diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14 #include "userdiff.h"
  15 #include "sigchain.h"
  16 #include "submodule.h"
  17 #include "ll-merge.h"
  18
  19 #ifdef NO_FAST_WORKING_DIRECTORY
  20 #define FAST_WORKING_DIRECTORY 0
  21 #else
  22 #define FAST_WORKING_DIRECTORY 1
  23 #endif
  24
  25 static int diff_detect_rename_default;
  26 static int diff_rename_limit_default = 200;
  27 static int diff_suppress_blank_empty;
  28 int diff_use_color_default = -1;
  29 static const char *diff_word_regex_cfg;
  30 static const char *external_diff_cmd_cfg;
  31 int diff_auto_refresh_index = 1;
  32 static int diff_mnemonic_prefix;
  33 static int diff_no_prefix;
  34
  35 static char diff_colors[][COLOR_MAXLEN] = {
  36         GIT_COLOR_RESET,
  37         GIT_COLOR_NORMAL,       /* PLAIN */
  38         GIT_COLOR_BOLD,         /* METAINFO */
  39         GIT_COLOR_CYAN,         /* FRAGINFO */
  40         GIT_COLOR_RED,          /* OLD */
  41         GIT_COLOR_GREEN,        /* NEW */
  42         GIT_COLOR_YELLOW,       /* COMMIT */
  43         GIT_COLOR_BG_RED,       /* WHITESPACE */
  44         GIT_COLOR_NORMAL,       /* FUNCINFO */
  45 };
  46
  47 static int parse_diff_color_slot(const char *var, int ofs)
  48 {
  49         if (!strcasecmp(var+ofs, "plain"))
  50                 return DIFF_PLAIN;
  51         if (!strcasecmp(var+ofs, "meta"))
  52                 return DIFF_METAINFO;
  53         if (!strcasecmp(var+ofs, "frag"))
  54                 return DIFF_FRAGINFO;
  55         if (!strcasecmp(var+ofs, "old"))
  56                 return DIFF_FILE_OLD;
  57         if (!strcasecmp(var+ofs, "new"))
  58                 return DIFF_FILE_NEW;
  59         if (!strcasecmp(var+ofs, "commit"))
  60                 return DIFF_COMMIT;
  61         if (!strcasecmp(var+ofs, "whitespace"))
  62                 return DIFF_WHITESPACE;
  63         if (!strcasecmp(var+ofs, "func"))
  64                 return DIFF_FUNCINFO;
  65         return -1;
  66 }
  67
  68 static int git_config_rename(const char *var, const char *value)
  69 {
  70         if (!value)
  71                 return DIFF_DETECT_RENAME;
  72         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
  73                 return  DIFF_DETECT_COPY;
  74         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
  75 }
  76
  77 /*
  78  * These are to give UI layer defaults.
  79  * The core-level commands such as git-diff-files should
  80  * never be affected by the setting of diff.renames
  81  * the user happens to have in the configuration file.
  82  */
  83 int git_diff_ui_config(const char *var, const char *value, void *cb)
  84 {
  85         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
  86                 diff_use_color_default = git_config_colorbool(var, value, -1);
  87                 return 0;
  88         }
  89         if (!strcmp(var, "diff.renames")) {
  90                 diff_detect_rename_default = git_config_rename(var, value);
  91                 return 0;
  92         }
  93         if (!strcmp(var, "diff.autorefreshindex")) {
  94                 diff_auto_refresh_index = git_config_bool(var, value);
  95                 return 0;
  96         }
  97         if (!strcmp(var, "diff.mnemonicprefix")) {
  98                 diff_mnemonic_prefix = git_config_bool(var, value);
  99                 return 0;
 100         }
 101         if (!strcmp(var, "diff.noprefix")) {
 102                 diff_no_prefix = git_config_bool(var, value);
 103                 return 0;
 104         }
 105         if (!strcmp(var, "diff.external"))
 106                 return git_config_string(&external_diff_cmd_cfg, var, value);
 107         if (!strcmp(var, "diff.wordregex"))
 108                 return git_config_string(&diff_word_regex_cfg, var, value);
 109
 110         return git_diff_basic_config(var, value, cb);
 111 }
 112
 113 int git_diff_basic_config(const char *var, const char *value, void *cb)
 114 {
 115         if (!strcmp(var, "diff.renamelimit")) {
 116                 diff_rename_limit_default = git_config_int(var, value);
 117                 return 0;
 118         }
 119
 120         switch (userdiff_config(var, value)) {
 121                 case 0: break;
 122                 case -1: return -1;
 123                 default: return 0;
 124         }
 125
 126         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 127                 int slot = parse_diff_color_slot(var, 11);
 128                 if (slot < 0)
 129                         return 0;
 130                 if (!value)
 131                         return config_error_nonbool(var);
 132                 color_parse(value, var, diff_colors[slot]);
 133                 return 0;
 134         }
 135
 136         /* like GNU diff's --suppress-blank-empty option  */
 137         if (!strcmp(var, "diff.suppressblankempty") ||
 138                         /* for backwards compatibility */
 139                         !strcmp(var, "diff.suppress-blank-empty")) {
 140                 diff_suppress_blank_empty = git_config_bool(var, value);
 141                 return 0;
 142         }
 143
 144         return git_color_default_config(var, value, cb);
 145 }
 146
 147 static char *quote_two(const char *one, const char *two)
 148 {
 149         int need_one = quote_c_style(one, NULL, NULL, 1);
 150         int need_two = quote_c_style(two, NULL, NULL, 1);
 151         struct strbuf res = STRBUF_INIT;
 152
 153         if (need_one + need_two) {
 154                 strbuf_addch(&res, '"');
 155                 quote_c_style(one, &res, NULL, 1);
 156                 quote_c_style(two, &res, NULL, 1);
 157                 strbuf_addch(&res, '"');
 158         } else {
 159                 strbuf_addstr(&res, one);
 160                 strbuf_addstr(&res, two);
 161         }
 162         return strbuf_detach(&res, NULL);
 163 }
 164
 165 static const char *external_diff(void)
 166 {
 167         static const char *external_diff_cmd = NULL;
 168         static int done_preparing = 0;
 169
 170         if (done_preparing)
 171                 return external_diff_cmd;
 172         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 173         if (!external_diff_cmd)
 174                 external_diff_cmd = external_diff_cmd_cfg;
 175         done_preparing = 1;
 176         return external_diff_cmd;
 177 }
 178
 179 static struct diff_tempfile {
 180         const char *name; /* filename external diff should read from */
 181         char hex[41];
 182         char mode[10];
 183         char tmp_path[PATH_MAX];
 184 } diff_temp[2];
 185
 186 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 187
 188 struct emit_callback {
 189         int color_diff;
 190         unsigned ws_rule;
 191         int blank_at_eof_in_preimage;
 192         int blank_at_eof_in_postimage;
 193         int lno_in_preimage;
 194         int lno_in_postimage;
 195         sane_truncate_fn truncate;
 196         const char **label_path;
 197         struct diff_words_data *diff_words;
 198         struct diff_options *opt;
 199         int *found_changesp;
 200         struct strbuf *header;
 201 };
 202
 203 static int count_lines(const char *data, int size)
 204 {
 205         int count, ch, completely_empty = 1, nl_just_seen = 0;
 206         count = 0;
 207         while (0 < size--) {
 208                 ch = *data++;
 209                 if (ch == '\n') {
 210                         count++;
 211                         nl_just_seen = 1;
 212                         completely_empty = 0;
 213                 }
 214                 else {
 215                         nl_just_seen = 0;
 216                         completely_empty = 0;
 217                 }
 218         }
 219         if (completely_empty)
 220                 return 0;
 221         if (!nl_just_seen)
 222                 count++; /* no trailing newline */
 223         return count;
 224 }
 225
 226 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 227 {
 228         if (!DIFF_FILE_VALID(one)) {
 229                 mf->ptr = (char *)""; /* does not matter */
 230                 mf->size = 0;
 231                 return 0;
 232         }
 233         else if (diff_populate_filespec(one, 0))
 234                 return -1;
 235
 236         mf->ptr = one->data;
 237         mf->size = one->size;
 238         return 0;
 239 }
 240
 241 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 242 {
 243         char *ptr = mf->ptr;
 244         long size = mf->size;
 245         int cnt = 0;
 246
 247         if (!size)
 248                 return cnt;
 249         ptr += size - 1; /* pointing at the very end */
 250         if (*ptr != '\n')
 251                 ; /* incomplete line */
 252         else
 253                 ptr--; /* skip the last LF */
 254         while (mf->ptr < ptr) {
 255                 char *prev_eol;
 256                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 257                         if (*prev_eol == '\n')
 258                                 break;
 259                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 260                         break;
 261                 cnt++;
 262                 ptr = prev_eol - 1;
 263         }
 264         return cnt;
 265 }
 266
 267 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 268                                struct emit_callback *ecbdata)
 269 {
 270         int l1, l2, at;
 271         unsigned ws_rule = ecbdata->ws_rule;
 272         l1 = count_trailing_blank(mf1, ws_rule);
 273         l2 = count_trailing_blank(mf2, ws_rule);
 274         if (l2 <= l1) {
 275                 ecbdata->blank_at_eof_in_preimage = 0;
 276                 ecbdata->blank_at_eof_in_postimage = 0;
 277                 return;
 278         }
 279         at = count_lines(mf1->ptr, mf1->size);
 280         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 281
 282         at = count_lines(mf2->ptr, mf2->size);
 283         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 284 }
 285
 286 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 287                         int first, const char *line, int len)
 288 {
 289         int has_trailing_newline, has_trailing_carriage_return;
 290         int nofirst;
 291         FILE *file = o->file;
 292
 293         if (o->output_prefix) {
 294                 struct strbuf *msg = NULL;
 295                 msg = o->output_prefix(o, o->output_prefix_data);
 296                 assert(msg);
 297                 fwrite(msg->buf, msg->len, 1, file);
 298         }
 299
 300         if (len == 0) {
 301                 has_trailing_newline = (first == '\n');
 302                 has_trailing_carriage_return = (!has_trailing_newline &&
 303                                                 (first == '\r'));
 304                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 305         } else {
 306                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 307                 if (has_trailing_newline)
 308                         len--;
 309                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 310                 if (has_trailing_carriage_return)
 311                         len--;
 312                 nofirst = 0;
 313         }
 314
 315         if (len || !nofirst) {
 316                 fputs(set, file);
 317                 if (!nofirst)
 318                         fputc(first, file);
 319                 fwrite(line, len, 1, file);
 320                 fputs(reset, file);
 321         }
 322         if (has_trailing_carriage_return)
 323                 fputc('\r', file);
 324         if (has_trailing_newline)
 325                 fputc('\n', file);
 326 }
 327
 328 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 329                       const char *line, int len)
 330 {
 331         emit_line_0(o, set, reset, line[0], line+1, len-1);
 332 }
 333
 334 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 335 {
 336         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 337               ecbdata->blank_at_eof_in_preimage &&
 338               ecbdata->blank_at_eof_in_postimage &&
 339               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 340               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 341                 return 0;
 342         return ws_blank_line(line, len, ecbdata->ws_rule);
 343 }
 344
 345 static void emit_add_line(const char *reset,
 346                           struct emit_callback *ecbdata,
 347                           const char *line, int len)
 348 {
 349         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 350         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 351
 352         if (!*ws)
 353                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
 354         else if (new_blank_line_at_eof(ecbdata, line, len))
 355                 /* Blank line at EOF - paint '+' as well */
 356                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
 357         else {
 358                 /* Emit just the prefix, then the rest. */
 359                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
 360                 ws_check_emit(line, len, ecbdata->ws_rule,
 361                               ecbdata->opt->file, set, reset, ws);
 362         }
 363 }
 364
 365 static void emit_hunk_header(struct emit_callback *ecbdata,
 366                              const char *line, int len)
 367 {
 368         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 369         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 370         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 371         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 372         static const char atat[2] = { '@', '@' };
 373         const char *cp, *ep;
 374         struct strbuf msgbuf = STRBUF_INIT;
 375         int org_len = len;
 376         int i = 1;
 377
 378         /*
 379          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 380          * it always is at least 10 bytes long.
 381          */
 382         if (len < 10 ||
 383             memcmp(line, atat, 2) ||
 384             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 385                 emit_line(ecbdata->opt, plain, reset, line, len);
 386                 return;
 387         }
 388         ep += 2; /* skip over @@ */
 389
 390         /* The hunk header in fraginfo color */
 391         strbuf_add(&msgbuf, frag, strlen(frag));
 392         strbuf_add(&msgbuf, line, ep - line);
 393         strbuf_add(&msgbuf, reset, strlen(reset));
 394
 395         /*
 396          * trailing "\r\n"
 397          */
 398         for ( ; i < 3; i++)
 399                 if (line[len - i] == '\r' || line[len - i] == '\n')
 400                         len--;
 401
 402         /* blank before the func header */
 403         for (cp = ep; ep - line < len; ep++)
 404                 if (*ep != ' ' && *ep != '\t')
 405                         break;
 406         if (ep != cp) {
 407                 strbuf_add(&msgbuf, plain, strlen(plain));
 408                 strbuf_add(&msgbuf, cp, ep - cp);
 409                 strbuf_add(&msgbuf, reset, strlen(reset));
 410         }
 411
 412         if (ep < line + len) {
 413                 strbuf_add(&msgbuf, func, strlen(func));
 414                 strbuf_add(&msgbuf, ep, line + len - ep);
 415                 strbuf_add(&msgbuf, reset, strlen(reset));
 416         }
 417
 418         strbuf_add(&msgbuf, line + len, org_len - len);
 419         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 420         strbuf_release(&msgbuf);
 421 }
 422
 423 static struct diff_tempfile *claim_diff_tempfile(void) {
 424         int i;
 425         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 426                 if (!diff_temp[i].name)
 427                         return diff_temp + i;
 428         die("BUG: diff is failing to clean up its tempfiles");
 429 }
 430
 431 static int remove_tempfile_installed;
 432
 433 static void remove_tempfile(void)
 434 {
 435         int i;
 436         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 437                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 438                         unlink_or_warn(diff_temp[i].name);
 439                 diff_temp[i].name = NULL;
 440         }
 441 }
 442
 443 static void remove_tempfile_on_signal(int signo)
 444 {
 445         remove_tempfile();
 446         sigchain_pop(signo);
 447         raise(signo);
 448 }
 449
 450 static void print_line_count(FILE *file, int count)
 451 {
 452         switch (count) {
 453         case 0:
 454                 fprintf(file, "0,0");
 455                 break;
 456         case 1:
 457                 fprintf(file, "1");
 458                 break;
 459         default:
 460                 fprintf(file, "1,%d", count);
 461                 break;
 462         }
 463 }
 464
 465 static void emit_rewrite_lines(struct emit_callback *ecb,
 466                                int prefix, const char *data, int size)
 467 {
 468         const char *endp = NULL;
 469         static const char *nneof = " No newline at end of file\n";
 470         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 471         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 472
 473         while (0 < size) {
 474                 int len;
 475
 476                 endp = memchr(data, '\n', size);
 477                 len = endp ? (endp - data + 1) : size;
 478                 if (prefix != '+') {
 479                         ecb->lno_in_preimage++;
 480                         emit_line_0(ecb->opt, old, reset, '-',
 481                                     data, len);
 482                 } else {
 483                         ecb->lno_in_postimage++;
 484                         emit_add_line(reset, ecb, data, len);
 485                 }
 486                 size -= len;
 487                 data += len;
 488         }
 489         if (!endp) {
 490                 const char *plain = diff_get_color(ecb->color_diff,
 491                                                    DIFF_PLAIN);
 492                 emit_line_0(ecb->opt, plain, reset, '\\',
 493                             nneof, strlen(nneof));
 494         }
 495 }
 496
 497 static void emit_rewrite_diff(const char *name_a,
 498                               const char *name_b,
 499                               struct diff_filespec *one,
 500                               struct diff_filespec *two,
 501                               struct userdiff_driver *textconv_one,
 502                               struct userdiff_driver *textconv_two,
 503                               struct diff_options *o)
 504 {
 505         int lc_a, lc_b;
 506         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 507         const char *name_a_tab, *name_b_tab;
 508         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 509         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 510         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 511         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 512         const char *a_prefix, *b_prefix;
 513         char *data_one, *data_two;
 514         size_t size_one, size_two;
 515         struct emit_callback ecbdata;
 516         char *line_prefix = "";
 517         struct strbuf *msgbuf;
 518
 519         if (o && o->output_prefix) {
 520                 msgbuf = o->output_prefix(o, o->output_prefix_data);
 521                 line_prefix = msgbuf->buf;
 522         }
 523
 524         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 525                 a_prefix = o->b_prefix;
 526                 b_prefix = o->a_prefix;
 527         } else {
 528                 a_prefix = o->a_prefix;
 529                 b_prefix = o->b_prefix;
 530         }
 531
 532         name_a += (*name_a == '/');
 533         name_b += (*name_b == '/');
 534         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 535         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 536
 537         strbuf_reset(&a_name);
 538         strbuf_reset(&b_name);
 539         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 540         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 541
 542         size_one = fill_textconv(textconv_one, one, &data_one);
 543         size_two = fill_textconv(textconv_two, two, &data_two);
 544
 545         memset(&ecbdata, 0, sizeof(ecbdata));
 546         ecbdata.color_diff = color_diff;
 547         ecbdata.found_changesp = &o->found_changes;
 548         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 549         ecbdata.opt = o;
 550         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 551                 mmfile_t mf1, mf2;
 552                 mf1.ptr = (char *)data_one;
 553                 mf2.ptr = (char *)data_two;
 554                 mf1.size = size_one;
 555                 mf2.size = size_two;
 556                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 557         }
 558         ecbdata.lno_in_preimage = 1;
 559         ecbdata.lno_in_postimage = 1;
 560
 561         lc_a = count_lines(data_one, size_one);
 562         lc_b = count_lines(data_two, size_two);
 563         fprintf(o->file,
 564                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 565                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 566                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 567                 line_prefix, fraginfo);
 568         print_line_count(o->file, lc_a);
 569         fprintf(o->file, " +");
 570         print_line_count(o->file, lc_b);
 571         fprintf(o->file, " @@%s\n", reset);
 572         if (lc_a)
 573                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 574         if (lc_b)
 575                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 576         if (textconv_one)
 577                 free((char *)data_one);
 578         if (textconv_two)
 579                 free((char *)data_two);
 580 }
 581
 582 struct diff_words_buffer {
 583         mmfile_t text;
 584         long alloc;
 585         struct diff_words_orig {
 586                 const char *begin, *end;
 587         } *orig;
 588         int orig_nr, orig_alloc;
 589 };
 590
 591 static void diff_words_append(char *line, unsigned long len,
 592                 struct diff_words_buffer *buffer)
 593 {
 594         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 595         line++;
 596         len--;
 597         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 598         buffer->text.size += len;
 599         buffer->text.ptr[buffer->text.size] = '\0';
 600 }
 601
 602 struct diff_words_style_elem
 603 {
 604         const char *prefix;
 605         const char *suffix;
 606         const char *color; /* NULL; filled in by the setup code if
 607                             * color is enabled */
 608 };
 609
 610 struct diff_words_style
 611 {
 612         enum diff_words_type type;
 613         struct diff_words_style_elem new, old, ctx;
 614         const char *newline;
 615 };
 616
 617 struct diff_words_style diff_words_styles[] = {
 618         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 619         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 620         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 621 };
 622
 623 struct diff_words_data {
 624         struct diff_words_buffer minus, plus;
 625         const char *current_plus;
 626         int last_minus;
 627         struct diff_options *opt;
 628         regex_t *word_regex;
 629         enum diff_words_type type;
 630         struct diff_words_style *style;
 631 };
 632
 633 static int fn_out_diff_words_write_helper(FILE *fp,
 634                                           struct diff_words_style_elem *st_el,
 635                                           const char *newline,
 636                                           size_t count, const char *buf,
 637                                           const char *line_prefix)
 638 {
 639         int print = 0;
 640
 641         while (count) {
 642                 char *p = memchr(buf, '\n', count);
 643                 if (print)
 644                         fputs(line_prefix, fp);
 645                 if (p != buf) {
 646                         if (st_el->color && fputs(st_el->color, fp) < 0)
 647                                 return -1;
 648                         if (fputs(st_el->prefix, fp) < 0 ||
 649                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 650                             fputs(st_el->suffix, fp) < 0)
 651                                 return -1;
 652                         if (st_el->color && *st_el->color
 653                             && fputs(GIT_COLOR_RESET, fp) < 0)
 654                                 return -1;
 655                 }
 656                 if (!p)
 657                         return 0;
 658                 if (fputs(newline, fp) < 0)
 659                         return -1;
 660                 count -= p + 1 - buf;
 661                 buf = p + 1;
 662                 print = 1;
 663         }
 664         return 0;
 665 }
 666
 667 /*
 668  * '--color-words' algorithm can be described as:
 669  *
 670  *   1. collect a the minus/plus lines of a diff hunk, divided into
 671  *      minus-lines and plus-lines;
 672  *
 673  *   2. break both minus-lines and plus-lines into words and
 674  *      place them into two mmfile_t with one word for each line;
 675  *
 676  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 677  *
 678  * And for the common parts of the both file, we output the plus side text.
 679  * diff_words->current_plus is used to trace the current position of the plus file
 680  * which printed. diff_words->last_minus is used to trace the last minus word
 681  * printed.
 682  *
 683  * For '--graph' to work with '--color-words', we need to output the graph prefix
 684  * on each line of color words output. Generally, there are two conditions on
 685  * which we should output the prefix.
 686  *
 687  *   1. diff_words->last_minus == 0 &&
 688  *      diff_words->current_plus == diff_words->plus.text.ptr
 689  *
 690  *      that is: the plus text must start as a new line, and if there is no minus
 691  *      word printed, a graph prefix must be printed.
 692  *
 693  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 694  *      *(diff_words->current_plus - 1) == '\n'
 695  *
 696  *      that is: a graph prefix must be printed following a '\n'
 697  */
 698 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 699 {
 700         if ((diff_words->last_minus == 0 &&
 701                 diff_words->current_plus == diff_words->plus.text.ptr) ||
 702                 (diff_words->current_plus > diff_words->plus.text.ptr &&
 703                 *(diff_words->current_plus - 1) == '\n')) {
 704                 return 1;
 705         } else {
 706                 return 0;
 707         }
 708 }
 709
 710 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 711 {
 712         struct diff_words_data *diff_words = priv;
 713         struct diff_words_style *style = diff_words->style;
 714         int minus_first, minus_len, plus_first, plus_len;
 715         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 716         struct diff_options *opt = diff_words->opt;
 717         struct strbuf *msgbuf;
 718         char *line_prefix = "";
 719
 720         if (line[0] != '@' || parse_hunk_header(line, len,
 721                         &minus_first, &minus_len, &plus_first, &plus_len))
 722                 return;
 723
 724         assert(opt);
 725         if (opt->output_prefix) {
 726                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 727                 line_prefix = msgbuf->buf;
 728         }
 729
 730         /* POSIX requires that first be decremented by one if len == 0... */
 731         if (minus_len) {
 732                 minus_begin = diff_words->minus.orig[minus_first].begin;
 733                 minus_end =
 734                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 735         } else
 736                 minus_begin = minus_end =
 737                         diff_words->minus.orig[minus_first].end;
 738
 739         if (plus_len) {
 740                 plus_begin = diff_words->plus.orig[plus_first].begin;
 741                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 742         } else
 743                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 744
 745         if (color_words_output_graph_prefix(diff_words)) {
 746                 fputs(line_prefix, diff_words->opt->file);
 747         }
 748         if (diff_words->current_plus != plus_begin) {
 749                 fn_out_diff_words_write_helper(diff_words->opt->file,
 750                                 &style->ctx, style->newline,
 751                                 plus_begin - diff_words->current_plus,
 752                                 diff_words->current_plus, line_prefix);
 753                 if (*(plus_begin - 1) == '\n')
 754                         fputs(line_prefix, diff_words->opt->file);
 755         }
 756         if (minus_begin != minus_end) {
 757                 fn_out_diff_words_write_helper(diff_words->opt->file,
 758                                 &style->old, style->newline,
 759                                 minus_end - minus_begin, minus_begin,
 760                                 line_prefix);
 761         }
 762         if (plus_begin != plus_end) {
 763                 fn_out_diff_words_write_helper(diff_words->opt->file,
 764                                 &style->new, style->newline,
 765                                 plus_end - plus_begin, plus_begin,
 766                                 line_prefix);
 767         }
 768
 769         diff_words->current_plus = plus_end;
 770         diff_words->last_minus = minus_first;
 771 }
 772
 773 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 774 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 775                 int *begin, int *end)
 776 {
 777         if (word_regex && *begin < buffer->size) {
 778                 regmatch_t match[1];
 779                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 780                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 781                                         '\n', match[0].rm_eo - match[0].rm_so);
 782                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 783                         *begin += match[0].rm_so;
 784                         return *begin >= *end;
 785                 }
 786                 return -1;
 787         }
 788
 789         /* find the next word */
 790         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 791                 (*begin)++;
 792         if (*begin >= buffer->size)
 793                 return -1;
 794
 795         /* find the end of the word */
 796         *end = *begin + 1;
 797         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 798                 (*end)++;
 799
 800         return 0;
 801 }
 802
 803 /*
 804  * This function splits the words in buffer->text, stores the list with
 805  * newline separator into out, and saves the offsets of the original words
 806  * in buffer->orig.
 807  */
 808 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 809                 regex_t *word_regex)
 810 {
 811         int i, j;
 812         long alloc = 0;
 813
 814         out->size = 0;
 815         out->ptr = NULL;
 816
 817         /* fake an empty "0th" word */
 818         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 819         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 820         buffer->orig_nr = 1;
 821
 822         for (i = 0; i < buffer->text.size; i++) {
 823                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 824                         return;
 825
 826                 /* store original boundaries */
 827                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 828                                 buffer->orig_alloc);
 829                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 830                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 831                 buffer->orig_nr++;
 832
 833                 /* store one word */
 834                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 835                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 836                 out->ptr[out->size + j - i] = '\n';
 837                 out->size += j - i + 1;
 838
 839                 i = j - 1;
 840         }
 841 }
 842
 843 /* this executes the word diff on the accumulated buffers */
 844 static void diff_words_show(struct diff_words_data *diff_words)
 845 {
 846         xpparam_t xpp;
 847         xdemitconf_t xecfg;
 848         mmfile_t minus, plus;
 849         struct diff_words_style *style = diff_words->style;
 850
 851         struct diff_options *opt = diff_words->opt;
 852         struct strbuf *msgbuf;
 853         char *line_prefix = "";
 854
 855         assert(opt);
 856         if (opt->output_prefix) {
 857                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 858                 line_prefix = msgbuf->buf;
 859         }
 860
 861         /* special case: only removal */
 862         if (!diff_words->plus.text.size) {
 863                 fputs(line_prefix, diff_words->opt->file);
 864                 fn_out_diff_words_write_helper(diff_words->opt->file,
 865                         &style->old, style->newline,
 866                         diff_words->minus.text.size,
 867                         diff_words->minus.text.ptr, line_prefix);
 868                 diff_words->minus.text.size = 0;
 869                 return;
 870         }
 871
 872         diff_words->current_plus = diff_words->plus.text.ptr;
 873         diff_words->last_minus = 0;
 874
 875         memset(&xpp, 0, sizeof(xpp));
 876         memset(&xecfg, 0, sizeof(xecfg));
 877         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 878         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 879         xpp.flags = 0;
 880         /* as only the hunk header will be parsed, we need a 0-context */
 881         xecfg.ctxlen = 0;
 882         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 883                       &xpp, &xecfg);
 884         free(minus.ptr);
 885         free(plus.ptr);
 886         if (diff_words->current_plus != diff_words->plus.text.ptr +
 887                         diff_words->plus.text.size) {
 888                 if (color_words_output_graph_prefix(diff_words))
 889                         fputs(line_prefix, diff_words->opt->file);
 890                 fn_out_diff_words_write_helper(diff_words->opt->file,
 891                         &style->ctx, style->newline,
 892                         diff_words->plus.text.ptr + diff_words->plus.text.size
 893                         - diff_words->current_plus, diff_words->current_plus,
 894                         line_prefix);
 895         }
 896         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 897 }
 898
 899 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
 900 static void diff_words_flush(struct emit_callback *ecbdata)
 901 {
 902         if (ecbdata->diff_words->minus.text.size ||
 903             ecbdata->diff_words->plus.text.size)
 904                 diff_words_show(ecbdata->diff_words);
 905 }
 906
 907 static void free_diff_words_data(struct emit_callback *ecbdata)
 908 {
 909         if (ecbdata->diff_words) {
 910                 diff_words_flush(ecbdata);
 911                 free (ecbdata->diff_words->minus.text.ptr);
 912                 free (ecbdata->diff_words->minus.orig);
 913                 free (ecbdata->diff_words->plus.text.ptr);
 914                 free (ecbdata->diff_words->plus.orig);
 915                 free(ecbdata->diff_words->word_regex);
 916                 free(ecbdata->diff_words);
 917                 ecbdata->diff_words = NULL;
 918         }
 919 }
 920
 921 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 922 {
 923         if (diff_use_color)
 924                 return diff_colors[ix];
 925         return "";
 926 }
 927
 928 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 929 {
 930         const char *cp;
 931         unsigned long allot;
 932         size_t l = len;
 933
 934         if (ecb->truncate)
 935                 return ecb->truncate(line, len);
 936         cp = line;
 937         allot = l;
 938         while (0 < l) {
 939                 (void) utf8_width(&cp, &l);
 940                 if (!cp)
 941                         break; /* truncated in the middle? */
 942         }
 943         return allot - l;
 944 }
 945
 946 static void find_lno(const char *line, struct emit_callback *ecbdata)
 947 {
 948         const char *p;
 949         ecbdata->lno_in_preimage = 0;
 950         ecbdata->lno_in_postimage = 0;
 951         p = strchr(line, '-');
 952         if (!p)
 953                 return; /* cannot happen */
 954         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
 955         p = strchr(p, '+');
 956         if (!p)
 957                 return; /* cannot happen */
 958         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
 959 }
 960
 961 static void fn_out_consume(void *priv, char *line, unsigned long len)
 962 {
 963         struct emit_callback *ecbdata = priv;
 964         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 965         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 966         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 967         struct diff_options *o = ecbdata->opt;
 968         char *line_prefix = "";
 969         struct strbuf *msgbuf;
 970
 971         if (o && o->output_prefix) {
 972                 msgbuf = o->output_prefix(o, o->output_prefix_data);
 973                 line_prefix = msgbuf->buf;
 974         }
 975
 976         if (ecbdata->header) {
 977                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
 978                 strbuf_reset(ecbdata->header);
 979                 ecbdata->header = NULL;
 980         }
 981         *(ecbdata->found_changesp) = 1;
 982
 983         if (ecbdata->label_path[0]) {
 984                 const char *name_a_tab, *name_b_tab;
 985
 986                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 987                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 988
 989                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
 990                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
 991                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
 992                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
 993                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 994         }
 995
 996         if (diff_suppress_blank_empty
 997             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 998                 line[0] = '\n';
 999                 len = 1;
1000         }
1001
1002         if (line[0] == '@') {
1003                 if (ecbdata->diff_words)
1004                         diff_words_flush(ecbdata);
1005                 len = sane_truncate_line(ecbdata, line, len);
1006                 find_lno(line, ecbdata);
1007                 emit_hunk_header(ecbdata, line, len);
1008                 if (line[len-1] != '\n')
1009                         putc('\n', ecbdata->opt->file);
1010                 return;
1011         }
1012
1013         if (len < 1) {
1014                 emit_line(ecbdata->opt, reset, reset, line, len);
1015                 if (ecbdata->diff_words
1016                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1017                         fputs("~\n", ecbdata->opt->file);
1018                 return;
1019         }
1020
1021         if (ecbdata->diff_words) {
1022                 if (line[0] == '-') {
1023                         diff_words_append(line, len,
1024                                           &ecbdata->diff_words->minus);
1025                         return;
1026                 } else if (line[0] == '+') {
1027                         diff_words_append(line, len,
1028                                           &ecbdata->diff_words->plus);
1029                         return;
1030                 }
1031                 diff_words_flush(ecbdata);
1032                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1033                         emit_line(ecbdata->opt, plain, reset, line, len);
1034                         fputs("~\n", ecbdata->opt->file);
1035                 } else {
1036                         /* don't print the prefix character */
1037                         emit_line(ecbdata->opt, plain, reset, line+1, len-1);
1038                 }
1039                 return;
1040         }
1041
1042         if (line[0] != '+') {
1043                 const char *color =
1044                         diff_get_color(ecbdata->color_diff,
1045                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1046                 ecbdata->lno_in_preimage++;
1047                 if (line[0] == ' ')
1048                         ecbdata->lno_in_postimage++;
1049                 emit_line(ecbdata->opt, color, reset, line, len);
1050         } else {
1051                 ecbdata->lno_in_postimage++;
1052                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1053         }
1054 }
1055
1056 static char *pprint_rename(const char *a, const char *b)
1057 {
1058         const char *old = a;
1059         const char *new = b;
1060         struct strbuf name = STRBUF_INIT;
1061         int pfx_length, sfx_length;
1062         int len_a = strlen(a);
1063         int len_b = strlen(b);
1064         int a_midlen, b_midlen;
1065         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1066         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1067
1068         if (qlen_a || qlen_b) {
1069                 quote_c_style(a, &name, NULL, 0);
1070                 strbuf_addstr(&name, " => ");
1071                 quote_c_style(b, &name, NULL, 0);
1072                 return strbuf_detach(&name, NULL);
1073         }
1074
1075         /* Find common prefix */
1076         pfx_length = 0;
1077         while (*old && *new && *old == *new) {
1078                 if (*old == '/')
1079                         pfx_length = old - a + 1;
1080                 old++;
1081                 new++;
1082         }
1083
1084         /* Find common suffix */
1085         old = a + len_a;
1086         new = b + len_b;
1087         sfx_length = 0;
1088         while (a <= old && b <= new && *old == *new) {
1089                 if (*old == '/')
1090                         sfx_length = len_a - (old - a);
1091                 old--;
1092                 new--;
1093         }
1094
1095         /*
1096          * pfx{mid-a => mid-b}sfx
1097          * {pfx-a => pfx-b}sfx
1098          * pfx{sfx-a => sfx-b}
1099          * name-a => name-b
1100          */
1101         a_midlen = len_a - pfx_length - sfx_length;
1102         b_midlen = len_b - pfx_length - sfx_length;
1103         if (a_midlen < 0)
1104                 a_midlen = 0;
1105         if (b_midlen < 0)
1106                 b_midlen = 0;
1107
1108         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1109         if (pfx_length + sfx_length) {
1110                 strbuf_add(&name, a, pfx_length);
1111                 strbuf_addch(&name, '{');
1112         }
1113         strbuf_add(&name, a + pfx_length, a_midlen);
1114         strbuf_addstr(&name, " => ");
1115         strbuf_add(&name, b + pfx_length, b_midlen);
1116         if (pfx_length + sfx_length) {
1117                 strbuf_addch(&name, '}');
1118                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1119         }
1120         return strbuf_detach(&name, NULL);
1121 }
1122
1123 struct diffstat_t {
1124         int nr;
1125         int alloc;
1126         struct diffstat_file {
1127                 char *from_name;
1128                 char *name;
1129                 char *print_name;
1130                 unsigned is_unmerged:1;
1131                 unsigned is_binary:1;
1132                 unsigned is_renamed:1;
1133                 uintmax_t added, deleted;
1134         } **files;
1135 };
1136
1137 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1138                                           const char *name_a,
1139                                           const char *name_b)
1140 {
1141         struct diffstat_file *x;
1142         x = xcalloc(sizeof (*x), 1);
1143         if (diffstat->nr == diffstat->alloc) {
1144                 diffstat->alloc = alloc_nr(diffstat->alloc);
1145                 diffstat->files = xrealloc(diffstat->files,
1146                                 diffstat->alloc * sizeof(x));
1147         }
1148         diffstat->files[diffstat->nr++] = x;
1149         if (name_b) {
1150                 x->from_name = xstrdup(name_a);
1151                 x->name = xstrdup(name_b);
1152                 x->is_renamed = 1;
1153         }
1154         else {
1155                 x->from_name = NULL;
1156                 x->name = xstrdup(name_a);
1157         }
1158         return x;
1159 }
1160
1161 static void diffstat_consume(void *priv, char *line, unsigned long len)
1162 {
1163         struct diffstat_t *diffstat = priv;
1164         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1165
1166         if (line[0] == '+')
1167                 x->added++;
1168         else if (line[0] == '-')
1169                 x->deleted++;
1170 }
1171
1172 const char mime_boundary_leader[] = "------------";
1173
1174 static int scale_linear(int it, int width, int max_change)
1175 {
1176         /*
1177          * make sure that at least one '-' is printed if there were deletions,
1178          * and likewise for '+'.
1179          */
1180         if (max_change < 2)
1181                 return it;
1182         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1183 }
1184
1185 static void show_name(FILE *file,
1186                       const char *prefix, const char *name, int len)
1187 {
1188         fprintf(file, " %s%-*s |", prefix, len, name);
1189 }
1190
1191 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1192 {
1193         if (cnt <= 0)
1194                 return;
1195         fprintf(file, "%s", set);
1196         while (cnt--)
1197                 putc(ch, file);
1198         fprintf(file, "%s", reset);
1199 }
1200
1201 static void fill_print_name(struct diffstat_file *file)
1202 {
1203         char *pname;
1204
1205         if (file->print_name)
1206                 return;
1207
1208         if (!file->is_renamed) {
1209                 struct strbuf buf = STRBUF_INIT;
1210                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1211                         pname = strbuf_detach(&buf, NULL);
1212                 } else {
1213                         pname = file->name;
1214                         strbuf_release(&buf);
1215                 }
1216         } else {
1217                 pname = pprint_rename(file->from_name, file->name);
1218         }
1219         file->print_name = pname;
1220 }
1221
1222 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1223 {
1224         int i, len, add, del, adds = 0, dels = 0;
1225         uintmax_t max_change = 0, max_len = 0;
1226         int total_files = data->nr;
1227         int width, name_width;
1228         const char *reset, *set, *add_c, *del_c;
1229         const char *line_prefix = "";
1230         struct strbuf *msg = NULL;
1231
1232         if (data->nr == 0)
1233                 return;
1234
1235         if (options->output_prefix) {
1236                 msg = options->output_prefix(options, options->output_prefix_data);
1237                 line_prefix = msg->buf;
1238         }
1239
1240         width = options->stat_width ? options->stat_width : 80;
1241         name_width = options->stat_name_width ? options->stat_name_width : 50;
1242
1243         /* Sanity: give at least 5 columns to the graph,
1244          * but leave at least 10 columns for the name.
1245          */
1246         if (width < 25)
1247                 width = 25;
1248         if (name_width < 10)
1249                 name_width = 10;
1250         else if (width < name_width + 15)
1251                 name_width = width - 15;
1252
1253         /* Find the longest filename and max number of changes */
1254         reset = diff_get_color_opt(options, DIFF_RESET);
1255         set   = diff_get_color_opt(options, DIFF_PLAIN);
1256         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1257         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1258
1259         for (i = 0; i < data->nr; i++) {
1260                 struct diffstat_file *file = data->files[i];
1261                 uintmax_t change = file->added + file->deleted;
1262                 fill_print_name(file);
1263                 len = strlen(file->print_name);
1264                 if (max_len < len)
1265                         max_len = len;
1266
1267                 if (file->is_binary || file->is_unmerged)
1268                         continue;
1269                 if (max_change < change)
1270                         max_change = change;
1271         }
1272
1273         /* Compute the width of the graph part;
1274          * 10 is for one blank at the beginning of the line plus
1275          * " | count " between the name and the graph.
1276          *
1277          * From here on, name_width is the width of the name area,
1278          * and width is the width of the graph area.
1279          */
1280         name_width = (name_width < max_len) ? name_width : max_len;
1281         if (width < (name_width + 10) + max_change)
1282                 width = width - (name_width + 10);
1283         else
1284                 width = max_change;
1285
1286         for (i = 0; i < data->nr; i++) {
1287                 const char *prefix = "";
1288                 char *name = data->files[i]->print_name;
1289                 uintmax_t added = data->files[i]->added;
1290                 uintmax_t deleted = data->files[i]->deleted;
1291                 int name_len;
1292
1293                 /*
1294                  * "scale" the filename
1295                  */
1296                 len = name_width;
1297                 name_len = strlen(name);
1298                 if (name_width < name_len) {
1299                         char *slash;
1300                         prefix = "...";
1301                         len -= 3;
1302                         name += name_len - len;
1303                         slash = strchr(name, '/');
1304                         if (slash)
1305                                 name = slash;
1306                 }
1307
1308                 if (data->files[i]->is_binary) {
1309                         fprintf(options->file, "%s", line_prefix);
1310                         show_name(options->file, prefix, name, len);
1311                         fprintf(options->file, "  Bin ");
1312                         fprintf(options->file, "%s%"PRIuMAX"%s",
1313                                 del_c, deleted, reset);
1314                         fprintf(options->file, " -> ");
1315                         fprintf(options->file, "%s%"PRIuMAX"%s",
1316                                 add_c, added, reset);
1317                         fprintf(options->file, " bytes");
1318                         fprintf(options->file, "\n");
1319                         continue;
1320                 }
1321                 else if (data->files[i]->is_unmerged) {
1322                         fprintf(options->file, "%s", line_prefix);
1323                         show_name(options->file, prefix, name, len);
1324                         fprintf(options->file, "  Unmerged\n");
1325                         continue;
1326                 }
1327                 else if (!data->files[i]->is_renamed &&
1328                          (added + deleted == 0)) {
1329                         total_files--;
1330                         continue;
1331                 }
1332
1333                 /*
1334                  * scale the add/delete
1335                  */
1336                 add = added;
1337                 del = deleted;
1338                 adds += add;
1339                 dels += del;
1340
1341                 if (width <= max_change) {
1342                         add = scale_linear(add, width, max_change);
1343                         del = scale_linear(del, width, max_change);
1344                 }
1345                 fprintf(options->file, "%s", line_prefix);
1346                 show_name(options->file, prefix, name, len);
1347                 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1348                                 added + deleted ? " " : "");
1349                 show_graph(options->file, '+', add, add_c, reset);
1350                 show_graph(options->file, '-', del, del_c, reset);
1351                 fprintf(options->file, "\n");
1352         }
1353         fprintf(options->file, "%s", line_prefix);
1354         fprintf(options->file,
1355                " %d files changed, %d insertions(+), %d deletions(-)\n",
1356                total_files, adds, dels);
1357 }
1358
1359 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1360 {
1361         int i, adds = 0, dels = 0, total_files = data->nr;
1362
1363         if (data->nr == 0)
1364                 return;
1365
1366         for (i = 0; i < data->nr; i++) {
1367                 if (!data->files[i]->is_binary &&
1368                     !data->files[i]->is_unmerged) {
1369                         int added = data->files[i]->added;
1370                         int deleted= data->files[i]->deleted;
1371                         if (!data->files[i]->is_renamed &&
1372                             (added + deleted == 0)) {
1373                                 total_files--;
1374                         } else {
1375                                 adds += added;
1376                                 dels += deleted;
1377                         }
1378                 }
1379         }
1380         if (options->output_prefix) {
1381                 struct strbuf *msg = NULL;
1382                 msg = options->output_prefix(options,
1383                                 options->output_prefix_data);
1384                 fprintf(options->file, "%s", msg->buf);
1385         }
1386         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1387                total_files, adds, dels);
1388 }
1389
1390 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1391 {
1392         int i;
1393
1394         if (data->nr == 0)
1395                 return;
1396
1397         for (i = 0; i < data->nr; i++) {
1398                 struct diffstat_file *file = data->files[i];
1399
1400                 if (options->output_prefix) {
1401                         struct strbuf *msg = NULL;
1402                         msg = options->output_prefix(options,
1403                                         options->output_prefix_data);
1404                         fprintf(options->file, "%s", msg->buf);
1405                 }
1406
1407                 if (file->is_binary)
1408                         fprintf(options->file, "-\t-\t");
1409                 else
1410                         fprintf(options->file,
1411                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1412                                 file->added, file->deleted);
1413                 if (options->line_termination) {
1414                         fill_print_name(file);
1415                         if (!file->is_renamed)
1416                                 write_name_quoted(file->name, options->file,
1417                                                   options->line_termination);
1418                         else {
1419                                 fputs(file->print_name, options->file);
1420                                 putc(options->line_termination, options->file);
1421                         }
1422                 } else {
1423                         if (file->is_renamed) {
1424                                 putc('\0', options->file);
1425                                 write_name_quoted(file->from_name, options->file, '\0');
1426                         }
1427                         write_name_quoted(file->name, options->file, '\0');
1428                 }
1429         }
1430 }
1431
1432 struct dirstat_file {
1433         const char *name;
1434         unsigned long changed;
1435 };
1436
1437 struct dirstat_dir {
1438         struct dirstat_file *files;
1439         int alloc, nr, percent, cumulative;
1440 };
1441
1442 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1443                 unsigned long changed, const char *base, int baselen)
1444 {
1445         unsigned long this_dir = 0;
1446         unsigned int sources = 0;
1447         const char *line_prefix = "";
1448         struct strbuf *msg = NULL;
1449
1450         if (opt->output_prefix) {
1451                 msg = opt->output_prefix(opt, opt->output_prefix_data);
1452                 line_prefix = msg->buf;
1453         }
1454
1455         while (dir->nr) {
1456                 struct dirstat_file *f = dir->files;
1457                 int namelen = strlen(f->name);
1458                 unsigned long this;
1459                 char *slash;
1460
1461                 if (namelen < baselen)
1462                         break;
1463                 if (memcmp(f->name, base, baselen))
1464                         break;
1465                 slash = strchr(f->name + baselen, '/');
1466                 if (slash) {
1467                         int newbaselen = slash + 1 - f->name;
1468                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1469                         sources++;
1470                 } else {
1471                         this = f->changed;
1472                         dir->files++;
1473                         dir->nr--;
1474                         sources += 2;
1475                 }
1476                 this_dir += this;
1477         }
1478
1479         /*
1480          * We don't report dirstat's for
1481          *  - the top level
1482          *  - or cases where everything came from a single directory
1483          *    under this directory (sources == 1).
1484          */
1485         if (baselen && sources != 1) {
1486                 int permille = this_dir * 1000 / changed;
1487                 if (permille) {
1488                         int percent = permille / 10;
1489                         if (percent >= dir->percent) {
1490                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1491                                         percent, permille % 10, baselen, base);
1492                                 if (!dir->cumulative)
1493                                         return 0;
1494                         }
1495                 }
1496         }
1497         return this_dir;
1498 }
1499
1500 static int dirstat_compare(const void *_a, const void *_b)
1501 {
1502         const struct dirstat_file *a = _a;
1503         const struct dirstat_file *b = _b;
1504         return strcmp(a->name, b->name);
1505 }
1506
1507 static void show_dirstat(struct diff_options *options)
1508 {
1509         int i;
1510         unsigned long changed;
1511         struct dirstat_dir dir;
1512         struct diff_queue_struct *q = &diff_queued_diff;
1513
1514         dir.files = NULL;
1515         dir.alloc = 0;
1516         dir.nr = 0;
1517         dir.percent = options->dirstat_percent;
1518         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1519
1520         changed = 0;
1521         for (i = 0; i < q->nr; i++) {
1522                 struct diff_filepair *p = q->queue[i];
1523                 const char *name;
1524                 unsigned long copied, added, damage;
1525
1526                 name = p->one->path ? p->one->path : p->two->path;
1527
1528                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1529                         diff_populate_filespec(p->one, 0);
1530                         diff_populate_filespec(p->two, 0);
1531                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1532                                                &copied, &added);
1533                         diff_free_filespec_data(p->one);
1534                         diff_free_filespec_data(p->two);
1535                 } else if (DIFF_FILE_VALID(p->one)) {
1536                         diff_populate_filespec(p->one, 1);
1537                         copied = added = 0;
1538                         diff_free_filespec_data(p->one);
1539                 } else if (DIFF_FILE_VALID(p->two)) {
1540                         diff_populate_filespec(p->two, 1);
1541                         copied = 0;
1542                         added = p->two->size;
1543                         diff_free_filespec_data(p->two);
1544                 } else
1545                         continue;
1546
1547                 /*
1548                  * Original minus copied is the removed material,
1549                  * added is the new material.  They are both damages
1550                  * made to the preimage. In --dirstat-by-file mode, count
1551                  * damaged files, not damaged lines. This is done by
1552                  * counting only a single damaged line per file.
1553                  */
1554                 damage = (p->one->size - copied) + added;
1555                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1556                         damage = 1;
1557
1558                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1559                 dir.files[dir.nr].name = name;
1560                 dir.files[dir.nr].changed = damage;
1561                 changed += damage;
1562                 dir.nr++;
1563         }
1564
1565         /* This can happen even with many files, if everything was renames */
1566         if (!changed)
1567                 return;
1568
1569         /* Show all directories with more than x% of the changes */
1570         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1571         gather_dirstat(options, &dir, changed, "", 0);
1572 }
1573
1574 static void free_diffstat_info(struct diffstat_t *diffstat)
1575 {
1576         int i;
1577         for (i = 0; i < diffstat->nr; i++) {
1578                 struct diffstat_file *f = diffstat->files[i];
1579                 if (f->name != f->print_name)
1580                         free(f->print_name);
1581                 free(f->name);
1582                 free(f->from_name);
1583                 free(f);
1584         }
1585         free(diffstat->files);
1586 }
1587
1588 struct checkdiff_t {
1589         const char *filename;
1590         int lineno;
1591         int conflict_marker_size;
1592         struct diff_options *o;
1593         unsigned ws_rule;
1594         unsigned status;
1595 };
1596
1597 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1598 {
1599         char firstchar;
1600         int cnt;
1601
1602         if (len < marker_size + 1)
1603                 return 0;
1604         firstchar = line[0];
1605         switch (firstchar) {
1606         case '=': case '>': case '<': case '|':
1607                 break;
1608         default:
1609                 return 0;
1610         }
1611         for (cnt = 1; cnt < marker_size; cnt++)
1612                 if (line[cnt] != firstchar)
1613                         return 0;
1614         /* line[1] thru line[marker_size-1] are same as firstchar */
1615         if (len < marker_size + 1 || !isspace(line[marker_size]))
1616                 return 0;
1617         return 1;
1618 }
1619
1620 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1621 {
1622         struct checkdiff_t *data = priv;
1623         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1624         int marker_size = data->conflict_marker_size;
1625         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1626         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1627         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1628         char *err;
1629         char *line_prefix = "";
1630         struct strbuf *msgbuf;
1631
1632         assert(data->o);
1633         if (data->o->output_prefix) {
1634                 msgbuf = data->o->output_prefix(data->o,
1635                         data->o->output_prefix_data);
1636                 line_prefix = msgbuf->buf;
1637         }
1638
1639         if (line[0] == '+') {
1640                 unsigned bad;
1641                 data->lineno++;
1642                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1643                         data->status |= 1;
1644                         fprintf(data->o->file,
1645                                 "%s%s:%d: leftover conflict marker\n",
1646                                 line_prefix, data->filename, data->lineno);
1647                 }
1648                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1649                 if (!bad)
1650                         return;
1651                 data->status |= bad;
1652                 err = whitespace_error_string(bad);
1653                 fprintf(data->o->file, "%s%s:%d: %s.\n",
1654                         line_prefix, data->filename, data->lineno, err);
1655                 free(err);
1656                 emit_line(data->o, set, reset, line, 1);
1657                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1658                               data->o->file, set, reset, ws);
1659         } else if (line[0] == ' ') {
1660                 data->lineno++;
1661         } else if (line[0] == '@') {
1662                 char *plus = strchr(line, '+');
1663                 if (plus)
1664                         data->lineno = strtol(plus, NULL, 10) - 1;
1665                 else
1666                         die("invalid diff");
1667         }
1668 }
1669
1670 static unsigned char *deflate_it(char *data,
1671                                  unsigned long size,
1672                                  unsigned long *result_size)
1673 {
1674         int bound;
1675         unsigned char *deflated;
1676         z_stream stream;
1677
1678         memset(&stream, 0, sizeof(stream));
1679         deflateInit(&stream, zlib_compression_level);
1680         bound = deflateBound(&stream, size);
1681         deflated = xmalloc(bound);
1682         stream.next_out = deflated;
1683         stream.avail_out = bound;
1684
1685         stream.next_in = (unsigned char *)data;
1686         stream.avail_in = size;
1687         while (deflate(&stream, Z_FINISH) == Z_OK)
1688                 ; /* nothing */
1689         deflateEnd(&stream);
1690         *result_size = stream.total_out;
1691         return deflated;
1692 }
1693
1694 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1695 {
1696         void *cp;
1697         void *delta;
1698         void *deflated;
1699         void *data;
1700         unsigned long orig_size;
1701         unsigned long delta_size;
1702         unsigned long deflate_size;
1703         unsigned long data_size;
1704
1705         /* We could do deflated delta, or we could do just deflated two,
1706          * whichever is smaller.
1707          */
1708         delta = NULL;
1709         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1710         if (one->size && two->size) {
1711                 delta = diff_delta(one->ptr, one->size,
1712                                    two->ptr, two->size,
1713                                    &delta_size, deflate_size);
1714                 if (delta) {
1715                         void *to_free = delta;
1716                         orig_size = delta_size;
1717                         delta = deflate_it(delta, delta_size, &delta_size);
1718                         free(to_free);
1719                 }
1720         }
1721
1722         if (delta && delta_size < deflate_size) {
1723                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
1724                 free(deflated);
1725                 data = delta;
1726                 data_size = delta_size;
1727         }
1728         else {
1729                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
1730                 free(delta);
1731                 data = deflated;
1732                 data_size = deflate_size;
1733         }
1734
1735         /* emit data encoded in base85 */
1736         cp = data;
1737         while (data_size) {
1738                 int bytes = (52 < data_size) ? 52 : data_size;
1739                 char line[70];
1740                 data_size -= bytes;
1741                 if (bytes <= 26)
1742                         line[0] = bytes + 'A' - 1;
1743                 else
1744                         line[0] = bytes - 26 + 'a' - 1;
1745                 encode_85(line + 1, cp, bytes);
1746                 cp = (char *) cp + bytes;
1747                 fprintf(file, "%s", prefix);
1748                 fputs(line, file);
1749                 fputc('\n', file);
1750         }
1751         fprintf(file, "%s\n", prefix);
1752         free(data);
1753 }
1754
1755 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1756 {
1757         fprintf(file, "%sGIT binary patch\n", prefix);
1758         emit_binary_diff_body(file, one, two, prefix);
1759         emit_binary_diff_body(file, two, one, prefix);
1760 }
1761
1762 static void diff_filespec_load_driver(struct diff_filespec *one)
1763 {
1764         if (!one->driver)
1765                 one->driver = userdiff_find_by_path(one->path);
1766         if (!one->driver)
1767                 one->driver = userdiff_find_by_name("default");
1768 }
1769
1770 int diff_filespec_is_binary(struct diff_filespec *one)
1771 {
1772         if (one->is_binary == -1) {
1773                 diff_filespec_load_driver(one);
1774                 if (one->driver->binary != -1)
1775                         one->is_binary = one->driver->binary;
1776                 else {
1777                         if (!one->data && DIFF_FILE_VALID(one))
1778                                 diff_populate_filespec(one, 0);
1779                         if (one->data)
1780                                 one->is_binary = buffer_is_binary(one->data,
1781                                                 one->size);
1782                         if (one->is_binary == -1)
1783                                 one->is_binary = 0;
1784                 }
1785         }
1786         return one->is_binary;
1787 }
1788
1789 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1790 {
1791         diff_filespec_load_driver(one);
1792         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1793 }
1794
1795 static const char *userdiff_word_regex(struct diff_filespec *one)
1796 {
1797         diff_filespec_load_driver(one);
1798         return one->driver->word_regex;
1799 }
1800
1801 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1802 {
1803         if (!options->a_prefix)
1804                 options->a_prefix = a;
1805         if (!options->b_prefix)
1806                 options->b_prefix = b;
1807 }
1808
1809 struct userdiff_driver *get_textconv(struct diff_filespec *one)
1810 {
1811         if (!DIFF_FILE_VALID(one))
1812                 return NULL;
1813         if (!S_ISREG(one->mode))
1814                 return NULL;
1815         diff_filespec_load_driver(one);
1816         if (!one->driver->textconv)
1817                 return NULL;
1818
1819         if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
1820                 struct notes_cache *c = xmalloc(sizeof(*c));
1821                 struct strbuf name = STRBUF_INIT;
1822
1823                 strbuf_addf(&name, "textconv/%s", one->driver->name);
1824                 notes_cache_init(c, name.buf, one->driver->textconv);
1825                 one->driver->textconv_cache = c;
1826         }
1827
1828         return one->driver;
1829 }
1830
1831 static void builtin_diff(const char *name_a,
1832                          const char *name_b,
1833                          struct diff_filespec *one,
1834                          struct diff_filespec *two,
1835                          const char *xfrm_msg,
1836                          int must_show_header,
1837                          struct diff_options *o,
1838                          int complete_rewrite)
1839 {
1840         mmfile_t mf1, mf2;
1841         const char *lbl[2];
1842         char *a_one, *b_two;
1843         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1844         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1845         const char *a_prefix, *b_prefix;
1846         struct userdiff_driver *textconv_one = NULL;
1847         struct userdiff_driver *textconv_two = NULL;
1848         struct strbuf header = STRBUF_INIT;
1849         struct strbuf *msgbuf;
1850         char *line_prefix = "";
1851
1852         if (o->output_prefix) {
1853                 msgbuf = o->output_prefix(o, o->output_prefix_data);
1854                 line_prefix = msgbuf->buf;
1855         }
1856
1857         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1858                         (!one->mode || S_ISGITLINK(one->mode)) &&
1859                         (!two->mode || S_ISGITLINK(two->mode))) {
1860                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1861                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1862                 show_submodule_summary(o->file, one ? one->path : two->path,
1863                                 one->sha1, two->sha1, two->dirty_submodule,
1864                                 del, add, reset);
1865                 return;
1866         }
1867
1868         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1869                 textconv_one = get_textconv(one);
1870                 textconv_two = get_textconv(two);
1871         }
1872
1873         diff_set_mnemonic_prefix(o, "a/", "b/");
1874         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1875                 a_prefix = o->b_prefix;
1876                 b_prefix = o->a_prefix;
1877         } else {
1878                 a_prefix = o->a_prefix;
1879                 b_prefix = o->b_prefix;
1880         }
1881
1882         /* Never use a non-valid filename anywhere if at all possible */
1883         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1884         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1885
1886         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1887         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1888         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1889         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1890         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
1891         if (lbl[0][0] == '/') {
1892                 /* /dev/null */
1893                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
1894                 if (xfrm_msg)
1895                         strbuf_addstr(&header, xfrm_msg);
1896                 must_show_header = 1;
1897         }
1898         else if (lbl[1][0] == '/') {
1899                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
1900                 if (xfrm_msg)
1901                         strbuf_addstr(&header, xfrm_msg);
1902                 must_show_header = 1;
1903         }
1904         else {
1905                 if (one->mode != two->mode) {
1906                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
1907                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
1908                         must_show_header = 1;
1909                 }
1910                 if (xfrm_msg)
1911                         strbuf_addstr(&header, xfrm_msg);
1912
1913                 /*
1914                  * we do not run diff between different kind
1915                  * of objects.
1916                  */
1917                 if ((one->mode ^ two->mode) & S_IFMT)
1918                         goto free_ab_and_return;
1919                 if (complete_rewrite &&
1920                     (textconv_one || !diff_filespec_is_binary(one)) &&
1921                     (textconv_two || !diff_filespec_is_binary(two))) {
1922                         fprintf(o->file, "%s", header.buf);
1923                         strbuf_reset(&header);
1924                         emit_rewrite_diff(name_a, name_b, one, two,
1925                                                 textconv_one, textconv_two, o);
1926                         o->found_changes = 1;
1927                         goto free_ab_and_return;
1928                 }
1929         }
1930
1931         if (!DIFF_OPT_TST(o, TEXT) &&
1932             ( (!textconv_one && diff_filespec_is_binary(one)) ||
1933               (!textconv_two && diff_filespec_is_binary(two)) )) {
1934                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1935                         die("unable to read files to diff");
1936                 /* Quite common confusing case */
1937                 if (mf1.size == mf2.size &&
1938                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
1939                         if (must_show_header)
1940                                 fprintf(o->file, "%s", header.buf);
1941                         goto free_ab_and_return;
1942                 }
1943                 fprintf(o->file, "%s", header.buf);
1944                 strbuf_reset(&header);
1945                 if (DIFF_OPT_TST(o, BINARY))
1946                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
1947                 else
1948                         fprintf(o->file, "%sBinary files %s and %s differ\n",
1949                                 line_prefix, lbl[0], lbl[1]);
1950                 o->found_changes = 1;
1951         }
1952         else {
1953                 /* Crazy xdl interfaces.. */
1954                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1955                 xpparam_t xpp;
1956                 xdemitconf_t xecfg;
1957                 struct emit_callback ecbdata;
1958                 const struct userdiff_funcname *pe;
1959
1960                 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
1961                         fprintf(o->file, "%s", header.buf);
1962                         strbuf_reset(&header);
1963                 }
1964
1965                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
1966                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
1967
1968                 pe = diff_funcname_pattern(one);
1969                 if (!pe)
1970                         pe = diff_funcname_pattern(two);
1971
1972                 memset(&xpp, 0, sizeof(xpp));
1973                 memset(&xecfg, 0, sizeof(xecfg));
1974                 memset(&ecbdata, 0, sizeof(ecbdata));
1975                 ecbdata.label_path = lbl;
1976                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1977                 ecbdata.found_changesp = &o->found_changes;
1978                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1979                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1980                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1981                 ecbdata.opt = o;
1982                 ecbdata.header = header.len ? &header : NULL;
1983                 xpp.flags = o->xdl_opts;
1984                 xecfg.ctxlen = o->context;
1985                 xecfg.interhunkctxlen = o->interhunkcontext;
1986                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1987                 if (pe)
1988                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1989                 if (!diffopts)
1990                         ;
1991                 else if (!prefixcmp(diffopts, "--unified="))
1992                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1993                 else if (!prefixcmp(diffopts, "-u"))
1994                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1995                 if (o->word_diff) {
1996                         int i;
1997
1998                         ecbdata.diff_words =
1999                                 xcalloc(1, sizeof(struct diff_words_data));
2000                         ecbdata.diff_words->type = o->word_diff;
2001                         ecbdata.diff_words->opt = o;
2002                         if (!o->word_regex)
2003                                 o->word_regex = userdiff_word_regex(one);
2004                         if (!o->word_regex)
2005                                 o->word_regex = userdiff_word_regex(two);
2006                         if (!o->word_regex)
2007                                 o->word_regex = diff_word_regex_cfg;
2008                         if (o->word_regex) {
2009                                 ecbdata.diff_words->word_regex = (regex_t *)
2010                                         xmalloc(sizeof(regex_t));
2011                                 if (regcomp(ecbdata.diff_words->word_regex,
2012                                                 o->word_regex,
2013                                                 REG_EXTENDED | REG_NEWLINE))
2014                                         die ("Invalid regular expression: %s",
2015                                                         o->word_regex);
2016                         }
2017                         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2018                                 if (o->word_diff == diff_words_styles[i].type) {
2019                                         ecbdata.diff_words->style =
2020                                                 &diff_words_styles[i];
2021                                         break;
2022                                 }
2023                         }
2024                         if (DIFF_OPT_TST(o, COLOR_DIFF)) {
2025                                 struct diff_words_style *st = ecbdata.diff_words->style;
2026                                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2027                                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2028                                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
2029                         }
2030                 }
2031                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2032                               &xpp, &xecfg);
2033                 if (o->word_diff)
2034                         free_diff_words_data(&ecbdata);
2035                 if (textconv_one)
2036                         free(mf1.ptr);
2037                 if (textconv_two)
2038                         free(mf2.ptr);
2039                 xdiff_clear_find_func(&xecfg);
2040         }
2041
2042  free_ab_and_return:
2043         strbuf_release(&header);
2044         diff_free_filespec_data(one);
2045         diff_free_filespec_data(two);
2046         free(a_one);
2047         free(b_two);
2048         return;
2049 }
2050
2051 static void builtin_diffstat(const char *name_a, const char *name_b,
2052                              struct diff_filespec *one,
2053                              struct diff_filespec *two,
2054                              struct diffstat_t *diffstat,
2055                              struct diff_options *o,
2056                              int complete_rewrite)
2057 {
2058         mmfile_t mf1, mf2;
2059         struct diffstat_file *data;
2060
2061         data = diffstat_add(diffstat, name_a, name_b);
2062
2063         if (!one || !two) {
2064                 data->is_unmerged = 1;
2065                 return;
2066         }
2067         if (complete_rewrite) {
2068                 diff_populate_filespec(one, 0);
2069                 diff_populate_filespec(two, 0);
2070                 data->deleted = count_lines(one->data, one->size);
2071                 data->added = count_lines(two->data, two->size);
2072                 goto free_and_return;
2073         }
2074         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2075                 die("unable to read files to diff");
2076
2077         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2078                 data->is_binary = 1;
2079                 data->added = mf2.size;
2080                 data->deleted = mf1.size;
2081         } else {
2082                 /* Crazy xdl interfaces.. */
2083                 xpparam_t xpp;
2084                 xdemitconf_t xecfg;
2085
2086                 memset(&xpp, 0, sizeof(xpp));
2087                 memset(&xecfg, 0, sizeof(xecfg));
2088                 xpp.flags = o->xdl_opts;
2089                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2090                               &xpp, &xecfg);
2091         }
2092
2093  free_and_return:
2094         diff_free_filespec_data(one);
2095         diff_free_filespec_data(two);
2096 }
2097
2098 static void builtin_checkdiff(const char *name_a, const char *name_b,
2099                               const char *attr_path,
2100                               struct diff_filespec *one,
2101                               struct diff_filespec *two,
2102                               struct diff_options *o)
2103 {
2104         mmfile_t mf1, mf2;
2105         struct checkdiff_t data;
2106
2107         if (!two)
2108                 return;
2109
2110         memset(&data, 0, sizeof(data));
2111         data.filename = name_b ? name_b : name_a;
2112         data.lineno = 0;
2113         data.o = o;
2114         data.ws_rule = whitespace_rule(attr_path);
2115         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2116
2117         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2118                 die("unable to read files to diff");
2119
2120         /*
2121          * All the other codepaths check both sides, but not checking
2122          * the "old" side here is deliberate.  We are checking the newly
2123          * introduced changes, and as long as the "new" side is text, we
2124          * can and should check what it introduces.
2125          */
2126         if (diff_filespec_is_binary(two))
2127                 goto free_and_return;
2128         else {
2129                 /* Crazy xdl interfaces.. */
2130                 xpparam_t xpp;
2131                 xdemitconf_t xecfg;
2132
2133                 memset(&xpp, 0, sizeof(xpp));
2134                 memset(&xecfg, 0, sizeof(xecfg));
2135                 xecfg.ctxlen = 1; /* at least one context line */
2136                 xpp.flags = 0;
2137                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2138                               &xpp, &xecfg);
2139
2140                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2141                         struct emit_callback ecbdata;
2142                         int blank_at_eof;
2143
2144                         ecbdata.ws_rule = data.ws_rule;
2145                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2146                         blank_at_eof = ecbdata.blank_at_eof_in_preimage;
2147
2148                         if (blank_at_eof) {
2149                                 static char *err;
2150                                 if (!err)
2151                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2152                                 fprintf(o->file, "%s:%d: %s.\n",
2153                                         data.filename, blank_at_eof, err);
2154                                 data.status = 1; /* report errors */
2155                         }
2156                 }
2157         }
2158  free_and_return:
2159         diff_free_filespec_data(one);
2160         diff_free_filespec_data(two);
2161         if (data.status)
2162                 DIFF_OPT_SET(o, CHECK_FAILED);
2163 }
2164
2165 struct diff_filespec *alloc_filespec(const char *path)
2166 {
2167         int namelen = strlen(path);
2168         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2169
2170         memset(spec, 0, sizeof(*spec));
2171         spec->path = (char *)(spec + 1);
2172         memcpy(spec->path, path, namelen+1);
2173         spec->count = 1;
2174         spec->is_binary = -1;
2175         return spec;
2176 }
2177
2178 void free_filespec(struct diff_filespec *spec)
2179 {
2180         if (!--spec->count) {
2181                 diff_free_filespec_data(spec);
2182                 free(spec);
2183         }
2184 }
2185
2186 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2187                    unsigned short mode)
2188 {
2189         if (mode) {
2190                 spec->mode = canon_mode(mode);
2191                 hashcpy(spec->sha1, sha1);
2192                 spec->sha1_valid = !is_null_sha1(sha1);
2193         }
2194 }
2195
2196 /*
2197  * Given a name and sha1 pair, if the index tells us the file in
2198  * the work tree has that object contents, return true, so that
2199  * prepare_temp_file() does not have to inflate and extract.
2200  */
2201 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2202 {
2203         struct cache_entry *ce;
2204         struct stat st;
2205         int pos, len;
2206
2207         /*
2208          * We do not read the cache ourselves here, because the
2209          * benchmark with my previous version that always reads cache
2210          * shows that it makes things worse for diff-tree comparing
2211          * two linux-2.6 kernel trees in an already checked out work
2212          * tree.  This is because most diff-tree comparisons deal with
2213          * only a small number of files, while reading the cache is
2214          * expensive for a large project, and its cost outweighs the
2215          * savings we get by not inflating the object to a temporary
2216          * file.  Practically, this code only helps when we are used
2217          * by diff-cache --cached, which does read the cache before
2218          * calling us.
2219          */
2220         if (!active_cache)
2221                 return 0;
2222
2223         /* We want to avoid the working directory if our caller
2224          * doesn't need the data in a normal file, this system
2225          * is rather slow with its stat/open/mmap/close syscalls,
2226          * and the object is contained in a pack file.  The pack
2227          * is probably already open and will be faster to obtain
2228          * the data through than the working directory.  Loose
2229          * objects however would tend to be slower as they need
2230          * to be individually opened and inflated.
2231          */
2232         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2233                 return 0;
2234
2235         len = strlen(name);
2236         pos = cache_name_pos(name, len);
2237         if (pos < 0)
2238                 return 0;
2239         ce = active_cache[pos];
2240
2241         /*
2242          * This is not the sha1 we are looking for, or
2243          * unreusable because it is not a regular file.
2244          */
2245         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2246                 return 0;
2247
2248         /*
2249          * If ce is marked as "assume unchanged", there is no
2250          * guarantee that work tree matches what we are looking for.
2251          */
2252         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2253                 return 0;
2254
2255         /*
2256          * If ce matches the file in the work tree, we can reuse it.
2257          */
2258         if (ce_uptodate(ce) ||
2259             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2260                 return 1;
2261
2262         return 0;
2263 }
2264
2265 static int populate_from_stdin(struct diff_filespec *s)
2266 {
2267         struct strbuf buf = STRBUF_INIT;
2268         size_t size = 0;
2269
2270         if (strbuf_read(&buf, 0, 0) < 0)
2271                 return error("error while reading from stdin %s",
2272                                      strerror(errno));
2273
2274         s->should_munmap = 0;
2275         s->data = strbuf_detach(&buf, &size);
2276         s->size = size;
2277         s->should_free = 1;
2278         return 0;
2279 }
2280
2281 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2282 {
2283         int len;
2284         char *data = xmalloc(100), *dirty = "";
2285
2286         /* Are we looking at the work tree? */
2287         if (s->dirty_submodule)
2288                 dirty = "-dirty";
2289
2290         len = snprintf(data, 100,
2291                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2292         s->data = data;
2293         s->size = len;
2294         s->should_free = 1;
2295         if (size_only) {
2296                 s->data = NULL;
2297                 free(data);
2298         }
2299         return 0;
2300 }
2301
2302 /*
2303  * While doing rename detection and pickaxe operation, we may need to
2304  * grab the data for the blob (or file) for our own in-core comparison.
2305  * diff_filespec has data and size fields for this purpose.
2306  */
2307 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2308 {
2309         int err = 0;
2310         if (!DIFF_FILE_VALID(s))
2311                 die("internal error: asking to populate invalid file.");
2312         if (S_ISDIR(s->mode))
2313                 return -1;
2314
2315         if (s->data)
2316                 return 0;
2317
2318         if (size_only && 0 < s->size)
2319                 return 0;
2320
2321         if (S_ISGITLINK(s->mode))
2322                 return diff_populate_gitlink(s, size_only);
2323
2324         if (!s->sha1_valid ||
2325             reuse_worktree_file(s->path, s->sha1, 0)) {
2326                 struct strbuf buf = STRBUF_INIT;
2327                 struct stat st;
2328                 int fd;
2329
2330                 if (!strcmp(s->path, "-"))
2331                         return populate_from_stdin(s);
2332
2333                 if (lstat(s->path, &st) < 0) {
2334                         if (errno == ENOENT) {
2335                         err_empty:
2336                                 err = -1;
2337                         empty:
2338                                 s->data = (char *)"";
2339                                 s->size = 0;
2340                                 return err;
2341                         }
2342                 }
2343                 s->size = xsize_t(st.st_size);
2344                 if (!s->size)
2345                         goto empty;
2346                 if (S_ISLNK(st.st_mode)) {
2347                         struct strbuf sb = STRBUF_INIT;
2348
2349                         if (strbuf_readlink(&sb, s->path, s->size))
2350                                 goto err_empty;
2351                         s->size = sb.len;
2352                         s->data = strbuf_detach(&sb, NULL);
2353                         s->should_free = 1;
2354                         return 0;
2355                 }
2356                 if (size_only)
2357                         return 0;
2358                 fd = open(s->path, O_RDONLY);
2359                 if (fd < 0)
2360                         goto err_empty;
2361                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2362                 close(fd);
2363                 s->should_munmap = 1;
2364
2365                 /*
2366                  * Convert from working tree format to canonical git format
2367                  */
2368                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2369                         size_t size = 0;
2370                         munmap(s->data, s->size);
2371                         s->should_munmap = 0;
2372                         s->data = strbuf_detach(&buf, &size);
2373                         s->size = size;
2374                         s->should_free = 1;
2375                 }
2376         }
2377         else {
2378                 enum object_type type;
2379                 if (size_only)
2380                         type = sha1_object_info(s->sha1, &s->size);
2381                 else {
2382                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2383                         s->should_free = 1;
2384                 }
2385         }
2386         return 0;
2387 }
2388
2389 void diff_free_filespec_blob(struct diff_filespec *s)
2390 {
2391         if (s->should_free)
2392                 free(s->data);
2393         else if (s->should_munmap)
2394                 munmap(s->data, s->size);
2395
2396         if (s->should_free || s->should_munmap) {
2397                 s->should_free = s->should_munmap = 0;
2398                 s->data = NULL;
2399         }
2400 }
2401
2402 void diff_free_filespec_data(struct diff_filespec *s)
2403 {
2404         diff_free_filespec_blob(s);
2405         free(s->cnt_data);
2406         s->cnt_data = NULL;
2407 }
2408
2409 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2410                            void *blob,
2411                            unsigned long size,
2412                            const unsigned char *sha1,
2413                            int mode)
2414 {
2415         int fd;
2416         struct strbuf buf = STRBUF_INIT;
2417         struct strbuf template = STRBUF_INIT;
2418         char *path_dup = xstrdup(path);
2419         const char *base = basename(path_dup);
2420
2421         /* Generate "XXXXXX_basename.ext" */
2422         strbuf_addstr(&template, "XXXXXX_");
2423         strbuf_addstr(&template, base);
2424
2425         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2426                         strlen(base) + 1);
2427         if (fd < 0)
2428                 die_errno("unable to create temp-file");
2429         if (convert_to_working_tree(path,
2430                         (const char *)blob, (size_t)size, &buf)) {
2431                 blob = buf.buf;
2432                 size = buf.len;
2433         }
2434         if (write_in_full(fd, blob, size) != size)
2435                 die_errno("unable to write temp-file");
2436         close(fd);
2437         temp->name = temp->tmp_path;
2438         strcpy(temp->hex, sha1_to_hex(sha1));
2439         temp->hex[40] = 0;
2440         sprintf(temp->mode, "%06o", mode);
2441         strbuf_release(&buf);
2442         strbuf_release(&template);
2443         free(path_dup);
2444 }
2445
2446 static struct diff_tempfile *prepare_temp_file(const char *name,
2447                 struct diff_filespec *one)
2448 {
2449         struct diff_tempfile *temp = claim_diff_tempfile();
2450
2451         if (!DIFF_FILE_VALID(one)) {
2452         not_a_valid_file:
2453                 /* A '-' entry produces this for file-2, and
2454                  * a '+' entry produces this for file-1.
2455                  */
2456                 temp->name = "/dev/null";
2457                 strcpy(temp->hex, ".");
2458                 strcpy(temp->mode, ".");
2459                 return temp;
2460         }
2461
2462         if (!remove_tempfile_installed) {
2463                 atexit(remove_tempfile);
2464                 sigchain_push_common(remove_tempfile_on_signal);
2465                 remove_tempfile_installed = 1;
2466         }
2467
2468         if (!one->sha1_valid ||
2469             reuse_worktree_file(name, one->sha1, 1)) {
2470                 struct stat st;
2471                 if (lstat(name, &st) < 0) {
2472                         if (errno == ENOENT)
2473                                 goto not_a_valid_file;
2474                         die_errno("stat(%s)", name);
2475                 }
2476                 if (S_ISLNK(st.st_mode)) {
2477                         struct strbuf sb = STRBUF_INIT;
2478                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2479                                 die_errno("readlink(%s)", name);
2480                         prep_temp_blob(name, temp, sb.buf, sb.len,
2481                                        (one->sha1_valid ?
2482                                         one->sha1 : null_sha1),
2483                                        (one->sha1_valid ?
2484                                         one->mode : S_IFLNK));
2485                         strbuf_release(&sb);
2486                 }
2487                 else {
2488                         /* we can borrow from the file in the work tree */
2489                         temp->name = name;
2490                         if (!one->sha1_valid)
2491                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2492                         else
2493                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2494                         /* Even though we may sometimes borrow the
2495                          * contents from the work tree, we always want
2496                          * one->mode.  mode is trustworthy even when
2497                          * !(one->sha1_valid), as long as
2498                          * DIFF_FILE_VALID(one).
2499                          */
2500                         sprintf(temp->mode, "%06o", one->mode);
2501                 }
2502                 return temp;
2503         }
2504         else {
2505                 if (diff_populate_filespec(one, 0))
2506                         die("cannot read data blob for %s", one->path);
2507                 prep_temp_blob(name, temp, one->data, one->size,
2508                                one->sha1, one->mode);
2509         }
2510         return temp;
2511 }
2512
2513 /* An external diff command takes:
2514  *
2515  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2516  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2517  *
2518  */
2519 static void run_external_diff(const char *pgm,
2520                               const char *name,
2521                               const char *other,
2522                               struct diff_filespec *one,
2523                               struct diff_filespec *two,
2524                               const char *xfrm_msg,
2525                               int complete_rewrite)
2526 {
2527         const char *spawn_arg[10];
2528         int retval;
2529         const char **arg = &spawn_arg[0];
2530
2531         if (one && two) {
2532                 struct diff_tempfile *temp_one, *temp_two;
2533                 const char *othername = (other ? other : name);
2534                 temp_one = prepare_temp_file(name, one);
2535                 temp_two = prepare_temp_file(othername, two);
2536                 *arg++ = pgm;
2537                 *arg++ = name;
2538                 *arg++ = temp_one->name;
2539                 *arg++ = temp_one->hex;
2540                 *arg++ = temp_one->mode;
2541                 *arg++ = temp_two->name;
2542                 *arg++ = temp_two->hex;
2543                 *arg++ = temp_two->mode;
2544                 if (other) {
2545                         *arg++ = other;
2546                         *arg++ = xfrm_msg;
2547                 }
2548         } else {
2549                 *arg++ = pgm;
2550                 *arg++ = name;
2551         }
2552         *arg = NULL;
2553         fflush(NULL);
2554         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2555         remove_tempfile();
2556         if (retval) {
2557                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2558                 exit(1);
2559         }
2560 }
2561
2562 static int similarity_index(struct diff_filepair *p)
2563 {
2564         return p->score * 100 / MAX_SCORE;
2565 }
2566
2567 static void fill_metainfo(struct strbuf *msg,
2568                           const char *name,
2569                           const char *other,
2570                           struct diff_filespec *one,
2571                           struct diff_filespec *two,
2572                           struct diff_options *o,
2573                           struct diff_filepair *p,
2574                           int *must_show_header,
2575                           int use_color)
2576 {
2577         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2578         const char *reset = diff_get_color(use_color, DIFF_RESET);
2579         struct strbuf *msgbuf;
2580         char *line_prefix = "";
2581
2582         *must_show_header = 1;
2583         if (o->output_prefix) {
2584                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2585                 line_prefix = msgbuf->buf;
2586         }
2587         strbuf_init(msg, PATH_MAX * 2 + 300);
2588         switch (p->status) {
2589         case DIFF_STATUS_COPIED:
2590                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2591                             line_prefix, set, similarity_index(p));
2592                 strbuf_addf(msg, "%s\n%s%scopy from ",
2593                             reset,  line_prefix, set);
2594                 quote_c_style(name, msg, NULL, 0);
2595                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2596                 quote_c_style(other, msg, NULL, 0);
2597                 strbuf_addf(msg, "%s\n", reset);
2598                 break;
2599         case DIFF_STATUS_RENAMED:
2600                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2601                             line_prefix, set, similarity_index(p));
2602                 strbuf_addf(msg, "%s\n%s%srename from ",
2603                             reset, line_prefix, set);
2604                 quote_c_style(name, msg, NULL, 0);
2605                 strbuf_addf(msg, "%s\n%s%srename to ",
2606                             reset, line_prefix, set);
2607                 quote_c_style(other, msg, NULL, 0);
2608                 strbuf_addf(msg, "%s\n", reset);
2609                 break;
2610         case DIFF_STATUS_MODIFIED:
2611                 if (p->score) {
2612                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2613                                     line_prefix,
2614                                     set, similarity_index(p), reset);
2615                         break;
2616                 }
2617                 /* fallthru */
2618         default:
2619                 *must_show_header = 0;
2620         }
2621         if (one && two && hashcmp(one->sha1, two->sha1)) {
2622                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2623
2624                 if (DIFF_OPT_TST(o, BINARY)) {
2625                         mmfile_t mf;
2626                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2627                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2628                                 abbrev = 40;
2629                 }
2630                 strbuf_addf(msg, "%s%sindex %s..", set,
2631                             line_prefix,
2632                             find_unique_abbrev(one->sha1, abbrev));
2633                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2634                 if (one->mode == two->mode)
2635                         strbuf_addf(msg, " %06o", one->mode);
2636                 strbuf_addf(msg, "%s\n", reset);
2637         }
2638 }
2639
2640 static void run_diff_cmd(const char *pgm,
2641                          const char *name,
2642                          const char *other,
2643                          const char *attr_path,
2644                          struct diff_filespec *one,
2645                          struct diff_filespec *two,
2646                          struct strbuf *msg,
2647                          struct diff_options *o,
2648                          struct diff_filepair *p)
2649 {
2650         const char *xfrm_msg = NULL;
2651         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2652         int must_show_header = 0;
2653
2654         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2655                 pgm = NULL;
2656         else {
2657                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2658                 if (drv && drv->external)
2659                         pgm = drv->external;
2660         }
2661
2662         if (msg) {
2663                 /*
2664                  * don't use colors when the header is intended for an
2665                  * external diff driver
2666                  */
2667                 fill_metainfo(msg, name, other, one, two, o, p,
2668                               &must_show_header,
2669                               DIFF_OPT_TST(o, COLOR_DIFF) && !pgm);
2670                 xfrm_msg = msg->len ? msg->buf : NULL;
2671         }
2672
2673         if (pgm) {
2674                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2675                                   complete_rewrite);
2676                 return;
2677         }
2678         if (one && two)
2679                 builtin_diff(name, other ? other : name,
2680                              one, two, xfrm_msg, must_show_header,
2681                              o, complete_rewrite);
2682         else
2683                 fprintf(o->file, "* Unmerged path %s\n", name);
2684 }
2685
2686 static void diff_fill_sha1_info(struct diff_filespec *one)
2687 {
2688         if (DIFF_FILE_VALID(one)) {
2689                 if (!one->sha1_valid) {
2690                         struct stat st;
2691                         if (!strcmp(one->path, "-")) {
2692                                 hashcpy(one->sha1, null_sha1);
2693                                 return;
2694                         }
2695                         if (lstat(one->path, &st) < 0)
2696                                 die_errno("stat '%s'", one->path);
2697                         if (index_path(one->sha1, one->path, &st, 0))
2698                                 die("cannot hash %s", one->path);
2699                 }
2700         }
2701         else
2702                 hashclr(one->sha1);
2703 }
2704
2705 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2706 {
2707         /* Strip the prefix but do not molest /dev/null and absolute paths */
2708         if (*namep && **namep != '/')
2709                 *namep += prefix_length;
2710         if (*otherp && **otherp != '/')
2711                 *otherp += prefix_length;
2712 }
2713
2714 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2715 {
2716         const char *pgm = external_diff();
2717         struct strbuf msg;
2718         struct diff_filespec *one = p->one;
2719         struct diff_filespec *two = p->two;
2720         const char *name;
2721         const char *other;
2722         const char *attr_path;
2723
2724         name  = p->one->path;
2725         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2726         attr_path = name;
2727         if (o->prefix_length)
2728                 strip_prefix(o->prefix_length, &name, &other);
2729
2730         if (DIFF_PAIR_UNMERGED(p)) {
2731                 run_diff_cmd(pgm, name, NULL, attr_path,
2732                              NULL, NULL, NULL, o, p);
2733                 return;
2734         }
2735
2736         diff_fill_sha1_info(one);
2737         diff_fill_sha1_info(two);
2738
2739         if (!pgm &&
2740             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2741             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2742                 /*
2743                  * a filepair that changes between file and symlink
2744                  * needs to be split into deletion and creation.
2745                  */
2746                 struct diff_filespec *null = alloc_filespec(two->path);
2747                 run_diff_cmd(NULL, name, other, attr_path,
2748                              one, null, &msg, o, p);
2749                 free(null);
2750                 strbuf_release(&msg);
2751
2752                 null = alloc_filespec(one->path);
2753                 run_diff_cmd(NULL, name, other, attr_path,
2754                              null, two, &msg, o, p);
2755                 free(null);
2756         }
2757         else
2758                 run_diff_cmd(pgm, name, other, attr_path,
2759                              one, two, &msg, o, p);
2760
2761         strbuf_release(&msg);
2762 }
2763
2764 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2765                          struct diffstat_t *diffstat)
2766 {
2767         const char *name;
2768         const char *other;
2769         int complete_rewrite = 0;
2770
2771         if (DIFF_PAIR_UNMERGED(p)) {
2772                 /* unmerged */
2773                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2774                 return;
2775         }
2776
2777         name = p->one->path;
2778         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2779
2780         if (o->prefix_length)
2781                 strip_prefix(o->prefix_length, &name, &other);
2782
2783         diff_fill_sha1_info(p->one);
2784         diff_fill_sha1_info(p->two);
2785
2786         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2787                 complete_rewrite = 1;
2788         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2789 }
2790
2791 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2792 {
2793         const char *name;
2794         const char *other;
2795         const char *attr_path;
2796
2797         if (DIFF_PAIR_UNMERGED(p)) {
2798                 /* unmerged */
2799                 return;
2800         }
2801
2802         name = p->one->path;
2803         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2804         attr_path = other ? other : name;
2805
2806         if (o->prefix_length)
2807                 strip_prefix(o->prefix_length, &name, &other);
2808
2809         diff_fill_sha1_info(p->one);
2810         diff_fill_sha1_info(p->two);
2811
2812         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2813 }
2814
2815 void diff_setup(struct diff_options *options)
2816 {
2817         memset(options, 0, sizeof(*options));
2818         memset(&diff_queued_diff, 0, sizeof(diff_queued_diff));
2819
2820         options->file = stdout;
2821
2822         options->line_termination = '\n';
2823         options->break_opt = -1;
2824         options->rename_limit = -1;
2825         options->dirstat_percent = 3;
2826         options->context = 3;
2827
2828         options->change = diff_change;
2829         options->add_remove = diff_addremove;
2830         if (diff_use_color_default > 0)
2831                 DIFF_OPT_SET(options, COLOR_DIFF);
2832         options->detect_rename = diff_detect_rename_default;
2833
2834         if (diff_no_prefix) {
2835                 options->a_prefix = options->b_prefix = "";
2836         } else if (!diff_mnemonic_prefix) {
2837                 options->a_prefix = "a/";
2838                 options->b_prefix = "b/";
2839         }
2840 }
2841
2842 int diff_setup_done(struct diff_options *options)
2843 {
2844         int count = 0;
2845
2846         if (options->output_format & DIFF_FORMAT_NAME)
2847                 count++;
2848         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2849                 count++;
2850         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2851                 count++;
2852         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2853                 count++;
2854         if (count > 1)
2855                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2856
2857         /*
2858          * Most of the time we can say "there are changes"
2859          * only by checking if there are changed paths, but
2860          * --ignore-whitespace* options force us to look
2861          * inside contents.
2862          */
2863
2864         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2865             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2866             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2867                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2868         else
2869                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2870
2871         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2872                 options->detect_rename = DIFF_DETECT_COPY;
2873
2874         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2875                 options->prefix = NULL;
2876         if (options->prefix)
2877                 options->prefix_length = strlen(options->prefix);
2878         else
2879                 options->prefix_length = 0;
2880
2881         if (options->output_format & (DIFF_FORMAT_NAME |
2882                                       DIFF_FORMAT_NAME_STATUS |
2883                                       DIFF_FORMAT_CHECKDIFF |
2884                                       DIFF_FORMAT_NO_OUTPUT))
2885                 options->output_format &= ~(DIFF_FORMAT_RAW |
2886                                             DIFF_FORMAT_NUMSTAT |
2887                                             DIFF_FORMAT_DIFFSTAT |
2888                                             DIFF_FORMAT_SHORTSTAT |
2889                                             DIFF_FORMAT_DIRSTAT |
2890                                             DIFF_FORMAT_SUMMARY |
2891                                             DIFF_FORMAT_PATCH);
2892
2893         /*
2894          * These cases always need recursive; we do not drop caller-supplied
2895          * recursive bits for other formats here.
2896          */
2897         if (options->output_format & (DIFF_FORMAT_PATCH |
2898                                       DIFF_FORMAT_NUMSTAT |
2899                                       DIFF_FORMAT_DIFFSTAT |
2900                                       DIFF_FORMAT_SHORTSTAT |
2901                                       DIFF_FORMAT_DIRSTAT |
2902                                       DIFF_FORMAT_SUMMARY |
2903                                       DIFF_FORMAT_CHECKDIFF))
2904                 DIFF_OPT_SET(options, RECURSIVE);
2905         /*
2906          * Also pickaxe would not work very well if you do not say recursive
2907          */
2908         if (options->pickaxe)
2909                 DIFF_OPT_SET(options, RECURSIVE);
2910         /*
2911          * When patches are generated, submodules diffed against the work tree
2912          * must be checked for dirtiness too so it can be shown in the output
2913          */
2914         if (options->output_format & DIFF_FORMAT_PATCH)
2915                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2916
2917         if (options->detect_rename && options->rename_limit < 0)
2918                 options->rename_limit = diff_rename_limit_default;
2919         if (options->setup & DIFF_SETUP_USE_CACHE) {
2920                 if (!active_cache)
2921                         /* read-cache does not die even when it fails
2922                          * so it is safe for us to do this here.  Also
2923                          * it does not smudge active_cache or active_nr
2924                          * when it fails, so we do not have to worry about
2925                          * cleaning it up ourselves either.
2926                          */
2927                         read_cache();
2928         }
2929         if (options->abbrev <= 0 || 40 < options->abbrev)
2930                 options->abbrev = 40; /* full */
2931
2932         /*
2933          * It does not make sense to show the first hit we happened
2934          * to have found.  It does not make sense not to return with
2935          * exit code in such a case either.
2936          */
2937         if (DIFF_OPT_TST(options, QUICK)) {
2938                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2939                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2940         }
2941
2942         return 0;
2943 }
2944
2945 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2946 {
2947         char c, *eq;
2948         int len;
2949
2950         if (*arg != '-')
2951                 return 0;
2952         c = *++arg;
2953         if (!c)
2954                 return 0;
2955         if (c == arg_short) {
2956                 c = *++arg;
2957                 if (!c)
2958                         return 1;
2959                 if (val && isdigit(c)) {
2960                         char *end;
2961                         int n = strtoul(arg, &end, 10);
2962                         if (*end)
2963                                 return 0;
2964                         *val = n;
2965                         return 1;
2966                 }
2967                 return 0;
2968         }
2969         if (c != '-')
2970                 return 0;
2971         arg++;
2972         eq = strchr(arg, '=');
2973         if (eq)
2974                 len = eq - arg;
2975         else
2976                 len = strlen(arg);
2977         if (!len || strncmp(arg, arg_long, len))
2978                 return 0;
2979         if (eq) {
2980                 int n;
2981                 char *end;
2982                 if (!isdigit(*++eq))
2983                         return 0;
2984                 n = strtoul(eq, &end, 10);
2985                 if (*end)
2986                         return 0;
2987                 *val = n;
2988         }
2989         return 1;
2990 }
2991
2992 static int diff_scoreopt_parse(const char *opt);
2993
2994 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2995 {
2996         const char *arg = av[0];
2997
2998         /* Output format options */
2999         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3000                 options->output_format |= DIFF_FORMAT_PATCH;
3001         else if (opt_arg(arg, 'U', "unified", &options->context))
3002                 options->output_format |= DIFF_FORMAT_PATCH;
3003         else if (!strcmp(arg, "--raw"))
3004                 options->output_format |= DIFF_FORMAT_RAW;
3005         else if (!strcmp(arg, "--patch-with-raw"))
3006                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3007         else if (!strcmp(arg, "--numstat"))
3008                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3009         else if (!strcmp(arg, "--shortstat"))
3010                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3011         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
3012                 options->output_format |= DIFF_FORMAT_DIRSTAT;
3013         else if (!strcmp(arg, "--cumulative")) {
3014                 options->output_format |= DIFF_FORMAT_DIRSTAT;
3015                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
3016         } else if (opt_arg(arg, 0, "dirstat-by-file",
3017                            &options->dirstat_percent)) {
3018                 options->output_format |= DIFF_FORMAT_DIRSTAT;
3019                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
3020         }
3021         else if (!strcmp(arg, "--check"))
3022                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3023         else if (!strcmp(arg, "--summary"))
3024                 options->output_format |= DIFF_FORMAT_SUMMARY;
3025         else if (!strcmp(arg, "--patch-with-stat"))
3026                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3027         else if (!strcmp(arg, "--name-only"))
3028                 options->output_format |= DIFF_FORMAT_NAME;
3029         else if (!strcmp(arg, "--name-status"))
3030                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3031         else if (!strcmp(arg, "-s"))
3032                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3033         else if (!prefixcmp(arg, "--stat")) {
3034                 char *end;
3035                 int width = options->stat_width;
3036                 int name_width = options->stat_name_width;
3037                 arg += 6;
3038                 end = (char *)arg;
3039
3040                 switch (*arg) {
3041                 case '-':
3042                         if (!prefixcmp(arg, "-width="))
3043                                 width = strtoul(arg + 7, &end, 10);
3044                         else if (!prefixcmp(arg, "-name-width="))
3045                                 name_width = strtoul(arg + 12, &end, 10);
3046                         break;
3047                 case '=':
3048                         width = strtoul(arg+1, &end, 10);
3049                         if (*end == ',')
3050                                 name_width = strtoul(end+1, &end, 10);
3051                 }
3052
3053                 /* Important! This checks all the error cases! */
3054                 if (*end)
3055                         return 0;
3056                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3057                 options->stat_name_width = name_width;
3058                 options->stat_width = width;
3059         }
3060
3061         /* renames options */
3062         else if (!prefixcmp(arg, "-B")) {
3063                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3064                         return -1;
3065         }
3066         else if (!prefixcmp(arg, "-M")) {
3067                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3068                         return -1;
3069                 options->detect_rename = DIFF_DETECT_RENAME;
3070         }
3071         else if (!prefixcmp(arg, "-C")) {
3072                 if (options->detect_rename == DIFF_DETECT_COPY)
3073                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3074                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3075                         return -1;
3076                 options->detect_rename = DIFF_DETECT_COPY;
3077         }
3078         else if (!strcmp(arg, "--no-renames"))
3079                 options->detect_rename = 0;
3080         else if (!strcmp(arg, "--relative"))
3081                 DIFF_OPT_SET(options, RELATIVE_NAME);
3082         else if (!prefixcmp(arg, "--relative=")) {
3083                 DIFF_OPT_SET(options, RELATIVE_NAME);
3084                 options->prefix = arg + 11;
3085         }
3086
3087         /* xdiff options */
3088         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3089                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3090         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3091                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3092         else if (!strcmp(arg, "--ignore-space-at-eol"))
3093                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3094         else if (!strcmp(arg, "--patience"))
3095                 DIFF_XDL_SET(options, PATIENCE_DIFF);
3096
3097         /* flags options */
3098         else if (!strcmp(arg, "--binary")) {
3099                 options->output_format |= DIFF_FORMAT_PATCH;
3100                 DIFF_OPT_SET(options, BINARY);
3101         }
3102         else if (!strcmp(arg, "--full-index"))
3103                 DIFF_OPT_SET(options, FULL_INDEX);
3104         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3105                 DIFF_OPT_SET(options, TEXT);
3106         else if (!strcmp(arg, "-R"))
3107                 DIFF_OPT_SET(options, REVERSE_DIFF);
3108         else if (!strcmp(arg, "--find-copies-harder"))
3109                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3110         else if (!strcmp(arg, "--follow"))
3111                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3112         else if (!strcmp(arg, "--color"))
3113                 DIFF_OPT_SET(options, COLOR_DIFF);
3114         else if (!prefixcmp(arg, "--color=")) {
3115                 int value = git_config_colorbool(NULL, arg+8, -1);
3116                 if (value == 0)
3117                         DIFF_OPT_CLR(options, COLOR_DIFF);
3118                 else if (value > 0)
3119                         DIFF_OPT_SET(options, COLOR_DIFF);
3120                 else
3121                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3122         }
3123         else if (!strcmp(arg, "--no-color"))
3124                 DIFF_OPT_CLR(options, COLOR_DIFF);
3125         else if (!strcmp(arg, "--color-words")) {
3126                 DIFF_OPT_SET(options, COLOR_DIFF);
3127                 options->word_diff = DIFF_WORDS_COLOR;
3128         }
3129         else if (!prefixcmp(arg, "--color-words=")) {
3130                 DIFF_OPT_SET(options, COLOR_DIFF);
3131                 options->word_diff = DIFF_WORDS_COLOR;
3132                 options->word_regex = arg + 14;
3133         }
3134         else if (!strcmp(arg, "--word-diff")) {
3135                 if (options->word_diff == DIFF_WORDS_NONE)
3136                         options->word_diff = DIFF_WORDS_PLAIN;
3137         }
3138         else if (!prefixcmp(arg, "--word-diff=")) {
3139                 const char *type = arg + 12;
3140                 if (!strcmp(type, "plain"))
3141                         options->word_diff = DIFF_WORDS_PLAIN;
3142                 else if (!strcmp(type, "color")) {
3143                         DIFF_OPT_SET(options, COLOR_DIFF);
3144                         options->word_diff = DIFF_WORDS_COLOR;
3145                 }
3146                 else if (!strcmp(type, "porcelain"))
3147                         options->word_diff = DIFF_WORDS_PORCELAIN;
3148                 else if (!strcmp(type, "none"))
3149                         options->word_diff = DIFF_WORDS_NONE;
3150                 else
3151                         die("bad --word-diff argument: %s", type);
3152         }
3153         else if (!prefixcmp(arg, "--word-diff-regex=")) {
3154                 if (options->word_diff == DIFF_WORDS_NONE)
3155                         options->word_diff = DIFF_WORDS_PLAIN;
3156                 options->word_regex = arg + 18;
3157         }
3158         else if (!strcmp(arg, "--exit-code"))
3159                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3160         else if (!strcmp(arg, "--quiet"))
3161                 DIFF_OPT_SET(options, QUICK);
3162         else if (!strcmp(arg, "--ext-diff"))
3163                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3164         else if (!strcmp(arg, "--no-ext-diff"))
3165                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3166         else if (!strcmp(arg, "--textconv"))
3167                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3168         else if (!strcmp(arg, "--no-textconv"))
3169                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3170         else if (!strcmp(arg, "--ignore-submodules"))
3171                 handle_ignore_submodules_arg(options, "all");
3172         else if (!prefixcmp(arg, "--ignore-submodules="))
3173                 handle_ignore_submodules_arg(options, arg + 20);
3174         else if (!strcmp(arg, "--submodule"))
3175                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3176         else if (!prefixcmp(arg, "--submodule=")) {
3177                 if (!strcmp(arg + 12, "log"))
3178                         DIFF_OPT_SET(options, SUBMODULE_LOG);
3179         }
3180
3181         /* misc options */
3182         else if (!strcmp(arg, "-z"))
3183                 options->line_termination = 0;
3184         else if (!prefixcmp(arg, "-l"))
3185                 options->rename_limit = strtoul(arg+2, NULL, 10);
3186         else if (!prefixcmp(arg, "-S"))
3187                 options->pickaxe = arg + 2;
3188         else if (!strcmp(arg, "--pickaxe-all"))
3189                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
3190         else if (!strcmp(arg, "--pickaxe-regex"))
3191                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
3192         else if (!prefixcmp(arg, "-O"))
3193                 options->orderfile = arg + 2;
3194         else if (!prefixcmp(arg, "--diff-filter="))
3195                 options->filter = arg + 14;
3196         else if (!strcmp(arg, "--abbrev"))
3197                 options->abbrev = DEFAULT_ABBREV;
3198         else if (!prefixcmp(arg, "--abbrev=")) {
3199                 options->abbrev = strtoul(arg + 9, NULL, 10);
3200                 if (options->abbrev < MINIMUM_ABBREV)
3201                         options->abbrev = MINIMUM_ABBREV;
3202                 else if (40 < options->abbrev)
3203                         options->abbrev = 40;
3204         }
3205         else if (!prefixcmp(arg, "--src-prefix="))
3206                 options->a_prefix = arg + 13;
3207         else if (!prefixcmp(arg, "--dst-prefix="))
3208                 options->b_prefix = arg + 13;
3209         else if (!strcmp(arg, "--no-prefix"))
3210                 options->a_prefix = options->b_prefix = "";
3211         else if (opt_arg(arg, '\0', "inter-hunk-context",
3212                          &options->interhunkcontext))
3213                 ;
3214         else if (!prefixcmp(arg, "--output=")) {
3215                 options->file = fopen(arg + strlen("--output="), "w");
3216                 if (!options->file)
3217                         die_errno("Could not open '%s'", arg + strlen("--output="));
3218                 options->close_file = 1;
3219         } else
3220                 return 0;
3221         return 1;
3222 }
3223
3224 static int parse_num(const char **cp_p)
3225 {
3226         unsigned long num, scale;
3227         int ch, dot;
3228         const char *cp = *cp_p;
3229
3230         num = 0;
3231         scale = 1;
3232         dot = 0;
3233         for (;;) {
3234                 ch = *cp;
3235                 if ( !dot && ch == '.' ) {
3236                         scale = 1;
3237                         dot = 1;
3238                 } else if ( ch == '%' ) {
3239                         scale = dot ? scale*100 : 100;
3240                         cp++;   /* % is always at the end */
3241                         break;
3242                 } else if ( ch >= '0' && ch <= '9' ) {
3243                         if ( scale < 100000 ) {
3244                                 scale *= 10;
3245                                 num = (num*10) + (ch-'0');
3246                         }
3247                 } else {
3248                         break;
3249                 }
3250                 cp++;
3251         }
3252         *cp_p = cp;
3253
3254         /* user says num divided by scale and we say internally that
3255          * is MAX_SCORE * num / scale.
3256          */
3257         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3258 }
3259
3260 static int diff_scoreopt_parse(const char *opt)
3261 {
3262         int opt1, opt2, cmd;
3263
3264         if (*opt++ != '-')
3265                 return -1;
3266         cmd = *opt++;
3267         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3268                 return -1; /* that is not a -M, -C nor -B option */
3269
3270         opt1 = parse_num(&opt);
3271         if (cmd != 'B')
3272                 opt2 = 0;
3273         else {
3274                 if (*opt == 0)
3275                         opt2 = 0;
3276                 else if (*opt != '/')
3277                         return -1; /* we expect -B80/99 or -B80 */
3278                 else {
3279                         opt++;
3280                         opt2 = parse_num(&opt);
3281                 }
3282         }
3283         if (*opt != 0)
3284                 return -1;
3285         return opt1 | (opt2 << 16);
3286 }
3287
3288 struct diff_queue_struct diff_queued_diff;
3289
3290 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3291 {
3292         if (queue->alloc <= queue->nr) {
3293                 queue->alloc = alloc_nr(queue->alloc);
3294                 queue->queue = xrealloc(queue->queue,
3295                                         sizeof(dp) * queue->alloc);
3296         }
3297         queue->queue[queue->nr++] = dp;
3298 }
3299
3300 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3301                                  struct diff_filespec *one,
3302                                  struct diff_filespec *two)
3303 {
3304         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3305         dp->one = one;
3306         dp->two = two;
3307         if (queue)
3308                 diff_q(queue, dp);
3309         return dp;
3310 }
3311
3312 void diff_free_filepair(struct diff_filepair *p)
3313 {
3314         free_filespec(p->one);
3315         free_filespec(p->two);
3316         free(p);
3317 }
3318
3319 /* This is different from find_unique_abbrev() in that
3320  * it stuffs the result with dots for alignment.
3321  */
3322 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3323 {
3324         int abblen;
3325         const char *abbrev;
3326         if (len == 40)
3327                 return sha1_to_hex(sha1);
3328
3329         abbrev = find_unique_abbrev(sha1, len);
3330         abblen = strlen(abbrev);
3331         if (abblen < 37) {
3332                 static char hex[41];
3333                 if (len < abblen && abblen <= len + 2)
3334                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3335                 else
3336                         sprintf(hex, "%s...", abbrev);
3337                 return hex;
3338         }
3339         return sha1_to_hex(sha1);
3340 }
3341
3342 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3343 {
3344         int line_termination = opt->line_termination;
3345         int inter_name_termination = line_termination ? '\t' : '\0';
3346         if (opt->output_prefix) {
3347                 struct strbuf *msg = NULL;
3348                 msg = opt->output_prefix(opt, opt->output_prefix_data);
3349                 fprintf(opt->file, "%s", msg->buf);
3350         }
3351
3352         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3353                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3354                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3355                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3356         }
3357         if (p->score) {
3358                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3359                         inter_name_termination);
3360         } else {
3361                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3362         }
3363
3364         if (p->status == DIFF_STATUS_COPIED ||
3365             p->status == DIFF_STATUS_RENAMED) {
3366                 const char *name_a, *name_b;
3367                 name_a = p->one->path;
3368                 name_b = p->two->path;
3369                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3370                 write_name_quoted(name_a, opt->file, inter_name_termination);
3371                 write_name_quoted(name_b, opt->file, line_termination);
3372         } else {
3373                 const char *name_a, *name_b;
3374                 name_a = p->one->mode ? p->one->path : p->two->path;
3375                 name_b = NULL;
3376                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3377                 write_name_quoted(name_a, opt->file, line_termination);
3378         }
3379 }
3380
3381 int diff_unmodified_pair(struct diff_filepair *p)
3382 {
3383         /* This function is written stricter than necessary to support
3384          * the currently implemented transformers, but the idea is to
3385          * let transformers to produce diff_filepairs any way they want,
3386          * and filter and clean them up here before producing the output.
3387          */
3388         struct diff_filespec *one = p->one, *two = p->two;
3389
3390         if (DIFF_PAIR_UNMERGED(p))
3391                 return 0; /* unmerged is interesting */
3392
3393         /* deletion, addition, mode or type change
3394          * and rename are all interesting.
3395          */
3396         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3397             DIFF_PAIR_MODE_CHANGED(p) ||
3398             strcmp(one->path, two->path))
3399                 return 0;
3400
3401         /* both are valid and point at the same path.  that is, we are
3402          * dealing with a change.
3403          */
3404         if (one->sha1_valid && two->sha1_valid &&
3405             !hashcmp(one->sha1, two->sha1) &&
3406             !one->dirty_submodule && !two->dirty_submodule)
3407                 return 1; /* no change */
3408         if (!one->sha1_valid && !two->sha1_valid)
3409                 return 1; /* both look at the same file on the filesystem. */
3410         return 0;
3411 }
3412
3413 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3414 {
3415         if (diff_unmodified_pair(p))
3416                 return;
3417
3418         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3419             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3420                 return; /* no tree diffs in patch format */
3421
3422         run_diff(p, o);
3423 }
3424
3425 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3426                             struct diffstat_t *diffstat)
3427 {
3428         if (diff_unmodified_pair(p))
3429                 return;
3430
3431         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3432             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3433                 return; /* no tree diffs in patch format */
3434
3435         run_diffstat(p, o, diffstat);
3436 }
3437
3438 static void diff_flush_checkdiff(struct diff_filepair *p,
3439                 struct diff_options *o)
3440 {
3441         if (diff_unmodified_pair(p))
3442                 return;
3443
3444         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3445             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3446                 return; /* no tree diffs in patch format */
3447
3448         run_checkdiff(p, o);
3449 }
3450
3451 int diff_queue_is_empty(void)
3452 {
3453         struct diff_queue_struct *q = &diff_queued_diff;
3454         int i;
3455         for (i = 0; i < q->nr; i++)
3456                 if (!diff_unmodified_pair(q->queue[i]))
3457                         return 0;
3458         return 1;
3459 }
3460
3461 #if DIFF_DEBUG
3462 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3463 {
3464         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3465                 x, one ? one : "",
3466                 s->path,
3467                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3468                 s->mode,
3469                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3470         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3471                 x, one ? one : "",
3472                 s->size, s->xfrm_flags);
3473 }
3474
3475 void diff_debug_filepair(const struct diff_filepair *p, int i)
3476 {
3477         diff_debug_filespec(p->one, i, "one");
3478         diff_debug_filespec(p->two, i, "two");
3479         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3480                 p->score, p->status ? p->status : '?',
3481                 p->one->rename_used, p->broken_pair);
3482 }
3483
3484 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3485 {
3486         int i;
3487         if (msg)
3488                 fprintf(stderr, "%s\n", msg);
3489         fprintf(stderr, "q->nr = %d\n", q->nr);
3490         for (i = 0; i < q->nr; i++) {
3491                 struct diff_filepair *p = q->queue[i];
3492                 diff_debug_filepair(p, i);
3493         }
3494 }
3495 #endif
3496
3497 static void diff_resolve_rename_copy(void)
3498 {
3499         int i;
3500         struct diff_filepair *p;
3501         struct diff_queue_struct *q = &diff_queued_diff;
3502
3503         diff_debug_queue("resolve-rename-copy", q);
3504
3505         for (i = 0; i < q->nr; i++) {
3506                 p = q->queue[i];
3507                 p->status = 0; /* undecided */
3508                 if (DIFF_PAIR_UNMERGED(p))
3509                         p->status = DIFF_STATUS_UNMERGED;
3510                 else if (!DIFF_FILE_VALID(p->one))
3511                         p->status = DIFF_STATUS_ADDED;
3512                 else if (!DIFF_FILE_VALID(p->two))
3513                         p->status = DIFF_STATUS_DELETED;
3514                 else if (DIFF_PAIR_TYPE_CHANGED(p))
3515                         p->status = DIFF_STATUS_TYPE_CHANGED;
3516
3517                 /* from this point on, we are dealing with a pair
3518                  * whose both sides are valid and of the same type, i.e.
3519                  * either in-place edit or rename/copy edit.
3520                  */
3521                 else if (DIFF_PAIR_RENAME(p)) {
3522                         /*
3523                          * A rename might have re-connected a broken
3524                          * pair up, causing the pathnames to be the
3525                          * same again. If so, that's not a rename at
3526                          * all, just a modification..
3527                          *
3528                          * Otherwise, see if this source was used for
3529                          * multiple renames, in which case we decrement
3530                          * the count, and call it a copy.
3531                          */
3532                         if (!strcmp(p->one->path, p->two->path))
3533                                 p->status = DIFF_STATUS_MODIFIED;
3534                         else if (--p->one->rename_used > 0)
3535                                 p->status = DIFF_STATUS_COPIED;
3536                         else
3537                                 p->status = DIFF_STATUS_RENAMED;
3538                 }
3539                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3540                          p->one->mode != p->two->mode ||
3541                          p->one->dirty_submodule ||
3542                          p->two->dirty_submodule ||
3543                          is_null_sha1(p->one->sha1))
3544                         p->status = DIFF_STATUS_MODIFIED;
3545                 else {
3546                         /* This is a "no-change" entry and should not
3547                          * happen anymore, but prepare for broken callers.
3548                          */
3549                         error("feeding unmodified %s to diffcore",
3550                               p->one->path);
3551                         p->status = DIFF_STATUS_UNKNOWN;
3552                 }
3553         }
3554         diff_debug_queue("resolve-rename-copy done", q);
3555 }
3556
3557 static int check_pair_status(struct diff_filepair *p)
3558 {
3559         switch (p->status) {
3560         case DIFF_STATUS_UNKNOWN:
3561                 return 0;
3562         case 0:
3563                 die("internal error in diff-resolve-rename-copy");
3564         default:
3565                 return 1;
3566         }
3567 }
3568
3569 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3570 {
3571         int fmt = opt->output_format;
3572
3573         if (fmt & DIFF_FORMAT_CHECKDIFF)
3574                 diff_flush_checkdiff(p, opt);
3575         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3576                 diff_flush_raw(p, opt);
3577         else if (fmt & DIFF_FORMAT_NAME) {
3578                 const char *name_a, *name_b;
3579                 name_a = p->two->path;
3580                 name_b = NULL;
3581                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3582                 write_name_quoted(name_a, opt->file, opt->line_termination);
3583         }
3584 }
3585
3586 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3587 {
3588         if (fs->mode)
3589                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3590         else
3591                 fprintf(file, " %s ", newdelete);
3592         write_name_quoted(fs->path, file, '\n');
3593 }
3594
3595
3596 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
3597                 const char *line_prefix)
3598 {
3599         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3600                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
3601                         p->two->mode, show_name ? ' ' : '\n');
3602                 if (show_name) {
3603                         write_name_quoted(p->two->path, file, '\n');
3604                 }
3605         }
3606 }
3607
3608 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
3609                         const char *line_prefix)
3610 {
3611         char *names = pprint_rename(p->one->path, p->two->path);
3612
3613         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3614         free(names);
3615         show_mode_change(file, p, 0, line_prefix);
3616 }
3617
3618 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
3619 {
3620         FILE *file = opt->file;
3621         char *line_prefix = "";
3622
3623         if (opt->output_prefix) {
3624                 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
3625                 line_prefix = buf->buf;
3626         }
3627
3628         switch(p->status) {
3629         case DIFF_STATUS_DELETED:
3630                 fputs(line_prefix, file);
3631                 show_file_mode_name(file, "delete", p->one);
3632                 break;
3633         case DIFF_STATUS_ADDED:
3634                 fputs(line_prefix, file);
3635                 show_file_mode_name(file, "create", p->two);
3636                 break;
3637         case DIFF_STATUS_COPIED:
3638                 fputs(line_prefix, file);
3639                 show_rename_copy(file, "copy", p, line_prefix);
3640                 break;
3641         case DIFF_STATUS_RENAMED:
3642                 fputs(line_prefix, file);
3643                 show_rename_copy(file, "rename", p, line_prefix);
3644                 break;
3645         default:
3646                 if (p->score) {
3647                         fprintf(file, "%s rewrite ", line_prefix);
3648                         write_name_quoted(p->two->path, file, ' ');
3649                         fprintf(file, "(%d%%)\n", similarity_index(p));
3650                 }
3651                 show_mode_change(file, p, !p->score, line_prefix);
3652                 break;
3653         }
3654 }
3655
3656 struct patch_id_t {
3657         git_SHA_CTX *ctx;
3658         int patchlen;
3659 };
3660
3661 static int remove_space(char *line, int len)
3662 {
3663         int i;
3664         char *dst = line;
3665         unsigned char c;
3666
3667         for (i = 0; i < len; i++)
3668                 if (!isspace((c = line[i])))
3669                         *dst++ = c;
3670
3671         return dst - line;
3672 }
3673
3674 static void patch_id_consume(void *priv, char *line, unsigned long len)
3675 {
3676         struct patch_id_t *data = priv;
3677         int new_len;
3678
3679         /* Ignore line numbers when computing the SHA1 of the patch */
3680         if (!prefixcmp(line, "@@ -"))
3681                 return;
3682
3683         new_len = remove_space(line, len);
3684
3685         git_SHA1_Update(data->ctx, line, new_len);
3686         data->patchlen += new_len;
3687 }
3688
3689 /* returns 0 upon success, and writes result into sha1 */
3690 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3691 {
3692         struct diff_queue_struct *q = &diff_queued_diff;
3693         int i;
3694         git_SHA_CTX ctx;
3695         struct patch_id_t data;
3696         char buffer[PATH_MAX * 4 + 20];
3697
3698         git_SHA1_Init(&ctx);
3699         memset(&data, 0, sizeof(struct patch_id_t));
3700         data.ctx = &ctx;
3701
3702         for (i = 0; i < q->nr; i++) {
3703                 xpparam_t xpp;
3704                 xdemitconf_t xecfg;
3705                 mmfile_t mf1, mf2;
3706                 struct diff_filepair *p = q->queue[i];
3707                 int len1, len2;
3708
3709                 memset(&xpp, 0, sizeof(xpp));
3710                 memset(&xecfg, 0, sizeof(xecfg));
3711                 if (p->status == 0)
3712                         return error("internal diff status error");
3713                 if (p->status == DIFF_STATUS_UNKNOWN)
3714                         continue;
3715                 if (diff_unmodified_pair(p))
3716                         continue;
3717                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3718                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3719                         continue;
3720                 if (DIFF_PAIR_UNMERGED(p))
3721                         continue;
3722
3723                 diff_fill_sha1_info(p->one);
3724                 diff_fill_sha1_info(p->two);
3725                 if (fill_mmfile(&mf1, p->one) < 0 ||
3726                                 fill_mmfile(&mf2, p->two) < 0)
3727                         return error("unable to read files to diff");
3728
3729                 len1 = remove_space(p->one->path, strlen(p->one->path));
3730                 len2 = remove_space(p->two->path, strlen(p->two->path));
3731                 if (p->one->mode == 0)
3732                         len1 = snprintf(buffer, sizeof(buffer),
3733                                         "diff--gita/%.*sb/%.*s"
3734                                         "newfilemode%06o"
3735                                         "---/dev/null"
3736                                         "+++b/%.*s",
3737                                         len1, p->one->path,
3738                                         len2, p->two->path,
3739                                         p->two->mode,
3740                                         len2, p->two->path);
3741                 else if (p->two->mode == 0)
3742                         len1 = snprintf(buffer, sizeof(buffer),
3743                                         "diff--gita/%.*sb/%.*s"
3744                                         "deletedfilemode%06o"
3745                                         "---a/%.*s"
3746                                         "+++/dev/null",
3747                                         len1, p->one->path,
3748                                         len2, p->two->path,
3749                                         p->one->mode,
3750                                         len1, p->one->path);
3751                 else
3752                         len1 = snprintf(buffer, sizeof(buffer),
3753                                         "diff--gita/%.*sb/%.*s"
3754                                         "---a/%.*s"
3755                                         "+++b/%.*s",
3756                                         len1, p->one->path,
3757                                         len2, p->two->path,
3758                                         len1, p->one->path,
3759                                         len2, p->two->path);
3760                 git_SHA1_Update(&ctx, buffer, len1);
3761
3762                 xpp.flags = 0;
3763                 xecfg.ctxlen = 3;
3764                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3765                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3766                               &xpp, &xecfg);
3767         }
3768
3769         git_SHA1_Final(sha1, &ctx);
3770         return 0;
3771 }
3772
3773 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3774 {
3775         struct diff_queue_struct *q = &diff_queued_diff;
3776         int i;
3777         int result = diff_get_patch_id(options, sha1);
3778
3779         for (i = 0; i < q->nr; i++)
3780                 diff_free_filepair(q->queue[i]);
3781
3782         free(q->queue);
3783         DIFF_QUEUE_CLEAR(q);
3784
3785         return result;
3786 }
3787
3788 static int is_summary_empty(const struct diff_queue_struct *q)
3789 {
3790         int i;
3791
3792         for (i = 0; i < q->nr; i++) {
3793                 const struct diff_filepair *p = q->queue[i];
3794
3795                 switch (p->status) {
3796                 case DIFF_STATUS_DELETED:
3797                 case DIFF_STATUS_ADDED:
3798                 case DIFF_STATUS_COPIED:
3799                 case DIFF_STATUS_RENAMED:
3800                         return 0;
3801                 default:
3802                         if (p->score)
3803                                 return 0;
3804                         if (p->one->mode && p->two->mode &&
3805                             p->one->mode != p->two->mode)
3806                                 return 0;
3807                         break;
3808                 }
3809         }
3810         return 1;
3811 }
3812
3813 void diff_flush(struct diff_options *options)
3814 {
3815         struct diff_queue_struct *q = &diff_queued_diff;
3816         int i, output_format = options->output_format;
3817         int separator = 0;
3818
3819         /*
3820          * Order: raw, stat, summary, patch
3821          * or:    name/name-status/checkdiff (other bits clear)
3822          */
3823         if (!q->nr)
3824                 goto free_queue;
3825
3826         if (output_format & (DIFF_FORMAT_RAW |
3827                              DIFF_FORMAT_NAME |
3828                              DIFF_FORMAT_NAME_STATUS |
3829                              DIFF_FORMAT_CHECKDIFF)) {
3830                 for (i = 0; i < q->nr; i++) {
3831                         struct diff_filepair *p = q->queue[i];
3832                         if (check_pair_status(p))
3833                                 flush_one_pair(p, options);
3834                 }
3835                 separator++;
3836         }
3837
3838         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3839                 struct diffstat_t diffstat;
3840
3841                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3842                 for (i = 0; i < q->nr; i++) {
3843                         struct diff_filepair *p = q->queue[i];
3844                         if (check_pair_status(p))
3845                                 diff_flush_stat(p, options, &diffstat);
3846                 }
3847                 if (output_format & DIFF_FORMAT_NUMSTAT)
3848                         show_numstat(&diffstat, options);
3849                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3850                         show_stats(&diffstat, options);
3851                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3852                         show_shortstats(&diffstat, options);
3853                 free_diffstat_info(&diffstat);
3854                 separator++;
3855         }
3856         if (output_format & DIFF_FORMAT_DIRSTAT)
3857                 show_dirstat(options);
3858
3859         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3860                 for (i = 0; i < q->nr; i++) {
3861                         diff_summary(options, q->queue[i]);
3862                 }
3863                 separator++;
3864         }
3865
3866         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3867             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3868             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3869                 /*
3870                  * run diff_flush_patch for the exit status. setting
3871                  * options->file to /dev/null should be safe, becaue we
3872                  * aren't supposed to produce any output anyway.
3873                  */
3874                 if (options->close_file)
3875                         fclose(options->file);
3876                 options->file = fopen("/dev/null", "w");
3877                 if (!options->file)
3878                         die_errno("Could not open /dev/null");
3879                 options->close_file = 1;
3880                 for (i = 0; i < q->nr; i++) {
3881                         struct diff_filepair *p = q->queue[i];
3882                         if (check_pair_status(p))
3883                                 diff_flush_patch(p, options);
3884                         if (options->found_changes)
3885                                 break;
3886                 }
3887         }
3888
3889         if (output_format & DIFF_FORMAT_PATCH) {
3890                 if (separator) {
3891                         putc(options->line_termination, options->file);
3892                         if (options->stat_sep) {
3893                                 /* attach patch instead of inline */
3894                                 fputs(options->stat_sep, options->file);
3895                         }
3896                 }
3897
3898                 for (i = 0; i < q->nr; i++) {
3899                         struct diff_filepair *p = q->queue[i];
3900                         if (check_pair_status(p))
3901                                 diff_flush_patch(p, options);
3902                 }
3903         }
3904
3905         if (output_format & DIFF_FORMAT_CALLBACK)
3906                 options->format_callback(q, options, options->format_callback_data);
3907
3908         for (i = 0; i < q->nr; i++)
3909                 diff_free_filepair(q->queue[i]);
3910 free_queue:
3911         free(q->queue);
3912         DIFF_QUEUE_CLEAR(q);
3913         if (options->close_file)
3914                 fclose(options->file);
3915
3916         /*
3917          * Report the content-level differences with HAS_CHANGES;
3918          * diff_addremove/diff_change does not set the bit when
3919          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3920          */
3921         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3922                 if (options->found_changes)
3923                         DIFF_OPT_SET(options, HAS_CHANGES);
3924                 else
3925                         DIFF_OPT_CLR(options, HAS_CHANGES);
3926         }
3927 }
3928
3929 static void diffcore_apply_filter(const char *filter)
3930 {
3931         int i;
3932         struct diff_queue_struct *q = &diff_queued_diff;
3933         struct diff_queue_struct outq;
3934         DIFF_QUEUE_CLEAR(&outq);
3935
3936         if (!filter)
3937                 return;
3938
3939         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3940                 int found;
3941                 for (i = found = 0; !found && i < q->nr; i++) {
3942                         struct diff_filepair *p = q->queue[i];
3943                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3944                              ((p->score &&
3945                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3946                               (!p->score &&
3947                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3948                             ((p->status != DIFF_STATUS_MODIFIED) &&
3949                              strchr(filter, p->status)))
3950                                 found++;
3951                 }
3952                 if (found)
3953                         return;
3954
3955                 /* otherwise we will clear the whole queue
3956                  * by copying the empty outq at the end of this
3957                  * function, but first clear the current entries
3958                  * in the queue.
3959                  */
3960                 for (i = 0; i < q->nr; i++)
3961                         diff_free_filepair(q->queue[i]);
3962         }
3963         else {
3964                 /* Only the matching ones */
3965                 for (i = 0; i < q->nr; i++) {
3966                         struct diff_filepair *p = q->queue[i];
3967
3968                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3969                              ((p->score &&
3970                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3971                               (!p->score &&
3972                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3973                             ((p->status != DIFF_STATUS_MODIFIED) &&
3974                              strchr(filter, p->status)))
3975                                 diff_q(&outq, p);
3976                         else
3977                                 diff_free_filepair(p);
3978                 }
3979         }
3980         free(q->queue);
3981         *q = outq;
3982 }
3983
3984 /* Check whether two filespecs with the same mode and size are identical */
3985 static int diff_filespec_is_identical(struct diff_filespec *one,
3986                                       struct diff_filespec *two)
3987 {
3988         if (S_ISGITLINK(one->mode))
3989                 return 0;
3990         if (diff_populate_filespec(one, 0))
3991                 return 0;
3992         if (diff_populate_filespec(two, 0))
3993                 return 0;
3994         return !memcmp(one->data, two->data, one->size);
3995 }
3996
3997 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3998 {
3999         int i;
4000         struct diff_queue_struct *q = &diff_queued_diff;
4001         struct diff_queue_struct outq;
4002         DIFF_QUEUE_CLEAR(&outq);
4003
4004         for (i = 0; i < q->nr; i++) {
4005                 struct diff_filepair *p = q->queue[i];
4006
4007                 /*
4008                  * 1. Entries that come from stat info dirtiness
4009                  *    always have both sides (iow, not create/delete),
4010                  *    one side of the object name is unknown, with
4011                  *    the same mode and size.  Keep the ones that
4012                  *    do not match these criteria.  They have real
4013                  *    differences.
4014                  *
4015                  * 2. At this point, the file is known to be modified,
4016                  *    with the same mode and size, and the object
4017                  *    name of one side is unknown.  Need to inspect
4018                  *    the identical contents.
4019                  */
4020                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4021                     !DIFF_FILE_VALID(p->two) ||
4022                     (p->one->sha1_valid && p->two->sha1_valid) ||
4023                     (p->one->mode != p->two->mode) ||
4024                     diff_populate_filespec(p->one, 1) ||
4025                     diff_populate_filespec(p->two, 1) ||
4026                     (p->one->size != p->two->size) ||
4027                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4028                         diff_q(&outq, p);
4029                 else {
4030                         /*
4031                          * The caller can subtract 1 from skip_stat_unmatch
4032                          * to determine how many paths were dirty only
4033                          * due to stat info mismatch.
4034                          */
4035                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4036                                 diffopt->skip_stat_unmatch++;
4037                         diff_free_filepair(p);
4038                 }
4039         }
4040         free(q->queue);
4041         *q = outq;
4042 }
4043
4044 static int diffnamecmp(const void *a_, const void *b_)
4045 {
4046         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4047         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4048         const char *name_a, *name_b;
4049
4050         name_a = a->one ? a->one->path : a->two->path;
4051         name_b = b->one ? b->one->path : b->two->path;
4052         return strcmp(name_a, name_b);
4053 }
4054
4055 void diffcore_fix_diff_index(struct diff_options *options)
4056 {
4057         struct diff_queue_struct *q = &diff_queued_diff;
4058         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4059 }
4060
4061 void diffcore_std(struct diff_options *options)
4062 {
4063         /* We never run this function more than one time, because the
4064          * rename/copy detection logic can only run once.
4065          */
4066         if (diff_queued_diff.run)
4067                 return;
4068
4069         if (options->skip_stat_unmatch)
4070                 diffcore_skip_stat_unmatch(options);
4071         if (options->break_opt != -1)
4072                 diffcore_break(options->break_opt);
4073         if (options->detect_rename)
4074                 diffcore_rename(options);
4075         if (options->break_opt != -1)
4076                 diffcore_merge_broken();
4077         if (options->pickaxe)
4078                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
4079         if (options->orderfile)
4080                 diffcore_order(options->orderfile);
4081         diff_resolve_rename_copy();
4082         diffcore_apply_filter(options->filter);
4083
4084         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4085                 DIFF_OPT_SET(options, HAS_CHANGES);
4086         else
4087                 DIFF_OPT_CLR(options, HAS_CHANGES);
4088
4089         diff_queued_diff.run = 1;
4090 }
4091
4092 int diff_result_code(struct diff_options *opt, int status)
4093 {
4094         int result = 0;
4095         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4096             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4097                 return status;
4098         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4099             DIFF_OPT_TST(opt, HAS_CHANGES))
4100                 result |= 01;
4101         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4102             DIFF_OPT_TST(opt, CHECK_FAILED))
4103                 result |= 02;
4104         return result;
4105 }
4106
4107 void diff_addremove(struct diff_options *options,
4108                     int addremove, unsigned mode,
4109                     const unsigned char *sha1,
4110                     const char *concatpath, unsigned dirty_submodule)
4111 {
4112         struct diff_filespec *one, *two;
4113
4114         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
4115                 return;
4116
4117         /* This may look odd, but it is a preparation for
4118          * feeding "there are unchanged files which should
4119          * not produce diffs, but when you are doing copy
4120          * detection you would need them, so here they are"
4121          * entries to the diff-core.  They will be prefixed
4122          * with something like '=' or '*' (I haven't decided
4123          * which but should not make any difference).
4124          * Feeding the same new and old to diff_change()
4125          * also has the same effect.
4126          * Before the final output happens, they are pruned after
4127          * merged into rename/copy pairs as appropriate.
4128          */
4129         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4130                 addremove = (addremove == '+' ? '-' :
4131                              addremove == '-' ? '+' : addremove);
4132
4133         if (options->prefix &&
4134             strncmp(concatpath, options->prefix, options->prefix_length))
4135                 return;
4136
4137         one = alloc_filespec(concatpath);
4138         two = alloc_filespec(concatpath);
4139
4140         if (addremove != '+')
4141                 fill_filespec(one, sha1, mode);
4142         if (addremove != '-') {
4143                 fill_filespec(two, sha1, mode);
4144                 two->dirty_submodule = dirty_submodule;
4145         }
4146
4147         diff_queue(&diff_queued_diff, one, two);
4148         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4149                 DIFF_OPT_SET(options, HAS_CHANGES);
4150 }
4151
4152 void diff_change(struct diff_options *options,
4153                  unsigned old_mode, unsigned new_mode,
4154                  const unsigned char *old_sha1,
4155                  const unsigned char *new_sha1,
4156                  const char *concatpath,
4157                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4158 {
4159         struct diff_filespec *one, *two;
4160
4161         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
4162                         && S_ISGITLINK(new_mode))
4163                 return;
4164
4165         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4166                 unsigned tmp;
4167                 const unsigned char *tmp_c;
4168                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4169                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4170                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4171                         new_dirty_submodule = tmp;
4172         }
4173
4174         if (options->prefix &&
4175             strncmp(concatpath, options->prefix, options->prefix_length))
4176                 return;
4177
4178         one = alloc_filespec(concatpath);
4179         two = alloc_filespec(concatpath);
4180         fill_filespec(one, old_sha1, old_mode);
4181         fill_filespec(two, new_sha1, new_mode);
4182         one->dirty_submodule = old_dirty_submodule;
4183         two->dirty_submodule = new_dirty_submodule;
4184
4185         diff_queue(&diff_queued_diff, one, two);
4186         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4187                 DIFF_OPT_SET(options, HAS_CHANGES);
4188 }
4189
4190 void diff_unmerge(struct diff_options *options,
4191                   const char *path,
4192                   unsigned mode, const unsigned char *sha1)
4193 {
4194         struct diff_filespec *one, *two;
4195
4196         if (options->prefix &&
4197             strncmp(path, options->prefix, options->prefix_length))
4198                 return;
4199
4200         one = alloc_filespec(path);
4201         two = alloc_filespec(path);
4202         fill_filespec(one, sha1, mode);
4203         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
4204 }
4205
4206 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4207                 size_t *outsize)
4208 {
4209         struct diff_tempfile *temp;
4210         const char *argv[3];
4211         const char **arg = argv;
4212         struct child_process child;
4213         struct strbuf buf = STRBUF_INIT;
4214         int err = 0;
4215
4216         temp = prepare_temp_file(spec->path, spec);
4217         *arg++ = pgm;
4218         *arg++ = temp->name;
4219         *arg = NULL;
4220
4221         memset(&child, 0, sizeof(child));
4222         child.use_shell = 1;
4223         child.argv = argv;
4224         child.out = -1;
4225         if (start_command(&child)) {
4226                 remove_tempfile();
4227                 return NULL;
4228         }
4229
4230         if (strbuf_read(&buf, child.out, 0) < 0)
4231                 err = error("error reading from textconv command '%s'", pgm);
4232         close(child.out);
4233
4234         if (finish_command(&child) || err) {
4235                 strbuf_release(&buf);
4236                 remove_tempfile();
4237                 return NULL;
4238         }
4239         remove_tempfile();
4240
4241         return strbuf_detach(&buf, outsize);
4242 }
4243
4244 size_t fill_textconv(struct userdiff_driver *driver,
4245                      struct diff_filespec *df,
4246                      char **outbuf)
4247 {
4248         size_t size;
4249
4250         if (!driver || !driver->textconv) {
4251                 if (!DIFF_FILE_VALID(df)) {
4252                         *outbuf = "";
4253                         return 0;
4254                 }
4255                 if (diff_populate_filespec(df, 0))
4256                         die("unable to read files to diff");
4257                 *outbuf = df->data;
4258                 return df->size;
4259         }
4260
4261         if (driver->textconv_cache) {
4262                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4263                                           &size);
4264                 if (*outbuf)
4265                         return size;
4266         }
4267
4268         *outbuf = run_textconv(driver->textconv, df, &size);
4269         if (!*outbuf)
4270                 die("unable to read files to diff");
4271
4272         if (driver->textconv_cache) {
4273                 /* ignore errors, as we might be in a readonly repository */
4274                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4275                                 size);
4276                 /*
4277                  * we could save up changes and flush them all at the end,
4278                  * but we would need an extra call after all diffing is done.
4279                  * Since generating a cache entry is the slow path anyway,
4280                  * this extra overhead probably isn't a big deal.
4281                  */
4282                 notes_cache_write(driver->textconv_cache);
4283         }
4284
4285         return size;
4286 }