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
  15 #ifdef NO_FAST_WORKING_DIRECTORY
  16 #define FAST_WORKING_DIRECTORY 0
  17 #else
  18 #define FAST_WORKING_DIRECTORY 1
  19 #endif
  20
  21 static int diff_detect_rename_default;
  22 static int diff_rename_limit_default = 100;
  23 static int diff_use_color_default;
  24 static const char *external_diff_cmd_cfg;
  25 int diff_auto_refresh_index = 1;
  26
  27 static char diff_colors[][COLOR_MAXLEN] = {
  28         "\033[m",       /* reset */
  29         "",             /* PLAIN (normal) */
  30         "\033[1m",      /* METAINFO (bold) */
  31         "\033[36m",     /* FRAGINFO (cyan) */
  32         "\033[31m",     /* OLD (red) */
  33         "\033[32m",     /* NEW (green) */
  34         "\033[33m",     /* COMMIT (yellow) */
  35         "\033[41m",     /* WHITESPACE (red background) */
  36 };
  37
  38 static int parse_diff_color_slot(const char *var, int ofs)
  39 {
  40         if (!strcasecmp(var+ofs, "plain"))
  41                 return DIFF_PLAIN;
  42         if (!strcasecmp(var+ofs, "meta"))
  43                 return DIFF_METAINFO;
  44         if (!strcasecmp(var+ofs, "frag"))
  45                 return DIFF_FRAGINFO;
  46         if (!strcasecmp(var+ofs, "old"))
  47                 return DIFF_FILE_OLD;
  48         if (!strcasecmp(var+ofs, "new"))
  49                 return DIFF_FILE_NEW;
  50         if (!strcasecmp(var+ofs, "commit"))
  51                 return DIFF_COMMIT;
  52         if (!strcasecmp(var+ofs, "whitespace"))
  53                 return DIFF_WHITESPACE;
  54         die("bad config variable '%s'", var);
  55 }
  56
  57 static struct ll_diff_driver {
  58         const char *name;
  59         struct ll_diff_driver *next;
  60         char *cmd;
  61 } *user_diff, **user_diff_tail;
  62
  63 /*
  64  * Currently there is only "diff.<drivername>.command" variable;
  65  * because there are "diff.color.<slot>" variables, we are parsing
  66  * this in a bit convoluted way to allow low level diff driver
  67  * called "color".
  68  */
  69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  70 {
  71         const char *name;
  72         int namelen;
  73         struct ll_diff_driver *drv;
  74
  75         name = var + 5;
  76         namelen = ep - name;
  77         for (drv = user_diff; drv; drv = drv->next)
  78                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  79                         break;
  80         if (!drv) {
  81                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
  82                 drv->name = xmemdupz(name, namelen);
  83                 if (!user_diff_tail)
  84                         user_diff_tail = &user_diff;
  85                 *user_diff_tail = drv;
  86                 user_diff_tail = &(drv->next);
  87         }
  88
  89         if (!value)
  90                 return error("%s: lacks value", var);
  91         drv->cmd = strdup(value);
  92         return 0;
  93 }
  94
  95 /*
  96  * 'diff.<what>.funcname' attribute can be specified in the configuration
  97  * to define a customized regexp to find the beginning of a function to
  98  * be used for hunk header lines of "diff -p" style output.
  99  */
 100 static struct funcname_pattern {
 101         char *name;
 102         char *pattern;
 103         struct funcname_pattern *next;
 104 } *funcname_pattern_list;
 105
 106 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 107 {
 108         const char *name;
 109         int namelen;
 110         struct funcname_pattern *pp;
 111
 112         name = var + 5; /* "diff." */
 113         namelen = ep - name;
 114
 115         for (pp = funcname_pattern_list; pp; pp = pp->next)
 116                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 117                         break;
 118         if (!pp) {
 119                 pp = xcalloc(1, sizeof(*pp));
 120                 pp->name = xmemdupz(name, namelen);
 121                 pp->next = funcname_pattern_list;
 122                 funcname_pattern_list = pp;
 123         }
 124         if (pp->pattern)
 125                 free(pp->pattern);
 126         pp->pattern = xstrdup(value);
 127         return 0;
 128 }
 129
 130 /*
 131  * These are to give UI layer defaults.
 132  * The core-level commands such as git-diff-files should
 133  * never be affected by the setting of diff.renames
 134  * the user happens to have in the configuration file.
 135  */
 136 int git_diff_ui_config(const char *var, const char *value)
 137 {
 138         if (!strcmp(var, "diff.renamelimit")) {
 139                 diff_rename_limit_default = git_config_int(var, value);
 140                 return 0;
 141         }
 142         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 143                 diff_use_color_default = git_config_colorbool(var, value, -1);
 144                 return 0;
 145         }
 146         if (!strcmp(var, "diff.renames")) {
 147                 if (!value)
 148                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 149                 else if (!strcasecmp(value, "copies") ||
 150                          !strcasecmp(value, "copy"))
 151                         diff_detect_rename_default = DIFF_DETECT_COPY;
 152                 else if (git_config_bool(var,value))
 153                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 154                 return 0;
 155         }
 156         if (!strcmp(var, "diff.autorefreshindex")) {
 157                 diff_auto_refresh_index = git_config_bool(var, value);
 158                 return 0;
 159         }
 160         if (!strcmp(var, "diff.external")) {
 161                 external_diff_cmd_cfg = xstrdup(value);
 162                 return 0;
 163         }
 164         if (!prefixcmp(var, "diff.")) {
 165                 const char *ep = strrchr(var, '.');
 166
 167                 if (ep != var + 4) {
 168                         if (!strcmp(ep, ".command"))
 169                                 return parse_lldiff_command(var, ep, value);
 170                 }
 171         }
 172
 173         return git_diff_basic_config(var, value);
 174 }
 175
 176 int git_diff_basic_config(const char *var, const char *value)
 177 {
 178         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 179                 int slot = parse_diff_color_slot(var, 11);
 180                 color_parse(value, var, diff_colors[slot]);
 181                 return 0;
 182         }
 183
 184         if (!prefixcmp(var, "diff.")) {
 185                 const char *ep = strrchr(var, '.');
 186                 if (ep != var + 4) {
 187                         if (!strcmp(ep, ".funcname"))
 188                                 return parse_funcname_pattern(var, ep, value);
 189                 }
 190         }
 191
 192         return git_default_config(var, value);
 193 }
 194
 195 static char *quote_two(const char *one, const char *two)
 196 {
 197         int need_one = quote_c_style(one, NULL, NULL, 1);
 198         int need_two = quote_c_style(two, NULL, NULL, 1);
 199         struct strbuf res;
 200
 201         strbuf_init(&res, 0);
 202         if (need_one + need_two) {
 203                 strbuf_addch(&res, '"');
 204                 quote_c_style(one, &res, NULL, 1);
 205                 quote_c_style(two, &res, NULL, 1);
 206                 strbuf_addch(&res, '"');
 207         } else {
 208                 strbuf_addstr(&res, one);
 209                 strbuf_addstr(&res, two);
 210         }
 211         return strbuf_detach(&res, NULL);
 212 }
 213
 214 static const char *external_diff(void)
 215 {
 216         static const char *external_diff_cmd = NULL;
 217         static int done_preparing = 0;
 218
 219         if (done_preparing)
 220                 return external_diff_cmd;
 221         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 222         if (!external_diff_cmd)
 223                 external_diff_cmd = external_diff_cmd_cfg;
 224         done_preparing = 1;
 225         return external_diff_cmd;
 226 }
 227
 228 static struct diff_tempfile {
 229         const char *name; /* filename external diff should read from */
 230         char hex[41];
 231         char mode[10];
 232         char tmp_path[PATH_MAX];
 233 } diff_temp[2];
 234
 235 static int count_lines(const char *data, int size)
 236 {
 237         int count, ch, completely_empty = 1, nl_just_seen = 0;
 238         count = 0;
 239         while (0 < size--) {
 240                 ch = *data++;
 241                 if (ch == '\n') {
 242                         count++;
 243                         nl_just_seen = 1;
 244                         completely_empty = 0;
 245                 }
 246                 else {
 247                         nl_just_seen = 0;
 248                         completely_empty = 0;
 249                 }
 250         }
 251         if (completely_empty)
 252                 return 0;
 253         if (!nl_just_seen)
 254                 count++; /* no trailing newline */
 255         return count;
 256 }
 257
 258 static void print_line_count(int count)
 259 {
 260         switch (count) {
 261         case 0:
 262                 printf("0,0");
 263                 break;
 264         case 1:
 265                 printf("1");
 266                 break;
 267         default:
 268                 printf("1,%d", count);
 269                 break;
 270         }
 271 }
 272
 273 static void copy_file(int prefix, const char *data, int size,
 274                 const char *set, const char *reset)
 275 {
 276         int ch, nl_just_seen = 1;
 277         while (0 < size--) {
 278                 ch = *data++;
 279                 if (nl_just_seen) {
 280                         fputs(set, stdout);
 281                         putchar(prefix);
 282                 }
 283                 if (ch == '\n') {
 284                         nl_just_seen = 1;
 285                         fputs(reset, stdout);
 286                 } else
 287                         nl_just_seen = 0;
 288                 putchar(ch);
 289         }
 290         if (!nl_just_seen)
 291                 printf("%s\n\\ No newline at end of file\n", reset);
 292 }
 293
 294 static void emit_rewrite_diff(const char *name_a,
 295                               const char *name_b,
 296                               struct diff_filespec *one,
 297                               struct diff_filespec *two,
 298                               struct diff_options *o)
 299 {
 300         int lc_a, lc_b;
 301         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 302         const char *name_a_tab, *name_b_tab;
 303         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 304         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 305         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 306         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 307         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 308         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 309
 310         name_a += (*name_a == '/');
 311         name_b += (*name_b == '/');
 312         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 313         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 314
 315         strbuf_reset(&a_name);
 316         strbuf_reset(&b_name);
 317         quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
 318         quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
 319
 320         diff_populate_filespec(one, 0);
 321         diff_populate_filespec(two, 0);
 322         lc_a = count_lines(one->data, one->size);
 323         lc_b = count_lines(two->data, two->size);
 324         printf("%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 325                metainfo, a_name.buf, name_a_tab, reset,
 326                metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 327         print_line_count(lc_a);
 328         printf(" +");
 329         print_line_count(lc_b);
 330         printf(" @@%s\n", reset);
 331         if (lc_a)
 332                 copy_file('-', one->data, one->size, old, reset);
 333         if (lc_b)
 334                 copy_file('+', two->data, two->size, new, reset);
 335 }
 336
 337 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 338 {
 339         if (!DIFF_FILE_VALID(one)) {
 340                 mf->ptr = (char *)""; /* does not matter */
 341                 mf->size = 0;
 342                 return 0;
 343         }
 344         else if (diff_populate_filespec(one, 0))
 345                 return -1;
 346         mf->ptr = one->data;
 347         mf->size = one->size;
 348         return 0;
 349 }
 350
 351 struct diff_words_buffer {
 352         mmfile_t text;
 353         long alloc;
 354         long current; /* output pointer */
 355         int suppressed_newline;
 356 };
 357
 358 static void diff_words_append(char *line, unsigned long len,
 359                 struct diff_words_buffer *buffer)
 360 {
 361         if (buffer->text.size + len > buffer->alloc) {
 362                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 363                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 364         }
 365         line++;
 366         len--;
 367         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 368         buffer->text.size += len;
 369 }
 370
 371 struct diff_words_data {
 372         struct xdiff_emit_state xm;
 373         struct diff_words_buffer minus, plus;
 374 };
 375
 376 static void print_word(struct diff_words_buffer *buffer, int len, int color,
 377                 int suppress_newline)
 378 {
 379         const char *ptr;
 380         int eol = 0;
 381
 382         if (len == 0)
 383                 return;
 384
 385         ptr  = buffer->text.ptr + buffer->current;
 386         buffer->current += len;
 387
 388         if (ptr[len - 1] == '\n') {
 389                 eol = 1;
 390                 len--;
 391         }
 392
 393         fputs(diff_get_color(1, color), stdout);
 394         fwrite(ptr, len, 1, stdout);
 395         fputs(diff_get_color(1, DIFF_RESET), stdout);
 396
 397         if (eol) {
 398                 if (suppress_newline)
 399                         buffer->suppressed_newline = 1;
 400                 else
 401                         putchar('\n');
 402         }
 403 }
 404
 405 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 406 {
 407         struct diff_words_data *diff_words = priv;
 408
 409         if (diff_words->minus.suppressed_newline) {
 410                 if (line[0] != '+')
 411                         putchar('\n');
 412                 diff_words->minus.suppressed_newline = 0;
 413         }
 414
 415         len--;
 416         switch (line[0]) {
 417                 case '-':
 418                         print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
 419                         break;
 420                 case '+':
 421                         print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
 422                         break;
 423                 case ' ':
 424                         print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
 425                         diff_words->minus.current += len;
 426                         break;
 427         }
 428 }
 429
 430 /* this executes the word diff on the accumulated buffers */
 431 static void diff_words_show(struct diff_words_data *diff_words)
 432 {
 433         xpparam_t xpp;
 434         xdemitconf_t xecfg;
 435         xdemitcb_t ecb;
 436         mmfile_t minus, plus;
 437         int i;
 438
 439         memset(&xecfg, 0, sizeof(xecfg));
 440         minus.size = diff_words->minus.text.size;
 441         minus.ptr = xmalloc(minus.size);
 442         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 443         for (i = 0; i < minus.size; i++)
 444                 if (isspace(minus.ptr[i]))
 445                         minus.ptr[i] = '\n';
 446         diff_words->minus.current = 0;
 447
 448         plus.size = diff_words->plus.text.size;
 449         plus.ptr = xmalloc(plus.size);
 450         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 451         for (i = 0; i < plus.size; i++)
 452                 if (isspace(plus.ptr[i]))
 453                         plus.ptr[i] = '\n';
 454         diff_words->plus.current = 0;
 455
 456         xpp.flags = XDF_NEED_MINIMAL;
 457         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 458         ecb.outf = xdiff_outf;
 459         ecb.priv = diff_words;
 460         diff_words->xm.consume = fn_out_diff_words_aux;
 461         xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 462
 463         free(minus.ptr);
 464         free(plus.ptr);
 465         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 466
 467         if (diff_words->minus.suppressed_newline) {
 468                 putchar('\n');
 469                 diff_words->minus.suppressed_newline = 0;
 470         }
 471 }
 472
 473 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 474
 475 struct emit_callback {
 476         struct xdiff_emit_state xm;
 477         int nparents, color_diff;
 478         unsigned ws_rule;
 479         sane_truncate_fn truncate;
 480         const char **label_path;
 481         struct diff_words_data *diff_words;
 482         int *found_changesp;
 483 };
 484
 485 static void free_diff_words_data(struct emit_callback *ecbdata)
 486 {
 487         if (ecbdata->diff_words) {
 488                 /* flush buffers */
 489                 if (ecbdata->diff_words->minus.text.size ||
 490                                 ecbdata->diff_words->plus.text.size)
 491                         diff_words_show(ecbdata->diff_words);
 492
 493                 if (ecbdata->diff_words->minus.text.ptr)
 494                         free (ecbdata->diff_words->minus.text.ptr);
 495                 if (ecbdata->diff_words->plus.text.ptr)
 496                         free (ecbdata->diff_words->plus.text.ptr);
 497                 free(ecbdata->diff_words);
 498                 ecbdata->diff_words = NULL;
 499         }
 500 }
 501
 502 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 503 {
 504         if (diff_use_color)
 505                 return diff_colors[ix];
 506         return "";
 507 }
 508
 509 static void emit_line(const char *set, const char *reset, const char *line, int len)
 510 {
 511         fputs(set, stdout);
 512         fwrite(line, len, 1, stdout);
 513         fputs(reset, stdout);
 514 }
 515
 516 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 517 {
 518         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 519         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 520
 521         if (!*ws)
 522                 emit_line(set, reset, line, len);
 523         else {
 524                 /* Emit just the prefix, then the rest. */
 525                 emit_line(set, reset, line, ecbdata->nparents);
 526                 (void)check_and_emit_line(line + ecbdata->nparents,
 527                     len - ecbdata->nparents, ecbdata->ws_rule,
 528                     stdout, set, reset, ws);
 529         }
 530 }
 531
 532 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 533 {
 534         const char *cp;
 535         unsigned long allot;
 536         size_t l = len;
 537
 538         if (ecb->truncate)
 539                 return ecb->truncate(line, len);
 540         cp = line;
 541         allot = l;
 542         while (0 < l) {
 543                 (void) utf8_width(&cp, &l);
 544                 if (!cp)
 545                         break; /* truncated in the middle? */
 546         }
 547         return allot - l;
 548 }
 549
 550 static void fn_out_consume(void *priv, char *line, unsigned long len)
 551 {
 552         int i;
 553         int color;
 554         struct emit_callback *ecbdata = priv;
 555         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 556         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 557         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 558
 559         *(ecbdata->found_changesp) = 1;
 560
 561         if (ecbdata->label_path[0]) {
 562                 const char *name_a_tab, *name_b_tab;
 563
 564                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 565                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 566
 567                 printf("%s--- %s%s%s\n",
 568                        meta, ecbdata->label_path[0], reset, name_a_tab);
 569                 printf("%s+++ %s%s%s\n",
 570                        meta, ecbdata->label_path[1], reset, name_b_tab);
 571                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 572         }
 573
 574         /* This is not really necessary for now because
 575          * this codepath only deals with two-way diffs.
 576          */
 577         for (i = 0; i < len && line[i] == '@'; i++)
 578                 ;
 579         if (2 <= i && i < len && line[i] == ' ') {
 580                 ecbdata->nparents = i - 1;
 581                 len = sane_truncate_line(ecbdata, line, len);
 582                 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 583                           reset, line, len);
 584                 if (line[len-1] != '\n')
 585                         putchar('\n');
 586                 return;
 587         }
 588
 589         if (len < ecbdata->nparents) {
 590                 emit_line(reset, reset, line, len);
 591                 return;
 592         }
 593
 594         color = DIFF_PLAIN;
 595         if (ecbdata->diff_words && ecbdata->nparents != 1)
 596                 /* fall back to normal diff */
 597                 free_diff_words_data(ecbdata);
 598         if (ecbdata->diff_words) {
 599                 if (line[0] == '-') {
 600                         diff_words_append(line, len,
 601                                           &ecbdata->diff_words->minus);
 602                         return;
 603                 } else if (line[0] == '+') {
 604                         diff_words_append(line, len,
 605                                           &ecbdata->diff_words->plus);
 606                         return;
 607                 }
 608                 if (ecbdata->diff_words->minus.text.size ||
 609                     ecbdata->diff_words->plus.text.size)
 610                         diff_words_show(ecbdata->diff_words);
 611                 line++;
 612                 len--;
 613                 emit_line(plain, reset, line, len);
 614                 return;
 615         }
 616         for (i = 0; i < ecbdata->nparents && len; i++) {
 617                 if (line[i] == '-')
 618                         color = DIFF_FILE_OLD;
 619                 else if (line[i] == '+')
 620                         color = DIFF_FILE_NEW;
 621         }
 622
 623         if (color != DIFF_FILE_NEW) {
 624                 emit_line(diff_get_color(ecbdata->color_diff, color),
 625                           reset, line, len);
 626                 return;
 627         }
 628         emit_add_line(reset, ecbdata, line, len);
 629 }
 630
 631 static char *pprint_rename(const char *a, const char *b)
 632 {
 633         const char *old = a;
 634         const char *new = b;
 635         struct strbuf name;
 636         int pfx_length, sfx_length;
 637         int len_a = strlen(a);
 638         int len_b = strlen(b);
 639         int a_midlen, b_midlen;
 640         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 641         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 642
 643         strbuf_init(&name, 0);
 644         if (qlen_a || qlen_b) {
 645                 quote_c_style(a, &name, NULL, 0);
 646                 strbuf_addstr(&name, " => ");
 647                 quote_c_style(b, &name, NULL, 0);
 648                 return strbuf_detach(&name, NULL);
 649         }
 650
 651         /* Find common prefix */
 652         pfx_length = 0;
 653         while (*old && *new && *old == *new) {
 654                 if (*old == '/')
 655                         pfx_length = old - a + 1;
 656                 old++;
 657                 new++;
 658         }
 659
 660         /* Find common suffix */
 661         old = a + len_a;
 662         new = b + len_b;
 663         sfx_length = 0;
 664         while (a <= old && b <= new && *old == *new) {
 665                 if (*old == '/')
 666                         sfx_length = len_a - (old - a);
 667                 old--;
 668                 new--;
 669         }
 670
 671         /*
 672          * pfx{mid-a => mid-b}sfx
 673          * {pfx-a => pfx-b}sfx
 674          * pfx{sfx-a => sfx-b}
 675          * name-a => name-b
 676          */
 677         a_midlen = len_a - pfx_length - sfx_length;
 678         b_midlen = len_b - pfx_length - sfx_length;
 679         if (a_midlen < 0)
 680                 a_midlen = 0;
 681         if (b_midlen < 0)
 682                 b_midlen = 0;
 683
 684         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 685         if (pfx_length + sfx_length) {
 686                 strbuf_add(&name, a, pfx_length);
 687                 strbuf_addch(&name, '{');
 688         }
 689         strbuf_add(&name, a + pfx_length, a_midlen);
 690         strbuf_addstr(&name, " => ");
 691         strbuf_add(&name, b + pfx_length, b_midlen);
 692         if (pfx_length + sfx_length) {
 693                 strbuf_addch(&name, '}');
 694                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 695         }
 696         return strbuf_detach(&name, NULL);
 697 }
 698
 699 struct diffstat_t {
 700         struct xdiff_emit_state xm;
 701
 702         int nr;
 703         int alloc;
 704         struct diffstat_file {
 705                 char *from_name;
 706                 char *name;
 707                 char *print_name;
 708                 unsigned is_unmerged:1;
 709                 unsigned is_binary:1;
 710                 unsigned is_renamed:1;
 711                 unsigned int added, deleted;
 712         } **files;
 713 };
 714
 715 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 716                                           const char *name_a,
 717                                           const char *name_b)
 718 {
 719         struct diffstat_file *x;
 720         x = xcalloc(sizeof (*x), 1);
 721         if (diffstat->nr == diffstat->alloc) {
 722                 diffstat->alloc = alloc_nr(diffstat->alloc);
 723                 diffstat->files = xrealloc(diffstat->files,
 724                                 diffstat->alloc * sizeof(x));
 725         }
 726         diffstat->files[diffstat->nr++] = x;
 727         if (name_b) {
 728                 x->from_name = xstrdup(name_a);
 729                 x->name = xstrdup(name_b);
 730                 x->is_renamed = 1;
 731         }
 732         else {
 733                 x->from_name = NULL;
 734                 x->name = xstrdup(name_a);
 735         }
 736         return x;
 737 }
 738
 739 static void diffstat_consume(void *priv, char *line, unsigned long len)
 740 {
 741         struct diffstat_t *diffstat = priv;
 742         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 743
 744         if (line[0] == '+')
 745                 x->added++;
 746         else if (line[0] == '-')
 747                 x->deleted++;
 748 }
 749
 750 const char mime_boundary_leader[] = "------------";
 751
 752 static int scale_linear(int it, int width, int max_change)
 753 {
 754         /*
 755          * make sure that at least one '-' is printed if there were deletions,
 756          * and likewise for '+'.
 757          */
 758         if (max_change < 2)
 759                 return it;
 760         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 761 }
 762
 763 static void show_name(const char *prefix, const char *name, int len,
 764                       const char *reset, const char *set)
 765 {
 766         printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
 767 }
 768
 769 static void show_graph(char ch, int cnt, const char *set, const char *reset)
 770 {
 771         if (cnt <= 0)
 772                 return;
 773         printf("%s", set);
 774         while (cnt--)
 775                 putchar(ch);
 776         printf("%s", reset);
 777 }
 778
 779 static void fill_print_name(struct diffstat_file *file)
 780 {
 781         char *pname;
 782
 783         if (file->print_name)
 784                 return;
 785
 786         if (!file->is_renamed) {
 787                 struct strbuf buf;
 788                 strbuf_init(&buf, 0);
 789                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 790                         pname = strbuf_detach(&buf, NULL);
 791                 } else {
 792                         pname = file->name;
 793                         strbuf_release(&buf);
 794                 }
 795         } else {
 796                 pname = pprint_rename(file->from_name, file->name);
 797         }
 798         file->print_name = pname;
 799 }
 800
 801 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 802 {
 803         int i, len, add, del, total, adds = 0, dels = 0;
 804         int max_change = 0, max_len = 0;
 805         int total_files = data->nr;
 806         int width, name_width;
 807         const char *reset, *set, *add_c, *del_c;
 808
 809         if (data->nr == 0)
 810                 return;
 811
 812         width = options->stat_width ? options->stat_width : 80;
 813         name_width = options->stat_name_width ? options->stat_name_width : 50;
 814
 815         /* Sanity: give at least 5 columns to the graph,
 816          * but leave at least 10 columns for the name.
 817          */
 818         if (width < name_width + 15) {
 819                 if (name_width <= 25)
 820                         width = name_width + 15;
 821                 else
 822                         name_width = width - 15;
 823         }
 824
 825         /* Find the longest filename and max number of changes */
 826         reset = diff_get_color_opt(options, DIFF_RESET);
 827         set   = diff_get_color_opt(options, DIFF_PLAIN);
 828         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 829         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 830
 831         for (i = 0; i < data->nr; i++) {
 832                 struct diffstat_file *file = data->files[i];
 833                 int change = file->added + file->deleted;
 834                 fill_print_name(file);
 835                 len = strlen(file->print_name);
 836                 if (max_len < len)
 837                         max_len = len;
 838
 839                 if (file->is_binary || file->is_unmerged)
 840                         continue;
 841                 if (max_change < change)
 842                         max_change = change;
 843         }
 844
 845         /* Compute the width of the graph part;
 846          * 10 is for one blank at the beginning of the line plus
 847          * " | count " between the name and the graph.
 848          *
 849          * From here on, name_width is the width of the name area,
 850          * and width is the width of the graph area.
 851          */
 852         name_width = (name_width < max_len) ? name_width : max_len;
 853         if (width < (name_width + 10) + max_change)
 854                 width = width - (name_width + 10);
 855         else
 856                 width = max_change;
 857
 858         for (i = 0; i < data->nr; i++) {
 859                 const char *prefix = "";
 860                 char *name = data->files[i]->print_name;
 861                 int added = data->files[i]->added;
 862                 int deleted = data->files[i]->deleted;
 863                 int name_len;
 864
 865                 /*
 866                  * "scale" the filename
 867                  */
 868                 len = name_width;
 869                 name_len = strlen(name);
 870                 if (name_width < name_len) {
 871                         char *slash;
 872                         prefix = "...";
 873                         len -= 3;
 874                         name += name_len - len;
 875                         slash = strchr(name, '/');
 876                         if (slash)
 877                                 name = slash;
 878                 }
 879
 880                 if (data->files[i]->is_binary) {
 881                         show_name(prefix, name, len, reset, set);
 882                         printf("  Bin ");
 883                         printf("%s%d%s", del_c, deleted, reset);
 884                         printf(" -> ");
 885                         printf("%s%d%s", add_c, added, reset);
 886                         printf(" bytes");
 887                         printf("\n");
 888                         continue;
 889                 }
 890                 else if (data->files[i]->is_unmerged) {
 891                         show_name(prefix, name, len, reset, set);
 892                         printf("  Unmerged\n");
 893                         continue;
 894                 }
 895                 else if (!data->files[i]->is_renamed &&
 896                          (added + deleted == 0)) {
 897                         total_files--;
 898                         continue;
 899                 }
 900
 901                 /*
 902                  * scale the add/delete
 903                  */
 904                 add = added;
 905                 del = deleted;
 906                 total = add + del;
 907                 adds += add;
 908                 dels += del;
 909
 910                 if (width <= max_change) {
 911                         add = scale_linear(add, width, max_change);
 912                         del = scale_linear(del, width, max_change);
 913                         total = add + del;
 914                 }
 915                 show_name(prefix, name, len, reset, set);
 916                 printf("%5d ", added + deleted);
 917                 show_graph('+', add, add_c, reset);
 918                 show_graph('-', del, del_c, reset);
 919                 putchar('\n');
 920         }
 921         printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 922                set, total_files, adds, dels, reset);
 923 }
 924
 925 static void show_shortstats(struct diffstat_t* data)
 926 {
 927         int i, adds = 0, dels = 0, total_files = data->nr;
 928
 929         if (data->nr == 0)
 930                 return;
 931
 932         for (i = 0; i < data->nr; i++) {
 933                 if (!data->files[i]->is_binary &&
 934                     !data->files[i]->is_unmerged) {
 935                         int added = data->files[i]->added;
 936                         int deleted= data->files[i]->deleted;
 937                         if (!data->files[i]->is_renamed &&
 938                             (added + deleted == 0)) {
 939                                 total_files--;
 940                         } else {
 941                                 adds += added;
 942                                 dels += deleted;
 943                         }
 944                 }
 945         }
 946         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 947                total_files, adds, dels);
 948 }
 949
 950 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 951 {
 952         int i;
 953
 954         if (data->nr == 0)
 955                 return;
 956
 957         for (i = 0; i < data->nr; i++) {
 958                 struct diffstat_file *file = data->files[i];
 959
 960                 if (file->is_binary)
 961                         printf("-\t-\t");
 962                 else
 963                         printf("%d\t%d\t", file->added, file->deleted);
 964                 if (options->line_termination) {
 965                         fill_print_name(file);
 966                         if (!file->is_renamed)
 967                                 write_name_quoted(file->name, stdout,
 968                                                   options->line_termination);
 969                         else {
 970                                 fputs(file->print_name, stdout);
 971                                 putchar(options->line_termination);
 972                         }
 973                 } else {
 974                         if (file->is_renamed) {
 975                                 putchar('\0');
 976                                 write_name_quoted(file->from_name, stdout, '\0');
 977                         }
 978                         write_name_quoted(file->name, stdout, '\0');
 979                 }
 980         }
 981 }
 982
 983 static void free_diffstat_info(struct diffstat_t *diffstat)
 984 {
 985         int i;
 986         for (i = 0; i < diffstat->nr; i++) {
 987                 struct diffstat_file *f = diffstat->files[i];
 988                 if (f->name != f->print_name)
 989                         free(f->print_name);
 990                 free(f->name);
 991                 free(f->from_name);
 992                 free(f);
 993         }
 994         free(diffstat->files);
 995 }
 996
 997 struct checkdiff_t {
 998         struct xdiff_emit_state xm;
 999         const char *filename;
1000         int lineno, color_diff;
1001         unsigned ws_rule;
1002         unsigned status;
1003 };
1004
1005 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1006 {
1007         struct checkdiff_t *data = priv;
1008         const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1009         const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1010         const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1011         char *err;
1012
1013         if (line[0] == '+') {
1014                 data->status = check_and_emit_line(line + 1, len - 1,
1015                     data->ws_rule, NULL, NULL, NULL, NULL);
1016                 if (!data->status)
1017                         return;
1018                 err = whitespace_error_string(data->status);
1019                 printf("%s:%d: %s.\n", data->filename, data->lineno, err);
1020                 free(err);
1021                 emit_line(set, reset, line, 1);
1022                 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1023                     stdout, set, reset, ws);
1024                 data->lineno++;
1025         } else if (line[0] == ' ')
1026                 data->lineno++;
1027         else if (line[0] == '@') {
1028                 char *plus = strchr(line, '+');
1029                 if (plus)
1030                         data->lineno = strtol(plus, NULL, 10);
1031                 else
1032                         die("invalid diff");
1033         }
1034 }
1035
1036 static unsigned char *deflate_it(char *data,
1037                                  unsigned long size,
1038                                  unsigned long *result_size)
1039 {
1040         int bound;
1041         unsigned char *deflated;
1042         z_stream stream;
1043
1044         memset(&stream, 0, sizeof(stream));
1045         deflateInit(&stream, zlib_compression_level);
1046         bound = deflateBound(&stream, size);
1047         deflated = xmalloc(bound);
1048         stream.next_out = deflated;
1049         stream.avail_out = bound;
1050
1051         stream.next_in = (unsigned char *)data;
1052         stream.avail_in = size;
1053         while (deflate(&stream, Z_FINISH) == Z_OK)
1054                 ; /* nothing */
1055         deflateEnd(&stream);
1056         *result_size = stream.total_out;
1057         return deflated;
1058 }
1059
1060 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1061 {
1062         void *cp;
1063         void *delta;
1064         void *deflated;
1065         void *data;
1066         unsigned long orig_size;
1067         unsigned long delta_size;
1068         unsigned long deflate_size;
1069         unsigned long data_size;
1070
1071         /* We could do deflated delta, or we could do just deflated two,
1072          * whichever is smaller.
1073          */
1074         delta = NULL;
1075         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1076         if (one->size && two->size) {
1077                 delta = diff_delta(one->ptr, one->size,
1078                                    two->ptr, two->size,
1079                                    &delta_size, deflate_size);
1080                 if (delta) {
1081                         void *to_free = delta;
1082                         orig_size = delta_size;
1083                         delta = deflate_it(delta, delta_size, &delta_size);
1084                         free(to_free);
1085                 }
1086         }
1087
1088         if (delta && delta_size < deflate_size) {
1089                 printf("delta %lu\n", orig_size);
1090                 free(deflated);
1091                 data = delta;
1092                 data_size = delta_size;
1093         }
1094         else {
1095                 printf("literal %lu\n", two->size);
1096                 free(delta);
1097                 data = deflated;
1098                 data_size = deflate_size;
1099         }
1100
1101         /* emit data encoded in base85 */
1102         cp = data;
1103         while (data_size) {
1104                 int bytes = (52 < data_size) ? 52 : data_size;
1105                 char line[70];
1106                 data_size -= bytes;
1107                 if (bytes <= 26)
1108                         line[0] = bytes + 'A' - 1;
1109                 else
1110                         line[0] = bytes - 26 + 'a' - 1;
1111                 encode_85(line + 1, cp, bytes);
1112                 cp = (char *) cp + bytes;
1113                 puts(line);
1114         }
1115         printf("\n");
1116         free(data);
1117 }
1118
1119 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1120 {
1121         printf("GIT binary patch\n");
1122         emit_binary_diff_body(one, two);
1123         emit_binary_diff_body(two, one);
1124 }
1125
1126 static void setup_diff_attr_check(struct git_attr_check *check)
1127 {
1128         static struct git_attr *attr_diff;
1129
1130         if (!attr_diff) {
1131                 attr_diff = git_attr("diff", 4);
1132         }
1133         check[0].attr = attr_diff;
1134 }
1135
1136 static void diff_filespec_check_attr(struct diff_filespec *one)
1137 {
1138         struct git_attr_check attr_diff_check;
1139         int check_from_data = 0;
1140
1141         if (one->checked_attr)
1142                 return;
1143
1144         setup_diff_attr_check(&attr_diff_check);
1145         one->is_binary = 0;
1146         one->funcname_pattern_ident = NULL;
1147
1148         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1149                 const char *value;
1150
1151                 /* binaryness */
1152                 value = attr_diff_check.value;
1153                 if (ATTR_TRUE(value))
1154                         ;
1155                 else if (ATTR_FALSE(value))
1156                         one->is_binary = 1;
1157                 else
1158                         check_from_data = 1;
1159
1160                 /* funcname pattern ident */
1161                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1162                         ;
1163                 else
1164                         one->funcname_pattern_ident = value;
1165         }
1166
1167         if (check_from_data) {
1168                 if (!one->data && DIFF_FILE_VALID(one))
1169                         diff_populate_filespec(one, 0);
1170
1171                 if (one->data)
1172                         one->is_binary = buffer_is_binary(one->data, one->size);
1173         }
1174 }
1175
1176 int diff_filespec_is_binary(struct diff_filespec *one)
1177 {
1178         diff_filespec_check_attr(one);
1179         return one->is_binary;
1180 }
1181
1182 static const char *funcname_pattern(const char *ident)
1183 {
1184         struct funcname_pattern *pp;
1185
1186         for (pp = funcname_pattern_list; pp; pp = pp->next)
1187                 if (!strcmp(ident, pp->name))
1188                         return pp->pattern;
1189         return NULL;
1190 }
1191
1192 static struct builtin_funcname_pattern {
1193         const char *name;
1194         const char *pattern;
1195 } builtin_funcname_pattern[] = {
1196         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1197                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1198                         "^[     ]*\\(\\([       ]*"
1199                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1200                         "[      ]*([^;]*$\\)" },
1201         { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1202 };
1203
1204 static const char *diff_funcname_pattern(struct diff_filespec *one)
1205 {
1206         const char *ident, *pattern;
1207         int i;
1208
1209         diff_filespec_check_attr(one);
1210         ident = one->funcname_pattern_ident;
1211
1212         if (!ident)
1213                 /*
1214                  * If the config file has "funcname.default" defined, that
1215                  * regexp is used; otherwise NULL is returned and xemit uses
1216                  * the built-in default.
1217                  */
1218                 return funcname_pattern("default");
1219
1220         /* Look up custom "funcname.$ident" regexp from config. */
1221         pattern = funcname_pattern(ident);
1222         if (pattern)
1223                 return pattern;
1224
1225         /*
1226          * And define built-in fallback patterns here.  Note that
1227          * these can be overridden by the user's config settings.
1228          */
1229         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1230                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1231                         return builtin_funcname_pattern[i].pattern;
1232
1233         return NULL;
1234 }
1235
1236 static void builtin_diff(const char *name_a,
1237                          const char *name_b,
1238                          struct diff_filespec *one,
1239                          struct diff_filespec *two,
1240                          const char *xfrm_msg,
1241                          struct diff_options *o,
1242                          int complete_rewrite)
1243 {
1244         mmfile_t mf1, mf2;
1245         const char *lbl[2];
1246         char *a_one, *b_two;
1247         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1248         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1249
1250         a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1251         b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1252         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1253         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1254         printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1255         if (lbl[0][0] == '/') {
1256                 /* /dev/null */
1257                 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1258                 if (xfrm_msg && xfrm_msg[0])
1259                         printf("%s%s%s\n", set, xfrm_msg, reset);
1260         }
1261         else if (lbl[1][0] == '/') {
1262                 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1263                 if (xfrm_msg && xfrm_msg[0])
1264                         printf("%s%s%s\n", set, xfrm_msg, reset);
1265         }
1266         else {
1267                 if (one->mode != two->mode) {
1268                         printf("%sold mode %06o%s\n", set, one->mode, reset);
1269                         printf("%snew mode %06o%s\n", set, two->mode, reset);
1270                 }
1271                 if (xfrm_msg && xfrm_msg[0])
1272                         printf("%s%s%s\n", set, xfrm_msg, reset);
1273                 /*
1274                  * we do not run diff between different kind
1275                  * of objects.
1276                  */
1277                 if ((one->mode ^ two->mode) & S_IFMT)
1278                         goto free_ab_and_return;
1279                 if (complete_rewrite) {
1280                         emit_rewrite_diff(name_a, name_b, one, two, o);
1281                         o->found_changes = 1;
1282                         goto free_ab_and_return;
1283                 }
1284         }
1285
1286         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1287                 die("unable to read files to diff");
1288
1289         if (!DIFF_OPT_TST(o, TEXT) &&
1290             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1291                 /* Quite common confusing case */
1292                 if (mf1.size == mf2.size &&
1293                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1294                         goto free_ab_and_return;
1295                 if (DIFF_OPT_TST(o, BINARY))
1296                         emit_binary_diff(&mf1, &mf2);
1297                 else
1298                         printf("Binary files %s and %s differ\n",
1299                                lbl[0], lbl[1]);
1300                 o->found_changes = 1;
1301         }
1302         else {
1303                 /* Crazy xdl interfaces.. */
1304                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1305                 xpparam_t xpp;
1306                 xdemitconf_t xecfg;
1307                 xdemitcb_t ecb;
1308                 struct emit_callback ecbdata;
1309                 const char *funcname_pattern;
1310
1311                 funcname_pattern = diff_funcname_pattern(one);
1312                 if (!funcname_pattern)
1313                         funcname_pattern = diff_funcname_pattern(two);
1314
1315                 memset(&xecfg, 0, sizeof(xecfg));
1316                 memset(&ecbdata, 0, sizeof(ecbdata));
1317                 ecbdata.label_path = lbl;
1318                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1319                 ecbdata.found_changesp = &o->found_changes;
1320                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1321                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1322                 xecfg.ctxlen = o->context;
1323                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1324                 if (funcname_pattern)
1325                         xdiff_set_find_func(&xecfg, funcname_pattern);
1326                 if (!diffopts)
1327                         ;
1328                 else if (!prefixcmp(diffopts, "--unified="))
1329                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1330                 else if (!prefixcmp(diffopts, "-u"))
1331                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1332                 ecb.outf = xdiff_outf;
1333                 ecb.priv = &ecbdata;
1334                 ecbdata.xm.consume = fn_out_consume;
1335                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1336                         ecbdata.diff_words =
1337                                 xcalloc(1, sizeof(struct diff_words_data));
1338                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1339                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1340                         free_diff_words_data(&ecbdata);
1341         }
1342
1343  free_ab_and_return:
1344         diff_free_filespec_data(one);
1345         diff_free_filespec_data(two);
1346         free(a_one);
1347         free(b_two);
1348         return;
1349 }
1350
1351 static void builtin_diffstat(const char *name_a, const char *name_b,
1352                              struct diff_filespec *one,
1353                              struct diff_filespec *two,
1354                              struct diffstat_t *diffstat,
1355                              struct diff_options *o,
1356                              int complete_rewrite)
1357 {
1358         mmfile_t mf1, mf2;
1359         struct diffstat_file *data;
1360
1361         data = diffstat_add(diffstat, name_a, name_b);
1362
1363         if (!one || !two) {
1364                 data->is_unmerged = 1;
1365                 return;
1366         }
1367         if (complete_rewrite) {
1368                 diff_populate_filespec(one, 0);
1369                 diff_populate_filespec(two, 0);
1370                 data->deleted = count_lines(one->data, one->size);
1371                 data->added = count_lines(two->data, two->size);
1372                 goto free_and_return;
1373         }
1374         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1375                 die("unable to read files to diff");
1376
1377         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1378                 data->is_binary = 1;
1379                 data->added = mf2.size;
1380                 data->deleted = mf1.size;
1381         } else {
1382                 /* Crazy xdl interfaces.. */
1383                 xpparam_t xpp;
1384                 xdemitconf_t xecfg;
1385                 xdemitcb_t ecb;
1386
1387                 memset(&xecfg, 0, sizeof(xecfg));
1388                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1389                 ecb.outf = xdiff_outf;
1390                 ecb.priv = diffstat;
1391                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1392         }
1393
1394  free_and_return:
1395         diff_free_filespec_data(one);
1396         diff_free_filespec_data(two);
1397 }
1398
1399 static void builtin_checkdiff(const char *name_a, const char *name_b,
1400                              struct diff_filespec *one,
1401                              struct diff_filespec *two, struct diff_options *o)
1402 {
1403         mmfile_t mf1, mf2;
1404         struct checkdiff_t data;
1405
1406         if (!two)
1407                 return;
1408
1409         memset(&data, 0, sizeof(data));
1410         data.xm.consume = checkdiff_consume;
1411         data.filename = name_b ? name_b : name_a;
1412         data.lineno = 0;
1413         data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1414         data.ws_rule = whitespace_rule(data.filename);
1415
1416         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1417                 die("unable to read files to diff");
1418
1419         if (diff_filespec_is_binary(two))
1420                 goto free_and_return;
1421         else {
1422                 /* Crazy xdl interfaces.. */
1423                 xpparam_t xpp;
1424                 xdemitconf_t xecfg;
1425                 xdemitcb_t ecb;
1426
1427                 memset(&xecfg, 0, sizeof(xecfg));
1428                 xpp.flags = XDF_NEED_MINIMAL;
1429                 ecb.outf = xdiff_outf;
1430                 ecb.priv = &data;
1431                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1432         }
1433  free_and_return:
1434         diff_free_filespec_data(one);
1435         diff_free_filespec_data(two);
1436         if (data.status)
1437                 DIFF_OPT_SET(o, CHECK_FAILED);
1438 }
1439
1440 struct diff_filespec *alloc_filespec(const char *path)
1441 {
1442         int namelen = strlen(path);
1443         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1444
1445         memset(spec, 0, sizeof(*spec));
1446         spec->path = (char *)(spec + 1);
1447         memcpy(spec->path, path, namelen+1);
1448         spec->count = 1;
1449         return spec;
1450 }
1451
1452 void free_filespec(struct diff_filespec *spec)
1453 {
1454         if (!--spec->count) {
1455                 diff_free_filespec_data(spec);
1456                 free(spec);
1457         }
1458 }
1459
1460 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1461                    unsigned short mode)
1462 {
1463         if (mode) {
1464                 spec->mode = canon_mode(mode);
1465                 hashcpy(spec->sha1, sha1);
1466                 spec->sha1_valid = !is_null_sha1(sha1);
1467         }
1468 }
1469
1470 /*
1471  * Given a name and sha1 pair, if the index tells us the file in
1472  * the work tree has that object contents, return true, so that
1473  * prepare_temp_file() does not have to inflate and extract.
1474  */
1475 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1476 {
1477         struct cache_entry *ce;
1478         struct stat st;
1479         int pos, len;
1480
1481         /* We do not read the cache ourselves here, because the
1482          * benchmark with my previous version that always reads cache
1483          * shows that it makes things worse for diff-tree comparing
1484          * two linux-2.6 kernel trees in an already checked out work
1485          * tree.  This is because most diff-tree comparisons deal with
1486          * only a small number of files, while reading the cache is
1487          * expensive for a large project, and its cost outweighs the
1488          * savings we get by not inflating the object to a temporary
1489          * file.  Practically, this code only helps when we are used
1490          * by diff-cache --cached, which does read the cache before
1491          * calling us.
1492          */
1493         if (!active_cache)
1494                 return 0;
1495
1496         /* We want to avoid the working directory if our caller
1497          * doesn't need the data in a normal file, this system
1498          * is rather slow with its stat/open/mmap/close syscalls,
1499          * and the object is contained in a pack file.  The pack
1500          * is probably already open and will be faster to obtain
1501          * the data through than the working directory.  Loose
1502          * objects however would tend to be slower as they need
1503          * to be individually opened and inflated.
1504          */
1505         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1506                 return 0;
1507
1508         len = strlen(name);
1509         pos = cache_name_pos(name, len);
1510         if (pos < 0)
1511                 return 0;
1512         ce = active_cache[pos];
1513
1514         /*
1515          * This is not the sha1 we are looking for, or
1516          * unreusable because it is not a regular file.
1517          */
1518         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1519                 return 0;
1520
1521         /*
1522          * If ce matches the file in the work tree, we can reuse it.
1523          */
1524         if (ce_uptodate(ce) ||
1525             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1526                 return 1;
1527
1528         return 0;
1529 }
1530
1531 static int populate_from_stdin(struct diff_filespec *s)
1532 {
1533         struct strbuf buf;
1534         size_t size = 0;
1535
1536         strbuf_init(&buf, 0);
1537         if (strbuf_read(&buf, 0, 0) < 0)
1538                 return error("error while reading from stdin %s",
1539                                      strerror(errno));
1540
1541         s->should_munmap = 0;
1542         s->data = strbuf_detach(&buf, &size);
1543         s->size = size;
1544         s->should_free = 1;
1545         return 0;
1546 }
1547
1548 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1549 {
1550         int len;
1551         char *data = xmalloc(100);
1552         len = snprintf(data, 100,
1553                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1554         s->data = data;
1555         s->size = len;
1556         s->should_free = 1;
1557         if (size_only) {
1558                 s->data = NULL;
1559                 free(data);
1560         }
1561         return 0;
1562 }
1563
1564 /*
1565  * While doing rename detection and pickaxe operation, we may need to
1566  * grab the data for the blob (or file) for our own in-core comparison.
1567  * diff_filespec has data and size fields for this purpose.
1568  */
1569 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1570 {
1571         int err = 0;
1572         if (!DIFF_FILE_VALID(s))
1573                 die("internal error: asking to populate invalid file.");
1574         if (S_ISDIR(s->mode))
1575                 return -1;
1576
1577         if (s->data)
1578                 return 0;
1579
1580         if (size_only && 0 < s->size)
1581                 return 0;
1582
1583         if (S_ISGITLINK(s->mode))
1584                 return diff_populate_gitlink(s, size_only);
1585
1586         if (!s->sha1_valid ||
1587             reuse_worktree_file(s->path, s->sha1, 0)) {
1588                 struct strbuf buf;
1589                 struct stat st;
1590                 int fd;
1591
1592                 if (!strcmp(s->path, "-"))
1593                         return populate_from_stdin(s);
1594
1595                 if (lstat(s->path, &st) < 0) {
1596                         if (errno == ENOENT) {
1597                         err_empty:
1598                                 err = -1;
1599                         empty:
1600                                 s->data = (char *)"";
1601                                 s->size = 0;
1602                                 return err;
1603                         }
1604                 }
1605                 s->size = xsize_t(st.st_size);
1606                 if (!s->size)
1607                         goto empty;
1608                 if (size_only)
1609                         return 0;
1610                 if (S_ISLNK(st.st_mode)) {
1611                         int ret;
1612                         s->data = xmalloc(s->size);
1613                         s->should_free = 1;
1614                         ret = readlink(s->path, s->data, s->size);
1615                         if (ret < 0) {
1616                                 free(s->data);
1617                                 goto err_empty;
1618                         }
1619                         return 0;
1620                 }
1621                 fd = open(s->path, O_RDONLY);
1622                 if (fd < 0)
1623                         goto err_empty;
1624                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1625                 close(fd);
1626                 s->should_munmap = 1;
1627
1628                 /*
1629                  * Convert from working tree format to canonical git format
1630                  */
1631                 strbuf_init(&buf, 0);
1632                 if (convert_to_git(s->path, s->data, s->size, &buf)) {
1633                         size_t size = 0;
1634                         munmap(s->data, s->size);
1635                         s->should_munmap = 0;
1636                         s->data = strbuf_detach(&buf, &size);
1637                         s->size = size;
1638                         s->should_free = 1;
1639                 }
1640         }
1641         else {
1642                 enum object_type type;
1643                 if (size_only)
1644                         type = sha1_object_info(s->sha1, &s->size);
1645                 else {
1646                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1647                         s->should_free = 1;
1648                 }
1649         }
1650         return 0;
1651 }
1652
1653 void diff_free_filespec_blob(struct diff_filespec *s)
1654 {
1655         if (s->should_free)
1656                 free(s->data);
1657         else if (s->should_munmap)
1658                 munmap(s->data, s->size);
1659
1660         if (s->should_free || s->should_munmap) {
1661                 s->should_free = s->should_munmap = 0;
1662                 s->data = NULL;
1663         }
1664 }
1665
1666 void diff_free_filespec_data(struct diff_filespec *s)
1667 {
1668         diff_free_filespec_blob(s);
1669         free(s->cnt_data);
1670         s->cnt_data = NULL;
1671 }
1672
1673 static void prep_temp_blob(struct diff_tempfile *temp,
1674                            void *blob,
1675                            unsigned long size,
1676                            const unsigned char *sha1,
1677                            int mode)
1678 {
1679         int fd;
1680
1681         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1682         if (fd < 0)
1683                 die("unable to create temp-file: %s", strerror(errno));
1684         if (write_in_full(fd, blob, size) != size)
1685                 die("unable to write temp-file");
1686         close(fd);
1687         temp->name = temp->tmp_path;
1688         strcpy(temp->hex, sha1_to_hex(sha1));
1689         temp->hex[40] = 0;
1690         sprintf(temp->mode, "%06o", mode);
1691 }
1692
1693 static void prepare_temp_file(const char *name,
1694                               struct diff_tempfile *temp,
1695                               struct diff_filespec *one)
1696 {
1697         if (!DIFF_FILE_VALID(one)) {
1698         not_a_valid_file:
1699                 /* A '-' entry produces this for file-2, and
1700                  * a '+' entry produces this for file-1.
1701                  */
1702                 temp->name = "/dev/null";
1703                 strcpy(temp->hex, ".");
1704                 strcpy(temp->mode, ".");
1705                 return;
1706         }
1707
1708         if (!one->sha1_valid ||
1709             reuse_worktree_file(name, one->sha1, 1)) {
1710                 struct stat st;
1711                 if (lstat(name, &st) < 0) {
1712                         if (errno == ENOENT)
1713                                 goto not_a_valid_file;
1714                         die("stat(%s): %s", name, strerror(errno));
1715                 }
1716                 if (S_ISLNK(st.st_mode)) {
1717                         int ret;
1718                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1719                         size_t sz = xsize_t(st.st_size);
1720                         if (sizeof(buf) <= st.st_size)
1721                                 die("symlink too long: %s", name);
1722                         ret = readlink(name, buf, sz);
1723                         if (ret < 0)
1724                                 die("readlink(%s)", name);
1725                         prep_temp_blob(temp, buf, sz,
1726                                        (one->sha1_valid ?
1727                                         one->sha1 : null_sha1),
1728                                        (one->sha1_valid ?
1729                                         one->mode : S_IFLNK));
1730                 }
1731                 else {
1732                         /* we can borrow from the file in the work tree */
1733                         temp->name = name;
1734                         if (!one->sha1_valid)
1735                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1736                         else
1737                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1738                         /* Even though we may sometimes borrow the
1739                          * contents from the work tree, we always want
1740                          * one->mode.  mode is trustworthy even when
1741                          * !(one->sha1_valid), as long as
1742                          * DIFF_FILE_VALID(one).
1743                          */
1744                         sprintf(temp->mode, "%06o", one->mode);
1745                 }
1746                 return;
1747         }
1748         else {
1749                 if (diff_populate_filespec(one, 0))
1750                         die("cannot read data blob for %s", one->path);
1751                 prep_temp_blob(temp, one->data, one->size,
1752                                one->sha1, one->mode);
1753         }
1754 }
1755
1756 static void remove_tempfile(void)
1757 {
1758         int i;
1759
1760         for (i = 0; i < 2; i++)
1761                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1762                         unlink(diff_temp[i].name);
1763                         diff_temp[i].name = NULL;
1764                 }
1765 }
1766
1767 static void remove_tempfile_on_signal(int signo)
1768 {
1769         remove_tempfile();
1770         signal(SIGINT, SIG_DFL);
1771         raise(signo);
1772 }
1773
1774 /* An external diff command takes:
1775  *
1776  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1777  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1778  *
1779  */
1780 static void run_external_diff(const char *pgm,
1781                               const char *name,
1782                               const char *other,
1783                               struct diff_filespec *one,
1784                               struct diff_filespec *two,
1785                               const char *xfrm_msg,
1786                               int complete_rewrite)
1787 {
1788         const char *spawn_arg[10];
1789         struct diff_tempfile *temp = diff_temp;
1790         int retval;
1791         static int atexit_asked = 0;
1792         const char *othername;
1793         const char **arg = &spawn_arg[0];
1794
1795         othername = (other? other : name);
1796         if (one && two) {
1797                 prepare_temp_file(name, &temp[0], one);
1798                 prepare_temp_file(othername, &temp[1], two);
1799                 if (! atexit_asked &&
1800                     (temp[0].name == temp[0].tmp_path ||
1801                      temp[1].name == temp[1].tmp_path)) {
1802                         atexit_asked = 1;
1803                         atexit(remove_tempfile);
1804                 }
1805                 signal(SIGINT, remove_tempfile_on_signal);
1806         }
1807
1808         if (one && two) {
1809                 *arg++ = pgm;
1810                 *arg++ = name;
1811                 *arg++ = temp[0].name;
1812                 *arg++ = temp[0].hex;
1813                 *arg++ = temp[0].mode;
1814                 *arg++ = temp[1].name;
1815                 *arg++ = temp[1].hex;
1816                 *arg++ = temp[1].mode;
1817                 if (other) {
1818                         *arg++ = other;
1819                         *arg++ = xfrm_msg;
1820                 }
1821         } else {
1822                 *arg++ = pgm;
1823                 *arg++ = name;
1824         }
1825         *arg = NULL;
1826         fflush(NULL);
1827         retval = run_command_v_opt(spawn_arg, 0);
1828         remove_tempfile();
1829         if (retval) {
1830                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1831                 exit(1);
1832         }
1833 }
1834
1835 static const char *external_diff_attr(const char *name)
1836 {
1837         struct git_attr_check attr_diff_check;
1838
1839         setup_diff_attr_check(&attr_diff_check);
1840         if (!git_checkattr(name, 1, &attr_diff_check)) {
1841                 const char *value = attr_diff_check.value;
1842                 if (!ATTR_TRUE(value) &&
1843                     !ATTR_FALSE(value) &&
1844                     !ATTR_UNSET(value)) {
1845                         struct ll_diff_driver *drv;
1846
1847                         for (drv = user_diff; drv; drv = drv->next)
1848                                 if (!strcmp(drv->name, value))
1849                                         return drv->cmd;
1850                 }
1851         }
1852         return NULL;
1853 }
1854
1855 static void run_diff_cmd(const char *pgm,
1856                          const char *name,
1857                          const char *other,
1858                          struct diff_filespec *one,
1859                          struct diff_filespec *two,
1860                          const char *xfrm_msg,
1861                          struct diff_options *o,
1862                          int complete_rewrite)
1863 {
1864         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
1865                 pgm = NULL;
1866         else {
1867                 const char *cmd = external_diff_attr(name);
1868                 if (cmd)
1869                         pgm = cmd;
1870         }
1871
1872         if (pgm) {
1873                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1874                                   complete_rewrite);
1875                 return;
1876         }
1877         if (one && two)
1878                 builtin_diff(name, other ? other : name,
1879                              one, two, xfrm_msg, o, complete_rewrite);
1880         else
1881                 printf("* Unmerged path %s\n", name);
1882 }
1883
1884 static void diff_fill_sha1_info(struct diff_filespec *one)
1885 {
1886         if (DIFF_FILE_VALID(one)) {
1887                 if (!one->sha1_valid) {
1888                         struct stat st;
1889                         if (!strcmp(one->path, "-")) {
1890                                 hashcpy(one->sha1, null_sha1);
1891                                 return;
1892                         }
1893                         if (lstat(one->path, &st) < 0)
1894                                 die("stat %s", one->path);
1895                         if (index_path(one->sha1, one->path, &st, 0))
1896                                 die("cannot hash %s\n", one->path);
1897                 }
1898         }
1899         else
1900                 hashclr(one->sha1);
1901 }
1902
1903 static int similarity_index(struct diff_filepair *p)
1904 {
1905         return p->score * 100 / MAX_SCORE;
1906 }
1907
1908 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1909 {
1910         const char *pgm = external_diff();
1911         struct strbuf msg;
1912         char *xfrm_msg;
1913         struct diff_filespec *one = p->one;
1914         struct diff_filespec *two = p->two;
1915         const char *name;
1916         const char *other;
1917         int complete_rewrite = 0;
1918
1919
1920         if (DIFF_PAIR_UNMERGED(p)) {
1921                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1922                 return;
1923         }
1924
1925         name  = p->one->path;
1926         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1927         diff_fill_sha1_info(one);
1928         diff_fill_sha1_info(two);
1929
1930         strbuf_init(&msg, PATH_MAX * 2 + 300);
1931         switch (p->status) {
1932         case DIFF_STATUS_COPIED:
1933                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1934                 strbuf_addstr(&msg, "\ncopy from ");
1935                 quote_c_style(name, &msg, NULL, 0);
1936                 strbuf_addstr(&msg, "\ncopy to ");
1937                 quote_c_style(other, &msg, NULL, 0);
1938                 strbuf_addch(&msg, '\n');
1939                 break;
1940         case DIFF_STATUS_RENAMED:
1941                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1942                 strbuf_addstr(&msg, "\nrename from ");
1943                 quote_c_style(name, &msg, NULL, 0);
1944                 strbuf_addstr(&msg, "\nrename to ");
1945                 quote_c_style(other, &msg, NULL, 0);
1946                 strbuf_addch(&msg, '\n');
1947                 break;
1948         case DIFF_STATUS_MODIFIED:
1949                 if (p->score) {
1950                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
1951                                         similarity_index(p));
1952                         complete_rewrite = 1;
1953                         break;
1954                 }
1955                 /* fallthru */
1956         default:
1957                 /* nothing */
1958                 ;
1959         }
1960
1961         if (hashcmp(one->sha1, two->sha1)) {
1962                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
1963
1964                 if (DIFF_OPT_TST(o, BINARY)) {
1965                         mmfile_t mf;
1966                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
1967                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
1968                                 abbrev = 40;
1969                 }
1970                 strbuf_addf(&msg, "index %.*s..%.*s",
1971                                 abbrev, sha1_to_hex(one->sha1),
1972                                 abbrev, sha1_to_hex(two->sha1));
1973                 if (one->mode == two->mode)
1974                         strbuf_addf(&msg, " %06o", one->mode);
1975                 strbuf_addch(&msg, '\n');
1976         }
1977
1978         if (msg.len)
1979                 strbuf_setlen(&msg, msg.len - 1);
1980         xfrm_msg = msg.len ? msg.buf : NULL;
1981
1982         if (!pgm &&
1983             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1984             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1985                 /* a filepair that changes between file and symlink
1986                  * needs to be split into deletion and creation.
1987                  */
1988                 struct diff_filespec *null = alloc_filespec(two->path);
1989                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1990                 free(null);
1991                 null = alloc_filespec(one->path);
1992                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1993                 free(null);
1994         }
1995         else
1996                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1997                              complete_rewrite);
1998
1999         strbuf_release(&msg);
2000 }
2001
2002 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2003                          struct diffstat_t *diffstat)
2004 {
2005         const char *name;
2006         const char *other;
2007         int complete_rewrite = 0;
2008
2009         if (DIFF_PAIR_UNMERGED(p)) {
2010                 /* unmerged */
2011                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2012                 return;
2013         }
2014
2015         name = p->one->path;
2016         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2017
2018         diff_fill_sha1_info(p->one);
2019         diff_fill_sha1_info(p->two);
2020
2021         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2022                 complete_rewrite = 1;
2023         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2024 }
2025
2026 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2027 {
2028         const char *name;
2029         const char *other;
2030
2031         if (DIFF_PAIR_UNMERGED(p)) {
2032                 /* unmerged */
2033                 return;
2034         }
2035
2036         name = p->one->path;
2037         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2038
2039         diff_fill_sha1_info(p->one);
2040         diff_fill_sha1_info(p->two);
2041
2042         builtin_checkdiff(name, other, p->one, p->two, o);
2043 }
2044
2045 void diff_setup(struct diff_options *options)
2046 {
2047         memset(options, 0, sizeof(*options));
2048         options->line_termination = '\n';
2049         options->break_opt = -1;
2050         options->rename_limit = -1;
2051         options->context = 3;
2052         options->msg_sep = "";
2053
2054         options->change = diff_change;
2055         options->add_remove = diff_addremove;
2056         if (diff_use_color_default)
2057                 DIFF_OPT_SET(options, COLOR_DIFF);
2058         else
2059                 DIFF_OPT_CLR(options, COLOR_DIFF);
2060         options->detect_rename = diff_detect_rename_default;
2061
2062         options->a_prefix = "a/";
2063         options->b_prefix = "b/";
2064 }
2065
2066 int diff_setup_done(struct diff_options *options)
2067 {
2068         int count = 0;
2069
2070         if (options->output_format & DIFF_FORMAT_NAME)
2071                 count++;
2072         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2073                 count++;
2074         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2075                 count++;
2076         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2077                 count++;
2078         if (count > 1)
2079                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2080
2081         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2082                 options->detect_rename = DIFF_DETECT_COPY;
2083
2084         if (options->output_format & (DIFF_FORMAT_NAME |
2085                                       DIFF_FORMAT_NAME_STATUS |
2086                                       DIFF_FORMAT_CHECKDIFF |
2087                                       DIFF_FORMAT_NO_OUTPUT))
2088                 options->output_format &= ~(DIFF_FORMAT_RAW |
2089                                             DIFF_FORMAT_NUMSTAT |
2090                                             DIFF_FORMAT_DIFFSTAT |
2091                                             DIFF_FORMAT_SHORTSTAT |
2092                                             DIFF_FORMAT_SUMMARY |
2093                                             DIFF_FORMAT_PATCH);
2094
2095         /*
2096          * These cases always need recursive; we do not drop caller-supplied
2097          * recursive bits for other formats here.
2098          */
2099         if (options->output_format & (DIFF_FORMAT_PATCH |
2100                                       DIFF_FORMAT_NUMSTAT |
2101                                       DIFF_FORMAT_DIFFSTAT |
2102                                       DIFF_FORMAT_SHORTSTAT |
2103                                       DIFF_FORMAT_SUMMARY |
2104                                       DIFF_FORMAT_CHECKDIFF))
2105                 DIFF_OPT_SET(options, RECURSIVE);
2106         /*
2107          * Also pickaxe would not work very well if you do not say recursive
2108          */
2109         if (options->pickaxe)
2110                 DIFF_OPT_SET(options, RECURSIVE);
2111
2112         if (options->detect_rename && options->rename_limit < 0)
2113                 options->rename_limit = diff_rename_limit_default;
2114         if (options->setup & DIFF_SETUP_USE_CACHE) {
2115                 if (!active_cache)
2116                         /* read-cache does not die even when it fails
2117                          * so it is safe for us to do this here.  Also
2118                          * it does not smudge active_cache or active_nr
2119                          * when it fails, so we do not have to worry about
2120                          * cleaning it up ourselves either.
2121                          */
2122                         read_cache();
2123         }
2124         if (options->abbrev <= 0 || 40 < options->abbrev)
2125                 options->abbrev = 40; /* full */
2126
2127         /*
2128          * It does not make sense to show the first hit we happened
2129          * to have found.  It does not make sense not to return with
2130          * exit code in such a case either.
2131          */
2132         if (DIFF_OPT_TST(options, QUIET)) {
2133                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2134                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2135         }
2136
2137         /*
2138          * If we postprocess in diffcore, we cannot simply return
2139          * upon the first hit.  We need to run diff as usual.
2140          */
2141         if (options->pickaxe || options->filter)
2142                 DIFF_OPT_CLR(options, QUIET);
2143
2144         return 0;
2145 }
2146
2147 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2148 {
2149         char c, *eq;
2150         int len;
2151
2152         if (*arg != '-')
2153                 return 0;
2154         c = *++arg;
2155         if (!c)
2156                 return 0;
2157         if (c == arg_short) {
2158                 c = *++arg;
2159                 if (!c)
2160                         return 1;
2161                 if (val && isdigit(c)) {
2162                         char *end;
2163                         int n = strtoul(arg, &end, 10);
2164                         if (*end)
2165                                 return 0;
2166                         *val = n;
2167                         return 1;
2168                 }
2169                 return 0;
2170         }
2171         if (c != '-')
2172                 return 0;
2173         arg++;
2174         eq = strchr(arg, '=');
2175         if (eq)
2176                 len = eq - arg;
2177         else
2178                 len = strlen(arg);
2179         if (!len || strncmp(arg, arg_long, len))
2180                 return 0;
2181         if (eq) {
2182                 int n;
2183                 char *end;
2184                 if (!isdigit(*++eq))
2185                         return 0;
2186                 n = strtoul(eq, &end, 10);
2187                 if (*end)
2188                         return 0;
2189                 *val = n;
2190         }
2191         return 1;
2192 }
2193
2194 static int diff_scoreopt_parse(const char *opt);
2195
2196 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2197 {
2198         const char *arg = av[0];
2199
2200         /* Output format options */
2201         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2202                 options->output_format |= DIFF_FORMAT_PATCH;
2203         else if (opt_arg(arg, 'U', "unified", &options->context))
2204                 options->output_format |= DIFF_FORMAT_PATCH;
2205         else if (!strcmp(arg, "--raw"))
2206                 options->output_format |= DIFF_FORMAT_RAW;
2207         else if (!strcmp(arg, "--patch-with-raw"))
2208                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2209         else if (!strcmp(arg, "--numstat"))
2210                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2211         else if (!strcmp(arg, "--shortstat"))
2212                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2213         else if (!strcmp(arg, "--check"))
2214                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2215         else if (!strcmp(arg, "--summary"))
2216                 options->output_format |= DIFF_FORMAT_SUMMARY;
2217         else if (!strcmp(arg, "--patch-with-stat"))
2218                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2219         else if (!strcmp(arg, "--name-only"))
2220                 options->output_format |= DIFF_FORMAT_NAME;
2221         else if (!strcmp(arg, "--name-status"))
2222                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2223         else if (!strcmp(arg, "-s"))
2224                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2225         else if (!prefixcmp(arg, "--stat")) {
2226                 char *end;
2227                 int width = options->stat_width;
2228                 int name_width = options->stat_name_width;
2229                 arg += 6;
2230                 end = (char *)arg;
2231
2232                 switch (*arg) {
2233                 case '-':
2234                         if (!prefixcmp(arg, "-width="))
2235                                 width = strtoul(arg + 7, &end, 10);
2236                         else if (!prefixcmp(arg, "-name-width="))
2237                                 name_width = strtoul(arg + 12, &end, 10);
2238                         break;
2239                 case '=':
2240                         width = strtoul(arg+1, &end, 10);
2241                         if (*end == ',')
2242                                 name_width = strtoul(end+1, &end, 10);
2243                 }
2244
2245                 /* Important! This checks all the error cases! */
2246                 if (*end)
2247                         return 0;
2248                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2249                 options->stat_name_width = name_width;
2250                 options->stat_width = width;
2251         }
2252
2253         /* renames options */
2254         else if (!prefixcmp(arg, "-B")) {
2255                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2256                         return -1;
2257         }
2258         else if (!prefixcmp(arg, "-M")) {
2259                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2260                         return -1;
2261                 options->detect_rename = DIFF_DETECT_RENAME;
2262         }
2263         else if (!prefixcmp(arg, "-C")) {
2264                 if (options->detect_rename == DIFF_DETECT_COPY)
2265                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2266                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2267                         return -1;
2268                 options->detect_rename = DIFF_DETECT_COPY;
2269         }
2270         else if (!strcmp(arg, "--no-renames"))
2271                 options->detect_rename = 0;
2272
2273         /* xdiff options */
2274         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2275                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2276         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2277                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2278         else if (!strcmp(arg, "--ignore-space-at-eol"))
2279                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2280
2281         /* flags options */
2282         else if (!strcmp(arg, "--binary")) {
2283                 options->output_format |= DIFF_FORMAT_PATCH;
2284                 DIFF_OPT_SET(options, BINARY);
2285         }
2286         else if (!strcmp(arg, "--full-index"))
2287                 DIFF_OPT_SET(options, FULL_INDEX);
2288         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2289                 DIFF_OPT_SET(options, TEXT);
2290         else if (!strcmp(arg, "-R"))
2291                 DIFF_OPT_SET(options, REVERSE_DIFF);
2292         else if (!strcmp(arg, "--find-copies-harder"))
2293                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2294         else if (!strcmp(arg, "--follow"))
2295                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2296         else if (!strcmp(arg, "--color"))
2297                 DIFF_OPT_SET(options, COLOR_DIFF);
2298         else if (!strcmp(arg, "--no-color"))
2299                 DIFF_OPT_CLR(options, COLOR_DIFF);
2300         else if (!strcmp(arg, "--color-words"))
2301                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2302         else if (!strcmp(arg, "--exit-code"))
2303                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2304         else if (!strcmp(arg, "--quiet"))
2305                 DIFF_OPT_SET(options, QUIET);
2306         else if (!strcmp(arg, "--ext-diff"))
2307                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2308         else if (!strcmp(arg, "--no-ext-diff"))
2309                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2310
2311         /* misc options */
2312         else if (!strcmp(arg, "-z"))
2313                 options->line_termination = 0;
2314         else if (!prefixcmp(arg, "-l"))
2315                 options->rename_limit = strtoul(arg+2, NULL, 10);
2316         else if (!prefixcmp(arg, "-S"))
2317                 options->pickaxe = arg + 2;
2318         else if (!strcmp(arg, "--pickaxe-all"))
2319                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2320         else if (!strcmp(arg, "--pickaxe-regex"))
2321                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2322         else if (!prefixcmp(arg, "-O"))
2323                 options->orderfile = arg + 2;
2324         else if (!prefixcmp(arg, "--diff-filter="))
2325                 options->filter = arg + 14;
2326         else if (!strcmp(arg, "--abbrev"))
2327                 options->abbrev = DEFAULT_ABBREV;
2328         else if (!prefixcmp(arg, "--abbrev=")) {
2329                 options->abbrev = strtoul(arg + 9, NULL, 10);
2330                 if (options->abbrev < MINIMUM_ABBREV)
2331                         options->abbrev = MINIMUM_ABBREV;
2332                 else if (40 < options->abbrev)
2333                         options->abbrev = 40;
2334         }
2335         else if (!prefixcmp(arg, "--src-prefix="))
2336                 options->a_prefix = arg + 13;
2337         else if (!prefixcmp(arg, "--dst-prefix="))
2338                 options->b_prefix = arg + 13;
2339         else if (!strcmp(arg, "--no-prefix"))
2340                 options->a_prefix = options->b_prefix = "";
2341         else
2342                 return 0;
2343         return 1;
2344 }
2345
2346 static int parse_num(const char **cp_p)
2347 {
2348         unsigned long num, scale;
2349         int ch, dot;
2350         const char *cp = *cp_p;
2351
2352         num = 0;
2353         scale = 1;
2354         dot = 0;
2355         for(;;) {
2356                 ch = *cp;
2357                 if ( !dot && ch == '.' ) {
2358                         scale = 1;
2359                         dot = 1;
2360                 } else if ( ch == '%' ) {
2361                         scale = dot ? scale*100 : 100;
2362                         cp++;   /* % is always at the end */
2363                         break;
2364                 } else if ( ch >= '0' && ch <= '9' ) {
2365                         if ( scale < 100000 ) {
2366                                 scale *= 10;
2367                                 num = (num*10) + (ch-'0');
2368                         }
2369                 } else {
2370                         break;
2371                 }
2372                 cp++;
2373         }
2374         *cp_p = cp;
2375
2376         /* user says num divided by scale and we say internally that
2377          * is MAX_SCORE * num / scale.
2378          */
2379         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2380 }
2381
2382 static int diff_scoreopt_parse(const char *opt)
2383 {
2384         int opt1, opt2, cmd;
2385
2386         if (*opt++ != '-')
2387                 return -1;
2388         cmd = *opt++;
2389         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2390                 return -1; /* that is not a -M, -C nor -B option */
2391
2392         opt1 = parse_num(&opt);
2393         if (cmd != 'B')
2394                 opt2 = 0;
2395         else {
2396                 if (*opt == 0)
2397                         opt2 = 0;
2398                 else if (*opt != '/')
2399                         return -1; /* we expect -B80/99 or -B80 */
2400                 else {
2401                         opt++;
2402                         opt2 = parse_num(&opt);
2403                 }
2404         }
2405         if (*opt != 0)
2406                 return -1;
2407         return opt1 | (opt2 << 16);
2408 }
2409
2410 struct diff_queue_struct diff_queued_diff;
2411
2412 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2413 {
2414         if (queue->alloc <= queue->nr) {
2415                 queue->alloc = alloc_nr(queue->alloc);
2416                 queue->queue = xrealloc(queue->queue,
2417                                         sizeof(dp) * queue->alloc);
2418         }
2419         queue->queue[queue->nr++] = dp;
2420 }
2421
2422 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2423                                  struct diff_filespec *one,
2424                                  struct diff_filespec *two)
2425 {
2426         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2427         dp->one = one;
2428         dp->two = two;
2429         if (queue)
2430                 diff_q(queue, dp);
2431         return dp;
2432 }
2433
2434 void diff_free_filepair(struct diff_filepair *p)
2435 {
2436         free_filespec(p->one);
2437         free_filespec(p->two);
2438         free(p);
2439 }
2440
2441 /* This is different from find_unique_abbrev() in that
2442  * it stuffs the result with dots for alignment.
2443  */
2444 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2445 {
2446         int abblen;
2447         const char *abbrev;
2448         if (len == 40)
2449                 return sha1_to_hex(sha1);
2450
2451         abbrev = find_unique_abbrev(sha1, len);
2452         if (!abbrev)
2453                 return sha1_to_hex(sha1);
2454         abblen = strlen(abbrev);
2455         if (abblen < 37) {
2456                 static char hex[41];
2457                 if (len < abblen && abblen <= len + 2)
2458                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2459                 else
2460                         sprintf(hex, "%s...", abbrev);
2461                 return hex;
2462         }
2463         return sha1_to_hex(sha1);
2464 }
2465
2466 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2467 {
2468         int line_termination = opt->line_termination;
2469         int inter_name_termination = line_termination ? '\t' : '\0';
2470
2471         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2472                 printf(":%06o %06o %s ", p->one->mode, p->two->mode,
2473                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
2474                 printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2475         }
2476         if (p->score) {
2477                 printf("%c%03d%c", p->status, similarity_index(p),
2478                            inter_name_termination);
2479         } else {
2480                 printf("%c%c", p->status, inter_name_termination);
2481         }
2482
2483         if (p->status == DIFF_STATUS_COPIED || p->status == DIFF_STATUS_RENAMED) {
2484                 write_name_quoted(p->one->path, stdout, inter_name_termination);
2485                 write_name_quoted(p->two->path, stdout, line_termination);
2486         } else {
2487                 const char *path = p->one->mode ? p->one->path : p->two->path;
2488                 write_name_quoted(path, stdout, line_termination);
2489         }
2490 }
2491
2492 int diff_unmodified_pair(struct diff_filepair *p)
2493 {
2494         /* This function is written stricter than necessary to support
2495          * the currently implemented transformers, but the idea is to
2496          * let transformers to produce diff_filepairs any way they want,
2497          * and filter and clean them up here before producing the output.
2498          */
2499         struct diff_filespec *one = p->one, *two = p->two;
2500
2501         if (DIFF_PAIR_UNMERGED(p))
2502                 return 0; /* unmerged is interesting */
2503
2504         /* deletion, addition, mode or type change
2505          * and rename are all interesting.
2506          */
2507         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2508             DIFF_PAIR_MODE_CHANGED(p) ||
2509             strcmp(one->path, two->path))
2510                 return 0;
2511
2512         /* both are valid and point at the same path.  that is, we are
2513          * dealing with a change.
2514          */
2515         if (one->sha1_valid && two->sha1_valid &&
2516             !hashcmp(one->sha1, two->sha1))
2517                 return 1; /* no change */
2518         if (!one->sha1_valid && !two->sha1_valid)
2519                 return 1; /* both look at the same file on the filesystem. */
2520         return 0;
2521 }
2522
2523 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2524 {
2525         if (diff_unmodified_pair(p))
2526                 return;
2527
2528         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2529             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2530                 return; /* no tree diffs in patch format */
2531
2532         run_diff(p, o);
2533 }
2534
2535 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2536                             struct diffstat_t *diffstat)
2537 {
2538         if (diff_unmodified_pair(p))
2539                 return;
2540
2541         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2542             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2543                 return; /* no tree diffs in patch format */
2544
2545         run_diffstat(p, o, diffstat);
2546 }
2547
2548 static void diff_flush_checkdiff(struct diff_filepair *p,
2549                 struct diff_options *o)
2550 {
2551         if (diff_unmodified_pair(p))
2552                 return;
2553
2554         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2555             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2556                 return; /* no tree diffs in patch format */
2557
2558         run_checkdiff(p, o);
2559 }
2560
2561 int diff_queue_is_empty(void)
2562 {
2563         struct diff_queue_struct *q = &diff_queued_diff;
2564         int i;
2565         for (i = 0; i < q->nr; i++)
2566                 if (!diff_unmodified_pair(q->queue[i]))
2567                         return 0;
2568         return 1;
2569 }
2570
2571 #if DIFF_DEBUG
2572 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2573 {
2574         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2575                 x, one ? one : "",
2576                 s->path,
2577                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2578                 s->mode,
2579                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2580         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2581                 x, one ? one : "",
2582                 s->size, s->xfrm_flags);
2583 }
2584
2585 void diff_debug_filepair(const struct diff_filepair *p, int i)
2586 {
2587         diff_debug_filespec(p->one, i, "one");
2588         diff_debug_filespec(p->two, i, "two");
2589         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2590                 p->score, p->status ? p->status : '?',
2591                 p->one->rename_used, p->broken_pair);
2592 }
2593
2594 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2595 {
2596         int i;
2597         if (msg)
2598                 fprintf(stderr, "%s\n", msg);
2599         fprintf(stderr, "q->nr = %d\n", q->nr);
2600         for (i = 0; i < q->nr; i++) {
2601                 struct diff_filepair *p = q->queue[i];
2602                 diff_debug_filepair(p, i);
2603         }
2604 }
2605 #endif
2606
2607 static void diff_resolve_rename_copy(void)
2608 {
2609         int i;
2610         struct diff_filepair *p;
2611         struct diff_queue_struct *q = &diff_queued_diff;
2612
2613         diff_debug_queue("resolve-rename-copy", q);
2614
2615         for (i = 0; i < q->nr; i++) {
2616                 p = q->queue[i];
2617                 p->status = 0; /* undecided */
2618                 if (DIFF_PAIR_UNMERGED(p))
2619                         p->status = DIFF_STATUS_UNMERGED;
2620                 else if (!DIFF_FILE_VALID(p->one))
2621                         p->status = DIFF_STATUS_ADDED;
2622                 else if (!DIFF_FILE_VALID(p->two))
2623                         p->status = DIFF_STATUS_DELETED;
2624                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2625                         p->status = DIFF_STATUS_TYPE_CHANGED;
2626
2627                 /* from this point on, we are dealing with a pair
2628                  * whose both sides are valid and of the same type, i.e.
2629                  * either in-place edit or rename/copy edit.
2630                  */
2631                 else if (DIFF_PAIR_RENAME(p)) {
2632                         /*
2633                          * A rename might have re-connected a broken
2634                          * pair up, causing the pathnames to be the
2635                          * same again. If so, that's not a rename at
2636                          * all, just a modification..
2637                          *
2638                          * Otherwise, see if this source was used for
2639                          * multiple renames, in which case we decrement
2640                          * the count, and call it a copy.
2641                          */
2642                         if (!strcmp(p->one->path, p->two->path))
2643                                 p->status = DIFF_STATUS_MODIFIED;
2644                         else if (--p->one->rename_used > 0)
2645                                 p->status = DIFF_STATUS_COPIED;
2646                         else
2647                                 p->status = DIFF_STATUS_RENAMED;
2648                 }
2649                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2650                          p->one->mode != p->two->mode ||
2651                          is_null_sha1(p->one->sha1))
2652                         p->status = DIFF_STATUS_MODIFIED;
2653                 else {
2654                         /* This is a "no-change" entry and should not
2655                          * happen anymore, but prepare for broken callers.
2656                          */
2657                         error("feeding unmodified %s to diffcore",
2658                               p->one->path);
2659                         p->status = DIFF_STATUS_UNKNOWN;
2660                 }
2661         }
2662         diff_debug_queue("resolve-rename-copy done", q);
2663 }
2664
2665 static int check_pair_status(struct diff_filepair *p)
2666 {
2667         switch (p->status) {
2668         case DIFF_STATUS_UNKNOWN:
2669                 return 0;
2670         case 0:
2671                 die("internal error in diff-resolve-rename-copy");
2672         default:
2673                 return 1;
2674         }
2675 }
2676
2677 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2678 {
2679         int fmt = opt->output_format;
2680
2681         if (fmt & DIFF_FORMAT_CHECKDIFF)
2682                 diff_flush_checkdiff(p, opt);
2683         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2684                 diff_flush_raw(p, opt);
2685         else if (fmt & DIFF_FORMAT_NAME)
2686                 write_name_quoted(p->two->path, stdout, opt->line_termination);
2687 }
2688
2689 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2690 {
2691         if (fs->mode)
2692                 printf(" %s mode %06o ", newdelete, fs->mode);
2693         else
2694                 printf(" %s ", newdelete);
2695         write_name_quoted(fs->path, stdout, '\n');
2696 }
2697
2698
2699 static void show_mode_change(struct diff_filepair *p, int show_name)
2700 {
2701         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2702                 printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
2703                         show_name ? ' ' : '\n');
2704                 if (show_name) {
2705                         write_name_quoted(p->two->path, stdout, '\n');
2706                 }
2707         }
2708 }
2709
2710 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2711 {
2712         char *names = pprint_rename(p->one->path, p->two->path);
2713
2714         printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2715         free(names);
2716         show_mode_change(p, 0);
2717 }
2718
2719 static void diff_summary(struct diff_filepair *p)
2720 {
2721         switch(p->status) {
2722         case DIFF_STATUS_DELETED:
2723                 show_file_mode_name("delete", p->one);
2724                 break;
2725         case DIFF_STATUS_ADDED:
2726                 show_file_mode_name("create", p->two);
2727                 break;
2728         case DIFF_STATUS_COPIED:
2729                 show_rename_copy("copy", p);
2730                 break;
2731         case DIFF_STATUS_RENAMED:
2732                 show_rename_copy("rename", p);
2733                 break;
2734         default:
2735                 if (p->score) {
2736                         fputs(" rewrite ", stdout);
2737                         write_name_quoted(p->two->path, stdout, ' ');
2738                         printf("(%d%%)\n", similarity_index(p));
2739                 }
2740                 show_mode_change(p, !p->score);
2741                 break;
2742         }
2743 }
2744
2745 struct patch_id_t {
2746         struct xdiff_emit_state xm;
2747         SHA_CTX *ctx;
2748         int patchlen;
2749 };
2750
2751 static int remove_space(char *line, int len)
2752 {
2753         int i;
2754         char *dst = line;
2755         unsigned char c;
2756
2757         for (i = 0; i < len; i++)
2758                 if (!isspace((c = line[i])))
2759                         *dst++ = c;
2760
2761         return dst - line;
2762 }
2763
2764 static void patch_id_consume(void *priv, char *line, unsigned long len)
2765 {
2766         struct patch_id_t *data = priv;
2767         int new_len;
2768
2769         /* Ignore line numbers when computing the SHA1 of the patch */
2770         if (!prefixcmp(line, "@@ -"))
2771                 return;
2772
2773         new_len = remove_space(line, len);
2774
2775         SHA1_Update(data->ctx, line, new_len);
2776         data->patchlen += new_len;
2777 }
2778
2779 /* returns 0 upon success, and writes result into sha1 */
2780 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2781 {
2782         struct diff_queue_struct *q = &diff_queued_diff;
2783         int i;
2784         SHA_CTX ctx;
2785         struct patch_id_t data;
2786         char buffer[PATH_MAX * 4 + 20];
2787
2788         SHA1_Init(&ctx);
2789         memset(&data, 0, sizeof(struct patch_id_t));
2790         data.ctx = &ctx;
2791         data.xm.consume = patch_id_consume;
2792
2793         for (i = 0; i < q->nr; i++) {
2794                 xpparam_t xpp;
2795                 xdemitconf_t xecfg;
2796                 xdemitcb_t ecb;
2797                 mmfile_t mf1, mf2;
2798                 struct diff_filepair *p = q->queue[i];
2799                 int len1, len2;
2800
2801                 memset(&xecfg, 0, sizeof(xecfg));
2802                 if (p->status == 0)
2803                         return error("internal diff status error");
2804                 if (p->status == DIFF_STATUS_UNKNOWN)
2805                         continue;
2806                 if (diff_unmodified_pair(p))
2807                         continue;
2808                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2809                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2810                         continue;
2811                 if (DIFF_PAIR_UNMERGED(p))
2812                         continue;
2813
2814                 diff_fill_sha1_info(p->one);
2815                 diff_fill_sha1_info(p->two);
2816                 if (fill_mmfile(&mf1, p->one) < 0 ||
2817                                 fill_mmfile(&mf2, p->two) < 0)
2818                         return error("unable to read files to diff");
2819
2820                 len1 = remove_space(p->one->path, strlen(p->one->path));
2821                 len2 = remove_space(p->two->path, strlen(p->two->path));
2822                 if (p->one->mode == 0)
2823                         len1 = snprintf(buffer, sizeof(buffer),
2824                                         "diff--gita/%.*sb/%.*s"
2825                                         "newfilemode%06o"
2826                                         "---/dev/null"
2827                                         "+++b/%.*s",
2828                                         len1, p->one->path,
2829                                         len2, p->two->path,
2830                                         p->two->mode,
2831                                         len2, p->two->path);
2832                 else if (p->two->mode == 0)
2833                         len1 = snprintf(buffer, sizeof(buffer),
2834                                         "diff--gita/%.*sb/%.*s"
2835                                         "deletedfilemode%06o"
2836                                         "---a/%.*s"
2837                                         "+++/dev/null",
2838                                         len1, p->one->path,
2839                                         len2, p->two->path,
2840                                         p->one->mode,
2841                                         len1, p->one->path);
2842                 else
2843                         len1 = snprintf(buffer, sizeof(buffer),
2844                                         "diff--gita/%.*sb/%.*s"
2845                                         "---a/%.*s"
2846                                         "+++b/%.*s",
2847                                         len1, p->one->path,
2848                                         len2, p->two->path,
2849                                         len1, p->one->path,
2850                                         len2, p->two->path);
2851                 SHA1_Update(&ctx, buffer, len1);
2852
2853                 xpp.flags = XDF_NEED_MINIMAL;
2854                 xecfg.ctxlen = 3;
2855                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2856                 ecb.outf = xdiff_outf;
2857                 ecb.priv = &data;
2858                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2859         }
2860
2861         SHA1_Final(sha1, &ctx);
2862         return 0;
2863 }
2864
2865 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2866 {
2867         struct diff_queue_struct *q = &diff_queued_diff;
2868         int i;
2869         int result = diff_get_patch_id(options, sha1);
2870
2871         for (i = 0; i < q->nr; i++)
2872                 diff_free_filepair(q->queue[i]);
2873
2874         free(q->queue);
2875         q->queue = NULL;
2876         q->nr = q->alloc = 0;
2877
2878         return result;
2879 }
2880
2881 static int is_summary_empty(const struct diff_queue_struct *q)
2882 {
2883         int i;
2884
2885         for (i = 0; i < q->nr; i++) {
2886                 const struct diff_filepair *p = q->queue[i];
2887
2888                 switch (p->status) {
2889                 case DIFF_STATUS_DELETED:
2890                 case DIFF_STATUS_ADDED:
2891                 case DIFF_STATUS_COPIED:
2892                 case DIFF_STATUS_RENAMED:
2893                         return 0;
2894                 default:
2895                         if (p->score)
2896                                 return 0;
2897                         if (p->one->mode && p->two->mode &&
2898                             p->one->mode != p->two->mode)
2899                                 return 0;
2900                         break;
2901                 }
2902         }
2903         return 1;
2904 }
2905
2906 void diff_flush(struct diff_options *options)
2907 {
2908         struct diff_queue_struct *q = &diff_queued_diff;
2909         int i, output_format = options->output_format;
2910         int separator = 0;
2911
2912         /*
2913          * Order: raw, stat, summary, patch
2914          * or:    name/name-status/checkdiff (other bits clear)
2915          */
2916         if (!q->nr)
2917                 goto free_queue;
2918
2919         if (output_format & (DIFF_FORMAT_RAW |
2920                              DIFF_FORMAT_NAME |
2921                              DIFF_FORMAT_NAME_STATUS |
2922                              DIFF_FORMAT_CHECKDIFF)) {
2923                 for (i = 0; i < q->nr; i++) {
2924                         struct diff_filepair *p = q->queue[i];
2925                         if (check_pair_status(p))
2926                                 flush_one_pair(p, options);
2927                 }
2928                 separator++;
2929         }
2930
2931         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2932                 struct diffstat_t diffstat;
2933
2934                 memset(&diffstat, 0, sizeof(struct diffstat_t));
2935                 diffstat.xm.consume = diffstat_consume;
2936                 for (i = 0; i < q->nr; i++) {
2937                         struct diff_filepair *p = q->queue[i];
2938                         if (check_pair_status(p))
2939                                 diff_flush_stat(p, options, &diffstat);
2940                 }
2941                 if (output_format & DIFF_FORMAT_NUMSTAT)
2942                         show_numstat(&diffstat, options);
2943                 if (output_format & DIFF_FORMAT_DIFFSTAT)
2944                         show_stats(&diffstat, options);
2945                 if (output_format & DIFF_FORMAT_SHORTSTAT)
2946                         show_shortstats(&diffstat);
2947                 free_diffstat_info(&diffstat);
2948                 separator++;
2949         }
2950
2951         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2952                 for (i = 0; i < q->nr; i++)
2953                         diff_summary(q->queue[i]);
2954                 separator++;
2955         }
2956
2957         if (output_format & DIFF_FORMAT_PATCH) {
2958                 if (separator) {
2959                         if (options->stat_sep) {
2960                                 /* attach patch instead of inline */
2961                                 fputs(options->stat_sep, stdout);
2962                         } else {
2963                                 putchar(options->line_termination);
2964                         }
2965                 }
2966
2967                 for (i = 0; i < q->nr; i++) {
2968                         struct diff_filepair *p = q->queue[i];
2969                         if (check_pair_status(p))
2970                                 diff_flush_patch(p, options);
2971                 }
2972         }
2973
2974         if (output_format & DIFF_FORMAT_CALLBACK)
2975                 options->format_callback(q, options, options->format_callback_data);
2976
2977         for (i = 0; i < q->nr; i++)
2978                 diff_free_filepair(q->queue[i]);
2979 free_queue:
2980         free(q->queue);
2981         q->queue = NULL;
2982         q->nr = q->alloc = 0;
2983 }
2984
2985 static void diffcore_apply_filter(const char *filter)
2986 {
2987         int i;
2988         struct diff_queue_struct *q = &diff_queued_diff;
2989         struct diff_queue_struct outq;
2990         outq.queue = NULL;
2991         outq.nr = outq.alloc = 0;
2992
2993         if (!filter)
2994                 return;
2995
2996         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2997                 int found;
2998                 for (i = found = 0; !found && i < q->nr; i++) {
2999                         struct diff_filepair *p = q->queue[i];
3000                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3001                              ((p->score &&
3002                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3003                               (!p->score &&
3004                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3005                             ((p->status != DIFF_STATUS_MODIFIED) &&
3006                              strchr(filter, p->status)))
3007                                 found++;
3008                 }
3009                 if (found)
3010                         return;
3011
3012                 /* otherwise we will clear the whole queue
3013                  * by copying the empty outq at the end of this
3014                  * function, but first clear the current entries
3015                  * in the queue.
3016                  */
3017                 for (i = 0; i < q->nr; i++)
3018                         diff_free_filepair(q->queue[i]);
3019         }
3020         else {
3021                 /* Only the matching ones */
3022                 for (i = 0; i < q->nr; i++) {
3023                         struct diff_filepair *p = q->queue[i];
3024
3025                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3026                              ((p->score &&
3027                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3028                               (!p->score &&
3029                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3030                             ((p->status != DIFF_STATUS_MODIFIED) &&
3031                              strchr(filter, p->status)))
3032                                 diff_q(&outq, p);
3033                         else
3034                                 diff_free_filepair(p);
3035                 }
3036         }
3037         free(q->queue);
3038         *q = outq;
3039 }
3040
3041 /* Check whether two filespecs with the same mode and size are identical */
3042 static int diff_filespec_is_identical(struct diff_filespec *one,
3043                                       struct diff_filespec *two)
3044 {
3045         if (S_ISGITLINK(one->mode)) {
3046                 diff_fill_sha1_info(one);
3047                 diff_fill_sha1_info(two);
3048                 return !hashcmp(one->sha1, two->sha1);
3049         }
3050         if (diff_populate_filespec(one, 0))
3051                 return 0;
3052         if (diff_populate_filespec(two, 0))
3053                 return 0;
3054         return !memcmp(one->data, two->data, one->size);
3055 }
3056
3057 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3058 {
3059         int i;
3060         struct diff_queue_struct *q = &diff_queued_diff;
3061         struct diff_queue_struct outq;
3062         outq.queue = NULL;
3063         outq.nr = outq.alloc = 0;
3064
3065         for (i = 0; i < q->nr; i++) {
3066                 struct diff_filepair *p = q->queue[i];
3067
3068                 /*
3069                  * 1. Entries that come from stat info dirtyness
3070                  *    always have both sides (iow, not create/delete),
3071                  *    one side of the object name is unknown, with
3072                  *    the same mode and size.  Keep the ones that
3073                  *    do not match these criteria.  They have real
3074                  *    differences.
3075                  *
3076                  * 2. At this point, the file is known to be modified,
3077                  *    with the same mode and size, and the object
3078                  *    name of one side is unknown.  Need to inspect
3079                  *    the identical contents.
3080                  */
3081                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3082                     !DIFF_FILE_VALID(p->two) ||
3083                     (p->one->sha1_valid && p->two->sha1_valid) ||
3084                     (p->one->mode != p->two->mode) ||
3085                     diff_populate_filespec(p->one, 1) ||
3086                     diff_populate_filespec(p->two, 1) ||
3087                     (p->one->size != p->two->size) ||
3088                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3089                         diff_q(&outq, p);
3090                 else {
3091                         /*
3092                          * The caller can subtract 1 from skip_stat_unmatch
3093                          * to determine how many paths were dirty only
3094                          * due to stat info mismatch.
3095                          */
3096                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3097                                 diffopt->skip_stat_unmatch++;
3098                         diff_free_filepair(p);
3099                 }
3100         }
3101         free(q->queue);
3102         *q = outq;
3103 }
3104
3105 void diffcore_std(struct diff_options *options)
3106 {
3107         if (DIFF_OPT_TST(options, QUIET))
3108                 return;
3109
3110         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3111                 diffcore_skip_stat_unmatch(options);
3112         if (options->break_opt != -1)
3113                 diffcore_break(options->break_opt);
3114         if (options->detect_rename)
3115                 diffcore_rename(options);
3116         if (options->break_opt != -1)
3117                 diffcore_merge_broken();
3118         if (options->pickaxe)
3119                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3120         if (options->orderfile)
3121                 diffcore_order(options->orderfile);
3122         diff_resolve_rename_copy();
3123         diffcore_apply_filter(options->filter);
3124
3125         if (diff_queued_diff.nr)
3126                 DIFF_OPT_SET(options, HAS_CHANGES);
3127         else
3128                 DIFF_OPT_CLR(options, HAS_CHANGES);
3129 }
3130
3131 int diff_result_code(struct diff_options *opt, int status)
3132 {
3133         int result = 0;
3134         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3135             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3136                 return status;
3137         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3138             DIFF_OPT_TST(opt, HAS_CHANGES))
3139                 result |= 01;
3140         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3141             DIFF_OPT_TST(opt, CHECK_FAILED))
3142                 result |= 02;
3143         return result;
3144 }
3145
3146 void diff_addremove(struct diff_options *options,
3147                     int addremove, unsigned mode,
3148                     const unsigned char *sha1,
3149                     const char *base, const char *path)
3150 {
3151         char concatpath[PATH_MAX];
3152         struct diff_filespec *one, *two;
3153
3154         /* This may look odd, but it is a preparation for
3155          * feeding "there are unchanged files which should
3156          * not produce diffs, but when you are doing copy
3157          * detection you would need them, so here they are"
3158          * entries to the diff-core.  They will be prefixed
3159          * with something like '=' or '*' (I haven't decided
3160          * which but should not make any difference).
3161          * Feeding the same new and old to diff_change()
3162          * also has the same effect.
3163          * Before the final output happens, they are pruned after
3164          * merged into rename/copy pairs as appropriate.
3165          */
3166         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3167                 addremove = (addremove == '+' ? '-' :
3168                              addremove == '-' ? '+' : addremove);
3169
3170         if (!path) path = "";
3171         sprintf(concatpath, "%s%s", base, path);
3172         one = alloc_filespec(concatpath);
3173         two = alloc_filespec(concatpath);
3174
3175         if (addremove != '+')
3176                 fill_filespec(one, sha1, mode);
3177         if (addremove != '-')
3178                 fill_filespec(two, sha1, mode);
3179
3180         diff_queue(&diff_queued_diff, one, two);
3181         DIFF_OPT_SET(options, HAS_CHANGES);
3182 }
3183
3184 void diff_change(struct diff_options *options,
3185                  unsigned old_mode, unsigned new_mode,
3186                  const unsigned char *old_sha1,
3187                  const unsigned char *new_sha1,
3188                  const char *base, const char *path)
3189 {
3190         char concatpath[PATH_MAX];
3191         struct diff_filespec *one, *two;
3192
3193         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3194                 unsigned tmp;
3195                 const unsigned char *tmp_c;
3196                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3197                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3198         }
3199         if (!path) path = "";
3200         sprintf(concatpath, "%s%s", base, path);
3201         one = alloc_filespec(concatpath);
3202         two = alloc_filespec(concatpath);
3203         fill_filespec(one, old_sha1, old_mode);
3204         fill_filespec(two, new_sha1, new_mode);
3205
3206         diff_queue(&diff_queued_diff, one, two);
3207         DIFF_OPT_SET(options, HAS_CHANGES);
3208 }
3209
3210 void diff_unmerge(struct diff_options *options,
3211                   const char *path,
3212                   unsigned mode, const unsigned char *sha1)
3213 {
3214         struct diff_filespec *one, *two;
3215         one = alloc_filespec(path);
3216         two = alloc_filespec(path);
3217         fill_filespec(one, sha1, mode);
3218         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3219 }