diff.con commit Add the option "--ignore-submodules" to "git status" (46a958b)
   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        if (textconv_one)
 555                free((char *)data_one);
 556        if (textconv_two)
 557                free((char *)data_two);
 558}
 559
 560struct diff_words_buffer {
 561        mmfile_t text;
 562        long alloc;
 563        struct diff_words_orig {
 564                const char *begin, *end;
 565        } *orig;
 566        int orig_nr, orig_alloc;
 567};
 568
 569static void diff_words_append(char *line, unsigned long len,
 570                struct diff_words_buffer *buffer)
 571{
 572        ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 573        line++;
 574        len--;
 575        memcpy(buffer->text.ptr + buffer->text.size, line, len);
 576        buffer->text.size += len;
 577        buffer->text.ptr[buffer->text.size] = '\0';
 578}
 579
 580struct diff_words_data {
 581        struct diff_words_buffer minus, plus;
 582        const char *current_plus;
 583        FILE *file;
 584        regex_t *word_regex;
 585};
 586
 587static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 588{
 589        struct diff_words_data *diff_words = priv;
 590        int minus_first, minus_len, plus_first, plus_len;
 591        const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 592
 593        if (line[0] != '@' || parse_hunk_header(line, len,
 594                        &minus_first, &minus_len, &plus_first, &plus_len))
 595                return;
 596
 597        /* POSIX requires that first be decremented by one if len == 0... */
 598        if (minus_len) {
 599                minus_begin = diff_words->minus.orig[minus_first].begin;
 600                minus_end =
 601                        diff_words->minus.orig[minus_first + minus_len - 1].end;
 602        } else
 603                minus_begin = minus_end =
 604                        diff_words->minus.orig[minus_first].end;
 605
 606        if (plus_len) {
 607                plus_begin = diff_words->plus.orig[plus_first].begin;
 608                plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 609        } else
 610                plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 611
 612        if (diff_words->current_plus != plus_begin)
 613                fwrite(diff_words->current_plus,
 614                                plus_begin - diff_words->current_plus, 1,
 615                                diff_words->file);
 616        if (minus_begin != minus_end)
 617                color_fwrite_lines(diff_words->file,
 618                                diff_get_color(1, DIFF_FILE_OLD),
 619                                minus_end - minus_begin, minus_begin);
 620        if (plus_begin != plus_end)
 621                color_fwrite_lines(diff_words->file,
 622                                diff_get_color(1, DIFF_FILE_NEW),
 623                                plus_end - plus_begin, plus_begin);
 624
 625        diff_words->current_plus = plus_end;
 626}
 627
 628/* This function starts looking at *begin, and returns 0 iff a word was found. */
 629static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 630                int *begin, int *end)
 631{
 632        if (word_regex && *begin < buffer->size) {
 633                regmatch_t match[1];
 634                if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 635                        char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 636                                        '\n', match[0].rm_eo - match[0].rm_so);
 637                        *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 638                        *begin += match[0].rm_so;
 639                        return *begin >= *end;
 640                }
 641                return -1;
 642        }
 643
 644        /* find the next word */
 645        while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 646                (*begin)++;
 647        if (*begin >= buffer->size)
 648                return -1;
 649
 650        /* find the end of the word */
 651        *end = *begin + 1;
 652        while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 653                (*end)++;
 654
 655        return 0;
 656}
 657
 658/*
 659 * This function splits the words in buffer->text, stores the list with
 660 * newline separator into out, and saves the offsets of the original words
 661 * in buffer->orig.
 662 */
 663static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 664                regex_t *word_regex)
 665{
 666        int i, j;
 667        long alloc = 0;
 668
 669        out->size = 0;
 670        out->ptr = NULL;
 671
 672        /* fake an empty "0th" word */
 673        ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 674        buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 675        buffer->orig_nr = 1;
 676
 677        for (i = 0; i < buffer->text.size; i++) {
 678                if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 679                        return;
 680
 681                /* store original boundaries */
 682                ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 683                                buffer->orig_alloc);
 684                buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 685                buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 686                buffer->orig_nr++;
 687
 688                /* store one word */
 689                ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 690                memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 691                out->ptr[out->size + j - i] = '\n';
 692                out->size += j - i + 1;
 693
 694                i = j - 1;
 695        }
 696}
 697
 698/* this executes the word diff on the accumulated buffers */
 699static void diff_words_show(struct diff_words_data *diff_words)
 700{
 701        xpparam_t xpp;
 702        xdemitconf_t xecfg;
 703        mmfile_t minus, plus;
 704
 705        /* special case: only removal */
 706        if (!diff_words->plus.text.size) {
 707                color_fwrite_lines(diff_words->file,
 708                        diff_get_color(1, DIFF_FILE_OLD),
 709                        diff_words->minus.text.size, diff_words->minus.text.ptr);
 710                diff_words->minus.text.size = 0;
 711                return;
 712        }
 713
 714        diff_words->current_plus = diff_words->plus.text.ptr;
 715
 716        memset(&xpp, 0, sizeof(xpp));
 717        memset(&xecfg, 0, sizeof(xecfg));
 718        diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 719        diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 720        xpp.flags = XDF_NEED_MINIMAL;
 721        /* as only the hunk header will be parsed, we need a 0-context */
 722        xecfg.ctxlen = 0;
 723        xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 724                      &xpp, &xecfg);
 725        free(minus.ptr);
 726        free(plus.ptr);
 727        if (diff_words->current_plus != diff_words->plus.text.ptr +
 728                        diff_words->plus.text.size)
 729                fwrite(diff_words->current_plus,
 730                        diff_words->plus.text.ptr + diff_words->plus.text.size
 731                        - diff_words->current_plus, 1,
 732                        diff_words->file);
 733        diff_words->minus.text.size = diff_words->plus.text.size = 0;
 734}
 735
 736/* In "color-words" mode, show word-diff of words accumulated in the buffer */
 737static void diff_words_flush(struct emit_callback *ecbdata)
 738{
 739        if (ecbdata->diff_words->minus.text.size ||
 740            ecbdata->diff_words->plus.text.size)
 741                diff_words_show(ecbdata->diff_words);
 742}
 743
 744static void free_diff_words_data(struct emit_callback *ecbdata)
 745{
 746        if (ecbdata->diff_words) {
 747                diff_words_flush(ecbdata);
 748                free (ecbdata->diff_words->minus.text.ptr);
 749                free (ecbdata->diff_words->minus.orig);
 750                free (ecbdata->diff_words->plus.text.ptr);
 751                free (ecbdata->diff_words->plus.orig);
 752                free(ecbdata->diff_words->word_regex);
 753                free(ecbdata->diff_words);
 754                ecbdata->diff_words = NULL;
 755        }
 756}
 757
 758const char *diff_get_color(int diff_use_color, enum color_diff ix)
 759{
 760        if (diff_use_color)
 761                return diff_colors[ix];
 762        return "";
 763}
 764
 765static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 766{
 767        const char *cp;
 768        unsigned long allot;
 769        size_t l = len;
 770
 771        if (ecb->truncate)
 772                return ecb->truncate(line, len);
 773        cp = line;
 774        allot = l;
 775        while (0 < l) {
 776                (void) utf8_width(&cp, &l);
 777                if (!cp)
 778                        break; /* truncated in the middle? */
 779        }
 780        return allot - l;
 781}
 782
 783static void find_lno(const char *line, struct emit_callback *ecbdata)
 784{
 785        const char *p;
 786        ecbdata->lno_in_preimage = 0;
 787        ecbdata->lno_in_postimage = 0;
 788        p = strchr(line, '-');
 789        if (!p)
 790                return; /* cannot happen */
 791        ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
 792        p = strchr(p, '+');
 793        if (!p)
 794                return; /* cannot happen */
 795        ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
 796}
 797
 798static void fn_out_consume(void *priv, char *line, unsigned long len)
 799{
 800        struct emit_callback *ecbdata = priv;
 801        const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 802        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 803        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 804
 805        if (ecbdata->header) {
 806                fprintf(ecbdata->file, "%s", ecbdata->header->buf);
 807                strbuf_reset(ecbdata->header);
 808                ecbdata->header = NULL;
 809        }
 810        *(ecbdata->found_changesp) = 1;
 811
 812        if (ecbdata->label_path[0]) {
 813                const char *name_a_tab, *name_b_tab;
 814
 815                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 816                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 817
 818                fprintf(ecbdata->file, "%s--- %s%s%s\n",
 819                        meta, ecbdata->label_path[0], reset, name_a_tab);
 820                fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 821                        meta, ecbdata->label_path[1], reset, name_b_tab);
 822                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 823        }
 824
 825        if (diff_suppress_blank_empty
 826            && len == 2 && line[0] == ' ' && line[1] == '\n') {
 827                line[0] = '\n';
 828                len = 1;
 829        }
 830
 831        if (line[0] == '@') {
 832                if (ecbdata->diff_words)
 833                        diff_words_flush(ecbdata);
 834                len = sane_truncate_line(ecbdata, line, len);
 835                find_lno(line, ecbdata);
 836                emit_hunk_header(ecbdata, line, len);
 837                if (line[len-1] != '\n')
 838                        putc('\n', ecbdata->file);
 839                return;
 840        }
 841
 842        if (len < 1) {
 843                emit_line(ecbdata->file, reset, reset, line, len);
 844                return;
 845        }
 846
 847        if (ecbdata->diff_words) {
 848                if (line[0] == '-') {
 849                        diff_words_append(line, len,
 850                                          &ecbdata->diff_words->minus);
 851                        return;
 852                } else if (line[0] == '+') {
 853                        diff_words_append(line, len,
 854                                          &ecbdata->diff_words->plus);
 855                        return;
 856                }
 857                diff_words_flush(ecbdata);
 858                line++;
 859                len--;
 860                emit_line(ecbdata->file, plain, reset, line, len);
 861                return;
 862        }
 863
 864        if (line[0] != '+') {
 865                const char *color =
 866                        diff_get_color(ecbdata->color_diff,
 867                                       line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
 868                ecbdata->lno_in_preimage++;
 869                if (line[0] == ' ')
 870                        ecbdata->lno_in_postimage++;
 871                emit_line(ecbdata->file, color, reset, line, len);
 872        } else {
 873                ecbdata->lno_in_postimage++;
 874                emit_add_line(reset, ecbdata, line + 1, len - 1);
 875        }
 876}
 877
 878static char *pprint_rename(const char *a, const char *b)
 879{
 880        const char *old = a;
 881        const char *new = b;
 882        struct strbuf name = STRBUF_INIT;
 883        int pfx_length, sfx_length;
 884        int len_a = strlen(a);
 885        int len_b = strlen(b);
 886        int a_midlen, b_midlen;
 887        int qlen_a = quote_c_style(a, NULL, NULL, 0);
 888        int qlen_b = quote_c_style(b, NULL, NULL, 0);
 889
 890        if (qlen_a || qlen_b) {
 891                quote_c_style(a, &name, NULL, 0);
 892                strbuf_addstr(&name, " => ");
 893                quote_c_style(b, &name, NULL, 0);
 894                return strbuf_detach(&name, NULL);
 895        }
 896
 897        /* Find common prefix */
 898        pfx_length = 0;
 899        while (*old && *new && *old == *new) {
 900                if (*old == '/')
 901                        pfx_length = old - a + 1;
 902                old++;
 903                new++;
 904        }
 905
 906        /* Find common suffix */
 907        old = a + len_a;
 908        new = b + len_b;
 909        sfx_length = 0;
 910        while (a <= old && b <= new && *old == *new) {
 911                if (*old == '/')
 912                        sfx_length = len_a - (old - a);
 913                old--;
 914                new--;
 915        }
 916
 917        /*
 918         * pfx{mid-a => mid-b}sfx
 919         * {pfx-a => pfx-b}sfx
 920         * pfx{sfx-a => sfx-b}
 921         * name-a => name-b
 922         */
 923        a_midlen = len_a - pfx_length - sfx_length;
 924        b_midlen = len_b - pfx_length - sfx_length;
 925        if (a_midlen < 0)
 926                a_midlen = 0;
 927        if (b_midlen < 0)
 928                b_midlen = 0;
 929
 930        strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 931        if (pfx_length + sfx_length) {
 932                strbuf_add(&name, a, pfx_length);
 933                strbuf_addch(&name, '{');
 934        }
 935        strbuf_add(&name, a + pfx_length, a_midlen);
 936        strbuf_addstr(&name, " => ");
 937        strbuf_add(&name, b + pfx_length, b_midlen);
 938        if (pfx_length + sfx_length) {
 939                strbuf_addch(&name, '}');
 940                strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 941        }
 942        return strbuf_detach(&name, NULL);
 943}
 944
 945struct diffstat_t {
 946        int nr;
 947        int alloc;
 948        struct diffstat_file {
 949                char *from_name;
 950                char *name;
 951                char *print_name;
 952                unsigned is_unmerged:1;
 953                unsigned is_binary:1;
 954                unsigned is_renamed:1;
 955                uintmax_t added, deleted;
 956        } **files;
 957};
 958
 959static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 960                                          const char *name_a,
 961                                          const char *name_b)
 962{
 963        struct diffstat_file *x;
 964        x = xcalloc(sizeof (*x), 1);
 965        if (diffstat->nr == diffstat->alloc) {
 966                diffstat->alloc = alloc_nr(diffstat->alloc);
 967                diffstat->files = xrealloc(diffstat->files,
 968                                diffstat->alloc * sizeof(x));
 969        }
 970        diffstat->files[diffstat->nr++] = x;
 971        if (name_b) {
 972                x->from_name = xstrdup(name_a);
 973                x->name = xstrdup(name_b);
 974                x->is_renamed = 1;
 975        }
 976        else {
 977                x->from_name = NULL;
 978                x->name = xstrdup(name_a);
 979        }
 980        return x;
 981}
 982
 983static void diffstat_consume(void *priv, char *line, unsigned long len)
 984{
 985        struct diffstat_t *diffstat = priv;
 986        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 987
 988        if (line[0] == '+')
 989                x->added++;
 990        else if (line[0] == '-')
 991                x->deleted++;
 992}
 993
 994const char mime_boundary_leader[] = "------------";
 995
 996static int scale_linear(int it, int width, int max_change)
 997{
 998        /*
 999         * make sure that at least one '-' is printed if there were deletions,
1000         * and likewise for '+'.
1001         */
1002        if (max_change < 2)
1003                return it;
1004        return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1005}
1006
1007static void show_name(FILE *file,
1008                      const char *prefix, const char *name, int len)
1009{
1010        fprintf(file, " %s%-*s |", prefix, len, name);
1011}
1012
1013static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1014{
1015        if (cnt <= 0)
1016                return;
1017        fprintf(file, "%s", set);
1018        while (cnt--)
1019                putc(ch, file);
1020        fprintf(file, "%s", reset);
1021}
1022
1023static void fill_print_name(struct diffstat_file *file)
1024{
1025        char *pname;
1026
1027        if (file->print_name)
1028                return;
1029
1030        if (!file->is_renamed) {
1031                struct strbuf buf = STRBUF_INIT;
1032                if (quote_c_style(file->name, &buf, NULL, 0)) {
1033                        pname = strbuf_detach(&buf, NULL);
1034                } else {
1035                        pname = file->name;
1036                        strbuf_release(&buf);
1037                }
1038        } else {
1039                pname = pprint_rename(file->from_name, file->name);
1040        }
1041        file->print_name = pname;
1042}
1043
1044static void show_stats(struct diffstat_t *data, struct diff_options *options)
1045{
1046        int i, len, add, del, adds = 0, dels = 0;
1047        uintmax_t max_change = 0, max_len = 0;
1048        int total_files = data->nr;
1049        int width, name_width;
1050        const char *reset, *set, *add_c, *del_c;
1051
1052        if (data->nr == 0)
1053                return;
1054
1055        width = options->stat_width ? options->stat_width : 80;
1056        name_width = options->stat_name_width ? options->stat_name_width : 50;
1057
1058        /* Sanity: give at least 5 columns to the graph,
1059         * but leave at least 10 columns for the name.
1060         */
1061        if (width < 25)
1062                width = 25;
1063        if (name_width < 10)
1064                name_width = 10;
1065        else if (width < name_width + 15)
1066                name_width = width - 15;
1067
1068        /* Find the longest filename and max number of changes */
1069        reset = diff_get_color_opt(options, DIFF_RESET);
1070        set   = diff_get_color_opt(options, DIFF_PLAIN);
1071        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1072        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1073
1074        for (i = 0; i < data->nr; i++) {
1075                struct diffstat_file *file = data->files[i];
1076                uintmax_t change = file->added + file->deleted;
1077                fill_print_name(file);
1078                len = strlen(file->print_name);
1079                if (max_len < len)
1080                        max_len = len;
1081
1082                if (file->is_binary || file->is_unmerged)
1083                        continue;
1084                if (max_change < change)
1085                        max_change = change;
1086        }
1087
1088        /* Compute the width of the graph part;
1089         * 10 is for one blank at the beginning of the line plus
1090         * " | count " between the name and the graph.
1091         *
1092         * From here on, name_width is the width of the name area,
1093         * and width is the width of the graph area.
1094         */
1095        name_width = (name_width < max_len) ? name_width : max_len;
1096        if (width < (name_width + 10) + max_change)
1097                width = width - (name_width + 10);
1098        else
1099                width = max_change;
1100
1101        for (i = 0; i < data->nr; i++) {
1102                const char *prefix = "";
1103                char *name = data->files[i]->print_name;
1104                uintmax_t added = data->files[i]->added;
1105                uintmax_t deleted = data->files[i]->deleted;
1106                int name_len;
1107
1108                /*
1109                 * "scale" the filename
1110                 */
1111                len = name_width;
1112                name_len = strlen(name);
1113                if (name_width < name_len) {
1114                        char *slash;
1115                        prefix = "...";
1116                        len -= 3;
1117                        name += name_len - len;
1118                        slash = strchr(name, '/');
1119                        if (slash)
1120                                name = slash;
1121                }
1122
1123                if (data->files[i]->is_binary) {
1124                        show_name(options->file, prefix, name, len);
1125                        fprintf(options->file, "  Bin ");
1126                        fprintf(options->file, "%s%"PRIuMAX"%s",
1127                                del_c, deleted, reset);
1128                        fprintf(options->file, " -> ");
1129                        fprintf(options->file, "%s%"PRIuMAX"%s",
1130                                add_c, added, reset);
1131                        fprintf(options->file, " bytes");
1132                        fprintf(options->file, "\n");
1133                        continue;
1134                }
1135                else if (data->files[i]->is_unmerged) {
1136                        show_name(options->file, prefix, name, len);
1137                        fprintf(options->file, "  Unmerged\n");
1138                        continue;
1139                }
1140                else if (!data->files[i]->is_renamed &&
1141                         (added + deleted == 0)) {
1142                        total_files--;
1143                        continue;
1144                }
1145
1146                /*
1147                 * scale the add/delete
1148                 */
1149                add = added;
1150                del = deleted;
1151                adds += add;
1152                dels += del;
1153
1154                if (width <= max_change) {
1155                        add = scale_linear(add, width, max_change);
1156                        del = scale_linear(del, width, max_change);
1157                }
1158                show_name(options->file, prefix, name, len);
1159                fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1160                                added + deleted ? " " : "");
1161                show_graph(options->file, '+', add, add_c, reset);
1162                show_graph(options->file, '-', del, del_c, reset);
1163                fprintf(options->file, "\n");
1164        }
1165        fprintf(options->file,
1166               " %d files changed, %d insertions(+), %d deletions(-)\n",
1167               total_files, adds, dels);
1168}
1169
1170static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1171{
1172        int i, adds = 0, dels = 0, total_files = data->nr;
1173
1174        if (data->nr == 0)
1175                return;
1176
1177        for (i = 0; i < data->nr; i++) {
1178                if (!data->files[i]->is_binary &&
1179                    !data->files[i]->is_unmerged) {
1180                        int added = data->files[i]->added;
1181                        int deleted= data->files[i]->deleted;
1182                        if (!data->files[i]->is_renamed &&
1183                            (added + deleted == 0)) {
1184                                total_files--;
1185                        } else {
1186                                adds += added;
1187                                dels += deleted;
1188                        }
1189                }
1190        }
1191        fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1192               total_files, adds, dels);
1193}
1194
1195static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1196{
1197        int i;
1198
1199        if (data->nr == 0)
1200                return;
1201
1202        for (i = 0; i < data->nr; i++) {
1203                struct diffstat_file *file = data->files[i];
1204
1205                if (file->is_binary)
1206                        fprintf(options->file, "-\t-\t");
1207                else
1208                        fprintf(options->file,
1209                                "%"PRIuMAX"\t%"PRIuMAX"\t",
1210                                file->added, file->deleted);
1211                if (options->line_termination) {
1212                        fill_print_name(file);
1213                        if (!file->is_renamed)
1214                                write_name_quoted(file->name, options->file,
1215                                                  options->line_termination);
1216                        else {
1217                                fputs(file->print_name, options->file);
1218                                putc(options->line_termination, options->file);
1219                        }
1220                } else {
1221                        if (file->is_renamed) {
1222                                putc('\0', options->file);
1223                                write_name_quoted(file->from_name, options->file, '\0');
1224                        }
1225                        write_name_quoted(file->name, options->file, '\0');
1226                }
1227        }
1228}
1229
1230struct dirstat_file {
1231        const char *name;
1232        unsigned long changed;
1233};
1234
1235struct dirstat_dir {
1236        struct dirstat_file *files;
1237        int alloc, nr, percent, cumulative;
1238};
1239
1240static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1241{
1242        unsigned long this_dir = 0;
1243        unsigned int sources = 0;
1244
1245        while (dir->nr) {
1246                struct dirstat_file *f = dir->files;
1247                int namelen = strlen(f->name);
1248                unsigned long this;
1249                char *slash;
1250
1251                if (namelen < baselen)
1252                        break;
1253                if (memcmp(f->name, base, baselen))
1254                        break;
1255                slash = strchr(f->name + baselen, '/');
1256                if (slash) {
1257                        int newbaselen = slash + 1 - f->name;
1258                        this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1259                        sources++;
1260                } else {
1261                        this = f->changed;
1262                        dir->files++;
1263                        dir->nr--;
1264                        sources += 2;
1265                }
1266                this_dir += this;
1267        }
1268
1269        /*
1270         * We don't report dirstat's for
1271         *  - the top level
1272         *  - or cases where everything came from a single directory
1273         *    under this directory (sources == 1).
1274         */
1275        if (baselen && sources != 1) {
1276                int permille = this_dir * 1000 / changed;
1277                if (permille) {
1278                        int percent = permille / 10;
1279                        if (percent >= dir->percent) {
1280                                fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1281                                if (!dir->cumulative)
1282                                        return 0;
1283                        }
1284                }
1285        }
1286        return this_dir;
1287}
1288
1289static int dirstat_compare(const void *_a, const void *_b)
1290{
1291        const struct dirstat_file *a = _a;
1292        const struct dirstat_file *b = _b;
1293        return strcmp(a->name, b->name);
1294}
1295
1296static void show_dirstat(struct diff_options *options)
1297{
1298        int i;
1299        unsigned long changed;
1300        struct dirstat_dir dir;
1301        struct diff_queue_struct *q = &diff_queued_diff;
1302
1303        dir.files = NULL;
1304        dir.alloc = 0;
1305        dir.nr = 0;
1306        dir.percent = options->dirstat_percent;
1307        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1308
1309        changed = 0;
1310        for (i = 0; i < q->nr; i++) {
1311                struct diff_filepair *p = q->queue[i];
1312                const char *name;
1313                unsigned long copied, added, damage;
1314
1315                name = p->one->path ? p->one->path : p->two->path;
1316
1317                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1318                        diff_populate_filespec(p->one, 0);
1319                        diff_populate_filespec(p->two, 0);
1320                        diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1321                                               &copied, &added);
1322                        diff_free_filespec_data(p->one);
1323                        diff_free_filespec_data(p->two);
1324                } else if (DIFF_FILE_VALID(p->one)) {
1325                        diff_populate_filespec(p->one, 1);
1326                        copied = added = 0;
1327                        diff_free_filespec_data(p->one);
1328                } else if (DIFF_FILE_VALID(p->two)) {
1329                        diff_populate_filespec(p->two, 1);
1330                        copied = 0;
1331                        added = p->two->size;
1332                        diff_free_filespec_data(p->two);
1333                } else
1334                        continue;
1335
1336                /*
1337                 * Original minus copied is the removed material,
1338                 * added is the new material.  They are both damages
1339                 * made to the preimage. In --dirstat-by-file mode, count
1340                 * damaged files, not damaged lines. This is done by
1341                 * counting only a single damaged line per file.
1342                 */
1343                damage = (p->one->size - copied) + added;
1344                if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1345                        damage = 1;
1346
1347                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1348                dir.files[dir.nr].name = name;
1349                dir.files[dir.nr].changed = damage;
1350                changed += damage;
1351                dir.nr++;
1352        }
1353
1354        /* This can happen even with many files, if everything was renames */
1355        if (!changed)
1356                return;
1357
1358        /* Show all directories with more than x% of the changes */
1359        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1360        gather_dirstat(options->file, &dir, changed, "", 0);
1361}
1362
1363static void free_diffstat_info(struct diffstat_t *diffstat)
1364{
1365        int i;
1366        for (i = 0; i < diffstat->nr; i++) {
1367                struct diffstat_file *f = diffstat->files[i];
1368                if (f->name != f->print_name)
1369                        free(f->print_name);
1370                free(f->name);
1371                free(f->from_name);
1372                free(f);
1373        }
1374        free(diffstat->files);
1375}
1376
1377struct checkdiff_t {
1378        const char *filename;
1379        int lineno;
1380        int conflict_marker_size;
1381        struct diff_options *o;
1382        unsigned ws_rule;
1383        unsigned status;
1384};
1385
1386static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1387{
1388        char firstchar;
1389        int cnt;
1390
1391        if (len < marker_size + 1)
1392                return 0;
1393        firstchar = line[0];
1394        switch (firstchar) {
1395        case '=': case '>': case '<': case '|':
1396                break;
1397        default:
1398                return 0;
1399        }
1400        for (cnt = 1; cnt < marker_size; cnt++)
1401                if (line[cnt] != firstchar)
1402                        return 0;
1403        /* line[1] thru line[marker_size-1] are same as firstchar */
1404        if (len < marker_size + 1 || !isspace(line[marker_size]))
1405                return 0;
1406        return 1;
1407}
1408
1409static void checkdiff_consume(void *priv, char *line, unsigned long len)
1410{
1411        struct checkdiff_t *data = priv;
1412        int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1413        int marker_size = data->conflict_marker_size;
1414        const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1415        const char *reset = diff_get_color(color_diff, DIFF_RESET);
1416        const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1417        char *err;
1418
1419        if (line[0] == '+') {
1420                unsigned bad;
1421                data->lineno++;
1422                if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1423                        data->status |= 1;
1424                        fprintf(data->o->file,
1425                                "%s:%d: leftover conflict marker\n",
1426                                data->filename, data->lineno);
1427                }
1428                bad = ws_check(line + 1, len - 1, data->ws_rule);
1429                if (!bad)
1430                        return;
1431                data->status |= bad;
1432                err = whitespace_error_string(bad);
1433                fprintf(data->o->file, "%s:%d: %s.\n",
1434                        data->filename, data->lineno, err);
1435                free(err);
1436                emit_line(data->o->file, set, reset, line, 1);
1437                ws_check_emit(line + 1, len - 1, data->ws_rule,
1438                              data->o->file, set, reset, ws);
1439        } else if (line[0] == ' ') {
1440                data->lineno++;
1441        } else if (line[0] == '@') {
1442                char *plus = strchr(line, '+');
1443                if (plus)
1444                        data->lineno = strtol(plus, NULL, 10) - 1;
1445                else
1446                        die("invalid diff");
1447        }
1448}
1449
1450static unsigned char *deflate_it(char *data,
1451                                 unsigned long size,
1452                                 unsigned long *result_size)
1453{
1454        int bound;
1455        unsigned char *deflated;
1456        z_stream stream;
1457
1458        memset(&stream, 0, sizeof(stream));
1459        deflateInit(&stream, zlib_compression_level);
1460        bound = deflateBound(&stream, size);
1461        deflated = xmalloc(bound);
1462        stream.next_out = deflated;
1463        stream.avail_out = bound;
1464
1465        stream.next_in = (unsigned char *)data;
1466        stream.avail_in = size;
1467        while (deflate(&stream, Z_FINISH) == Z_OK)
1468                ; /* nothing */
1469        deflateEnd(&stream);
1470        *result_size = stream.total_out;
1471        return deflated;
1472}
1473
1474static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1475{
1476        void *cp;
1477        void *delta;
1478        void *deflated;
1479        void *data;
1480        unsigned long orig_size;
1481        unsigned long delta_size;
1482        unsigned long deflate_size;
1483        unsigned long data_size;
1484
1485        /* We could do deflated delta, or we could do just deflated two,
1486         * whichever is smaller.
1487         */
1488        delta = NULL;
1489        deflated = deflate_it(two->ptr, two->size, &deflate_size);
1490        if (one->size && two->size) {
1491                delta = diff_delta(one->ptr, one->size,
1492                                   two->ptr, two->size,
1493                                   &delta_size, deflate_size);
1494                if (delta) {
1495                        void *to_free = delta;
1496                        orig_size = delta_size;
1497                        delta = deflate_it(delta, delta_size, &delta_size);
1498                        free(to_free);
1499                }
1500        }
1501
1502        if (delta && delta_size < deflate_size) {
1503                fprintf(file, "delta %lu\n", orig_size);
1504                free(deflated);
1505                data = delta;
1506                data_size = delta_size;
1507        }
1508        else {
1509                fprintf(file, "literal %lu\n", two->size);
1510                free(delta);
1511                data = deflated;
1512                data_size = deflate_size;
1513        }
1514
1515        /* emit data encoded in base85 */
1516        cp = data;
1517        while (data_size) {
1518                int bytes = (52 < data_size) ? 52 : data_size;
1519                char line[70];
1520                data_size -= bytes;
1521                if (bytes <= 26)
1522                        line[0] = bytes + 'A' - 1;
1523                else
1524                        line[0] = bytes - 26 + 'a' - 1;
1525                encode_85(line + 1, cp, bytes);
1526                cp = (char *) cp + bytes;
1527                fputs(line, file);
1528                fputc('\n', file);
1529        }
1530        fprintf(file, "\n");
1531        free(data);
1532}
1533
1534static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1535{
1536        fprintf(file, "GIT binary patch\n");
1537        emit_binary_diff_body(file, one, two);
1538        emit_binary_diff_body(file, two, one);
1539}
1540
1541static void diff_filespec_load_driver(struct diff_filespec *one)
1542{
1543        if (!one->driver)
1544                one->driver = userdiff_find_by_path(one->path);
1545        if (!one->driver)
1546                one->driver = userdiff_find_by_name("default");
1547}
1548
1549int diff_filespec_is_binary(struct diff_filespec *one)
1550{
1551        if (one->is_binary == -1) {
1552                diff_filespec_load_driver(one);
1553                if (one->driver->binary != -1)
1554                        one->is_binary = one->driver->binary;
1555                else {
1556                        if (!one->data && DIFF_FILE_VALID(one))
1557                                diff_populate_filespec(one, 0);
1558                        if (one->data)
1559                                one->is_binary = buffer_is_binary(one->data,
1560                                                one->size);
1561                        if (one->is_binary == -1)
1562                                one->is_binary = 0;
1563                }
1564        }
1565        return one->is_binary;
1566}
1567
1568static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1569{
1570        diff_filespec_load_driver(one);
1571        return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1572}
1573
1574static const char *userdiff_word_regex(struct diff_filespec *one)
1575{
1576        diff_filespec_load_driver(one);
1577        return one->driver->word_regex;
1578}
1579
1580void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1581{
1582        if (!options->a_prefix)
1583                options->a_prefix = a;
1584        if (!options->b_prefix)
1585                options->b_prefix = b;
1586}
1587
1588static const char *get_textconv(struct diff_filespec *one)
1589{
1590        if (!DIFF_FILE_VALID(one))
1591                return NULL;
1592        if (!S_ISREG(one->mode))
1593                return NULL;
1594        diff_filespec_load_driver(one);
1595        return one->driver->textconv;
1596}
1597
1598static void builtin_diff(const char *name_a,
1599                         const char *name_b,
1600                         struct diff_filespec *one,
1601                         struct diff_filespec *two,
1602                         const char *xfrm_msg,
1603                         struct diff_options *o,
1604                         int complete_rewrite)
1605{
1606        mmfile_t mf1, mf2;
1607        const char *lbl[2];
1608        char *a_one, *b_two;
1609        const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1610        const char *reset = diff_get_color_opt(o, DIFF_RESET);
1611        const char *a_prefix, *b_prefix;
1612        const char *textconv_one = NULL, *textconv_two = NULL;
1613        struct strbuf header = STRBUF_INIT;
1614
1615        if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1616                        (!one->mode || S_ISGITLINK(one->mode)) &&
1617                        (!two->mode || S_ISGITLINK(two->mode))) {
1618                const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1619                const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1620                show_submodule_summary(o->file, one ? one->path : two->path,
1621                                one->sha1, two->sha1, two->dirty_submodule,
1622                                del, add, reset);
1623                return;
1624        }
1625
1626        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1627                textconv_one = get_textconv(one);
1628                textconv_two = get_textconv(two);
1629        }
1630
1631        diff_set_mnemonic_prefix(o, "a/", "b/");
1632        if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1633                a_prefix = o->b_prefix;
1634                b_prefix = o->a_prefix;
1635        } else {
1636                a_prefix = o->a_prefix;
1637                b_prefix = o->b_prefix;
1638        }
1639
1640        /* Never use a non-valid filename anywhere if at all possible */
1641        name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1642        name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1643
1644        a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1645        b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1646        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1647        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1648        strbuf_addf(&header, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1649        if (lbl[0][0] == '/') {
1650                /* /dev/null */
1651                strbuf_addf(&header, "%snew file mode %06o%s\n", set, two->mode, reset);
1652                if (xfrm_msg && xfrm_msg[0])
1653                        strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1654        }
1655        else if (lbl[1][0] == '/') {
1656                strbuf_addf(&header, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1657                if (xfrm_msg && xfrm_msg[0])
1658                        strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1659        }
1660        else {
1661                if (one->mode != two->mode) {
1662                        strbuf_addf(&header, "%sold mode %06o%s\n", set, one->mode, reset);
1663                        strbuf_addf(&header, "%snew mode %06o%s\n", set, two->mode, reset);
1664                }
1665                if (xfrm_msg && xfrm_msg[0])
1666                        strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1667
1668                /*
1669                 * we do not run diff between different kind
1670                 * of objects.
1671                 */
1672                if ((one->mode ^ two->mode) & S_IFMT)
1673                        goto free_ab_and_return;
1674                if (complete_rewrite &&
1675                    (textconv_one || !diff_filespec_is_binary(one)) &&
1676                    (textconv_two || !diff_filespec_is_binary(two))) {
1677                        fprintf(o->file, "%s", header.buf);
1678                        strbuf_reset(&header);
1679                        emit_rewrite_diff(name_a, name_b, one, two,
1680                                                textconv_one, textconv_two, o);
1681                        o->found_changes = 1;
1682                        goto free_ab_and_return;
1683                }
1684        }
1685
1686        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1687                die("unable to read files to diff");
1688
1689        if (!DIFF_OPT_TST(o, TEXT) &&
1690            ( (diff_filespec_is_binary(one) && !textconv_one) ||
1691              (diff_filespec_is_binary(two) && !textconv_two) )) {
1692                /* Quite common confusing case */
1693                if (mf1.size == mf2.size &&
1694                    !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1695                        goto free_ab_and_return;
1696                fprintf(o->file, "%s", header.buf);
1697                strbuf_reset(&header);
1698                if (DIFF_OPT_TST(o, BINARY))
1699                        emit_binary_diff(o->file, &mf1, &mf2);
1700                else
1701                        fprintf(o->file, "Binary files %s and %s differ\n",
1702                                lbl[0], lbl[1]);
1703                o->found_changes = 1;
1704        }
1705        else {
1706                /* Crazy xdl interfaces.. */
1707                const char *diffopts = getenv("GIT_DIFF_OPTS");
1708                xpparam_t xpp;
1709                xdemitconf_t xecfg;
1710                struct emit_callback ecbdata;
1711                const struct userdiff_funcname *pe;
1712
1713                if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS)) {
1714                        fprintf(o->file, "%s", header.buf);
1715                        strbuf_reset(&header);
1716                }
1717
1718                if (textconv_one) {
1719                        size_t size;
1720                        mf1.ptr = run_textconv(textconv_one, one, &size);
1721                        if (!mf1.ptr)
1722                                die("unable to read files to diff");
1723                        mf1.size = size;
1724                }
1725                if (textconv_two) {
1726                        size_t size;
1727                        mf2.ptr = run_textconv(textconv_two, two, &size);
1728                        if (!mf2.ptr)
1729                                die("unable to read files to diff");
1730                        mf2.size = size;
1731                }
1732
1733                pe = diff_funcname_pattern(one);
1734                if (!pe)
1735                        pe = diff_funcname_pattern(two);
1736
1737                memset(&xpp, 0, sizeof(xpp));
1738                memset(&xecfg, 0, sizeof(xecfg));
1739                memset(&ecbdata, 0, sizeof(ecbdata));
1740                ecbdata.label_path = lbl;
1741                ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1742                ecbdata.found_changesp = &o->found_changes;
1743                ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1744                if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1745                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
1746                ecbdata.file = o->file;
1747                ecbdata.header = header.len ? &header : NULL;
1748                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1749                xecfg.ctxlen = o->context;
1750                xecfg.interhunkctxlen = o->interhunkcontext;
1751                xecfg.flags = XDL_EMIT_FUNCNAMES;
1752                if (pe)
1753                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1754                if (!diffopts)
1755                        ;
1756                else if (!prefixcmp(diffopts, "--unified="))
1757                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1758                else if (!prefixcmp(diffopts, "-u"))
1759                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1760                if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1761                        ecbdata.diff_words =
1762                                xcalloc(1, sizeof(struct diff_words_data));
1763                        ecbdata.diff_words->file = o->file;
1764                        if (!o->word_regex)
1765                                o->word_regex = userdiff_word_regex(one);
1766                        if (!o->word_regex)
1767                                o->word_regex = userdiff_word_regex(two);
1768                        if (!o->word_regex)
1769                                o->word_regex = diff_word_regex_cfg;
1770                        if (o->word_regex) {
1771                                ecbdata.diff_words->word_regex = (regex_t *)
1772                                        xmalloc(sizeof(regex_t));
1773                                if (regcomp(ecbdata.diff_words->word_regex,
1774                                                o->word_regex,
1775                                                REG_EXTENDED | REG_NEWLINE))
1776                                        die ("Invalid regular expression: %s",
1777                                                        o->word_regex);
1778                        }
1779                }
1780                xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1781                              &xpp, &xecfg);
1782                if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1783                        free_diff_words_data(&ecbdata);
1784                if (textconv_one)
1785                        free(mf1.ptr);
1786                if (textconv_two)
1787                        free(mf2.ptr);
1788                xdiff_clear_find_func(&xecfg);
1789        }
1790
1791 free_ab_and_return:
1792        strbuf_release(&header);
1793        diff_free_filespec_data(one);
1794        diff_free_filespec_data(two);
1795        free(a_one);
1796        free(b_two);
1797        return;
1798}
1799
1800static void builtin_diffstat(const char *name_a, const char *name_b,
1801                             struct diff_filespec *one,
1802                             struct diff_filespec *two,
1803                             struct diffstat_t *diffstat,
1804                             struct diff_options *o,
1805                             int complete_rewrite)
1806{
1807        mmfile_t mf1, mf2;
1808        struct diffstat_file *data;
1809
1810        data = diffstat_add(diffstat, name_a, name_b);
1811
1812        if (!one || !two) {
1813                data->is_unmerged = 1;
1814                return;
1815        }
1816        if (complete_rewrite) {
1817                diff_populate_filespec(one, 0);
1818                diff_populate_filespec(two, 0);
1819                data->deleted = count_lines(one->data, one->size);
1820                data->added = count_lines(two->data, two->size);
1821                goto free_and_return;
1822        }
1823        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1824                die("unable to read files to diff");
1825
1826        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1827                data->is_binary = 1;
1828                data->added = mf2.size;
1829                data->deleted = mf1.size;
1830        } else {
1831                /* Crazy xdl interfaces.. */
1832                xpparam_t xpp;
1833                xdemitconf_t xecfg;
1834
1835                memset(&xpp, 0, sizeof(xpp));
1836                memset(&xecfg, 0, sizeof(xecfg));
1837                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1838                xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1839                              &xpp, &xecfg);
1840        }
1841
1842 free_and_return:
1843        diff_free_filespec_data(one);
1844        diff_free_filespec_data(two);
1845}
1846
1847static void builtin_checkdiff(const char *name_a, const char *name_b,
1848                              const char *attr_path,
1849                              struct diff_filespec *one,
1850                              struct diff_filespec *two,
1851                              struct diff_options *o)
1852{
1853        mmfile_t mf1, mf2;
1854        struct checkdiff_t data;
1855
1856        if (!two)
1857                return;
1858
1859        memset(&data, 0, sizeof(data));
1860        data.filename = name_b ? name_b : name_a;
1861        data.lineno = 0;
1862        data.o = o;
1863        data.ws_rule = whitespace_rule(attr_path);
1864        data.conflict_marker_size = ll_merge_marker_size(attr_path);
1865
1866        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1867                die("unable to read files to diff");
1868
1869        /*
1870         * All the other codepaths check both sides, but not checking
1871         * the "old" side here is deliberate.  We are checking the newly
1872         * introduced changes, and as long as the "new" side is text, we
1873         * can and should check what it introduces.
1874         */
1875        if (diff_filespec_is_binary(two))
1876                goto free_and_return;
1877        else {
1878                /* Crazy xdl interfaces.. */
1879                xpparam_t xpp;
1880                xdemitconf_t xecfg;
1881
1882                memset(&xpp, 0, sizeof(xpp));
1883                memset(&xecfg, 0, sizeof(xecfg));
1884                xecfg.ctxlen = 1; /* at least one context line */
1885                xpp.flags = XDF_NEED_MINIMAL;
1886                xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1887                              &xpp, &xecfg);
1888
1889                if (data.ws_rule & WS_BLANK_AT_EOF) {
1890                        struct emit_callback ecbdata;
1891                        int blank_at_eof;
1892
1893                        ecbdata.ws_rule = data.ws_rule;
1894                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
1895                        blank_at_eof = ecbdata.blank_at_eof_in_preimage;
1896
1897                        if (blank_at_eof) {
1898                                static char *err;
1899                                if (!err)
1900                                        err = whitespace_error_string(WS_BLANK_AT_EOF);
1901                                fprintf(o->file, "%s:%d: %s.\n",
1902                                        data.filename, blank_at_eof, err);
1903                                data.status = 1; /* report errors */
1904                        }
1905                }
1906        }
1907 free_and_return:
1908        diff_free_filespec_data(one);
1909        diff_free_filespec_data(two);
1910        if (data.status)
1911                DIFF_OPT_SET(o, CHECK_FAILED);
1912}
1913
1914struct diff_filespec *alloc_filespec(const char *path)
1915{
1916        int namelen = strlen(path);
1917        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1918
1919        memset(spec, 0, sizeof(*spec));
1920        spec->path = (char *)(spec + 1);
1921        memcpy(spec->path, path, namelen+1);
1922        spec->count = 1;
1923        spec->is_binary = -1;
1924        return spec;
1925}
1926
1927void free_filespec(struct diff_filespec *spec)
1928{
1929        if (!--spec->count) {
1930                diff_free_filespec_data(spec);
1931                free(spec);
1932        }
1933}
1934
1935void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1936                   unsigned short mode)
1937{
1938        if (mode) {
1939                spec->mode = canon_mode(mode);
1940                hashcpy(spec->sha1, sha1);
1941                spec->sha1_valid = !is_null_sha1(sha1);
1942        }
1943}
1944
1945/*
1946 * Given a name and sha1 pair, if the index tells us the file in
1947 * the work tree has that object contents, return true, so that
1948 * prepare_temp_file() does not have to inflate and extract.
1949 */
1950static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1951{
1952        struct cache_entry *ce;
1953        struct stat st;
1954        int pos, len;
1955
1956        /*
1957         * We do not read the cache ourselves here, because the
1958         * benchmark with my previous version that always reads cache
1959         * shows that it makes things worse for diff-tree comparing
1960         * two linux-2.6 kernel trees in an already checked out work
1961         * tree.  This is because most diff-tree comparisons deal with
1962         * only a small number of files, while reading the cache is
1963         * expensive for a large project, and its cost outweighs the
1964         * savings we get by not inflating the object to a temporary
1965         * file.  Practically, this code only helps when we are used
1966         * by diff-cache --cached, which does read the cache before
1967         * calling us.
1968         */
1969        if (!active_cache)
1970                return 0;
1971
1972        /* We want to avoid the working directory if our caller
1973         * doesn't need the data in a normal file, this system
1974         * is rather slow with its stat/open/mmap/close syscalls,
1975         * and the object is contained in a pack file.  The pack
1976         * is probably already open and will be faster to obtain
1977         * the data through than the working directory.  Loose
1978         * objects however would tend to be slower as they need
1979         * to be individually opened and inflated.
1980         */
1981        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1982                return 0;
1983
1984        len = strlen(name);
1985        pos = cache_name_pos(name, len);
1986        if (pos < 0)
1987                return 0;
1988        ce = active_cache[pos];
1989
1990        /*
1991         * This is not the sha1 we are looking for, or
1992         * unreusable because it is not a regular file.
1993         */
1994        if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1995                return 0;
1996
1997        /*
1998         * If ce is marked as "assume unchanged", there is no
1999         * guarantee that work tree matches what we are looking for.
2000         */
2001        if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2002                return 0;
2003
2004        /*
2005         * If ce matches the file in the work tree, we can reuse it.
2006         */
2007        if (ce_uptodate(ce) ||
2008            (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2009                return 1;
2010
2011        return 0;
2012}
2013
2014static int populate_from_stdin(struct diff_filespec *s)
2015{
2016        struct strbuf buf = STRBUF_INIT;
2017        size_t size = 0;
2018
2019        if (strbuf_read(&buf, 0, 0) < 0)
2020                return error("error while reading from stdin %s",
2021                                     strerror(errno));
2022
2023        s->should_munmap = 0;
2024        s->data = strbuf_detach(&buf, &size);
2025        s->size = size;
2026        s->should_free = 1;
2027        return 0;
2028}
2029
2030static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2031{
2032        int len;
2033        char *data = xmalloc(100), *dirty = "";
2034
2035        /* Are we looking at the work tree? */
2036        if (s->dirty_submodule)
2037                dirty = "-dirty";
2038
2039        len = snprintf(data, 100,
2040                       "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2041        s->data = data;
2042        s->size = len;
2043        s->should_free = 1;
2044        if (size_only) {
2045                s->data = NULL;
2046                free(data);
2047        }
2048        return 0;
2049}
2050
2051/*
2052 * While doing rename detection and pickaxe operation, we may need to
2053 * grab the data for the blob (or file) for our own in-core comparison.
2054 * diff_filespec has data and size fields for this purpose.
2055 */
2056int diff_populate_filespec(struct diff_filespec *s, int size_only)
2057{
2058        int err = 0;
2059        if (!DIFF_FILE_VALID(s))
2060                die("internal error: asking to populate invalid file.");
2061        if (S_ISDIR(s->mode))
2062                return -1;
2063
2064        if (s->data)
2065                return 0;
2066
2067        if (size_only && 0 < s->size)
2068                return 0;
2069
2070        if (S_ISGITLINK(s->mode))
2071                return diff_populate_gitlink(s, size_only);
2072
2073        if (!s->sha1_valid ||
2074            reuse_worktree_file(s->path, s->sha1, 0)) {
2075                struct strbuf buf = STRBUF_INIT;
2076                struct stat st;
2077                int fd;
2078
2079                if (!strcmp(s->path, "-"))
2080                        return populate_from_stdin(s);
2081
2082                if (lstat(s->path, &st) < 0) {
2083                        if (errno == ENOENT) {
2084                        err_empty:
2085                                err = -1;
2086                        empty:
2087                                s->data = (char *)"";
2088                                s->size = 0;
2089                                return err;
2090                        }
2091                }
2092                s->size = xsize_t(st.st_size);
2093                if (!s->size)
2094                        goto empty;
2095                if (S_ISLNK(st.st_mode)) {
2096                        struct strbuf sb = STRBUF_INIT;
2097
2098                        if (strbuf_readlink(&sb, s->path, s->size))
2099                                goto err_empty;
2100                        s->size = sb.len;
2101                        s->data = strbuf_detach(&sb, NULL);
2102                        s->should_free = 1;
2103                        return 0;
2104                }
2105                if (size_only)
2106                        return 0;
2107                fd = open(s->path, O_RDONLY);
2108                if (fd < 0)
2109                        goto err_empty;
2110                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2111                close(fd);
2112                s->should_munmap = 1;
2113
2114                /*
2115                 * Convert from working tree format to canonical git format
2116                 */
2117                if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2118                        size_t size = 0;
2119                        munmap(s->data, s->size);
2120                        s->should_munmap = 0;
2121                        s->data = strbuf_detach(&buf, &size);
2122                        s->size = size;
2123                        s->should_free = 1;
2124                }
2125        }
2126        else {
2127                enum object_type type;
2128                if (size_only)
2129                        type = sha1_object_info(s->sha1, &s->size);
2130                else {
2131                        s->data = read_sha1_file(s->sha1, &type, &s->size);
2132                        s->should_free = 1;
2133                }
2134        }
2135        return 0;
2136}
2137
2138void diff_free_filespec_blob(struct diff_filespec *s)
2139{
2140        if (s->should_free)
2141                free(s->data);
2142        else if (s->should_munmap)
2143                munmap(s->data, s->size);
2144
2145        if (s->should_free || s->should_munmap) {
2146                s->should_free = s->should_munmap = 0;
2147                s->data = NULL;
2148        }
2149}
2150
2151void diff_free_filespec_data(struct diff_filespec *s)
2152{
2153        diff_free_filespec_blob(s);
2154        free(s->cnt_data);
2155        s->cnt_data = NULL;
2156}
2157
2158static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2159                           void *blob,
2160                           unsigned long size,
2161                           const unsigned char *sha1,
2162                           int mode)
2163{
2164        int fd;
2165        struct strbuf buf = STRBUF_INIT;
2166        struct strbuf template = STRBUF_INIT;
2167        char *path_dup = xstrdup(path);
2168        const char *base = basename(path_dup);
2169
2170        /* Generate "XXXXXX_basename.ext" */
2171        strbuf_addstr(&template, "XXXXXX_");
2172        strbuf_addstr(&template, base);
2173
2174        fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2175                        strlen(base) + 1);
2176        if (fd < 0)
2177                die_errno("unable to create temp-file");
2178        if (convert_to_working_tree(path,
2179                        (const char *)blob, (size_t)size, &buf)) {
2180                blob = buf.buf;
2181                size = buf.len;
2182        }
2183        if (write_in_full(fd, blob, size) != size)
2184                die_errno("unable to write temp-file");
2185        close(fd);
2186        temp->name = temp->tmp_path;
2187        strcpy(temp->hex, sha1_to_hex(sha1));
2188        temp->hex[40] = 0;
2189        sprintf(temp->mode, "%06o", mode);
2190        strbuf_release(&buf);
2191        strbuf_release(&template);
2192        free(path_dup);
2193}
2194
2195static struct diff_tempfile *prepare_temp_file(const char *name,
2196                struct diff_filespec *one)
2197{
2198        struct diff_tempfile *temp = claim_diff_tempfile();
2199
2200        if (!DIFF_FILE_VALID(one)) {
2201        not_a_valid_file:
2202                /* A '-' entry produces this for file-2, and
2203                 * a '+' entry produces this for file-1.
2204                 */
2205                temp->name = "/dev/null";
2206                strcpy(temp->hex, ".");
2207                strcpy(temp->mode, ".");
2208                return temp;
2209        }
2210
2211        if (!remove_tempfile_installed) {
2212                atexit(remove_tempfile);
2213                sigchain_push_common(remove_tempfile_on_signal);
2214                remove_tempfile_installed = 1;
2215        }
2216
2217        if (!one->sha1_valid ||
2218            reuse_worktree_file(name, one->sha1, 1)) {
2219                struct stat st;
2220                if (lstat(name, &st) < 0) {
2221                        if (errno == ENOENT)
2222                                goto not_a_valid_file;
2223                        die_errno("stat(%s)", name);
2224                }
2225                if (S_ISLNK(st.st_mode)) {
2226                        struct strbuf sb = STRBUF_INIT;
2227                        if (strbuf_readlink(&sb, name, st.st_size) < 0)
2228                                die_errno("readlink(%s)", name);
2229                        prep_temp_blob(name, temp, sb.buf, sb.len,
2230                                       (one->sha1_valid ?
2231                                        one->sha1 : null_sha1),
2232                                       (one->sha1_valid ?
2233                                        one->mode : S_IFLNK));
2234                        strbuf_release(&sb);
2235                }
2236                else {
2237                        /* we can borrow from the file in the work tree */
2238                        temp->name = name;
2239                        if (!one->sha1_valid)
2240                                strcpy(temp->hex, sha1_to_hex(null_sha1));
2241                        else
2242                                strcpy(temp->hex, sha1_to_hex(one->sha1));
2243                        /* Even though we may sometimes borrow the
2244                         * contents from the work tree, we always want
2245                         * one->mode.  mode is trustworthy even when
2246                         * !(one->sha1_valid), as long as
2247                         * DIFF_FILE_VALID(one).
2248                         */
2249                        sprintf(temp->mode, "%06o", one->mode);
2250                }
2251                return temp;
2252        }
2253        else {
2254                if (diff_populate_filespec(one, 0))
2255                        die("cannot read data blob for %s", one->path);
2256                prep_temp_blob(name, temp, one->data, one->size,
2257                               one->sha1, one->mode);
2258        }
2259        return temp;
2260}
2261
2262/* An external diff command takes:
2263 *
2264 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2265 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2266 *
2267 */
2268static void run_external_diff(const char *pgm,
2269                              const char *name,
2270                              const char *other,
2271                              struct diff_filespec *one,
2272                              struct diff_filespec *two,
2273                              const char *xfrm_msg,
2274                              int complete_rewrite)
2275{
2276        const char *spawn_arg[10];
2277        int retval;
2278        const char **arg = &spawn_arg[0];
2279
2280        if (one && two) {
2281                struct diff_tempfile *temp_one, *temp_two;
2282                const char *othername = (other ? other : name);
2283                temp_one = prepare_temp_file(name, one);
2284                temp_two = prepare_temp_file(othername, two);
2285                *arg++ = pgm;
2286                *arg++ = name;
2287                *arg++ = temp_one->name;
2288                *arg++ = temp_one->hex;
2289                *arg++ = temp_one->mode;
2290                *arg++ = temp_two->name;
2291                *arg++ = temp_two->hex;
2292                *arg++ = temp_two->mode;
2293                if (other) {
2294                        *arg++ = other;
2295                        *arg++ = xfrm_msg;
2296                }
2297        } else {
2298                *arg++ = pgm;
2299                *arg++ = name;
2300        }
2301        *arg = NULL;
2302        fflush(NULL);
2303        retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2304        remove_tempfile();
2305        if (retval) {
2306                fprintf(stderr, "external diff died, stopping at %s.\n", name);
2307                exit(1);
2308        }
2309}
2310
2311static int similarity_index(struct diff_filepair *p)
2312{
2313        return p->score * 100 / MAX_SCORE;
2314}
2315
2316static void fill_metainfo(struct strbuf *msg,
2317                          const char *name,
2318                          const char *other,
2319                          struct diff_filespec *one,
2320                          struct diff_filespec *two,
2321                          struct diff_options *o,
2322                          struct diff_filepair *p)
2323{
2324        strbuf_init(msg, PATH_MAX * 2 + 300);
2325        switch (p->status) {
2326        case DIFF_STATUS_COPIED:
2327                strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2328                strbuf_addstr(msg, "\ncopy from ");
2329                quote_c_style(name, msg, NULL, 0);
2330                strbuf_addstr(msg, "\ncopy to ");
2331                quote_c_style(other, msg, NULL, 0);
2332                strbuf_addch(msg, '\n');
2333                break;
2334        case DIFF_STATUS_RENAMED:
2335                strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2336                strbuf_addstr(msg, "\nrename from ");
2337                quote_c_style(name, msg, NULL, 0);
2338                strbuf_addstr(msg, "\nrename to ");
2339                quote_c_style(other, msg, NULL, 0);
2340                strbuf_addch(msg, '\n');
2341                break;
2342        case DIFF_STATUS_MODIFIED:
2343                if (p->score) {
2344                        strbuf_addf(msg, "dissimilarity index %d%%\n",
2345                                    similarity_index(p));
2346                        break;
2347                }
2348                /* fallthru */
2349        default:
2350                /* nothing */
2351                ;
2352        }
2353        if (one && two && hashcmp(one->sha1, two->sha1)) {
2354                int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2355
2356                if (DIFF_OPT_TST(o, BINARY)) {
2357                        mmfile_t mf;
2358                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2359                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2360                                abbrev = 40;
2361                }
2362                strbuf_addf(msg, "index %.*s..%.*s",
2363                            abbrev, sha1_to_hex(one->sha1),
2364                            abbrev, sha1_to_hex(two->sha1));
2365                if (one->mode == two->mode)
2366                        strbuf_addf(msg, " %06o", one->mode);
2367                strbuf_addch(msg, '\n');
2368        }
2369        if (msg->len)
2370                strbuf_setlen(msg, msg->len - 1);
2371}
2372
2373static void run_diff_cmd(const char *pgm,
2374                         const char *name,
2375                         const char *other,
2376                         const char *attr_path,
2377                         struct diff_filespec *one,
2378                         struct diff_filespec *two,
2379                         struct strbuf *msg,
2380                         struct diff_options *o,
2381                         struct diff_filepair *p)
2382{
2383        const char *xfrm_msg = NULL;
2384        int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2385
2386        if (msg) {
2387                fill_metainfo(msg, name, other, one, two, o, p);
2388                xfrm_msg = msg->len ? msg->buf : NULL;
2389        }
2390
2391        if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2392                pgm = NULL;
2393        else {
2394                struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2395                if (drv && drv->external)
2396                        pgm = drv->external;
2397        }
2398
2399        if (pgm) {
2400                run_external_diff(pgm, name, other, one, two, xfrm_msg,
2401                                  complete_rewrite);
2402                return;
2403        }
2404        if (one && two)
2405                builtin_diff(name, other ? other : name,
2406                             one, two, xfrm_msg, o, complete_rewrite);
2407        else
2408                fprintf(o->file, "* Unmerged path %s\n", name);
2409}
2410
2411static void diff_fill_sha1_info(struct diff_filespec *one)
2412{
2413        if (DIFF_FILE_VALID(one)) {
2414                if (!one->sha1_valid) {
2415                        struct stat st;
2416                        if (!strcmp(one->path, "-")) {
2417                                hashcpy(one->sha1, null_sha1);
2418                                return;
2419                        }
2420                        if (lstat(one->path, &st) < 0)
2421                                die_errno("stat '%s'", one->path);
2422                        if (index_path(one->sha1, one->path, &st, 0))
2423                                die("cannot hash %s", one->path);
2424                }
2425        }
2426        else
2427                hashclr(one->sha1);
2428}
2429
2430static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2431{
2432        /* Strip the prefix but do not molest /dev/null and absolute paths */
2433        if (*namep && **namep != '/')
2434                *namep += prefix_length;
2435        if (*otherp && **otherp != '/')
2436                *otherp += prefix_length;
2437}
2438
2439static void run_diff(struct diff_filepair *p, struct diff_options *o)
2440{
2441        const char *pgm = external_diff();
2442        struct strbuf msg;
2443        struct diff_filespec *one = p->one;
2444        struct diff_filespec *two = p->two;
2445        const char *name;
2446        const char *other;
2447        const char *attr_path;
2448
2449        name  = p->one->path;
2450        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2451        attr_path = name;
2452        if (o->prefix_length)
2453                strip_prefix(o->prefix_length, &name, &other);
2454
2455        if (DIFF_PAIR_UNMERGED(p)) {
2456                run_diff_cmd(pgm, name, NULL, attr_path,
2457                             NULL, NULL, NULL, o, p);
2458                return;
2459        }
2460
2461        diff_fill_sha1_info(one);
2462        diff_fill_sha1_info(two);
2463
2464        if (!pgm &&
2465            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2466            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2467                /*
2468                 * a filepair that changes between file and symlink
2469                 * needs to be split into deletion and creation.
2470                 */
2471                struct diff_filespec *null = alloc_filespec(two->path);
2472                run_diff_cmd(NULL, name, other, attr_path,
2473                             one, null, &msg, o, p);
2474                free(null);
2475                strbuf_release(&msg);
2476
2477                null = alloc_filespec(one->path);
2478                run_diff_cmd(NULL, name, other, attr_path,
2479                             null, two, &msg, o, p);
2480                free(null);
2481        }
2482        else
2483                run_diff_cmd(pgm, name, other, attr_path,
2484                             one, two, &msg, o, p);
2485
2486        strbuf_release(&msg);
2487}
2488
2489static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2490                         struct diffstat_t *diffstat)
2491{
2492        const char *name;
2493        const char *other;
2494        int complete_rewrite = 0;
2495
2496        if (DIFF_PAIR_UNMERGED(p)) {
2497                /* unmerged */
2498                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2499                return;
2500        }
2501
2502        name = p->one->path;
2503        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2504
2505        if (o->prefix_length)
2506                strip_prefix(o->prefix_length, &name, &other);
2507
2508        diff_fill_sha1_info(p->one);
2509        diff_fill_sha1_info(p->two);
2510
2511        if (p->status == DIFF_STATUS_MODIFIED && p->score)
2512                complete_rewrite = 1;
2513        builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2514}
2515
2516static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2517{
2518        const char *name;
2519        const char *other;
2520        const char *attr_path;
2521
2522        if (DIFF_PAIR_UNMERGED(p)) {
2523                /* unmerged */
2524                return;
2525        }
2526
2527        name = p->one->path;
2528        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2529        attr_path = other ? other : name;
2530
2531        if (o->prefix_length)
2532                strip_prefix(o->prefix_length, &name, &other);
2533
2534        diff_fill_sha1_info(p->one);
2535        diff_fill_sha1_info(p->two);
2536
2537        builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2538}
2539
2540void diff_setup(struct diff_options *options)
2541{
2542        memset(options, 0, sizeof(*options));
2543
2544        options->file = stdout;
2545
2546        options->line_termination = '\n';
2547        options->break_opt = -1;
2548        options->rename_limit = -1;
2549        options->dirstat_percent = 3;
2550        options->context = 3;
2551
2552        options->change = diff_change;
2553        options->add_remove = diff_addremove;
2554        if (diff_use_color_default > 0)
2555                DIFF_OPT_SET(options, COLOR_DIFF);
2556        options->detect_rename = diff_detect_rename_default;
2557
2558        if (!diff_mnemonic_prefix) {
2559                options->a_prefix = "a/";
2560                options->b_prefix = "b/";
2561        }
2562}
2563
2564int diff_setup_done(struct diff_options *options)
2565{
2566        int count = 0;
2567
2568        if (options->output_format & DIFF_FORMAT_NAME)
2569                count++;
2570        if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2571                count++;
2572        if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2573                count++;
2574        if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2575                count++;
2576        if (count > 1)
2577                die("--name-only, --name-status, --check and -s are mutually exclusive");
2578
2579        /*
2580         * Most of the time we can say "there are changes"
2581         * only by checking if there are changed paths, but
2582         * --ignore-whitespace* options force us to look
2583         * inside contents.
2584         */
2585
2586        if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2587            DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2588            DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2589                DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2590        else
2591                DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2592
2593        if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2594                options->detect_rename = DIFF_DETECT_COPY;
2595
2596        if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2597                options->prefix = NULL;
2598        if (options->prefix)
2599                options->prefix_length = strlen(options->prefix);
2600        else
2601                options->prefix_length = 0;
2602
2603        if (options->output_format & (DIFF_FORMAT_NAME |
2604                                      DIFF_FORMAT_NAME_STATUS |
2605                                      DIFF_FORMAT_CHECKDIFF |
2606                                      DIFF_FORMAT_NO_OUTPUT))
2607                options->output_format &= ~(DIFF_FORMAT_RAW |
2608                                            DIFF_FORMAT_NUMSTAT |
2609                                            DIFF_FORMAT_DIFFSTAT |
2610                                            DIFF_FORMAT_SHORTSTAT |
2611                                            DIFF_FORMAT_DIRSTAT |
2612                                            DIFF_FORMAT_SUMMARY |
2613                                            DIFF_FORMAT_PATCH);
2614
2615        /*
2616         * These cases always need recursive; we do not drop caller-supplied
2617         * recursive bits for other formats here.
2618         */
2619        if (options->output_format & (DIFF_FORMAT_PATCH |
2620                                      DIFF_FORMAT_NUMSTAT |
2621                                      DIFF_FORMAT_DIFFSTAT |
2622                                      DIFF_FORMAT_SHORTSTAT |
2623                                      DIFF_FORMAT_DIRSTAT |
2624                                      DIFF_FORMAT_SUMMARY |
2625                                      DIFF_FORMAT_CHECKDIFF))
2626                DIFF_OPT_SET(options, RECURSIVE);
2627        /*
2628         * Also pickaxe would not work very well if you do not say recursive
2629         */
2630        if (options->pickaxe)
2631                DIFF_OPT_SET(options, RECURSIVE);
2632        /*
2633         * When patches are generated, submodules diffed against the work tree
2634         * must be checked for dirtiness too so it can be shown in the output
2635         */
2636        if (options->output_format & DIFF_FORMAT_PATCH)
2637                DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2638
2639        if (options->detect_rename && options->rename_limit < 0)
2640                options->rename_limit = diff_rename_limit_default;
2641        if (options->setup & DIFF_SETUP_USE_CACHE) {
2642                if (!active_cache)
2643                        /* read-cache does not die even when it fails
2644                         * so it is safe for us to do this here.  Also
2645                         * it does not smudge active_cache or active_nr
2646                         * when it fails, so we do not have to worry about
2647                         * cleaning it up ourselves either.
2648                         */
2649                        read_cache();
2650        }
2651        if (options->abbrev <= 0 || 40 < options->abbrev)
2652                options->abbrev = 40; /* full */
2653
2654        /*
2655         * It does not make sense to show the first hit we happened
2656         * to have found.  It does not make sense not to return with
2657         * exit code in such a case either.
2658         */
2659        if (DIFF_OPT_TST(options, QUICK)) {
2660                options->output_format = DIFF_FORMAT_NO_OUTPUT;
2661                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2662        }
2663
2664        return 0;
2665}
2666
2667static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2668{
2669        char c, *eq;
2670        int len;
2671
2672        if (*arg != '-')
2673                return 0;
2674        c = *++arg;
2675        if (!c)
2676                return 0;
2677        if (c == arg_short) {
2678                c = *++arg;
2679                if (!c)
2680                        return 1;
2681                if (val && isdigit(c)) {
2682                        char *end;
2683                        int n = strtoul(arg, &end, 10);
2684                        if (*end)
2685                                return 0;
2686                        *val = n;
2687                        return 1;
2688                }
2689                return 0;
2690        }
2691        if (c != '-')
2692                return 0;
2693        arg++;
2694        eq = strchr(arg, '=');
2695        if (eq)
2696                len = eq - arg;
2697        else
2698                len = strlen(arg);
2699        if (!len || strncmp(arg, arg_long, len))
2700                return 0;
2701        if (eq) {
2702                int n;
2703                char *end;
2704                if (!isdigit(*++eq))
2705                        return 0;
2706                n = strtoul(eq, &end, 10);
2707                if (*end)
2708                        return 0;
2709                *val = n;
2710        }
2711        return 1;
2712}
2713
2714static int diff_scoreopt_parse(const char *opt);
2715
2716int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2717{
2718        const char *arg = av[0];
2719
2720        /* Output format options */
2721        if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2722                options->output_format |= DIFF_FORMAT_PATCH;
2723        else if (opt_arg(arg, 'U', "unified", &options->context))
2724                options->output_format |= DIFF_FORMAT_PATCH;
2725        else if (!strcmp(arg, "--raw"))
2726                options->output_format |= DIFF_FORMAT_RAW;
2727        else if (!strcmp(arg, "--patch-with-raw"))
2728                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2729        else if (!strcmp(arg, "--numstat"))
2730                options->output_format |= DIFF_FORMAT_NUMSTAT;
2731        else if (!strcmp(arg, "--shortstat"))
2732                options->output_format |= DIFF_FORMAT_SHORTSTAT;
2733        else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2734                options->output_format |= DIFF_FORMAT_DIRSTAT;
2735        else if (!strcmp(arg, "--cumulative")) {
2736                options->output_format |= DIFF_FORMAT_DIRSTAT;
2737                DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2738        } else if (opt_arg(arg, 0, "dirstat-by-file",
2739                           &options->dirstat_percent)) {
2740                options->output_format |= DIFF_FORMAT_DIRSTAT;
2741                DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2742        }
2743        else if (!strcmp(arg, "--check"))
2744                options->output_format |= DIFF_FORMAT_CHECKDIFF;
2745        else if (!strcmp(arg, "--summary"))
2746                options->output_format |= DIFF_FORMAT_SUMMARY;
2747        else if (!strcmp(arg, "--patch-with-stat"))
2748                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2749        else if (!strcmp(arg, "--name-only"))
2750                options->output_format |= DIFF_FORMAT_NAME;
2751        else if (!strcmp(arg, "--name-status"))
2752                options->output_format |= DIFF_FORMAT_NAME_STATUS;
2753        else if (!strcmp(arg, "-s"))
2754                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2755        else if (!prefixcmp(arg, "--stat")) {
2756                char *end;
2757                int width = options->stat_width;
2758                int name_width = options->stat_name_width;
2759                arg += 6;
2760                end = (char *)arg;
2761
2762                switch (*arg) {
2763                case '-':
2764                        if (!prefixcmp(arg, "-width="))
2765                                width = strtoul(arg + 7, &end, 10);
2766                        else if (!prefixcmp(arg, "-name-width="))
2767                                name_width = strtoul(arg + 12, &end, 10);
2768                        break;
2769                case '=':
2770                        width = strtoul(arg+1, &end, 10);
2771                        if (*end == ',')
2772                                name_width = strtoul(end+1, &end, 10);
2773                }
2774
2775                /* Important! This checks all the error cases! */
2776                if (*end)
2777                        return 0;
2778                options->output_format |= DIFF_FORMAT_DIFFSTAT;
2779                options->stat_name_width = name_width;
2780                options->stat_width = width;
2781        }
2782
2783        /* renames options */
2784        else if (!prefixcmp(arg, "-B")) {
2785                if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2786                        return -1;
2787        }
2788        else if (!prefixcmp(arg, "-M")) {
2789                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2790                        return -1;
2791                options->detect_rename = DIFF_DETECT_RENAME;
2792        }
2793        else if (!prefixcmp(arg, "-C")) {
2794                if (options->detect_rename == DIFF_DETECT_COPY)
2795                        DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2796                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2797                        return -1;
2798                options->detect_rename = DIFF_DETECT_COPY;
2799        }
2800        else if (!strcmp(arg, "--no-renames"))
2801                options->detect_rename = 0;
2802        else if (!strcmp(arg, "--relative"))
2803                DIFF_OPT_SET(options, RELATIVE_NAME);
2804        else if (!prefixcmp(arg, "--relative=")) {
2805                DIFF_OPT_SET(options, RELATIVE_NAME);
2806                options->prefix = arg + 11;
2807        }
2808
2809        /* xdiff options */
2810        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2811                DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2812        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2813                DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2814        else if (!strcmp(arg, "--ignore-space-at-eol"))
2815                DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2816        else if (!strcmp(arg, "--patience"))
2817                DIFF_XDL_SET(options, PATIENCE_DIFF);
2818
2819        /* flags options */
2820        else if (!strcmp(arg, "--binary")) {
2821                options->output_format |= DIFF_FORMAT_PATCH;
2822                DIFF_OPT_SET(options, BINARY);
2823        }
2824        else if (!strcmp(arg, "--full-index"))
2825                DIFF_OPT_SET(options, FULL_INDEX);
2826        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2827                DIFF_OPT_SET(options, TEXT);
2828        else if (!strcmp(arg, "-R"))
2829                DIFF_OPT_SET(options, REVERSE_DIFF);
2830        else if (!strcmp(arg, "--find-copies-harder"))
2831                DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2832        else if (!strcmp(arg, "--follow"))
2833                DIFF_OPT_SET(options, FOLLOW_RENAMES);
2834        else if (!strcmp(arg, "--color"))
2835                DIFF_OPT_SET(options, COLOR_DIFF);
2836        else if (!prefixcmp(arg, "--color=")) {
2837                int value = git_config_colorbool(NULL, arg+8, -1);
2838                if (value == 0)
2839                        DIFF_OPT_CLR(options, COLOR_DIFF);
2840                else if (value > 0)
2841                        DIFF_OPT_SET(options, COLOR_DIFF);
2842                else
2843                        return error("option `color' expects \"always\", \"auto\", or \"never\"");
2844        }
2845        else if (!strcmp(arg, "--no-color"))
2846                DIFF_OPT_CLR(options, COLOR_DIFF);
2847        else if (!strcmp(arg, "--color-words")) {
2848                DIFF_OPT_SET(options, COLOR_DIFF);
2849                DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2850        }
2851        else if (!prefixcmp(arg, "--color-words=")) {
2852                DIFF_OPT_SET(options, COLOR_DIFF);
2853                DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2854                options->word_regex = arg + 14;
2855        }
2856        else if (!strcmp(arg, "--exit-code"))
2857                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2858        else if (!strcmp(arg, "--quiet"))
2859                DIFF_OPT_SET(options, QUICK);
2860        else if (!strcmp(arg, "--ext-diff"))
2861                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2862        else if (!strcmp(arg, "--no-ext-diff"))
2863                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2864        else if (!strcmp(arg, "--textconv"))
2865                DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2866        else if (!strcmp(arg, "--no-textconv"))
2867                DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2868        else if (!strcmp(arg, "--ignore-submodules"))
2869                handle_ignore_submodules_arg(options, "all");
2870        else if (!prefixcmp(arg, "--ignore-submodules="))
2871                handle_ignore_submodules_arg(options, arg + 20);
2872        else if (!strcmp(arg, "--submodule"))
2873                DIFF_OPT_SET(options, SUBMODULE_LOG);
2874        else if (!prefixcmp(arg, "--submodule=")) {
2875                if (!strcmp(arg + 12, "log"))
2876                        DIFF_OPT_SET(options, SUBMODULE_LOG);
2877        }
2878
2879        /* misc options */
2880        else if (!strcmp(arg, "-z"))
2881                options->line_termination = 0;
2882        else if (!prefixcmp(arg, "-l"))
2883                options->rename_limit = strtoul(arg+2, NULL, 10);
2884        else if (!prefixcmp(arg, "-S"))
2885                options->pickaxe = arg + 2;
2886        else if (!strcmp(arg, "--pickaxe-all"))
2887                options->pickaxe_opts = DIFF_PICKAXE_ALL;
2888        else if (!strcmp(arg, "--pickaxe-regex"))
2889                options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2890        else if (!prefixcmp(arg, "-O"))
2891                options->orderfile = arg + 2;
2892        else if (!prefixcmp(arg, "--diff-filter="))
2893                options->filter = arg + 14;
2894        else if (!strcmp(arg, "--abbrev"))
2895                options->abbrev = DEFAULT_ABBREV;
2896        else if (!prefixcmp(arg, "--abbrev=")) {
2897                options->abbrev = strtoul(arg + 9, NULL, 10);
2898                if (options->abbrev < MINIMUM_ABBREV)
2899                        options->abbrev = MINIMUM_ABBREV;
2900                else if (40 < options->abbrev)
2901                        options->abbrev = 40;
2902        }
2903        else if (!prefixcmp(arg, "--src-prefix="))
2904                options->a_prefix = arg + 13;
2905        else if (!prefixcmp(arg, "--dst-prefix="))
2906                options->b_prefix = arg + 13;
2907        else if (!strcmp(arg, "--no-prefix"))
2908                options->a_prefix = options->b_prefix = "";
2909        else if (opt_arg(arg, '\0', "inter-hunk-context",
2910                         &options->interhunkcontext))
2911                ;
2912        else if (!prefixcmp(arg, "--output=")) {
2913                options->file = fopen(arg + strlen("--output="), "w");
2914                if (!options->file)
2915                        die_errno("Could not open '%s'", arg + strlen("--output="));
2916                options->close_file = 1;
2917        } else
2918                return 0;
2919        return 1;
2920}
2921
2922static int parse_num(const char **cp_p)
2923{
2924        unsigned long num, scale;
2925        int ch, dot;
2926        const char *cp = *cp_p;
2927
2928        num = 0;
2929        scale = 1;
2930        dot = 0;
2931        for (;;) {
2932                ch = *cp;
2933                if ( !dot && ch == '.' ) {
2934                        scale = 1;
2935                        dot = 1;
2936                } else if ( ch == '%' ) {
2937                        scale = dot ? scale*100 : 100;
2938                        cp++;   /* % is always at the end */
2939                        break;
2940                } else if ( ch >= '0' && ch <= '9' ) {
2941                        if ( scale < 100000 ) {
2942                                scale *= 10;
2943                                num = (num*10) + (ch-'0');
2944                        }
2945                } else {
2946                        break;
2947                }
2948                cp++;
2949        }
2950        *cp_p = cp;
2951
2952        /* user says num divided by scale and we say internally that
2953         * is MAX_SCORE * num / scale.
2954         */
2955        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2956}
2957
2958static int diff_scoreopt_parse(const char *opt)
2959{
2960        int opt1, opt2, cmd;
2961
2962        if (*opt++ != '-')
2963                return -1;
2964        cmd = *opt++;
2965        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2966                return -1; /* that is not a -M, -C nor -B option */
2967
2968        opt1 = parse_num(&opt);
2969        if (cmd != 'B')
2970                opt2 = 0;
2971        else {
2972                if (*opt == 0)
2973                        opt2 = 0;
2974                else if (*opt != '/')
2975                        return -1; /* we expect -B80/99 or -B80 */
2976                else {
2977                        opt++;
2978                        opt2 = parse_num(&opt);
2979                }
2980        }
2981        if (*opt != 0)
2982                return -1;
2983        return opt1 | (opt2 << 16);
2984}
2985
2986struct diff_queue_struct diff_queued_diff;
2987
2988void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2989{
2990        if (queue->alloc <= queue->nr) {
2991                queue->alloc = alloc_nr(queue->alloc);
2992                queue->queue = xrealloc(queue->queue,
2993                                        sizeof(dp) * queue->alloc);
2994        }
2995        queue->queue[queue->nr++] = dp;
2996}
2997
2998struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2999                                 struct diff_filespec *one,
3000                                 struct diff_filespec *two)
3001{
3002        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3003        dp->one = one;
3004        dp->two = two;
3005        if (queue)
3006                diff_q(queue, dp);
3007        return dp;
3008}
3009
3010void diff_free_filepair(struct diff_filepair *p)
3011{
3012        free_filespec(p->one);
3013        free_filespec(p->two);
3014        free(p);
3015}
3016
3017/* This is different from find_unique_abbrev() in that
3018 * it stuffs the result with dots for alignment.
3019 */
3020const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3021{
3022        int abblen;
3023        const char *abbrev;
3024        if (len == 40)
3025                return sha1_to_hex(sha1);
3026
3027        abbrev = find_unique_abbrev(sha1, len);
3028        abblen = strlen(abbrev);
3029        if (abblen < 37) {
3030                static char hex[41];
3031                if (len < abblen && abblen <= len + 2)
3032                        sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3033                else
3034                        sprintf(hex, "%s...", abbrev);
3035                return hex;
3036        }
3037        return sha1_to_hex(sha1);
3038}
3039
3040static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3041{
3042        int line_termination = opt->line_termination;
3043        int inter_name_termination = line_termination ? '\t' : '\0';
3044
3045        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3046                fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3047                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
3048                fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3049        }
3050        if (p->score) {
3051                fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3052                        inter_name_termination);
3053        } else {
3054                fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3055        }
3056
3057        if (p->status == DIFF_STATUS_COPIED ||
3058            p->status == DIFF_STATUS_RENAMED) {
3059                const char *name_a, *name_b;
3060                name_a = p->one->path;
3061                name_b = p->two->path;
3062                strip_prefix(opt->prefix_length, &name_a, &name_b);
3063                write_name_quoted(name_a, opt->file, inter_name_termination);
3064                write_name_quoted(name_b, opt->file, line_termination);
3065        } else {
3066                const char *name_a, *name_b;
3067                name_a = p->one->mode ? p->one->path : p->two->path;
3068                name_b = NULL;
3069                strip_prefix(opt->prefix_length, &name_a, &name_b);
3070                write_name_quoted(name_a, opt->file, line_termination);
3071        }
3072}
3073
3074int diff_unmodified_pair(struct diff_filepair *p)
3075{
3076        /* This function is written stricter than necessary to support
3077         * the currently implemented transformers, but the idea is to
3078         * let transformers to produce diff_filepairs any way they want,
3079         * and filter and clean them up here before producing the output.
3080         */
3081        struct diff_filespec *one = p->one, *two = p->two;
3082
3083        if (DIFF_PAIR_UNMERGED(p))
3084                return 0; /* unmerged is interesting */
3085
3086        /* deletion, addition, mode or type change
3087         * and rename are all interesting.
3088         */
3089        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3090            DIFF_PAIR_MODE_CHANGED(p) ||
3091            strcmp(one->path, two->path))
3092                return 0;
3093
3094        /* both are valid and point at the same path.  that is, we are
3095         * dealing with a change.
3096         */
3097        if (one->sha1_valid && two->sha1_valid &&
3098            !hashcmp(one->sha1, two->sha1) &&
3099            !one->dirty_submodule && !two->dirty_submodule)
3100                return 1; /* no change */
3101        if (!one->sha1_valid && !two->sha1_valid)
3102                return 1; /* both look at the same file on the filesystem. */
3103        return 0;
3104}
3105
3106static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3107{
3108        if (diff_unmodified_pair(p))
3109                return;
3110
3111        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3112            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3113                return; /* no tree diffs in patch format */
3114
3115        run_diff(p, o);
3116}
3117
3118static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3119                            struct diffstat_t *diffstat)
3120{
3121        if (diff_unmodified_pair(p))
3122                return;
3123
3124        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3125            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3126                return; /* no tree diffs in patch format */
3127
3128        run_diffstat(p, o, diffstat);
3129}
3130
3131static void diff_flush_checkdiff(struct diff_filepair *p,
3132                struct diff_options *o)
3133{
3134        if (diff_unmodified_pair(p))
3135                return;
3136
3137        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3138            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3139                return; /* no tree diffs in patch format */
3140
3141        run_checkdiff(p, o);
3142}
3143
3144int diff_queue_is_empty(void)
3145{
3146        struct diff_queue_struct *q = &diff_queued_diff;
3147        int i;
3148        for (i = 0; i < q->nr; i++)
3149                if (!diff_unmodified_pair(q->queue[i]))
3150                        return 0;
3151        return 1;
3152}
3153
3154#if DIFF_DEBUG
3155void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3156{
3157        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3158                x, one ? one : "",
3159                s->path,
3160                DIFF_FILE_VALID(s) ? "valid" : "invalid",
3161                s->mode,
3162                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3163        fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3164                x, one ? one : "",
3165                s->size, s->xfrm_flags);
3166}
3167
3168void diff_debug_filepair(const struct diff_filepair *p, int i)
3169{
3170        diff_debug_filespec(p->one, i, "one");
3171        diff_debug_filespec(p->two, i, "two");
3172        fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3173                p->score, p->status ? p->status : '?',
3174                p->one->rename_used, p->broken_pair);
3175}
3176
3177void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3178{
3179        int i;
3180        if (msg)
3181                fprintf(stderr, "%s\n", msg);
3182        fprintf(stderr, "q->nr = %d\n", q->nr);
3183        for (i = 0; i < q->nr; i++) {
3184                struct diff_filepair *p = q->queue[i];
3185                diff_debug_filepair(p, i);
3186        }
3187}
3188#endif
3189
3190static void diff_resolve_rename_copy(void)
3191{
3192        int i;
3193        struct diff_filepair *p;
3194        struct diff_queue_struct *q = &diff_queued_diff;
3195
3196        diff_debug_queue("resolve-rename-copy", q);
3197
3198        for (i = 0; i < q->nr; i++) {
3199                p = q->queue[i];
3200                p->status = 0; /* undecided */
3201                if (DIFF_PAIR_UNMERGED(p))
3202                        p->status = DIFF_STATUS_UNMERGED;
3203                else if (!DIFF_FILE_VALID(p->one))
3204                        p->status = DIFF_STATUS_ADDED;
3205                else if (!DIFF_FILE_VALID(p->two))
3206                        p->status = DIFF_STATUS_DELETED;
3207                else if (DIFF_PAIR_TYPE_CHANGED(p))
3208                        p->status = DIFF_STATUS_TYPE_CHANGED;
3209
3210                /* from this point on, we are dealing with a pair
3211                 * whose both sides are valid and of the same type, i.e.
3212                 * either in-place edit or rename/copy edit.
3213                 */
3214                else if (DIFF_PAIR_RENAME(p)) {
3215                        /*
3216                         * A rename might have re-connected a broken
3217                         * pair up, causing the pathnames to be the
3218                         * same again. If so, that's not a rename at
3219                         * all, just a modification..
3220                         *
3221                         * Otherwise, see if this source was used for
3222                         * multiple renames, in which case we decrement
3223                         * the count, and call it a copy.
3224                         */
3225                        if (!strcmp(p->one->path, p->two->path))
3226                                p->status = DIFF_STATUS_MODIFIED;
3227                        else if (--p->one->rename_used > 0)
3228                                p->status = DIFF_STATUS_COPIED;
3229                        else
3230                                p->status = DIFF_STATUS_RENAMED;
3231                }
3232                else if (hashcmp(p->one->sha1, p->two->sha1) ||
3233                         p->one->mode != p->two->mode ||
3234                         p->one->dirty_submodule ||
3235                         p->two->dirty_submodule ||
3236                         is_null_sha1(p->one->sha1))
3237                        p->status = DIFF_STATUS_MODIFIED;
3238                else {
3239                        /* This is a "no-change" entry and should not
3240                         * happen anymore, but prepare for broken callers.
3241                         */
3242                        error("feeding unmodified %s to diffcore",
3243                              p->one->path);
3244                        p->status = DIFF_STATUS_UNKNOWN;
3245                }
3246        }
3247        diff_debug_queue("resolve-rename-copy done", q);
3248}
3249
3250static int check_pair_status(struct diff_filepair *p)
3251{
3252        switch (p->status) {
3253        case DIFF_STATUS_UNKNOWN:
3254                return 0;
3255        case 0:
3256                die("internal error in diff-resolve-rename-copy");
3257        default:
3258                return 1;
3259        }
3260}
3261
3262static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3263{
3264        int fmt = opt->output_format;
3265
3266        if (fmt & DIFF_FORMAT_CHECKDIFF)
3267                diff_flush_checkdiff(p, opt);
3268        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3269                diff_flush_raw(p, opt);
3270        else if (fmt & DIFF_FORMAT_NAME) {
3271                const char *name_a, *name_b;
3272                name_a = p->two->path;
3273                name_b = NULL;
3274                strip_prefix(opt->prefix_length, &name_a, &name_b);
3275                write_name_quoted(name_a, opt->file, opt->line_termination);
3276        }
3277}
3278
3279static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3280{
3281        if (fs->mode)
3282                fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3283        else
3284                fprintf(file, " %s ", newdelete);
3285        write_name_quoted(fs->path, file, '\n');
3286}
3287
3288
3289static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3290{
3291        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3292                fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3293                        show_name ? ' ' : '\n');
3294                if (show_name) {
3295                        write_name_quoted(p->two->path, file, '\n');
3296                }
3297        }
3298}
3299
3300static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3301{
3302        char *names = pprint_rename(p->one->path, p->two->path);
3303
3304        fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3305        free(names);
3306        show_mode_change(file, p, 0);
3307}
3308
3309static void diff_summary(FILE *file, struct diff_filepair *p)
3310{
3311        switch(p->status) {
3312        case DIFF_STATUS_DELETED:
3313                show_file_mode_name(file, "delete", p->one);
3314                break;
3315        case DIFF_STATUS_ADDED:
3316                show_file_mode_name(file, "create", p->two);
3317                break;
3318        case DIFF_STATUS_COPIED:
3319                show_rename_copy(file, "copy", p);
3320                break;
3321        case DIFF_STATUS_RENAMED:
3322                show_rename_copy(file, "rename", p);
3323                break;
3324        default:
3325                if (p->score) {
3326                        fputs(" rewrite ", file);
3327                        write_name_quoted(p->two->path, file, ' ');
3328                        fprintf(file, "(%d%%)\n", similarity_index(p));
3329                }
3330                show_mode_change(file, p, !p->score);
3331                break;
3332        }
3333}
3334
3335struct patch_id_t {
3336        git_SHA_CTX *ctx;
3337        int patchlen;
3338};
3339
3340static int remove_space(char *line, int len)
3341{
3342        int i;
3343        char *dst = line;
3344        unsigned char c;
3345
3346        for (i = 0; i < len; i++)
3347                if (!isspace((c = line[i])))
3348                        *dst++ = c;
3349
3350        return dst - line;
3351}
3352
3353static void patch_id_consume(void *priv, char *line, unsigned long len)
3354{
3355        struct patch_id_t *data = priv;
3356        int new_len;
3357
3358        /* Ignore line numbers when computing the SHA1 of the patch */
3359        if (!prefixcmp(line, "@@ -"))
3360                return;
3361
3362        new_len = remove_space(line, len);
3363
3364        git_SHA1_Update(data->ctx, line, new_len);
3365        data->patchlen += new_len;
3366}
3367
3368/* returns 0 upon success, and writes result into sha1 */
3369static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3370{
3371        struct diff_queue_struct *q = &diff_queued_diff;
3372        int i;
3373        git_SHA_CTX ctx;
3374        struct patch_id_t data;
3375        char buffer[PATH_MAX * 4 + 20];
3376
3377        git_SHA1_Init(&ctx);
3378        memset(&data, 0, sizeof(struct patch_id_t));
3379        data.ctx = &ctx;
3380
3381        for (i = 0; i < q->nr; i++) {
3382                xpparam_t xpp;
3383                xdemitconf_t xecfg;
3384                mmfile_t mf1, mf2;
3385                struct diff_filepair *p = q->queue[i];
3386                int len1, len2;
3387
3388                memset(&xpp, 0, sizeof(xpp));
3389                memset(&xecfg, 0, sizeof(xecfg));
3390                if (p->status == 0)
3391                        return error("internal diff status error");
3392                if (p->status == DIFF_STATUS_UNKNOWN)
3393                        continue;
3394                if (diff_unmodified_pair(p))
3395                        continue;
3396                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3397                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3398                        continue;
3399                if (DIFF_PAIR_UNMERGED(p))
3400                        continue;
3401
3402                diff_fill_sha1_info(p->one);
3403                diff_fill_sha1_info(p->two);
3404                if (fill_mmfile(&mf1, p->one) < 0 ||
3405                                fill_mmfile(&mf2, p->two) < 0)
3406                        return error("unable to read files to diff");
3407
3408                len1 = remove_space(p->one->path, strlen(p->one->path));
3409                len2 = remove_space(p->two->path, strlen(p->two->path));
3410                if (p->one->mode == 0)
3411                        len1 = snprintf(buffer, sizeof(buffer),
3412                                        "diff--gita/%.*sb/%.*s"
3413                                        "newfilemode%06o"
3414                                        "---/dev/null"
3415                                        "+++b/%.*s",
3416                                        len1, p->one->path,
3417                                        len2, p->two->path,
3418                                        p->two->mode,
3419                                        len2, p->two->path);
3420                else if (p->two->mode == 0)
3421                        len1 = snprintf(buffer, sizeof(buffer),
3422                                        "diff--gita/%.*sb/%.*s"
3423                                        "deletedfilemode%06o"
3424                                        "---a/%.*s"
3425                                        "+++/dev/null",
3426                                        len1, p->one->path,
3427                                        len2, p->two->path,
3428                                        p->one->mode,
3429                                        len1, p->one->path);
3430                else
3431                        len1 = snprintf(buffer, sizeof(buffer),
3432                                        "diff--gita/%.*sb/%.*s"
3433                                        "---a/%.*s"
3434                                        "+++b/%.*s",
3435                                        len1, p->one->path,
3436                                        len2, p->two->path,
3437                                        len1, p->one->path,
3438                                        len2, p->two->path);
3439                git_SHA1_Update(&ctx, buffer, len1);
3440
3441                xpp.flags = XDF_NEED_MINIMAL;
3442                xecfg.ctxlen = 3;
3443                xecfg.flags = XDL_EMIT_FUNCNAMES;
3444                xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3445                              &xpp, &xecfg);
3446        }
3447
3448        git_SHA1_Final(sha1, &ctx);
3449        return 0;
3450}
3451
3452int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3453{
3454        struct diff_queue_struct *q = &diff_queued_diff;
3455        int i;
3456        int result = diff_get_patch_id(options, sha1);
3457
3458        for (i = 0; i < q->nr; i++)
3459                diff_free_filepair(q->queue[i]);
3460
3461        free(q->queue);
3462        q->queue = NULL;
3463        q->nr = q->alloc = 0;
3464
3465        return result;
3466}
3467
3468static int is_summary_empty(const struct diff_queue_struct *q)
3469{
3470        int i;
3471
3472        for (i = 0; i < q->nr; i++) {
3473                const struct diff_filepair *p = q->queue[i];
3474
3475                switch (p->status) {
3476                case DIFF_STATUS_DELETED:
3477                case DIFF_STATUS_ADDED:
3478                case DIFF_STATUS_COPIED:
3479                case DIFF_STATUS_RENAMED:
3480                        return 0;
3481                default:
3482                        if (p->score)
3483                                return 0;
3484                        if (p->one->mode && p->two->mode &&
3485                            p->one->mode != p->two->mode)
3486                                return 0;
3487                        break;
3488                }
3489        }
3490        return 1;
3491}
3492
3493void diff_flush(struct diff_options *options)
3494{
3495        struct diff_queue_struct *q = &diff_queued_diff;
3496        int i, output_format = options->output_format;
3497        int separator = 0;
3498
3499        /*
3500         * Order: raw, stat, summary, patch
3501         * or:    name/name-status/checkdiff (other bits clear)
3502         */
3503        if (!q->nr)
3504                goto free_queue;
3505
3506        if (output_format & (DIFF_FORMAT_RAW |
3507                             DIFF_FORMAT_NAME |
3508                             DIFF_FORMAT_NAME_STATUS |
3509                             DIFF_FORMAT_CHECKDIFF)) {
3510                for (i = 0; i < q->nr; i++) {
3511                        struct diff_filepair *p = q->queue[i];
3512                        if (check_pair_status(p))
3513                                flush_one_pair(p, options);
3514                }
3515                separator++;
3516        }
3517
3518        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3519                struct diffstat_t diffstat;
3520
3521                memset(&diffstat, 0, sizeof(struct diffstat_t));
3522                for (i = 0; i < q->nr; i++) {
3523                        struct diff_filepair *p = q->queue[i];
3524                        if (check_pair_status(p))
3525                                diff_flush_stat(p, options, &diffstat);
3526                }
3527                if (output_format & DIFF_FORMAT_NUMSTAT)
3528                        show_numstat(&diffstat, options);
3529                if (output_format & DIFF_FORMAT_DIFFSTAT)
3530                        show_stats(&diffstat, options);
3531                if (output_format & DIFF_FORMAT_SHORTSTAT)
3532                        show_shortstats(&diffstat, options);
3533                free_diffstat_info(&diffstat);
3534                separator++;
3535        }
3536        if (output_format & DIFF_FORMAT_DIRSTAT)
3537                show_dirstat(options);
3538
3539        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3540                for (i = 0; i < q->nr; i++)
3541                        diff_summary(options->file, q->queue[i]);
3542                separator++;
3543        }
3544
3545        if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3546            DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3547            DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3548                /*
3549                 * run diff_flush_patch for the exit status. setting
3550                 * options->file to /dev/null should be safe, becaue we
3551                 * aren't supposed to produce any output anyway.
3552                 */
3553                if (options->close_file)
3554                        fclose(options->file);
3555                options->file = fopen("/dev/null", "w");
3556                if (!options->file)
3557                        die_errno("Could not open /dev/null");
3558                options->close_file = 1;
3559                for (i = 0; i < q->nr; i++) {
3560                        struct diff_filepair *p = q->queue[i];
3561                        if (check_pair_status(p))
3562                                diff_flush_patch(p, options);
3563                        if (options->found_changes)
3564                                break;
3565                }
3566        }
3567
3568        if (output_format & DIFF_FORMAT_PATCH) {
3569                if (separator) {
3570                        putc(options->line_termination, options->file);
3571                        if (options->stat_sep) {
3572                                /* attach patch instead of inline */
3573                                fputs(options->stat_sep, options->file);
3574                        }
3575                }
3576
3577                for (i = 0; i < q->nr; i++) {
3578                        struct diff_filepair *p = q->queue[i];
3579                        if (check_pair_status(p))
3580                                diff_flush_patch(p, options);
3581                }
3582        }
3583
3584        if (output_format & DIFF_FORMAT_CALLBACK)
3585                options->format_callback(q, options, options->format_callback_data);
3586
3587        for (i = 0; i < q->nr; i++)
3588                diff_free_filepair(q->queue[i]);
3589free_queue:
3590        free(q->queue);
3591        q->queue = NULL;
3592        q->nr = q->alloc = 0;
3593        if (options->close_file)
3594                fclose(options->file);
3595
3596        /*
3597         * Report the content-level differences with HAS_CHANGES;
3598         * diff_addremove/diff_change does not set the bit when
3599         * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3600         */
3601        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3602                if (options->found_changes)
3603                        DIFF_OPT_SET(options, HAS_CHANGES);
3604                else
3605                        DIFF_OPT_CLR(options, HAS_CHANGES);
3606        }
3607}
3608
3609static void diffcore_apply_filter(const char *filter)
3610{
3611        int i;
3612        struct diff_queue_struct *q = &diff_queued_diff;
3613        struct diff_queue_struct outq;
3614        outq.queue = NULL;
3615        outq.nr = outq.alloc = 0;
3616
3617        if (!filter)
3618                return;
3619
3620        if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3621                int found;
3622                for (i = found = 0; !found && i < q->nr; i++) {
3623                        struct diff_filepair *p = q->queue[i];
3624                        if (((p->status == DIFF_STATUS_MODIFIED) &&
3625                             ((p->score &&
3626                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3627                              (!p->score &&
3628                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3629                            ((p->status != DIFF_STATUS_MODIFIED) &&
3630                             strchr(filter, p->status)))
3631                                found++;
3632                }
3633                if (found)
3634                        return;
3635
3636                /* otherwise we will clear the whole queue
3637                 * by copying the empty outq at the end of this
3638                 * function, but first clear the current entries
3639                 * in the queue.
3640                 */
3641                for (i = 0; i < q->nr; i++)
3642                        diff_free_filepair(q->queue[i]);
3643        }
3644        else {
3645                /* Only the matching ones */
3646                for (i = 0; i < q->nr; i++) {
3647                        struct diff_filepair *p = q->queue[i];
3648
3649                        if (((p->status == DIFF_STATUS_MODIFIED) &&
3650                             ((p->score &&
3651                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3652                              (!p->score &&
3653                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3654                            ((p->status != DIFF_STATUS_MODIFIED) &&
3655                             strchr(filter, p->status)))
3656                                diff_q(&outq, p);
3657                        else
3658                                diff_free_filepair(p);
3659                }
3660        }
3661        free(q->queue);
3662        *q = outq;
3663}
3664
3665/* Check whether two filespecs with the same mode and size are identical */
3666static int diff_filespec_is_identical(struct diff_filespec *one,
3667                                      struct diff_filespec *two)
3668{
3669        if (S_ISGITLINK(one->mode))
3670                return 0;
3671        if (diff_populate_filespec(one, 0))
3672                return 0;
3673        if (diff_populate_filespec(two, 0))
3674                return 0;
3675        return !memcmp(one->data, two->data, one->size);
3676}
3677
3678static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3679{
3680        int i;
3681        struct diff_queue_struct *q = &diff_queued_diff;
3682        struct diff_queue_struct outq;
3683        outq.queue = NULL;
3684        outq.nr = outq.alloc = 0;
3685
3686        for (i = 0; i < q->nr; i++) {
3687                struct diff_filepair *p = q->queue[i];
3688
3689                /*
3690                 * 1. Entries that come from stat info dirtiness
3691                 *    always have both sides (iow, not create/delete),
3692                 *    one side of the object name is unknown, with
3693                 *    the same mode and size.  Keep the ones that
3694                 *    do not match these criteria.  They have real
3695                 *    differences.
3696                 *
3697                 * 2. At this point, the file is known to be modified,
3698                 *    with the same mode and size, and the object
3699                 *    name of one side is unknown.  Need to inspect
3700                 *    the identical contents.
3701                 */
3702                if (!DIFF_FILE_VALID(p->one) || /* (1) */
3703                    !DIFF_FILE_VALID(p->two) ||
3704                    (p->one->sha1_valid && p->two->sha1_valid) ||
3705                    (p->one->mode != p->two->mode) ||
3706                    diff_populate_filespec(p->one, 1) ||
3707                    diff_populate_filespec(p->two, 1) ||
3708                    (p->one->size != p->two->size) ||
3709                    !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3710                        diff_q(&outq, p);
3711                else {
3712                        /*
3713                         * The caller can subtract 1 from skip_stat_unmatch
3714                         * to determine how many paths were dirty only
3715                         * due to stat info mismatch.
3716                         */
3717                        if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3718                                diffopt->skip_stat_unmatch++;
3719                        diff_free_filepair(p);
3720                }
3721        }
3722        free(q->queue);
3723        *q = outq;
3724}
3725
3726static int diffnamecmp(const void *a_, const void *b_)
3727{
3728        const struct diff_filepair *a = *((const struct diff_filepair **)a_);
3729        const struct diff_filepair *b = *((const struct diff_filepair **)b_);
3730        const char *name_a, *name_b;
3731
3732        name_a = a->one ? a->one->path : a->two->path;
3733        name_b = b->one ? b->one->path : b->two->path;
3734        return strcmp(name_a, name_b);
3735}
3736
3737void diffcore_fix_diff_index(struct diff_options *options)
3738{
3739        struct diff_queue_struct *q = &diff_queued_diff;
3740        qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
3741}
3742
3743void diffcore_std(struct diff_options *options)
3744{
3745        if (options->skip_stat_unmatch)
3746                diffcore_skip_stat_unmatch(options);
3747        if (options->break_opt != -1)
3748                diffcore_break(options->break_opt);
3749        if (options->detect_rename)
3750                diffcore_rename(options);
3751        if (options->break_opt != -1)
3752                diffcore_merge_broken();
3753        if (options->pickaxe)
3754                diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3755        if (options->orderfile)
3756                diffcore_order(options->orderfile);
3757        diff_resolve_rename_copy();
3758        diffcore_apply_filter(options->filter);
3759
3760        if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3761                DIFF_OPT_SET(options, HAS_CHANGES);
3762        else
3763                DIFF_OPT_CLR(options, HAS_CHANGES);
3764}
3765
3766int diff_result_code(struct diff_options *opt, int status)
3767{
3768        int result = 0;
3769        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3770            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3771                return status;
3772        if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3773            DIFF_OPT_TST(opt, HAS_CHANGES))
3774                result |= 01;
3775        if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3776            DIFF_OPT_TST(opt, CHECK_FAILED))
3777                result |= 02;
3778        return result;
3779}
3780
3781void diff_addremove(struct diff_options *options,
3782                    int addremove, unsigned mode,
3783                    const unsigned char *sha1,
3784                    const char *concatpath, unsigned dirty_submodule)
3785{
3786        struct diff_filespec *one, *two;
3787
3788        if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3789                return;
3790
3791        /* This may look odd, but it is a preparation for
3792         * feeding "there are unchanged files which should
3793         * not produce diffs, but when you are doing copy
3794         * detection you would need them, so here they are"
3795         * entries to the diff-core.  They will be prefixed
3796         * with something like '=' or '*' (I haven't decided
3797         * which but should not make any difference).
3798         * Feeding the same new and old to diff_change()
3799         * also has the same effect.
3800         * Before the final output happens, they are pruned after
3801         * merged into rename/copy pairs as appropriate.
3802         */
3803        if (DIFF_OPT_TST(options, REVERSE_DIFF))
3804                addremove = (addremove == '+' ? '-' :
3805                             addremove == '-' ? '+' : addremove);
3806
3807        if (options->prefix &&
3808            strncmp(concatpath, options->prefix, options->prefix_length))
3809                return;
3810
3811        one = alloc_filespec(concatpath);
3812        two = alloc_filespec(concatpath);
3813
3814        if (addremove != '+')
3815                fill_filespec(one, sha1, mode);
3816        if (addremove != '-') {
3817                fill_filespec(two, sha1, mode);
3818                two->dirty_submodule = dirty_submodule;
3819        }
3820
3821        diff_queue(&diff_queued_diff, one, two);
3822        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3823                DIFF_OPT_SET(options, HAS_CHANGES);
3824}
3825
3826void diff_change(struct diff_options *options,
3827                 unsigned old_mode, unsigned new_mode,
3828                 const unsigned char *old_sha1,
3829                 const unsigned char *new_sha1,
3830                 const char *concatpath,
3831                 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
3832{
3833        struct diff_filespec *one, *two;
3834
3835        if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3836                        && S_ISGITLINK(new_mode))
3837                return;
3838
3839        if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3840                unsigned tmp;
3841                const unsigned char *tmp_c;
3842                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3843                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3844                tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
3845                        new_dirty_submodule = tmp;
3846        }
3847
3848        if (options->prefix &&
3849            strncmp(concatpath, options->prefix, options->prefix_length))
3850                return;
3851
3852        one = alloc_filespec(concatpath);
3853        two = alloc_filespec(concatpath);
3854        fill_filespec(one, old_sha1, old_mode);
3855        fill_filespec(two, new_sha1, new_mode);
3856        one->dirty_submodule = old_dirty_submodule;
3857        two->dirty_submodule = new_dirty_submodule;
3858
3859        diff_queue(&diff_queued_diff, one, two);
3860        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3861                DIFF_OPT_SET(options, HAS_CHANGES);
3862}
3863
3864void diff_unmerge(struct diff_options *options,
3865                  const char *path,
3866                  unsigned mode, const unsigned char *sha1)
3867{
3868        struct diff_filespec *one, *two;
3869
3870        if (options->prefix &&
3871            strncmp(path, options->prefix, options->prefix_length))
3872                return;
3873
3874        one = alloc_filespec(path);
3875        two = alloc_filespec(path);
3876        fill_filespec(one, sha1, mode);
3877        diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3878}
3879
3880static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3881                size_t *outsize)
3882{
3883        struct diff_tempfile *temp;
3884        const char *argv[3];
3885        const char **arg = argv;
3886        struct child_process child;
3887        struct strbuf buf = STRBUF_INIT;
3888        int err = 0;
3889
3890        temp = prepare_temp_file(spec->path, spec);
3891        *arg++ = pgm;
3892        *arg++ = temp->name;
3893        *arg = NULL;
3894
3895        memset(&child, 0, sizeof(child));
3896        child.use_shell = 1;
3897        child.argv = argv;
3898        child.out = -1;
3899        if (start_command(&child)) {
3900                remove_tempfile();
3901                return NULL;
3902        }
3903
3904        if (strbuf_read(&buf, child.out, 0) < 0)
3905                err = error("error reading from textconv command '%s'", pgm);
3906        close(child.out);
3907
3908        if (finish_command(&child) || err) {
3909                strbuf_release(&buf);
3910                remove_tempfile();
3911                return NULL;
3912        }
3913        remove_tempfile();
3914
3915        return strbuf_detach(&buf, outsize);
3916}