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 #include "fetch-object.h"
  29
  30 #ifdef NO_FAST_WORKING_DIRECTORY
  31 #define FAST_WORKING_DIRECTORY 0
  32 #else
  33 #define FAST_WORKING_DIRECTORY 1
  34 #endif
  35
  36 static int diff_detect_rename_default;
  37 static int diff_indent_heuristic = 1;
  38 static int diff_rename_limit_default = 400;
  39 static int diff_suppress_blank_empty;
  40 static int diff_use_color_default = -1;
  41 static int diff_color_moved_default;
  42 static int diff_color_moved_ws_default;
  43 static int diff_context_default = 3;
  44 static int diff_interhunk_context_default;
  45 static const char *diff_word_regex_cfg;
  46 static const char *external_diff_cmd_cfg;
  47 static const char *diff_order_file_cfg;
  48 int diff_auto_refresh_index = 1;
  49 static int diff_mnemonic_prefix;
  50 static int diff_no_prefix;
  51 static int diff_stat_graph_width;
  52 static int diff_dirstat_permille_default = 30;
  53 static struct diff_options default_diff_options;
  54 static long diff_algorithm;
  55 static unsigned ws_error_highlight_default = WSEH_NEW;
  56
  57 static char diff_colors[][COLOR_MAXLEN] = {
  58         GIT_COLOR_RESET,
  59         GIT_COLOR_NORMAL,       /* CONTEXT */
  60         GIT_COLOR_BOLD,         /* METAINFO */
  61         GIT_COLOR_CYAN,         /* FRAGINFO */
  62         GIT_COLOR_RED,          /* OLD */
  63         GIT_COLOR_GREEN,        /* NEW */
  64         GIT_COLOR_YELLOW,       /* COMMIT */
  65         GIT_COLOR_BG_RED,       /* WHITESPACE */
  66         GIT_COLOR_NORMAL,       /* FUNCINFO */
  67         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  68         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  69         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  70         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  71         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  72         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  73         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  74         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  75         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  76         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  77         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  78         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  79         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  80         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  81 };
  82
  83 static const char *color_diff_slots[] = {
  84         [DIFF_CONTEXT]                = "context",
  85         [DIFF_METAINFO]               = "meta",
  86         [DIFF_FRAGINFO]               = "frag",
  87         [DIFF_FILE_OLD]               = "old",
  88         [DIFF_FILE_NEW]               = "new",
  89         [DIFF_COMMIT]                 = "commit",
  90         [DIFF_WHITESPACE]             = "whitespace",
  91         [DIFF_FUNCINFO]               = "func",
  92         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  93         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  94         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  95         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  96         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  97         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  98         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  99         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
 100         [DIFF_CONTEXT_DIM]            = "contextDimmed",
 101         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 102         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 103         [DIFF_CONTEXT_BOLD]           = "contextBold",
 104         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 105         [DIFF_FILE_NEW_BOLD]          = "newBold",
 106 };
 107
 108 define_list_config_array_extra(color_diff_slots, {"plain"});
 109
 110 static int parse_diff_color_slot(const char *var)
 111 {
 112         if (!strcasecmp(var, "plain"))
 113                 return DIFF_CONTEXT;
 114         return LOOKUP_CONFIG(color_diff_slots, var);
 115 }
 116
 117 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 118                                 struct strbuf *errmsg)
 119 {
 120         char *params_copy = xstrdup(params_string);
 121         struct string_list params = STRING_LIST_INIT_NODUP;
 122         int ret = 0;
 123         int i;
 124
 125         if (*params_copy)
 126                 string_list_split_in_place(&params, params_copy, ',', -1);
 127         for (i = 0; i < params.nr; i++) {
 128                 const char *p = params.items[i].string;
 129                 if (!strcmp(p, "changes")) {
 130                         options->flags.dirstat_by_line = 0;
 131                         options->flags.dirstat_by_file = 0;
 132                 } else if (!strcmp(p, "lines")) {
 133                         options->flags.dirstat_by_line = 1;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "files")) {
 136                         options->flags.dirstat_by_line = 0;
 137                         options->flags.dirstat_by_file = 1;
 138                 } else if (!strcmp(p, "noncumulative")) {
 139                         options->flags.dirstat_cumulative = 0;
 140                 } else if (!strcmp(p, "cumulative")) {
 141                         options->flags.dirstat_cumulative = 1;
 142                 } else if (isdigit(*p)) {
 143                         char *end;
 144                         int permille = strtoul(p, &end, 10) * 10;
 145                         if (*end == '.' && isdigit(*++end)) {
 146                                 /* only use first digit */
 147                                 permille += *end - '0';
 148                                 /* .. and ignore any further digits */
 149                                 while (isdigit(*++end))
 150                                         ; /* nothing */
 151                         }
 152                         if (!*end)
 153                                 options->dirstat_permille = permille;
 154                         else {
 155                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 156                                             p);
 157                                 ret++;
 158                         }
 159                 } else {
 160                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 161                         ret++;
 162                 }
 163
 164         }
 165         string_list_clear(&params, 0);
 166         free(params_copy);
 167         return ret;
 168 }
 169
 170 static int parse_submodule_params(struct diff_options *options, const char *value)
 171 {
 172         if (!strcmp(value, "log"))
 173                 options->submodule_format = DIFF_SUBMODULE_LOG;
 174         else if (!strcmp(value, "short"))
 175                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 176         else if (!strcmp(value, "diff"))
 177                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 178         /*
 179          * Please update $__git_diff_submodule_formats in
 180          * git-completion.bash when you add new formats.
 181          */
 182         else
 183                 return -1;
 184         return 0;
 185 }
 186
 187 int git_config_rename(const char *var, const char *value)
 188 {
 189         if (!value)
 190                 return DIFF_DETECT_RENAME;
 191         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 192                 return  DIFF_DETECT_COPY;
 193         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 194 }
 195
 196 long parse_algorithm_value(const char *value)
 197 {
 198         if (!value)
 199                 return -1;
 200         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 201                 return 0;
 202         else if (!strcasecmp(value, "minimal"))
 203                 return XDF_NEED_MINIMAL;
 204         else if (!strcasecmp(value, "patience"))
 205                 return XDF_PATIENCE_DIFF;
 206         else if (!strcasecmp(value, "histogram"))
 207                 return XDF_HISTOGRAM_DIFF;
 208         /*
 209          * Please update $__git_diff_algorithms in git-completion.bash
 210          * when you add new algorithms.
 211          */
 212         return -1;
 213 }
 214
 215 static int parse_one_token(const char **arg, const char *token)
 216 {
 217         const char *rest;
 218         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 219                 *arg = rest;
 220                 return 1;
 221         }
 222         return 0;
 223 }
 224
 225 static int parse_ws_error_highlight(const char *arg)
 226 {
 227         const char *orig_arg = arg;
 228         unsigned val = 0;
 229
 230         while (*arg) {
 231                 if (parse_one_token(&arg, "none"))
 232                         val = 0;
 233                 else if (parse_one_token(&arg, "default"))
 234                         val = WSEH_NEW;
 235                 else if (parse_one_token(&arg, "all"))
 236                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 237                 else if (parse_one_token(&arg, "new"))
 238                         val |= WSEH_NEW;
 239                 else if (parse_one_token(&arg, "old"))
 240                         val |= WSEH_OLD;
 241                 else if (parse_one_token(&arg, "context"))
 242                         val |= WSEH_CONTEXT;
 243                 else {
 244                         return -1 - (int)(arg - orig_arg);
 245                 }
 246                 if (*arg)
 247                         arg++;
 248         }
 249         return val;
 250 }
 251
 252 /*
 253  * These are to give UI layer defaults.
 254  * The core-level commands such as git-diff-files should
 255  * never be affected by the setting of diff.renames
 256  * the user happens to have in the configuration file.
 257  */
 258 void init_diff_ui_defaults(void)
 259 {
 260         diff_detect_rename_default = DIFF_DETECT_RENAME;
 261 }
 262
 263 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 264 {
 265         if (!strcmp(var, "diff.indentheuristic"))
 266                 diff_indent_heuristic = git_config_bool(var, value);
 267         return 0;
 268 }
 269
 270 static int parse_color_moved(const char *arg)
 271 {
 272         switch (git_parse_maybe_bool(arg)) {
 273         case 0:
 274                 return COLOR_MOVED_NO;
 275         case 1:
 276                 return COLOR_MOVED_DEFAULT;
 277         default:
 278                 break;
 279         }
 280
 281         if (!strcmp(arg, "no"))
 282                 return COLOR_MOVED_NO;
 283         else if (!strcmp(arg, "plain"))
 284                 return COLOR_MOVED_PLAIN;
 285         else if (!strcmp(arg, "blocks"))
 286                 return COLOR_MOVED_BLOCKS;
 287         else if (!strcmp(arg, "zebra"))
 288                 return COLOR_MOVED_ZEBRA;
 289         else if (!strcmp(arg, "default"))
 290                 return COLOR_MOVED_DEFAULT;
 291         else if (!strcmp(arg, "dimmed-zebra"))
 292                 return COLOR_MOVED_ZEBRA_DIM;
 293         else if (!strcmp(arg, "dimmed_zebra"))
 294                 return COLOR_MOVED_ZEBRA_DIM;
 295         else
 296                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 297 }
 298
 299 static unsigned parse_color_moved_ws(const char *arg)
 300 {
 301         int ret = 0;
 302         struct string_list l = STRING_LIST_INIT_DUP;
 303         struct string_list_item *i;
 304
 305         string_list_split(&l, arg, ',', -1);
 306
 307         for_each_string_list_item(i, &l) {
 308                 struct strbuf sb = STRBUF_INIT;
 309                 strbuf_addstr(&sb, i->string);
 310                 strbuf_trim(&sb);
 311
 312                 if (!strcmp(sb.buf, "no"))
 313                         ret = 0;
 314                 else if (!strcmp(sb.buf, "ignore-space-change"))
 315                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 316                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 317                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 318                 else if (!strcmp(sb.buf, "ignore-all-space"))
 319                         ret |= XDF_IGNORE_WHITESPACE;
 320                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 321                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 322                 else {
 323                         ret |= COLOR_MOVED_WS_ERROR;
 324                         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);
 325                 }
 326
 327                 strbuf_release(&sb);
 328         }
 329
 330         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 331             (ret & XDF_WHITESPACE_FLAGS)) {
 332                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 333                 ret |= COLOR_MOVED_WS_ERROR;
 334         }
 335
 336         string_list_clear(&l, 0);
 337
 338         return ret;
 339 }
 340
 341 int git_diff_ui_config(const char *var, const char *value, void *cb)
 342 {
 343         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 344                 diff_use_color_default = git_config_colorbool(var, value);
 345                 return 0;
 346         }
 347         if (!strcmp(var, "diff.colormoved")) {
 348                 int cm = parse_color_moved(value);
 349                 if (cm < 0)
 350                         return -1;
 351                 diff_color_moved_default = cm;
 352                 return 0;
 353         }
 354         if (!strcmp(var, "diff.colormovedws")) {
 355                 unsigned cm = parse_color_moved_ws(value);
 356                 if (cm & COLOR_MOVED_WS_ERROR)
 357                         return -1;
 358                 diff_color_moved_ws_default = cm;
 359                 return 0;
 360         }
 361         if (!strcmp(var, "diff.context")) {
 362                 diff_context_default = git_config_int(var, value);
 363                 if (diff_context_default < 0)
 364                         return -1;
 365                 return 0;
 366         }
 367         if (!strcmp(var, "diff.interhunkcontext")) {
 368                 diff_interhunk_context_default = git_config_int(var, value);
 369                 if (diff_interhunk_context_default < 0)
 370                         return -1;
 371                 return 0;
 372         }
 373         if (!strcmp(var, "diff.renames")) {
 374                 diff_detect_rename_default = git_config_rename(var, value);
 375                 return 0;
 376         }
 377         if (!strcmp(var, "diff.autorefreshindex")) {
 378                 diff_auto_refresh_index = git_config_bool(var, value);
 379                 return 0;
 380         }
 381         if (!strcmp(var, "diff.mnemonicprefix")) {
 382                 diff_mnemonic_prefix = git_config_bool(var, value);
 383                 return 0;
 384         }
 385         if (!strcmp(var, "diff.noprefix")) {
 386                 diff_no_prefix = git_config_bool(var, value);
 387                 return 0;
 388         }
 389         if (!strcmp(var, "diff.statgraphwidth")) {
 390                 diff_stat_graph_width = git_config_int(var, value);
 391                 return 0;
 392         }
 393         if (!strcmp(var, "diff.external"))
 394                 return git_config_string(&external_diff_cmd_cfg, var, value);
 395         if (!strcmp(var, "diff.wordregex"))
 396                 return git_config_string(&diff_word_regex_cfg, var, value);
 397         if (!strcmp(var, "diff.orderfile"))
 398                 return git_config_pathname(&diff_order_file_cfg, var, value);
 399
 400         if (!strcmp(var, "diff.ignoresubmodules"))
 401                 handle_ignore_submodules_arg(&default_diff_options, value);
 402
 403         if (!strcmp(var, "diff.submodule")) {
 404                 if (parse_submodule_params(&default_diff_options, value))
 405                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 406                                 value);
 407                 return 0;
 408         }
 409
 410         if (!strcmp(var, "diff.algorithm")) {
 411                 diff_algorithm = parse_algorithm_value(value);
 412                 if (diff_algorithm < 0)
 413                         return -1;
 414                 return 0;
 415         }
 416
 417         if (!strcmp(var, "diff.wserrorhighlight")) {
 418                 int val = parse_ws_error_highlight(value);
 419                 if (val < 0)
 420                         return -1;
 421                 ws_error_highlight_default = val;
 422                 return 0;
 423         }
 424
 425         if (git_color_config(var, value, cb) < 0)
 426                 return -1;
 427
 428         return git_diff_basic_config(var, value, cb);
 429 }
 430
 431 int git_diff_basic_config(const char *var, const char *value, void *cb)
 432 {
 433         const char *name;
 434
 435         if (!strcmp(var, "diff.renamelimit")) {
 436                 diff_rename_limit_default = git_config_int(var, value);
 437                 return 0;
 438         }
 439
 440         if (userdiff_config(var, value) < 0)
 441                 return -1;
 442
 443         if (skip_prefix(var, "diff.color.", &name) ||
 444             skip_prefix(var, "color.diff.", &name)) {
 445                 int slot = parse_diff_color_slot(name);
 446                 if (slot < 0)
 447                         return 0;
 448                 if (!value)
 449                         return config_error_nonbool(var);
 450                 return color_parse(value, diff_colors[slot]);
 451         }
 452
 453         /* like GNU diff's --suppress-blank-empty option  */
 454         if (!strcmp(var, "diff.suppressblankempty") ||
 455                         /* for backwards compatibility */
 456                         !strcmp(var, "diff.suppress-blank-empty")) {
 457                 diff_suppress_blank_empty = git_config_bool(var, value);
 458                 return 0;
 459         }
 460
 461         if (!strcmp(var, "diff.dirstat")) {
 462                 struct strbuf errmsg = STRBUF_INIT;
 463                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 464                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 465                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 466                                 errmsg.buf);
 467                 strbuf_release(&errmsg);
 468                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 469                 return 0;
 470         }
 471
 472         if (git_diff_heuristic_config(var, value, cb) < 0)
 473                 return -1;
 474
 475         return git_default_config(var, value, cb);
 476 }
 477
 478 static char *quote_two(const char *one, const char *two)
 479 {
 480         int need_one = quote_c_style(one, NULL, NULL, 1);
 481         int need_two = quote_c_style(two, NULL, NULL, 1);
 482         struct strbuf res = STRBUF_INIT;
 483
 484         if (need_one + need_two) {
 485                 strbuf_addch(&res, '"');
 486                 quote_c_style(one, &res, NULL, 1);
 487                 quote_c_style(two, &res, NULL, 1);
 488                 strbuf_addch(&res, '"');
 489         } else {
 490                 strbuf_addstr(&res, one);
 491                 strbuf_addstr(&res, two);
 492         }
 493         return strbuf_detach(&res, NULL);
 494 }
 495
 496 static const char *external_diff(void)
 497 {
 498         static const char *external_diff_cmd = NULL;
 499         static int done_preparing = 0;
 500
 501         if (done_preparing)
 502                 return external_diff_cmd;
 503         external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
 504         if (!external_diff_cmd)
 505                 external_diff_cmd = external_diff_cmd_cfg;
 506         done_preparing = 1;
 507         return external_diff_cmd;
 508 }
 509
 510 /*
 511  * Keep track of files used for diffing. Sometimes such an entry
 512  * refers to a temporary file, sometimes to an existing file, and
 513  * sometimes to "/dev/null".
 514  */
 515 static struct diff_tempfile {
 516         /*
 517          * filename external diff should read from, or NULL if this
 518          * entry is currently not in use:
 519          */
 520         const char *name;
 521
 522         char hex[GIT_MAX_HEXSZ + 1];
 523         char mode[10];
 524
 525         /*
 526          * If this diff_tempfile instance refers to a temporary file,
 527          * this tempfile object is used to manage its lifetime.
 528          */
 529         struct tempfile *tempfile;
 530 } diff_temp[2];
 531
 532 struct emit_callback {
 533         int color_diff;
 534         unsigned ws_rule;
 535         int blank_at_eof_in_preimage;
 536         int blank_at_eof_in_postimage;
 537         int lno_in_preimage;
 538         int lno_in_postimage;
 539         const char **label_path;
 540         struct diff_words_data *diff_words;
 541         struct diff_options *opt;
 542         struct strbuf *header;
 543 };
 544
 545 static int count_lines(const char *data, int size)
 546 {
 547         int count, ch, completely_empty = 1, nl_just_seen = 0;
 548         count = 0;
 549         while (0 < size--) {
 550                 ch = *data++;
 551                 if (ch == '\n') {
 552                         count++;
 553                         nl_just_seen = 1;
 554                         completely_empty = 0;
 555                 }
 556                 else {
 557                         nl_just_seen = 0;
 558                         completely_empty = 0;
 559                 }
 560         }
 561         if (completely_empty)
 562                 return 0;
 563         if (!nl_just_seen)
 564                 count++; /* no trailing newline */
 565         return count;
 566 }
 567
 568 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 569                        struct diff_filespec *one)
 570 {
 571         if (!DIFF_FILE_VALID(one)) {
 572                 mf->ptr = (char *)""; /* does not matter */
 573                 mf->size = 0;
 574                 return 0;
 575         }
 576         else if (diff_populate_filespec(r, one, 0))
 577                 return -1;
 578
 579         mf->ptr = one->data;
 580         mf->size = one->size;
 581         return 0;
 582 }
 583
 584 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 585 static unsigned long diff_filespec_size(struct repository *r,
 586                                         struct diff_filespec *one)
 587 {
 588         if (!DIFF_FILE_VALID(one))
 589                 return 0;
 590         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 591         return one->size;
 592 }
 593
 594 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 595 {
 596         char *ptr = mf->ptr;
 597         long size = mf->size;
 598         int cnt = 0;
 599
 600         if (!size)
 601                 return cnt;
 602         ptr += size - 1; /* pointing at the very end */
 603         if (*ptr != '\n')
 604                 ; /* incomplete line */
 605         else
 606                 ptr--; /* skip the last LF */
 607         while (mf->ptr < ptr) {
 608                 char *prev_eol;
 609                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 610                         if (*prev_eol == '\n')
 611                                 break;
 612                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 613                         break;
 614                 cnt++;
 615                 ptr = prev_eol - 1;
 616         }
 617         return cnt;
 618 }
 619
 620 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 621                                struct emit_callback *ecbdata)
 622 {
 623         int l1, l2, at;
 624         unsigned ws_rule = ecbdata->ws_rule;
 625         l1 = count_trailing_blank(mf1, ws_rule);
 626         l2 = count_trailing_blank(mf2, ws_rule);
 627         if (l2 <= l1) {
 628                 ecbdata->blank_at_eof_in_preimage = 0;
 629                 ecbdata->blank_at_eof_in_postimage = 0;
 630                 return;
 631         }
 632         at = count_lines(mf1->ptr, mf1->size);
 633         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 634
 635         at = count_lines(mf2->ptr, mf2->size);
 636         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 637 }
 638
 639 static void emit_line_0(struct diff_options *o,
 640                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 641                         int first, const char *line, int len)
 642 {
 643         int has_trailing_newline, has_trailing_carriage_return;
 644         int needs_reset = 0; /* at the end of the line */
 645         FILE *file = o->file;
 646
 647         fputs(diff_line_prefix(o), file);
 648
 649         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 650         if (has_trailing_newline)
 651                 len--;
 652
 653         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 654         if (has_trailing_carriage_return)
 655                 len--;
 656
 657         if (!len && !first)
 658                 goto end_of_line;
 659
 660         if (reverse && want_color(o->use_color)) {
 661                 fputs(GIT_COLOR_REVERSE, file);
 662                 needs_reset = 1;
 663         }
 664
 665         if (set_sign) {
 666                 fputs(set_sign, file);
 667                 needs_reset = 1;
 668         }
 669
 670         if (first)
 671                 fputc(first, file);
 672
 673         if (!len)
 674                 goto end_of_line;
 675
 676         if (set) {
 677                 if (set_sign && set != set_sign)
 678                         fputs(reset, file);
 679                 fputs(set, file);
 680                 needs_reset = 1;
 681         }
 682         fwrite(line, len, 1, file);
 683         needs_reset = 1; /* 'line' may contain color codes. */
 684
 685 end_of_line:
 686         if (needs_reset)
 687                 fputs(reset, file);
 688         if (has_trailing_carriage_return)
 689                 fputc('\r', file);
 690         if (has_trailing_newline)
 691                 fputc('\n', file);
 692 }
 693
 694 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 695                       const char *line, int len)
 696 {
 697         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 698 }
 699
 700 enum diff_symbol {
 701         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 702         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 703         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 704         DIFF_SYMBOL_BINARY_DIFF_BODY,
 705         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 706         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 707         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 708         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 709         DIFF_SYMBOL_STATS_LINE,
 710         DIFF_SYMBOL_WORD_DIFF,
 711         DIFF_SYMBOL_STAT_SEP,
 712         DIFF_SYMBOL_SUMMARY,
 713         DIFF_SYMBOL_SUBMODULE_ADD,
 714         DIFF_SYMBOL_SUBMODULE_DEL,
 715         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 716         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 717         DIFF_SYMBOL_SUBMODULE_HEADER,
 718         DIFF_SYMBOL_SUBMODULE_ERROR,
 719         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 720         DIFF_SYMBOL_REWRITE_DIFF,
 721         DIFF_SYMBOL_BINARY_FILES,
 722         DIFF_SYMBOL_HEADER,
 723         DIFF_SYMBOL_FILEPAIR_PLUS,
 724         DIFF_SYMBOL_FILEPAIR_MINUS,
 725         DIFF_SYMBOL_WORDS_PORCELAIN,
 726         DIFF_SYMBOL_WORDS,
 727         DIFF_SYMBOL_CONTEXT,
 728         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 729         DIFF_SYMBOL_PLUS,
 730         DIFF_SYMBOL_MINUS,
 731         DIFF_SYMBOL_NO_LF_EOF,
 732         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 733         DIFF_SYMBOL_CONTEXT_MARKER,
 734         DIFF_SYMBOL_SEPARATOR
 735 };
 736 /*
 737  * Flags for content lines:
 738  * 0..12 are whitespace rules
 739  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 740  * 16 is marking if the line is blank at EOF
 741  */
 742 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 743 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 744 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 745 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 746 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 747
 748 /*
 749  * This struct is used when we need to buffer the output of the diff output.
 750  *
 751  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 752  * into the pre/post image file. This pointer could be a union with the
 753  * line pointer. By storing an offset into the file instead of the literal line,
 754  * we can decrease the memory footprint for the buffered output. At first we
 755  * may want to only have indirection for the content lines, but we could also
 756  * enhance the state for emitting prefabricated lines, e.g. the similarity
 757  * score line or hunk/file headers would only need to store a number or path
 758  * and then the output can be constructed later on depending on state.
 759  */
 760 struct emitted_diff_symbol {
 761         const char *line;
 762         int len;
 763         int flags;
 764         int indent_off;   /* Offset to first non-whitespace character */
 765         int indent_width; /* The visual width of the indentation */
 766         enum diff_symbol s;
 767 };
 768 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 769
 770 struct emitted_diff_symbols {
 771         struct emitted_diff_symbol *buf;
 772         int nr, alloc;
 773 };
 774 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 775
 776 static void append_emitted_diff_symbol(struct diff_options *o,
 777                                        struct emitted_diff_symbol *e)
 778 {
 779         struct emitted_diff_symbol *f;
 780
 781         ALLOC_GROW(o->emitted_symbols->buf,
 782                    o->emitted_symbols->nr + 1,
 783                    o->emitted_symbols->alloc);
 784         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 785
 786         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 787         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 788 }
 789
 790 struct moved_entry {
 791         struct hashmap_entry ent;
 792         const struct emitted_diff_symbol *es;
 793         struct moved_entry *next_line;
 794 };
 795
 796 struct moved_block {
 797         struct moved_entry *match;
 798         int wsd; /* The whitespace delta of this block */
 799 };
 800
 801 static void moved_block_clear(struct moved_block *b)
 802 {
 803         memset(b, 0, sizeof(*b));
 804 }
 805
 806 #define INDENT_BLANKLINE INT_MIN
 807
 808 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 809 {
 810         unsigned int off = 0, i;
 811         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 812         const char *s = es->line;
 813         const int len = es->len;
 814
 815         /* skip any \v \f \r at start of indentation */
 816         while (s[off] == '\f' || s[off] == '\v' ||
 817                (s[off] == '\r' && off < len - 1))
 818                 off++;
 819
 820         /* calculate the visual width of indentation */
 821         while(1) {
 822                 if (s[off] == ' ') {
 823                         width++;
 824                         off++;
 825                 } else if (s[off] == '\t') {
 826                         width += tab_width - (width % tab_width);
 827                         while (s[++off] == '\t')
 828                                 width += tab_width;
 829                 } else {
 830                         break;
 831                 }
 832         }
 833
 834         /* check if this line is blank */
 835         for (i = off; i < len; i++)
 836                 if (!isspace(s[i]))
 837                     break;
 838
 839         if (i == len) {
 840                 es->indent_width = INDENT_BLANKLINE;
 841                 es->indent_off = len;
 842         } else {
 843                 es->indent_off = off;
 844                 es->indent_width = width;
 845         }
 846 }
 847
 848 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 849                             const struct emitted_diff_symbol *b,
 850                             int *out)
 851 {
 852         int a_len = a->len,
 853             b_len = b->len,
 854             a_off = a->indent_off,
 855             a_width = a->indent_width,
 856             b_off = b->indent_off,
 857             b_width = b->indent_width;
 858         int delta;
 859
 860         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
 861                 *out = INDENT_BLANKLINE;
 862                 return 1;
 863         }
 864
 865         if (a->s == DIFF_SYMBOL_PLUS)
 866                 delta = a_width - b_width;
 867         else
 868                 delta = b_width - a_width;
 869
 870         if (a_len - a_off != b_len - b_off ||
 871             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 872                 return 0;
 873
 874         *out = delta;
 875
 876         return 1;
 877 }
 878
 879 static int cmp_in_block_with_wsd(const struct diff_options *o,
 880                                  const struct moved_entry *cur,
 881                                  const struct moved_entry *match,
 882                                  struct moved_block *pmb,
 883                                  int n)
 884 {
 885         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 886         int al = cur->es->len, bl = match->es->len, cl = l->len;
 887         const char *a = cur->es->line,
 888                    *b = match->es->line,
 889                    *c = l->line;
 890         int a_off = cur->es->indent_off,
 891             a_width = cur->es->indent_width,
 892             c_off = l->indent_off,
 893             c_width = l->indent_width;
 894         int delta;
 895
 896         /*
 897          * We need to check if 'cur' is equal to 'match'.  As those
 898          * are from the same (+/-) side, we do not need to adjust for
 899          * indent changes. However these were found using fuzzy
 900          * matching so we do have to check if they are equal. Here we
 901          * just check the lengths. We delay calling memcmp() to check
 902          * the contents until later as if the length comparison for a
 903          * and c fails we can avoid the call all together.
 904          */
 905         if (al != bl)
 906                 return 1;
 907
 908         /* If 'l' and 'cur' are both blank then they match. */
 909         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
 910                 return 0;
 911
 912         /*
 913          * The indent changes of the block are known and stored in pmb->wsd;
 914          * however we need to check if the indent changes of the current line
 915          * match those of the current block and that the text of 'l' and 'cur'
 916          * after the indentation match.
 917          */
 918         if (cur->es->s == DIFF_SYMBOL_PLUS)
 919                 delta = a_width - c_width;
 920         else
 921                 delta = c_width - a_width;
 922
 923         /*
 924          * If the previous lines of this block were all blank then set its
 925          * whitespace delta.
 926          */
 927         if (pmb->wsd == INDENT_BLANKLINE)
 928                 pmb->wsd = delta;
 929
 930         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 931                  !memcmp(a, b, al) && !
 932                  memcmp(a + a_off, c + c_off, al - a_off));
 933 }
 934
 935 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 936                            const void *entry,
 937                            const void *entry_or_key,
 938                            const void *keydata)
 939 {
 940         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 941         const struct moved_entry *a = entry;
 942         const struct moved_entry *b = entry_or_key;
 943         unsigned flags = diffopt->color_moved_ws_handling
 944                          & XDF_WHITESPACE_FLAGS;
 945
 946         if (diffopt->color_moved_ws_handling &
 947             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 948                 /*
 949                  * As there is not specific white space config given,
 950                  * we'd need to check for a new block, so ignore all
 951                  * white space. The setup of the white space
 952                  * configuration for the next block is done else where
 953                  */
 954                 flags |= XDF_IGNORE_WHITESPACE;
 955
 956         return !xdiff_compare_lines(a->es->line, a->es->len,
 957                                     b->es->line, b->es->len,
 958                                     flags);
 959 }
 960
 961 static struct moved_entry *prepare_entry(struct diff_options *o,
 962                                          int line_no)
 963 {
 964         struct moved_entry *ret = xmalloc(sizeof(*ret));
 965         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 966         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 967         unsigned int hash = xdiff_hash_string(l->line, l->len, flags);
 968
 969         hashmap_entry_init(&ret->ent, hash);
 970         ret->es = l;
 971         ret->next_line = NULL;
 972
 973         return ret;
 974 }
 975
 976 static void add_lines_to_move_detection(struct diff_options *o,
 977                                         struct hashmap *add_lines,
 978                                         struct hashmap *del_lines)
 979 {
 980         struct moved_entry *prev_line = NULL;
 981
 982         int n;
 983         for (n = 0; n < o->emitted_symbols->nr; n++) {
 984                 struct hashmap *hm;
 985                 struct moved_entry *key;
 986
 987                 switch (o->emitted_symbols->buf[n].s) {
 988                 case DIFF_SYMBOL_PLUS:
 989                         hm = add_lines;
 990                         break;
 991                 case DIFF_SYMBOL_MINUS:
 992                         hm = del_lines;
 993                         break;
 994                 default:
 995                         prev_line = NULL;
 996                         continue;
 997                 }
 998
 999                 if (o->color_moved_ws_handling &
1000                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1001                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
1002                 key = prepare_entry(o, n);
1003                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1004                         prev_line->next_line = key;
1005
1006                 hashmap_add(hm, &key->ent);
1007                 prev_line = key;
1008         }
1009 }
1010
1011 static void pmb_advance_or_null(struct diff_options *o,
1012                                 struct moved_entry *match,
1013                                 struct hashmap *hm,
1014                                 struct moved_block *pmb,
1015                                 int pmb_nr)
1016 {
1017         int i;
1018         for (i = 0; i < pmb_nr; i++) {
1019                 struct moved_entry *prev = pmb[i].match;
1020                 struct moved_entry *cur = (prev && prev->next_line) ?
1021                                 prev->next_line : NULL;
1022                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
1023                         pmb[i].match = cur;
1024                 } else {
1025                         pmb[i].match = NULL;
1026                 }
1027         }
1028 }
1029
1030 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1031                                             struct moved_entry *match,
1032                                             struct hashmap *hm,
1033                                             struct moved_block *pmb,
1034                                             int pmb_nr, int n)
1035 {
1036         int i;
1037         char *got_match = xcalloc(1, pmb_nr);
1038
1039         hashmap_for_each_entry_from(hm, match, struct moved_entry, ent) {
1040                 for (i = 0; i < pmb_nr; i++) {
1041                         struct moved_entry *prev = pmb[i].match;
1042                         struct moved_entry *cur = (prev && prev->next_line) ?
1043                                         prev->next_line : NULL;
1044                         if (!cur)
1045                                 continue;
1046                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1047                                 got_match[i] |= 1;
1048                 }
1049         }
1050
1051         for (i = 0; i < pmb_nr; i++) {
1052                 if (got_match[i]) {
1053                         /* Advance to the next line */
1054                         pmb[i].match = pmb[i].match->next_line;
1055                 } else {
1056                         moved_block_clear(&pmb[i]);
1057                 }
1058         }
1059
1060         free(got_match);
1061 }
1062
1063 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1064                                          int pmb_nr)
1065 {
1066         int lp, rp;
1067
1068         /* Shrink the set of potential block to the remaining running */
1069         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1070                 while (lp < pmb_nr && pmb[lp].match)
1071                         lp++;
1072                 /* lp points at the first NULL now */
1073
1074                 while (rp > -1 && !pmb[rp].match)
1075                         rp--;
1076                 /* rp points at the last non-NULL */
1077
1078                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1079                         pmb[lp] = pmb[rp];
1080                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1081                         rp--;
1082                         lp++;
1083                 }
1084         }
1085
1086         /* Remember the number of running sets */
1087         return rp + 1;
1088 }
1089
1090 /*
1091  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1092  *
1093  * Otherwise, if the last block has fewer alphanumeric characters than
1094  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1095  * that block.
1096  *
1097  * The last block consists of the (n - block_length)'th line up to but not
1098  * including the nth line.
1099  *
1100  * Returns 0 if the last block is empty or is unset by this function, non zero
1101  * otherwise.
1102  *
1103  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1104  * Think of a way to unify them.
1105  */
1106 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1107 {
1108         int i, alnum_count = 0;
1109         if (o->color_moved == COLOR_MOVED_PLAIN)
1110                 return block_length;
1111         for (i = 1; i < block_length + 1; i++) {
1112                 const char *c = o->emitted_symbols->buf[n - i].line;
1113                 for (; *c; c++) {
1114                         if (!isalnum(*c))
1115                                 continue;
1116                         alnum_count++;
1117                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1118                                 return 1;
1119                 }
1120         }
1121         for (i = 1; i < block_length + 1; i++)
1122                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1123         return 0;
1124 }
1125
1126 /* Find blocks of moved code, delegate actual coloring decision to helper */
1127 static void mark_color_as_moved(struct diff_options *o,
1128                                 struct hashmap *add_lines,
1129                                 struct hashmap *del_lines)
1130 {
1131         struct moved_block *pmb = NULL; /* potentially moved blocks */
1132         int pmb_nr = 0, pmb_alloc = 0;
1133         int n, flipped_block = 0, block_length = 0;
1134
1135
1136         for (n = 0; n < o->emitted_symbols->nr; n++) {
1137                 struct hashmap *hm = NULL;
1138                 struct moved_entry *key;
1139                 struct moved_entry *match = NULL;
1140                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1141                 enum diff_symbol last_symbol = 0;
1142
1143                 switch (l->s) {
1144                 case DIFF_SYMBOL_PLUS:
1145                         hm = del_lines;
1146                         key = prepare_entry(o, n);
1147                         match = hashmap_get_entry(hm, key, NULL,
1148                                                 struct moved_entry, ent);
1149                         free(key);
1150                         break;
1151                 case DIFF_SYMBOL_MINUS:
1152                         hm = add_lines;
1153                         key = prepare_entry(o, n);
1154                         match = hashmap_get_entry(hm, key, NULL,
1155                                                 struct moved_entry, ent);
1156                         free(key);
1157                         break;
1158                 default:
1159                         flipped_block = 0;
1160                 }
1161
1162                 if (!match) {
1163                         int i;
1164
1165                         adjust_last_block(o, n, block_length);
1166                         for(i = 0; i < pmb_nr; i++)
1167                                 moved_block_clear(&pmb[i]);
1168                         pmb_nr = 0;
1169                         block_length = 0;
1170                         flipped_block = 0;
1171                         last_symbol = l->s;
1172                         continue;
1173                 }
1174
1175                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1176                         last_symbol = l->s;
1177                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1178                         continue;
1179                 }
1180
1181                 if (o->color_moved_ws_handling &
1182                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1183                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1184                 else
1185                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1186
1187                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1188
1189                 if (pmb_nr == 0) {
1190                         /*
1191                          * The current line is the start of a new block.
1192                          * Setup the set of potential blocks.
1193                          */
1194                         hashmap_for_each_entry_from(hm, match,
1195                                                 struct moved_entry, ent) {
1196                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1197                                 if (o->color_moved_ws_handling &
1198                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1199                                         if (compute_ws_delta(l, match->es,
1200                                                              &pmb[pmb_nr].wsd))
1201                                                 pmb[pmb_nr++].match = match;
1202                                 } else {
1203                                         pmb[pmb_nr].wsd = 0;
1204                                         pmb[pmb_nr++].match = match;
1205                                 }
1206                         }
1207
1208                         if (adjust_last_block(o, n, block_length) &&
1209                             pmb_nr && last_symbol != l->s)
1210                                 flipped_block = (flipped_block + 1) % 2;
1211                         else
1212                                 flipped_block = 0;
1213
1214                         block_length = 0;
1215                 }
1216
1217                 if (pmb_nr) {
1218                         block_length++;
1219                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1220                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1221                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1222                 }
1223                 last_symbol = l->s;
1224         }
1225         adjust_last_block(o, n, block_length);
1226
1227         for(n = 0; n < pmb_nr; n++)
1228                 moved_block_clear(&pmb[n]);
1229         free(pmb);
1230 }
1231
1232 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1233   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1234 static void dim_moved_lines(struct diff_options *o)
1235 {
1236         int n;
1237         for (n = 0; n < o->emitted_symbols->nr; n++) {
1238                 struct emitted_diff_symbol *prev = (n != 0) ?
1239                                 &o->emitted_symbols->buf[n - 1] : NULL;
1240                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1241                 struct emitted_diff_symbol *next =
1242                                 (n < o->emitted_symbols->nr - 1) ?
1243                                 &o->emitted_symbols->buf[n + 1] : NULL;
1244
1245                 /* Not a plus or minus line? */
1246                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1247                         continue;
1248
1249                 /* Not a moved line? */
1250                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1251                         continue;
1252
1253                 /*
1254                  * If prev or next are not a plus or minus line,
1255                  * pretend they don't exist
1256                  */
1257                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1258                             prev->s != DIFF_SYMBOL_MINUS)
1259                         prev = NULL;
1260                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1261                             next->s != DIFF_SYMBOL_MINUS)
1262                         next = NULL;
1263
1264                 /* Inside a block? */
1265                 if ((prev &&
1266                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1267                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1268                     (next &&
1269                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1270                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1271                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1272                         continue;
1273                 }
1274
1275                 /* Check if we are at an interesting bound: */
1276                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1277                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1278                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1279                         continue;
1280                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1281                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1282                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1283                         continue;
1284
1285                 /*
1286                  * The boundary to prev and next are not interesting,
1287                  * so this line is not interesting as a whole
1288                  */
1289                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1290         }
1291 }
1292
1293 static void emit_line_ws_markup(struct diff_options *o,
1294                                 const char *set_sign, const char *set,
1295                                 const char *reset,
1296                                 int sign_index, const char *line, int len,
1297                                 unsigned ws_rule, int blank_at_eof)
1298 {
1299         const char *ws = NULL;
1300         int sign = o->output_indicators[sign_index];
1301
1302         if (o->ws_error_highlight & ws_rule) {
1303                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1304                 if (!*ws)
1305                         ws = NULL;
1306         }
1307
1308         if (!ws && !set_sign)
1309                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1310         else if (!ws) {
1311                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1312         } else if (blank_at_eof)
1313                 /* Blank line at EOF - paint '+' as well */
1314                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1315         else {
1316                 /* Emit just the prefix, then the rest. */
1317                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1318                             sign, "", 0);
1319                 ws_check_emit(line, len, ws_rule,
1320                               o->file, set, reset, ws);
1321         }
1322 }
1323
1324 static void emit_diff_symbol_from_struct(struct diff_options *o,
1325                                          struct emitted_diff_symbol *eds)
1326 {
1327         static const char *nneof = " No newline at end of file\n";
1328         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1329         struct strbuf sb = STRBUF_INIT;
1330
1331         enum diff_symbol s = eds->s;
1332         const char *line = eds->line;
1333         int len = eds->len;
1334         unsigned flags = eds->flags;
1335
1336         switch (s) {
1337         case DIFF_SYMBOL_NO_LF_EOF:
1338                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1339                 reset = diff_get_color_opt(o, DIFF_RESET);
1340                 putc('\n', o->file);
1341                 emit_line_0(o, context, NULL, 0, reset, '\\',
1342                             nneof, strlen(nneof));
1343                 break;
1344         case DIFF_SYMBOL_SUBMODULE_HEADER:
1345         case DIFF_SYMBOL_SUBMODULE_ERROR:
1346         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1347         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1348         case DIFF_SYMBOL_SUMMARY:
1349         case DIFF_SYMBOL_STATS_LINE:
1350         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1351         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1352                 emit_line(o, "", "", line, len);
1353                 break;
1354         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1355         case DIFF_SYMBOL_CONTEXT_MARKER:
1356                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1357                 reset = diff_get_color_opt(o, DIFF_RESET);
1358                 emit_line(o, context, reset, line, len);
1359                 break;
1360         case DIFF_SYMBOL_SEPARATOR:
1361                 fprintf(o->file, "%s%c",
1362                         diff_line_prefix(o),
1363                         o->line_termination);
1364                 break;
1365         case DIFF_SYMBOL_CONTEXT:
1366                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1367                 reset = diff_get_color_opt(o, DIFF_RESET);
1368                 set_sign = NULL;
1369                 if (o->flags.dual_color_diffed_diffs) {
1370                         char c = !len ? 0 : line[0];
1371
1372                         if (c == '+')
1373                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1374                         else if (c == '@')
1375                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1376                         else if (c == '-')
1377                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1378                 }
1379                 emit_line_ws_markup(o, set_sign, set, reset,
1380                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1381                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1382                 break;
1383         case DIFF_SYMBOL_PLUS:
1384                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1385                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1386                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1387                 case DIFF_SYMBOL_MOVED_LINE |
1388                      DIFF_SYMBOL_MOVED_LINE_ALT |
1389                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1390                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1391                         break;
1392                 case DIFF_SYMBOL_MOVED_LINE |
1393                      DIFF_SYMBOL_MOVED_LINE_ALT:
1394                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1395                         break;
1396                 case DIFF_SYMBOL_MOVED_LINE |
1397                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1398                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1399                         break;
1400                 case DIFF_SYMBOL_MOVED_LINE:
1401                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1402                         break;
1403                 default:
1404                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1405                 }
1406                 reset = diff_get_color_opt(o, DIFF_RESET);
1407                 if (!o->flags.dual_color_diffed_diffs)
1408                         set_sign = NULL;
1409                 else {
1410                         char c = !len ? 0 : line[0];
1411
1412                         set_sign = set;
1413                         if (c == '-')
1414                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1415                         else if (c == '@')
1416                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1417                         else if (c == '+')
1418                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1419                         else
1420                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1421                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1422                 }
1423                 emit_line_ws_markup(o, set_sign, set, reset,
1424                                     OUTPUT_INDICATOR_NEW, line, len,
1425                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1426                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1427                 break;
1428         case DIFF_SYMBOL_MINUS:
1429                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1430                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1431                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1432                 case DIFF_SYMBOL_MOVED_LINE |
1433                      DIFF_SYMBOL_MOVED_LINE_ALT |
1434                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1435                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1436                         break;
1437                 case DIFF_SYMBOL_MOVED_LINE |
1438                      DIFF_SYMBOL_MOVED_LINE_ALT:
1439                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1440                         break;
1441                 case DIFF_SYMBOL_MOVED_LINE |
1442                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1443                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1444                         break;
1445                 case DIFF_SYMBOL_MOVED_LINE:
1446                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1447                         break;
1448                 default:
1449                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1450                 }
1451                 reset = diff_get_color_opt(o, DIFF_RESET);
1452                 if (!o->flags.dual_color_diffed_diffs)
1453                         set_sign = NULL;
1454                 else {
1455                         char c = !len ? 0 : line[0];
1456
1457                         set_sign = set;
1458                         if (c == '+')
1459                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1460                         else if (c == '@')
1461                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1462                         else if (c == '-')
1463                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1464                         else
1465                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1466                 }
1467                 emit_line_ws_markup(o, set_sign, set, reset,
1468                                     OUTPUT_INDICATOR_OLD, line, len,
1469                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1470                 break;
1471         case DIFF_SYMBOL_WORDS_PORCELAIN:
1472                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1473                 reset = diff_get_color_opt(o, DIFF_RESET);
1474                 emit_line(o, context, reset, line, len);
1475                 fputs("~\n", o->file);
1476                 break;
1477         case DIFF_SYMBOL_WORDS:
1478                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1479                 reset = diff_get_color_opt(o, DIFF_RESET);
1480                 /*
1481                  * Skip the prefix character, if any.  With
1482                  * diff_suppress_blank_empty, there may be
1483                  * none.
1484                  */
1485                 if (line[0] != '\n') {
1486                         line++;
1487                         len--;
1488                 }
1489                 emit_line(o, context, reset, line, len);
1490                 break;
1491         case DIFF_SYMBOL_FILEPAIR_PLUS:
1492                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1493                 reset = diff_get_color_opt(o, DIFF_RESET);
1494                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1495                         line, reset,
1496                         strchr(line, ' ') ? "\t" : "");
1497                 break;
1498         case DIFF_SYMBOL_FILEPAIR_MINUS:
1499                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1500                 reset = diff_get_color_opt(o, DIFF_RESET);
1501                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1502                         line, reset,
1503                         strchr(line, ' ') ? "\t" : "");
1504                 break;
1505         case DIFF_SYMBOL_BINARY_FILES:
1506         case DIFF_SYMBOL_HEADER:
1507                 fprintf(o->file, "%s", line);
1508                 break;
1509         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1510                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1511                 break;
1512         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1513                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1514                 break;
1515         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1516                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1517                 break;
1518         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1519                 fputs(diff_line_prefix(o), o->file);
1520                 fputc('\n', o->file);
1521                 break;
1522         case DIFF_SYMBOL_REWRITE_DIFF:
1523                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1524                 reset = diff_get_color_opt(o, DIFF_RESET);
1525                 emit_line(o, fraginfo, reset, line, len);
1526                 break;
1527         case DIFF_SYMBOL_SUBMODULE_ADD:
1528                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1529                 reset = diff_get_color_opt(o, DIFF_RESET);
1530                 emit_line(o, set, reset, line, len);
1531                 break;
1532         case DIFF_SYMBOL_SUBMODULE_DEL:
1533                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1534                 reset = diff_get_color_opt(o, DIFF_RESET);
1535                 emit_line(o, set, reset, line, len);
1536                 break;
1537         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1538                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1539                         diff_line_prefix(o), line);
1540                 break;
1541         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1542                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1543                         diff_line_prefix(o), line);
1544                 break;
1545         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1546                 emit_line(o, "", "", " 0 files changed\n",
1547                           strlen(" 0 files changed\n"));
1548                 break;
1549         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1550                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1551                 break;
1552         case DIFF_SYMBOL_WORD_DIFF:
1553                 fprintf(o->file, "%.*s", len, line);
1554                 break;
1555         case DIFF_SYMBOL_STAT_SEP:
1556                 fputs(o->stat_sep, o->file);
1557                 break;
1558         default:
1559                 BUG("unknown diff symbol");
1560         }
1561         strbuf_release(&sb);
1562 }
1563
1564 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1565                              const char *line, int len, unsigned flags)
1566 {
1567         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1568
1569         if (o->emitted_symbols)
1570                 append_emitted_diff_symbol(o, &e);
1571         else
1572                 emit_diff_symbol_from_struct(o, &e);
1573 }
1574
1575 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1576 {
1577         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1578 }
1579
1580 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1581 {
1582         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1583 }
1584
1585 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1586 {
1587         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1588                          path, strlen(path), 0);
1589 }
1590
1591 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1592 {
1593         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1594                          path, strlen(path), 0);
1595 }
1596
1597 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1598 {
1599         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1600                          header, strlen(header), 0);
1601 }
1602
1603 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1604 {
1605         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1606 }
1607
1608 void diff_emit_submodule_pipethrough(struct diff_options *o,
1609                                      const char *line, int len)
1610 {
1611         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1612 }
1613
1614 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1615 {
1616         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1617               ecbdata->blank_at_eof_in_preimage &&
1618               ecbdata->blank_at_eof_in_postimage &&
1619               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1620               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1621                 return 0;
1622         return ws_blank_line(line, len, ecbdata->ws_rule);
1623 }
1624
1625 static void emit_add_line(struct emit_callback *ecbdata,
1626                           const char *line, int len)
1627 {
1628         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1629         if (new_blank_line_at_eof(ecbdata, line, len))
1630                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1631
1632         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1633 }
1634
1635 static void emit_del_line(struct emit_callback *ecbdata,
1636                           const char *line, int len)
1637 {
1638         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1639         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1640 }
1641
1642 static void emit_context_line(struct emit_callback *ecbdata,
1643                               const char *line, int len)
1644 {
1645         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1646         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1647 }
1648
1649 static void emit_hunk_header(struct emit_callback *ecbdata,
1650                              const char *line, int len)
1651 {
1652         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1653         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1654         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1655         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1656         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1657         static const char atat[2] = { '@', '@' };
1658         const char *cp, *ep;
1659         struct strbuf msgbuf = STRBUF_INIT;
1660         int org_len = len;
1661         int i = 1;
1662
1663         /*
1664          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1665          * it always is at least 10 bytes long.
1666          */
1667         if (len < 10 ||
1668             memcmp(line, atat, 2) ||
1669             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1670                 emit_diff_symbol(ecbdata->opt,
1671                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1672                 return;
1673         }
1674         ep += 2; /* skip over @@ */
1675
1676         /* The hunk header in fraginfo color */
1677         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1678                 strbuf_addstr(&msgbuf, reverse);
1679         strbuf_addstr(&msgbuf, frag);
1680         if (ecbdata->opt->flags.suppress_hunk_header_line_count)
1681                 strbuf_add(&msgbuf, atat, sizeof(atat));
1682         else
1683                 strbuf_add(&msgbuf, line, ep - line);
1684         strbuf_addstr(&msgbuf, reset);
1685
1686         /*
1687          * trailing "\r\n"
1688          */
1689         for ( ; i < 3; i++)
1690                 if (line[len - i] == '\r' || line[len - i] == '\n')
1691                         len--;
1692
1693         /* blank before the func header */
1694         for (cp = ep; ep - line < len; ep++)
1695                 if (*ep != ' ' && *ep != '\t')
1696                         break;
1697         if (ep != cp) {
1698                 strbuf_addstr(&msgbuf, context);
1699                 strbuf_add(&msgbuf, cp, ep - cp);
1700                 strbuf_addstr(&msgbuf, reset);
1701         }
1702
1703         if (ep < line + len) {
1704                 strbuf_addstr(&msgbuf, func);
1705                 strbuf_add(&msgbuf, ep, line + len - ep);
1706                 strbuf_addstr(&msgbuf, reset);
1707         }
1708
1709         strbuf_add(&msgbuf, line + len, org_len - len);
1710         strbuf_complete_line(&msgbuf);
1711         emit_diff_symbol(ecbdata->opt,
1712                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1713         strbuf_release(&msgbuf);
1714 }
1715
1716 static struct diff_tempfile *claim_diff_tempfile(void)
1717 {
1718         int i;
1719         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1720                 if (!diff_temp[i].name)
1721                         return diff_temp + i;
1722         BUG("diff is failing to clean up its tempfiles");
1723 }
1724
1725 static void remove_tempfile(void)
1726 {
1727         int i;
1728         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1729                 if (is_tempfile_active(diff_temp[i].tempfile))
1730                         delete_tempfile(&diff_temp[i].tempfile);
1731                 diff_temp[i].name = NULL;
1732         }
1733 }
1734
1735 static void add_line_count(struct strbuf *out, int count)
1736 {
1737         switch (count) {
1738         case 0:
1739                 strbuf_addstr(out, "0,0");
1740                 break;
1741         case 1:
1742                 strbuf_addstr(out, "1");
1743                 break;
1744         default:
1745                 strbuf_addf(out, "1,%d", count);
1746                 break;
1747         }
1748 }
1749
1750 static void emit_rewrite_lines(struct emit_callback *ecb,
1751                                int prefix, const char *data, int size)
1752 {
1753         const char *endp = NULL;
1754
1755         while (0 < size) {
1756                 int len;
1757
1758                 endp = memchr(data, '\n', size);
1759                 len = endp ? (endp - data + 1) : size;
1760                 if (prefix != '+') {
1761                         ecb->lno_in_preimage++;
1762                         emit_del_line(ecb, data, len);
1763                 } else {
1764                         ecb->lno_in_postimage++;
1765                         emit_add_line(ecb, data, len);
1766                 }
1767                 size -= len;
1768                 data += len;
1769         }
1770         if (!endp)
1771                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1772 }
1773
1774 static void emit_rewrite_diff(const char *name_a,
1775                               const char *name_b,
1776                               struct diff_filespec *one,
1777                               struct diff_filespec *two,
1778                               struct userdiff_driver *textconv_one,
1779                               struct userdiff_driver *textconv_two,
1780                               struct diff_options *o)
1781 {
1782         int lc_a, lc_b;
1783         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1784         const char *a_prefix, *b_prefix;
1785         char *data_one, *data_two;
1786         size_t size_one, size_two;
1787         struct emit_callback ecbdata;
1788         struct strbuf out = STRBUF_INIT;
1789
1790         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1791                 a_prefix = o->b_prefix;
1792                 b_prefix = o->a_prefix;
1793         } else {
1794                 a_prefix = o->a_prefix;
1795                 b_prefix = o->b_prefix;
1796         }
1797
1798         name_a += (*name_a == '/');
1799         name_b += (*name_b == '/');
1800
1801         strbuf_reset(&a_name);
1802         strbuf_reset(&b_name);
1803         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1804         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1805
1806         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1807         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1808
1809         memset(&ecbdata, 0, sizeof(ecbdata));
1810         ecbdata.color_diff = want_color(o->use_color);
1811         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1812         ecbdata.opt = o;
1813         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1814                 mmfile_t mf1, mf2;
1815                 mf1.ptr = (char *)data_one;
1816                 mf2.ptr = (char *)data_two;
1817                 mf1.size = size_one;
1818                 mf2.size = size_two;
1819                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1820         }
1821         ecbdata.lno_in_preimage = 1;
1822         ecbdata.lno_in_postimage = 1;
1823
1824         lc_a = count_lines(data_one, size_one);
1825         lc_b = count_lines(data_two, size_two);
1826
1827         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1828                          a_name.buf, a_name.len, 0);
1829         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1830                          b_name.buf, b_name.len, 0);
1831
1832         strbuf_addstr(&out, "@@ -");
1833         if (!o->irreversible_delete)
1834                 add_line_count(&out, lc_a);
1835         else
1836                 strbuf_addstr(&out, "?,?");
1837         strbuf_addstr(&out, " +");
1838         add_line_count(&out, lc_b);
1839         strbuf_addstr(&out, " @@\n");
1840         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1841         strbuf_release(&out);
1842
1843         if (lc_a && !o->irreversible_delete)
1844                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1845         if (lc_b)
1846                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1847         if (textconv_one)
1848                 free((char *)data_one);
1849         if (textconv_two)
1850                 free((char *)data_two);
1851 }
1852
1853 struct diff_words_buffer {
1854         mmfile_t text;
1855         unsigned long alloc;
1856         struct diff_words_orig {
1857                 const char *begin, *end;
1858         } *orig;
1859         int orig_nr, orig_alloc;
1860 };
1861
1862 static void diff_words_append(char *line, unsigned long len,
1863                 struct diff_words_buffer *buffer)
1864 {
1865         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1866         line++;
1867         len--;
1868         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1869         buffer->text.size += len;
1870         buffer->text.ptr[buffer->text.size] = '\0';
1871 }
1872
1873 struct diff_words_style_elem {
1874         const char *prefix;
1875         const char *suffix;
1876         const char *color; /* NULL; filled in by the setup code if
1877                             * color is enabled */
1878 };
1879
1880 struct diff_words_style {
1881         enum diff_words_type type;
1882         struct diff_words_style_elem new_word, old_word, ctx;
1883         const char *newline;
1884 };
1885
1886 static struct diff_words_style diff_words_styles[] = {
1887         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1888         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1889         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1890 };
1891
1892 struct diff_words_data {
1893         struct diff_words_buffer minus, plus;
1894         const char *current_plus;
1895         int last_minus;
1896         struct diff_options *opt;
1897         regex_t *word_regex;
1898         enum diff_words_type type;
1899         struct diff_words_style *style;
1900 };
1901
1902 static int fn_out_diff_words_write_helper(struct diff_options *o,
1903                                           struct diff_words_style_elem *st_el,
1904                                           const char *newline,
1905                                           size_t count, const char *buf)
1906 {
1907         int print = 0;
1908         struct strbuf sb = STRBUF_INIT;
1909
1910         while (count) {
1911                 char *p = memchr(buf, '\n', count);
1912                 if (print)
1913                         strbuf_addstr(&sb, diff_line_prefix(o));
1914
1915                 if (p != buf) {
1916                         const char *reset = st_el->color && *st_el->color ?
1917                                             GIT_COLOR_RESET : NULL;
1918                         if (st_el->color && *st_el->color)
1919                                 strbuf_addstr(&sb, st_el->color);
1920                         strbuf_addstr(&sb, st_el->prefix);
1921                         strbuf_add(&sb, buf, p ? p - buf : count);
1922                         strbuf_addstr(&sb, st_el->suffix);
1923                         if (reset)
1924                                 strbuf_addstr(&sb, reset);
1925                 }
1926                 if (!p)
1927                         goto out;
1928
1929                 strbuf_addstr(&sb, newline);
1930                 count -= p + 1 - buf;
1931                 buf = p + 1;
1932                 print = 1;
1933                 if (count) {
1934                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1935                                          sb.buf, sb.len, 0);
1936                         strbuf_reset(&sb);
1937                 }
1938         }
1939
1940 out:
1941         if (sb.len)
1942                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1943                                  sb.buf, sb.len, 0);
1944         strbuf_release(&sb);
1945         return 0;
1946 }
1947
1948 /*
1949  * '--color-words' algorithm can be described as:
1950  *
1951  *   1. collect the minus/plus lines of a diff hunk, divided into
1952  *      minus-lines and plus-lines;
1953  *
1954  *   2. break both minus-lines and plus-lines into words and
1955  *      place them into two mmfile_t with one word for each line;
1956  *
1957  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1958  *
1959  * And for the common parts of the both file, we output the plus side text.
1960  * diff_words->current_plus is used to trace the current position of the plus file
1961  * which printed. diff_words->last_minus is used to trace the last minus word
1962  * printed.
1963  *
1964  * For '--graph' to work with '--color-words', we need to output the graph prefix
1965  * on each line of color words output. Generally, there are two conditions on
1966  * which we should output the prefix.
1967  *
1968  *   1. diff_words->last_minus == 0 &&
1969  *      diff_words->current_plus == diff_words->plus.text.ptr
1970  *
1971  *      that is: the plus text must start as a new line, and if there is no minus
1972  *      word printed, a graph prefix must be printed.
1973  *
1974  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1975  *      *(diff_words->current_plus - 1) == '\n'
1976  *
1977  *      that is: a graph prefix must be printed following a '\n'
1978  */
1979 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1980 {
1981         if ((diff_words->last_minus == 0 &&
1982                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1983                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1984                 *(diff_words->current_plus - 1) == '\n')) {
1985                 return 1;
1986         } else {
1987                 return 0;
1988         }
1989 }
1990
1991 static void fn_out_diff_words_aux(void *priv,
1992                                   long minus_first, long minus_len,
1993                                   long plus_first, long plus_len,
1994                                   const char *func, long funclen)
1995 {
1996         struct diff_words_data *diff_words = priv;
1997         struct diff_words_style *style = diff_words->style;
1998         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1999         struct diff_options *opt = diff_words->opt;
2000         const char *line_prefix;
2001
2002         assert(opt);
2003         line_prefix = diff_line_prefix(opt);
2004
2005         /* POSIX requires that first be decremented by one if len == 0... */
2006         if (minus_len) {
2007                 minus_begin = diff_words->minus.orig[minus_first].begin;
2008                 minus_end =
2009                         diff_words->minus.orig[minus_first + minus_len - 1].end;
2010         } else
2011                 minus_begin = minus_end =
2012                         diff_words->minus.orig[minus_first].end;
2013
2014         if (plus_len) {
2015                 plus_begin = diff_words->plus.orig[plus_first].begin;
2016                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2017         } else
2018                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2019
2020         if (color_words_output_graph_prefix(diff_words)) {
2021                 fputs(line_prefix, diff_words->opt->file);
2022         }
2023         if (diff_words->current_plus != plus_begin) {
2024                 fn_out_diff_words_write_helper(diff_words->opt,
2025                                 &style->ctx, style->newline,
2026                                 plus_begin - diff_words->current_plus,
2027                                 diff_words->current_plus);
2028         }
2029         if (minus_begin != minus_end) {
2030                 fn_out_diff_words_write_helper(diff_words->opt,
2031                                 &style->old_word, style->newline,
2032                                 minus_end - minus_begin, minus_begin);
2033         }
2034         if (plus_begin != plus_end) {
2035                 fn_out_diff_words_write_helper(diff_words->opt,
2036                                 &style->new_word, style->newline,
2037                                 plus_end - plus_begin, plus_begin);
2038         }
2039
2040         diff_words->current_plus = plus_end;
2041         diff_words->last_minus = minus_first;
2042 }
2043
2044 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2045 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2046                 int *begin, int *end)
2047 {
2048         if (word_regex && *begin < buffer->size) {
2049                 regmatch_t match[1];
2050                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2051                                  buffer->size - *begin, 1, match, 0)) {
2052                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2053                                         '\n', match[0].rm_eo - match[0].rm_so);
2054                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2055                         *begin += match[0].rm_so;
2056                         return *begin >= *end;
2057                 }
2058                 return -1;
2059         }
2060
2061         /* find the next word */
2062         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2063                 (*begin)++;
2064         if (*begin >= buffer->size)
2065                 return -1;
2066
2067         /* find the end of the word */
2068         *end = *begin + 1;
2069         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2070                 (*end)++;
2071
2072         return 0;
2073 }
2074
2075 /*
2076  * This function splits the words in buffer->text, stores the list with
2077  * newline separator into out, and saves the offsets of the original words
2078  * in buffer->orig.
2079  */
2080 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2081                 regex_t *word_regex)
2082 {
2083         int i, j;
2084         long alloc = 0;
2085
2086         out->size = 0;
2087         out->ptr = NULL;
2088
2089         /* fake an empty "0th" word */
2090         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2091         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2092         buffer->orig_nr = 1;
2093
2094         for (i = 0; i < buffer->text.size; i++) {
2095                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2096                         return;
2097
2098                 /* store original boundaries */
2099                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2100                                 buffer->orig_alloc);
2101                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2102                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2103                 buffer->orig_nr++;
2104
2105                 /* store one word */
2106                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2107                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2108                 out->ptr[out->size + j - i] = '\n';
2109                 out->size += j - i + 1;
2110
2111                 i = j - 1;
2112         }
2113 }
2114
2115 /* this executes the word diff on the accumulated buffers */
2116 static void diff_words_show(struct diff_words_data *diff_words)
2117 {
2118         xpparam_t xpp;
2119         xdemitconf_t xecfg;
2120         mmfile_t minus, plus;
2121         struct diff_words_style *style = diff_words->style;
2122
2123         struct diff_options *opt = diff_words->opt;
2124         const char *line_prefix;
2125
2126         assert(opt);
2127         line_prefix = diff_line_prefix(opt);
2128
2129         /* special case: only removal */
2130         if (!diff_words->plus.text.size) {
2131                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2132                                  line_prefix, strlen(line_prefix), 0);
2133                 fn_out_diff_words_write_helper(diff_words->opt,
2134                         &style->old_word, style->newline,
2135                         diff_words->minus.text.size,
2136                         diff_words->minus.text.ptr);
2137                 diff_words->minus.text.size = 0;
2138                 return;
2139         }
2140
2141         diff_words->current_plus = diff_words->plus.text.ptr;
2142         diff_words->last_minus = 0;
2143
2144         memset(&xpp, 0, sizeof(xpp));
2145         memset(&xecfg, 0, sizeof(xecfg));
2146         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2147         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2148         xpp.flags = 0;
2149         /* as only the hunk header will be parsed, we need a 0-context */
2150         xecfg.ctxlen = 0;
2151         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2152                           diff_words, &xpp, &xecfg))
2153                 die("unable to generate word diff");
2154         free(minus.ptr);
2155         free(plus.ptr);
2156         if (diff_words->current_plus != diff_words->plus.text.ptr +
2157                         diff_words->plus.text.size) {
2158                 if (color_words_output_graph_prefix(diff_words))
2159                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2160                                          line_prefix, strlen(line_prefix), 0);
2161                 fn_out_diff_words_write_helper(diff_words->opt,
2162                         &style->ctx, style->newline,
2163                         diff_words->plus.text.ptr + diff_words->plus.text.size
2164                         - diff_words->current_plus, diff_words->current_plus);
2165         }
2166         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2167 }
2168
2169 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2170 static void diff_words_flush(struct emit_callback *ecbdata)
2171 {
2172         struct diff_options *wo = ecbdata->diff_words->opt;
2173
2174         if (ecbdata->diff_words->minus.text.size ||
2175             ecbdata->diff_words->plus.text.size)
2176                 diff_words_show(ecbdata->diff_words);
2177
2178         if (wo->emitted_symbols) {
2179                 struct diff_options *o = ecbdata->opt;
2180                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2181                 int i;
2182
2183                 /*
2184                  * NEEDSWORK:
2185                  * Instead of appending each, concat all words to a line?
2186                  */
2187                 for (i = 0; i < wol->nr; i++)
2188                         append_emitted_diff_symbol(o, &wol->buf[i]);
2189
2190                 for (i = 0; i < wol->nr; i++)
2191                         free((void *)wol->buf[i].line);
2192
2193                 wol->nr = 0;
2194         }
2195 }
2196
2197 static void diff_filespec_load_driver(struct diff_filespec *one,
2198                                       struct index_state *istate)
2199 {
2200         /* Use already-loaded driver */
2201         if (one->driver)
2202                 return;
2203
2204         if (S_ISREG(one->mode))
2205                 one->driver = userdiff_find_by_path(istate, one->path);
2206
2207         /* Fallback to default settings */
2208         if (!one->driver)
2209                 one->driver = userdiff_find_by_name("default");
2210 }
2211
2212 static const char *userdiff_word_regex(struct diff_filespec *one,
2213                                        struct index_state *istate)
2214 {
2215         diff_filespec_load_driver(one, istate);
2216         return one->driver->word_regex;
2217 }
2218
2219 static void init_diff_words_data(struct emit_callback *ecbdata,
2220                                  struct diff_options *orig_opts,
2221                                  struct diff_filespec *one,
2222                                  struct diff_filespec *two)
2223 {
2224         int i;
2225         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2226         memcpy(o, orig_opts, sizeof(struct diff_options));
2227
2228         ecbdata->diff_words =
2229                 xcalloc(1, sizeof(struct diff_words_data));
2230         ecbdata->diff_words->type = o->word_diff;
2231         ecbdata->diff_words->opt = o;
2232
2233         if (orig_opts->emitted_symbols)
2234                 o->emitted_symbols =
2235                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2236
2237         if (!o->word_regex)
2238                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2239         if (!o->word_regex)
2240                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2241         if (!o->word_regex)
2242                 o->word_regex = diff_word_regex_cfg;
2243         if (o->word_regex) {
2244                 ecbdata->diff_words->word_regex = (regex_t *)
2245                         xmalloc(sizeof(regex_t));
2246                 if (regcomp(ecbdata->diff_words->word_regex,
2247                             o->word_regex,
2248                             REG_EXTENDED | REG_NEWLINE))
2249                         die("invalid regular expression: %s",
2250                             o->word_regex);
2251         }
2252         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2253                 if (o->word_diff == diff_words_styles[i].type) {
2254                         ecbdata->diff_words->style =
2255                                 &diff_words_styles[i];
2256                         break;
2257                 }
2258         }
2259         if (want_color(o->use_color)) {
2260                 struct diff_words_style *st = ecbdata->diff_words->style;
2261                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2262                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2263                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2264         }
2265 }
2266
2267 static void free_diff_words_data(struct emit_callback *ecbdata)
2268 {
2269         if (ecbdata->diff_words) {
2270                 diff_words_flush(ecbdata);
2271                 free (ecbdata->diff_words->opt->emitted_symbols);
2272                 free (ecbdata->diff_words->opt);
2273                 free (ecbdata->diff_words->minus.text.ptr);
2274                 free (ecbdata->diff_words->minus.orig);
2275                 free (ecbdata->diff_words->plus.text.ptr);
2276                 free (ecbdata->diff_words->plus.orig);
2277                 if (ecbdata->diff_words->word_regex) {
2278                         regfree(ecbdata->diff_words->word_regex);
2279                         free(ecbdata->diff_words->word_regex);
2280                 }
2281                 FREE_AND_NULL(ecbdata->diff_words);
2282         }
2283 }
2284
2285 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2286 {
2287         if (want_color(diff_use_color))
2288                 return diff_colors[ix];
2289         return "";
2290 }
2291
2292 const char *diff_line_prefix(struct diff_options *opt)
2293 {
2294         struct strbuf *msgbuf;
2295         if (!opt->output_prefix)
2296                 return "";
2297
2298         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2299         return msgbuf->buf;
2300 }
2301
2302 static unsigned long sane_truncate_line(char *line, unsigned long len)
2303 {
2304         const char *cp;
2305         unsigned long allot;
2306         size_t l = len;
2307
2308         cp = line;
2309         allot = l;
2310         while (0 < l) {
2311                 (void) utf8_width(&cp, &l);
2312                 if (!cp)
2313                         break; /* truncated in the middle? */
2314         }
2315         return allot - l;
2316 }
2317
2318 static void find_lno(const char *line, struct emit_callback *ecbdata)
2319 {
2320         const char *p;
2321         ecbdata->lno_in_preimage = 0;
2322         ecbdata->lno_in_postimage = 0;
2323         p = strchr(line, '-');
2324         if (!p)
2325                 return; /* cannot happen */
2326         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2327         p = strchr(p, '+');
2328         if (!p)
2329                 return; /* cannot happen */
2330         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2331 }
2332
2333 static void fn_out_consume(void *priv, char *line, unsigned long len)
2334 {
2335         struct emit_callback *ecbdata = priv;
2336         struct diff_options *o = ecbdata->opt;
2337
2338         o->found_changes = 1;
2339
2340         if (ecbdata->header) {
2341                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2342                                  ecbdata->header->buf, ecbdata->header->len, 0);
2343                 strbuf_reset(ecbdata->header);
2344                 ecbdata->header = NULL;
2345         }
2346
2347         if (ecbdata->label_path[0]) {
2348                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2349                                  ecbdata->label_path[0],
2350                                  strlen(ecbdata->label_path[0]), 0);
2351                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2352                                  ecbdata->label_path[1],
2353                                  strlen(ecbdata->label_path[1]), 0);
2354                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2355         }
2356
2357         if (diff_suppress_blank_empty
2358             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2359                 line[0] = '\n';
2360                 len = 1;
2361         }
2362
2363         if (line[0] == '@') {
2364                 if (ecbdata->diff_words)
2365                         diff_words_flush(ecbdata);
2366                 len = sane_truncate_line(line, len);
2367                 find_lno(line, ecbdata);
2368                 emit_hunk_header(ecbdata, line, len);
2369                 return;
2370         }
2371
2372         if (ecbdata->diff_words) {
2373                 enum diff_symbol s =
2374                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2375                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2376                 if (line[0] == '-') {
2377                         diff_words_append(line, len,
2378                                           &ecbdata->diff_words->minus);
2379                         return;
2380                 } else if (line[0] == '+') {
2381                         diff_words_append(line, len,
2382                                           &ecbdata->diff_words->plus);
2383                         return;
2384                 } else if (starts_with(line, "\\ ")) {
2385                         /*
2386                          * Eat the "no newline at eof" marker as if we
2387                          * saw a "+" or "-" line with nothing on it,
2388                          * and return without diff_words_flush() to
2389                          * defer processing. If this is the end of
2390                          * preimage, more "+" lines may come after it.
2391                          */
2392                         return;
2393                 }
2394                 diff_words_flush(ecbdata);
2395                 emit_diff_symbol(o, s, line, len, 0);
2396                 return;
2397         }
2398
2399         switch (line[0]) {
2400         case '+':
2401                 ecbdata->lno_in_postimage++;
2402                 emit_add_line(ecbdata, line + 1, len - 1);
2403                 break;
2404         case '-':
2405                 ecbdata->lno_in_preimage++;
2406                 emit_del_line(ecbdata, line + 1, len - 1);
2407                 break;
2408         case ' ':
2409                 ecbdata->lno_in_postimage++;
2410                 ecbdata->lno_in_preimage++;
2411                 emit_context_line(ecbdata, line + 1, len - 1);
2412                 break;
2413         default:
2414                 /* incomplete line at the end */
2415                 ecbdata->lno_in_preimage++;
2416                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2417                                  line, len, 0);
2418                 break;
2419         }
2420 }
2421
2422 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2423 {
2424         const char *old_name = a;
2425         const char *new_name = b;
2426         int pfx_length, sfx_length;
2427         int pfx_adjust_for_slash;
2428         int len_a = strlen(a);
2429         int len_b = strlen(b);
2430         int a_midlen, b_midlen;
2431         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2432         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2433
2434         if (qlen_a || qlen_b) {
2435                 quote_c_style(a, name, NULL, 0);
2436                 strbuf_addstr(name, " => ");
2437                 quote_c_style(b, name, NULL, 0);
2438                 return;
2439         }
2440
2441         /* Find common prefix */
2442         pfx_length = 0;
2443         while (*old_name && *new_name && *old_name == *new_name) {
2444                 if (*old_name == '/')
2445                         pfx_length = old_name - a + 1;
2446                 old_name++;
2447                 new_name++;
2448         }
2449
2450         /* Find common suffix */
2451         old_name = a + len_a;
2452         new_name = b + len_b;
2453         sfx_length = 0;
2454         /*
2455          * If there is a common prefix, it must end in a slash.  In
2456          * that case we let this loop run 1 into the prefix to see the
2457          * same slash.
2458          *
2459          * If there is no common prefix, we cannot do this as it would
2460          * underrun the input strings.
2461          */
2462         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2463         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2464                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2465                *old_name == *new_name) {
2466                 if (*old_name == '/')
2467                         sfx_length = len_a - (old_name - a);
2468                 old_name--;
2469                 new_name--;
2470         }
2471
2472         /*
2473          * pfx{mid-a => mid-b}sfx
2474          * {pfx-a => pfx-b}sfx
2475          * pfx{sfx-a => sfx-b}
2476          * name-a => name-b
2477          */
2478         a_midlen = len_a - pfx_length - sfx_length;
2479         b_midlen = len_b - pfx_length - sfx_length;
2480         if (a_midlen < 0)
2481                 a_midlen = 0;
2482         if (b_midlen < 0)
2483                 b_midlen = 0;
2484
2485         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2486         if (pfx_length + sfx_length) {
2487                 strbuf_add(name, a, pfx_length);
2488                 strbuf_addch(name, '{');
2489         }
2490         strbuf_add(name, a + pfx_length, a_midlen);
2491         strbuf_addstr(name, " => ");
2492         strbuf_add(name, b + pfx_length, b_midlen);
2493         if (pfx_length + sfx_length) {
2494                 strbuf_addch(name, '}');
2495                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2496         }
2497 }
2498
2499 struct diffstat_t {
2500         int nr;
2501         int alloc;
2502         struct diffstat_file {
2503                 char *from_name;
2504                 char *name;
2505                 char *print_name;
2506                 const char *comments;
2507                 unsigned is_unmerged:1;
2508                 unsigned is_binary:1;
2509                 unsigned is_renamed:1;
2510                 unsigned is_interesting:1;
2511                 uintmax_t added, deleted;
2512         } **files;
2513 };
2514
2515 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2516                                           const char *name_a,
2517                                           const char *name_b)
2518 {
2519         struct diffstat_file *x;
2520         x = xcalloc(1, sizeof(*x));
2521         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2522         diffstat->files[diffstat->nr++] = x;
2523         if (name_b) {
2524                 x->from_name = xstrdup(name_a);
2525                 x->name = xstrdup(name_b);
2526                 x->is_renamed = 1;
2527         }
2528         else {
2529                 x->from_name = NULL;
2530                 x->name = xstrdup(name_a);
2531         }
2532         return x;
2533 }
2534
2535 static void diffstat_consume(void *priv, char *line, unsigned long len)
2536 {
2537         struct diffstat_t *diffstat = priv;
2538         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2539
2540         if (line[0] == '+')
2541                 x->added++;
2542         else if (line[0] == '-')
2543                 x->deleted++;
2544 }
2545
2546 const char mime_boundary_leader[] = "------------";
2547
2548 static int scale_linear(int it, int width, int max_change)
2549 {
2550         if (!it)
2551                 return 0;
2552         /*
2553          * make sure that at least one '-' or '+' is printed if
2554          * there is any change to this path. The easiest way is to
2555          * scale linearly as if the alloted width is one column shorter
2556          * than it is, and then add 1 to the result.
2557          */
2558         return 1 + (it * (width - 1) / max_change);
2559 }
2560
2561 static void show_graph(struct strbuf *out, char ch, int cnt,
2562                        const char *set, const char *reset)
2563 {
2564         if (cnt <= 0)
2565                 return;
2566         strbuf_addstr(out, set);
2567         strbuf_addchars(out, ch, cnt);
2568         strbuf_addstr(out, reset);
2569 }
2570
2571 static void fill_print_name(struct diffstat_file *file)
2572 {
2573         struct strbuf pname = STRBUF_INIT;
2574
2575         if (file->print_name)
2576                 return;
2577
2578         if (file->is_renamed)
2579                 pprint_rename(&pname, file->from_name, file->name);
2580         else
2581                 quote_c_style(file->name, &pname, NULL, 0);
2582
2583         if (file->comments)
2584                 strbuf_addf(&pname, " (%s)", file->comments);
2585
2586         file->print_name = strbuf_detach(&pname, NULL);
2587 }
2588
2589 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2590                 int files, int insertions, int deletions)
2591 {
2592         struct strbuf sb = STRBUF_INIT;
2593
2594         if (!files) {
2595                 assert(insertions == 0 && deletions == 0);
2596                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2597                                  NULL, 0, 0);
2598                 return;
2599         }
2600
2601         strbuf_addf(&sb,
2602                     (files == 1) ? " %d file changed" : " %d files changed",
2603                     files);
2604
2605         /*
2606          * For binary diff, the caller may want to print "x files
2607          * changed" with insertions == 0 && deletions == 0.
2608          *
2609          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2610          * is probably less confusing (i.e skip over "2 files changed
2611          * but nothing about added/removed lines? Is this a bug in Git?").
2612          */
2613         if (insertions || deletions == 0) {
2614                 strbuf_addf(&sb,
2615                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2616                             insertions);
2617         }
2618
2619         if (deletions || insertions == 0) {
2620                 strbuf_addf(&sb,
2621                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2622                             deletions);
2623         }
2624         strbuf_addch(&sb, '\n');
2625         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2626                          sb.buf, sb.len, 0);
2627         strbuf_release(&sb);
2628 }
2629
2630 void print_stat_summary(FILE *fp, int files,
2631                         int insertions, int deletions)
2632 {
2633         struct diff_options o;
2634         memset(&o, 0, sizeof(o));
2635         o.file = fp;
2636
2637         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2638 }
2639
2640 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2641 {
2642         int i, len, add, del, adds = 0, dels = 0;
2643         uintmax_t max_change = 0, max_len = 0;
2644         int total_files = data->nr, count;
2645         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2646         const char *reset, *add_c, *del_c;
2647         int extra_shown = 0;
2648         const char *line_prefix = diff_line_prefix(options);
2649         struct strbuf out = STRBUF_INIT;
2650
2651         if (data->nr == 0)
2652                 return;
2653
2654         count = options->stat_count ? options->stat_count : data->nr;
2655
2656         reset = diff_get_color_opt(options, DIFF_RESET);
2657         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2658         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2659
2660         /*
2661          * Find the longest filename and max number of changes
2662          */
2663         for (i = 0; (i < count) && (i < data->nr); i++) {
2664                 struct diffstat_file *file = data->files[i];
2665                 uintmax_t change = file->added + file->deleted;
2666
2667                 if (!file->is_interesting && (change == 0)) {
2668                         count++; /* not shown == room for one more */
2669                         continue;
2670                 }
2671                 fill_print_name(file);
2672                 len = strlen(file->print_name);
2673                 if (max_len < len)
2674                         max_len = len;
2675
2676                 if (file->is_unmerged) {
2677                         /* "Unmerged" is 8 characters */
2678                         bin_width = bin_width < 8 ? 8 : bin_width;
2679                         continue;
2680                 }
2681                 if (file->is_binary) {
2682                         /* "Bin XXX -> YYY bytes" */
2683                         int w = 14 + decimal_width(file->added)
2684                                 + decimal_width(file->deleted);
2685                         bin_width = bin_width < w ? w : bin_width;
2686                         /* Display change counts aligned with "Bin" */
2687                         number_width = 3;
2688                         continue;
2689                 }
2690
2691                 if (max_change < change)
2692                         max_change = change;
2693         }
2694         count = i; /* where we can stop scanning in data->files[] */
2695
2696         /*
2697          * We have width = stat_width or term_columns() columns total.
2698          * We want a maximum of min(max_len, stat_name_width) for the name part.
2699          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2700          * We also need 1 for " " and 4 + decimal_width(max_change)
2701          * for " | NNNN " and one the empty column at the end, altogether
2702          * 6 + decimal_width(max_change).
2703          *
2704          * If there's not enough space, we will use the smaller of
2705          * stat_name_width (if set) and 5/8*width for the filename,
2706          * and the rest for constant elements + graph part, but no more
2707          * than stat_graph_width for the graph part.
2708          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2709          * for the standard terminal size).
2710          *
2711          * In other words: stat_width limits the maximum width, and
2712          * stat_name_width fixes the maximum width of the filename,
2713          * and is also used to divide available columns if there
2714          * aren't enough.
2715          *
2716          * Binary files are displayed with "Bin XXX -> YYY bytes"
2717          * instead of the change count and graph. This part is treated
2718          * similarly to the graph part, except that it is not
2719          * "scaled". If total width is too small to accommodate the
2720          * guaranteed minimum width of the filename part and the
2721          * separators and this message, this message will "overflow"
2722          * making the line longer than the maximum width.
2723          */
2724
2725         if (options->stat_width == -1)
2726                 width = term_columns() - strlen(line_prefix);
2727         else
2728                 width = options->stat_width ? options->stat_width : 80;
2729         number_width = decimal_width(max_change) > number_width ?
2730                 decimal_width(max_change) : number_width;
2731
2732         if (options->stat_graph_width == -1)
2733                 options->stat_graph_width = diff_stat_graph_width;
2734
2735         /*
2736          * Guarantee 3/8*16==6 for the graph part
2737          * and 5/8*16==10 for the filename part
2738          */
2739         if (width < 16 + 6 + number_width)
2740                 width = 16 + 6 + number_width;
2741
2742         /*
2743          * First assign sizes that are wanted, ignoring available width.
2744          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2745          * starting from "XXX" should fit in graph_width.
2746          */
2747         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2748         if (options->stat_graph_width &&
2749             options->stat_graph_width < graph_width)
2750                 graph_width = options->stat_graph_width;
2751
2752         name_width = (options->stat_name_width > 0 &&
2753                       options->stat_name_width < max_len) ?
2754                 options->stat_name_width : max_len;
2755
2756         /*
2757          * Adjust adjustable widths not to exceed maximum width
2758          */
2759         if (name_width + number_width + 6 + graph_width > width) {
2760                 if (graph_width > width * 3/8 - number_width - 6) {
2761                         graph_width = width * 3/8 - number_width - 6;
2762                         if (graph_width < 6)
2763                                 graph_width = 6;
2764                 }
2765
2766                 if (options->stat_graph_width &&
2767                     graph_width > options->stat_graph_width)
2768                         graph_width = options->stat_graph_width;
2769                 if (name_width > width - number_width - 6 - graph_width)
2770                         name_width = width - number_width - 6 - graph_width;
2771                 else
2772                         graph_width = width - number_width - 6 - name_width;
2773         }
2774
2775         /*
2776          * From here name_width is the width of the name area,
2777          * and graph_width is the width of the graph area.
2778          * max_change is used to scale graph properly.
2779          */
2780         for (i = 0; i < count; i++) {
2781                 const char *prefix = "";
2782                 struct diffstat_file *file = data->files[i];
2783                 char *name = file->print_name;
2784                 uintmax_t added = file->added;
2785                 uintmax_t deleted = file->deleted;
2786                 int name_len;
2787
2788                 if (!file->is_interesting && (added + deleted == 0))
2789                         continue;
2790
2791                 /*
2792                  * "scale" the filename
2793                  */
2794                 len = name_width;
2795                 name_len = strlen(name);
2796                 if (name_width < name_len) {
2797                         char *slash;
2798                         prefix = "...";
2799                         len -= 3;
2800                         name += name_len - len;
2801                         slash = strchr(name, '/');
2802                         if (slash)
2803                                 name = slash;
2804                 }
2805
2806                 if (file->is_binary) {
2807                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2808                         strbuf_addf(&out, " %*s", number_width, "Bin");
2809                         if (!added && !deleted) {
2810                                 strbuf_addch(&out, '\n');
2811                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2812                                                  out.buf, out.len, 0);
2813                                 strbuf_reset(&out);
2814                                 continue;
2815                         }
2816                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2817                                 del_c, deleted, reset);
2818                         strbuf_addstr(&out, " -> ");
2819                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2820                                 add_c, added, reset);
2821                         strbuf_addstr(&out, " bytes\n");
2822                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2823                                          out.buf, out.len, 0);
2824                         strbuf_reset(&out);
2825                         continue;
2826                 }
2827                 else if (file->is_unmerged) {
2828                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2829                         strbuf_addstr(&out, " Unmerged\n");
2830                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2831                                          out.buf, out.len, 0);
2832                         strbuf_reset(&out);
2833                         continue;
2834                 }
2835
2836                 /*
2837                  * scale the add/delete
2838                  */
2839                 add = added;
2840                 del = deleted;
2841
2842                 if (graph_width <= max_change) {
2843                         int total = scale_linear(add + del, graph_width, max_change);
2844                         if (total < 2 && add && del)
2845                                 /* width >= 2 due to the sanity check */
2846                                 total = 2;
2847                         if (add < del) {
2848                                 add = scale_linear(add, graph_width, max_change);
2849                                 del = total - add;
2850                         } else {
2851                                 del = scale_linear(del, graph_width, max_change);
2852                                 add = total - del;
2853                         }
2854                 }
2855                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2856                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2857                         number_width, added + deleted,
2858                         added + deleted ? " " : "");
2859                 show_graph(&out, '+', add, add_c, reset);
2860                 show_graph(&out, '-', del, del_c, reset);
2861                 strbuf_addch(&out, '\n');
2862                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2863                                  out.buf, out.len, 0);
2864                 strbuf_reset(&out);
2865         }
2866
2867         for (i = 0; i < data->nr; i++) {
2868                 struct diffstat_file *file = data->files[i];
2869                 uintmax_t added = file->added;
2870                 uintmax_t deleted = file->deleted;
2871
2872                 if (file->is_unmerged ||
2873                     (!file->is_interesting && (added + deleted == 0))) {
2874                         total_files--;
2875                         continue;
2876                 }
2877
2878                 if (!file->is_binary) {
2879                         adds += added;
2880                         dels += deleted;
2881                 }
2882                 if (i < count)
2883                         continue;
2884                 if (!extra_shown)
2885                         emit_diff_symbol(options,
2886                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2887                                          NULL, 0, 0);
2888                 extra_shown = 1;
2889         }
2890
2891         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2892         strbuf_release(&out);
2893 }
2894
2895 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2896 {
2897         int i, adds = 0, dels = 0, total_files = data->nr;
2898
2899         if (data->nr == 0)
2900                 return;
2901
2902         for (i = 0; i < data->nr; i++) {
2903                 int added = data->files[i]->added;
2904                 int deleted = data->files[i]->deleted;
2905
2906                 if (data->files[i]->is_unmerged ||
2907                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2908                         total_files--;
2909                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2910                         adds += added;
2911                         dels += deleted;
2912                 }
2913         }
2914         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2915 }
2916
2917 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2918 {
2919         int i;
2920
2921         if (data->nr == 0)
2922                 return;
2923
2924         for (i = 0; i < data->nr; i++) {
2925                 struct diffstat_file *file = data->files[i];
2926
2927                 fprintf(options->file, "%s", diff_line_prefix(options));
2928
2929                 if (file->is_binary)
2930                         fprintf(options->file, "-\t-\t");
2931                 else
2932                         fprintf(options->file,
2933                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2934                                 file->added, file->deleted);
2935                 if (options->line_termination) {
2936                         fill_print_name(file);
2937                         if (!file->is_renamed)
2938                                 write_name_quoted(file->name, options->file,
2939                                                   options->line_termination);
2940                         else {
2941                                 fputs(file->print_name, options->file);
2942                                 putc(options->line_termination, options->file);
2943                         }
2944                 } else {
2945                         if (file->is_renamed) {
2946                                 putc('\0', options->file);
2947                                 write_name_quoted(file->from_name, options->file, '\0');
2948                         }
2949                         write_name_quoted(file->name, options->file, '\0');
2950                 }
2951         }
2952 }
2953
2954 struct dirstat_file {
2955         const char *name;
2956         unsigned long changed;
2957 };
2958
2959 struct dirstat_dir {
2960         struct dirstat_file *files;
2961         int alloc, nr, permille, cumulative;
2962 };
2963
2964 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2965                 unsigned long changed, const char *base, int baselen)
2966 {
2967         unsigned long sum_changes = 0;
2968         unsigned int sources = 0;
2969         const char *line_prefix = diff_line_prefix(opt);
2970
2971         while (dir->nr) {
2972                 struct dirstat_file *f = dir->files;
2973                 int namelen = strlen(f->name);
2974                 unsigned long changes;
2975                 char *slash;
2976
2977                 if (namelen < baselen)
2978                         break;
2979                 if (memcmp(f->name, base, baselen))
2980                         break;
2981                 slash = strchr(f->name + baselen, '/');
2982                 if (slash) {
2983                         int newbaselen = slash + 1 - f->name;
2984                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2985                         sources++;
2986                 } else {
2987                         changes = f->changed;
2988                         dir->files++;
2989                         dir->nr--;
2990                         sources += 2;
2991                 }
2992                 sum_changes += changes;
2993         }
2994
2995         /*
2996          * We don't report dirstat's for
2997          *  - the top level
2998          *  - or cases where everything came from a single directory
2999          *    under this directory (sources == 1).
3000          */
3001         if (baselen && sources != 1) {
3002                 if (sum_changes) {
3003                         int permille = sum_changes * 1000 / changed;
3004                         if (permille >= dir->permille) {
3005                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
3006                                         permille / 10, permille % 10, baselen, base);
3007                                 if (!dir->cumulative)
3008                                         return 0;
3009                         }
3010                 }
3011         }
3012         return sum_changes;
3013 }
3014
3015 static int dirstat_compare(const void *_a, const void *_b)
3016 {
3017         const struct dirstat_file *a = _a;
3018         const struct dirstat_file *b = _b;
3019         return strcmp(a->name, b->name);
3020 }
3021
3022 static void show_dirstat(struct diff_options *options)
3023 {
3024         int i;
3025         unsigned long changed;
3026         struct dirstat_dir dir;
3027         struct diff_queue_struct *q = &diff_queued_diff;
3028
3029         dir.files = NULL;
3030         dir.alloc = 0;
3031         dir.nr = 0;
3032         dir.permille = options->dirstat_permille;
3033         dir.cumulative = options->flags.dirstat_cumulative;
3034
3035         changed = 0;
3036         for (i = 0; i < q->nr; i++) {
3037                 struct diff_filepair *p = q->queue[i];
3038                 const char *name;
3039                 unsigned long copied, added, damage;
3040
3041                 name = p->two->path ? p->two->path : p->one->path;
3042
3043                 if (p->one->oid_valid && p->two->oid_valid &&
3044                     oideq(&p->one->oid, &p->two->oid)) {
3045                         /*
3046                          * The SHA1 has not changed, so pre-/post-content is
3047                          * identical. We can therefore skip looking at the
3048                          * file contents altogether.
3049                          */
3050                         damage = 0;
3051                         goto found_damage;
3052                 }
3053
3054                 if (options->flags.dirstat_by_file) {
3055                         /*
3056                          * In --dirstat-by-file mode, we don't really need to
3057                          * look at the actual file contents at all.
3058                          * The fact that the SHA1 changed is enough for us to
3059                          * add this file to the list of results
3060                          * (with each file contributing equal damage).
3061                          */
3062                         damage = 1;
3063                         goto found_damage;
3064                 }
3065
3066                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3067                         diff_populate_filespec(options->repo, p->one, 0);
3068                         diff_populate_filespec(options->repo, p->two, 0);
3069                         diffcore_count_changes(options->repo,
3070                                                p->one, p->two, NULL, NULL,
3071                                                &copied, &added);
3072                         diff_free_filespec_data(p->one);
3073                         diff_free_filespec_data(p->two);
3074                 } else if (DIFF_FILE_VALID(p->one)) {
3075                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3076                         copied = added = 0;
3077                         diff_free_filespec_data(p->one);
3078                 } else if (DIFF_FILE_VALID(p->two)) {
3079                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3080                         copied = 0;
3081                         added = p->two->size;
3082                         diff_free_filespec_data(p->two);
3083                 } else
3084                         continue;
3085
3086                 /*
3087                  * Original minus copied is the removed material,
3088                  * added is the new material.  They are both damages
3089                  * made to the preimage.
3090                  * If the resulting damage is zero, we know that
3091                  * diffcore_count_changes() considers the two entries to
3092                  * be identical, but since the oid changed, we
3093                  * know that there must have been _some_ kind of change,
3094                  * so we force all entries to have damage > 0.
3095                  */
3096                 damage = (p->one->size - copied) + added;
3097                 if (!damage)
3098                         damage = 1;
3099
3100 found_damage:
3101                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3102                 dir.files[dir.nr].name = name;
3103                 dir.files[dir.nr].changed = damage;
3104                 changed += damage;
3105                 dir.nr++;
3106         }
3107
3108         /* This can happen even with many files, if everything was renames */
3109         if (!changed)
3110                 return;
3111
3112         /* Show all directories with more than x% of the changes */
3113         QSORT(dir.files, dir.nr, dirstat_compare);
3114         gather_dirstat(options, &dir, changed, "", 0);
3115 }
3116
3117 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3118 {
3119         int i;
3120         unsigned long changed;
3121         struct dirstat_dir dir;
3122
3123         if (data->nr == 0)
3124                 return;
3125
3126         dir.files = NULL;
3127         dir.alloc = 0;
3128         dir.nr = 0;
3129         dir.permille = options->dirstat_permille;
3130         dir.cumulative = options->flags.dirstat_cumulative;
3131
3132         changed = 0;
3133         for (i = 0; i < data->nr; i++) {
3134                 struct diffstat_file *file = data->files[i];
3135                 unsigned long damage = file->added + file->deleted;
3136                 if (file->is_binary)
3137                         /*
3138                          * binary files counts bytes, not lines. Must find some
3139                          * way to normalize binary bytes vs. textual lines.
3140                          * The following heuristic assumes that there are 64
3141                          * bytes per "line".
3142                          * This is stupid and ugly, but very cheap...
3143                          */
3144                         damage = DIV_ROUND_UP(damage, 64);
3145                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3146                 dir.files[dir.nr].name = file->name;
3147                 dir.files[dir.nr].changed = damage;
3148                 changed += damage;
3149                 dir.nr++;
3150         }
3151
3152         /* This can happen even with many files, if everything was renames */
3153         if (!changed)
3154                 return;
3155
3156         /* Show all directories with more than x% of the changes */
3157         QSORT(dir.files, dir.nr, dirstat_compare);
3158         gather_dirstat(options, &dir, changed, "", 0);
3159 }
3160
3161 static void free_diffstat_info(struct diffstat_t *diffstat)
3162 {
3163         int i;
3164         for (i = 0; i < diffstat->nr; i++) {
3165                 struct diffstat_file *f = diffstat->files[i];
3166                 free(f->print_name);
3167                 free(f->name);
3168                 free(f->from_name);
3169                 free(f);
3170         }
3171         free(diffstat->files);
3172 }
3173
3174 struct checkdiff_t {
3175         const char *filename;
3176         int lineno;
3177         int conflict_marker_size;
3178         struct diff_options *o;
3179         unsigned ws_rule;
3180         unsigned status;
3181 };
3182
3183 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3184 {
3185         char firstchar;
3186         int cnt;
3187
3188         if (len < marker_size + 1)
3189                 return 0;
3190         firstchar = line[0];
3191         switch (firstchar) {
3192         case '=': case '>': case '<': case '|':
3193                 break;
3194         default:
3195                 return 0;
3196         }
3197         for (cnt = 1; cnt < marker_size; cnt++)
3198                 if (line[cnt] != firstchar)
3199                         return 0;
3200         /* line[1] thru line[marker_size-1] are same as firstchar */
3201         if (len < marker_size + 1 || !isspace(line[marker_size]))
3202                 return 0;
3203         return 1;
3204 }
3205
3206 static void checkdiff_consume_hunk(void *priv,
3207                                    long ob, long on, long nb, long nn,
3208                                    const char *func, long funclen)
3209
3210 {
3211         struct checkdiff_t *data = priv;
3212         data->lineno = nb - 1;
3213 }
3214
3215 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3216 {
3217         struct checkdiff_t *data = priv;
3218         int marker_size = data->conflict_marker_size;
3219         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3220         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3221         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3222         char *err;
3223         const char *line_prefix;
3224
3225         assert(data->o);
3226         line_prefix = diff_line_prefix(data->o);
3227
3228         if (line[0] == '+') {
3229                 unsigned bad;
3230                 data->lineno++;
3231                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3232                         data->status |= 1;
3233                         fprintf(data->o->file,
3234                                 "%s%s:%d: leftover conflict marker\n",
3235                                 line_prefix, data->filename, data->lineno);
3236                 }
3237                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3238                 if (!bad)
3239                         return;
3240                 data->status |= bad;
3241                 err = whitespace_error_string(bad);
3242                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3243                         line_prefix, data->filename, data->lineno, err);
3244                 free(err);
3245                 emit_line(data->o, set, reset, line, 1);
3246                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3247                               data->o->file, set, reset, ws);
3248         } else if (line[0] == ' ') {
3249                 data->lineno++;
3250         }
3251 }
3252
3253 static unsigned char *deflate_it(char *data,
3254                                  unsigned long size,
3255                                  unsigned long *result_size)
3256 {
3257         int bound;
3258         unsigned char *deflated;
3259         git_zstream stream;
3260
3261         git_deflate_init(&stream, zlib_compression_level);
3262         bound = git_deflate_bound(&stream, size);
3263         deflated = xmalloc(bound);
3264         stream.next_out = deflated;
3265         stream.avail_out = bound;
3266
3267         stream.next_in = (unsigned char *)data;
3268         stream.avail_in = size;
3269         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3270                 ; /* nothing */
3271         git_deflate_end(&stream);
3272         *result_size = stream.total_out;
3273         return deflated;
3274 }
3275
3276 static void emit_binary_diff_body(struct diff_options *o,
3277                                   mmfile_t *one, mmfile_t *two)
3278 {
3279         void *cp;
3280         void *delta;
3281         void *deflated;
3282         void *data;
3283         unsigned long orig_size;
3284         unsigned long delta_size;
3285         unsigned long deflate_size;
3286         unsigned long data_size;
3287
3288         /* We could do deflated delta, or we could do just deflated two,
3289          * whichever is smaller.
3290          */
3291         delta = NULL;
3292         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3293         if (one->size && two->size) {
3294                 delta = diff_delta(one->ptr, one->size,
3295                                    two->ptr, two->size,
3296                                    &delta_size, deflate_size);
3297                 if (delta) {
3298                         void *to_free = delta;
3299                         orig_size = delta_size;
3300                         delta = deflate_it(delta, delta_size, &delta_size);
3301                         free(to_free);
3302                 }
3303         }
3304
3305         if (delta && delta_size < deflate_size) {
3306                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3307                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3308                                  s, strlen(s), 0);
3309                 free(s);
3310                 free(deflated);
3311                 data = delta;
3312                 data_size = delta_size;
3313         } else {
3314                 char *s = xstrfmt("%lu", two->size);
3315                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3316                                  s, strlen(s), 0);
3317                 free(s);
3318                 free(delta);
3319                 data = deflated;
3320                 data_size = deflate_size;
3321         }
3322
3323         /* emit data encoded in base85 */
3324         cp = data;
3325         while (data_size) {
3326                 int len;
3327                 int bytes = (52 < data_size) ? 52 : data_size;
3328                 char line[71];
3329                 data_size -= bytes;
3330                 if (bytes <= 26)
3331                         line[0] = bytes + 'A' - 1;
3332                 else
3333                         line[0] = bytes - 26 + 'a' - 1;
3334                 encode_85(line + 1, cp, bytes);
3335                 cp = (char *) cp + bytes;
3336
3337                 len = strlen(line);
3338                 line[len++] = '\n';
3339                 line[len] = '\0';
3340
3341                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3342                                  line, len, 0);
3343         }
3344         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3345         free(data);
3346 }
3347
3348 static void emit_binary_diff(struct diff_options *o,
3349                              mmfile_t *one, mmfile_t *two)
3350 {
3351         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3352         emit_binary_diff_body(o, one, two);
3353         emit_binary_diff_body(o, two, one);
3354 }
3355
3356 int diff_filespec_is_binary(struct repository *r,
3357                             struct diff_filespec *one)
3358 {
3359         if (one->is_binary == -1) {
3360                 diff_filespec_load_driver(one, r->index);
3361                 if (one->driver->binary != -1)
3362                         one->is_binary = one->driver->binary;
3363                 else {
3364                         if (!one->data && DIFF_FILE_VALID(one))
3365                                 diff_populate_filespec(r, one, CHECK_BINARY);
3366                         if (one->is_binary == -1 && one->data)
3367                                 one->is_binary = buffer_is_binary(one->data,
3368                                                 one->size);
3369                         if (one->is_binary == -1)
3370                                 one->is_binary = 0;
3371                 }
3372         }
3373         return one->is_binary;
3374 }
3375
3376 static const struct userdiff_funcname *
3377 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3378 {
3379         diff_filespec_load_driver(one, o->repo->index);
3380         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3381 }
3382
3383 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3384 {
3385         if (!options->a_prefix)
3386                 options->a_prefix = a;
3387         if (!options->b_prefix)
3388                 options->b_prefix = b;
3389 }
3390
3391 struct userdiff_driver *get_textconv(struct repository *r,
3392                                      struct diff_filespec *one)
3393 {
3394         if (!DIFF_FILE_VALID(one))
3395                 return NULL;
3396
3397         diff_filespec_load_driver(one, r->index);
3398         return userdiff_get_textconv(r, one->driver);
3399 }
3400
3401 static void builtin_diff(const char *name_a,
3402                          const char *name_b,
3403                          struct diff_filespec *one,
3404                          struct diff_filespec *two,
3405                          const char *xfrm_msg,
3406                          int must_show_header,
3407                          struct diff_options *o,
3408                          int complete_rewrite)
3409 {
3410         mmfile_t mf1, mf2;
3411         const char *lbl[2];
3412         char *a_one, *b_two;
3413         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3414         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3415         const char *a_prefix, *b_prefix;
3416         struct userdiff_driver *textconv_one = NULL;
3417         struct userdiff_driver *textconv_two = NULL;
3418         struct strbuf header = STRBUF_INIT;
3419         const char *line_prefix = diff_line_prefix(o);
3420
3421         diff_set_mnemonic_prefix(o, "a/", "b/");
3422         if (o->flags.reverse_diff) {
3423                 a_prefix = o->b_prefix;
3424                 b_prefix = o->a_prefix;
3425         } else {
3426                 a_prefix = o->a_prefix;
3427                 b_prefix = o->b_prefix;
3428         }
3429
3430         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3431             (!one->mode || S_ISGITLINK(one->mode)) &&
3432             (!two->mode || S_ISGITLINK(two->mode))) {
3433                 show_submodule_summary(o, one->path ? one->path : two->path,
3434                                 &one->oid, &two->oid,
3435                                 two->dirty_submodule);
3436                 return;
3437         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3438                    (!one->mode || S_ISGITLINK(one->mode)) &&
3439                    (!two->mode || S_ISGITLINK(two->mode))) {
3440                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3441                                 &one->oid, &two->oid,
3442                                 two->dirty_submodule);
3443                 return;
3444         }
3445
3446         if (o->flags.allow_textconv) {
3447                 textconv_one = get_textconv(o->repo, one);
3448                 textconv_two = get_textconv(o->repo, two);
3449         }
3450
3451         /* Never use a non-valid filename anywhere if at all possible */
3452         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3453         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3454
3455         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3456         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3457         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3458         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3459         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3460         if (lbl[0][0] == '/') {
3461                 /* /dev/null */
3462                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3463                 if (xfrm_msg)
3464                         strbuf_addstr(&header, xfrm_msg);
3465                 must_show_header = 1;
3466         }
3467         else if (lbl[1][0] == '/') {
3468                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3469                 if (xfrm_msg)
3470                         strbuf_addstr(&header, xfrm_msg);
3471                 must_show_header = 1;
3472         }
3473         else {
3474                 if (one->mode != two->mode) {
3475                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3476                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3477                         must_show_header = 1;
3478                 }
3479                 if (xfrm_msg)
3480                         strbuf_addstr(&header, xfrm_msg);
3481
3482                 /*
3483                  * we do not run diff between different kind
3484                  * of objects.
3485                  */
3486                 if ((one->mode ^ two->mode) & S_IFMT)
3487                         goto free_ab_and_return;
3488                 if (complete_rewrite &&
3489                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3490                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3491                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3492                                          header.buf, header.len, 0);
3493                         strbuf_reset(&header);
3494                         emit_rewrite_diff(name_a, name_b, one, two,
3495                                           textconv_one, textconv_two, o);
3496                         o->found_changes = 1;
3497                         goto free_ab_and_return;
3498                 }
3499         }
3500
3501         if (o->irreversible_delete && lbl[1][0] == '/') {
3502                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3503                                  header.len, 0);
3504                 strbuf_reset(&header);
3505                 goto free_ab_and_return;
3506         } else if (!o->flags.text &&
3507                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3508                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3509                 struct strbuf sb = STRBUF_INIT;
3510                 if (!one->data && !two->data &&
3511                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3512                     !o->flags.binary) {
3513                         if (oideq(&one->oid, &two->oid)) {
3514                                 if (must_show_header)
3515                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3516                                                          header.buf, header.len,
3517                                                          0);
3518                                 goto free_ab_and_return;
3519                         }
3520                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3521                                          header.buf, header.len, 0);
3522                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3523                                     diff_line_prefix(o), lbl[0], lbl[1]);
3524                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3525                                          sb.buf, sb.len, 0);
3526                         strbuf_release(&sb);
3527                         goto free_ab_and_return;
3528                 }
3529                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3530                     fill_mmfile(o->repo, &mf2, two) < 0)
3531                         die("unable to read files to diff");
3532                 /* Quite common confusing case */
3533                 if (mf1.size == mf2.size &&
3534                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3535                         if (must_show_header)
3536                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3537                                                  header.buf, header.len, 0);
3538                         goto free_ab_and_return;
3539                 }
3540                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3541                 strbuf_reset(&header);
3542                 if (o->flags.binary)
3543                         emit_binary_diff(o, &mf1, &mf2);
3544                 else {
3545                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3546                                     diff_line_prefix(o), lbl[0], lbl[1]);
3547                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3548                                          sb.buf, sb.len, 0);
3549                         strbuf_release(&sb);
3550                 }
3551                 o->found_changes = 1;
3552         } else {
3553                 /* Crazy xdl interfaces.. */
3554                 const char *diffopts;
3555                 const char *v;
3556                 xpparam_t xpp;
3557                 xdemitconf_t xecfg;
3558                 struct emit_callback ecbdata;
3559                 const struct userdiff_funcname *pe;
3560
3561                 if (must_show_header) {
3562                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3563                                          header.buf, header.len, 0);
3564                         strbuf_reset(&header);
3565                 }
3566
3567                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3568                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3569
3570                 pe = diff_funcname_pattern(o, one);
3571                 if (!pe)
3572                         pe = diff_funcname_pattern(o, two);
3573
3574                 memset(&xpp, 0, sizeof(xpp));
3575                 memset(&xecfg, 0, sizeof(xecfg));
3576                 memset(&ecbdata, 0, sizeof(ecbdata));
3577                 if (o->flags.suppress_diff_headers)
3578                         lbl[0] = NULL;
3579                 ecbdata.label_path = lbl;
3580                 ecbdata.color_diff = want_color(o->use_color);
3581                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3582                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3583                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3584                 ecbdata.opt = o;
3585                 if (header.len && !o->flags.suppress_diff_headers)
3586                         ecbdata.header = &header;
3587                 xpp.flags = o->xdl_opts;
3588                 xpp.anchors = o->anchors;
3589                 xpp.anchors_nr = o->anchors_nr;
3590                 xecfg.ctxlen = o->context;
3591                 xecfg.interhunkctxlen = o->interhunkcontext;
3592                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3593                 if (o->flags.funccontext)
3594                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3595                 if (pe)
3596                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3597
3598                 diffopts = getenv("GIT_DIFF_OPTS");
3599                 if (!diffopts)
3600                         ;
3601                 else if (skip_prefix(diffopts, "--unified=", &v))
3602                         xecfg.ctxlen = strtoul(v, NULL, 10);
3603                 else if (skip_prefix(diffopts, "-u", &v))
3604                         xecfg.ctxlen = strtoul(v, NULL, 10);
3605
3606                 if (o->word_diff)
3607                         init_diff_words_data(&ecbdata, o, one, two);
3608                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3609                                   &ecbdata, &xpp, &xecfg))
3610                         die("unable to generate diff for %s", one->path);
3611                 if (o->word_diff)
3612                         free_diff_words_data(&ecbdata);
3613                 if (textconv_one)
3614                         free(mf1.ptr);
3615                 if (textconv_two)
3616                         free(mf2.ptr);
3617                 xdiff_clear_find_func(&xecfg);
3618         }
3619
3620  free_ab_and_return:
3621         strbuf_release(&header);
3622         diff_free_filespec_data(one);
3623         diff_free_filespec_data(two);
3624         free(a_one);
3625         free(b_two);
3626         return;
3627 }
3628
3629 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3630 {
3631         if (!is_renamed) {
3632                 if (p->status == DIFF_STATUS_ADDED) {
3633                         if (S_ISLNK(p->two->mode))
3634                                 return "new +l";
3635                         else if ((p->two->mode & 0777) == 0755)
3636                                 return "new +x";
3637                         else
3638                                 return "new";
3639                 } else if (p->status == DIFF_STATUS_DELETED)
3640                         return "gone";
3641         }
3642         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3643                 return "mode -l";
3644         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3645                 return "mode +l";
3646         else if ((p->one->mode & 0777) == 0644 &&
3647                  (p->two->mode & 0777) == 0755)
3648                 return "mode +x";
3649         else if ((p->one->mode & 0777) == 0755 &&
3650                  (p->two->mode & 0777) == 0644)
3651                 return "mode -x";
3652         return NULL;
3653 }
3654
3655 static void builtin_diffstat(const char *name_a, const char *name_b,
3656                              struct diff_filespec *one,
3657                              struct diff_filespec *two,
3658                              struct diffstat_t *diffstat,
3659                              struct diff_options *o,
3660                              struct diff_filepair *p)
3661 {
3662         mmfile_t mf1, mf2;
3663         struct diffstat_file *data;
3664         int same_contents;
3665         int complete_rewrite = 0;
3666
3667         if (!DIFF_PAIR_UNMERGED(p)) {
3668                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3669                         complete_rewrite = 1;
3670         }
3671
3672         data = diffstat_add(diffstat, name_a, name_b);
3673         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3674         if (o->flags.stat_with_summary)
3675                 data->comments = get_compact_summary(p, data->is_renamed);
3676
3677         if (!one || !two) {
3678                 data->is_unmerged = 1;
3679                 return;
3680         }
3681
3682         same_contents = oideq(&one->oid, &two->oid);
3683
3684         if (diff_filespec_is_binary(o->repo, one) ||
3685             diff_filespec_is_binary(o->repo, two)) {
3686                 data->is_binary = 1;
3687                 if (same_contents) {
3688                         data->added = 0;
3689                         data->deleted = 0;
3690                 } else {
3691                         data->added = diff_filespec_size(o->repo, two);
3692                         data->deleted = diff_filespec_size(o->repo, one);
3693                 }
3694         }
3695
3696         else if (complete_rewrite) {
3697                 diff_populate_filespec(o->repo, one, 0);
3698                 diff_populate_filespec(o->repo, two, 0);
3699                 data->deleted = count_lines(one->data, one->size);
3700                 data->added = count_lines(two->data, two->size);
3701         }
3702
3703         else if (!same_contents) {
3704                 /* Crazy xdl interfaces.. */
3705                 xpparam_t xpp;
3706                 xdemitconf_t xecfg;
3707
3708                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3709                     fill_mmfile(o->repo, &mf2, two) < 0)
3710                         die("unable to read files to diff");
3711
3712                 memset(&xpp, 0, sizeof(xpp));
3713                 memset(&xecfg, 0, sizeof(xecfg));
3714                 xpp.flags = o->xdl_opts;
3715                 xpp.anchors = o->anchors;
3716                 xpp.anchors_nr = o->anchors_nr;
3717                 xecfg.ctxlen = o->context;
3718                 xecfg.interhunkctxlen = o->interhunkcontext;
3719                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3720                                   diffstat_consume, diffstat, &xpp, &xecfg))
3721                         die("unable to generate diffstat for %s", one->path);
3722         }
3723
3724         diff_free_filespec_data(one);
3725         diff_free_filespec_data(two);
3726 }
3727
3728 static void builtin_checkdiff(const char *name_a, const char *name_b,
3729                               const char *attr_path,
3730                               struct diff_filespec *one,
3731                               struct diff_filespec *two,
3732                               struct diff_options *o)
3733 {
3734         mmfile_t mf1, mf2;
3735         struct checkdiff_t data;
3736
3737         if (!two)
3738                 return;
3739
3740         memset(&data, 0, sizeof(data));
3741         data.filename = name_b ? name_b : name_a;
3742         data.lineno = 0;
3743         data.o = o;
3744         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3745         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3746
3747         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3748             fill_mmfile(o->repo, &mf2, two) < 0)
3749                 die("unable to read files to diff");
3750
3751         /*
3752          * All the other codepaths check both sides, but not checking
3753          * the "old" side here is deliberate.  We are checking the newly
3754          * introduced changes, and as long as the "new" side is text, we
3755          * can and should check what it introduces.
3756          */
3757         if (diff_filespec_is_binary(o->repo, two))
3758                 goto free_and_return;
3759         else {
3760                 /* Crazy xdl interfaces.. */
3761                 xpparam_t xpp;
3762                 xdemitconf_t xecfg;
3763
3764                 memset(&xpp, 0, sizeof(xpp));
3765                 memset(&xecfg, 0, sizeof(xecfg));
3766                 xecfg.ctxlen = 1; /* at least one context line */
3767                 xpp.flags = 0;
3768                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3769                                   checkdiff_consume, &data,
3770                                   &xpp, &xecfg))
3771                         die("unable to generate checkdiff for %s", one->path);
3772
3773                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3774                         struct emit_callback ecbdata;
3775                         int blank_at_eof;
3776
3777                         ecbdata.ws_rule = data.ws_rule;
3778                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3779                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3780
3781                         if (blank_at_eof) {
3782                                 static char *err;
3783                                 if (!err)
3784                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3785                                 fprintf(o->file, "%s:%d: %s.\n",
3786                                         data.filename, blank_at_eof, err);
3787                                 data.status = 1; /* report errors */
3788                         }
3789                 }
3790         }
3791  free_and_return:
3792         diff_free_filespec_data(one);
3793         diff_free_filespec_data(two);
3794         if (data.status)
3795                 o->flags.check_failed = 1;
3796 }
3797
3798 struct diff_filespec *alloc_filespec(const char *path)
3799 {
3800         struct diff_filespec *spec;
3801
3802         FLEXPTR_ALLOC_STR(spec, path, path);
3803         spec->count = 1;
3804         spec->is_binary = -1;
3805         return spec;
3806 }
3807
3808 void free_filespec(struct diff_filespec *spec)
3809 {
3810         if (!--spec->count) {
3811                 diff_free_filespec_data(spec);
3812                 free(spec);
3813         }
3814 }
3815
3816 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3817                    int oid_valid, unsigned short mode)
3818 {
3819         if (mode) {
3820                 spec->mode = canon_mode(mode);
3821                 oidcpy(&spec->oid, oid);
3822                 spec->oid_valid = oid_valid;
3823         }
3824 }
3825
3826 /*
3827  * Given a name and sha1 pair, if the index tells us the file in
3828  * the work tree has that object contents, return true, so that
3829  * prepare_temp_file() does not have to inflate and extract.
3830  */
3831 static int reuse_worktree_file(struct index_state *istate,
3832                                const char *name,
3833                                const struct object_id *oid,
3834                                int want_file)
3835 {
3836         const struct cache_entry *ce;
3837         struct stat st;
3838         int pos, len;
3839
3840         /*
3841          * We do not read the cache ourselves here, because the
3842          * benchmark with my previous version that always reads cache
3843          * shows that it makes things worse for diff-tree comparing
3844          * two linux-2.6 kernel trees in an already checked out work
3845          * tree.  This is because most diff-tree comparisons deal with
3846          * only a small number of files, while reading the cache is
3847          * expensive for a large project, and its cost outweighs the
3848          * savings we get by not inflating the object to a temporary
3849          * file.  Practically, this code only helps when we are used
3850          * by diff-cache --cached, which does read the cache before
3851          * calling us.
3852          */
3853         if (!istate->cache)
3854                 return 0;
3855
3856         /* We want to avoid the working directory if our caller
3857          * doesn't need the data in a normal file, this system
3858          * is rather slow with its stat/open/mmap/close syscalls,
3859          * and the object is contained in a pack file.  The pack
3860          * is probably already open and will be faster to obtain
3861          * the data through than the working directory.  Loose
3862          * objects however would tend to be slower as they need
3863          * to be individually opened and inflated.
3864          */
3865         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3866                 return 0;
3867
3868         /*
3869          * Similarly, if we'd have to convert the file contents anyway, that
3870          * makes the optimization not worthwhile.
3871          */
3872         if (!want_file && would_convert_to_git(istate, name))
3873                 return 0;
3874
3875         len = strlen(name);
3876         pos = index_name_pos(istate, name, len);
3877         if (pos < 0)
3878                 return 0;
3879         ce = istate->cache[pos];
3880
3881         /*
3882          * This is not the sha1 we are looking for, or
3883          * unreusable because it is not a regular file.
3884          */
3885         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3886                 return 0;
3887
3888         /*
3889          * If ce is marked as "assume unchanged", there is no
3890          * guarantee that work tree matches what we are looking for.
3891          */
3892         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3893                 return 0;
3894
3895         /*
3896          * If ce matches the file in the work tree, we can reuse it.
3897          */
3898         if (ce_uptodate(ce) ||
3899             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3900                 return 1;
3901
3902         return 0;
3903 }
3904
3905 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3906 {
3907         struct strbuf buf = STRBUF_INIT;
3908         char *dirty = "";
3909
3910         /* Are we looking at the work tree? */
3911         if (s->dirty_submodule)
3912                 dirty = "-dirty";
3913
3914         strbuf_addf(&buf, "Subproject commit %s%s\n",
3915                     oid_to_hex(&s->oid), dirty);
3916         s->size = buf.len;
3917         if (size_only) {
3918                 s->data = NULL;
3919                 strbuf_release(&buf);
3920         } else {
3921                 s->data = strbuf_detach(&buf, NULL);
3922                 s->should_free = 1;
3923         }
3924         return 0;
3925 }
3926
3927 /*
3928  * While doing rename detection and pickaxe operation, we may need to
3929  * grab the data for the blob (or file) for our own in-core comparison.
3930  * diff_filespec has data and size fields for this purpose.
3931  */
3932 int diff_populate_filespec(struct repository *r,
3933                            struct diff_filespec *s,
3934                            unsigned int flags)
3935 {
3936         int size_only = flags & CHECK_SIZE_ONLY;
3937         int err = 0;
3938         int conv_flags = global_conv_flags_eol;
3939         /*
3940          * demote FAIL to WARN to allow inspecting the situation
3941          * instead of refusing.
3942          */
3943         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3944                 conv_flags = CONV_EOL_RNDTRP_WARN;
3945
3946         if (!DIFF_FILE_VALID(s))
3947                 die("internal error: asking to populate invalid file.");
3948         if (S_ISDIR(s->mode))
3949                 return -1;
3950
3951         if (s->data)
3952                 return 0;
3953
3954         if (size_only && 0 < s->size)
3955                 return 0;
3956
3957         if (S_ISGITLINK(s->mode))
3958                 return diff_populate_gitlink(s, size_only);
3959
3960         if (!s->oid_valid ||
3961             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3962                 struct strbuf buf = STRBUF_INIT;
3963                 struct stat st;
3964                 int fd;
3965
3966                 if (lstat(s->path, &st) < 0) {
3967                 err_empty:
3968                         err = -1;
3969                 empty:
3970                         s->data = (char *)"";
3971                         s->size = 0;
3972                         return err;
3973                 }
3974                 s->size = xsize_t(st.st_size);
3975                 if (!s->size)
3976                         goto empty;
3977                 if (S_ISLNK(st.st_mode)) {
3978                         struct strbuf sb = STRBUF_INIT;
3979
3980                         if (strbuf_readlink(&sb, s->path, s->size))
3981                                 goto err_empty;
3982                         s->size = sb.len;
3983                         s->data = strbuf_detach(&sb, NULL);
3984                         s->should_free = 1;
3985                         return 0;
3986                 }
3987
3988                 /*
3989                  * Even if the caller would be happy with getting
3990                  * only the size, we cannot return early at this
3991                  * point if the path requires us to run the content
3992                  * conversion.
3993                  */
3994                 if (size_only && !would_convert_to_git(r->index, s->path))
3995                         return 0;
3996
3997                 /*
3998                  * Note: this check uses xsize_t(st.st_size) that may
3999                  * not be the true size of the blob after it goes
4000                  * through convert_to_git().  This may not strictly be
4001                  * correct, but the whole point of big_file_threshold
4002                  * and is_binary check being that we want to avoid
4003                  * opening the file and inspecting the contents, this
4004                  * is probably fine.
4005                  */
4006                 if ((flags & CHECK_BINARY) &&
4007                     s->size > big_file_threshold && s->is_binary == -1) {
4008                         s->is_binary = 1;
4009                         return 0;
4010                 }
4011                 fd = open(s->path, O_RDONLY);
4012                 if (fd < 0)
4013                         goto err_empty;
4014                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4015                 close(fd);
4016                 s->should_munmap = 1;
4017
4018                 /*
4019                  * Convert from working tree format to canonical git format
4020                  */
4021                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4022                         size_t size = 0;
4023                         munmap(s->data, s->size);
4024                         s->should_munmap = 0;
4025                         s->data = strbuf_detach(&buf, &size);
4026                         s->size = size;
4027                         s->should_free = 1;
4028                 }
4029         }
4030         else {
4031                 enum object_type type;
4032                 if (size_only || (flags & CHECK_BINARY)) {
4033                         type = oid_object_info(r, &s->oid, &s->size);
4034                         if (type < 0)
4035                                 die("unable to read %s",
4036                                     oid_to_hex(&s->oid));
4037                         if (size_only)
4038                                 return 0;
4039                         if (s->size > big_file_threshold && s->is_binary == -1) {
4040                                 s->is_binary = 1;
4041                                 return 0;
4042                         }
4043                 }
4044                 s->data = read_object_file(&s->oid, &type, &s->size);
4045                 if (!s->data)
4046                         die("unable to read %s", oid_to_hex(&s->oid));
4047                 s->should_free = 1;
4048         }
4049         return 0;
4050 }
4051
4052 void diff_free_filespec_blob(struct diff_filespec *s)
4053 {
4054         if (s->should_free)
4055                 free(s->data);
4056         else if (s->should_munmap)
4057                 munmap(s->data, s->size);
4058
4059         if (s->should_free || s->should_munmap) {
4060                 s->should_free = s->should_munmap = 0;
4061                 s->data = NULL;
4062         }
4063 }
4064
4065 void diff_free_filespec_data(struct diff_filespec *s)
4066 {
4067         diff_free_filespec_blob(s);
4068         FREE_AND_NULL(s->cnt_data);
4069 }
4070
4071 static void prep_temp_blob(struct index_state *istate,
4072                            const char *path, struct diff_tempfile *temp,
4073                            void *blob,
4074                            unsigned long size,
4075                            const struct object_id *oid,
4076                            int mode)
4077 {
4078         struct strbuf buf = STRBUF_INIT;
4079         struct strbuf tempfile = STRBUF_INIT;
4080         char *path_dup = xstrdup(path);
4081         const char *base = basename(path_dup);
4082
4083         /* Generate "XXXXXX_basename.ext" */
4084         strbuf_addstr(&tempfile, "XXXXXX_");
4085         strbuf_addstr(&tempfile, base);
4086
4087         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4088         if (!temp->tempfile)
4089                 die_errno("unable to create temp-file");
4090         if (convert_to_working_tree(istate, path,
4091                         (const char *)blob, (size_t)size, &buf)) {
4092                 blob = buf.buf;
4093                 size = buf.len;
4094         }
4095         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4096             close_tempfile_gently(temp->tempfile))
4097                 die_errno("unable to write temp-file");
4098         temp->name = get_tempfile_path(temp->tempfile);
4099         oid_to_hex_r(temp->hex, oid);
4100         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4101         strbuf_release(&buf);
4102         strbuf_release(&tempfile);
4103         free(path_dup);
4104 }
4105
4106 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4107                                                const char *name,
4108                                                struct diff_filespec *one)
4109 {
4110         struct diff_tempfile *temp = claim_diff_tempfile();
4111
4112         if (!DIFF_FILE_VALID(one)) {
4113         not_a_valid_file:
4114                 /* A '-' entry produces this for file-2, and
4115                  * a '+' entry produces this for file-1.
4116                  */
4117                 temp->name = "/dev/null";
4118                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4119                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4120                 return temp;
4121         }
4122
4123         if (!S_ISGITLINK(one->mode) &&
4124             (!one->oid_valid ||
4125              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4126                 struct stat st;
4127                 if (lstat(name, &st) < 0) {
4128                         if (errno == ENOENT)
4129                                 goto not_a_valid_file;
4130                         die_errno("stat(%s)", name);
4131                 }
4132                 if (S_ISLNK(st.st_mode)) {
4133                         struct strbuf sb = STRBUF_INIT;
4134                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4135                                 die_errno("readlink(%s)", name);
4136                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4137                                        (one->oid_valid ?
4138                                         &one->oid : &null_oid),
4139                                        (one->oid_valid ?
4140                                         one->mode : S_IFLNK));
4141                         strbuf_release(&sb);
4142                 }
4143                 else {
4144                         /* we can borrow from the file in the work tree */
4145                         temp->name = name;
4146                         if (!one->oid_valid)
4147                                 oid_to_hex_r(temp->hex, &null_oid);
4148                         else
4149                                 oid_to_hex_r(temp->hex, &one->oid);
4150                         /* Even though we may sometimes borrow the
4151                          * contents from the work tree, we always want
4152                          * one->mode.  mode is trustworthy even when
4153                          * !(one->oid_valid), as long as
4154                          * DIFF_FILE_VALID(one).
4155                          */
4156                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4157                 }
4158                 return temp;
4159         }
4160         else {
4161                 if (diff_populate_filespec(r, one, 0))
4162                         die("cannot read data blob for %s", one->path);
4163                 prep_temp_blob(r->index, name, temp,
4164                                one->data, one->size,
4165                                &one->oid, one->mode);
4166         }
4167         return temp;
4168 }
4169
4170 static void add_external_diff_name(struct repository *r,
4171                                    struct argv_array *argv,
4172                                    const char *name,
4173                                    struct diff_filespec *df)
4174 {
4175         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4176         argv_array_push(argv, temp->name);
4177         argv_array_push(argv, temp->hex);
4178         argv_array_push(argv, temp->mode);
4179 }
4180
4181 /* An external diff command takes:
4182  *
4183  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4184  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4185  *
4186  */
4187 static void run_external_diff(const char *pgm,
4188                               const char *name,
4189                               const char *other,
4190                               struct diff_filespec *one,
4191                               struct diff_filespec *two,
4192                               const char *xfrm_msg,
4193                               struct diff_options *o)
4194 {
4195         struct argv_array argv = ARGV_ARRAY_INIT;
4196         struct argv_array env = ARGV_ARRAY_INIT;
4197         struct diff_queue_struct *q = &diff_queued_diff;
4198
4199         argv_array_push(&argv, pgm);
4200         argv_array_push(&argv, name);
4201
4202         if (one && two) {
4203                 add_external_diff_name(o->repo, &argv, name, one);
4204                 if (!other)
4205                         add_external_diff_name(o->repo, &argv, name, two);
4206                 else {
4207                         add_external_diff_name(o->repo, &argv, other, two);
4208                         argv_array_push(&argv, other);
4209                         argv_array_push(&argv, xfrm_msg);
4210                 }
4211         }
4212
4213         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4214         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4215
4216         diff_free_filespec_data(one);
4217         diff_free_filespec_data(two);
4218         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4219                 die(_("external diff died, stopping at %s"), name);
4220
4221         remove_tempfile();
4222         argv_array_clear(&argv);
4223         argv_array_clear(&env);
4224 }
4225
4226 static int similarity_index(struct diff_filepair *p)
4227 {
4228         return p->score * 100 / MAX_SCORE;
4229 }
4230
4231 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4232 {
4233         if (startup_info->have_repository)
4234                 return find_unique_abbrev(oid, abbrev);
4235         else {
4236                 char *hex = oid_to_hex(oid);
4237                 if (abbrev < 0)
4238                         abbrev = FALLBACK_DEFAULT_ABBREV;
4239                 if (abbrev > the_hash_algo->hexsz)
4240                         BUG("oid abbreviation out of range: %d", abbrev);
4241                 if (abbrev)
4242                         hex[abbrev] = '\0';
4243                 return hex;
4244         }
4245 }
4246
4247 static void fill_metainfo(struct strbuf *msg,
4248                           const char *name,
4249                           const char *other,
4250                           struct diff_filespec *one,
4251                           struct diff_filespec *two,
4252                           struct diff_options *o,
4253                           struct diff_filepair *p,
4254                           int *must_show_header,
4255                           int use_color)
4256 {
4257         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4258         const char *reset = diff_get_color(use_color, DIFF_RESET);
4259         const char *line_prefix = diff_line_prefix(o);
4260
4261         *must_show_header = 1;
4262         strbuf_init(msg, PATH_MAX * 2 + 300);
4263         switch (p->status) {
4264         case DIFF_STATUS_COPIED:
4265                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4266                             line_prefix, set, similarity_index(p));
4267                 strbuf_addf(msg, "%s\n%s%scopy from ",
4268                             reset,  line_prefix, set);
4269                 quote_c_style(name, msg, NULL, 0);
4270                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4271                 quote_c_style(other, msg, NULL, 0);
4272                 strbuf_addf(msg, "%s\n", reset);
4273                 break;
4274         case DIFF_STATUS_RENAMED:
4275                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4276                             line_prefix, set, similarity_index(p));
4277                 strbuf_addf(msg, "%s\n%s%srename from ",
4278                             reset, line_prefix, set);
4279                 quote_c_style(name, msg, NULL, 0);
4280                 strbuf_addf(msg, "%s\n%s%srename to ",
4281                             reset, line_prefix, set);
4282                 quote_c_style(other, msg, NULL, 0);
4283                 strbuf_addf(msg, "%s\n", reset);
4284                 break;
4285         case DIFF_STATUS_MODIFIED:
4286                 if (p->score) {
4287                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4288                                     line_prefix,
4289                                     set, similarity_index(p), reset);
4290                         break;
4291                 }
4292                 /* fallthru */
4293         default:
4294                 *must_show_header = 0;
4295         }
4296         if (one && two && !oideq(&one->oid, &two->oid)) {
4297                 const unsigned hexsz = the_hash_algo->hexsz;
4298                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4299
4300                 if (o->flags.binary) {
4301                         mmfile_t mf;
4302                         if ((!fill_mmfile(o->repo, &mf, one) &&
4303                              diff_filespec_is_binary(o->repo, one)) ||
4304                             (!fill_mmfile(o->repo, &mf, two) &&
4305                              diff_filespec_is_binary(o->repo, two)))
4306                                 abbrev = hexsz;
4307                 }
4308                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4309                             diff_abbrev_oid(&one->oid, abbrev),
4310                             diff_abbrev_oid(&two->oid, abbrev));
4311                 if (one->mode == two->mode)
4312                         strbuf_addf(msg, " %06o", one->mode);
4313                 strbuf_addf(msg, "%s\n", reset);
4314         }
4315 }
4316
4317 static void run_diff_cmd(const char *pgm,
4318                          const char *name,
4319                          const char *other,
4320                          const char *attr_path,
4321                          struct diff_filespec *one,
4322                          struct diff_filespec *two,
4323                          struct strbuf *msg,
4324                          struct diff_options *o,
4325                          struct diff_filepair *p)
4326 {
4327         const char *xfrm_msg = NULL;
4328         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4329         int must_show_header = 0;
4330
4331
4332         if (o->flags.allow_external) {
4333                 struct userdiff_driver *drv;
4334
4335                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4336                 if (drv && drv->external)
4337                         pgm = drv->external;
4338         }
4339
4340         if (msg) {
4341                 /*
4342                  * don't use colors when the header is intended for an
4343                  * external diff driver
4344                  */
4345                 fill_metainfo(msg, name, other, one, two, o, p,
4346                               &must_show_header,
4347                               want_color(o->use_color) && !pgm);
4348                 xfrm_msg = msg->len ? msg->buf : NULL;
4349         }
4350
4351         if (pgm) {
4352                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4353                 return;
4354         }
4355         if (one && two)
4356                 builtin_diff(name, other ? other : name,
4357                              one, two, xfrm_msg, must_show_header,
4358                              o, complete_rewrite);
4359         else
4360                 fprintf(o->file, "* Unmerged path %s\n", name);
4361 }
4362
4363 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4364 {
4365         if (DIFF_FILE_VALID(one)) {
4366                 if (!one->oid_valid) {
4367                         struct stat st;
4368                         if (one->is_stdin) {
4369                                 oidclr(&one->oid);
4370                                 return;
4371                         }
4372                         if (lstat(one->path, &st) < 0)
4373                                 die_errno("stat '%s'", one->path);
4374                         if (index_path(istate, &one->oid, one->path, &st, 0))
4375                                 die("cannot hash %s", one->path);
4376                 }
4377         }
4378         else
4379                 oidclr(&one->oid);
4380 }
4381
4382 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4383 {
4384         /* Strip the prefix but do not molest /dev/null and absolute paths */
4385         if (*namep && !is_absolute_path(*namep)) {
4386                 *namep += prefix_length;
4387                 if (**namep == '/')
4388                         ++*namep;
4389         }
4390         if (*otherp && !is_absolute_path(*otherp)) {
4391                 *otherp += prefix_length;
4392                 if (**otherp == '/')
4393                         ++*otherp;
4394         }
4395 }
4396
4397 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4398 {
4399         const char *pgm = external_diff();
4400         struct strbuf msg;
4401         struct diff_filespec *one = p->one;
4402         struct diff_filespec *two = p->two;
4403         const char *name;
4404         const char *other;
4405         const char *attr_path;
4406
4407         name  = one->path;
4408         other = (strcmp(name, two->path) ? two->path : NULL);
4409         attr_path = name;
4410         if (o->prefix_length)
4411                 strip_prefix(o->prefix_length, &name, &other);
4412
4413         if (!o->flags.allow_external)
4414                 pgm = NULL;
4415
4416         if (DIFF_PAIR_UNMERGED(p)) {
4417                 run_diff_cmd(pgm, name, NULL, attr_path,
4418                              NULL, NULL, NULL, o, p);
4419                 return;
4420         }
4421
4422         diff_fill_oid_info(one, o->repo->index);
4423         diff_fill_oid_info(two, o->repo->index);
4424
4425         if (!pgm &&
4426             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4427             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4428                 /*
4429                  * a filepair that changes between file and symlink
4430                  * needs to be split into deletion and creation.
4431                  */
4432                 struct diff_filespec *null = alloc_filespec(two->path);
4433                 run_diff_cmd(NULL, name, other, attr_path,
4434                              one, null, &msg,
4435                              o, p);
4436                 free(null);
4437                 strbuf_release(&msg);
4438
4439                 null = alloc_filespec(one->path);
4440                 run_diff_cmd(NULL, name, other, attr_path,
4441                              null, two, &msg, o, p);
4442                 free(null);
4443         }
4444         else
4445                 run_diff_cmd(pgm, name, other, attr_path,
4446                              one, two, &msg, o, p);
4447
4448         strbuf_release(&msg);
4449 }
4450
4451 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4452                          struct diffstat_t *diffstat)
4453 {
4454         const char *name;
4455         const char *other;
4456
4457         if (DIFF_PAIR_UNMERGED(p)) {
4458                 /* unmerged */
4459                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4460                                  diffstat, o, p);
4461                 return;
4462         }
4463
4464         name = p->one->path;
4465         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4466
4467         if (o->prefix_length)
4468                 strip_prefix(o->prefix_length, &name, &other);
4469
4470         diff_fill_oid_info(p->one, o->repo->index);
4471         diff_fill_oid_info(p->two, o->repo->index);
4472
4473         builtin_diffstat(name, other, p->one, p->two,
4474                          diffstat, o, p);
4475 }
4476
4477 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4478 {
4479         const char *name;
4480         const char *other;
4481         const char *attr_path;
4482
4483         if (DIFF_PAIR_UNMERGED(p)) {
4484                 /* unmerged */
4485                 return;
4486         }
4487
4488         name = p->one->path;
4489         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4490         attr_path = other ? other : name;
4491
4492         if (o->prefix_length)
4493                 strip_prefix(o->prefix_length, &name, &other);
4494
4495         diff_fill_oid_info(p->one, o->repo->index);
4496         diff_fill_oid_info(p->two, o->repo->index);
4497
4498         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4499 }
4500
4501 static void prep_parse_options(struct diff_options *options);
4502
4503 void repo_diff_setup(struct repository *r, struct diff_options *options)
4504 {
4505         memcpy(options, &default_diff_options, sizeof(*options));
4506
4507         options->file = stdout;
4508         options->repo = r;
4509
4510         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4511         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4512         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4513         options->abbrev = DEFAULT_ABBREV;
4514         options->line_termination = '\n';
4515         options->break_opt = -1;
4516         options->rename_limit = -1;
4517         options->dirstat_permille = diff_dirstat_permille_default;
4518         options->context = diff_context_default;
4519         options->interhunkcontext = diff_interhunk_context_default;
4520         options->ws_error_highlight = ws_error_highlight_default;
4521         options->flags.rename_empty = 1;
4522         options->objfind = NULL;
4523
4524         /* pathchange left =NULL by default */
4525         options->change = diff_change;
4526         options->add_remove = diff_addremove;
4527         options->use_color = diff_use_color_default;
4528         options->detect_rename = diff_detect_rename_default;
4529         options->xdl_opts |= diff_algorithm;
4530         if (diff_indent_heuristic)
4531                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4532
4533         options->orderfile = diff_order_file_cfg;
4534
4535         if (diff_no_prefix) {
4536                 options->a_prefix = options->b_prefix = "";
4537         } else if (!diff_mnemonic_prefix) {
4538                 options->a_prefix = "a/";
4539                 options->b_prefix = "b/";
4540         }
4541
4542         options->color_moved = diff_color_moved_default;
4543         options->color_moved_ws_handling = diff_color_moved_ws_default;
4544
4545         prep_parse_options(options);
4546 }
4547
4548 void diff_setup_done(struct diff_options *options)
4549 {
4550         unsigned check_mask = DIFF_FORMAT_NAME |
4551                               DIFF_FORMAT_NAME_STATUS |
4552                               DIFF_FORMAT_CHECKDIFF |
4553                               DIFF_FORMAT_NO_OUTPUT;
4554         /*
4555          * This must be signed because we're comparing against a potentially
4556          * negative value.
4557          */
4558         const int hexsz = the_hash_algo->hexsz;
4559
4560         if (options->set_default)
4561                 options->set_default(options);
4562
4563         if (HAS_MULTI_BITS(options->output_format & check_mask))
4564                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4565
4566         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4567                 die(_("-G, -S and --find-object are mutually exclusive"));
4568
4569         /*
4570          * Most of the time we can say "there are changes"
4571          * only by checking if there are changed paths, but
4572          * --ignore-whitespace* options force us to look
4573          * inside contents.
4574          */
4575
4576         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4577                 options->flags.diff_from_contents = 1;
4578         else
4579                 options->flags.diff_from_contents = 0;
4580
4581         if (options->flags.find_copies_harder)
4582                 options->detect_rename = DIFF_DETECT_COPY;
4583
4584         if (!options->flags.relative_name)
4585                 options->prefix = NULL;
4586         if (options->prefix)
4587                 options->prefix_length = strlen(options->prefix);
4588         else
4589                 options->prefix_length = 0;
4590
4591         if (options->output_format & (DIFF_FORMAT_NAME |
4592                                       DIFF_FORMAT_NAME_STATUS |
4593                                       DIFF_FORMAT_CHECKDIFF |
4594                                       DIFF_FORMAT_NO_OUTPUT))
4595                 options->output_format &= ~(DIFF_FORMAT_RAW |
4596                                             DIFF_FORMAT_NUMSTAT |
4597                                             DIFF_FORMAT_DIFFSTAT |
4598                                             DIFF_FORMAT_SHORTSTAT |
4599                                             DIFF_FORMAT_DIRSTAT |
4600                                             DIFF_FORMAT_SUMMARY |
4601                                             DIFF_FORMAT_PATCH);
4602
4603         /*
4604          * These cases always need recursive; we do not drop caller-supplied
4605          * recursive bits for other formats here.
4606          */
4607         if (options->output_format & (DIFF_FORMAT_PATCH |
4608                                       DIFF_FORMAT_NUMSTAT |
4609                                       DIFF_FORMAT_DIFFSTAT |
4610                                       DIFF_FORMAT_SHORTSTAT |
4611                                       DIFF_FORMAT_DIRSTAT |
4612                                       DIFF_FORMAT_SUMMARY |
4613                                       DIFF_FORMAT_CHECKDIFF))
4614                 options->flags.recursive = 1;
4615         /*
4616          * Also pickaxe would not work very well if you do not say recursive
4617          */
4618         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4619                 options->flags.recursive = 1;
4620         /*
4621          * When patches are generated, submodules diffed against the work tree
4622          * must be checked for dirtiness too so it can be shown in the output
4623          */
4624         if (options->output_format & DIFF_FORMAT_PATCH)
4625                 options->flags.dirty_submodules = 1;
4626
4627         if (options->detect_rename && options->rename_limit < 0)
4628                 options->rename_limit = diff_rename_limit_default;
4629         if (hexsz < options->abbrev)
4630                 options->abbrev = hexsz; /* full */
4631
4632         /*
4633          * It does not make sense to show the first hit we happened
4634          * to have found.  It does not make sense not to return with
4635          * exit code in such a case either.
4636          */
4637         if (options->flags.quick) {
4638                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4639                 options->flags.exit_with_status = 1;
4640         }
4641
4642         options->diff_path_counter = 0;
4643
4644         if (options->flags.follow_renames && options->pathspec.nr != 1)
4645                 die(_("--follow requires exactly one pathspec"));
4646
4647         if (!options->use_color || external_diff())
4648                 options->color_moved = 0;
4649
4650         FREE_AND_NULL(options->parseopts);
4651 }
4652
4653 int parse_long_opt(const char *opt, const char **argv,
4654                    const char **optarg)
4655 {
4656         const char *arg = argv[0];
4657         if (!skip_prefix(arg, "--", &arg))
4658                 return 0;
4659         if (!skip_prefix(arg, opt, &arg))
4660                 return 0;
4661         if (*arg == '=') { /* stuck form: --option=value */
4662                 *optarg = arg + 1;
4663                 return 1;
4664         }
4665         if (*arg != '\0')
4666                 return 0;
4667         /* separate form: --option value */
4668         if (!argv[1])
4669                 die("Option '--%s' requires a value", opt);
4670         *optarg = argv[1];
4671         return 2;
4672 }
4673
4674 static int diff_opt_stat(const struct option *opt, const char *value, int unset)
4675 {
4676         struct diff_options *options = opt->value;
4677         int width = options->stat_width;
4678         int name_width = options->stat_name_width;
4679         int graph_width = options->stat_graph_width;
4680         int count = options->stat_count;
4681         char *end;
4682
4683         BUG_ON_OPT_NEG(unset);
4684
4685         if (!strcmp(opt->long_name, "stat")) {
4686                 if (value) {
4687                         width = strtoul(value, &end, 10);
4688                         if (*end == ',')
4689                                 name_width = strtoul(end+1, &end, 10);
4690                         if (*end == ',')
4691                                 count = strtoul(end+1, &end, 10);
4692                         if (*end)
4693                                 return error(_("invalid --stat value: %s"), value);
4694                 }
4695         } else if (!strcmp(opt->long_name, "stat-width")) {
4696                 width = strtoul(value, &end, 10);
4697                 if (*end)
4698                         return error(_("%s expects a numerical value"),
4699                                      opt->long_name);
4700         } else if (!strcmp(opt->long_name, "stat-name-width")) {
4701                 name_width = strtoul(value, &end, 10);
4702                 if (*end)
4703                         return error(_("%s expects a numerical value"),
4704                                      opt->long_name);
4705         } else if (!strcmp(opt->long_name, "stat-graph-width")) {
4706                 graph_width = strtoul(value, &end, 10);
4707                 if (*end)
4708                         return error(_("%s expects a numerical value"),
4709                                      opt->long_name);
4710         } else if (!strcmp(opt->long_name, "stat-count")) {
4711                 count = strtoul(value, &end, 10);
4712                 if (*end)
4713                         return error(_("%s expects a numerical value"),
4714                                      opt->long_name);
4715         } else
4716                 BUG("%s should not get here", opt->long_name);
4717
4718         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4719         options->stat_name_width = name_width;
4720         options->stat_graph_width = graph_width;
4721         options->stat_width = width;
4722         options->stat_count = count;
4723         return 0;
4724 }
4725
4726 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4727 {
4728         struct strbuf errmsg = STRBUF_INIT;
4729         if (parse_dirstat_params(options, params, &errmsg))
4730                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4731                     errmsg.buf);
4732         strbuf_release(&errmsg);
4733         /*
4734          * The caller knows a dirstat-related option is given from the command
4735          * line; allow it to say "return this_function();"
4736          */
4737         options->output_format |= DIFF_FORMAT_DIRSTAT;
4738         return 1;
4739 }
4740
4741 static const char diff_status_letters[] = {
4742         DIFF_STATUS_ADDED,
4743         DIFF_STATUS_COPIED,
4744         DIFF_STATUS_DELETED,
4745         DIFF_STATUS_MODIFIED,
4746         DIFF_STATUS_RENAMED,
4747         DIFF_STATUS_TYPE_CHANGED,
4748         DIFF_STATUS_UNKNOWN,
4749         DIFF_STATUS_UNMERGED,
4750         DIFF_STATUS_FILTER_AON,
4751         DIFF_STATUS_FILTER_BROKEN,
4752         '\0',
4753 };
4754
4755 static unsigned int filter_bit['Z' + 1];
4756
4757 static void prepare_filter_bits(void)
4758 {
4759         int i;
4760
4761         if (!filter_bit[DIFF_STATUS_ADDED]) {
4762                 for (i = 0; diff_status_letters[i]; i++)
4763                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4764         }
4765 }
4766
4767 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4768 {
4769         return opt->filter & filter_bit[(int) status];
4770 }
4771
4772 unsigned diff_filter_bit(char status)
4773 {
4774         prepare_filter_bits();
4775         return filter_bit[(int) status];
4776 }
4777
4778 static int diff_opt_diff_filter(const struct option *option,
4779                                 const char *optarg, int unset)
4780 {
4781         struct diff_options *opt = option->value;
4782         int i, optch;
4783
4784         BUG_ON_OPT_NEG(unset);
4785         prepare_filter_bits();
4786
4787         /*
4788          * If there is a negation e.g. 'd' in the input, and we haven't
4789          * initialized the filter field with another --diff-filter, start
4790          * from full set of bits, except for AON.
4791          */
4792         if (!opt->filter) {
4793                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4794                         if (optch < 'a' || 'z' < optch)
4795                                 continue;
4796                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4797                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4798                         break;
4799                 }
4800         }
4801
4802         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4803                 unsigned int bit;
4804                 int negate;
4805
4806                 if ('a' <= optch && optch <= 'z') {
4807                         negate = 1;
4808                         optch = toupper(optch);
4809                 } else {
4810                         negate = 0;
4811                 }
4812
4813                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4814                 if (!bit)
4815                         return error(_("unknown change class '%c' in --diff-filter=%s"),
4816                                      optarg[i], optarg);
4817                 if (negate)
4818                         opt->filter &= ~bit;
4819                 else
4820                         opt->filter |= bit;
4821         }
4822         return 0;
4823 }
4824
4825 static void enable_patch_output(int *fmt)
4826 {
4827         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4828         *fmt |= DIFF_FORMAT_PATCH;
4829 }
4830
4831 static int diff_opt_ws_error_highlight(const struct option *option,
4832                                        const char *arg, int unset)
4833 {
4834         struct diff_options *opt = option->value;
4835         int val = parse_ws_error_highlight(arg);
4836
4837         BUG_ON_OPT_NEG(unset);
4838         if (val < 0)
4839                 return error(_("unknown value after ws-error-highlight=%.*s"),
4840                              -1 - val, arg);
4841         opt->ws_error_highlight = val;
4842         return 0;
4843 }
4844
4845 static int diff_opt_find_object(const struct option *option,
4846                                 const char *arg, int unset)
4847 {
4848         struct diff_options *opt = option->value;
4849         struct object_id oid;
4850
4851         BUG_ON_OPT_NEG(unset);
4852         if (get_oid(arg, &oid))
4853                 return error(_("unable to resolve '%s'"), arg);
4854
4855         if (!opt->objfind)
4856                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4857
4858         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4859         opt->flags.recursive = 1;
4860         opt->flags.tree_in_recursive = 1;
4861         oidset_insert(opt->objfind, &oid);
4862         return 0;
4863 }
4864
4865 static int diff_opt_anchored(const struct option *opt,
4866                              const char *arg, int unset)
4867 {
4868         struct diff_options *options = opt->value;
4869
4870         BUG_ON_OPT_NEG(unset);
4871         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4872         ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4873                    options->anchors_alloc);
4874         options->anchors[options->anchors_nr++] = xstrdup(arg);
4875         return 0;
4876 }
4877
4878 static int diff_opt_binary(const struct option *opt,
4879                            const char *arg, int unset)
4880 {
4881         struct diff_options *options = opt->value;
4882
4883         BUG_ON_OPT_NEG(unset);
4884         BUG_ON_OPT_ARG(arg);
4885         enable_patch_output(&options->output_format);
4886         options->flags.binary = 1;
4887         return 0;
4888 }
4889
4890 static int diff_opt_break_rewrites(const struct option *opt,
4891                                    const char *arg, int unset)
4892 {
4893         int *break_opt = opt->value;
4894         int opt1, opt2;
4895
4896         BUG_ON_OPT_NEG(unset);
4897         if (!arg)
4898                 arg = "";
4899         opt1 = parse_rename_score(&arg);
4900         if (*arg == 0)
4901                 opt2 = 0;
4902         else if (*arg != '/')
4903                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4904         else {
4905                 arg++;
4906                 opt2 = parse_rename_score(&arg);
4907         }
4908         if (*arg != 0)
4909                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4910         *break_opt = opt1 | (opt2 << 16);
4911         return 0;
4912 }
4913
4914 static int diff_opt_char(const struct option *opt,
4915                          const char *arg, int unset)
4916 {
4917         char *value = opt->value;
4918
4919         BUG_ON_OPT_NEG(unset);
4920         if (arg[1])
4921                 return error(_("%s expects a character, got '%s'"),
4922                              opt->long_name, arg);
4923         *value = arg[0];
4924         return 0;
4925 }
4926
4927 static int diff_opt_color_moved(const struct option *opt,
4928                                 const char *arg, int unset)
4929 {
4930         struct diff_options *options = opt->value;
4931
4932         if (unset) {
4933                 options->color_moved = COLOR_MOVED_NO;
4934         } else if (!arg) {
4935                 if (diff_color_moved_default)
4936                         options->color_moved = diff_color_moved_default;
4937                 if (options->color_moved == COLOR_MOVED_NO)
4938                         options->color_moved = COLOR_MOVED_DEFAULT;
4939         } else {
4940                 int cm = parse_color_moved(arg);
4941                 if (cm < 0)
4942                         return error(_("bad --color-moved argument: %s"), arg);
4943                 options->color_moved = cm;
4944         }
4945         return 0;
4946 }
4947
4948 static int diff_opt_color_moved_ws(const struct option *opt,
4949                                    const char *arg, int unset)
4950 {
4951         struct diff_options *options = opt->value;
4952         unsigned cm;
4953
4954         if (unset) {
4955                 options->color_moved_ws_handling = 0;
4956                 return 0;
4957         }
4958
4959         cm = parse_color_moved_ws(arg);
4960         if (cm & COLOR_MOVED_WS_ERROR)
4961                 return error(_("invalid mode '%s' in --color-moved-ws"), arg);
4962         options->color_moved_ws_handling = cm;
4963         return 0;
4964 }
4965
4966 static int diff_opt_color_words(const struct option *opt,
4967                                 const char *arg, int unset)
4968 {
4969         struct diff_options *options = opt->value;
4970
4971         BUG_ON_OPT_NEG(unset);
4972         options->use_color = 1;
4973         options->word_diff = DIFF_WORDS_COLOR;
4974         options->word_regex = arg;
4975         return 0;
4976 }
4977
4978 static int diff_opt_compact_summary(const struct option *opt,
4979                                     const char *arg, int unset)
4980 {
4981         struct diff_options *options = opt->value;
4982
4983         BUG_ON_OPT_ARG(arg);
4984         if (unset) {
4985                 options->flags.stat_with_summary = 0;
4986         } else {
4987                 options->flags.stat_with_summary = 1;
4988                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4989         }
4990         return 0;
4991 }
4992
4993 static int diff_opt_diff_algorithm(const struct option *opt,
4994                                    const char *arg, int unset)
4995 {
4996         struct diff_options *options = opt->value;
4997         long value = parse_algorithm_value(arg);
4998
4999         BUG_ON_OPT_NEG(unset);
5000         if (value < 0)
5001                 return error(_("option diff-algorithm accepts \"myers\", "
5002                                "\"minimal\", \"patience\" and \"histogram\""));
5003
5004         /* clear out previous settings */
5005         DIFF_XDL_CLR(options, NEED_MINIMAL);
5006         options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5007         options->xdl_opts |= value;
5008         return 0;
5009 }
5010
5011 static int diff_opt_dirstat(const struct option *opt,
5012                             const char *arg, int unset)
5013 {
5014         struct diff_options *options = opt->value;
5015
5016         BUG_ON_OPT_NEG(unset);
5017         if (!strcmp(opt->long_name, "cumulative")) {
5018                 if (arg)
5019                         BUG("how come --cumulative take a value?");
5020                 arg = "cumulative";
5021         } else if (!strcmp(opt->long_name, "dirstat-by-file"))
5022                 parse_dirstat_opt(options, "files");
5023         parse_dirstat_opt(options, arg ? arg : "");
5024         return 0;
5025 }
5026
5027 static int diff_opt_find_copies(const struct option *opt,
5028                                 const char *arg, int unset)
5029 {
5030         struct diff_options *options = opt->value;
5031
5032         BUG_ON_OPT_NEG(unset);
5033         if (!arg)
5034                 arg = "";
5035         options->rename_score = parse_rename_score(&arg);
5036         if (*arg != 0)
5037                 return error(_("invalid argument to %s"), opt->long_name);
5038
5039         if (options->detect_rename == DIFF_DETECT_COPY)
5040                 options->flags.find_copies_harder = 1;
5041         else
5042                 options->detect_rename = DIFF_DETECT_COPY;
5043
5044         return 0;
5045 }
5046
5047 static int diff_opt_find_renames(const struct option *opt,
5048                                  const char *arg, int unset)
5049 {
5050         struct diff_options *options = opt->value;
5051
5052         BUG_ON_OPT_NEG(unset);
5053         if (!arg)
5054                 arg = "";
5055         options->rename_score = parse_rename_score(&arg);
5056         if (*arg != 0)
5057                 return error(_("invalid argument to %s"), opt->long_name);
5058
5059         options->detect_rename = DIFF_DETECT_RENAME;
5060         return 0;
5061 }
5062
5063 static int diff_opt_follow(const struct option *opt,
5064                            const char *arg, int unset)
5065 {
5066         struct diff_options *options = opt->value;
5067
5068         BUG_ON_OPT_ARG(arg);
5069         if (unset) {
5070                 options->flags.follow_renames = 0;
5071                 options->flags.default_follow_renames = 0;
5072         } else {
5073                 options->flags.follow_renames = 1;
5074         }
5075         return 0;
5076 }
5077
5078 static int diff_opt_ignore_submodules(const struct option *opt,
5079                                       const char *arg, int unset)
5080 {
5081         struct diff_options *options = opt->value;
5082
5083         BUG_ON_OPT_NEG(unset);
5084         if (!arg)
5085                 arg = "all";
5086         options->flags.override_submodule_config = 1;
5087         handle_ignore_submodules_arg(options, arg);
5088         return 0;
5089 }
5090
5091 static int diff_opt_line_prefix(const struct option *opt,
5092                                 const char *optarg, int unset)
5093 {
5094         struct diff_options *options = opt->value;
5095
5096         BUG_ON_OPT_NEG(unset);
5097         options->line_prefix = optarg;
5098         options->line_prefix_length = strlen(options->line_prefix);
5099         graph_setup_line_prefix(options);
5100         return 0;
5101 }
5102
5103 static int diff_opt_no_prefix(const struct option *opt,
5104                               const char *optarg, int unset)
5105 {
5106         struct diff_options *options = opt->value;
5107
5108         BUG_ON_OPT_NEG(unset);
5109         BUG_ON_OPT_ARG(optarg);
5110         options->a_prefix = "";
5111         options->b_prefix = "";
5112         return 0;
5113 }
5114
5115 static enum parse_opt_result diff_opt_output(struct parse_opt_ctx_t *ctx,
5116                                              const struct option *opt,
5117                                              const char *arg, int unset)
5118 {
5119         struct diff_options *options = opt->value;
5120         char *path;
5121
5122         BUG_ON_OPT_NEG(unset);
5123         path = prefix_filename(ctx->prefix, arg);
5124         options->file = xfopen(path, "w");
5125         options->close_file = 1;
5126         if (options->use_color != GIT_COLOR_ALWAYS)
5127                 options->use_color = GIT_COLOR_NEVER;
5128         free(path);
5129         return 0;
5130 }
5131
5132 static int diff_opt_patience(const struct option *opt,
5133                              const char *arg, int unset)
5134 {
5135         struct diff_options *options = opt->value;
5136         int i;
5137
5138         BUG_ON_OPT_NEG(unset);
5139         BUG_ON_OPT_ARG(arg);
5140         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5141         /*
5142          * Both --patience and --anchored use PATIENCE_DIFF
5143          * internally, so remove any anchors previously
5144          * specified.
5145          */
5146         for (i = 0; i < options->anchors_nr; i++)
5147                 free(options->anchors[i]);
5148         options->anchors_nr = 0;
5149         return 0;
5150 }
5151
5152 static int diff_opt_pickaxe_regex(const struct option *opt,
5153                                   const char *arg, int unset)
5154 {
5155         struct diff_options *options = opt->value;
5156
5157         BUG_ON_OPT_NEG(unset);
5158         options->pickaxe = arg;
5159         options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5160         return 0;
5161 }
5162
5163 static int diff_opt_pickaxe_string(const struct option *opt,
5164                                    const char *arg, int unset)
5165 {
5166         struct diff_options *options = opt->value;
5167
5168         BUG_ON_OPT_NEG(unset);
5169         options->pickaxe = arg;
5170         options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5171         return 0;
5172 }
5173
5174 static int diff_opt_relative(const struct option *opt,
5175                              const char *arg, int unset)
5176 {
5177         struct diff_options *options = opt->value;
5178
5179         BUG_ON_OPT_NEG(unset);
5180         options->flags.relative_name = 1;
5181         if (arg)
5182                 options->prefix = arg;
5183         return 0;
5184 }
5185
5186 static int diff_opt_submodule(const struct option *opt,
5187                               const char *arg, int unset)
5188 {
5189         struct diff_options *options = opt->value;
5190
5191         BUG_ON_OPT_NEG(unset);
5192         if (!arg)
5193                 arg = "log";
5194         if (parse_submodule_params(options, arg))
5195                 return error(_("failed to parse --submodule option parameter: '%s'"),
5196                              arg);
5197         return 0;
5198 }
5199
5200 static int diff_opt_textconv(const struct option *opt,
5201                              const char *arg, int unset)
5202 {
5203         struct diff_options *options = opt->value;
5204
5205         BUG_ON_OPT_ARG(arg);
5206         if (unset) {
5207                 options->flags.allow_textconv = 0;
5208         } else {
5209                 options->flags.allow_textconv = 1;
5210                 options->flags.textconv_set_via_cmdline = 1;
5211         }
5212         return 0;
5213 }
5214
5215 static int diff_opt_unified(const struct option *opt,
5216                             const char *arg, int unset)
5217 {
5218         struct diff_options *options = opt->value;
5219         char *s;
5220
5221         BUG_ON_OPT_NEG(unset);
5222
5223         if (arg) {
5224                 options->context = strtol(arg, &s, 10);
5225                 if (*s)
5226                         return error(_("%s expects a numerical value"), "--unified");
5227         }
5228         enable_patch_output(&options->output_format);
5229
5230         return 0;
5231 }
5232
5233 static int diff_opt_word_diff(const struct option *opt,
5234                               const char *arg, int unset)
5235 {
5236         struct diff_options *options = opt->value;
5237
5238         BUG_ON_OPT_NEG(unset);
5239         if (arg) {
5240                 if (!strcmp(arg, "plain"))
5241                         options->word_diff = DIFF_WORDS_PLAIN;
5242                 else if (!strcmp(arg, "color")) {
5243                         options->use_color = 1;
5244                         options->word_diff = DIFF_WORDS_COLOR;
5245                 }
5246                 else if (!strcmp(arg, "porcelain"))
5247                         options->word_diff = DIFF_WORDS_PORCELAIN;
5248                 else if (!strcmp(arg, "none"))
5249                         options->word_diff = DIFF_WORDS_NONE;
5250                 else
5251                         return error(_("bad --word-diff argument: %s"), arg);
5252         } else {
5253                 if (options->word_diff == DIFF_WORDS_NONE)
5254                         options->word_diff = DIFF_WORDS_PLAIN;
5255         }
5256         return 0;
5257 }
5258
5259 static int diff_opt_word_diff_regex(const struct option *opt,
5260                                     const char *arg, int unset)
5261 {
5262         struct diff_options *options = opt->value;
5263
5264         BUG_ON_OPT_NEG(unset);
5265         if (options->word_diff == DIFF_WORDS_NONE)
5266                 options->word_diff = DIFF_WORDS_PLAIN;
5267         options->word_regex = arg;
5268         return 0;
5269 }
5270
5271 static void prep_parse_options(struct diff_options *options)
5272 {
5273         struct option parseopts[] = {
5274                 OPT_GROUP(N_("Diff output format options")),
5275                 OPT_BITOP('p', "patch", &options->output_format,
5276                           N_("generate patch"),
5277                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5278                 OPT_BIT_F('s', "no-patch", &options->output_format,
5279                           N_("suppress diff output"),
5280                           DIFF_FORMAT_NO_OUTPUT, PARSE_OPT_NONEG),
5281                 OPT_BITOP('u', NULL, &options->output_format,
5282                           N_("generate patch"),
5283                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5284                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
5285                                N_("generate diffs with <n> lines context"),
5286                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_unified),
5287                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
5288                          N_("generate diffs with <n> lines context")),
5289                 OPT_BIT_F(0, "raw", &options->output_format,
5290                           N_("generate the diff in raw format"),
5291                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
5292                 OPT_BITOP(0, "patch-with-raw", &options->output_format,
5293                           N_("synonym for '-p --raw'"),
5294                           DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW,
5295                           DIFF_FORMAT_NO_OUTPUT),
5296                 OPT_BITOP(0, "patch-with-stat", &options->output_format,
5297                           N_("synonym for '-p --stat'"),
5298                           DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT,
5299                           DIFF_FORMAT_NO_OUTPUT),
5300                 OPT_BIT_F(0, "numstat", &options->output_format,
5301                           N_("machine friendly --stat"),
5302                           DIFF_FORMAT_NUMSTAT, PARSE_OPT_NONEG),
5303                 OPT_BIT_F(0, "shortstat", &options->output_format,
5304                           N_("output only the last line of --stat"),
5305                           DIFF_FORMAT_SHORTSTAT, PARSE_OPT_NONEG),
5306                 OPT_CALLBACK_F('X', "dirstat", options, N_("<param1,param2>..."),
5307                                N_("output the distribution of relative amount of changes for each sub-directory"),
5308                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5309                                diff_opt_dirstat),
5310                 OPT_CALLBACK_F(0, "cumulative", options, NULL,
5311                                N_("synonym for --dirstat=cumulative"),
5312                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5313                                diff_opt_dirstat),
5314                 OPT_CALLBACK_F(0, "dirstat-by-file", options, N_("<param1,param2>..."),
5315                                N_("synonym for --dirstat=files,param1,param2..."),
5316                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5317                                diff_opt_dirstat),
5318                 OPT_BIT_F(0, "check", &options->output_format,
5319                           N_("warn if changes introduce conflict markers or whitespace errors"),
5320                           DIFF_FORMAT_CHECKDIFF, PARSE_OPT_NONEG),
5321                 OPT_BIT_F(0, "summary", &options->output_format,
5322                           N_("condensed summary such as creations, renames and mode changes"),
5323                           DIFF_FORMAT_SUMMARY, PARSE_OPT_NONEG),
5324                 OPT_BIT_F(0, "name-only", &options->output_format,
5325                           N_("show only names of changed files"),
5326                           DIFF_FORMAT_NAME, PARSE_OPT_NONEG),
5327                 OPT_BIT_F(0, "name-status", &options->output_format,
5328                           N_("show only names and status of changed files"),
5329                           DIFF_FORMAT_NAME_STATUS, PARSE_OPT_NONEG),
5330                 OPT_CALLBACK_F(0, "stat", options, N_("<width>[,<name-width>[,<count>]]"),
5331                                N_("generate diffstat"),
5332                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_stat),
5333                 OPT_CALLBACK_F(0, "stat-width", options, N_("<width>"),
5334                                N_("generate diffstat with a given width"),
5335                                PARSE_OPT_NONEG, diff_opt_stat),
5336                 OPT_CALLBACK_F(0, "stat-name-width", options, N_("<width>"),
5337                                N_("generate diffstat with a given name width"),
5338                                PARSE_OPT_NONEG, diff_opt_stat),
5339                 OPT_CALLBACK_F(0, "stat-graph-width", options, N_("<width>"),
5340                                N_("generate diffstat with a given graph width"),
5341                                PARSE_OPT_NONEG, diff_opt_stat),
5342                 OPT_CALLBACK_F(0, "stat-count", options, N_("<count>"),
5343                                N_("generate diffstat with limited lines"),
5344                                PARSE_OPT_NONEG, diff_opt_stat),
5345                 OPT_CALLBACK_F(0, "compact-summary", options, NULL,
5346                                N_("generate compact summary in diffstat"),
5347                                PARSE_OPT_NOARG, diff_opt_compact_summary),
5348                 OPT_CALLBACK_F(0, "binary", options, NULL,
5349                                N_("output a binary diff that can be applied"),
5350                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_binary),
5351                 OPT_BOOL(0, "full-index", &options->flags.full_index,
5352                          N_("show full pre- and post-image object names on the \"index\" lines")),
5353                 OPT_COLOR_FLAG(0, "color", &options->use_color,
5354                                N_("show colored diff")),
5355                 OPT_CALLBACK_F(0, "ws-error-highlight", options, N_("<kind>"),
5356                                N_("highlight whitespace errors in the 'context', 'old' or 'new' lines in the diff"),
5357                                PARSE_OPT_NONEG, diff_opt_ws_error_highlight),
5358                 OPT_SET_INT('z', NULL, &options->line_termination,
5359                             N_("do not munge pathnames and use NULs as output field terminators in --raw or --numstat"),
5360                             0),
5361                 OPT__ABBREV(&options->abbrev),
5362                 OPT_STRING_F(0, "src-prefix", &options->a_prefix, N_("<prefix>"),
5363                              N_("show the given source prefix instead of \"a/\""),
5364                              PARSE_OPT_NONEG),
5365                 OPT_STRING_F(0, "dst-prefix", &options->b_prefix, N_("<prefix>"),
5366                              N_("show the given destination prefix instead of \"b/\""),
5367                              PARSE_OPT_NONEG),
5368                 OPT_CALLBACK_F(0, "line-prefix", options, N_("<prefix>"),
5369                                N_("prepend an additional prefix to every line of output"),
5370                                PARSE_OPT_NONEG, diff_opt_line_prefix),
5371                 OPT_CALLBACK_F(0, "no-prefix", options, NULL,
5372                                N_("do not show any source or destination prefix"),
5373                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_no_prefix),
5374                 OPT_INTEGER_F(0, "inter-hunk-context", &options->interhunkcontext,
5375                               N_("show context between diff hunks up to the specified number of lines"),
5376                               PARSE_OPT_NONEG),
5377                 OPT_CALLBACK_F(0, "output-indicator-new",
5378                                &options->output_indicators[OUTPUT_INDICATOR_NEW],
5379                                N_("<char>"),
5380                                N_("specify the character to indicate a new line instead of '+'"),
5381                                PARSE_OPT_NONEG, diff_opt_char),
5382                 OPT_CALLBACK_F(0, "output-indicator-old",
5383                                &options->output_indicators[OUTPUT_INDICATOR_OLD],
5384                                N_("<char>"),
5385                                N_("specify the character to indicate an old line instead of '-'"),
5386                                PARSE_OPT_NONEG, diff_opt_char),
5387                 OPT_CALLBACK_F(0, "output-indicator-context",
5388                                &options->output_indicators[OUTPUT_INDICATOR_CONTEXT],
5389                                N_("<char>"),
5390                                N_("specify the character to indicate a context instead of ' '"),
5391                                PARSE_OPT_NONEG, diff_opt_char),
5392
5393                 OPT_GROUP(N_("Diff rename options")),
5394                 OPT_CALLBACK_F('B', "break-rewrites", &options->break_opt, N_("<n>[/<m>]"),
5395                                N_("break complete rewrite changes into pairs of delete and create"),
5396                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5397                                diff_opt_break_rewrites),
5398                 OPT_CALLBACK_F('M', "find-renames", options, N_("<n>"),
5399                                N_("detect renames"),
5400                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5401                                diff_opt_find_renames),
5402                 OPT_SET_INT_F('D', "irreversible-delete", &options->irreversible_delete,
5403                               N_("omit the preimage for deletes"),
5404                               1, PARSE_OPT_NONEG),
5405                 OPT_CALLBACK_F('C', "find-copies", options, N_("<n>"),
5406                                N_("detect copies"),
5407                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5408                                diff_opt_find_copies),
5409                 OPT_BOOL(0, "find-copies-harder", &options->flags.find_copies_harder,
5410                          N_("use unmodified files as source to find copies")),
5411                 OPT_SET_INT_F(0, "no-renames", &options->detect_rename,
5412                               N_("disable rename detection"),
5413                               0, PARSE_OPT_NONEG),
5414                 OPT_BOOL(0, "rename-empty", &options->flags.rename_empty,
5415                          N_("use empty blobs as rename source")),
5416                 OPT_CALLBACK_F(0, "follow", options, NULL,
5417                                N_("continue listing the history of a file beyond renames"),
5418                                PARSE_OPT_NOARG, diff_opt_follow),
5419                 OPT_INTEGER('l', NULL, &options->rename_limit,
5420                             N_("prevent rename/copy detection if the number of rename/copy targets exceeds given limit")),
5421
5422                 OPT_GROUP(N_("Diff algorithm options")),
5423                 OPT_BIT(0, "minimal", &options->xdl_opts,
5424                         N_("produce the smallest possible diff"),
5425                         XDF_NEED_MINIMAL),
5426                 OPT_BIT_F('w', "ignore-all-space", &options->xdl_opts,
5427                           N_("ignore whitespace when comparing lines"),
5428                           XDF_IGNORE_WHITESPACE, PARSE_OPT_NONEG),
5429                 OPT_BIT_F('b', "ignore-space-change", &options->xdl_opts,
5430                           N_("ignore changes in amount of whitespace"),
5431                           XDF_IGNORE_WHITESPACE_CHANGE, PARSE_OPT_NONEG),
5432                 OPT_BIT_F(0, "ignore-space-at-eol", &options->xdl_opts,
5433                           N_("ignore changes in whitespace at EOL"),
5434                           XDF_IGNORE_WHITESPACE_AT_EOL, PARSE_OPT_NONEG),
5435                 OPT_BIT_F(0, "ignore-cr-at-eol", &options->xdl_opts,
5436                           N_("ignore carrier-return at the end of line"),
5437                           XDF_IGNORE_CR_AT_EOL, PARSE_OPT_NONEG),
5438                 OPT_BIT_F(0, "ignore-blank-lines", &options->xdl_opts,
5439                           N_("ignore changes whose lines are all blank"),
5440                           XDF_IGNORE_BLANK_LINES, PARSE_OPT_NONEG),
5441                 OPT_BIT(0, "indent-heuristic", &options->xdl_opts,
5442                         N_("heuristic to shift diff hunk boundaries for easy reading"),
5443                         XDF_INDENT_HEURISTIC),
5444                 OPT_CALLBACK_F(0, "patience", options, NULL,
5445                                N_("generate diff using the \"patience diff\" algorithm"),
5446                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5447                                diff_opt_patience),
5448                 OPT_BITOP(0, "histogram", &options->xdl_opts,
5449                           N_("generate diff using the \"histogram diff\" algorithm"),
5450                           XDF_HISTOGRAM_DIFF, XDF_DIFF_ALGORITHM_MASK),
5451                 OPT_CALLBACK_F(0, "diff-algorithm", options, N_("<algorithm>"),
5452                                N_("choose a diff algorithm"),
5453                                PARSE_OPT_NONEG, diff_opt_diff_algorithm),
5454                 OPT_CALLBACK_F(0, "anchored", options, N_("<text>"),
5455                                N_("generate diff using the \"anchored diff\" algorithm"),
5456                                PARSE_OPT_NONEG, diff_opt_anchored),
5457                 OPT_CALLBACK_F(0, "word-diff", options, N_("<mode>"),
5458                                N_("show word diff, using <mode> to delimit changed words"),
5459                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_word_diff),
5460                 OPT_CALLBACK_F(0, "word-diff-regex", options, N_("<regex>"),
5461                                N_("use <regex> to decide what a word is"),
5462                                PARSE_OPT_NONEG, diff_opt_word_diff_regex),
5463                 OPT_CALLBACK_F(0, "color-words", options, N_("<regex>"),
5464                                N_("equivalent to --word-diff=color --word-diff-regex=<regex>"),
5465                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_color_words),
5466                 OPT_CALLBACK_F(0, "color-moved", options, N_("<mode>"),
5467                                N_("moved lines of code are colored differently"),
5468                                PARSE_OPT_OPTARG, diff_opt_color_moved),
5469                 OPT_CALLBACK_F(0, "color-moved-ws", options, N_("<mode>"),
5470                                N_("how white spaces are ignored in --color-moved"),
5471                                0, diff_opt_color_moved_ws),
5472
5473                 OPT_GROUP(N_("Other diff options")),
5474                 OPT_CALLBACK_F(0, "relative", options, N_("<prefix>"),
5475                                N_("when run from subdir, exclude changes outside and show relative paths"),
5476                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5477                                diff_opt_relative),
5478                 OPT_BOOL('a', "text", &options->flags.text,
5479                          N_("treat all files as text")),
5480                 OPT_BOOL('R', NULL, &options->flags.reverse_diff,
5481                          N_("swap two inputs, reverse the diff")),
5482                 OPT_BOOL(0, "exit-code", &options->flags.exit_with_status,
5483                          N_("exit with 1 if there were differences, 0 otherwise")),
5484                 OPT_BOOL(0, "quiet", &options->flags.quick,
5485                          N_("disable all output of the program")),
5486                 OPT_BOOL(0, "ext-diff", &options->flags.allow_external,
5487                          N_("allow an external diff helper to be executed")),
5488                 OPT_CALLBACK_F(0, "textconv", options, NULL,
5489                                N_("run external text conversion filters when comparing binary files"),
5490                                PARSE_OPT_NOARG, diff_opt_textconv),
5491                 OPT_CALLBACK_F(0, "ignore-submodules", options, N_("<when>"),
5492                                N_("ignore changes to submodules in the diff generation"),
5493                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5494                                diff_opt_ignore_submodules),
5495                 OPT_CALLBACK_F(0, "submodule", options, N_("<format>"),
5496                                N_("specify how differences in submodules are shown"),
5497                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5498                                diff_opt_submodule),
5499                 OPT_SET_INT_F(0, "ita-invisible-in-index", &options->ita_invisible_in_index,
5500                               N_("hide 'git add -N' entries from the index"),
5501                               1, PARSE_OPT_NONEG),
5502                 OPT_SET_INT_F(0, "ita-visible-in-index", &options->ita_invisible_in_index,
5503                               N_("treat 'git add -N' entries as real in the index"),
5504                               0, PARSE_OPT_NONEG),
5505                 OPT_CALLBACK_F('S', NULL, options, N_("<string>"),
5506                                N_("look for differences that change the number of occurrences of the specified string"),
5507                                0, diff_opt_pickaxe_string),
5508                 OPT_CALLBACK_F('G', NULL, options, N_("<regex>"),
5509                                N_("look for differences that change the number of occurrences of the specified regex"),
5510                                0, diff_opt_pickaxe_regex),
5511                 OPT_BIT_F(0, "pickaxe-all", &options->pickaxe_opts,
5512                           N_("show all changes in the changeset with -S or -G"),
5513                           DIFF_PICKAXE_ALL, PARSE_OPT_NONEG),
5514                 OPT_BIT_F(0, "pickaxe-regex", &options->pickaxe_opts,
5515                           N_("treat <string> in -S as extended POSIX regular expression"),
5516                           DIFF_PICKAXE_REGEX, PARSE_OPT_NONEG),
5517                 OPT_FILENAME('O', NULL, &options->orderfile,
5518                              N_("control the order in which files appear in the output")),
5519                 OPT_CALLBACK_F(0, "find-object", options, N_("<object-id>"),
5520                                N_("look for differences that change the number of occurrences of the specified object"),
5521                                PARSE_OPT_NONEG, diff_opt_find_object),
5522                 OPT_CALLBACK_F(0, "diff-filter", options, N_("[(A|C|D|M|R|T|U|X|B)...[*]]"),
5523                                N_("select files by diff type"),
5524                                PARSE_OPT_NONEG, diff_opt_diff_filter),
5525                 { OPTION_CALLBACK, 0, "output", options, N_("<file>"),
5526                   N_("Output to a specific file"),
5527                   PARSE_OPT_NONEG, NULL, 0, diff_opt_output },
5528
5529                 OPT_END()
5530         };
5531
5532         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
5533         memcpy(options->parseopts, parseopts, sizeof(parseopts));
5534 }
5535
5536 int diff_opt_parse(struct diff_options *options,
5537                    const char **av, int ac, const char *prefix)
5538 {
5539         if (!prefix)
5540                 prefix = "";
5541
5542         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
5543                            PARSE_OPT_KEEP_DASHDASH |
5544                            PARSE_OPT_KEEP_UNKNOWN |
5545                            PARSE_OPT_NO_INTERNAL_HELP |
5546                            PARSE_OPT_ONE_SHOT |
5547                            PARSE_OPT_STOP_AT_NON_OPTION);
5548
5549         return ac;
5550 }
5551
5552 int parse_rename_score(const char **cp_p)
5553 {
5554         unsigned long num, scale;
5555         int ch, dot;
5556         const char *cp = *cp_p;
5557
5558         num = 0;
5559         scale = 1;
5560         dot = 0;
5561         for (;;) {
5562                 ch = *cp;
5563                 if ( !dot && ch == '.' ) {
5564                         scale = 1;
5565                         dot = 1;
5566                 } else if ( ch == '%' ) {
5567                         scale = dot ? scale*100 : 100;
5568                         cp++;   /* % is always at the end */
5569                         break;
5570                 } else if ( ch >= '0' && ch <= '9' ) {
5571                         if ( scale < 100000 ) {
5572                                 scale *= 10;
5573                                 num = (num*10) + (ch-'0');
5574                         }
5575                 } else {
5576                         break;
5577                 }
5578                 cp++;
5579         }
5580         *cp_p = cp;
5581
5582         /* user says num divided by scale and we say internally that
5583          * is MAX_SCORE * num / scale.
5584          */
5585         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5586 }
5587
5588 struct diff_queue_struct diff_queued_diff;
5589
5590 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5591 {
5592         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5593         queue->queue[queue->nr++] = dp;
5594 }
5595
5596 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5597                                  struct diff_filespec *one,
5598                                  struct diff_filespec *two)
5599 {
5600         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5601         dp->one = one;
5602         dp->two = two;
5603         if (queue)
5604                 diff_q(queue, dp);
5605         return dp;
5606 }
5607
5608 void diff_free_filepair(struct diff_filepair *p)
5609 {
5610         free_filespec(p->one);
5611         free_filespec(p->two);
5612         free(p);
5613 }
5614
5615 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5616 {
5617         int abblen;
5618         const char *abbrev;
5619
5620         /* Do we want all 40 hex characters? */
5621         if (len == the_hash_algo->hexsz)
5622                 return oid_to_hex(oid);
5623
5624         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5625         abbrev = diff_abbrev_oid(oid, len);
5626
5627         if (!print_sha1_ellipsis())
5628                 return abbrev;
5629
5630         abblen = strlen(abbrev);
5631
5632         /*
5633          * In well-behaved cases, where the abbreviated result is the
5634          * same as the requested length, append three dots after the
5635          * abbreviation (hence the whole logic is limited to the case
5636          * where abblen < 37); when the actual abbreviated result is a
5637          * bit longer than the requested length, we reduce the number
5638          * of dots so that they match the well-behaved ones.  However,
5639          * if the actual abbreviation is longer than the requested
5640          * length by more than three, we give up on aligning, and add
5641          * three dots anyway, to indicate that the output is not the
5642          * full object name.  Yes, this may be suboptimal, but this
5643          * appears only in "diff --raw --abbrev" output and it is not
5644          * worth the effort to change it now.  Note that this would
5645          * likely to work fine when the automatic sizing of default
5646          * abbreviation length is used--we would be fed -1 in "len" in
5647          * that case, and will end up always appending three-dots, but
5648          * the automatic sizing is supposed to give abblen that ensures
5649          * uniqueness across all objects (statistically speaking).
5650          */
5651         if (abblen < the_hash_algo->hexsz - 3) {
5652                 static char hex[GIT_MAX_HEXSZ + 1];
5653                 if (len < abblen && abblen <= len + 2)
5654                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5655                 else
5656                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5657                 return hex;
5658         }
5659
5660         return oid_to_hex(oid);
5661 }
5662
5663 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5664 {
5665         int line_termination = opt->line_termination;
5666         int inter_name_termination = line_termination ? '\t' : '\0';
5667
5668         fprintf(opt->file, "%s", diff_line_prefix(opt));
5669         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5670                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5671                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5672                 fprintf(opt->file, "%s ",
5673                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5674         }
5675         if (p->score) {
5676                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5677                         inter_name_termination);
5678         } else {
5679                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5680         }
5681
5682         if (p->status == DIFF_STATUS_COPIED ||
5683             p->status == DIFF_STATUS_RENAMED) {
5684                 const char *name_a, *name_b;
5685                 name_a = p->one->path;
5686                 name_b = p->two->path;
5687                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5688                 write_name_quoted(name_a, opt->file, inter_name_termination);
5689                 write_name_quoted(name_b, opt->file, line_termination);
5690         } else {
5691                 const char *name_a, *name_b;
5692                 name_a = p->one->mode ? p->one->path : p->two->path;
5693                 name_b = NULL;
5694                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5695                 write_name_quoted(name_a, opt->file, line_termination);
5696         }
5697 }
5698
5699 int diff_unmodified_pair(struct diff_filepair *p)
5700 {
5701         /* This function is written stricter than necessary to support
5702          * the currently implemented transformers, but the idea is to
5703          * let transformers to produce diff_filepairs any way they want,
5704          * and filter and clean them up here before producing the output.
5705          */
5706         struct diff_filespec *one = p->one, *two = p->two;
5707
5708         if (DIFF_PAIR_UNMERGED(p))
5709                 return 0; /* unmerged is interesting */
5710
5711         /* deletion, addition, mode or type change
5712          * and rename are all interesting.
5713          */
5714         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5715             DIFF_PAIR_MODE_CHANGED(p) ||
5716             strcmp(one->path, two->path))
5717                 return 0;
5718
5719         /* both are valid and point at the same path.  that is, we are
5720          * dealing with a change.
5721          */
5722         if (one->oid_valid && two->oid_valid &&
5723             oideq(&one->oid, &two->oid) &&
5724             !one->dirty_submodule && !two->dirty_submodule)
5725                 return 1; /* no change */
5726         if (!one->oid_valid && !two->oid_valid)
5727                 return 1; /* both look at the same file on the filesystem. */
5728         return 0;
5729 }
5730
5731 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5732 {
5733         if (diff_unmodified_pair(p))
5734                 return;
5735
5736         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5737             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5738                 return; /* no tree diffs in patch format */
5739
5740         run_diff(p, o);
5741 }
5742
5743 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5744                             struct diffstat_t *diffstat)
5745 {
5746         if (diff_unmodified_pair(p))
5747                 return;
5748
5749         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5750             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5751                 return; /* no useful stat for tree diffs */
5752
5753         run_diffstat(p, o, diffstat);
5754 }
5755
5756 static void diff_flush_checkdiff(struct diff_filepair *p,
5757                 struct diff_options *o)
5758 {
5759         if (diff_unmodified_pair(p))
5760                 return;
5761
5762         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5763             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5764                 return; /* nothing to check in tree diffs */
5765
5766         run_checkdiff(p, o);
5767 }
5768
5769 int diff_queue_is_empty(void)
5770 {
5771         struct diff_queue_struct *q = &diff_queued_diff;
5772         int i;
5773         for (i = 0; i < q->nr; i++)
5774                 if (!diff_unmodified_pair(q->queue[i]))
5775                         return 0;
5776         return 1;
5777 }
5778
5779 #if DIFF_DEBUG
5780 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5781 {
5782         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5783                 x, one ? one : "",
5784                 s->path,
5785                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5786                 s->mode,
5787                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5788         fprintf(stderr, "queue[%d] %s size %lu\n",
5789                 x, one ? one : "",
5790                 s->size);
5791 }
5792
5793 void diff_debug_filepair(const struct diff_filepair *p, int i)
5794 {
5795         diff_debug_filespec(p->one, i, "one");
5796         diff_debug_filespec(p->two, i, "two");
5797         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5798                 p->score, p->status ? p->status : '?',
5799                 p->one->rename_used, p->broken_pair);
5800 }
5801
5802 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5803 {
5804         int i;
5805         if (msg)
5806                 fprintf(stderr, "%s\n", msg);
5807         fprintf(stderr, "q->nr = %d\n", q->nr);
5808         for (i = 0; i < q->nr; i++) {
5809                 struct diff_filepair *p = q->queue[i];
5810                 diff_debug_filepair(p, i);
5811         }
5812 }
5813 #endif
5814
5815 static void diff_resolve_rename_copy(void)
5816 {
5817         int i;
5818         struct diff_filepair *p;
5819         struct diff_queue_struct *q = &diff_queued_diff;
5820
5821         diff_debug_queue("resolve-rename-copy", q);
5822
5823         for (i = 0; i < q->nr; i++) {
5824                 p = q->queue[i];
5825                 p->status = 0; /* undecided */
5826                 if (DIFF_PAIR_UNMERGED(p))
5827                         p->status = DIFF_STATUS_UNMERGED;
5828                 else if (!DIFF_FILE_VALID(p->one))
5829                         p->status = DIFF_STATUS_ADDED;
5830                 else if (!DIFF_FILE_VALID(p->two))
5831                         p->status = DIFF_STATUS_DELETED;
5832                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5833                         p->status = DIFF_STATUS_TYPE_CHANGED;
5834
5835                 /* from this point on, we are dealing with a pair
5836                  * whose both sides are valid and of the same type, i.e.
5837                  * either in-place edit or rename/copy edit.
5838                  */
5839                 else if (DIFF_PAIR_RENAME(p)) {
5840                         /*
5841                          * A rename might have re-connected a broken
5842                          * pair up, causing the pathnames to be the
5843                          * same again. If so, that's not a rename at
5844                          * all, just a modification..
5845                          *
5846                          * Otherwise, see if this source was used for
5847                          * multiple renames, in which case we decrement
5848                          * the count, and call it a copy.
5849                          */
5850                         if (!strcmp(p->one->path, p->two->path))
5851                                 p->status = DIFF_STATUS_MODIFIED;
5852                         else if (--p->one->rename_used > 0)
5853                                 p->status = DIFF_STATUS_COPIED;
5854                         else
5855                                 p->status = DIFF_STATUS_RENAMED;
5856                 }
5857                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5858                          p->one->mode != p->two->mode ||
5859                          p->one->dirty_submodule ||
5860                          p->two->dirty_submodule ||
5861                          is_null_oid(&p->one->oid))
5862                         p->status = DIFF_STATUS_MODIFIED;
5863                 else {
5864                         /* This is a "no-change" entry and should not
5865                          * happen anymore, but prepare for broken callers.
5866                          */
5867                         error("feeding unmodified %s to diffcore",
5868                               p->one->path);
5869                         p->status = DIFF_STATUS_UNKNOWN;
5870                 }
5871         }
5872         diff_debug_queue("resolve-rename-copy done", q);
5873 }
5874
5875 static int check_pair_status(struct diff_filepair *p)
5876 {
5877         switch (p->status) {
5878         case DIFF_STATUS_UNKNOWN:
5879                 return 0;
5880         case 0:
5881                 die("internal error in diff-resolve-rename-copy");
5882         default:
5883                 return 1;
5884         }
5885 }
5886
5887 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5888 {
5889         int fmt = opt->output_format;
5890
5891         if (fmt & DIFF_FORMAT_CHECKDIFF)
5892                 diff_flush_checkdiff(p, opt);
5893         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5894                 diff_flush_raw(p, opt);
5895         else if (fmt & DIFF_FORMAT_NAME) {
5896                 const char *name_a, *name_b;
5897                 name_a = p->two->path;
5898                 name_b = NULL;
5899                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5900                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5901                 write_name_quoted(name_a, opt->file, opt->line_termination);
5902         }
5903 }
5904
5905 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5906 {
5907         struct strbuf sb = STRBUF_INIT;
5908         if (fs->mode)
5909                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5910         else
5911                 strbuf_addf(&sb, " %s ", newdelete);
5912
5913         quote_c_style(fs->path, &sb, NULL, 0);
5914         strbuf_addch(&sb, '\n');
5915         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5916                          sb.buf, sb.len, 0);
5917         strbuf_release(&sb);
5918 }
5919
5920 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5921                 int show_name)
5922 {
5923         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5924                 struct strbuf sb = STRBUF_INIT;
5925                 strbuf_addf(&sb, " mode change %06o => %06o",
5926                             p->one->mode, p->two->mode);
5927                 if (show_name) {
5928                         strbuf_addch(&sb, ' ');
5929                         quote_c_style(p->two->path, &sb, NULL, 0);
5930                 }
5931                 strbuf_addch(&sb, '\n');
5932                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5933                                  sb.buf, sb.len, 0);
5934                 strbuf_release(&sb);
5935         }
5936 }
5937
5938 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5939                 struct diff_filepair *p)
5940 {
5941         struct strbuf sb = STRBUF_INIT;
5942         struct strbuf names = STRBUF_INIT;
5943
5944         pprint_rename(&names, p->one->path, p->two->path);
5945         strbuf_addf(&sb, " %s %s (%d%%)\n",
5946                     renamecopy, names.buf, similarity_index(p));
5947         strbuf_release(&names);
5948         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5949                                  sb.buf, sb.len, 0);
5950         show_mode_change(opt, p, 0);
5951         strbuf_release(&sb);
5952 }
5953
5954 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5955 {
5956         switch(p->status) {
5957         case DIFF_STATUS_DELETED:
5958                 show_file_mode_name(opt, "delete", p->one);
5959                 break;
5960         case DIFF_STATUS_ADDED:
5961                 show_file_mode_name(opt, "create", p->two);
5962                 break;
5963         case DIFF_STATUS_COPIED:
5964                 show_rename_copy(opt, "copy", p);
5965                 break;
5966         case DIFF_STATUS_RENAMED:
5967                 show_rename_copy(opt, "rename", p);
5968                 break;
5969         default:
5970                 if (p->score) {
5971                         struct strbuf sb = STRBUF_INIT;
5972                         strbuf_addstr(&sb, " rewrite ");
5973                         quote_c_style(p->two->path, &sb, NULL, 0);
5974                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5975                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5976                                          sb.buf, sb.len, 0);
5977                         strbuf_release(&sb);
5978                 }
5979                 show_mode_change(opt, p, !p->score);
5980                 break;
5981         }
5982 }
5983
5984 struct patch_id_t {
5985         git_SHA_CTX *ctx;
5986         int patchlen;
5987 };
5988
5989 static int remove_space(char *line, int len)
5990 {
5991         int i;
5992         char *dst = line;
5993         unsigned char c;
5994
5995         for (i = 0; i < len; i++)
5996                 if (!isspace((c = line[i])))
5997                         *dst++ = c;
5998
5999         return dst - line;
6000 }
6001
6002 void flush_one_hunk(struct object_id *result, git_SHA_CTX *ctx)
6003 {
6004         unsigned char hash[GIT_MAX_RAWSZ];
6005         unsigned short carry = 0;
6006         int i;
6007
6008         git_SHA1_Final(hash, ctx);
6009         git_SHA1_Init(ctx);
6010         /* 20-byte sum, with carry */
6011         for (i = 0; i < GIT_SHA1_RAWSZ; ++i) {
6012                 carry += result->hash[i] + hash[i];
6013                 result->hash[i] = carry;
6014                 carry >>= 8;
6015         }
6016 }
6017
6018 static void patch_id_consume(void *priv, char *line, unsigned long len)
6019 {
6020         struct patch_id_t *data = priv;
6021         int new_len;
6022
6023         new_len = remove_space(line, len);
6024
6025         git_SHA1_Update(data->ctx, line, new_len);
6026         data->patchlen += new_len;
6027 }
6028
6029 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
6030 {
6031         git_SHA1_Update(ctx, str, strlen(str));
6032 }
6033
6034 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
6035 {
6036         /* large enough for 2^32 in octal */
6037         char buf[12];
6038         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
6039         git_SHA1_Update(ctx, buf, len);
6040 }
6041
6042 /* returns 0 upon success, and writes result into oid */
6043 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6044 {
6045         struct diff_queue_struct *q = &diff_queued_diff;
6046         int i;
6047         git_SHA_CTX ctx;
6048         struct patch_id_t data;
6049
6050         git_SHA1_Init(&ctx);
6051         memset(&data, 0, sizeof(struct patch_id_t));
6052         data.ctx = &ctx;
6053         oidclr(oid);
6054
6055         for (i = 0; i < q->nr; i++) {
6056                 xpparam_t xpp;
6057                 xdemitconf_t xecfg;
6058                 mmfile_t mf1, mf2;
6059                 struct diff_filepair *p = q->queue[i];
6060                 int len1, len2;
6061
6062                 memset(&xpp, 0, sizeof(xpp));
6063                 memset(&xecfg, 0, sizeof(xecfg));
6064                 if (p->status == 0)
6065                         return error("internal diff status error");
6066                 if (p->status == DIFF_STATUS_UNKNOWN)
6067                         continue;
6068                 if (diff_unmodified_pair(p))
6069                         continue;
6070                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
6071                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
6072                         continue;
6073                 if (DIFF_PAIR_UNMERGED(p))
6074                         continue;
6075
6076                 diff_fill_oid_info(p->one, options->repo->index);
6077                 diff_fill_oid_info(p->two, options->repo->index);
6078
6079                 len1 = remove_space(p->one->path, strlen(p->one->path));
6080                 len2 = remove_space(p->two->path, strlen(p->two->path));
6081                 patch_id_add_string(&ctx, "diff--git");
6082                 patch_id_add_string(&ctx, "a/");
6083                 git_SHA1_Update(&ctx, p->one->path, len1);
6084                 patch_id_add_string(&ctx, "b/");
6085                 git_SHA1_Update(&ctx, p->two->path, len2);
6086
6087                 if (p->one->mode == 0) {
6088                         patch_id_add_string(&ctx, "newfilemode");
6089                         patch_id_add_mode(&ctx, p->two->mode);
6090                         patch_id_add_string(&ctx, "---/dev/null");
6091                         patch_id_add_string(&ctx, "+++b/");
6092                         git_SHA1_Update(&ctx, p->two->path, len2);
6093                 } else if (p->two->mode == 0) {
6094                         patch_id_add_string(&ctx, "deletedfilemode");
6095                         patch_id_add_mode(&ctx, p->one->mode);
6096                         patch_id_add_string(&ctx, "---a/");
6097                         git_SHA1_Update(&ctx, p->one->path, len1);
6098                         patch_id_add_string(&ctx, "+++/dev/null");
6099                 } else {
6100                         patch_id_add_string(&ctx, "---a/");
6101                         git_SHA1_Update(&ctx, p->one->path, len1);
6102                         patch_id_add_string(&ctx, "+++b/");
6103                         git_SHA1_Update(&ctx, p->two->path, len2);
6104                 }
6105
6106                 if (diff_header_only)
6107                         continue;
6108
6109                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
6110                     fill_mmfile(options->repo, &mf2, p->two) < 0)
6111                         return error("unable to read files to diff");
6112
6113                 if (diff_filespec_is_binary(options->repo, p->one) ||
6114                     diff_filespec_is_binary(options->repo, p->two)) {
6115                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
6116                                         GIT_SHA1_HEXSZ);
6117                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
6118                                         GIT_SHA1_HEXSZ);
6119                         continue;
6120                 }
6121
6122                 xpp.flags = 0;
6123                 xecfg.ctxlen = 3;
6124                 xecfg.flags = 0;
6125                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
6126                                   patch_id_consume, &data, &xpp, &xecfg))
6127                         return error("unable to generate patch-id diff for %s",
6128                                      p->one->path);
6129
6130                 if (stable)
6131                         flush_one_hunk(oid, &ctx);
6132         }
6133
6134         if (!stable)
6135                 git_SHA1_Final(oid->hash, &ctx);
6136
6137         return 0;
6138 }
6139
6140 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6141 {
6142         struct diff_queue_struct *q = &diff_queued_diff;
6143         int i;
6144         int result = diff_get_patch_id(options, oid, diff_header_only, stable);
6145
6146         for (i = 0; i < q->nr; i++)
6147                 diff_free_filepair(q->queue[i]);
6148
6149         free(q->queue);
6150         DIFF_QUEUE_CLEAR(q);
6151
6152         return result;
6153 }
6154
6155 static int is_summary_empty(const struct diff_queue_struct *q)
6156 {
6157         int i;
6158
6159         for (i = 0; i < q->nr; i++) {
6160                 const struct diff_filepair *p = q->queue[i];
6161
6162                 switch (p->status) {
6163                 case DIFF_STATUS_DELETED:
6164                 case DIFF_STATUS_ADDED:
6165                 case DIFF_STATUS_COPIED:
6166                 case DIFF_STATUS_RENAMED:
6167                         return 0;
6168                 default:
6169                         if (p->score)
6170                                 return 0;
6171                         if (p->one->mode && p->two->mode &&
6172                             p->one->mode != p->two->mode)
6173                                 return 0;
6174                         break;
6175                 }
6176         }
6177         return 1;
6178 }
6179
6180 static const char rename_limit_warning[] =
6181 N_("inexact rename detection was skipped due to too many files.");
6182
6183 static const char degrade_cc_to_c_warning[] =
6184 N_("only found copies from modified paths due to too many files.");
6185
6186 static const char rename_limit_advice[] =
6187 N_("you may want to set your %s variable to at least "
6188    "%d and retry the command.");
6189
6190 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
6191 {
6192         fflush(stdout);
6193         if (degraded_cc)
6194                 warning(_(degrade_cc_to_c_warning));
6195         else if (needed)
6196                 warning(_(rename_limit_warning));
6197         else
6198                 return;
6199         if (0 < needed)
6200                 warning(_(rename_limit_advice), varname, needed);
6201 }
6202
6203 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
6204 {
6205         int i;
6206         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
6207         struct diff_queue_struct *q = &diff_queued_diff;
6208
6209         if (WSEH_NEW & WS_RULE_MASK)
6210                 BUG("WS rules bit mask overlaps with diff symbol flags");
6211
6212         if (o->color_moved)
6213                 o->emitted_symbols = &esm;
6214
6215         for (i = 0; i < q->nr; i++) {
6216                 struct diff_filepair *p = q->queue[i];
6217                 if (check_pair_status(p))
6218                         diff_flush_patch(p, o);
6219         }
6220
6221         if (o->emitted_symbols) {
6222                 if (o->color_moved) {
6223                         struct hashmap add_lines, del_lines;
6224
6225                         if (o->color_moved_ws_handling &
6226                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
6227                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
6228
6229                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
6230                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
6231
6232                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6233                         mark_color_as_moved(o, &add_lines, &del_lines);
6234                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6235                                 dim_moved_lines(o);
6236
6237                         hashmap_free(&add_lines, 1);
6238                         hashmap_free(&del_lines, 1);
6239                 }
6240
6241                 for (i = 0; i < esm.nr; i++)
6242                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6243
6244                 for (i = 0; i < esm.nr; i++)
6245                         free((void *)esm.buf[i].line);
6246                 esm.nr = 0;
6247
6248                 o->emitted_symbols = NULL;
6249         }
6250 }
6251
6252 void diff_flush(struct diff_options *options)
6253 {
6254         struct diff_queue_struct *q = &diff_queued_diff;
6255         int i, output_format = options->output_format;
6256         int separator = 0;
6257         int dirstat_by_line = 0;
6258
6259         /*
6260          * Order: raw, stat, summary, patch
6261          * or:    name/name-status/checkdiff (other bits clear)
6262          */
6263         if (!q->nr)
6264                 goto free_queue;
6265
6266         if (output_format & (DIFF_FORMAT_RAW |
6267                              DIFF_FORMAT_NAME |
6268                              DIFF_FORMAT_NAME_STATUS |
6269                              DIFF_FORMAT_CHECKDIFF)) {
6270                 for (i = 0; i < q->nr; i++) {
6271                         struct diff_filepair *p = q->queue[i];
6272                         if (check_pair_status(p))
6273                                 flush_one_pair(p, options);
6274                 }
6275                 separator++;
6276         }
6277
6278         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6279                 dirstat_by_line = 1;
6280
6281         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6282             dirstat_by_line) {
6283                 struct diffstat_t diffstat;
6284
6285                 memset(&diffstat, 0, sizeof(struct diffstat_t));
6286                 for (i = 0; i < q->nr; i++) {
6287                         struct diff_filepair *p = q->queue[i];
6288                         if (check_pair_status(p))
6289                                 diff_flush_stat(p, options, &diffstat);
6290                 }
6291                 if (output_format & DIFF_FORMAT_NUMSTAT)
6292                         show_numstat(&diffstat, options);
6293                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6294                         show_stats(&diffstat, options);
6295                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6296                         show_shortstats(&diffstat, options);
6297                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6298                         show_dirstat_by_line(&diffstat, options);
6299                 free_diffstat_info(&diffstat);
6300                 separator++;
6301         }
6302         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6303                 show_dirstat(options);
6304
6305         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6306                 for (i = 0; i < q->nr; i++) {
6307                         diff_summary(options, q->queue[i]);
6308                 }
6309                 separator++;
6310         }
6311
6312         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6313             options->flags.exit_with_status &&
6314             options->flags.diff_from_contents) {
6315                 /*
6316                  * run diff_flush_patch for the exit status. setting
6317                  * options->file to /dev/null should be safe, because we
6318                  * aren't supposed to produce any output anyway.
6319                  */
6320                 if (options->close_file)
6321                         fclose(options->file);
6322                 options->file = xfopen("/dev/null", "w");
6323                 options->close_file = 1;
6324                 options->color_moved = 0;
6325                 for (i = 0; i < q->nr; i++) {
6326                         struct diff_filepair *p = q->queue[i];
6327                         if (check_pair_status(p))
6328                                 diff_flush_patch(p, options);
6329                         if (options->found_changes)
6330                                 break;
6331                 }
6332         }
6333
6334         if (output_format & DIFF_FORMAT_PATCH) {
6335                 if (separator) {
6336                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6337                         if (options->stat_sep)
6338                                 /* attach patch instead of inline */
6339                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6340                                                  NULL, 0, 0);
6341                 }
6342
6343                 diff_flush_patch_all_file_pairs(options);
6344         }
6345
6346         if (output_format & DIFF_FORMAT_CALLBACK)
6347                 options->format_callback(q, options, options->format_callback_data);
6348
6349         for (i = 0; i < q->nr; i++)
6350                 diff_free_filepair(q->queue[i]);
6351 free_queue:
6352         free(q->queue);
6353         DIFF_QUEUE_CLEAR(q);
6354         if (options->close_file)
6355                 fclose(options->file);
6356
6357         /*
6358          * Report the content-level differences with HAS_CHANGES;
6359          * diff_addremove/diff_change does not set the bit when
6360          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6361          */
6362         if (options->flags.diff_from_contents) {
6363                 if (options->found_changes)
6364                         options->flags.has_changes = 1;
6365                 else
6366                         options->flags.has_changes = 0;
6367         }
6368 }
6369
6370 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6371 {
6372         return (((p->status == DIFF_STATUS_MODIFIED) &&
6373                  ((p->score &&
6374                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6375                   (!p->score &&
6376                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6377                 ((p->status != DIFF_STATUS_MODIFIED) &&
6378                  filter_bit_tst(p->status, options)));
6379 }
6380
6381 static void diffcore_apply_filter(struct diff_options *options)
6382 {
6383         int i;
6384         struct diff_queue_struct *q = &diff_queued_diff;
6385         struct diff_queue_struct outq;
6386
6387         DIFF_QUEUE_CLEAR(&outq);
6388
6389         if (!options->filter)
6390                 return;
6391
6392         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6393                 int found;
6394                 for (i = found = 0; !found && i < q->nr; i++) {
6395                         if (match_filter(options, q->queue[i]))
6396                                 found++;
6397                 }
6398                 if (found)
6399                         return;
6400
6401                 /* otherwise we will clear the whole queue
6402                  * by copying the empty outq at the end of this
6403                  * function, but first clear the current entries
6404                  * in the queue.
6405                  */
6406                 for (i = 0; i < q->nr; i++)
6407                         diff_free_filepair(q->queue[i]);
6408         }
6409         else {
6410                 /* Only the matching ones */
6411                 for (i = 0; i < q->nr; i++) {
6412                         struct diff_filepair *p = q->queue[i];
6413                         if (match_filter(options, p))
6414                                 diff_q(&outq, p);
6415                         else
6416                                 diff_free_filepair(p);
6417                 }
6418         }
6419         free(q->queue);
6420         *q = outq;
6421 }
6422
6423 /* Check whether two filespecs with the same mode and size are identical */
6424 static int diff_filespec_is_identical(struct repository *r,
6425                                       struct diff_filespec *one,
6426                                       struct diff_filespec *two)
6427 {
6428         if (S_ISGITLINK(one->mode))
6429                 return 0;
6430         if (diff_populate_filespec(r, one, 0))
6431                 return 0;
6432         if (diff_populate_filespec(r, two, 0))
6433                 return 0;
6434         return !memcmp(one->data, two->data, one->size);
6435 }
6436
6437 static int diff_filespec_check_stat_unmatch(struct repository *r,
6438                                             struct diff_filepair *p)
6439 {
6440         if (p->done_skip_stat_unmatch)
6441                 return p->skip_stat_unmatch_result;
6442
6443         p->done_skip_stat_unmatch = 1;
6444         p->skip_stat_unmatch_result = 0;
6445         /*
6446          * 1. Entries that come from stat info dirtiness
6447          *    always have both sides (iow, not create/delete),
6448          *    one side of the object name is unknown, with
6449          *    the same mode and size.  Keep the ones that
6450          *    do not match these criteria.  They have real
6451          *    differences.
6452          *
6453          * 2. At this point, the file is known to be modified,
6454          *    with the same mode and size, and the object
6455          *    name of one side is unknown.  Need to inspect
6456          *    the identical contents.
6457          */
6458         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6459             !DIFF_FILE_VALID(p->two) ||
6460             (p->one->oid_valid && p->two->oid_valid) ||
6461             (p->one->mode != p->two->mode) ||
6462             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6463             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6464             (p->one->size != p->two->size) ||
6465             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6466                 p->skip_stat_unmatch_result = 1;
6467         return p->skip_stat_unmatch_result;
6468 }
6469
6470 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6471 {
6472         int i;
6473         struct diff_queue_struct *q = &diff_queued_diff;
6474         struct diff_queue_struct outq;
6475         DIFF_QUEUE_CLEAR(&outq);
6476
6477         for (i = 0; i < q->nr; i++) {
6478                 struct diff_filepair *p = q->queue[i];
6479
6480                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6481                         diff_q(&outq, p);
6482                 else {
6483                         /*
6484                          * The caller can subtract 1 from skip_stat_unmatch
6485                          * to determine how many paths were dirty only
6486                          * due to stat info mismatch.
6487                          */
6488                         if (!diffopt->flags.no_index)
6489                                 diffopt->skip_stat_unmatch++;
6490                         diff_free_filepair(p);
6491                 }
6492         }
6493         free(q->queue);
6494         *q = outq;
6495 }
6496
6497 static int diffnamecmp(const void *a_, const void *b_)
6498 {
6499         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6500         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6501         const char *name_a, *name_b;
6502
6503         name_a = a->one ? a->one->path : a->two->path;
6504         name_b = b->one ? b->one->path : b->two->path;
6505         return strcmp(name_a, name_b);
6506 }
6507
6508 void diffcore_fix_diff_index(void)
6509 {
6510         struct diff_queue_struct *q = &diff_queued_diff;
6511         QSORT(q->queue, q->nr, diffnamecmp);
6512 }
6513
6514 static void add_if_missing(struct repository *r,
6515                            struct oid_array *to_fetch,
6516                            const struct diff_filespec *filespec)
6517 {
6518         if (filespec && filespec->oid_valid &&
6519             oid_object_info_extended(r, &filespec->oid, NULL,
6520                                      OBJECT_INFO_FOR_PREFETCH))
6521                 oid_array_append(to_fetch, &filespec->oid);
6522 }
6523
6524 void diffcore_std(struct diff_options *options)
6525 {
6526         if (options->repo == the_repository &&
6527             repository_format_partial_clone) {
6528                 /*
6529                  * Prefetch the diff pairs that are about to be flushed.
6530                  */
6531                 int i;
6532                 struct diff_queue_struct *q = &diff_queued_diff;
6533                 struct oid_array to_fetch = OID_ARRAY_INIT;
6534
6535                 for (i = 0; i < q->nr; i++) {
6536                         struct diff_filepair *p = q->queue[i];
6537                         add_if_missing(options->repo, &to_fetch, p->one);
6538                         add_if_missing(options->repo, &to_fetch, p->two);
6539                 }
6540                 if (to_fetch.nr)
6541                         /*
6542                          * NEEDSWORK: Consider deduplicating the OIDs sent.
6543                          */
6544                         fetch_objects(repository_format_partial_clone,
6545                                       to_fetch.oid, to_fetch.nr);
6546                 oid_array_clear(&to_fetch);
6547         }
6548
6549         /* NOTE please keep the following in sync with diff_tree_combined() */
6550         if (options->skip_stat_unmatch)
6551                 diffcore_skip_stat_unmatch(options);
6552         if (!options->found_follow) {
6553                 /* See try_to_follow_renames() in tree-diff.c */
6554                 if (options->break_opt != -1)
6555                         diffcore_break(options->repo,
6556                                        options->break_opt);
6557                 if (options->detect_rename)
6558                         diffcore_rename(options);
6559                 if (options->break_opt != -1)
6560                         diffcore_merge_broken();
6561         }
6562         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6563                 diffcore_pickaxe(options);
6564         if (options->orderfile)
6565                 diffcore_order(options->orderfile);
6566         if (!options->found_follow)
6567                 /* See try_to_follow_renames() in tree-diff.c */
6568                 diff_resolve_rename_copy();
6569         diffcore_apply_filter(options);
6570
6571         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6572                 options->flags.has_changes = 1;
6573         else
6574                 options->flags.has_changes = 0;
6575
6576         options->found_follow = 0;
6577 }
6578
6579 int diff_result_code(struct diff_options *opt, int status)
6580 {
6581         int result = 0;
6582
6583         diff_warn_rename_limit("diff.renameLimit",
6584                                opt->needed_rename_limit,
6585                                opt->degraded_cc_to_c);
6586         if (!opt->flags.exit_with_status &&
6587             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6588                 return status;
6589         if (opt->flags.exit_with_status &&
6590             opt->flags.has_changes)
6591                 result |= 01;
6592         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6593             opt->flags.check_failed)
6594                 result |= 02;
6595         return result;
6596 }
6597
6598 int diff_can_quit_early(struct diff_options *opt)
6599 {
6600         return (opt->flags.quick &&
6601                 !opt->filter &&
6602                 opt->flags.has_changes);
6603 }
6604
6605 /*
6606  * Shall changes to this submodule be ignored?
6607  *
6608  * Submodule changes can be configured to be ignored separately for each path,
6609  * but that configuration can be overridden from the command line.
6610  */
6611 static int is_submodule_ignored(const char *path, struct diff_options *options)
6612 {
6613         int ignored = 0;
6614         struct diff_flags orig_flags = options->flags;
6615         if (!options->flags.override_submodule_config)
6616                 set_diffopt_flags_from_submodule_config(options, path);
6617         if (options->flags.ignore_submodules)
6618                 ignored = 1;
6619         options->flags = orig_flags;
6620         return ignored;
6621 }
6622
6623 void diff_addremove(struct diff_options *options,
6624                     int addremove, unsigned mode,
6625                     const struct object_id *oid,
6626                     int oid_valid,
6627                     const char *concatpath, unsigned dirty_submodule)
6628 {
6629         struct diff_filespec *one, *two;
6630
6631         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6632                 return;
6633
6634         /* This may look odd, but it is a preparation for
6635          * feeding "there are unchanged files which should
6636          * not produce diffs, but when you are doing copy
6637          * detection you would need them, so here they are"
6638          * entries to the diff-core.  They will be prefixed
6639          * with something like '=' or '*' (I haven't decided
6640          * which but should not make any difference).
6641          * Feeding the same new and old to diff_change()
6642          * also has the same effect.
6643          * Before the final output happens, they are pruned after
6644          * merged into rename/copy pairs as appropriate.
6645          */
6646         if (options->flags.reverse_diff)
6647                 addremove = (addremove == '+' ? '-' :
6648                              addremove == '-' ? '+' : addremove);
6649
6650         if (options->prefix &&
6651             strncmp(concatpath, options->prefix, options->prefix_length))
6652                 return;
6653
6654         one = alloc_filespec(concatpath);
6655         two = alloc_filespec(concatpath);
6656
6657         if (addremove != '+')
6658                 fill_filespec(one, oid, oid_valid, mode);
6659         if (addremove != '-') {
6660                 fill_filespec(two, oid, oid_valid, mode);
6661                 two->dirty_submodule = dirty_submodule;
6662         }
6663
6664         diff_queue(&diff_queued_diff, one, two);
6665         if (!options->flags.diff_from_contents)
6666                 options->flags.has_changes = 1;
6667 }
6668
6669 void diff_change(struct diff_options *options,
6670                  unsigned old_mode, unsigned new_mode,
6671                  const struct object_id *old_oid,
6672                  const struct object_id *new_oid,
6673                  int old_oid_valid, int new_oid_valid,
6674                  const char *concatpath,
6675                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6676 {
6677         struct diff_filespec *one, *two;
6678         struct diff_filepair *p;
6679
6680         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6681             is_submodule_ignored(concatpath, options))
6682                 return;
6683
6684         if (options->flags.reverse_diff) {
6685                 SWAP(old_mode, new_mode);
6686                 SWAP(old_oid, new_oid);
6687                 SWAP(old_oid_valid, new_oid_valid);
6688                 SWAP(old_dirty_submodule, new_dirty_submodule);
6689         }
6690
6691         if (options->prefix &&
6692             strncmp(concatpath, options->prefix, options->prefix_length))
6693                 return;
6694
6695         one = alloc_filespec(concatpath);
6696         two = alloc_filespec(concatpath);
6697         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6698         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6699         one->dirty_submodule = old_dirty_submodule;
6700         two->dirty_submodule = new_dirty_submodule;
6701         p = diff_queue(&diff_queued_diff, one, two);
6702
6703         if (options->flags.diff_from_contents)
6704                 return;
6705
6706         if (options->flags.quick && options->skip_stat_unmatch &&
6707             !diff_filespec_check_stat_unmatch(options->repo, p))
6708                 return;
6709
6710         options->flags.has_changes = 1;
6711 }
6712
6713 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6714 {
6715         struct diff_filepair *pair;
6716         struct diff_filespec *one, *two;
6717
6718         if (options->prefix &&
6719             strncmp(path, options->prefix, options->prefix_length))
6720                 return NULL;
6721
6722         one = alloc_filespec(path);
6723         two = alloc_filespec(path);
6724         pair = diff_queue(&diff_queued_diff, one, two);
6725         pair->is_unmerged = 1;
6726         return pair;
6727 }
6728
6729 static char *run_textconv(struct repository *r,
6730                           const char *pgm,
6731                           struct diff_filespec *spec,
6732                           size_t *outsize)
6733 {
6734         struct diff_tempfile *temp;
6735         const char *argv[3];
6736         const char **arg = argv;
6737         struct child_process child = CHILD_PROCESS_INIT;
6738         struct strbuf buf = STRBUF_INIT;
6739         int err = 0;
6740
6741         temp = prepare_temp_file(r, spec->path, spec);
6742         *arg++ = pgm;
6743         *arg++ = temp->name;
6744         *arg = NULL;
6745
6746         child.use_shell = 1;
6747         child.argv = argv;
6748         child.out = -1;
6749         if (start_command(&child)) {
6750                 remove_tempfile();
6751                 return NULL;
6752         }
6753
6754         if (strbuf_read(&buf, child.out, 0) < 0)
6755                 err = error("error reading from textconv command '%s'", pgm);
6756         close(child.out);
6757
6758         if (finish_command(&child) || err) {
6759                 strbuf_release(&buf);
6760                 remove_tempfile();
6761                 return NULL;
6762         }
6763         remove_tempfile();
6764
6765         return strbuf_detach(&buf, outsize);
6766 }
6767
6768 size_t fill_textconv(struct repository *r,
6769                      struct userdiff_driver *driver,
6770                      struct diff_filespec *df,
6771                      char **outbuf)
6772 {
6773         size_t size;
6774
6775         if (!driver) {
6776                 if (!DIFF_FILE_VALID(df)) {
6777                         *outbuf = "";
6778                         return 0;
6779                 }
6780                 if (diff_populate_filespec(r, df, 0))
6781                         die("unable to read files to diff");
6782                 *outbuf = df->data;
6783                 return df->size;
6784         }
6785
6786         if (!driver->textconv)
6787                 BUG("fill_textconv called with non-textconv driver");
6788
6789         if (driver->textconv_cache && df->oid_valid) {
6790                 *outbuf = notes_cache_get(driver->textconv_cache,
6791                                           &df->oid,
6792                                           &size);
6793                 if (*outbuf)
6794                         return size;
6795         }
6796
6797         *outbuf = run_textconv(r, driver->textconv, df, &size);
6798         if (!*outbuf)
6799                 die("unable to read files to diff");
6800
6801         if (driver->textconv_cache && df->oid_valid) {
6802                 /* ignore errors, as we might be in a readonly repository */
6803                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6804                                 size);
6805                 /*
6806                  * we could save up changes and flush them all at the end,
6807                  * but we would need an extra call after all diffing is done.
6808                  * Since generating a cache entry is the slow path anyway,
6809                  * this extra overhead probably isn't a big deal.
6810                  */
6811                 notes_cache_write(driver->textconv_cache);
6812         }
6813
6814         return size;
6815 }
6816
6817 int textconv_object(struct repository *r,
6818                     const char *path,
6819                     unsigned mode,
6820                     const struct object_id *oid,
6821                     int oid_valid,
6822                     char **buf,
6823                     unsigned long *buf_size)
6824 {
6825         struct diff_filespec *df;
6826         struct userdiff_driver *textconv;
6827
6828         df = alloc_filespec(path);
6829         fill_filespec(df, oid, oid_valid, mode);
6830         textconv = get_textconv(r, df);
6831         if (!textconv) {
6832                 free_filespec(df);
6833                 return 0;
6834         }
6835
6836         *buf_size = fill_textconv(r, textconv, df, buf);
6837         free_filespec(df);
6838         return 1;
6839 }
6840
6841 void setup_diff_pager(struct diff_options *opt)
6842 {
6843         /*
6844          * If the user asked for our exit code, then either they want --quiet
6845          * or --exit-code. We should definitely not bother with a pager in the
6846          * former case, as we will generate no output. Since we still properly
6847          * report our exit code even when a pager is run, we _could_ run a
6848          * pager with --exit-code. But since we have not done so historically,
6849          * and because it is easy to find people oneline advising "git diff
6850          * --exit-code" in hooks and other scripts, we do not do so.
6851          */
6852         if (!opt->flags.exit_with_status &&
6853             check_pager_config("diff") != 0)
6854                 setup_pager();
6855 }