git.oblomov.eu Git - git/blob - diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14
  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         const 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         return git_config_string(&(drv->cmd), var, value);
  90 }
  91
  92 /*
  93  * 'diff.<what>.funcname' attribute can be specified in the configuration
  94  * to define a customized regexp to find the beginning of a function to
  95  * be used for hunk header lines of "diff -p" style output.
  96  */
  97 static struct funcname_pattern {
  98         char *name;
  99         char *pattern;
 100         struct funcname_pattern *next;
 101 } *funcname_pattern_list;
 102
 103 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 104 {
 105         const char *name;
 106         int namelen;
 107         struct funcname_pattern *pp;
 108
 109         name = var + 5; /* "diff." */
 110         namelen = ep - name;
 111
 112         for (pp = funcname_pattern_list; pp; pp = pp->next)
 113                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 114                         break;
 115         if (!pp) {
 116                 pp = xcalloc(1, sizeof(*pp));
 117                 pp->name = xmemdupz(name, namelen);
 118                 pp->next = funcname_pattern_list;
 119                 funcname_pattern_list = pp;
 120         }
 121         if (pp->pattern)
 122                 free(pp->pattern);
 123         pp->pattern = xstrdup(value);
 124         return 0;
 125 }
 126
 127 /*
 128  * These are to give UI layer defaults.
 129  * The core-level commands such as git-diff-files should
 130  * never be affected by the setting of diff.renames
 131  * the user happens to have in the configuration file.
 132  */
 133 int git_diff_ui_config(const char *var, const char *value)
 134 {
 135         if (!strcmp(var, "diff.renamelimit")) {
 136                 diff_rename_limit_default = git_config_int(var, value);
 137                 return 0;
 138         }
 139         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 140                 diff_use_color_default = git_config_colorbool(var, value, -1);
 141                 return 0;
 142         }
 143         if (!strcmp(var, "diff.renames")) {
 144                 if (!value)
 145                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 146                 else if (!strcasecmp(value, "copies") ||
 147                          !strcasecmp(value, "copy"))
 148                         diff_detect_rename_default = DIFF_DETECT_COPY;
 149                 else if (git_config_bool(var,value))
 150                         diff_detect_rename_default = DIFF_DETECT_RENAME;
 151                 return 0;
 152         }
 153         if (!strcmp(var, "diff.autorefreshindex")) {
 154                 diff_auto_refresh_index = git_config_bool(var, value);
 155                 return 0;
 156         }
 157         if (!strcmp(var, "diff.external")) {
 158                 if (!value)
 159                         return config_error_nonbool(var);
 160                 external_diff_cmd_cfg = xstrdup(value);
 161                 return 0;
 162         }
 163         if (!prefixcmp(var, "diff.")) {
 164                 const char *ep = strrchr(var, '.');
 165
 166                 if (ep != var + 4 && !strcmp(ep, ".command"))
 167                         return parse_lldiff_command(var, ep, value);
 168         }
 169
 170         return git_diff_basic_config(var, value);
 171 }
 172
 173 int git_diff_basic_config(const char *var, const char *value)
 174 {
 175         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 176                 int slot = parse_diff_color_slot(var, 11);
 177                 if (!value)
 178                         return config_error_nonbool(var);
 179                 color_parse(value, var, diff_colors[slot]);
 180                 return 0;
 181         }
 182
 183         if (!prefixcmp(var, "diff.")) {
 184                 const char *ep = strrchr(var, '.');
 185                 if (ep != var + 4) {
 186                         if (!strcmp(ep, ".funcname")) {
 187                                 if (!value)
 188                                         return config_error_nonbool(var);
 189                                 return parse_funcname_pattern(var, ep, value);
 190                         }
 191                 }
 192         }
 193
 194         return git_default_config(var, value);
 195 }
 196
 197 static char *quote_two(const char *one, const char *two)
 198 {
 199         int need_one = quote_c_style(one, NULL, NULL, 1);
 200         int need_two = quote_c_style(two, NULL, NULL, 1);
 201         struct strbuf res;
 202
 203         strbuf_init(&res, 0);
 204         if (need_one + need_two) {
 205                 strbuf_addch(&res, '"');
 206                 quote_c_style(one, &res, NULL, 1);
 207                 quote_c_style(two, &res, NULL, 1);
 208                 strbuf_addch(&res, '"');
 209         } else {
 210                 strbuf_addstr(&res, one);
 211                 strbuf_addstr(&res, two);
 212         }
 213         return strbuf_detach(&res, NULL);
 214 }
 215
 216 static const char *external_diff(void)
 217 {
 218         static const char *external_diff_cmd = NULL;
 219         static int done_preparing = 0;
 220
 221         if (done_preparing)
 222                 return external_diff_cmd;
 223         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 224         if (!external_diff_cmd)
 225                 external_diff_cmd = external_diff_cmd_cfg;
 226         done_preparing = 1;
 227         return external_diff_cmd;
 228 }
 229
 230 static struct diff_tempfile {
 231         const char *name; /* filename external diff should read from */
 232         char hex[41];
 233         char mode[10];
 234         char tmp_path[PATH_MAX];
 235 } diff_temp[2];
 236
 237 static int count_lines(const char *data, int size)
 238 {
 239         int count, ch, completely_empty = 1, nl_just_seen = 0;
 240         count = 0;
 241         while (0 < size--) {
 242                 ch = *data++;
 243                 if (ch == '\n') {
 244                         count++;
 245                         nl_just_seen = 1;
 246                         completely_empty = 0;
 247                 }
 248                 else {
 249                         nl_just_seen = 0;
 250                         completely_empty = 0;
 251                 }
 252         }
 253         if (completely_empty)
 254                 return 0;
 255         if (!nl_just_seen)
 256                 count++; /* no trailing newline */
 257         return count;
 258 }
 259
 260 static void print_line_count(int count)
 261 {
 262         switch (count) {
 263         case 0:
 264                 printf("0,0");
 265                 break;
 266         case 1:
 267                 printf("1");
 268                 break;
 269         default:
 270                 printf("1,%d", count);
 271                 break;
 272         }
 273 }
 274
 275 static void copy_file_with_prefix(int prefix, const char *data, int size,
 276                                   const char *set, const char *reset)
 277 {
 278         int ch, nl_just_seen = 1;
 279         while (0 < size--) {
 280                 ch = *data++;
 281                 if (nl_just_seen) {
 282                         fputs(set, stdout);
 283                         putchar(prefix);
 284                 }
 285                 if (ch == '\n') {
 286                         nl_just_seen = 1;
 287                         fputs(reset, stdout);
 288                 } else
 289                         nl_just_seen = 0;
 290                 putchar(ch);
 291         }
 292         if (!nl_just_seen)
 293                 printf("%s\n\\ No newline at end of file\n", reset);
 294 }
 295
 296 static void emit_rewrite_diff(const char *name_a,
 297                               const char *name_b,
 298                               struct diff_filespec *one,
 299                               struct diff_filespec *two,
 300                               struct diff_options *o)
 301 {
 302         int lc_a, lc_b;
 303         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 304         const char *name_a_tab, *name_b_tab;
 305         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 306         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 307         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 308         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 309         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 310         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 311
 312         name_a += (*name_a == '/');
 313         name_b += (*name_b == '/');
 314         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 315         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 316
 317         strbuf_reset(&a_name);
 318         strbuf_reset(&b_name);
 319         quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
 320         quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
 321
 322         diff_populate_filespec(one, 0);
 323         diff_populate_filespec(two, 0);
 324         lc_a = count_lines(one->data, one->size);
 325         lc_b = count_lines(two->data, two->size);
 326         printf("%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 327                metainfo, a_name.buf, name_a_tab, reset,
 328                metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 329         print_line_count(lc_a);
 330         printf(" +");
 331         print_line_count(lc_b);
 332         printf(" @@%s\n", reset);
 333         if (lc_a)
 334                 copy_file_with_prefix('-', one->data, one->size, old, reset);
 335         if (lc_b)
 336                 copy_file_with_prefix('+', two->data, two->size, new, reset);
 337 }
 338
 339 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 340 {
 341         if (!DIFF_FILE_VALID(one)) {
 342                 mf->ptr = (char *)""; /* does not matter */
 343                 mf->size = 0;
 344                 return 0;
 345         }
 346         else if (diff_populate_filespec(one, 0))
 347                 return -1;
 348         mf->ptr = one->data;
 349         mf->size = one->size;
 350         return 0;
 351 }
 352
 353 struct diff_words_buffer {
 354         mmfile_t text;
 355         long alloc;
 356         long current; /* output pointer */
 357         int suppressed_newline;
 358 };
 359
 360 static void diff_words_append(char *line, unsigned long len,
 361                 struct diff_words_buffer *buffer)
 362 {
 363         if (buffer->text.size + len > buffer->alloc) {
 364                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 365                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 366         }
 367         line++;
 368         len--;
 369         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 370         buffer->text.size += len;
 371 }
 372
 373 struct diff_words_data {
 374         struct xdiff_emit_state xm;
 375         struct diff_words_buffer minus, plus;
 376 };
 377
 378 static void print_word(struct diff_words_buffer *buffer, int len, int color,
 379                 int suppress_newline)
 380 {
 381         const char *ptr;
 382         int eol = 0;
 383
 384         if (len == 0)
 385                 return;
 386
 387         ptr  = buffer->text.ptr + buffer->current;
 388         buffer->current += len;
 389
 390         if (ptr[len - 1] == '\n') {
 391                 eol = 1;
 392                 len--;
 393         }
 394
 395         fputs(diff_get_color(1, color), stdout);
 396         fwrite(ptr, len, 1, stdout);
 397         fputs(diff_get_color(1, DIFF_RESET), stdout);
 398
 399         if (eol) {
 400                 if (suppress_newline)
 401                         buffer->suppressed_newline = 1;
 402                 else
 403                         putchar('\n');
 404         }
 405 }
 406
 407 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 408 {
 409         struct diff_words_data *diff_words = priv;
 410
 411         if (diff_words->minus.suppressed_newline) {
 412                 if (line[0] != '+')
 413                         putchar('\n');
 414                 diff_words->minus.suppressed_newline = 0;
 415         }
 416
 417         len--;
 418         switch (line[0]) {
 419                 case '-':
 420                         print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
 421                         break;
 422                 case '+':
 423                         print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
 424                         break;
 425                 case ' ':
 426                         print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
 427                         diff_words->minus.current += len;
 428                         break;
 429         }
 430 }
 431
 432 /* this executes the word diff on the accumulated buffers */
 433 static void diff_words_show(struct diff_words_data *diff_words)
 434 {
 435         xpparam_t xpp;
 436         xdemitconf_t xecfg;
 437         xdemitcb_t ecb;
 438         mmfile_t minus, plus;
 439         int i;
 440
 441         memset(&xecfg, 0, sizeof(xecfg));
 442         minus.size = diff_words->minus.text.size;
 443         minus.ptr = xmalloc(minus.size);
 444         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 445         for (i = 0; i < minus.size; i++)
 446                 if (isspace(minus.ptr[i]))
 447                         minus.ptr[i] = '\n';
 448         diff_words->minus.current = 0;
 449
 450         plus.size = diff_words->plus.text.size;
 451         plus.ptr = xmalloc(plus.size);
 452         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 453         for (i = 0; i < plus.size; i++)
 454                 if (isspace(plus.ptr[i]))
 455                         plus.ptr[i] = '\n';
 456         diff_words->plus.current = 0;
 457
 458         xpp.flags = XDF_NEED_MINIMAL;
 459         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 460         ecb.outf = xdiff_outf;
 461         ecb.priv = diff_words;
 462         diff_words->xm.consume = fn_out_diff_words_aux;
 463         xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 464
 465         free(minus.ptr);
 466         free(plus.ptr);
 467         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 468
 469         if (diff_words->minus.suppressed_newline) {
 470                 putchar('\n');
 471                 diff_words->minus.suppressed_newline = 0;
 472         }
 473 }
 474
 475 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 476
 477 struct emit_callback {
 478         struct xdiff_emit_state xm;
 479         int nparents, color_diff;
 480         unsigned ws_rule;
 481         sane_truncate_fn truncate;
 482         const char **label_path;
 483         struct diff_words_data *diff_words;
 484         int *found_changesp;
 485 };
 486
 487 static void free_diff_words_data(struct emit_callback *ecbdata)
 488 {
 489         if (ecbdata->diff_words) {
 490                 /* flush buffers */
 491                 if (ecbdata->diff_words->minus.text.size ||
 492                                 ecbdata->diff_words->plus.text.size)
 493                         diff_words_show(ecbdata->diff_words);
 494
 495                 if (ecbdata->diff_words->minus.text.ptr)
 496                         free (ecbdata->diff_words->minus.text.ptr);
 497                 if (ecbdata->diff_words->plus.text.ptr)
 498                         free (ecbdata->diff_words->plus.text.ptr);
 499                 free(ecbdata->diff_words);
 500                 ecbdata->diff_words = NULL;
 501         }
 502 }
 503
 504 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 505 {
 506         if (diff_use_color)
 507                 return diff_colors[ix];
 508         return "";
 509 }
 510
 511 static void emit_line(const char *set, const char *reset, const char *line, int len)
 512 {
 513         fputs(set, stdout);
 514         fwrite(line, len, 1, stdout);
 515         fputs(reset, stdout);
 516 }
 517
 518 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 519 {
 520         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 521         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 522
 523         if (!*ws)
 524                 emit_line(set, reset, line, len);
 525         else {
 526                 /* Emit just the prefix, then the rest. */
 527                 emit_line(set, reset, line, ecbdata->nparents);
 528                 (void)check_and_emit_line(line + ecbdata->nparents,
 529                     len - ecbdata->nparents, ecbdata->ws_rule,
 530                     stdout, set, reset, ws);
 531         }
 532 }
 533
 534 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 535 {
 536         const char *cp;
 537         unsigned long allot;
 538         size_t l = len;
 539
 540         if (ecb->truncate)
 541                 return ecb->truncate(line, len);
 542         cp = line;
 543         allot = l;
 544         while (0 < l) {
 545                 (void) utf8_width(&cp, &l);
 546                 if (!cp)
 547                         break; /* truncated in the middle? */
 548         }
 549         return allot - l;
 550 }
 551
 552 static void fn_out_consume(void *priv, char *line, unsigned long len)
 553 {
 554         int i;
 555         int color;
 556         struct emit_callback *ecbdata = priv;
 557         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 558         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 559         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 560
 561         *(ecbdata->found_changesp) = 1;
 562
 563         if (ecbdata->label_path[0]) {
 564                 const char *name_a_tab, *name_b_tab;
 565
 566                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 567                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 568
 569                 printf("%s--- %s%s%s\n",
 570                        meta, ecbdata->label_path[0], reset, name_a_tab);
 571                 printf("%s+++ %s%s%s\n",
 572                        meta, ecbdata->label_path[1], reset, name_b_tab);
 573                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 574         }
 575
 576         /* This is not really necessary for now because
 577          * this codepath only deals with two-way diffs.
 578          */
 579         for (i = 0; i < len && line[i] == '@'; i++)
 580                 ;
 581         if (2 <= i && i < len && line[i] == ' ') {
 582                 ecbdata->nparents = i - 1;
 583                 len = sane_truncate_line(ecbdata, line, len);
 584                 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 585                           reset, line, len);
 586                 if (line[len-1] != '\n')
 587                         putchar('\n');
 588                 return;
 589         }
 590
 591         if (len < ecbdata->nparents) {
 592                 emit_line(reset, reset, line, len);
 593                 return;
 594         }
 595
 596         color = DIFF_PLAIN;
 597         if (ecbdata->diff_words && ecbdata->nparents != 1)
 598                 /* fall back to normal diff */
 599                 free_diff_words_data(ecbdata);
 600         if (ecbdata->diff_words) {
 601                 if (line[0] == '-') {
 602                         diff_words_append(line, len,
 603                                           &ecbdata->diff_words->minus);
 604                         return;
 605                 } else if (line[0] == '+') {
 606                         diff_words_append(line, len,
 607                                           &ecbdata->diff_words->plus);
 608                         return;
 609                 }
 610                 if (ecbdata->diff_words->minus.text.size ||
 611                     ecbdata->diff_words->plus.text.size)
 612                         diff_words_show(ecbdata->diff_words);
 613                 line++;
 614                 len--;
 615                 emit_line(plain, reset, line, len);
 616                 return;
 617         }
 618         for (i = 0; i < ecbdata->nparents && len; i++) {
 619                 if (line[i] == '-')
 620                         color = DIFF_FILE_OLD;
 621                 else if (line[i] == '+')
 622                         color = DIFF_FILE_NEW;
 623         }
 624
 625         if (color != DIFF_FILE_NEW) {
 626                 emit_line(diff_get_color(ecbdata->color_diff, color),
 627                           reset, line, len);
 628                 return;
 629         }
 630         emit_add_line(reset, ecbdata, line, len);
 631 }
 632
 633 static char *pprint_rename(const char *a, const char *b)
 634 {
 635         const char *old = a;
 636         const char *new = b;
 637         struct strbuf name;
 638         int pfx_length, sfx_length;
 639         int len_a = strlen(a);
 640         int len_b = strlen(b);
 641         int a_midlen, b_midlen;
 642         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 643         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 644
 645         strbuf_init(&name, 0);
 646         if (qlen_a || qlen_b) {
 647                 quote_c_style(a, &name, NULL, 0);
 648                 strbuf_addstr(&name, " => ");
 649                 quote_c_style(b, &name, NULL, 0);
 650                 return strbuf_detach(&name, NULL);
 651         }
 652
 653         /* Find common prefix */
 654         pfx_length = 0;
 655         while (*old && *new && *old == *new) {
 656                 if (*old == '/')
 657                         pfx_length = old - a + 1;
 658                 old++;
 659                 new++;
 660         }
 661
 662         /* Find common suffix */
 663         old = a + len_a;
 664         new = b + len_b;
 665         sfx_length = 0;
 666         while (a <= old && b <= new && *old == *new) {
 667                 if (*old == '/')
 668                         sfx_length = len_a - (old - a);
 669                 old--;
 670                 new--;
 671         }
 672
 673         /*
 674          * pfx{mid-a => mid-b}sfx
 675          * {pfx-a => pfx-b}sfx
 676          * pfx{sfx-a => sfx-b}
 677          * name-a => name-b
 678          */
 679         a_midlen = len_a - pfx_length - sfx_length;
 680         b_midlen = len_b - pfx_length - sfx_length;
 681         if (a_midlen < 0)
 682                 a_midlen = 0;
 683         if (b_midlen < 0)
 684                 b_midlen = 0;
 685
 686         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 687         if (pfx_length + sfx_length) {
 688                 strbuf_add(&name, a, pfx_length);
 689                 strbuf_addch(&name, '{');
 690         }
 691         strbuf_add(&name, a + pfx_length, a_midlen);
 692         strbuf_addstr(&name, " => ");
 693         strbuf_add(&name, b + pfx_length, b_midlen);
 694         if (pfx_length + sfx_length) {
 695                 strbuf_addch(&name, '}');
 696                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 697         }
 698         return strbuf_detach(&name, NULL);
 699 }
 700
 701 struct diffstat_t {
 702         struct xdiff_emit_state xm;
 703
 704         int nr;
 705         int alloc;
 706         struct diffstat_file {
 707                 char *from_name;
 708                 char *name;
 709                 char *print_name;
 710                 unsigned is_unmerged:1;
 711                 unsigned is_binary:1;
 712                 unsigned is_renamed:1;
 713                 unsigned int added, deleted;
 714         } **files;
 715 };
 716
 717 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 718                                           const char *name_a,
 719                                           const char *name_b)
 720 {
 721         struct diffstat_file *x;
 722         x = xcalloc(sizeof (*x), 1);
 723         if (diffstat->nr == diffstat->alloc) {
 724                 diffstat->alloc = alloc_nr(diffstat->alloc);
 725                 diffstat->files = xrealloc(diffstat->files,
 726                                 diffstat->alloc * sizeof(x));
 727         }
 728         diffstat->files[diffstat->nr++] = x;
 729         if (name_b) {
 730                 x->from_name = xstrdup(name_a);
 731                 x->name = xstrdup(name_b);
 732                 x->is_renamed = 1;
 733         }
 734         else {
 735                 x->from_name = NULL;
 736                 x->name = xstrdup(name_a);
 737         }
 738         return x;
 739 }
 740
 741 static void diffstat_consume(void *priv, char *line, unsigned long len)
 742 {
 743         struct diffstat_t *diffstat = priv;
 744         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 745
 746         if (line[0] == '+')
 747                 x->added++;
 748         else if (line[0] == '-')
 749                 x->deleted++;
 750 }
 751
 752 const char mime_boundary_leader[] = "------------";
 753
 754 static int scale_linear(int it, int width, int max_change)
 755 {
 756         /*
 757          * make sure that at least one '-' is printed if there were deletions,
 758          * and likewise for '+'.
 759          */
 760         if (max_change < 2)
 761                 return it;
 762         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 763 }
 764
 765 static void show_name(const char *prefix, const char *name, int len,
 766                       const char *reset, const char *set)
 767 {
 768         printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
 769 }
 770
 771 static void show_graph(char ch, int cnt, const char *set, const char *reset)
 772 {
 773         if (cnt <= 0)
 774                 return;
 775         printf("%s", set);
 776         while (cnt--)
 777                 putchar(ch);
 778         printf("%s", reset);
 779 }
 780
 781 static void fill_print_name(struct diffstat_file *file)
 782 {
 783         char *pname;
 784
 785         if (file->print_name)
 786                 return;
 787
 788         if (!file->is_renamed) {
 789                 struct strbuf buf;
 790                 strbuf_init(&buf, 0);
 791                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 792                         pname = strbuf_detach(&buf, NULL);
 793                 } else {
 794                         pname = file->name;
 795                         strbuf_release(&buf);
 796                 }
 797         } else {
 798                 pname = pprint_rename(file->from_name, file->name);
 799         }
 800         file->print_name = pname;
 801 }
 802
 803 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 804 {
 805         int i, len, add, del, total, adds = 0, dels = 0;
 806         int max_change = 0, max_len = 0;
 807         int total_files = data->nr;
 808         int width, name_width;
 809         const char *reset, *set, *add_c, *del_c;
 810
 811         if (data->nr == 0)
 812                 return;
 813
 814         width = options->stat_width ? options->stat_width : 80;
 815         name_width = options->stat_name_width ? options->stat_name_width : 50;
 816
 817         /* Sanity: give at least 5 columns to the graph,
 818          * but leave at least 10 columns for the name.
 819          */
 820         if (width < name_width + 15) {
 821                 if (name_width <= 25)
 822                         width = name_width + 15;
 823                 else
 824                         name_width = width - 15;
 825         }
 826
 827         /* Find the longest filename and max number of changes */
 828         reset = diff_get_color_opt(options, DIFF_RESET);
 829         set   = diff_get_color_opt(options, DIFF_PLAIN);
 830         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 831         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 832
 833         for (i = 0; i < data->nr; i++) {
 834                 struct diffstat_file *file = data->files[i];
 835                 int change = file->added + file->deleted;
 836                 fill_print_name(file);
 837                 len = strlen(file->print_name);
 838                 if (max_len < len)
 839                         max_len = len;
 840
 841                 if (file->is_binary || file->is_unmerged)
 842                         continue;
 843                 if (max_change < change)
 844                         max_change = change;
 845         }
 846
 847         /* Compute the width of the graph part;
 848          * 10 is for one blank at the beginning of the line plus
 849          * " | count " between the name and the graph.
 850          *
 851          * From here on, name_width is the width of the name area,
 852          * and width is the width of the graph area.
 853          */
 854         name_width = (name_width < max_len) ? name_width : max_len;
 855         if (width < (name_width + 10) + max_change)
 856                 width = width - (name_width + 10);
 857         else
 858                 width = max_change;
 859
 860         for (i = 0; i < data->nr; i++) {
 861                 const char *prefix = "";
 862                 char *name = data->files[i]->print_name;
 863                 int added = data->files[i]->added;
 864                 int deleted = data->files[i]->deleted;
 865                 int name_len;
 866
 867                 /*
 868                  * "scale" the filename
 869                  */
 870                 len = name_width;
 871                 name_len = strlen(name);
 872                 if (name_width < name_len) {
 873                         char *slash;
 874                         prefix = "...";
 875                         len -= 3;
 876                         name += name_len - len;
 877                         slash = strchr(name, '/');
 878                         if (slash)
 879                                 name = slash;
 880                 }
 881
 882                 if (data->files[i]->is_binary) {
 883                         show_name(prefix, name, len, reset, set);
 884                         printf("  Bin ");
 885                         printf("%s%d%s", del_c, deleted, reset);
 886                         printf(" -> ");
 887                         printf("%s%d%s", add_c, added, reset);
 888                         printf(" bytes");
 889                         printf("\n");
 890                         continue;
 891                 }
 892                 else if (data->files[i]->is_unmerged) {
 893                         show_name(prefix, name, len, reset, set);
 894                         printf("  Unmerged\n");
 895                         continue;
 896                 }
 897                 else if (!data->files[i]->is_renamed &&
 898                          (added + deleted == 0)) {
 899                         total_files--;
 900                         continue;
 901                 }
 902
 903                 /*
 904                  * scale the add/delete
 905                  */
 906                 add = added;
 907                 del = deleted;
 908                 total = add + del;
 909                 adds += add;
 910                 dels += del;
 911
 912                 if (width <= max_change) {
 913                         add = scale_linear(add, width, max_change);
 914                         del = scale_linear(del, width, max_change);
 915                         total = add + del;
 916                 }
 917                 show_name(prefix, name, len, reset, set);
 918                 printf("%5d ", added + deleted);
 919                 show_graph('+', add, add_c, reset);
 920                 show_graph('-', del, del_c, reset);
 921                 putchar('\n');
 922         }
 923         printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 924                set, total_files, adds, dels, reset);
 925 }
 926
 927 static void show_shortstats(struct diffstat_t* data)
 928 {
 929         int i, adds = 0, dels = 0, total_files = data->nr;
 930
 931         if (data->nr == 0)
 932                 return;
 933
 934         for (i = 0; i < data->nr; i++) {
 935                 if (!data->files[i]->is_binary &&
 936                     !data->files[i]->is_unmerged) {
 937                         int added = data->files[i]->added;
 938                         int deleted= data->files[i]->deleted;
 939                         if (!data->files[i]->is_renamed &&
 940                             (added + deleted == 0)) {
 941                                 total_files--;
 942                         } else {
 943                                 adds += added;
 944                                 dels += deleted;
 945                         }
 946                 }
 947         }
 948         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 949                total_files, adds, dels);
 950 }
 951
 952 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 953 {
 954         int i;
 955
 956         if (data->nr == 0)
 957                 return;
 958
 959         for (i = 0; i < data->nr; i++) {
 960                 struct diffstat_file *file = data->files[i];
 961
 962                 if (file->is_binary)
 963                         printf("-\t-\t");
 964                 else
 965                         printf("%d\t%d\t", file->added, file->deleted);
 966                 if (options->line_termination) {
 967                         fill_print_name(file);
 968                         if (!file->is_renamed)
 969                                 write_name_quoted(file->name, stdout,
 970                                                   options->line_termination);
 971                         else {
 972                                 fputs(file->print_name, stdout);
 973                                 putchar(options->line_termination);
 974                         }
 975                 } else {
 976                         if (file->is_renamed) {
 977                                 putchar('\0');
 978                                 write_name_quoted(file->from_name, stdout, '\0');
 979                         }
 980                         write_name_quoted(file->name, stdout, '\0');
 981                 }
 982         }
 983 }
 984
 985 static void free_diffstat_info(struct diffstat_t *diffstat)
 986 {
 987         int i;
 988         for (i = 0; i < diffstat->nr; i++) {
 989                 struct diffstat_file *f = diffstat->files[i];
 990                 if (f->name != f->print_name)
 991                         free(f->print_name);
 992                 free(f->name);
 993                 free(f->from_name);
 994                 free(f);
 995         }
 996         free(diffstat->files);
 997 }
 998
 999 struct checkdiff_t {
1000         struct xdiff_emit_state xm;
1001         const char *filename;
1002         int lineno, color_diff;
1003         unsigned ws_rule;
1004         unsigned status;
1005 };
1006
1007 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1008 {
1009         struct checkdiff_t *data = priv;
1010         const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1011         const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1012         const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1013         char *err;
1014
1015         if (line[0] == '+') {
1016                 data->lineno++;
1017                 data->status = check_and_emit_line(line + 1, len - 1,
1018                     data->ws_rule, NULL, NULL, NULL, NULL);
1019                 if (!data->status)
1020                         return;
1021                 err = whitespace_error_string(data->status);
1022                 printf("%s:%d: %s.\n", data->filename, data->lineno, err);
1023                 free(err);
1024                 emit_line(set, reset, line, 1);
1025                 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1026                     stdout, set, reset, ws);
1027         } else if (line[0] == ' ')
1028                 data->lineno++;
1029         else if (line[0] == '@') {
1030                 char *plus = strchr(line, '+');
1031                 if (plus)
1032                         data->lineno = strtol(plus, NULL, 10) - 1;
1033                 else
1034                         die("invalid diff");
1035         }
1036 }
1037
1038 static unsigned char *deflate_it(char *data,
1039                                  unsigned long size,
1040                                  unsigned long *result_size)
1041 {
1042         int bound;
1043         unsigned char *deflated;
1044         z_stream stream;
1045
1046         memset(&stream, 0, sizeof(stream));
1047         deflateInit(&stream, zlib_compression_level);
1048         bound = deflateBound(&stream, size);
1049         deflated = xmalloc(bound);
1050         stream.next_out = deflated;
1051         stream.avail_out = bound;
1052
1053         stream.next_in = (unsigned char *)data;
1054         stream.avail_in = size;
1055         while (deflate(&stream, Z_FINISH) == Z_OK)
1056                 ; /* nothing */
1057         deflateEnd(&stream);
1058         *result_size = stream.total_out;
1059         return deflated;
1060 }
1061
1062 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1063 {
1064         void *cp;
1065         void *delta;
1066         void *deflated;
1067         void *data;
1068         unsigned long orig_size;
1069         unsigned long delta_size;
1070         unsigned long deflate_size;
1071         unsigned long data_size;
1072
1073         /* We could do deflated delta, or we could do just deflated two,
1074          * whichever is smaller.
1075          */
1076         delta = NULL;
1077         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1078         if (one->size && two->size) {
1079                 delta = diff_delta(one->ptr, one->size,
1080                                    two->ptr, two->size,
1081                                    &delta_size, deflate_size);
1082                 if (delta) {
1083                         void *to_free = delta;
1084                         orig_size = delta_size;
1085                         delta = deflate_it(delta, delta_size, &delta_size);
1086                         free(to_free);
1087                 }
1088         }
1089
1090         if (delta && delta_size < deflate_size) {
1091                 printf("delta %lu\n", orig_size);
1092                 free(deflated);
1093                 data = delta;
1094                 data_size = delta_size;
1095         }
1096         else {
1097                 printf("literal %lu\n", two->size);
1098                 free(delta);
1099                 data = deflated;
1100                 data_size = deflate_size;
1101         }
1102
1103         /* emit data encoded in base85 */
1104         cp = data;
1105         while (data_size) {
1106                 int bytes = (52 < data_size) ? 52 : data_size;
1107                 char line[70];
1108                 data_size -= bytes;
1109                 if (bytes <= 26)
1110                         line[0] = bytes + 'A' - 1;
1111                 else
1112                         line[0] = bytes - 26 + 'a' - 1;
1113                 encode_85(line + 1, cp, bytes);
1114                 cp = (char *) cp + bytes;
1115                 puts(line);
1116         }
1117         printf("\n");
1118         free(data);
1119 }
1120
1121 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1122 {
1123         printf("GIT binary patch\n");
1124         emit_binary_diff_body(one, two);
1125         emit_binary_diff_body(two, one);
1126 }
1127
1128 static void setup_diff_attr_check(struct git_attr_check *check)
1129 {
1130         static struct git_attr *attr_diff;
1131
1132         if (!attr_diff) {
1133                 attr_diff = git_attr("diff", 4);
1134         }
1135         check[0].attr = attr_diff;
1136 }
1137
1138 static void diff_filespec_check_attr(struct diff_filespec *one)
1139 {
1140         struct git_attr_check attr_diff_check;
1141         int check_from_data = 0;
1142
1143         if (one->checked_attr)
1144                 return;
1145
1146         setup_diff_attr_check(&attr_diff_check);
1147         one->is_binary = 0;
1148         one->funcname_pattern_ident = NULL;
1149
1150         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1151                 const char *value;
1152
1153                 /* binaryness */
1154                 value = attr_diff_check.value;
1155                 if (ATTR_TRUE(value))
1156                         ;
1157                 else if (ATTR_FALSE(value))
1158                         one->is_binary = 1;
1159                 else
1160                         check_from_data = 1;
1161
1162                 /* funcname pattern ident */
1163                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1164                         ;
1165                 else
1166                         one->funcname_pattern_ident = value;
1167         }
1168
1169         if (check_from_data) {
1170                 if (!one->data && DIFF_FILE_VALID(one))
1171                         diff_populate_filespec(one, 0);
1172
1173                 if (one->data)
1174                         one->is_binary = buffer_is_binary(one->data, one->size);
1175         }
1176 }
1177
1178 int diff_filespec_is_binary(struct diff_filespec *one)
1179 {
1180         diff_filespec_check_attr(one);
1181         return one->is_binary;
1182 }
1183
1184 static const char *funcname_pattern(const char *ident)
1185 {
1186         struct funcname_pattern *pp;
1187
1188         for (pp = funcname_pattern_list; pp; pp = pp->next)
1189                 if (!strcmp(ident, pp->name))
1190                         return pp->pattern;
1191         return NULL;
1192 }
1193
1194 static struct builtin_funcname_pattern {
1195         const char *name;
1196         const char *pattern;
1197 } builtin_funcname_pattern[] = {
1198         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1199                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1200                         "^[     ]*\\(\\([       ]*"
1201                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1202                         "[      ]*([^;]*$\\)" },
1203         { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1204 };
1205
1206 static const char *diff_funcname_pattern(struct diff_filespec *one)
1207 {
1208         const char *ident, *pattern;
1209         int i;
1210
1211         diff_filespec_check_attr(one);
1212         ident = one->funcname_pattern_ident;
1213
1214         if (!ident)
1215                 /*
1216                  * If the config file has "funcname.default" defined, that
1217                  * regexp is used; otherwise NULL is returned and xemit uses
1218                  * the built-in default.
1219                  */
1220                 return funcname_pattern("default");
1221
1222         /* Look up custom "funcname.$ident" regexp from config. */
1223         pattern = funcname_pattern(ident);
1224         if (pattern)
1225                 return pattern;
1226
1227         /*
1228          * And define built-in fallback patterns here.  Note that
1229          * these can be overridden by the user's config settings.
1230          */
1231         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1232                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1233                         return builtin_funcname_pattern[i].pattern;
1234
1235         return NULL;
1236 }
1237
1238 static void builtin_diff(const char *name_a,
1239                          const char *name_b,
1240                          struct diff_filespec *one,
1241                          struct diff_filespec *two,
1242                          const char *xfrm_msg,
1243                          struct diff_options *o,
1244                          int complete_rewrite)
1245 {
1246         mmfile_t mf1, mf2;
1247         const char *lbl[2];
1248         char *a_one, *b_two;
1249         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1250         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1251
1252         a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1253         b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1254         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1255         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1256         printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1257         if (lbl[0][0] == '/') {
1258                 /* /dev/null */
1259                 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1260                 if (xfrm_msg && xfrm_msg[0])
1261                         printf("%s%s%s\n", set, xfrm_msg, reset);
1262         }
1263         else if (lbl[1][0] == '/') {
1264                 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1265                 if (xfrm_msg && xfrm_msg[0])
1266                         printf("%s%s%s\n", set, xfrm_msg, reset);
1267         }
1268         else {
1269                 if (one->mode != two->mode) {
1270                         printf("%sold mode %06o%s\n", set, one->mode, reset);
1271                         printf("%snew mode %06o%s\n", set, two->mode, reset);
1272                 }
1273                 if (xfrm_msg && xfrm_msg[0])
1274                         printf("%s%s%s\n", set, xfrm_msg, reset);
1275                 /*
1276                  * we do not run diff between different kind
1277                  * of objects.
1278                  */
1279                 if ((one->mode ^ two->mode) & S_IFMT)
1280                         goto free_ab_and_return;
1281                 if (complete_rewrite) {
1282                         emit_rewrite_diff(name_a, name_b, one, two, o);
1283                         o->found_changes = 1;
1284                         goto free_ab_and_return;
1285                 }
1286         }
1287
1288         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1289                 die("unable to read files to diff");
1290
1291         if (!DIFF_OPT_TST(o, TEXT) &&
1292             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1293                 /* Quite common confusing case */
1294                 if (mf1.size == mf2.size &&
1295                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1296                         goto free_ab_and_return;
1297                 if (DIFF_OPT_TST(o, BINARY))
1298                         emit_binary_diff(&mf1, &mf2);
1299                 else
1300                         printf("Binary files %s and %s differ\n",
1301                                lbl[0], lbl[1]);
1302                 o->found_changes = 1;
1303         }
1304         else {
1305                 /* Crazy xdl interfaces.. */
1306                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1307                 xpparam_t xpp;
1308                 xdemitconf_t xecfg;
1309                 xdemitcb_t ecb;
1310                 struct emit_callback ecbdata;
1311                 const char *funcname_pattern;
1312
1313                 funcname_pattern = diff_funcname_pattern(one);
1314                 if (!funcname_pattern)
1315                         funcname_pattern = diff_funcname_pattern(two);
1316
1317                 memset(&xecfg, 0, sizeof(xecfg));
1318                 memset(&ecbdata, 0, sizeof(ecbdata));
1319                 ecbdata.label_path = lbl;
1320                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1321                 ecbdata.found_changesp = &o->found_changes;
1322                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1323                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1324                 xecfg.ctxlen = o->context;
1325                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1326                 if (funcname_pattern)
1327                         xdiff_set_find_func(&xecfg, funcname_pattern);
1328                 if (!diffopts)
1329                         ;
1330                 else if (!prefixcmp(diffopts, "--unified="))
1331                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1332                 else if (!prefixcmp(diffopts, "-u"))
1333                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1334                 ecb.outf = xdiff_outf;
1335                 ecb.priv = &ecbdata;
1336                 ecbdata.xm.consume = fn_out_consume;
1337                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1338                         ecbdata.diff_words =
1339                                 xcalloc(1, sizeof(struct diff_words_data));
1340                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1341                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1342                         free_diff_words_data(&ecbdata);
1343         }
1344
1345  free_ab_and_return:
1346         diff_free_filespec_data(one);
1347         diff_free_filespec_data(two);
1348         free(a_one);
1349         free(b_two);
1350         return;
1351 }
1352
1353 static void builtin_diffstat(const char *name_a, const char *name_b,
1354                              struct diff_filespec *one,
1355                              struct diff_filespec *two,
1356                              struct diffstat_t *diffstat,
1357                              struct diff_options *o,
1358                              int complete_rewrite)
1359 {
1360         mmfile_t mf1, mf2;
1361         struct diffstat_file *data;
1362
1363         data = diffstat_add(diffstat, name_a, name_b);
1364
1365         if (!one || !two) {
1366                 data->is_unmerged = 1;
1367                 return;
1368         }
1369         if (complete_rewrite) {
1370                 diff_populate_filespec(one, 0);
1371                 diff_populate_filespec(two, 0);
1372                 data->deleted = count_lines(one->data, one->size);
1373                 data->added = count_lines(two->data, two->size);
1374                 goto free_and_return;
1375         }
1376         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1377                 die("unable to read files to diff");
1378
1379         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1380                 data->is_binary = 1;
1381                 data->added = mf2.size;
1382                 data->deleted = mf1.size;
1383         } else {
1384                 /* Crazy xdl interfaces.. */
1385                 xpparam_t xpp;
1386                 xdemitconf_t xecfg;
1387                 xdemitcb_t ecb;
1388
1389                 memset(&xecfg, 0, sizeof(xecfg));
1390                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1391                 ecb.outf = xdiff_outf;
1392                 ecb.priv = diffstat;
1393                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1394         }
1395
1396  free_and_return:
1397         diff_free_filespec_data(one);
1398         diff_free_filespec_data(two);
1399 }
1400
1401 static void builtin_checkdiff(const char *name_a, const char *name_b,
1402                              struct diff_filespec *one,
1403                              struct diff_filespec *two, struct diff_options *o)
1404 {
1405         mmfile_t mf1, mf2;
1406         struct checkdiff_t data;
1407
1408         if (!two)
1409                 return;
1410
1411         memset(&data, 0, sizeof(data));
1412         data.xm.consume = checkdiff_consume;
1413         data.filename = name_b ? name_b : name_a;
1414         data.lineno = 0;
1415         data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1416         data.ws_rule = whitespace_rule(data.filename);
1417
1418         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1419                 die("unable to read files to diff");
1420
1421         if (diff_filespec_is_binary(two))
1422                 goto free_and_return;
1423         else {
1424                 /* Crazy xdl interfaces.. */
1425                 xpparam_t xpp;
1426                 xdemitconf_t xecfg;
1427                 xdemitcb_t ecb;
1428
1429                 memset(&xecfg, 0, sizeof(xecfg));
1430                 xpp.flags = XDF_NEED_MINIMAL;
1431                 ecb.outf = xdiff_outf;
1432                 ecb.priv = &data;
1433                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1434         }
1435  free_and_return:
1436         diff_free_filespec_data(one);
1437         diff_free_filespec_data(two);
1438         if (data.status)
1439                 DIFF_OPT_SET(o, CHECK_FAILED);
1440 }
1441
1442 struct diff_filespec *alloc_filespec(const char *path)
1443 {
1444         int namelen = strlen(path);
1445         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1446
1447         memset(spec, 0, sizeof(*spec));
1448         spec->path = (char *)(spec + 1);
1449         memcpy(spec->path, path, namelen+1);
1450         spec->count = 1;
1451         return spec;
1452 }
1453
1454 void free_filespec(struct diff_filespec *spec)
1455 {
1456         if (!--spec->count) {
1457                 diff_free_filespec_data(spec);
1458                 free(spec);
1459         }
1460 }
1461
1462 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1463                    unsigned short mode)
1464 {
1465         if (mode) {
1466                 spec->mode = canon_mode(mode);
1467                 hashcpy(spec->sha1, sha1);
1468                 spec->sha1_valid = !is_null_sha1(sha1);
1469         }
1470 }
1471
1472 /*
1473  * Given a name and sha1 pair, if the index tells us the file in
1474  * the work tree has that object contents, return true, so that
1475  * prepare_temp_file() does not have to inflate and extract.
1476  */
1477 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1478 {
1479         struct cache_entry *ce;
1480         struct stat st;
1481         int pos, len;
1482
1483         /* We do not read the cache ourselves here, because the
1484          * benchmark with my previous version that always reads cache
1485          * shows that it makes things worse for diff-tree comparing
1486          * two linux-2.6 kernel trees in an already checked out work
1487          * tree.  This is because most diff-tree comparisons deal with
1488          * only a small number of files, while reading the cache is
1489          * expensive for a large project, and its cost outweighs the
1490          * savings we get by not inflating the object to a temporary
1491          * file.  Practically, this code only helps when we are used
1492          * by diff-cache --cached, which does read the cache before
1493          * calling us.
1494          */
1495         if (!active_cache)
1496                 return 0;
1497
1498         /* We want to avoid the working directory if our caller
1499          * doesn't need the data in a normal file, this system
1500          * is rather slow with its stat/open/mmap/close syscalls,
1501          * and the object is contained in a pack file.  The pack
1502          * is probably already open and will be faster to obtain
1503          * the data through than the working directory.  Loose
1504          * objects however would tend to be slower as they need
1505          * to be individually opened and inflated.
1506          */
1507         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1508                 return 0;
1509
1510         len = strlen(name);
1511         pos = cache_name_pos(name, len);
1512         if (pos < 0)
1513                 return 0;
1514         ce = active_cache[pos];
1515         if ((lstat(name, &st) < 0) ||
1516             !S_ISREG(st.st_mode) || /* careful! */
1517             ce_match_stat(ce, &st, 0) ||
1518             hashcmp(sha1, ce->sha1))
1519                 return 0;
1520         /* we return 1 only when we can stat, it is a regular file,
1521          * stat information matches, and sha1 recorded in the cache
1522          * matches.  I.e. we know the file in the work tree really is
1523          * the same as the <name, sha1> pair.
1524          */
1525         return 1;
1526 }
1527
1528 static int populate_from_stdin(struct diff_filespec *s)
1529 {
1530         struct strbuf buf;
1531         size_t size = 0;
1532
1533         strbuf_init(&buf, 0);
1534         if (strbuf_read(&buf, 0, 0) < 0)
1535                 return error("error while reading from stdin %s",
1536                                      strerror(errno));
1537
1538         s->should_munmap = 0;
1539         s->data = strbuf_detach(&buf, &size);
1540         s->size = size;
1541         s->should_free = 1;
1542         return 0;
1543 }
1544
1545 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1546 {
1547         int len;
1548         char *data = xmalloc(100);
1549         len = snprintf(data, 100,
1550                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1551         s->data = data;
1552         s->size = len;
1553         s->should_free = 1;
1554         if (size_only) {
1555                 s->data = NULL;
1556                 free(data);
1557         }
1558         return 0;
1559 }
1560
1561 /*
1562  * While doing rename detection and pickaxe operation, we may need to
1563  * grab the data for the blob (or file) for our own in-core comparison.
1564  * diff_filespec has data and size fields for this purpose.
1565  */
1566 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1567 {
1568         int err = 0;
1569         if (!DIFF_FILE_VALID(s))
1570                 die("internal error: asking to populate invalid file.");
1571         if (S_ISDIR(s->mode))
1572                 return -1;
1573
1574         if (s->data)
1575                 return 0;
1576
1577         if (size_only && 0 < s->size)
1578                 return 0;
1579
1580         if (S_ISGITLINK(s->mode))
1581                 return diff_populate_gitlink(s, size_only);
1582
1583         if (!s->sha1_valid ||
1584             reuse_worktree_file(s->path, s->sha1, 0)) {
1585                 struct strbuf buf;
1586                 struct stat st;
1587                 int fd;
1588
1589                 if (!strcmp(s->path, "-"))
1590                         return populate_from_stdin(s);
1591
1592                 if (lstat(s->path, &st) < 0) {
1593                         if (errno == ENOENT) {
1594                         err_empty:
1595                                 err = -1;
1596                         empty:
1597                                 s->data = (char *)"";
1598                                 s->size = 0;
1599                                 return err;
1600                         }
1601                 }
1602                 s->size = xsize_t(st.st_size);
1603                 if (!s->size)
1604                         goto empty;
1605                 if (size_only)
1606                         return 0;
1607                 if (S_ISLNK(st.st_mode)) {
1608                         int ret;
1609                         s->data = xmalloc(s->size);
1610                         s->should_free = 1;
1611                         ret = readlink(s->path, s->data, s->size);
1612                         if (ret < 0) {
1613                                 free(s->data);
1614                                 goto err_empty;
1615                         }
1616                         return 0;
1617                 }
1618                 fd = open(s->path, O_RDONLY);
1619                 if (fd < 0)
1620                         goto err_empty;
1621                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1622                 close(fd);
1623                 s->should_munmap = 1;
1624
1625                 /*
1626                  * Convert from working tree format to canonical git format
1627                  */
1628                 strbuf_init(&buf, 0);
1629                 if (convert_to_git(s->path, s->data, s->size, &buf)) {
1630                         size_t size = 0;
1631                         munmap(s->data, s->size);
1632                         s->should_munmap = 0;
1633                         s->data = strbuf_detach(&buf, &size);
1634                         s->size = size;
1635                         s->should_free = 1;
1636                 }
1637         }
1638         else {
1639                 enum object_type type;
1640                 if (size_only)
1641                         type = sha1_object_info(s->sha1, &s->size);
1642                 else {
1643                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1644                         s->should_free = 1;
1645                 }
1646         }
1647         return 0;
1648 }
1649
1650 void diff_free_filespec_blob(struct diff_filespec *s)
1651 {
1652         if (s->should_free)
1653                 free(s->data);
1654         else if (s->should_munmap)
1655                 munmap(s->data, s->size);
1656
1657         if (s->should_free || s->should_munmap) {
1658                 s->should_free = s->should_munmap = 0;
1659                 s->data = NULL;
1660         }
1661 }
1662
1663 void diff_free_filespec_data(struct diff_filespec *s)
1664 {
1665         diff_free_filespec_blob(s);
1666         free(s->cnt_data);
1667         s->cnt_data = NULL;
1668 }
1669
1670 static void prep_temp_blob(struct diff_tempfile *temp,
1671                            void *blob,
1672                            unsigned long size,
1673                            const unsigned char *sha1,
1674                            int mode)
1675 {
1676         int fd;
1677
1678         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1679         if (fd < 0)
1680                 die("unable to create temp-file: %s", strerror(errno));
1681         if (write_in_full(fd, blob, size) != size)
1682                 die("unable to write temp-file");
1683         close(fd);
1684         temp->name = temp->tmp_path;
1685         strcpy(temp->hex, sha1_to_hex(sha1));
1686         temp->hex[40] = 0;
1687         sprintf(temp->mode, "%06o", mode);
1688 }
1689
1690 static void prepare_temp_file(const char *name,
1691                               struct diff_tempfile *temp,
1692                               struct diff_filespec *one)
1693 {
1694         if (!DIFF_FILE_VALID(one)) {
1695         not_a_valid_file:
1696                 /* A '-' entry produces this for file-2, and
1697                  * a '+' entry produces this for file-1.
1698                  */
1699                 temp->name = "/dev/null";
1700                 strcpy(temp->hex, ".");
1701                 strcpy(temp->mode, ".");
1702                 return;
1703         }
1704
1705         if (!one->sha1_valid ||
1706             reuse_worktree_file(name, one->sha1, 1)) {
1707                 struct stat st;
1708                 if (lstat(name, &st) < 0) {
1709                         if (errno == ENOENT)
1710                                 goto not_a_valid_file;
1711                         die("stat(%s): %s", name, strerror(errno));
1712                 }
1713                 if (S_ISLNK(st.st_mode)) {
1714                         int ret;
1715                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1716                         size_t sz = xsize_t(st.st_size);
1717                         if (sizeof(buf) <= st.st_size)
1718                                 die("symlink too long: %s", name);
1719                         ret = readlink(name, buf, sz);
1720                         if (ret < 0)
1721                                 die("readlink(%s)", name);
1722                         prep_temp_blob(temp, buf, sz,
1723                                        (one->sha1_valid ?
1724                                         one->sha1 : null_sha1),
1725                                        (one->sha1_valid ?
1726                                         one->mode : S_IFLNK));
1727                 }
1728                 else {
1729                         /* we can borrow from the file in the work tree */
1730                         temp->name = name;
1731                         if (!one->sha1_valid)
1732                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1733                         else
1734                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1735                         /* Even though we may sometimes borrow the
1736                          * contents from the work tree, we always want
1737                          * one->mode.  mode is trustworthy even when
1738                          * !(one->sha1_valid), as long as
1739                          * DIFF_FILE_VALID(one).
1740                          */
1741                         sprintf(temp->mode, "%06o", one->mode);
1742                 }
1743                 return;
1744         }
1745         else {
1746                 if (diff_populate_filespec(one, 0))
1747                         die("cannot read data blob for %s", one->path);
1748                 prep_temp_blob(temp, one->data, one->size,
1749                                one->sha1, one->mode);
1750         }
1751 }
1752
1753 static void remove_tempfile(void)
1754 {
1755         int i;
1756
1757         for (i = 0; i < 2; i++)
1758                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1759                         unlink(diff_temp[i].name);
1760                         diff_temp[i].name = NULL;
1761                 }
1762 }
1763
1764 static void remove_tempfile_on_signal(int signo)
1765 {
1766         remove_tempfile();
1767         signal(SIGINT, SIG_DFL);
1768         raise(signo);
1769 }
1770
1771 /* An external diff command takes:
1772  *
1773  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1774  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1775  *
1776  */
1777 static void run_external_diff(const char *pgm,
1778                               const char *name,
1779                               const char *other,
1780                               struct diff_filespec *one,
1781                               struct diff_filespec *two,
1782                               const char *xfrm_msg,
1783                               int complete_rewrite)
1784 {
1785         const char *spawn_arg[10];
1786         struct diff_tempfile *temp = diff_temp;
1787         int retval;
1788         static int atexit_asked = 0;
1789         const char *othername;
1790         const char **arg = &spawn_arg[0];
1791
1792         othername = (other? other : name);
1793         if (one && two) {
1794                 prepare_temp_file(name, &temp[0], one);
1795                 prepare_temp_file(othername, &temp[1], two);
1796                 if (! atexit_asked &&
1797                     (temp[0].name == temp[0].tmp_path ||
1798                      temp[1].name == temp[1].tmp_path)) {
1799                         atexit_asked = 1;
1800                         atexit(remove_tempfile);
1801                 }
1802                 signal(SIGINT, remove_tempfile_on_signal);
1803         }
1804
1805         if (one && two) {
1806                 *arg++ = pgm;
1807                 *arg++ = name;
1808                 *arg++ = temp[0].name;
1809                 *arg++ = temp[0].hex;
1810                 *arg++ = temp[0].mode;
1811                 *arg++ = temp[1].name;
1812                 *arg++ = temp[1].hex;
1813                 *arg++ = temp[1].mode;
1814                 if (other) {
1815                         *arg++ = other;
1816                         *arg++ = xfrm_msg;
1817                 }
1818         } else {
1819                 *arg++ = pgm;
1820                 *arg++ = name;
1821         }
1822         *arg = NULL;
1823         fflush(NULL);
1824         retval = run_command_v_opt(spawn_arg, 0);
1825         remove_tempfile();
1826         if (retval) {
1827                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1828                 exit(1);
1829         }
1830 }
1831
1832 static const char *external_diff_attr(const char *name)
1833 {
1834         struct git_attr_check attr_diff_check;
1835
1836         setup_diff_attr_check(&attr_diff_check);
1837         if (!git_checkattr(name, 1, &attr_diff_check)) {
1838                 const char *value = attr_diff_check.value;
1839                 if (!ATTR_TRUE(value) &&
1840                     !ATTR_FALSE(value) &&
1841                     !ATTR_UNSET(value)) {
1842                         struct ll_diff_driver *drv;
1843
1844                         for (drv = user_diff; drv; drv = drv->next)
1845                                 if (!strcmp(drv->name, value))
1846                                         return drv->cmd;
1847                 }
1848         }
1849         return NULL;
1850 }
1851
1852 static void run_diff_cmd(const char *pgm,
1853                          const char *name,
1854                          const char *other,
1855                          struct diff_filespec *one,
1856                          struct diff_filespec *two,
1857                          const char *xfrm_msg,
1858                          struct diff_options *o,
1859                          int complete_rewrite)
1860 {
1861         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
1862                 pgm = NULL;
1863         else {
1864                 const char *cmd = external_diff_attr(name);
1865                 if (cmd)
1866                         pgm = cmd;
1867         }
1868
1869         if (pgm) {
1870                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1871                                   complete_rewrite);
1872                 return;
1873         }
1874         if (one && two)
1875                 builtin_diff(name, other ? other : name,
1876                              one, two, xfrm_msg, o, complete_rewrite);
1877         else
1878                 printf("* Unmerged path %s\n", name);
1879 }
1880
1881 static void diff_fill_sha1_info(struct diff_filespec *one)
1882 {
1883         if (DIFF_FILE_VALID(one)) {
1884                 if (!one->sha1_valid) {
1885                         struct stat st;
1886                         if (!strcmp(one->path, "-")) {
1887                                 hashcpy(one->sha1, null_sha1);
1888                                 return;
1889                         }
1890                         if (lstat(one->path, &st) < 0)
1891                                 die("stat %s", one->path);
1892                         if (index_path(one->sha1, one->path, &st, 0))
1893                                 die("cannot hash %s\n", one->path);
1894                 }
1895         }
1896         else
1897                 hashclr(one->sha1);
1898 }
1899
1900 static int similarity_index(struct diff_filepair *p)
1901 {
1902         return p->score * 100 / MAX_SCORE;
1903 }
1904
1905 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1906 {
1907         const char *pgm = external_diff();
1908         struct strbuf msg;
1909         char *xfrm_msg;
1910         struct diff_filespec *one = p->one;
1911         struct diff_filespec *two = p->two;
1912         const char *name;
1913         const char *other;
1914         int complete_rewrite = 0;
1915
1916
1917         if (DIFF_PAIR_UNMERGED(p)) {
1918                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1919                 return;
1920         }
1921
1922         name  = p->one->path;
1923         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1924         diff_fill_sha1_info(one);
1925         diff_fill_sha1_info(two);
1926
1927         strbuf_init(&msg, PATH_MAX * 2 + 300);
1928         switch (p->status) {
1929         case DIFF_STATUS_COPIED:
1930                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1931                 strbuf_addstr(&msg, "\ncopy from ");
1932                 quote_c_style(name, &msg, NULL, 0);
1933                 strbuf_addstr(&msg, "\ncopy to ");
1934                 quote_c_style(other, &msg, NULL, 0);
1935                 strbuf_addch(&msg, '\n');
1936                 break;
1937         case DIFF_STATUS_RENAMED:
1938                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1939                 strbuf_addstr(&msg, "\nrename from ");
1940                 quote_c_style(name, &msg, NULL, 0);
1941                 strbuf_addstr(&msg, "\nrename to ");
1942                 quote_c_style(other, &msg, NULL, 0);
1943                 strbuf_addch(&msg, '\n');
1944                 break;
1945         case DIFF_STATUS_MODIFIED:
1946                 if (p->score) {
1947                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
1948                                         similarity_index(p));
1949                         complete_rewrite = 1;
1950                         break;
1951                 }
1952                 /* fallthru */
1953         default:
1954                 /* nothing */
1955                 ;
1956         }
1957
1958         if (hashcmp(one->sha1, two->sha1)) {
1959                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
1960
1961                 if (DIFF_OPT_TST(o, BINARY)) {
1962                         mmfile_t mf;
1963                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
1964                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
1965                                 abbrev = 40;
1966                 }
1967                 strbuf_addf(&msg, "index %.*s..%.*s",
1968                                 abbrev, sha1_to_hex(one->sha1),
1969                                 abbrev, sha1_to_hex(two->sha1));
1970                 if (one->mode == two->mode)
1971                         strbuf_addf(&msg, " %06o", one->mode);
1972                 strbuf_addch(&msg, '\n');
1973         }
1974
1975         if (msg.len)
1976                 strbuf_setlen(&msg, msg.len - 1);
1977         xfrm_msg = msg.len ? msg.buf : NULL;
1978
1979         if (!pgm &&
1980             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1981             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1982                 /* a filepair that changes between file and symlink
1983                  * needs to be split into deletion and creation.
1984                  */
1985                 struct diff_filespec *null = alloc_filespec(two->path);
1986                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1987                 free(null);
1988                 null = alloc_filespec(one->path);
1989                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1990                 free(null);
1991         }
1992         else
1993                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1994                              complete_rewrite);
1995
1996         strbuf_release(&msg);
1997 }
1998
1999 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2000                          struct diffstat_t *diffstat)
2001 {
2002         const char *name;
2003         const char *other;
2004         int complete_rewrite = 0;
2005
2006         if (DIFF_PAIR_UNMERGED(p)) {
2007                 /* unmerged */
2008                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2009                 return;
2010         }
2011
2012         name = p->one->path;
2013         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2014
2015         diff_fill_sha1_info(p->one);
2016         diff_fill_sha1_info(p->two);
2017
2018         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2019                 complete_rewrite = 1;
2020         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2021 }
2022
2023 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2024 {
2025         const char *name;
2026         const char *other;
2027
2028         if (DIFF_PAIR_UNMERGED(p)) {
2029                 /* unmerged */
2030                 return;
2031         }
2032
2033         name = p->one->path;
2034         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2035
2036         diff_fill_sha1_info(p->one);
2037         diff_fill_sha1_info(p->two);
2038
2039         builtin_checkdiff(name, other, p->one, p->two, o);
2040 }
2041
2042 void diff_setup(struct diff_options *options)
2043 {
2044         memset(options, 0, sizeof(*options));
2045         options->line_termination = '\n';
2046         options->break_opt = -1;
2047         options->rename_limit = -1;
2048         options->context = 3;
2049         options->msg_sep = "";
2050
2051         options->change = diff_change;
2052         options->add_remove = diff_addremove;
2053         if (diff_use_color_default)
2054                 DIFF_OPT_SET(options, COLOR_DIFF);
2055         else
2056                 DIFF_OPT_CLR(options, COLOR_DIFF);
2057         options->detect_rename = diff_detect_rename_default;
2058
2059         options->a_prefix = "a/";
2060         options->b_prefix = "b/";
2061 }
2062
2063 int diff_setup_done(struct diff_options *options)
2064 {
2065         int count = 0;
2066
2067         if (options->output_format & DIFF_FORMAT_NAME)
2068                 count++;
2069         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2070                 count++;
2071         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2072                 count++;
2073         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2074                 count++;
2075         if (count > 1)
2076                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2077
2078         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2079                 options->detect_rename = DIFF_DETECT_COPY;
2080
2081         if (options->output_format & (DIFF_FORMAT_NAME |
2082                                       DIFF_FORMAT_NAME_STATUS |
2083                                       DIFF_FORMAT_CHECKDIFF |
2084                                       DIFF_FORMAT_NO_OUTPUT))
2085                 options->output_format &= ~(DIFF_FORMAT_RAW |
2086                                             DIFF_FORMAT_NUMSTAT |
2087                                             DIFF_FORMAT_DIFFSTAT |
2088                                             DIFF_FORMAT_SHORTSTAT |
2089                                             DIFF_FORMAT_SUMMARY |
2090                                             DIFF_FORMAT_PATCH);
2091
2092         /*
2093          * These cases always need recursive; we do not drop caller-supplied
2094          * recursive bits for other formats here.
2095          */
2096         if (options->output_format & (DIFF_FORMAT_PATCH |
2097                                       DIFF_FORMAT_NUMSTAT |
2098                                       DIFF_FORMAT_DIFFSTAT |
2099                                       DIFF_FORMAT_SHORTSTAT |
2100                                       DIFF_FORMAT_SUMMARY |
2101                                       DIFF_FORMAT_CHECKDIFF))
2102                 DIFF_OPT_SET(options, RECURSIVE);
2103         /*
2104          * Also pickaxe would not work very well if you do not say recursive
2105          */
2106         if (options->pickaxe)
2107                 DIFF_OPT_SET(options, RECURSIVE);
2108
2109         if (options->detect_rename && options->rename_limit < 0)
2110                 options->rename_limit = diff_rename_limit_default;
2111         if (options->setup & DIFF_SETUP_USE_CACHE) {
2112                 if (!active_cache)
2113                         /* read-cache does not die even when it fails
2114                          * so it is safe for us to do this here.  Also
2115                          * it does not smudge active_cache or active_nr
2116                          * when it fails, so we do not have to worry about
2117                          * cleaning it up ourselves either.
2118                          */
2119                         read_cache();
2120         }
2121         if (options->abbrev <= 0 || 40 < options->abbrev)
2122                 options->abbrev = 40; /* full */
2123
2124         /*
2125          * It does not make sense to show the first hit we happened
2126          * to have found.  It does not make sense not to return with
2127          * exit code in such a case either.
2128          */
2129         if (DIFF_OPT_TST(options, QUIET)) {
2130                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2131                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2132         }
2133
2134         /*
2135          * If we postprocess in diffcore, we cannot simply return
2136          * upon the first hit.  We need to run diff as usual.
2137          */
2138         if (options->pickaxe || options->filter)
2139                 DIFF_OPT_CLR(options, QUIET);
2140
2141         return 0;
2142 }
2143
2144 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2145 {
2146         char c, *eq;
2147         int len;
2148
2149         if (*arg != '-')
2150                 return 0;
2151         c = *++arg;
2152         if (!c)
2153                 return 0;
2154         if (c == arg_short) {
2155                 c = *++arg;
2156                 if (!c)
2157                         return 1;
2158                 if (val && isdigit(c)) {
2159                         char *end;
2160                         int n = strtoul(arg, &end, 10);
2161                         if (*end)
2162                                 return 0;
2163                         *val = n;
2164                         return 1;
2165                 }
2166                 return 0;
2167         }
2168         if (c != '-')
2169                 return 0;
2170         arg++;
2171         eq = strchr(arg, '=');
2172         if (eq)
2173                 len = eq - arg;
2174         else
2175                 len = strlen(arg);
2176         if (!len || strncmp(arg, arg_long, len))
2177                 return 0;
2178         if (eq) {
2179                 int n;
2180                 char *end;
2181                 if (!isdigit(*++eq))
2182                         return 0;
2183                 n = strtoul(eq, &end, 10);
2184                 if (*end)
2185                         return 0;
2186                 *val = n;
2187         }
2188         return 1;
2189 }
2190
2191 static int diff_scoreopt_parse(const char *opt);
2192
2193 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2194 {
2195         const char *arg = av[0];
2196
2197         /* Output format options */
2198         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2199                 options->output_format |= DIFF_FORMAT_PATCH;
2200         else if (opt_arg(arg, 'U', "unified", &options->context))
2201                 options->output_format |= DIFF_FORMAT_PATCH;
2202         else if (!strcmp(arg, "--raw"))
2203                 options->output_format |= DIFF_FORMAT_RAW;
2204         else if (!strcmp(arg, "--patch-with-raw"))
2205                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2206         else if (!strcmp(arg, "--numstat"))
2207                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2208         else if (!strcmp(arg, "--shortstat"))
2209                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2210         else if (!strcmp(arg, "--check"))
2211                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2212         else if (!strcmp(arg, "--summary"))
2213                 options->output_format |= DIFF_FORMAT_SUMMARY;
2214         else if (!strcmp(arg, "--patch-with-stat"))
2215                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2216         else if (!strcmp(arg, "--name-only"))
2217                 options->output_format |= DIFF_FORMAT_NAME;
2218         else if (!strcmp(arg, "--name-status"))
2219                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2220         else if (!strcmp(arg, "-s"))
2221                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2222         else if (!prefixcmp(arg, "--stat")) {
2223                 char *end;
2224                 int width = options->stat_width;
2225                 int name_width = options->stat_name_width;
2226                 arg += 6;
2227                 end = (char *)arg;
2228
2229                 switch (*arg) {
2230                 case '-':
2231                         if (!prefixcmp(arg, "-width="))
2232                                 width = strtoul(arg + 7, &end, 10);
2233                         else if (!prefixcmp(arg, "-name-width="))
2234                                 name_width = strtoul(arg + 12, &end, 10);
2235                         break;
2236                 case '=':
2237                         width = strtoul(arg+1, &end, 10);
2238                         if (*end == ',')
2239                                 name_width = strtoul(end+1, &end, 10);
2240                 }
2241
2242                 /* Important! This checks all the error cases! */
2243                 if (*end)
2244                         return 0;
2245                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2246                 options->stat_name_width = name_width;
2247                 options->stat_width = width;
2248         }
2249
2250         /* renames options */
2251         else if (!prefixcmp(arg, "-B")) {
2252                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2253                         return -1;
2254         }
2255         else if (!prefixcmp(arg, "-M")) {
2256                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2257                         return -1;
2258                 options->detect_rename = DIFF_DETECT_RENAME;
2259         }
2260         else if (!prefixcmp(arg, "-C")) {
2261                 if (options->detect_rename == DIFF_DETECT_COPY)
2262                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2263                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2264                         return -1;
2265                 options->detect_rename = DIFF_DETECT_COPY;
2266         }
2267         else if (!strcmp(arg, "--no-renames"))
2268                 options->detect_rename = 0;
2269
2270         /* xdiff options */
2271         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2272                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2273         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2274                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2275         else if (!strcmp(arg, "--ignore-space-at-eol"))
2276                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2277
2278         /* flags options */
2279         else if (!strcmp(arg, "--binary")) {
2280                 options->output_format |= DIFF_FORMAT_PATCH;
2281                 DIFF_OPT_SET(options, BINARY);
2282         }
2283         else if (!strcmp(arg, "--full-index"))
2284                 DIFF_OPT_SET(options, FULL_INDEX);
2285         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2286                 DIFF_OPT_SET(options, TEXT);
2287         else if (!strcmp(arg, "-R"))
2288                 DIFF_OPT_SET(options, REVERSE_DIFF);
2289         else if (!strcmp(arg, "--find-copies-harder"))
2290                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2291         else if (!strcmp(arg, "--follow"))
2292                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2293         else if (!strcmp(arg, "--color"))
2294                 DIFF_OPT_SET(options, COLOR_DIFF);
2295         else if (!strcmp(arg, "--no-color"))
2296                 DIFF_OPT_CLR(options, COLOR_DIFF);
2297         else if (!strcmp(arg, "--color-words"))
2298                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2299         else if (!strcmp(arg, "--exit-code"))
2300                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2301         else if (!strcmp(arg, "--quiet"))
2302                 DIFF_OPT_SET(options, QUIET);
2303         else if (!strcmp(arg, "--ext-diff"))
2304                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2305         else if (!strcmp(arg, "--no-ext-diff"))
2306                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2307
2308         /* misc options */
2309         else if (!strcmp(arg, "-z"))
2310                 options->line_termination = 0;
2311         else if (!prefixcmp(arg, "-l"))
2312                 options->rename_limit = strtoul(arg+2, NULL, 10);
2313         else if (!prefixcmp(arg, "-S"))
2314                 options->pickaxe = arg + 2;
2315         else if (!strcmp(arg, "--pickaxe-all"))
2316                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2317         else if (!strcmp(arg, "--pickaxe-regex"))
2318                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2319         else if (!prefixcmp(arg, "-O"))
2320                 options->orderfile = arg + 2;
2321         else if (!prefixcmp(arg, "--diff-filter="))
2322                 options->filter = arg + 14;
2323         else if (!strcmp(arg, "--abbrev"))
2324                 options->abbrev = DEFAULT_ABBREV;
2325         else if (!prefixcmp(arg, "--abbrev=")) {
2326                 options->abbrev = strtoul(arg + 9, NULL, 10);
2327                 if (options->abbrev < MINIMUM_ABBREV)
2328                         options->abbrev = MINIMUM_ABBREV;
2329                 else if (40 < options->abbrev)
2330                         options->abbrev = 40;
2331         }
2332         else if (!prefixcmp(arg, "--src-prefix="))
2333                 options->a_prefix = arg + 13;
2334         else if (!prefixcmp(arg, "--dst-prefix="))
2335                 options->b_prefix = arg + 13;
2336         else if (!strcmp(arg, "--no-prefix"))
2337                 options->a_prefix = options->b_prefix = "";
2338         else
2339                 return 0;
2340         return 1;
2341 }
2342
2343 static int parse_num(const char **cp_p)
2344 {
2345         unsigned long num, scale;
2346         int ch, dot;
2347         const char *cp = *cp_p;
2348
2349         num = 0;
2350         scale = 1;
2351         dot = 0;
2352         for(;;) {
2353                 ch = *cp;
2354                 if ( !dot && ch == '.' ) {
2355                         scale = 1;
2356                         dot = 1;
2357                 } else if ( ch == '%' ) {
2358                         scale = dot ? scale*100 : 100;
2359                         cp++;   /* % is always at the end */
2360                         break;
2361                 } else if ( ch >= '0' && ch <= '9' ) {
2362                         if ( scale < 100000 ) {
2363                                 scale *= 10;
2364                                 num = (num*10) + (ch-'0');
2365                         }
2366                 } else {
2367                         break;
2368                 }
2369                 cp++;
2370         }
2371         *cp_p = cp;
2372
2373         /* user says num divided by scale and we say internally that
2374          * is MAX_SCORE * num / scale.
2375          */
2376         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2377 }
2378
2379 static int diff_scoreopt_parse(const char *opt)
2380 {
2381         int opt1, opt2, cmd;
2382
2383         if (*opt++ != '-')
2384                 return -1;
2385         cmd = *opt++;
2386         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2387                 return -1; /* that is not a -M, -C nor -B option */
2388
2389         opt1 = parse_num(&opt);
2390         if (cmd != 'B')
2391                 opt2 = 0;
2392         else {
2393                 if (*opt == 0)
2394                         opt2 = 0;
2395                 else if (*opt != '/')
2396                         return -1; /* we expect -B80/99 or -B80 */
2397                 else {
2398                         opt++;
2399                         opt2 = parse_num(&opt);
2400                 }
2401         }
2402         if (*opt != 0)
2403                 return -1;
2404         return opt1 | (opt2 << 16);
2405 }
2406
2407 struct diff_queue_struct diff_queued_diff;
2408
2409 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2410 {
2411         if (queue->alloc <= queue->nr) {
2412                 queue->alloc = alloc_nr(queue->alloc);
2413                 queue->queue = xrealloc(queue->queue,
2414                                         sizeof(dp) * queue->alloc);
2415         }
2416         queue->queue[queue->nr++] = dp;
2417 }
2418
2419 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2420                                  struct diff_filespec *one,
2421                                  struct diff_filespec *two)
2422 {
2423         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2424         dp->one = one;
2425         dp->two = two;
2426         if (queue)
2427                 diff_q(queue, dp);
2428         return dp;
2429 }
2430
2431 void diff_free_filepair(struct diff_filepair *p)
2432 {
2433         free_filespec(p->one);
2434         free_filespec(p->two);
2435         free(p);
2436 }
2437
2438 /* This is different from find_unique_abbrev() in that
2439  * it stuffs the result with dots for alignment.
2440  */
2441 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2442 {
2443         int abblen;
2444         const char *abbrev;
2445         if (len == 40)
2446                 return sha1_to_hex(sha1);
2447
2448         abbrev = find_unique_abbrev(sha1, len);
2449         if (!abbrev)
2450                 return sha1_to_hex(sha1);
2451         abblen = strlen(abbrev);
2452         if (abblen < 37) {
2453                 static char hex[41];
2454                 if (len < abblen && abblen <= len + 2)
2455                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2456                 else
2457                         sprintf(hex, "%s...", abbrev);
2458                 return hex;
2459         }
2460         return sha1_to_hex(sha1);
2461 }
2462
2463 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2464 {
2465         int line_termination = opt->line_termination;
2466         int inter_name_termination = line_termination ? '\t' : '\0';
2467
2468         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2469                 printf(":%06o %06o %s ", p->one->mode, p->two->mode,
2470                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
2471                 printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2472         }
2473         if (p->score) {
2474                 printf("%c%03d%c", p->status, similarity_index(p),
2475                            inter_name_termination);
2476         } else {
2477                 printf("%c%c", p->status, inter_name_termination);
2478         }
2479
2480         if (p->status == DIFF_STATUS_COPIED || p->status == DIFF_STATUS_RENAMED) {
2481                 write_name_quoted(p->one->path, stdout, inter_name_termination);
2482                 write_name_quoted(p->two->path, stdout, line_termination);
2483         } else {
2484                 const char *path = p->one->mode ? p->one->path : p->two->path;
2485                 write_name_quoted(path, stdout, line_termination);
2486         }
2487 }
2488
2489 int diff_unmodified_pair(struct diff_filepair *p)
2490 {
2491         /* This function is written stricter than necessary to support
2492          * the currently implemented transformers, but the idea is to
2493          * let transformers to produce diff_filepairs any way they want,
2494          * and filter and clean them up here before producing the output.
2495          */
2496         struct diff_filespec *one = p->one, *two = p->two;
2497
2498         if (DIFF_PAIR_UNMERGED(p))
2499                 return 0; /* unmerged is interesting */
2500
2501         /* deletion, addition, mode or type change
2502          * and rename are all interesting.
2503          */
2504         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2505             DIFF_PAIR_MODE_CHANGED(p) ||
2506             strcmp(one->path, two->path))
2507                 return 0;
2508
2509         /* both are valid and point at the same path.  that is, we are
2510          * dealing with a change.
2511          */
2512         if (one->sha1_valid && two->sha1_valid &&
2513             !hashcmp(one->sha1, two->sha1))
2514                 return 1; /* no change */
2515         if (!one->sha1_valid && !two->sha1_valid)
2516                 return 1; /* both look at the same file on the filesystem. */
2517         return 0;
2518 }
2519
2520 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2521 {
2522         if (diff_unmodified_pair(p))
2523                 return;
2524
2525         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2526             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2527                 return; /* no tree diffs in patch format */
2528
2529         run_diff(p, o);
2530 }
2531
2532 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2533                             struct diffstat_t *diffstat)
2534 {
2535         if (diff_unmodified_pair(p))
2536                 return;
2537
2538         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2539             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2540                 return; /* no tree diffs in patch format */
2541
2542         run_diffstat(p, o, diffstat);
2543 }
2544
2545 static void diff_flush_checkdiff(struct diff_filepair *p,
2546                 struct diff_options *o)
2547 {
2548         if (diff_unmodified_pair(p))
2549                 return;
2550
2551         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2552             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2553                 return; /* no tree diffs in patch format */
2554
2555         run_checkdiff(p, o);
2556 }
2557
2558 int diff_queue_is_empty(void)
2559 {
2560         struct diff_queue_struct *q = &diff_queued_diff;
2561         int i;
2562         for (i = 0; i < q->nr; i++)
2563                 if (!diff_unmodified_pair(q->queue[i]))
2564                         return 0;
2565         return 1;
2566 }
2567
2568 #if DIFF_DEBUG
2569 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2570 {
2571         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2572                 x, one ? one : "",
2573                 s->path,
2574                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2575                 s->mode,
2576                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2577         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2578                 x, one ? one : "",
2579                 s->size, s->xfrm_flags);
2580 }
2581
2582 void diff_debug_filepair(const struct diff_filepair *p, int i)
2583 {
2584         diff_debug_filespec(p->one, i, "one");
2585         diff_debug_filespec(p->two, i, "two");
2586         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2587                 p->score, p->status ? p->status : '?',
2588                 p->one->rename_used, p->broken_pair);
2589 }
2590
2591 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2592 {
2593         int i;
2594         if (msg)
2595                 fprintf(stderr, "%s\n", msg);
2596         fprintf(stderr, "q->nr = %d\n", q->nr);
2597         for (i = 0; i < q->nr; i++) {
2598                 struct diff_filepair *p = q->queue[i];
2599                 diff_debug_filepair(p, i);
2600         }
2601 }
2602 #endif
2603
2604 static void diff_resolve_rename_copy(void)
2605 {
2606         int i;
2607         struct diff_filepair *p;
2608         struct diff_queue_struct *q = &diff_queued_diff;
2609
2610         diff_debug_queue("resolve-rename-copy", q);
2611
2612         for (i = 0; i < q->nr; i++) {
2613                 p = q->queue[i];
2614                 p->status = 0; /* undecided */
2615                 if (DIFF_PAIR_UNMERGED(p))
2616                         p->status = DIFF_STATUS_UNMERGED;
2617                 else if (!DIFF_FILE_VALID(p->one))
2618                         p->status = DIFF_STATUS_ADDED;
2619                 else if (!DIFF_FILE_VALID(p->two))
2620                         p->status = DIFF_STATUS_DELETED;
2621                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2622                         p->status = DIFF_STATUS_TYPE_CHANGED;
2623
2624                 /* from this point on, we are dealing with a pair
2625                  * whose both sides are valid and of the same type, i.e.
2626                  * either in-place edit or rename/copy edit.
2627                  */
2628                 else if (DIFF_PAIR_RENAME(p)) {
2629                         /*
2630                          * A rename might have re-connected a broken
2631                          * pair up, causing the pathnames to be the
2632                          * same again. If so, that's not a rename at
2633                          * all, just a modification..
2634                          *
2635                          * Otherwise, see if this source was used for
2636                          * multiple renames, in which case we decrement
2637                          * the count, and call it a copy.
2638                          */
2639                         if (!strcmp(p->one->path, p->two->path))
2640                                 p->status = DIFF_STATUS_MODIFIED;
2641                         else if (--p->one->rename_used > 0)
2642                                 p->status = DIFF_STATUS_COPIED;
2643                         else
2644                                 p->status = DIFF_STATUS_RENAMED;
2645                 }
2646                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2647                          p->one->mode != p->two->mode ||
2648                          is_null_sha1(p->one->sha1))
2649                         p->status = DIFF_STATUS_MODIFIED;
2650                 else {
2651                         /* This is a "no-change" entry and should not
2652                          * happen anymore, but prepare for broken callers.
2653                          */
2654                         error("feeding unmodified %s to diffcore",
2655                               p->one->path);
2656                         p->status = DIFF_STATUS_UNKNOWN;
2657                 }
2658         }
2659         diff_debug_queue("resolve-rename-copy done", q);
2660 }
2661
2662 static int check_pair_status(struct diff_filepair *p)
2663 {
2664         switch (p->status) {
2665         case DIFF_STATUS_UNKNOWN:
2666                 return 0;
2667         case 0:
2668                 die("internal error in diff-resolve-rename-copy");
2669         default:
2670                 return 1;
2671         }
2672 }
2673
2674 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2675 {
2676         int fmt = opt->output_format;
2677
2678         if (fmt & DIFF_FORMAT_CHECKDIFF)
2679                 diff_flush_checkdiff(p, opt);
2680         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2681                 diff_flush_raw(p, opt);
2682         else if (fmt & DIFF_FORMAT_NAME)
2683                 write_name_quoted(p->two->path, stdout, opt->line_termination);
2684 }
2685
2686 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2687 {
2688         if (fs->mode)
2689                 printf(" %s mode %06o ", newdelete, fs->mode);
2690         else
2691                 printf(" %s ", newdelete);
2692         write_name_quoted(fs->path, stdout, '\n');
2693 }
2694
2695
2696 static void show_mode_change(struct diff_filepair *p, int show_name)
2697 {
2698         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2699                 printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
2700                         show_name ? ' ' : '\n');
2701                 if (show_name) {
2702                         write_name_quoted(p->two->path, stdout, '\n');
2703                 }
2704         }
2705 }
2706
2707 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2708 {
2709         char *names = pprint_rename(p->one->path, p->two->path);
2710
2711         printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2712         free(names);
2713         show_mode_change(p, 0);
2714 }
2715
2716 static void diff_summary(struct diff_filepair *p)
2717 {
2718         switch(p->status) {
2719         case DIFF_STATUS_DELETED:
2720                 show_file_mode_name("delete", p->one);
2721                 break;
2722         case DIFF_STATUS_ADDED:
2723                 show_file_mode_name("create", p->two);
2724                 break;
2725         case DIFF_STATUS_COPIED:
2726                 show_rename_copy("copy", p);
2727                 break;
2728         case DIFF_STATUS_RENAMED:
2729                 show_rename_copy("rename", p);
2730                 break;
2731         default:
2732                 if (p->score) {
2733                         fputs(" rewrite ", stdout);
2734                         write_name_quoted(p->two->path, stdout, ' ');
2735                         printf("(%d%%)\n", similarity_index(p));
2736                 }
2737                 show_mode_change(p, !p->score);
2738                 break;
2739         }
2740 }
2741
2742 struct patch_id_t {
2743         struct xdiff_emit_state xm;
2744         SHA_CTX *ctx;
2745         int patchlen;
2746 };
2747
2748 static int remove_space(char *line, int len)
2749 {
2750         int i;
2751         char *dst = line;
2752         unsigned char c;
2753
2754         for (i = 0; i < len; i++)
2755                 if (!isspace((c = line[i])))
2756                         *dst++ = c;
2757
2758         return dst - line;
2759 }
2760
2761 static void patch_id_consume(void *priv, char *line, unsigned long len)
2762 {
2763         struct patch_id_t *data = priv;
2764         int new_len;
2765
2766         /* Ignore line numbers when computing the SHA1 of the patch */
2767         if (!prefixcmp(line, "@@ -"))
2768                 return;
2769
2770         new_len = remove_space(line, len);
2771
2772         SHA1_Update(data->ctx, line, new_len);
2773         data->patchlen += new_len;
2774 }
2775
2776 /* returns 0 upon success, and writes result into sha1 */
2777 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2778 {
2779         struct diff_queue_struct *q = &diff_queued_diff;
2780         int i;
2781         SHA_CTX ctx;
2782         struct patch_id_t data;
2783         char buffer[PATH_MAX * 4 + 20];
2784
2785         SHA1_Init(&ctx);
2786         memset(&data, 0, sizeof(struct patch_id_t));
2787         data.ctx = &ctx;
2788         data.xm.consume = patch_id_consume;
2789
2790         for (i = 0; i < q->nr; i++) {
2791                 xpparam_t xpp;
2792                 xdemitconf_t xecfg;
2793                 xdemitcb_t ecb;
2794                 mmfile_t mf1, mf2;
2795                 struct diff_filepair *p = q->queue[i];
2796                 int len1, len2;
2797
2798                 memset(&xecfg, 0, sizeof(xecfg));
2799                 if (p->status == 0)
2800                         return error("internal diff status error");
2801                 if (p->status == DIFF_STATUS_UNKNOWN)
2802                         continue;
2803                 if (diff_unmodified_pair(p))
2804                         continue;
2805                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2806                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2807                         continue;
2808                 if (DIFF_PAIR_UNMERGED(p))
2809                         continue;
2810
2811                 diff_fill_sha1_info(p->one);
2812                 diff_fill_sha1_info(p->two);
2813                 if (fill_mmfile(&mf1, p->one) < 0 ||
2814                                 fill_mmfile(&mf2, p->two) < 0)
2815                         return error("unable to read files to diff");
2816
2817                 len1 = remove_space(p->one->path, strlen(p->one->path));
2818                 len2 = remove_space(p->two->path, strlen(p->two->path));
2819                 if (p->one->mode == 0)
2820                         len1 = snprintf(buffer, sizeof(buffer),
2821                                         "diff--gita/%.*sb/%.*s"
2822                                         "newfilemode%06o"
2823                                         "---/dev/null"
2824                                         "+++b/%.*s",
2825                                         len1, p->one->path,
2826                                         len2, p->two->path,
2827                                         p->two->mode,
2828                                         len2, p->two->path);
2829                 else if (p->two->mode == 0)
2830                         len1 = snprintf(buffer, sizeof(buffer),
2831                                         "diff--gita/%.*sb/%.*s"
2832                                         "deletedfilemode%06o"
2833                                         "---a/%.*s"
2834                                         "+++/dev/null",
2835                                         len1, p->one->path,
2836                                         len2, p->two->path,
2837                                         p->one->mode,
2838                                         len1, p->one->path);
2839                 else
2840                         len1 = snprintf(buffer, sizeof(buffer),
2841                                         "diff--gita/%.*sb/%.*s"
2842                                         "---a/%.*s"
2843                                         "+++b/%.*s",
2844                                         len1, p->one->path,
2845                                         len2, p->two->path,
2846                                         len1, p->one->path,
2847                                         len2, p->two->path);
2848                 SHA1_Update(&ctx, buffer, len1);
2849
2850                 xpp.flags = XDF_NEED_MINIMAL;
2851                 xecfg.ctxlen = 3;
2852                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2853                 ecb.outf = xdiff_outf;
2854                 ecb.priv = &data;
2855                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2856         }
2857
2858         SHA1_Final(sha1, &ctx);
2859         return 0;
2860 }
2861
2862 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2863 {
2864         struct diff_queue_struct *q = &diff_queued_diff;
2865         int i;
2866         int result = diff_get_patch_id(options, sha1);
2867
2868         for (i = 0; i < q->nr; i++)
2869                 diff_free_filepair(q->queue[i]);
2870
2871         free(q->queue);
2872         q->queue = NULL;
2873         q->nr = q->alloc = 0;
2874
2875         return result;
2876 }
2877
2878 static int is_summary_empty(const struct diff_queue_struct *q)
2879 {
2880         int i;
2881
2882         for (i = 0; i < q->nr; i++) {
2883                 const struct diff_filepair *p = q->queue[i];
2884
2885                 switch (p->status) {
2886                 case DIFF_STATUS_DELETED:
2887                 case DIFF_STATUS_ADDED:
2888                 case DIFF_STATUS_COPIED:
2889                 case DIFF_STATUS_RENAMED:
2890                         return 0;
2891                 default:
2892                         if (p->score)
2893                                 return 0;
2894                         if (p->one->mode && p->two->mode &&
2895                             p->one->mode != p->two->mode)
2896                                 return 0;
2897                         break;
2898                 }
2899         }
2900         return 1;
2901 }
2902
2903 void diff_flush(struct diff_options *options)
2904 {
2905         struct diff_queue_struct *q = &diff_queued_diff;
2906         int i, output_format = options->output_format;
2907         int separator = 0;
2908
2909         /*
2910          * Order: raw, stat, summary, patch
2911          * or:    name/name-status/checkdiff (other bits clear)
2912          */
2913         if (!q->nr)
2914                 goto free_queue;
2915
2916         if (output_format & (DIFF_FORMAT_RAW |
2917                              DIFF_FORMAT_NAME |
2918                              DIFF_FORMAT_NAME_STATUS |
2919                              DIFF_FORMAT_CHECKDIFF)) {
2920                 for (i = 0; i < q->nr; i++) {
2921                         struct diff_filepair *p = q->queue[i];
2922                         if (check_pair_status(p))
2923                                 flush_one_pair(p, options);
2924                 }
2925                 separator++;
2926         }
2927
2928         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2929                 struct diffstat_t diffstat;
2930
2931                 memset(&diffstat, 0, sizeof(struct diffstat_t));
2932                 diffstat.xm.consume = diffstat_consume;
2933                 for (i = 0; i < q->nr; i++) {
2934                         struct diff_filepair *p = q->queue[i];
2935                         if (check_pair_status(p))
2936                                 diff_flush_stat(p, options, &diffstat);
2937                 }
2938                 if (output_format & DIFF_FORMAT_NUMSTAT)
2939                         show_numstat(&diffstat, options);
2940                 if (output_format & DIFF_FORMAT_DIFFSTAT)
2941                         show_stats(&diffstat, options);
2942                 if (output_format & DIFF_FORMAT_SHORTSTAT)
2943                         show_shortstats(&diffstat);
2944                 free_diffstat_info(&diffstat);
2945                 separator++;
2946         }
2947
2948         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2949                 for (i = 0; i < q->nr; i++)
2950                         diff_summary(q->queue[i]);
2951                 separator++;
2952         }
2953
2954         if (output_format & DIFF_FORMAT_PATCH) {
2955                 if (separator) {
2956                         if (options->stat_sep) {
2957                                 /* attach patch instead of inline */
2958                                 fputs(options->stat_sep, stdout);
2959                         } else {
2960                                 putchar(options->line_termination);
2961                         }
2962                 }
2963
2964                 for (i = 0; i < q->nr; i++) {
2965                         struct diff_filepair *p = q->queue[i];
2966                         if (check_pair_status(p))
2967                                 diff_flush_patch(p, options);
2968                 }
2969         }
2970
2971         if (output_format & DIFF_FORMAT_CALLBACK)
2972                 options->format_callback(q, options, options->format_callback_data);
2973
2974         for (i = 0; i < q->nr; i++)
2975                 diff_free_filepair(q->queue[i]);
2976 free_queue:
2977         free(q->queue);
2978         q->queue = NULL;
2979         q->nr = q->alloc = 0;
2980 }
2981
2982 static void diffcore_apply_filter(const char *filter)
2983 {
2984         int i;
2985         struct diff_queue_struct *q = &diff_queued_diff;
2986         struct diff_queue_struct outq;
2987         outq.queue = NULL;
2988         outq.nr = outq.alloc = 0;
2989
2990         if (!filter)
2991                 return;
2992
2993         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2994                 int found;
2995                 for (i = found = 0; !found && i < q->nr; i++) {
2996                         struct diff_filepair *p = q->queue[i];
2997                         if (((p->status == DIFF_STATUS_MODIFIED) &&
2998                              ((p->score &&
2999                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3000                               (!p->score &&
3001                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3002                             ((p->status != DIFF_STATUS_MODIFIED) &&
3003                              strchr(filter, p->status)))
3004                                 found++;
3005                 }
3006                 if (found)
3007                         return;
3008
3009                 /* otherwise we will clear the whole queue
3010                  * by copying the empty outq at the end of this
3011                  * function, but first clear the current entries
3012                  * in the queue.
3013                  */
3014                 for (i = 0; i < q->nr; i++)
3015                         diff_free_filepair(q->queue[i]);
3016         }
3017         else {
3018                 /* Only the matching ones */
3019                 for (i = 0; i < q->nr; i++) {
3020                         struct diff_filepair *p = q->queue[i];
3021
3022                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3023                              ((p->score &&
3024                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3025                               (!p->score &&
3026                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3027                             ((p->status != DIFF_STATUS_MODIFIED) &&
3028                              strchr(filter, p->status)))
3029                                 diff_q(&outq, p);
3030                         else
3031                                 diff_free_filepair(p);
3032                 }
3033         }
3034         free(q->queue);
3035         *q = outq;
3036 }
3037
3038 /* Check whether two filespecs with the same mode and size are identical */
3039 static int diff_filespec_is_identical(struct diff_filespec *one,
3040                                       struct diff_filespec *two)
3041 {
3042         if (S_ISGITLINK(one->mode)) {
3043                 diff_fill_sha1_info(one);
3044                 diff_fill_sha1_info(two);
3045                 return !hashcmp(one->sha1, two->sha1);
3046         }
3047         if (diff_populate_filespec(one, 0))
3048                 return 0;
3049         if (diff_populate_filespec(two, 0))
3050                 return 0;
3051         return !memcmp(one->data, two->data, one->size);
3052 }
3053
3054 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3055 {
3056         int i;
3057         struct diff_queue_struct *q = &diff_queued_diff;
3058         struct diff_queue_struct outq;
3059         outq.queue = NULL;
3060         outq.nr = outq.alloc = 0;
3061
3062         for (i = 0; i < q->nr; i++) {
3063                 struct diff_filepair *p = q->queue[i];
3064
3065                 /*
3066                  * 1. Entries that come from stat info dirtyness
3067                  *    always have both sides (iow, not create/delete),
3068                  *    one side of the object name is unknown, with
3069                  *    the same mode and size.  Keep the ones that
3070                  *    do not match these criteria.  They have real
3071                  *    differences.
3072                  *
3073                  * 2. At this point, the file is known to be modified,
3074                  *    with the same mode and size, and the object
3075                  *    name of one side is unknown.  Need to inspect
3076                  *    the identical contents.
3077                  */
3078                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3079                     !DIFF_FILE_VALID(p->two) ||
3080                     (p->one->sha1_valid && p->two->sha1_valid) ||
3081                     (p->one->mode != p->two->mode) ||
3082                     diff_populate_filespec(p->one, 1) ||
3083                     diff_populate_filespec(p->two, 1) ||
3084                     (p->one->size != p->two->size) ||
3085                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3086                         diff_q(&outq, p);
3087                 else {
3088                         /*
3089                          * The caller can subtract 1 from skip_stat_unmatch
3090                          * to determine how many paths were dirty only
3091                          * due to stat info mismatch.
3092                          */
3093                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3094                                 diffopt->skip_stat_unmatch++;
3095                         diff_free_filepair(p);
3096                 }
3097         }
3098         free(q->queue);
3099         *q = outq;
3100 }
3101
3102 void diffcore_std(struct diff_options *options)
3103 {
3104         if (DIFF_OPT_TST(options, QUIET))
3105                 return;
3106
3107         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3108                 diffcore_skip_stat_unmatch(options);
3109         if (options->break_opt != -1)
3110                 diffcore_break(options->break_opt);
3111         if (options->detect_rename)
3112                 diffcore_rename(options);
3113         if (options->break_opt != -1)
3114                 diffcore_merge_broken();
3115         if (options->pickaxe)
3116                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3117         if (options->orderfile)
3118                 diffcore_order(options->orderfile);
3119         diff_resolve_rename_copy();
3120         diffcore_apply_filter(options->filter);
3121
3122         if (diff_queued_diff.nr)
3123                 DIFF_OPT_SET(options, HAS_CHANGES);
3124         else
3125                 DIFF_OPT_CLR(options, HAS_CHANGES);
3126 }
3127
3128 int diff_result_code(struct diff_options *opt, int status)
3129 {
3130         int result = 0;
3131         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3132             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3133                 return status;
3134         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3135             DIFF_OPT_TST(opt, HAS_CHANGES))
3136                 result |= 01;
3137         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3138             DIFF_OPT_TST(opt, CHECK_FAILED))
3139                 result |= 02;
3140         return result;
3141 }
3142
3143 void diff_addremove(struct diff_options *options,
3144                     int addremove, unsigned mode,
3145                     const unsigned char *sha1,
3146                     const char *base, const char *path)
3147 {
3148         char concatpath[PATH_MAX];
3149         struct diff_filespec *one, *two;
3150
3151         /* This may look odd, but it is a preparation for
3152          * feeding "there are unchanged files which should
3153          * not produce diffs, but when you are doing copy
3154          * detection you would need them, so here they are"
3155          * entries to the diff-core.  They will be prefixed
3156          * with something like '=' or '*' (I haven't decided
3157          * which but should not make any difference).
3158          * Feeding the same new and old to diff_change()
3159          * also has the same effect.
3160          * Before the final output happens, they are pruned after
3161          * merged into rename/copy pairs as appropriate.
3162          */
3163         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3164                 addremove = (addremove == '+' ? '-' :
3165                              addremove == '-' ? '+' : addremove);
3166
3167         if (!path) path = "";
3168         sprintf(concatpath, "%s%s", base, path);
3169         one = alloc_filespec(concatpath);
3170         two = alloc_filespec(concatpath);
3171
3172         if (addremove != '+')
3173                 fill_filespec(one, sha1, mode);
3174         if (addremove != '-')
3175                 fill_filespec(two, sha1, mode);
3176
3177         diff_queue(&diff_queued_diff, one, two);
3178         DIFF_OPT_SET(options, HAS_CHANGES);
3179 }
3180
3181 void diff_change(struct diff_options *options,
3182                  unsigned old_mode, unsigned new_mode,
3183                  const unsigned char *old_sha1,
3184                  const unsigned char *new_sha1,
3185                  const char *base, const char *path)
3186 {
3187         char concatpath[PATH_MAX];
3188         struct diff_filespec *one, *two;
3189
3190         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3191                 unsigned tmp;
3192                 const unsigned char *tmp_c;
3193                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3194                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3195         }
3196         if (!path) path = "";
3197         sprintf(concatpath, "%s%s", base, path);
3198         one = alloc_filespec(concatpath);
3199         two = alloc_filespec(concatpath);
3200         fill_filespec(one, old_sha1, old_mode);
3201         fill_filespec(two, new_sha1, new_mode);
3202
3203         diff_queue(&diff_queued_diff, one, two);
3204         DIFF_OPT_SET(options, HAS_CHANGES);
3205 }
3206
3207 void diff_unmerge(struct diff_options *options,
3208                   const char *path,
3209                   unsigned mode, const unsigned char *sha1)
3210 {
3211         struct diff_filespec *one, *two;
3212         one = alloc_filespec(path);
3213         two = alloc_filespec(path);
3214         fill_filespec(one, sha1, mode);
3215         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3216 }