size = diff_words->plus.text.size;
 423         plus.ptr = xmalloc(plus.size);
 424         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 425         for (i = 0; i < plus.size; i++)
 426                 if (isspace(plus.ptr[i]))
 427                         plus.ptr[i] = '\n';
 428         diff_words->plus.current = 0;
 429
 430         xpp.flags = XDF_NEED_MINIMAL;
 431         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 432         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 433                       &xpp, &xecfg, &ecb);
 434         free(minus.ptr);
 435         free(plus.ptr);
 436         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 437
 438         if (diff_words->minus.suppressed_newline) {
 439                 putc('\n', diff_words->file);
 440                 diff_words->minus.suppressed_newline = 0;
 441         }
 442 }
 443
 444 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 445
 446 struct emit_callback {
 447         int nparents, color_diff;
 448         unsigned ws_rule;
 449         sane_truncate_fn truncate;
 450         const char **label_path;
 451         struct diff_words_data *diff_words;
 452         int *found_changesp;
 453         FILE *file;
 454 };
 455
 456 static void free_diff_words_data(struct emit_callback *ecbdata)
 457 {
 458         if (ecbdata->diff_words) {
 459                 /* flush buffers */
 460                 if (ecbdata->diff_words->minus.text.size ||
 461                                 ecbdata->diff_words->plus.text.size)
 462                         diff_words_show(ecbdata->diff_words);
 463
 464                 free (ecbdata->diff_words->minus.text.ptr);
 465                 free (ecbdata->diff_words->plus.text.ptr);
 466                 free(ecbdata->diff_words);
 467                 ecbdata->diff_words = NULL;
 468         }
 469 }
 470
 471 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 472 {
 473         if (diff_use_color)
 474                 return diff_colors[ix];
 475         return "";
 476 }
 477
 478 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 479 {
 480         int has_trailing_newline, has_trailing_carriage_return;
 481
 482         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 483         if (has_trailing_newline)
 484                 len--;
 485         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 486         if (has_trailing_carriage_return)
 487                 len--;
 488
 489         fputs(set, file);
 490         fwrite(line, len, 1, file);
 491         fputs(reset, file);
 492         if (has_trailing_carriage_return)
 493                 fputc('\r', file);
 494         if (has_trailing_newline)
 495                 fputc('\n', file);
 496 }
 497
 498 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 499 {
 500         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 501         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 502
 503         if (!*ws)
 504                 emit_line(ecbdata->file, set, reset, line, len);
 505         else {
 506                 /* Emit just the prefix, then the rest. */
 507                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 508                 ws_check_emit(line + ecbdata->nparents,
 509                               len - ecbdata->nparents, ecbdata->ws_rule,
 510                               ecbdata->file, set, reset, ws);
 511         }
 512 }
 513
 514 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 515 {
 516         const char *cp;
 517         unsigned long allot;
 518         size_t l = len;
 519
 520         if (ecb->truncate)
 521                 return ecb->truncate(line, len);
 522         cp = line;
 523         allot = l;
 524         while (0 < l) {
 525                 (void) utf8_width(&cp, &l);
 526                 if (!cp)
 527                         break; /* truncated in the middle? */
 528         }
 529         return allot - l;
 530 }
 531
 532 static void fn_out_consume(void *priv, char *line, unsigned long len)
 533 {
 534         int i;
 535         int color;
 536         struct emit_callback *ecbdata = priv;
 537         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 538         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 539         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 540
 541         *(ecbdata->found_changesp) = 1;
 542
 543         if (ecbdata->label_path[0]) {
 544                 const char *name_a_tab, *name_b_tab;
 545
 546                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 547                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 548
 549                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 550                         meta, ecbdata->label_path[0], reset, name_a_tab);
 551                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 552                         meta, ecbdata->label_path[1], reset, name_b_tab);
 553                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 554         }
 555
 556         if (diff_suppress_blank_empty
 557             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 558                 line[0] = '\n';
 559                 len = 1;
 560         }
 561
 562         /* This is not really necessary for now because
 563          * this codepath only deals with two-way diffs.
 564          */
 565         for (i = 0; i < len && line[i] == '@'; i++)
 566                 ;
 567         if (2 <= i && i < len && line[i] == ' ') {
 568                 ecbdata->nparents = i - 1;
 569                 len = sane_truncate_line(ecbdata, line, len);
 570                 emit_line(ecbdata->file,
 571                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 572                           reset, line, len);
 573                 if (line[len-1] != '\n')
 574                         putc('\n', ecbdata->file);
 575                 return;
 576         }
 577
 578         if (len < ecbdata->nparents) {
 579                 emit_line(ecbdata->file, reset, reset, line, len);
 580                 return;
 581         }
 582
 583         color = DIFF_PLAIN;
 584         if (ecbdata->diff_words && ecbdata->nparents != 1)
 585                 /* fall back to normal diff */
 586                 free_diff_words_data(ecbdata);
 587         if (ecbdata->diff_words) {
 588                 if (line[0] == '-') {
 589                         diff_words_append(line, len,
 590                                           &ecbdata->diff_words->minus);
 591                         return;
 592                 } else if (line[0] == '+') {
 593                         diff_words_append(line, len,
 594                                           &ecbdata->diff_words->plus);
 595                         return;
 596                 }
 597                 if (ecbdata->diff_words->minus.text.size ||
 598                     ecbdata->diff_words->plus.text.size)
 599                         diff_words_show(ecbdata->diff_words);
 600                 line++;
 601                 len--;
 602                 emit_line(ecbdata->file, plain, reset, line, len);
 603                 return;
 604         }
 605         for (i = 0; i < ecbdata->nparents && len; i++) {
 606                 if (line[i] == '-')
 607                         color = DIFF_FILE_OLD;
 608                 else if (line[i] == '+')
 609                         color = DIFF_FILE_NEW;
 610         }
 611
 612         if (color != DIFF_FILE_NEW) {
 613                 emit_line(ecbdata->file,
 614                           diff_get_color(ecbdata->color_diff, color),
 615                           reset, line, len);
 616                 return;
 617         }
 618         emit_add_line(reset, ecbdata, line, len);
 619 }
 620
 621 static char *pprint_rename(const char *a, const char *b)
 622 {
 623         const char *old = a;
 624         const char *new = b;
 625         struct strbuf name = STRBUF_INIT;
 626         int pfx_length, sfx_length;
 627         int len_a = strlen(a);
 628         int len_b = strlen(b);
 629         int a_midlen, b_midlen;
 630         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 631         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 632
 633         if (qlen_a || qlen_b) {
 634                 quote_c_style(a, &name, NULL, 0);
 635                 strbuf_addstr(&name, " => ");
 636                 quote_c_style(b, &name, NULL, 0);
 637                 return strbuf_detach(&name, NULL);
 638         }
 639
 640         /* Find common prefix */
 641         pfx_length = 0;
 642         while (*old && *new && *old == *new) {
 643                 if (*old == '/')
 644                         pfx_length = old - a + 1;
 645                 old++;
 646                 new++;
 647         }
 648
 649         /* Find common suffix */
 650         old = a + len_a;
 651         new = b + len_b;
 652         sfx_length = 0;
 653         while (a <= old && b <= new && *old == *new) {
 654                 if (*old == '/')
 655                         sfx_length = len_a - (old - a);
 656                 old--;
 657                 new--;
 658         }
 659
 660         /*
 661          * pfx{mid-a => mid-b}sfx
 662          * {pfx-a => pfx-b}sfx
 663          * pfx{sfx-a => sfx-b}
 664          * name-a => name-b
 665          */
 666         a_midlen = len_a - pfx_length - sfx_length;
 667         b_midlen = len_b - pfx_length - sfx_length;
 668         if (a_midlen < 0)
 669                 a_midlen = 0;
 670         if (b_midlen < 0)
 671                 b_midlen = 0;
 672
 673         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 674         if (pfx_length + sfx_length) {
 675                 strbuf_add(&name, a, pfx_length);
 676                 strbuf_addch(&name, '{');
 677         }
 678         strbuf_add(&name, a + pfx_length, a_midlen);
 679         strbuf_addstr(&name, " => ");
 680         strbuf_add(&name, b + pfx_length, b_midlen);
 681         if (pfx_length + sfx_length) {
 682                 strbuf_addch(&name, '}');
 683                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 684         }
 685         return strbuf_detach(&name, NULL);
 686 }
 687
 688 struct diffstat_t {
 689         int nr;
 690         int alloc;
 691         struct diffstat_file {
 692                 char *from_name;
 693                 char *name;
 694                 char *print_name;
 695                 unsigned is_unmerged:1;
 696                 unsigned is_binary:1;
 697                 unsigned is_renamed:1;
 698                 unsigned int added, deleted;
 699         } **files;
 700 };
 701
 702 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 703                                           const char *name_a,
 704                                           const char *name_b)
 705 {
 706         struct diffstat_file *x;
 707         x = xcalloc(sizeof (*x), 1);
 708         if (diffstat->nr == diffstat->alloc) {
 709                 diffstat->alloc = alloc_nr(diffstat->alloc);
 710                 diffstat->files = xrealloc(diffstat->files,
 711                                 diffstat->alloc * sizeof(x));
 712         }
 713         diffstat->files[diffstat->nr++] = x;
 714         if (name_b) {
 715                 x->from_name = xstrdup(name_a);
 716                 x->name = xstrdup(name_b);
 717                 x->is_renamed = 1;
 718         }
 719         else {
 720                 x->from_name = NULL;
 721                 x->name = xstrdup(name_a);
 722         }
 723         return x;
 724 }
 725
 726 static void diffstat_consume(void *priv, char *line, unsigned long len)
 727 {
 728         struct diffstat_t *diffstat = priv;
 729         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 730
 731         if (line[0] == '+')
 732                 x->added++;
 733         else if (line[0] == '-')
 734                 x->deleted++;
 735 }
 736
 737 const char mime_boundary_leader[] = "------------";
 738
 739 static int scale_linear(int it, int width, int max_change)
 740 {
 741         /*
 742          * make sure that at least one '-' is printed if there were deletions,
 743          * and likewise for '+'.
 744          */
 745         if (max_change < 2)
 746                 return it;
 747         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 748 }
 749
 750 static void show_name(FILE *file,
 751                       const char *prefix, const char *name, int len,
 752                       const char *reset, const char *set)
 753 {
 754         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 755 }
 756
 757 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 758 {
 759         if (cnt <= 0)
 760                 return;
 761         fprintf(file, "%s", set);
 762         while (cnt--)
 763                 putc(ch, file);
 764         fprintf(file, "%s", reset);
 765 }
 766
 767 static void fill_print_name(struct diffstat_file *file)
 768 {
 769         char *pname;
 770
 771         if (file->print_name)
 772                 return;
 773
 774         if (!file->is_renamed) {
 775                 struct strbuf buf = STRBUF_INIT;
 776                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 777                         pname = strbuf_detach(&buf, NULL);
 778                 } else {
 779                         pname = file->name;
 780                         strbuf_release(&buf);
 781                 }
 782         } else {
 783                 pname = pprint_rename(file->from_name, file->name);
 784         }
 785         file->print_name = pname;
 786 }
 787
 788 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 789 {
 790         int i, len, add, del, total, adds = 0, dels = 0;
 791         int max_change = 0, max_len = 0;
 792         int total_files = data->nr;
 793         int width, name_width;
 794         const char *reset, *set, *add_c, *del_c;
 795
 796         if (data->nr == 0)
 797                 return;
 798
 799         width = options->stat_width ? options->stat_width : 80;
 800         name_width = options->stat_name_width ? options->stat_name_width : 50;
 801
 802         /* Sanity: give at least 5 columns to the graph,
 803          * but leave at least 10 columns for the name.
 804          */
 805         if (width < 25)
 806                 width = 25;
 807         if (name_width < 10)
 808                 name_width = 10;
 809         else if (width < name_width + 15)
 810                 name_width = width - 15;
 811
 812         /* Find the longest filename and max number of changes */
 813         reset = diff_get_color_opt(options, DIFF_RESET);
 814         set   = diff_get_color_opt(options, DIFF_PLAIN);
 815         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 816         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 817
 818         for (i = 0; i < data->nr; i++) {
 819                 struct diffstat_file *file = data->files[i];
 820                 int change = file->added + file->deleted;
 821                 fill_print_name(file);
 822                 len = strlen(file->print_name);
 823                 if (max_len < len)
 824                         max_len = len;
 825
 826                 if (file->is_binary || file->is_unmerged)
 827                         continue;
 828                 if (max_change < change)
 829                         max_change = change;
 830         }
 831
 832         /* Compute the width of the graph part;
 833          * 10 is for one blank at the beginning of the line plus
 834          * " | count " between the name and the graph.
 835          *
 836          * From here on, name_width is the width of the name area,
 837          * and width is the width of the graph area.
 838          */
 839         name_width = (name_width < max_len) ? name_width : max_len;
 840         if (width < (name_width + 10) + max_change)
 841                 width = width - (name_width + 10);
 842         else
 843                 width = max_change;
 844
 845         for (i = 0; i < data->nr; i++) {
 846                 const char *prefix = "";
 847                 char *name = data->files[i]->print_name;
 848                 int added = data->files[i]->added;
 849                 int deleted = data->files[i]->deleted;
 850                 int name_len;
 851
 852                 /*
 853                  * "scale" the filename
 854                  */
 855                 len = name_width;
 856                 name_len = strlen(name);
 857                 if (name_width < name_len) {
 858                         char *slash;
 859                         prefix = "...";
 860                         len -= 3;
 861                         name += name_len - len;
 862                         slash = strchr(name, '/');
 863                         if (slash)
 864                                 name = slash;
 865                 }
 866
 867                 if (data->files[i]->is_binary) {
 868                         show_name(options->file, prefix, name, len, reset, set);
 869                         fprintf(options->file, "  Bin ");
 870                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 871                         fprintf(options->file, " -> ");
 872                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 873                         fprintf(options->file, " bytes");
 874                         fprintf(options->file, "\n");
 875                         continue;
 876                 }
 877                 else if (data->files[i]->is_unmerged) {
 878                         show_name(options->file, prefix, name, len, reset, set);
 879                         fprintf(options->file, "  Unmerged\n");
 880                         continue;
 881                 }
 882                 else if (!data->files[i]->is_renamed &&
 883                          (added + deleted == 0)) {
 884                         total_files--;
 885                         continue;
 886                 }
 887
 888                 /*
 889                  * scale the add/delete
 890                  */
 891                 add = added;
 892                 del = deleted;
 893                 total = add + del;
 894                 adds += add;
 895                 dels += del;
 896
 897                 if (width <= max_change) {
 898                         add = scale_linear(add, width, max_change);
 899                         del = scale_linear(del, width, max_change);
 900                         total = add + del;
 901                 }
 902                 show_name(options->file, prefix, name, len, reset, set);
 903                 fprintf(options->file, "%5d%s", added + deleted,
 904                                 added + deleted ? " " : "");
 905                 show_graph(options->file, '+', add, add_c, reset);
 906                 show_graph(options->file, '-', del, del_c, reset);
 907                 fprintf(options->file, "\n");
 908         }
 909         fprintf(options->file,
 910                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 911                set, total_files, adds, dels, reset);
 912 }
 913
 914 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 915 {
 916         int i, adds = 0, dels = 0, total_files = data->nr;
 917
 918         if (data->nr == 0)
 919                 return;
 920
 921         for (i = 0; i < data->nr; i++) {
 922                 if (!data->files[i]->is_binary &&
 923                     !data->files[i]->is_unmerged) {
 924                         int added = data->files[i]->added;
 925                         int deleted= data->files[i]->deleted;
 926                         if (!data->files[i]->is_renamed &&
 927                             (added + deleted == 0)) {
 928                                 total_files--;
 929                         } else {
 930                                 adds += added;
 931                                 dels += deleted;
 932                         }
 933                 }
 934         }
 935         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 936                total_files, adds, dels);
 937 }
 938
 939 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 940 {
 941         int i;
 942
 943         if (data->nr == 0)
 944                 return;
 945
 946         for (i = 0; i < data->nr; i++) {
 947                 struct diffstat_file *file = data->files[i];
 948
 949                 if (file->is_binary)
 950                         fprintf(options->file, "-\t-\t");
 951                 else
 952                         fprintf(options->file,
 953                                 "%d\t%d\t", file->added, file->deleted);
 954                 if (options->line_termination) {
 955                         fill_print_name(file);
 956                         if (!file->is_renamed)
 957                                 write_name_quoted(file->name, options->file,
 958                                                   options->line_termination);
 959                         else {
 960                                 fputs(file->print_name, options->file);
 961                                 putc(options->line_termination, options->file);
 962                         }
 963                 } else {
 964                         if (file->is_renamed) {
 965                                 putc('\0', options->file);
 966                                 write_name_quoted(file->from_name, options->file, '\0');
 967                         }
 968                         write_name_quoted(file->name, options->file, '\0');
 969                 }
 970         }
 971 }
 972
 973 struct dirstat_file {
 974         const char *name;
 975         unsigned long changed;
 976 };
 977
 978 struct dirstat_dir {
 979         struct dirstat_file *files;
 980         int alloc, nr, percent, cumulative;
 981 };
 982
 983 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
 984 {
 985         unsigned long this_dir = 0;
 986         unsigned int sources = 0;
 987
 988         while (dir->nr) {
 989                 struct dirstat_file *f = dir->files;
 990                 int namelen = strlen(f->name);
 991                 unsigned long this;
 992                 char *slash;
 993
 994                 if (namelen < baselen)
 995                         break;
 996                 if (memcmp(f->name, base, baselen))
 997                         break;
 998                 slash = strchr(f->name + baselen, '/');
 999                 if (slash) {
1000                         int newbaselen = slash + 1 - f->name;
1001                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1002                         sources++;
1003                 } else {
1004                         this = f->changed;
1005                         dir->files++;
1006                         dir->nr--;
1007                         sources += 2;
1008                 }
1009                 this_dir += this;
1010         }
1011
1012         /*
1013          * We don't report dirstat's for
1014          *  - the top level
1015          *  - or cases where everything came from a single directory
1016          *    under this directory (sources == 1).
1017          */
1018         if (baselen && sources != 1) {
1019                 int permille = this_dir * 1000 / changed;
1020                 if (permille) {
1021                         int percent = permille / 10;
1022                         if (percent >= dir->percent) {
1023                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1024                                 if (!dir->cumulative)
1025                                         return 0;
1026                         }
1027                 }
1028         }
1029         return this_dir;
1030 }
1031
1032 static int dirstat_compare(const void *_a, const void *_b)
1033 {
1034         const struct dirstat_file *a = _a;
1035         const struct dirstat_file *b = _b;
1036         return strcmp(a->name, b->name);
1037 }
1038
1039 static void show_dirstat(struct diff_options *options)
1040 {
1041         int i;
1042         unsigned long changed;
1043         struct dirstat_dir dir;
1044         struct diff_queue_struct *q = &diff_queued_diff;
1045
1046         dir.files = NULL;
1047         dir.alloc = 0;
1048         dir.nr = 0;
1049         dir.percent = options->dirstat_percent;
1050         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1051
1052         changed = 0;
1053         for (i = 0; i < q->nr; i++) {
1054                 struct diff_filepair *p = q->queue[i];
1055                 const char *name;
1056                 unsigned long copied, added, damage;
1057
1058                 name = p->one->path ? p->one->path : p->two->path;
1059
1060                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1061                         diff_populate_filespec(p->one, 0);
1062                         diff_populate_filespec(p->two, 0);
1063                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1064                                                &copied, &added);
1065                         diff_free_filespec_data(p->one);
1066                         diff_free_filespec_data(p->two);
1067                 } else if (DIFF_FILE_VALID(p->one)) {
1068                         diff_populate_filespec(p->one, 1);
1069                         copied = added = 0;
1070                         diff_free_filespec_data(p->one);
1071                 } else if (DIFF_FILE_VALID(p->two)) {
1072                         diff_populate_filespec(p->two, 1);
1073                         copied = 0;
1074                         added = p->two->size;
1075                         diff_free_filespec_data(p->two);
1076                 } else
1077                         continue;
1078
1079                 /*
1080                  * Original minus copied is the removed material,
1081                  * added is the new material.  They are both damages
1082                  * made to the preimage. In --dirstat-by-file mode, count
1083                  * damaged files, not damaged lines. This is done by
1084                  * counting only a single damaged line per file.
1085                  */
1086                 damage = (p->one->size - copied) + added;
1087                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1088                         damage = 1;
1089
1090                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1091                 dir.files[dir.nr].name = name;
1092                 dir.files[dir.nr].changed = damage;
1093                 changed += damage;
1094                 dir.nr++;
1095         }
1096
1097         /* This can happen even with many files, if everything was renames */
1098         if (!changed)
1099                 return;
1100
1101         /* Show all directories with more than x% of the changes */
1102         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1103         gather_dirstat(options->file, &dir, changed, "", 0);
1104 }
1105
1106 static void free_diffstat_info(struct diffstat_t *diffstat)
1107 {
1108         int i;
1109         for (i = 0; i < diffstat->nr; i++) {
1110                 struct diffstat_file *f = diffstat->files[i];
1111                 if (f->name != f->print_name)
1112                         free(f->print_name);
1113                 free(f->name);
1114                 free(f->from_name);
1115                 free(f);
1116         }
1117         free(diffstat->files);
1118 }
1119
1120 struct checkdiff_t {
1121         const char *filename;
1122         int lineno;
1123         struct diff_options *o;
1124         unsigned ws_rule;
1125         unsigned status;
1126         int trailing_blanks_start;
1127 };
1128
1129 static int is_conflict_marker(const char *line, unsigned long len)
1130 {
1131         char firstchar;
1132         int cnt;
1133
1134         if (len < 8)
1135                 return 0;
1136         firstchar = line[0];
1137         switch (firstchar) {
1138         case '=': case '>': case '<':
1139                 break;
1140         default:
1141                 return 0;
1142         }
1143         for (cnt = 1; cnt < 7; cnt++)
1144                 if (line[cnt] != firstchar)
1145                         return 0;
1146         /* line[0] thru line[6] are same as firstchar */
1147         if (firstchar == '=') {
1148                 /* divider between ours and theirs? */
1149                 if (len != 8 || line[7] != '\n')
1150                         return 0;
1151         } else if (len < 8 || !isspace(line[7])) {
1152                 /* not divider before ours nor after theirs */
1153                 return 0;
1154         }
1155         return 1;
1156 }
1157
1158 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1159 {
1160         struct checkdiff_t *data = priv;
1161         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1162         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1163         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1164         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1165         char *err;
1166
1167         if (line[0] == '+') {
1168                 unsigned bad;
1169                 data->lineno++;
1170                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1171                         data->trailing_blanks_start = 0;
1172                 else if (!data->trailing_blanks_start)
1173                         data->trailing_blanks_start = data->lineno;
1174                 if (is_conflict_marker(line + 1, len - 1)) {
1175                         data->status |= 1;
1176                         fprintf(data->o->file,
1177                                 "%s:%d: leftover conflict marker\n",
1178                                 data->filename, data->lineno);
1179                 }
1180                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1181                 if (!bad)
1182                         return;
1183                 data->status |= bad;
1184                 err = whitespace_error_string(bad);
1185                 fprintf(data->o->file, "%s:%d: %s.\n",
1186                         data->filename, data->lineno, err);
1187                 free(err);
1188                 emit_line(data->o->file, set, reset, line, 1);
1189                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1190                               data->o->file, set, reset, ws);
1191         } else if (line[0] == ' ') {
1192                 data->lineno++;
1193                 data->trailing_blanks_start = 0;
1194         } else if (line[0] == '@') {
1195                 char *plus = strchr(line, '+');
1196                 if (plus)
1197                         data->lineno = strtol(plus, NULL, 10) - 1;
1198                 else
1199                         die("invalid diff");
1200                 data->trailing_blanks_start = 0;
1201         }
1202 }
1203
1204 static unsigned char *deflate_it(char *data,
1205                                  unsigned long size,
1206                                  unsigned long *result_size)
1207 {
1208         int bound;
1209         unsigned char *deflated;
1210         z_stream stream;
1211
1212         memset(&stream, 0, sizeof(stream));
1213         deflateInit(&stream, zlib_compression_level);
1214         bound = deflateBound(&stream, size);
1215         deflated = xmalloc(bound);
1216         stream.next_out = deflated;
1217         stream.avail_out = bound;
1218
1219         stream.next_in = (unsigned char *)data;
1220         stream.avail_in = size;
1221         while (deflate(&stream, Z_FINISH) == Z_OK)
1222                 ; /* nothing */
1223         deflateEnd(&stream);
1224         *result_size = stream.total_out;
1225         return deflated;
1226 }
1227
1228 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1229 {
1230         void *cp;
1231         void *delta;
1232         void *deflated;
1233         void *data;
1234         unsigned long orig_size;
1235         unsigned long delta_size;
1236         unsigned long deflate_size;
1237         unsigned long data_size;
1238
1239         /* We could do deflated delta, or we could do just deflated two,
1240          * whichever is smaller.
1241          */
1242         delta = NULL;
1243         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1244         if (one->size && two->size) {
1245                 delta = diff_delta(one->ptr, one->size,
1246                                    two->ptr, two->size,
1247                                    &delta_size, deflate_size);
1248                 if (delta) {
1249                         void *to_free = delta;
1250                         orig_size = delta_size;
1251                         delta = deflate_it(delta, delta_size, &delta_size);
1252                         free(to_free);
1253                 }
1254         }
1255
1256         if (delta && delta_size < deflate_size) {
1257                 fprintf(file, "delta %lu\n", orig_size);
1258                 free(deflated);
1259                 data = delta;
1260                 data_size = delta_size;
1261         }
1262         else {
1263                 fprintf(file, "literal %lu\n", two->size);
1264                 free(delta);
1265                 data = deflated;
1266                 data_size = deflate_size;
1267         }
1268
1269         /* emit data encoded in base85 */
1270         cp = data;
1271         while (data_size) {
1272                 int bytes = (52 < data_size) ? 52 : data_size;
1273                 char line[70];
1274                 data_size -= bytes;
1275                 if (bytes <= 26)
1276                         line[0] = bytes + 'A' - 1;
1277                 else
1278                         line[0] = bytes - 26 + 'a' - 1;
1279                 encode_85(line + 1, cp, bytes);
1280                 cp = (char *) cp + bytes;
1281                 fputs(line, file);
1282                 fputc('\n', file);
1283         }
1284         fprintf(file, "\n");
1285         free(data);
1286 }
1287
1288 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1289 {
1290         fprintf(file, "GIT binary patch\n");
1291         emit_binary_diff_body(file, one, two);
1292         emit_binary_diff_body(file, two, one);
1293 }
1294
1295 static void diff_filespec_load_driver(struct diff_filespec *one)
1296 {
1297         if (!one->driver)
1298                 one->driver = userdiff_find_by_path(one->path);
1299         if (!one->driver)
1300                 one->driver = userdiff_find_by_name("default");
1301 }
1302
1303 int diff_filespec_is_binary(struct diff_filespec *one)
1304 {
1305         if (one->is_binary == -1) {
1306                 diff_filespec_load_driver(one);
1307                 if (one->driver->binary != -1)
1308                         one->is_binary = one->driver->binary;
1309                 else {
1310                         if (!one->data && DIFF_FILE_VALID(one))
1311                                 diff_populate_filespec(one, 0);
1312                         if (one->data)
1313                                 one->is_binary = buffer_is_binary(one->data,
1314                                                 one->size);
1315                         if (one->is_binary == -1)
1316                                 one->is_binary = 0;
1317                 }
1318         }
1319         return one->is_binary;
1320 }
1321
1322 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1323 {
1324         diff_filespec_load_driver(one);
1325         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1326 }
1327
1328 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1329 {
1330         if (!options->a_prefix)
1331                 options->a_prefix = a;
1332         if (!options->b_prefix)
1333                 options->b_prefix = b;
1334 }
1335
1336 static const char *get_textconv(struct diff_filespec *one)
1337 {
1338         if (!DIFF_FILE_VALID(one))
1339                 return NULL;
1340         if (!S_ISREG(one->mode))
1341                 return NULL;
1342         diff_filespec_load_driver(one);
1343         return one->driver->textconv;
1344 }
1345
1346 static void builtin_diff(const char *name_a,
1347                          const char *name_b,
1348                          struct diff_filespec *one,
1349                          struct diff_filespec *two,
1350                          const char *xfrm_msg,
1351                          struct diff_options *o,
1352                          int complete_rewrite)
1353 {
1354         mmfile_t mf1, mf2;
1355         const char *lbl[2];
1356         char *a_one, *b_two;
1357         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1358         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1359         const char *a_prefix, *b_prefix;
1360         const char *textconv_one = NULL, *textconv_two = NULL;
1361
1362         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1363                 textconv_one = get_textconv(one);
1364                 textconv_two = get_textconv(two);
1365         }
1366
1367         diff_set_mnemonic_prefix(o, "a/", "b/");
1368         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1369                 a_prefix = o->b_prefix;
1370                 b_prefix = o->a_prefix;
1371         } else {
1372                 a_prefix = o->a_prefix;
1373                 b_prefix = o->b_prefix;
1374         }
1375
1376         /* Never use a non-valid filename anywhere if at all possible */
1377         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1378         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1379
1380         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1381         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1382         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1383         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1384         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1385         if (lbl[0][0] == '/') {
1386                 /* /dev/null */
1387                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1388                 if (xfrm_msg && xfrm_msg[0])
1389                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1390         }
1391         else if (lbl[1][0] == '/') {
1392                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1393                 if (xfrm_msg && xfrm_msg[0])
1394                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1395         }
1396         else {
1397                 if (one->mode != two->mode) {
1398                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1399                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1400                 }
1401                 if (xfrm_msg && xfrm_msg[0])
1402                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1403                 /*
1404                  * we do not run diff between different kind
1405                  * of objects.
1406                  */
1407                 if ((one->mode ^ two->mode) & S_IFMT)
1408                         goto free_ab_and_return;
1409                 if (complete_rewrite &&
1410                     (textconv_one || !diff_filespec_is_binary(one)) &&
1411                     (textconv_two || !diff_filespec_is_binary(two))) {
1412                         emit_rewrite_diff(name_a, name_b, one, two,
1413                                                 textconv_one, textconv_two, o);
1414                         o->found_changes = 1;
1415                         goto free_ab_and_return;
1416                 }
1417         }
1418
1419         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1420                 die("unable to read files to diff");
1421
1422         if (!DIFF_OPT_TST(o, TEXT) &&
1423             ( (diff_filespec_is_binary(one) && !textconv_one) ||
1424               (diff_filespec_is_binary(two) && !textconv_two) )) {
1425                 /* Quite common confusing case */
1426                 if (mf1.size == mf2.size &&
1427                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1428                         goto free_ab_and_return;
1429                 if (DIFF_OPT_TST(o, BINARY))
1430                         emit_binary_diff(o->file, &mf1, &mf2);
1431                 else
1432                         fprintf(o->file, "Binary files %s and %s differ\n",
1433                                 lbl[0], lbl[1]);
1434                 o->found_changes = 1;
1435         }
1436         else {
1437                 /* Crazy xdl interfaces.. */
1438                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1439                 xpparam_t xpp;
1440                 xdemitconf_t xecfg;
1441                 xdemitcb_t ecb;
1442                 struct emit_callback ecbdata;
1443                 const struct userdiff_funcname *pe;
1444
1445                 if (textconv_one) {
1446                         size_t size;
1447                         mf1.ptr = run_textconv(textconv_one, one, &size);
1448                         if (!mf1.ptr)
1449                                 die("unable to read files to diff");
1450                         mf1.size = size;
1451                 }
1452                 if (textconv_two) {
1453                         size_t size;
1454                         mf2.ptr = run_textconv(textconv_two, two, &size);
1455                         if (!mf2.ptr)
1456                                 die("unable to read files to diff");
1457                         mf2.size = size;
1458                 }
1459
1460                 pe = diff_funcname_pattern(one);
1461                 if (!pe)
1462                         pe = diff_funcname_pattern(two);
1463
1464                 memset(&xpp, 0, sizeof(xpp));
1465                 memset(&xecfg, 0, sizeof(xecfg));
1466                 memset(&ecbdata, 0, sizeof(ecbdata));
1467                 ecbdata.label_path = lbl;
1468                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1469                 ecbdata.found_changesp = &o->found_changes;
1470                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1471                 ecbdata.file = o->file;
1472                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1473                 xecfg.ctxlen = o->context;
1474                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1475                 if (pe)
1476                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1477                 if (!diffopts)
1478                         ;
1479                 else if (!prefixcmp(diffopts, "--unified="))
1480                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1481                 else if (!prefixcmp(diffopts, "-u"))
1482                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1483                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1484                         ecbdata.diff_words =
1485                                 xcalloc(1, sizeof(struct diff_words_data));
1486                         ecbdata.diff_words->file = o->file;
1487                 }
1488                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1489                               &xpp, &xecfg, &ecb);
1490                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1491                         free_diff_words_data(&ecbdata);
1492                 if (textconv_one)
1493                         free(mf1.ptr);
1494                 if (textconv_two)
1495                         free(mf2.ptr);
1496         }
1497
1498  free_ab_and_return:
1499         diff_free_filespec_data(one);
1500         diff_free_filespec_data(two);
1501         free(a_one);
1502         free(b_two);
1503         return;
1504 }
1505
1506 static void builtin_diffstat(const char *name_a, const char *name_b,
1507                              struct diff_filespec *one,
1508                              struct diff_filespec *two,
1509                              struct diffstat_t *diffstat,
1510                              struct diff_options *o,
1511                              int complete_rewrite)
1512 {
1513         mmfile_t mf1, mf2;
1514         struct diffstat_file *data;
1515
1516         data = diffstat_add(diffstat, name_a, name_b);
1517
1518         if (!one || !two) {
1519                 data->is_unmerged = 1;
1520                 return;
1521         }
1522         if (complete_rewrite) {
1523                 diff_populate_filespec(one, 0);
1524                 diff_populate_filespec(two, 0);
1525                 data->deleted = count_lines(one->data, one->size);
1526                 data->added = count_lines(two->data, two->size);
1527                 goto free_and_return;
1528         }
1529         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1530                 die("unable to read files to diff");
1531
1532         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1533                 data->is_binary = 1;
1534                 data->added = mf2.size;
1535                 data->deleted = mf1.size;
1536         } else {
1537                 /* Crazy xdl interfaces.. */
1538                 xpparam_t xpp;
1539                 xdemitconf_t xecfg;
1540                 xdemitcb_t ecb;
1541
1542                 memset(&xpp, 0, sizeof(xpp));
1543                 memset(&xecfg, 0, sizeof(xecfg));
1544                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1545                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1546                               &xpp, &xecfg, &ecb);
1547         }
1548
1549  free_and_return:
1550         diff_free_filespec_data(one);
1551         diff_free_filespec_data(two);
1552 }
1553
1554 static void builtin_checkdiff(const char *name_a, const char *name_b,
1555                               const char *attr_path,
1556                               struct diff_filespec *one,
1557                               struct diff_filespec *two,
1558                               struct diff_options *o)
1559 {
1560         mmfile_t mf1, mf2;
1561         struct checkdiff_t data;
1562
1563         if (!two)
1564                 return;
1565
1566         memset(&data, 0, sizeof(data));
1567         data.filename = name_b ? name_b : name_a;
1568         data.lineno = 0;
1569         data.o = o;
1570         data.ws_rule = whitespace_rule(attr_path);
1571
1572         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1573                 die("unable to read files to diff");
1574
1575         /*
1576          * All the other codepaths check both sides, but not checking
1577          * the "old" side here is deliberate.  We are checking the newly
1578          * introduced changes, and as long as the "new" side is text, we
1579          * can and should check what it introduces.
1580          */
1581         if (diff_filespec_is_binary(two))
1582                 goto free_and_return;
1583         else {
1584                 /* Crazy xdl interfaces.. */
1585                 xpparam_t xpp;
1586                 xdemitconf_t xecfg;
1587                 xdemitcb_t ecb;
1588
1589                 memset(&xpp, 0, sizeof(xpp));
1590                 memset(&xecfg, 0, sizeof(xecfg));
1591                 xecfg.ctxlen = 1; /* at least one context line */
1592                 xpp.flags = XDF_NEED_MINIMAL;
1593                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1594                               &xpp, &xecfg, &ecb);
1595
1596                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1597                     data.trailing_blanks_start) {
1598                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1599                                 data.filename, data.trailing_blanks_start);
1600                         data.status = 1; /* report errors */
1601                 }
1602         }
1603  free_and_return:
1604         diff_free_filespec_data(one);
1605         diff_free_filespec_data(two);
1606         if (data.status)
1607                 DIFF_OPT_SET(o, CHECK_FAILED);
1608 }
1609
1610 struct diff_filespec *alloc_filespec(const char *path)
1611 {
1612         int namelen = strlen(path);
1613         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1614
1615         memset(spec, 0, sizeof(*spec));
1616         spec->path = (char *)(spec + 1);
1617         memcpy(spec->path, path, namelen+1);
1618         spec->count = 1;
1619         spec->is_binary = -1;
1620         return spec;
1621 }
1622
1623 void free_filespec(struct diff_filespec *spec)
1624 {
1625         if (!--spec->count) {
1626                 diff_free_filespec_data(spec);
1627                 free(spec);
1628         }
1629 }
1630
1631 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1632                    unsigned short mode)
1633 {
1634         if (mode) {
1635                 spec->mode = canon_mode(mode);
1636                 hashcpy(spec->sha1, sha1);
1637                 spec->sha1_valid = !is_null_sha1(sha1);
1638         }
1639 }
1640
1641 /*
1642  * Given a name and sha1 pair, if the index tells us the file in
1643  * the work tree has that object contents, return true, so that
1644  * prepare_temp_file() does not have to inflate and extract.
1645  */
1646 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1647 {
1648         struct cache_entry *ce;
1649         struct stat st;
1650         int pos, len;
1651
1652         /* We do not read the cache ourselves here, because the
1653          * benchmark with my previous version that always reads cache
1654          * shows that it makes things worse for diff-tree comparing
1655          * two linux-2.6 kernel trees in an already checked out work
1656          * tree.  This is because most diff-tree comparisons deal with
1657          * only a small number of files, while reading the cache is
1658          * expensive for a large project, and its cost outweighs the
1659          * savings we get by not inflating the object to a temporary
1660          * file.  Practically, this code only helps when we are used
1661          * by diff-cache --cached, which does read the cache before
1662          * calling us.
1663          */
1664         if (!active_cache)
1665                 return 0;
1666
1667         /* We want to avoid the working directory if our caller
1668          * doesn't need the data in a normal file, this system
1669          * is rather slow with its stat/open/mmap/close syscalls,
1670          * and the object is contained in a pack file.  The pack
1671          * is probably already open and will be faster to obtain
1672          * the data through than the working directory.  Loose
1673          * objects however would tend to be slower as they need
1674          * to be individually opened and inflated.
1675          */
1676         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1677                 return 0;
1678
1679         len = strlen(name);
1680         pos = cache_name_pos(name, len);
1681         if (pos < 0)
1682                 return 0;
1683         ce = active_cache[pos];
1684
1685         /*
1686          * This is not the sha1 we are looking for, or
1687          * unreusable because it is not a regular file.
1688          */
1689         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1690                 return 0;
1691
1692         /*
1693          * If ce matches the file in the work tree, we can reuse it.
1694          */
1695         if (ce_uptodate(ce) ||
1696             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1697                 return 1;
1698
1699         return 0;
1700 }
1701
1702 static int populate_from_stdin(struct diff_filespec *s)
1703 {
1704         struct strbuf buf = STRBUF_INIT;
1705         size_t size = 0;
1706
1707         if (strbuf_read(&buf, 0, 0) < 0)
1708                 return error("error while reading from stdin %s",
1709                                      strerror(errno));
1710
1711         s->should_munmap = 0;
1712         s->data = strbuf_detach(&buf, &size);
1713         s->size = size;
1714         s->should_free = 1;
1715         return 0;
1716 }
1717
1718 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1719 {
1720         int len;
1721         char *data = xmalloc(100);
1722         len = snprintf(data, 100,
1723                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1724         s->data = data;
1725         s->size = len;
1726         s->should_free = 1;
1727         if (size_only) {
1728                 s->data = NULL;
1729                 free(data);
1730         }
1731         return 0;
1732 }
1733
1734 /*
1735  * While doing rename detection and pickaxe operation, we may need to
1736  * grab the data for the blob (or file) for our own in-core comparison.
1737  * diff_filespec has data and size fields for this purpose.
1738  */
1739 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1740 {
1741         int err = 0;
1742         if (!DIFF_FILE_VALID(s))
1743                 die("internal error: asking to populate invalid file.");
1744         if (S_ISDIR(s->mode))
1745                 return -1;
1746
1747         if (s->data)
1748                 return 0;
1749
1750         if (size_only && 0 < s->size)
1751                 return 0;
1752
1753         if (S_ISGITLINK(s->mode))
1754                 return diff_populate_gitlink(s, size_only);
1755
1756         if (!s->sha1_valid ||
1757             reuse_worktree_file(s->path, s->sha1, 0)) {
1758                 struct strbuf buf = STRBUF_INIT;
1759                 struct stat st;
1760                 int fd;
1761
1762                 if (!strcmp(s->path, "-"))
1763                         return populate_from_stdin(s);
1764
1765                 if (lstat(s->path, &st) < 0) {
1766                         if (errno == ENOENT) {
1767                         err_empty:
1768                                 err = -1;
1769                         empty:
1770                                 s->data = (char *)"";
1771                                 s->size = 0;
1772                                 return err;
1773                         }
1774                 }
1775                 s->size = xsize_t(st.st_size);
1776                 if (!s->size)
1777                         goto empty;
1778                 if (S_ISLNK(st.st_mode)) {
1779                         struct strbuf sb = STRBUF_INIT;
1780
1781                         if (strbuf_readlink(&sb, s->path, s->size))
1782                                 goto err_empty;
1783                         s->size = sb.len;
1784                         s->data = strbuf_detach(&sb, NULL);
1785                         s->should_free = 1;
1786                         return 0;
1787                 }
1788                 if (size_only)
1789                         return 0;
1790                 fd = open(s->path, O_RDONLY);
1791                 if (fd < 0)
1792                         goto err_empty;
1793                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1794                 close(fd);
1795                 s->should_munmap = 1;
1796
1797                 /*
1798                  * Convert from working tree format to canonical git format
1799                  */
1800                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1801                         size_t size = 0;
1802                         munmap(s->data, s->size);
1803                         s->should_munmap = 0;
1804                         s->data = strbuf_detach(&buf, &size);
1805                         s->size = size;
1806                         s->should_free = 1;
1807                 }
1808         }
1809         else {
1810                 enum object_type type;
1811                 if (size_only)
1812                         type = sha1_object_info(s->sha1, &s->size);
1813                 else {
1814                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1815                         s->should_free = 1;
1816                 }
1817         }
1818         return 0;
1819 }
1820
1821 void diff_free_filespec_blob(struct diff_filespec *s)
1822 {
1823         if (s->should_free)
1824                 free(s->data);
1825         else if (s->should_munmap)
1826                 munmap(s->data, s->size);
1827
1828         if (s->should_free || s->should_munmap) {
1829                 s->should_free = s->should_munmap = 0;
1830                 s->data = NULL;
1831         }
1832 }
1833
1834 void diff_free_filespec_data(struct diff_filespec *s)
1835 {
1836         diff_free_filespec_blob(s);
1837         free(s->cnt_data);
1838         s->cnt_data = NULL;
1839 }
1840
1841 static void prep_temp_blob(struct diff_tempfile *temp,
1842                            void *blob,
1843                            unsigned long size,
1844                            const unsigned char *sha1,
1845                            int mode)
1846 {
1847         int fd;
1848
1849         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1850         if (fd < 0)
1851                 die("unable to create temp-file: %s", strerror(errno));
1852         if (write_in_full(fd, blob, size) != size)
1853                 die("unable to write temp-file");
1854         close(fd);
1855         temp->name = temp->tmp_path;
1856         strcpy(temp->hex, sha1_to_hex(sha1));
1857         temp->hex[40] = 0;
1858         sprintf(temp->mode, "%06o", mode);
1859 }
1860
1861 static void prepare_temp_file(const char *name,
1862                               struct diff_tempfile *temp,
1863                               struct diff_filespec *one)
1864 {
1865         if (!DIFF_FILE_VALID(one)) {
1866         not_a_valid_file:
1867                 /* A '-' entry produces this for file-2, and
1868                  * a '+' entry produces this for file-1.
1869                  */
1870                 temp->name = "/dev/null";
1871                 strcpy(temp->hex, ".");
1872                 strcpy(temp->mode, ".");
1873                 return;
1874         }
1875
1876         if (!one->sha1_valid ||
1877             reuse_worktree_file(name, one->sha1, 1)) {
1878                 struct stat st;
1879                 if (lstat(name, &st) < 0) {
1880                         if (errno == ENOENT)
1881                                 goto not_a_valid_file;
1882                         die("stat(%s): %s", name, strerror(errno));
1883                 }
1884                 if (S_ISLNK(st.st_mode)) {
1885                         int ret;
1886                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1887                         ret = readlink(name, buf, sizeof(buf));
1888                         if (ret < 0)
1889                                 die("readlink(%s)", name);
1890                         if (ret == sizeof(buf))
1891                                 die("symlink too long: %s", name);
1892                         prep_temp_blob(temp, buf, ret,
1893                                        (one->sha1_valid ?
1894                                         one->sha1 : null_sha1),
1895                                        (one->sha1_valid ?
1896                                         one->mode : S_IFLNK));
1897                 }
1898                 else {
1899                         /* we can borrow from the file in the work tree */
1900                         temp->name = name;
1901                         if (!one->sha1_valid)
1902                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1903                         else
1904                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1905                         /* Even though we may sometimes borrow the
1906                          * contents from the work tree, we always want
1907                          * one->mode.  mode is trustworthy even when
1908                          * !(one->sha1_valid), as long as
1909                          * DIFF_FILE_VALID(one).
1910                          */
1911                         sprintf(temp->mode, "%06o", one->mode);
1912                 }
1913                 return;
1914         }
1915         else {
1916                 if (diff_populate_filespec(one, 0))
1917                         die("cannot read data blob for %s", one->path);
1918                 prep_temp_blob(temp, one->data, one->size,
1919                                one->sha1, one->mode);
1920         }
1921 }
1922
1923 static void remove_tempfile(void)
1924 {
1925         int i;
1926
1927         for (i = 0; i < 2; i++)
1928                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1929                         unlink(diff_temp[i].name);
1930                         diff_temp[i].name = NULL;
1931                 }
1932 }
1933
1934 static void remove_tempfile_on_signal(int signo)
1935 {
1936         remove_tempfile();
1937         signal(SIGINT, SIG_DFL);
1938         raise(signo);
1939 }
1940
1941 /* An external diff command takes:
1942  *
1943  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1944  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1945  *
1946  */
1947 static void run_external_diff(const char *pgm,
1948                               const char *name,
1949                               const char *other,
1950                               struct diff_filespec *one,
1951                               struct diff_filespec *two,
1952                               const char *xfrm_msg,
1953                               int complete_rewrite)
1954 {
1955         const char *spawn_arg[10];
1956         struct diff_tempfile *temp = diff_temp;
1957         int retval;
1958         static int atexit_asked = 0;
1959         const char *othername;
1960         const char **arg = &spawn_arg[0];
1961
1962         othername = (other? other : name);
1963         if (one && two) {
1964                 prepare_temp_file(name, &temp[0], one);
1965                 prepare_temp_file(othername, &temp[1], two);
1966                 if (! atexit_asked &&
1967                     (temp[0].name == temp[0].tmp_path ||
1968                      temp[1].name == temp[1].tmp_path)) {
1969                         atexit_asked = 1;
1970                         atexit(remove_tempfile);
1971                 }
1972                 signal(SIGINT, remove_tempfile_on_signal);
1973         }
1974
1975         if (one && two) {
1976                 *arg++ = pgm;
1977                 *arg++ = name;
1978                 *arg++ = temp[0].name;
1979                 *arg++ = temp[0].hex;
1980                 *arg++ = temp[0].mode;
1981                 *arg++ = temp[1].name;
1982                 *arg++ = temp[1].hex;
1983                 *arg++ = temp[1].mode;
1984                 if (other) {
1985                         *arg++ = other;
1986                         *arg++ = xfrm_msg;
1987                 }
1988         } else {
1989                 *arg++ = pgm;
1990                 *arg++ = name;
1991         }
1992         *arg = NULL;
1993         fflush(NULL);
1994         retval = run_command_v_opt(spawn_arg, 0);
1995         remove_tempfile();
1996         if (retval) {
1997                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1998                 exit(1);
1999         }
2000 }
2001
2002 static void run_diff_cmd(const char *pgm,
2003                          const char *name,
2004                          const char *other,
2005                          const char *attr_path,
2006                          struct diff_filespec *one,
2007                          struct diff_filespec *two,
2008                          const char *xfrm_msg,
2009                          struct diff_options *o,
2010                          int complete_rewrite)
2011 {
2012         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2013                 pgm = NULL;
2014         else {
2015                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2016                 if (drv && drv->external)
2017                         pgm = drv->external;
2018         }
2019
2020         if (pgm) {
2021                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2022                                   complete_rewrite);
2023                 return;
2024         }
2025         if (one && two)
2026                 builtin_diff(name, other ? other : name,
2027                              one, two, xfrm_msg, o, complete_rewrite);
2028         else
2029                 fprintf(o->file, "* Unmerged path %s\n", name);
2030 }
2031
2032 static void diff_fill_sha1_info(struct diff_filespec *one)
2033 {
2034         if (DIFF_FILE_VALID(one)) {
2035                 if (!one->sha1_valid) {
2036                         struct stat st;
2037                         if (!strcmp(one->path, "-")) {
2038                                 hashcpy(one->sha1, null_sha1);
2039                                 return;
2040                         }
2041                         if (lstat(one->path, &st) < 0)
2042                                 die("stat %s", one->path);
2043                         if (index_path(one->sha1, one->path, &st, 0))
2044                                 die("cannot hash %s\n", one->path);
2045                 }
2046         }
2047         else
2048                 hashclr(one->sha1);
2049 }
2050
2051 static int similarity_index(struct diff_filepair *p)
2052 {
2053         return p->score * 100 / MAX_SCORE;
2054 }
2055
2056 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2057 {
2058         /* Strip the prefix but do not molest /dev/null and absolute paths */
2059         if (*namep && **namep != '/')
2060                 *namep += prefix_length;
2061         if (*otherp && **otherp != '/')
2062                 *otherp += prefix_length;
2063 }
2064
2065 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2066 {
2067         const char *pgm = external_diff();
2068         struct strbuf msg;
2069         char *xfrm_msg;
2070         struct diff_filespec *one = p->one;
2071         struct diff_filespec *two = p->two;
2072         const char *name;
2073         const char *other;
2074         const char *attr_path;
2075         int complete_rewrite = 0;
2076
2077         name  = p->one->path;
2078         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2079         attr_path = name;
2080         if (o->prefix_length)
2081                 strip_prefix(o->prefix_length, &name, &other);
2082
2083         if (DIFF_PAIR_UNMERGED(p)) {
2084                 run_diff_cmd(pgm, name, NULL, attr_path,
2085                              NULL, NULL, NULL, o, 0);
2086                 return;
2087         }
2088
2089         diff_fill_sha1_info(one);
2090         diff_fill_sha1_info(two);
2091
2092         strbuf_init(&msg, PATH_MAX * 2 + 300);
2093         switch (p->status) {
2094         case DIFF_STATUS_COPIED:
2095                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2096                 strbuf_addstr(&msg, "\ncopy from ");
2097                 quote_c_style(name, &msg, NULL, 0);
2098                 strbuf_addstr(&msg, "\ncopy to ");
2099                 quote_c_style(other, &msg, NULL, 0);
2100                 strbuf_addch(&msg, '\n');
2101                 break;
2102         case DIFF_STATUS_RENAMED:
2103                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2104                 strbuf_addstr(&msg, "\nrename from ");
2105                 quote_c_style(name, &msg, NULL, 0);
2106                 strbuf_addstr(&msg, "\nrename to ");
2107                 quote_c_style(other, &msg, NULL, 0);
2108                 strbuf_addch(&msg, '\n');
2109                 break;
2110         case DIFF_STATUS_MODIFIED:
2111                 if (p->score) {
2112                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2113                                         similarity_index(p));
2114                         complete_rewrite = 1;
2115                         break;
2116                 }
2117                 /* fallthru */
2118         default:
2119                 /* nothing */
2120                 ;
2121         }
2122
2123         if (hashcmp(one->sha1, two->sha1)) {
2124                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2125
2126                 if (DIFF_OPT_TST(o, BINARY)) {
2127                         mmfile_t mf;
2128                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2129                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2130                                 abbrev = 40;
2131                 }
2132                 strbuf_addf(&msg, "index %.*s..%.*s",
2133                                 abbrev, sha1_to_hex(one->sha1),
2134                                 abbrev, sha1_to_hex(two->sha1));
2135                 if (one->mode == two->mode)
2136                         strbuf_addf(&msg, " %06o", one->mode);
2137                 strbuf_addch(&msg, '\n');
2138         }
2139
2140         if (msg.len)
2141                 strbuf_setlen(&msg, msg.len - 1);
2142         xfrm_msg = msg.len ? msg.buf : NULL;
2143
2144         if (!pgm &&
2145             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2146             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2147                 /* a filepair that changes between file and symlink
2148                  * needs to be split into deletion and creation.
2149                  */
2150                 struct diff_filespec *null = alloc_filespec(two->path);
2151                 run_diff_cmd(NULL, name, other, attr_path,
2152                              one, null, xfrm_msg, o, 0);
2153                 free(null);
2154                 null = alloc_filespec(one->path);
2155                 run_diff_cmd(NULL, name, other, attr_path,
2156                              null, two, xfrm_msg, o, 0);
2157                 free(null);
2158         }
2159         else
2160                 run_diff_cmd(pgm, name, other, attr_path,
2161                              one, two, xfrm_msg, o, complete_rewrite);
2162
2163         strbuf_release(&msg);
2164 }
2165
2166 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2167                          struct diffstat_t *diffstat)
2168 {
2169         const char *name;
2170         const char *other;
2171         int complete_rewrite = 0;
2172
2173         if (DIFF_PAIR_UNMERGED(p)) {
2174                 /* unmerged */
2175                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2176                 return;
2177         }
2178
2179         name = p->one->path;
2180         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2181
2182         if (o->prefix_length)
2183                 strip_prefix(o->prefix_length, &name, &other);
2184
2185         diff_fill_sha1_info(p->one);
2186         diff_fill_sha1_info(p->two);
2187
2188         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2189                 complete_rewrite = 1;
2190         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2191 }
2192
2193 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2194 {
2195         const char *name;
2196         const char *other;
2197         const char *attr_path;
2198
2199         if (DIFF_PAIR_UNMERGED(p)) {
2200                 /* unmerged */
2201                 return;
2202         }
2203
2204         name = p->one->path;
2205         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2206         attr_path = other ? other : name;
2207
2208         if (o->prefix_length)
2209                 strip_prefix(o->prefix_length, &name, &other);
2210
2211         diff_fill_sha1_info(p->one);
2212         diff_fill_sha1_info(p->two);
2213
2214         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2215 }
2216
2217 void diff_setup(struct diff_options *options)
2218 {
2219         memset(options, 0, sizeof(*options));
2220
2221         options->file = stdout;
2222
2223         options->line_termination = '\n';
2224         options->break_opt = -1;
2225         options->rename_limit = -1;
2226         options->dirstat_percent = 3;
2227         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2228         options->context = 3;
2229
2230         options->change = diff_change;
2231         options->add_remove = diff_addremove;
2232         if (diff_use_color_default > 0)
2233                 DIFF_OPT_SET(options, COLOR_DIFF);
2234         else
2235                 DIFF_OPT_CLR(options, COLOR_DIFF);
2236         options->detect_rename = diff_detect_rename_default;
2237
2238         if (!diff_mnemonic_prefix) {
2239                 options->a_prefix = "a/";
2240                 options->b_prefix = "b/";
2241         }
2242 }
2243
2244 int diff_setup_done(struct diff_options *options)
2245 {
2246         int count = 0;
2247
2248         if (options->output_format & DIFF_FORMAT_NAME)
2249                 count++;
2250         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2251                 count++;
2252         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2253                 count++;
2254         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2255                 count++;
2256         if (count > 1)
2257                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2258
2259         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2260                 options->detect_rename = DIFF_DETECT_COPY;
2261
2262         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2263                 options->prefix = NULL;
2264         if (options->prefix)
2265                 options->prefix_length = strlen(options->prefix);
2266         else
2267                 options->prefix_length = 0;
2268
2269         if (options->output_format & (DIFF_FORMAT_NAME |
2270                                       DIFF_FORMAT_NAME_STATUS |
2271                                       DIFF_FORMAT_CHECKDIFF |
2272                                       DIFF_FORMAT_NO_OUTPUT))
2273                 options->output_format &= ~(DIFF_FORMAT_RAW |
2274                                             DIFF_FORMAT_NUMSTAT |
2275                                             DIFF_FORMAT_DIFFSTAT |
2276                                             DIFF_FORMAT_SHORTSTAT |
2277                                             DIFF_FORMAT_DIRSTAT |
2278                                             DIFF_FORMAT_SUMMARY |
2279                                             DIFF_FORMAT_PATCH);
2280
2281         /*
2282          * These cases always need recursive; we do not drop caller-supplied
2283          * recursive bits for other formats here.
2284          */
2285         if (options->output_format & (DIFF_FORMAT_PATCH |
2286                                       DIFF_FORMAT_NUMSTAT |
2287                                       DIFF_FORMAT_DIFFSTAT |
2288                                       DIFF_FORMAT_SHORTSTAT |
2289                                       DIFF_FORMAT_DIRSTAT |
2290                                       DIFF_FORMAT_SUMMARY |
2291                                       DIFF_FORMAT_CHECKDIFF))
2292                 DIFF_OPT_SET(options, RECURSIVE);
2293         /*
2294          * Also pickaxe would not work very well if you do not say recursive
2295          */
2296         if (options->pickaxe)
2297                 DIFF_OPT_SET(options, RECURSIVE);
2298
2299         if (options->detect_rename && options->rename_limit < 0)
2300                 options->rename_limit = diff_rename_limit_default;
2301         if (options->setup & DIFF_SETUP_USE_CACHE) {
2302                 if (!active_cache)
2303                         /* read-cache does not die even when it fails
2304                          * so it is safe for us to do this here.  Also
2305                          * it does not smudge active_cache or active_nr
2306                          * when it fails, so we do not have to worry about
2307                          * cleaning it up ourselves either.
2308                          */
2309                         read_cache();
2310         }
2311         if (options->abbrev <= 0 || 40 < options->abbrev)
2312                 options->abbrev = 40; /* full */
2313
2314         /*
2315          * It does not make sense to show the first hit we happened
2316          * to have found.  It does not make sense not to return with
2317          * exit code in such a case either.
2318          */
2319         if (DIFF_OPT_TST(options, QUIET)) {
2320                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2321                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2322         }
2323
2324         return 0;
2325 }
2326
2327 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2328 {
2329         char c, *eq;
2330         int len;
2331
2332         if (*arg != '-')
2333                 return 0;
2334         c = *++arg;
2335         if (!c)
2336                 return 0;
2337         if (c == arg_short) {
2338                 c = *++arg;
2339                 if (!c)
2340                         return 1;
2341                 if (val && isdigit(c)) {
2342                         char *end;
2343                         int n = strtoul(arg, &end, 10);
2344                         if (*end)
2345                                 return 0;
2346                         *val = n;
2347                         return 1;
2348                 }
2349                 return 0;
2350         }
2351         if (c != '-')
2352                 return 0;
2353         arg++;
2354         eq = strchr(arg, '=');
2355         if (eq)
2356                 len = eq - arg;
2357         else
2358                 len = strlen(arg);
2359         if (!len || strncmp(arg, arg_long, len))
2360                 return 0;
2361         if (eq) {
2362                 int n;
2363                 char *end;
2364                 if (!isdigit(*++eq))
2365                         return 0;
2366                 n = strtoul(eq, &end, 10);
2367                 if (*end)
2368                         return 0;
2369                 *val = n;
2370         }
2371         return 1;
2372 }
2373
2374 static int diff_scoreopt_parse(const char *opt);
2375
2376 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2377 {
2378         const char *arg = av[0];
2379
2380         /* Output format options */
2381         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2382                 options->output_format |= DIFF_FORMAT_PATCH;
2383         else if (opt_arg(arg, 'U', "unified", &options->context))
2384                 options->output_format |= DIFF_FORMAT_PATCH;
2385         else if (!strcmp(arg, "--raw"))
2386                 options->output_format |= DIFF_FORMAT_RAW;
2387         else if (!strcmp(arg, "--patch-with-raw"))
2388                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2389         else if (!strcmp(arg, "--numstat"))
2390                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2391         else if (!strcmp(arg, "--shortstat"))
2392                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2393         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2394                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2395         else if (!strcmp(arg, "--cumulative")) {
2396                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2397                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2398         } else if (opt_arg(arg, 0, "dirstat-by-file",
2399                            &options->dirstat_percent)) {
2400                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2401                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2402         }
2403         else if (!strcmp(arg, "--check"))
2404                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2405         else if (!strcmp(arg, "--summary"))
2406                 options->output_format |= DIFF_FORMAT_SUMMARY;
2407         else if (!strcmp(arg, "--patch-with-stat"))
2408                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2409         else if (!strcmp(arg, "--name-only"))
2410                 options->output_format |= DIFF_FORMAT_NAME;
2411         else if (!strcmp(arg, "--name-status"))
2412                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2413         else if (!strcmp(arg, "-s"))
2414                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2415         else if (!prefixcmp(arg, "--stat")) {
2416                 char *end;
2417                 int width = options->stat_width;
2418                 int name_width = options->stat_name_width;
2419                 arg += 6;
2420                 end = (char *)arg;
2421
2422                 switch (*arg) {
2423                 case '-':
2424                         if (!prefixcmp(arg, "-width="))
2425                                 width = strtoul(arg + 7, &end, 10);
2426                         else if (!prefixcmp(arg, "-name-width="))
2427                                 name_width = strtoul(arg + 12, &end, 10);
2428                         break;
2429                 case '=':
2430                         width = strtoul(arg+1, &end, 10);
2431                         if (*end == ',')
2432                                 name_width = strtoul(end+1, &end, 10);
2433                 }
2434
2435                 /* Important! This checks all the error cases! */
2436                 if (*end)
2437                         return 0;
2438                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2439                 options->stat_name_width = name_width;
2440                 options->stat_width = width;
2441         }
2442
2443         /* renames options */
2444         else if (!prefixcmp(arg, "-B")) {
2445                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2446                         return -1;
2447         }
2448         else if (!prefixcmp(arg, "-M")) {
2449                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2450                         return -1;
2451                 options->detect_rename = DIFF_DETECT_RENAME;
2452         }
2453         else if (!prefixcmp(arg, "-C")) {
2454                 if (options->detect_rename == DIFF_DETECT_COPY)
2455                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2456                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2457                         return -1;
2458                 options->detect_rename = DIFF_DETECT_COPY;
2459         }
2460         else if (!strcmp(arg, "--no-renames"))
2461                 options->detect_rename = 0;
2462         else if (!strcmp(arg, "--relative"))
2463                 DIFF_OPT_SET(options, RELATIVE_NAME);
2464         else if (!prefixcmp(arg, "--relative=")) {
2465                 DIFF_OPT_SET(options, RELATIVE_NAME);
2466                 options->prefix = arg + 11;
2467         }
2468
2469         /* xdiff options */
2470         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2471                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2472         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2473                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2474         else if (!strcmp(arg, "--ignore-space-at-eol"))
2475                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2476
2477         /* flags options */
2478         else if (!strcmp(arg, "--binary")) {
2479                 options->output_format |= DIFF_FORMAT_PATCH;
2480                 DIFF_OPT_SET(options, BINARY);
2481         }
2482         else if (!strcmp(arg, "--full-index"))
2483                 DIFF_OPT_SET(options, FULL_INDEX);
2484         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2485                 DIFF_OPT_SET(options, TEXT);
2486         else if (!strcmp(arg, "-R"))
2487                 DIFF_OPT_SET(options, REVERSE_DIFF);
2488         else if (!strcmp(arg, "--find-copies-harder"))
2489                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2490         else if (!strcmp(arg, "--follow"))
2491                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2492         else if (!strcmp(arg, "--color"))
2493                 DIFF_OPT_SET(options, COLOR_DIFF);
2494         else if (!strcmp(arg, "--no-color"))
2495                 DIFF_OPT_CLR(options, COLOR_DIFF);
2496         else if (!strcmp(arg, "--color-words"))
2497                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2498         else if (!strcmp(arg, "--exit-code"))
2499                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2500         else if (!strcmp(arg, "--quiet"))
2501                 DIFF_OPT_SET(options, QUIET);
2502         else if (!strcmp(arg, "--ext-diff"))
2503                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2504         else if (!strcmp(arg, "--no-ext-diff"))
2505                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2506         else if (!strcmp(arg, "--textconv"))
2507                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2508         else if (!strcmp(arg, "--no-textconv"))
2509                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2510         else if (!strcmp(arg, "--ignore-submodules"))
2511                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2512
2513         /* misc options */
2514         else if (!strcmp(arg, "-z"))
2515                 options->line_termination = 0;
2516         else if (!prefixcmp(arg, "-l"))
2517                 options->rename_limit = strtoul(arg+2, NULL, 10);
2518         else if (!prefixcmp(arg, "-S"))
2519                 options->pickaxe = arg + 2;
2520         else if (!strcmp(arg, "--pickaxe-all"))
2521                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2522         else if (!strcmp(arg, "--pickaxe-regex"))
2523                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2524         else if (!prefixcmp(arg, "-O"))
2525                 options->orderfile = arg + 2;
2526         else if (!prefixcmp(arg, "--diff-filter="))
2527                 options->filter = arg + 14;
2528         else if (!strcmp(arg, "--abbrev"))
2529                 options->abbrev = DEFAULT_ABBREV;
2530         else if (!prefixcmp(arg, "--abbrev=")) {
2531                 options->abbrev = strtoul(arg + 9, NULL, 10);
2532                 if (options->abbrev < MINIMUM_ABBREV)
2533                         options->abbrev = MINIMUM_ABBREV;
2534                 else if (40 < options->abbrev)
2535                         options->abbrev = 40;
2536         }
2537         else if (!prefixcmp(arg, "--src-prefix="))
2538                 options->a_prefix = arg + 13;
2539         else if (!prefixcmp(arg, "--dst-prefix="))
2540                 options->b_prefix = arg + 13;
2541         else if (!strcmp(arg, "--no-prefix"))
2542                 options->a_prefix = options->b_prefix = "";
2543         else if (!prefixcmp(arg, "--output=")) {
2544                 options->file = fopen(arg + strlen("--output="), "w");
2545                 options->close_file = 1;
2546         } else
2547                 return 0;
2548         return 1;
2549 }
2550
2551 static int parse_num(const char **cp_p)
2552 {
2553         unsigned long num, scale;
2554         int ch, dot;
2555         const char *cp = *cp_p;
2556
2557         num = 0;
2558         scale = 1;
2559         dot = 0;
2560         for(;;) {
2561                 ch = *cp;
2562                 if ( !dot && ch == '.' ) {
2563                         scale = 1;
2564                         dot = 1;
2565                 } else if ( ch == '%' ) {
2566                         scale = dot ? scale*100 : 100;
2567                         cp++;   /* % is always at the end */
2568                         break;
2569                 } else if ( ch >= '0' && ch <= '9' ) {
2570                         if ( scale < 100000 ) {
2571                                 scale *= 10;
2572                                 num = (num*10) + (ch-'0');
2573                         }
2574                 } else {
2575                         break;
2576                 }
2577                 cp++;
2578         }
2579         *cp_p = cp;
2580
2581         /* user says num divided by scale and we say internally that
2582          * is MAX_SCORE * num / scale.
2583          */
2584         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2585 }
2586
2587 static int diff_scoreopt_parse(const char *opt)
2588 {
2589         int opt1, opt2, cmd;
2590
2591         if (*opt++ != '-')
2592                 return -1;
2593         cmd = *opt++;
2594         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2595                 return -1; /* that is not a -M, -C nor -B option */
2596
2597         opt1 = parse_num(&opt);
2598         if (cmd != 'B')
2599                 opt2 = 0;
2600         else {
2601                 if (*opt == 0)
2602                         opt2 = 0;
2603                 else if (*opt != '/')
2604                         return -1; /* we expect -B80/99 or -B80 */
2605                 else {
2606                         opt++;
2607                         opt2 = parse_num(&opt);
2608                 }
2609         }
2610         if (*opt != 0)
2611                 return -1;
2612         return opt1 | (opt2 << 16);
2613 }
2614
2615 struct diff_queue_struct diff_queued_diff;
2616
2617 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2618 {
2619         if (queue->alloc <= queue->nr) {
2620                 queue->alloc = alloc_nr(queue->alloc);
2621                 queue->queue = xrealloc(queue->queue,
2622                                         sizeof(dp) * queue->alloc);
2623         }
2624         queue->queue[queue->nr++] = dp;
2625 }
2626
2627 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2628                                  struct diff_filespec *one,
2629                                  struct diff_filespec *two)
2630 {
2631         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2632         dp->one = one;
2633         dp->two = two;
2634         if (queue)
2635                 diff_q(queue, dp);
2636         return dp;
2637 }
2638
2639 void diff_free_filepair(struct diff_filepair *p)
2640 {
2641         free_filespec(p->one);
2642         free_filespec(p->two);
2643         free(p);
2644 }
2645
2646 /* This is different from find_unique_abbrev() in that
2647  * it stuffs the result with dots for alignment.
2648  */
2649 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2650 {
2651         int abblen;
2652         const char *abbrev;
2653         if (len == 40)
2654                 return sha1_to_hex(sha1);
2655
2656         abbrev = find_unique_abbrev(sha1, len);
2657         abblen = strlen(abbrev);
2658         if (abblen < 37) {
2659                 static char hex[41];
2660                 if (len < abblen && abblen <= len + 2)
2661                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2662                 else
2663                         sprintf(hex, "%s...", abbrev);
2664                 return hex;
2665         }
2666         return sha1_to_hex(sha1);
2667 }
2668
2669 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2670 {
2671         int line_termination = opt->line_termination;
2672         int inter_name_termination = line_termination ? '\t' : '\0';
2673
2674         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2675                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2676                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2677                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2678         }
2679         if (p->score) {
2680                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2681                         inter_name_termination);
2682         } else {
2683                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2684         }
2685
2686         if (p->status == DIFF_STATUS_COPIED ||
2687             p->status == DIFF_STATUS_RENAMED) {
2688                 const char *name_a, *name_b;
2689                 name_a = p->one->path;
2690                 name_b = p->two->path;
2691                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2692                 write_name_quoted(name_a, opt->file, inter_name_termination);
2693                 write_name_quoted(name_b, opt->file, line_termination);
2694         } else {
2695                 const char *name_a, *name_b;
2696                 name_a = p->one->mode ? p->one->path : p->two->path;
2697                 name_b = NULL;
2698                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2699                 write_name_quoted(name_a, opt->file, line_termination);
2700         }
2701 }
2702
2703 int diff_unmodified_pair(struct diff_filepair *p)
2704 {
2705         /* This function is written stricter than necessary to support
2706          * the currently implemented transformers, but the idea is to
2707          * let transformers to produce diff_filepairs any way they want,
2708          * and filter and clean them up here before producing the output.
2709          */
2710         struct diff_filespec *one = p->one, *two = p->two;
2711
2712         if (DIFF_PAIR_UNMERGED(p))
2713                 return 0; /* unmerged is interesting */
2714
2715         /* deletion, addition, mode or type change
2716          * and rename are all interesting.
2717          */
2718         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2719             DIFF_PAIR_MODE_CHANGED(p) ||
2720             strcmp(one->path, two->path))
2721                 return 0;
2722
2723         /* both are valid and point at the same path.  that is, we are
2724          * dealing with a change.
2725          */
2726         if (one->sha1_valid && two->sha1_valid &&
2727             !hashcmp(one->sha1, two->sha1))
2728                 return 1; /* no change */
2729         if (!one->sha1_valid && !two->sha1_valid)
2730                 return 1; /* both look at the same file on the filesystem. */
2731         return 0;
2732 }
2733
2734 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2735 {
2736         if (diff_unmodified_pair(p))
2737                 return;
2738
2739         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2740             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2741                 return; /* no tree diffs in patch format */
2742
2743         run_diff(p, o);
2744 }
2745
2746 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2747                             struct diffstat_t *diffstat)
2748 {
2749         if (diff_unmodified_pair(p))
2750                 return;
2751
2752         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2753             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2754                 return; /* no tree diffs in patch format */
2755
2756         run_diffstat(p, o, diffstat);
2757 }
2758
2759 static void diff_flush_checkdiff(struct diff_filepair *p,
2760                 struct diff_options *o)
2761 {
2762         if (diff_unmodified_pair(p))
2763                 return;
2764
2765         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2766             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2767                 return; /* no tree diffs in patch format */
2768
2769         run_checkdiff(p, o);
2770 }
2771
2772 int diff_queue_is_empty(void)
2773 {
2774         struct diff_queue_struct *q = &diff_queued_diff;
2775         int i;
2776         for (i = 0; i < q->nr; i++)
2777                 if (!diff_unmodified_pair(q->queue[i]))
2778                         return 0;
2779         return 1;
2780 }
2781
2782 #if DIFF_DEBUG
2783 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2784 {
2785         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2786                 x, one ? one : "",
2787                 s->path,
2788                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2789                 s->mode,
2790                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2791         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2792                 x, one ? one : "",
2793                 s->size, s->xfrm_flags);
2794 }
2795
2796 void diff_debug_filepair(const struct diff_filepair *p, int i)
2797 {
2798         diff_debug_filespec(p->one, i, "one");
2799         diff_debug_filespec(p->two, i, "two");
2800         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2801                 p->score, p->status ? p->status : '?',
2802                 p->one->rename_used, p->broken_pair);
2803 }
2804
2805 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2806 {
2807         int i;
2808         if (msg)
2809                 fprintf(stderr, "%s\n", msg);
2810         fprintf(stderr, "q->nr = %d\n", q->nr);
2811         for (i = 0; i < q->nr; i++) {
2812                 struct diff_filepair *p = q->queue[i];
2813                 diff_debug_filepair(p, i);
2814         }
2815 }
2816 #endif
2817
2818 static void diff_resolve_rename_copy(void)
2819 {
2820         int i;
2821         struct diff_filepair *p;
2822         struct diff_queue_struct *q = &diff_queued_diff;
2823
2824         diff_debug_queue("resolve-rename-copy", q);
2825
2826         for (i = 0; i < q->nr; i++) {
2827                 p = q->queue[i];
2828                 p->status = 0; /* undecided */
2829                 if (DIFF_PAIR_UNMERGED(p))
2830                         p->status = DIFF_STATUS_UNMERGED;
2831                 else if (!DIFF_FILE_VALID(p->one))
2832                         p->status = DIFF_STATUS_ADDED;
2833                 else if (!DIFF_FILE_VALID(p->two))
2834                         p->status = DIFF_STATUS_DELETED;
2835                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2836                         p->status = DIFF_STATUS_TYPE_CHANGED;
2837
2838                 /* from this point on, we are dealing with a pair
2839                  * whose both sides are valid and of the same type, i.e.
2840                  * either in-place edit or rename/copy edit.
2841                  */
2842                 else if (DIFF_PAIR_RENAME(p)) {
2843                         /*
2844                          * A rename might have re-connected a broken
2845                          * pair up, causing the pathnames to be the
2846                          * same again. If so, that's not a rename at
2847                          * all, just a modification..
2848                          *
2849                          * Otherwise, see if this source was used for
2850                          * multiple renames, in which case we decrement
2851                          * the count, and call it a copy.
2852                          */
2853                         if (!strcmp(p->one->path, p->two->path))
2854                                 p->status = DIFF_STATUS_MODIFIED;
2855                         else if (--p->one->rename_used > 0)
2856                                 p->status = DIFF_STATUS_COPIED;
2857                         else
2858                                 p->status = DIFF_STATUS_RENAMED;
2859                 }
2860                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2861                          p->one->mode != p->two->mode ||
2862                          is_null_sha1(p->one->sha1))
2863                         p->status = DIFF_STATUS_MODIFIED;
2864                 else {
2865                         /* This is a "no-change" entry and should not
2866                          * happen anymore, but prepare for broken callers.
2867                          */
2868                         error("feeding unmodified %s to diffcore",
2869                               p->one->path);
2870                         p->status = DIFF_STATUS_UNKNOWN;
2871                 }
2872         }
2873         diff_debug_queue("resolve-rename-copy done", q);
2874 }
2875
2876 static int check_pair_status(struct diff_filepair *p)
2877 {
2878         switch (p->status) {
2879         case DIFF_STATUS_UNKNOWN:
2880                 return 0;
2881         case 0:
2882                 die("internal error in diff-resolve-rename-copy");
2883         default:
2884                 return 1;
2885         }
2886 }
2887
2888 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2889 {
2890         int fmt = opt->output_format;
2891
2892         if (fmt & DIFF_FORMAT_CHECKDIFF)
2893                 diff_flush_checkdiff(p, opt);
2894         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2895                 diff_flush_raw(p, opt);
2896         else if (fmt & DIFF_FORMAT_NAME) {
2897                 const char *name_a, *name_b;
2898                 name_a = p->two->path;
2899                 name_b = NULL;
2900                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2901                 write_name_quoted(name_a, opt->file, opt->line_termination);
2902         }
2903 }
2904
2905 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2906 {
2907         if (fs->mode)
2908                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2909         else
2910                 fprintf(file, " %s ", newdelete);
2911         write_name_quoted(fs->path, file, '\n');
2912 }
2913
2914
2915 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2916 {
2917         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2918                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2919                         show_name ? ' ' : '\n');
2920                 if (show_name) {
2921                         write_name_quoted(p->two->path, file, '\n');
2922                 }
2923         }
2924 }
2925
2926 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2927 {
2928         char *names = pprint_rename(p->one->path, p->two->path);
2929
2930         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2931         free(names);
2932         show_mode_change(file, p, 0);
2933 }
2934
2935 static void diff_summary(FILE *file, struct diff_filepair *p)
2936 {
2937         switch(p->status) {
2938         case DIFF_STATUS_DELETED:
2939                 show_file_mode_name(file, "delete", p->one);
2940                 break;
2941         case DIFF_STATUS_ADDED:
2942                 show_file_mode_name(file, "create", p->two);
2943                 break;
2944         case DIFF_STATUS_COPIED:
2945                 show_rename_copy(file, "copy", p);
2946                 break;
2947         case DIFF_STATUS_RENAMED:
2948                 show_rename_copy(file, "rename", p);
2949                 break;
2950         default:
2951                 if (p->score) {
2952                         fputs(" rewrite ", file);
2953                         write_name_quoted(p->two->path, file, ' ');
2954                         fprintf(file, "(%d%%)\n", similarity_index(p));
2955                 }
2956                 show_mode_change(file, p, !p->score);
2957                 break;
2958         }
2959 }
2960
2961 struct patch_id_t {
2962         git_SHA_CTX *ctx;
2963         int patchlen;
2964 };
2965
2966 static int remove_space(char *line, int len)
2967 {
2968         int i;
2969         char *dst = line;
2970         unsigned char c;
2971
2972         for (i = 0; i < len; i++)
2973                 if (!isspace((c = line[i])))
2974                         *dst++ = c;
2975
2976         return dst - line;
2977 }
2978
2979 static void patch_id_consume(void *priv, char *line, unsigned long len)
2980 {
2981         struct patch_id_t *data = priv;
2982         int new_len;
2983
2984         /* Ignore line numbers when computing the SHA1 of the patch */
2985         if (!prefixcmp(line, "@@ -"))
2986                 return;
2987
2988         new_len = remove_space(line, len);
2989
2990         git_SHA1_Update(data->ctx, line, new_len);
2991         data->patchlen += new_len;
2992 }
2993
2994 /* returns 0 upon success, and writes result into sha1 */
2995 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2996 {
2997         struct diff_queue_struct *q = &diff_queued_diff;
2998         int i;
2999         git_SHA_CTX ctx;
3000         struct patch_id_t data;
3001         char buffer[PATH_MAX * 4 + 20];
3002
3003         git_SHA1_Init(&ctx);
3004         memset(&data, 0, sizeof(struct patch_id_t));
3005         data.ctx = &ctx;
3006
3007         for (i = 0; i < q->nr; i++) {
3008                 xpparam_t xpp;
3009                 xdemitconf_t xecfg;
3010                 xdemitcb_t ecb;
3011                 mmfile_t mf1, mf2;
3012                 struct diff_filepair *p = q->queue[i];
3013                 int len1, len2;
3014
3015                 memset(&xpp, 0, sizeof(xpp));
3016                 memset(&xecfg, 0, sizeof(xecfg));
3017                 if (p->status == 0)
3018                         return error("internal diff status error");
3019                 if (p->status == DIFF_STATUS_UNKNOWN)
3020                         continue;
3021                 if (diff_unmodified_pair(p))
3022                         continue;
3023                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3024                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3025                         continue;
3026                 if (DIFF_PAIR_UNMERGED(p))
3027                         continue;
3028
3029                 diff_fill_sha1_info(p->one);
3030                 diff_fill_sha1_info(p->two);
3031                 if (fill_mmfile(&mf1, p->one) < 0 ||
3032                                 fill_mmfile(&mf2, p->two) < 0)
3033                         return error("unable to read files to diff");
3034
3035                 len1 = remove_space(p->one->path, strlen(p->one->path));
3036                 len2 = remove_space(p->two->path, strlen(p->two->path));
3037                 if (p->one->mode == 0)
3038                         len1 = snprintf(buffer, sizeof(buffer),
3039                                         "diff--gita/%.*sb/%.*s"
3040                                         "newfilemode%06o"
3041                                         "---/dev/null"
3042                                         "+++b/%.*s",
3043                                         len1, p->one->path,
3044                                         len2, p->two->path,
3045                                         p->two->mode,
3046                                         len2, p->two->path);
3047                 else if (p->two->mode == 0)
3048                         len1 = snprintf(buffer, sizeof(buffer),
3049                                         "diff--gita/%.*sb/%.*s"
3050                                         "deletedfilemode%06o"
3051                                         "---a/%.*s"
3052                                         "+++/dev/null",
3053                                         len1, p->one->path,
3054                                         len2, p->two->path,
3055                                         p->one->mode,
3056                                         len1, p->one->path);
3057                 else
3058                         len1 = snprintf(buffer, sizeof(buffer),
3059                                         "diff--gita/%.*sb/%.*s"
3060                                         "---a/%.*s"
3061                                         "+++b/%.*s",
3062                                         len1, p->one->path,
3063                                         len2, p->two->path,
3064                                         len1, p->one->path,
3065                                         len2, p->two->path);
3066                 git_SHA1_Update(&ctx, buffer, len1);
3067
3068                 xpp.flags = XDF_NEED_MINIMAL;
3069                 xecfg.ctxlen = 3;
3070                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3071                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3072                               &xpp, &xecfg, &ecb);
3073         }
3074
3075         git_SHA1_Final(sha1, &ctx);
3076         return 0;
3077 }
3078
3079 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3080 {
3081         struct diff_queue_struct *q = &diff_queued_diff;
3082         int i;
3083         int result = diff_get_patch_id(options, sha1);
3084
3085         for (i = 0; i < q->nr; i++)
3086                 diff_free_filepair(q->queue[i]);
3087
3088         free(q->queue);
3089         q->queue = NULL;
3090         q->nr = q->alloc = 0;
3091
3092         return result;
3093 }
3094
3095 static int is_summary_empty(const struct diff_queue_struct *q)
3096 {
3097         int i;
3098
3099         for (i = 0; i < q->nr; i++) {
3100                 const struct diff_filepair *p = q->queue[i];
3101
3102                 switch (p->status) {
3103                 case DIFF_STATUS_DELETED:
3104                 case DIFF_STATUS_ADDED:
3105                 case DIFF_STATUS_COPIED:
3106                 case DIFF_STATUS_RENAMED:
3107                         return 0;
3108                 default:
3109                         if (p->score)
3110                                 return 0;
3111                         if (p->one->mode && p->two->mode &&
3112                             p->one->mode != p->two->mode)
3113                                 return 0;
3114                         break;
3115                 }
3116         }
3117         return 1;
3118 }
3119
3120 void diff_flush(struct diff_options *options)
3121 {
3122         struct diff_queue_struct *q = &diff_queued_diff;
3123         int i, output_format = options->output_format;
3124         int separator = 0;
3125
3126         /*
3127          * Order: raw, stat, summary, patch
3128          * or:    name/name-status/checkdiff (other bits clear)
3129          */
3130         if (!q->nr)
3131                 goto free_queue;
3132
3133         if (output_format & (DIFF_FORMAT_RAW |
3134                              DIFF_FORMAT_NAME |
3135                              DIFF_FORMAT_NAME_STATUS |
3136                              DIFF_FORMAT_CHECKDIFF)) {
3137                 for (i = 0; i < q->nr; i++) {
3138                         struct diff_filepair *p = q->queue[i];
3139                         if (check_pair_status(p))
3140                                 flush_one_pair(p, options);
3141                 }
3142                 separator++;
3143         }
3144
3145         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3146                 struct diffstat_t diffstat;
3147
3148                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3149                 for (i = 0; i < q->nr; i++) {
3150                         struct diff_filepair *p = q->queue[i];
3151                         if (check_pair_status(p))
3152                                 diff_flush_stat(p, options, &diffstat);
3153                 }
3154                 if (output_format & DIFF_FORMAT_NUMSTAT)
3155                         show_numstat(&diffstat, options);
3156                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3157                         show_stats(&diffstat, options);
3158                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3159                         show_shortstats(&diffstat, options);
3160                 free_diffstat_info(&diffstat);
3161                 separator++;
3162         }
3163         if (output_format & DIFF_FORMAT_DIRSTAT)
3164                 show_dirstat(options);
3165
3166         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3167                 for (i = 0; i < q->nr; i++)
3168                         diff_summary(options->file, q->queue[i]);
3169                 separator++;
3170         }
3171
3172         if (output_format & DIFF_FORMAT_PATCH) {
3173                 if (separator) {
3174                         putc(options->line_termination, options->file);
3175                         if (options->stat_sep) {
3176                                 /* attach patch instead of inline */
3177                                 fputs(options->stat_sep, options->file);
3178                         }
3179                 }
3180
3181                 for (i = 0; i < q->nr; i++) {
3182                         struct diff_filepair *p = q->queue[i];
3183                         if (check_pair_status(p))
3184                                 diff_flush_patch(p, options);
3185                 }
3186         }
3187
3188         if (output_format & DIFF_FORMAT_CALLBACK)
3189                 options->format_callback(q, options, options->format_callback_data);
3190
3191         for (i = 0; i < q->nr; i++)
3192                 diff_free_filepair(q->queue[i]);
3193 free_queue:
3194         free(q->queue);
3195         q->queue = NULL;
3196         q->nr = q->alloc = 0;
3197         if (options->close_file)
3198                 fclose(options->file);
3199 }
3200
3201 static void diffcore_apply_filter(const char *filter)
3202 {
3203         int i;
3204         struct diff_queue_struct *q = &diff_queued_diff;
3205         struct diff_queue_struct outq;
3206         outq.queue = NULL;
3207         outq.nr = outq.alloc = 0;
3208
3209         if (!filter)
3210                 return;
3211
3212         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3213                 int found;
3214                 for (i = found = 0; !found && i < q->nr; i++) {
3215                         struct diff_filepair *p = q->queue[i];
3216                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3217                              ((p->score &&
3218                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3219                               (!p->score &&
3220                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3221                             ((p->status != DIFF_STATUS_MODIFIED) &&
3222                              strchr(filter, p->status)))
3223                                 found++;
3224                 }
3225                 if (found)
3226                         return;
3227
3228                 /* otherwise we will clear the whole queue
3229                  * by copying the empty outq at the end of this
3230                  * function, but first clear the current entries
3231                  * in the queue.
3232                  */
3233                 for (i = 0; i < q->nr; i++)
3234                         diff_free_filepair(q->queue[i]);
3235         }
3236         else {
3237                 /* Only the matching ones */
3238                 for (i = 0; i < q->nr; i++) {
3239                         struct diff_filepair *p = q->queue[i];
3240
3241                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3242                              ((p->score &&
3243                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3244                               (!p->score &&
3245                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3246                             ((p->status != DIFF_STATUS_MODIFIED) &&
3247                              strchr(filter, p->status)))
3248                                 diff_q(&outq, p);
3249                         else
3250                                 diff_free_filepair(p);
3251                 }
3252         }
3253         free(q->queue);
3254         *q = outq;
3255 }
3256
3257 /* Check whether two filespecs with the same mode and size are identical */
3258 static int diff_filespec_is_identical(struct diff_filespec *one,
3259                                       struct diff_filespec *two)
3260 {
3261         if (S_ISGITLINK(one->mode))
3262                 return 0;
3263         if (diff_populate_filespec(one, 0))
3264                 return 0;
3265         if (diff_populate_filespec(two, 0))
3266                 return 0;
3267         return !memcmp(one->data, two->data, one->size);
3268 }
3269
3270 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3271 {
3272         int i;
3273         struct diff_queue_struct *q = &diff_queued_diff;
3274         struct diff_queue_struct outq;
3275         outq.queue = NULL;
3276         outq.nr = outq.alloc = 0;
3277
3278         for (i = 0; i < q->nr; i++) {
3279                 struct diff_filepair *p = q->queue[i];
3280
3281                 /*
3282                  * 1. Entries that come from stat info dirtyness
3283                  *    always have both sides (iow, not create/delete),
3284                  *    one side of the object name is unknown, with
3285                  *    the same mode and size.  Keep the ones that
3286                  *    do not match these criteria.  They have real
3287                  *    differences.
3288                  *
3289                  * 2. At this point, the file is known to be modified,
3290                  *    with the same mode and size, and the object
3291                  *    name of one side is unknown.  Need to inspect
3292                  *    the identical contents.
3293                  */
3294                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3295                     !DIFF_FILE_VALID(p->two) ||
3296                     (p->one->sha1_valid && p->two->sha1_valid) ||
3297                     (p->one->mode != p->two->mode) ||
3298                     diff_populate_filespec(p->one, 1) ||
3299                     diff_populate_filespec(p->two, 1) ||
3300                     (p->one->size != p->two->size) ||
3301                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3302                         diff_q(&outq, p);
3303                 else {
3304                         /*
3305                          * The caller can subtract 1 from skip_stat_unmatch
3306                          * to determine how many paths were dirty only
3307                          * due to stat info mismatch.
3308                          */
3309                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3310                                 diffopt->skip_stat_unmatch++;
3311                         diff_free_filepair(p);
3312                 }
3313         }
3314         free(q->queue);
3315         *q = outq;
3316 }
3317
3318 void diffcore_std(struct diff_options *options)
3319 {
3320         if (options->skip_stat_unmatch)
3321                 diffcore_skip_stat_unmatch(options);
3322         if (options->break_opt != -1)
3323                 diffcore_break(options->break_opt);
3324         if (options->detect_rename)
3325                 diffcore_rename(options);
3326         if (options->break_opt != -1)
3327                 diffcore_merge_broken();
3328         if (options->pickaxe)
3329                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3330         if (options->orderfile)
3331                 diffcore_order(options->orderfile);
3332         diff_resolve_rename_copy();
3333         diffcore_apply_filter(options->filter);
3334
3335         if (diff_queued_diff.nr)
3336                 DIFF_OPT_SET(options, HAS_CHANGES);
3337         else
3338                 DIFF_OPT_CLR(options, HAS_CHANGES);
3339 }
3340
3341 int diff_result_code(struct diff_options *opt, int status)
3342 {
3343         int result = 0;
3344         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3345             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3346                 return status;
3347         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3348             DIFF_OPT_TST(opt, HAS_CHANGES))
3349                 result |= 01;
3350         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3351             DIFF_OPT_TST(opt, CHECK_FAILED))
3352                 result |= 02;
3353         return result;
3354 }
3355
3356 void diff_addremove(struct diff_options *options,
3357                     int addremove, unsigned mode,
3358                     const unsigned char *sha1,
3359                     const char *concatpath)
3360 {
3361         struct diff_filespec *one, *two;
3362
3363         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3364                 return;
3365
3366         /* This may look odd, but it is a preparation for
3367          * feeding "there are unchanged files which should
3368          * not produce diffs, but when you are doing copy
3369          * detection you would need them, so here they are"
3370          * entries to the diff-core.  They will be prefixed
3371          * with something like '=' or '*' (I haven't decided
3372          * which but should not make any difference).
3373          * Feeding the same new and old to diff_change()
3374          * also has the same effect.
3375          * Before the final output happens, they are pruned after
3376          * merged into rename/copy pairs as appropriate.
3377          */
3378         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3379                 addremove = (addremove == '+' ? '-' :
3380                              addremove == '-' ? '+' : addremove);
3381
3382         if (options->prefix &&
3383             strncmp(concatpath, options->prefix, options->prefix_length))
3384                 return;
3385
3386         one = alloc_filespec(concatpath);
3387         two = alloc_filespec(concatpath);
3388
3389         if (addremove != '+')
3390                 fill_filespec(one, sha1, mode);
3391         if (addremove != '-')
3392                 fill_filespec(two, sha1, mode);
3393
3394         diff_queue(&diff_queued_diff, one, two);
3395         DIFF_OPT_SET(options, HAS_CHANGES);
3396 }
3397
3398 void diff_change(struct diff_options *options,
3399                  unsigned old_mode, unsigned new_mode,
3400                  const unsigned char *old_sha1,
3401                  const unsigned char *new_sha1,
3402                  const char *concatpath)
3403 {
3404         struct diff_filespec *one, *two;
3405
3406         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3407                         && S_ISGITLINK(new_mode))
3408                 return;
3409
3410         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3411                 unsigned tmp;
3412                 const unsigned char *tmp_c;
3413                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3414                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3415         }
3416
3417         if (options->prefix &&
3418             strncmp(concatpath, options->prefix, options->prefix_length))
3419                 return;
3420
3421         one = alloc_filespec(concatpath);
3422         two = alloc_filespec(concatpath);
3423         fill_filespec(one, old_sha1, old_mode);
3424         fill_filespec(two, new_sha1, new_mode);
3425
3426         diff_queue(&diff_queued_diff, one, two);
3427         DIFF_OPT_SET(options, HAS_CHANGES);
3428 }
3429
3430 void diff_unmerge(struct diff_options *options,
3431                   const char *path,
3432                   unsigned mode, const unsigned char *sha1)
3433 {
3434         struct diff_filespec *one, *two;
3435
3436         if (options->prefix &&
3437             strncmp(path, options->prefix, options->prefix_length))
3438                 return;
3439
3440         one = alloc_filespec(path);
3441         two = alloc_filespec(path);
3442         fill_filespec(one, sha1, mode);
3443         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3444 }
3445
3446 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3447                 size_t *outsize)
3448 {
3449         struct diff_tempfile temp;
3450         const char *argv[3];
3451         const char **arg = argv;
3452         struct child_process child;
3453         struct strbuf buf = STRBUF_INIT;
3454
3455         prepare_temp_file(spec->path, &temp, spec);
3456         *arg++ = pgm;
3457         *arg++ = temp.name;
3458         *arg = NULL;
3459
3460         memset(&child, 0, sizeof(child));
3461         child.argv = argv;
3462         child.out = -1;
3463         if (start_command(&child) != 0 ||
3464             strbuf_read(&buf, child.out, 0) < 0 ||
3465             finish_command(&child) != 0) {
3466                 if (temp.name == temp.tmp_path)
3467                         unlink(temp.name);
3468                 error("error running textconv command '%s'", pgm);
3469                 return NULL;
3470         }
3471         if (temp.name == temp.tmp_path)
3472                 unlink(temp.name);
3473
3474         return strbuf_detach(&buf, outsize);
3475 }