diff.con commit Merge branch 'jk/lib-terminal-lazy' (66d9133)
   1/*
   2 * Copyright (C) 2005 Junio C Hamano
   3 */
   4#include "cache.h"
   5#include "quote.h"
   6#include "diff.h"
   7#include "diffcore.h"
   8#include "delta.h"
   9#include "xdiff-interface.h"
  10#include "color.h"
  11#include "attr.h"
  12#include "run-command.h"
  13#include "utf8.h"
  14#include "userdiff.h"
  15#include "sigchain.h"
  16#include "submodule.h"
  17#include "ll-merge.h"
  18#include "string-list.h"
  19
  20#ifdef NO_FAST_WORKING_DIRECTORY
  21#define FAST_WORKING_DIRECTORY 0
  22#else
  23#define FAST_WORKING_DIRECTORY 1
  24#endif
  25
  26static int diff_detect_rename_default;
  27static int diff_rename_limit_default = 400;
  28static int diff_suppress_blank_empty;
  29static int diff_use_color_default = -1;
  30static int diff_context_default = 3;
  31static const char *diff_word_regex_cfg;
  32static const char *external_diff_cmd_cfg;
  33static const char *diff_order_file_cfg;
  34int diff_auto_refresh_index = 1;
  35static int diff_mnemonic_prefix;
  36static int diff_no_prefix;
  37static int diff_stat_graph_width;
  38static int diff_dirstat_permille_default = 30;
  39static struct diff_options default_diff_options;
  40static long diff_algorithm;
  41
  42static char diff_colors[][COLOR_MAXLEN] = {
  43        GIT_COLOR_RESET,
  44        GIT_COLOR_NORMAL,       /* PLAIN */
  45        GIT_COLOR_BOLD,         /* METAINFO */
  46        GIT_COLOR_CYAN,         /* FRAGINFO */
  47        GIT_COLOR_RED,          /* OLD */
  48        GIT_COLOR_GREEN,        /* NEW */
  49        GIT_COLOR_YELLOW,       /* COMMIT */
  50        GIT_COLOR_BG_RED,       /* WHITESPACE */
  51        GIT_COLOR_NORMAL,       /* FUNCINFO */
  52};
  53
  54static int parse_diff_color_slot(const char *var, int ofs)
  55{
  56        if (!strcasecmp(var+ofs, "plain"))
  57                return DIFF_PLAIN;
  58        if (!strcasecmp(var+ofs, "meta"))
  59                return DIFF_METAINFO;
  60        if (!strcasecmp(var+ofs, "frag"))
  61                return DIFF_FRAGINFO;
  62        if (!strcasecmp(var+ofs, "old"))
  63                return DIFF_FILE_OLD;
  64        if (!strcasecmp(var+ofs, "new"))
  65                return DIFF_FILE_NEW;
  66        if (!strcasecmp(var+ofs, "commit"))
  67                return DIFF_COMMIT;
  68        if (!strcasecmp(var+ofs, "whitespace"))
  69                return DIFF_WHITESPACE;
  70        if (!strcasecmp(var+ofs, "func"))
  71                return DIFF_FUNCINFO;
  72        return -1;
  73}
  74
  75static int parse_dirstat_params(struct diff_options *options, const char *params_string,
  76                                struct strbuf *errmsg)
  77{
  78        char *params_copy = xstrdup(params_string);
  79        struct string_list params = STRING_LIST_INIT_NODUP;
  80        int ret = 0;
  81        int i;
  82
  83        if (*params_copy)
  84                string_list_split_in_place(&params, params_copy, ',', -1);
  85        for (i = 0; i < params.nr; i++) {
  86                const char *p = params.items[i].string;
  87                if (!strcmp(p, "changes")) {
  88                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  89                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  90                } else if (!strcmp(p, "lines")) {
  91                        DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  92                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  93                } else if (!strcmp(p, "files")) {
  94                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  95                        DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  96                } else if (!strcmp(p, "noncumulative")) {
  97                        DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
  98                } else if (!strcmp(p, "cumulative")) {
  99                        DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 100                } else if (isdigit(*p)) {
 101                        char *end;
 102                        int permille = strtoul(p, &end, 10) * 10;
 103                        if (*end == '.' && isdigit(*++end)) {
 104                                /* only use first digit */
 105                                permille += *end - '0';
 106                                /* .. and ignore any further digits */
 107                                while (isdigit(*++end))
 108                                        ; /* nothing */
 109                        }
 110                        if (!*end)
 111                                options->dirstat_permille = permille;
 112                        else {
 113                                strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 114                                            p);
 115                                ret++;
 116                        }
 117                } else {
 118                        strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 119                        ret++;
 120                }
 121
 122        }
 123        string_list_clear(&params, 0);
 124        free(params_copy);
 125        return ret;
 126}
 127
 128static int parse_submodule_params(struct diff_options *options, const char *value)
 129{
 130        if (!strcmp(value, "log"))
 131                DIFF_OPT_SET(options, SUBMODULE_LOG);
 132        else if (!strcmp(value, "short"))
 133                DIFF_OPT_CLR(options, SUBMODULE_LOG);
 134        else
 135                return -1;
 136        return 0;
 137}
 138
 139static int git_config_rename(const char *var, const char *value)
 140{
 141        if (!value)
 142                return DIFF_DETECT_RENAME;
 143        if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 144                return  DIFF_DETECT_COPY;
 145        return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 146}
 147
 148long parse_algorithm_value(const char *value)
 149{
 150        if (!value)
 151                return -1;
 152        else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 153                return 0;
 154        else if (!strcasecmp(value, "minimal"))
 155                return XDF_NEED_MINIMAL;
 156        else if (!strcasecmp(value, "patience"))
 157                return XDF_PATIENCE_DIFF;
 158        else if (!strcasecmp(value, "histogram"))
 159                return XDF_HISTOGRAM_DIFF;
 160        return -1;
 161}
 162
 163/*
 164 * These are to give UI layer defaults.
 165 * The core-level commands such as git-diff-files should
 166 * never be affected by the setting of diff.renames
 167 * the user happens to have in the configuration file.
 168 */
 169int git_diff_ui_config(const char *var, const char *value, void *cb)
 170{
 171        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 172                diff_use_color_default = git_config_colorbool(var, value);
 173                return 0;
 174        }
 175        if (!strcmp(var, "diff.context")) {
 176                diff_context_default = git_config_int(var, value);
 177                if (diff_context_default < 0)
 178                        return -1;
 179                return 0;
 180        }
 181        if (!strcmp(var, "diff.renames")) {
 182                diff_detect_rename_default = git_config_rename(var, value);
 183                return 0;
 184        }
 185        if (!strcmp(var, "diff.autorefreshindex")) {
 186                diff_auto_refresh_index = git_config_bool(var, value);
 187                return 0;
 188        }
 189        if (!strcmp(var, "diff.mnemonicprefix")) {
 190                diff_mnemonic_prefix = git_config_bool(var, value);
 191                return 0;
 192        }
 193        if (!strcmp(var, "diff.noprefix")) {
 194                diff_no_prefix = git_config_bool(var, value);
 195                return 0;
 196        }
 197        if (!strcmp(var, "diff.statgraphwidth")) {
 198                diff_stat_graph_width = git_config_int(var, value);
 199                return 0;
 200        }
 201        if (!strcmp(var, "diff.external"))
 202                return git_config_string(&external_diff_cmd_cfg, var, value);
 203        if (!strcmp(var, "diff.wordregex"))
 204                return git_config_string(&diff_word_regex_cfg, var, value);
 205        if (!strcmp(var, "diff.orderfile"))
 206                return git_config_pathname(&diff_order_file_cfg, var, value);
 207
 208        if (!strcmp(var, "diff.ignoresubmodules"))
 209                handle_ignore_submodules_arg(&default_diff_options, value);
 210
 211        if (!strcmp(var, "diff.submodule")) {
 212                if (parse_submodule_params(&default_diff_options, value))
 213                        warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 214                                value);
 215                return 0;
 216        }
 217
 218        if (!strcmp(var, "diff.algorithm")) {
 219                diff_algorithm = parse_algorithm_value(value);
 220                if (diff_algorithm < 0)
 221                        return -1;
 222                return 0;
 223        }
 224
 225        if (git_color_config(var, value, cb) < 0)
 226                return -1;
 227
 228        return git_diff_basic_config(var, value, cb);
 229}
 230
 231int git_diff_basic_config(const char *var, const char *value, void *cb)
 232{
 233        if (!strcmp(var, "diff.renamelimit")) {
 234                diff_rename_limit_default = git_config_int(var, value);
 235                return 0;
 236        }
 237
 238        if (userdiff_config(var, value) < 0)
 239                return -1;
 240
 241        if (starts_with(var, "diff.color.") || starts_with(var, "color.diff.")) {
 242                int slot = parse_diff_color_slot(var, 11);
 243                if (slot < 0)
 244                        return 0;
 245                if (!value)
 246                        return config_error_nonbool(var);
 247                color_parse(value, var, diff_colors[slot]);
 248                return 0;
 249        }
 250
 251        /* like GNU diff's --suppress-blank-empty option  */
 252        if (!strcmp(var, "diff.suppressblankempty") ||
 253                        /* for backwards compatibility */
 254                        !strcmp(var, "diff.suppress-blank-empty")) {
 255                diff_suppress_blank_empty = git_config_bool(var, value);
 256                return 0;
 257        }
 258
 259        if (!strcmp(var, "diff.dirstat")) {
 260                struct strbuf errmsg = STRBUF_INIT;
 261                default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 262                if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 263                        warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 264                                errmsg.buf);
 265                strbuf_release(&errmsg);
 266                diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 267                return 0;
 268        }
 269
 270        if (starts_with(var, "submodule."))
 271                return parse_submodule_config_option(var, value);
 272
 273        return git_default_config(var, value, cb);
 274}
 275
 276static char *quote_two(const char *one, const char *two)
 277{
 278        int need_one = quote_c_style(one, NULL, NULL, 1);
 279        int need_two = quote_c_style(two, NULL, NULL, 1);
 280        struct strbuf res = STRBUF_INIT;
 281
 282        if (need_one + need_two) {
 283                strbuf_addch(&res, '"');
 284                quote_c_style(one, &res, NULL, 1);
 285                quote_c_style(two, &res, NULL, 1);
 286                strbuf_addch(&res, '"');
 287        } else {
 288                strbuf_addstr(&res, one);
 289                strbuf_addstr(&res, two);
 290        }
 291        return strbuf_detach(&res, NULL);
 292}
 293
 294static const char *external_diff(void)
 295{
 296        static const char *external_diff_cmd = NULL;
 297        static int done_preparing = 0;
 298
 299        if (done_preparing)
 300                return external_diff_cmd;
 301        external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 302        if (!external_diff_cmd)
 303                external_diff_cmd = external_diff_cmd_cfg;
 304        done_preparing = 1;
 305        return external_diff_cmd;
 306}
 307
 308static struct diff_tempfile {
 309        const char *name; /* filename external diff should read from */
 310        char hex[41];
 311        char mode[10];
 312        char tmp_path[PATH_MAX];
 313} diff_temp[2];
 314
 315typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 316
 317struct emit_callback {
 318        int color_diff;
 319        unsigned ws_rule;
 320        int blank_at_eof_in_preimage;
 321        int blank_at_eof_in_postimage;
 322        int lno_in_preimage;
 323        int lno_in_postimage;
 324        sane_truncate_fn truncate;
 325        const char **label_path;
 326        struct diff_words_data *diff_words;
 327        struct diff_options *opt;
 328        int *found_changesp;
 329        struct strbuf *header;
 330};
 331
 332static int count_lines(const char *data, int size)
 333{
 334        int count, ch, completely_empty = 1, nl_just_seen = 0;
 335        count = 0;
 336        while (0 < size--) {
 337                ch = *data++;
 338                if (ch == '\n') {
 339                        count++;
 340                        nl_just_seen = 1;
 341                        completely_empty = 0;
 342                }
 343                else {
 344                        nl_just_seen = 0;
 345                        completely_empty = 0;
 346                }
 347        }
 348        if (completely_empty)
 349                return 0;
 350        if (!nl_just_seen)
 351                count++; /* no trailing newline */
 352        return count;
 353}
 354
 355static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 356{
 357        if (!DIFF_FILE_VALID(one)) {
 358                mf->ptr = (char *)""; /* does not matter */
 359                mf->size = 0;
 360                return 0;
 361        }
 362        else if (diff_populate_filespec(one, 0))
 363                return -1;
 364
 365        mf->ptr = one->data;
 366        mf->size = one->size;
 367        return 0;
 368}
 369
 370/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 371static unsigned long diff_filespec_size(struct diff_filespec *one)
 372{
 373        if (!DIFF_FILE_VALID(one))
 374                return 0;
 375        diff_populate_filespec(one, 1);
 376        return one->size;
 377}
 378
 379static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 380{
 381        char *ptr = mf->ptr;
 382        long size = mf->size;
 383        int cnt = 0;
 384
 385        if (!size)
 386                return cnt;
 387        ptr += size - 1; /* pointing at the very end */
 388        if (*ptr != '\n')
 389                ; /* incomplete line */
 390        else
 391                ptr--; /* skip the last LF */
 392        while (mf->ptr < ptr) {
 393                char *prev_eol;
 394                for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 395                        if (*prev_eol == '\n')
 396                                break;
 397                if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 398                        break;
 399                cnt++;
 400                ptr = prev_eol - 1;
 401        }
 402        return cnt;
 403}
 404
 405static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 406                               struct emit_callback *ecbdata)
 407{
 408        int l1, l2, at;
 409        unsigned ws_rule = ecbdata->ws_rule;
 410        l1 = count_trailing_blank(mf1, ws_rule);
 411        l2 = count_trailing_blank(mf2, ws_rule);
 412        if (l2 <= l1) {
 413                ecbdata->blank_at_eof_in_preimage = 0;
 414                ecbdata->blank_at_eof_in_postimage = 0;
 415                return;
 416        }
 417        at = count_lines(mf1->ptr, mf1->size);
 418        ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 419
 420        at = count_lines(mf2->ptr, mf2->size);
 421        ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 422}
 423
 424static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 425                        int first, const char *line, int len)
 426{
 427        int has_trailing_newline, has_trailing_carriage_return;
 428        int nofirst;
 429        FILE *file = o->file;
 430
 431        fputs(diff_line_prefix(o), file);
 432
 433        if (len == 0) {
 434                has_trailing_newline = (first == '\n');
 435                has_trailing_carriage_return = (!has_trailing_newline &&
 436                                                (first == '\r'));
 437                nofirst = has_trailing_newline || has_trailing_carriage_return;
 438        } else {
 439                has_trailing_newline = (len > 0 && line[len-1] == '\n');
 440                if (has_trailing_newline)
 441                        len--;
 442                has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 443                if (has_trailing_carriage_return)
 444                        len--;
 445                nofirst = 0;
 446        }
 447
 448        if (len || !nofirst) {
 449                fputs(set, file);
 450                if (!nofirst)
 451                        fputc(first, file);
 452                fwrite(line, len, 1, file);
 453                fputs(reset, file);
 454        }
 455        if (has_trailing_carriage_return)
 456                fputc('\r', file);
 457        if (has_trailing_newline)
 458                fputc('\n', file);
 459}
 460
 461static void emit_line(struct diff_options *o, const char *set, const char *reset,
 462                      const char *line, int len)
 463{
 464        emit_line_0(o, set, reset, line[0], line+1, len-1);
 465}
 466
 467static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 468{
 469        if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 470              ecbdata->blank_at_eof_in_preimage &&
 471              ecbdata->blank_at_eof_in_postimage &&
 472              ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 473              ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 474                return 0;
 475        return ws_blank_line(line, len, ecbdata->ws_rule);
 476}
 477
 478static void emit_add_line(const char *reset,
 479                          struct emit_callback *ecbdata,
 480                          const char *line, int len)
 481{
 482        const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 483        const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 484
 485        if (!*ws)
 486                emit_line_0(ecbdata->opt, set, reset, '+', line, len);
 487        else if (new_blank_line_at_eof(ecbdata, line, len))
 488                /* Blank line at EOF - paint '+' as well */
 489                emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
 490        else {
 491                /* Emit just the prefix, then the rest. */
 492                emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
 493                ws_check_emit(line, len, ecbdata->ws_rule,
 494                              ecbdata->opt->file, set, reset, ws);
 495        }
 496}
 497
 498static void emit_hunk_header(struct emit_callback *ecbdata,
 499                             const char *line, int len)
 500{
 501        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 502        const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 503        const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 504        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 505        static const char atat[2] = { '@', '@' };
 506        const char *cp, *ep;
 507        struct strbuf msgbuf = STRBUF_INIT;
 508        int org_len = len;
 509        int i = 1;
 510
 511        /*
 512         * As a hunk header must begin with "@@ -<old>, +<new> @@",
 513         * it always is at least 10 bytes long.
 514         */
 515        if (len < 10 ||
 516            memcmp(line, atat, 2) ||
 517            !(ep = memmem(line + 2, len - 2, atat, 2))) {
 518                emit_line(ecbdata->opt, plain, reset, line, len);
 519                return;
 520        }
 521        ep += 2; /* skip over @@ */
 522
 523        /* The hunk header in fraginfo color */
 524        strbuf_add(&msgbuf, frag, strlen(frag));
 525        strbuf_add(&msgbuf, line, ep - line);
 526        strbuf_add(&msgbuf, reset, strlen(reset));
 527
 528        /*
 529         * trailing "\r\n"
 530         */
 531        for ( ; i < 3; i++)
 532                if (line[len - i] == '\r' || line[len - i] == '\n')
 533                        len--;
 534
 535        /* blank before the func header */
 536        for (cp = ep; ep - line < len; ep++)
 537                if (*ep != ' ' && *ep != '\t')
 538                        break;
 539        if (ep != cp) {
 540                strbuf_add(&msgbuf, plain, strlen(plain));
 541                strbuf_add(&msgbuf, cp, ep - cp);
 542                strbuf_add(&msgbuf, reset, strlen(reset));
 543        }
 544
 545        if (ep < line + len) {
 546                strbuf_add(&msgbuf, func, strlen(func));
 547                strbuf_add(&msgbuf, ep, line + len - ep);
 548                strbuf_add(&msgbuf, reset, strlen(reset));
 549        }
 550
 551        strbuf_add(&msgbuf, line + len, org_len - len);
 552        emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 553        strbuf_release(&msgbuf);
 554}
 555
 556static struct diff_tempfile *claim_diff_tempfile(void) {
 557        int i;
 558        for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 559                if (!diff_temp[i].name)
 560                        return diff_temp + i;
 561        die("BUG: diff is failing to clean up its tempfiles");
 562}
 563
 564static int remove_tempfile_installed;
 565
 566static void remove_tempfile(void)
 567{
 568        int i;
 569        for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 570                if (diff_temp[i].name == diff_temp[i].tmp_path)
 571                        unlink_or_warn(diff_temp[i].name);
 572                diff_temp[i].name = NULL;
 573        }
 574}
 575
 576static void remove_tempfile_on_signal(int signo)
 577{
 578        remove_tempfile();
 579        sigchain_pop(signo);
 580        raise(signo);
 581}
 582
 583static void print_line_count(FILE *file, int count)
 584{
 585        switch (count) {
 586        case 0:
 587                fprintf(file, "0,0");
 588                break;
 589        case 1:
 590                fprintf(file, "1");
 591                break;
 592        default:
 593                fprintf(file, "1,%d", count);
 594                break;
 595        }
 596}
 597
 598static void emit_rewrite_lines(struct emit_callback *ecb,
 599                               int prefix, const char *data, int size)
 600{
 601        const char *endp = NULL;
 602        static const char *nneof = " No newline at end of file\n";
 603        const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 604        const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 605
 606        while (0 < size) {
 607                int len;
 608
 609                endp = memchr(data, '\n', size);
 610                len = endp ? (endp - data + 1) : size;
 611                if (prefix != '+') {
 612                        ecb->lno_in_preimage++;
 613                        emit_line_0(ecb->opt, old, reset, '-',
 614                                    data, len);
 615                } else {
 616                        ecb->lno_in_postimage++;
 617                        emit_add_line(reset, ecb, data, len);
 618                }
 619                size -= len;
 620                data += len;
 621        }
 622        if (!endp) {
 623                const char *plain = diff_get_color(ecb->color_diff,
 624                                                   DIFF_PLAIN);
 625                putc('\n', ecb->opt->file);
 626                emit_line_0(ecb->opt, plain, reset, '\\',
 627                            nneof, strlen(nneof));
 628        }
 629}
 630
 631static void emit_rewrite_diff(const char *name_a,
 632                              const char *name_b,
 633                              struct diff_filespec *one,
 634                              struct diff_filespec *two,
 635                              struct userdiff_driver *textconv_one,
 636                              struct userdiff_driver *textconv_two,
 637                              struct diff_options *o)
 638{
 639        int lc_a, lc_b;
 640        const char *name_a_tab, *name_b_tab;
 641        const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 642        const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 643        const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 644        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 645        const char *a_prefix, *b_prefix;
 646        char *data_one, *data_two;
 647        size_t size_one, size_two;
 648        struct emit_callback ecbdata;
 649        const char *line_prefix = diff_line_prefix(o);
 650
 651        if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 652                a_prefix = o->b_prefix;
 653                b_prefix = o->a_prefix;
 654        } else {
 655                a_prefix = o->a_prefix;
 656                b_prefix = o->b_prefix;
 657        }
 658
 659        name_a += (*name_a == '/');
 660        name_b += (*name_b == '/');
 661        name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 662        name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 663
 664        strbuf_reset(&a_name);
 665        strbuf_reset(&b_name);
 666        quote_two_c_style(&a_name, a_prefix, name_a, 0);
 667        quote_two_c_style(&b_name, b_prefix, name_b, 0);
 668
 669        size_one = fill_textconv(textconv_one, one, &data_one);
 670        size_two = fill_textconv(textconv_two, two, &data_two);
 671
 672        memset(&ecbdata, 0, sizeof(ecbdata));
 673        ecbdata.color_diff = want_color(o->use_color);
 674        ecbdata.found_changesp = &o->found_changes;
 675        ecbdata.ws_rule = whitespace_rule(name_b);
 676        ecbdata.opt = o;
 677        if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 678                mmfile_t mf1, mf2;
 679                mf1.ptr = (char *)data_one;
 680                mf2.ptr = (char *)data_two;
 681                mf1.size = size_one;
 682                mf2.size = size_two;
 683                check_blank_at_eof(&mf1, &mf2, &ecbdata);
 684        }
 685        ecbdata.lno_in_preimage = 1;
 686        ecbdata.lno_in_postimage = 1;
 687
 688        lc_a = count_lines(data_one, size_one);
 689        lc_b = count_lines(data_two, size_two);
 690        fprintf(o->file,
 691                "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 692                line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 693                line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 694                line_prefix, fraginfo);
 695        if (!o->irreversible_delete)
 696                print_line_count(o->file, lc_a);
 697        else
 698                fprintf(o->file, "?,?");
 699        fprintf(o->file, " +");
 700        print_line_count(o->file, lc_b);
 701        fprintf(o->file, " @@%s\n", reset);
 702        if (lc_a && !o->irreversible_delete)
 703                emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 704        if (lc_b)
 705                emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 706        if (textconv_one)
 707                free((char *)data_one);
 708        if (textconv_two)
 709                free((char *)data_two);
 710}
 711
 712struct diff_words_buffer {
 713        mmfile_t text;
 714        long alloc;
 715        struct diff_words_orig {
 716                const char *begin, *end;
 717        } *orig;
 718        int orig_nr, orig_alloc;
 719};
 720
 721static void diff_words_append(char *line, unsigned long len,
 722                struct diff_words_buffer *buffer)
 723{
 724        ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 725        line++;
 726        len--;
 727        memcpy(buffer->text.ptr + buffer->text.size, line, len);
 728        buffer->text.size += len;
 729        buffer->text.ptr[buffer->text.size] = '\0';
 730}
 731
 732struct diff_words_style_elem {
 733        const char *prefix;
 734        const char *suffix;
 735        const char *color; /* NULL; filled in by the setup code if
 736                            * color is enabled */
 737};
 738
 739struct diff_words_style {
 740        enum diff_words_type type;
 741        struct diff_words_style_elem new, old, ctx;
 742        const char *newline;
 743};
 744
 745static struct diff_words_style diff_words_styles[] = {
 746        { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 747        { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 748        { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 749};
 750
 751struct diff_words_data {
 752        struct diff_words_buffer minus, plus;
 753        const char *current_plus;
 754        int last_minus;
 755        struct diff_options *opt;
 756        regex_t *word_regex;
 757        enum diff_words_type type;
 758        struct diff_words_style *style;
 759};
 760
 761static int fn_out_diff_words_write_helper(FILE *fp,
 762                                          struct diff_words_style_elem *st_el,
 763                                          const char *newline,
 764                                          size_t count, const char *buf,
 765                                          const char *line_prefix)
 766{
 767        int print = 0;
 768
 769        while (count) {
 770                char *p = memchr(buf, '\n', count);
 771                if (print)
 772                        fputs(line_prefix, fp);
 773                if (p != buf) {
 774                        if (st_el->color && fputs(st_el->color, fp) < 0)
 775                                return -1;
 776                        if (fputs(st_el->prefix, fp) < 0 ||
 777                            fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 778                            fputs(st_el->suffix, fp) < 0)
 779                                return -1;
 780                        if (st_el->color && *st_el->color
 781                            && fputs(GIT_COLOR_RESET, fp) < 0)
 782                                return -1;
 783                }
 784                if (!p)
 785                        return 0;
 786                if (fputs(newline, fp) < 0)
 787                        return -1;
 788                count -= p + 1 - buf;
 789                buf = p + 1;
 790                print = 1;
 791        }
 792        return 0;
 793}
 794
 795/*
 796 * '--color-words' algorithm can be described as:
 797 *
 798 *   1. collect a the minus/plus lines of a diff hunk, divided into
 799 *      minus-lines and plus-lines;
 800 *
 801 *   2. break both minus-lines and plus-lines into words and
 802 *      place them into two mmfile_t with one word for each line;
 803 *
 804 *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 805 *
 806 * And for the common parts of the both file, we output the plus side text.
 807 * diff_words->current_plus is used to trace the current position of the plus file
 808 * which printed. diff_words->last_minus is used to trace the last minus word
 809 * printed.
 810 *
 811 * For '--graph' to work with '--color-words', we need to output the graph prefix
 812 * on each line of color words output. Generally, there are two conditions on
 813 * which we should output the prefix.
 814 *
 815 *   1. diff_words->last_minus == 0 &&
 816 *      diff_words->current_plus == diff_words->plus.text.ptr
 817 *
 818 *      that is: the plus text must start as a new line, and if there is no minus
 819 *      word printed, a graph prefix must be printed.
 820 *
 821 *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 822 *      *(diff_words->current_plus - 1) == '\n'
 823 *
 824 *      that is: a graph prefix must be printed following a '\n'
 825 */
 826static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 827{
 828        if ((diff_words->last_minus == 0 &&
 829                diff_words->current_plus == diff_words->plus.text.ptr) ||
 830                (diff_words->current_plus > diff_words->plus.text.ptr &&
 831                *(diff_words->current_plus - 1) == '\n')) {
 832                return 1;
 833        } else {
 834                return 0;
 835        }
 836}
 837
 838static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 839{
 840        struct diff_words_data *diff_words = priv;
 841        struct diff_words_style *style = diff_words->style;
 842        int minus_first, minus_len, plus_first, plus_len;
 843        const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 844        struct diff_options *opt = diff_words->opt;
 845        const char *line_prefix;
 846
 847        if (line[0] != '@' || parse_hunk_header(line, len,
 848                        &minus_first, &minus_len, &plus_first, &plus_len))
 849                return;
 850
 851        assert(opt);
 852        line_prefix = diff_line_prefix(opt);
 853
 854        /* POSIX requires that first be decremented by one if len == 0... */
 855        if (minus_len) {
 856                minus_begin = diff_words->minus.orig[minus_first].begin;
 857                minus_end =
 858                        diff_words->minus.orig[minus_first + minus_len - 1].end;
 859        } else
 860                minus_begin = minus_end =
 861                        diff_words->minus.orig[minus_first].end;
 862
 863        if (plus_len) {
 864                plus_begin = diff_words->plus.orig[plus_first].begin;
 865                plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 866        } else
 867                plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 868
 869        if (color_words_output_graph_prefix(diff_words)) {
 870                fputs(line_prefix, diff_words->opt->file);
 871        }
 872        if (diff_words->current_plus != plus_begin) {
 873                fn_out_diff_words_write_helper(diff_words->opt->file,
 874                                &style->ctx, style->newline,
 875                                plus_begin - diff_words->current_plus,
 876                                diff_words->current_plus, line_prefix);
 877                if (*(plus_begin - 1) == '\n')
 878                        fputs(line_prefix, diff_words->opt->file);
 879        }
 880        if (minus_begin != minus_end) {
 881                fn_out_diff_words_write_helper(diff_words->opt->file,
 882                                &style->old, style->newline,
 883                                minus_end - minus_begin, minus_begin,
 884                                line_prefix);
 885        }
 886        if (plus_begin != plus_end) {
 887                fn_out_diff_words_write_helper(diff_words->opt->file,
 888                                &style->new, style->newline,
 889                                plus_end - plus_begin, plus_begin,
 890                                line_prefix);
 891        }
 892
 893        diff_words->current_plus = plus_end;
 894        diff_words->last_minus = minus_first;
 895}
 896
 897/* This function starts looking at *begin, and returns 0 iff a word was found. */
 898static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 899                int *begin, int *end)
 900{
 901        if (word_regex && *begin < buffer->size) {
 902                regmatch_t match[1];
 903                if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 904                        char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 905                                        '\n', match[0].rm_eo - match[0].rm_so);
 906                        *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 907                        *begin += match[0].rm_so;
 908                        return *begin >= *end;
 909                }
 910                return -1;
 911        }
 912
 913        /* find the next word */
 914        while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 915                (*begin)++;
 916        if (*begin >= buffer->size)
 917                return -1;
 918
 919        /* find the end of the word */
 920        *end = *begin + 1;
 921        while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 922                (*end)++;
 923
 924        return 0;
 925}
 926
 927/*
 928 * This function splits the words in buffer->text, stores the list with
 929 * newline separator into out, and saves the offsets of the original words
 930 * in buffer->orig.
 931 */
 932static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 933                regex_t *word_regex)
 934{
 935        int i, j;
 936        long alloc = 0;
 937
 938        out->size = 0;
 939        out->ptr = NULL;
 940
 941        /* fake an empty "0th" word */
 942        ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 943        buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 944        buffer->orig_nr = 1;
 945
 946        for (i = 0; i < buffer->text.size; i++) {
 947                if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 948                        return;
 949
 950                /* store original boundaries */
 951                ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 952                                buffer->orig_alloc);
 953                buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 954                buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 955                buffer->orig_nr++;
 956
 957                /* store one word */
 958                ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 959                memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 960                out->ptr[out->size + j - i] = '\n';
 961                out->size += j - i + 1;
 962
 963                i = j - 1;
 964        }
 965}
 966
 967/* this executes the word diff on the accumulated buffers */
 968static void diff_words_show(struct diff_words_data *diff_words)
 969{
 970        xpparam_t xpp;
 971        xdemitconf_t xecfg;
 972        mmfile_t minus, plus;
 973        struct diff_words_style *style = diff_words->style;
 974
 975        struct diff_options *opt = diff_words->opt;
 976        const char *line_prefix;
 977
 978        assert(opt);
 979        line_prefix = diff_line_prefix(opt);
 980
 981        /* special case: only removal */
 982        if (!diff_words->plus.text.size) {
 983                fputs(line_prefix, diff_words->opt->file);
 984                fn_out_diff_words_write_helper(diff_words->opt->file,
 985                        &style->old, style->newline,
 986                        diff_words->minus.text.size,
 987                        diff_words->minus.text.ptr, line_prefix);
 988                diff_words->minus.text.size = 0;
 989                return;
 990        }
 991
 992        diff_words->current_plus = diff_words->plus.text.ptr;
 993        diff_words->last_minus = 0;
 994
 995        memset(&xpp, 0, sizeof(xpp));
 996        memset(&xecfg, 0, sizeof(xecfg));
 997        diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 998        diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 999        xpp.flags = 0;
1000        /* as only the hunk header will be parsed, we need a 0-context */
1001        xecfg.ctxlen = 0;
1002        xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1003                      &xpp, &xecfg);
1004        free(minus.ptr);
1005        free(plus.ptr);
1006        if (diff_words->current_plus != diff_words->plus.text.ptr +
1007                        diff_words->plus.text.size) {
1008                if (color_words_output_graph_prefix(diff_words))
1009                        fputs(line_prefix, diff_words->opt->file);
1010                fn_out_diff_words_write_helper(diff_words->opt->file,
1011                        &style->ctx, style->newline,
1012                        diff_words->plus.text.ptr + diff_words->plus.text.size
1013                        - diff_words->current_plus, diff_words->current_plus,
1014                        line_prefix);
1015        }
1016        diff_words->minus.text.size = diff_words->plus.text.size = 0;
1017}
1018
1019/* In "color-words" mode, show word-diff of words accumulated in the buffer */
1020static void diff_words_flush(struct emit_callback *ecbdata)
1021{
1022        if (ecbdata->diff_words->minus.text.size ||
1023            ecbdata->diff_words->plus.text.size)
1024                diff_words_show(ecbdata->diff_words);
1025}
1026
1027static void diff_filespec_load_driver(struct diff_filespec *one)
1028{
1029        /* Use already-loaded driver */
1030        if (one->driver)
1031                return;
1032
1033        if (S_ISREG(one->mode))
1034                one->driver = userdiff_find_by_path(one->path);
1035
1036        /* Fallback to default settings */
1037        if (!one->driver)
1038                one->driver = userdiff_find_by_name("default");
1039}
1040
1041static const char *userdiff_word_regex(struct diff_filespec *one)
1042{
1043        diff_filespec_load_driver(one);
1044        return one->driver->word_regex;
1045}
1046
1047static void init_diff_words_data(struct emit_callback *ecbdata,
1048                                 struct diff_options *orig_opts,
1049                                 struct diff_filespec *one,
1050                                 struct diff_filespec *two)
1051{
1052        int i;
1053        struct diff_options *o = xmalloc(sizeof(struct diff_options));
1054        memcpy(o, orig_opts, sizeof(struct diff_options));
1055
1056        ecbdata->diff_words =
1057                xcalloc(1, sizeof(struct diff_words_data));
1058        ecbdata->diff_words->type = o->word_diff;
1059        ecbdata->diff_words->opt = o;
1060        if (!o->word_regex)
1061                o->word_regex = userdiff_word_regex(one);
1062        if (!o->word_regex)
1063                o->word_regex = userdiff_word_regex(two);
1064        if (!o->word_regex)
1065                o->word_regex = diff_word_regex_cfg;
1066        if (o->word_regex) {
1067                ecbdata->diff_words->word_regex = (regex_t *)
1068                        xmalloc(sizeof(regex_t));
1069                if (regcomp(ecbdata->diff_words->word_regex,
1070                            o->word_regex,
1071                            REG_EXTENDED | REG_NEWLINE))
1072                        die ("Invalid regular expression: %s",
1073                             o->word_regex);
1074        }
1075        for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1076                if (o->word_diff == diff_words_styles[i].type) {
1077                        ecbdata->diff_words->style =
1078                                &diff_words_styles[i];
1079                        break;
1080                }
1081        }
1082        if (want_color(o->use_color)) {
1083                struct diff_words_style *st = ecbdata->diff_words->style;
1084                st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1085                st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1086                st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1087        }
1088}
1089
1090static void free_diff_words_data(struct emit_callback *ecbdata)
1091{
1092        if (ecbdata->diff_words) {
1093                diff_words_flush(ecbdata);
1094                free (ecbdata->diff_words->opt);
1095                free (ecbdata->diff_words->minus.text.ptr);
1096                free (ecbdata->diff_words->minus.orig);
1097                free (ecbdata->diff_words->plus.text.ptr);
1098                free (ecbdata->diff_words->plus.orig);
1099                if (ecbdata->diff_words->word_regex) {
1100                        regfree(ecbdata->diff_words->word_regex);
1101                        free(ecbdata->diff_words->word_regex);
1102                }
1103                free(ecbdata->diff_words);
1104                ecbdata->diff_words = NULL;
1105        }
1106}
1107
1108const char *diff_get_color(int diff_use_color, enum color_diff ix)
1109{
1110        if (want_color(diff_use_color))
1111                return diff_colors[ix];
1112        return "";
1113}
1114
1115const char *diff_line_prefix(struct diff_options *opt)
1116{
1117        struct strbuf *msgbuf;
1118        if (!opt->output_prefix)
1119                return "";
1120
1121        msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1122        return msgbuf->buf;
1123}
1124
1125static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1126{
1127        const char *cp;
1128        unsigned long allot;
1129        size_t l = len;
1130
1131        if (ecb->truncate)
1132                return ecb->truncate(line, len);
1133        cp = line;
1134        allot = l;
1135        while (0 < l) {
1136                (void) utf8_width(&cp, &l);
1137                if (!cp)
1138                        break; /* truncated in the middle? */
1139        }
1140        return allot - l;
1141}
1142
1143static void find_lno(const char *line, struct emit_callback *ecbdata)
1144{
1145        const char *p;
1146        ecbdata->lno_in_preimage = 0;
1147        ecbdata->lno_in_postimage = 0;
1148        p = strchr(line, '-');
1149        if (!p)
1150                return; /* cannot happen */
1151        ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1152        p = strchr(p, '+');
1153        if (!p)
1154                return; /* cannot happen */
1155        ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1156}
1157
1158static void fn_out_consume(void *priv, char *line, unsigned long len)
1159{
1160        struct emit_callback *ecbdata = priv;
1161        const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1162        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1163        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1164        struct diff_options *o = ecbdata->opt;
1165        const char *line_prefix = diff_line_prefix(o);
1166
1167        if (ecbdata->header) {
1168                fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1169                strbuf_reset(ecbdata->header);
1170                ecbdata->header = NULL;
1171        }
1172        *(ecbdata->found_changesp) = 1;
1173
1174        if (ecbdata->label_path[0]) {
1175                const char *name_a_tab, *name_b_tab;
1176
1177                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1178                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1179
1180                fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1181                        line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1182                fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1183                        line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1184                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1185        }
1186
1187        if (diff_suppress_blank_empty
1188            && len == 2 && line[0] == ' ' && line[1] == '\n') {
1189                line[0] = '\n';
1190                len = 1;
1191        }
1192
1193        if (line[0] == '@') {
1194                if (ecbdata->diff_words)
1195                        diff_words_flush(ecbdata);
1196                len = sane_truncate_line(ecbdata, line, len);
1197                find_lno(line, ecbdata);
1198                emit_hunk_header(ecbdata, line, len);
1199                if (line[len-1] != '\n')
1200                        putc('\n', ecbdata->opt->file);
1201                return;
1202        }
1203
1204        if (len < 1) {
1205                emit_line(ecbdata->opt, reset, reset, line, len);
1206                if (ecbdata->diff_words
1207                    && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1208                        fputs("~\n", ecbdata->opt->file);
1209                return;
1210        }
1211
1212        if (ecbdata->diff_words) {
1213                if (line[0] == '-') {
1214                        diff_words_append(line, len,
1215                                          &ecbdata->diff_words->minus);
1216                        return;
1217                } else if (line[0] == '+') {
1218                        diff_words_append(line, len,
1219                                          &ecbdata->diff_words->plus);
1220                        return;
1221                } else if (starts_with(line, "\\ ")) {
1222                        /*
1223                         * Eat the "no newline at eof" marker as if we
1224                         * saw a "+" or "-" line with nothing on it,
1225                         * and return without diff_words_flush() to
1226                         * defer processing. If this is the end of
1227                         * preimage, more "+" lines may come after it.
1228                         */
1229                        return;
1230                }
1231                diff_words_flush(ecbdata);
1232                if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1233                        emit_line(ecbdata->opt, plain, reset, line, len);
1234                        fputs("~\n", ecbdata->opt->file);
1235                } else {
1236                        /*
1237                         * Skip the prefix character, if any.  With
1238                         * diff_suppress_blank_empty, there may be
1239                         * none.
1240                         */
1241                        if (line[0] != '\n') {
1242                              line++;
1243                              len--;
1244                        }
1245                        emit_line(ecbdata->opt, plain, reset, line, len);
1246                }
1247                return;
1248        }
1249
1250        if (line[0] != '+') {
1251                const char *color =
1252                        diff_get_color(ecbdata->color_diff,
1253                                       line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1254                ecbdata->lno_in_preimage++;
1255                if (line[0] == ' ')
1256                        ecbdata->lno_in_postimage++;
1257                emit_line(ecbdata->opt, color, reset, line, len);
1258        } else {
1259                ecbdata->lno_in_postimage++;
1260                emit_add_line(reset, ecbdata, line + 1, len - 1);
1261        }
1262}
1263
1264static char *pprint_rename(const char *a, const char *b)
1265{
1266        const char *old = a;
1267        const char *new = b;
1268        struct strbuf name = STRBUF_INIT;
1269        int pfx_length, sfx_length;
1270        int pfx_adjust_for_slash;
1271        int len_a = strlen(a);
1272        int len_b = strlen(b);
1273        int a_midlen, b_midlen;
1274        int qlen_a = quote_c_style(a, NULL, NULL, 0);
1275        int qlen_b = quote_c_style(b, NULL, NULL, 0);
1276
1277        if (qlen_a || qlen_b) {
1278                quote_c_style(a, &name, NULL, 0);
1279                strbuf_addstr(&name, " => ");
1280                quote_c_style(b, &name, NULL, 0);
1281                return strbuf_detach(&name, NULL);
1282        }
1283
1284        /* Find common prefix */
1285        pfx_length = 0;
1286        while (*old && *new && *old == *new) {
1287                if (*old == '/')
1288                        pfx_length = old - a + 1;
1289                old++;
1290                new++;
1291        }
1292
1293        /* Find common suffix */
1294        old = a + len_a;
1295        new = b + len_b;
1296        sfx_length = 0;
1297        /*
1298         * If there is a common prefix, it must end in a slash.  In
1299         * that case we let this loop run 1 into the prefix to see the
1300         * same slash.
1301         *
1302         * If there is no common prefix, we cannot do this as it would
1303         * underrun the input strings.
1304         */
1305        pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1306        while (a + pfx_length - pfx_adjust_for_slash <= old &&
1307               b + pfx_length - pfx_adjust_for_slash <= new &&
1308               *old == *new) {
1309                if (*old == '/')
1310                        sfx_length = len_a - (old - a);
1311                old--;
1312                new--;
1313        }
1314
1315        /*
1316         * pfx{mid-a => mid-b}sfx
1317         * {pfx-a => pfx-b}sfx
1318         * pfx{sfx-a => sfx-b}
1319         * name-a => name-b
1320         */
1321        a_midlen = len_a - pfx_length - sfx_length;
1322        b_midlen = len_b - pfx_length - sfx_length;
1323        if (a_midlen < 0)
1324                a_midlen = 0;
1325        if (b_midlen < 0)
1326                b_midlen = 0;
1327
1328        strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1329        if (pfx_length + sfx_length) {
1330                strbuf_add(&name, a, pfx_length);
1331                strbuf_addch(&name, '{');
1332        }
1333        strbuf_add(&name, a + pfx_length, a_midlen);
1334        strbuf_addstr(&name, " => ");
1335        strbuf_add(&name, b + pfx_length, b_midlen);
1336        if (pfx_length + sfx_length) {
1337                strbuf_addch(&name, '}');
1338                strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1339        }
1340        return strbuf_detach(&name, NULL);
1341}
1342
1343struct diffstat_t {
1344        int nr;
1345        int alloc;
1346        struct diffstat_file {
1347                char *from_name;
1348                char *name;
1349                char *print_name;
1350                unsigned is_unmerged:1;
1351                unsigned is_binary:1;
1352                unsigned is_renamed:1;
1353                unsigned is_interesting:1;
1354                uintmax_t added, deleted;
1355        } **files;
1356};
1357
1358static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1359                                          const char *name_a,
1360                                          const char *name_b)
1361{
1362        struct diffstat_file *x;
1363        x = xcalloc(sizeof (*x), 1);
1364        ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1365        diffstat->files[diffstat->nr++] = x;
1366        if (name_b) {
1367                x->from_name = xstrdup(name_a);
1368                x->name = xstrdup(name_b);
1369                x->is_renamed = 1;
1370        }
1371        else {
1372                x->from_name = NULL;
1373                x->name = xstrdup(name_a);
1374        }
1375        return x;
1376}
1377
1378static void diffstat_consume(void *priv, char *line, unsigned long len)
1379{
1380        struct diffstat_t *diffstat = priv;
1381        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1382
1383        if (line[0] == '+')
1384                x->added++;
1385        else if (line[0] == '-')
1386                x->deleted++;
1387}
1388
1389const char mime_boundary_leader[] = "------------";
1390
1391static int scale_linear(int it, int width, int max_change)
1392{
1393        if (!it)
1394                return 0;
1395        /*
1396         * make sure that at least one '-' or '+' is printed if
1397         * there is any change to this path. The easiest way is to
1398         * scale linearly as if the alloted width is one column shorter
1399         * than it is, and then add 1 to the result.
1400         */
1401        return 1 + (it * (width - 1) / max_change);
1402}
1403
1404static void show_name(FILE *file,
1405                      const char *prefix, const char *name, int len)
1406{
1407        fprintf(file, " %s%-*s |", prefix, len, name);
1408}
1409
1410static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1411{
1412        if (cnt <= 0)
1413                return;
1414        fprintf(file, "%s", set);
1415        while (cnt--)
1416                putc(ch, file);
1417        fprintf(file, "%s", reset);
1418}
1419
1420static void fill_print_name(struct diffstat_file *file)
1421{
1422        char *pname;
1423
1424        if (file->print_name)
1425                return;
1426
1427        if (!file->is_renamed) {
1428                struct strbuf buf = STRBUF_INIT;
1429                if (quote_c_style(file->name, &buf, NULL, 0)) {
1430                        pname = strbuf_detach(&buf, NULL);
1431                } else {
1432                        pname = file->name;
1433                        strbuf_release(&buf);
1434                }
1435        } else {
1436                pname = pprint_rename(file->from_name, file->name);
1437        }
1438        file->print_name = pname;
1439}
1440
1441int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1442{
1443        struct strbuf sb = STRBUF_INIT;
1444        int ret;
1445
1446        if (!files) {
1447                assert(insertions == 0 && deletions == 0);
1448                return fprintf(fp, "%s\n", " 0 files changed");
1449        }
1450
1451        strbuf_addf(&sb,
1452                    (files == 1) ? " %d file changed" : " %d files changed",
1453                    files);
1454
1455        /*
1456         * For binary diff, the caller may want to print "x files
1457         * changed" with insertions == 0 && deletions == 0.
1458         *
1459         * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1460         * is probably less confusing (i.e skip over "2 files changed
1461         * but nothing about added/removed lines? Is this a bug in Git?").
1462         */
1463        if (insertions || deletions == 0) {
1464                /*
1465                 * TRANSLATORS: "+" in (+) is a line addition marker;
1466                 * do not translate it.
1467                 */
1468                strbuf_addf(&sb,
1469                            (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1470                            insertions);
1471        }
1472
1473        if (deletions || insertions == 0) {
1474                /*
1475                 * TRANSLATORS: "-" in (-) is a line removal marker;
1476                 * do not translate it.
1477                 */
1478                strbuf_addf(&sb,
1479                            (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1480                            deletions);
1481        }
1482        strbuf_addch(&sb, '\n');
1483        ret = fputs(sb.buf, fp);
1484        strbuf_release(&sb);
1485        return ret;
1486}
1487
1488static void show_stats(struct diffstat_t *data, struct diff_options *options)
1489{
1490        int i, len, add, del, adds = 0, dels = 0;
1491        uintmax_t max_change = 0, max_len = 0;
1492        int total_files = data->nr, count;
1493        int width, name_width, graph_width, number_width = 0, bin_width = 0;
1494        const char *reset, *add_c, *del_c;
1495        const char *line_prefix = "";
1496        int extra_shown = 0;
1497
1498        if (data->nr == 0)
1499                return;
1500
1501        line_prefix = diff_line_prefix(options);
1502        count = options->stat_count ? options->stat_count : data->nr;
1503
1504        reset = diff_get_color_opt(options, DIFF_RESET);
1505        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1506        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1507
1508        /*
1509         * Find the longest filename and max number of changes
1510         */
1511        for (i = 0; (i < count) && (i < data->nr); i++) {
1512                struct diffstat_file *file = data->files[i];
1513                uintmax_t change = file->added + file->deleted;
1514
1515                if (!file->is_interesting && (change == 0)) {
1516                        count++; /* not shown == room for one more */
1517                        continue;
1518                }
1519                fill_print_name(file);
1520                len = strlen(file->print_name);
1521                if (max_len < len)
1522                        max_len = len;
1523
1524                if (file->is_unmerged) {
1525                        /* "Unmerged" is 8 characters */
1526                        bin_width = bin_width < 8 ? 8 : bin_width;
1527                        continue;
1528                }
1529                if (file->is_binary) {
1530                        /* "Bin XXX -> YYY bytes" */
1531                        int w = 14 + decimal_width(file->added)
1532                                + decimal_width(file->deleted);
1533                        bin_width = bin_width < w ? w : bin_width;
1534                        /* Display change counts aligned with "Bin" */
1535                        number_width = 3;
1536                        continue;
1537                }
1538
1539                if (max_change < change)
1540                        max_change = change;
1541        }
1542        count = i; /* where we can stop scanning in data->files[] */
1543
1544        /*
1545         * We have width = stat_width or term_columns() columns total.
1546         * We want a maximum of min(max_len, stat_name_width) for the name part.
1547         * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1548         * We also need 1 for " " and 4 + decimal_width(max_change)
1549         * for " | NNNN " and one the empty column at the end, altogether
1550         * 6 + decimal_width(max_change).
1551         *
1552         * If there's not enough space, we will use the smaller of
1553         * stat_name_width (if set) and 5/8*width for the filename,
1554         * and the rest for constant elements + graph part, but no more
1555         * than stat_graph_width for the graph part.
1556         * (5/8 gives 50 for filename and 30 for the constant parts + graph
1557         * for the standard terminal size).
1558         *
1559         * In other words: stat_width limits the maximum width, and
1560         * stat_name_width fixes the maximum width of the filename,
1561         * and is also used to divide available columns if there
1562         * aren't enough.
1563         *
1564         * Binary files are displayed with "Bin XXX -> YYY bytes"
1565         * instead of the change count and graph. This part is treated
1566         * similarly to the graph part, except that it is not
1567         * "scaled". If total width is too small to accommodate the
1568         * guaranteed minimum width of the filename part and the
1569         * separators and this message, this message will "overflow"
1570         * making the line longer than the maximum width.
1571         */
1572
1573        if (options->stat_width == -1)
1574                width = term_columns() - options->output_prefix_length;
1575        else
1576                width = options->stat_width ? options->stat_width : 80;
1577        number_width = decimal_width(max_change) > number_width ?
1578                decimal_width(max_change) : number_width;
1579
1580        if (options->stat_graph_width == -1)
1581                options->stat_graph_width = diff_stat_graph_width;
1582
1583        /*
1584         * Guarantee 3/8*16==6 for the graph part
1585         * and 5/8*16==10 for the filename part
1586         */
1587        if (width < 16 + 6 + number_width)
1588                width = 16 + 6 + number_width;
1589
1590        /*
1591         * First assign sizes that are wanted, ignoring available width.
1592         * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1593         * starting from "XXX" should fit in graph_width.
1594         */
1595        graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1596        if (options->stat_graph_width &&
1597            options->stat_graph_width < graph_width)
1598                graph_width = options->stat_graph_width;
1599
1600        name_width = (options->stat_name_width > 0 &&
1601                      options->stat_name_width < max_len) ?
1602                options->stat_name_width : max_len;
1603
1604        /*
1605         * Adjust adjustable widths not to exceed maximum width
1606         */
1607        if (name_width + number_width + 6 + graph_width > width) {
1608                if (graph_width > width * 3/8 - number_width - 6) {
1609                        graph_width = width * 3/8 - number_width - 6;
1610                        if (graph_width < 6)
1611                                graph_width = 6;
1612                }
1613
1614                if (options->stat_graph_width &&
1615                    graph_width > options->stat_graph_width)
1616                        graph_width = options->stat_graph_width;
1617                if (name_width > width - number_width - 6 - graph_width)
1618                        name_width = width - number_width - 6 - graph_width;
1619                else
1620                        graph_width = width - number_width - 6 - name_width;
1621        }
1622
1623        /*
1624         * From here name_width is the width of the name area,
1625         * and graph_width is the width of the graph area.
1626         * max_change is used to scale graph properly.
1627         */
1628        for (i = 0; i < count; i++) {
1629                const char *prefix = "";
1630                struct diffstat_file *file = data->files[i];
1631                char *name = file->print_name;
1632                uintmax_t added = file->added;
1633                uintmax_t deleted = file->deleted;
1634                int name_len;
1635
1636                if (!file->is_interesting && (added + deleted == 0))
1637                        continue;
1638
1639                /*
1640                 * "scale" the filename
1641                 */
1642                len = name_width;
1643                name_len = strlen(name);
1644                if (name_width < name_len) {
1645                        char *slash;
1646                        prefix = "...";
1647                        len -= 3;
1648                        name += name_len - len;
1649                        slash = strchr(name, '/');
1650                        if (slash)
1651                                name = slash;
1652                }
1653
1654                if (file->is_binary) {
1655                        fprintf(options->file, "%s", line_prefix);
1656                        show_name(options->file, prefix, name, len);
1657                        fprintf(options->file, " %*s", number_width, "Bin");
1658                        if (!added && !deleted) {
1659                                putc('\n', options->file);
1660                                continue;
1661                        }
1662                        fprintf(options->file, " %s%"PRIuMAX"%s",
1663                                del_c, deleted, reset);
1664                        fprintf(options->file, " -> ");
1665                        fprintf(options->file, "%s%"PRIuMAX"%s",
1666                                add_c, added, reset);
1667                        fprintf(options->file, " bytes");
1668                        fprintf(options->file, "\n");
1669                        continue;
1670                }
1671                else if (file->is_unmerged) {
1672                        fprintf(options->file, "%s", line_prefix);
1673                        show_name(options->file, prefix, name, len);
1674                        fprintf(options->file, " Unmerged\n");
1675                        continue;
1676                }
1677
1678                /*
1679                 * scale the add/delete
1680                 */
1681                add = added;
1682                del = deleted;
1683
1684                if (graph_width <= max_change) {
1685                        int total = scale_linear(add + del, graph_width, max_change);
1686                        if (total < 2 && add && del)
1687                                /* width >= 2 due to the sanity check */
1688                                total = 2;
1689                        if (add < del) {
1690                                add = scale_linear(add, graph_width, max_change);
1691                                del = total - add;
1692                        } else {
1693                                del = scale_linear(del, graph_width, max_change);
1694                                add = total - del;
1695                        }
1696                }
1697                fprintf(options->file, "%s", line_prefix);
1698                show_name(options->file, prefix, name, len);
1699                fprintf(options->file, " %*"PRIuMAX"%s",
1700                        number_width, added + deleted,
1701                        added + deleted ? " " : "");
1702                show_graph(options->file, '+', add, add_c, reset);
1703                show_graph(options->file, '-', del, del_c, reset);
1704                fprintf(options->file, "\n");
1705        }
1706
1707        for (i = 0; i < data->nr; i++) {
1708                struct diffstat_file *file = data->files[i];
1709                uintmax_t added = file->added;
1710                uintmax_t deleted = file->deleted;
1711
1712                if (file->is_unmerged ||
1713                    (!file->is_interesting && (added + deleted == 0))) {
1714                        total_files--;
1715                        continue;
1716                }
1717
1718                if (!file->is_binary) {
1719                        adds += added;
1720                        dels += deleted;
1721                }
1722                if (i < count)
1723                        continue;
1724                if (!extra_shown)
1725                        fprintf(options->file, "%s ...\n", line_prefix);
1726                extra_shown = 1;
1727        }
1728        fprintf(options->file, "%s", line_prefix);
1729        print_stat_summary(options->file, total_files, adds, dels);
1730}
1731
1732static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1733{
1734        int i, adds = 0, dels = 0, total_files = data->nr;
1735
1736        if (data->nr == 0)
1737                return;
1738
1739        for (i = 0; i < data->nr; i++) {
1740                int added = data->files[i]->added;
1741                int deleted= data->files[i]->deleted;
1742
1743                if (data->files[i]->is_unmerged ||
1744                    (!data->files[i]->is_interesting && (added + deleted == 0))) {
1745                        total_files--;
1746                } else if (!data->files[i]->is_binary) { /* don't count bytes */
1747                        adds += added;
1748                        dels += deleted;
1749                }
1750        }
1751        fprintf(options->file, "%s", diff_line_prefix(options));
1752        print_stat_summary(options->file, total_files, adds, dels);
1753}
1754
1755static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1756{
1757        int i;
1758
1759        if (data->nr == 0)
1760                return;
1761
1762        for (i = 0; i < data->nr; i++) {
1763                struct diffstat_file *file = data->files[i];
1764
1765                fprintf(options->file, "%s", diff_line_prefix(options));
1766
1767                if (file->is_binary)
1768                        fprintf(options->file, "-\t-\t");
1769                else
1770                        fprintf(options->file,
1771                                "%"PRIuMAX"\t%"PRIuMAX"\t",
1772                                file->added, file->deleted);
1773                if (options->line_termination) {
1774                        fill_print_name(file);
1775                        if (!file->is_renamed)
1776                                write_name_quoted(file->name, options->file,
1777                                                  options->line_termination);
1778                        else {
1779                                fputs(file->print_name, options->file);
1780                                putc(options->line_termination, options->file);
1781                        }
1782                } else {
1783                        if (file->is_renamed) {
1784                                putc('\0', options->file);
1785                                write_name_quoted(file->from_name, options->file, '\0');
1786                        }
1787                        write_name_quoted(file->name, options->file, '\0');
1788                }
1789        }
1790}
1791
1792struct dirstat_file {
1793        const char *name;
1794        unsigned long changed;
1795};
1796
1797struct dirstat_dir {
1798        struct dirstat_file *files;
1799        int alloc, nr, permille, cumulative;
1800};
1801
1802static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1803                unsigned long changed, const char *base, int baselen)
1804{
1805        unsigned long this_dir = 0;
1806        unsigned int sources = 0;
1807        const char *line_prefix = diff_line_prefix(opt);
1808
1809        while (dir->nr) {
1810                struct dirstat_file *f = dir->files;
1811                int namelen = strlen(f->name);
1812                unsigned long this;
1813                char *slash;
1814
1815                if (namelen < baselen)
1816                        break;
1817                if (memcmp(f->name, base, baselen))
1818                        break;
1819                slash = strchr(f->name + baselen, '/');
1820                if (slash) {
1821                        int newbaselen = slash + 1 - f->name;
1822                        this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1823                        sources++;
1824                } else {
1825                        this = f->changed;
1826                        dir->files++;
1827                        dir->nr--;
1828                        sources += 2;
1829                }
1830                this_dir += this;
1831        }
1832
1833        /*
1834         * We don't report dirstat's for
1835         *  - the top level
1836         *  - or cases where everything came from a single directory
1837         *    under this directory (sources == 1).
1838         */
1839        if (baselen && sources != 1) {
1840                if (this_dir) {
1841                        int permille = this_dir * 1000 / changed;
1842                        if (permille >= dir->permille) {
1843                                fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1844                                        permille / 10, permille % 10, baselen, base);
1845                                if (!dir->cumulative)
1846                                        return 0;
1847                        }
1848                }
1849        }
1850        return this_dir;
1851}
1852
1853static int dirstat_compare(const void *_a, const void *_b)
1854{
1855        const struct dirstat_file *a = _a;
1856        const struct dirstat_file *b = _b;
1857        return strcmp(a->name, b->name);
1858}
1859
1860static void show_dirstat(struct diff_options *options)
1861{
1862        int i;
1863        unsigned long changed;
1864        struct dirstat_dir dir;
1865        struct diff_queue_struct *q = &diff_queued_diff;
1866
1867        dir.files = NULL;
1868        dir.alloc = 0;
1869        dir.nr = 0;
1870        dir.permille = options->dirstat_permille;
1871        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1872
1873        changed = 0;
1874        for (i = 0; i < q->nr; i++) {
1875                struct diff_filepair *p = q->queue[i];
1876                const char *name;
1877                unsigned long copied, added, damage;
1878                int content_changed;
1879
1880                name = p->two->path ? p->two->path : p->one->path;
1881
1882                if (p->one->sha1_valid && p->two->sha1_valid)
1883                        content_changed = hashcmp(p->one->sha1, p->two->sha1);
1884                else
1885                        content_changed = 1;
1886
1887                if (!content_changed) {
1888                        /*
1889                         * The SHA1 has not changed, so pre-/post-content is
1890                         * identical. We can therefore skip looking at the
1891                         * file contents altogether.
1892                         */
1893                        damage = 0;
1894                        goto found_damage;
1895                }
1896
1897                if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1898                        /*
1899                         * In --dirstat-by-file mode, we don't really need to
1900                         * look at the actual file contents at all.
1901                         * The fact that the SHA1 changed is enough for us to
1902                         * add this file to the list of results
1903                         * (with each file contributing equal damage).
1904                         */
1905                        damage = 1;
1906                        goto found_damage;
1907                }
1908
1909                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1910                        diff_populate_filespec(p->one, 0);
1911                        diff_populate_filespec(p->two, 0);
1912                        diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1913                                               &copied, &added);
1914                        diff_free_filespec_data(p->one);
1915                        diff_free_filespec_data(p->two);
1916                } else if (DIFF_FILE_VALID(p->one)) {
1917                        diff_populate_filespec(p->one, 1);
1918                        copied = added = 0;
1919                        diff_free_filespec_data(p->one);
1920                } else if (DIFF_FILE_VALID(p->two)) {
1921                        diff_populate_filespec(p->two, 1);
1922                        copied = 0;
1923                        added = p->two->size;
1924                        diff_free_filespec_data(p->two);
1925                } else
1926                        continue;
1927
1928                /*
1929                 * Original minus copied is the removed material,
1930                 * added is the new material.  They are both damages
1931                 * made to the preimage.
1932                 * If the resulting damage is zero, we know that
1933                 * diffcore_count_changes() considers the two entries to
1934                 * be identical, but since content_changed is true, we
1935                 * know that there must have been _some_ kind of change,
1936                 * so we force all entries to have damage > 0.
1937                 */
1938                damage = (p->one->size - copied) + added;
1939                if (!damage)
1940                        damage = 1;
1941
1942found_damage:
1943                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1944                dir.files[dir.nr].name = name;
1945                dir.files[dir.nr].changed = damage;
1946                changed += damage;
1947                dir.nr++;
1948        }
1949
1950        /* This can happen even with many files, if everything was renames */
1951        if (!changed)
1952                return;
1953
1954        /* Show all directories with more than x% of the changes */
1955        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1956        gather_dirstat(options, &dir, changed, "", 0);
1957}
1958
1959static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1960{
1961        int i;
1962        unsigned long changed;
1963        struct dirstat_dir dir;
1964
1965        if (data->nr == 0)
1966                return;
1967
1968        dir.files = NULL;
1969        dir.alloc = 0;
1970        dir.nr = 0;
1971        dir.permille = options->dirstat_permille;
1972        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1973
1974        changed = 0;
1975        for (i = 0; i < data->nr; i++) {
1976                struct diffstat_file *file = data->files[i];
1977                unsigned long damage = file->added + file->deleted;
1978                if (file->is_binary)
1979                        /*
1980                         * binary files counts bytes, not lines. Must find some
1981                         * way to normalize binary bytes vs. textual lines.
1982                         * The following heuristic assumes that there are 64
1983                         * bytes per "line".
1984                         * This is stupid and ugly, but very cheap...
1985                         */
1986                        damage = (damage + 63) / 64;
1987                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1988                dir.files[dir.nr].name = file->name;
1989                dir.files[dir.nr].changed = damage;
1990                changed += damage;
1991                dir.nr++;
1992        }
1993
1994        /* This can happen even with many files, if everything was renames */
1995        if (!changed)
1996                return;
1997
1998        /* Show all directories with more than x% of the changes */
1999        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2000        gather_dirstat(options, &dir, changed, "", 0);
2001}
2002
2003static void free_diffstat_info(struct diffstat_t *diffstat)
2004{
2005        int i;
2006        for (i = 0; i < diffstat->nr; i++) {
2007                struct diffstat_file *f = diffstat->files[i];
2008                if (f->name != f->print_name)
2009                        free(f->print_name);
2010                free(f->name);
2011                free(f->from_name);
2012                free(f);
2013        }
2014        free(diffstat->files);
2015}
2016
2017struct checkdiff_t {
2018        const char *filename;
2019        int lineno;
2020        int conflict_marker_size;
2021        struct diff_options *o;
2022        unsigned ws_rule;
2023        unsigned status;
2024};
2025
2026static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2027{
2028        char firstchar;
2029        int cnt;
2030
2031        if (len < marker_size + 1)
2032                return 0;
2033        firstchar = line[0];
2034        switch (firstchar) {
2035        case '=': case '>': case '<': case '|':
2036                break;
2037        default:
2038                return 0;
2039        }
2040        for (cnt = 1; cnt < marker_size; cnt++)
2041                if (line[cnt] != firstchar)
2042                        return 0;
2043        /* line[1] thru line[marker_size-1] are same as firstchar */
2044        if (len < marker_size + 1 || !isspace(line[marker_size]))
2045                return 0;
2046        return 1;
2047}
2048
2049static void checkdiff_consume(void *priv, char *line, unsigned long len)
2050{
2051        struct checkdiff_t *data = priv;
2052        int marker_size = data->conflict_marker_size;
2053        const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2054        const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2055        const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2056        char *err;
2057        const char *line_prefix;
2058
2059        assert(data->o);
2060        line_prefix = diff_line_prefix(data->o);
2061
2062        if (line[0] == '+') {
2063                unsigned bad;
2064                data->lineno++;
2065                if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2066                        data->status |= 1;
2067                        fprintf(data->o->file,
2068                                "%s%s:%d: leftover conflict marker\n",
2069                                line_prefix, data->filename, data->lineno);
2070                }
2071                bad = ws_check(line + 1, len - 1, data->ws_rule);
2072                if (!bad)
2073                        return;
2074                data->status |= bad;
2075                err = whitespace_error_string(bad);
2076                fprintf(data->o->file, "%s%s:%d: %s.\n",
2077                        line_prefix, data->filename, data->lineno, err);
2078                free(err);
2079                emit_line(data->o, set, reset, line, 1);
2080                ws_check_emit(line + 1, len - 1, data->ws_rule,
2081                              data->o->file, set, reset, ws);
2082        } else if (line[0] == ' ') {
2083                data->lineno++;
2084        } else if (line[0] == '@') {
2085                char *plus = strchr(line, '+');
2086                if (plus)
2087                        data->lineno = strtol(plus, NULL, 10) - 1;
2088                else
2089                        die("invalid diff");
2090        }
2091}
2092
2093static unsigned char *deflate_it(char *data,
2094                                 unsigned long size,
2095                                 unsigned long *result_size)
2096{
2097        int bound;
2098        unsigned char *deflated;
2099        git_zstream stream;
2100
2101        memset(&stream, 0, sizeof(stream));
2102        git_deflate_init(&stream, zlib_compression_level);
2103        bound = git_deflate_bound(&stream, size);
2104        deflated = xmalloc(bound);
2105        stream.next_out = deflated;
2106        stream.avail_out = bound;
2107
2108        stream.next_in = (unsigned char *)data;
2109        stream.avail_in = size;
2110        while (git_deflate(&stream, Z_FINISH) == Z_OK)
2111                ; /* nothing */
2112        git_deflate_end(&stream);
2113        *result_size = stream.total_out;
2114        return deflated;
2115}
2116
2117static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2118                                  const char *prefix)
2119{
2120        void *cp;
2121        void *delta;
2122        void *deflated;
2123        void *data;
2124        unsigned long orig_size;
2125        unsigned long delta_size;
2126        unsigned long deflate_size;
2127        unsigned long data_size;
2128
2129        /* We could do deflated delta, or we could do just deflated two,
2130         * whichever is smaller.
2131         */
2132        delta = NULL;
2133        deflated = deflate_it(two->ptr, two->size, &deflate_size);
2134        if (one->size && two->size) {
2135                delta = diff_delta(one->ptr, one->size,
2136                                   two->ptr, two->size,
2137                                   &delta_size, deflate_size);
2138                if (delta) {
2139                        void *to_free = delta;
2140                        orig_size = delta_size;
2141                        delta = deflate_it(delta, delta_size, &delta_size);
2142                        free(to_free);
2143                }
2144        }
2145
2146        if (delta && delta_size < deflate_size) {
2147                fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2148                free(deflated);
2149                data = delta;
2150                data_size = delta_size;
2151        }
2152        else {
2153                fprintf(file, "%sliteral %lu\n", prefix, two->size);
2154                free(delta);
2155                data = deflated;
2156                data_size = deflate_size;
2157        }
2158
2159        /* emit data encoded in base85 */
2160        cp = data;
2161        while (data_size) {
2162                int bytes = (52 < data_size) ? 52 : data_size;
2163                char line[70];
2164                data_size -= bytes;
2165                if (bytes <= 26)
2166                        line[0] = bytes + 'A' - 1;
2167                else
2168                        line[0] = bytes - 26 + 'a' - 1;
2169                encode_85(line + 1, cp, bytes);
2170                cp = (char *) cp + bytes;
2171                fprintf(file, "%s", prefix);
2172                fputs(line, file);
2173                fputc('\n', file);
2174        }
2175        fprintf(file, "%s\n", prefix);
2176        free(data);
2177}
2178
2179static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2180                             const char *prefix)
2181{
2182        fprintf(file, "%sGIT binary patch\n", prefix);
2183        emit_binary_diff_body(file, one, two, prefix);
2184        emit_binary_diff_body(file, two, one, prefix);
2185}
2186
2187int diff_filespec_is_binary(struct diff_filespec *one)
2188{
2189        if (one->is_binary == -1) {
2190                diff_filespec_load_driver(one);
2191                if (one->driver->binary != -1)
2192                        one->is_binary = one->driver->binary;
2193                else {
2194                        if (!one->data && DIFF_FILE_VALID(one))
2195                                diff_populate_filespec(one, 0);
2196                        if (one->data)
2197                                one->is_binary = buffer_is_binary(one->data,
2198                                                one->size);
2199                        if (one->is_binary == -1)
2200                                one->is_binary = 0;
2201                }
2202        }
2203        return one->is_binary;
2204}
2205
2206static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2207{
2208        diff_filespec_load_driver(one);
2209        return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2210}
2211
2212void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2213{
2214        if (!options->a_prefix)
2215                options->a_prefix = a;
2216        if (!options->b_prefix)
2217                options->b_prefix = b;
2218}
2219
2220struct userdiff_driver *get_textconv(struct diff_filespec *one)
2221{
2222        if (!DIFF_FILE_VALID(one))
2223                return NULL;
2224
2225        diff_filespec_load_driver(one);
2226        return userdiff_get_textconv(one->driver);
2227}
2228
2229static void builtin_diff(const char *name_a,
2230                         const char *name_b,
2231                         struct diff_filespec *one,
2232                         struct diff_filespec *two,
2233                         const char *xfrm_msg,
2234                         int must_show_header,
2235                         struct diff_options *o,
2236                         int complete_rewrite)
2237{
2238        mmfile_t mf1, mf2;
2239        const char *lbl[2];
2240        char *a_one, *b_two;
2241        const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2242        const char *reset = diff_get_color_opt(o, DIFF_RESET);
2243        const char *a_prefix, *b_prefix;
2244        struct userdiff_driver *textconv_one = NULL;
2245        struct userdiff_driver *textconv_two = NULL;
2246        struct strbuf header = STRBUF_INIT;
2247        const char *line_prefix = diff_line_prefix(o);
2248
2249        if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2250                        (!one->mode || S_ISGITLINK(one->mode)) &&
2251                        (!two->mode || S_ISGITLINK(two->mode))) {
2252                const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2253                const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2254                show_submodule_summary(o->file, one->path ? one->path : two->path,
2255                                line_prefix,
2256                                one->sha1, two->sha1, two->dirty_submodule,
2257                                meta, del, add, reset);
2258                return;
2259        }
2260
2261        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2262                textconv_one = get_textconv(one);
2263                textconv_two = get_textconv(two);
2264        }
2265
2266        diff_set_mnemonic_prefix(o, "a/", "b/");
2267        if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2268                a_prefix = o->b_prefix;
2269                b_prefix = o->a_prefix;
2270        } else {
2271                a_prefix = o->a_prefix;
2272                b_prefix = o->b_prefix;
2273        }
2274
2275        /* Never use a non-valid filename anywhere if at all possible */
2276        name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2277        name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2278
2279        a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2280        b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2281        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2282        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2283        strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2284        if (lbl[0][0] == '/') {
2285                /* /dev/null */
2286                strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2287                if (xfrm_msg)
2288                        strbuf_addstr(&header, xfrm_msg);
2289                must_show_header = 1;
2290        }
2291        else if (lbl[1][0] == '/') {
2292                strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2293                if (xfrm_msg)
2294                        strbuf_addstr(&header, xfrm_msg);
2295                must_show_header = 1;
2296        }
2297        else {
2298                if (one->mode != two->mode) {
2299                        strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2300                        strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2301                        must_show_header = 1;
2302                }
2303                if (xfrm_msg)
2304                        strbuf_addstr(&header, xfrm_msg);
2305
2306                /*
2307                 * we do not run diff between different kind
2308                 * of objects.
2309                 */
2310                if ((one->mode ^ two->mode) & S_IFMT)
2311                        goto free_ab_and_return;
2312                if (complete_rewrite &&
2313                    (textconv_one || !diff_filespec_is_binary(one)) &&
2314                    (textconv_two || !diff_filespec_is_binary(two))) {
2315                        fprintf(o->file, "%s", header.buf);
2316                        strbuf_reset(&header);
2317                        emit_rewrite_diff(name_a, name_b, one, two,
2318                                                textconv_one, textconv_two, o);
2319                        o->found_changes = 1;
2320                        goto free_ab_and_return;
2321                }
2322        }
2323
2324        if (o->irreversible_delete && lbl[1][0] == '/') {
2325                fprintf(o->file, "%s", header.buf);
2326                strbuf_reset(&header);
2327                goto free_ab_and_return;
2328        } else if (!DIFF_OPT_TST(o, TEXT) &&
2329            ( (!textconv_one && diff_filespec_is_binary(one)) ||
2330              (!textconv_two && diff_filespec_is_binary(two)) )) {
2331                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2332                        die("unable to read files to diff");
2333                /* Quite common confusing case */
2334                if (mf1.size == mf2.size &&
2335                    !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2336                        if (must_show_header)
2337                                fprintf(o->file, "%s", header.buf);
2338                        goto free_ab_and_return;
2339                }
2340                fprintf(o->file, "%s", header.buf);
2341                strbuf_reset(&header);
2342                if (DIFF_OPT_TST(o, BINARY))
2343                        emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2344                else
2345                        fprintf(o->file, "%sBinary files %s and %s differ\n",
2346                                line_prefix, lbl[0], lbl[1]);
2347                o->found_changes = 1;
2348        } else {
2349                /* Crazy xdl interfaces.. */
2350                const char *diffopts = getenv("GIT_DIFF_OPTS");
2351                xpparam_t xpp;
2352                xdemitconf_t xecfg;
2353                struct emit_callback ecbdata;
2354                const struct userdiff_funcname *pe;
2355
2356                if (must_show_header) {
2357                        fprintf(o->file, "%s", header.buf);
2358                        strbuf_reset(&header);
2359                }
2360
2361                mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2362                mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2363
2364                pe = diff_funcname_pattern(one);
2365                if (!pe)
2366                        pe = diff_funcname_pattern(two);
2367
2368                memset(&xpp, 0, sizeof(xpp));
2369                memset(&xecfg, 0, sizeof(xecfg));
2370                memset(&ecbdata, 0, sizeof(ecbdata));
2371                ecbdata.label_path = lbl;
2372                ecbdata.color_diff = want_color(o->use_color);
2373                ecbdata.found_changesp = &o->found_changes;
2374                ecbdata.ws_rule = whitespace_rule(name_b);
2375                if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2376                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2377                ecbdata.opt = o;
2378                ecbdata.header = header.len ? &header : NULL;
2379                xpp.flags = o->xdl_opts;
2380                xecfg.ctxlen = o->context;
2381                xecfg.interhunkctxlen = o->interhunkcontext;
2382                xecfg.flags = XDL_EMIT_FUNCNAMES;
2383                if (DIFF_OPT_TST(o, FUNCCONTEXT))
2384                        xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2385                if (pe)
2386                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2387                if (!diffopts)
2388                        ;
2389                else if (starts_with(diffopts, "--unified="))
2390                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2391                else if (starts_with(diffopts, "-u"))
2392                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2393                if (o->word_diff)
2394                        init_diff_words_data(&ecbdata, o, one, two);
2395                xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2396                              &xpp, &xecfg);
2397                if (o->word_diff)
2398                        free_diff_words_data(&ecbdata);
2399                if (textconv_one)
2400                        free(mf1.ptr);
2401                if (textconv_two)
2402                        free(mf2.ptr);
2403                xdiff_clear_find_func(&xecfg);
2404        }
2405
2406 free_ab_and_return:
2407        strbuf_release(&header);
2408        diff_free_filespec_data(one);
2409        diff_free_filespec_data(two);
2410        free(a_one);
2411        free(b_two);
2412        return;
2413}
2414
2415static void builtin_diffstat(const char *name_a, const char *name_b,
2416                             struct diff_filespec *one,
2417                             struct diff_filespec *two,
2418                             struct diffstat_t *diffstat,
2419                             struct diff_options *o,
2420                             struct diff_filepair *p)
2421{
2422        mmfile_t mf1, mf2;
2423        struct diffstat_file *data;
2424        int same_contents;
2425        int complete_rewrite = 0;
2426
2427        if (!DIFF_PAIR_UNMERGED(p)) {
2428                if (p->status == DIFF_STATUS_MODIFIED && p->score)
2429                        complete_rewrite = 1;
2430        }
2431
2432        data = diffstat_add(diffstat, name_a, name_b);
2433        data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2434
2435        if (!one || !two) {
2436                data->is_unmerged = 1;
2437                return;
2438        }
2439
2440        same_contents = !hashcmp(one->sha1, two->sha1);
2441
2442        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2443                data->is_binary = 1;
2444                if (same_contents) {
2445                        data->added = 0;
2446                        data->deleted = 0;
2447                } else {
2448                        data->added = diff_filespec_size(two);
2449                        data->deleted = diff_filespec_size(one);
2450                }
2451        }
2452
2453        else if (complete_rewrite) {
2454                diff_populate_filespec(one, 0);
2455                diff_populate_filespec(two, 0);
2456                data->deleted = count_lines(one->data, one->size);
2457                data->added = count_lines(two->data, two->size);
2458        }
2459
2460        else if (!same_contents) {
2461                /* Crazy xdl interfaces.. */
2462                xpparam_t xpp;
2463                xdemitconf_t xecfg;
2464
2465                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2466                        die("unable to read files to diff");
2467
2468                memset(&xpp, 0, sizeof(xpp));
2469                memset(&xecfg, 0, sizeof(xecfg));
2470                xpp.flags = o->xdl_opts;
2471                xecfg.ctxlen = o->context;
2472                xecfg.interhunkctxlen = o->interhunkcontext;
2473                xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2474                              &xpp, &xecfg);
2475        }
2476
2477        diff_free_filespec_data(one);
2478        diff_free_filespec_data(two);
2479}
2480
2481static void builtin_checkdiff(const char *name_a, const char *name_b,
2482                              const char *attr_path,
2483                              struct diff_filespec *one,
2484                              struct diff_filespec *two,
2485                              struct diff_options *o)
2486{
2487        mmfile_t mf1, mf2;
2488        struct checkdiff_t data;
2489
2490        if (!two)
2491                return;
2492
2493        memset(&data, 0, sizeof(data));
2494        data.filename = name_b ? name_b : name_a;
2495        data.lineno = 0;
2496        data.o = o;
2497        data.ws_rule = whitespace_rule(attr_path);
2498        data.conflict_marker_size = ll_merge_marker_size(attr_path);
2499
2500        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2501                die("unable to read files to diff");
2502
2503        /*
2504         * All the other codepaths check both sides, but not checking
2505         * the "old" side here is deliberate.  We are checking the newly
2506         * introduced changes, and as long as the "new" side is text, we
2507         * can and should check what it introduces.
2508         */
2509        if (diff_filespec_is_binary(two))
2510                goto free_and_return;
2511        else {
2512                /* Crazy xdl interfaces.. */
2513                xpparam_t xpp;
2514                xdemitconf_t xecfg;
2515
2516                memset(&xpp, 0, sizeof(xpp));
2517                memset(&xecfg, 0, sizeof(xecfg));
2518                xecfg.ctxlen = 1; /* at least one context line */
2519                xpp.flags = 0;
2520                xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2521                              &xpp, &xecfg);
2522
2523                if (data.ws_rule & WS_BLANK_AT_EOF) {
2524                        struct emit_callback ecbdata;
2525                        int blank_at_eof;
2526
2527                        ecbdata.ws_rule = data.ws_rule;
2528                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2529                        blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2530
2531                        if (blank_at_eof) {
2532                                static char *err;
2533                                if (!err)
2534                                        err = whitespace_error_string(WS_BLANK_AT_EOF);
2535                                fprintf(o->file, "%s:%d: %s.\n",
2536                                        data.filename, blank_at_eof, err);
2537                                data.status = 1; /* report errors */
2538                        }
2539                }
2540        }
2541 free_and_return:
2542        diff_free_filespec_data(one);
2543        diff_free_filespec_data(two);
2544        if (data.status)
2545                DIFF_OPT_SET(o, CHECK_FAILED);
2546}
2547
2548struct diff_filespec *alloc_filespec(const char *path)
2549{
2550        int namelen = strlen(path);
2551        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2552
2553        memset(spec, 0, sizeof(*spec));
2554        spec->path = (char *)(spec + 1);
2555        memcpy(spec->path, path, namelen+1);
2556        spec->count = 1;
2557        spec->is_binary = -1;
2558        return spec;
2559}
2560
2561void free_filespec(struct diff_filespec *spec)
2562{
2563        if (!--spec->count) {
2564                diff_free_filespec_data(spec);
2565                free(spec);
2566        }
2567}
2568
2569void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2570                   int sha1_valid, unsigned short mode)
2571{
2572        if (mode) {
2573                spec->mode = canon_mode(mode);
2574                hashcpy(spec->sha1, sha1);
2575                spec->sha1_valid = sha1_valid;
2576        }
2577}
2578
2579/*
2580 * Given a name and sha1 pair, if the index tells us the file in
2581 * the work tree has that object contents, return true, so that
2582 * prepare_temp_file() does not have to inflate and extract.
2583 */
2584static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2585{
2586        const struct cache_entry *ce;
2587        struct stat st;
2588        int pos, len;
2589
2590        /*
2591         * We do not read the cache ourselves here, because the
2592         * benchmark with my previous version that always reads cache
2593         * shows that it makes things worse for diff-tree comparing
2594         * two linux-2.6 kernel trees in an already checked out work
2595         * tree.  This is because most diff-tree comparisons deal with
2596         * only a small number of files, while reading the cache is
2597         * expensive for a large project, and its cost outweighs the
2598         * savings we get by not inflating the object to a temporary
2599         * file.  Practically, this code only helps when we are used
2600         * by diff-cache --cached, which does read the cache before
2601         * calling us.
2602         */
2603        if (!active_cache)
2604                return 0;
2605
2606        /* We want to avoid the working directory if our caller
2607         * doesn't need the data in a normal file, this system
2608         * is rather slow with its stat/open/mmap/close syscalls,
2609         * and the object is contained in a pack file.  The pack
2610         * is probably already open and will be faster to obtain
2611         * the data through than the working directory.  Loose
2612         * objects however would tend to be slower as they need
2613         * to be individually opened and inflated.
2614         */
2615        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2616                return 0;
2617
2618        len = strlen(name);
2619        pos = cache_name_pos(name, len);
2620        if (pos < 0)
2621                return 0;
2622        ce = active_cache[pos];
2623
2624        /*
2625         * This is not the sha1 we are looking for, or
2626         * unreusable because it is not a regular file.
2627         */
2628        if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2629                return 0;
2630
2631        /*
2632         * If ce is marked as "assume unchanged", there is no
2633         * guarantee that work tree matches what we are looking for.
2634         */
2635        if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2636                return 0;
2637
2638        /*
2639         * If ce matches the file in the work tree, we can reuse it.
2640         */
2641        if (ce_uptodate(ce) ||
2642            (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2643                return 1;
2644
2645        return 0;
2646}
2647
2648static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2649{
2650        int len;
2651        char *data = xmalloc(100), *dirty = "";
2652
2653        /* Are we looking at the work tree? */
2654        if (s->dirty_submodule)
2655                dirty = "-dirty";
2656
2657        len = snprintf(data, 100,
2658                       "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2659        s->data = data;
2660        s->size = len;
2661        s->should_free = 1;
2662        if (size_only) {
2663                s->data = NULL;
2664                free(data);
2665        }
2666        return 0;
2667}
2668
2669/*
2670 * While doing rename detection and pickaxe operation, we may need to
2671 * grab the data for the blob (or file) for our own in-core comparison.
2672 * diff_filespec has data and size fields for this purpose.
2673 */
2674int diff_populate_filespec(struct diff_filespec *s, int size_only)
2675{
2676        int err = 0;
2677        /*
2678         * demote FAIL to WARN to allow inspecting the situation
2679         * instead of refusing.
2680         */
2681        enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2682                                    ? SAFE_CRLF_WARN
2683                                    : safe_crlf);
2684
2685        if (!DIFF_FILE_VALID(s))
2686                die("internal error: asking to populate invalid file.");
2687        if (S_ISDIR(s->mode))
2688                return -1;
2689
2690        if (s->data)
2691                return 0;
2692
2693        if (size_only && 0 < s->size)
2694                return 0;
2695
2696        if (S_ISGITLINK(s->mode))
2697                return diff_populate_gitlink(s, size_only);
2698
2699        if (!s->sha1_valid ||
2700            reuse_worktree_file(s->path, s->sha1, 0)) {
2701                struct strbuf buf = STRBUF_INIT;
2702                struct stat st;
2703                int fd;
2704
2705                if (lstat(s->path, &st) < 0) {
2706                        if (errno == ENOENT) {
2707                        err_empty:
2708                                err = -1;
2709                        empty:
2710                                s->data = (char *)"";
2711                                s->size = 0;
2712                                return err;
2713                        }
2714                }
2715                s->size = xsize_t(st.st_size);
2716                if (!s->size)
2717                        goto empty;
2718                if (S_ISLNK(st.st_mode)) {
2719                        struct strbuf sb = STRBUF_INIT;
2720
2721                        if (strbuf_readlink(&sb, s->path, s->size))
2722                                goto err_empty;
2723                        s->size = sb.len;
2724                        s->data = strbuf_detach(&sb, NULL);
2725                        s->should_free = 1;
2726                        return 0;
2727                }
2728                if (size_only)
2729                        return 0;
2730                fd = open(s->path, O_RDONLY);
2731                if (fd < 0)
2732                        goto err_empty;
2733                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2734                close(fd);
2735                s->should_munmap = 1;
2736
2737                /*
2738                 * Convert from working tree format to canonical git format
2739                 */
2740                if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2741                        size_t size = 0;
2742                        munmap(s->data, s->size);
2743                        s->should_munmap = 0;
2744                        s->data = strbuf_detach(&buf, &size);
2745                        s->size = size;
2746                        s->should_free = 1;
2747                }
2748        }
2749        else {
2750                enum object_type type;
2751                if (size_only) {
2752                        type = sha1_object_info(s->sha1, &s->size);
2753                        if (type < 0)
2754                                die("unable to read %s", sha1_to_hex(s->sha1));
2755                } else {
2756                        s->data = read_sha1_file(s->sha1, &type, &s->size);
2757                        if (!s->data)
2758                                die("unable to read %s", sha1_to_hex(s->sha1));
2759                        s->should_free = 1;
2760                }
2761        }
2762        return 0;
2763}
2764
2765void diff_free_filespec_blob(struct diff_filespec *s)
2766{
2767        if (s->should_free)
2768                free(s->data);
2769        else if (s->should_munmap)
2770                munmap(s->data, s->size);
2771
2772        if (s->should_free || s->should_munmap) {
2773                s->should_free = s->should_munmap = 0;
2774                s->data = NULL;
2775        }
2776}
2777
2778void diff_free_filespec_data(struct diff_filespec *s)
2779{
2780        diff_free_filespec_blob(s);
2781        free(s->cnt_data);
2782        s->cnt_data = NULL;
2783}
2784
2785static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2786                           void *blob,
2787                           unsigned long size,
2788                           const unsigned char *sha1,
2789                           int mode)
2790{
2791        int fd;
2792        struct strbuf buf = STRBUF_INIT;
2793        struct strbuf template = STRBUF_INIT;
2794        char *path_dup = xstrdup(path);
2795        const char *base = basename(path_dup);
2796
2797        /* Generate "XXXXXX_basename.ext" */
2798        strbuf_addstr(&template, "XXXXXX_");
2799        strbuf_addstr(&template, base);
2800
2801        fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2802                        strlen(base) + 1);
2803        if (fd < 0)
2804                die_errno("unable to create temp-file");
2805        if (convert_to_working_tree(path,
2806                        (const char *)blob, (size_t)size, &buf)) {
2807                blob = buf.buf;
2808                size = buf.len;
2809        }
2810        if (write_in_full(fd, blob, size) != size)
2811                die_errno("unable to write temp-file");
2812        close(fd);
2813        temp->name = temp->tmp_path;
2814        strcpy(temp->hex, sha1_to_hex(sha1));
2815        temp->hex[40] = 0;
2816        sprintf(temp->mode, "%06o", mode);
2817        strbuf_release(&buf);
2818        strbuf_release(&template);
2819        free(path_dup);
2820}
2821
2822static struct diff_tempfile *prepare_temp_file(const char *name,
2823                struct diff_filespec *one)
2824{
2825        struct diff_tempfile *temp = claim_diff_tempfile();
2826
2827        if (!DIFF_FILE_VALID(one)) {
2828        not_a_valid_file:
2829                /* A '-' entry produces this for file-2, and
2830                 * a '+' entry produces this for file-1.
2831                 */
2832                temp->name = "/dev/null";
2833                strcpy(temp->hex, ".");
2834                strcpy(temp->mode, ".");
2835                return temp;
2836        }
2837
2838        if (!remove_tempfile_installed) {
2839                atexit(remove_tempfile);
2840                sigchain_push_common(remove_tempfile_on_signal);
2841                remove_tempfile_installed = 1;
2842        }
2843
2844        if (!S_ISGITLINK(one->mode) &&
2845            (!one->sha1_valid ||
2846             reuse_worktree_file(name, one->sha1, 1))) {
2847                struct stat st;
2848                if (lstat(name, &st) < 0) {
2849                        if (errno == ENOENT)
2850                                goto not_a_valid_file;
2851                        die_errno("stat(%s)", name);
2852                }
2853                if (S_ISLNK(st.st_mode)) {
2854                        struct strbuf sb = STRBUF_INIT;
2855                        if (strbuf_readlink(&sb, name, st.st_size) < 0)
2856                                die_errno("readlink(%s)", name);
2857                        prep_temp_blob(name, temp, sb.buf, sb.len,
2858                                       (one->sha1_valid ?
2859                                        one->sha1 : null_sha1),
2860                                       (one->sha1_valid ?
2861                                        one->mode : S_IFLNK));
2862                        strbuf_release(&sb);
2863                }
2864                else {
2865                        /* we can borrow from the file in the work tree */
2866                        temp->name = name;
2867                        if (!one->sha1_valid)
2868                                strcpy(temp->hex, sha1_to_hex(null_sha1));
2869                        else
2870                                strcpy(temp->hex, sha1_to_hex(one->sha1));
2871                        /* Even though we may sometimes borrow the
2872                         * contents from the work tree, we always want
2873                         * one->mode.  mode is trustworthy even when
2874                         * !(one->sha1_valid), as long as
2875                         * DIFF_FILE_VALID(one).
2876                         */
2877                        sprintf(temp->mode, "%06o", one->mode);
2878                }
2879                return temp;
2880        }
2881        else {
2882                if (diff_populate_filespec(one, 0))
2883                        die("cannot read data blob for %s", one->path);
2884                prep_temp_blob(name, temp, one->data, one->size,
2885                               one->sha1, one->mode);
2886        }
2887        return temp;
2888}
2889
2890/* An external diff command takes:
2891 *
2892 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2893 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2894 *
2895 */
2896static void run_external_diff(const char *pgm,
2897                              const char *name,
2898                              const char *other,
2899                              struct diff_filespec *one,
2900                              struct diff_filespec *two,
2901                              const char *xfrm_msg,
2902                              int complete_rewrite,
2903                              struct diff_options *o)
2904{
2905        const char *spawn_arg[10];
2906        int retval;
2907        const char **arg = &spawn_arg[0];
2908        struct diff_queue_struct *q = &diff_queued_diff;
2909        const char *env[3] = { NULL };
2910        char env_counter[50];
2911        char env_total[50];
2912
2913        if (one && two) {
2914                struct diff_tempfile *temp_one, *temp_two;
2915                const char *othername = (other ? other : name);
2916                temp_one = prepare_temp_file(name, one);
2917                temp_two = prepare_temp_file(othername, two);
2918                *arg++ = pgm;
2919                *arg++ = name;
2920                *arg++ = temp_one->name;
2921                *arg++ = temp_one->hex;
2922                *arg++ = temp_one->mode;
2923                *arg++ = temp_two->name;
2924                *arg++ = temp_two->hex;
2925                *arg++ = temp_two->mode;
2926                if (other) {
2927                        *arg++ = other;
2928                        *arg++ = xfrm_msg;
2929                }
2930        } else {
2931                *arg++ = pgm;
2932                *arg++ = name;
2933        }
2934        *arg = NULL;
2935        fflush(NULL);
2936
2937        env[0] = env_counter;
2938        snprintf(env_counter, sizeof(env_counter), "GIT_DIFF_PATH_COUNTER=%d",
2939                 ++o->diff_path_counter);
2940        env[1] = env_total;
2941        snprintf(env_total, sizeof(env_total), "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2942
2943        retval = run_command_v_opt_cd_env(spawn_arg, RUN_USING_SHELL, NULL, env);
2944        remove_tempfile();
2945        if (retval) {
2946                fprintf(stderr, "external diff died, stopping at %s.\n", name);
2947                exit(1);
2948        }
2949}
2950
2951static int similarity_index(struct diff_filepair *p)
2952{
2953        return p->score * 100 / MAX_SCORE;
2954}
2955
2956static void fill_metainfo(struct strbuf *msg,
2957                          const char *name,
2958                          const char *other,
2959                          struct diff_filespec *one,
2960                          struct diff_filespec *two,
2961                          struct diff_options *o,
2962                          struct diff_filepair *p,
2963                          int *must_show_header,
2964                          int use_color)
2965{
2966        const char *set = diff_get_color(use_color, DIFF_METAINFO);
2967        const char *reset = diff_get_color(use_color, DIFF_RESET);
2968        const char *line_prefix = diff_line_prefix(o);
2969
2970        *must_show_header = 1;
2971        strbuf_init(msg, PATH_MAX * 2 + 300);
2972        switch (p->status) {
2973        case DIFF_STATUS_COPIED:
2974                strbuf_addf(msg, "%s%ssimilarity index %d%%",
2975                            line_prefix, set, similarity_index(p));
2976                strbuf_addf(msg, "%s\n%s%scopy from ",
2977                            reset,  line_prefix, set);
2978                quote_c_style(name, msg, NULL, 0);
2979                strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2980                quote_c_style(other, msg, NULL, 0);
2981                strbuf_addf(msg, "%s\n", reset);
2982                break;
2983        case DIFF_STATUS_RENAMED:
2984                strbuf_addf(msg, "%s%ssimilarity index %d%%",
2985                            line_prefix, set, similarity_index(p));
2986                strbuf_addf(msg, "%s\n%s%srename from ",
2987                            reset, line_prefix, set);
2988                quote_c_style(name, msg, NULL, 0);
2989                strbuf_addf(msg, "%s\n%s%srename to ",
2990                            reset, line_prefix, set);
2991                quote_c_style(other, msg, NULL, 0);
2992                strbuf_addf(msg, "%s\n", reset);
2993                break;
2994        case DIFF_STATUS_MODIFIED:
2995                if (p->score) {
2996                        strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2997                                    line_prefix,
2998                                    set, similarity_index(p), reset);
2999                        break;
3000                }
3001                /* fallthru */
3002        default:
3003                *must_show_header = 0;
3004        }
3005        if (one && two && hashcmp(one->sha1, two->sha1)) {
3006                int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3007
3008                if (DIFF_OPT_TST(o, BINARY)) {
3009                        mmfile_t mf;
3010                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3011                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3012                                abbrev = 40;
3013                }
3014                strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3015                            find_unique_abbrev(one->sha1, abbrev));
3016                strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3017                if (one->mode == two->mode)
3018                        strbuf_addf(msg, " %06o", one->mode);
3019                strbuf_addf(msg, "%s\n", reset);
3020        }
3021}
3022
3023static void run_diff_cmd(const char *pgm,
3024                         const char *name,
3025                         const char *other,
3026                         const char *attr_path,
3027                         struct diff_filespec *one,
3028                         struct diff_filespec *two,
3029                         struct strbuf *msg,
3030                         struct diff_options *o,
3031                         struct diff_filepair *p)
3032{
3033        const char *xfrm_msg = NULL;
3034        int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3035        int must_show_header = 0;
3036
3037
3038        if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3039                struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3040                if (drv && drv->external)
3041                        pgm = drv->external;
3042        }
3043
3044        if (msg) {
3045                /*
3046                 * don't use colors when the header is intended for an
3047                 * external diff driver
3048                 */
3049                fill_metainfo(msg, name, other, one, two, o, p,
3050                              &must_show_header,
3051                              want_color(o->use_color) && !pgm);
3052                xfrm_msg = msg->len ? msg->buf : NULL;
3053        }
3054
3055        if (pgm) {
3056                run_external_diff(pgm, name, other, one, two, xfrm_msg,
3057                                  complete_rewrite, o);
3058                return;
3059        }
3060        if (one && two)
3061                builtin_diff(name, other ? other : name,
3062                             one, two, xfrm_msg, must_show_header,
3063                             o, complete_rewrite);
3064        else
3065                fprintf(o->file, "* Unmerged path %s\n", name);
3066}
3067
3068static void diff_fill_sha1_info(struct diff_filespec *one)
3069{
3070        if (DIFF_FILE_VALID(one)) {
3071                if (!one->sha1_valid) {
3072                        struct stat st;
3073                        if (one->is_stdin) {
3074                                hashcpy(one->sha1, null_sha1);
3075                                return;
3076                        }
3077                        if (lstat(one->path, &st) < 0)
3078                                die_errno("stat '%s'", one->path);
3079                        if (index_path(one->sha1, one->path, &st, 0))
3080                                die("cannot hash %s", one->path);
3081                }
3082        }
3083        else
3084                hashclr(one->sha1);
3085}
3086
3087static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3088{
3089        /* Strip the prefix but do not molest /dev/null and absolute paths */
3090        if (*namep && **namep != '/') {
3091                *namep += prefix_length;
3092                if (**namep == '/')
3093                        ++*namep;
3094        }
3095        if (*otherp && **otherp != '/') {
3096                *otherp += prefix_length;
3097                if (**otherp == '/')
3098                        ++*otherp;
3099        }
3100}
3101
3102static void run_diff(struct diff_filepair *p, struct diff_options *o)
3103{
3104        const char *pgm = external_diff();
3105        struct strbuf msg;
3106        struct diff_filespec *one = p->one;
3107        struct diff_filespec *two = p->two;
3108        const char *name;
3109        const char *other;
3110        const char *attr_path;
3111
3112        name  = p->one->path;
3113        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3114        attr_path = name;
3115        if (o->prefix_length)
3116                strip_prefix(o->prefix_length, &name, &other);
3117
3118        if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3119                pgm = NULL;
3120
3121        if (DIFF_PAIR_UNMERGED(p)) {
3122                run_diff_cmd(pgm, name, NULL, attr_path,
3123                             NULL, NULL, NULL, o, p);
3124                return;
3125        }
3126
3127        diff_fill_sha1_info(one);
3128        diff_fill_sha1_info(two);
3129
3130        if (!pgm &&
3131            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3132            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3133                /*
3134                 * a filepair that changes between file and symlink
3135                 * needs to be split into deletion and creation.
3136                 */
3137                struct diff_filespec *null = alloc_filespec(two->path);
3138                run_diff_cmd(NULL, name, other, attr_path,
3139                             one, null, &msg, o, p);
3140                free(null);
3141                strbuf_release(&msg);
3142
3143                null = alloc_filespec(one->path);
3144                run_diff_cmd(NULL, name, other, attr_path,
3145                             null, two, &msg, o, p);
3146                free(null);
3147        }
3148        else
3149                run_diff_cmd(pgm, name, other, attr_path,
3150                             one, two, &msg, o, p);
3151
3152        strbuf_release(&msg);
3153}
3154
3155static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3156                         struct diffstat_t *diffstat)
3157{
3158        const char *name;
3159        const char *other;
3160
3161        if (DIFF_PAIR_UNMERGED(p)) {
3162                /* unmerged */
3163                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3164                return;
3165        }
3166
3167        name = p->one->path;
3168        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3169
3170        if (o->prefix_length)
3171                strip_prefix(o->prefix_length, &name, &other);
3172
3173        diff_fill_sha1_info(p->one);
3174        diff_fill_sha1_info(p->two);
3175
3176        builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3177}
3178
3179static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3180{
3181        const char *name;
3182        const char *other;
3183        const char *attr_path;
3184
3185        if (DIFF_PAIR_UNMERGED(p)) {
3186                /* unmerged */
3187                return;
3188        }
3189
3190        name = p->one->path;
3191        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3192        attr_path = other ? other : name;
3193
3194        if (o->prefix_length)
3195                strip_prefix(o->prefix_length, &name, &other);
3196
3197        diff_fill_sha1_info(p->one);
3198        diff_fill_sha1_info(p->two);
3199
3200        builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3201}
3202
3203void diff_setup(struct diff_options *options)
3204{
3205        memcpy(options, &default_diff_options, sizeof(*options));
3206
3207        options->file = stdout;
3208
3209        options->line_termination = '\n';
3210        options->break_opt = -1;
3211        options->rename_limit = -1;
3212        options->dirstat_permille = diff_dirstat_permille_default;
3213        options->context = diff_context_default;
3214        DIFF_OPT_SET(options, RENAME_EMPTY);
3215
3216        options->change = diff_change;
3217        options->add_remove = diff_addremove;
3218        options->use_color = diff_use_color_default;
3219        options->detect_rename = diff_detect_rename_default;
3220        options->xdl_opts |= diff_algorithm;
3221
3222        options->orderfile = diff_order_file_cfg;
3223
3224        if (diff_no_prefix) {
3225                options->a_prefix = options->b_prefix = "";
3226        } else if (!diff_mnemonic_prefix) {
3227                options->a_prefix = "a/";
3228                options->b_prefix = "b/";
3229        }
3230}
3231
3232void diff_setup_done(struct diff_options *options)
3233{
3234        int count = 0;
3235
3236        if (options->set_default)
3237                options->set_default(options);
3238
3239        if (options->output_format & DIFF_FORMAT_NAME)
3240                count++;
3241        if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3242                count++;
3243        if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3244                count++;
3245        if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3246                count++;
3247        if (count > 1)
3248                die("--name-only, --name-status, --check and -s are mutually exclusive");
3249
3250        /*
3251         * Most of the time we can say "there are changes"
3252         * only by checking if there are changed paths, but
3253         * --ignore-whitespace* options force us to look
3254         * inside contents.
3255         */
3256
3257        if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3258            DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3259            DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3260                DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3261        else
3262                DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3263
3264        if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3265                options->detect_rename = DIFF_DETECT_COPY;
3266
3267        if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3268                options->prefix = NULL;
3269        if (options->prefix)
3270                options->prefix_length = strlen(options->prefix);
3271        else
3272                options->prefix_length = 0;
3273
3274        if (options->output_format & (DIFF_FORMAT_NAME |
3275                                      DIFF_FORMAT_NAME_STATUS |
3276                                      DIFF_FORMAT_CHECKDIFF |
3277                                      DIFF_FORMAT_NO_OUTPUT))
3278                options->output_format &= ~(DIFF_FORMAT_RAW |
3279                                            DIFF_FORMAT_NUMSTAT |
3280                                            DIFF_FORMAT_DIFFSTAT |
3281                                            DIFF_FORMAT_SHORTSTAT |
3282                                            DIFF_FORMAT_DIRSTAT |
3283                                            DIFF_FORMAT_SUMMARY |
3284                                            DIFF_FORMAT_PATCH);
3285
3286        /*
3287         * These cases always need recursive; we do not drop caller-supplied
3288         * recursive bits for other formats here.
3289         */
3290        if (options->output_format & (DIFF_FORMAT_PATCH |
3291                                      DIFF_FORMAT_NUMSTAT |
3292                                      DIFF_FORMAT_DIFFSTAT |
3293                                      DIFF_FORMAT_SHORTSTAT |
3294                                      DIFF_FORMAT_DIRSTAT |
3295                                      DIFF_FORMAT_SUMMARY |
3296                                      DIFF_FORMAT_CHECKDIFF))
3297                DIFF_OPT_SET(options, RECURSIVE);
3298        /*
3299         * Also pickaxe would not work very well if you do not say recursive
3300         */
3301        if (options->pickaxe)
3302                DIFF_OPT_SET(options, RECURSIVE);
3303        /*
3304         * When patches are generated, submodules diffed against the work tree
3305         * must be checked for dirtiness too so it can be shown in the output
3306         */
3307        if (options->output_format & DIFF_FORMAT_PATCH)
3308                DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3309
3310        if (options->detect_rename && options->rename_limit < 0)
3311                options->rename_limit = diff_rename_limit_default;
3312        if (options->setup & DIFF_SETUP_USE_CACHE) {
3313                if (!active_cache)
3314                        /* read-cache does not die even when it fails
3315                         * so it is safe for us to do this here.  Also
3316                         * it does not smudge active_cache or active_nr
3317                         * when it fails, so we do not have to worry about
3318                         * cleaning it up ourselves either.
3319                         */
3320                        read_cache();
3321        }
3322        if (options->abbrev <= 0 || 40 < options->abbrev)
3323                options->abbrev = 40; /* full */
3324
3325        /*
3326         * It does not make sense to show the first hit we happened
3327         * to have found.  It does not make sense not to return with
3328         * exit code in such a case either.
3329         */
3330        if (DIFF_OPT_TST(options, QUICK)) {
3331                options->output_format = DIFF_FORMAT_NO_OUTPUT;
3332                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3333        }
3334
3335        options->diff_path_counter = 0;
3336}
3337
3338static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3339{
3340        char c, *eq;
3341        int len;
3342
3343        if (*arg != '-')
3344                return 0;
3345        c = *++arg;
3346        if (!c)
3347                return 0;
3348        if (c == arg_short) {
3349                c = *++arg;
3350                if (!c)
3351                        return 1;
3352                if (val && isdigit(c)) {
3353                        char *end;
3354                        int n = strtoul(arg, &end, 10);
3355                        if (*end)
3356                                return 0;
3357                        *val = n;
3358                        return 1;
3359                }
3360                return 0;
3361        }
3362        if (c != '-')
3363                return 0;
3364        arg++;
3365        eq = strchrnul(arg, '=');
3366        len = eq - arg;
3367        if (!len || strncmp(arg, arg_long, len))
3368                return 0;
3369        if (*eq) {
3370                int n;
3371                char *end;
3372                if (!isdigit(*++eq))
3373                        return 0;
3374                n = strtoul(eq, &end, 10);
3375                if (*end)
3376                        return 0;
3377                *val = n;
3378        }
3379        return 1;
3380}
3381
3382static int diff_scoreopt_parse(const char *opt);
3383
3384static inline int short_opt(char opt, const char **argv,
3385                            const char **optarg)
3386{
3387        const char *arg = argv[0];
3388        if (arg[0] != '-' || arg[1] != opt)
3389                return 0;
3390        if (arg[2] != '\0') {
3391                *optarg = arg + 2;
3392                return 1;
3393        }
3394        if (!argv[1])
3395                die("Option '%c' requires a value", opt);
3396        *optarg = argv[1];
3397        return 2;
3398}
3399
3400int parse_long_opt(const char *opt, const char **argv,
3401                   const char **optarg)
3402{
3403        const char *arg = argv[0];
3404        if (arg[0] != '-' || arg[1] != '-')
3405                return 0;
3406        arg += strlen("--");
3407        if (!starts_with(arg, opt))
3408                return 0;
3409        arg += strlen(opt);
3410        if (*arg == '=') { /* stuck form: --option=value */
3411                *optarg = arg + 1;
3412                return 1;
3413        }
3414        if (*arg != '\0')
3415                return 0;
3416        /* separate form: --option value */
3417        if (!argv[1])
3418                die("Option '--%s' requires a value", opt);
3419        *optarg = argv[1];
3420        return 2;
3421}
3422
3423static int stat_opt(struct diff_options *options, const char **av)
3424{
3425        const char *arg = av[0];
3426        char *end;
3427        int width = options->stat_width;
3428        int name_width = options->stat_name_width;
3429        int graph_width = options->stat_graph_width;
3430        int count = options->stat_count;
3431        int argcount = 1;
3432
3433        arg += strlen("--stat");
3434        end = (char *)arg;
3435
3436        switch (*arg) {
3437        case '-':
3438                if (starts_with(arg, "-width")) {
3439                        arg += strlen("-width");
3440                        if (*arg == '=')
3441                                width = strtoul(arg + 1, &end, 10);
3442                        else if (!*arg && !av[1])
3443                                die("Option '--stat-width' requires a value");
3444                        else if (!*arg) {
3445                                width = strtoul(av[1], &end, 10);
3446                                argcount = 2;
3447                        }
3448                } else if (starts_with(arg, "-name-width")) {
3449                        arg += strlen("-name-width");
3450                        if (*arg == '=')
3451                                name_width = strtoul(arg + 1, &end, 10);
3452                        else if (!*arg && !av[1])
3453                                die("Option '--stat-name-width' requires a value");
3454                        else if (!*arg) {
3455                                name_width = strtoul(av[1], &end, 10);
3456                                argcount = 2;
3457                        }
3458                } else if (starts_with(arg, "-graph-width")) {
3459                        arg += strlen("-graph-width");
3460                        if (*arg == '=')
3461                                graph_width = strtoul(arg + 1, &end, 10);
3462                        else if (!*arg && !av[1])
3463                                die("Option '--stat-graph-width' requires a value");
3464                        else if (!*arg) {
3465                                graph_width = strtoul(av[1], &end, 10);
3466                                argcount = 2;
3467                        }
3468                } else if (starts_with(arg, "-count")) {
3469                        arg += strlen("-count");
3470                        if (*arg == '=')
3471                                count = strtoul(arg + 1, &end, 10);
3472                        else if (!*arg && !av[1])
3473                                die("Option '--stat-count' requires a value");
3474                        else if (!*arg) {
3475                                count = strtoul(av[1], &end, 10);
3476                                argcount = 2;
3477                        }
3478                }
3479                break;
3480        case '=':
3481                width = strtoul(arg+1, &end, 10);
3482                if (*end == ',')
3483                        name_width = strtoul(end+1, &end, 10);
3484                if (*end == ',')
3485                        count = strtoul(end+1, &end, 10);
3486        }
3487
3488        /* Important! This checks all the error cases! */
3489        if (*end)
3490                return 0;
3491        options->output_format |= DIFF_FORMAT_DIFFSTAT;
3492        options->stat_name_width = name_width;
3493        options->stat_graph_width = graph_width;
3494        options->stat_width = width;
3495        options->stat_count = count;
3496        return argcount;
3497}
3498
3499static int parse_dirstat_opt(struct diff_options *options, const char *params)
3500{
3501        struct strbuf errmsg = STRBUF_INIT;
3502        if (parse_dirstat_params(options, params, &errmsg))
3503                die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3504                    errmsg.buf);
3505        strbuf_release(&errmsg);
3506        /*
3507         * The caller knows a dirstat-related option is given from the command
3508         * line; allow it to say "return this_function();"
3509         */
3510        options->output_format |= DIFF_FORMAT_DIRSTAT;
3511        return 1;
3512}
3513
3514static int parse_submodule_opt(struct diff_options *options, const char *value)
3515{
3516        if (parse_submodule_params(options, value))
3517                die(_("Failed to parse --submodule option parameter: '%s'"),
3518                        value);
3519        return 1;
3520}
3521
3522static const char diff_status_letters[] = {
3523        DIFF_STATUS_ADDED,
3524        DIFF_STATUS_COPIED,
3525        DIFF_STATUS_DELETED,
3526        DIFF_STATUS_MODIFIED,
3527        DIFF_STATUS_RENAMED,
3528        DIFF_STATUS_TYPE_CHANGED,
3529        DIFF_STATUS_UNKNOWN,
3530        DIFF_STATUS_UNMERGED,
3531        DIFF_STATUS_FILTER_AON,
3532        DIFF_STATUS_FILTER_BROKEN,
3533        '\0',
3534};
3535
3536static unsigned int filter_bit['Z' + 1];
3537
3538static void prepare_filter_bits(void)
3539{
3540        int i;
3541
3542        if (!filter_bit[DIFF_STATUS_ADDED]) {
3543                for (i = 0; diff_status_letters[i]; i++)
3544                        filter_bit[(int) diff_status_letters[i]] = (1 << i);
3545        }
3546}
3547
3548static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3549{
3550        return opt->filter & filter_bit[(int) status];
3551}
3552
3553static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3554{
3555        int i, optch;
3556
3557        prepare_filter_bits();
3558
3559        /*
3560         * If there is a negation e.g. 'd' in the input, and we haven't
3561         * initialized the filter field with another --diff-filter, start
3562         * from full set of bits, except for AON.
3563         */
3564        if (!opt->filter) {
3565                for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3566                        if (optch < 'a' || 'z' < optch)
3567                                continue;
3568                        opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3569                        opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3570                        break;
3571                }
3572        }
3573
3574        for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3575                unsigned int bit;
3576                int negate;
3577
3578                if ('a' <= optch && optch <= 'z') {
3579                        negate = 1;
3580                        optch = toupper(optch);
3581                } else {
3582                        negate = 0;
3583                }
3584
3585                bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3586                if (!bit)
3587                        return optarg[i];
3588                if (negate)
3589                        opt->filter &= ~bit;
3590                else
3591                        opt->filter |= bit;
3592        }
3593        return 0;
3594}
3595
3596static void enable_patch_output(int *fmt) {
3597        *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3598        *fmt |= DIFF_FORMAT_PATCH;
3599}
3600
3601int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3602{
3603        const char *arg = av[0];
3604        const char *optarg;
3605        int argcount;
3606
3607        /* Output format options */
3608        if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3609            || opt_arg(arg, 'U', "unified", &options->context))
3610                enable_patch_output(&options->output_format);
3611        else if (!strcmp(arg, "--raw"))
3612                options->output_format |= DIFF_FORMAT_RAW;
3613        else if (!strcmp(arg, "--patch-with-raw")) {
3614                enable_patch_output(&options->output_format);
3615                options->output_format |= DIFF_FORMAT_RAW;
3616        } else if (!strcmp(arg, "--numstat"))
3617                options->output_format |= DIFF_FORMAT_NUMSTAT;
3618        else if (!strcmp(arg, "--shortstat"))
3619                options->output_format |= DIFF_FORMAT_SHORTSTAT;
3620        else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3621                return parse_dirstat_opt(options, "");
3622        else if (starts_with(arg, "-X"))
3623                return parse_dirstat_opt(options, arg + 2);
3624        else if (starts_with(arg, "--dirstat="))
3625                return parse_dirstat_opt(options, arg + 10);
3626        else if (!strcmp(arg, "--cumulative"))
3627                return parse_dirstat_opt(options, "cumulative");
3628        else if (!strcmp(arg, "--dirstat-by-file"))
3629                return parse_dirstat_opt(options, "files");
3630        else if (starts_with(arg, "--dirstat-by-file=")) {
3631                parse_dirstat_opt(options, "files");
3632                return parse_dirstat_opt(options, arg + 18);
3633        }
3634        else if (!strcmp(arg, "--check"))
3635                options->output_format |= DIFF_FORMAT_CHECKDIFF;
3636        else if (!strcmp(arg, "--summary"))
3637                options->output_format |= DIFF_FORMAT_SUMMARY;
3638        else if (!strcmp(arg, "--patch-with-stat")) {
3639                enable_patch_output(&options->output_format);
3640                options->output_format |= DIFF_FORMAT_DIFFSTAT;
3641        } else if (!strcmp(arg, "--name-only"))
3642                options->output_format |= DIFF_FORMAT_NAME;
3643        else if (!strcmp(arg, "--name-status"))
3644                options->output_format |= DIFF_FORMAT_NAME_STATUS;
3645        else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3646                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3647        else if (starts_with(arg, "--stat"))
3648                /* --stat, --stat-width, --stat-name-width, or --stat-count */
3649                return stat_opt(options, av);
3650
3651        /* renames options */
3652        else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3653                 !strcmp(arg, "--break-rewrites")) {
3654                if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3655                        return error("invalid argument to -B: %s", arg+2);
3656        }
3657        else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3658                 !strcmp(arg, "--find-renames")) {
3659                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3660                        return error("invalid argument to -M: %s", arg+2);
3661                options->detect_rename = DIFF_DETECT_RENAME;
3662        }
3663        else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3664                options->irreversible_delete = 1;
3665        }
3666        else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3667                 !strcmp(arg, "--find-copies")) {
3668                if (options->detect_rename == DIFF_DETECT_COPY)
3669                        DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3670                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3671                        return error("invalid argument to -C: %s", arg+2);
3672                options->detect_rename = DIFF_DETECT_COPY;
3673        }
3674        else if (!strcmp(arg, "--no-renames"))
3675                options->detect_rename = 0;
3676        else if (!strcmp(arg, "--rename-empty"))
3677                DIFF_OPT_SET(options, RENAME_EMPTY);
3678        else if (!strcmp(arg, "--no-rename-empty"))
3679                DIFF_OPT_CLR(options, RENAME_EMPTY);
3680        else if (!strcmp(arg, "--relative"))
3681                DIFF_OPT_SET(options, RELATIVE_NAME);
3682        else if (starts_with(arg, "--relative=")) {
3683                DIFF_OPT_SET(options, RELATIVE_NAME);
3684                options->prefix = arg + 11;
3685        }
3686
3687        /* xdiff options */
3688        else if (!strcmp(arg, "--minimal"))
3689                DIFF_XDL_SET(options, NEED_MINIMAL);
3690        else if (!strcmp(arg, "--no-minimal"))
3691                DIFF_XDL_CLR(options, NEED_MINIMAL);
3692        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3693                DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3694        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3695                DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3696        else if (!strcmp(arg, "--ignore-space-at-eol"))
3697                DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3698        else if (!strcmp(arg, "--ignore-blank-lines"))
3699                DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3700        else if (!strcmp(arg, "--patience"))
3701                options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3702        else if (!strcmp(arg, "--histogram"))
3703                options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3704        else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3705                long value = parse_algorithm_value(optarg);
3706                if (value < 0)
3707                        return error("option diff-algorithm accepts \"myers\", "
3708                                     "\"minimal\", \"patience\" and \"histogram\"");
3709                /* clear out previous settings */
3710                DIFF_XDL_CLR(options, NEED_MINIMAL);
3711                options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3712                options->xdl_opts |= value;
3713                return argcount;
3714        }
3715
3716        /* flags options */
3717        else if (!strcmp(arg, "--binary")) {
3718                enable_patch_output(&options->output_format);
3719                DIFF_OPT_SET(options, BINARY);
3720        }
3721        else if (!strcmp(arg, "--full-index"))
3722                DIFF_OPT_SET(options, FULL_INDEX);
3723        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3724                DIFF_OPT_SET(options, TEXT);
3725        else if (!strcmp(arg, "-R"))
3726                DIFF_OPT_SET(options, REVERSE_DIFF);
3727        else if (!strcmp(arg, "--find-copies-harder"))
3728                DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3729        else if (!strcmp(arg, "--follow"))
3730                DIFF_OPT_SET(options, FOLLOW_RENAMES);
3731        else if (!strcmp(arg, "--no-follow"))
3732                DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3733        else if (!strcmp(arg, "--color"))
3734                options->use_color = 1;
3735        else if (starts_with(arg, "--color=")) {
3736                int value = git_config_colorbool(NULL, arg+8);
3737                if (value < 0)
3738                        return error("option `color' expects \"always\", \"auto\", or \"never\"");
3739                options->use_color = value;
3740        }
3741        else if (!strcmp(arg, "--no-color"))
3742                options->use_color = 0;
3743        else if (!strcmp(arg, "--color-words")) {
3744                options->use_color = 1;
3745                options->word_diff = DIFF_WORDS_COLOR;
3746        }
3747        else if (starts_with(arg, "--color-words=")) {
3748                options->use_color = 1;
3749                options->word_diff = DIFF_WORDS_COLOR;
3750                options->word_regex = arg + 14;
3751        }
3752        else if (!strcmp(arg, "--word-diff")) {
3753                if (options->word_diff == DIFF_WORDS_NONE)
3754                        options->word_diff = DIFF_WORDS_PLAIN;
3755        }
3756        else if (starts_with(arg, "--word-diff=")) {
3757                const char *type = arg + 12;
3758                if (!strcmp(type, "plain"))
3759                        options->word_diff = DIFF_WORDS_PLAIN;
3760                else if (!strcmp(type, "color")) {
3761                        options->use_color = 1;
3762                        options->word_diff = DIFF_WORDS_COLOR;
3763                }
3764                else if (!strcmp(type, "porcelain"))
3765                        options->word_diff = DIFF_WORDS_PORCELAIN;
3766                else if (!strcmp(type, "none"))
3767                        options->word_diff = DIFF_WORDS_NONE;
3768                else
3769                        die("bad --word-diff argument: %s", type);
3770        }
3771        else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3772                if (options->word_diff == DIFF_WORDS_NONE)
3773                        options->word_diff = DIFF_WORDS_PLAIN;
3774                options->word_regex = optarg;
3775                return argcount;
3776        }
3777        else if (!strcmp(arg, "--exit-code"))
3778                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3779        else if (!strcmp(arg, "--quiet"))
3780                DIFF_OPT_SET(options, QUICK);
3781        else if (!strcmp(arg, "--ext-diff"))
3782                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3783        else if (!strcmp(arg, "--no-ext-diff"))
3784                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3785        else if (!strcmp(arg, "--textconv"))
3786                DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3787        else if (!strcmp(arg, "--no-textconv"))
3788                DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3789        else if (!strcmp(arg, "--ignore-submodules")) {
3790                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3791                handle_ignore_submodules_arg(options, "all");
3792        } else if (starts_with(arg, "--ignore-submodules=")) {
3793                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3794                handle_ignore_submodules_arg(options, arg + 20);
3795        } else if (!strcmp(arg, "--submodule"))
3796                DIFF_OPT_SET(options, SUBMODULE_LOG);
3797        else if (starts_with(arg, "--submodule="))
3798                return parse_submodule_opt(options, arg + 12);
3799
3800        /* misc options */
3801        else if (!strcmp(arg, "-z"))
3802                options->line_termination = 0;
3803        else if ((argcount = short_opt('l', av, &optarg))) {
3804                options->rename_limit = strtoul(optarg, NULL, 10);
3805                return argcount;
3806        }
3807        else if ((argcount = short_opt('S', av, &optarg))) {
3808                options->pickaxe = optarg;
3809                options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3810                return argcount;
3811        } else if ((argcount = short_opt('G', av, &optarg))) {
3812                options->pickaxe = optarg;
3813                options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3814                return argcount;
3815        }
3816        else if (!strcmp(arg, "--pickaxe-all"))
3817                options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3818        else if (!strcmp(arg, "--pickaxe-regex"))
3819                options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3820        else if ((argcount = short_opt('O', av, &optarg))) {
3821                options->orderfile = optarg;
3822                return argcount;
3823        }
3824        else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3825                int offending = parse_diff_filter_opt(optarg, options);
3826                if (offending)
3827                        die("unknown change class '%c' in --diff-filter=%s",
3828                            offending, optarg);
3829                return argcount;
3830        }
3831        else if (!strcmp(arg, "--abbrev"))
3832                options->abbrev = DEFAULT_ABBREV;
3833        else if (starts_with(arg, "--abbrev=")) {
3834                options->abbrev = strtoul(arg + 9, NULL, 10);
3835                if (options->abbrev < MINIMUM_ABBREV)
3836                        options->abbrev = MINIMUM_ABBREV;
3837                else if (40 < options->abbrev)
3838                        options->abbrev = 40;
3839        }
3840        else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3841                options->a_prefix = optarg;
3842                return argcount;
3843        }
3844        else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3845                options->b_prefix = optarg;
3846                return argcount;
3847        }
3848        else if (!strcmp(arg, "--no-prefix"))
3849                options->a_prefix = options->b_prefix = "";
3850        else if (opt_arg(arg, '\0', "inter-hunk-context",
3851                         &options->interhunkcontext))
3852                ;
3853        else if (!strcmp(arg, "-W"))
3854                DIFF_OPT_SET(options, FUNCCONTEXT);
3855        else if (!strcmp(arg, "--function-context"))
3856                DIFF_OPT_SET(options, FUNCCONTEXT);
3857        else if (!strcmp(arg, "--no-function-context"))
3858                DIFF_OPT_CLR(options, FUNCCONTEXT);
3859        else if ((argcount = parse_long_opt("output", av, &optarg))) {
3860                options->file = fopen(optarg, "w");
3861                if (!options->file)
3862                        die_errno("Could not open '%s'", optarg);
3863                options->close_file = 1;
3864                return argcount;
3865        } else
3866                return 0;
3867        return 1;
3868}
3869
3870int parse_rename_score(const char **cp_p)
3871{
3872        unsigned long num, scale;
3873        int ch, dot;
3874        const char *cp = *cp_p;
3875
3876        num = 0;
3877        scale = 1;
3878        dot = 0;
3879        for (;;) {
3880                ch = *cp;
3881                if ( !dot && ch == '.' ) {
3882                        scale = 1;
3883                        dot = 1;
3884                } else if ( ch == '%' ) {
3885                        scale = dot ? scale*100 : 100;
3886                        cp++;   /* % is always at the end */
3887                        break;
3888                } else if ( ch >= '0' && ch <= '9' ) {
3889                        if ( scale < 100000 ) {
3890                                scale *= 10;
3891                                num = (num*10) + (ch-'0');
3892                        }
3893                } else {
3894                        break;
3895                }
3896                cp++;
3897        }
3898        *cp_p = cp;
3899
3900        /* user says num divided by scale and we say internally that
3901         * is MAX_SCORE * num / scale.
3902         */
3903        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3904}
3905
3906static int diff_scoreopt_parse(const char *opt)
3907{
3908        int opt1, opt2, cmd;
3909
3910        if (*opt++ != '-')
3911                return -1;
3912        cmd = *opt++;
3913        if (cmd == '-') {
3914                /* convert the long-form arguments into short-form versions */
3915                if (starts_with(opt, "break-rewrites")) {
3916                        opt += strlen("break-rewrites");
3917                        if (*opt == 0 || *opt++ == '=')
3918                                cmd = 'B';
3919                } else if (starts_with(opt, "find-copies")) {
3920                        opt += strlen("find-copies");
3921                        if (*opt == 0 || *opt++ == '=')
3922                                cmd = 'C';
3923                } else if (starts_with(opt, "find-renames")) {
3924                        opt += strlen("find-renames");
3925                        if (*opt == 0 || *opt++ == '=')
3926                                cmd = 'M';
3927                }
3928        }
3929        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3930                return -1; /* that is not a -M, -C nor -B option */
3931
3932        opt1 = parse_rename_score(&opt);
3933        if (cmd != 'B')
3934                opt2 = 0;
3935        else {
3936                if (*opt == 0)
3937                        opt2 = 0;
3938                else if (*opt != '/')
3939                        return -1; /* we expect -B80/99 or -B80 */
3940                else {
3941                        opt++;
3942                        opt2 = parse_rename_score(&opt);
3943                }
3944        }
3945        if (*opt != 0)
3946                return -1;
3947        return opt1 | (opt2 << 16);
3948}
3949
3950struct diff_queue_struct diff_queued_diff;
3951
3952void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3953{
3954        ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
3955        queue->queue[queue->nr++] = dp;
3956}
3957
3958struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3959                                 struct diff_filespec *one,
3960                                 struct diff_filespec *two)
3961{
3962        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3963        dp->one = one;
3964        dp->two = two;
3965        if (queue)
3966                diff_q(queue, dp);
3967        return dp;
3968}
3969
3970void diff_free_filepair(struct diff_filepair *p)
3971{
3972        free_filespec(p->one);
3973        free_filespec(p->two);
3974        free(p);
3975}
3976
3977/* This is different from find_unique_abbrev() in that
3978 * it stuffs the result with dots for alignment.
3979 */
3980const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3981{
3982        int abblen;
3983        const char *abbrev;
3984        if (len == 40)
3985                return sha1_to_hex(sha1);
3986
3987        abbrev = find_unique_abbrev(sha1, len);
3988        abblen = strlen(abbrev);
3989        if (abblen < 37) {
3990                static char hex[41];
3991                if (len < abblen && abblen <= len + 2)
3992                        sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3993                else
3994                        sprintf(hex, "%s...", abbrev);
3995                return hex;
3996        }
3997        return sha1_to_hex(sha1);
3998}
3999
4000static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4001{
4002        int line_termination = opt->line_termination;
4003        int inter_name_termination = line_termination ? '\t' : '\0';
4004
4005        fprintf(opt->file, "%s", diff_line_prefix(opt));
4006        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4007                fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4008                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
4009                fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4010        }
4011        if (p->score) {
4012                fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4013                        inter_name_termination);
4014        } else {
4015                fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4016        }
4017
4018        if (p->status == DIFF_STATUS_COPIED ||
4019            p->status == DIFF_STATUS_RENAMED) {
4020                const char *name_a, *name_b;
4021                name_a = p->one->path;
4022                name_b = p->two->path;
4023                strip_prefix(opt->prefix_length, &name_a, &name_b);
4024                write_name_quoted(name_a, opt->file, inter_name_termination);
4025                write_name_quoted(name_b, opt->file, line_termination);
4026        } else {
4027                const char *name_a, *name_b;
4028                name_a = p->one->mode ? p->one->path : p->two->path;
4029                name_b = NULL;
4030                strip_prefix(opt->prefix_length, &name_a, &name_b);
4031                write_name_quoted(name_a, opt->file, line_termination);
4032        }
4033}
4034
4035int diff_unmodified_pair(struct diff_filepair *p)
4036{
4037        /* This function is written stricter than necessary to support
4038         * the currently implemented transformers, but the idea is to
4039         * let transformers to produce diff_filepairs any way they want,
4040         * and filter and clean them up here before producing the output.
4041         */
4042        struct diff_filespec *one = p->one, *two = p->two;
4043
4044        if (DIFF_PAIR_UNMERGED(p))
4045                return 0; /* unmerged is interesting */
4046
4047        /* deletion, addition, mode or type change
4048         * and rename are all interesting.
4049         */
4050        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4051            DIFF_PAIR_MODE_CHANGED(p) ||
4052            strcmp(one->path, two->path))
4053                return 0;
4054
4055        /* both are valid and point at the same path.  that is, we are
4056         * dealing with a change.
4057         */
4058        if (one->sha1_valid && two->sha1_valid &&
4059            !hashcmp(one->sha1, two->sha1) &&
4060            !one->dirty_submodule && !two->dirty_submodule)
4061                return 1; /* no change */
4062        if (!one->sha1_valid && !two->sha1_valid)
4063                return 1; /* both look at the same file on the filesystem. */
4064        return 0;
4065}
4066
4067static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4068{
4069        if (diff_unmodified_pair(p))
4070                return;
4071
4072        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4073            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4074                return; /* no tree diffs in patch format */
4075
4076        run_diff(p, o);
4077}
4078
4079static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4080                            struct diffstat_t *diffstat)
4081{
4082        if (diff_unmodified_pair(p))
4083                return;
4084
4085        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4086            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4087                return; /* no useful stat for tree diffs */
4088
4089        run_diffstat(p, o, diffstat);
4090}
4091
4092static void diff_flush_checkdiff(struct diff_filepair *p,
4093                struct diff_options *o)
4094{
4095        if (diff_unmodified_pair(p))
4096                return;
4097
4098        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4099            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4100                return; /* nothing to check in tree diffs */
4101
4102        run_checkdiff(p, o);
4103}
4104
4105int diff_queue_is_empty(void)
4106{
4107        struct diff_queue_struct *q = &diff_queued_diff;
4108        int i;
4109        for (i = 0; i < q->nr; i++)
4110                if (!diff_unmodified_pair(q->queue[i]))
4111                        return 0;
4112        return 1;
4113}
4114
4115#if DIFF_DEBUG
4116void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4117{
4118        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4119                x, one ? one : "",
4120                s->path,
4121                DIFF_FILE_VALID(s) ? "valid" : "invalid",
4122                s->mode,
4123                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4124        fprintf(stderr, "queue[%d] %s size %lu\n",
4125                x, one ? one : "",
4126                s->size);
4127}
4128
4129void diff_debug_filepair(const struct diff_filepair *p, int i)
4130{
4131        diff_debug_filespec(p->one, i, "one");
4132        diff_debug_filespec(p->two, i, "two");
4133        fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4134                p->score, p->status ? p->status : '?',
4135                p->one->rename_used, p->broken_pair);
4136}
4137
4138void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4139{
4140        int i;
4141        if (msg)
4142                fprintf(stderr, "%s\n", msg);
4143        fprintf(stderr, "q->nr = %d\n", q->nr);
4144        for (i = 0; i < q->nr; i++) {
4145                struct diff_filepair *p = q->queue[i];
4146                diff_debug_filepair(p, i);
4147        }
4148}
4149#endif
4150
4151static void diff_resolve_rename_copy(void)
4152{
4153        int i;
4154        struct diff_filepair *p;
4155        struct diff_queue_struct *q = &diff_queued_diff;
4156
4157        diff_debug_queue("resolve-rename-copy", q);
4158
4159        for (i = 0; i < q->nr; i++) {
4160                p = q->queue[i];
4161                p->status = 0; /* undecided */
4162                if (DIFF_PAIR_UNMERGED(p))
4163                        p->status = DIFF_STATUS_UNMERGED;
4164                else if (!DIFF_FILE_VALID(p->one))
4165                        p->status = DIFF_STATUS_ADDED;
4166                else if (!DIFF_FILE_VALID(p->two))
4167                        p->status = DIFF_STATUS_DELETED;
4168                else if (DIFF_PAIR_TYPE_CHANGED(p))
4169                        p->status = DIFF_STATUS_TYPE_CHANGED;
4170
4171                /* from this point on, we are dealing with a pair
4172                 * whose both sides are valid and of the same type, i.e.
4173                 * either in-place edit or rename/copy edit.
4174                 */
4175                else if (DIFF_PAIR_RENAME(p)) {
4176                        /*
4177                         * A rename might have re-connected a broken
4178                         * pair up, causing the pathnames to be the
4179                         * same again. If so, that's not a rename at
4180                         * all, just a modification..
4181                         *
4182                         * Otherwise, see if this source was used for
4183                         * multiple renames, in which case we decrement
4184                         * the count, and call it a copy.
4185                         */
4186                        if (!strcmp(p->one->path, p->two->path))
4187                                p->status = DIFF_STATUS_MODIFIED;
4188                        else if (--p->one->rename_used > 0)
4189                                p->status = DIFF_STATUS_COPIED;
4190                        else
4191                                p->status = DIFF_STATUS_RENAMED;
4192                }
4193                else if (hashcmp(p->one->sha1, p->two->sha1) ||
4194                         p->one->mode != p->two->mode ||
4195                         p->one->dirty_submodule ||
4196                         p->two->dirty_submodule ||
4197                         is_null_sha1(p->one->sha1))
4198                        p->status = DIFF_STATUS_MODIFIED;
4199                else {
4200                        /* This is a "no-change" entry and should not
4201                         * happen anymore, but prepare for broken callers.
4202                         */
4203                        error("feeding unmodified %s to diffcore",
4204                              p->one->path);
4205                        p->status = DIFF_STATUS_UNKNOWN;
4206                }
4207        }
4208        diff_debug_queue("resolve-rename-copy done", q);
4209}
4210
4211static int check_pair_status(struct diff_filepair *p)
4212{
4213        switch (p->status) {
4214        case DIFF_STATUS_UNKNOWN:
4215                return 0;
4216        case 0:
4217                die("internal error in diff-resolve-rename-copy");
4218        default:
4219                return 1;
4220        }
4221}
4222
4223static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4224{
4225        int fmt = opt->output_format;
4226
4227        if (fmt & DIFF_FORMAT_CHECKDIFF)
4228                diff_flush_checkdiff(p, opt);
4229        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4230                diff_flush_raw(p, opt);
4231        else if (fmt & DIFF_FORMAT_NAME) {
4232                const char *name_a, *name_b;
4233                name_a = p->two->path;
4234                name_b = NULL;
4235                strip_prefix(opt->prefix_length, &name_a, &name_b);
4236                write_name_quoted(name_a, opt->file, opt->line_termination);
4237        }
4238}
4239
4240static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4241{
4242        if (fs->mode)
4243                fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4244        else
4245                fprintf(file, " %s ", newdelete);
4246        write_name_quoted(fs->path, file, '\n');
4247}
4248
4249
4250static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4251                const char *line_prefix)
4252{
4253        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4254                fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4255                        p->two->mode, show_name ? ' ' : '\n');
4256                if (show_name) {
4257                        write_name_quoted(p->two->path, file, '\n');
4258                }
4259        }
4260}
4261
4262static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4263                        const char *line_prefix)
4264{
4265        char *names = pprint_rename(p->one->path, p->two->path);
4266
4267        fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4268        free(names);
4269        show_mode_change(file, p, 0, line_prefix);
4270}
4271
4272static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4273{
4274        FILE *file = opt->file;
4275        const char *line_prefix = diff_line_prefix(opt);
4276
4277        switch(p->status) {
4278        case DIFF_STATUS_DELETED:
4279                fputs(line_prefix, file);
4280                show_file_mode_name(file, "delete", p->one);
4281                break;
4282        case DIFF_STATUS_ADDED:
4283                fputs(line_prefix, file);
4284                show_file_mode_name(file, "create", p->two);
4285                break;
4286        case DIFF_STATUS_COPIED:
4287                fputs(line_prefix, file);
4288                show_rename_copy(file, "copy", p, line_prefix);
4289                break;
4290        case DIFF_STATUS_RENAMED:
4291                fputs(line_prefix, file);
4292                show_rename_copy(file, "rename", p, line_prefix);
4293                break;
4294        default:
4295                if (p->score) {
4296                        fprintf(file, "%s rewrite ", line_prefix);
4297                        write_name_quoted(p->two->path, file, ' ');
4298                        fprintf(file, "(%d%%)\n", similarity_index(p));
4299                }
4300                show_mode_change(file, p, !p->score, line_prefix);
4301                break;
4302        }
4303}
4304
4305struct patch_id_t {
4306        git_SHA_CTX *ctx;
4307        int patchlen;
4308};
4309
4310static int remove_space(char *line, int len)
4311{
4312        int i;
4313        char *dst = line;
4314        unsigned char c;
4315
4316        for (i = 0; i < len; i++)
4317                if (!isspace((c = line[i])))
4318                        *dst++ = c;
4319
4320        return dst - line;
4321}
4322
4323static void patch_id_consume(void *priv, char *line, unsigned long len)
4324{
4325        struct patch_id_t *data = priv;
4326        int new_len;
4327
4328        /* Ignore line numbers when computing the SHA1 of the patch */
4329        if (starts_with(line, "@@ -"))
4330                return;
4331
4332        new_len = remove_space(line, len);
4333
4334        git_SHA1_Update(data->ctx, line, new_len);
4335        data->patchlen += new_len;
4336}
4337
4338/* returns 0 upon success, and writes result into sha1 */
4339static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4340{
4341        struct diff_queue_struct *q = &diff_queued_diff;
4342        int i;
4343        git_SHA_CTX ctx;
4344        struct patch_id_t data;
4345        char buffer[PATH_MAX * 4 + 20];
4346
4347        git_SHA1_Init(&ctx);
4348        memset(&data, 0, sizeof(struct patch_id_t));
4349        data.ctx = &ctx;
4350
4351        for (i = 0; i < q->nr; i++) {
4352                xpparam_t xpp;
4353                xdemitconf_t xecfg;
4354                mmfile_t mf1, mf2;
4355                struct diff_filepair *p = q->queue[i];
4356                int len1, len2;
4357
4358                memset(&xpp, 0, sizeof(xpp));
4359                memset(&xecfg, 0, sizeof(xecfg));
4360                if (p->status == 0)
4361                        return error("internal diff status error");
4362                if (p->status == DIFF_STATUS_UNKNOWN)
4363                        continue;
4364                if (diff_unmodified_pair(p))
4365                        continue;
4366                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4367                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4368                        continue;
4369                if (DIFF_PAIR_UNMERGED(p))
4370                        continue;
4371
4372                diff_fill_sha1_info(p->one);
4373                diff_fill_sha1_info(p->two);
4374                if (fill_mmfile(&mf1, p->one) < 0 ||
4375                                fill_mmfile(&mf2, p->two) < 0)
4376                        return error("unable to read files to diff");
4377
4378                len1 = remove_space(p->one->path, strlen(p->one->path));
4379                len2 = remove_space(p->two->path, strlen(p->two->path));
4380                if (p->one->mode == 0)
4381                        len1 = snprintf(buffer, sizeof(buffer),
4382                                        "diff--gita/%.*sb/%.*s"
4383                                        "newfilemode%06o"
4384                                        "---/dev/null"
4385                                        "+++b/%.*s",
4386                                        len1, p->one->path,
4387                                        len2, p->two->path,
4388                                        p->two->mode,
4389                                        len2, p->two->path);
4390                else if (p->two->mode == 0)
4391                        len1 = snprintf(buffer, sizeof(buffer),
4392                                        "diff--gita/%.*sb/%.*s"
4393                                        "deletedfilemode%06o"
4394                                        "---a/%.*s"
4395                                        "+++/dev/null",
4396                                        len1, p->one->path,
4397                                        len2, p->two->path,
4398                                        p->one->mode,
4399                                        len1, p->one->path);
4400                else
4401                        len1 = snprintf(buffer, sizeof(buffer),
4402                                        "diff--gita/%.*sb/%.*s"
4403                                        "---a/%.*s"
4404                                        "+++b/%.*s",
4405                                        len1, p->one->path,
4406                                        len2, p->two->path,
4407                                        len1, p->one->path,
4408                                        len2, p->two->path);
4409                git_SHA1_Update(&ctx, buffer, len1);
4410
4411                if (diff_filespec_is_binary(p->one) ||
4412                    diff_filespec_is_binary(p->two)) {
4413                        git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4414                        git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4415                        continue;
4416                }
4417
4418                xpp.flags = 0;
4419                xecfg.ctxlen = 3;
4420                xecfg.flags = 0;
4421                xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4422                              &xpp, &xecfg);
4423        }
4424
4425        git_SHA1_Final(sha1, &ctx);
4426        return 0;
4427}
4428
4429int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4430{
4431        struct diff_queue_struct *q = &diff_queued_diff;
4432        int i;
4433        int result = diff_get_patch_id(options, sha1);
4434
4435        for (i = 0; i < q->nr; i++)
4436                diff_free_filepair(q->queue[i]);
4437
4438        free(q->queue);
4439        DIFF_QUEUE_CLEAR(q);
4440
4441        return result;
4442}
4443
4444static int is_summary_empty(const struct diff_queue_struct *q)
4445{
4446        int i;
4447
4448        for (i = 0; i < q->nr; i++) {
4449                const struct diff_filepair *p = q->queue[i];
4450
4451                switch (p->status) {
4452                case DIFF_STATUS_DELETED:
4453                case DIFF_STATUS_ADDED:
4454                case DIFF_STATUS_COPIED:
4455                case DIFF_STATUS_RENAMED:
4456                        return 0;
4457                default:
4458                        if (p->score)
4459                                return 0;
4460                        if (p->one->mode && p->two->mode &&
4461                            p->one->mode != p->two->mode)
4462                                return 0;
4463                        break;
4464                }
4465        }
4466        return 1;
4467}
4468
4469static const char rename_limit_warning[] =
4470"inexact rename detection was skipped due to too many files.";
4471
4472static const char degrade_cc_to_c_warning[] =
4473"only found copies from modified paths due to too many files.";
4474
4475static const char rename_limit_advice[] =
4476"you may want to set your %s variable to at least "
4477"%d and retry the command.";
4478
4479void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4480{
4481        if (degraded_cc)
4482                warning(degrade_cc_to_c_warning);
4483        else if (needed)
4484                warning(rename_limit_warning);
4485        else
4486                return;
4487        if (0 < needed && needed < 32767)
4488                warning(rename_limit_advice, varname, needed);
4489}
4490
4491void diff_flush(struct diff_options *options)
4492{
4493        struct diff_queue_struct *q = &diff_queued_diff;
4494        int i, output_format = options->output_format;
4495        int separator = 0;
4496        int dirstat_by_line = 0;
4497
4498        /*
4499         * Order: raw, stat, summary, patch
4500         * or:    name/name-status/checkdiff (other bits clear)
4501         */
4502        if (!q->nr)
4503                goto free_queue;
4504
4505        if (output_format & (DIFF_FORMAT_RAW |
4506                             DIFF_FORMAT_NAME |
4507                             DIFF_FORMAT_NAME_STATUS |
4508                             DIFF_FORMAT_CHECKDIFF)) {
4509                for (i = 0; i < q->nr; i++) {
4510                        struct diff_filepair *p = q->queue[i];
4511                        if (check_pair_status(p))
4512                                flush_one_pair(p, options);
4513                }
4514                separator++;
4515        }
4516
4517        if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4518                dirstat_by_line = 1;
4519
4520        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4521            dirstat_by_line) {
4522                struct diffstat_t diffstat;
4523
4524                memset(&diffstat, 0, sizeof(struct diffstat_t));
4525                for (i = 0; i < q->nr; i++) {
4526                        struct diff_filepair *p = q->queue[i];
4527                        if (check_pair_status(p))
4528                                diff_flush_stat(p, options, &diffstat);
4529                }
4530                if (output_format & DIFF_FORMAT_NUMSTAT)
4531                        show_numstat(&diffstat, options);
4532                if (output_format & DIFF_FORMAT_DIFFSTAT)
4533                        show_stats(&diffstat, options);
4534                if (output_format & DIFF_FORMAT_SHORTSTAT)
4535                        show_shortstats(&diffstat, options);
4536                if (output_format & DIFF_FORMAT_DIRSTAT)
4537                        show_dirstat_by_line(&diffstat, options);
4538                free_diffstat_info(&diffstat);
4539                separator++;
4540        }
4541        if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4542                show_dirstat(options);
4543
4544        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4545                for (i = 0; i < q->nr; i++) {
4546                        diff_summary(options, q->queue[i]);
4547                }
4548                separator++;
4549        }
4550
4551        if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4552            DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4553            DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4554                /*
4555                 * run diff_flush_patch for the exit status. setting
4556                 * options->file to /dev/null should be safe, because we
4557                 * aren't supposed to produce any output anyway.
4558                 */
4559                if (options->close_file)
4560                        fclose(options->file);
4561                options->file = fopen("/dev/null", "w");
4562                if (!options->file)
4563                        die_errno("Could not open /dev/null");
4564                options->close_file = 1;
4565                for (i = 0; i < q->nr; i++) {
4566                        struct diff_filepair *p = q->queue[i];
4567                        if (check_pair_status(p))
4568                                diff_flush_patch(p, options);
4569                        if (options->found_changes)
4570                                break;
4571                }
4572        }
4573
4574        if (output_format & DIFF_FORMAT_PATCH) {
4575                if (separator) {
4576                        fprintf(options->file, "%s%c",
4577                                diff_line_prefix(options),
4578                                options->line_termination);
4579                        if (options->stat_sep) {
4580                                /* attach patch instead of inline */
4581                                fputs(options->stat_sep, options->file);
4582                        }
4583                }
4584
4585                for (i = 0; i < q->nr; i++) {
4586                        struct diff_filepair *p = q->queue[i];
4587                        if (check_pair_status(p))
4588                                diff_flush_patch(p, options);
4589                }
4590        }
4591
4592        if (output_format & DIFF_FORMAT_CALLBACK)
4593                options->format_callback(q, options, options->format_callback_data);
4594
4595        for (i = 0; i < q->nr; i++)
4596                diff_free_filepair(q->queue[i]);
4597free_queue:
4598        free(q->queue);
4599        DIFF_QUEUE_CLEAR(q);
4600        if (options->close_file)
4601                fclose(options->file);
4602
4603        /*
4604         * Report the content-level differences with HAS_CHANGES;
4605         * diff_addremove/diff_change does not set the bit when
4606         * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4607         */
4608        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4609                if (options->found_changes)
4610                        DIFF_OPT_SET(options, HAS_CHANGES);
4611                else
4612                        DIFF_OPT_CLR(options, HAS_CHANGES);
4613        }
4614}
4615
4616static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4617{
4618        return (((p->status == DIFF_STATUS_MODIFIED) &&
4619                 ((p->score &&
4620                   filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4621                  (!p->score &&
4622                   filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4623                ((p->status != DIFF_STATUS_MODIFIED) &&
4624                 filter_bit_tst(p->status, options)));
4625}
4626
4627static void diffcore_apply_filter(struct diff_options *options)
4628{
4629        int i;
4630        struct diff_queue_struct *q = &diff_queued_diff;
4631        struct diff_queue_struct outq;
4632
4633        DIFF_QUEUE_CLEAR(&outq);
4634
4635        if (!options->filter)
4636                return;
4637
4638        if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4639                int found;
4640                for (i = found = 0; !found && i < q->nr; i++) {
4641                        if (match_filter(options, q->queue[i]))
4642                                found++;
4643                }
4644                if (found)
4645                        return;
4646
4647                /* otherwise we will clear the whole queue
4648                 * by copying the empty outq at the end of this
4649                 * function, but first clear the current entries
4650                 * in the queue.
4651                 */
4652                for (i = 0; i < q->nr; i++)
4653                        diff_free_filepair(q->queue[i]);
4654        }
4655        else {
4656                /* Only the matching ones */
4657                for (i = 0; i < q->nr; i++) {
4658                        struct diff_filepair *p = q->queue[i];
4659                        if (match_filter(options, p))
4660                                diff_q(&outq, p);
4661                        else
4662                                diff_free_filepair(p);
4663                }
4664        }
4665        free(q->queue);
4666        *q = outq;
4667}
4668
4669/* Check whether two filespecs with the same mode and size are identical */
4670static int diff_filespec_is_identical(struct diff_filespec *one,
4671                                      struct diff_filespec *two)
4672{
4673        if (S_ISGITLINK(one->mode))
4674                return 0;
4675        if (diff_populate_filespec(one, 0))
4676                return 0;
4677        if (diff_populate_filespec(two, 0))
4678                return 0;
4679        return !memcmp(one->data, two->data, one->size);
4680}
4681
4682static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4683{
4684        if (p->done_skip_stat_unmatch)
4685                return p->skip_stat_unmatch_result;
4686
4687        p->done_skip_stat_unmatch = 1;
4688        p->skip_stat_unmatch_result = 0;
4689        /*
4690         * 1. Entries that come from stat info dirtiness
4691         *    always have both sides (iow, not create/delete),
4692         *    one side of the object name is unknown, with
4693         *    the same mode and size.  Keep the ones that
4694         *    do not match these criteria.  They have real
4695         *    differences.
4696         *
4697         * 2. At this point, the file is known to be modified,
4698         *    with the same mode and size, and the object
4699         *    name of one side is unknown.  Need to inspect
4700         *    the identical contents.
4701         */
4702        if (!DIFF_FILE_VALID(p->one) || /* (1) */
4703            !DIFF_FILE_VALID(p->two) ||
4704            (p->one->sha1_valid && p->two->sha1_valid) ||
4705            (p->one->mode != p->two->mode) ||
4706            diff_populate_filespec(p->one, 1) ||
4707            diff_populate_filespec(p->two, 1) ||
4708            (p->one->size != p->two->size) ||
4709            !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4710                p->skip_stat_unmatch_result = 1;
4711        return p->skip_stat_unmatch_result;
4712}
4713
4714static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4715{
4716        int i;
4717        struct diff_queue_struct *q = &diff_queued_diff;
4718        struct diff_queue_struct outq;
4719        DIFF_QUEUE_CLEAR(&outq);
4720
4721        for (i = 0; i < q->nr; i++) {
4722                struct diff_filepair *p = q->queue[i];
4723
4724                if (diff_filespec_check_stat_unmatch(p))
4725                        diff_q(&outq, p);
4726                else {
4727                        /*
4728                         * The caller can subtract 1 from skip_stat_unmatch
4729                         * to determine how many paths were dirty only
4730                         * due to stat info mismatch.
4731                         */
4732                        if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4733                                diffopt->skip_stat_unmatch++;
4734                        diff_free_filepair(p);
4735                }
4736        }
4737        free(q->queue);
4738        *q = outq;
4739}
4740
4741static int diffnamecmp(const void *a_, const void *b_)
4742{
4743        const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4744        const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4745        const char *name_a, *name_b;
4746
4747        name_a = a->one ? a->one->path : a->two->path;
4748        name_b = b->one ? b->one->path : b->two->path;
4749        return strcmp(name_a, name_b);
4750}
4751
4752void diffcore_fix_diff_index(struct diff_options *options)
4753{
4754        struct diff_queue_struct *q = &diff_queued_diff;
4755        qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4756}
4757
4758void diffcore_std(struct diff_options *options)
4759{
4760        if (options->skip_stat_unmatch)
4761                diffcore_skip_stat_unmatch(options);
4762        if (!options->found_follow) {
4763                /* See try_to_follow_renames() in tree-diff.c */
4764                if (options->break_opt != -1)
4765                        diffcore_break(options->break_opt);
4766                if (options->detect_rename)
4767                        diffcore_rename(options);
4768                if (options->break_opt != -1)
4769                        diffcore_merge_broken();
4770        }
4771        if (options->pickaxe)
4772                diffcore_pickaxe(options);
4773        if (options->orderfile)
4774                diffcore_order(options->orderfile);
4775        if (!options->found_follow)
4776                /* See try_to_follow_renames() in tree-diff.c */
4777                diff_resolve_rename_copy();
4778        diffcore_apply_filter(options);
4779
4780        if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4781                DIFF_OPT_SET(options, HAS_CHANGES);
4782        else
4783                DIFF_OPT_CLR(options, HAS_CHANGES);
4784
4785        options->found_follow = 0;
4786}
4787
4788int diff_result_code(struct diff_options *opt, int status)
4789{
4790        int result = 0;
4791
4792        diff_warn_rename_limit("diff.renameLimit",
4793                               opt->needed_rename_limit,
4794                               opt->degraded_cc_to_c);
4795        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4796            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4797                return status;
4798        if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4799            DIFF_OPT_TST(opt, HAS_CHANGES))
4800                result |= 01;
4801        if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4802            DIFF_OPT_TST(opt, CHECK_FAILED))
4803                result |= 02;
4804        return result;
4805}
4806
4807int diff_can_quit_early(struct diff_options *opt)
4808{
4809        return (DIFF_OPT_TST(opt, QUICK) &&
4810                !opt->filter &&
4811                DIFF_OPT_TST(opt, HAS_CHANGES));
4812}
4813
4814/*
4815 * Shall changes to this submodule be ignored?
4816 *
4817 * Submodule changes can be configured to be ignored separately for each path,
4818 * but that configuration can be overridden from the command line.
4819 */
4820static int is_submodule_ignored(const char *path, struct diff_options *options)
4821{
4822        int ignored = 0;
4823        unsigned orig_flags = options->flags;
4824        if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4825                set_diffopt_flags_from_submodule_config(options, path);
4826        if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4827                ignored = 1;
4828        options->flags = orig_flags;
4829        return ignored;
4830}
4831
4832void diff_addremove(struct diff_options *options,
4833                    int addremove, unsigned mode,
4834                    const unsigned char *sha1,
4835                    int sha1_valid,
4836                    const char *concatpath, unsigned dirty_submodule)
4837{
4838        struct diff_filespec *one, *two;
4839
4840        if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4841                return;
4842
4843        /* This may look odd, but it is a preparation for
4844         * feeding "there are unchanged files which should
4845         * not produce diffs, but when you are doing copy
4846         * detection you would need them, so here they are"
4847         * entries to the diff-core.  They will be prefixed
4848         * with something like '=' or '*' (I haven't decided
4849         * which but should not make any difference).
4850         * Feeding the same new and old to diff_change()
4851         * also has the same effect.
4852         * Before the final output happens, they are pruned after
4853         * merged into rename/copy pairs as appropriate.
4854         */
4855        if (DIFF_OPT_TST(options, REVERSE_DIFF))
4856                addremove = (addremove == '+' ? '-' :
4857                             addremove == '-' ? '+' : addremove);
4858
4859        if (options->prefix &&
4860            strncmp(concatpath, options->prefix, options->prefix_length))
4861                return;
4862
4863        one = alloc_filespec(concatpath);
4864        two = alloc_filespec(concatpath);
4865
4866        if (addremove != '+')
4867                fill_filespec(one, sha1, sha1_valid, mode);
4868        if (addremove != '-') {
4869                fill_filespec(two, sha1, sha1_valid, mode);
4870                two->dirty_submodule = dirty_submodule;
4871        }
4872
4873        diff_queue(&diff_queued_diff, one, two);
4874        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4875                DIFF_OPT_SET(options, HAS_CHANGES);
4876}
4877
4878void diff_change(struct diff_options *options,
4879                 unsigned old_mode, unsigned new_mode,
4880                 const unsigned char *old_sha1,
4881                 const unsigned char *new_sha1,
4882                 int old_sha1_valid, int new_sha1_valid,
4883                 const char *concatpath,
4884                 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4885{
4886        struct diff_filespec *one, *two;
4887        struct diff_filepair *p;
4888
4889        if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4890            is_submodule_ignored(concatpath, options))
4891                return;
4892
4893        if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4894                unsigned tmp;
4895                const unsigned char *tmp_c;
4896                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4897                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4898                tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4899                        new_sha1_valid = tmp;
4900                tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4901                        new_dirty_submodule = tmp;
4902        }
4903
4904        if (options->prefix &&
4905            strncmp(concatpath, options->prefix, options->prefix_length))
4906                return;
4907
4908        one = alloc_filespec(concatpath);
4909        two = alloc_filespec(concatpath);
4910        fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4911        fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4912        one->dirty_submodule = old_dirty_submodule;
4913        two->dirty_submodule = new_dirty_submodule;
4914        p = diff_queue(&diff_queued_diff, one, two);
4915
4916        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4917                return;
4918
4919        if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
4920            !diff_filespec_check_stat_unmatch(p))
4921                return;
4922
4923        DIFF_OPT_SET(options, HAS_CHANGES);
4924}
4925
4926struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4927{
4928        struct diff_filepair *pair;
4929        struct diff_filespec *one, *two;
4930
4931        if (options->prefix &&
4932            strncmp(path, options->prefix, options->prefix_length))
4933                return NULL;
4934
4935        one = alloc_filespec(path);
4936        two = alloc_filespec(path);
4937        pair = diff_queue(&diff_queued_diff, one, two);
4938        pair->is_unmerged = 1;
4939        return pair;
4940}
4941
4942static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4943                size_t *outsize)
4944{
4945        struct diff_tempfile *temp;
4946        const char *argv[3];
4947        const char **arg = argv;
4948        struct child_process child;
4949        struct strbuf buf = STRBUF_INIT;
4950        int err = 0;
4951
4952        temp = prepare_temp_file(spec->path, spec);
4953        *arg++ = pgm;
4954        *arg++ = temp->name;
4955        *arg = NULL;
4956
4957        memset(&child, 0, sizeof(child));
4958        child.use_shell = 1;
4959        child.argv = argv;
4960        child.out = -1;
4961        if (start_command(&child)) {
4962                remove_tempfile();
4963                return NULL;
4964        }
4965
4966        if (strbuf_read(&buf, child.out, 0) < 0)
4967                err = error("error reading from textconv command '%s'", pgm);
4968        close(child.out);
4969
4970        if (finish_command(&child) || err) {
4971                strbuf_release(&buf);
4972                remove_tempfile();
4973                return NULL;
4974        }
4975        remove_tempfile();
4976
4977        return strbuf_detach(&buf, outsize);
4978}
4979
4980size_t fill_textconv(struct userdiff_driver *driver,
4981                     struct diff_filespec *df,
4982                     char **outbuf)
4983{
4984        size_t size;
4985
4986        if (!driver || !driver->textconv) {
4987                if (!DIFF_FILE_VALID(df)) {
4988                        *outbuf = "";
4989                        return 0;
4990                }
4991                if (diff_populate_filespec(df, 0))
4992                        die("unable to read files to diff");
4993                *outbuf = df->data;
4994                return df->size;
4995        }
4996
4997        if (driver->textconv_cache && df->sha1_valid) {
4998                *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4999                                          &size);
5000                if (*outbuf)
5001                        return size;
5002        }
5003
5004        *outbuf = run_textconv(driver->textconv, df, &size);
5005        if (!*outbuf)
5006                die("unable to read files to diff");
5007
5008        if (driver->textconv_cache && df->sha1_valid) {
5009                /* ignore errors, as we might be in a readonly repository */
5010                notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5011                                size);
5012                /*
5013                 * we could save up changes and flush them all at the end,
5014                 * but we would need an extra call after all diffing is done.
5015                 * Since generating a cache entry is the slow path anyway,
5016                 * this extra overhead probably isn't a big deal.
5017                 */
5018                notes_cache_write(driver->textconv_cache);
5019        }
5020
5021        return size;
5022}
5023
5024void setup_diff_pager(struct diff_options *opt)
5025{
5026        /*
5027         * If the user asked for our exit code, then either they want --quiet
5028         * or --exit-code. We should definitely not bother with a pager in the
5029         * former case, as we will generate no output. Since we still properly
5030         * report our exit code even when a pager is run, we _could_ run a
5031         * pager with --exit-code. But since we have not done so historically,
5032         * and because it is easy to find people oneline advising "git diff
5033         * --exit-code" in hooks and other scripts, we do not do so.
5034         */
5035        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5036            check_pager_config("diff") != 0)
5037                setup_pager();
5038}