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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "parse-options.h"
  27 #include "help.h"
  28
  29 #ifdef NO_FAST_WORKING_DIRECTORY
  30 #define FAST_WORKING_DIRECTORY 0
  31 #else
  32 #define FAST_WORKING_DIRECTORY 1
  33 #endif
  34
  35 static int diff_detect_rename_default;
  36 static int diff_indent_heuristic = 1;
  37 static int diff_rename_limit_default = 400;
  38 static int diff_suppress_blank_empty;
  39 static int diff_use_color_default = -1;
  40 static int diff_color_moved_default;
  41 static int diff_color_moved_ws_default;
  42 static int diff_context_default = 3;
  43 static int diff_interhunk_context_default;
  44 static const char *diff_word_regex_cfg;
  45 static const char *external_diff_cmd_cfg;
  46 static const char *diff_order_file_cfg;
  47 int diff_auto_refresh_index = 1;
  48 static int diff_mnemonic_prefix;
  49 static int diff_no_prefix;
  50 static int diff_stat_graph_width;
  51 static int diff_dirstat_permille_default = 30;
  52 static struct diff_options default_diff_options;
  53 static long diff_algorithm;
  54 static unsigned ws_error_highlight_default = WSEH_NEW;
  55
  56 static char diff_colors[][COLOR_MAXLEN] = {
  57         GIT_COLOR_RESET,
  58         GIT_COLOR_NORMAL,       /* CONTEXT */
  59         GIT_COLOR_BOLD,         /* METAINFO */
  60         GIT_COLOR_CYAN,         /* FRAGINFO */
  61         GIT_COLOR_RED,          /* OLD */
  62         GIT_COLOR_GREEN,        /* NEW */
  63         GIT_COLOR_YELLOW,       /* COMMIT */
  64         GIT_COLOR_BG_RED,       /* WHITESPACE */
  65         GIT_COLOR_NORMAL,       /* FUNCINFO */
  66         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  67         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  68         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  69         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  70         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  71         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  72         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  73         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  74         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  75         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  76         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  77         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  78         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  79         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  80 };
  81
  82 static const char *color_diff_slots[] = {
  83         [DIFF_CONTEXT]                = "context",
  84         [DIFF_METAINFO]               = "meta",
  85         [DIFF_FRAGINFO]               = "frag",
  86         [DIFF_FILE_OLD]               = "old",
  87         [DIFF_FILE_NEW]               = "new",
  88         [DIFF_COMMIT]                 = "commit",
  89         [DIFF_WHITESPACE]             = "whitespace",
  90         [DIFF_FUNCINFO]               = "func",
  91         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  92         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  93         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  94         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  95         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  96         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  97         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  98         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  99         [DIFF_CONTEXT_DIM]            = "contextDimmed",
 100         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 101         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 102         [DIFF_CONTEXT_BOLD]           = "contextBold",
 103         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 104         [DIFF_FILE_NEW_BOLD]          = "newBold",
 105 };
 106
 107 define_list_config_array_extra(color_diff_slots, {"plain"});
 108
 109 static int parse_diff_color_slot(const char *var)
 110 {
 111         if (!strcasecmp(var, "plain"))
 112                 return DIFF_CONTEXT;
 113         return LOOKUP_CONFIG(color_diff_slots, var);
 114 }
 115
 116 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 117                                 struct strbuf *errmsg)
 118 {
 119         char *params_copy = xstrdup(params_string);
 120         struct string_list params = STRING_LIST_INIT_NODUP;
 121         int ret = 0;
 122         int i;
 123
 124         if (*params_copy)
 125                 string_list_split_in_place(&params, params_copy, ',', -1);
 126         for (i = 0; i < params.nr; i++) {
 127                 const char *p = params.items[i].string;
 128                 if (!strcmp(p, "changes")) {
 129                         options->flags.dirstat_by_line = 0;
 130                         options->flags.dirstat_by_file = 0;
 131                 } else if (!strcmp(p, "lines")) {
 132                         options->flags.dirstat_by_line = 1;
 133                         options->flags.dirstat_by_file = 0;
 134                 } else if (!strcmp(p, "files")) {
 135                         options->flags.dirstat_by_line = 0;
 136                         options->flags.dirstat_by_file = 1;
 137                 } else if (!strcmp(p, "noncumulative")) {
 138                         options->flags.dirstat_cumulative = 0;
 139                 } else if (!strcmp(p, "cumulative")) {
 140                         options->flags.dirstat_cumulative = 1;
 141                 } else if (isdigit(*p)) {
 142                         char *end;
 143                         int permille = strtoul(p, &end, 10) * 10;
 144                         if (*end == '.' && isdigit(*++end)) {
 145                                 /* only use first digit */
 146                                 permille += *end - '0';
 147                                 /* .. and ignore any further digits */
 148                                 while (isdigit(*++end))
 149                                         ; /* nothing */
 150                         }
 151                         if (!*end)
 152                                 options->dirstat_permille = permille;
 153                         else {
 154                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 155                                             p);
 156                                 ret++;
 157                         }
 158                 } else {
 159                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 160                         ret++;
 161                 }
 162
 163         }
 164         string_list_clear(&params, 0);
 165         free(params_copy);
 166         return ret;
 167 }
 168
 169 static int parse_submodule_params(struct diff_options *options, const char *value)
 170 {
 171         if (!strcmp(value, "log"))
 172                 options->submodule_format = DIFF_SUBMODULE_LOG;
 173         else if (!strcmp(value, "short"))
 174                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 175         else if (!strcmp(value, "diff"))
 176                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 177         else
 178                 return -1;
 179         return 0;
 180 }
 181
 182 int git_config_rename(const char *var, const char *value)
 183 {
 184         if (!value)
 185                 return DIFF_DETECT_RENAME;
 186         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 187                 return  DIFF_DETECT_COPY;
 188         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 189 }
 190
 191 long parse_algorithm_value(const char *value)
 192 {
 193         if (!value)
 194                 return -1;
 195         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 196                 return 0;
 197         else if (!strcasecmp(value, "minimal"))
 198                 return XDF_NEED_MINIMAL;
 199         else if (!strcasecmp(value, "patience"))
 200                 return XDF_PATIENCE_DIFF;
 201         else if (!strcasecmp(value, "histogram"))
 202                 return XDF_HISTOGRAM_DIFF;
 203         return -1;
 204 }
 205
 206 static int parse_one_token(const char **arg, const char *token)
 207 {
 208         const char *rest;
 209         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 210                 *arg = rest;
 211                 return 1;
 212         }
 213         return 0;
 214 }
 215
 216 static int parse_ws_error_highlight(const char *arg)
 217 {
 218         const char *orig_arg = arg;
 219         unsigned val = 0;
 220
 221         while (*arg) {
 222                 if (parse_one_token(&arg, "none"))
 223                         val = 0;
 224                 else if (parse_one_token(&arg, "default"))
 225                         val = WSEH_NEW;
 226                 else if (parse_one_token(&arg, "all"))
 227                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 228                 else if (parse_one_token(&arg, "new"))
 229                         val |= WSEH_NEW;
 230                 else if (parse_one_token(&arg, "old"))
 231                         val |= WSEH_OLD;
 232                 else if (parse_one_token(&arg, "context"))
 233                         val |= WSEH_CONTEXT;
 234                 else {
 235                         return -1 - (int)(arg - orig_arg);
 236                 }
 237                 if (*arg)
 238                         arg++;
 239         }
 240         return val;
 241 }
 242
 243 /*
 244  * These are to give UI layer defaults.
 245  * The core-level commands such as git-diff-files should
 246  * never be affected by the setting of diff.renames
 247  * the user happens to have in the configuration file.
 248  */
 249 void init_diff_ui_defaults(void)
 250 {
 251         diff_detect_rename_default = DIFF_DETECT_RENAME;
 252 }
 253
 254 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 255 {
 256         if (!strcmp(var, "diff.indentheuristic"))
 257                 diff_indent_heuristic = git_config_bool(var, value);
 258         return 0;
 259 }
 260
 261 static int parse_color_moved(const char *arg)
 262 {
 263         switch (git_parse_maybe_bool(arg)) {
 264         case 0:
 265                 return COLOR_MOVED_NO;
 266         case 1:
 267                 return COLOR_MOVED_DEFAULT;
 268         default:
 269                 break;
 270         }
 271
 272         if (!strcmp(arg, "no"))
 273                 return COLOR_MOVED_NO;
 274         else if (!strcmp(arg, "plain"))
 275                 return COLOR_MOVED_PLAIN;
 276         else if (!strcmp(arg, "blocks"))
 277                 return COLOR_MOVED_BLOCKS;
 278         else if (!strcmp(arg, "zebra"))
 279                 return COLOR_MOVED_ZEBRA;
 280         else if (!strcmp(arg, "default"))
 281                 return COLOR_MOVED_DEFAULT;
 282         else if (!strcmp(arg, "dimmed-zebra"))
 283                 return COLOR_MOVED_ZEBRA_DIM;
 284         else if (!strcmp(arg, "dimmed_zebra"))
 285                 return COLOR_MOVED_ZEBRA_DIM;
 286         else
 287                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 288 }
 289
 290 static unsigned parse_color_moved_ws(const char *arg)
 291 {
 292         int ret = 0;
 293         struct string_list l = STRING_LIST_INIT_DUP;
 294         struct string_list_item *i;
 295
 296         string_list_split(&l, arg, ',', -1);
 297
 298         for_each_string_list_item(i, &l) {
 299                 struct strbuf sb = STRBUF_INIT;
 300                 strbuf_addstr(&sb, i->string);
 301                 strbuf_trim(&sb);
 302
 303                 if (!strcmp(sb.buf, "ignore-space-change"))
 304                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 305                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 306                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 307                 else if (!strcmp(sb.buf, "ignore-all-space"))
 308                         ret |= XDF_IGNORE_WHITESPACE;
 309                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 310                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 311                 else {
 312                         ret |= COLOR_MOVED_WS_ERROR;
 313                         error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
 314                 }
 315
 316                 strbuf_release(&sb);
 317         }
 318
 319         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 320             (ret & XDF_WHITESPACE_FLAGS)) {
 321                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
 322                 ret |= COLOR_MOVED_WS_ERROR;
 323         }
 324
 325         string_list_clear(&l, 0);
 326
 327         return ret;
 328 }
 329
 330 int git_diff_ui_config(const char *var, const char *value, void *cb)
 331 {
 332         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 333                 diff_use_color_default = git_config_colorbool(var, value);
 334                 return 0;
 335         }
 336         if (!strcmp(var, "diff.colormoved")) {
 337                 int cm = parse_color_moved(value);
 338                 if (cm < 0)
 339                         return -1;
 340                 diff_color_moved_default = cm;
 341                 return 0;
 342         }
 343         if (!strcmp(var, "diff.colormovedws")) {
 344                 unsigned cm = parse_color_moved_ws(value);
 345                 if (cm & COLOR_MOVED_WS_ERROR)
 346                         return -1;
 347                 diff_color_moved_ws_default = cm;
 348                 return 0;
 349         }
 350         if (!strcmp(var, "diff.context")) {
 351                 diff_context_default = git_config_int(var, value);
 352                 if (diff_context_default < 0)
 353                         return -1;
 354                 return 0;
 355         }
 356         if (!strcmp(var, "diff.interhunkcontext")) {
 357                 diff_interhunk_context_default = git_config_int(var, value);
 358                 if (diff_interhunk_context_default < 0)
 359                         return -1;
 360                 return 0;
 361         }
 362         if (!strcmp(var, "diff.renames")) {
 363                 diff_detect_rename_default = git_config_rename(var, value);
 364                 return 0;
 365         }
 366         if (!strcmp(var, "diff.autorefreshindex")) {
 367                 diff_auto_refresh_index = git_config_bool(var, value);
 368                 return 0;
 369         }
 370         if (!strcmp(var, "diff.mnemonicprefix")) {
 371                 diff_mnemonic_prefix = git_config_bool(var, value);
 372                 return 0;
 373         }
 374         if (!strcmp(var, "diff.noprefix")) {
 375                 diff_no_prefix = git_config_bool(var, value);
 376                 return 0;
 377         }
 378         if (!strcmp(var, "diff.statgraphwidth")) {
 379                 diff_stat_graph_width = git_config_int(var, value);
 380                 return 0;
 381         }
 382         if (!strcmp(var, "diff.external"))
 383                 return git_config_string(&external_diff_cmd_cfg, var, value);
 384         if (!strcmp(var, "diff.wordregex"))
 385                 return git_config_string(&diff_word_regex_cfg, var, value);
 386         if (!strcmp(var, "diff.orderfile"))
 387                 return git_config_pathname(&diff_order_file_cfg, var, value);
 388
 389         if (!strcmp(var, "diff.ignoresubmodules"))
 390                 handle_ignore_submodules_arg(&default_diff_options, value);
 391
 392         if (!strcmp(var, "diff.submodule")) {
 393                 if (parse_submodule_params(&default_diff_options, value))
 394                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 395                                 value);
 396                 return 0;
 397         }
 398
 399         if (!strcmp(var, "diff.algorithm")) {
 400                 diff_algorithm = parse_algorithm_value(value);
 401                 if (diff_algorithm < 0)
 402                         return -1;
 403                 return 0;
 404         }
 405
 406         if (!strcmp(var, "diff.wserrorhighlight")) {
 407                 int val = parse_ws_error_highlight(value);
 408                 if (val < 0)
 409                         return -1;
 410                 ws_error_highlight_default = val;
 411                 return 0;
 412         }
 413
 414         if (git_color_config(var, value, cb) < 0)
 415                 return -1;
 416
 417         return git_diff_basic_config(var, value, cb);
 418 }
 419
 420 int git_diff_basic_config(const char *var, const char *value, void *cb)
 421 {
 422         const char *name;
 423
 424         if (!strcmp(var, "diff.renamelimit")) {
 425                 diff_rename_limit_default = git_config_int(var, value);
 426                 return 0;
 427         }
 428
 429         if (userdiff_config(var, value) < 0)
 430                 return -1;
 431
 432         if (skip_prefix(var, "diff.color.", &name) ||
 433             skip_prefix(var, "color.diff.", &name)) {
 434                 int slot = parse_diff_color_slot(name);
 435                 if (slot < 0)
 436                         return 0;
 437                 if (!value)
 438                         return config_error_nonbool(var);
 439                 return color_parse(value, diff_colors[slot]);
 440         }
 441
 442         /* like GNU diff's --suppress-blank-empty option  */
 443         if (!strcmp(var, "diff.suppressblankempty") ||
 444                         /* for backwards compatibility */
 445                         !strcmp(var, "diff.suppress-blank-empty")) {
 446                 diff_suppress_blank_empty = git_config_bool(var, value);
 447                 return 0;
 448         }
 449
 450         if (!strcmp(var, "diff.dirstat")) {
 451                 struct strbuf errmsg = STRBUF_INIT;
 452                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 453                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 454                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 455                                 errmsg.buf);
 456                 strbuf_release(&errmsg);
 457                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 458                 return 0;
 459         }
 460
 461         if (git_diff_heuristic_config(var, value, cb) < 0)
 462                 return -1;
 463
 464         return git_default_config(var, value, cb);
 465 }
 466
 467 static char *quote_two(const char *one, const char *two)
 468 {
 469         int need_one = quote_c_style(one, NULL, NULL, 1);
 470         int need_two = quote_c_style(two, NULL, NULL, 1);
 471         struct strbuf res = STRBUF_INIT;
 472
 473         if (need_one + need_two) {
 474                 strbuf_addch(&res, '"');
 475                 quote_c_style(one, &res, NULL, 1);
 476                 quote_c_style(two, &res, NULL, 1);
 477                 strbuf_addch(&res, '"');
 478         } else {
 479                 strbuf_addstr(&res, one);
 480                 strbuf_addstr(&res, two);
 481         }
 482         return strbuf_detach(&res, NULL);
 483 }
 484
 485 static const char *external_diff(void)
 486 {
 487         static const char *external_diff_cmd = NULL;
 488         static int done_preparing = 0;
 489
 490         if (done_preparing)
 491                 return external_diff_cmd;
 492         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 493         if (!external_diff_cmd)
 494                 external_diff_cmd = external_diff_cmd_cfg;
 495         done_preparing = 1;
 496         return external_diff_cmd;
 497 }
 498
 499 /*
 500  * Keep track of files used for diffing. Sometimes such an entry
 501  * refers to a temporary file, sometimes to an existing file, and
 502  * sometimes to "/dev/null".
 503  */
 504 static struct diff_tempfile {
 505         /*
 506          * filename external diff should read from, or NULL if this
 507          * entry is currently not in use:
 508          */
 509         const char *name;
 510
 511         char hex[GIT_MAX_HEXSZ + 1];
 512         char mode[10];
 513
 514         /*
 515          * If this diff_tempfile instance refers to a temporary file,
 516          * this tempfile object is used to manage its lifetime.
 517          */
 518         struct tempfile *tempfile;
 519 } diff_temp[2];
 520
 521 struct emit_callback {
 522         int color_diff;
 523         unsigned ws_rule;
 524         int blank_at_eof_in_preimage;
 525         int blank_at_eof_in_postimage;
 526         int lno_in_preimage;
 527         int lno_in_postimage;
 528         const char **label_path;
 529         struct diff_words_data *diff_words;
 530         struct diff_options *opt;
 531         struct strbuf *header;
 532 };
 533
 534 static int count_lines(const char *data, int size)
 535 {
 536         int count, ch, completely_empty = 1, nl_just_seen = 0;
 537         count = 0;
 538         while (0 < size--) {
 539                 ch = *data++;
 540                 if (ch == '\n') {
 541                         count++;
 542                         nl_just_seen = 1;
 543                         completely_empty = 0;
 544                 }
 545                 else {
 546                         nl_just_seen = 0;
 547                         completely_empty = 0;
 548                 }
 549         }
 550         if (completely_empty)
 551                 return 0;
 552         if (!nl_just_seen)
 553                 count++; /* no trailing newline */
 554         return count;
 555 }
 556
 557 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 558                        struct diff_filespec *one)
 559 {
 560         if (!DIFF_FILE_VALID(one)) {
 561                 mf->ptr = (char *)""; /* does not matter */
 562                 mf->size = 0;
 563                 return 0;
 564         }
 565         else if (diff_populate_filespec(r, one, 0))
 566                 return -1;
 567
 568         mf->ptr = one->data;
 569         mf->size = one->size;
 570         return 0;
 571 }
 572
 573 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 574 static unsigned long diff_filespec_size(struct repository *r,
 575                                         struct diff_filespec *one)
 576 {
 577         if (!DIFF_FILE_VALID(one))
 578                 return 0;
 579         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 580         return one->size;
 581 }
 582
 583 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 584 {
 585         char *ptr = mf->ptr;
 586         long size = mf->size;
 587         int cnt = 0;
 588
 589         if (!size)
 590                 return cnt;
 591         ptr += size - 1; /* pointing at the very end */
 592         if (*ptr != '\n')
 593                 ; /* incomplete line */
 594         else
 595                 ptr--; /* skip the last LF */
 596         while (mf->ptr < ptr) {
 597                 char *prev_eol;
 598                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 599                         if (*prev_eol == '\n')
 600                                 break;
 601                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 602                         break;
 603                 cnt++;
 604                 ptr = prev_eol - 1;
 605         }
 606         return cnt;
 607 }
 608
 609 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 610                                struct emit_callback *ecbdata)
 611 {
 612         int l1, l2, at;
 613         unsigned ws_rule = ecbdata->ws_rule;
 614         l1 = count_trailing_blank(mf1, ws_rule);
 615         l2 = count_trailing_blank(mf2, ws_rule);
 616         if (l2 <= l1) {
 617                 ecbdata->blank_at_eof_in_preimage = 0;
 618                 ecbdata->blank_at_eof_in_postimage = 0;
 619                 return;
 620         }
 621         at = count_lines(mf1->ptr, mf1->size);
 622         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 623
 624         at = count_lines(mf2->ptr, mf2->size);
 625         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 626 }
 627
 628 static void emit_line_0(struct diff_options *o,
 629                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 630                         int first, const char *line, int len)
 631 {
 632         int has_trailing_newline, has_trailing_carriage_return;
 633         int needs_reset = 0; /* at the end of the line */
 634         FILE *file = o->file;
 635
 636         fputs(diff_line_prefix(o), file);
 637
 638         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 639         if (has_trailing_newline)
 640                 len--;
 641
 642         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 643         if (has_trailing_carriage_return)
 644                 len--;
 645
 646         if (!len && !first)
 647                 goto end_of_line;
 648
 649         if (reverse && want_color(o->use_color)) {
 650                 fputs(GIT_COLOR_REVERSE, file);
 651                 needs_reset = 1;
 652         }
 653
 654         if (set_sign) {
 655                 fputs(set_sign, file);
 656                 needs_reset = 1;
 657         }
 658
 659         if (first)
 660                 fputc(first, file);
 661
 662         if (!len)
 663                 goto end_of_line;
 664
 665         if (set) {
 666                 if (set_sign && set != set_sign)
 667                         fputs(reset, file);
 668                 fputs(set, file);
 669                 needs_reset = 1;
 670         }
 671         fwrite(line, len, 1, file);
 672         needs_reset = 1; /* 'line' may contain color codes. */
 673
 674 end_of_line:
 675         if (needs_reset)
 676                 fputs(reset, file);
 677         if (has_trailing_carriage_return)
 678                 fputc('\r', file);
 679         if (has_trailing_newline)
 680                 fputc('\n', file);
 681 }
 682
 683 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 684                       const char *line, int len)
 685 {
 686         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 687 }
 688
 689 enum diff_symbol {
 690         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 691         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 692         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 693         DIFF_SYMBOL_BINARY_DIFF_BODY,
 694         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 695         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 696         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 697         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 698         DIFF_SYMBOL_STATS_LINE,
 699         DIFF_SYMBOL_WORD_DIFF,
 700         DIFF_SYMBOL_STAT_SEP,
 701         DIFF_SYMBOL_SUMMARY,
 702         DIFF_SYMBOL_SUBMODULE_ADD,
 703         DIFF_SYMBOL_SUBMODULE_DEL,
 704         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 705         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 706         DIFF_SYMBOL_SUBMODULE_HEADER,
 707         DIFF_SYMBOL_SUBMODULE_ERROR,
 708         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 709         DIFF_SYMBOL_REWRITE_DIFF,
 710         DIFF_SYMBOL_BINARY_FILES,
 711         DIFF_SYMBOL_HEADER,
 712         DIFF_SYMBOL_FILEPAIR_PLUS,
 713         DIFF_SYMBOL_FILEPAIR_MINUS,
 714         DIFF_SYMBOL_WORDS_PORCELAIN,
 715         DIFF_SYMBOL_WORDS,
 716         DIFF_SYMBOL_CONTEXT,
 717         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 718         DIFF_SYMBOL_PLUS,
 719         DIFF_SYMBOL_MINUS,
 720         DIFF_SYMBOL_NO_LF_EOF,
 721         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 722         DIFF_SYMBOL_CONTEXT_MARKER,
 723         DIFF_SYMBOL_SEPARATOR
 724 };
 725 /*
 726  * Flags for content lines:
 727  * 0..12 are whitespace rules
 728  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 729  * 16 is marking if the line is blank at EOF
 730  */
 731 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 732 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 733 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 734 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 735 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 736
 737 /*
 738  * This struct is used when we need to buffer the output of the diff output.
 739  *
 740  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 741  * into the pre/post image file. This pointer could be a union with the
 742  * line pointer. By storing an offset into the file instead of the literal line,
 743  * we can decrease the memory footprint for the buffered output. At first we
 744  * may want to only have indirection for the content lines, but we could also
 745  * enhance the state for emitting prefabricated lines, e.g. the similarity
 746  * score line or hunk/file headers would only need to store a number or path
 747  * and then the output can be constructed later on depending on state.
 748  */
 749 struct emitted_diff_symbol {
 750         const char *line;
 751         int len;
 752         int flags;
 753         enum diff_symbol s;
 754 };
 755 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 756
 757 struct emitted_diff_symbols {
 758         struct emitted_diff_symbol *buf;
 759         int nr, alloc;
 760 };
 761 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 762
 763 static void append_emitted_diff_symbol(struct diff_options *o,
 764                                        struct emitted_diff_symbol *e)
 765 {
 766         struct emitted_diff_symbol *f;
 767
 768         ALLOC_GROW(o->emitted_symbols->buf,
 769                    o->emitted_symbols->nr + 1,
 770                    o->emitted_symbols->alloc);
 771         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 772
 773         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 774         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 775 }
 776
 777 struct moved_entry {
 778         struct hashmap_entry ent;
 779         const struct emitted_diff_symbol *es;
 780         struct moved_entry *next_line;
 781 };
 782
 783 /**
 784  * The struct ws_delta holds white space differences between moved lines, i.e.
 785  * between '+' and '-' lines that have been detected to be a move.
 786  * The string contains the difference in leading white spaces, before the
 787  * rest of the line is compared using the white space config for move
 788  * coloring. The current_longer indicates if the first string in the
 789  * comparision is longer than the second.
 790  */
 791 struct ws_delta {
 792         char *string;
 793         unsigned int current_longer : 1;
 794 };
 795 #define WS_DELTA_INIT { NULL, 0 }
 796
 797 struct moved_block {
 798         struct moved_entry *match;
 799         struct ws_delta wsd;
 800 };
 801
 802 static void moved_block_clear(struct moved_block *b)
 803 {
 804         FREE_AND_NULL(b->wsd.string);
 805         b->match = NULL;
 806 }
 807
 808 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 809                              const struct emitted_diff_symbol *b,
 810                              struct ws_delta *out)
 811 {
 812         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 813         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 814         int d = longer->len - shorter->len;
 815
 816         if (strncmp(longer->line + d, shorter->line, shorter->len))
 817                 return 0;
 818
 819         out->string = xmemdupz(longer->line, d);
 820         out->current_longer = (a == longer);
 821
 822         return 1;
 823 }
 824
 825 static int cmp_in_block_with_wsd(const struct diff_options *o,
 826                                  const struct moved_entry *cur,
 827                                  const struct moved_entry *match,
 828                                  struct moved_block *pmb,
 829                                  int n)
 830 {
 831         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 832         int al = cur->es->len, cl = l->len;
 833         const char *a = cur->es->line,
 834                    *b = match->es->line,
 835                    *c = l->line;
 836
 837         int wslen;
 838
 839         /*
 840          * We need to check if 'cur' is equal to 'match'.
 841          * As those are from the same (+/-) side, we do not need to adjust for
 842          * indent changes. However these were found using fuzzy matching
 843          * so we do have to check if they are equal.
 844          */
 845         if (strcmp(a, b))
 846                 return 1;
 847
 848         if (!pmb->wsd.string)
 849                 /*
 850                  * The white space delta is not active? This can happen
 851                  * when we exit early in this function.
 852                  */
 853                 return 1;
 854
 855         /*
 856          * The indent changes of the block are known and stored in
 857          * pmb->wsd; however we need to check if the indent changes of the
 858          * current line are still the same as before.
 859          *
 860          * To do so we need to compare 'l' to 'cur', adjusting the
 861          * one of them for the white spaces, depending which was longer.
 862          */
 863
 864         wslen = strlen(pmb->wsd.string);
 865         if (pmb->wsd.current_longer) {
 866                 c += wslen;
 867                 cl -= wslen;
 868         } else {
 869                 a += wslen;
 870                 al -= wslen;
 871         }
 872
 873         if (al != cl || memcmp(a, c, al))
 874                 return 1;
 875
 876         return 0;
 877 }
 878
 879 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 880                            const void *entry,
 881                            const void *entry_or_key,
 882                            const void *keydata)
 883 {
 884         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 885         const struct moved_entry *a = entry;
 886         const struct moved_entry *b = entry_or_key;
 887         unsigned flags = diffopt->color_moved_ws_handling
 888                          & XDF_WHITESPACE_FLAGS;
 889
 890         if (diffopt->color_moved_ws_handling &
 891             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 892                 /*
 893                  * As there is not specific white space config given,
 894                  * we'd need to check for a new block, so ignore all
 895                  * white space. The setup of the white space
 896                  * configuration for the next block is done else where
 897                  */
 898                 flags |= XDF_IGNORE_WHITESPACE;
 899
 900         return !xdiff_compare_lines(a->es->line, a->es->len,
 901                                     b->es->line, b->es->len,
 902                                     flags);
 903 }
 904
 905 static struct moved_entry *prepare_entry(struct diff_options *o,
 906                                          int line_no)
 907 {
 908         struct moved_entry *ret = xmalloc(sizeof(*ret));
 909         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 910         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 911
 912         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 913         ret->es = l;
 914         ret->next_line = NULL;
 915
 916         return ret;
 917 }
 918
 919 static void add_lines_to_move_detection(struct diff_options *o,
 920                                         struct hashmap *add_lines,
 921                                         struct hashmap *del_lines)
 922 {
 923         struct moved_entry *prev_line = NULL;
 924
 925         int n;
 926         for (n = 0; n < o->emitted_symbols->nr; n++) {
 927                 struct hashmap *hm;
 928                 struct moved_entry *key;
 929
 930                 switch (o->emitted_symbols->buf[n].s) {
 931                 case DIFF_SYMBOL_PLUS:
 932                         hm = add_lines;
 933                         break;
 934                 case DIFF_SYMBOL_MINUS:
 935                         hm = del_lines;
 936                         break;
 937                 default:
 938                         prev_line = NULL;
 939                         continue;
 940                 }
 941
 942                 key = prepare_entry(o, n);
 943                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 944                         prev_line->next_line = key;
 945
 946                 hashmap_add(hm, key);
 947                 prev_line = key;
 948         }
 949 }
 950
 951 static void pmb_advance_or_null(struct diff_options *o,
 952                                 struct moved_entry *match,
 953                                 struct hashmap *hm,
 954                                 struct moved_block *pmb,
 955                                 int pmb_nr)
 956 {
 957         int i;
 958         for (i = 0; i < pmb_nr; i++) {
 959                 struct moved_entry *prev = pmb[i].match;
 960                 struct moved_entry *cur = (prev && prev->next_line) ?
 961                                 prev->next_line : NULL;
 962                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 963                         pmb[i].match = cur;
 964                 } else {
 965                         pmb[i].match = NULL;
 966                 }
 967         }
 968 }
 969
 970 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 971                                             struct moved_entry *match,
 972                                             struct hashmap *hm,
 973                                             struct moved_block *pmb,
 974                                             int pmb_nr, int n)
 975 {
 976         int i;
 977         char *got_match = xcalloc(1, pmb_nr);
 978
 979         for (; match; match = hashmap_get_next(hm, match)) {
 980                 for (i = 0; i < pmb_nr; i++) {
 981                         struct moved_entry *prev = pmb[i].match;
 982                         struct moved_entry *cur = (prev && prev->next_line) ?
 983                                         prev->next_line : NULL;
 984                         if (!cur)
 985                                 continue;
 986                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
 987                                 got_match[i] |= 1;
 988                 }
 989         }
 990
 991         for (i = 0; i < pmb_nr; i++) {
 992                 if (got_match[i]) {
 993                         /* Advance to the next line */
 994                         pmb[i].match = pmb[i].match->next_line;
 995                 } else {
 996                         moved_block_clear(&pmb[i]);
 997                 }
 998         }
 999
1000         free(got_match);
1001 }
1002
1003 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1004                                          int pmb_nr)
1005 {
1006         int lp, rp;
1007
1008         /* Shrink the set of potential block to the remaining running */
1009         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1010                 while (lp < pmb_nr && pmb[lp].match)
1011                         lp++;
1012                 /* lp points at the first NULL now */
1013
1014                 while (rp > -1 && !pmb[rp].match)
1015                         rp--;
1016                 /* rp points at the last non-NULL */
1017
1018                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1019                         pmb[lp] = pmb[rp];
1020                         pmb[rp].match = NULL;
1021                         pmb[rp].wsd.string = NULL;
1022                         rp--;
1023                         lp++;
1024                 }
1025         }
1026
1027         /* Remember the number of running sets */
1028         return rp + 1;
1029 }
1030
1031 /*
1032  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1033  *
1034  * Otherwise, if the last block has fewer alphanumeric characters than
1035  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1036  * that block.
1037  *
1038  * The last block consists of the (n - block_length)'th line up to but not
1039  * including the nth line.
1040  *
1041  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1042  * Think of a way to unify them.
1043  */
1044 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1045 {
1046         int i, alnum_count = 0;
1047         if (o->color_moved == COLOR_MOVED_PLAIN)
1048                 return;
1049         for (i = 1; i < block_length + 1; i++) {
1050                 const char *c = o->emitted_symbols->buf[n - i].line;
1051                 for (; *c; c++) {
1052                         if (!isalnum(*c))
1053                                 continue;
1054                         alnum_count++;
1055                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1056                                 return;
1057                 }
1058         }
1059         for (i = 1; i < block_length + 1; i++)
1060                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1061 }
1062
1063 /* Find blocks of moved code, delegate actual coloring decision to helper */
1064 static void mark_color_as_moved(struct diff_options *o,
1065                                 struct hashmap *add_lines,
1066                                 struct hashmap *del_lines)
1067 {
1068         struct moved_block *pmb = NULL; /* potentially moved blocks */
1069         int pmb_nr = 0, pmb_alloc = 0;
1070         int n, flipped_block = 1, block_length = 0;
1071
1072
1073         for (n = 0; n < o->emitted_symbols->nr; n++) {
1074                 struct hashmap *hm = NULL;
1075                 struct moved_entry *key;
1076                 struct moved_entry *match = NULL;
1077                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1078
1079                 switch (l->s) {
1080                 case DIFF_SYMBOL_PLUS:
1081                         hm = del_lines;
1082                         key = prepare_entry(o, n);
1083                         match = hashmap_get(hm, key, NULL);
1084                         free(key);
1085                         break;
1086                 case DIFF_SYMBOL_MINUS:
1087                         hm = add_lines;
1088                         key = prepare_entry(o, n);
1089                         match = hashmap_get(hm, key, NULL);
1090                         free(key);
1091                         break;
1092                 default:
1093                         flipped_block = 1;
1094                 }
1095
1096                 if (!match) {
1097                         int i;
1098
1099                         adjust_last_block(o, n, block_length);
1100                         for(i = 0; i < pmb_nr; i++)
1101                                 moved_block_clear(&pmb[i]);
1102                         pmb_nr = 0;
1103                         block_length = 0;
1104                         continue;
1105                 }
1106
1107                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1108
1109                 if (o->color_moved == COLOR_MOVED_PLAIN)
1110                         continue;
1111
1112                 if (o->color_moved_ws_handling &
1113                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1114                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1115                 else
1116                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1117
1118                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1119
1120                 if (pmb_nr == 0) {
1121                         /*
1122                          * The current line is the start of a new block.
1123                          * Setup the set of potential blocks.
1124                          */
1125                         for (; match; match = hashmap_get_next(hm, match)) {
1126                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1127                                 if (o->color_moved_ws_handling &
1128                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1129                                         if (compute_ws_delta(l, match->es,
1130                                                              &pmb[pmb_nr].wsd))
1131                                                 pmb[pmb_nr++].match = match;
1132                                 } else {
1133                                         pmb[pmb_nr].wsd.string = NULL;
1134                                         pmb[pmb_nr++].match = match;
1135                                 }
1136                         }
1137
1138                         flipped_block = (flipped_block + 1) % 2;
1139
1140                         adjust_last_block(o, n, block_length);
1141                         block_length = 0;
1142                 }
1143
1144                 block_length++;
1145
1146                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1147                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1148         }
1149         adjust_last_block(o, n, block_length);
1150
1151         for(n = 0; n < pmb_nr; n++)
1152                 moved_block_clear(&pmb[n]);
1153         free(pmb);
1154 }
1155
1156 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1157   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1158 static void dim_moved_lines(struct diff_options *o)
1159 {
1160         int n;
1161         for (n = 0; n < o->emitted_symbols->nr; n++) {
1162                 struct emitted_diff_symbol *prev = (n != 0) ?
1163                                 &o->emitted_symbols->buf[n - 1] : NULL;
1164                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1165                 struct emitted_diff_symbol *next =
1166                                 (n < o->emitted_symbols->nr - 1) ?
1167                                 &o->emitted_symbols->buf[n + 1] : NULL;
1168
1169                 /* Not a plus or minus line? */
1170                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1171                         continue;
1172
1173                 /* Not a moved line? */
1174                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1175                         continue;
1176
1177                 /*
1178                  * If prev or next are not a plus or minus line,
1179                  * pretend they don't exist
1180                  */
1181                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1182                             prev->s != DIFF_SYMBOL_MINUS)
1183                         prev = NULL;
1184                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1185                             next->s != DIFF_SYMBOL_MINUS)
1186                         next = NULL;
1187
1188                 /* Inside a block? */
1189                 if ((prev &&
1190                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1191                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1192                     (next &&
1193                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1194                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1195                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1196                         continue;
1197                 }
1198
1199                 /* Check if we are at an interesting bound: */
1200                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1201                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1202                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1203                         continue;
1204                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1205                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1206                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1207                         continue;
1208
1209                 /*
1210                  * The boundary to prev and next are not interesting,
1211                  * so this line is not interesting as a whole
1212                  */
1213                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1214         }
1215 }
1216
1217 static void emit_line_ws_markup(struct diff_options *o,
1218                                 const char *set_sign, const char *set,
1219                                 const char *reset,
1220                                 int sign_index, const char *line, int len,
1221                                 unsigned ws_rule, int blank_at_eof)
1222 {
1223         const char *ws = NULL;
1224         int sign = o->output_indicators[sign_index];
1225
1226         if (o->ws_error_highlight & ws_rule) {
1227                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1228                 if (!*ws)
1229                         ws = NULL;
1230         }
1231
1232         if (!ws && !set_sign)
1233                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1234         else if (!ws) {
1235                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1236         } else if (blank_at_eof)
1237                 /* Blank line at EOF - paint '+' as well */
1238                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1239         else {
1240                 /* Emit just the prefix, then the rest. */
1241                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1242                             sign, "", 0);
1243                 ws_check_emit(line, len, ws_rule,
1244                               o->file, set, reset, ws);
1245         }
1246 }
1247
1248 static void emit_diff_symbol_from_struct(struct diff_options *o,
1249                                          struct emitted_diff_symbol *eds)
1250 {
1251         static const char *nneof = " No newline at end of file\n";
1252         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1253         struct strbuf sb = STRBUF_INIT;
1254
1255         enum diff_symbol s = eds->s;
1256         const char *line = eds->line;
1257         int len = eds->len;
1258         unsigned flags = eds->flags;
1259
1260         switch (s) {
1261         case DIFF_SYMBOL_NO_LF_EOF:
1262                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1263                 reset = diff_get_color_opt(o, DIFF_RESET);
1264                 putc('\n', o->file);
1265                 emit_line_0(o, context, NULL, 0, reset, '\\',
1266                             nneof, strlen(nneof));
1267                 break;
1268         case DIFF_SYMBOL_SUBMODULE_HEADER:
1269         case DIFF_SYMBOL_SUBMODULE_ERROR:
1270         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1271         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1272         case DIFF_SYMBOL_SUMMARY:
1273         case DIFF_SYMBOL_STATS_LINE:
1274         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1275         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1276                 emit_line(o, "", "", line, len);
1277                 break;
1278         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1279         case DIFF_SYMBOL_CONTEXT_MARKER:
1280                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1281                 reset = diff_get_color_opt(o, DIFF_RESET);
1282                 emit_line(o, context, reset, line, len);
1283                 break;
1284         case DIFF_SYMBOL_SEPARATOR:
1285                 fprintf(o->file, "%s%c",
1286                         diff_line_prefix(o),
1287                         o->line_termination);
1288                 break;
1289         case DIFF_SYMBOL_CONTEXT:
1290                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1291                 reset = diff_get_color_opt(o, DIFF_RESET);
1292                 set_sign = NULL;
1293                 if (o->flags.dual_color_diffed_diffs) {
1294                         char c = !len ? 0 : line[0];
1295
1296                         if (c == '+')
1297                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1298                         else if (c == '@')
1299                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1300                         else if (c == '-')
1301                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1302                 }
1303                 emit_line_ws_markup(o, set_sign, set, reset,
1304                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1305                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1306                 break;
1307         case DIFF_SYMBOL_PLUS:
1308                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1309                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1310                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1311                 case DIFF_SYMBOL_MOVED_LINE |
1312                      DIFF_SYMBOL_MOVED_LINE_ALT |
1313                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1314                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1315                         break;
1316                 case DIFF_SYMBOL_MOVED_LINE |
1317                      DIFF_SYMBOL_MOVED_LINE_ALT:
1318                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1319                         break;
1320                 case DIFF_SYMBOL_MOVED_LINE |
1321                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1322                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1323                         break;
1324                 case DIFF_SYMBOL_MOVED_LINE:
1325                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1326                         break;
1327                 default:
1328                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1329                 }
1330                 reset = diff_get_color_opt(o, DIFF_RESET);
1331                 if (!o->flags.dual_color_diffed_diffs)
1332                         set_sign = NULL;
1333                 else {
1334                         char c = !len ? 0 : line[0];
1335
1336                         set_sign = set;
1337                         if (c == '-')
1338                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1339                         else if (c == '@')
1340                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1341                         else if (c == '+')
1342                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1343                         else
1344                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1345                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1346                 }
1347                 emit_line_ws_markup(o, set_sign, set, reset,
1348                                     OUTPUT_INDICATOR_NEW, line, len,
1349                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1350                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1351                 break;
1352         case DIFF_SYMBOL_MINUS:
1353                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1354                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1355                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1356                 case DIFF_SYMBOL_MOVED_LINE |
1357                      DIFF_SYMBOL_MOVED_LINE_ALT |
1358                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1359                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1360                         break;
1361                 case DIFF_SYMBOL_MOVED_LINE |
1362                      DIFF_SYMBOL_MOVED_LINE_ALT:
1363                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1364                         break;
1365                 case DIFF_SYMBOL_MOVED_LINE |
1366                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1367                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1368                         break;
1369                 case DIFF_SYMBOL_MOVED_LINE:
1370                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1371                         break;
1372                 default:
1373                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1374                 }
1375                 reset = diff_get_color_opt(o, DIFF_RESET);
1376                 if (!o->flags.dual_color_diffed_diffs)
1377                         set_sign = NULL;
1378                 else {
1379                         char c = !len ? 0 : line[0];
1380
1381                         set_sign = set;
1382                         if (c == '+')
1383                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1384                         else if (c == '@')
1385                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1386                         else if (c == '-')
1387                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1388                         else
1389                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1390                 }
1391                 emit_line_ws_markup(o, set_sign, set, reset,
1392                                     OUTPUT_INDICATOR_OLD, line, len,
1393                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1394                 break;
1395         case DIFF_SYMBOL_WORDS_PORCELAIN:
1396                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1397                 reset = diff_get_color_opt(o, DIFF_RESET);
1398                 emit_line(o, context, reset, line, len);
1399                 fputs("~\n", o->file);
1400                 break;
1401         case DIFF_SYMBOL_WORDS:
1402                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1403                 reset = diff_get_color_opt(o, DIFF_RESET);
1404                 /*
1405                  * Skip the prefix character, if any.  With
1406                  * diff_suppress_blank_empty, there may be
1407                  * none.
1408                  */
1409                 if (line[0] != '\n') {
1410                         line++;
1411                         len--;
1412                 }
1413                 emit_line(o, context, reset, line, len);
1414                 break;
1415         case DIFF_SYMBOL_FILEPAIR_PLUS:
1416                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1417                 reset = diff_get_color_opt(o, DIFF_RESET);
1418                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1419                         line, reset,
1420                         strchr(line, ' ') ? "\t" : "");
1421                 break;
1422         case DIFF_SYMBOL_FILEPAIR_MINUS:
1423                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1424                 reset = diff_get_color_opt(o, DIFF_RESET);
1425                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1426                         line, reset,
1427                         strchr(line, ' ') ? "\t" : "");
1428                 break;
1429         case DIFF_SYMBOL_BINARY_FILES:
1430         case DIFF_SYMBOL_HEADER:
1431                 fprintf(o->file, "%s", line);
1432                 break;
1433         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1434                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1435                 break;
1436         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1437                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1438                 break;
1439         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1440                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1441                 break;
1442         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1443                 fputs(diff_line_prefix(o), o->file);
1444                 fputc('\n', o->file);
1445                 break;
1446         case DIFF_SYMBOL_REWRITE_DIFF:
1447                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1448                 reset = diff_get_color_opt(o, DIFF_RESET);
1449                 emit_line(o, fraginfo, reset, line, len);
1450                 break;
1451         case DIFF_SYMBOL_SUBMODULE_ADD:
1452                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1453                 reset = diff_get_color_opt(o, DIFF_RESET);
1454                 emit_line(o, set, reset, line, len);
1455                 break;
1456         case DIFF_SYMBOL_SUBMODULE_DEL:
1457                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1458                 reset = diff_get_color_opt(o, DIFF_RESET);
1459                 emit_line(o, set, reset, line, len);
1460                 break;
1461         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1462                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1463                         diff_line_prefix(o), line);
1464                 break;
1465         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1466                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1467                         diff_line_prefix(o), line);
1468                 break;
1469         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1470                 emit_line(o, "", "", " 0 files changed\n",
1471                           strlen(" 0 files changed\n"));
1472                 break;
1473         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1474                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1475                 break;
1476         case DIFF_SYMBOL_WORD_DIFF:
1477                 fprintf(o->file, "%.*s", len, line);
1478                 break;
1479         case DIFF_SYMBOL_STAT_SEP:
1480                 fputs(o->stat_sep, o->file);
1481                 break;
1482         default:
1483                 BUG("unknown diff symbol");
1484         }
1485         strbuf_release(&sb);
1486 }
1487
1488 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1489                              const char *line, int len, unsigned flags)
1490 {
1491         struct emitted_diff_symbol e = {line, len, flags, s};
1492
1493         if (o->emitted_symbols)
1494                 append_emitted_diff_symbol(o, &e);
1495         else
1496                 emit_diff_symbol_from_struct(o, &e);
1497 }
1498
1499 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1500 {
1501         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1502 }
1503
1504 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1505 {
1506         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1507 }
1508
1509 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1510 {
1511         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1512                          path, strlen(path), 0);
1513 }
1514
1515 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1516 {
1517         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1518                          path, strlen(path), 0);
1519 }
1520
1521 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1522 {
1523         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1524                          header, strlen(header), 0);
1525 }
1526
1527 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1528 {
1529         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1530 }
1531
1532 void diff_emit_submodule_pipethrough(struct diff_options *o,
1533                                      const char *line, int len)
1534 {
1535         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1536 }
1537
1538 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1539 {
1540         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1541               ecbdata->blank_at_eof_in_preimage &&
1542               ecbdata->blank_at_eof_in_postimage &&
1543               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1544               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1545                 return 0;
1546         return ws_blank_line(line, len, ecbdata->ws_rule);
1547 }
1548
1549 static void emit_add_line(const char *reset,
1550                           struct emit_callback *ecbdata,
1551                           const char *line, int len)
1552 {
1553         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1554         if (new_blank_line_at_eof(ecbdata, line, len))
1555                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1556
1557         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1558 }
1559
1560 static void emit_del_line(const char *reset,
1561                           struct emit_callback *ecbdata,
1562                           const char *line, int len)
1563 {
1564         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1565         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1566 }
1567
1568 static void emit_context_line(const char *reset,
1569                               struct emit_callback *ecbdata,
1570                               const char *line, int len)
1571 {
1572         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1573         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1574 }
1575
1576 static void emit_hunk_header(struct emit_callback *ecbdata,
1577                              const char *line, int len)
1578 {
1579         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1580         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1581         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1582         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1583         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1584         static const char atat[2] = { '@', '@' };
1585         const char *cp, *ep;
1586         struct strbuf msgbuf = STRBUF_INIT;
1587         int org_len = len;
1588         int i = 1;
1589
1590         /*
1591          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1592          * it always is at least 10 bytes long.
1593          */
1594         if (len < 10 ||
1595             memcmp(line, atat, 2) ||
1596             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1597                 emit_diff_symbol(ecbdata->opt,
1598                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1599                 return;
1600         }
1601         ep += 2; /* skip over @@ */
1602
1603         /* The hunk header in fraginfo color */
1604         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1605                 strbuf_addstr(&msgbuf, reverse);
1606         strbuf_addstr(&msgbuf, frag);
1607         strbuf_add(&msgbuf, line, ep - line);
1608         strbuf_addstr(&msgbuf, reset);
1609
1610         /*
1611          * trailing "\r\n"
1612          */
1613         for ( ; i < 3; i++)
1614                 if (line[len - i] == '\r' || line[len - i] == '\n')
1615                         len--;
1616
1617         /* blank before the func header */
1618         for (cp = ep; ep - line < len; ep++)
1619                 if (*ep != ' ' && *ep != '\t')
1620                         break;
1621         if (ep != cp) {
1622                 strbuf_addstr(&msgbuf, context);
1623                 strbuf_add(&msgbuf, cp, ep - cp);
1624                 strbuf_addstr(&msgbuf, reset);
1625         }
1626
1627         if (ep < line + len) {
1628                 strbuf_addstr(&msgbuf, func);
1629                 strbuf_add(&msgbuf, ep, line + len - ep);
1630                 strbuf_addstr(&msgbuf, reset);
1631         }
1632
1633         strbuf_add(&msgbuf, line + len, org_len - len);
1634         strbuf_complete_line(&msgbuf);
1635         emit_diff_symbol(ecbdata->opt,
1636                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1637         strbuf_release(&msgbuf);
1638 }
1639
1640 static struct diff_tempfile *claim_diff_tempfile(void)
1641 {
1642         int i;
1643         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1644                 if (!diff_temp[i].name)
1645                         return diff_temp + i;
1646         BUG("diff is failing to clean up its tempfiles");
1647 }
1648
1649 static void remove_tempfile(void)
1650 {
1651         int i;
1652         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1653                 if (is_tempfile_active(diff_temp[i].tempfile))
1654                         delete_tempfile(&diff_temp[i].tempfile);
1655                 diff_temp[i].name = NULL;
1656         }
1657 }
1658
1659 static void add_line_count(struct strbuf *out, int count)
1660 {
1661         switch (count) {
1662         case 0:
1663                 strbuf_addstr(out, "0,0");
1664                 break;
1665         case 1:
1666                 strbuf_addstr(out, "1");
1667                 break;
1668         default:
1669                 strbuf_addf(out, "1,%d", count);
1670                 break;
1671         }
1672 }
1673
1674 static void emit_rewrite_lines(struct emit_callback *ecb,
1675                                int prefix, const char *data, int size)
1676 {
1677         const char *endp = NULL;
1678         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1679
1680         while (0 < size) {
1681                 int len;
1682
1683                 endp = memchr(data, '\n', size);
1684                 len = endp ? (endp - data + 1) : size;
1685                 if (prefix != '+') {
1686                         ecb->lno_in_preimage++;
1687                         emit_del_line(reset, ecb, data, len);
1688                 } else {
1689                         ecb->lno_in_postimage++;
1690                         emit_add_line(reset, ecb, data, len);
1691                 }
1692                 size -= len;
1693                 data += len;
1694         }
1695         if (!endp)
1696                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1697 }
1698
1699 static void emit_rewrite_diff(const char *name_a,
1700                               const char *name_b,
1701                               struct diff_filespec *one,
1702                               struct diff_filespec *two,
1703                               struct userdiff_driver *textconv_one,
1704                               struct userdiff_driver *textconv_two,
1705                               struct diff_options *o)
1706 {
1707         int lc_a, lc_b;
1708         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1709         const char *a_prefix, *b_prefix;
1710         char *data_one, *data_two;
1711         size_t size_one, size_two;
1712         struct emit_callback ecbdata;
1713         struct strbuf out = STRBUF_INIT;
1714
1715         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1716                 a_prefix = o->b_prefix;
1717                 b_prefix = o->a_prefix;
1718         } else {
1719                 a_prefix = o->a_prefix;
1720                 b_prefix = o->b_prefix;
1721         }
1722
1723         name_a += (*name_a == '/');
1724         name_b += (*name_b == '/');
1725
1726         strbuf_reset(&a_name);
1727         strbuf_reset(&b_name);
1728         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1729         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1730
1731         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1732         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1733
1734         memset(&ecbdata, 0, sizeof(ecbdata));
1735         ecbdata.color_diff = want_color(o->use_color);
1736         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1737         ecbdata.opt = o;
1738         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1739                 mmfile_t mf1, mf2;
1740                 mf1.ptr = (char *)data_one;
1741                 mf2.ptr = (char *)data_two;
1742                 mf1.size = size_one;
1743                 mf2.size = size_two;
1744                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1745         }
1746         ecbdata.lno_in_preimage = 1;
1747         ecbdata.lno_in_postimage = 1;
1748
1749         lc_a = count_lines(data_one, size_one);
1750         lc_b = count_lines(data_two, size_two);
1751
1752         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1753                          a_name.buf, a_name.len, 0);
1754         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1755                          b_name.buf, b_name.len, 0);
1756
1757         strbuf_addstr(&out, "@@ -");
1758         if (!o->irreversible_delete)
1759                 add_line_count(&out, lc_a);
1760         else
1761                 strbuf_addstr(&out, "?,?");
1762         strbuf_addstr(&out, " +");
1763         add_line_count(&out, lc_b);
1764         strbuf_addstr(&out, " @@\n");
1765         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1766         strbuf_release(&out);
1767
1768         if (lc_a && !o->irreversible_delete)
1769                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1770         if (lc_b)
1771                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1772         if (textconv_one)
1773                 free((char *)data_one);
1774         if (textconv_two)
1775                 free((char *)data_two);
1776 }
1777
1778 struct diff_words_buffer {
1779         mmfile_t text;
1780         unsigned long alloc;
1781         struct diff_words_orig {
1782                 const char *begin, *end;
1783         } *orig;
1784         int orig_nr, orig_alloc;
1785 };
1786
1787 static void diff_words_append(char *line, unsigned long len,
1788                 struct diff_words_buffer *buffer)
1789 {
1790         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1791         line++;
1792         len--;
1793         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1794         buffer->text.size += len;
1795         buffer->text.ptr[buffer->text.size] = '\0';
1796 }
1797
1798 struct diff_words_style_elem {
1799         const char *prefix;
1800         const char *suffix;
1801         const char *color; /* NULL; filled in by the setup code if
1802                             * color is enabled */
1803 };
1804
1805 struct diff_words_style {
1806         enum diff_words_type type;
1807         struct diff_words_style_elem new_word, old_word, ctx;
1808         const char *newline;
1809 };
1810
1811 static struct diff_words_style diff_words_styles[] = {
1812         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1813         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1814         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1815 };
1816
1817 struct diff_words_data {
1818         struct diff_words_buffer minus, plus;
1819         const char *current_plus;
1820         int last_minus;
1821         struct diff_options *opt;
1822         regex_t *word_regex;
1823         enum diff_words_type type;
1824         struct diff_words_style *style;
1825 };
1826
1827 static int fn_out_diff_words_write_helper(struct diff_options *o,
1828                                           struct diff_words_style_elem *st_el,
1829                                           const char *newline,
1830                                           size_t count, const char *buf)
1831 {
1832         int print = 0;
1833         struct strbuf sb = STRBUF_INIT;
1834
1835         while (count) {
1836                 char *p = memchr(buf, '\n', count);
1837                 if (print)
1838                         strbuf_addstr(&sb, diff_line_prefix(o));
1839
1840                 if (p != buf) {
1841                         const char *reset = st_el->color && *st_el->color ?
1842                                             GIT_COLOR_RESET : NULL;
1843                         if (st_el->color && *st_el->color)
1844                                 strbuf_addstr(&sb, st_el->color);
1845                         strbuf_addstr(&sb, st_el->prefix);
1846                         strbuf_add(&sb, buf, p ? p - buf : count);
1847                         strbuf_addstr(&sb, st_el->suffix);
1848                         if (reset)
1849                                 strbuf_addstr(&sb, reset);
1850                 }
1851                 if (!p)
1852                         goto out;
1853
1854                 strbuf_addstr(&sb, newline);
1855                 count -= p + 1 - buf;
1856                 buf = p + 1;
1857                 print = 1;
1858                 if (count) {
1859                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1860                                          sb.buf, sb.len, 0);
1861                         strbuf_reset(&sb);
1862                 }
1863         }
1864
1865 out:
1866         if (sb.len)
1867                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1868                                  sb.buf, sb.len, 0);
1869         strbuf_release(&sb);
1870         return 0;
1871 }
1872
1873 /*
1874  * '--color-words' algorithm can be described as:
1875  *
1876  *   1. collect the minus/plus lines of a diff hunk, divided into
1877  *      minus-lines and plus-lines;
1878  *
1879  *   2. break both minus-lines and plus-lines into words and
1880  *      place them into two mmfile_t with one word for each line;
1881  *
1882  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1883  *
1884  * And for the common parts of the both file, we output the plus side text.
1885  * diff_words->current_plus is used to trace the current position of the plus file
1886  * which printed. diff_words->last_minus is used to trace the last minus word
1887  * printed.
1888  *
1889  * For '--graph' to work with '--color-words', we need to output the graph prefix
1890  * on each line of color words output. Generally, there are two conditions on
1891  * which we should output the prefix.
1892  *
1893  *   1. diff_words->last_minus == 0 &&
1894  *      diff_words->current_plus == diff_words->plus.text.ptr
1895  *
1896  *      that is: the plus text must start as a new line, and if there is no minus
1897  *      word printed, a graph prefix must be printed.
1898  *
1899  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1900  *      *(diff_words->current_plus - 1) == '\n'
1901  *
1902  *      that is: a graph prefix must be printed following a '\n'
1903  */
1904 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1905 {
1906         if ((diff_words->last_minus == 0 &&
1907                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1908                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1909                 *(diff_words->current_plus - 1) == '\n')) {
1910                 return 1;
1911         } else {
1912                 return 0;
1913         }
1914 }
1915
1916 static void fn_out_diff_words_aux(void *priv,
1917                                   long minus_first, long minus_len,
1918                                   long plus_first, long plus_len,
1919                                   const char *func, long funclen)
1920 {
1921         struct diff_words_data *diff_words = priv;
1922         struct diff_words_style *style = diff_words->style;
1923         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1924         struct diff_options *opt = diff_words->opt;
1925         const char *line_prefix;
1926
1927         assert(opt);
1928         line_prefix = diff_line_prefix(opt);
1929
1930         /* POSIX requires that first be decremented by one if len == 0... */
1931         if (minus_len) {
1932                 minus_begin = diff_words->minus.orig[minus_first].begin;
1933                 minus_end =
1934                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1935         } else
1936                 minus_begin = minus_end =
1937                         diff_words->minus.orig[minus_first].end;
1938
1939         if (plus_len) {
1940                 plus_begin = diff_words->plus.orig[plus_first].begin;
1941                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1942         } else
1943                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1944
1945         if (color_words_output_graph_prefix(diff_words)) {
1946                 fputs(line_prefix, diff_words->opt->file);
1947         }
1948         if (diff_words->current_plus != plus_begin) {
1949                 fn_out_diff_words_write_helper(diff_words->opt,
1950                                 &style->ctx, style->newline,
1951                                 plus_begin - diff_words->current_plus,
1952                                 diff_words->current_plus);
1953         }
1954         if (minus_begin != minus_end) {
1955                 fn_out_diff_words_write_helper(diff_words->opt,
1956                                 &style->old_word, style->newline,
1957                                 minus_end - minus_begin, minus_begin);
1958         }
1959         if (plus_begin != plus_end) {
1960                 fn_out_diff_words_write_helper(diff_words->opt,
1961                                 &style->new_word, style->newline,
1962                                 plus_end - plus_begin, plus_begin);
1963         }
1964
1965         diff_words->current_plus = plus_end;
1966         diff_words->last_minus = minus_first;
1967 }
1968
1969 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1970 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1971                 int *begin, int *end)
1972 {
1973         if (word_regex && *begin < buffer->size) {
1974                 regmatch_t match[1];
1975                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1976                                  buffer->size - *begin, 1, match, 0)) {
1977                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1978                                         '\n', match[0].rm_eo - match[0].rm_so);
1979                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1980                         *begin += match[0].rm_so;
1981                         return *begin >= *end;
1982                 }
1983                 return -1;
1984         }
1985
1986         /* find the next word */
1987         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1988                 (*begin)++;
1989         if (*begin >= buffer->size)
1990                 return -1;
1991
1992         /* find the end of the word */
1993         *end = *begin + 1;
1994         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1995                 (*end)++;
1996
1997         return 0;
1998 }
1999
2000 /*
2001  * This function splits the words in buffer->text, stores the list with
2002  * newline separator into out, and saves the offsets of the original words
2003  * in buffer->orig.
2004  */
2005 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2006                 regex_t *word_regex)
2007 {
2008         int i, j;
2009         long alloc = 0;
2010
2011         out->size = 0;
2012         out->ptr = NULL;
2013
2014         /* fake an empty "0th" word */
2015         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2016         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2017         buffer->orig_nr = 1;
2018
2019         for (i = 0; i < buffer->text.size; i++) {
2020                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2021                         return;
2022
2023                 /* store original boundaries */
2024                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2025                                 buffer->orig_alloc);
2026                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2027                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2028                 buffer->orig_nr++;
2029
2030                 /* store one word */
2031                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2032                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2033                 out->ptr[out->size + j - i] = '\n';
2034                 out->size += j - i + 1;
2035
2036                 i = j - 1;
2037         }
2038 }
2039
2040 /* this executes the word diff on the accumulated buffers */
2041 static void diff_words_show(struct diff_words_data *diff_words)
2042 {
2043         xpparam_t xpp;
2044         xdemitconf_t xecfg;
2045         mmfile_t minus, plus;
2046         struct diff_words_style *style = diff_words->style;
2047
2048         struct diff_options *opt = diff_words->opt;
2049         const char *line_prefix;
2050
2051         assert(opt);
2052         line_prefix = diff_line_prefix(opt);
2053
2054         /* special case: only removal */
2055         if (!diff_words->plus.text.size) {
2056                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2057                                  line_prefix, strlen(line_prefix), 0);
2058                 fn_out_diff_words_write_helper(diff_words->opt,
2059                         &style->old_word, style->newline,
2060                         diff_words->minus.text.size,
2061                         diff_words->minus.text.ptr);
2062                 diff_words->minus.text.size = 0;
2063                 return;
2064         }
2065
2066         diff_words->current_plus = diff_words->plus.text.ptr;
2067         diff_words->last_minus = 0;
2068
2069         memset(&xpp, 0, sizeof(xpp));
2070         memset(&xecfg, 0, sizeof(xecfg));
2071         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2072         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2073         xpp.flags = 0;
2074         /* as only the hunk header will be parsed, we need a 0-context */
2075         xecfg.ctxlen = 0;
2076         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2077                           diff_words, &xpp, &xecfg))
2078                 die("unable to generate word diff");
2079         free(minus.ptr);
2080         free(plus.ptr);
2081         if (diff_words->current_plus != diff_words->plus.text.ptr +
2082                         diff_words->plus.text.size) {
2083                 if (color_words_output_graph_prefix(diff_words))
2084                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2085                                          line_prefix, strlen(line_prefix), 0);
2086                 fn_out_diff_words_write_helper(diff_words->opt,
2087                         &style->ctx, style->newline,
2088                         diff_words->plus.text.ptr + diff_words->plus.text.size
2089                         - diff_words->current_plus, diff_words->current_plus);
2090         }
2091         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2092 }
2093
2094 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2095 static void diff_words_flush(struct emit_callback *ecbdata)
2096 {
2097         struct diff_options *wo = ecbdata->diff_words->opt;
2098
2099         if (ecbdata->diff_words->minus.text.size ||
2100             ecbdata->diff_words->plus.text.size)
2101                 diff_words_show(ecbdata->diff_words);
2102
2103         if (wo->emitted_symbols) {
2104                 struct diff_options *o = ecbdata->opt;
2105                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2106                 int i;
2107
2108                 /*
2109                  * NEEDSWORK:
2110                  * Instead of appending each, concat all words to a line?
2111                  */
2112                 for (i = 0; i < wol->nr; i++)
2113                         append_emitted_diff_symbol(o, &wol->buf[i]);
2114
2115                 for (i = 0; i < wol->nr; i++)
2116                         free((void *)wol->buf[i].line);
2117
2118                 wol->nr = 0;
2119         }
2120 }
2121
2122 static void diff_filespec_load_driver(struct diff_filespec *one,
2123                                       struct index_state *istate)
2124 {
2125         /* Use already-loaded driver */
2126         if (one->driver)
2127                 return;
2128
2129         if (S_ISREG(one->mode))
2130                 one->driver = userdiff_find_by_path(istate, one->path);
2131
2132         /* Fallback to default settings */
2133         if (!one->driver)
2134                 one->driver = userdiff_find_by_name("default");
2135 }
2136
2137 static const char *userdiff_word_regex(struct diff_filespec *one,
2138                                        struct index_state *istate)
2139 {
2140         diff_filespec_load_driver(one, istate);
2141         return one->driver->word_regex;
2142 }
2143
2144 static void init_diff_words_data(struct emit_callback *ecbdata,
2145                                  struct diff_options *orig_opts,
2146                                  struct diff_filespec *one,
2147                                  struct diff_filespec *two)
2148 {
2149         int i;
2150         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2151         memcpy(o, orig_opts, sizeof(struct diff_options));
2152
2153         ecbdata->diff_words =
2154                 xcalloc(1, sizeof(struct diff_words_data));
2155         ecbdata->diff_words->type = o->word_diff;
2156         ecbdata->diff_words->opt = o;
2157
2158         if (orig_opts->emitted_symbols)
2159                 o->emitted_symbols =
2160                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2161
2162         if (!o->word_regex)
2163                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2164         if (!o->word_regex)
2165                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2166         if (!o->word_regex)
2167                 o->word_regex = diff_word_regex_cfg;
2168         if (o->word_regex) {
2169                 ecbdata->diff_words->word_regex = (regex_t *)
2170                         xmalloc(sizeof(regex_t));
2171                 if (regcomp(ecbdata->diff_words->word_regex,
2172                             o->word_regex,
2173                             REG_EXTENDED | REG_NEWLINE))
2174                         die("invalid regular expression: %s",
2175                             o->word_regex);
2176         }
2177         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2178                 if (o->word_diff == diff_words_styles[i].type) {
2179                         ecbdata->diff_words->style =
2180                                 &diff_words_styles[i];
2181                         break;
2182                 }
2183         }
2184         if (want_color(o->use_color)) {
2185                 struct diff_words_style *st = ecbdata->diff_words->style;
2186                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2187                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2188                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2189         }
2190 }
2191
2192 static void free_diff_words_data(struct emit_callback *ecbdata)
2193 {
2194         if (ecbdata->diff_words) {
2195                 diff_words_flush(ecbdata);
2196                 free (ecbdata->diff_words->opt->emitted_symbols);
2197                 free (ecbdata->diff_words->opt);
2198                 free (ecbdata->diff_words->minus.text.ptr);
2199                 free (ecbdata->diff_words->minus.orig);
2200                 free (ecbdata->diff_words->plus.text.ptr);
2201                 free (ecbdata->diff_words->plus.orig);
2202                 if (ecbdata->diff_words->word_regex) {
2203                         regfree(ecbdata->diff_words->word_regex);
2204                         free(ecbdata->diff_words->word_regex);
2205                 }
2206                 FREE_AND_NULL(ecbdata->diff_words);
2207         }
2208 }
2209
2210 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2211 {
2212         if (want_color(diff_use_color))
2213                 return diff_colors[ix];
2214         return "";
2215 }
2216
2217 const char *diff_line_prefix(struct diff_options *opt)
2218 {
2219         struct strbuf *msgbuf;
2220         if (!opt->output_prefix)
2221                 return "";
2222
2223         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2224         return msgbuf->buf;
2225 }
2226
2227 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2228 {
2229         const char *cp;
2230         unsigned long allot;
2231         size_t l = len;
2232
2233         cp = line;
2234         allot = l;
2235         while (0 < l) {
2236                 (void) utf8_width(&cp, &l);
2237                 if (!cp)
2238                         break; /* truncated in the middle? */
2239         }
2240         return allot - l;
2241 }
2242
2243 static void find_lno(const char *line, struct emit_callback *ecbdata)
2244 {
2245         const char *p;
2246         ecbdata->lno_in_preimage = 0;
2247         ecbdata->lno_in_postimage = 0;
2248         p = strchr(line, '-');
2249         if (!p)
2250                 return; /* cannot happen */
2251         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2252         p = strchr(p, '+');
2253         if (!p)
2254                 return; /* cannot happen */
2255         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2256 }
2257
2258 static void fn_out_consume(void *priv, char *line, unsigned long len)
2259 {
2260         struct emit_callback *ecbdata = priv;
2261         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2262         struct diff_options *o = ecbdata->opt;
2263
2264         o->found_changes = 1;
2265
2266         if (ecbdata->header) {
2267                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2268                                  ecbdata->header->buf, ecbdata->header->len, 0);
2269                 strbuf_reset(ecbdata->header);
2270                 ecbdata->header = NULL;
2271         }
2272
2273         if (ecbdata->label_path[0]) {
2274                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2275                                  ecbdata->label_path[0],
2276                                  strlen(ecbdata->label_path[0]), 0);
2277                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2278                                  ecbdata->label_path[1],
2279                                  strlen(ecbdata->label_path[1]), 0);
2280                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2281         }
2282
2283         if (diff_suppress_blank_empty
2284             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2285                 line[0] = '\n';
2286                 len = 1;
2287         }
2288
2289         if (line[0] == '@') {
2290                 if (ecbdata->diff_words)
2291                         diff_words_flush(ecbdata);
2292                 len = sane_truncate_line(ecbdata, line, len);
2293                 find_lno(line, ecbdata);
2294                 emit_hunk_header(ecbdata, line, len);
2295                 return;
2296         }
2297
2298         if (ecbdata->diff_words) {
2299                 enum diff_symbol s =
2300                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2301                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2302                 if (line[0] == '-') {
2303                         diff_words_append(line, len,
2304                                           &ecbdata->diff_words->minus);
2305                         return;
2306                 } else if (line[0] == '+') {
2307                         diff_words_append(line, len,
2308                                           &ecbdata->diff_words->plus);
2309                         return;
2310                 } else if (starts_with(line, "\\ ")) {
2311                         /*
2312                          * Eat the "no newline at eof" marker as if we
2313                          * saw a "+" or "-" line with nothing on it,
2314                          * and return without diff_words_flush() to
2315                          * defer processing. If this is the end of
2316                          * preimage, more "+" lines may come after it.
2317                          */
2318                         return;
2319                 }
2320                 diff_words_flush(ecbdata);
2321                 emit_diff_symbol(o, s, line, len, 0);
2322                 return;
2323         }
2324
2325         switch (line[0]) {
2326         case '+':
2327                 ecbdata->lno_in_postimage++;
2328                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2329                 break;
2330         case '-':
2331                 ecbdata->lno_in_preimage++;
2332                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2333                 break;
2334         case ' ':
2335                 ecbdata->lno_in_postimage++;
2336                 ecbdata->lno_in_preimage++;
2337                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2338                 break;
2339         default:
2340                 /* incomplete line at the end */
2341                 ecbdata->lno_in_preimage++;
2342                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2343                                  line, len, 0);
2344                 break;
2345         }
2346 }
2347
2348 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2349 {
2350         const char *old_name = a;
2351         const char *new_name = b;
2352         int pfx_length, sfx_length;
2353         int pfx_adjust_for_slash;
2354         int len_a = strlen(a);
2355         int len_b = strlen(b);
2356         int a_midlen, b_midlen;
2357         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2358         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2359
2360         if (qlen_a || qlen_b) {
2361                 quote_c_style(a, name, NULL, 0);
2362                 strbuf_addstr(name, " => ");
2363                 quote_c_style(b, name, NULL, 0);
2364                 return;
2365         }
2366
2367         /* Find common prefix */
2368         pfx_length = 0;
2369         while (*old_name && *new_name && *old_name == *new_name) {
2370                 if (*old_name == '/')
2371                         pfx_length = old_name - a + 1;
2372                 old_name++;
2373                 new_name++;
2374         }
2375
2376         /* Find common suffix */
2377         old_name = a + len_a;
2378         new_name = b + len_b;
2379         sfx_length = 0;
2380         /*
2381          * If there is a common prefix, it must end in a slash.  In
2382          * that case we let this loop run 1 into the prefix to see the
2383          * same slash.
2384          *
2385          * If there is no common prefix, we cannot do this as it would
2386          * underrun the input strings.
2387          */
2388         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2389         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2390                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2391                *old_name == *new_name) {
2392                 if (*old_name == '/')
2393                         sfx_length = len_a - (old_name - a);
2394                 old_name--;
2395                 new_name--;
2396         }
2397
2398         /*
2399          * pfx{mid-a => mid-b}sfx
2400          * {pfx-a => pfx-b}sfx
2401          * pfx{sfx-a => sfx-b}
2402          * name-a => name-b
2403          */
2404         a_midlen = len_a - pfx_length - sfx_length;
2405         b_midlen = len_b - pfx_length - sfx_length;
2406         if (a_midlen < 0)
2407                 a_midlen = 0;
2408         if (b_midlen < 0)
2409                 b_midlen = 0;
2410
2411         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2412         if (pfx_length + sfx_length) {
2413                 strbuf_add(name, a, pfx_length);
2414                 strbuf_addch(name, '{');
2415         }
2416         strbuf_add(name, a + pfx_length, a_midlen);
2417         strbuf_addstr(name, " => ");
2418         strbuf_add(name, b + pfx_length, b_midlen);
2419         if (pfx_length + sfx_length) {
2420                 strbuf_addch(name, '}');
2421                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2422         }
2423 }
2424
2425 struct diffstat_t {
2426         int nr;
2427         int alloc;
2428         struct diffstat_file {
2429                 char *from_name;
2430                 char *name;
2431                 char *print_name;
2432                 const char *comments;
2433                 unsigned is_unmerged:1;
2434                 unsigned is_binary:1;
2435                 unsigned is_renamed:1;
2436                 unsigned is_interesting:1;
2437                 uintmax_t added, deleted;
2438         } **files;
2439 };
2440
2441 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2442                                           const char *name_a,
2443                                           const char *name_b)
2444 {
2445         struct diffstat_file *x;
2446         x = xcalloc(1, sizeof(*x));
2447         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2448         diffstat->files[diffstat->nr++] = x;
2449         if (name_b) {
2450                 x->from_name = xstrdup(name_a);
2451                 x->name = xstrdup(name_b);
2452                 x->is_renamed = 1;
2453         }
2454         else {
2455                 x->from_name = NULL;
2456                 x->name = xstrdup(name_a);
2457         }
2458         return x;
2459 }
2460
2461 static void diffstat_consume(void *priv, char *line, unsigned long len)
2462 {
2463         struct diffstat_t *diffstat = priv;
2464         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2465
2466         if (line[0] == '+')
2467                 x->added++;
2468         else if (line[0] == '-')
2469                 x->deleted++;
2470 }
2471
2472 const char mime_boundary_leader[] = "------------";
2473
2474 static int scale_linear(int it, int width, int max_change)
2475 {
2476         if (!it)
2477                 return 0;
2478         /*
2479          * make sure that at least one '-' or '+' is printed if
2480          * there is any change to this path. The easiest way is to
2481          * scale linearly as if the alloted width is one column shorter
2482          * than it is, and then add 1 to the result.
2483          */
2484         return 1 + (it * (width - 1) / max_change);
2485 }
2486
2487 static void show_graph(struct strbuf *out, char ch, int cnt,
2488                        const char *set, const char *reset)
2489 {
2490         if (cnt <= 0)
2491                 return;
2492         strbuf_addstr(out, set);
2493         strbuf_addchars(out, ch, cnt);
2494         strbuf_addstr(out, reset);
2495 }
2496
2497 static void fill_print_name(struct diffstat_file *file)
2498 {
2499         struct strbuf pname = STRBUF_INIT;
2500
2501         if (file->print_name)
2502                 return;
2503
2504         if (file->is_renamed)
2505                 pprint_rename(&pname, file->from_name, file->name);
2506         else
2507                 quote_c_style(file->name, &pname, NULL, 0);
2508
2509         if (file->comments)
2510                 strbuf_addf(&pname, " (%s)", file->comments);
2511
2512         file->print_name = strbuf_detach(&pname, NULL);
2513 }
2514
2515 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2516                 int files, int insertions, int deletions)
2517 {
2518         struct strbuf sb = STRBUF_INIT;
2519
2520         if (!files) {
2521                 assert(insertions == 0 && deletions == 0);
2522                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2523                                  NULL, 0, 0);
2524                 return;
2525         }
2526
2527         strbuf_addf(&sb,
2528                     (files == 1) ? " %d file changed" : " %d files changed",
2529                     files);
2530
2531         /*
2532          * For binary diff, the caller may want to print "x files
2533          * changed" with insertions == 0 && deletions == 0.
2534          *
2535          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2536          * is probably less confusing (i.e skip over "2 files changed
2537          * but nothing about added/removed lines? Is this a bug in Git?").
2538          */
2539         if (insertions || deletions == 0) {
2540                 strbuf_addf(&sb,
2541                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2542                             insertions);
2543         }
2544
2545         if (deletions || insertions == 0) {
2546                 strbuf_addf(&sb,
2547                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2548                             deletions);
2549         }
2550         strbuf_addch(&sb, '\n');
2551         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2552                          sb.buf, sb.len, 0);
2553         strbuf_release(&sb);
2554 }
2555
2556 void print_stat_summary(FILE *fp, int files,
2557                         int insertions, int deletions)
2558 {
2559         struct diff_options o;
2560         memset(&o, 0, sizeof(o));
2561         o.file = fp;
2562
2563         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2564 }
2565
2566 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2567 {
2568         int i, len, add, del, adds = 0, dels = 0;
2569         uintmax_t max_change = 0, max_len = 0;
2570         int total_files = data->nr, count;
2571         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2572         const char *reset, *add_c, *del_c;
2573         int extra_shown = 0;
2574         const char *line_prefix = diff_line_prefix(options);
2575         struct strbuf out = STRBUF_INIT;
2576
2577         if (data->nr == 0)
2578                 return;
2579
2580         count = options->stat_count ? options->stat_count : data->nr;
2581
2582         reset = diff_get_color_opt(options, DIFF_RESET);
2583         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2584         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2585
2586         /*
2587          * Find the longest filename and max number of changes
2588          */
2589         for (i = 0; (i < count) && (i < data->nr); i++) {
2590                 struct diffstat_file *file = data->files[i];
2591                 uintmax_t change = file->added + file->deleted;
2592
2593                 if (!file->is_interesting && (change == 0)) {
2594                         count++; /* not shown == room for one more */
2595                         continue;
2596                 }
2597                 fill_print_name(file);
2598                 len = strlen(file->print_name);
2599                 if (max_len < len)
2600                         max_len = len;
2601
2602                 if (file->is_unmerged) {
2603                         /* "Unmerged" is 8 characters */
2604                         bin_width = bin_width < 8 ? 8 : bin_width;
2605                         continue;
2606                 }
2607                 if (file->is_binary) {
2608                         /* "Bin XXX -> YYY bytes" */
2609                         int w = 14 + decimal_width(file->added)
2610                                 + decimal_width(file->deleted);
2611                         bin_width = bin_width < w ? w : bin_width;
2612                         /* Display change counts aligned with "Bin" */
2613                         number_width = 3;
2614                         continue;
2615                 }
2616
2617                 if (max_change < change)
2618                         max_change = change;
2619         }
2620         count = i; /* where we can stop scanning in data->files[] */
2621
2622         /*
2623          * We have width = stat_width or term_columns() columns total.
2624          * We want a maximum of min(max_len, stat_name_width) for the name part.
2625          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2626          * We also need 1 for " " and 4 + decimal_width(max_change)
2627          * for " | NNNN " and one the empty column at the end, altogether
2628          * 6 + decimal_width(max_change).
2629          *
2630          * If there's not enough space, we will use the smaller of
2631          * stat_name_width (if set) and 5/8*width for the filename,
2632          * and the rest for constant elements + graph part, but no more
2633          * than stat_graph_width for the graph part.
2634          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2635          * for the standard terminal size).
2636          *
2637          * In other words: stat_width limits the maximum width, and
2638          * stat_name_width fixes the maximum width of the filename,
2639          * and is also used to divide available columns if there
2640          * aren't enough.
2641          *
2642          * Binary files are displayed with "Bin XXX -> YYY bytes"
2643          * instead of the change count and graph. This part is treated
2644          * similarly to the graph part, except that it is not
2645          * "scaled". If total width is too small to accommodate the
2646          * guaranteed minimum width of the filename part and the
2647          * separators and this message, this message will "overflow"
2648          * making the line longer than the maximum width.
2649          */
2650
2651         if (options->stat_width == -1)
2652                 width = term_columns() - strlen(line_prefix);
2653         else
2654                 width = options->stat_width ? options->stat_width : 80;
2655         number_width = decimal_width(max_change) > number_width ?
2656                 decimal_width(max_change) : number_width;
2657
2658         if (options->stat_graph_width == -1)
2659                 options->stat_graph_width = diff_stat_graph_width;
2660
2661         /*
2662          * Guarantee 3/8*16==6 for the graph part
2663          * and 5/8*16==10 for the filename part
2664          */
2665         if (width < 16 + 6 + number_width)
2666                 width = 16 + 6 + number_width;
2667
2668         /*
2669          * First assign sizes that are wanted, ignoring available width.
2670          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2671          * starting from "XXX" should fit in graph_width.
2672          */
2673         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2674         if (options->stat_graph_width &&
2675             options->stat_graph_width < graph_width)
2676                 graph_width = options->stat_graph_width;
2677
2678         name_width = (options->stat_name_width > 0 &&
2679                       options->stat_name_width < max_len) ?
2680                 options->stat_name_width : max_len;
2681
2682         /*
2683          * Adjust adjustable widths not to exceed maximum width
2684          */
2685         if (name_width + number_width + 6 + graph_width > width) {
2686                 if (graph_width > width * 3/8 - number_width - 6) {
2687                         graph_width = width * 3/8 - number_width - 6;
2688                         if (graph_width < 6)
2689                                 graph_width = 6;
2690                 }
2691
2692                 if (options->stat_graph_width &&
2693                     graph_width > options->stat_graph_width)
2694                         graph_width = options->stat_graph_width;
2695                 if (name_width > width - number_width - 6 - graph_width)
2696                         name_width = width - number_width - 6 - graph_width;
2697                 else
2698                         graph_width = width - number_width - 6 - name_width;
2699         }
2700
2701         /*
2702          * From here name_width is the width of the name area,
2703          * and graph_width is the width of the graph area.
2704          * max_change is used to scale graph properly.
2705          */
2706         for (i = 0; i < count; i++) {
2707                 const char *prefix = "";
2708                 struct diffstat_file *file = data->files[i];
2709                 char *name = file->print_name;
2710                 uintmax_t added = file->added;
2711                 uintmax_t deleted = file->deleted;
2712                 int name_len;
2713
2714                 if (!file->is_interesting && (added + deleted == 0))
2715                         continue;
2716
2717                 /*
2718                  * "scale" the filename
2719                  */
2720                 len = name_width;
2721                 name_len = strlen(name);
2722                 if (name_width < name_len) {
2723                         char *slash;
2724                         prefix = "...";
2725                         len -= 3;
2726                         name += name_len - len;
2727                         slash = strchr(name, '/');
2728                         if (slash)
2729                                 name = slash;
2730                 }
2731
2732                 if (file->is_binary) {
2733                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2734                         strbuf_addf(&out, " %*s", number_width, "Bin");
2735                         if (!added && !deleted) {
2736                                 strbuf_addch(&out, '\n');
2737                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2738                                                  out.buf, out.len, 0);
2739                                 strbuf_reset(&out);
2740                                 continue;
2741                         }
2742                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2743                                 del_c, deleted, reset);
2744                         strbuf_addstr(&out, " -> ");
2745                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2746                                 add_c, added, reset);
2747                         strbuf_addstr(&out, " bytes\n");
2748                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2749                                          out.buf, out.len, 0);
2750                         strbuf_reset(&out);
2751                         continue;
2752                 }
2753                 else if (file->is_unmerged) {
2754                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2755                         strbuf_addstr(&out, " Unmerged\n");
2756                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2757                                          out.buf, out.len, 0);
2758                         strbuf_reset(&out);
2759                         continue;
2760                 }
2761
2762                 /*
2763                  * scale the add/delete
2764                  */
2765                 add = added;
2766                 del = deleted;
2767
2768                 if (graph_width <= max_change) {
2769                         int total = scale_linear(add + del, graph_width, max_change);
2770                         if (total < 2 && add && del)
2771                                 /* width >= 2 due to the sanity check */
2772                                 total = 2;
2773                         if (add < del) {
2774                                 add = scale_linear(add, graph_width, max_change);
2775                                 del = total - add;
2776                         } else {
2777                                 del = scale_linear(del, graph_width, max_change);
2778                                 add = total - del;
2779                         }
2780                 }
2781                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2782                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2783                         number_width, added + deleted,
2784                         added + deleted ? " " : "");
2785                 show_graph(&out, '+', add, add_c, reset);
2786                 show_graph(&out, '-', del, del_c, reset);
2787                 strbuf_addch(&out, '\n');
2788                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2789                                  out.buf, out.len, 0);
2790                 strbuf_reset(&out);
2791         }
2792
2793         for (i = 0; i < data->nr; i++) {
2794                 struct diffstat_file *file = data->files[i];
2795                 uintmax_t added = file->added;
2796                 uintmax_t deleted = file->deleted;
2797
2798                 if (file->is_unmerged ||
2799                     (!file->is_interesting && (added + deleted == 0))) {
2800                         total_files--;
2801                         continue;
2802                 }
2803
2804                 if (!file->is_binary) {
2805                         adds += added;
2806                         dels += deleted;
2807                 }
2808                 if (i < count)
2809                         continue;
2810                 if (!extra_shown)
2811                         emit_diff_symbol(options,
2812                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2813                                          NULL, 0, 0);
2814                 extra_shown = 1;
2815         }
2816
2817         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2818         strbuf_release(&out);
2819 }
2820
2821 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2822 {
2823         int i, adds = 0, dels = 0, total_files = data->nr;
2824
2825         if (data->nr == 0)
2826                 return;
2827
2828         for (i = 0; i < data->nr; i++) {
2829                 int added = data->files[i]->added;
2830                 int deleted = data->files[i]->deleted;
2831
2832                 if (data->files[i]->is_unmerged ||
2833                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2834                         total_files--;
2835                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2836                         adds += added;
2837                         dels += deleted;
2838                 }
2839         }
2840         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2841 }
2842
2843 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2844 {
2845         int i;
2846
2847         if (data->nr == 0)
2848                 return;
2849
2850         for (i = 0; i < data->nr; i++) {
2851                 struct diffstat_file *file = data->files[i];
2852
2853                 fprintf(options->file, "%s", diff_line_prefix(options));
2854
2855                 if (file->is_binary)
2856                         fprintf(options->file, "-\t-\t");
2857                 else
2858                         fprintf(options->file,
2859                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2860                                 file->added, file->deleted);
2861                 if (options->line_termination) {
2862                         fill_print_name(file);
2863                         if (!file->is_renamed)
2864                                 write_name_quoted(file->name, options->file,
2865                                                   options->line_termination);
2866                         else {
2867                                 fputs(file->print_name, options->file);
2868                                 putc(options->line_termination, options->file);
2869                         }
2870                 } else {
2871                         if (file->is_renamed) {
2872                                 putc('\0', options->file);
2873                                 write_name_quoted(file->from_name, options->file, '\0');
2874                         }
2875                         write_name_quoted(file->name, options->file, '\0');
2876                 }
2877         }
2878 }
2879
2880 struct dirstat_file {
2881         const char *name;
2882         unsigned long changed;
2883 };
2884
2885 struct dirstat_dir {
2886         struct dirstat_file *files;
2887         int alloc, nr, permille, cumulative;
2888 };
2889
2890 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2891                 unsigned long changed, const char *base, int baselen)
2892 {
2893         unsigned long sum_changes = 0;
2894         unsigned int sources = 0;
2895         const char *line_prefix = diff_line_prefix(opt);
2896
2897         while (dir->nr) {
2898                 struct dirstat_file *f = dir->files;
2899                 int namelen = strlen(f->name);
2900                 unsigned long changes;
2901                 char *slash;
2902
2903                 if (namelen < baselen)
2904                         break;
2905                 if (memcmp(f->name, base, baselen))
2906                         break;
2907                 slash = strchr(f->name + baselen, '/');
2908                 if (slash) {
2909                         int newbaselen = slash + 1 - f->name;
2910                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2911                         sources++;
2912                 } else {
2913                         changes = f->changed;
2914                         dir->files++;
2915                         dir->nr--;
2916                         sources += 2;
2917                 }
2918                 sum_changes += changes;
2919         }
2920
2921         /*
2922          * We don't report dirstat's for
2923          *  - the top level
2924          *  - or cases where everything came from a single directory
2925          *    under this directory (sources == 1).
2926          */
2927         if (baselen && sources != 1) {
2928                 if (sum_changes) {
2929                         int permille = sum_changes * 1000 / changed;
2930                         if (permille >= dir->permille) {
2931                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2932                                         permille / 10, permille % 10, baselen, base);
2933                                 if (!dir->cumulative)
2934                                         return 0;
2935                         }
2936                 }
2937         }
2938         return sum_changes;
2939 }
2940
2941 static int dirstat_compare(const void *_a, const void *_b)
2942 {
2943         const struct dirstat_file *a = _a;
2944         const struct dirstat_file *b = _b;
2945         return strcmp(a->name, b->name);
2946 }
2947
2948 static void show_dirstat(struct diff_options *options)
2949 {
2950         int i;
2951         unsigned long changed;
2952         struct dirstat_dir dir;
2953         struct diff_queue_struct *q = &diff_queued_diff;
2954
2955         dir.files = NULL;
2956         dir.alloc = 0;
2957         dir.nr = 0;
2958         dir.permille = options->dirstat_permille;
2959         dir.cumulative = options->flags.dirstat_cumulative;
2960
2961         changed = 0;
2962         for (i = 0; i < q->nr; i++) {
2963                 struct diff_filepair *p = q->queue[i];
2964                 const char *name;
2965                 unsigned long copied, added, damage;
2966
2967                 name = p->two->path ? p->two->path : p->one->path;
2968
2969                 if (p->one->oid_valid && p->two->oid_valid &&
2970                     oideq(&p->one->oid, &p->two->oid)) {
2971                         /*
2972                          * The SHA1 has not changed, so pre-/post-content is
2973                          * identical. We can therefore skip looking at the
2974                          * file contents altogether.
2975                          */
2976                         damage = 0;
2977                         goto found_damage;
2978                 }
2979
2980                 if (options->flags.dirstat_by_file) {
2981                         /*
2982                          * In --dirstat-by-file mode, we don't really need to
2983                          * look at the actual file contents at all.
2984                          * The fact that the SHA1 changed is enough for us to
2985                          * add this file to the list of results
2986                          * (with each file contributing equal damage).
2987                          */
2988                         damage = 1;
2989                         goto found_damage;
2990                 }
2991
2992                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2993                         diff_populate_filespec(options->repo, p->one, 0);
2994                         diff_populate_filespec(options->repo, p->two, 0);
2995                         diffcore_count_changes(options->repo,
2996                                                p->one, p->two, NULL, NULL,
2997                                                &copied, &added);
2998                         diff_free_filespec_data(p->one);
2999                         diff_free_filespec_data(p->two);
3000                 } else if (DIFF_FILE_VALID(p->one)) {
3001                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3002                         copied = added = 0;
3003                         diff_free_filespec_data(p->one);
3004                 } else if (DIFF_FILE_VALID(p->two)) {
3005                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3006                         copied = 0;
3007                         added = p->two->size;
3008                         diff_free_filespec_data(p->two);
3009                 } else
3010                         continue;
3011
3012                 /*
3013                  * Original minus copied is the removed material,
3014                  * added is the new material.  They are both damages
3015                  * made to the preimage.
3016                  * If the resulting damage is zero, we know that
3017                  * diffcore_count_changes() considers the two entries to
3018                  * be identical, but since the oid changed, we
3019                  * know that there must have been _some_ kind of change,
3020                  * so we force all entries to have damage > 0.
3021                  */
3022                 damage = (p->one->size - copied) + added;
3023                 if (!damage)
3024                         damage = 1;
3025
3026 found_damage:
3027                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3028                 dir.files[dir.nr].name = name;
3029                 dir.files[dir.nr].changed = damage;
3030                 changed += damage;
3031                 dir.nr++;
3032         }
3033
3034         /* This can happen even with many files, if everything was renames */
3035         if (!changed)
3036                 return;
3037
3038         /* Show all directories with more than x% of the changes */
3039         QSORT(dir.files, dir.nr, dirstat_compare);
3040         gather_dirstat(options, &dir, changed, "", 0);
3041 }
3042
3043 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3044 {
3045         int i;
3046         unsigned long changed;
3047         struct dirstat_dir dir;
3048
3049         if (data->nr == 0)
3050                 return;
3051
3052         dir.files = NULL;
3053         dir.alloc = 0;
3054         dir.nr = 0;
3055         dir.permille = options->dirstat_permille;
3056         dir.cumulative = options->flags.dirstat_cumulative;
3057
3058         changed = 0;
3059         for (i = 0; i < data->nr; i++) {
3060                 struct diffstat_file *file = data->files[i];
3061                 unsigned long damage = file->added + file->deleted;
3062                 if (file->is_binary)
3063                         /*
3064                          * binary files counts bytes, not lines. Must find some
3065                          * way to normalize binary bytes vs. textual lines.
3066                          * The following heuristic assumes that there are 64
3067                          * bytes per "line".
3068                          * This is stupid and ugly, but very cheap...
3069                          */
3070                         damage = DIV_ROUND_UP(damage, 64);
3071                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3072                 dir.files[dir.nr].name = file->name;
3073                 dir.files[dir.nr].changed = damage;
3074                 changed += damage;
3075                 dir.nr++;
3076         }
3077
3078         /* This can happen even with many files, if everything was renames */
3079         if (!changed)
3080                 return;
3081
3082         /* Show all directories with more than x% of the changes */
3083         QSORT(dir.files, dir.nr, dirstat_compare);
3084         gather_dirstat(options, &dir, changed, "", 0);
3085 }
3086
3087 static void free_diffstat_info(struct diffstat_t *diffstat)
3088 {
3089         int i;
3090         for (i = 0; i < diffstat->nr; i++) {
3091                 struct diffstat_file *f = diffstat->files[i];
3092                 free(f->print_name);
3093                 free(f->name);
3094                 free(f->from_name);
3095                 free(f);
3096         }
3097         free(diffstat->files);
3098 }
3099
3100 struct checkdiff_t {
3101         const char *filename;
3102         int lineno;
3103         int conflict_marker_size;
3104         struct diff_options *o;
3105         unsigned ws_rule;
3106         unsigned status;
3107 };
3108
3109 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3110 {
3111         char firstchar;
3112         int cnt;
3113
3114         if (len < marker_size + 1)
3115                 return 0;
3116         firstchar = line[0];
3117         switch (firstchar) {
3118         case '=': case '>': case '<': case '|':
3119                 break;
3120         default:
3121                 return 0;
3122         }
3123         for (cnt = 1; cnt < marker_size; cnt++)
3124                 if (line[cnt] != firstchar)
3125                         return 0;
3126         /* line[1] thru line[marker_size-1] are same as firstchar */
3127         if (len < marker_size + 1 || !isspace(line[marker_size]))
3128                 return 0;
3129         return 1;
3130 }
3131
3132 static void checkdiff_consume_hunk(void *priv,
3133                                    long ob, long on, long nb, long nn,
3134                                    const char *func, long funclen)
3135
3136 {
3137         struct checkdiff_t *data = priv;
3138         data->lineno = nb - 1;
3139 }
3140
3141 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3142 {
3143         struct checkdiff_t *data = priv;
3144         int marker_size = data->conflict_marker_size;
3145         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3146         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3147         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3148         char *err;
3149         const char *line_prefix;
3150
3151         assert(data->o);
3152         line_prefix = diff_line_prefix(data->o);
3153
3154         if (line[0] == '+') {
3155                 unsigned bad;
3156                 data->lineno++;
3157                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3158                         data->status |= 1;
3159                         fprintf(data->o->file,
3160                                 "%s%s:%d: leftover conflict marker\n",
3161                                 line_prefix, data->filename, data->lineno);
3162                 }
3163                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3164                 if (!bad)
3165                         return;
3166                 data->status |= bad;
3167                 err = whitespace_error_string(bad);
3168                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3169                         line_prefix, data->filename, data->lineno, err);
3170                 free(err);
3171                 emit_line(data->o, set, reset, line, 1);
3172                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3173                               data->o->file, set, reset, ws);
3174         } else if (line[0] == ' ') {
3175                 data->lineno++;
3176         }
3177 }
3178
3179 static unsigned char *deflate_it(char *data,
3180                                  unsigned long size,
3181                                  unsigned long *result_size)
3182 {
3183         int bound;
3184         unsigned char *deflated;
3185         git_zstream stream;
3186
3187         git_deflate_init(&stream, zlib_compression_level);
3188         bound = git_deflate_bound(&stream, size);
3189         deflated = xmalloc(bound);
3190         stream.next_out = deflated;
3191         stream.avail_out = bound;
3192
3193         stream.next_in = (unsigned char *)data;
3194         stream.avail_in = size;
3195         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3196                 ; /* nothing */
3197         git_deflate_end(&stream);
3198         *result_size = stream.total_out;
3199         return deflated;
3200 }
3201
3202 static void emit_binary_diff_body(struct diff_options *o,
3203                                   mmfile_t *one, mmfile_t *two)
3204 {
3205         void *cp;
3206         void *delta;
3207         void *deflated;
3208         void *data;
3209         unsigned long orig_size;
3210         unsigned long delta_size;
3211         unsigned long deflate_size;
3212         unsigned long data_size;
3213
3214         /* We could do deflated delta, or we could do just deflated two,
3215          * whichever is smaller.
3216          */
3217         delta = NULL;
3218         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3219         if (one->size && two->size) {
3220                 delta = diff_delta(one->ptr, one->size,
3221                                    two->ptr, two->size,
3222                                    &delta_size, deflate_size);
3223                 if (delta) {
3224                         void *to_free = delta;
3225                         orig_size = delta_size;
3226                         delta = deflate_it(delta, delta_size, &delta_size);
3227                         free(to_free);
3228                 }
3229         }
3230
3231         if (delta && delta_size < deflate_size) {
3232                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3233                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3234                                  s, strlen(s), 0);
3235                 free(s);
3236                 free(deflated);
3237                 data = delta;
3238                 data_size = delta_size;
3239         } else {
3240                 char *s = xstrfmt("%lu", two->size);
3241                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3242                                  s, strlen(s), 0);
3243                 free(s);
3244                 free(delta);
3245                 data = deflated;
3246                 data_size = deflate_size;
3247         }
3248
3249         /* emit data encoded in base85 */
3250         cp = data;
3251         while (data_size) {
3252                 int len;
3253                 int bytes = (52 < data_size) ? 52 : data_size;
3254                 char line[71];
3255                 data_size -= bytes;
3256                 if (bytes <= 26)
3257                         line[0] = bytes + 'A' - 1;
3258                 else
3259                         line[0] = bytes - 26 + 'a' - 1;
3260                 encode_85(line + 1, cp, bytes);
3261                 cp = (char *) cp + bytes;
3262
3263                 len = strlen(line);
3264                 line[len++] = '\n';
3265                 line[len] = '\0';
3266
3267                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3268                                  line, len, 0);
3269         }
3270         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3271         free(data);
3272 }
3273
3274 static void emit_binary_diff(struct diff_options *o,
3275                              mmfile_t *one, mmfile_t *two)
3276 {
3277         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3278         emit_binary_diff_body(o, one, two);
3279         emit_binary_diff_body(o, two, one);
3280 }
3281
3282 int diff_filespec_is_binary(struct repository *r,
3283                             struct diff_filespec *one)
3284 {
3285         if (one->is_binary == -1) {
3286                 diff_filespec_load_driver(one, r->index);
3287                 if (one->driver->binary != -1)
3288                         one->is_binary = one->driver->binary;
3289                 else {
3290                         if (!one->data && DIFF_FILE_VALID(one))
3291                                 diff_populate_filespec(r, one, CHECK_BINARY);
3292                         if (one->is_binary == -1 && one->data)
3293                                 one->is_binary = buffer_is_binary(one->data,
3294                                                 one->size);
3295                         if (one->is_binary == -1)
3296                                 one->is_binary = 0;
3297                 }
3298         }
3299         return one->is_binary;
3300 }
3301
3302 static const struct userdiff_funcname *
3303 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3304 {
3305         diff_filespec_load_driver(one, o->repo->index);
3306         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3307 }
3308
3309 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3310 {
3311         if (!options->a_prefix)
3312                 options->a_prefix = a;
3313         if (!options->b_prefix)
3314                 options->b_prefix = b;
3315 }
3316
3317 struct userdiff_driver *get_textconv(struct repository *r,
3318                                      struct diff_filespec *one)
3319 {
3320         if (!DIFF_FILE_VALID(one))
3321                 return NULL;
3322
3323         diff_filespec_load_driver(one, r->index);
3324         return userdiff_get_textconv(r, one->driver);
3325 }
3326
3327 static void builtin_diff(const char *name_a,
3328                          const char *name_b,
3329                          struct diff_filespec *one,
3330                          struct diff_filespec *two,
3331                          const char *xfrm_msg,
3332                          int must_show_header,
3333                          struct diff_options *o,
3334                          int complete_rewrite)
3335 {
3336         mmfile_t mf1, mf2;
3337         const char *lbl[2];
3338         char *a_one, *b_two;
3339         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3340         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3341         const char *a_prefix, *b_prefix;
3342         struct userdiff_driver *textconv_one = NULL;
3343         struct userdiff_driver *textconv_two = NULL;
3344         struct strbuf header = STRBUF_INIT;
3345         const char *line_prefix = diff_line_prefix(o);
3346
3347         diff_set_mnemonic_prefix(o, "a/", "b/");
3348         if (o->flags.reverse_diff) {
3349                 a_prefix = o->b_prefix;
3350                 b_prefix = o->a_prefix;
3351         } else {
3352                 a_prefix = o->a_prefix;
3353                 b_prefix = o->b_prefix;
3354         }
3355
3356         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3357             (!one->mode || S_ISGITLINK(one->mode)) &&
3358             (!two->mode || S_ISGITLINK(two->mode))) {
3359                 show_submodule_summary(o, one->path ? one->path : two->path,
3360                                 &one->oid, &two->oid,
3361                                 two->dirty_submodule);
3362                 return;
3363         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3364                    (!one->mode || S_ISGITLINK(one->mode)) &&
3365                    (!two->mode || S_ISGITLINK(two->mode))) {
3366                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3367                                 &one->oid, &two->oid,
3368                                 two->dirty_submodule);
3369                 return;
3370         }
3371
3372         if (o->flags.allow_textconv) {
3373                 textconv_one = get_textconv(o->repo, one);
3374                 textconv_two = get_textconv(o->repo, two);
3375         }
3376
3377         /* Never use a non-valid filename anywhere if at all possible */
3378         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3379         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3380
3381         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3382         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3383         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3384         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3385         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3386         if (lbl[0][0] == '/') {
3387                 /* /dev/null */
3388                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3389                 if (xfrm_msg)
3390                         strbuf_addstr(&header, xfrm_msg);
3391                 must_show_header = 1;
3392         }
3393         else if (lbl[1][0] == '/') {
3394                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3395                 if (xfrm_msg)
3396                         strbuf_addstr(&header, xfrm_msg);
3397                 must_show_header = 1;
3398         }
3399         else {
3400                 if (one->mode != two->mode) {
3401                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3402                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3403                         must_show_header = 1;
3404                 }
3405                 if (xfrm_msg)
3406                         strbuf_addstr(&header, xfrm_msg);
3407
3408                 /*
3409                  * we do not run diff between different kind
3410                  * of objects.
3411                  */
3412                 if ((one->mode ^ two->mode) & S_IFMT)
3413                         goto free_ab_and_return;
3414                 if (complete_rewrite &&
3415                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3416                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3417                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3418                                          header.buf, header.len, 0);
3419                         strbuf_reset(&header);
3420                         emit_rewrite_diff(name_a, name_b, one, two,
3421                                           textconv_one, textconv_two, o);
3422                         o->found_changes = 1;
3423                         goto free_ab_and_return;
3424                 }
3425         }
3426
3427         if (o->irreversible_delete && lbl[1][0] == '/') {
3428                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3429                                  header.len, 0);
3430                 strbuf_reset(&header);
3431                 goto free_ab_and_return;
3432         } else if (!o->flags.text &&
3433                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3434                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3435                 struct strbuf sb = STRBUF_INIT;
3436                 if (!one->data && !two->data &&
3437                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3438                     !o->flags.binary) {
3439                         if (oideq(&one->oid, &two->oid)) {
3440                                 if (must_show_header)
3441                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3442                                                          header.buf, header.len,
3443                                                          0);
3444                                 goto free_ab_and_return;
3445                         }
3446                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3447                                          header.buf, header.len, 0);
3448                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3449                                     diff_line_prefix(o), lbl[0], lbl[1]);
3450                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3451                                          sb.buf, sb.len, 0);
3452                         strbuf_release(&sb);
3453                         goto free_ab_and_return;
3454                 }
3455                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3456                     fill_mmfile(o->repo, &mf2, two) < 0)
3457                         die("unable to read files to diff");
3458                 /* Quite common confusing case */
3459                 if (mf1.size == mf2.size &&
3460                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3461                         if (must_show_header)
3462                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3463                                                  header.buf, header.len, 0);
3464                         goto free_ab_and_return;
3465                 }
3466                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3467                 strbuf_reset(&header);
3468                 if (o->flags.binary)
3469                         emit_binary_diff(o, &mf1, &mf2);
3470                 else {
3471                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3472                                     diff_line_prefix(o), lbl[0], lbl[1]);
3473                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3474                                          sb.buf, sb.len, 0);
3475                         strbuf_release(&sb);
3476                 }
3477                 o->found_changes = 1;
3478         } else {
3479                 /* Crazy xdl interfaces.. */
3480                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3481                 const char *v;
3482                 xpparam_t xpp;
3483                 xdemitconf_t xecfg;
3484                 struct emit_callback ecbdata;
3485                 const struct userdiff_funcname *pe;
3486
3487                 if (must_show_header) {
3488                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3489                                          header.buf, header.len, 0);
3490                         strbuf_reset(&header);
3491                 }
3492
3493                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3494                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3495
3496                 pe = diff_funcname_pattern(o, one);
3497                 if (!pe)
3498                         pe = diff_funcname_pattern(o, two);
3499
3500                 memset(&xpp, 0, sizeof(xpp));
3501                 memset(&xecfg, 0, sizeof(xecfg));
3502                 memset(&ecbdata, 0, sizeof(ecbdata));
3503                 if (o->flags.suppress_diff_headers)
3504                         lbl[0] = NULL;
3505                 ecbdata.label_path = lbl;
3506                 ecbdata.color_diff = want_color(o->use_color);
3507                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3508                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3509                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3510                 ecbdata.opt = o;
3511                 if (header.len && !o->flags.suppress_diff_headers)
3512                         ecbdata.header = &header;
3513                 xpp.flags = o->xdl_opts;
3514                 xpp.anchors = o->anchors;
3515                 xpp.anchors_nr = o->anchors_nr;
3516                 xecfg.ctxlen = o->context;
3517                 xecfg.interhunkctxlen = o->interhunkcontext;
3518                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3519                 if (o->flags.funccontext)
3520                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3521                 if (pe)
3522                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3523                 if (!diffopts)
3524                         ;
3525                 else if (skip_prefix(diffopts, "--unified=", &v))
3526                         xecfg.ctxlen = strtoul(v, NULL, 10);
3527                 else if (skip_prefix(diffopts, "-u", &v))
3528                         xecfg.ctxlen = strtoul(v, NULL, 10);
3529                 if (o->word_diff)
3530                         init_diff_words_data(&ecbdata, o, one, two);
3531                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3532                                   &ecbdata, &xpp, &xecfg))
3533                         die("unable to generate diff for %s", one->path);
3534                 if (o->word_diff)
3535                         free_diff_words_data(&ecbdata);
3536                 if (textconv_one)
3537                         free(mf1.ptr);
3538                 if (textconv_two)
3539                         free(mf2.ptr);
3540                 xdiff_clear_find_func(&xecfg);
3541         }
3542
3543  free_ab_and_return:
3544         strbuf_release(&header);
3545         diff_free_filespec_data(one);
3546         diff_free_filespec_data(two);
3547         free(a_one);
3548         free(b_two);
3549         return;
3550 }
3551
3552 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3553 {
3554         if (!is_renamed) {
3555                 if (p->status == DIFF_STATUS_ADDED) {
3556                         if (S_ISLNK(p->two->mode))
3557                                 return "new +l";
3558                         else if ((p->two->mode & 0777) == 0755)
3559                                 return "new +x";
3560                         else
3561                                 return "new";
3562                 } else if (p->status == DIFF_STATUS_DELETED)
3563                         return "gone";
3564         }
3565         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3566                 return "mode -l";
3567         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3568                 return "mode +l";
3569         else if ((p->one->mode & 0777) == 0644 &&
3570                  (p->two->mode & 0777) == 0755)
3571                 return "mode +x";
3572         else if ((p->one->mode & 0777) == 0755 &&
3573                  (p->two->mode & 0777) == 0644)
3574                 return "mode -x";
3575         return NULL;
3576 }
3577
3578 static void builtin_diffstat(const char *name_a, const char *name_b,
3579                              struct diff_filespec *one,
3580                              struct diff_filespec *two,
3581                              struct diffstat_t *diffstat,
3582                              struct diff_options *o,
3583                              struct diff_filepair *p)
3584 {
3585         mmfile_t mf1, mf2;
3586         struct diffstat_file *data;
3587         int same_contents;
3588         int complete_rewrite = 0;
3589
3590         if (!DIFF_PAIR_UNMERGED(p)) {
3591                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3592                         complete_rewrite = 1;
3593         }
3594
3595         data = diffstat_add(diffstat, name_a, name_b);
3596         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3597         if (o->flags.stat_with_summary)
3598                 data->comments = get_compact_summary(p, data->is_renamed);
3599
3600         if (!one || !two) {
3601                 data->is_unmerged = 1;
3602                 return;
3603         }
3604
3605         same_contents = oideq(&one->oid, &two->oid);
3606
3607         if (diff_filespec_is_binary(o->repo, one) ||
3608             diff_filespec_is_binary(o->repo, two)) {
3609                 data->is_binary = 1;
3610                 if (same_contents) {
3611                         data->added = 0;
3612                         data->deleted = 0;
3613                 } else {
3614                         data->added = diff_filespec_size(o->repo, two);
3615                         data->deleted = diff_filespec_size(o->repo, one);
3616                 }
3617         }
3618
3619         else if (complete_rewrite) {
3620                 diff_populate_filespec(o->repo, one, 0);
3621                 diff_populate_filespec(o->repo, two, 0);
3622                 data->deleted = count_lines(one->data, one->size);
3623                 data->added = count_lines(two->data, two->size);
3624         }
3625
3626         else if (!same_contents) {
3627                 /* Crazy xdl interfaces.. */
3628                 xpparam_t xpp;
3629                 xdemitconf_t xecfg;
3630
3631                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3632                     fill_mmfile(o->repo, &mf2, two) < 0)
3633                         die("unable to read files to diff");
3634
3635                 memset(&xpp, 0, sizeof(xpp));
3636                 memset(&xecfg, 0, sizeof(xecfg));
3637                 xpp.flags = o->xdl_opts;
3638                 xpp.anchors = o->anchors;
3639                 xpp.anchors_nr = o->anchors_nr;
3640                 xecfg.ctxlen = o->context;
3641                 xecfg.interhunkctxlen = o->interhunkcontext;
3642                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3643                                   diffstat_consume, diffstat, &xpp, &xecfg))
3644                         die("unable to generate diffstat for %s", one->path);
3645         }
3646
3647         diff_free_filespec_data(one);
3648         diff_free_filespec_data(two);
3649 }
3650
3651 static void builtin_checkdiff(const char *name_a, const char *name_b,
3652                               const char *attr_path,
3653                               struct diff_filespec *one,
3654                               struct diff_filespec *two,
3655                               struct diff_options *o)
3656 {
3657         mmfile_t mf1, mf2;
3658         struct checkdiff_t data;
3659
3660         if (!two)
3661                 return;
3662
3663         memset(&data, 0, sizeof(data));
3664         data.filename = name_b ? name_b : name_a;
3665         data.lineno = 0;
3666         data.o = o;
3667         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3668         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3669
3670         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3671             fill_mmfile(o->repo, &mf2, two) < 0)
3672                 die("unable to read files to diff");
3673
3674         /*
3675          * All the other codepaths check both sides, but not checking
3676          * the "old" side here is deliberate.  We are checking the newly
3677          * introduced changes, and as long as the "new" side is text, we
3678          * can and should check what it introduces.
3679          */
3680         if (diff_filespec_is_binary(o->repo, two))
3681                 goto free_and_return;
3682         else {
3683                 /* Crazy xdl interfaces.. */
3684                 xpparam_t xpp;
3685                 xdemitconf_t xecfg;
3686
3687                 memset(&xpp, 0, sizeof(xpp));
3688                 memset(&xecfg, 0, sizeof(xecfg));
3689                 xecfg.ctxlen = 1; /* at least one context line */
3690                 xpp.flags = 0;
3691                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3692                                   checkdiff_consume, &data,
3693                                   &xpp, &xecfg))
3694                         die("unable to generate checkdiff for %s", one->path);
3695
3696                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3697                         struct emit_callback ecbdata;
3698                         int blank_at_eof;
3699
3700                         ecbdata.ws_rule = data.ws_rule;
3701                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3702                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3703
3704                         if (blank_at_eof) {
3705                                 static char *err;
3706                                 if (!err)
3707                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3708                                 fprintf(o->file, "%s:%d: %s.\n",
3709                                         data.filename, blank_at_eof, err);
3710                                 data.status = 1; /* report errors */
3711                         }
3712                 }
3713         }
3714  free_and_return:
3715         diff_free_filespec_data(one);
3716         diff_free_filespec_data(two);
3717         if (data.status)
3718                 o->flags.check_failed = 1;
3719 }
3720
3721 struct diff_filespec *alloc_filespec(const char *path)
3722 {
3723         struct diff_filespec *spec;
3724
3725         FLEXPTR_ALLOC_STR(spec, path, path);
3726         spec->count = 1;
3727         spec->is_binary = -1;
3728         return spec;
3729 }
3730
3731 void free_filespec(struct diff_filespec *spec)
3732 {
3733         if (!--spec->count) {
3734                 diff_free_filespec_data(spec);
3735                 free(spec);
3736         }
3737 }
3738
3739 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3740                    int oid_valid, unsigned short mode)
3741 {
3742         if (mode) {
3743                 spec->mode = canon_mode(mode);
3744                 oidcpy(&spec->oid, oid);
3745                 spec->oid_valid = oid_valid;
3746         }
3747 }
3748
3749 /*
3750  * Given a name and sha1 pair, if the index tells us the file in
3751  * the work tree has that object contents, return true, so that
3752  * prepare_temp_file() does not have to inflate and extract.
3753  */
3754 static int reuse_worktree_file(struct index_state *istate,
3755                                const char *name,
3756                                const struct object_id *oid,
3757                                int want_file)
3758 {
3759         const struct cache_entry *ce;
3760         struct stat st;
3761         int pos, len;
3762
3763         /*
3764          * We do not read the cache ourselves here, because the
3765          * benchmark with my previous version that always reads cache
3766          * shows that it makes things worse for diff-tree comparing
3767          * two linux-2.6 kernel trees in an already checked out work
3768          * tree.  This is because most diff-tree comparisons deal with
3769          * only a small number of files, while reading the cache is
3770          * expensive for a large project, and its cost outweighs the
3771          * savings we get by not inflating the object to a temporary
3772          * file.  Practically, this code only helps when we are used
3773          * by diff-cache --cached, which does read the cache before
3774          * calling us.
3775          */
3776         if (!istate->cache)
3777                 return 0;
3778
3779         /* We want to avoid the working directory if our caller
3780          * doesn't need the data in a normal file, this system
3781          * is rather slow with its stat/open/mmap/close syscalls,
3782          * and the object is contained in a pack file.  The pack
3783          * is probably already open and will be faster to obtain
3784          * the data through than the working directory.  Loose
3785          * objects however would tend to be slower as they need
3786          * to be individually opened and inflated.
3787          */
3788         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3789                 return 0;
3790
3791         /*
3792          * Similarly, if we'd have to convert the file contents anyway, that
3793          * makes the optimization not worthwhile.
3794          */
3795         if (!want_file && would_convert_to_git(istate, name))
3796                 return 0;
3797
3798         len = strlen(name);
3799         pos = index_name_pos(istate, name, len);
3800         if (pos < 0)
3801                 return 0;
3802         ce = istate->cache[pos];
3803
3804         /*
3805          * This is not the sha1 we are looking for, or
3806          * unreusable because it is not a regular file.
3807          */
3808         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3809                 return 0;
3810
3811         /*
3812          * If ce is marked as "assume unchanged", there is no
3813          * guarantee that work tree matches what we are looking for.
3814          */
3815         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3816                 return 0;
3817
3818         /*
3819          * If ce matches the file in the work tree, we can reuse it.
3820          */
3821         if (ce_uptodate(ce) ||
3822             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3823                 return 1;
3824
3825         return 0;
3826 }
3827
3828 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3829 {
3830         struct strbuf buf = STRBUF_INIT;
3831         char *dirty = "";
3832
3833         /* Are we looking at the work tree? */
3834         if (s->dirty_submodule)
3835                 dirty = "-dirty";
3836
3837         strbuf_addf(&buf, "Subproject commit %s%s\n",
3838                     oid_to_hex(&s->oid), dirty);
3839         s->size = buf.len;
3840         if (size_only) {
3841                 s->data = NULL;
3842                 strbuf_release(&buf);
3843         } else {
3844                 s->data = strbuf_detach(&buf, NULL);
3845                 s->should_free = 1;
3846         }
3847         return 0;
3848 }
3849
3850 /*
3851  * While doing rename detection and pickaxe operation, we may need to
3852  * grab the data for the blob (or file) for our own in-core comparison.
3853  * diff_filespec has data and size fields for this purpose.
3854  */
3855 int diff_populate_filespec(struct repository *r,
3856                            struct diff_filespec *s,
3857                            unsigned int flags)
3858 {
3859         int size_only = flags & CHECK_SIZE_ONLY;
3860         int err = 0;
3861         int conv_flags = global_conv_flags_eol;
3862         /*
3863          * demote FAIL to WARN to allow inspecting the situation
3864          * instead of refusing.
3865          */
3866         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3867                 conv_flags = CONV_EOL_RNDTRP_WARN;
3868
3869         if (!DIFF_FILE_VALID(s))
3870                 die("internal error: asking to populate invalid file.");
3871         if (S_ISDIR(s->mode))
3872                 return -1;
3873
3874         if (s->data)
3875                 return 0;
3876
3877         if (size_only && 0 < s->size)
3878                 return 0;
3879
3880         if (S_ISGITLINK(s->mode))
3881                 return diff_populate_gitlink(s, size_only);
3882
3883         if (!s->oid_valid ||
3884             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3885                 struct strbuf buf = STRBUF_INIT;
3886                 struct stat st;
3887                 int fd;
3888
3889                 if (lstat(s->path, &st) < 0) {
3890                 err_empty:
3891                         err = -1;
3892                 empty:
3893                         s->data = (char *)"";
3894                         s->size = 0;
3895                         return err;
3896                 }
3897                 s->size = xsize_t(st.st_size);
3898                 if (!s->size)
3899                         goto empty;
3900                 if (S_ISLNK(st.st_mode)) {
3901                         struct strbuf sb = STRBUF_INIT;
3902
3903                         if (strbuf_readlink(&sb, s->path, s->size))
3904                                 goto err_empty;
3905                         s->size = sb.len;
3906                         s->data = strbuf_detach(&sb, NULL);
3907                         s->should_free = 1;
3908                         return 0;
3909                 }
3910
3911                 /*
3912                  * Even if the caller would be happy with getting
3913                  * only the size, we cannot return early at this
3914                  * point if the path requires us to run the content
3915                  * conversion.
3916                  */
3917                 if (size_only && !would_convert_to_git(r->index, s->path))
3918                         return 0;
3919
3920                 /*
3921                  * Note: this check uses xsize_t(st.st_size) that may
3922                  * not be the true size of the blob after it goes
3923                  * through convert_to_git().  This may not strictly be
3924                  * correct, but the whole point of big_file_threshold
3925                  * and is_binary check being that we want to avoid
3926                  * opening the file and inspecting the contents, this
3927                  * is probably fine.
3928                  */
3929                 if ((flags & CHECK_BINARY) &&
3930                     s->size > big_file_threshold && s->is_binary == -1) {
3931                         s->is_binary = 1;
3932                         return 0;
3933                 }
3934                 fd = open(s->path, O_RDONLY);
3935                 if (fd < 0)
3936                         goto err_empty;
3937                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3938                 close(fd);
3939                 s->should_munmap = 1;
3940
3941                 /*
3942                  * Convert from working tree format to canonical git format
3943                  */
3944                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
3945                         size_t size = 0;
3946                         munmap(s->data, s->size);
3947                         s->should_munmap = 0;
3948                         s->data = strbuf_detach(&buf, &size);
3949                         s->size = size;
3950                         s->should_free = 1;
3951                 }
3952         }
3953         else {
3954                 enum object_type type;
3955                 if (size_only || (flags & CHECK_BINARY)) {
3956                         type = oid_object_info(r, &s->oid, &s->size);
3957                         if (type < 0)
3958                                 die("unable to read %s",
3959                                     oid_to_hex(&s->oid));
3960                         if (size_only)
3961                                 return 0;
3962                         if (s->size > big_file_threshold && s->is_binary == -1) {
3963                                 s->is_binary = 1;
3964                                 return 0;
3965                         }
3966                 }
3967                 s->data = read_object_file(&s->oid, &type, &s->size);
3968                 if (!s->data)
3969                         die("unable to read %s", oid_to_hex(&s->oid));
3970                 s->should_free = 1;
3971         }
3972         return 0;
3973 }
3974
3975 void diff_free_filespec_blob(struct diff_filespec *s)
3976 {
3977         if (s->should_free)
3978                 free(s->data);
3979         else if (s->should_munmap)
3980                 munmap(s->data, s->size);
3981
3982         if (s->should_free || s->should_munmap) {
3983                 s->should_free = s->should_munmap = 0;
3984                 s->data = NULL;
3985         }
3986 }
3987
3988 void diff_free_filespec_data(struct diff_filespec *s)
3989 {
3990         diff_free_filespec_blob(s);
3991         FREE_AND_NULL(s->cnt_data);
3992 }
3993
3994 static void prep_temp_blob(struct index_state *istate,
3995                            const char *path, struct diff_tempfile *temp,
3996                            void *blob,
3997                            unsigned long size,
3998                            const struct object_id *oid,
3999                            int mode)
4000 {
4001         struct strbuf buf = STRBUF_INIT;
4002         struct strbuf tempfile = STRBUF_INIT;
4003         char *path_dup = xstrdup(path);
4004         const char *base = basename(path_dup);
4005
4006         /* Generate "XXXXXX_basename.ext" */
4007         strbuf_addstr(&tempfile, "XXXXXX_");
4008         strbuf_addstr(&tempfile, base);
4009
4010         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4011         if (!temp->tempfile)
4012                 die_errno("unable to create temp-file");
4013         if (convert_to_working_tree(istate, path,
4014                         (const char *)blob, (size_t)size, &buf)) {
4015                 blob = buf.buf;
4016                 size = buf.len;
4017         }
4018         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4019             close_tempfile_gently(temp->tempfile))
4020                 die_errno("unable to write temp-file");
4021         temp->name = get_tempfile_path(temp->tempfile);
4022         oid_to_hex_r(temp->hex, oid);
4023         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4024         strbuf_release(&buf);
4025         strbuf_release(&tempfile);
4026         free(path_dup);
4027 }
4028
4029 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4030                                                const char *name,
4031                                                struct diff_filespec *one)
4032 {
4033         struct diff_tempfile *temp = claim_diff_tempfile();
4034
4035         if (!DIFF_FILE_VALID(one)) {
4036         not_a_valid_file:
4037                 /* A '-' entry produces this for file-2, and
4038                  * a '+' entry produces this for file-1.
4039                  */
4040                 temp->name = "/dev/null";
4041                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4042                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4043                 return temp;
4044         }
4045
4046         if (!S_ISGITLINK(one->mode) &&
4047             (!one->oid_valid ||
4048              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4049                 struct stat st;
4050                 if (lstat(name, &st) < 0) {
4051                         if (errno == ENOENT)
4052                                 goto not_a_valid_file;
4053                         die_errno("stat(%s)", name);
4054                 }
4055                 if (S_ISLNK(st.st_mode)) {
4056                         struct strbuf sb = STRBUF_INIT;
4057                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4058                                 die_errno("readlink(%s)", name);
4059                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4060                                        (one->oid_valid ?
4061                                         &one->oid : &null_oid),
4062                                        (one->oid_valid ?
4063                                         one->mode : S_IFLNK));
4064                         strbuf_release(&sb);
4065                 }
4066                 else {
4067                         /* we can borrow from the file in the work tree */
4068                         temp->name = name;
4069                         if (!one->oid_valid)
4070                                 oid_to_hex_r(temp->hex, &null_oid);
4071                         else
4072                                 oid_to_hex_r(temp->hex, &one->oid);
4073                         /* Even though we may sometimes borrow the
4074                          * contents from the work tree, we always want
4075                          * one->mode.  mode is trustworthy even when
4076                          * !(one->oid_valid), as long as
4077                          * DIFF_FILE_VALID(one).
4078                          */
4079                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4080                 }
4081                 return temp;
4082         }
4083         else {
4084                 if (diff_populate_filespec(r, one, 0))
4085                         die("cannot read data blob for %s", one->path);
4086                 prep_temp_blob(r->index, name, temp,
4087                                one->data, one->size,
4088                                &one->oid, one->mode);
4089         }
4090         return temp;
4091 }
4092
4093 static void add_external_diff_name(struct repository *r,
4094                                    struct argv_array *argv,
4095                                    const char *name,
4096                                    struct diff_filespec *df)
4097 {
4098         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4099         argv_array_push(argv, temp->name);
4100         argv_array_push(argv, temp->hex);
4101         argv_array_push(argv, temp->mode);
4102 }
4103
4104 /* An external diff command takes:
4105  *
4106  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4107  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4108  *
4109  */
4110 static void run_external_diff(const char *pgm,
4111                               const char *name,
4112                               const char *other,
4113                               struct diff_filespec *one,
4114                               struct diff_filespec *two,
4115                               const char *xfrm_msg,
4116                               int complete_rewrite,
4117                               struct diff_options *o)
4118 {
4119         struct argv_array argv = ARGV_ARRAY_INIT;
4120         struct argv_array env = ARGV_ARRAY_INIT;
4121         struct diff_queue_struct *q = &diff_queued_diff;
4122
4123         argv_array_push(&argv, pgm);
4124         argv_array_push(&argv, name);
4125
4126         if (one && two) {
4127                 add_external_diff_name(o->repo, &argv, name, one);
4128                 if (!other)
4129                         add_external_diff_name(o->repo, &argv, name, two);
4130                 else {
4131                         add_external_diff_name(o->repo, &argv, other, two);
4132                         argv_array_push(&argv, other);
4133                         argv_array_push(&argv, xfrm_msg);
4134                 }
4135         }
4136
4137         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4138         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4139
4140         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4141                 die(_("external diff died, stopping at %s"), name);
4142
4143         remove_tempfile();
4144         argv_array_clear(&argv);
4145         argv_array_clear(&env);
4146 }
4147
4148 static int similarity_index(struct diff_filepair *p)
4149 {
4150         return p->score * 100 / MAX_SCORE;
4151 }
4152
4153 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4154 {
4155         if (startup_info->have_repository)
4156                 return find_unique_abbrev(oid, abbrev);
4157         else {
4158                 char *hex = oid_to_hex(oid);
4159                 if (abbrev < 0)
4160                         abbrev = FALLBACK_DEFAULT_ABBREV;
4161                 if (abbrev > the_hash_algo->hexsz)
4162                         BUG("oid abbreviation out of range: %d", abbrev);
4163                 if (abbrev)
4164                         hex[abbrev] = '\0';
4165                 return hex;
4166         }
4167 }
4168
4169 static void fill_metainfo(struct strbuf *msg,
4170                           const char *name,
4171                           const char *other,
4172                           struct diff_filespec *one,
4173                           struct diff_filespec *two,
4174                           struct diff_options *o,
4175                           struct diff_filepair *p,
4176                           int *must_show_header,
4177                           int use_color)
4178 {
4179         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4180         const char *reset = diff_get_color(use_color, DIFF_RESET);
4181         const char *line_prefix = diff_line_prefix(o);
4182
4183         *must_show_header = 1;
4184         strbuf_init(msg, PATH_MAX * 2 + 300);
4185         switch (p->status) {
4186         case DIFF_STATUS_COPIED:
4187                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4188                             line_prefix, set, similarity_index(p));
4189                 strbuf_addf(msg, "%s\n%s%scopy from ",
4190                             reset,  line_prefix, set);
4191                 quote_c_style(name, msg, NULL, 0);
4192                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4193                 quote_c_style(other, msg, NULL, 0);
4194                 strbuf_addf(msg, "%s\n", reset);
4195                 break;
4196         case DIFF_STATUS_RENAMED:
4197                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4198                             line_prefix, set, similarity_index(p));
4199                 strbuf_addf(msg, "%s\n%s%srename from ",
4200                             reset, line_prefix, set);
4201                 quote_c_style(name, msg, NULL, 0);
4202                 strbuf_addf(msg, "%s\n%s%srename to ",
4203                             reset, line_prefix, set);
4204                 quote_c_style(other, msg, NULL, 0);
4205                 strbuf_addf(msg, "%s\n", reset);
4206                 break;
4207         case DIFF_STATUS_MODIFIED:
4208                 if (p->score) {
4209                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4210                                     line_prefix,
4211                                     set, similarity_index(p), reset);
4212                         break;
4213                 }
4214                 /* fallthru */
4215         default:
4216                 *must_show_header = 0;
4217         }
4218         if (one && two && !oideq(&one->oid, &two->oid)) {
4219                 const unsigned hexsz = the_hash_algo->hexsz;
4220                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4221
4222                 if (o->flags.binary) {
4223                         mmfile_t mf;
4224                         if ((!fill_mmfile(o->repo, &mf, one) &&
4225                              diff_filespec_is_binary(o->repo, one)) ||
4226                             (!fill_mmfile(o->repo, &mf, two) &&
4227                              diff_filespec_is_binary(o->repo, two)))
4228                                 abbrev = hexsz;
4229                 }
4230                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4231                             diff_abbrev_oid(&one->oid, abbrev),
4232                             diff_abbrev_oid(&two->oid, abbrev));
4233                 if (one->mode == two->mode)
4234                         strbuf_addf(msg, " %06o", one->mode);
4235                 strbuf_addf(msg, "%s\n", reset);
4236         }
4237 }
4238
4239 static void run_diff_cmd(const char *pgm,
4240                          const char *name,
4241                          const char *other,
4242                          const char *attr_path,
4243                          struct diff_filespec *one,
4244                          struct diff_filespec *two,
4245                          struct strbuf *msg,
4246                          struct diff_options *o,
4247                          struct diff_filepair *p)
4248 {
4249         const char *xfrm_msg = NULL;
4250         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4251         int must_show_header = 0;
4252
4253
4254         if (o->flags.allow_external) {
4255                 struct userdiff_driver *drv;
4256
4257                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4258                 if (drv && drv->external)
4259                         pgm = drv->external;
4260         }
4261
4262         if (msg) {
4263                 /*
4264                  * don't use colors when the header is intended for an
4265                  * external diff driver
4266                  */
4267                 fill_metainfo(msg, name, other, one, two, o, p,
4268                               &must_show_header,
4269                               want_color(o->use_color) && !pgm);
4270                 xfrm_msg = msg->len ? msg->buf : NULL;
4271         }
4272
4273         if (pgm) {
4274                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4275                                   complete_rewrite, o);
4276                 return;
4277         }
4278         if (one && two)
4279                 builtin_diff(name, other ? other : name,
4280                              one, two, xfrm_msg, must_show_header,
4281                              o, complete_rewrite);
4282         else
4283                 fprintf(o->file, "* Unmerged path %s\n", name);
4284 }
4285
4286 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4287 {
4288         if (DIFF_FILE_VALID(one)) {
4289                 if (!one->oid_valid) {
4290                         struct stat st;
4291                         if (one->is_stdin) {
4292                                 oidclr(&one->oid);
4293                                 return;
4294                         }
4295                         if (lstat(one->path, &st) < 0)
4296                                 die_errno("stat '%s'", one->path);
4297                         if (index_path(istate, &one->oid, one->path, &st, 0))
4298                                 die("cannot hash %s", one->path);
4299                 }
4300         }
4301         else
4302                 oidclr(&one->oid);
4303 }
4304
4305 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4306 {
4307         /* Strip the prefix but do not molest /dev/null and absolute paths */
4308         if (*namep && !is_absolute_path(*namep)) {
4309                 *namep += prefix_length;
4310                 if (**namep == '/')
4311                         ++*namep;
4312         }
4313         if (*otherp && !is_absolute_path(*otherp)) {
4314                 *otherp += prefix_length;
4315                 if (**otherp == '/')
4316                         ++*otherp;
4317         }
4318 }
4319
4320 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4321 {
4322         const char *pgm = external_diff();
4323         struct strbuf msg;
4324         struct diff_filespec *one = p->one;
4325         struct diff_filespec *two = p->two;
4326         const char *name;
4327         const char *other;
4328         const char *attr_path;
4329
4330         name  = one->path;
4331         other = (strcmp(name, two->path) ? two->path : NULL);
4332         attr_path = name;
4333         if (o->prefix_length)
4334                 strip_prefix(o->prefix_length, &name, &other);
4335
4336         if (!o->flags.allow_external)
4337                 pgm = NULL;
4338
4339         if (DIFF_PAIR_UNMERGED(p)) {
4340                 run_diff_cmd(pgm, name, NULL, attr_path,
4341                              NULL, NULL, NULL, o, p);
4342                 return;
4343         }
4344
4345         diff_fill_oid_info(one, o->repo->index);
4346         diff_fill_oid_info(two, o->repo->index);
4347
4348         if (!pgm &&
4349             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4350             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4351                 /*
4352                  * a filepair that changes between file and symlink
4353                  * needs to be split into deletion and creation.
4354                  */
4355                 struct diff_filespec *null = alloc_filespec(two->path);
4356                 run_diff_cmd(NULL, name, other, attr_path,
4357                              one, null, &msg,
4358                              o, p);
4359                 free(null);
4360                 strbuf_release(&msg);
4361
4362                 null = alloc_filespec(one->path);
4363                 run_diff_cmd(NULL, name, other, attr_path,
4364                              null, two, &msg, o, p);
4365                 free(null);
4366         }
4367         else
4368                 run_diff_cmd(pgm, name, other, attr_path,
4369                              one, two, &msg, o, p);
4370
4371         strbuf_release(&msg);
4372 }
4373
4374 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4375                          struct diffstat_t *diffstat)
4376 {
4377         const char *name;
4378         const char *other;
4379
4380         if (DIFF_PAIR_UNMERGED(p)) {
4381                 /* unmerged */
4382                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4383                                  diffstat, o, p);
4384                 return;
4385         }
4386
4387         name = p->one->path;
4388         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4389
4390         if (o->prefix_length)
4391                 strip_prefix(o->prefix_length, &name, &other);
4392
4393         diff_fill_oid_info(p->one, o->repo->index);
4394         diff_fill_oid_info(p->two, o->repo->index);
4395
4396         builtin_diffstat(name, other, p->one, p->two,
4397                          diffstat, o, p);
4398 }
4399
4400 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4401 {
4402         const char *name;
4403         const char *other;
4404         const char *attr_path;
4405
4406         if (DIFF_PAIR_UNMERGED(p)) {
4407                 /* unmerged */
4408                 return;
4409         }
4410
4411         name = p->one->path;
4412         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4413         attr_path = other ? other : name;
4414
4415         if (o->prefix_length)
4416                 strip_prefix(o->prefix_length, &name, &other);
4417
4418         diff_fill_oid_info(p->one, o->repo->index);
4419         diff_fill_oid_info(p->two, o->repo->index);
4420
4421         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4422 }
4423
4424 static void prep_parse_options(struct diff_options *options);
4425
4426 void repo_diff_setup(struct repository *r, struct diff_options *options)
4427 {
4428         memcpy(options, &default_diff_options, sizeof(*options));
4429
4430         options->file = stdout;
4431         options->repo = r;
4432
4433         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4434         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4435         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4436         options->abbrev = DEFAULT_ABBREV;
4437         options->line_termination = '\n';
4438         options->break_opt = -1;
4439         options->rename_limit = -1;
4440         options->dirstat_permille = diff_dirstat_permille_default;
4441         options->context = diff_context_default;
4442         options->interhunkcontext = diff_interhunk_context_default;
4443         options->ws_error_highlight = ws_error_highlight_default;
4444         options->flags.rename_empty = 1;
4445         options->objfind = NULL;
4446
4447         /* pathchange left =NULL by default */
4448         options->change = diff_change;
4449         options->add_remove = diff_addremove;
4450         options->use_color = diff_use_color_default;
4451         options->detect_rename = diff_detect_rename_default;
4452         options->xdl_opts |= diff_algorithm;
4453         if (diff_indent_heuristic)
4454                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4455
4456         options->orderfile = diff_order_file_cfg;
4457
4458         if (diff_no_prefix) {
4459                 options->a_prefix = options->b_prefix = "";
4460         } else if (!diff_mnemonic_prefix) {
4461                 options->a_prefix = "a/";
4462                 options->b_prefix = "b/";
4463         }
4464
4465         options->color_moved = diff_color_moved_default;
4466         options->color_moved_ws_handling = diff_color_moved_ws_default;
4467
4468         prep_parse_options(options);
4469 }
4470
4471 void diff_setup_done(struct diff_options *options)
4472 {
4473         unsigned check_mask = DIFF_FORMAT_NAME |
4474                               DIFF_FORMAT_NAME_STATUS |
4475                               DIFF_FORMAT_CHECKDIFF |
4476                               DIFF_FORMAT_NO_OUTPUT;
4477         /*
4478          * This must be signed because we're comparing against a potentially
4479          * negative value.
4480          */
4481         const int hexsz = the_hash_algo->hexsz;
4482
4483         if (options->set_default)
4484                 options->set_default(options);
4485
4486         if (HAS_MULTI_BITS(options->output_format & check_mask))
4487                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4488
4489         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4490                 die(_("-G, -S and --find-object are mutually exclusive"));
4491
4492         /*
4493          * Most of the time we can say "there are changes"
4494          * only by checking if there are changed paths, but
4495          * --ignore-whitespace* options force us to look
4496          * inside contents.
4497          */
4498
4499         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4500                 options->flags.diff_from_contents = 1;
4501         else
4502                 options->flags.diff_from_contents = 0;
4503
4504         if (options->flags.find_copies_harder)
4505                 options->detect_rename = DIFF_DETECT_COPY;
4506
4507         if (!options->flags.relative_name)
4508                 options->prefix = NULL;
4509         if (options->prefix)
4510                 options->prefix_length = strlen(options->prefix);
4511         else
4512                 options->prefix_length = 0;
4513
4514         if (options->output_format & (DIFF_FORMAT_NAME |
4515                                       DIFF_FORMAT_NAME_STATUS |
4516                                       DIFF_FORMAT_CHECKDIFF |
4517                                       DIFF_FORMAT_NO_OUTPUT))
4518                 options->output_format &= ~(DIFF_FORMAT_RAW |
4519                                             DIFF_FORMAT_NUMSTAT |
4520                                             DIFF_FORMAT_DIFFSTAT |
4521                                             DIFF_FORMAT_SHORTSTAT |
4522                                             DIFF_FORMAT_DIRSTAT |
4523                                             DIFF_FORMAT_SUMMARY |
4524                                             DIFF_FORMAT_PATCH);
4525
4526         /*
4527          * These cases always need recursive; we do not drop caller-supplied
4528          * recursive bits for other formats here.
4529          */
4530         if (options->output_format & (DIFF_FORMAT_PATCH |
4531                                       DIFF_FORMAT_NUMSTAT |
4532                                       DIFF_FORMAT_DIFFSTAT |
4533                                       DIFF_FORMAT_SHORTSTAT |
4534                                       DIFF_FORMAT_DIRSTAT |
4535                                       DIFF_FORMAT_SUMMARY |
4536                                       DIFF_FORMAT_CHECKDIFF))
4537                 options->flags.recursive = 1;
4538         /*
4539          * Also pickaxe would not work very well if you do not say recursive
4540          */
4541         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4542                 options->flags.recursive = 1;
4543         /*
4544          * When patches are generated, submodules diffed against the work tree
4545          * must be checked for dirtiness too so it can be shown in the output
4546          */
4547         if (options->output_format & DIFF_FORMAT_PATCH)
4548                 options->flags.dirty_submodules = 1;
4549
4550         if (options->detect_rename && options->rename_limit < 0)
4551                 options->rename_limit = diff_rename_limit_default;
4552         if (hexsz < options->abbrev)
4553                 options->abbrev = hexsz; /* full */
4554
4555         /*
4556          * It does not make sense to show the first hit we happened
4557          * to have found.  It does not make sense not to return with
4558          * exit code in such a case either.
4559          */
4560         if (options->flags.quick) {
4561                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4562                 options->flags.exit_with_status = 1;
4563         }
4564
4565         options->diff_path_counter = 0;
4566
4567         if (options->flags.follow_renames && options->pathspec.nr != 1)
4568                 die(_("--follow requires exactly one pathspec"));
4569
4570         if (!options->use_color || external_diff())
4571                 options->color_moved = 0;
4572
4573         FREE_AND_NULL(options->parseopts);
4574 }
4575
4576 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4577 {
4578         char c, *eq;
4579         int len;
4580
4581         if (*arg != '-')
4582                 return 0;
4583         c = *++arg;
4584         if (!c)
4585                 return 0;
4586         if (c == arg_short) {
4587                 c = *++arg;
4588                 if (!c)
4589                         return 1;
4590                 if (val && isdigit(c)) {
4591                         char *end;
4592                         int n = strtoul(arg, &end, 10);
4593                         if (*end)
4594                                 return 0;
4595                         *val = n;
4596                         return 1;
4597                 }
4598                 return 0;
4599         }
4600         if (c != '-')
4601                 return 0;
4602         arg++;
4603         eq = strchrnul(arg, '=');
4604         len = eq - arg;
4605         if (!len || strncmp(arg, arg_long, len))
4606                 return 0;
4607         if (*eq) {
4608                 int n;
4609                 char *end;
4610                 if (!isdigit(*++eq))
4611                         return 0;
4612                 n = strtoul(eq, &end, 10);
4613                 if (*end)
4614                         return 0;
4615                 *val = n;
4616         }
4617         return 1;
4618 }
4619
4620 static inline int short_opt(char opt, const char **argv,
4621                             const char **optarg)
4622 {
4623         const char *arg = argv[0];
4624         if (arg[0] != '-' || arg[1] != opt)
4625                 return 0;
4626         if (arg[2] != '\0') {
4627                 *optarg = arg + 2;
4628                 return 1;
4629         }
4630         if (!argv[1])
4631                 die("Option '%c' requires a value", opt);
4632         *optarg = argv[1];
4633         return 2;
4634 }
4635
4636 int parse_long_opt(const char *opt, const char **argv,
4637                    const char **optarg)
4638 {
4639         const char *arg = argv[0];
4640         if (!skip_prefix(arg, "--", &arg))
4641                 return 0;
4642         if (!skip_prefix(arg, opt, &arg))
4643                 return 0;
4644         if (*arg == '=') { /* stuck form: --option=value */
4645                 *optarg = arg + 1;
4646                 return 1;
4647         }
4648         if (*arg != '\0')
4649                 return 0;
4650         /* separate form: --option value */
4651         if (!argv[1])
4652                 die("Option '--%s' requires a value", opt);
4653         *optarg = argv[1];
4654         return 2;
4655 }
4656
4657 static int diff_opt_stat(const struct option *opt, const char *value, int unset)
4658 {
4659         struct diff_options *options = opt->value;
4660         int width = options->stat_width;
4661         int name_width = options->stat_name_width;
4662         int graph_width = options->stat_graph_width;
4663         int count = options->stat_count;
4664         char *end;
4665
4666         BUG_ON_OPT_NEG(unset);
4667
4668         if (!strcmp(opt->long_name, "stat")) {
4669                 if (value) {
4670                         width = strtoul(value, &end, 10);
4671                         if (*end == ',')
4672                                 name_width = strtoul(end+1, &end, 10);
4673                         if (*end == ',')
4674                                 count = strtoul(end+1, &end, 10);
4675                         if (*end)
4676                                 return error(_("invalid --stat value: %s"), value);
4677                 }
4678         } else if (!strcmp(opt->long_name, "stat-width")) {
4679                 width = strtoul(value, &end, 10);
4680                 if (*end)
4681                         return error(_("%s expects a numerical value"),
4682                                      opt->long_name);
4683         } else if (!strcmp(opt->long_name, "stat-name-width")) {
4684                 name_width = strtoul(value, &end, 10);
4685                 if (*end)
4686                         return error(_("%s expects a numerical value"),
4687                                      opt->long_name);
4688         } else if (!strcmp(opt->long_name, "stat-graph-width")) {
4689                 graph_width = strtoul(value, &end, 10);
4690                 if (*end)
4691                         return error(_("%s expects a numerical value"),
4692                                      opt->long_name);
4693         } else if (!strcmp(opt->long_name, "stat-count")) {
4694                 count = strtoul(value, &end, 10);
4695                 if (*end)
4696                         return error(_("%s expects a numerical value"),
4697                                      opt->long_name);
4698         } else
4699                 BUG("%s should not get here", opt->long_name);
4700
4701         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4702         options->stat_name_width = name_width;
4703         options->stat_graph_width = graph_width;
4704         options->stat_width = width;
4705         options->stat_count = count;
4706         return 0;
4707 }
4708
4709 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4710 {
4711         struct strbuf errmsg = STRBUF_INIT;
4712         if (parse_dirstat_params(options, params, &errmsg))
4713                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4714                     errmsg.buf);
4715         strbuf_release(&errmsg);
4716         /*
4717          * The caller knows a dirstat-related option is given from the command
4718          * line; allow it to say "return this_function();"
4719          */
4720         options->output_format |= DIFF_FORMAT_DIRSTAT;
4721         return 1;
4722 }
4723
4724 static int parse_submodule_opt(struct diff_options *options, const char *value)
4725 {
4726         if (parse_submodule_params(options, value))
4727                 die(_("Failed to parse --submodule option parameter: '%s'"),
4728                         value);
4729         return 1;
4730 }
4731
4732 static const char diff_status_letters[] = {
4733         DIFF_STATUS_ADDED,
4734         DIFF_STATUS_COPIED,
4735         DIFF_STATUS_DELETED,
4736         DIFF_STATUS_MODIFIED,
4737         DIFF_STATUS_RENAMED,
4738         DIFF_STATUS_TYPE_CHANGED,
4739         DIFF_STATUS_UNKNOWN,
4740         DIFF_STATUS_UNMERGED,
4741         DIFF_STATUS_FILTER_AON,
4742         DIFF_STATUS_FILTER_BROKEN,
4743         '\0',
4744 };
4745
4746 static unsigned int filter_bit['Z' + 1];
4747
4748 static void prepare_filter_bits(void)
4749 {
4750         int i;
4751
4752         if (!filter_bit[DIFF_STATUS_ADDED]) {
4753                 for (i = 0; diff_status_letters[i]; i++)
4754                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4755         }
4756 }
4757
4758 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4759 {
4760         return opt->filter & filter_bit[(int) status];
4761 }
4762
4763 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4764 {
4765         int i, optch;
4766
4767         prepare_filter_bits();
4768
4769         /*
4770          * If there is a negation e.g. 'd' in the input, and we haven't
4771          * initialized the filter field with another --diff-filter, start
4772          * from full set of bits, except for AON.
4773          */
4774         if (!opt->filter) {
4775                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4776                         if (optch < 'a' || 'z' < optch)
4777                                 continue;
4778                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4779                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4780                         break;
4781                 }
4782         }
4783
4784         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4785                 unsigned int bit;
4786                 int negate;
4787
4788                 if ('a' <= optch && optch <= 'z') {
4789                         negate = 1;
4790                         optch = toupper(optch);
4791                 } else {
4792                         negate = 0;
4793                 }
4794
4795                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4796                 if (!bit)
4797                         return optarg[i];
4798                 if (negate)
4799                         opt->filter &= ~bit;
4800                 else
4801                         opt->filter |= bit;
4802         }
4803         return 0;
4804 }
4805
4806 static void enable_patch_output(int *fmt)
4807 {
4808         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4809         *fmt |= DIFF_FORMAT_PATCH;
4810 }
4811
4812 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4813 {
4814         int val = parse_ws_error_highlight(arg);
4815
4816         if (val < 0) {
4817                 error("unknown value after ws-error-highlight=%.*s",
4818                       -1 - val, arg);
4819                 return 0;
4820         }
4821         opt->ws_error_highlight = val;
4822         return 1;
4823 }
4824
4825 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4826 {
4827         struct object_id oid;
4828
4829         if (get_oid(arg, &oid))
4830                 return error("unable to resolve '%s'", arg);
4831
4832         if (!opt->objfind)
4833                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4834
4835         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4836         opt->flags.recursive = 1;
4837         opt->flags.tree_in_recursive = 1;
4838         oidset_insert(opt->objfind, &oid);
4839         return 1;
4840 }
4841
4842 static int diff_opt_anchored(const struct option *opt,
4843                              const char *arg, int unset)
4844 {
4845         struct diff_options *options = opt->value;
4846
4847         BUG_ON_OPT_NEG(unset);
4848         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4849         ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4850                    options->anchors_alloc);
4851         options->anchors[options->anchors_nr++] = xstrdup(arg);
4852         return 0;
4853 }
4854
4855 static int diff_opt_binary(const struct option *opt,
4856                            const char *arg, int unset)
4857 {
4858         struct diff_options *options = opt->value;
4859
4860         BUG_ON_OPT_NEG(unset);
4861         BUG_ON_OPT_ARG(arg);
4862         enable_patch_output(&options->output_format);
4863         options->flags.binary = 1;
4864         return 0;
4865 }
4866
4867 static int diff_opt_break_rewrites(const struct option *opt,
4868                                    const char *arg, int unset)
4869 {
4870         int *break_opt = opt->value;
4871         int opt1, opt2;
4872
4873         BUG_ON_OPT_NEG(unset);
4874         if (!arg)
4875                 arg = "";
4876         opt1 = parse_rename_score(&arg);
4877         if (*arg == 0)
4878                 opt2 = 0;
4879         else if (*arg != '/')
4880                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4881         else {
4882                 arg++;
4883                 opt2 = parse_rename_score(&arg);
4884         }
4885         if (*arg != 0)
4886                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4887         *break_opt = opt1 | (opt2 << 16);
4888         return 0;
4889 }
4890
4891 static int diff_opt_char(const struct option *opt,
4892                          const char *arg, int unset)
4893 {
4894         char *value = opt->value;
4895
4896         BUG_ON_OPT_NEG(unset);
4897         if (arg[1])
4898                 return error(_("%s expects a character, got '%s'"),
4899                              opt->long_name, arg);
4900         *value = arg[0];
4901         return 0;
4902 }
4903
4904 static int diff_opt_compact_summary(const struct option *opt,
4905                                     const char *arg, int unset)
4906 {
4907         struct diff_options *options = opt->value;
4908
4909         BUG_ON_OPT_ARG(arg);
4910         if (unset) {
4911                 options->flags.stat_with_summary = 0;
4912         } else {
4913                 options->flags.stat_with_summary = 1;
4914                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4915         }
4916         return 0;
4917 }
4918
4919 static int diff_opt_diff_algorithm(const struct option *opt,
4920                                    const char *arg, int unset)
4921 {
4922         struct diff_options *options = opt->value;
4923         long value = parse_algorithm_value(arg);
4924
4925         BUG_ON_OPT_NEG(unset);
4926         if (value < 0)
4927                 return error(_("option diff-algorithm accepts \"myers\", "
4928                                "\"minimal\", \"patience\" and \"histogram\""));
4929
4930         /* clear out previous settings */
4931         DIFF_XDL_CLR(options, NEED_MINIMAL);
4932         options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4933         options->xdl_opts |= value;
4934         return 0;
4935 }
4936
4937 static int diff_opt_dirstat(const struct option *opt,
4938                             const char *arg, int unset)
4939 {
4940         struct diff_options *options = opt->value;
4941
4942         BUG_ON_OPT_NEG(unset);
4943         if (!strcmp(opt->long_name, "cumulative")) {
4944                 if (arg)
4945                         BUG("how come --cumulative take a value?");
4946                 arg = "cumulative";
4947         } else if (!strcmp(opt->long_name, "dirstat-by-file"))
4948                 parse_dirstat_opt(options, "files");
4949         parse_dirstat_opt(options, arg ? arg : "");
4950         return 0;
4951 }
4952
4953 static int diff_opt_find_copies(const struct option *opt,
4954                                 const char *arg, int unset)
4955 {
4956         struct diff_options *options = opt->value;
4957
4958         BUG_ON_OPT_NEG(unset);
4959         if (!arg)
4960                 arg = "";
4961         options->rename_score = parse_rename_score(&arg);
4962         if (*arg != 0)
4963                 return error(_("invalid argument to %s"), opt->long_name);
4964
4965         if (options->detect_rename == DIFF_DETECT_COPY)
4966                 options->flags.find_copies_harder = 1;
4967         else
4968                 options->detect_rename = DIFF_DETECT_COPY;
4969
4970         return 0;
4971 }
4972
4973 static int diff_opt_find_renames(const struct option *opt,
4974                                  const char *arg, int unset)
4975 {
4976         struct diff_options *options = opt->value;
4977
4978         BUG_ON_OPT_NEG(unset);
4979         if (!arg)
4980                 arg = "";
4981         options->rename_score = parse_rename_score(&arg);
4982         if (*arg != 0)
4983                 return error(_("invalid argument to %s"), opt->long_name);
4984
4985         options->detect_rename = DIFF_DETECT_RENAME;
4986         return 0;
4987 }
4988
4989 static enum parse_opt_result diff_opt_output(struct parse_opt_ctx_t *ctx,
4990                                              const struct option *opt,
4991                                              const char *arg, int unset)
4992 {
4993         struct diff_options *options = opt->value;
4994         char *path;
4995
4996         BUG_ON_OPT_NEG(unset);
4997         path = prefix_filename(ctx->prefix, arg);
4998         options->file = xfopen(path, "w");
4999         options->close_file = 1;
5000         if (options->use_color != GIT_COLOR_ALWAYS)
5001                 options->use_color = GIT_COLOR_NEVER;
5002         free(path);
5003         return 0;
5004 }
5005
5006 static int diff_opt_patience(const struct option *opt,
5007                              const char *arg, int unset)
5008 {
5009         struct diff_options *options = opt->value;
5010         int i;
5011
5012         BUG_ON_OPT_NEG(unset);
5013         BUG_ON_OPT_ARG(arg);
5014         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5015         /*
5016          * Both --patience and --anchored use PATIENCE_DIFF
5017          * internally, so remove any anchors previously
5018          * specified.
5019          */
5020         for (i = 0; i < options->anchors_nr; i++)
5021                 free(options->anchors[i]);
5022         options->anchors_nr = 0;
5023         return 0;
5024 }
5025
5026 static int diff_opt_relative(const struct option *opt,
5027                              const char *arg, int unset)
5028 {
5029         struct diff_options *options = opt->value;
5030
5031         BUG_ON_OPT_NEG(unset);
5032         options->flags.relative_name = 1;
5033         if (arg)
5034                 options->prefix = arg;
5035         return 0;
5036 }
5037
5038 static int diff_opt_unified(const struct option *opt,
5039                             const char *arg, int unset)
5040 {
5041         struct diff_options *options = opt->value;
5042         char *s;
5043
5044         BUG_ON_OPT_NEG(unset);
5045
5046         options->context = strtol(arg, &s, 10);
5047         if (*s)
5048                 return error(_("%s expects a numerical value"), "--unified");
5049         enable_patch_output(&options->output_format);
5050
5051         return 0;
5052 }
5053
5054 static void prep_parse_options(struct diff_options *options)
5055 {
5056         struct option parseopts[] = {
5057                 OPT_GROUP(N_("Diff output format options")),
5058                 OPT_BITOP('p', "patch", &options->output_format,
5059                           N_("generate patch"),
5060                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5061                 OPT_BIT_F('s', "no-patch", &options->output_format,
5062                           N_("suppress diff output"),
5063                           DIFF_FORMAT_NO_OUTPUT, PARSE_OPT_NONEG),
5064                 OPT_BITOP('u', NULL, &options->output_format,
5065                           N_("generate patch"),
5066                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5067                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
5068                                N_("generate diffs with <n> lines context"),
5069                                PARSE_OPT_NONEG, diff_opt_unified),
5070                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
5071                          N_("generate diffs with <n> lines context")),
5072                 OPT_BIT_F(0, "raw", &options->output_format,
5073                           N_("generate the diff in raw format"),
5074                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
5075                 OPT_BITOP(0, "patch-with-raw", &options->output_format,
5076                           N_("synonym for '-p --raw'"),
5077                           DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW,
5078                           DIFF_FORMAT_NO_OUTPUT),
5079                 OPT_BITOP(0, "patch-with-stat", &options->output_format,
5080                           N_("synonym for '-p --stat'"),
5081                           DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT,
5082                           DIFF_FORMAT_NO_OUTPUT),
5083                 OPT_BIT_F(0, "numstat", &options->output_format,
5084                           N_("machine friendly --stat"),
5085                           DIFF_FORMAT_NUMSTAT, PARSE_OPT_NONEG),
5086                 OPT_BIT_F(0, "shortstat", &options->output_format,
5087                           N_("output only the last line of --stat"),
5088                           DIFF_FORMAT_SHORTSTAT, PARSE_OPT_NONEG),
5089                 OPT_CALLBACK_F('X', "dirstat", options, N_("<param1,param2>..."),
5090                                N_("output the distribution of relative amount of changes for each sub-directory"),
5091                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5092                                diff_opt_dirstat),
5093                 OPT_CALLBACK_F(0, "cumulative", options, NULL,
5094                                N_("synonym for --dirstat=cumulative"),
5095                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5096                                diff_opt_dirstat),
5097                 OPT_CALLBACK_F(0, "dirstat-by-file", options, N_("<param1,param2>..."),
5098                                N_("synonym for --dirstat=files,param1,param2..."),
5099                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5100                                diff_opt_dirstat),
5101                 OPT_BIT_F(0, "check", &options->output_format,
5102                           N_("warn if changes introduce conflict markers or whitespace errors"),
5103                           DIFF_FORMAT_CHECKDIFF, PARSE_OPT_NONEG),
5104                 OPT_BIT_F(0, "summary", &options->output_format,
5105                           N_("condensed summary such as creations, renames and mode changes"),
5106                           DIFF_FORMAT_SUMMARY, PARSE_OPT_NONEG),
5107                 OPT_BIT_F(0, "name-only", &options->output_format,
5108                           N_("show only names of changed files"),
5109                           DIFF_FORMAT_NAME, PARSE_OPT_NONEG),
5110                 OPT_BIT_F(0, "name-status", &options->output_format,
5111                           N_("show only names and status of changed files"),
5112                           DIFF_FORMAT_NAME_STATUS, PARSE_OPT_NONEG),
5113                 OPT_CALLBACK_F(0, "stat", options, N_("<width>[,<name-width>[,<count>]]"),
5114                                N_("generate diffstat"),
5115                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_stat),
5116                 OPT_CALLBACK_F(0, "stat-width", options, N_("<width>"),
5117                                N_("generate diffstat with a given width"),
5118                                PARSE_OPT_NONEG, diff_opt_stat),
5119                 OPT_CALLBACK_F(0, "stat-name-width", options, N_("<width>"),
5120                                N_("generate diffstat with a given name width"),
5121                                PARSE_OPT_NONEG, diff_opt_stat),
5122                 OPT_CALLBACK_F(0, "stat-graph-width", options, N_("<width>"),
5123                                N_("generate diffstat with a given graph width"),
5124                                PARSE_OPT_NONEG, diff_opt_stat),
5125                 OPT_CALLBACK_F(0, "stat-count", options, N_("<count>"),
5126                                N_("generate diffstat with limited lines"),
5127                                PARSE_OPT_NONEG, diff_opt_stat),
5128                 OPT_CALLBACK_F(0, "compact-summary", options, NULL,
5129                                N_("generate compact summary in diffstat"),
5130                                PARSE_OPT_NOARG, diff_opt_compact_summary),
5131                 OPT_CALLBACK_F(0, "binary", options, NULL,
5132                                N_("output a binary diff that can be applied"),
5133                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_binary),
5134                 OPT_BOOL(0, "full-index", &options->flags.full_index,
5135                          N_("show full pre- and post-image object names on the \"index\" lines")),
5136                 OPT_CALLBACK_F(0, "output-indicator-new",
5137                                &options->output_indicators[OUTPUT_INDICATOR_NEW],
5138                                N_("<char>"),
5139                                N_("specify the character to indicate a new line instead of '+'"),
5140                                PARSE_OPT_NONEG, diff_opt_char),
5141                 OPT_CALLBACK_F(0, "output-indicator-old",
5142                                &options->output_indicators[OUTPUT_INDICATOR_OLD],
5143                                N_("<char>"),
5144                                N_("specify the character to indicate an old line instead of '-'"),
5145                                PARSE_OPT_NONEG, diff_opt_char),
5146                 OPT_CALLBACK_F(0, "output-indicator-context",
5147                                &options->output_indicators[OUTPUT_INDICATOR_CONTEXT],
5148                                N_("<char>"),
5149                                N_("specify the character to indicate a context instead of ' '"),
5150                                PARSE_OPT_NONEG, diff_opt_char),
5151
5152                 OPT_GROUP(N_("Diff rename options")),
5153                 OPT_CALLBACK_F('B', "break-rewrites", &options->break_opt, N_("<n>[/<m>]"),
5154                                N_("break complete rewrite changes into pairs of delete and create"),
5155                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5156                                diff_opt_break_rewrites),
5157                 OPT_CALLBACK_F('M', "find-renames", options, N_("<n>"),
5158                                N_("detect renames"),
5159                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5160                                diff_opt_find_renames),
5161                 OPT_SET_INT_F('D', "irreversible-delete", &options->irreversible_delete,
5162                               N_("omit the preimage for deletes"),
5163                               1, PARSE_OPT_NONEG),
5164                 OPT_CALLBACK_F('C', "find-copies", options, N_("<n>"),
5165                                N_("detect copies"),
5166                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5167                                diff_opt_find_copies),
5168                 OPT_BOOL(0, "find-copies-harder", &options->flags.find_copies_harder,
5169                          N_("use unmodified files as source to find copies")),
5170                 OPT_SET_INT_F(0, "no-renames", &options->detect_rename,
5171                               N_("disable rename detection"),
5172                               0, PARSE_OPT_NONEG),
5173                 OPT_BOOL(0, "rename-empty", &options->flags.rename_empty,
5174                          N_("use empty blobs as rename source")),
5175
5176                 OPT_GROUP(N_("Diff algorithm options")),
5177                 OPT_BIT(0, "minimal", &options->xdl_opts,
5178                         N_("produce the smallest possible diff"),
5179                         XDF_NEED_MINIMAL),
5180                 OPT_BIT_F('w', "ignore-all-space", &options->xdl_opts,
5181                           N_("ignore whitespace when comparing lines"),
5182                           XDF_IGNORE_WHITESPACE, PARSE_OPT_NONEG),
5183                 OPT_BIT_F('b', "ignore-space-change", &options->xdl_opts,
5184                           N_("ignore changes in amount of whitespace"),
5185                           XDF_IGNORE_WHITESPACE_CHANGE, PARSE_OPT_NONEG),
5186                 OPT_BIT_F(0, "ignore-space-at-eol", &options->xdl_opts,
5187                           N_("ignore changes in whitespace at EOL"),
5188                           XDF_IGNORE_WHITESPACE_AT_EOL, PARSE_OPT_NONEG),
5189                 OPT_BIT_F(0, "ignore-cr-at-eol", &options->xdl_opts,
5190                           N_("ignore carrier-return at the end of line"),
5191                           XDF_IGNORE_CR_AT_EOL, PARSE_OPT_NONEG),
5192                 OPT_BIT_F(0, "ignore-blank-lines", &options->xdl_opts,
5193                           N_("ignore changes whose lines are all blank"),
5194                           XDF_IGNORE_BLANK_LINES, PARSE_OPT_NONEG),
5195                 OPT_BIT(0, "indent-heuristic", &options->xdl_opts,
5196                         N_("heuristic to shift diff hunk boundaries for easy reading"),
5197                         XDF_INDENT_HEURISTIC),
5198                 OPT_CALLBACK_F(0, "patience", options, NULL,
5199                                N_("generate diff using the \"patience diff\" algorithm"),
5200                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5201                                diff_opt_patience),
5202                 OPT_BITOP(0, "histogram", &options->xdl_opts,
5203                           N_("generate diff using the \"histogram diff\" algorithm"),
5204                           XDF_HISTOGRAM_DIFF, XDF_DIFF_ALGORITHM_MASK),
5205                 OPT_CALLBACK_F(0, "diff-algorithm", options, N_("<algorithm>"),
5206                                N_("choose a diff algorithm"),
5207                                PARSE_OPT_NONEG, diff_opt_diff_algorithm),
5208                 OPT_CALLBACK_F(0, "anchored", options, N_("<text>"),
5209                                N_("generate diff using the \"anchored diff\" algorithm"),
5210                                PARSE_OPT_NONEG, diff_opt_anchored),
5211
5212                 OPT_GROUP(N_("Diff other options")),
5213                 OPT_CALLBACK_F(0, "relative", options, N_("<prefix>"),
5214                                N_("when run from subdir, exclude changes outside and show relative paths"),
5215                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5216                                diff_opt_relative),
5217                 { OPTION_CALLBACK, 0, "output", options, N_("<file>"),
5218                   N_("Output to a specific file"),
5219                   PARSE_OPT_NONEG, NULL, 0, diff_opt_output },
5220
5221                 OPT_END()
5222         };
5223
5224         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
5225         memcpy(options->parseopts, parseopts, sizeof(parseopts));
5226 }
5227
5228 int diff_opt_parse(struct diff_options *options,
5229                    const char **av, int ac, const char *prefix)
5230 {
5231         const char *arg = av[0];
5232         const char *optarg;
5233         int argcount;
5234
5235         if (!prefix)
5236                 prefix = "";
5237
5238         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
5239                            PARSE_OPT_KEEP_DASHDASH |
5240                            PARSE_OPT_KEEP_UNKNOWN |
5241                            PARSE_OPT_NO_INTERNAL_HELP |
5242                            PARSE_OPT_ONE_SHOT |
5243                            PARSE_OPT_STOP_AT_NON_OPTION);
5244
5245         if (ac)
5246                 return ac;
5247
5248         /* flags options */
5249         if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5250                 options->flags.text = 1;
5251         else if (!strcmp(arg, "-R"))
5252                 options->flags.reverse_diff = 1;
5253         else if (!strcmp(arg, "--follow"))
5254                 options->flags.follow_renames = 1;
5255         else if (!strcmp(arg, "--no-follow")) {
5256                 options->flags.follow_renames = 0;
5257                 options->flags.default_follow_renames = 0;
5258         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5259                 int value = git_config_colorbool(NULL, arg);
5260                 if (value < 0)
5261                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5262                 options->use_color = value;
5263         }
5264         else if (!strcmp(arg, "--no-color"))
5265                 options->use_color = 0;
5266         else if (!strcmp(arg, "--color-moved")) {
5267                 if (diff_color_moved_default)
5268                         options->color_moved = diff_color_moved_default;
5269                 if (options->color_moved == COLOR_MOVED_NO)
5270                         options->color_moved = COLOR_MOVED_DEFAULT;
5271         } else if (!strcmp(arg, "--no-color-moved"))
5272                 options->color_moved = COLOR_MOVED_NO;
5273         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5274                 int cm = parse_color_moved(arg);
5275                 if (cm < 0)
5276                         return error("bad --color-moved argument: %s", arg);
5277                 options->color_moved = cm;
5278         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5279                 unsigned cm = parse_color_moved_ws(arg);
5280                 if (cm & COLOR_MOVED_WS_ERROR)
5281                         return -1;
5282                 options->color_moved_ws_handling = cm;
5283         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5284                 options->use_color = 1;
5285                 options->word_diff = DIFF_WORDS_COLOR;
5286         }
5287         else if (!strcmp(arg, "--word-diff")) {
5288                 if (options->word_diff == DIFF_WORDS_NONE)
5289                         options->word_diff = DIFF_WORDS_PLAIN;
5290         }
5291         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5292                 if (!strcmp(arg, "plain"))
5293                         options->word_diff = DIFF_WORDS_PLAIN;
5294                 else if (!strcmp(arg, "color")) {
5295                         options->use_color = 1;
5296                         options->word_diff = DIFF_WORDS_COLOR;
5297                 }
5298                 else if (!strcmp(arg, "porcelain"))
5299                         options->word_diff = DIFF_WORDS_PORCELAIN;
5300                 else if (!strcmp(arg, "none"))
5301                         options->word_diff = DIFF_WORDS_NONE;
5302                 else
5303                         die("bad --word-diff argument: %s", arg);
5304         }
5305         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5306                 if (options->word_diff == DIFF_WORDS_NONE)
5307                         options->word_diff = DIFF_WORDS_PLAIN;
5308                 options->word_regex = optarg;
5309                 return argcount;
5310         }
5311         else if (!strcmp(arg, "--exit-code"))
5312                 options->flags.exit_with_status = 1;
5313         else if (!strcmp(arg, "--quiet"))
5314                 options->flags.quick = 1;
5315         else if (!strcmp(arg, "--ext-diff"))
5316                 options->flags.allow_external = 1;
5317         else if (!strcmp(arg, "--no-ext-diff"))
5318                 options->flags.allow_external = 0;
5319         else if (!strcmp(arg, "--textconv")) {
5320                 options->flags.allow_textconv = 1;
5321                 options->flags.textconv_set_via_cmdline = 1;
5322         } else if (!strcmp(arg, "--no-textconv"))
5323                 options->flags.allow_textconv = 0;
5324         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5325                 options->flags.override_submodule_config = 1;
5326                 handle_ignore_submodules_arg(options, arg);
5327         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5328                 return parse_submodule_opt(options, arg);
5329         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5330                 return parse_ws_error_highlight_opt(options, arg);
5331         else if (!strcmp(arg, "--ita-invisible-in-index"))
5332                 options->ita_invisible_in_index = 1;
5333         else if (!strcmp(arg, "--ita-visible-in-index"))
5334                 options->ita_invisible_in_index = 0;
5335
5336         /* misc options */
5337         else if (!strcmp(arg, "-z"))
5338                 options->line_termination = 0;
5339         else if ((argcount = short_opt('l', av, &optarg))) {
5340                 options->rename_limit = strtoul(optarg, NULL, 10);
5341                 return argcount;
5342         }
5343         else if ((argcount = short_opt('S', av, &optarg))) {
5344                 options->pickaxe = optarg;
5345                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5346                 return argcount;
5347         } else if ((argcount = short_opt('G', av, &optarg))) {
5348                 options->pickaxe = optarg;
5349                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5350                 return argcount;
5351         }
5352         else if (!strcmp(arg, "--pickaxe-all"))
5353                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5354         else if (!strcmp(arg, "--pickaxe-regex"))
5355                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5356         else if ((argcount = short_opt('O', av, &optarg))) {
5357                 options->orderfile = prefix_filename(prefix, optarg);
5358                 return argcount;
5359         } else if (skip_prefix(arg, "--find-object=", &arg))
5360                 return parse_objfind_opt(options, arg);
5361         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5362                 int offending = parse_diff_filter_opt(optarg, options);
5363                 if (offending)
5364                         die("unknown change class '%c' in --diff-filter=%s",
5365                             offending, optarg);
5366                 return argcount;
5367         }
5368         else if (!strcmp(arg, "--no-abbrev"))
5369                 options->abbrev = 0;
5370         else if (!strcmp(arg, "--abbrev"))
5371                 options->abbrev = DEFAULT_ABBREV;
5372         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5373                 options->abbrev = strtoul(arg, NULL, 10);
5374                 if (options->abbrev < MINIMUM_ABBREV)
5375                         options->abbrev = MINIMUM_ABBREV;
5376                 else if (the_hash_algo->hexsz < options->abbrev)
5377                         options->abbrev = the_hash_algo->hexsz;
5378         }
5379         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5380                 options->a_prefix = optarg;
5381                 return argcount;
5382         }
5383         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5384                 options->line_prefix = optarg;
5385                 options->line_prefix_length = strlen(options->line_prefix);
5386                 graph_setup_line_prefix(options);
5387                 return argcount;
5388         }
5389         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5390                 options->b_prefix = optarg;
5391                 return argcount;
5392         }
5393         else if (!strcmp(arg, "--no-prefix"))
5394                 options->a_prefix = options->b_prefix = "";
5395         else if (opt_arg(arg, '\0', "inter-hunk-context",
5396                          &options->interhunkcontext))
5397                 ;
5398         else
5399                 return 0;
5400         return 1;
5401 }
5402
5403 int parse_rename_score(const char **cp_p)
5404 {
5405         unsigned long num, scale;
5406         int ch, dot;
5407         const char *cp = *cp_p;
5408
5409         num = 0;
5410         scale = 1;
5411         dot = 0;
5412         for (;;) {
5413                 ch = *cp;
5414                 if ( !dot && ch == '.' ) {
5415                         scale = 1;
5416                         dot = 1;
5417                 } else if ( ch == '%' ) {
5418                         scale = dot ? scale*100 : 100;
5419                         cp++;   /* % is always at the end */
5420                         break;
5421                 } else if ( ch >= '0' && ch <= '9' ) {
5422                         if ( scale < 100000 ) {
5423                                 scale *= 10;
5424                                 num = (num*10) + (ch-'0');
5425                         }
5426                 } else {
5427                         break;
5428                 }
5429                 cp++;
5430         }
5431         *cp_p = cp;
5432
5433         /* user says num divided by scale and we say internally that
5434          * is MAX_SCORE * num / scale.
5435          */
5436         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5437 }
5438
5439 struct diff_queue_struct diff_queued_diff;
5440
5441 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5442 {
5443         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5444         queue->queue[queue->nr++] = dp;
5445 }
5446
5447 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5448                                  struct diff_filespec *one,
5449                                  struct diff_filespec *two)
5450 {
5451         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5452         dp->one = one;
5453         dp->two = two;
5454         if (queue)
5455                 diff_q(queue, dp);
5456         return dp;
5457 }
5458
5459 void diff_free_filepair(struct diff_filepair *p)
5460 {
5461         free_filespec(p->one);
5462         free_filespec(p->two);
5463         free(p);
5464 }
5465
5466 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5467 {
5468         int abblen;
5469         const char *abbrev;
5470
5471         /* Do we want all 40 hex characters? */
5472         if (len == the_hash_algo->hexsz)
5473                 return oid_to_hex(oid);
5474
5475         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5476         abbrev = diff_abbrev_oid(oid, len);
5477
5478         if (!print_sha1_ellipsis())
5479                 return abbrev;
5480
5481         abblen = strlen(abbrev);
5482
5483         /*
5484          * In well-behaved cases, where the abbreviated result is the
5485          * same as the requested length, append three dots after the
5486          * abbreviation (hence the whole logic is limited to the case
5487          * where abblen < 37); when the actual abbreviated result is a
5488          * bit longer than the requested length, we reduce the number
5489          * of dots so that they match the well-behaved ones.  However,
5490          * if the actual abbreviation is longer than the requested
5491          * length by more than three, we give up on aligning, and add
5492          * three dots anyway, to indicate that the output is not the
5493          * full object name.  Yes, this may be suboptimal, but this
5494          * appears only in "diff --raw --abbrev" output and it is not
5495          * worth the effort to change it now.  Note that this would
5496          * likely to work fine when the automatic sizing of default
5497          * abbreviation length is used--we would be fed -1 in "len" in
5498          * that case, and will end up always appending three-dots, but
5499          * the automatic sizing is supposed to give abblen that ensures
5500          * uniqueness across all objects (statistically speaking).
5501          */
5502         if (abblen < the_hash_algo->hexsz - 3) {
5503                 static char hex[GIT_MAX_HEXSZ + 1];
5504                 if (len < abblen && abblen <= len + 2)
5505                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5506                 else
5507                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5508                 return hex;
5509         }
5510
5511         return oid_to_hex(oid);
5512 }
5513
5514 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5515 {
5516         int line_termination = opt->line_termination;
5517         int inter_name_termination = line_termination ? '\t' : '\0';
5518
5519         fprintf(opt->file, "%s", diff_line_prefix(opt));
5520         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5521                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5522                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5523                 fprintf(opt->file, "%s ",
5524                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5525         }
5526         if (p->score) {
5527                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5528                         inter_name_termination);
5529         } else {
5530                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5531         }
5532
5533         if (p->status == DIFF_STATUS_COPIED ||
5534             p->status == DIFF_STATUS_RENAMED) {
5535                 const char *name_a, *name_b;
5536                 name_a = p->one->path;
5537                 name_b = p->two->path;
5538                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5539                 write_name_quoted(name_a, opt->file, inter_name_termination);
5540                 write_name_quoted(name_b, opt->file, line_termination);
5541         } else {
5542                 const char *name_a, *name_b;
5543                 name_a = p->one->mode ? p->one->path : p->two->path;
5544                 name_b = NULL;
5545                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5546                 write_name_quoted(name_a, opt->file, line_termination);
5547         }
5548 }
5549
5550 int diff_unmodified_pair(struct diff_filepair *p)
5551 {
5552         /* This function is written stricter than necessary to support
5553          * the currently implemented transformers, but the idea is to
5554          * let transformers to produce diff_filepairs any way they want,
5555          * and filter and clean them up here before producing the output.
5556          */
5557         struct diff_filespec *one = p->one, *two = p->two;
5558
5559         if (DIFF_PAIR_UNMERGED(p))
5560                 return 0; /* unmerged is interesting */
5561
5562         /* deletion, addition, mode or type change
5563          * and rename are all interesting.
5564          */
5565         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5566             DIFF_PAIR_MODE_CHANGED(p) ||
5567             strcmp(one->path, two->path))
5568                 return 0;
5569
5570         /* both are valid and point at the same path.  that is, we are
5571          * dealing with a change.
5572          */
5573         if (one->oid_valid && two->oid_valid &&
5574             oideq(&one->oid, &two->oid) &&
5575             !one->dirty_submodule && !two->dirty_submodule)
5576                 return 1; /* no change */
5577         if (!one->oid_valid && !two->oid_valid)
5578                 return 1; /* both look at the same file on the filesystem. */
5579         return 0;
5580 }
5581
5582 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5583 {
5584         if (diff_unmodified_pair(p))
5585                 return;
5586
5587         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5588             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5589                 return; /* no tree diffs in patch format */
5590
5591         run_diff(p, o);
5592 }
5593
5594 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5595                             struct diffstat_t *diffstat)
5596 {
5597         if (diff_unmodified_pair(p))
5598                 return;
5599
5600         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5601             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5602                 return; /* no useful stat for tree diffs */
5603
5604         run_diffstat(p, o, diffstat);
5605 }
5606
5607 static void diff_flush_checkdiff(struct diff_filepair *p,
5608                 struct diff_options *o)
5609 {
5610         if (diff_unmodified_pair(p))
5611                 return;
5612
5613         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5614             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5615                 return; /* nothing to check in tree diffs */
5616
5617         run_checkdiff(p, o);
5618 }
5619
5620 int diff_queue_is_empty(void)
5621 {
5622         struct diff_queue_struct *q = &diff_queued_diff;
5623         int i;
5624         for (i = 0; i < q->nr; i++)
5625                 if (!diff_unmodified_pair(q->queue[i]))
5626                         return 0;
5627         return 1;
5628 }
5629
5630 #if DIFF_DEBUG
5631 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5632 {
5633         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5634                 x, one ? one : "",
5635                 s->path,
5636                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5637                 s->mode,
5638                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5639         fprintf(stderr, "queue[%d] %s size %lu\n",
5640                 x, one ? one : "",
5641                 s->size);
5642 }
5643
5644 void diff_debug_filepair(const struct diff_filepair *p, int i)
5645 {
5646         diff_debug_filespec(p->one, i, "one");
5647         diff_debug_filespec(p->two, i, "two");
5648         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5649                 p->score, p->status ? p->status : '?',
5650                 p->one->rename_used, p->broken_pair);
5651 }
5652
5653 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5654 {
5655         int i;
5656         if (msg)
5657                 fprintf(stderr, "%s\n", msg);
5658         fprintf(stderr, "q->nr = %d\n", q->nr);
5659         for (i = 0; i < q->nr; i++) {
5660                 struct diff_filepair *p = q->queue[i];
5661                 diff_debug_filepair(p, i);
5662         }
5663 }
5664 #endif
5665
5666 static void diff_resolve_rename_copy(void)
5667 {
5668         int i;
5669         struct diff_filepair *p;
5670         struct diff_queue_struct *q = &diff_queued_diff;
5671
5672         diff_debug_queue("resolve-rename-copy", q);
5673
5674         for (i = 0; i < q->nr; i++) {
5675                 p = q->queue[i];
5676                 p->status = 0; /* undecided */
5677                 if (DIFF_PAIR_UNMERGED(p))
5678                         p->status = DIFF_STATUS_UNMERGED;
5679                 else if (!DIFF_FILE_VALID(p->one))
5680                         p->status = DIFF_STATUS_ADDED;
5681                 else if (!DIFF_FILE_VALID(p->two))
5682                         p->status = DIFF_STATUS_DELETED;
5683                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5684                         p->status = DIFF_STATUS_TYPE_CHANGED;
5685
5686                 /* from this point on, we are dealing with a pair
5687                  * whose both sides are valid and of the same type, i.e.
5688                  * either in-place edit or rename/copy edit.
5689                  */
5690                 else if (DIFF_PAIR_RENAME(p)) {
5691                         /*
5692                          * A rename might have re-connected a broken
5693                          * pair up, causing the pathnames to be the
5694                          * same again. If so, that's not a rename at
5695                          * all, just a modification..
5696                          *
5697                          * Otherwise, see if this source was used for
5698                          * multiple renames, in which case we decrement
5699                          * the count, and call it a copy.
5700                          */
5701                         if (!strcmp(p->one->path, p->two->path))
5702                                 p->status = DIFF_STATUS_MODIFIED;
5703                         else if (--p->one->rename_used > 0)
5704                                 p->status = DIFF_STATUS_COPIED;
5705                         else
5706                                 p->status = DIFF_STATUS_RENAMED;
5707                 }
5708                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5709                          p->one->mode != p->two->mode ||
5710                          p->one->dirty_submodule ||
5711                          p->two->dirty_submodule ||
5712                          is_null_oid(&p->one->oid))
5713                         p->status = DIFF_STATUS_MODIFIED;
5714                 else {
5715                         /* This is a "no-change" entry and should not
5716                          * happen anymore, but prepare for broken callers.
5717                          */
5718                         error("feeding unmodified %s to diffcore",
5719                               p->one->path);
5720                         p->status = DIFF_STATUS_UNKNOWN;
5721                 }
5722         }
5723         diff_debug_queue("resolve-rename-copy done", q);
5724 }
5725
5726 static int check_pair_status(struct diff_filepair *p)
5727 {
5728         switch (p->status) {
5729         case DIFF_STATUS_UNKNOWN:
5730                 return 0;
5731         case 0:
5732                 die("internal error in diff-resolve-rename-copy");
5733         default:
5734                 return 1;
5735         }
5736 }
5737
5738 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5739 {
5740         int fmt = opt->output_format;
5741
5742         if (fmt & DIFF_FORMAT_CHECKDIFF)
5743                 diff_flush_checkdiff(p, opt);
5744         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5745                 diff_flush_raw(p, opt);
5746         else if (fmt & DIFF_FORMAT_NAME) {
5747                 const char *name_a, *name_b;
5748                 name_a = p->two->path;
5749                 name_b = NULL;
5750                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5751                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5752                 write_name_quoted(name_a, opt->file, opt->line_termination);
5753         }
5754 }
5755
5756 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5757 {
5758         struct strbuf sb = STRBUF_INIT;
5759         if (fs->mode)
5760                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5761         else
5762                 strbuf_addf(&sb, " %s ", newdelete);
5763
5764         quote_c_style(fs->path, &sb, NULL, 0);
5765         strbuf_addch(&sb, '\n');
5766         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5767                          sb.buf, sb.len, 0);
5768         strbuf_release(&sb);
5769 }
5770
5771 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5772                 int show_name)
5773 {
5774         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5775                 struct strbuf sb = STRBUF_INIT;
5776                 strbuf_addf(&sb, " mode change %06o => %06o",
5777                             p->one->mode, p->two->mode);
5778                 if (show_name) {
5779                         strbuf_addch(&sb, ' ');
5780                         quote_c_style(p->two->path, &sb, NULL, 0);
5781                 }
5782                 strbuf_addch(&sb, '\n');
5783                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5784                                  sb.buf, sb.len, 0);
5785                 strbuf_release(&sb);
5786         }
5787 }
5788
5789 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5790                 struct diff_filepair *p)
5791 {
5792         struct strbuf sb = STRBUF_INIT;
5793         struct strbuf names = STRBUF_INIT;
5794
5795         pprint_rename(&names, p->one->path, p->two->path);
5796         strbuf_addf(&sb, " %s %s (%d%%)\n",
5797                     renamecopy, names.buf, similarity_index(p));
5798         strbuf_release(&names);
5799         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5800                                  sb.buf, sb.len, 0);
5801         show_mode_change(opt, p, 0);
5802         strbuf_release(&sb);
5803 }
5804
5805 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5806 {
5807         switch(p->status) {
5808         case DIFF_STATUS_DELETED:
5809                 show_file_mode_name(opt, "delete", p->one);
5810                 break;
5811         case DIFF_STATUS_ADDED:
5812                 show_file_mode_name(opt, "create", p->two);
5813                 break;
5814         case DIFF_STATUS_COPIED:
5815                 show_rename_copy(opt, "copy", p);
5816                 break;
5817         case DIFF_STATUS_RENAMED:
5818                 show_rename_copy(opt, "rename", p);
5819                 break;
5820         default:
5821                 if (p->score) {
5822                         struct strbuf sb = STRBUF_INIT;
5823                         strbuf_addstr(&sb, " rewrite ");
5824                         quote_c_style(p->two->path, &sb, NULL, 0);
5825                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5826                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5827                                          sb.buf, sb.len, 0);
5828                         strbuf_release(&sb);
5829                 }
5830                 show_mode_change(opt, p, !p->score);
5831                 break;
5832         }
5833 }
5834
5835 struct patch_id_t {
5836         git_SHA_CTX *ctx;
5837         int patchlen;
5838 };
5839
5840 static int remove_space(char *line, int len)
5841 {
5842         int i;
5843         char *dst = line;
5844         unsigned char c;
5845
5846         for (i = 0; i < len; i++)
5847                 if (!isspace((c = line[i])))
5848                         *dst++ = c;
5849
5850         return dst - line;
5851 }
5852
5853 static void patch_id_consume(void *priv, char *line, unsigned long len)
5854 {
5855         struct patch_id_t *data = priv;
5856         int new_len;
5857
5858         new_len = remove_space(line, len);
5859
5860         git_SHA1_Update(data->ctx, line, new_len);
5861         data->patchlen += new_len;
5862 }
5863
5864 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5865 {
5866         git_SHA1_Update(ctx, str, strlen(str));
5867 }
5868
5869 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5870 {
5871         /* large enough for 2^32 in octal */
5872         char buf[12];
5873         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5874         git_SHA1_Update(ctx, buf, len);
5875 }
5876
5877 /* returns 0 upon success, and writes result into sha1 */
5878 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5879 {
5880         struct diff_queue_struct *q = &diff_queued_diff;
5881         int i;
5882         git_SHA_CTX ctx;
5883         struct patch_id_t data;
5884
5885         git_SHA1_Init(&ctx);
5886         memset(&data, 0, sizeof(struct patch_id_t));
5887         data.ctx = &ctx;
5888
5889         for (i = 0; i < q->nr; i++) {
5890                 xpparam_t xpp;
5891                 xdemitconf_t xecfg;
5892                 mmfile_t mf1, mf2;
5893                 struct diff_filepair *p = q->queue[i];
5894                 int len1, len2;
5895
5896                 memset(&xpp, 0, sizeof(xpp));
5897                 memset(&xecfg, 0, sizeof(xecfg));
5898                 if (p->status == 0)
5899                         return error("internal diff status error");
5900                 if (p->status == DIFF_STATUS_UNKNOWN)
5901                         continue;
5902                 if (diff_unmodified_pair(p))
5903                         continue;
5904                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5905                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5906                         continue;
5907                 if (DIFF_PAIR_UNMERGED(p))
5908                         continue;
5909
5910                 diff_fill_oid_info(p->one, options->repo->index);
5911                 diff_fill_oid_info(p->two, options->repo->index);
5912
5913                 len1 = remove_space(p->one->path, strlen(p->one->path));
5914                 len2 = remove_space(p->two->path, strlen(p->two->path));
5915                 patch_id_add_string(&ctx, "diff--git");
5916                 patch_id_add_string(&ctx, "a/");
5917                 git_SHA1_Update(&ctx, p->one->path, len1);
5918                 patch_id_add_string(&ctx, "b/");
5919                 git_SHA1_Update(&ctx, p->two->path, len2);
5920
5921                 if (p->one->mode == 0) {
5922                         patch_id_add_string(&ctx, "newfilemode");
5923                         patch_id_add_mode(&ctx, p->two->mode);
5924                         patch_id_add_string(&ctx, "---/dev/null");
5925                         patch_id_add_string(&ctx, "+++b/");
5926                         git_SHA1_Update(&ctx, p->two->path, len2);
5927                 } else if (p->two->mode == 0) {
5928                         patch_id_add_string(&ctx, "deletedfilemode");
5929                         patch_id_add_mode(&ctx, p->one->mode);
5930                         patch_id_add_string(&ctx, "---a/");
5931                         git_SHA1_Update(&ctx, p->one->path, len1);
5932                         patch_id_add_string(&ctx, "+++/dev/null");
5933                 } else {
5934                         patch_id_add_string(&ctx, "---a/");
5935                         git_SHA1_Update(&ctx, p->one->path, len1);
5936                         patch_id_add_string(&ctx, "+++b/");
5937                         git_SHA1_Update(&ctx, p->two->path, len2);
5938                 }
5939
5940                 if (diff_header_only)
5941                         continue;
5942
5943                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5944                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5945                         return error("unable to read files to diff");
5946
5947                 if (diff_filespec_is_binary(options->repo, p->one) ||
5948                     diff_filespec_is_binary(options->repo, p->two)) {
5949                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5950                                         GIT_SHA1_HEXSZ);
5951                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5952                                         GIT_SHA1_HEXSZ);
5953                         continue;
5954                 }
5955
5956                 xpp.flags = 0;
5957                 xecfg.ctxlen = 3;
5958                 xecfg.flags = 0;
5959                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5960                                   patch_id_consume, &data, &xpp, &xecfg))
5961                         return error("unable to generate patch-id diff for %s",
5962                                      p->one->path);
5963         }
5964
5965         git_SHA1_Final(oid->hash, &ctx);
5966         return 0;
5967 }
5968
5969 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5970 {
5971         struct diff_queue_struct *q = &diff_queued_diff;
5972         int i;
5973         int result = diff_get_patch_id(options, oid, diff_header_only);
5974
5975         for (i = 0; i < q->nr; i++)
5976                 diff_free_filepair(q->queue[i]);
5977
5978         free(q->queue);
5979         DIFF_QUEUE_CLEAR(q);
5980
5981         return result;
5982 }
5983
5984 static int is_summary_empty(const struct diff_queue_struct *q)
5985 {
5986         int i;
5987
5988         for (i = 0; i < q->nr; i++) {
5989                 const struct diff_filepair *p = q->queue[i];
5990
5991                 switch (p->status) {
5992                 case DIFF_STATUS_DELETED:
5993                 case DIFF_STATUS_ADDED:
5994                 case DIFF_STATUS_COPIED:
5995                 case DIFF_STATUS_RENAMED:
5996                         return 0;
5997                 default:
5998                         if (p->score)
5999                                 return 0;
6000                         if (p->one->mode && p->two->mode &&
6001                             p->one->mode != p->two->mode)
6002                                 return 0;
6003                         break;
6004                 }
6005         }
6006         return 1;
6007 }
6008
6009 static const char rename_limit_warning[] =
6010 N_("inexact rename detection was skipped due to too many files.");
6011
6012 static const char degrade_cc_to_c_warning[] =
6013 N_("only found copies from modified paths due to too many files.");
6014
6015 static const char rename_limit_advice[] =
6016 N_("you may want to set your %s variable to at least "
6017    "%d and retry the command.");
6018
6019 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
6020 {
6021         fflush(stdout);
6022         if (degraded_cc)
6023                 warning(_(degrade_cc_to_c_warning));
6024         else if (needed)
6025                 warning(_(rename_limit_warning));
6026         else
6027                 return;
6028         if (0 < needed)
6029                 warning(_(rename_limit_advice), varname, needed);
6030 }
6031
6032 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
6033 {
6034         int i;
6035         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
6036         struct diff_queue_struct *q = &diff_queued_diff;
6037
6038         if (WSEH_NEW & WS_RULE_MASK)
6039                 BUG("WS rules bit mask overlaps with diff symbol flags");
6040
6041         if (o->color_moved)
6042                 o->emitted_symbols = &esm;
6043
6044         for (i = 0; i < q->nr; i++) {
6045                 struct diff_filepair *p = q->queue[i];
6046                 if (check_pair_status(p))
6047                         diff_flush_patch(p, o);
6048         }
6049
6050         if (o->emitted_symbols) {
6051                 if (o->color_moved) {
6052                         struct hashmap add_lines, del_lines;
6053
6054                         if (o->color_moved_ws_handling &
6055                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
6056                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
6057
6058                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
6059                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
6060
6061                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6062                         mark_color_as_moved(o, &add_lines, &del_lines);
6063                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6064                                 dim_moved_lines(o);
6065
6066                         hashmap_free(&add_lines, 1);
6067                         hashmap_free(&del_lines, 1);
6068                 }
6069
6070                 for (i = 0; i < esm.nr; i++)
6071                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6072
6073                 for (i = 0; i < esm.nr; i++)
6074                         free((void *)esm.buf[i].line);
6075         }
6076         esm.nr = 0;
6077 }
6078
6079 void diff_flush(struct diff_options *options)
6080 {
6081         struct diff_queue_struct *q = &diff_queued_diff;
6082         int i, output_format = options->output_format;
6083         int separator = 0;
6084         int dirstat_by_line = 0;
6085
6086         /*
6087          * Order: raw, stat, summary, patch
6088          * or:    name/name-status/checkdiff (other bits clear)
6089          */
6090         if (!q->nr)
6091                 goto free_queue;
6092
6093         if (output_format & (DIFF_FORMAT_RAW |
6094                              DIFF_FORMAT_NAME |
6095                              DIFF_FORMAT_NAME_STATUS |
6096                              DIFF_FORMAT_CHECKDIFF)) {
6097                 for (i = 0; i < q->nr; i++) {
6098                         struct diff_filepair *p = q->queue[i];
6099                         if (check_pair_status(p))
6100                                 flush_one_pair(p, options);
6101                 }
6102                 separator++;
6103         }
6104
6105         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6106                 dirstat_by_line = 1;
6107
6108         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6109             dirstat_by_line) {
6110                 struct diffstat_t diffstat;
6111
6112                 memset(&diffstat, 0, sizeof(struct diffstat_t));
6113                 for (i = 0; i < q->nr; i++) {
6114                         struct diff_filepair *p = q->queue[i];
6115                         if (check_pair_status(p))
6116                                 diff_flush_stat(p, options, &diffstat);
6117                 }
6118                 if (output_format & DIFF_FORMAT_NUMSTAT)
6119                         show_numstat(&diffstat, options);
6120                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6121                         show_stats(&diffstat, options);
6122                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6123                         show_shortstats(&diffstat, options);
6124                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6125                         show_dirstat_by_line(&diffstat, options);
6126                 free_diffstat_info(&diffstat);
6127                 separator++;
6128         }
6129         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6130                 show_dirstat(options);
6131
6132         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6133                 for (i = 0; i < q->nr; i++) {
6134                         diff_summary(options, q->queue[i]);
6135                 }
6136                 separator++;
6137         }
6138
6139         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6140             options->flags.exit_with_status &&
6141             options->flags.diff_from_contents) {
6142                 /*
6143                  * run diff_flush_patch for the exit status. setting
6144                  * options->file to /dev/null should be safe, because we
6145                  * aren't supposed to produce any output anyway.
6146                  */
6147                 if (options->close_file)
6148                         fclose(options->file);
6149                 options->file = xfopen("/dev/null", "w");
6150                 options->close_file = 1;
6151                 options->color_moved = 0;
6152                 for (i = 0; i < q->nr; i++) {
6153                         struct diff_filepair *p = q->queue[i];
6154                         if (check_pair_status(p))
6155                                 diff_flush_patch(p, options);
6156                         if (options->found_changes)
6157                                 break;
6158                 }
6159         }
6160
6161         if (output_format & DIFF_FORMAT_PATCH) {
6162                 if (separator) {
6163                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6164                         if (options->stat_sep)
6165                                 /* attach patch instead of inline */
6166                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6167                                                  NULL, 0, 0);
6168                 }
6169
6170                 diff_flush_patch_all_file_pairs(options);
6171         }
6172
6173         if (output_format & DIFF_FORMAT_CALLBACK)
6174                 options->format_callback(q, options, options->format_callback_data);
6175
6176         for (i = 0; i < q->nr; i++)
6177                 diff_free_filepair(q->queue[i]);
6178 free_queue:
6179         free(q->queue);
6180         DIFF_QUEUE_CLEAR(q);
6181         if (options->close_file)
6182                 fclose(options->file);
6183
6184         /*
6185          * Report the content-level differences with HAS_CHANGES;
6186          * diff_addremove/diff_change does not set the bit when
6187          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6188          */
6189         if (options->flags.diff_from_contents) {
6190                 if (options->found_changes)
6191                         options->flags.has_changes = 1;
6192                 else
6193                         options->flags.has_changes = 0;
6194         }
6195 }
6196
6197 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6198 {
6199         return (((p->status == DIFF_STATUS_MODIFIED) &&
6200                  ((p->score &&
6201                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6202                   (!p->score &&
6203                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6204                 ((p->status != DIFF_STATUS_MODIFIED) &&
6205                  filter_bit_tst(p->status, options)));
6206 }
6207
6208 static void diffcore_apply_filter(struct diff_options *options)
6209 {
6210         int i;
6211         struct diff_queue_struct *q = &diff_queued_diff;
6212         struct diff_queue_struct outq;
6213
6214         DIFF_QUEUE_CLEAR(&outq);
6215
6216         if (!options->filter)
6217                 return;
6218
6219         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6220                 int found;
6221                 for (i = found = 0; !found && i < q->nr; i++) {
6222                         if (match_filter(options, q->queue[i]))
6223                                 found++;
6224                 }
6225                 if (found)
6226                         return;
6227
6228                 /* otherwise we will clear the whole queue
6229                  * by copying the empty outq at the end of this
6230                  * function, but first clear the current entries
6231                  * in the queue.
6232                  */
6233                 for (i = 0; i < q->nr; i++)
6234                         diff_free_filepair(q->queue[i]);
6235         }
6236         else {
6237                 /* Only the matching ones */
6238                 for (i = 0; i < q->nr; i++) {
6239                         struct diff_filepair *p = q->queue[i];
6240                         if (match_filter(options, p))
6241                                 diff_q(&outq, p);
6242                         else
6243                                 diff_free_filepair(p);
6244                 }
6245         }
6246         free(q->queue);
6247         *q = outq;
6248 }
6249
6250 /* Check whether two filespecs with the same mode and size are identical */
6251 static int diff_filespec_is_identical(struct repository *r,
6252                                       struct diff_filespec *one,
6253                                       struct diff_filespec *two)
6254 {
6255         if (S_ISGITLINK(one->mode))
6256                 return 0;
6257         if (diff_populate_filespec(r, one, 0))
6258                 return 0;
6259         if (diff_populate_filespec(r, two, 0))
6260                 return 0;
6261         return !memcmp(one->data, two->data, one->size);
6262 }
6263
6264 static int diff_filespec_check_stat_unmatch(struct repository *r,
6265                                             struct diff_filepair *p)
6266 {
6267         if (p->done_skip_stat_unmatch)
6268                 return p->skip_stat_unmatch_result;
6269
6270         p->done_skip_stat_unmatch = 1;
6271         p->skip_stat_unmatch_result = 0;
6272         /*
6273          * 1. Entries that come from stat info dirtiness
6274          *    always have both sides (iow, not create/delete),
6275          *    one side of the object name is unknown, with
6276          *    the same mode and size.  Keep the ones that
6277          *    do not match these criteria.  They have real
6278          *    differences.
6279          *
6280          * 2. At this point, the file is known to be modified,
6281          *    with the same mode and size, and the object
6282          *    name of one side is unknown.  Need to inspect
6283          *    the identical contents.
6284          */
6285         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6286             !DIFF_FILE_VALID(p->two) ||
6287             (p->one->oid_valid && p->two->oid_valid) ||
6288             (p->one->mode != p->two->mode) ||
6289             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6290             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6291             (p->one->size != p->two->size) ||
6292             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6293                 p->skip_stat_unmatch_result = 1;
6294         return p->skip_stat_unmatch_result;
6295 }
6296
6297 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6298 {
6299         int i;
6300         struct diff_queue_struct *q = &diff_queued_diff;
6301         struct diff_queue_struct outq;
6302         DIFF_QUEUE_CLEAR(&outq);
6303
6304         for (i = 0; i < q->nr; i++) {
6305                 struct diff_filepair *p = q->queue[i];
6306
6307                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6308                         diff_q(&outq, p);
6309                 else {
6310                         /*
6311                          * The caller can subtract 1 from skip_stat_unmatch
6312                          * to determine how many paths were dirty only
6313                          * due to stat info mismatch.
6314                          */
6315                         if (!diffopt->flags.no_index)
6316                                 diffopt->skip_stat_unmatch++;
6317                         diff_free_filepair(p);
6318                 }
6319         }
6320         free(q->queue);
6321         *q = outq;
6322 }
6323
6324 static int diffnamecmp(const void *a_, const void *b_)
6325 {
6326         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6327         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6328         const char *name_a, *name_b;
6329
6330         name_a = a->one ? a->one->path : a->two->path;
6331         name_b = b->one ? b->one->path : b->two->path;
6332         return strcmp(name_a, name_b);
6333 }
6334
6335 void diffcore_fix_diff_index(struct diff_options *options)
6336 {
6337         struct diff_queue_struct *q = &diff_queued_diff;
6338         QSORT(q->queue, q->nr, diffnamecmp);
6339 }
6340
6341 void diffcore_std(struct diff_options *options)
6342 {
6343         /* NOTE please keep the following in sync with diff_tree_combined() */
6344         if (options->skip_stat_unmatch)
6345                 diffcore_skip_stat_unmatch(options);
6346         if (!options->found_follow) {
6347                 /* See try_to_follow_renames() in tree-diff.c */
6348                 if (options->break_opt != -1)
6349                         diffcore_break(options->repo,
6350                                        options->break_opt);
6351                 if (options->detect_rename)
6352                         diffcore_rename(options);
6353                 if (options->break_opt != -1)
6354                         diffcore_merge_broken();
6355         }
6356         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6357                 diffcore_pickaxe(options);
6358         if (options->orderfile)
6359                 diffcore_order(options->orderfile);
6360         if (!options->found_follow)
6361                 /* See try_to_follow_renames() in tree-diff.c */
6362                 diff_resolve_rename_copy();
6363         diffcore_apply_filter(options);
6364
6365         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6366                 options->flags.has_changes = 1;
6367         else
6368                 options->flags.has_changes = 0;
6369
6370         options->found_follow = 0;
6371 }
6372
6373 int diff_result_code(struct diff_options *opt, int status)
6374 {
6375         int result = 0;
6376
6377         diff_warn_rename_limit("diff.renameLimit",
6378                                opt->needed_rename_limit,
6379                                opt->degraded_cc_to_c);
6380         if (!opt->flags.exit_with_status &&
6381             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6382                 return status;
6383         if (opt->flags.exit_with_status &&
6384             opt->flags.has_changes)
6385                 result |= 01;
6386         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6387             opt->flags.check_failed)
6388                 result |= 02;
6389         return result;
6390 }
6391
6392 int diff_can_quit_early(struct diff_options *opt)
6393 {
6394         return (opt->flags.quick &&
6395                 !opt->filter &&
6396                 opt->flags.has_changes);
6397 }
6398
6399 /*
6400  * Shall changes to this submodule be ignored?
6401  *
6402  * Submodule changes can be configured to be ignored separately for each path,
6403  * but that configuration can be overridden from the command line.
6404  */
6405 static int is_submodule_ignored(const char *path, struct diff_options *options)
6406 {
6407         int ignored = 0;
6408         struct diff_flags orig_flags = options->flags;
6409         if (!options->flags.override_submodule_config)
6410                 set_diffopt_flags_from_submodule_config(options, path);
6411         if (options->flags.ignore_submodules)
6412                 ignored = 1;
6413         options->flags = orig_flags;
6414         return ignored;
6415 }
6416
6417 void diff_addremove(struct diff_options *options,
6418                     int addremove, unsigned mode,
6419                     const struct object_id *oid,
6420                     int oid_valid,
6421                     const char *concatpath, unsigned dirty_submodule)
6422 {
6423         struct diff_filespec *one, *two;
6424
6425         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6426                 return;
6427
6428         /* This may look odd, but it is a preparation for
6429          * feeding "there are unchanged files which should
6430          * not produce diffs, but when you are doing copy
6431          * detection you would need them, so here they are"
6432          * entries to the diff-core.  They will be prefixed
6433          * with something like '=' or '*' (I haven't decided
6434          * which but should not make any difference).
6435          * Feeding the same new and old to diff_change()
6436          * also has the same effect.
6437          * Before the final output happens, they are pruned after
6438          * merged into rename/copy pairs as appropriate.
6439          */
6440         if (options->flags.reverse_diff)
6441                 addremove = (addremove == '+' ? '-' :
6442                              addremove == '-' ? '+' : addremove);
6443
6444         if (options->prefix &&
6445             strncmp(concatpath, options->prefix, options->prefix_length))
6446                 return;
6447
6448         one = alloc_filespec(concatpath);
6449         two = alloc_filespec(concatpath);
6450
6451         if (addremove != '+')
6452                 fill_filespec(one, oid, oid_valid, mode);
6453         if (addremove != '-') {
6454                 fill_filespec(two, oid, oid_valid, mode);
6455                 two->dirty_submodule = dirty_submodule;
6456         }
6457
6458         diff_queue(&diff_queued_diff, one, two);
6459         if (!options->flags.diff_from_contents)
6460                 options->flags.has_changes = 1;
6461 }
6462
6463 void diff_change(struct diff_options *options,
6464                  unsigned old_mode, unsigned new_mode,
6465                  const struct object_id *old_oid,
6466                  const struct object_id *new_oid,
6467                  int old_oid_valid, int new_oid_valid,
6468                  const char *concatpath,
6469                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6470 {
6471         struct diff_filespec *one, *two;
6472         struct diff_filepair *p;
6473
6474         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6475             is_submodule_ignored(concatpath, options))
6476                 return;
6477
6478         if (options->flags.reverse_diff) {
6479                 SWAP(old_mode, new_mode);
6480                 SWAP(old_oid, new_oid);
6481                 SWAP(old_oid_valid, new_oid_valid);
6482                 SWAP(old_dirty_submodule, new_dirty_submodule);
6483         }
6484
6485         if (options->prefix &&
6486             strncmp(concatpath, options->prefix, options->prefix_length))
6487                 return;
6488
6489         one = alloc_filespec(concatpath);
6490         two = alloc_filespec(concatpath);
6491         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6492         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6493         one->dirty_submodule = old_dirty_submodule;
6494         two->dirty_submodule = new_dirty_submodule;
6495         p = diff_queue(&diff_queued_diff, one, two);
6496
6497         if (options->flags.diff_from_contents)
6498                 return;
6499
6500         if (options->flags.quick && options->skip_stat_unmatch &&
6501             !diff_filespec_check_stat_unmatch(options->repo, p))
6502                 return;
6503
6504         options->flags.has_changes = 1;
6505 }
6506
6507 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6508 {
6509         struct diff_filepair *pair;
6510         struct diff_filespec *one, *two;
6511
6512         if (options->prefix &&
6513             strncmp(path, options->prefix, options->prefix_length))
6514                 return NULL;
6515
6516         one = alloc_filespec(path);
6517         two = alloc_filespec(path);
6518         pair = diff_queue(&diff_queued_diff, one, two);
6519         pair->is_unmerged = 1;
6520         return pair;
6521 }
6522
6523 static char *run_textconv(struct repository *r,
6524                           const char *pgm,
6525                           struct diff_filespec *spec,
6526                           size_t *outsize)
6527 {
6528         struct diff_tempfile *temp;
6529         const char *argv[3];
6530         const char **arg = argv;
6531         struct child_process child = CHILD_PROCESS_INIT;
6532         struct strbuf buf = STRBUF_INIT;
6533         int err = 0;
6534
6535         temp = prepare_temp_file(r, spec->path, spec);
6536         *arg++ = pgm;
6537         *arg++ = temp->name;
6538         *arg = NULL;
6539
6540         child.use_shell = 1;
6541         child.argv = argv;
6542         child.out = -1;
6543         if (start_command(&child)) {
6544                 remove_tempfile();
6545                 return NULL;
6546         }
6547
6548         if (strbuf_read(&buf, child.out, 0) < 0)
6549                 err = error("error reading from textconv command '%s'", pgm);
6550         close(child.out);
6551
6552         if (finish_command(&child) || err) {
6553                 strbuf_release(&buf);
6554                 remove_tempfile();
6555                 return NULL;
6556         }
6557         remove_tempfile();
6558
6559         return strbuf_detach(&buf, outsize);
6560 }
6561
6562 size_t fill_textconv(struct repository *r,
6563                      struct userdiff_driver *driver,
6564                      struct diff_filespec *df,
6565                      char **outbuf)
6566 {
6567         size_t size;
6568
6569         if (!driver) {
6570                 if (!DIFF_FILE_VALID(df)) {
6571                         *outbuf = "";
6572                         return 0;
6573                 }
6574                 if (diff_populate_filespec(r, df, 0))
6575                         die("unable to read files to diff");
6576                 *outbuf = df->data;
6577                 return df->size;
6578         }
6579
6580         if (!driver->textconv)
6581                 BUG("fill_textconv called with non-textconv driver");
6582
6583         if (driver->textconv_cache && df->oid_valid) {
6584                 *outbuf = notes_cache_get(driver->textconv_cache,
6585                                           &df->oid,
6586                                           &size);
6587                 if (*outbuf)
6588                         return size;
6589         }
6590
6591         *outbuf = run_textconv(r, driver->textconv, df, &size);
6592         if (!*outbuf)
6593                 die("unable to read files to diff");
6594
6595         if (driver->textconv_cache && df->oid_valid) {
6596                 /* ignore errors, as we might be in a readonly repository */
6597                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6598                                 size);
6599                 /*
6600                  * we could save up changes and flush them all at the end,
6601                  * but we would need an extra call after all diffing is done.
6602                  * Since generating a cache entry is the slow path anyway,
6603                  * this extra overhead probably isn't a big deal.
6604                  */
6605                 notes_cache_write(driver->textconv_cache);
6606         }
6607
6608         return size;
6609 }
6610
6611 int textconv_object(struct repository *r,
6612                     const char *path,
6613                     unsigned mode,
6614                     const struct object_id *oid,
6615                     int oid_valid,
6616                     char **buf,
6617                     unsigned long *buf_size)
6618 {
6619         struct diff_filespec *df;
6620         struct userdiff_driver *textconv;
6621
6622         df = alloc_filespec(path);
6623         fill_filespec(df, oid, oid_valid, mode);
6624         textconv = get_textconv(r, df);
6625         if (!textconv) {
6626                 free_filespec(df);
6627                 return 0;
6628         }
6629
6630         *buf_size = fill_textconv(r, textconv, df, buf);
6631         free_filespec(df);
6632         return 1;
6633 }
6634
6635 void setup_diff_pager(struct diff_options *opt)
6636 {
6637         /*
6638          * If the user asked for our exit code, then either they want --quiet
6639          * or --exit-code. We should definitely not bother with a pager in the
6640          * former case, as we will generate no output. Since we still properly
6641          * report our exit code even when a pager is run, we _could_ run a
6642          * pager with --exit-code. But since we have not done so historically,
6643          * and because it is easy to find people oneline advising "git diff
6644          * --exit-code" in hooks and other scripts, we do not do so.
6645          */
6646         if (!opt->flags.exit_with_status &&
6647             check_pager_config("diff") != 0)
6648                 setup_pager();
6649 }