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