diff.con commit ls-remote: fall-back to default remotes when no remote specified (9c00de5)
   1/*
   2 * Copyright (C) 2005 Junio C Hamano
   3 */
   4#include "cache.h"
   5#include "quote.h"
   6#include "diff.h"
   7#include "diffcore.h"
   8#include "delta.h"
   9#include "xdiff-interface.h"
  10#include "color.h"
  11#include "attr.h"
  12#include "run-command.h"
  13#include "utf8.h"
  14#include "userdiff.h"
  15#include "sigchain.h"
  16#include "submodule.h"
  17#include "ll-merge.h"
  18
  19#ifdef NO_FAST_WORKING_DIRECTORY
  20#define FAST_WORKING_DIRECTORY 0
  21#else
  22#define FAST_WORKING_DIRECTORY 1
  23#endif
  24
  25static int diff_detect_rename_default;
  26static int diff_rename_limit_default = 200;
  27static int diff_suppress_blank_empty;
  28int diff_use_color_default = -1;
  29static const char *diff_word_regex_cfg;
  30static const char *external_diff_cmd_cfg;
  31int diff_auto_refresh_index = 1;
  32static int diff_mnemonic_prefix;
  33
  34static char diff_colors[][COLOR_MAXLEN] = {
  35        GIT_COLOR_RESET,
  36        GIT_COLOR_NORMAL,       /* PLAIN */
  37        GIT_COLOR_BOLD,         /* METAINFO */
  38        GIT_COLOR_CYAN,         /* FRAGINFO */
  39        GIT_COLOR_RED,          /* OLD */
  40        GIT_COLOR_GREEN,        /* NEW */
  41        GIT_COLOR_YELLOW,       /* COMMIT */
  42        GIT_COLOR_BG_RED,       /* WHITESPACE */
  43        GIT_COLOR_NORMAL,       /* FUNCINFO */
  44};
  45
  46static void diff_filespec_load_driver(struct diff_filespec *one);
  47static char *run_textconv(const char *, struct diff_filespec *, size_t *);
  48
  49static int parse_diff_color_slot(const char *var, int ofs)
  50{
  51        if (!strcasecmp(var+ofs, "plain"))
  52                return DIFF_PLAIN;
  53        if (!strcasecmp(var+ofs, "meta"))
  54                return DIFF_METAINFO;
  55        if (!strcasecmp(var+ofs, "frag"))
  56                return DIFF_FRAGINFO;
  57        if (!strcasecmp(var+ofs, "old"))
  58                return DIFF_FILE_OLD;
  59        if (!strcasecmp(var+ofs, "new"))
  60                return DIFF_FILE_NEW;
  61        if (!strcasecmp(var+ofs, "commit"))
  62                return DIFF_COMMIT;
  63        if (!strcasecmp(var+ofs, "whitespace"))
  64                return DIFF_WHITESPACE;
  65        if (!strcasecmp(var+ofs, "func"))
  66                return DIFF_FUNCINFO;
  67        return -1;
  68}
  69
  70static int git_config_rename(const char *var, const char *value)
  71{
  72        if (!value)
  73                return DIFF_DETECT_RENAME;
  74        if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
  75                return  DIFF_DETECT_COPY;
  76        return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
  77}
  78
  79/*
  80 * These are to give UI layer defaults.
  81 * The core-level commands such as git-diff-files should
  82 * never be affected by the setting of diff.renames
  83 * the user happens to have in the configuration file.
  84 */
  85int git_diff_ui_config(const char *var, const char *value, void *cb)
  86{
  87        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
  88                diff_use_color_default = git_config_colorbool(var, value, -1);
  89                return 0;
  90        }
  91        if (!strcmp(var, "diff.renames")) {
  92                diff_detect_rename_default = git_config_rename(var, value);
  93                return 0;
  94        }
  95        if (!strcmp(var, "diff.autorefreshindex")) {
  96                diff_auto_refresh_index = git_config_bool(var, value);
  97                return 0;
  98        }
  99        if (!strcmp(var, "diff.mnemonicprefix")) {
 100                diff_mnemonic_prefix = git_config_bool(var, value);
 101                return 0;
 102        }
 103        if (!strcmp(var, "diff.external"))
 104                return git_config_string(&external_diff_cmd_cfg, var, value);
 105        if (!strcmp(var, "diff.wordregex"))
 106                return git_config_string(&diff_word_regex_cfg, var, value);
 107
 108        return git_diff_basic_config(var, value, cb);
 109}
 110
 111int git_diff_basic_config(const char *var, const char *value, void *cb)
 112{
 113        if (!strcmp(var, "diff.renamelimit")) {
 114                diff_rename_limit_default = git_config_int(var, value);
 115                return 0;
 116        }
 117
 118        switch (userdiff_config(var, value)) {
 119                case 0: break;
 120                case -1: return -1;
 121                default: return 0;
 122        }
 123
 124        if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 125                int slot = parse_diff_color_slot(var, 11);
 126                if (slot < 0)
 127                        return 0;
 128                if (!value)
 129                        return config_error_nonbool(var);
 130                color_parse(value, var, diff_colors[slot]);
 131                return 0;
 132        }
 133
 134        /* like GNU diff's --suppress-blank-empty option  */
 135        if (!strcmp(var, "diff.suppressblankempty") ||
 136                        /* for backwards compatibility */
 137                        !strcmp(var, "diff.suppress-blank-empty")) {
 138                diff_suppress_blank_empty = git_config_bool(var, value);
 139                return 0;
 140        }
 141
 142        return git_color_default_config(var, value, cb);
 143}
 144
 145static char *quote_two(const char *one, const char *two)
 146{
 147        int need_one = quote_c_style(one, NULL, NULL, 1);
 148        int need_two = quote_c_style(two, NULL, NULL, 1);
 149        struct strbuf res = STRBUF_INIT;
 150
 151        if (need_one + need_two) {
 152                strbuf_addch(&res, '"');
 153                quote_c_style(one, &res, NULL, 1);
 154                quote_c_style(two, &res, NULL, 1);
 155                strbuf_addch(&res, '"');
 156        } else {
 157                strbuf_addstr(&res, one);
 158                strbuf_addstr(&res, two);
 159        }
 160        return strbuf_detach(&res, NULL);
 161}
 162
 163static const char *external_diff(void)
 164{
 165        static const char *external_diff_cmd = NULL;
 166        static int done_preparing = 0;
 167
 168        if (done_preparing)
 169                return external_diff_cmd;
 170        external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 171        if (!external_diff_cmd)
 172                external_diff_cmd = external_diff_cmd_cfg;
 173        done_preparing = 1;
 174        return external_diff_cmd;
 175}
 176
 177static struct diff_tempfile {
 178        const char *name; /* filename external diff should read from */
 179        char hex[41];
 180        char mode[10];
 181        char tmp_path[PATH_MAX];
 182} diff_temp[2];
 183
 184typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 185
 186struct emit_callback {
 187        int color_diff;
 188        unsigned ws_rule;
 189        int blank_at_eof_in_preimage;
 190        int blank_at_eof_in_postimage;
 191        int lno_in_preimage;
 192        int lno_in_postimage;
 193        sane_truncate_fn truncate;
 194        const char **label_path;
 195        struct diff_words_data *diff_words;
 196        int *found_changesp;
 197        FILE *file;
 198        struct strbuf *header;
 199};
 200
 201static int count_lines(const char *data, int size)
 202{
 203        int count, ch, completely_empty = 1, nl_just_seen = 0;
 204        count = 0;
 205        while (0 < size--) {
 206                ch = *data++;
 207                if (ch == '\n') {
 208                        count++;
 209                        nl_just_seen = 1;
 210                        completely_empty = 0;
 211                }
 212                else {
 213                        nl_just_seen = 0;
 214                        completely_empty = 0;
 215                }
 216        }
 217        if (completely_empty)
 218                return 0;
 219        if (!nl_just_seen)
 220                count++; /* no trailing newline */
 221        return count;
 222}
 223
 224static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 225{
 226        if (!DIFF_FILE_VALID(one)) {
 227                mf->ptr = (char *)""; /* does not matter */
 228                mf->size = 0;
 229                return 0;
 230        }
 231        else if (diff_populate_filespec(one, 0))
 232                return -1;
 233
 234        mf->ptr = one->data;
 235        mf->size = one->size;
 236        return 0;
 237}
 238
 239static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 240{
 241        char *ptr = mf->ptr;
 242        long size = mf->size;
 243        int cnt = 0;
 244
 245        if (!size)
 246                return cnt;
 247        ptr += size - 1; /* pointing at the very end */
 248        if (*ptr != '\n')
 249                ; /* incomplete line */
 250        else
 251                ptr--; /* skip the last LF */
 252        while (mf->ptr < ptr) {
 253                char *prev_eol;
 254                for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 255                        if (*prev_eol == '\n')
 256                                break;
 257                if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 258                        break;
 259                cnt++;
 260                ptr = prev_eol - 1;
 261        }
 262        return cnt;
 263}
 264
 265static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 266                               struct emit_callback *ecbdata)
 267{
 268        int l1, l2, at;
 269        unsigned ws_rule = ecbdata->ws_rule;
 270        l1 = count_trailing_blank(mf1, ws_rule);
 271        l2 = count_trailing_blank(mf2, ws_rule);
 272        if (l2 <= l1) {
 273                ecbdata->blank_at_eof_in_preimage = 0;
 274                ecbdata->blank_at_eof_in_postimage = 0;
 275                return;
 276        }
 277        at = count_lines(mf1->ptr, mf1->size);
 278        ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 279
 280        at = count_lines(mf2->ptr, mf2->size);
 281        ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 282}
 283
 284static void emit_line_0(FILE *file, const char *set, const char *reset,
 285                        int first, const char *line, int len)
 286{
 287        int has_trailing_newline, has_trailing_carriage_return;
 288        int nofirst;
 289
 290        if (len == 0) {
 291                has_trailing_newline = (first == '\n');
 292                has_trailing_carriage_return = (!has_trailing_newline &&
 293                                                (first == '\r'));
 294                nofirst = has_trailing_newline || has_trailing_carriage_return;
 295        } else {
 296                has_trailing_newline = (len > 0 && line[len-1] == '\n');
 297                if (has_trailing_newline)
 298                        len--;
 299                has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 300                if (has_trailing_carriage_return)
 301                        len--;
 302                nofirst = 0;
 303        }
 304
 305        if (len || !nofirst) {
 306                fputs(set, file);
 307                if (!nofirst)
 308                        fputc(first, file);
 309                fwrite(line, len, 1, file);
 310                fputs(reset, file);
 311        }
 312        if (has_trailing_carriage_return)
 313                fputc('\r', file);
 314        if (has_trailing_newline)
 315                fputc('\n', file);
 316}
 317
 318static void emit_line(FILE *file, const char *set, const char *reset,
 319                      const char *line, int len)
 320{
 321        emit_line_0(file, set, reset, line[0], line+1, len-1);
 322}
 323
 324static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 325{
 326        if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 327              ecbdata->blank_at_eof_in_preimage &&
 328              ecbdata->blank_at_eof_in_postimage &&
 329              ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 330              ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 331                return 0;
 332        return ws_blank_line(line, len, ecbdata->ws_rule);
 333}
 334
 335static void emit_add_line(const char *reset,
 336                          struct emit_callback *ecbdata,
 337                          const char *line, int len)
 338{
 339        const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 340        const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 341
 342        if (!*ws)
 343                emit_line_0(ecbdata->file, set, reset, '+', line, len);
 344        else if (new_blank_line_at_eof(ecbdata, line, len))
 345                /* Blank line at EOF - paint '+' as well */
 346                emit_line_0(ecbdata->file, ws, reset, '+', line, len);
 347        else {
 348                /* Emit just the prefix, then the rest. */
 349                emit_line_0(ecbdata->file, set, reset, '+', "", 0);
 350                ws_check_emit(line, len, ecbdata->ws_rule,
 351                              ecbdata->file, set, reset, ws);
 352        }
 353}
 354
 355static void emit_hunk_header(struct emit_callback *ecbdata,
 356                             const char *line, int len)
 357{
 358        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 359        const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 360        const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 361        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 362        static const char atat[2] = { '@', '@' };
 363        const char *cp, *ep;
 364
 365        /*
 366         * As a hunk header must begin with "@@ -<old>, +<new> @@",
 367         * it always is at least 10 bytes long.
 368         */
 369        if (len < 10 ||
 370            memcmp(line, atat, 2) ||
 371            !(ep = memmem(line + 2, len - 2, atat, 2))) {
 372                emit_line(ecbdata->file, plain, reset, line, len);
 373                return;
 374        }
 375        ep += 2; /* skip over @@ */
 376
 377        /* The hunk header in fraginfo color */
 378        emit_line(ecbdata->file, frag, reset, line, ep - line);
 379
 380        /* blank before the func header */
 381        for (cp = ep; ep - line < len; ep++)
 382                if (*ep != ' ' && *ep != '\t')
 383                        break;
 384        if (ep != cp)
 385                emit_line(ecbdata->file, plain, reset, cp, ep - cp);
 386
 387        if (ep < line + len)
 388                emit_line(ecbdata->file, func, reset, ep, line + len - ep);
 389}
 390
 391static struct diff_tempfile *claim_diff_tempfile(void) {
 392        int i;
 393        for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 394                if (!diff_temp[i].name)
 395                        return diff_temp + i;
 396        die("BUG: diff is failing to clean up its tempfiles");
 397}
 398
 399static int remove_tempfile_installed;
 400
 401static void remove_tempfile(void)
 402{
 403        int i;
 404        for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 405                if (diff_temp[i].name == diff_temp[i].tmp_path)
 406                        unlink_or_warn(diff_temp[i].name);
 407                diff_temp[i].name = NULL;
 408        }
 409}
 410
 411static void remove_tempfile_on_signal(int signo)
 412{
 413        remove_tempfile();
 414        sigchain_pop(signo);
 415        raise(signo);
 416}
 417
 418static void print_line_count(FILE *file, int count)
 419{
 420        switch (count) {
 421        case 0:
 422                fprintf(file, "0,0");
 423                break;
 424        case 1:
 425                fprintf(file, "1");
 426                break;
 427        default:
 428                fprintf(file, "1,%d", count);
 429                break;
 430        }
 431}
 432
 433static void emit_rewrite_lines(struct emit_callback *ecb,
 434                               int prefix, const char *data, int size)
 435{
 436        const char *endp = NULL;
 437        static const char *nneof = " No newline at end of file\n";
 438        const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 439        const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 440
 441        while (0 < size) {
 442                int len;
 443
 444                endp = memchr(data, '\n', size);
 445                len = endp ? (endp - data + 1) : size;
 446                if (prefix != '+') {
 447                        ecb->lno_in_preimage++;
 448                        emit_line_0(ecb->file, old, reset, '-',
 449                                    data, len);
 450                } else {
 451                        ecb->lno_in_postimage++;
 452                        emit_add_line(reset, ecb, data, len);
 453                }
 454                size -= len;
 455                data += len;
 456        }
 457        if (!endp) {
 458                const char *plain = diff_get_color(ecb->color_diff,
 459                                                   DIFF_PLAIN);
 460                emit_line_0(ecb->file, plain, reset, '\\',
 461                            nneof, strlen(nneof));
 462        }
 463}
 464
 465static void emit_rewrite_diff(const char *name_a,
 466                              const char *name_b,
 467                              struct diff_filespec *one,
 468                              struct diff_filespec *two,
 469                              const char *textconv_one,
 470                              const char *textconv_two,
 471                              struct diff_options *o)
 472{
 473        int lc_a, lc_b;
 474        int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 475        const char *name_a_tab, *name_b_tab;
 476        const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 477        const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 478        const char *reset = diff_get_color(color_diff, DIFF_RESET);
 479        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 480        const char *a_prefix, *b_prefix;
 481        const char *data_one, *data_two;
 482        size_t size_one, size_two;
 483        struct emit_callback ecbdata;
 484
 485        if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 486                a_prefix = o->b_prefix;
 487                b_prefix = o->a_prefix;
 488        } else {
 489                a_prefix = o->a_prefix;
 490                b_prefix = o->b_prefix;
 491        }
 492
 493        name_a += (*name_a == '/');
 494        name_b += (*name_b == '/');
 495        name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 496        name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 497
 498        strbuf_reset(&a_name);
 499        strbuf_reset(&b_name);
 500        quote_two_c_style(&a_name, a_prefix, name_a, 0);
 501        quote_two_c_style(&b_name, b_prefix, name_b, 0);
 502
 503        diff_populate_filespec(one, 0);
 504        diff_populate_filespec(two, 0);
 505        if (textconv_one) {
 506                data_one = run_textconv(textconv_one, one, &size_one);
 507                if (!data_one)
 508                        die("unable to read files to diff");
 509        }
 510        else {
 511                data_one = one->data;
 512                size_one = one->size;
 513        }
 514        if (textconv_two) {
 515                data_two = run_textconv(textconv_two, two, &size_two);
 516                if (!data_two)
 517                        die("unable to read files to diff");
 518        }
 519        else {
 520                data_two = two->data;
 521                size_two = two->size;
 522        }
 523
 524        memset(&ecbdata, 0, sizeof(ecbdata));
 525        ecbdata.color_diff = color_diff;
 526        ecbdata.found_changesp = &o->found_changes;
 527        ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 528        ecbdata.file = o->file;
 529        if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 530                mmfile_t mf1, mf2;
 531                mf1.ptr = (char *)data_one;
 532                mf2.ptr = (char *)data_two;
 533                mf1.size = size_one;
 534                mf2.size = size_two;
 535                check_blank_at_eof(&mf1, &mf2, &ecbdata);
 536        }
 537        ecbdata.lno_in_preimage = 1;
 538        ecbdata.lno_in_postimage = 1;
 539
 540        lc_a = count_lines(data_one, size_one);
 541        lc_b = count_lines(data_two, size_two);
 542        fprintf(o->file,
 543                "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 544                metainfo, a_name.buf, name_a_tab, reset,
 545                metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 546        print_line_count(o->file, lc_a);
 547        fprintf(o->file, " +");
 548        print_line_count(o->file, lc_b);
 549        fprintf(o->file, " @@%s\n", reset);
 550        if (lc_a)
 551                emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 552        if (lc_b)
 553                emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 554}
 555
 556struct diff_words_buffer {
 557        mmfile_t text;
 558        long alloc;
 559        struct diff_words_orig {
 560                const char *begin, *end;
 561        } *orig;
 562        int orig_nr, orig_alloc;
 563};
 564
 565static void diff_words_append(char *line, unsigned long len,
 566                struct diff_words_buffer *buffer)
 567{
 568        ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 569        line++;
 570        len--;
 571        memcpy(buffer->text.ptr + buffer->text.size, line, len);
 572        buffer->text.size += len;
 573        buffer->text.ptr[buffer->text.size] = '\0';
 574}
 575
 576struct diff_words_data {
 577        struct diff_words_buffer minus, plus;
 578        const char *current_plus;
 579        FILE *file;
 580        regex_t *word_regex;
 581};
 582
 583static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 584{
 585        struct diff_words_data *diff_words = priv;
 586        int minus_first, minus_len, plus_first, plus_len;
 587        const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 588
 589        if (line[0] != '@' || parse_hunk_header(line, len,
 590                        &minus_first, &minus_len, &plus_first, &plus_len))
 591                return;
 592
 593        /* POSIX requires that first be decremented by one if len == 0... */
 594        if (minus_len) {
 595                minus_begin = diff_words->minus.orig[minus_first].begin;
 596                minus_end =
 597                        diff_words->minus.orig[minus_first + minus_len - 1].end;
 598        } else
 599                minus_begin = minus_end =
 600                        diff_words->minus.orig[minus_first].end;
 601
 602        if (plus_len) {
 603                plus_begin = diff_words->plus.orig[plus_first].begin;
 604                plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 605        } else
 606                plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 607
 608        if (diff_words->current_plus != plus_begin)
 609                fwrite(diff_words->current_plus,
 610                                plus_begin - diff_words->current_plus, 1,
 611                                diff_words->file);
 612        if (minus_begin != minus_end)
 613                color_fwrite_lines(diff_words->file,
 614                                diff_get_color(1, DIFF_FILE_OLD),
 615                                minus_end - minus_begin, minus_begin);
 616        if (plus_begin != plus_end)
 617                color_fwrite_lines(diff_words->file,
 618                                diff_get_color(1, DIFF_FILE_NEW),
 619                                plus_end - plus_begin, plus_begin);
 620
 621        diff_words->current_plus = plus_end;
 622}
 623
 624/* This function starts looking at *begin, and returns 0 iff a word was found. */
 625static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 626                int *begin, int *end)
 627{
 628        if (word_regex && *begin < buffer->size) {
 629                regmatch_t match[1];
 630                if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 631                        char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 632                                        '\n', match[0].rm_eo - match[0].rm_so);
 633                        *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 634                        *begin += match[0].rm_so;
 635                        return *begin >= *end;
 636                }
 637                return -1;
 638        }
 639
 640        /* find the next word */
 641        while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 642                (*begin)++;
 643        if (*begin >= buffer->size)
 644                return -1;
 645
 646        /* find the end of the word */
 647        *end = *begin + 1;
 648        while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 649                (*end)++;
 650
 651        return 0;
 652}
 653
 654/*
 655 * This function splits the words in buffer->text, stores the list with
 656 * newline separator into out, and saves the offsets of the original words
 657 * in buffer->orig.
 658 */
 659static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 660                regex_t *word_regex)
 661{
 662        int i, j;
 663        long alloc = 0;
 664
 665        out->size = 0;
 666        out->ptr = NULL;
 667
 668        /* fake an empty "0th" word */
 669        ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 670        buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 671        buffer->orig_nr = 1;
 672
 673        for (i = 0; i < buffer->text.size; i++) {
 674                if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 675                        return;
 676
 677                /* store original boundaries */
 678                ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 679                                buffer->orig_alloc);
 680                buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 681                buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 682                buffer->orig_nr++;
 683
 684                /* store one word */
 685                ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 686                memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 687                out->ptr[out->size + j - i] = '\n';
 688                out->size += j - i + 1;
 689
 690                i = j - 1;
 691        }
 692}
 693
 694/* this executes the word diff on the accumulated buffers */
 695static void diff_words_show(struct diff_words_data *diff_words)
 696{
 697        xpparam_t xpp;
 698        xdemitconf_t xecfg;
 699        xdemitcb_t ecb;
 700        mmfile_t minus, plus;
 701
 702        /* special case: only removal */
 703        if (!diff_words->plus.text.size) {
 704                color_fwrite_lines(diff_words->file,
 705                        diff_get_color(1, DIFF_FILE_OLD),
 706                        diff_words->minus.text.size, diff_words->minus.text.ptr);
 707                diff_words->minus.text.size = 0;
 708                return;
 709        }
 710
 711        diff_words->current_plus = diff_words->plus.text.ptr;
 712
 713        memset(&xpp, 0, sizeof(xpp));
 714        memset(&xecfg, 0, sizeof(xecfg));
 715        diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 716        diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 717        xpp.flags = XDF_NEED_MINIMAL;
 718        /* as only the hunk header will be parsed, we need a 0-context */
 719        xecfg.ctxlen = 0;
 720        xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 721                      &xpp, &xecfg, &ecb);
 722        free(minus.ptr);
 723        free(plus.ptr);
 724        if (diff_words->current_plus != diff_words->plus.text.ptr +
 725                        diff_words->plus.text.size)
 726                fwrite(diff_words->current_plus,
 727                        diff_words->plus.text.ptr + diff_words->plus.text.size
 728                        - diff_words->current_plus, 1,
 729                        diff_words->file);
 730        diff_words->minus.text.size = diff_words->plus.text.size = 0;
 731}
 732
 733/* In "color-words" mode, show word-diff of words accumulated in the buffer */
 734static void diff_words_flush(struct emit_callback *ecbdata)
 735{
 736        if (ecbdata->diff_words->minus.text.size ||
 737            ecbdata->diff_words->plus.text.size)
 738                diff_words_show(ecbdata->diff_words);
 739}
 740
 741static void free_diff_words_data(struct emit_callback *ecbdata)
 742{
 743        if (ecbdata->diff_words) {
 744                diff_words_flush(ecbdata);
 745                free (ecbdata->diff_words->minus.text.ptr);
 746                free (ecbdata->diff_words->minus.orig);
 747                free (ecbdata->diff_words->plus.text.ptr);
 748                free (ecbdata->diff_words->plus.orig);
 749                free(ecbdata->diff_words->word_regex);
 750                free(ecbdata->diff_words);
 751                ecbdata->diff_words = NULL;
 752        }
 753}
 754
 755const char *diff_get_color(int diff_use_color, enum color_diff ix)
 756{
 757        if (diff_use_color)
 758                return diff_colors[ix];
 759        return "";
 760}
 761
 762static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 763{
 764        const char *cp;
 765        unsigned long allot;
 766        size_t l = len;
 767
 768        if (ecb->truncate)
 769                return ecb->truncate(line, len);
 770        cp = line;
 771        allot = l;
 772        while (0 < l) {
 773                (void) utf8_width(&cp, &l);
 774                if (!cp)
 775                        break; /* truncated in the middle? */
 776        }
 777        return allot - l;
 778}
 779
 780static void find_lno(const char *line, struct emit_callback *ecbdata)
 781{
 782        const char *p;
 783        ecbdata->lno_in_preimage = 0;
 784        ecbdata->lno_in_postimage = 0;
 785        p = strchr(line, '-');
 786        if (!p)
 787                return; /* cannot happen */
 788        ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
 789        p = strchr(p, '+');
 790        if (!p)
 791                return; /* cannot happen */
 792        ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
 793}
 794
 795static void fn_out_consume(void *priv, char *line, unsigned long len)
 796{
 797        struct emit_callback *ecbdata = priv;
 798        const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 799        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 800        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 801
 802        if (ecbdata->header) {
 803                fprintf(ecbdata->file, "%s", ecbdata->header->buf);
 804                strbuf_reset(ecbdata->header);
 805                ecbdata->header = NULL;
 806        }
 807        *(ecbdata->found_changesp) = 1;
 808
 809        if (ecbdata->label_path[0]) {
 810                const char *name_a_tab, *name_b_tab;
 811
 812                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 813                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 814
 815                fprintf(ecbdata->file, "%s--- %s%s%s\n",
 816                        meta, ecbdata->label_path[0], reset, name_a_tab);
 817                fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 818                        meta, ecbdata->label_path[1], reset, name_b_tab);
 819                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 820        }
 821
 822        if (diff_suppress_blank_empty
 823            && len == 2 && line[0] == ' ' && line[1] == '\n') {
 824                line[0] = '\n';
 825                len = 1;
 826        }
 827
 828        if (line[0] == '@') {
 829                if (ecbdata->diff_words)
 830                        diff_words_flush(ecbdata);
 831                len = sane_truncate_line(ecbdata, line, len);
 832                find_lno(line, ecbdata);
 833                emit_hunk_header(ecbdata, line, len);
 834                if (line[len-1] != '\n')
 835                        putc('\n', ecbdata->file);
 836                return;
 837        }
 838
 839        if (len < 1) {
 840                emit_line(ecbdata->file, reset, reset, line, len);
 841                return;
 842        }
 843
 844        if (ecbdata->diff_words) {
 845                if (line[0] == '-') {
 846                        diff_words_append(line, len,
 847                                          &ecbdata->diff_words->minus);
 848                        return;
 849                } else if (line[0] == '+') {
 850                        diff_words_append(line, len,
 851                                          &ecbdata->diff_words->plus);
 852                        return;
 853                }
 854                diff_words_flush(ecbdata);
 855                line++;
 856                len--;
 857                emit_line(ecbdata->file, plain, reset, line, len);
 858                return;
 859        }
 860
 861        if (line[0] != '+') {
 862                const char *color =
 863                        diff_get_color(ecbdata->color_diff,
 864                                       line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
 865                ecbdata->lno_in_preimage++;
 866                if (line[0] == ' ')
 867                        ecbdata->lno_in_postimage++;
 868                emit_line(ecbdata->file, color, reset, line, len);
 869        } else {
 870                ecbdata->lno_in_postimage++;
 871                emit_add_line(reset, ecbdata, line + 1, len - 1);
 872        }
 873}
 874
 875static char *pprint_rename(const char *a, const char *b)
 876{
 877        const char *old = a;
 878        const char *new = b;
 879        struct strbuf name = STRBUF_INIT;
 880        int pfx_length, sfx_length;
 881        int len_a = strlen(a);
 882        int len_b = strlen(b);
 883        int a_midlen, b_midlen;
 884        int qlen_a = quote_c_style(a, NULL, NULL, 0);
 885        int qlen_b = quote_c_style(b, NULL, NULL, 0);
 886
 887        if (qlen_a || qlen_b) {
 888                quote_c_style(a, &name, NULL, 0);
 889                strbuf_addstr(&name, " => ");
 890                quote_c_style(b, &name, NULL, 0);
 891                return strbuf_detach(&name, NULL);
 892        }
 893
 894        /* Find common prefix */
 895        pfx_length = 0;
 896        while (*old && *new && *old == *new) {
 897                if (*old == '/')
 898                        pfx_length = old - a + 1;
 899                old++;
 900                new++;
 901        }
 902
 903        /* Find common suffix */
 904        old = a + len_a;
 905        new = b + len_b;
 906        sfx_length = 0;
 907        while (a <= old && b <= new && *old == *new) {
 908                if (*old == '/')
 909                        sfx_length = len_a - (old - a);
 910                old--;
 911                new--;
 912        }
 913
 914        /*
 915         * pfx{mid-a => mid-b}sfx
 916         * {pfx-a => pfx-b}sfx
 917         * pfx{sfx-a => sfx-b}
 918         * name-a => name-b
 919         */
 920        a_midlen = len_a - pfx_length - sfx_length;
 921        b_midlen = len_b - pfx_length - sfx_length;
 922        if (a_midlen < 0)
 923                a_midlen = 0;
 924        if (b_midlen < 0)
 925                b_midlen = 0;
 926
 927        strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 928        if (pfx_length + sfx_length) {
 929                strbuf_add(&name, a, pfx_length);
 930                strbuf_addch(&name, '{');
 931        }
 932        strbuf_add(&name, a + pfx_length, a_midlen);
 933        strbuf_addstr(&name, " => ");
 934        strbuf_add(&name, b + pfx_length, b_midlen);
 935        if (pfx_length + sfx_length) {
 936                strbuf_addch(&name, '}');
 937                strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 938        }
 939        return strbuf_detach(&name, NULL);
 940}
 941
 942struct diffstat_t {
 943        int nr;
 944        int alloc;
 945        struct diffstat_file {
 946                char *from_name;
 947                char *name;
 948                char *print_name;
 949                unsigned is_unmerged:1;
 950                unsigned is_binary:1;
 951                unsigned is_renamed:1;
 952                unsigned int added, deleted;
 953        } **files;
 954};
 955
 956static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 957                                          const char *name_a,
 958                                          const char *name_b)
 959{
 960        struct diffstat_file *x;
 961        x = xcalloc(sizeof (*x), 1);
 962        if (diffstat->nr == diffstat->alloc) {
 963                diffstat->alloc = alloc_nr(diffstat->alloc);
 964                diffstat->files = xrealloc(diffstat->files,
 965                                diffstat->alloc * sizeof(x));
 966        }
 967        diffstat->files[diffstat->nr++] = x;
 968        if (name_b) {
 969                x->from_name = xstrdup(name_a);
 970                x->name = xstrdup(name_b);
 971                x->is_renamed = 1;
 972        }
 973        else {
 974                x->from_name = NULL;
 975                x->name = xstrdup(name_a);
 976        }
 977        return x;
 978}
 979
 980static void diffstat_consume(void *priv, char *line, unsigned long len)
 981{
 982        struct diffstat_t *diffstat = priv;
 983        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 984
 985        if (line[0] == '+')
 986                x->added++;
 987        else if (line[0] == '-')
 988                x->deleted++;
 989}
 990
 991const char mime_boundary_leader[] = "------------";
 992
 993static int scale_linear(int it, int width, int max_change)
 994{
 995        /*
 996         * make sure that at least one '-' is printed if there were deletions,
 997         * and likewise for '+'.
 998         */
 999        if (max_change < 2)
1000                return it;
1001        return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1002}
1003
1004static void show_name(FILE *file,
1005                      const char *prefix, const char *name, int len)
1006{
1007        fprintf(file, " %s%-*s |", prefix, len, name);
1008}
1009
1010static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1011{
1012        if (cnt <= 0)
1013                return;
1014        fprintf(file, "%s", set);
1015        while (cnt--)
1016                putc(ch, file);
1017        fprintf(file, "%s", reset);
1018}
1019
1020static void fill_print_name(struct diffstat_file *file)
1021{
1022        char *pname;
1023
1024        if (file->print_name)
1025                return;
1026
1027        if (!file->is_renamed) {
1028                struct strbuf buf = STRBUF_INIT;
1029                if (quote_c_style(file->name, &buf, NULL, 0)) {
1030                        pname = strbuf_detach(&buf, NULL);
1031                } else {
1032                        pname = file->name;
1033                        strbuf_release(&buf);
1034                }
1035        } else {
1036                pname = pprint_rename(file->from_name, file->name);
1037        }
1038        file->print_name = pname;
1039}
1040
1041static void show_stats(struct diffstat_t *data, struct diff_options *options)
1042{
1043        int i, len, add, del, adds = 0, dels = 0;
1044        int max_change = 0, max_len = 0;
1045        int total_files = data->nr;
1046        int width, name_width;
1047        const char *reset, *set, *add_c, *del_c;
1048
1049        if (data->nr == 0)
1050                return;
1051
1052        width = options->stat_width ? options->stat_width : 80;
1053        name_width = options->stat_name_width ? options->stat_name_width : 50;
1054
1055        /* Sanity: give at least 5 columns to the graph,
1056         * but leave at least 10 columns for the name.
1057         */
1058        if (width < 25)
1059                width = 25;
1060        if (name_width < 10)
1061                name_width = 10;
1062        else if (width < name_width + 15)
1063                name_width = width - 15;
1064
1065        /* Find the longest filename and max number of changes */
1066        reset = diff_get_color_opt(options, DIFF_RESET);
1067        set   = diff_get_color_opt(options, DIFF_PLAIN);
1068        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1069        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1070
1071        for (i = 0; i < data->nr; i++) {
1072                struct diffstat_file *file = data->files[i];
1073                int change = file->added + file->deleted;
1074                fill_print_name(file);
1075                len = strlen(file->print_name);
1076                if (max_len < len)
1077                        max_len = len;
1078
1079                if (file->is_binary || file->is_unmerged)
1080                        continue;
1081                if (max_change < change)
1082                        max_change = change;
1083        }
1084
1085        /* Compute the width of the graph part;
1086         * 10 is for one blank at the beginning of the line plus
1087         * " | count " between the name and the graph.
1088         *
1089         * From here on, name_width is the width of the name area,
1090         * and width is the width of the graph area.
1091         */
1092        name_width = (name_width < max_len) ? name_width : max_len;
1093        if (width < (name_width + 10) + max_change)
1094                width = width - (name_width + 10);
1095        else
1096                width = max_change;
1097
1098        for (i = 0; i < data->nr; i++) {
1099                const char *prefix = "";
1100                char *name = data->files[i]->print_name;
1101                int added = data->files[i]->added;
1102                int deleted = data->files[i]->deleted;
1103                int name_len;
1104
1105                /*
1106                 * "scale" the filename
1107                 */
1108                len = name_width;
1109                name_len = strlen(name);
1110                if (name_width < name_len) {
1111                        char *slash;
1112                        prefix = "...";
1113                        len -= 3;
1114                        name += name_len - len;
1115                        slash = strchr(name, '/');
1116                        if (slash)
1117                                name = slash;
1118                }
1119
1120                if (data->files[i]->is_binary) {
1121                        show_name(options->file, prefix, name, len);
1122                        fprintf(options->file, "  Bin ");
1123                        fprintf(options->file, "%s%d%s", del_c, deleted, reset);
1124                        fprintf(options->file, " -> ");
1125                        fprintf(options->file, "%s%d%s", add_c, added, reset);
1126                        fprintf(options->file, " bytes");
1127                        fprintf(options->file, "\n");
1128                        continue;
1129                }
1130                else if (data->files[i]->is_unmerged) {
1131                        show_name(options->file, prefix, name, len);
1132                        fprintf(options->file, "  Unmerged\n");
1133                        continue;
1134                }
1135                else if (!data->files[i]->is_renamed &&
1136                         (added + deleted == 0)) {
1137                        total_files--;
1138                        continue;
1139                }
1140
1141                /*
1142                 * scale the add/delete
1143                 */
1144                add = added;
1145                del = deleted;
1146                adds += add;
1147                dels += del;
1148
1149                if (width <= max_change) {
1150                        add = scale_linear(add, width, max_change);
1151                        del = scale_linear(del, width, max_change);
1152                }
1153                show_name(options->file, prefix, name, len);
1154                fprintf(options->file, "%5d%s", added + deleted,
1155                                added + deleted ? " " : "");
1156                show_graph(options->file, '+', add, add_c, reset);
1157                show_graph(options->file, '-', del, del_c, reset);
1158                fprintf(options->file, "\n");
1159        }
1160        fprintf(options->file,
1161               " %d files changed, %d insertions(+), %d deletions(-)\n",
1162               total_files, adds, dels);
1163}
1164
1165static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1166{
1167        int i, adds = 0, dels = 0, total_files = data->nr;
1168
1169        if (data->nr == 0)
1170                return;
1171
1172        for (i = 0; i < data->nr; i++) {
1173                if (!data->files[i]->is_binary &&
1174                    !data->files[i]->is_unmerged) {
1175                        int added = data->files[i]->added;
1176                        int deleted= data->files[i]->deleted;
1177                        if (!data->files[i]->is_renamed &&
1178                            (added + deleted == 0)) {
1179                                total_files--;
1180                        } else {
1181                                adds += added;
1182                                dels += deleted;
1183                        }
1184                }
1185        }
1186        fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1187               total_files, adds, dels);
1188}
1189
1190static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1191{
1192        int i;
1193
1194        if (data->nr == 0)
1195                return;
1196
1197        for (i = 0; i < data->nr; i++) {
1198                struct diffstat_file *file = data->files[i];
1199
1200                if (file->is_binary)
1201                        fprintf(options->file, "-\t-\t");
1202                else
1203                        fprintf(options->file,
1204                                "%d\t%d\t", file->added, file->deleted);
1205                if (options->line_termination) {
1206                        fill_print_name(file);
1207                        if (!file->is_renamed)
1208                                write_name_quoted(file->name, options->file,
1209                                                  options->line_termination);
1210                        else {
1211                                fputs(file->print_name, options->file);
1212                                putc(options->line_termination, options->file);
1213                        }
1214                } else {
1215                        if (file->is_renamed) {
1216                                putc('\0', options->file);
1217                                write_name_quoted(file->from_name, options->file, '\0');
1218                        }
1219                        write_name_quoted(file->name, options->file, '\0');
1220                }
1221        }
1222}
1223
1224struct dirstat_file {
1225        const char *name;
1226        unsigned long changed;
1227};
1228
1229struct dirstat_dir {
1230        struct dirstat_file *files;
1231        int alloc, nr, percent, cumulative;
1232};
1233
1234static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1235{
1236        unsigned long this_dir = 0;
1237        unsigned int sources = 0;
1238
1239        while (dir->nr) {
1240                struct dirstat_file *f = dir->files;
1241                int namelen = strlen(f->name);
1242                unsigned long this;
1243                char *slash;
1244
1245                if (namelen < baselen)
1246                        break;
1247                if (memcmp(f->name, base, baselen))
1248                        break;
1249                slash = strchr(f->name + baselen, '/');
1250                if (slash) {
1251                        int newbaselen = slash + 1 - f->name;
1252                        this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1253                        sources++;
1254                } else {
1255                        this = f->changed;
1256                        dir->files++;
1257                        dir->nr--;
1258                        sources += 2;
1259                }
1260                this_dir += this;
1261        }
1262
1263        /*
1264         * We don't report dirstat's for
1265         *  - the top level
1266         *  - or cases where everything came from a single directory
1267         *    under this directory (sources == 1).
1268         */
1269        if (baselen && sources != 1) {
1270                int permille = this_dir * 1000 / changed;
1271                if (permille) {
1272                        int percent = permille / 10;
1273                        if (percent >= dir->percent) {
1274                                fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1275                                if (!dir->cumulative)
1276                                        return 0;
1277                        }
1278                }
1279        }
1280        return this_dir;
1281}
1282
1283static int dirstat_compare(const void *_a, const void *_b)
1284{
1285        const struct dirstat_file *a = _a;
1286        const struct dirstat_file *b = _b;
1287        return strcmp(a->name, b->name);
1288}
1289
1290static void show_dirstat(struct diff_options *options)
1291{
1292        int i;
1293        unsigned long changed;
1294        struct dirstat_dir dir;
1295        struct diff_queue_struct *q = &diff_queued_diff;
1296
1297        dir.files = NULL;
1298        dir.alloc = 0;
1299        dir.nr = 0;
1300        dir.percent = options->dirstat_percent;
1301        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1302
1303        changed = 0;
1304        for (i = 0; i < q->nr; i++) {
1305                struct diff_filepair *p = q->queue[i];
1306                const char *name;
1307                unsigned long copied, added, damage;
1308
1309                name = p->one->path ? p->one->path : p->two->path;
1310
1311                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1312                        diff_populate_filespec(p->one, 0);
1313                        diff_populate_filespec(p->two, 0);
1314                        diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1315                                               &copied, &added);
1316                        diff_free_filespec_data(p->one);
1317                        diff_free_filespec_data(p->two);
1318                } else if (DIFF_FILE_VALID(p->one)) {
1319                        diff_populate_filespec(p->one, 1);
1320                        copied = added = 0;
1321                        diff_free_filespec_data(p->one);
1322                } else if (DIFF_FILE_VALID(p->two)) {
1323                        diff_populate_filespec(p->two, 1);
1324                        copied = 0;
1325                        added = p->two->size;
1326                        diff_free_filespec_data(p->two);
1327                } else
1328                        continue;
1329
1330                /*
1331                 * Original minus copied is the removed material,
1332                 * added is the new material.  They are both damages
1333                 * made to the preimage. In --dirstat-by-file mode, count
1334                 * damaged files, not damaged lines. This is done by
1335                 * counting only a single damaged line per file.
1336                 */
1337                damage = (p->one->size - copied) + added;
1338                if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1339                        damage = 1;
1340
1341                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1342                dir.files[dir.nr].name = name;
1343                dir.files[dir.nr].changed = damage;
1344                changed += damage;
1345                dir.nr++;
1346        }
1347
1348        /* This can happen even with many files, if everything was renames */
1349        if (!changed)
1350                return;
1351
1352        /* Show all directories with more than x% of the changes */
1353        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1354        gather_dirstat(options->file, &dir, changed, "", 0);
1355}
1356
1357static void free_diffstat_info(struct diffstat_t *diffstat)
1358{
1359        int i;
1360        for (i = 0; i < diffstat->nr; i++) {
1361                struct diffstat_file *f = diffstat->files[i];
1362                if (f->name != f->print_name)
1363                        free(f->print_name);
1364                free(f->name);
1365                free(f->from_name);
1366                free(f);
1367        }
1368        free(diffstat->files);
1369}
1370
1371struct checkdiff_t {
1372        const char *filename;
1373        int lineno;
1374        int conflict_marker_size;
1375        struct diff_options *o;
1376        unsigned ws_rule;
1377        unsigned status;
1378};
1379
1380static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1381{
1382        char firstchar;
1383        int cnt;
1384
1385        if (len < marker_size + 1)
1386                return 0;
1387        firstchar = line[0];
1388        switch (firstchar) {
1389        case '=': case '>': case '<': case '|':
1390                break;
1391        default:
1392                return 0;
1393        }
1394        for (cnt = 1; cnt < marker_size; cnt++)
1395                if (line[cnt] != firstchar)
1396                        return 0;
1397        /* line[1] thru line[marker_size-1] are same as firstchar */
1398        if (len < marker_size + 1 || !isspace(line[marker_size]))
1399                return 0;
1400        return 1;
1401}
1402
1403static void checkdiff_consume(void *priv, char *line, unsigned long len)
1404{
1405        struct checkdiff_t *data = priv;
1406        int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1407        int marker_size = data->conflict_marker_size;
1408        const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1409        const char *reset = diff_get_color(color_diff, DIFF_RESET);
1410        const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1411        char *err;
1412
1413        if (line[0] == '+') {
1414                unsigned bad;
1415                data->lineno++;
1416                if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1417                        data->status |= 1;
1418                        fprintf(data->o->file,
1419                                "%s:%d: leftover conflict marker\n",
1420                                data->filename, data->lineno);
1421                }
1422                bad = ws_check(line + 1, len - 1, data->ws_rule);
1423                if (!bad)
1424                        return;
1425                data->status |= bad;
1426                err = whitespace_error_string(bad);
1427                fprintf(data->o->file, "%s:%d: %s.\n",
1428                        data->filename, data->lineno, err);
1429                free(err);
1430                emit_line(data->o->file, set, reset, line, 1);
1431                ws_check_emit(line + 1, len - 1, data->ws_rule,
1432                              data->o->file, set, reset, ws);
1433        } else if (line[0] == ' ') {
1434                data->lineno++;
1435        } else if (line[0] == '@') {
1436                char *plus = strchr(line, '+');
1437                if (plus)
1438                        data->lineno = strtol(plus, NULL, 10) - 1;
1439                else
1440                        die("invalid diff");
1441        }
1442}
1443
1444static unsigned char *deflate_it(char *data,
1445                                 unsigned long size,
1446                                 unsigned long *result_size)
1447{
1448        int bound;
1449        unsigned char *deflated;
1450        z_stream stream;
1451
1452        memset(&stream, 0, sizeof(stream));
1453        deflateInit(&stream, zlib_compression_level);
1454        bound = deflateBound(&stream, size);
1455        deflated = xmalloc(bound);
1456        stream.next_out = deflated;
1457        stream.avail_out = bound;
1458
1459        stream.next_in = (unsigned char *)data;
1460        stream.avail_in = size;
1461        while (deflate(&stream, Z_FINISH) == Z_OK)
1462                ; /* nothing */
1463        deflateEnd(&stream);
1464        *result_size = stream.total_out;
1465        return deflated;
1466}
1467
1468static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1469{
1470        void *cp;
1471        void *delta;
1472        void *deflated;
1473        void *data;
1474        unsigned long orig_size;
1475        unsigned long delta_size;
1476        unsigned long deflate_size;
1477        unsigned long data_size;
1478
1479        /* We could do deflated delta, or we could do just deflated two,
1480         * whichever is smaller.
1481         */
1482        delta = NULL;
1483        deflated = deflate_it(two->ptr, two->size, &deflate_size);
1484        if (one->size && two->size) {
1485                delta = diff_delta(one->ptr, one->size,
1486                                   two->ptr, two->size,
1487                                   &delta_size, deflate_size);
1488                if (delta) {
1489                        void *to_free = delta;
1490                        orig_size = delta_size;
1491                        delta = deflate_it(delta, delta_size, &delta_size);
1492                        free(to_free);
1493                }
1494        }
1495
1496        if (delta && delta_size < deflate_size) {
1497                fprintf(file, "delta %lu\n", orig_size);
1498                free(deflated);
1499                data = delta;
1500                data_size = delta_size;
1501        }
1502        else {
1503                fprintf(file, "literal %lu\n", two->size);
1504                free(delta);
1505                data = deflated;
1506                data_size = deflate_size;
1507        }
1508
1509        /* emit data encoded in base85 */
1510        cp = data;
1511        while (data_size) {
1512                int bytes = (52 < data_size) ? 52 : data_size;
1513                char line[70];
1514                data_size -= bytes;
1515                if (bytes <= 26)
1516                        line[0] = bytes + 'A' - 1;
1517                else
1518                        line[0] = bytes - 26 + 'a' - 1;
1519                encode_85(line + 1, cp, bytes);
1520                cp = (char *) cp + bytes;
1521                fputs(line, file);
1522                fputc('\n', file);
1523        }
1524        fprintf(file, "\n");
1525        free(data);
1526}
1527
1528static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1529{
1530        fprintf(file, "GIT binary patch\n");
1531        emit_binary_diff_body(file, one, two);
1532        emit_binary_diff_body(file, two, one);
1533}
1534
1535static void diff_filespec_load_driver(struct diff_filespec *one)
1536{
1537        if (!one->driver)
1538                one->driver = userdiff_find_by_path(one->path);
1539        if (!one->driver)
1540                one->driver = userdiff_find_by_name("default");
1541}
1542
1543int diff_filespec_is_binary(struct diff_filespec *one)
1544{
1545        if (one->is_binary == -1) {
1546                diff_filespec_load_driver(one);
1547                if (one->driver->binary != -1)
1548                        one->is_binary = one->driver->binary;
1549                else {
1550                        if (!one->data && DIFF_FILE_VALID(one))
1551                                diff_populate_filespec(one, 0);
1552                        if (one->data)
1553                                one->is_binary = buffer_is_binary(one->data,
1554                                                one->size);
1555                        if (one->is_binary == -1)
1556                                one->is_binary = 0;
1557                }
1558        }
1559        return one->is_binary;
1560}
1561
1562static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1563{
1564        diff_filespec_load_driver(one);
1565        return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1566}
1567
1568static const char *userdiff_word_regex(struct diff_filespec *one)
1569{
1570        diff_filespec_load_driver(one);
1571        return one->driver->word_regex;
1572}
1573
1574void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1575{
1576        if (!options->a_prefix)
1577                options->a_prefix = a;
1578        if (!options->b_prefix)
1579                options->b_prefix = b;
1580}
1581
1582static const char *get_textconv(struct diff_filespec *one)
1583{
1584        if (!DIFF_FILE_VALID(one))
1585                return NULL;
1586        if (!S_ISREG(one->mode))
1587                return NULL;
1588        diff_filespec_load_driver(one);
1589        return one->driver->textconv;
1590}
1591
1592static void builtin_diff(const char *name_a,
1593                         const char *name_b,
1594                         struct diff_filespec *one,
1595                         struct diff_filespec *two,
1596                         const char *xfrm_msg,
1597                         struct diff_options *o,
1598                         int complete_rewrite)
1599{
1600        mmfile_t mf1, mf2;
1601        const char *lbl[2];
1602        char *a_one, *b_two;
1603        const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1604        const char *reset = diff_get_color_opt(o, DIFF_RESET);
1605        const char *a_prefix, *b_prefix;
1606        const char *textconv_one = NULL, *textconv_two = NULL;
1607        struct strbuf header = STRBUF_INIT;
1608
1609        if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1610                        (!one->mode || S_ISGITLINK(one->mode)) &&
1611                        (!two->mode || S_ISGITLINK(two->mode))) {
1612                const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1613                const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1614                show_submodule_summary(o->file, one ? one->path : two->path,
1615                                one->sha1, two->sha1, two->dirty_submodule,
1616                                del, add, reset);
1617                return;
1618        }
1619
1620        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1621                textconv_one = get_textconv(one);
1622                textconv_two = get_textconv(two);
1623        }
1624
1625        diff_set_mnemonic_prefix(o, "a/", "b/");
1626        if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1627                a_prefix = o->b_prefix;
1628                b_prefix = o->a_prefix;
1629        } else {
1630                a_prefix = o->a_prefix;
1631                b_prefix = o->b_prefix;
1632        }
1633
1634        /* Never use a non-valid filename anywhere if at all possible */
1635        name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1636        name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1637
1638        a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1639        b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1640        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1641        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1642        strbuf_addf(&header, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1643        if (lbl[0][0] == '/') {
1644                /* /dev/null */
1645                strbuf_addf(&header, "%snew file mode %06o%s\n", set, two->mode, reset);
1646                if (xfrm_msg && xfrm_msg[0])
1647                        strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1648        }
1649        else if (lbl[1][0] == '/') {
1650                strbuf_addf(&header, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1651                if (xfrm_msg && xfrm_msg[0])
1652                        strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1653        }
1654        else {
1655                if (one->mode != two->mode) {
1656                        strbuf_addf(&header, "%sold mode %06o%s\n", set, one->mode, reset);
1657                        strbuf_addf(&header, "%snew mode %06o%s\n", set, two->mode, reset);
1658                }
1659                if (xfrm_msg && xfrm_msg[0])
1660                        strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1661
1662                /*
1663                 * we do not run diff between different kind
1664                 * of objects.
1665                 */
1666                if ((one->mode ^ two->mode) & S_IFMT)
1667                        goto free_ab_and_return;
1668                if (complete_rewrite &&
1669                    (textconv_one || !diff_filespec_is_binary(one)) &&
1670                    (textconv_two || !diff_filespec_is_binary(two))) {
1671                        fprintf(o->file, "%s", header.buf);
1672                        strbuf_reset(&header);
1673                        emit_rewrite_diff(name_a, name_b, one, two,
1674                                                textconv_one, textconv_two, o);
1675                        o->found_changes = 1;
1676                        goto free_ab_and_return;
1677                }
1678        }
1679
1680        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1681                die("unable to read files to diff");
1682
1683        if (!DIFF_OPT_TST(o, TEXT) &&
1684            ( (diff_filespec_is_binary(one) && !textconv_one) ||
1685              (diff_filespec_is_binary(two) && !textconv_two) )) {
1686                /* Quite common confusing case */
1687                if (mf1.size == mf2.size &&
1688                    !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1689                        goto free_ab_and_return;
1690                fprintf(o->file, "%s", header.buf);
1691                strbuf_reset(&header);
1692                if (DIFF_OPT_TST(o, BINARY))
1693                        emit_binary_diff(o->file, &mf1, &mf2);
1694                else
1695                        fprintf(o->file, "Binary files %s and %s differ\n",
1696                                lbl[0], lbl[1]);
1697                o->found_changes = 1;
1698        }
1699        else {
1700                /* Crazy xdl interfaces.. */
1701                const char *diffopts = getenv("GIT_DIFF_OPTS");
1702                xpparam_t xpp;
1703                xdemitconf_t xecfg;
1704                xdemitcb_t ecb;
1705                struct emit_callback ecbdata;
1706                const struct userdiff_funcname *pe;
1707
1708                if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS)) {
1709                        fprintf(o->file, "%s", header.buf);
1710                        strbuf_reset(&header);
1711                }
1712
1713                if (textconv_one) {
1714                        size_t size;
1715                        mf1.ptr = run_textconv(textconv_one, one, &size);
1716                        if (!mf1.ptr)
1717                                die("unable to read files to diff");
1718                        mf1.size = size;
1719                }
1720                if (textconv_two) {
1721                        size_t size;
1722                        mf2.ptr = run_textconv(textconv_two, two, &size);
1723                        if (!mf2.ptr)
1724                                die("unable to read files to diff");
1725                        mf2.size = size;
1726                }
1727
1728                pe = diff_funcname_pattern(one);
1729                if (!pe)
1730                        pe = diff_funcname_pattern(two);
1731
1732                memset(&xpp, 0, sizeof(xpp));
1733                memset(&xecfg, 0, sizeof(xecfg));
1734                memset(&ecbdata, 0, sizeof(ecbdata));
1735                ecbdata.label_path = lbl;
1736                ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1737                ecbdata.found_changesp = &o->found_changes;
1738                ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1739                if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1740                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
1741                ecbdata.file = o->file;
1742                ecbdata.header = header.len ? &header : NULL;
1743                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1744                xecfg.ctxlen = o->context;
1745                xecfg.interhunkctxlen = o->interhunkcontext;
1746                xecfg.flags = XDL_EMIT_FUNCNAMES;
1747                if (pe)
1748                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1749                if (!diffopts)
1750                        ;
1751                else if (!prefixcmp(diffopts, "--unified="))
1752                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1753                else if (!prefixcmp(diffopts, "-u"))
1754                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1755                if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1756                        ecbdata.diff_words =
1757                                xcalloc(1, sizeof(struct diff_words_data));
1758                        ecbdata.diff_words->file = o->file;
1759                        if (!o->word_regex)
1760                                o->word_regex = userdiff_word_regex(one);
1761                        if (!o->word_regex)
1762                                o->word_regex = userdiff_word_regex(two);
1763                        if (!o->word_regex)
1764                                o->word_regex = diff_word_regex_cfg;
1765                        if (o->word_regex) {
1766                                ecbdata.diff_words->word_regex = (regex_t *)
1767                                        xmalloc(sizeof(regex_t));
1768                                if (regcomp(ecbdata.diff_words->word_regex,
1769                                                o->word_regex,
1770                                                REG_EXTENDED | REG_NEWLINE))
1771                                        die ("Invalid regular expression: %s",
1772                                                        o->word_regex);
1773                        }
1774                }
1775                xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1776                              &xpp, &xecfg, &ecb);
1777                if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1778                        free_diff_words_data(&ecbdata);
1779                if (textconv_one)
1780                        free(mf1.ptr);
1781                if (textconv_two)
1782                        free(mf2.ptr);
1783                xdiff_clear_find_func(&xecfg);
1784        }
1785
1786 free_ab_and_return:
1787        strbuf_release(&header);
1788        diff_free_filespec_data(one);
1789        diff_free_filespec_data(two);
1790        free(a_one);
1791        free(b_two);
1792        return;
1793}
1794
1795static void builtin_diffstat(const char *name_a, const char *name_b,
1796                             struct diff_filespec *one,
1797                             struct diff_filespec *two,
1798                             struct diffstat_t *diffstat,
1799                             struct diff_options *o,
1800                             int complete_rewrite)
1801{
1802        mmfile_t mf1, mf2;
1803        struct diffstat_file *data;
1804
1805        data = diffstat_add(diffstat, name_a, name_b);
1806
1807        if (!one || !two) {
1808                data->is_unmerged = 1;
1809                return;
1810        }
1811        if (complete_rewrite) {
1812                diff_populate_filespec(one, 0);
1813                diff_populate_filespec(two, 0);
1814                data->deleted = count_lines(one->data, one->size);
1815                data->added = count_lines(two->data, two->size);
1816                goto free_and_return;
1817        }
1818        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1819                die("unable to read files to diff");
1820
1821        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1822                data->is_binary = 1;
1823                data->added = mf2.size;
1824                data->deleted = mf1.size;
1825        } else {
1826                /* Crazy xdl interfaces.. */
1827                xpparam_t xpp;
1828                xdemitconf_t xecfg;
1829                xdemitcb_t ecb;
1830
1831                memset(&xpp, 0, sizeof(xpp));
1832                memset(&xecfg, 0, sizeof(xecfg));
1833                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1834                xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1835                              &xpp, &xecfg, &ecb);
1836        }
1837
1838 free_and_return:
1839        diff_free_filespec_data(one);
1840        diff_free_filespec_data(two);
1841}
1842
1843static void builtin_checkdiff(const char *name_a, const char *name_b,
1844                              const char *attr_path,
1845                              struct diff_filespec *one,
1846                              struct diff_filespec *two,
1847                              struct diff_options *o)
1848{
1849        mmfile_t mf1, mf2;
1850        struct checkdiff_t data;
1851
1852        if (!two)
1853                return;
1854
1855        memset(&data, 0, sizeof(data));
1856        data.filename = name_b ? name_b : name_a;
1857        data.lineno = 0;
1858        data.o = o;
1859        data.ws_rule = whitespace_rule(attr_path);
1860        data.conflict_marker_size = ll_merge_marker_size(attr_path);
1861
1862        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1863                die("unable to read files to diff");
1864
1865        /*
1866         * All the other codepaths check both sides, but not checking
1867         * the "old" side here is deliberate.  We are checking the newly
1868         * introduced changes, and as long as the "new" side is text, we
1869         * can and should check what it introduces.
1870         */
1871        if (diff_filespec_is_binary(two))
1872                goto free_and_return;
1873        else {
1874                /* Crazy xdl interfaces.. */
1875                xpparam_t xpp;
1876                xdemitconf_t xecfg;
1877                xdemitcb_t ecb;
1878
1879                memset(&xpp, 0, sizeof(xpp));
1880                memset(&xecfg, 0, sizeof(xecfg));
1881                xecfg.ctxlen = 1; /* at least one context line */
1882                xpp.flags = XDF_NEED_MINIMAL;
1883                xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1884                              &xpp, &xecfg, &ecb);
1885
1886                if (data.ws_rule & WS_BLANK_AT_EOF) {
1887                        struct emit_callback ecbdata;
1888                        int blank_at_eof;
1889
1890                        ecbdata.ws_rule = data.ws_rule;
1891                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
1892                        blank_at_eof = ecbdata.blank_at_eof_in_preimage;
1893
1894                        if (blank_at_eof) {
1895                                static char *err;
1896                                if (!err)
1897                                        err = whitespace_error_string(WS_BLANK_AT_EOF);
1898                                fprintf(o->file, "%s:%d: %s.\n",
1899                                        data.filename, blank_at_eof, err);
1900                                data.status = 1; /* report errors */
1901                        }
1902                }
1903        }
1904 free_and_return:
1905        diff_free_filespec_data(one);
1906        diff_free_filespec_data(two);
1907        if (data.status)
1908                DIFF_OPT_SET(o, CHECK_FAILED);
1909}
1910
1911struct diff_filespec *alloc_filespec(const char *path)
1912{
1913        int namelen = strlen(path);
1914        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1915
1916        memset(spec, 0, sizeof(*spec));
1917        spec->path = (char *)(spec + 1);
1918        memcpy(spec->path, path, namelen+1);
1919        spec->count = 1;
1920        spec->is_binary = -1;
1921        return spec;
1922}
1923
1924void free_filespec(struct diff_filespec *spec)
1925{
1926        if (!--spec->count) {
1927                diff_free_filespec_data(spec);
1928                free(spec);
1929        }
1930}
1931
1932void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1933                   unsigned short mode)
1934{
1935        if (mode) {
1936                spec->mode = canon_mode(mode);
1937                hashcpy(spec->sha1, sha1);
1938                spec->sha1_valid = !is_null_sha1(sha1);
1939        }
1940}
1941
1942/*
1943 * Given a name and sha1 pair, if the index tells us the file in
1944 * the work tree has that object contents, return true, so that
1945 * prepare_temp_file() does not have to inflate and extract.
1946 */
1947static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1948{
1949        struct cache_entry *ce;
1950        struct stat st;
1951        int pos, len;
1952
1953        /*
1954         * We do not read the cache ourselves here, because the
1955         * benchmark with my previous version that always reads cache
1956         * shows that it makes things worse for diff-tree comparing
1957         * two linux-2.6 kernel trees in an already checked out work
1958         * tree.  This is because most diff-tree comparisons deal with
1959         * only a small number of files, while reading the cache is
1960         * expensive for a large project, and its cost outweighs the
1961         * savings we get by not inflating the object to a temporary
1962         * file.  Practically, this code only helps when we are used
1963         * by diff-cache --cached, which does read the cache before
1964         * calling us.
1965         */
1966        if (!active_cache)
1967                return 0;
1968
1969        /* We want to avoid the working directory if our caller
1970         * doesn't need the data in a normal file, this system
1971         * is rather slow with its stat/open/mmap/close syscalls,
1972         * and the object is contained in a pack file.  The pack
1973         * is probably already open and will be faster to obtain
1974         * the data through than the working directory.  Loose
1975         * objects however would tend to be slower as they need
1976         * to be individually opened and inflated.
1977         */
1978        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1979                return 0;
1980
1981        len = strlen(name);
1982        pos = cache_name_pos(name, len);
1983        if (pos < 0)
1984                return 0;
1985        ce = active_cache[pos];
1986
1987        /*
1988         * This is not the sha1 we are looking for, or
1989         * unreusable because it is not a regular file.
1990         */
1991        if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1992                return 0;
1993
1994        /*
1995         * If ce is marked as "assume unchanged", there is no
1996         * guarantee that work tree matches what we are looking for.
1997         */
1998        if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
1999                return 0;
2000
2001        /*
2002         * If ce matches the file in the work tree, we can reuse it.
2003         */
2004        if (ce_uptodate(ce) ||
2005            (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2006                return 1;
2007
2008        return 0;
2009}
2010
2011static int populate_from_stdin(struct diff_filespec *s)
2012{
2013        struct strbuf buf = STRBUF_INIT;
2014        size_t size = 0;
2015
2016        if (strbuf_read(&buf, 0, 0) < 0)
2017                return error("error while reading from stdin %s",
2018                                     strerror(errno));
2019
2020        s->should_munmap = 0;
2021        s->data = strbuf_detach(&buf, &size);
2022        s->size = size;
2023        s->should_free = 1;
2024        return 0;
2025}
2026
2027static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2028{
2029        int len;
2030        char *data = xmalloc(100), *dirty = "";
2031
2032        /* Are we looking at the work tree? */
2033        if (s->dirty_submodule)
2034                dirty = "-dirty";
2035
2036        len = snprintf(data, 100,
2037                       "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2038        s->data = data;
2039        s->size = len;
2040        s->should_free = 1;
2041        if (size_only) {
2042                s->data = NULL;
2043                free(data);
2044        }
2045        return 0;
2046}
2047
2048/*
2049 * While doing rename detection and pickaxe operation, we may need to
2050 * grab the data for the blob (or file) for our own in-core comparison.
2051 * diff_filespec has data and size fields for this purpose.
2052 */
2053int diff_populate_filespec(struct diff_filespec *s, int size_only)
2054{
2055        int err = 0;
2056        if (!DIFF_FILE_VALID(s))
2057                die("internal error: asking to populate invalid file.");
2058        if (S_ISDIR(s->mode))
2059                return -1;
2060
2061        if (s->data)
2062                return 0;
2063
2064        if (size_only && 0 < s->size)
2065                return 0;
2066
2067        if (S_ISGITLINK(s->mode))
2068                return diff_populate_gitlink(s, size_only);
2069
2070        if (!s->sha1_valid ||
2071            reuse_worktree_file(s->path, s->sha1, 0)) {
2072                struct strbuf buf = STRBUF_INIT;
2073                struct stat st;
2074                int fd;
2075
2076                if (!strcmp(s->path, "-"))
2077                        return populate_from_stdin(s);
2078
2079                if (lstat(s->path, &st) < 0) {
2080                        if (errno == ENOENT) {
2081                        err_empty:
2082                                err = -1;
2083                        empty:
2084                                s->data = (char *)"";
2085                                s->size = 0;
2086                                return err;
2087                        }
2088                }
2089                s->size = xsize_t(st.st_size);
2090                if (!s->size)
2091                        goto empty;
2092                if (S_ISLNK(st.st_mode)) {
2093                        struct strbuf sb = STRBUF_INIT;
2094
2095                        if (strbuf_readlink(&sb, s->path, s->size))
2096                                goto err_empty;
2097                        s->size = sb.len;
2098                        s->data = strbuf_detach(&sb, NULL);
2099                        s->should_free = 1;
2100                        return 0;
2101                }
2102                if (size_only)
2103                        return 0;
2104                fd = open(s->path, O_RDONLY);
2105                if (fd < 0)
2106                        goto err_empty;
2107                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2108                close(fd);
2109                s->should_munmap = 1;
2110
2111                /*
2112                 * Convert from working tree format to canonical git format
2113                 */
2114                if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2115                        size_t size = 0;
2116                        munmap(s->data, s->size);
2117                        s->should_munmap = 0;
2118                        s->data = strbuf_detach(&buf, &size);
2119                        s->size = size;
2120                        s->should_free = 1;
2121                }
2122        }
2123        else {
2124                enum object_type type;
2125                if (size_only)
2126                        type = sha1_object_info(s->sha1, &s->size);
2127                else {
2128                        s->data = read_sha1_file(s->sha1, &type, &s->size);
2129                        s->should_free = 1;
2130                }
2131        }
2132        return 0;
2133}
2134
2135void diff_free_filespec_blob(struct diff_filespec *s)
2136{
2137        if (s->should_free)
2138                free(s->data);
2139        else if (s->should_munmap)
2140                munmap(s->data, s->size);
2141
2142        if (s->should_free || s->should_munmap) {
2143                s->should_free = s->should_munmap = 0;
2144                s->data = NULL;
2145        }
2146}
2147
2148void diff_free_filespec_data(struct diff_filespec *s)
2149{
2150        diff_free_filespec_blob(s);
2151        free(s->cnt_data);
2152        s->cnt_data = NULL;
2153}
2154
2155static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2156                           void *blob,
2157                           unsigned long size,
2158                           const unsigned char *sha1,
2159                           int mode)
2160{
2161        int fd;
2162        struct strbuf buf = STRBUF_INIT;
2163        struct strbuf template = STRBUF_INIT;
2164        char *path_dup = xstrdup(path);
2165        const char *base = basename(path_dup);
2166
2167        /* Generate "XXXXXX_basename.ext" */
2168        strbuf_addstr(&template, "XXXXXX_");
2169        strbuf_addstr(&template, base);
2170
2171        fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2172                        strlen(base) + 1);
2173        if (fd < 0)
2174                die_errno("unable to create temp-file");
2175        if (convert_to_working_tree(path,
2176                        (const char *)blob, (size_t)size, &buf)) {
2177                blob = buf.buf;
2178                size = buf.len;
2179        }
2180        if (write_in_full(fd, blob, size) != size)
2181                die_errno("unable to write temp-file");
2182        close(fd);
2183        temp->name = temp->tmp_path;
2184        strcpy(temp->hex, sha1_to_hex(sha1));
2185        temp->hex[40] = 0;
2186        sprintf(temp->mode, "%06o", mode);
2187        strbuf_release(&buf);
2188        strbuf_release(&template);
2189        free(path_dup);
2190}
2191
2192static struct diff_tempfile *prepare_temp_file(const char *name,
2193                struct diff_filespec *one)
2194{
2195        struct diff_tempfile *temp = claim_diff_tempfile();
2196
2197        if (!DIFF_FILE_VALID(one)) {
2198        not_a_valid_file:
2199                /* A '-' entry produces this for file-2, and
2200                 * a '+' entry produces this for file-1.
2201                 */
2202                temp->name = "/dev/null";
2203                strcpy(temp->hex, ".");
2204                strcpy(temp->mode, ".");
2205                return temp;
2206        }
2207
2208        if (!remove_tempfile_installed) {
2209                atexit(remove_tempfile);
2210                sigchain_push_common(remove_tempfile_on_signal);
2211                remove_tempfile_installed = 1;
2212        }
2213
2214        if (!one->sha1_valid ||
2215            reuse_worktree_file(name, one->sha1, 1)) {
2216                struct stat st;
2217                if (lstat(name, &st) < 0) {
2218                        if (errno == ENOENT)
2219                                goto not_a_valid_file;
2220                        die_errno("stat(%s)", name);
2221                }
2222                if (S_ISLNK(st.st_mode)) {
2223                        struct strbuf sb = STRBUF_INIT;
2224                        if (strbuf_readlink(&sb, name, st.st_size) < 0)
2225                                die_errno("readlink(%s)", name);
2226                        prep_temp_blob(name, temp, sb.buf, sb.len,
2227                                       (one->sha1_valid ?
2228                                        one->sha1 : null_sha1),
2229                                       (one->sha1_valid ?
2230                                        one->mode : S_IFLNK));
2231                        strbuf_release(&sb);
2232                }
2233                else {
2234                        /* we can borrow from the file in the work tree */
2235                        temp->name = name;
2236                        if (!one->sha1_valid)
2237                                strcpy(temp->hex, sha1_to_hex(null_sha1));
2238                        else
2239                                strcpy(temp->hex, sha1_to_hex(one->sha1));
2240                        /* Even though we may sometimes borrow the
2241                         * contents from the work tree, we always want
2242                         * one->mode.  mode is trustworthy even when
2243                         * !(one->sha1_valid), as long as
2244                         * DIFF_FILE_VALID(one).
2245                         */
2246                        sprintf(temp->mode, "%06o", one->mode);
2247                }
2248                return temp;
2249        }
2250        else {
2251                if (diff_populate_filespec(one, 0))
2252                        die("cannot read data blob for %s", one->path);
2253                prep_temp_blob(name, temp, one->data, one->size,
2254                               one->sha1, one->mode);
2255        }
2256        return temp;
2257}
2258
2259/* An external diff command takes:
2260 *
2261 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2262 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2263 *
2264 */
2265static void run_external_diff(const char *pgm,
2266                              const char *name,
2267                              const char *other,
2268                              struct diff_filespec *one,
2269                              struct diff_filespec *two,
2270                              const char *xfrm_msg,
2271                              int complete_rewrite)
2272{
2273        const char *spawn_arg[10];
2274        int retval;
2275        const char **arg = &spawn_arg[0];
2276
2277        if (one && two) {
2278                struct diff_tempfile *temp_one, *temp_two;
2279                const char *othername = (other ? other : name);
2280                temp_one = prepare_temp_file(name, one);
2281                temp_two = prepare_temp_file(othername, two);
2282                *arg++ = pgm;
2283                *arg++ = name;
2284                *arg++ = temp_one->name;
2285                *arg++ = temp_one->hex;
2286                *arg++ = temp_one->mode;
2287                *arg++ = temp_two->name;
2288                *arg++ = temp_two->hex;
2289                *arg++ = temp_two->mode;
2290                if (other) {
2291                        *arg++ = other;
2292                        *arg++ = xfrm_msg;
2293                }
2294        } else {
2295                *arg++ = pgm;
2296                *arg++ = name;
2297        }
2298        *arg = NULL;
2299        fflush(NULL);
2300        retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2301        remove_tempfile();
2302        if (retval) {
2303                fprintf(stderr, "external diff died, stopping at %s.\n", name);
2304                exit(1);
2305        }
2306}
2307
2308static int similarity_index(struct diff_filepair *p)
2309{
2310        return p->score * 100 / MAX_SCORE;
2311}
2312
2313static void fill_metainfo(struct strbuf *msg,
2314                          const char *name,
2315                          const char *other,
2316                          struct diff_filespec *one,
2317                          struct diff_filespec *two,
2318                          struct diff_options *o,
2319                          struct diff_filepair *p)
2320{
2321        strbuf_init(msg, PATH_MAX * 2 + 300);
2322        switch (p->status) {
2323        case DIFF_STATUS_COPIED:
2324                strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2325                strbuf_addstr(msg, "\ncopy from ");
2326                quote_c_style(name, msg, NULL, 0);
2327                strbuf_addstr(msg, "\ncopy to ");
2328                quote_c_style(other, msg, NULL, 0);
2329                strbuf_addch(msg, '\n');
2330                break;
2331        case DIFF_STATUS_RENAMED:
2332                strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2333                strbuf_addstr(msg, "\nrename from ");
2334                quote_c_style(name, msg, NULL, 0);
2335                strbuf_addstr(msg, "\nrename to ");
2336                quote_c_style(other, msg, NULL, 0);
2337                strbuf_addch(msg, '\n');
2338                break;
2339        case DIFF_STATUS_MODIFIED:
2340                if (p->score) {
2341                        strbuf_addf(msg, "dissimilarity index %d%%\n",
2342                                    similarity_index(p));
2343                        break;
2344                }
2345                /* fallthru */
2346        default:
2347                /* nothing */
2348                ;
2349        }
2350        if (one && two && hashcmp(one->sha1, two->sha1)) {
2351                int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2352
2353                if (DIFF_OPT_TST(o, BINARY)) {
2354                        mmfile_t mf;
2355                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2356                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2357                                abbrev = 40;
2358                }
2359                strbuf_addf(msg, "index %.*s..%.*s",
2360                            abbrev, sha1_to_hex(one->sha1),
2361                            abbrev, sha1_to_hex(two->sha1));
2362                if (one->mode == two->mode)
2363                        strbuf_addf(msg, " %06o", one->mode);
2364                strbuf_addch(msg, '\n');
2365        }
2366        if (msg->len)
2367                strbuf_setlen(msg, msg->len - 1);
2368}
2369
2370static void run_diff_cmd(const char *pgm,
2371                         const char *name,
2372                         const char *other,
2373                         const char *attr_path,
2374                         struct diff_filespec *one,
2375                         struct diff_filespec *two,
2376                         struct strbuf *msg,
2377                         struct diff_options *o,
2378                         struct diff_filepair *p)
2379{
2380        const char *xfrm_msg = NULL;
2381        int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2382
2383        if (msg) {
2384                fill_metainfo(msg, name, other, one, two, o, p);
2385                xfrm_msg = msg->len ? msg->buf : NULL;
2386        }
2387
2388        if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2389                pgm = NULL;
2390        else {
2391                struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2392                if (drv && drv->external)
2393                        pgm = drv->external;
2394        }
2395
2396        if (pgm) {
2397                run_external_diff(pgm, name, other, one, two, xfrm_msg,
2398                                  complete_rewrite);
2399                return;
2400        }
2401        if (one && two)
2402                builtin_diff(name, other ? other : name,
2403                             one, two, xfrm_msg, o, complete_rewrite);
2404        else
2405                fprintf(o->file, "* Unmerged path %s\n", name);
2406}
2407
2408static void diff_fill_sha1_info(struct diff_filespec *one)
2409{
2410        if (DIFF_FILE_VALID(one)) {
2411                if (!one->sha1_valid) {
2412                        struct stat st;
2413                        if (!strcmp(one->path, "-")) {
2414                                hashcpy(one->sha1, null_sha1);
2415                                return;
2416                        }
2417                        if (lstat(one->path, &st) < 0)
2418                                die_errno("stat '%s'", one->path);
2419                        if (index_path(one->sha1, one->path, &st, 0))
2420                                die("cannot hash %s", one->path);
2421                }
2422        }
2423        else
2424                hashclr(one->sha1);
2425}
2426
2427static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2428{
2429        /* Strip the prefix but do not molest /dev/null and absolute paths */
2430        if (*namep && **namep != '/')
2431                *namep += prefix_length;
2432        if (*otherp && **otherp != '/')
2433                *otherp += prefix_length;
2434}
2435
2436static void run_diff(struct diff_filepair *p, struct diff_options *o)
2437{
2438        const char *pgm = external_diff();
2439        struct strbuf msg;
2440        struct diff_filespec *one = p->one;
2441        struct diff_filespec *two = p->two;
2442        const char *name;
2443        const char *other;
2444        const char *attr_path;
2445
2446        name  = p->one->path;
2447        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2448        attr_path = name;
2449        if (o->prefix_length)
2450                strip_prefix(o->prefix_length, &name, &other);
2451
2452        if (DIFF_PAIR_UNMERGED(p)) {
2453                run_diff_cmd(pgm, name, NULL, attr_path,
2454                             NULL, NULL, NULL, o, p);
2455                return;
2456        }
2457
2458        diff_fill_sha1_info(one);
2459        diff_fill_sha1_info(two);
2460
2461        if (!pgm &&
2462            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2463            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2464                /*
2465                 * a filepair that changes between file and symlink
2466                 * needs to be split into deletion and creation.
2467                 */
2468                struct diff_filespec *null = alloc_filespec(two->path);
2469                run_diff_cmd(NULL, name, other, attr_path,
2470                             one, null, &msg, o, p);
2471                free(null);
2472                strbuf_release(&msg);
2473
2474                null = alloc_filespec(one->path);
2475                run_diff_cmd(NULL, name, other, attr_path,
2476                             null, two, &msg, o, p);
2477                free(null);
2478        }
2479        else
2480                run_diff_cmd(pgm, name, other, attr_path,
2481                             one, two, &msg, o, p);
2482
2483        strbuf_release(&msg);
2484}
2485
2486static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2487                         struct diffstat_t *diffstat)
2488{
2489        const char *name;
2490        const char *other;
2491        int complete_rewrite = 0;
2492
2493        if (DIFF_PAIR_UNMERGED(p)) {
2494                /* unmerged */
2495                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2496                return;
2497        }
2498
2499        name = p->one->path;
2500        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2501
2502        if (o->prefix_length)
2503                strip_prefix(o->prefix_length, &name, &other);
2504
2505        diff_fill_sha1_info(p->one);
2506        diff_fill_sha1_info(p->two);
2507
2508        if (p->status == DIFF_STATUS_MODIFIED && p->score)
2509                complete_rewrite = 1;
2510        builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2511}
2512
2513static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2514{
2515        const char *name;
2516        const char *other;
2517        const char *attr_path;
2518
2519        if (DIFF_PAIR_UNMERGED(p)) {
2520                /* unmerged */
2521                return;
2522        }
2523
2524        name = p->one->path;
2525        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2526        attr_path = other ? other : name;
2527
2528        if (o->prefix_length)
2529                strip_prefix(o->prefix_length, &name, &other);
2530
2531        diff_fill_sha1_info(p->one);
2532        diff_fill_sha1_info(p->two);
2533
2534        builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2535}
2536
2537void diff_setup(struct diff_options *options)
2538{
2539        memset(options, 0, sizeof(*options));
2540
2541        options->file = stdout;
2542
2543        options->line_termination = '\n';
2544        options->break_opt = -1;
2545        options->rename_limit = -1;
2546        options->dirstat_percent = 3;
2547        options->context = 3;
2548
2549        options->change = diff_change;
2550        options->add_remove = diff_addremove;
2551        if (diff_use_color_default > 0)
2552                DIFF_OPT_SET(options, COLOR_DIFF);
2553        options->detect_rename = diff_detect_rename_default;
2554
2555        if (!diff_mnemonic_prefix) {
2556                options->a_prefix = "a/";
2557                options->b_prefix = "b/";
2558        }
2559}
2560
2561int diff_setup_done(struct diff_options *options)
2562{
2563        int count = 0;
2564
2565        if (options->output_format & DIFF_FORMAT_NAME)
2566                count++;
2567        if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2568                count++;
2569        if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2570                count++;
2571        if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2572                count++;
2573        if (count > 1)
2574                die("--name-only, --name-status, --check and -s are mutually exclusive");
2575
2576        /*
2577         * Most of the time we can say "there are changes"
2578         * only by checking if there are changed paths, but
2579         * --ignore-whitespace* options force us to look
2580         * inside contents.
2581         */
2582
2583        if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2584            DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2585            DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2586                DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2587        else
2588                DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2589
2590        if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2591                options->detect_rename = DIFF_DETECT_COPY;
2592
2593        if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2594                options->prefix = NULL;
2595        if (options->prefix)
2596                options->prefix_length = strlen(options->prefix);
2597        else
2598                options->prefix_length = 0;
2599
2600        if (options->output_format & (DIFF_FORMAT_NAME |
2601                                      DIFF_FORMAT_NAME_STATUS |
2602                                      DIFF_FORMAT_CHECKDIFF |
2603                                      DIFF_FORMAT_NO_OUTPUT))
2604                options->output_format &= ~(DIFF_FORMAT_RAW |
2605                                            DIFF_FORMAT_NUMSTAT |
2606                                            DIFF_FORMAT_DIFFSTAT |
2607                                            DIFF_FORMAT_SHORTSTAT |
2608                                            DIFF_FORMAT_DIRSTAT |
2609                                            DIFF_FORMAT_SUMMARY |
2610                                            DIFF_FORMAT_PATCH);
2611
2612        /*
2613         * These cases always need recursive; we do not drop caller-supplied
2614         * recursive bits for other formats here.
2615         */
2616        if (options->output_format & (DIFF_FORMAT_PATCH |
2617                                      DIFF_FORMAT_NUMSTAT |
2618                                      DIFF_FORMAT_DIFFSTAT |
2619                                      DIFF_FORMAT_SHORTSTAT |
2620                                      DIFF_FORMAT_DIRSTAT |
2621                                      DIFF_FORMAT_SUMMARY |
2622                                      DIFF_FORMAT_CHECKDIFF))
2623                DIFF_OPT_SET(options, RECURSIVE);
2624        /*
2625         * Also pickaxe would not work very well if you do not say recursive
2626         */
2627        if (options->pickaxe)
2628                DIFF_OPT_SET(options, RECURSIVE);
2629        /*
2630         * When patches are generated, submodules diffed against the work tree
2631         * must be checked for dirtiness too so it can be shown in the output
2632         */
2633        if (options->output_format & DIFF_FORMAT_PATCH)
2634                DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2635
2636        if (options->detect_rename && options->rename_limit < 0)
2637                options->rename_limit = diff_rename_limit_default;
2638        if (options->setup & DIFF_SETUP_USE_CACHE) {
2639                if (!active_cache)
2640                        /* read-cache does not die even when it fails
2641                         * so it is safe for us to do this here.  Also
2642                         * it does not smudge active_cache or active_nr
2643                         * when it fails, so we do not have to worry about
2644                         * cleaning it up ourselves either.
2645                         */
2646                        read_cache();
2647        }
2648        if (options->abbrev <= 0 || 40 < options->abbrev)
2649                options->abbrev = 40; /* full */
2650
2651        /*
2652         * It does not make sense to show the first hit we happened
2653         * to have found.  It does not make sense not to return with
2654         * exit code in such a case either.
2655         */
2656        if (DIFF_OPT_TST(options, QUICK)) {
2657                options->output_format = DIFF_FORMAT_NO_OUTPUT;
2658                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2659        }
2660
2661        return 0;
2662}
2663
2664static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2665{
2666        char c, *eq;
2667        int len;
2668
2669        if (*arg != '-')
2670                return 0;
2671        c = *++arg;
2672        if (!c)
2673                return 0;
2674        if (c == arg_short) {
2675                c = *++arg;
2676                if (!c)
2677                        return 1;
2678                if (val && isdigit(c)) {
2679                        char *end;
2680                        int n = strtoul(arg, &end, 10);
2681                        if (*end)
2682                                return 0;
2683                        *val = n;
2684                        return 1;
2685                }
2686                return 0;
2687        }
2688        if (c != '-')
2689                return 0;
2690        arg++;
2691        eq = strchr(arg, '=');
2692        if (eq)
2693                len = eq - arg;
2694        else
2695                len = strlen(arg);
2696        if (!len || strncmp(arg, arg_long, len))
2697                return 0;
2698        if (eq) {
2699                int n;
2700                char *end;
2701                if (!isdigit(*++eq))
2702                        return 0;
2703                n = strtoul(eq, &end, 10);
2704                if (*end)
2705                        return 0;
2706                *val = n;
2707        }
2708        return 1;
2709}
2710
2711static int diff_scoreopt_parse(const char *opt);
2712
2713int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2714{
2715        const char *arg = av[0];
2716
2717        /* Output format options */
2718        if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2719                options->output_format |= DIFF_FORMAT_PATCH;
2720        else if (opt_arg(arg, 'U', "unified", &options->context))
2721                options->output_format |= DIFF_FORMAT_PATCH;
2722        else if (!strcmp(arg, "--raw"))
2723                options->output_format |= DIFF_FORMAT_RAW;
2724        else if (!strcmp(arg, "--patch-with-raw"))
2725                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2726        else if (!strcmp(arg, "--numstat"))
2727                options->output_format |= DIFF_FORMAT_NUMSTAT;
2728        else if (!strcmp(arg, "--shortstat"))
2729                options->output_format |= DIFF_FORMAT_SHORTSTAT;
2730        else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2731                options->output_format |= DIFF_FORMAT_DIRSTAT;
2732        else if (!strcmp(arg, "--cumulative")) {
2733                options->output_format |= DIFF_FORMAT_DIRSTAT;
2734                DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2735        } else if (opt_arg(arg, 0, "dirstat-by-file",
2736                           &options->dirstat_percent)) {
2737                options->output_format |= DIFF_FORMAT_DIRSTAT;
2738                DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2739        }
2740        else if (!strcmp(arg, "--check"))
2741                options->output_format |= DIFF_FORMAT_CHECKDIFF;
2742        else if (!strcmp(arg, "--summary"))
2743                options->output_format |= DIFF_FORMAT_SUMMARY;
2744        else if (!strcmp(arg, "--patch-with-stat"))
2745                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2746        else if (!strcmp(arg, "--name-only"))
2747                options->output_format |= DIFF_FORMAT_NAME;
2748        else if (!strcmp(arg, "--name-status"))
2749                options->output_format |= DIFF_FORMAT_NAME_STATUS;
2750        else if (!strcmp(arg, "-s"))
2751                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2752        else if (!prefixcmp(arg, "--stat")) {
2753                char *end;
2754                int width = options->stat_width;
2755                int name_width = options->stat_name_width;
2756                arg += 6;
2757                end = (char *)arg;
2758
2759                switch (*arg) {
2760                case '-':
2761                        if (!prefixcmp(arg, "-width="))
2762                                width = strtoul(arg + 7, &end, 10);
2763                        else if (!prefixcmp(arg, "-name-width="))
2764                                name_width = strtoul(arg + 12, &end, 10);
2765                        break;
2766                case '=':
2767                        width = strtoul(arg+1, &end, 10);
2768                        if (*end == ',')
2769                                name_width = strtoul(end+1, &end, 10);
2770                }
2771
2772                /* Important! This checks all the error cases! */
2773                if (*end)
2774                        return 0;
2775                options->output_format |= DIFF_FORMAT_DIFFSTAT;
2776                options->stat_name_width = name_width;
2777                options->stat_width = width;
2778        }
2779
2780        /* renames options */
2781        else if (!prefixcmp(arg, "-B")) {
2782                if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2783                        return -1;
2784        }
2785        else if (!prefixcmp(arg, "-M")) {
2786                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2787                        return -1;
2788                options->detect_rename = DIFF_DETECT_RENAME;
2789        }
2790        else if (!prefixcmp(arg, "-C")) {
2791                if (options->detect_rename == DIFF_DETECT_COPY)
2792                        DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2793                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2794                        return -1;
2795                options->detect_rename = DIFF_DETECT_COPY;
2796        }
2797        else if (!strcmp(arg, "--no-renames"))
2798                options->detect_rename = 0;
2799        else if (!strcmp(arg, "--relative"))
2800                DIFF_OPT_SET(options, RELATIVE_NAME);
2801        else if (!prefixcmp(arg, "--relative=")) {
2802                DIFF_OPT_SET(options, RELATIVE_NAME);
2803                options->prefix = arg + 11;
2804        }
2805
2806        /* xdiff options */
2807        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2808                DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2809        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2810                DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2811        else if (!strcmp(arg, "--ignore-space-at-eol"))
2812                DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2813        else if (!strcmp(arg, "--patience"))
2814                DIFF_XDL_SET(options, PATIENCE_DIFF);
2815
2816        /* flags options */
2817        else if (!strcmp(arg, "--binary")) {
2818                options->output_format |= DIFF_FORMAT_PATCH;
2819                DIFF_OPT_SET(options, BINARY);
2820        }
2821        else if (!strcmp(arg, "--full-index"))
2822                DIFF_OPT_SET(options, FULL_INDEX);
2823        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2824                DIFF_OPT_SET(options, TEXT);
2825        else if (!strcmp(arg, "-R"))
2826                DIFF_OPT_SET(options, REVERSE_DIFF);
2827        else if (!strcmp(arg, "--find-copies-harder"))
2828                DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2829        else if (!strcmp(arg, "--follow"))
2830                DIFF_OPT_SET(options, FOLLOW_RENAMES);
2831        else if (!strcmp(arg, "--color"))
2832                DIFF_OPT_SET(options, COLOR_DIFF);
2833        else if (!prefixcmp(arg, "--color=")) {
2834                int value = git_config_colorbool(NULL, arg+8, -1);
2835                if (value == 0)
2836                        DIFF_OPT_CLR(options, COLOR_DIFF);
2837                else if (value > 0)
2838                        DIFF_OPT_SET(options, COLOR_DIFF);
2839                else
2840                        return error("option `color' expects \"always\", \"auto\", or \"never\"");
2841        }
2842        else if (!strcmp(arg, "--no-color"))
2843                DIFF_OPT_CLR(options, COLOR_DIFF);
2844        else if (!strcmp(arg, "--color-words")) {
2845                DIFF_OPT_SET(options, COLOR_DIFF);
2846                DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2847        }
2848        else if (!prefixcmp(arg, "--color-words=")) {
2849                DIFF_OPT_SET(options, COLOR_DIFF);
2850                DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2851                options->word_regex = arg + 14;
2852        }
2853        else if (!strcmp(arg, "--exit-code"))
2854                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2855        else if (!strcmp(arg, "--quiet"))
2856                DIFF_OPT_SET(options, QUICK);
2857        else if (!strcmp(arg, "--ext-diff"))
2858                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2859        else if (!strcmp(arg, "--no-ext-diff"))
2860                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2861        else if (!strcmp(arg, "--textconv"))
2862                DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2863        else if (!strcmp(arg, "--no-textconv"))
2864                DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2865        else if (!strcmp(arg, "--ignore-submodules"))
2866                DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2867        else if (!strcmp(arg, "--submodule"))
2868                DIFF_OPT_SET(options, SUBMODULE_LOG);
2869        else if (!prefixcmp(arg, "--submodule=")) {
2870                if (!strcmp(arg + 12, "log"))
2871                        DIFF_OPT_SET(options, SUBMODULE_LOG);
2872        }
2873
2874        /* misc options */
2875        else if (!strcmp(arg, "-z"))
2876                options->line_termination = 0;
2877        else if (!prefixcmp(arg, "-l"))
2878                options->rename_limit = strtoul(arg+2, NULL, 10);
2879        else if (!prefixcmp(arg, "-S"))
2880                options->pickaxe = arg + 2;
2881        else if (!strcmp(arg, "--pickaxe-all"))
2882                options->pickaxe_opts = DIFF_PICKAXE_ALL;
2883        else if (!strcmp(arg, "--pickaxe-regex"))
2884                options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2885        else if (!prefixcmp(arg, "-O"))
2886                options->orderfile = arg + 2;
2887        else if (!prefixcmp(arg, "--diff-filter="))
2888                options->filter = arg + 14;
2889        else if (!strcmp(arg, "--abbrev"))
2890                options->abbrev = DEFAULT_ABBREV;
2891        else if (!prefixcmp(arg, "--abbrev=")) {
2892                options->abbrev = strtoul(arg + 9, NULL, 10);
2893                if (options->abbrev < MINIMUM_ABBREV)
2894                        options->abbrev = MINIMUM_ABBREV;
2895                else if (40 < options->abbrev)
2896                        options->abbrev = 40;
2897        }
2898        else if (!prefixcmp(arg, "--src-prefix="))
2899                options->a_prefix = arg + 13;
2900        else if (!prefixcmp(arg, "--dst-prefix="))
2901                options->b_prefix = arg + 13;
2902        else if (!strcmp(arg, "--no-prefix"))
2903                options->a_prefix = options->b_prefix = "";
2904        else if (opt_arg(arg, '\0', "inter-hunk-context",
2905                         &options->interhunkcontext))
2906                ;
2907        else if (!prefixcmp(arg, "--output=")) {
2908                options->file = fopen(arg + strlen("--output="), "w");
2909                if (!options->file)
2910                        die_errno("Could not open '%s'", arg + strlen("--output="));
2911                options->close_file = 1;
2912        } else
2913                return 0;
2914        return 1;
2915}
2916
2917static int parse_num(const char **cp_p)
2918{
2919        unsigned long num, scale;
2920        int ch, dot;
2921        const char *cp = *cp_p;
2922
2923        num = 0;
2924        scale = 1;
2925        dot = 0;
2926        for (;;) {
2927                ch = *cp;
2928                if ( !dot && ch == '.' ) {
2929                        scale = 1;
2930                        dot = 1;
2931                } else if ( ch == '%' ) {
2932                        scale = dot ? scale*100 : 100;
2933                        cp++;   /* % is always at the end */
2934                        break;
2935                } else if ( ch >= '0' && ch <= '9' ) {
2936                        if ( scale < 100000 ) {
2937                                scale *= 10;
2938                                num = (num*10) + (ch-'0');
2939                        }
2940                } else {
2941                        break;
2942                }
2943                cp++;
2944        }
2945        *cp_p = cp;
2946
2947        /* user says num divided by scale and we say internally that
2948         * is MAX_SCORE * num / scale.
2949         */
2950        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2951}
2952
2953static int diff_scoreopt_parse(const char *opt)
2954{
2955        int opt1, opt2, cmd;
2956
2957        if (*opt++ != '-')
2958                return -1;
2959        cmd = *opt++;
2960        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2961                return -1; /* that is not a -M, -C nor -B option */
2962
2963        opt1 = parse_num(&opt);
2964        if (cmd != 'B')
2965                opt2 = 0;
2966        else {
2967                if (*opt == 0)
2968                        opt2 = 0;
2969                else if (*opt != '/')
2970                        return -1; /* we expect -B80/99 or -B80 */
2971                else {
2972                        opt++;
2973                        opt2 = parse_num(&opt);
2974                }
2975        }
2976        if (*opt != 0)
2977                return -1;
2978        return opt1 | (opt2 << 16);
2979}
2980
2981struct diff_queue_struct diff_queued_diff;
2982
2983void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2984{
2985        if (queue->alloc <= queue->nr) {
2986                queue->alloc = alloc_nr(queue->alloc);
2987                queue->queue = xrealloc(queue->queue,
2988                                        sizeof(dp) * queue->alloc);
2989        }
2990        queue->queue[queue->nr++] = dp;
2991}
2992
2993struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2994                                 struct diff_filespec *one,
2995                                 struct diff_filespec *two)
2996{
2997        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2998        dp->one = one;
2999        dp->two = two;
3000        if (queue)
3001                diff_q(queue, dp);
3002        return dp;
3003}
3004
3005void diff_free_filepair(struct diff_filepair *p)
3006{
3007        free_filespec(p->one);
3008        free_filespec(p->two);
3009        free(p);
3010}
3011
3012/* This is different from find_unique_abbrev() in that
3013 * it stuffs the result with dots for alignment.
3014 */
3015const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3016{
3017        int abblen;
3018        const char *abbrev;
3019        if (len == 40)
3020                return sha1_to_hex(sha1);
3021
3022        abbrev = find_unique_abbrev(sha1, len);
3023        abblen = strlen(abbrev);
3024        if (abblen < 37) {
3025                static char hex[41];
3026                if (len < abblen && abblen <= len + 2)
3027                        sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3028                else
3029                        sprintf(hex, "%s...", abbrev);
3030                return hex;
3031        }
3032        return sha1_to_hex(sha1);
3033}
3034
3035static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3036{
3037        int line_termination = opt->line_termination;
3038        int inter_name_termination = line_termination ? '\t' : '\0';
3039
3040        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3041                fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3042                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
3043                fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3044        }
3045        if (p->score) {
3046                fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3047                        inter_name_termination);
3048        } else {
3049                fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3050        }
3051
3052        if (p->status == DIFF_STATUS_COPIED ||
3053            p->status == DIFF_STATUS_RENAMED) {
3054                const char *name_a, *name_b;
3055                name_a = p->one->path;
3056                name_b = p->two->path;
3057                strip_prefix(opt->prefix_length, &name_a, &name_b);
3058                write_name_quoted(name_a, opt->file, inter_name_termination);
3059                write_name_quoted(name_b, opt->file, line_termination);
3060        } else {
3061                const char *name_a, *name_b;
3062                name_a = p->one->mode ? p->one->path : p->two->path;
3063                name_b = NULL;
3064                strip_prefix(opt->prefix_length, &name_a, &name_b);
3065                write_name_quoted(name_a, opt->file, line_termination);
3066        }
3067}
3068
3069int diff_unmodified_pair(struct diff_filepair *p)
3070{
3071        /* This function is written stricter than necessary to support
3072         * the currently implemented transformers, but the idea is to
3073         * let transformers to produce diff_filepairs any way they want,
3074         * and filter and clean them up here before producing the output.
3075         */
3076        struct diff_filespec *one = p->one, *two = p->two;
3077
3078        if (DIFF_PAIR_UNMERGED(p))
3079                return 0; /* unmerged is interesting */
3080
3081        /* deletion, addition, mode or type change
3082         * and rename are all interesting.
3083         */
3084        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3085            DIFF_PAIR_MODE_CHANGED(p) ||
3086            strcmp(one->path, two->path))
3087                return 0;
3088
3089        /* both are valid and point at the same path.  that is, we are
3090         * dealing with a change.
3091         */
3092        if (one->sha1_valid && two->sha1_valid &&
3093            !hashcmp(one->sha1, two->sha1) &&
3094            !one->dirty_submodule && !two->dirty_submodule)
3095                return 1; /* no change */
3096        if (!one->sha1_valid && !two->sha1_valid)
3097                return 1; /* both look at the same file on the filesystem. */
3098        return 0;
3099}
3100
3101static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3102{
3103        if (diff_unmodified_pair(p))
3104                return;
3105
3106        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3107            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3108                return; /* no tree diffs in patch format */
3109
3110        run_diff(p, o);
3111}
3112
3113static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3114                            struct diffstat_t *diffstat)
3115{
3116        if (diff_unmodified_pair(p))
3117                return;
3118
3119        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3120            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3121                return; /* no tree diffs in patch format */
3122
3123        run_diffstat(p, o, diffstat);
3124}
3125
3126static void diff_flush_checkdiff(struct diff_filepair *p,
3127                struct diff_options *o)
3128{
3129        if (diff_unmodified_pair(p))
3130                return;
3131
3132        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3133            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3134                return; /* no tree diffs in patch format */
3135
3136        run_checkdiff(p, o);
3137}
3138
3139int diff_queue_is_empty(void)
3140{
3141        struct diff_queue_struct *q = &diff_queued_diff;
3142        int i;
3143        for (i = 0; i < q->nr; i++)
3144                if (!diff_unmodified_pair(q->queue[i]))
3145                        return 0;
3146        return 1;
3147}
3148
3149#if DIFF_DEBUG
3150void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3151{
3152        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3153                x, one ? one : "",
3154                s->path,
3155                DIFF_FILE_VALID(s) ? "valid" : "invalid",
3156                s->mode,
3157                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3158        fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3159                x, one ? one : "",
3160                s->size, s->xfrm_flags);
3161}
3162
3163void diff_debug_filepair(const struct diff_filepair *p, int i)
3164{
3165        diff_debug_filespec(p->one, i, "one");
3166        diff_debug_filespec(p->two, i, "two");
3167        fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3168                p->score, p->status ? p->status : '?',
3169                p->one->rename_used, p->broken_pair);
3170}
3171
3172void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3173{
3174        int i;
3175        if (msg)
3176                fprintf(stderr, "%s\n", msg);
3177        fprintf(stderr, "q->nr = %d\n", q->nr);
3178        for (i = 0; i < q->nr; i++) {
3179                struct diff_filepair *p = q->queue[i];
3180                diff_debug_filepair(p, i);
3181        }
3182}
3183#endif
3184
3185static void diff_resolve_rename_copy(void)
3186{
3187        int i;
3188        struct diff_filepair *p;
3189        struct diff_queue_struct *q = &diff_queued_diff;
3190
3191        diff_debug_queue("resolve-rename-copy", q);
3192
3193        for (i = 0; i < q->nr; i++) {
3194                p = q->queue[i];
3195                p->status = 0; /* undecided */
3196                if (DIFF_PAIR_UNMERGED(p))
3197                        p->status = DIFF_STATUS_UNMERGED;
3198                else if (!DIFF_FILE_VALID(p->one))
3199                        p->status = DIFF_STATUS_ADDED;
3200                else if (!DIFF_FILE_VALID(p->two))
3201                        p->status = DIFF_STATUS_DELETED;
3202                else if (DIFF_PAIR_TYPE_CHANGED(p))
3203                        p->status = DIFF_STATUS_TYPE_CHANGED;
3204
3205                /* from this point on, we are dealing with a pair
3206                 * whose both sides are valid and of the same type, i.e.
3207                 * either in-place edit or rename/copy edit.
3208                 */
3209                else if (DIFF_PAIR_RENAME(p)) {
3210                        /*
3211                         * A rename might have re-connected a broken
3212                         * pair up, causing the pathnames to be the
3213                         * same again. If so, that's not a rename at
3214                         * all, just a modification..
3215                         *
3216                         * Otherwise, see if this source was used for
3217                         * multiple renames, in which case we decrement
3218                         * the count, and call it a copy.
3219                         */
3220                        if (!strcmp(p->one->path, p->two->path))
3221                                p->status = DIFF_STATUS_MODIFIED;
3222                        else if (--p->one->rename_used > 0)
3223                                p->status = DIFF_STATUS_COPIED;
3224                        else
3225                                p->status = DIFF_STATUS_RENAMED;
3226                }
3227                else if (hashcmp(p->one->sha1, p->two->sha1) ||
3228                         p->one->mode != p->two->mode ||
3229                         p->one->dirty_submodule ||
3230                         p->two->dirty_submodule ||
3231                         is_null_sha1(p->one->sha1))
3232                        p->status = DIFF_STATUS_MODIFIED;
3233                else {
3234                        /* This is a "no-change" entry and should not
3235                         * happen anymore, but prepare for broken callers.
3236                         */
3237                        error("feeding unmodified %s to diffcore",
3238                              p->one->path);
3239                        p->status = DIFF_STATUS_UNKNOWN;
3240                }
3241        }
3242        diff_debug_queue("resolve-rename-copy done", q);
3243}
3244
3245static int check_pair_status(struct diff_filepair *p)
3246{
3247        switch (p->status) {
3248        case DIFF_STATUS_UNKNOWN:
3249                return 0;
3250        case 0:
3251                die("internal error in diff-resolve-rename-copy");
3252        default:
3253                return 1;
3254        }
3255}
3256
3257static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3258{
3259        int fmt = opt->output_format;
3260
3261        if (fmt & DIFF_FORMAT_CHECKDIFF)
3262                diff_flush_checkdiff(p, opt);
3263        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3264                diff_flush_raw(p, opt);
3265        else if (fmt & DIFF_FORMAT_NAME) {
3266                const char *name_a, *name_b;
3267                name_a = p->two->path;
3268                name_b = NULL;
3269                strip_prefix(opt->prefix_length, &name_a, &name_b);
3270                write_name_quoted(name_a, opt->file, opt->line_termination);
3271        }
3272}
3273
3274static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3275{
3276        if (fs->mode)
3277                fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3278        else
3279                fprintf(file, " %s ", newdelete);
3280        write_name_quoted(fs->path, file, '\n');
3281}
3282
3283
3284static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3285{
3286        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3287                fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3288                        show_name ? ' ' : '\n');
3289                if (show_name) {
3290                        write_name_quoted(p->two->path, file, '\n');
3291                }
3292        }
3293}
3294
3295static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3296{
3297        char *names = pprint_rename(p->one->path, p->two->path);
3298
3299        fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3300        free(names);
3301        show_mode_change(file, p, 0);
3302}
3303
3304static void diff_summary(FILE *file, struct diff_filepair *p)
3305{
3306        switch(p->status) {
3307        case DIFF_STATUS_DELETED:
3308                show_file_mode_name(file, "delete", p->one);
3309                break;
3310        case DIFF_STATUS_ADDED:
3311                show_file_mode_name(file, "create", p->two);
3312                break;
3313        case DIFF_STATUS_COPIED:
3314                show_rename_copy(file, "copy", p);
3315                break;
3316        case DIFF_STATUS_RENAMED:
3317                show_rename_copy(file, "rename", p);
3318                break;
3319        default:
3320                if (p->score) {
3321                        fputs(" rewrite ", file);
3322                        write_name_quoted(p->two->path, file, ' ');
3323                        fprintf(file, "(%d%%)\n", similarity_index(p));
3324                }
3325                show_mode_change(file, p, !p->score);
3326                break;
3327        }
3328}
3329
3330struct patch_id_t {
3331        git_SHA_CTX *ctx;
3332        int patchlen;
3333};
3334
3335static int remove_space(char *line, int len)
3336{
3337        int i;
3338        char *dst = line;
3339        unsigned char c;
3340
3341        for (i = 0; i < len; i++)
3342                if (!isspace((c = line[i])))
3343                        *dst++ = c;
3344
3345        return dst - line;
3346}
3347
3348static void patch_id_consume(void *priv, char *line, unsigned long len)
3349{
3350        struct patch_id_t *data = priv;
3351        int new_len;
3352
3353        /* Ignore line numbers when computing the SHA1 of the patch */
3354        if (!prefixcmp(line, "@@ -"))
3355                return;
3356
3357        new_len = remove_space(line, len);
3358
3359        git_SHA1_Update(data->ctx, line, new_len);
3360        data->patchlen += new_len;
3361}
3362
3363/* returns 0 upon success, and writes result into sha1 */
3364static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3365{
3366        struct diff_queue_struct *q = &diff_queued_diff;
3367        int i;
3368        git_SHA_CTX ctx;
3369        struct patch_id_t data;
3370        char buffer[PATH_MAX * 4 + 20];
3371
3372        git_SHA1_Init(&ctx);
3373        memset(&data, 0, sizeof(struct patch_id_t));
3374        data.ctx = &ctx;
3375
3376        for (i = 0; i < q->nr; i++) {
3377                xpparam_t xpp;
3378                xdemitconf_t xecfg;
3379                xdemitcb_t ecb;
3380                mmfile_t mf1, mf2;
3381                struct diff_filepair *p = q->queue[i];
3382                int len1, len2;
3383
3384                memset(&xpp, 0, sizeof(xpp));
3385                memset(&xecfg, 0, sizeof(xecfg));
3386                if (p->status == 0)
3387                        return error("internal diff status error");
3388                if (p->status == DIFF_STATUS_UNKNOWN)
3389                        continue;
3390                if (diff_unmodified_pair(p))
3391                        continue;
3392                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3393                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3394                        continue;
3395                if (DIFF_PAIR_UNMERGED(p))
3396                        continue;
3397
3398                diff_fill_sha1_info(p->one);
3399                diff_fill_sha1_info(p->two);
3400                if (fill_mmfile(&mf1, p->one) < 0 ||
3401                                fill_mmfile(&mf2, p->two) < 0)
3402                        return error("unable to read files to diff");
3403
3404                len1 = remove_space(p->one->path, strlen(p->one->path));
3405                len2 = remove_space(p->two->path, strlen(p->two->path));
3406                if (p->one->mode == 0)
3407                        len1 = snprintf(buffer, sizeof(buffer),
3408                                        "diff--gita/%.*sb/%.*s"
3409                                        "newfilemode%06o"
3410                                        "---/dev/null"
3411                                        "+++b/%.*s",
3412                                        len1, p->one->path,
3413                                        len2, p->two->path,
3414                                        p->two->mode,
3415                                        len2, p->two->path);
3416                else if (p->two->mode == 0)
3417                        len1 = snprintf(buffer, sizeof(buffer),
3418                                        "diff--gita/%.*sb/%.*s"
3419                                        "deletedfilemode%06o"
3420                                        "---a/%.*s"
3421                                        "+++/dev/null",
3422                                        len1, p->one->path,
3423                                        len2, p->two->path,
3424                                        p->one->mode,
3425                                        len1, p->one->path);
3426                else
3427                        len1 = snprintf(buffer, sizeof(buffer),
3428                                        "diff--gita/%.*sb/%.*s"
3429                                        "---a/%.*s"
3430                                        "+++b/%.*s",
3431                                        len1, p->one->path,
3432                                        len2, p->two->path,
3433                                        len1, p->one->path,
3434                                        len2, p->two->path);
3435                git_SHA1_Update(&ctx, buffer, len1);
3436
3437                xpp.flags = XDF_NEED_MINIMAL;
3438                xecfg.ctxlen = 3;
3439                xecfg.flags = XDL_EMIT_FUNCNAMES;
3440                xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3441                              &xpp, &xecfg, &ecb);
3442        }
3443
3444        git_SHA1_Final(sha1, &ctx);
3445        return 0;
3446}
3447
3448int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3449{
3450        struct diff_queue_struct *q = &diff_queued_diff;
3451        int i;
3452        int result = diff_get_patch_id(options, sha1);
3453
3454        for (i = 0; i < q->nr; i++)
3455                diff_free_filepair(q->queue[i]);
3456
3457        free(q->queue);
3458        q->queue = NULL;
3459        q->nr = q->alloc = 0;
3460
3461        return result;
3462}
3463
3464static int is_summary_empty(const struct diff_queue_struct *q)
3465{
3466        int i;
3467
3468        for (i = 0; i < q->nr; i++) {
3469                const struct diff_filepair *p = q->queue[i];
3470
3471                switch (p->status) {
3472                case DIFF_STATUS_DELETED:
3473                case DIFF_STATUS_ADDED:
3474                case DIFF_STATUS_COPIED:
3475                case DIFF_STATUS_RENAMED:
3476                        return 0;
3477                default:
3478                        if (p->score)
3479                                return 0;
3480                        if (p->one->mode && p->two->mode &&
3481                            p->one->mode != p->two->mode)
3482                                return 0;
3483                        break;
3484                }
3485        }
3486        return 1;
3487}
3488
3489void diff_flush(struct diff_options *options)
3490{
3491        struct diff_queue_struct *q = &diff_queued_diff;
3492        int i, output_format = options->output_format;
3493        int separator = 0;
3494
3495        /*
3496         * Order: raw, stat, summary, patch
3497         * or:    name/name-status/checkdiff (other bits clear)
3498         */
3499        if (!q->nr)
3500                goto free_queue;
3501
3502        if (output_format & (DIFF_FORMAT_RAW |
3503                             DIFF_FORMAT_NAME |
3504                             DIFF_FORMAT_NAME_STATUS |
3505                             DIFF_FORMAT_CHECKDIFF)) {
3506                for (i = 0; i < q->nr; i++) {
3507                        struct diff_filepair *p = q->queue[i];
3508                        if (check_pair_status(p))
3509                                flush_one_pair(p, options);
3510                }
3511                separator++;
3512        }
3513
3514        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3515                struct diffstat_t diffstat;
3516
3517                memset(&diffstat, 0, sizeof(struct diffstat_t));
3518                for (i = 0; i < q->nr; i++) {
3519                        struct diff_filepair *p = q->queue[i];
3520                        if (check_pair_status(p))
3521                                diff_flush_stat(p, options, &diffstat);
3522                }
3523                if (output_format & DIFF_FORMAT_NUMSTAT)
3524                        show_numstat(&diffstat, options);
3525                if (output_format & DIFF_FORMAT_DIFFSTAT)
3526                        show_stats(&diffstat, options);
3527                if (output_format & DIFF_FORMAT_SHORTSTAT)
3528                        show_shortstats(&diffstat, options);
3529                free_diffstat_info(&diffstat);
3530                separator++;
3531        }
3532        if (output_format & DIFF_FORMAT_DIRSTAT)
3533                show_dirstat(options);
3534
3535        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3536                for (i = 0; i < q->nr; i++)
3537                        diff_summary(options->file, q->queue[i]);
3538                separator++;
3539        }
3540
3541        if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3542            DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3543            DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3544                /*
3545                 * run diff_flush_patch for the exit status. setting
3546                 * options->file to /dev/null should be safe, becaue we
3547                 * aren't supposed to produce any output anyway.
3548                 */
3549                if (options->close_file)
3550                        fclose(options->file);
3551                options->file = fopen("/dev/null", "w");
3552                if (!options->file)
3553                        die_errno("Could not open /dev/null");
3554                options->close_file = 1;
3555                for (i = 0; i < q->nr; i++) {
3556                        struct diff_filepair *p = q->queue[i];
3557                        if (check_pair_status(p))
3558                                diff_flush_patch(p, options);
3559                        if (options->found_changes)
3560                                break;
3561                }
3562        }
3563
3564        if (output_format & DIFF_FORMAT_PATCH) {
3565                if (separator) {
3566                        putc(options->line_termination, options->file);
3567                        if (options->stat_sep) {
3568                                /* attach patch instead of inline */
3569                                fputs(options->stat_sep, options->file);
3570                        }
3571                }
3572
3573                for (i = 0; i < q->nr; i++) {
3574                        struct diff_filepair *p = q->queue[i];
3575                        if (check_pair_status(p))
3576                                diff_flush_patch(p, options);
3577                }
3578        }
3579
3580        if (output_format & DIFF_FORMAT_CALLBACK)
3581                options->format_callback(q, options, options->format_callback_data);
3582
3583        for (i = 0; i < q->nr; i++)
3584                diff_free_filepair(q->queue[i]);
3585free_queue:
3586        free(q->queue);
3587        q->queue = NULL;
3588        q->nr = q->alloc = 0;
3589        if (options->close_file)
3590                fclose(options->file);
3591
3592        /*
3593         * Report the content-level differences with HAS_CHANGES;
3594         * diff_addremove/diff_change does not set the bit when
3595         * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3596         */
3597        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3598                if (options->found_changes)
3599                        DIFF_OPT_SET(options, HAS_CHANGES);
3600                else
3601                        DIFF_OPT_CLR(options, HAS_CHANGES);
3602        }
3603}
3604
3605static void diffcore_apply_filter(const char *filter)
3606{
3607        int i;
3608        struct diff_queue_struct *q = &diff_queued_diff;
3609        struct diff_queue_struct outq;
3610        outq.queue = NULL;
3611        outq.nr = outq.alloc = 0;
3612
3613        if (!filter)
3614                return;
3615
3616        if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3617                int found;
3618                for (i = found = 0; !found && i < q->nr; i++) {
3619                        struct diff_filepair *p = q->queue[i];
3620                        if (((p->status == DIFF_STATUS_MODIFIED) &&
3621                             ((p->score &&
3622                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3623                              (!p->score &&
3624                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3625                            ((p->status != DIFF_STATUS_MODIFIED) &&
3626                             strchr(filter, p->status)))
3627                                found++;
3628                }
3629                if (found)
3630                        return;
3631
3632                /* otherwise we will clear the whole queue
3633                 * by copying the empty outq at the end of this
3634                 * function, but first clear the current entries
3635                 * in the queue.
3636                 */
3637                for (i = 0; i < q->nr; i++)
3638                        diff_free_filepair(q->queue[i]);
3639        }
3640        else {
3641                /* Only the matching ones */
3642                for (i = 0; i < q->nr; i++) {
3643                        struct diff_filepair *p = q->queue[i];
3644
3645                        if (((p->status == DIFF_STATUS_MODIFIED) &&
3646                             ((p->score &&
3647                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3648                              (!p->score &&
3649                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3650                            ((p->status != DIFF_STATUS_MODIFIED) &&
3651                             strchr(filter, p->status)))
3652                                diff_q(&outq, p);
3653                        else
3654                                diff_free_filepair(p);
3655                }
3656        }
3657        free(q->queue);
3658        *q = outq;
3659}
3660
3661/* Check whether two filespecs with the same mode and size are identical */
3662static int diff_filespec_is_identical(struct diff_filespec *one,
3663                                      struct diff_filespec *two)
3664{
3665        if (S_ISGITLINK(one->mode))
3666                return 0;
3667        if (diff_populate_filespec(one, 0))
3668                return 0;
3669        if (diff_populate_filespec(two, 0))
3670                return 0;
3671        return !memcmp(one->data, two->data, one->size);
3672}
3673
3674static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3675{
3676        int i;
3677        struct diff_queue_struct *q = &diff_queued_diff;
3678        struct diff_queue_struct outq;
3679        outq.queue = NULL;
3680        outq.nr = outq.alloc = 0;
3681
3682        for (i = 0; i < q->nr; i++) {
3683                struct diff_filepair *p = q->queue[i];
3684
3685                /*
3686                 * 1. Entries that come from stat info dirtiness
3687                 *    always have both sides (iow, not create/delete),
3688                 *    one side of the object name is unknown, with
3689                 *    the same mode and size.  Keep the ones that
3690                 *    do not match these criteria.  They have real
3691                 *    differences.
3692                 *
3693                 * 2. At this point, the file is known to be modified,
3694                 *    with the same mode and size, and the object
3695                 *    name of one side is unknown.  Need to inspect
3696                 *    the identical contents.
3697                 */
3698                if (!DIFF_FILE_VALID(p->one) || /* (1) */
3699                    !DIFF_FILE_VALID(p->two) ||
3700                    (p->one->sha1_valid && p->two->sha1_valid) ||
3701                    (p->one->mode != p->two->mode) ||
3702                    diff_populate_filespec(p->one, 1) ||
3703                    diff_populate_filespec(p->two, 1) ||
3704                    (p->one->size != p->two->size) ||
3705                    !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3706                        diff_q(&outq, p);
3707                else {
3708                        /*
3709                         * The caller can subtract 1 from skip_stat_unmatch
3710                         * to determine how many paths were dirty only
3711                         * due to stat info mismatch.
3712                         */
3713                        if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3714                                diffopt->skip_stat_unmatch++;
3715                        diff_free_filepair(p);
3716                }
3717        }
3718        free(q->queue);
3719        *q = outq;
3720}
3721
3722static int diffnamecmp(const void *a_, const void *b_)
3723{
3724        const struct diff_filepair *a = *((const struct diff_filepair **)a_);
3725        const struct diff_filepair *b = *((const struct diff_filepair **)b_);
3726        const char *name_a, *name_b;
3727
3728        name_a = a->one ? a->one->path : a->two->path;
3729        name_b = b->one ? b->one->path : b->two->path;
3730        return strcmp(name_a, name_b);
3731}
3732
3733void diffcore_fix_diff_index(struct diff_options *options)
3734{
3735        struct diff_queue_struct *q = &diff_queued_diff;
3736        qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
3737}
3738
3739void diffcore_std(struct diff_options *options)
3740{
3741        if (options->skip_stat_unmatch)
3742                diffcore_skip_stat_unmatch(options);
3743        if (options->break_opt != -1)
3744                diffcore_break(options->break_opt);
3745        if (options->detect_rename)
3746                diffcore_rename(options);
3747        if (options->break_opt != -1)
3748                diffcore_merge_broken();
3749        if (options->pickaxe)
3750                diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3751        if (options->orderfile)
3752                diffcore_order(options->orderfile);
3753        diff_resolve_rename_copy();
3754        diffcore_apply_filter(options->filter);
3755
3756        if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3757                DIFF_OPT_SET(options, HAS_CHANGES);
3758        else
3759                DIFF_OPT_CLR(options, HAS_CHANGES);
3760}
3761
3762int diff_result_code(struct diff_options *opt, int status)
3763{
3764        int result = 0;
3765        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3766            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3767                return status;
3768        if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3769            DIFF_OPT_TST(opt, HAS_CHANGES))
3770                result |= 01;
3771        if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3772            DIFF_OPT_TST(opt, CHECK_FAILED))
3773                result |= 02;
3774        return result;
3775}
3776
3777void diff_addremove(struct diff_options *options,
3778                    int addremove, unsigned mode,
3779                    const unsigned char *sha1,
3780                    const char *concatpath, unsigned dirty_submodule)
3781{
3782        struct diff_filespec *one, *two;
3783
3784        if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3785                return;
3786
3787        /* This may look odd, but it is a preparation for
3788         * feeding "there are unchanged files which should
3789         * not produce diffs, but when you are doing copy
3790         * detection you would need them, so here they are"
3791         * entries to the diff-core.  They will be prefixed
3792         * with something like '=' or '*' (I haven't decided
3793         * which but should not make any difference).
3794         * Feeding the same new and old to diff_change()
3795         * also has the same effect.
3796         * Before the final output happens, they are pruned after
3797         * merged into rename/copy pairs as appropriate.
3798         */
3799        if (DIFF_OPT_TST(options, REVERSE_DIFF))
3800                addremove = (addremove == '+' ? '-' :
3801                             addremove == '-' ? '+' : addremove);
3802
3803        if (options->prefix &&
3804            strncmp(concatpath, options->prefix, options->prefix_length))
3805                return;
3806
3807        one = alloc_filespec(concatpath);
3808        two = alloc_filespec(concatpath);
3809
3810        if (addremove != '+')
3811                fill_filespec(one, sha1, mode);
3812        if (addremove != '-') {
3813                fill_filespec(two, sha1, mode);
3814                two->dirty_submodule = dirty_submodule;
3815        }
3816
3817        diff_queue(&diff_queued_diff, one, two);
3818        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3819                DIFF_OPT_SET(options, HAS_CHANGES);
3820}
3821
3822void diff_change(struct diff_options *options,
3823                 unsigned old_mode, unsigned new_mode,
3824                 const unsigned char *old_sha1,
3825                 const unsigned char *new_sha1,
3826                 const char *concatpath,
3827                 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
3828{
3829        struct diff_filespec *one, *two;
3830
3831        if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3832                        && S_ISGITLINK(new_mode))
3833                return;
3834
3835        if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3836                unsigned tmp;
3837                const unsigned char *tmp_c;
3838                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3839                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3840                tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
3841                        new_dirty_submodule = tmp;
3842        }
3843
3844        if (options->prefix &&
3845            strncmp(concatpath, options->prefix, options->prefix_length))
3846                return;
3847
3848        one = alloc_filespec(concatpath);
3849        two = alloc_filespec(concatpath);
3850        fill_filespec(one, old_sha1, old_mode);
3851        fill_filespec(two, new_sha1, new_mode);
3852        one->dirty_submodule = old_dirty_submodule;
3853        two->dirty_submodule = new_dirty_submodule;
3854
3855        diff_queue(&diff_queued_diff, one, two);
3856        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3857                DIFF_OPT_SET(options, HAS_CHANGES);
3858}
3859
3860void diff_unmerge(struct diff_options *options,
3861                  const char *path,
3862                  unsigned mode, const unsigned char *sha1)
3863{
3864        struct diff_filespec *one, *two;
3865
3866        if (options->prefix &&
3867            strncmp(path, options->prefix, options->prefix_length))
3868                return;
3869
3870        one = alloc_filespec(path);
3871        two = alloc_filespec(path);
3872        fill_filespec(one, sha1, mode);
3873        diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3874}
3875
3876static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3877                size_t *outsize)
3878{
3879        struct diff_tempfile *temp;
3880        const char *argv[3];
3881        const char **arg = argv;
3882        struct child_process child;
3883        struct strbuf buf = STRBUF_INIT;
3884        int err = 0;
3885
3886        temp = prepare_temp_file(spec->path, spec);
3887        *arg++ = pgm;
3888        *arg++ = temp->name;
3889        *arg = NULL;
3890
3891        memset(&child, 0, sizeof(child));
3892        child.use_shell = 1;
3893        child.argv = argv;
3894        child.out = -1;
3895        if (start_command(&child)) {
3896                remove_tempfile();
3897                return NULL;
3898        }
3899
3900        if (strbuf_read(&buf, child.out, 0) < 0)
3901                err = error("error reading from textconv command '%s'", pgm);
3902        close(child.out);
3903
3904        if (finish_command(&child) || err) {
3905                strbuf_release(&buf);
3906                remove_tempfile();
3907                return NULL;
3908        }
3909        remove_tempfile();
3910
3911        return strbuf_detach(&buf, outsize);
3912}