diff.con commit Teach revision walking machinery to walk multiple times sequencially (bcc0a3e)
   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 = 400;
  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;
  33static int diff_no_prefix;
  34static int diff_dirstat_permille_default = 30;
  35static struct diff_options default_diff_options;
  36
  37static char diff_colors[][COLOR_MAXLEN] = {
  38        GIT_COLOR_RESET,
  39        GIT_COLOR_NORMAL,       /* PLAIN */
  40        GIT_COLOR_BOLD,         /* METAINFO */
  41        GIT_COLOR_CYAN,         /* FRAGINFO */
  42        GIT_COLOR_RED,          /* OLD */
  43        GIT_COLOR_GREEN,        /* NEW */
  44        GIT_COLOR_YELLOW,       /* COMMIT */
  45        GIT_COLOR_BG_RED,       /* WHITESPACE */
  46        GIT_COLOR_NORMAL,       /* FUNCINFO */
  47};
  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 parse_dirstat_params(struct diff_options *options, const char *params,
  71                                struct strbuf *errmsg)
  72{
  73        const char *p = params;
  74        int p_len, ret = 0;
  75
  76        while (*p) {
  77                p_len = strchrnul(p, ',') - p;
  78                if (!memcmp(p, "changes", p_len)) {
  79                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  80                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  81                } else if (!memcmp(p, "lines", p_len)) {
  82                        DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  83                        DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  84                } else if (!memcmp(p, "files", p_len)) {
  85                        DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  86                        DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  87                } else if (!memcmp(p, "noncumulative", p_len)) {
  88                        DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
  89                } else if (!memcmp(p, "cumulative", p_len)) {
  90                        DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
  91                } else if (isdigit(*p)) {
  92                        char *end;
  93                        int permille = strtoul(p, &end, 10) * 10;
  94                        if (*end == '.' && isdigit(*++end)) {
  95                                /* only use first digit */
  96                                permille += *end - '0';
  97                                /* .. and ignore any further digits */
  98                                while (isdigit(*++end))
  99                                        ; /* nothing */
 100                        }
 101                        if (end - p == p_len)
 102                                options->dirstat_permille = permille;
 103                        else {
 104                                strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%.*s'\n"),
 105                                            p_len, p);
 106                                ret++;
 107                        }
 108                } else {
 109                        strbuf_addf(errmsg, _("  Unknown dirstat parameter '%.*s'\n"),
 110                                    p_len, p);
 111                        ret++;
 112                }
 113
 114                p += p_len;
 115
 116                if (*p)
 117                        p++; /* more parameters, swallow separator */
 118        }
 119        return ret;
 120}
 121
 122static int git_config_rename(const char *var, const char *value)
 123{
 124        if (!value)
 125                return DIFF_DETECT_RENAME;
 126        if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 127                return  DIFF_DETECT_COPY;
 128        return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 129}
 130
 131/*
 132 * These are to give UI layer defaults.
 133 * The core-level commands such as git-diff-files should
 134 * never be affected by the setting of diff.renames
 135 * the user happens to have in the configuration file.
 136 */
 137int git_diff_ui_config(const char *var, const char *value, void *cb)
 138{
 139        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 140                diff_use_color_default = git_config_colorbool(var, value);
 141                return 0;
 142        }
 143        if (!strcmp(var, "diff.renames")) {
 144                diff_detect_rename_default = git_config_rename(var, value);
 145                return 0;
 146        }
 147        if (!strcmp(var, "diff.autorefreshindex")) {
 148                diff_auto_refresh_index = git_config_bool(var, value);
 149                return 0;
 150        }
 151        if (!strcmp(var, "diff.mnemonicprefix")) {
 152                diff_mnemonic_prefix = git_config_bool(var, value);
 153                return 0;
 154        }
 155        if (!strcmp(var, "diff.noprefix")) {
 156                diff_no_prefix = git_config_bool(var, value);
 157                return 0;
 158        }
 159        if (!strcmp(var, "diff.external"))
 160                return git_config_string(&external_diff_cmd_cfg, var, value);
 161        if (!strcmp(var, "diff.wordregex"))
 162                return git_config_string(&diff_word_regex_cfg, var, value);
 163
 164        if (!strcmp(var, "diff.ignoresubmodules"))
 165                handle_ignore_submodules_arg(&default_diff_options, value);
 166
 167        if (git_color_config(var, value, cb) < 0)
 168                return -1;
 169
 170        return git_diff_basic_config(var, value, cb);
 171}
 172
 173int git_diff_basic_config(const char *var, const char *value, void *cb)
 174{
 175        if (!strcmp(var, "diff.renamelimit")) {
 176                diff_rename_limit_default = git_config_int(var, value);
 177                return 0;
 178        }
 179
 180        switch (userdiff_config(var, value)) {
 181                case 0: break;
 182                case -1: return -1;
 183                default: return 0;
 184        }
 185
 186        if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 187                int slot = parse_diff_color_slot(var, 11);
 188                if (slot < 0)
 189                        return 0;
 190                if (!value)
 191                        return config_error_nonbool(var);
 192                color_parse(value, var, diff_colors[slot]);
 193                return 0;
 194        }
 195
 196        /* like GNU diff's --suppress-blank-empty option  */
 197        if (!strcmp(var, "diff.suppressblankempty") ||
 198                        /* for backwards compatibility */
 199                        !strcmp(var, "diff.suppress-blank-empty")) {
 200                diff_suppress_blank_empty = git_config_bool(var, value);
 201                return 0;
 202        }
 203
 204        if (!strcmp(var, "diff.dirstat")) {
 205                struct strbuf errmsg = STRBUF_INIT;
 206                default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 207                if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 208                        warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 209                                errmsg.buf);
 210                strbuf_release(&errmsg);
 211                diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 212                return 0;
 213        }
 214
 215        if (!prefixcmp(var, "submodule."))
 216                return parse_submodule_config_option(var, value);
 217
 218        return git_default_config(var, value, cb);
 219}
 220
 221static char *quote_two(const char *one, const char *two)
 222{
 223        int need_one = quote_c_style(one, NULL, NULL, 1);
 224        int need_two = quote_c_style(two, NULL, NULL, 1);
 225        struct strbuf res = STRBUF_INIT;
 226
 227        if (need_one + need_two) {
 228                strbuf_addch(&res, '"');
 229                quote_c_style(one, &res, NULL, 1);
 230                quote_c_style(two, &res, NULL, 1);
 231                strbuf_addch(&res, '"');
 232        } else {
 233                strbuf_addstr(&res, one);
 234                strbuf_addstr(&res, two);
 235        }
 236        return strbuf_detach(&res, NULL);
 237}
 238
 239static const char *external_diff(void)
 240{
 241        static const char *external_diff_cmd = NULL;
 242        static int done_preparing = 0;
 243
 244        if (done_preparing)
 245                return external_diff_cmd;
 246        external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 247        if (!external_diff_cmd)
 248                external_diff_cmd = external_diff_cmd_cfg;
 249        done_preparing = 1;
 250        return external_diff_cmd;
 251}
 252
 253static struct diff_tempfile {
 254        const char *name; /* filename external diff should read from */
 255        char hex[41];
 256        char mode[10];
 257        char tmp_path[PATH_MAX];
 258} diff_temp[2];
 259
 260typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 261
 262struct emit_callback {
 263        int color_diff;
 264        unsigned ws_rule;
 265        int blank_at_eof_in_preimage;
 266        int blank_at_eof_in_postimage;
 267        int lno_in_preimage;
 268        int lno_in_postimage;
 269        sane_truncate_fn truncate;
 270        const char **label_path;
 271        struct diff_words_data *diff_words;
 272        struct diff_options *opt;
 273        int *found_changesp;
 274        struct strbuf *header;
 275};
 276
 277static int count_lines(const char *data, int size)
 278{
 279        int count, ch, completely_empty = 1, nl_just_seen = 0;
 280        count = 0;
 281        while (0 < size--) {
 282                ch = *data++;
 283                if (ch == '\n') {
 284                        count++;
 285                        nl_just_seen = 1;
 286                        completely_empty = 0;
 287                }
 288                else {
 289                        nl_just_seen = 0;
 290                        completely_empty = 0;
 291                }
 292        }
 293        if (completely_empty)
 294                return 0;
 295        if (!nl_just_seen)
 296                count++; /* no trailing newline */
 297        return count;
 298}
 299
 300static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 301{
 302        if (!DIFF_FILE_VALID(one)) {
 303                mf->ptr = (char *)""; /* does not matter */
 304                mf->size = 0;
 305                return 0;
 306        }
 307        else if (diff_populate_filespec(one, 0))
 308                return -1;
 309
 310        mf->ptr = one->data;
 311        mf->size = one->size;
 312        return 0;
 313}
 314
 315/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 316static unsigned long diff_filespec_size(struct diff_filespec *one)
 317{
 318        if (!DIFF_FILE_VALID(one))
 319                return 0;
 320        diff_populate_filespec(one, 1);
 321        return one->size;
 322}
 323
 324static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 325{
 326        char *ptr = mf->ptr;
 327        long size = mf->size;
 328        int cnt = 0;
 329
 330        if (!size)
 331                return cnt;
 332        ptr += size - 1; /* pointing at the very end */
 333        if (*ptr != '\n')
 334                ; /* incomplete line */
 335        else
 336                ptr--; /* skip the last LF */
 337        while (mf->ptr < ptr) {
 338                char *prev_eol;
 339                for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 340                        if (*prev_eol == '\n')
 341                                break;
 342                if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 343                        break;
 344                cnt++;
 345                ptr = prev_eol - 1;
 346        }
 347        return cnt;
 348}
 349
 350static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 351                               struct emit_callback *ecbdata)
 352{
 353        int l1, l2, at;
 354        unsigned ws_rule = ecbdata->ws_rule;
 355        l1 = count_trailing_blank(mf1, ws_rule);
 356        l2 = count_trailing_blank(mf2, ws_rule);
 357        if (l2 <= l1) {
 358                ecbdata->blank_at_eof_in_preimage = 0;
 359                ecbdata->blank_at_eof_in_postimage = 0;
 360                return;
 361        }
 362        at = count_lines(mf1->ptr, mf1->size);
 363        ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 364
 365        at = count_lines(mf2->ptr, mf2->size);
 366        ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 367}
 368
 369static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 370                        int first, const char *line, int len)
 371{
 372        int has_trailing_newline, has_trailing_carriage_return;
 373        int nofirst;
 374        FILE *file = o->file;
 375
 376        if (o->output_prefix) {
 377                struct strbuf *msg = NULL;
 378                msg = o->output_prefix(o, o->output_prefix_data);
 379                assert(msg);
 380                fwrite(msg->buf, msg->len, 1, file);
 381        }
 382
 383        if (len == 0) {
 384                has_trailing_newline = (first == '\n');
 385                has_trailing_carriage_return = (!has_trailing_newline &&
 386                                                (first == '\r'));
 387                nofirst = has_trailing_newline || has_trailing_carriage_return;
 388        } else {
 389                has_trailing_newline = (len > 0 && line[len-1] == '\n');
 390                if (has_trailing_newline)
 391                        len--;
 392                has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 393                if (has_trailing_carriage_return)
 394                        len--;
 395                nofirst = 0;
 396        }
 397
 398        if (len || !nofirst) {
 399                fputs(set, file);
 400                if (!nofirst)
 401                        fputc(first, file);
 402                fwrite(line, len, 1, file);
 403                fputs(reset, file);
 404        }
 405        if (has_trailing_carriage_return)
 406                fputc('\r', file);
 407        if (has_trailing_newline)
 408                fputc('\n', file);
 409}
 410
 411static void emit_line(struct diff_options *o, const char *set, const char *reset,
 412                      const char *line, int len)
 413{
 414        emit_line_0(o, set, reset, line[0], line+1, len-1);
 415}
 416
 417static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 418{
 419        if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 420              ecbdata->blank_at_eof_in_preimage &&
 421              ecbdata->blank_at_eof_in_postimage &&
 422              ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 423              ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 424                return 0;
 425        return ws_blank_line(line, len, ecbdata->ws_rule);
 426}
 427
 428static void emit_add_line(const char *reset,
 429                          struct emit_callback *ecbdata,
 430                          const char *line, int len)
 431{
 432        const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 433        const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 434
 435        if (!*ws)
 436                emit_line_0(ecbdata->opt, set, reset, '+', line, len);
 437        else if (new_blank_line_at_eof(ecbdata, line, len))
 438                /* Blank line at EOF - paint '+' as well */
 439                emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
 440        else {
 441                /* Emit just the prefix, then the rest. */
 442                emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
 443                ws_check_emit(line, len, ecbdata->ws_rule,
 444                              ecbdata->opt->file, set, reset, ws);
 445        }
 446}
 447
 448static void emit_hunk_header(struct emit_callback *ecbdata,
 449                             const char *line, int len)
 450{
 451        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 452        const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 453        const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 454        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 455        static const char atat[2] = { '@', '@' };
 456        const char *cp, *ep;
 457        struct strbuf msgbuf = STRBUF_INIT;
 458        int org_len = len;
 459        int i = 1;
 460
 461        /*
 462         * As a hunk header must begin with "@@ -<old>, +<new> @@",
 463         * it always is at least 10 bytes long.
 464         */
 465        if (len < 10 ||
 466            memcmp(line, atat, 2) ||
 467            !(ep = memmem(line + 2, len - 2, atat, 2))) {
 468                emit_line(ecbdata->opt, plain, reset, line, len);
 469                return;
 470        }
 471        ep += 2; /* skip over @@ */
 472
 473        /* The hunk header in fraginfo color */
 474        strbuf_add(&msgbuf, frag, strlen(frag));
 475        strbuf_add(&msgbuf, line, ep - line);
 476        strbuf_add(&msgbuf, reset, strlen(reset));
 477
 478        /*
 479         * trailing "\r\n"
 480         */
 481        for ( ; i < 3; i++)
 482                if (line[len - i] == '\r' || line[len - i] == '\n')
 483                        len--;
 484
 485        /* blank before the func header */
 486        for (cp = ep; ep - line < len; ep++)
 487                if (*ep != ' ' && *ep != '\t')
 488                        break;
 489        if (ep != cp) {
 490                strbuf_add(&msgbuf, plain, strlen(plain));
 491                strbuf_add(&msgbuf, cp, ep - cp);
 492                strbuf_add(&msgbuf, reset, strlen(reset));
 493        }
 494
 495        if (ep < line + len) {
 496                strbuf_add(&msgbuf, func, strlen(func));
 497                strbuf_add(&msgbuf, ep, line + len - ep);
 498                strbuf_add(&msgbuf, reset, strlen(reset));
 499        }
 500
 501        strbuf_add(&msgbuf, line + len, org_len - len);
 502        emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 503        strbuf_release(&msgbuf);
 504}
 505
 506static struct diff_tempfile *claim_diff_tempfile(void) {
 507        int i;
 508        for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 509                if (!diff_temp[i].name)
 510                        return diff_temp + i;
 511        die("BUG: diff is failing to clean up its tempfiles");
 512}
 513
 514static int remove_tempfile_installed;
 515
 516static void remove_tempfile(void)
 517{
 518        int i;
 519        for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 520                if (diff_temp[i].name == diff_temp[i].tmp_path)
 521                        unlink_or_warn(diff_temp[i].name);
 522                diff_temp[i].name = NULL;
 523        }
 524}
 525
 526static void remove_tempfile_on_signal(int signo)
 527{
 528        remove_tempfile();
 529        sigchain_pop(signo);
 530        raise(signo);
 531}
 532
 533static void print_line_count(FILE *file, int count)
 534{
 535        switch (count) {
 536        case 0:
 537                fprintf(file, "0,0");
 538                break;
 539        case 1:
 540                fprintf(file, "1");
 541                break;
 542        default:
 543                fprintf(file, "1,%d", count);
 544                break;
 545        }
 546}
 547
 548static void emit_rewrite_lines(struct emit_callback *ecb,
 549                               int prefix, const char *data, int size)
 550{
 551        const char *endp = NULL;
 552        static const char *nneof = " No newline at end of file\n";
 553        const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 554        const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 555
 556        while (0 < size) {
 557                int len;
 558
 559                endp = memchr(data, '\n', size);
 560                len = endp ? (endp - data + 1) : size;
 561                if (prefix != '+') {
 562                        ecb->lno_in_preimage++;
 563                        emit_line_0(ecb->opt, old, reset, '-',
 564                                    data, len);
 565                } else {
 566                        ecb->lno_in_postimage++;
 567                        emit_add_line(reset, ecb, data, len);
 568                }
 569                size -= len;
 570                data += len;
 571        }
 572        if (!endp) {
 573                const char *plain = diff_get_color(ecb->color_diff,
 574                                                   DIFF_PLAIN);
 575                emit_line_0(ecb->opt, plain, reset, '\\',
 576                            nneof, strlen(nneof));
 577        }
 578}
 579
 580static void emit_rewrite_diff(const char *name_a,
 581                              const char *name_b,
 582                              struct diff_filespec *one,
 583                              struct diff_filespec *two,
 584                              struct userdiff_driver *textconv_one,
 585                              struct userdiff_driver *textconv_two,
 586                              struct diff_options *o)
 587{
 588        int lc_a, lc_b;
 589        const char *name_a_tab, *name_b_tab;
 590        const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 591        const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 592        const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 593        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 594        const char *a_prefix, *b_prefix;
 595        char *data_one, *data_two;
 596        size_t size_one, size_two;
 597        struct emit_callback ecbdata;
 598        char *line_prefix = "";
 599        struct strbuf *msgbuf;
 600
 601        if (o && o->output_prefix) {
 602                msgbuf = o->output_prefix(o, o->output_prefix_data);
 603                line_prefix = msgbuf->buf;
 604        }
 605
 606        if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 607                a_prefix = o->b_prefix;
 608                b_prefix = o->a_prefix;
 609        } else {
 610                a_prefix = o->a_prefix;
 611                b_prefix = o->b_prefix;
 612        }
 613
 614        name_a += (*name_a == '/');
 615        name_b += (*name_b == '/');
 616        name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 617        name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 618
 619        strbuf_reset(&a_name);
 620        strbuf_reset(&b_name);
 621        quote_two_c_style(&a_name, a_prefix, name_a, 0);
 622        quote_two_c_style(&b_name, b_prefix, name_b, 0);
 623
 624        size_one = fill_textconv(textconv_one, one, &data_one);
 625        size_two = fill_textconv(textconv_two, two, &data_two);
 626
 627        memset(&ecbdata, 0, sizeof(ecbdata));
 628        ecbdata.color_diff = want_color(o->use_color);
 629        ecbdata.found_changesp = &o->found_changes;
 630        ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 631        ecbdata.opt = o;
 632        if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 633                mmfile_t mf1, mf2;
 634                mf1.ptr = (char *)data_one;
 635                mf2.ptr = (char *)data_two;
 636                mf1.size = size_one;
 637                mf2.size = size_two;
 638                check_blank_at_eof(&mf1, &mf2, &ecbdata);
 639        }
 640        ecbdata.lno_in_preimage = 1;
 641        ecbdata.lno_in_postimage = 1;
 642
 643        lc_a = count_lines(data_one, size_one);
 644        lc_b = count_lines(data_two, size_two);
 645        fprintf(o->file,
 646                "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 647                line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 648                line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 649                line_prefix, fraginfo);
 650        if (!o->irreversible_delete)
 651                print_line_count(o->file, lc_a);
 652        else
 653                fprintf(o->file, "?,?");
 654        fprintf(o->file, " +");
 655        print_line_count(o->file, lc_b);
 656        fprintf(o->file, " @@%s\n", reset);
 657        if (lc_a && !o->irreversible_delete)
 658                emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 659        if (lc_b)
 660                emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 661        if (textconv_one)
 662                free((char *)data_one);
 663        if (textconv_two)
 664                free((char *)data_two);
 665}
 666
 667struct diff_words_buffer {
 668        mmfile_t text;
 669        long alloc;
 670        struct diff_words_orig {
 671                const char *begin, *end;
 672        } *orig;
 673        int orig_nr, orig_alloc;
 674};
 675
 676static void diff_words_append(char *line, unsigned long len,
 677                struct diff_words_buffer *buffer)
 678{
 679        ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 680        line++;
 681        len--;
 682        memcpy(buffer->text.ptr + buffer->text.size, line, len);
 683        buffer->text.size += len;
 684        buffer->text.ptr[buffer->text.size] = '\0';
 685}
 686
 687struct diff_words_style_elem {
 688        const char *prefix;
 689        const char *suffix;
 690        const char *color; /* NULL; filled in by the setup code if
 691                            * color is enabled */
 692};
 693
 694struct diff_words_style {
 695        enum diff_words_type type;
 696        struct diff_words_style_elem new, old, ctx;
 697        const char *newline;
 698};
 699
 700static struct diff_words_style diff_words_styles[] = {
 701        { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 702        { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 703        { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 704};
 705
 706struct diff_words_data {
 707        struct diff_words_buffer minus, plus;
 708        const char *current_plus;
 709        int last_minus;
 710        struct diff_options *opt;
 711        regex_t *word_regex;
 712        enum diff_words_type type;
 713        struct diff_words_style *style;
 714};
 715
 716static int fn_out_diff_words_write_helper(FILE *fp,
 717                                          struct diff_words_style_elem *st_el,
 718                                          const char *newline,
 719                                          size_t count, const char *buf,
 720                                          const char *line_prefix)
 721{
 722        int print = 0;
 723
 724        while (count) {
 725                char *p = memchr(buf, '\n', count);
 726                if (print)
 727                        fputs(line_prefix, fp);
 728                if (p != buf) {
 729                        if (st_el->color && fputs(st_el->color, fp) < 0)
 730                                return -1;
 731                        if (fputs(st_el->prefix, fp) < 0 ||
 732                            fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 733                            fputs(st_el->suffix, fp) < 0)
 734                                return -1;
 735                        if (st_el->color && *st_el->color
 736                            && fputs(GIT_COLOR_RESET, fp) < 0)
 737                                return -1;
 738                }
 739                if (!p)
 740                        return 0;
 741                if (fputs(newline, fp) < 0)
 742                        return -1;
 743                count -= p + 1 - buf;
 744                buf = p + 1;
 745                print = 1;
 746        }
 747        return 0;
 748}
 749
 750/*
 751 * '--color-words' algorithm can be described as:
 752 *
 753 *   1. collect a the minus/plus lines of a diff hunk, divided into
 754 *      minus-lines and plus-lines;
 755 *
 756 *   2. break both minus-lines and plus-lines into words and
 757 *      place them into two mmfile_t with one word for each line;
 758 *
 759 *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 760 *
 761 * And for the common parts of the both file, we output the plus side text.
 762 * diff_words->current_plus is used to trace the current position of the plus file
 763 * which printed. diff_words->last_minus is used to trace the last minus word
 764 * printed.
 765 *
 766 * For '--graph' to work with '--color-words', we need to output the graph prefix
 767 * on each line of color words output. Generally, there are two conditions on
 768 * which we should output the prefix.
 769 *
 770 *   1. diff_words->last_minus == 0 &&
 771 *      diff_words->current_plus == diff_words->plus.text.ptr
 772 *
 773 *      that is: the plus text must start as a new line, and if there is no minus
 774 *      word printed, a graph prefix must be printed.
 775 *
 776 *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 777 *      *(diff_words->current_plus - 1) == '\n'
 778 *
 779 *      that is: a graph prefix must be printed following a '\n'
 780 */
 781static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 782{
 783        if ((diff_words->last_minus == 0 &&
 784                diff_words->current_plus == diff_words->plus.text.ptr) ||
 785                (diff_words->current_plus > diff_words->plus.text.ptr &&
 786                *(diff_words->current_plus - 1) == '\n')) {
 787                return 1;
 788        } else {
 789                return 0;
 790        }
 791}
 792
 793static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 794{
 795        struct diff_words_data *diff_words = priv;
 796        struct diff_words_style *style = diff_words->style;
 797        int minus_first, minus_len, plus_first, plus_len;
 798        const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 799        struct diff_options *opt = diff_words->opt;
 800        struct strbuf *msgbuf;
 801        char *line_prefix = "";
 802
 803        if (line[0] != '@' || parse_hunk_header(line, len,
 804                        &minus_first, &minus_len, &plus_first, &plus_len))
 805                return;
 806
 807        assert(opt);
 808        if (opt->output_prefix) {
 809                msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 810                line_prefix = msgbuf->buf;
 811        }
 812
 813        /* POSIX requires that first be decremented by one if len == 0... */
 814        if (minus_len) {
 815                minus_begin = diff_words->minus.orig[minus_first].begin;
 816                minus_end =
 817                        diff_words->minus.orig[minus_first + minus_len - 1].end;
 818        } else
 819                minus_begin = minus_end =
 820                        diff_words->minus.orig[minus_first].end;
 821
 822        if (plus_len) {
 823                plus_begin = diff_words->plus.orig[plus_first].begin;
 824                plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 825        } else
 826                plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 827
 828        if (color_words_output_graph_prefix(diff_words)) {
 829                fputs(line_prefix, diff_words->opt->file);
 830        }
 831        if (diff_words->current_plus != plus_begin) {
 832                fn_out_diff_words_write_helper(diff_words->opt->file,
 833                                &style->ctx, style->newline,
 834                                plus_begin - diff_words->current_plus,
 835                                diff_words->current_plus, line_prefix);
 836                if (*(plus_begin - 1) == '\n')
 837                        fputs(line_prefix, diff_words->opt->file);
 838        }
 839        if (minus_begin != minus_end) {
 840                fn_out_diff_words_write_helper(diff_words->opt->file,
 841                                &style->old, style->newline,
 842                                minus_end - minus_begin, minus_begin,
 843                                line_prefix);
 844        }
 845        if (plus_begin != plus_end) {
 846                fn_out_diff_words_write_helper(diff_words->opt->file,
 847                                &style->new, style->newline,
 848                                plus_end - plus_begin, plus_begin,
 849                                line_prefix);
 850        }
 851
 852        diff_words->current_plus = plus_end;
 853        diff_words->last_minus = minus_first;
 854}
 855
 856/* This function starts looking at *begin, and returns 0 iff a word was found. */
 857static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 858                int *begin, int *end)
 859{
 860        if (word_regex && *begin < buffer->size) {
 861                regmatch_t match[1];
 862                if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 863                        char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 864                                        '\n', match[0].rm_eo - match[0].rm_so);
 865                        *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 866                        *begin += match[0].rm_so;
 867                        return *begin >= *end;
 868                }
 869                return -1;
 870        }
 871
 872        /* find the next word */
 873        while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 874                (*begin)++;
 875        if (*begin >= buffer->size)
 876                return -1;
 877
 878        /* find the end of the word */
 879        *end = *begin + 1;
 880        while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 881                (*end)++;
 882
 883        return 0;
 884}
 885
 886/*
 887 * This function splits the words in buffer->text, stores the list with
 888 * newline separator into out, and saves the offsets of the original words
 889 * in buffer->orig.
 890 */
 891static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 892                regex_t *word_regex)
 893{
 894        int i, j;
 895        long alloc = 0;
 896
 897        out->size = 0;
 898        out->ptr = NULL;
 899
 900        /* fake an empty "0th" word */
 901        ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 902        buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 903        buffer->orig_nr = 1;
 904
 905        for (i = 0; i < buffer->text.size; i++) {
 906                if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 907                        return;
 908
 909                /* store original boundaries */
 910                ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 911                                buffer->orig_alloc);
 912                buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 913                buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 914                buffer->orig_nr++;
 915
 916                /* store one word */
 917                ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 918                memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 919                out->ptr[out->size + j - i] = '\n';
 920                out->size += j - i + 1;
 921
 922                i = j - 1;
 923        }
 924}
 925
 926/* this executes the word diff on the accumulated buffers */
 927static void diff_words_show(struct diff_words_data *diff_words)
 928{
 929        xpparam_t xpp;
 930        xdemitconf_t xecfg;
 931        mmfile_t minus, plus;
 932        struct diff_words_style *style = diff_words->style;
 933
 934        struct diff_options *opt = diff_words->opt;
 935        struct strbuf *msgbuf;
 936        char *line_prefix = "";
 937
 938        assert(opt);
 939        if (opt->output_prefix) {
 940                msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
 941                line_prefix = msgbuf->buf;
 942        }
 943
 944        /* special case: only removal */
 945        if (!diff_words->plus.text.size) {
 946                fputs(line_prefix, diff_words->opt->file);
 947                fn_out_diff_words_write_helper(diff_words->opt->file,
 948                        &style->old, style->newline,
 949                        diff_words->minus.text.size,
 950                        diff_words->minus.text.ptr, line_prefix);
 951                diff_words->minus.text.size = 0;
 952                return;
 953        }
 954
 955        diff_words->current_plus = diff_words->plus.text.ptr;
 956        diff_words->last_minus = 0;
 957
 958        memset(&xpp, 0, sizeof(xpp));
 959        memset(&xecfg, 0, sizeof(xecfg));
 960        diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 961        diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 962        xpp.flags = 0;
 963        /* as only the hunk header will be parsed, we need a 0-context */
 964        xecfg.ctxlen = 0;
 965        xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 966                      &xpp, &xecfg);
 967        free(minus.ptr);
 968        free(plus.ptr);
 969        if (diff_words->current_plus != diff_words->plus.text.ptr +
 970                        diff_words->plus.text.size) {
 971                if (color_words_output_graph_prefix(diff_words))
 972                        fputs(line_prefix, diff_words->opt->file);
 973                fn_out_diff_words_write_helper(diff_words->opt->file,
 974                        &style->ctx, style->newline,
 975                        diff_words->plus.text.ptr + diff_words->plus.text.size
 976                        - diff_words->current_plus, diff_words->current_plus,
 977                        line_prefix);
 978        }
 979        diff_words->minus.text.size = diff_words->plus.text.size = 0;
 980}
 981
 982/* In "color-words" mode, show word-diff of words accumulated in the buffer */
 983static void diff_words_flush(struct emit_callback *ecbdata)
 984{
 985        if (ecbdata->diff_words->minus.text.size ||
 986            ecbdata->diff_words->plus.text.size)
 987                diff_words_show(ecbdata->diff_words);
 988}
 989
 990static void free_diff_words_data(struct emit_callback *ecbdata)
 991{
 992        if (ecbdata->diff_words) {
 993                diff_words_flush(ecbdata);
 994                free (ecbdata->diff_words->minus.text.ptr);
 995                free (ecbdata->diff_words->minus.orig);
 996                free (ecbdata->diff_words->plus.text.ptr);
 997                free (ecbdata->diff_words->plus.orig);
 998                if (ecbdata->diff_words->word_regex) {
 999                        regfree(ecbdata->diff_words->word_regex);
1000                        free(ecbdata->diff_words->word_regex);
1001                }
1002                free(ecbdata->diff_words);
1003                ecbdata->diff_words = NULL;
1004        }
1005}
1006
1007const char *diff_get_color(int diff_use_color, enum color_diff ix)
1008{
1009        if (want_color(diff_use_color))
1010                return diff_colors[ix];
1011        return "";
1012}
1013
1014static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1015{
1016        const char *cp;
1017        unsigned long allot;
1018        size_t l = len;
1019
1020        if (ecb->truncate)
1021                return ecb->truncate(line, len);
1022        cp = line;
1023        allot = l;
1024        while (0 < l) {
1025                (void) utf8_width(&cp, &l);
1026                if (!cp)
1027                        break; /* truncated in the middle? */
1028        }
1029        return allot - l;
1030}
1031
1032static void find_lno(const char *line, struct emit_callback *ecbdata)
1033{
1034        const char *p;
1035        ecbdata->lno_in_preimage = 0;
1036        ecbdata->lno_in_postimage = 0;
1037        p = strchr(line, '-');
1038        if (!p)
1039                return; /* cannot happen */
1040        ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1041        p = strchr(p, '+');
1042        if (!p)
1043                return; /* cannot happen */
1044        ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1045}
1046
1047static void fn_out_consume(void *priv, char *line, unsigned long len)
1048{
1049        struct emit_callback *ecbdata = priv;
1050        const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1051        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1052        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1053        struct diff_options *o = ecbdata->opt;
1054        char *line_prefix = "";
1055        struct strbuf *msgbuf;
1056
1057        if (o && o->output_prefix) {
1058                msgbuf = o->output_prefix(o, o->output_prefix_data);
1059                line_prefix = msgbuf->buf;
1060        }
1061
1062        if (ecbdata->header) {
1063                fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1064                strbuf_reset(ecbdata->header);
1065                ecbdata->header = NULL;
1066        }
1067        *(ecbdata->found_changesp) = 1;
1068
1069        if (ecbdata->label_path[0]) {
1070                const char *name_a_tab, *name_b_tab;
1071
1072                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1073                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1074
1075                fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1076                        line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1077                fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1078                        line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1079                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1080        }
1081
1082        if (diff_suppress_blank_empty
1083            && len == 2 && line[0] == ' ' && line[1] == '\n') {
1084                line[0] = '\n';
1085                len = 1;
1086        }
1087
1088        if (line[0] == '@') {
1089                if (ecbdata->diff_words)
1090                        diff_words_flush(ecbdata);
1091                len = sane_truncate_line(ecbdata, line, len);
1092                find_lno(line, ecbdata);
1093                emit_hunk_header(ecbdata, line, len);
1094                if (line[len-1] != '\n')
1095                        putc('\n', ecbdata->opt->file);
1096                return;
1097        }
1098
1099        if (len < 1) {
1100                emit_line(ecbdata->opt, reset, reset, line, len);
1101                if (ecbdata->diff_words
1102                    && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1103                        fputs("~\n", ecbdata->opt->file);
1104                return;
1105        }
1106
1107        if (ecbdata->diff_words) {
1108                if (line[0] == '-') {
1109                        diff_words_append(line, len,
1110                                          &ecbdata->diff_words->minus);
1111                        return;
1112                } else if (line[0] == '+') {
1113                        diff_words_append(line, len,
1114                                          &ecbdata->diff_words->plus);
1115                        return;
1116                } else if (!prefixcmp(line, "\\ ")) {
1117                        /*
1118                         * Eat the "no newline at eof" marker as if we
1119                         * saw a "+" or "-" line with nothing on it,
1120                         * and return without diff_words_flush() to
1121                         * defer processing. If this is the end of
1122                         * preimage, more "+" lines may come after it.
1123                         */
1124                        return;
1125                }
1126                diff_words_flush(ecbdata);
1127                if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1128                        emit_line(ecbdata->opt, plain, reset, line, len);
1129                        fputs("~\n", ecbdata->opt->file);
1130                } else {
1131                        /*
1132                         * Skip the prefix character, if any.  With
1133                         * diff_suppress_blank_empty, there may be
1134                         * none.
1135                         */
1136                        if (line[0] != '\n') {
1137                              line++;
1138                              len--;
1139                        }
1140                        emit_line(ecbdata->opt, plain, reset, line, len);
1141                }
1142                return;
1143        }
1144
1145        if (line[0] != '+') {
1146                const char *color =
1147                        diff_get_color(ecbdata->color_diff,
1148                                       line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1149                ecbdata->lno_in_preimage++;
1150                if (line[0] == ' ')
1151                        ecbdata->lno_in_postimage++;
1152                emit_line(ecbdata->opt, color, reset, line, len);
1153        } else {
1154                ecbdata->lno_in_postimage++;
1155                emit_add_line(reset, ecbdata, line + 1, len - 1);
1156        }
1157}
1158
1159static char *pprint_rename(const char *a, const char *b)
1160{
1161        const char *old = a;
1162        const char *new = b;
1163        struct strbuf name = STRBUF_INIT;
1164        int pfx_length, sfx_length;
1165        int len_a = strlen(a);
1166        int len_b = strlen(b);
1167        int a_midlen, b_midlen;
1168        int qlen_a = quote_c_style(a, NULL, NULL, 0);
1169        int qlen_b = quote_c_style(b, NULL, NULL, 0);
1170
1171        if (qlen_a || qlen_b) {
1172                quote_c_style(a, &name, NULL, 0);
1173                strbuf_addstr(&name, " => ");
1174                quote_c_style(b, &name, NULL, 0);
1175                return strbuf_detach(&name, NULL);
1176        }
1177
1178        /* Find common prefix */
1179        pfx_length = 0;
1180        while (*old && *new && *old == *new) {
1181                if (*old == '/')
1182                        pfx_length = old - a + 1;
1183                old++;
1184                new++;
1185        }
1186
1187        /* Find common suffix */
1188        old = a + len_a;
1189        new = b + len_b;
1190        sfx_length = 0;
1191        while (a <= old && b <= new && *old == *new) {
1192                if (*old == '/')
1193                        sfx_length = len_a - (old - a);
1194                old--;
1195                new--;
1196        }
1197
1198        /*
1199         * pfx{mid-a => mid-b}sfx
1200         * {pfx-a => pfx-b}sfx
1201         * pfx{sfx-a => sfx-b}
1202         * name-a => name-b
1203         */
1204        a_midlen = len_a - pfx_length - sfx_length;
1205        b_midlen = len_b - pfx_length - sfx_length;
1206        if (a_midlen < 0)
1207                a_midlen = 0;
1208        if (b_midlen < 0)
1209                b_midlen = 0;
1210
1211        strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1212        if (pfx_length + sfx_length) {
1213                strbuf_add(&name, a, pfx_length);
1214                strbuf_addch(&name, '{');
1215        }
1216        strbuf_add(&name, a + pfx_length, a_midlen);
1217        strbuf_addstr(&name, " => ");
1218        strbuf_add(&name, b + pfx_length, b_midlen);
1219        if (pfx_length + sfx_length) {
1220                strbuf_addch(&name, '}');
1221                strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1222        }
1223        return strbuf_detach(&name, NULL);
1224}
1225
1226struct diffstat_t {
1227        int nr;
1228        int alloc;
1229        struct diffstat_file {
1230                char *from_name;
1231                char *name;
1232                char *print_name;
1233                unsigned is_unmerged:1;
1234                unsigned is_binary:1;
1235                unsigned is_renamed:1;
1236                uintmax_t added, deleted;
1237        } **files;
1238};
1239
1240static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1241                                          const char *name_a,
1242                                          const char *name_b)
1243{
1244        struct diffstat_file *x;
1245        x = xcalloc(sizeof (*x), 1);
1246        if (diffstat->nr == diffstat->alloc) {
1247                diffstat->alloc = alloc_nr(diffstat->alloc);
1248                diffstat->files = xrealloc(diffstat->files,
1249                                diffstat->alloc * sizeof(x));
1250        }
1251        diffstat->files[diffstat->nr++] = x;
1252        if (name_b) {
1253                x->from_name = xstrdup(name_a);
1254                x->name = xstrdup(name_b);
1255                x->is_renamed = 1;
1256        }
1257        else {
1258                x->from_name = NULL;
1259                x->name = xstrdup(name_a);
1260        }
1261        return x;
1262}
1263
1264static void diffstat_consume(void *priv, char *line, unsigned long len)
1265{
1266        struct diffstat_t *diffstat = priv;
1267        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1268
1269        if (line[0] == '+')
1270                x->added++;
1271        else if (line[0] == '-')
1272                x->deleted++;
1273}
1274
1275const char mime_boundary_leader[] = "------------";
1276
1277static int scale_linear(int it, int width, int max_change)
1278{
1279        /*
1280         * make sure that at least one '-' is printed if there were deletions,
1281         * and likewise for '+'.
1282         */
1283        if (max_change < 2)
1284                return it;
1285        return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1286}
1287
1288static void show_name(FILE *file,
1289                      const char *prefix, const char *name, int len)
1290{
1291        fprintf(file, " %s%-*s |", prefix, len, name);
1292}
1293
1294static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1295{
1296        if (cnt <= 0)
1297                return;
1298        fprintf(file, "%s", set);
1299        while (cnt--)
1300                putc(ch, file);
1301        fprintf(file, "%s", reset);
1302}
1303
1304static void fill_print_name(struct diffstat_file *file)
1305{
1306        char *pname;
1307
1308        if (file->print_name)
1309                return;
1310
1311        if (!file->is_renamed) {
1312                struct strbuf buf = STRBUF_INIT;
1313                if (quote_c_style(file->name, &buf, NULL, 0)) {
1314                        pname = strbuf_detach(&buf, NULL);
1315                } else {
1316                        pname = file->name;
1317                        strbuf_release(&buf);
1318                }
1319        } else {
1320                pname = pprint_rename(file->from_name, file->name);
1321        }
1322        file->print_name = pname;
1323}
1324
1325int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1326{
1327        struct strbuf sb = STRBUF_INIT;
1328        int ret;
1329
1330        if (!files) {
1331                assert(insertions == 0 && deletions == 0);
1332                return fputs(_(" 0 files changed\n"), fp);
1333        }
1334
1335        strbuf_addf(&sb,
1336                    Q_(" %d file changed", " %d files changed", files),
1337                    files);
1338
1339        /*
1340         * For binary diff, the caller may want to print "x files
1341         * changed" with insertions == 0 && deletions == 0.
1342         *
1343         * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1344         * is probably less confusing (i.e skip over "2 files changed
1345         * but nothing about added/removed lines? Is this a bug in Git?").
1346         */
1347        if (insertions || deletions == 0) {
1348                /*
1349                 * TRANSLATORS: "+" in (+) is a line addition marker;
1350                 * do not translate it.
1351                 */
1352                strbuf_addf(&sb,
1353                            Q_(", %d insertion(+)", ", %d insertions(+)",
1354                               insertions),
1355                            insertions);
1356        }
1357
1358        if (deletions || insertions == 0) {
1359                /*
1360                 * TRANSLATORS: "-" in (-) is a line removal marker;
1361                 * do not translate it.
1362                 */
1363                strbuf_addf(&sb,
1364                            Q_(", %d deletion(-)", ", %d deletions(-)",
1365                               deletions),
1366                            deletions);
1367        }
1368        strbuf_addch(&sb, '\n');
1369        ret = fputs(sb.buf, fp);
1370        strbuf_release(&sb);
1371        return ret;
1372}
1373
1374static void show_stats(struct diffstat_t *data, struct diff_options *options)
1375{
1376        int i, len, add, del, adds = 0, dels = 0;
1377        uintmax_t max_change = 0, max_len = 0;
1378        int total_files = data->nr;
1379        int width, name_width, count;
1380        const char *reset, *add_c, *del_c;
1381        const char *line_prefix = "";
1382        int extra_shown = 0;
1383        struct strbuf *msg = NULL;
1384
1385        if (data->nr == 0)
1386                return;
1387
1388        if (options->output_prefix) {
1389                msg = options->output_prefix(options, options->output_prefix_data);
1390                line_prefix = msg->buf;
1391        }
1392
1393        width = options->stat_width ? options->stat_width : 80;
1394        name_width = options->stat_name_width ? options->stat_name_width : 50;
1395        count = options->stat_count ? options->stat_count : data->nr;
1396
1397        /* Sanity: give at least 5 columns to the graph,
1398         * but leave at least 10 columns for the name.
1399         */
1400        if (width < 25)
1401                width = 25;
1402        if (name_width < 10)
1403                name_width = 10;
1404        else if (width < name_width + 15)
1405                name_width = width - 15;
1406
1407        /* Find the longest filename and max number of changes */
1408        reset = diff_get_color_opt(options, DIFF_RESET);
1409        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1410        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1411
1412        for (i = 0; (i < count) && (i < data->nr); i++) {
1413                struct diffstat_file *file = data->files[i];
1414                uintmax_t change = file->added + file->deleted;
1415                if (!data->files[i]->is_renamed &&
1416                         (change == 0)) {
1417                        count++; /* not shown == room for one more */
1418                        continue;
1419                }
1420                fill_print_name(file);
1421                len = strlen(file->print_name);
1422                if (max_len < len)
1423                        max_len = len;
1424
1425                if (file->is_binary || file->is_unmerged)
1426                        continue;
1427                if (max_change < change)
1428                        max_change = change;
1429        }
1430        count = i; /* min(count, data->nr) */
1431
1432        /* Compute the width of the graph part;
1433         * 10 is for one blank at the beginning of the line plus
1434         * " | count " between the name and the graph.
1435         *
1436         * From here on, name_width is the width of the name area,
1437         * and width is the width of the graph area.
1438         */
1439        name_width = (name_width < max_len) ? name_width : max_len;
1440        if (width < (name_width + 10) + max_change)
1441                width = width - (name_width + 10);
1442        else
1443                width = max_change;
1444
1445        for (i = 0; i < count; i++) {
1446                const char *prefix = "";
1447                char *name = data->files[i]->print_name;
1448                uintmax_t added = data->files[i]->added;
1449                uintmax_t deleted = data->files[i]->deleted;
1450                int name_len;
1451
1452                if (!data->files[i]->is_renamed &&
1453                         (added + deleted == 0)) {
1454                        total_files--;
1455                        continue;
1456                }
1457                /*
1458                 * "scale" the filename
1459                 */
1460                len = name_width;
1461                name_len = strlen(name);
1462                if (name_width < name_len) {
1463                        char *slash;
1464                        prefix = "...";
1465                        len -= 3;
1466                        name += name_len - len;
1467                        slash = strchr(name, '/');
1468                        if (slash)
1469                                name = slash;
1470                }
1471
1472                if (data->files[i]->is_binary) {
1473                        fprintf(options->file, "%s", line_prefix);
1474                        show_name(options->file, prefix, name, len);
1475                        fprintf(options->file, "  Bin ");
1476                        fprintf(options->file, "%s%"PRIuMAX"%s",
1477                                del_c, deleted, reset);
1478                        fprintf(options->file, " -> ");
1479                        fprintf(options->file, "%s%"PRIuMAX"%s",
1480                                add_c, added, reset);
1481                        fprintf(options->file, " bytes");
1482                        fprintf(options->file, "\n");
1483                        continue;
1484                }
1485                else if (data->files[i]->is_unmerged) {
1486                        fprintf(options->file, "%s", line_prefix);
1487                        show_name(options->file, prefix, name, len);
1488                        fprintf(options->file, "  Unmerged\n");
1489                        continue;
1490                }
1491
1492                /*
1493                 * scale the add/delete
1494                 */
1495                add = added;
1496                del = deleted;
1497                adds += add;
1498                dels += del;
1499
1500                if (width <= max_change) {
1501                        add = scale_linear(add, width, max_change);
1502                        del = scale_linear(del, width, max_change);
1503                }
1504                fprintf(options->file, "%s", line_prefix);
1505                show_name(options->file, prefix, name, len);
1506                fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1507                                added + deleted ? " " : "");
1508                show_graph(options->file, '+', add, add_c, reset);
1509                show_graph(options->file, '-', del, del_c, reset);
1510                fprintf(options->file, "\n");
1511        }
1512        for (i = count; i < data->nr; i++) {
1513                uintmax_t added = data->files[i]->added;
1514                uintmax_t deleted = data->files[i]->deleted;
1515                if (!data->files[i]->is_renamed &&
1516                         (added + deleted == 0)) {
1517                        total_files--;
1518                        continue;
1519                }
1520                adds += added;
1521                dels += deleted;
1522                if (!extra_shown)
1523                        fprintf(options->file, "%s ...\n", line_prefix);
1524                extra_shown = 1;
1525        }
1526        fprintf(options->file, "%s", line_prefix);
1527        print_stat_summary(options->file, total_files, adds, dels);
1528}
1529
1530static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1531{
1532        int i, adds = 0, dels = 0, total_files = data->nr;
1533
1534        if (data->nr == 0)
1535                return;
1536
1537        for (i = 0; i < data->nr; i++) {
1538                if (!data->files[i]->is_binary &&
1539                    !data->files[i]->is_unmerged) {
1540                        int added = data->files[i]->added;
1541                        int deleted= data->files[i]->deleted;
1542                        if (!data->files[i]->is_renamed &&
1543                            (added + deleted == 0)) {
1544                                total_files--;
1545                        } else {
1546                                adds += added;
1547                                dels += deleted;
1548                        }
1549                }
1550        }
1551        if (options->output_prefix) {
1552                struct strbuf *msg = NULL;
1553                msg = options->output_prefix(options,
1554                                options->output_prefix_data);
1555                fprintf(options->file, "%s", msg->buf);
1556        }
1557        print_stat_summary(options->file, total_files, adds, dels);
1558}
1559
1560static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1561{
1562        int i;
1563
1564        if (data->nr == 0)
1565                return;
1566
1567        for (i = 0; i < data->nr; i++) {
1568                struct diffstat_file *file = data->files[i];
1569
1570                if (options->output_prefix) {
1571                        struct strbuf *msg = NULL;
1572                        msg = options->output_prefix(options,
1573                                        options->output_prefix_data);
1574                        fprintf(options->file, "%s", msg->buf);
1575                }
1576
1577                if (file->is_binary)
1578                        fprintf(options->file, "-\t-\t");
1579                else
1580                        fprintf(options->file,
1581                                "%"PRIuMAX"\t%"PRIuMAX"\t",
1582                                file->added, file->deleted);
1583                if (options->line_termination) {
1584                        fill_print_name(file);
1585                        if (!file->is_renamed)
1586                                write_name_quoted(file->name, options->file,
1587                                                  options->line_termination);
1588                        else {
1589                                fputs(file->print_name, options->file);
1590                                putc(options->line_termination, options->file);
1591                        }
1592                } else {
1593                        if (file->is_renamed) {
1594                                putc('\0', options->file);
1595                                write_name_quoted(file->from_name, options->file, '\0');
1596                        }
1597                        write_name_quoted(file->name, options->file, '\0');
1598                }
1599        }
1600}
1601
1602struct dirstat_file {
1603        const char *name;
1604        unsigned long changed;
1605};
1606
1607struct dirstat_dir {
1608        struct dirstat_file *files;
1609        int alloc, nr, permille, cumulative;
1610};
1611
1612static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1613                unsigned long changed, const char *base, int baselen)
1614{
1615        unsigned long this_dir = 0;
1616        unsigned int sources = 0;
1617        const char *line_prefix = "";
1618        struct strbuf *msg = NULL;
1619
1620        if (opt->output_prefix) {
1621                msg = opt->output_prefix(opt, opt->output_prefix_data);
1622                line_prefix = msg->buf;
1623        }
1624
1625        while (dir->nr) {
1626                struct dirstat_file *f = dir->files;
1627                int namelen = strlen(f->name);
1628                unsigned long this;
1629                char *slash;
1630
1631                if (namelen < baselen)
1632                        break;
1633                if (memcmp(f->name, base, baselen))
1634                        break;
1635                slash = strchr(f->name + baselen, '/');
1636                if (slash) {
1637                        int newbaselen = slash + 1 - f->name;
1638                        this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1639                        sources++;
1640                } else {
1641                        this = f->changed;
1642                        dir->files++;
1643                        dir->nr--;
1644                        sources += 2;
1645                }
1646                this_dir += this;
1647        }
1648
1649        /*
1650         * We don't report dirstat's for
1651         *  - the top level
1652         *  - or cases where everything came from a single directory
1653         *    under this directory (sources == 1).
1654         */
1655        if (baselen && sources != 1) {
1656                if (this_dir) {
1657                        int permille = this_dir * 1000 / changed;
1658                        if (permille >= dir->permille) {
1659                                fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1660                                        permille / 10, permille % 10, baselen, base);
1661                                if (!dir->cumulative)
1662                                        return 0;
1663                        }
1664                }
1665        }
1666        return this_dir;
1667}
1668
1669static int dirstat_compare(const void *_a, const void *_b)
1670{
1671        const struct dirstat_file *a = _a;
1672        const struct dirstat_file *b = _b;
1673        return strcmp(a->name, b->name);
1674}
1675
1676static void show_dirstat(struct diff_options *options)
1677{
1678        int i;
1679        unsigned long changed;
1680        struct dirstat_dir dir;
1681        struct diff_queue_struct *q = &diff_queued_diff;
1682
1683        dir.files = NULL;
1684        dir.alloc = 0;
1685        dir.nr = 0;
1686        dir.permille = options->dirstat_permille;
1687        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1688
1689        changed = 0;
1690        for (i = 0; i < q->nr; i++) {
1691                struct diff_filepair *p = q->queue[i];
1692                const char *name;
1693                unsigned long copied, added, damage;
1694                int content_changed;
1695
1696                name = p->two->path ? p->two->path : p->one->path;
1697
1698                if (p->one->sha1_valid && p->two->sha1_valid)
1699                        content_changed = hashcmp(p->one->sha1, p->two->sha1);
1700                else
1701                        content_changed = 1;
1702
1703                if (!content_changed) {
1704                        /*
1705                         * The SHA1 has not changed, so pre-/post-content is
1706                         * identical. We can therefore skip looking at the
1707                         * file contents altogether.
1708                         */
1709                        damage = 0;
1710                        goto found_damage;
1711                }
1712
1713                if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1714                        /*
1715                         * In --dirstat-by-file mode, we don't really need to
1716                         * look at the actual file contents at all.
1717                         * The fact that the SHA1 changed is enough for us to
1718                         * add this file to the list of results
1719                         * (with each file contributing equal damage).
1720                         */
1721                        damage = 1;
1722                        goto found_damage;
1723                }
1724
1725                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1726                        diff_populate_filespec(p->one, 0);
1727                        diff_populate_filespec(p->two, 0);
1728                        diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1729                                               &copied, &added);
1730                        diff_free_filespec_data(p->one);
1731                        diff_free_filespec_data(p->two);
1732                } else if (DIFF_FILE_VALID(p->one)) {
1733                        diff_populate_filespec(p->one, 1);
1734                        copied = added = 0;
1735                        diff_free_filespec_data(p->one);
1736                } else if (DIFF_FILE_VALID(p->two)) {
1737                        diff_populate_filespec(p->two, 1);
1738                        copied = 0;
1739                        added = p->two->size;
1740                        diff_free_filespec_data(p->two);
1741                } else
1742                        continue;
1743
1744                /*
1745                 * Original minus copied is the removed material,
1746                 * added is the new material.  They are both damages
1747                 * made to the preimage.
1748                 * If the resulting damage is zero, we know that
1749                 * diffcore_count_changes() considers the two entries to
1750                 * be identical, but since content_changed is true, we
1751                 * know that there must have been _some_ kind of change,
1752                 * so we force all entries to have damage > 0.
1753                 */
1754                damage = (p->one->size - copied) + added;
1755                if (!damage)
1756                        damage = 1;
1757
1758found_damage:
1759                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1760                dir.files[dir.nr].name = name;
1761                dir.files[dir.nr].changed = damage;
1762                changed += damage;
1763                dir.nr++;
1764        }
1765
1766        /* This can happen even with many files, if everything was renames */
1767        if (!changed)
1768                return;
1769
1770        /* Show all directories with more than x% of the changes */
1771        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1772        gather_dirstat(options, &dir, changed, "", 0);
1773}
1774
1775static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1776{
1777        int i;
1778        unsigned long changed;
1779        struct dirstat_dir dir;
1780
1781        if (data->nr == 0)
1782                return;
1783
1784        dir.files = NULL;
1785        dir.alloc = 0;
1786        dir.nr = 0;
1787        dir.permille = options->dirstat_permille;
1788        dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1789
1790        changed = 0;
1791        for (i = 0; i < data->nr; i++) {
1792                struct diffstat_file *file = data->files[i];
1793                unsigned long damage = file->added + file->deleted;
1794                if (file->is_binary)
1795                        /*
1796                         * binary files counts bytes, not lines. Must find some
1797                         * way to normalize binary bytes vs. textual lines.
1798                         * The following heuristic assumes that there are 64
1799                         * bytes per "line".
1800                         * This is stupid and ugly, but very cheap...
1801                         */
1802                        damage = (damage + 63) / 64;
1803                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1804                dir.files[dir.nr].name = file->name;
1805                dir.files[dir.nr].changed = damage;
1806                changed += damage;
1807                dir.nr++;
1808        }
1809
1810        /* This can happen even with many files, if everything was renames */
1811        if (!changed)
1812                return;
1813
1814        /* Show all directories with more than x% of the changes */
1815        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1816        gather_dirstat(options, &dir, changed, "", 0);
1817}
1818
1819static void free_diffstat_info(struct diffstat_t *diffstat)
1820{
1821        int i;
1822        for (i = 0; i < diffstat->nr; i++) {
1823                struct diffstat_file *f = diffstat->files[i];
1824                if (f->name != f->print_name)
1825                        free(f->print_name);
1826                free(f->name);
1827                free(f->from_name);
1828                free(f);
1829        }
1830        free(diffstat->files);
1831}
1832
1833struct checkdiff_t {
1834        const char *filename;
1835        int lineno;
1836        int conflict_marker_size;
1837        struct diff_options *o;
1838        unsigned ws_rule;
1839        unsigned status;
1840};
1841
1842static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1843{
1844        char firstchar;
1845        int cnt;
1846
1847        if (len < marker_size + 1)
1848                return 0;
1849        firstchar = line[0];
1850        switch (firstchar) {
1851        case '=': case '>': case '<': case '|':
1852                break;
1853        default:
1854                return 0;
1855        }
1856        for (cnt = 1; cnt < marker_size; cnt++)
1857                if (line[cnt] != firstchar)
1858                        return 0;
1859        /* line[1] thru line[marker_size-1] are same as firstchar */
1860        if (len < marker_size + 1 || !isspace(line[marker_size]))
1861                return 0;
1862        return 1;
1863}
1864
1865static void checkdiff_consume(void *priv, char *line, unsigned long len)
1866{
1867        struct checkdiff_t *data = priv;
1868        int marker_size = data->conflict_marker_size;
1869        const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
1870        const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
1871        const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
1872        char *err;
1873        char *line_prefix = "";
1874        struct strbuf *msgbuf;
1875
1876        assert(data->o);
1877        if (data->o->output_prefix) {
1878                msgbuf = data->o->output_prefix(data->o,
1879                        data->o->output_prefix_data);
1880                line_prefix = msgbuf->buf;
1881        }
1882
1883        if (line[0] == '+') {
1884                unsigned bad;
1885                data->lineno++;
1886                if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1887                        data->status |= 1;
1888                        fprintf(data->o->file,
1889                                "%s%s:%d: leftover conflict marker\n",
1890                                line_prefix, data->filename, data->lineno);
1891                }
1892                bad = ws_check(line + 1, len - 1, data->ws_rule);
1893                if (!bad)
1894                        return;
1895                data->status |= bad;
1896                err = whitespace_error_string(bad);
1897                fprintf(data->o->file, "%s%s:%d: %s.\n",
1898                        line_prefix, data->filename, data->lineno, err);
1899                free(err);
1900                emit_line(data->o, set, reset, line, 1);
1901                ws_check_emit(line + 1, len - 1, data->ws_rule,
1902                              data->o->file, set, reset, ws);
1903        } else if (line[0] == ' ') {
1904                data->lineno++;
1905        } else if (line[0] == '@') {
1906                char *plus = strchr(line, '+');
1907                if (plus)
1908                        data->lineno = strtol(plus, NULL, 10) - 1;
1909                else
1910                        die("invalid diff");
1911        }
1912}
1913
1914static unsigned char *deflate_it(char *data,
1915                                 unsigned long size,
1916                                 unsigned long *result_size)
1917{
1918        int bound;
1919        unsigned char *deflated;
1920        git_zstream stream;
1921
1922        memset(&stream, 0, sizeof(stream));
1923        git_deflate_init(&stream, zlib_compression_level);
1924        bound = git_deflate_bound(&stream, size);
1925        deflated = xmalloc(bound);
1926        stream.next_out = deflated;
1927        stream.avail_out = bound;
1928
1929        stream.next_in = (unsigned char *)data;
1930        stream.avail_in = size;
1931        while (git_deflate(&stream, Z_FINISH) == Z_OK)
1932                ; /* nothing */
1933        git_deflate_end(&stream);
1934        *result_size = stream.total_out;
1935        return deflated;
1936}
1937
1938static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1939{
1940        void *cp;
1941        void *delta;
1942        void *deflated;
1943        void *data;
1944        unsigned long orig_size;
1945        unsigned long delta_size;
1946        unsigned long deflate_size;
1947        unsigned long data_size;
1948
1949        /* We could do deflated delta, or we could do just deflated two,
1950         * whichever is smaller.
1951         */
1952        delta = NULL;
1953        deflated = deflate_it(two->ptr, two->size, &deflate_size);
1954        if (one->size && two->size) {
1955                delta = diff_delta(one->ptr, one->size,
1956                                   two->ptr, two->size,
1957                                   &delta_size, deflate_size);
1958                if (delta) {
1959                        void *to_free = delta;
1960                        orig_size = delta_size;
1961                        delta = deflate_it(delta, delta_size, &delta_size);
1962                        free(to_free);
1963                }
1964        }
1965
1966        if (delta && delta_size < deflate_size) {
1967                fprintf(file, "%sdelta %lu\n", prefix, orig_size);
1968                free(deflated);
1969                data = delta;
1970                data_size = delta_size;
1971        }
1972        else {
1973                fprintf(file, "%sliteral %lu\n", prefix, two->size);
1974                free(delta);
1975                data = deflated;
1976                data_size = deflate_size;
1977        }
1978
1979        /* emit data encoded in base85 */
1980        cp = data;
1981        while (data_size) {
1982                int bytes = (52 < data_size) ? 52 : data_size;
1983                char line[70];
1984                data_size -= bytes;
1985                if (bytes <= 26)
1986                        line[0] = bytes + 'A' - 1;
1987                else
1988                        line[0] = bytes - 26 + 'a' - 1;
1989                encode_85(line + 1, cp, bytes);
1990                cp = (char *) cp + bytes;
1991                fprintf(file, "%s", prefix);
1992                fputs(line, file);
1993                fputc('\n', file);
1994        }
1995        fprintf(file, "%s\n", prefix);
1996        free(data);
1997}
1998
1999static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2000{
2001        fprintf(file, "%sGIT binary patch\n", prefix);
2002        emit_binary_diff_body(file, one, two, prefix);
2003        emit_binary_diff_body(file, two, one, prefix);
2004}
2005
2006static void diff_filespec_load_driver(struct diff_filespec *one)
2007{
2008        /* Use already-loaded driver */
2009        if (one->driver)
2010                return;
2011
2012        if (S_ISREG(one->mode))
2013                one->driver = userdiff_find_by_path(one->path);
2014
2015        /* Fallback to default settings */
2016        if (!one->driver)
2017                one->driver = userdiff_find_by_name("default");
2018}
2019
2020int diff_filespec_is_binary(struct diff_filespec *one)
2021{
2022        if (one->is_binary == -1) {
2023                diff_filespec_load_driver(one);
2024                if (one->driver->binary != -1)
2025                        one->is_binary = one->driver->binary;
2026                else {
2027                        if (!one->data && DIFF_FILE_VALID(one))
2028                                diff_populate_filespec(one, 0);
2029                        if (one->data)
2030                                one->is_binary = buffer_is_binary(one->data,
2031                                                one->size);
2032                        if (one->is_binary == -1)
2033                                one->is_binary = 0;
2034                }
2035        }
2036        return one->is_binary;
2037}
2038
2039static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2040{
2041        diff_filespec_load_driver(one);
2042        return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2043}
2044
2045static const char *userdiff_word_regex(struct diff_filespec *one)
2046{
2047        diff_filespec_load_driver(one);
2048        return one->driver->word_regex;
2049}
2050
2051void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2052{
2053        if (!options->a_prefix)
2054                options->a_prefix = a;
2055        if (!options->b_prefix)
2056                options->b_prefix = b;
2057}
2058
2059struct userdiff_driver *get_textconv(struct diff_filespec *one)
2060{
2061        if (!DIFF_FILE_VALID(one))
2062                return NULL;
2063
2064        diff_filespec_load_driver(one);
2065        return userdiff_get_textconv(one->driver);
2066}
2067
2068static void builtin_diff(const char *name_a,
2069                         const char *name_b,
2070                         struct diff_filespec *one,
2071                         struct diff_filespec *two,
2072                         const char *xfrm_msg,
2073                         int must_show_header,
2074                         struct diff_options *o,
2075                         int complete_rewrite)
2076{
2077        mmfile_t mf1, mf2;
2078        const char *lbl[2];
2079        char *a_one, *b_two;
2080        const char *set = diff_get_color_opt(o, DIFF_METAINFO);
2081        const char *reset = diff_get_color_opt(o, DIFF_RESET);
2082        const char *a_prefix, *b_prefix;
2083        struct userdiff_driver *textconv_one = NULL;
2084        struct userdiff_driver *textconv_two = NULL;
2085        struct strbuf header = STRBUF_INIT;
2086        struct strbuf *msgbuf;
2087        char *line_prefix = "";
2088
2089        if (o->output_prefix) {
2090                msgbuf = o->output_prefix(o, o->output_prefix_data);
2091                line_prefix = msgbuf->buf;
2092        }
2093
2094        if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2095                        (!one->mode || S_ISGITLINK(one->mode)) &&
2096                        (!two->mode || S_ISGITLINK(two->mode))) {
2097                const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2098                const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2099                show_submodule_summary(o->file, one ? one->path : two->path,
2100                                one->sha1, two->sha1, two->dirty_submodule,
2101                                del, add, reset);
2102                return;
2103        }
2104
2105        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2106                textconv_one = get_textconv(one);
2107                textconv_two = get_textconv(two);
2108        }
2109
2110        diff_set_mnemonic_prefix(o, "a/", "b/");
2111        if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2112                a_prefix = o->b_prefix;
2113                b_prefix = o->a_prefix;
2114        } else {
2115                a_prefix = o->a_prefix;
2116                b_prefix = o->b_prefix;
2117        }
2118
2119        /* Never use a non-valid filename anywhere if at all possible */
2120        name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2121        name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2122
2123        a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2124        b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2125        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2126        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2127        strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
2128        if (lbl[0][0] == '/') {
2129                /* /dev/null */
2130                strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
2131                if (xfrm_msg)
2132                        strbuf_addstr(&header, xfrm_msg);
2133                must_show_header = 1;
2134        }
2135        else if (lbl[1][0] == '/') {
2136                strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
2137                if (xfrm_msg)
2138                        strbuf_addstr(&header, xfrm_msg);
2139                must_show_header = 1;
2140        }
2141        else {
2142                if (one->mode != two->mode) {
2143                        strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
2144                        strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
2145                        must_show_header = 1;
2146                }
2147                if (xfrm_msg)
2148                        strbuf_addstr(&header, xfrm_msg);
2149
2150                /*
2151                 * we do not run diff between different kind
2152                 * of objects.
2153                 */
2154                if ((one->mode ^ two->mode) & S_IFMT)
2155                        goto free_ab_and_return;
2156                if (complete_rewrite &&
2157                    (textconv_one || !diff_filespec_is_binary(one)) &&
2158                    (textconv_two || !diff_filespec_is_binary(two))) {
2159                        fprintf(o->file, "%s", header.buf);
2160                        strbuf_reset(&header);
2161                        emit_rewrite_diff(name_a, name_b, one, two,
2162                                                textconv_one, textconv_two, o);
2163                        o->found_changes = 1;
2164                        goto free_ab_and_return;
2165                }
2166        }
2167
2168        if (o->irreversible_delete && lbl[1][0] == '/') {
2169                fprintf(o->file, "%s", header.buf);
2170                strbuf_reset(&header);
2171                goto free_ab_and_return;
2172        } else if (!DIFF_OPT_TST(o, TEXT) &&
2173            ( (!textconv_one && diff_filespec_is_binary(one)) ||
2174              (!textconv_two && diff_filespec_is_binary(two)) )) {
2175                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2176                        die("unable to read files to diff");
2177                /* Quite common confusing case */
2178                if (mf1.size == mf2.size &&
2179                    !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2180                        if (must_show_header)
2181                                fprintf(o->file, "%s", header.buf);
2182                        goto free_ab_and_return;
2183                }
2184                fprintf(o->file, "%s", header.buf);
2185                strbuf_reset(&header);
2186                if (DIFF_OPT_TST(o, BINARY))
2187                        emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2188                else
2189                        fprintf(o->file, "%sBinary files %s and %s differ\n",
2190                                line_prefix, lbl[0], lbl[1]);
2191                o->found_changes = 1;
2192        } else {
2193                /* Crazy xdl interfaces.. */
2194                const char *diffopts = getenv("GIT_DIFF_OPTS");
2195                xpparam_t xpp;
2196                xdemitconf_t xecfg;
2197                struct emit_callback ecbdata;
2198                const struct userdiff_funcname *pe;
2199
2200                if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
2201                        fprintf(o->file, "%s", header.buf);
2202                        strbuf_reset(&header);
2203                }
2204
2205                mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2206                mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2207
2208                pe = diff_funcname_pattern(one);
2209                if (!pe)
2210                        pe = diff_funcname_pattern(two);
2211
2212                memset(&xpp, 0, sizeof(xpp));
2213                memset(&xecfg, 0, sizeof(xecfg));
2214                memset(&ecbdata, 0, sizeof(ecbdata));
2215                ecbdata.label_path = lbl;
2216                ecbdata.color_diff = want_color(o->use_color);
2217                ecbdata.found_changesp = &o->found_changes;
2218                ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2219                if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2220                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2221                ecbdata.opt = o;
2222                ecbdata.header = header.len ? &header : NULL;
2223                xpp.flags = o->xdl_opts;
2224                xecfg.ctxlen = o->context;
2225                xecfg.interhunkctxlen = o->interhunkcontext;
2226                xecfg.flags = XDL_EMIT_FUNCNAMES;
2227                if (DIFF_OPT_TST(o, FUNCCONTEXT))
2228                        xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2229                if (pe)
2230                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2231                if (!diffopts)
2232                        ;
2233                else if (!prefixcmp(diffopts, "--unified="))
2234                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2235                else if (!prefixcmp(diffopts, "-u"))
2236                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2237                if (o->word_diff) {
2238                        int i;
2239
2240                        ecbdata.diff_words =
2241                                xcalloc(1, sizeof(struct diff_words_data));
2242                        ecbdata.diff_words->type = o->word_diff;
2243                        ecbdata.diff_words->opt = o;
2244                        if (!o->word_regex)
2245                                o->word_regex = userdiff_word_regex(one);
2246                        if (!o->word_regex)
2247                                o->word_regex = userdiff_word_regex(two);
2248                        if (!o->word_regex)
2249                                o->word_regex = diff_word_regex_cfg;
2250                        if (o->word_regex) {
2251                                ecbdata.diff_words->word_regex = (regex_t *)
2252                                        xmalloc(sizeof(regex_t));
2253                                if (regcomp(ecbdata.diff_words->word_regex,
2254                                                o->word_regex,
2255                                                REG_EXTENDED | REG_NEWLINE))
2256                                        die ("Invalid regular expression: %s",
2257                                                        o->word_regex);
2258                        }
2259                        for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2260                                if (o->word_diff == diff_words_styles[i].type) {
2261                                        ecbdata.diff_words->style =
2262                                                &diff_words_styles[i];
2263                                        break;
2264                                }
2265                        }
2266                        if (want_color(o->use_color)) {
2267                                struct diff_words_style *st = ecbdata.diff_words->style;
2268                                st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2269                                st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2270                                st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
2271                        }
2272                }
2273                xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2274                              &xpp, &xecfg);
2275                if (o->word_diff)
2276                        free_diff_words_data(&ecbdata);
2277                if (textconv_one)
2278                        free(mf1.ptr);
2279                if (textconv_two)
2280                        free(mf2.ptr);
2281                xdiff_clear_find_func(&xecfg);
2282        }
2283
2284 free_ab_and_return:
2285        strbuf_release(&header);
2286        diff_free_filespec_data(one);
2287        diff_free_filespec_data(two);
2288        free(a_one);
2289        free(b_two);
2290        return;
2291}
2292
2293static void builtin_diffstat(const char *name_a, const char *name_b,
2294                             struct diff_filespec *one,
2295                             struct diff_filespec *two,
2296                             struct diffstat_t *diffstat,
2297                             struct diff_options *o,
2298                             int complete_rewrite)
2299{
2300        mmfile_t mf1, mf2;
2301        struct diffstat_file *data;
2302
2303        data = diffstat_add(diffstat, name_a, name_b);
2304
2305        if (!one || !two) {
2306                data->is_unmerged = 1;
2307                return;
2308        }
2309
2310        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2311                data->is_binary = 1;
2312                data->added = diff_filespec_size(two);
2313                data->deleted = diff_filespec_size(one);
2314        }
2315
2316        else if (complete_rewrite) {
2317                diff_populate_filespec(one, 0);
2318                diff_populate_filespec(two, 0);
2319                data->deleted = count_lines(one->data, one->size);
2320                data->added = count_lines(two->data, two->size);
2321        }
2322
2323        else {
2324                /* Crazy xdl interfaces.. */
2325                xpparam_t xpp;
2326                xdemitconf_t xecfg;
2327
2328                if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2329                        die("unable to read files to diff");
2330
2331                memset(&xpp, 0, sizeof(xpp));
2332                memset(&xecfg, 0, sizeof(xecfg));
2333                xpp.flags = o->xdl_opts;
2334                xecfg.ctxlen = o->context;
2335                xecfg.interhunkctxlen = o->interhunkcontext;
2336                xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2337                              &xpp, &xecfg);
2338        }
2339
2340        diff_free_filespec_data(one);
2341        diff_free_filespec_data(two);
2342}
2343
2344static void builtin_checkdiff(const char *name_a, const char *name_b,
2345                              const char *attr_path,
2346                              struct diff_filespec *one,
2347                              struct diff_filespec *two,
2348                              struct diff_options *o)
2349{
2350        mmfile_t mf1, mf2;
2351        struct checkdiff_t data;
2352
2353        if (!two)
2354                return;
2355
2356        memset(&data, 0, sizeof(data));
2357        data.filename = name_b ? name_b : name_a;
2358        data.lineno = 0;
2359        data.o = o;
2360        data.ws_rule = whitespace_rule(attr_path);
2361        data.conflict_marker_size = ll_merge_marker_size(attr_path);
2362
2363        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2364                die("unable to read files to diff");
2365
2366        /*
2367         * All the other codepaths check both sides, but not checking
2368         * the "old" side here is deliberate.  We are checking the newly
2369         * introduced changes, and as long as the "new" side is text, we
2370         * can and should check what it introduces.
2371         */
2372        if (diff_filespec_is_binary(two))
2373                goto free_and_return;
2374        else {
2375                /* Crazy xdl interfaces.. */
2376                xpparam_t xpp;
2377                xdemitconf_t xecfg;
2378
2379                memset(&xpp, 0, sizeof(xpp));
2380                memset(&xecfg, 0, sizeof(xecfg));
2381                xecfg.ctxlen = 1; /* at least one context line */
2382                xpp.flags = 0;
2383                xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2384                              &xpp, &xecfg);
2385
2386                if (data.ws_rule & WS_BLANK_AT_EOF) {
2387                        struct emit_callback ecbdata;
2388                        int blank_at_eof;
2389
2390                        ecbdata.ws_rule = data.ws_rule;
2391                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
2392                        blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2393
2394                        if (blank_at_eof) {
2395                                static char *err;
2396                                if (!err)
2397                                        err = whitespace_error_string(WS_BLANK_AT_EOF);
2398                                fprintf(o->file, "%s:%d: %s.\n",
2399                                        data.filename, blank_at_eof, err);
2400                                data.status = 1; /* report errors */
2401                        }
2402                }
2403        }
2404 free_and_return:
2405        diff_free_filespec_data(one);
2406        diff_free_filespec_data(two);
2407        if (data.status)
2408                DIFF_OPT_SET(o, CHECK_FAILED);
2409}
2410
2411struct diff_filespec *alloc_filespec(const char *path)
2412{
2413        int namelen = strlen(path);
2414        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2415
2416        memset(spec, 0, sizeof(*spec));
2417        spec->path = (char *)(spec + 1);
2418        memcpy(spec->path, path, namelen+1);
2419        spec->count = 1;
2420        spec->is_binary = -1;
2421        return spec;
2422}
2423
2424void free_filespec(struct diff_filespec *spec)
2425{
2426        if (!--spec->count) {
2427                diff_free_filespec_data(spec);
2428                free(spec);
2429        }
2430}
2431
2432void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2433                   unsigned short mode)
2434{
2435        if (mode) {
2436                spec->mode = canon_mode(mode);
2437                hashcpy(spec->sha1, sha1);
2438                spec->sha1_valid = !is_null_sha1(sha1);
2439        }
2440}
2441
2442/*
2443 * Given a name and sha1 pair, if the index tells us the file in
2444 * the work tree has that object contents, return true, so that
2445 * prepare_temp_file() does not have to inflate and extract.
2446 */
2447static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2448{
2449        struct cache_entry *ce;
2450        struct stat st;
2451        int pos, len;
2452
2453        /*
2454         * We do not read the cache ourselves here, because the
2455         * benchmark with my previous version that always reads cache
2456         * shows that it makes things worse for diff-tree comparing
2457         * two linux-2.6 kernel trees in an already checked out work
2458         * tree.  This is because most diff-tree comparisons deal with
2459         * only a small number of files, while reading the cache is
2460         * expensive for a large project, and its cost outweighs the
2461         * savings we get by not inflating the object to a temporary
2462         * file.  Practically, this code only helps when we are used
2463         * by diff-cache --cached, which does read the cache before
2464         * calling us.
2465         */
2466        if (!active_cache)
2467                return 0;
2468
2469        /* We want to avoid the working directory if our caller
2470         * doesn't need the data in a normal file, this system
2471         * is rather slow with its stat/open/mmap/close syscalls,
2472         * and the object is contained in a pack file.  The pack
2473         * is probably already open and will be faster to obtain
2474         * the data through than the working directory.  Loose
2475         * objects however would tend to be slower as they need
2476         * to be individually opened and inflated.
2477         */
2478        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2479                return 0;
2480
2481        len = strlen(name);
2482        pos = cache_name_pos(name, len);
2483        if (pos < 0)
2484                return 0;
2485        ce = active_cache[pos];
2486
2487        /*
2488         * This is not the sha1 we are looking for, or
2489         * unreusable because it is not a regular file.
2490         */
2491        if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2492                return 0;
2493
2494        /*
2495         * If ce is marked as "assume unchanged", there is no
2496         * guarantee that work tree matches what we are looking for.
2497         */
2498        if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2499                return 0;
2500
2501        /*
2502         * If ce matches the file in the work tree, we can reuse it.
2503         */
2504        if (ce_uptodate(ce) ||
2505            (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2506                return 1;
2507
2508        return 0;
2509}
2510
2511static int populate_from_stdin(struct diff_filespec *s)
2512{
2513        struct strbuf buf = STRBUF_INIT;
2514        size_t size = 0;
2515
2516        if (strbuf_read(&buf, 0, 0) < 0)
2517                return error("error while reading from stdin %s",
2518                                     strerror(errno));
2519
2520        s->should_munmap = 0;
2521        s->data = strbuf_detach(&buf, &size);
2522        s->size = size;
2523        s->should_free = 1;
2524        return 0;
2525}
2526
2527static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2528{
2529        int len;
2530        char *data = xmalloc(100), *dirty = "";
2531
2532        /* Are we looking at the work tree? */
2533        if (s->dirty_submodule)
2534                dirty = "-dirty";
2535
2536        len = snprintf(data, 100,
2537                       "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2538        s->data = data;
2539        s->size = len;
2540        s->should_free = 1;
2541        if (size_only) {
2542                s->data = NULL;
2543                free(data);
2544        }
2545        return 0;
2546}
2547
2548/*
2549 * While doing rename detection and pickaxe operation, we may need to
2550 * grab the data for the blob (or file) for our own in-core comparison.
2551 * diff_filespec has data and size fields for this purpose.
2552 */
2553int diff_populate_filespec(struct diff_filespec *s, int size_only)
2554{
2555        int err = 0;
2556        if (!DIFF_FILE_VALID(s))
2557                die("internal error: asking to populate invalid file.");
2558        if (S_ISDIR(s->mode))
2559                return -1;
2560
2561        if (s->data)
2562                return 0;
2563
2564        if (size_only && 0 < s->size)
2565                return 0;
2566
2567        if (S_ISGITLINK(s->mode))
2568                return diff_populate_gitlink(s, size_only);
2569
2570        if (!s->sha1_valid ||
2571            reuse_worktree_file(s->path, s->sha1, 0)) {
2572                struct strbuf buf = STRBUF_INIT;
2573                struct stat st;
2574                int fd;
2575
2576                if (!strcmp(s->path, "-"))
2577                        return populate_from_stdin(s);
2578
2579                if (lstat(s->path, &st) < 0) {
2580                        if (errno == ENOENT) {
2581                        err_empty:
2582                                err = -1;
2583                        empty:
2584                                s->data = (char *)"";
2585                                s->size = 0;
2586                                return err;
2587                        }
2588                }
2589                s->size = xsize_t(st.st_size);
2590                if (!s->size)
2591                        goto empty;
2592                if (S_ISLNK(st.st_mode)) {
2593                        struct strbuf sb = STRBUF_INIT;
2594
2595                        if (strbuf_readlink(&sb, s->path, s->size))
2596                                goto err_empty;
2597                        s->size = sb.len;
2598                        s->data = strbuf_detach(&sb, NULL);
2599                        s->should_free = 1;
2600                        return 0;
2601                }
2602                if (size_only)
2603                        return 0;
2604                fd = open(s->path, O_RDONLY);
2605                if (fd < 0)
2606                        goto err_empty;
2607                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2608                close(fd);
2609                s->should_munmap = 1;
2610
2611                /*
2612                 * Convert from working tree format to canonical git format
2613                 */
2614                if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2615                        size_t size = 0;
2616                        munmap(s->data, s->size);
2617                        s->should_munmap = 0;
2618                        s->data = strbuf_detach(&buf, &size);
2619                        s->size = size;
2620                        s->should_free = 1;
2621                }
2622        }
2623        else {
2624                enum object_type type;
2625                if (size_only) {
2626                        type = sha1_object_info(s->sha1, &s->size);
2627                        if (type < 0)
2628                                die("unable to read %s", sha1_to_hex(s->sha1));
2629                } else {
2630                        s->data = read_sha1_file(s->sha1, &type, &s->size);
2631                        if (!s->data)
2632                                die("unable to read %s", sha1_to_hex(s->sha1));
2633                        s->should_free = 1;
2634                }
2635        }
2636        return 0;
2637}
2638
2639void diff_free_filespec_blob(struct diff_filespec *s)
2640{
2641        if (s->should_free)
2642                free(s->data);
2643        else if (s->should_munmap)
2644                munmap(s->data, s->size);
2645
2646        if (s->should_free || s->should_munmap) {
2647                s->should_free = s->should_munmap = 0;
2648                s->data = NULL;
2649        }
2650}
2651
2652void diff_free_filespec_data(struct diff_filespec *s)
2653{
2654        diff_free_filespec_blob(s);
2655        free(s->cnt_data);
2656        s->cnt_data = NULL;
2657}
2658
2659static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2660                           void *blob,
2661                           unsigned long size,
2662                           const unsigned char *sha1,
2663                           int mode)
2664{
2665        int fd;
2666        struct strbuf buf = STRBUF_INIT;
2667        struct strbuf template = STRBUF_INIT;
2668        char *path_dup = xstrdup(path);
2669        const char *base = basename(path_dup);
2670
2671        /* Generate "XXXXXX_basename.ext" */
2672        strbuf_addstr(&template, "XXXXXX_");
2673        strbuf_addstr(&template, base);
2674
2675        fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2676                        strlen(base) + 1);
2677        if (fd < 0)
2678                die_errno("unable to create temp-file");
2679        if (convert_to_working_tree(path,
2680                        (const char *)blob, (size_t)size, &buf)) {
2681                blob = buf.buf;
2682                size = buf.len;
2683        }
2684        if (write_in_full(fd, blob, size) != size)
2685                die_errno("unable to write temp-file");
2686        close(fd);
2687        temp->name = temp->tmp_path;
2688        strcpy(temp->hex, sha1_to_hex(sha1));
2689        temp->hex[40] = 0;
2690        sprintf(temp->mode, "%06o", mode);
2691        strbuf_release(&buf);
2692        strbuf_release(&template);
2693        free(path_dup);
2694}
2695
2696static struct diff_tempfile *prepare_temp_file(const char *name,
2697                struct diff_filespec *one)
2698{
2699        struct diff_tempfile *temp = claim_diff_tempfile();
2700
2701        if (!DIFF_FILE_VALID(one)) {
2702        not_a_valid_file:
2703                /* A '-' entry produces this for file-2, and
2704                 * a '+' entry produces this for file-1.
2705                 */
2706                temp->name = "/dev/null";
2707                strcpy(temp->hex, ".");
2708                strcpy(temp->mode, ".");
2709                return temp;
2710        }
2711
2712        if (!remove_tempfile_installed) {
2713                atexit(remove_tempfile);
2714                sigchain_push_common(remove_tempfile_on_signal);
2715                remove_tempfile_installed = 1;
2716        }
2717
2718        if (!one->sha1_valid ||
2719            reuse_worktree_file(name, one->sha1, 1)) {
2720                struct stat st;
2721                if (lstat(name, &st) < 0) {
2722                        if (errno == ENOENT)
2723                                goto not_a_valid_file;
2724                        die_errno("stat(%s)", name);
2725                }
2726                if (S_ISLNK(st.st_mode)) {
2727                        struct strbuf sb = STRBUF_INIT;
2728                        if (strbuf_readlink(&sb, name, st.st_size) < 0)
2729                                die_errno("readlink(%s)", name);
2730                        prep_temp_blob(name, temp, sb.buf, sb.len,
2731                                       (one->sha1_valid ?
2732                                        one->sha1 : null_sha1),
2733                                       (one->sha1_valid ?
2734                                        one->mode : S_IFLNK));
2735                        strbuf_release(&sb);
2736                }
2737                else {
2738                        /* we can borrow from the file in the work tree */
2739                        temp->name = name;
2740                        if (!one->sha1_valid)
2741                                strcpy(temp->hex, sha1_to_hex(null_sha1));
2742                        else
2743                                strcpy(temp->hex, sha1_to_hex(one->sha1));
2744                        /* Even though we may sometimes borrow the
2745                         * contents from the work tree, we always want
2746                         * one->mode.  mode is trustworthy even when
2747                         * !(one->sha1_valid), as long as
2748                         * DIFF_FILE_VALID(one).
2749                         */
2750                        sprintf(temp->mode, "%06o", one->mode);
2751                }
2752                return temp;
2753        }
2754        else {
2755                if (diff_populate_filespec(one, 0))
2756                        die("cannot read data blob for %s", one->path);
2757                prep_temp_blob(name, temp, one->data, one->size,
2758                               one->sha1, one->mode);
2759        }
2760        return temp;
2761}
2762
2763/* An external diff command takes:
2764 *
2765 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2766 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2767 *
2768 */
2769static void run_external_diff(const char *pgm,
2770                              const char *name,
2771                              const char *other,
2772                              struct diff_filespec *one,
2773                              struct diff_filespec *two,
2774                              const char *xfrm_msg,
2775                              int complete_rewrite)
2776{
2777        const char *spawn_arg[10];
2778        int retval;
2779        const char **arg = &spawn_arg[0];
2780
2781        if (one && two) {
2782                struct diff_tempfile *temp_one, *temp_two;
2783                const char *othername = (other ? other : name);
2784                temp_one = prepare_temp_file(name, one);
2785                temp_two = prepare_temp_file(othername, two);
2786                *arg++ = pgm;
2787                *arg++ = name;
2788                *arg++ = temp_one->name;
2789                *arg++ = temp_one->hex;
2790                *arg++ = temp_one->mode;
2791                *arg++ = temp_two->name;
2792                *arg++ = temp_two->hex;
2793                *arg++ = temp_two->mode;
2794                if (other) {
2795                        *arg++ = other;
2796                        *arg++ = xfrm_msg;
2797                }
2798        } else {
2799                *arg++ = pgm;
2800                *arg++ = name;
2801        }
2802        *arg = NULL;
2803        fflush(NULL);
2804        retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2805        remove_tempfile();
2806        if (retval) {
2807                fprintf(stderr, "external diff died, stopping at %s.\n", name);
2808                exit(1);
2809        }
2810}
2811
2812static int similarity_index(struct diff_filepair *p)
2813{
2814        return p->score * 100 / MAX_SCORE;
2815}
2816
2817static void fill_metainfo(struct strbuf *msg,
2818                          const char *name,
2819                          const char *other,
2820                          struct diff_filespec *one,
2821                          struct diff_filespec *two,
2822                          struct diff_options *o,
2823                          struct diff_filepair *p,
2824                          int *must_show_header,
2825                          int use_color)
2826{
2827        const char *set = diff_get_color(use_color, DIFF_METAINFO);
2828        const char *reset = diff_get_color(use_color, DIFF_RESET);
2829        struct strbuf *msgbuf;
2830        char *line_prefix = "";
2831
2832        *must_show_header = 1;
2833        if (o->output_prefix) {
2834                msgbuf = o->output_prefix(o, o->output_prefix_data);
2835                line_prefix = msgbuf->buf;
2836        }
2837        strbuf_init(msg, PATH_MAX * 2 + 300);
2838        switch (p->status) {
2839        case DIFF_STATUS_COPIED:
2840                strbuf_addf(msg, "%s%ssimilarity index %d%%",
2841                            line_prefix, set, similarity_index(p));
2842                strbuf_addf(msg, "%s\n%s%scopy from ",
2843                            reset,  line_prefix, set);
2844                quote_c_style(name, msg, NULL, 0);
2845                strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2846                quote_c_style(other, msg, NULL, 0);
2847                strbuf_addf(msg, "%s\n", reset);
2848                break;
2849        case DIFF_STATUS_RENAMED:
2850                strbuf_addf(msg, "%s%ssimilarity index %d%%",
2851                            line_prefix, set, similarity_index(p));
2852                strbuf_addf(msg, "%s\n%s%srename from ",
2853                            reset, line_prefix, set);
2854                quote_c_style(name, msg, NULL, 0);
2855                strbuf_addf(msg, "%s\n%s%srename to ",
2856                            reset, line_prefix, set);
2857                quote_c_style(other, msg, NULL, 0);
2858                strbuf_addf(msg, "%s\n", reset);
2859                break;
2860        case DIFF_STATUS_MODIFIED:
2861                if (p->score) {
2862                        strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2863                                    line_prefix,
2864                                    set, similarity_index(p), reset);
2865                        break;
2866                }
2867                /* fallthru */
2868        default:
2869                *must_show_header = 0;
2870        }
2871        if (one && two && hashcmp(one->sha1, two->sha1)) {
2872                int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2873
2874                if (DIFF_OPT_TST(o, BINARY)) {
2875                        mmfile_t mf;
2876                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2877                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2878                                abbrev = 40;
2879                }
2880                strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2881                            find_unique_abbrev(one->sha1, abbrev));
2882                strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2883                if (one->mode == two->mode)
2884                        strbuf_addf(msg, " %06o", one->mode);
2885                strbuf_addf(msg, "%s\n", reset);
2886        }
2887}
2888
2889static void run_diff_cmd(const char *pgm,
2890                         const char *name,
2891                         const char *other,
2892                         const char *attr_path,
2893                         struct diff_filespec *one,
2894                         struct diff_filespec *two,
2895                         struct strbuf *msg,
2896                         struct diff_options *o,
2897                         struct diff_filepair *p)
2898{
2899        const char *xfrm_msg = NULL;
2900        int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2901        int must_show_header = 0;
2902
2903        if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2904                pgm = NULL;
2905        else {
2906                struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2907                if (drv && drv->external)
2908                        pgm = drv->external;
2909        }
2910
2911        if (msg) {
2912                /*
2913                 * don't use colors when the header is intended for an
2914                 * external diff driver
2915                 */
2916                fill_metainfo(msg, name, other, one, two, o, p,
2917                              &must_show_header,
2918                              want_color(o->use_color) && !pgm);
2919                xfrm_msg = msg->len ? msg->buf : NULL;
2920        }
2921
2922        if (pgm) {
2923                run_external_diff(pgm, name, other, one, two, xfrm_msg,
2924                                  complete_rewrite);
2925                return;
2926        }
2927        if (one && two)
2928                builtin_diff(name, other ? other : name,
2929                             one, two, xfrm_msg, must_show_header,
2930                             o, complete_rewrite);
2931        else
2932                fprintf(o->file, "* Unmerged path %s\n", name);
2933}
2934
2935static void diff_fill_sha1_info(struct diff_filespec *one)
2936{
2937        if (DIFF_FILE_VALID(one)) {
2938                if (!one->sha1_valid) {
2939                        struct stat st;
2940                        if (!strcmp(one->path, "-")) {
2941                                hashcpy(one->sha1, null_sha1);
2942                                return;
2943                        }
2944                        if (lstat(one->path, &st) < 0)
2945                                die_errno("stat '%s'", one->path);
2946                        if (index_path(one->sha1, one->path, &st, 0))
2947                                die("cannot hash %s", one->path);
2948                }
2949        }
2950        else
2951                hashclr(one->sha1);
2952}
2953
2954static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2955{
2956        /* Strip the prefix but do not molest /dev/null and absolute paths */
2957        if (*namep && **namep != '/') {
2958                *namep += prefix_length;
2959                if (**namep == '/')
2960                        ++*namep;
2961        }
2962        if (*otherp && **otherp != '/') {
2963                *otherp += prefix_length;
2964                if (**otherp == '/')
2965                        ++*otherp;
2966        }
2967}
2968
2969static void run_diff(struct diff_filepair *p, struct diff_options *o)
2970{
2971        const char *pgm = external_diff();
2972        struct strbuf msg;
2973        struct diff_filespec *one = p->one;
2974        struct diff_filespec *two = p->two;
2975        const char *name;
2976        const char *other;
2977        const char *attr_path;
2978
2979        name  = p->one->path;
2980        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2981        attr_path = name;
2982        if (o->prefix_length)
2983                strip_prefix(o->prefix_length, &name, &other);
2984
2985        if (DIFF_PAIR_UNMERGED(p)) {
2986                run_diff_cmd(pgm, name, NULL, attr_path,
2987                             NULL, NULL, NULL, o, p);
2988                return;
2989        }
2990
2991        diff_fill_sha1_info(one);
2992        diff_fill_sha1_info(two);
2993
2994        if (!pgm &&
2995            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2996            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2997                /*
2998                 * a filepair that changes between file and symlink
2999                 * needs to be split into deletion and creation.
3000                 */
3001                struct diff_filespec *null = alloc_filespec(two->path);
3002                run_diff_cmd(NULL, name, other, attr_path,
3003                             one, null, &msg, o, p);
3004                free(null);
3005                strbuf_release(&msg);
3006
3007                null = alloc_filespec(one->path);
3008                run_diff_cmd(NULL, name, other, attr_path,
3009                             null, two, &msg, o, p);
3010                free(null);
3011        }
3012        else
3013                run_diff_cmd(pgm, name, other, attr_path,
3014                             one, two, &msg, o, p);
3015
3016        strbuf_release(&msg);
3017}
3018
3019static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3020                         struct diffstat_t *diffstat)
3021{
3022        const char *name;
3023        const char *other;
3024        int complete_rewrite = 0;
3025
3026        if (DIFF_PAIR_UNMERGED(p)) {
3027                /* unmerged */
3028                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
3029                return;
3030        }
3031
3032        name = p->one->path;
3033        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3034
3035        if (o->prefix_length)
3036                strip_prefix(o->prefix_length, &name, &other);
3037
3038        diff_fill_sha1_info(p->one);
3039        diff_fill_sha1_info(p->two);
3040
3041        if (p->status == DIFF_STATUS_MODIFIED && p->score)
3042                complete_rewrite = 1;
3043        builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
3044}
3045
3046static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3047{
3048        const char *name;
3049        const char *other;
3050        const char *attr_path;
3051
3052        if (DIFF_PAIR_UNMERGED(p)) {
3053                /* unmerged */
3054                return;
3055        }
3056
3057        name = p->one->path;
3058        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3059        attr_path = other ? other : name;
3060
3061        if (o->prefix_length)
3062                strip_prefix(o->prefix_length, &name, &other);
3063
3064        diff_fill_sha1_info(p->one);
3065        diff_fill_sha1_info(p->two);
3066
3067        builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3068}
3069
3070void diff_setup(struct diff_options *options)
3071{
3072        memcpy(options, &default_diff_options, sizeof(*options));
3073
3074        options->file = stdout;
3075
3076        options->line_termination = '\n';
3077        options->break_opt = -1;
3078        options->rename_limit = -1;
3079        options->dirstat_permille = diff_dirstat_permille_default;
3080        options->context = 3;
3081
3082        options->change = diff_change;
3083        options->add_remove = diff_addremove;
3084        options->use_color = diff_use_color_default;
3085        options->detect_rename = diff_detect_rename_default;
3086
3087        if (diff_no_prefix) {
3088                options->a_prefix = options->b_prefix = "";
3089        } else if (!diff_mnemonic_prefix) {
3090                options->a_prefix = "a/";
3091                options->b_prefix = "b/";
3092        }
3093}
3094
3095int diff_setup_done(struct diff_options *options)
3096{
3097        int count = 0;
3098
3099        if (options->output_format & DIFF_FORMAT_NAME)
3100                count++;
3101        if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3102                count++;
3103        if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3104                count++;
3105        if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3106                count++;
3107        if (count > 1)
3108                die("--name-only, --name-status, --check and -s are mutually exclusive");
3109
3110        /*
3111         * Most of the time we can say "there are changes"
3112         * only by checking if there are changed paths, but
3113         * --ignore-whitespace* options force us to look
3114         * inside contents.
3115         */
3116
3117        if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3118            DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3119            DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3120                DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3121        else
3122                DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3123
3124        if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3125                options->detect_rename = DIFF_DETECT_COPY;
3126
3127        if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3128                options->prefix = NULL;
3129        if (options->prefix)
3130                options->prefix_length = strlen(options->prefix);
3131        else
3132                options->prefix_length = 0;
3133
3134        if (options->output_format & (DIFF_FORMAT_NAME |
3135                                      DIFF_FORMAT_NAME_STATUS |
3136                                      DIFF_FORMAT_CHECKDIFF |
3137                                      DIFF_FORMAT_NO_OUTPUT))
3138                options->output_format &= ~(DIFF_FORMAT_RAW |
3139                                            DIFF_FORMAT_NUMSTAT |
3140                                            DIFF_FORMAT_DIFFSTAT |
3141                                            DIFF_FORMAT_SHORTSTAT |
3142                                            DIFF_FORMAT_DIRSTAT |
3143                                            DIFF_FORMAT_SUMMARY |
3144                                            DIFF_FORMAT_PATCH);
3145
3146        /*
3147         * These cases always need recursive; we do not drop caller-supplied
3148         * recursive bits for other formats here.
3149         */
3150        if (options->output_format & (DIFF_FORMAT_PATCH |
3151                                      DIFF_FORMAT_NUMSTAT |
3152                                      DIFF_FORMAT_DIFFSTAT |
3153                                      DIFF_FORMAT_SHORTSTAT |
3154                                      DIFF_FORMAT_DIRSTAT |
3155                                      DIFF_FORMAT_SUMMARY |
3156                                      DIFF_FORMAT_CHECKDIFF))
3157                DIFF_OPT_SET(options, RECURSIVE);
3158        /*
3159         * Also pickaxe would not work very well if you do not say recursive
3160         */
3161        if (options->pickaxe)
3162                DIFF_OPT_SET(options, RECURSIVE);
3163        /*
3164         * When patches are generated, submodules diffed against the work tree
3165         * must be checked for dirtiness too so it can be shown in the output
3166         */
3167        if (options->output_format & DIFF_FORMAT_PATCH)
3168                DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3169
3170        if (options->detect_rename && options->rename_limit < 0)
3171                options->rename_limit = diff_rename_limit_default;
3172        if (options->setup & DIFF_SETUP_USE_CACHE) {
3173                if (!active_cache)
3174                        /* read-cache does not die even when it fails
3175                         * so it is safe for us to do this here.  Also
3176                         * it does not smudge active_cache or active_nr
3177                         * when it fails, so we do not have to worry about
3178                         * cleaning it up ourselves either.
3179                         */
3180                        read_cache();
3181        }
3182        if (options->abbrev <= 0 || 40 < options->abbrev)
3183                options->abbrev = 40; /* full */
3184
3185        /*
3186         * It does not make sense to show the first hit we happened
3187         * to have found.  It does not make sense not to return with
3188         * exit code in such a case either.
3189         */
3190        if (DIFF_OPT_TST(options, QUICK)) {
3191                options->output_format = DIFF_FORMAT_NO_OUTPUT;
3192                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3193        }
3194
3195        return 0;
3196}
3197
3198static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3199{
3200        char c, *eq;
3201        int len;
3202
3203        if (*arg != '-')
3204                return 0;
3205        c = *++arg;
3206        if (!c)
3207                return 0;
3208        if (c == arg_short) {
3209                c = *++arg;
3210                if (!c)
3211                        return 1;
3212                if (val && isdigit(c)) {
3213                        char *end;
3214                        int n = strtoul(arg, &end, 10);
3215                        if (*end)
3216                                return 0;
3217                        *val = n;
3218                        return 1;
3219                }
3220                return 0;
3221        }
3222        if (c != '-')
3223                return 0;
3224        arg++;
3225        eq = strchr(arg, '=');
3226        if (eq)
3227                len = eq - arg;
3228        else
3229                len = strlen(arg);
3230        if (!len || strncmp(arg, arg_long, len))
3231                return 0;
3232        if (eq) {
3233                int n;
3234                char *end;
3235                if (!isdigit(*++eq))
3236                        return 0;
3237                n = strtoul(eq, &end, 10);
3238                if (*end)
3239                        return 0;
3240                *val = n;
3241        }
3242        return 1;
3243}
3244
3245static int diff_scoreopt_parse(const char *opt);
3246
3247static inline int short_opt(char opt, const char **argv,
3248                            const char **optarg)
3249{
3250        const char *arg = argv[0];
3251        if (arg[0] != '-' || arg[1] != opt)
3252                return 0;
3253        if (arg[2] != '\0') {
3254                *optarg = arg + 2;
3255                return 1;
3256        }
3257        if (!argv[1])
3258                die("Option '%c' requires a value", opt);
3259        *optarg = argv[1];
3260        return 2;
3261}
3262
3263int parse_long_opt(const char *opt, const char **argv,
3264                   const char **optarg)
3265{
3266        const char *arg = argv[0];
3267        if (arg[0] != '-' || arg[1] != '-')
3268                return 0;
3269        arg += strlen("--");
3270        if (prefixcmp(arg, opt))
3271                return 0;
3272        arg += strlen(opt);
3273        if (*arg == '=') { /* sticked form: --option=value */
3274                *optarg = arg + 1;
3275                return 1;
3276        }
3277        if (*arg != '\0')
3278                return 0;
3279        /* separate form: --option value */
3280        if (!argv[1])
3281                die("Option '--%s' requires a value", opt);
3282        *optarg = argv[1];
3283        return 2;
3284}
3285
3286static int stat_opt(struct diff_options *options, const char **av)
3287{
3288        const char *arg = av[0];
3289        char *end;
3290        int width = options->stat_width;
3291        int name_width = options->stat_name_width;
3292        int count = options->stat_count;
3293        int argcount = 1;
3294
3295        arg += strlen("--stat");
3296        end = (char *)arg;
3297
3298        switch (*arg) {
3299        case '-':
3300                if (!prefixcmp(arg, "-width")) {
3301                        arg += strlen("-width");
3302                        if (*arg == '=')
3303                                width = strtoul(arg + 1, &end, 10);
3304                        else if (!*arg && !av[1])
3305                                die("Option '--stat-width' requires a value");
3306                        else if (!*arg) {
3307                                width = strtoul(av[1], &end, 10);
3308                                argcount = 2;
3309                        }
3310                } else if (!prefixcmp(arg, "-name-width")) {
3311                        arg += strlen("-name-width");
3312                        if (*arg == '=')
3313                                name_width = strtoul(arg + 1, &end, 10);
3314                        else if (!*arg && !av[1])
3315                                die("Option '--stat-name-width' requires a value");
3316                        else if (!*arg) {
3317                                name_width = strtoul(av[1], &end, 10);
3318                                argcount = 2;
3319                        }
3320                } else if (!prefixcmp(arg, "-count")) {
3321                        arg += strlen("-count");
3322                        if (*arg == '=')
3323                                count = strtoul(arg + 1, &end, 10);
3324                        else if (!*arg && !av[1])
3325                                die("Option '--stat-count' requires a value");
3326                        else if (!*arg) {
3327                                count = strtoul(av[1], &end, 10);
3328                                argcount = 2;
3329                        }
3330                }
3331                break;
3332        case '=':
3333                width = strtoul(arg+1, &end, 10);
3334                if (*end == ',')
3335                        name_width = strtoul(end+1, &end, 10);
3336                if (*end == ',')
3337                        count = strtoul(end+1, &end, 10);
3338        }
3339
3340        /* Important! This checks all the error cases! */
3341        if (*end)
3342                return 0;
3343        options->output_format |= DIFF_FORMAT_DIFFSTAT;
3344        options->stat_name_width = name_width;
3345        options->stat_width = width;
3346        options->stat_count = count;
3347        return argcount;
3348}
3349
3350static int parse_dirstat_opt(struct diff_options *options, const char *params)
3351{
3352        struct strbuf errmsg = STRBUF_INIT;
3353        if (parse_dirstat_params(options, params, &errmsg))
3354                die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3355                    errmsg.buf);
3356        strbuf_release(&errmsg);
3357        /*
3358         * The caller knows a dirstat-related option is given from the command
3359         * line; allow it to say "return this_function();"
3360         */
3361        options->output_format |= DIFF_FORMAT_DIRSTAT;
3362        return 1;
3363}
3364
3365int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3366{
3367        const char *arg = av[0];
3368        const char *optarg;
3369        int argcount;
3370
3371        /* Output format options */
3372        if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3373                options->output_format |= DIFF_FORMAT_PATCH;
3374        else if (opt_arg(arg, 'U', "unified", &options->context))
3375                options->output_format |= DIFF_FORMAT_PATCH;
3376        else if (!strcmp(arg, "--raw"))
3377                options->output_format |= DIFF_FORMAT_RAW;
3378        else if (!strcmp(arg, "--patch-with-raw"))
3379                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3380        else if (!strcmp(arg, "--numstat"))
3381                options->output_format |= DIFF_FORMAT_NUMSTAT;
3382        else if (!strcmp(arg, "--shortstat"))
3383                options->output_format |= DIFF_FORMAT_SHORTSTAT;
3384        else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3385                return parse_dirstat_opt(options, "");
3386        else if (!prefixcmp(arg, "-X"))
3387                return parse_dirstat_opt(options, arg + 2);
3388        else if (!prefixcmp(arg, "--dirstat="))
3389                return parse_dirstat_opt(options, arg + 10);
3390        else if (!strcmp(arg, "--cumulative"))
3391                return parse_dirstat_opt(options, "cumulative");
3392        else if (!strcmp(arg, "--dirstat-by-file"))
3393                return parse_dirstat_opt(options, "files");
3394        else if (!prefixcmp(arg, "--dirstat-by-file=")) {
3395                parse_dirstat_opt(options, "files");
3396                return parse_dirstat_opt(options, arg + 18);
3397        }
3398        else if (!strcmp(arg, "--check"))
3399                options->output_format |= DIFF_FORMAT_CHECKDIFF;
3400        else if (!strcmp(arg, "--summary"))
3401                options->output_format |= DIFF_FORMAT_SUMMARY;
3402        else if (!strcmp(arg, "--patch-with-stat"))
3403                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3404        else if (!strcmp(arg, "--name-only"))
3405                options->output_format |= DIFF_FORMAT_NAME;
3406        else if (!strcmp(arg, "--name-status"))
3407                options->output_format |= DIFF_FORMAT_NAME_STATUS;
3408        else if (!strcmp(arg, "-s"))
3409                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3410        else if (!prefixcmp(arg, "--stat"))
3411                /* --stat, --stat-width, --stat-name-width, or --stat-count */
3412                return stat_opt(options, av);
3413
3414        /* renames options */
3415        else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3416                 !strcmp(arg, "--break-rewrites")) {
3417                if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3418                        return error("invalid argument to -B: %s", arg+2);
3419        }
3420        else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3421                 !strcmp(arg, "--find-renames")) {
3422                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3423                        return error("invalid argument to -M: %s", arg+2);
3424                options->detect_rename = DIFF_DETECT_RENAME;
3425        }
3426        else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3427                options->irreversible_delete = 1;
3428        }
3429        else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3430                 !strcmp(arg, "--find-copies")) {
3431                if (options->detect_rename == DIFF_DETECT_COPY)
3432                        DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3433                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3434                        return error("invalid argument to -C: %s", arg+2);
3435                options->detect_rename = DIFF_DETECT_COPY;
3436        }
3437        else if (!strcmp(arg, "--no-renames"))
3438                options->detect_rename = 0;
3439        else if (!strcmp(arg, "--relative"))
3440                DIFF_OPT_SET(options, RELATIVE_NAME);
3441        else if (!prefixcmp(arg, "--relative=")) {
3442                DIFF_OPT_SET(options, RELATIVE_NAME);
3443                options->prefix = arg + 11;
3444        }
3445
3446        /* xdiff options */
3447        else if (!strcmp(arg, "--minimal"))
3448                DIFF_XDL_SET(options, NEED_MINIMAL);
3449        else if (!strcmp(arg, "--no-minimal"))
3450                DIFF_XDL_CLR(options, NEED_MINIMAL);
3451        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3452                DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3453        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3454                DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3455        else if (!strcmp(arg, "--ignore-space-at-eol"))
3456                DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3457        else if (!strcmp(arg, "--patience"))
3458                DIFF_XDL_SET(options, PATIENCE_DIFF);
3459        else if (!strcmp(arg, "--histogram"))
3460                DIFF_XDL_SET(options, HISTOGRAM_DIFF);
3461
3462        /* flags options */
3463        else if (!strcmp(arg, "--binary")) {
3464                options->output_format |= DIFF_FORMAT_PATCH;
3465                DIFF_OPT_SET(options, BINARY);
3466        }
3467        else if (!strcmp(arg, "--full-index"))
3468                DIFF_OPT_SET(options, FULL_INDEX);
3469        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3470                DIFF_OPT_SET(options, TEXT);
3471        else if (!strcmp(arg, "-R"))
3472                DIFF_OPT_SET(options, REVERSE_DIFF);
3473        else if (!strcmp(arg, "--find-copies-harder"))
3474                DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3475        else if (!strcmp(arg, "--follow"))
3476                DIFF_OPT_SET(options, FOLLOW_RENAMES);
3477        else if (!strcmp(arg, "--color"))
3478                options->use_color = 1;
3479        else if (!prefixcmp(arg, "--color=")) {
3480                int value = git_config_colorbool(NULL, arg+8);
3481                if (value < 0)
3482                        return error("option `color' expects \"always\", \"auto\", or \"never\"");
3483                options->use_color = value;
3484        }
3485        else if (!strcmp(arg, "--no-color"))
3486                options->use_color = 0;
3487        else if (!strcmp(arg, "--color-words")) {
3488                options->use_color = 1;
3489                options->word_diff = DIFF_WORDS_COLOR;
3490        }
3491        else if (!prefixcmp(arg, "--color-words=")) {
3492                options->use_color = 1;
3493                options->word_diff = DIFF_WORDS_COLOR;
3494                options->word_regex = arg + 14;
3495        }
3496        else if (!strcmp(arg, "--word-diff")) {
3497                if (options->word_diff == DIFF_WORDS_NONE)
3498                        options->word_diff = DIFF_WORDS_PLAIN;
3499        }
3500        else if (!prefixcmp(arg, "--word-diff=")) {
3501                const char *type = arg + 12;
3502                if (!strcmp(type, "plain"))
3503                        options->word_diff = DIFF_WORDS_PLAIN;
3504                else if (!strcmp(type, "color")) {
3505                        options->use_color = 1;
3506                        options->word_diff = DIFF_WORDS_COLOR;
3507                }
3508                else if (!strcmp(type, "porcelain"))
3509                        options->word_diff = DIFF_WORDS_PORCELAIN;
3510                else if (!strcmp(type, "none"))
3511                        options->word_diff = DIFF_WORDS_NONE;
3512                else
3513                        die("bad --word-diff argument: %s", type);
3514        }
3515        else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3516                if (options->word_diff == DIFF_WORDS_NONE)
3517                        options->word_diff = DIFF_WORDS_PLAIN;
3518                options->word_regex = optarg;
3519                return argcount;
3520        }
3521        else if (!strcmp(arg, "--exit-code"))
3522                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3523        else if (!strcmp(arg, "--quiet"))
3524                DIFF_OPT_SET(options, QUICK);
3525        else if (!strcmp(arg, "--ext-diff"))
3526                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3527        else if (!strcmp(arg, "--no-ext-diff"))
3528                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3529        else if (!strcmp(arg, "--textconv"))
3530                DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3531        else if (!strcmp(arg, "--no-textconv"))
3532                DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3533        else if (!strcmp(arg, "--ignore-submodules")) {
3534                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3535                handle_ignore_submodules_arg(options, "all");
3536        } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3537                DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3538                handle_ignore_submodules_arg(options, arg + 20);
3539        } else if (!strcmp(arg, "--submodule"))
3540                DIFF_OPT_SET(options, SUBMODULE_LOG);
3541        else if (!prefixcmp(arg, "--submodule=")) {
3542                if (!strcmp(arg + 12, "log"))
3543                        DIFF_OPT_SET(options, SUBMODULE_LOG);
3544        }
3545
3546        /* misc options */
3547        else if (!strcmp(arg, "-z"))
3548                options->line_termination = 0;
3549        else if ((argcount = short_opt('l', av, &optarg))) {
3550                options->rename_limit = strtoul(optarg, NULL, 10);
3551                return argcount;
3552        }
3553        else if ((argcount = short_opt('S', av, &optarg))) {
3554                options->pickaxe = optarg;
3555                options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3556                return argcount;
3557        } else if ((argcount = short_opt('G', av, &optarg))) {
3558                options->pickaxe = optarg;
3559                options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3560                return argcount;
3561        }
3562        else if (!strcmp(arg, "--pickaxe-all"))
3563                options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3564        else if (!strcmp(arg, "--pickaxe-regex"))
3565                options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3566        else if ((argcount = short_opt('O', av, &optarg))) {
3567                options->orderfile = optarg;
3568                return argcount;
3569        }
3570        else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3571                options->filter = optarg;
3572                return argcount;
3573        }
3574        else if (!strcmp(arg, "--abbrev"))
3575                options->abbrev = DEFAULT_ABBREV;
3576        else if (!prefixcmp(arg, "--abbrev=")) {
3577                options->abbrev = strtoul(arg + 9, NULL, 10);
3578                if (options->abbrev < MINIMUM_ABBREV)
3579                        options->abbrev = MINIMUM_ABBREV;
3580                else if (40 < options->abbrev)
3581                        options->abbrev = 40;
3582        }
3583        else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3584                options->a_prefix = optarg;
3585                return argcount;
3586        }
3587        else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3588                options->b_prefix = optarg;
3589                return argcount;
3590        }
3591        else if (!strcmp(arg, "--no-prefix"))
3592                options->a_prefix = options->b_prefix = "";
3593        else if (opt_arg(arg, '\0', "inter-hunk-context",
3594                         &options->interhunkcontext))
3595                ;
3596        else if (!strcmp(arg, "-W"))
3597                DIFF_OPT_SET(options, FUNCCONTEXT);
3598        else if (!strcmp(arg, "--function-context"))
3599                DIFF_OPT_SET(options, FUNCCONTEXT);
3600        else if (!strcmp(arg, "--no-function-context"))
3601                DIFF_OPT_CLR(options, FUNCCONTEXT);
3602        else if ((argcount = parse_long_opt("output", av, &optarg))) {
3603                options->file = fopen(optarg, "w");
3604                if (!options->file)
3605                        die_errno("Could not open '%s'", optarg);
3606                options->close_file = 1;
3607                return argcount;
3608        } else
3609                return 0;
3610        return 1;
3611}
3612
3613int parse_rename_score(const char **cp_p)
3614{
3615        unsigned long num, scale;
3616        int ch, dot;
3617        const char *cp = *cp_p;
3618
3619        num = 0;
3620        scale = 1;
3621        dot = 0;
3622        for (;;) {
3623                ch = *cp;
3624                if ( !dot && ch == '.' ) {
3625                        scale = 1;
3626                        dot = 1;
3627                } else if ( ch == '%' ) {
3628                        scale = dot ? scale*100 : 100;
3629                        cp++;   /* % is always at the end */
3630                        break;
3631                } else if ( ch >= '0' && ch <= '9' ) {
3632                        if ( scale < 100000 ) {
3633                                scale *= 10;
3634                                num = (num*10) + (ch-'0');
3635                        }
3636                } else {
3637                        break;
3638                }
3639                cp++;
3640        }
3641        *cp_p = cp;
3642
3643        /* user says num divided by scale and we say internally that
3644         * is MAX_SCORE * num / scale.
3645         */
3646        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3647}
3648
3649static int diff_scoreopt_parse(const char *opt)
3650{
3651        int opt1, opt2, cmd;
3652
3653        if (*opt++ != '-')
3654                return -1;
3655        cmd = *opt++;
3656        if (cmd == '-') {
3657                /* convert the long-form arguments into short-form versions */
3658                if (!prefixcmp(opt, "break-rewrites")) {
3659                        opt += strlen("break-rewrites");
3660                        if (*opt == 0 || *opt++ == '=')
3661                                cmd = 'B';
3662                } else if (!prefixcmp(opt, "find-copies")) {
3663                        opt += strlen("find-copies");
3664                        if (*opt == 0 || *opt++ == '=')
3665                                cmd = 'C';
3666                } else if (!prefixcmp(opt, "find-renames")) {
3667                        opt += strlen("find-renames");
3668                        if (*opt == 0 || *opt++ == '=')
3669                                cmd = 'M';
3670                }
3671        }
3672        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3673                return -1; /* that is not a -M, -C nor -B option */
3674
3675        opt1 = parse_rename_score(&opt);
3676        if (cmd != 'B')
3677                opt2 = 0;
3678        else {
3679                if (*opt == 0)
3680                        opt2 = 0;
3681                else if (*opt != '/')
3682                        return -1; /* we expect -B80/99 or -B80 */
3683                else {
3684                        opt++;
3685                        opt2 = parse_rename_score(&opt);
3686                }
3687        }
3688        if (*opt != 0)
3689                return -1;
3690        return opt1 | (opt2 << 16);
3691}
3692
3693struct diff_queue_struct diff_queued_diff;
3694
3695void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3696{
3697        if (queue->alloc <= queue->nr) {
3698                queue->alloc = alloc_nr(queue->alloc);
3699                queue->queue = xrealloc(queue->queue,
3700                                        sizeof(dp) * queue->alloc);
3701        }
3702        queue->queue[queue->nr++] = dp;
3703}
3704
3705struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3706                                 struct diff_filespec *one,
3707                                 struct diff_filespec *two)
3708{
3709        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3710        dp->one = one;
3711        dp->two = two;
3712        if (queue)
3713                diff_q(queue, dp);
3714        return dp;
3715}
3716
3717void diff_free_filepair(struct diff_filepair *p)
3718{
3719        free_filespec(p->one);
3720        free_filespec(p->two);
3721        free(p);
3722}
3723
3724/* This is different from find_unique_abbrev() in that
3725 * it stuffs the result with dots for alignment.
3726 */
3727const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3728{
3729        int abblen;
3730        const char *abbrev;
3731        if (len == 40)
3732                return sha1_to_hex(sha1);
3733
3734        abbrev = find_unique_abbrev(sha1, len);
3735        abblen = strlen(abbrev);
3736        if (abblen < 37) {
3737                static char hex[41];
3738                if (len < abblen && abblen <= len + 2)
3739                        sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3740                else
3741                        sprintf(hex, "%s...", abbrev);
3742                return hex;
3743        }
3744        return sha1_to_hex(sha1);
3745}
3746
3747static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3748{
3749        int line_termination = opt->line_termination;
3750        int inter_name_termination = line_termination ? '\t' : '\0';
3751        if (opt->output_prefix) {
3752                struct strbuf *msg = NULL;
3753                msg = opt->output_prefix(opt, opt->output_prefix_data);
3754                fprintf(opt->file, "%s", msg->buf);
3755        }
3756
3757        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3758                fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3759                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
3760                fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3761        }
3762        if (p->score) {
3763                fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3764                        inter_name_termination);
3765        } else {
3766                fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3767        }
3768
3769        if (p->status == DIFF_STATUS_COPIED ||
3770            p->status == DIFF_STATUS_RENAMED) {
3771                const char *name_a, *name_b;
3772                name_a = p->one->path;
3773                name_b = p->two->path;
3774                strip_prefix(opt->prefix_length, &name_a, &name_b);
3775                write_name_quoted(name_a, opt->file, inter_name_termination);
3776                write_name_quoted(name_b, opt->file, line_termination);
3777        } else {
3778                const char *name_a, *name_b;
3779                name_a = p->one->mode ? p->one->path : p->two->path;
3780                name_b = NULL;
3781                strip_prefix(opt->prefix_length, &name_a, &name_b);
3782                write_name_quoted(name_a, opt->file, line_termination);
3783        }
3784}
3785
3786int diff_unmodified_pair(struct diff_filepair *p)
3787{
3788        /* This function is written stricter than necessary to support
3789         * the currently implemented transformers, but the idea is to
3790         * let transformers to produce diff_filepairs any way they want,
3791         * and filter and clean them up here before producing the output.
3792         */
3793        struct diff_filespec *one = p->one, *two = p->two;
3794
3795        if (DIFF_PAIR_UNMERGED(p))
3796                return 0; /* unmerged is interesting */
3797
3798        /* deletion, addition, mode or type change
3799         * and rename are all interesting.
3800         */
3801        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3802            DIFF_PAIR_MODE_CHANGED(p) ||
3803            strcmp(one->path, two->path))
3804                return 0;
3805
3806        /* both are valid and point at the same path.  that is, we are
3807         * dealing with a change.
3808         */
3809        if (one->sha1_valid && two->sha1_valid &&
3810            !hashcmp(one->sha1, two->sha1) &&
3811            !one->dirty_submodule && !two->dirty_submodule)
3812                return 1; /* no change */
3813        if (!one->sha1_valid && !two->sha1_valid)
3814                return 1; /* both look at the same file on the filesystem. */
3815        return 0;
3816}
3817
3818static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3819{
3820        if (diff_unmodified_pair(p))
3821                return;
3822
3823        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3824            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3825                return; /* no tree diffs in patch format */
3826
3827        run_diff(p, o);
3828}
3829
3830static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3831                            struct diffstat_t *diffstat)
3832{
3833        if (diff_unmodified_pair(p))
3834                return;
3835
3836        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3837            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3838                return; /* no useful stat for tree diffs */
3839
3840        run_diffstat(p, o, diffstat);
3841}
3842
3843static void diff_flush_checkdiff(struct diff_filepair *p,
3844                struct diff_options *o)
3845{
3846        if (diff_unmodified_pair(p))
3847                return;
3848
3849        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3850            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3851                return; /* nothing to check in tree diffs */
3852
3853        run_checkdiff(p, o);
3854}
3855
3856int diff_queue_is_empty(void)
3857{
3858        struct diff_queue_struct *q = &diff_queued_diff;
3859        int i;
3860        for (i = 0; i < q->nr; i++)
3861                if (!diff_unmodified_pair(q->queue[i]))
3862                        return 0;
3863        return 1;
3864}
3865
3866#if DIFF_DEBUG
3867void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3868{
3869        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3870                x, one ? one : "",
3871                s->path,
3872                DIFF_FILE_VALID(s) ? "valid" : "invalid",
3873                s->mode,
3874                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3875        fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3876                x, one ? one : "",
3877                s->size, s->xfrm_flags);
3878}
3879
3880void diff_debug_filepair(const struct diff_filepair *p, int i)
3881{
3882        diff_debug_filespec(p->one, i, "one");
3883        diff_debug_filespec(p->two, i, "two");
3884        fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3885                p->score, p->status ? p->status : '?',
3886                p->one->rename_used, p->broken_pair);
3887}
3888
3889void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3890{
3891        int i;
3892        if (msg)
3893                fprintf(stderr, "%s\n", msg);
3894        fprintf(stderr, "q->nr = %d\n", q->nr);
3895        for (i = 0; i < q->nr; i++) {
3896                struct diff_filepair *p = q->queue[i];
3897                diff_debug_filepair(p, i);
3898        }
3899}
3900#endif
3901
3902static void diff_resolve_rename_copy(void)
3903{
3904        int i;
3905        struct diff_filepair *p;
3906        struct diff_queue_struct *q = &diff_queued_diff;
3907
3908        diff_debug_queue("resolve-rename-copy", q);
3909
3910        for (i = 0; i < q->nr; i++) {
3911                p = q->queue[i];
3912                p->status = 0; /* undecided */
3913                if (DIFF_PAIR_UNMERGED(p))
3914                        p->status = DIFF_STATUS_UNMERGED;
3915                else if (!DIFF_FILE_VALID(p->one))
3916                        p->status = DIFF_STATUS_ADDED;
3917                else if (!DIFF_FILE_VALID(p->two))
3918                        p->status = DIFF_STATUS_DELETED;
3919                else if (DIFF_PAIR_TYPE_CHANGED(p))
3920                        p->status = DIFF_STATUS_TYPE_CHANGED;
3921
3922                /* from this point on, we are dealing with a pair
3923                 * whose both sides are valid and of the same type, i.e.
3924                 * either in-place edit or rename/copy edit.
3925                 */
3926                else if (DIFF_PAIR_RENAME(p)) {
3927                        /*
3928                         * A rename might have re-connected a broken
3929                         * pair up, causing the pathnames to be the
3930                         * same again. If so, that's not a rename at
3931                         * all, just a modification..
3932                         *
3933                         * Otherwise, see if this source was used for
3934                         * multiple renames, in which case we decrement
3935                         * the count, and call it a copy.
3936                         */
3937                        if (!strcmp(p->one->path, p->two->path))
3938                                p->status = DIFF_STATUS_MODIFIED;
3939                        else if (--p->one->rename_used > 0)
3940                                p->status = DIFF_STATUS_COPIED;
3941                        else
3942                                p->status = DIFF_STATUS_RENAMED;
3943                }
3944                else if (hashcmp(p->one->sha1, p->two->sha1) ||
3945                         p->one->mode != p->two->mode ||
3946                         p->one->dirty_submodule ||
3947                         p->two->dirty_submodule ||
3948                         is_null_sha1(p->one->sha1))
3949                        p->status = DIFF_STATUS_MODIFIED;
3950                else {
3951                        /* This is a "no-change" entry and should not
3952                         * happen anymore, but prepare for broken callers.
3953                         */
3954                        error("feeding unmodified %s to diffcore",
3955                              p->one->path);
3956                        p->status = DIFF_STATUS_UNKNOWN;
3957                }
3958        }
3959        diff_debug_queue("resolve-rename-copy done", q);
3960}
3961
3962static int check_pair_status(struct diff_filepair *p)
3963{
3964        switch (p->status) {
3965        case DIFF_STATUS_UNKNOWN:
3966                return 0;
3967        case 0:
3968                die("internal error in diff-resolve-rename-copy");
3969        default:
3970                return 1;
3971        }
3972}
3973
3974static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3975{
3976        int fmt = opt->output_format;
3977
3978        if (fmt & DIFF_FORMAT_CHECKDIFF)
3979                diff_flush_checkdiff(p, opt);
3980        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3981                diff_flush_raw(p, opt);
3982        else if (fmt & DIFF_FORMAT_NAME) {
3983                const char *name_a, *name_b;
3984                name_a = p->two->path;
3985                name_b = NULL;
3986                strip_prefix(opt->prefix_length, &name_a, &name_b);
3987                write_name_quoted(name_a, opt->file, opt->line_termination);
3988        }
3989}
3990
3991static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3992{
3993        if (fs->mode)
3994                fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3995        else
3996                fprintf(file, " %s ", newdelete);
3997        write_name_quoted(fs->path, file, '\n');
3998}
3999
4000
4001static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4002                const char *line_prefix)
4003{
4004        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4005                fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4006                        p->two->mode, show_name ? ' ' : '\n');
4007                if (show_name) {
4008                        write_name_quoted(p->two->path, file, '\n');
4009                }
4010        }
4011}
4012
4013static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4014                        const char *line_prefix)
4015{
4016        char *names = pprint_rename(p->one->path, p->two->path);
4017
4018        fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4019        free(names);
4020        show_mode_change(file, p, 0, line_prefix);
4021}
4022
4023static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4024{
4025        FILE *file = opt->file;
4026        char *line_prefix = "";
4027
4028        if (opt->output_prefix) {
4029                struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
4030                line_prefix = buf->buf;
4031        }
4032
4033        switch(p->status) {
4034        case DIFF_STATUS_DELETED:
4035                fputs(line_prefix, file);
4036                show_file_mode_name(file, "delete", p->one);
4037                break;
4038        case DIFF_STATUS_ADDED:
4039                fputs(line_prefix, file);
4040                show_file_mode_name(file, "create", p->two);
4041                break;
4042        case DIFF_STATUS_COPIED:
4043                fputs(line_prefix, file);
4044                show_rename_copy(file, "copy", p, line_prefix);
4045                break;
4046        case DIFF_STATUS_RENAMED:
4047                fputs(line_prefix, file);
4048                show_rename_copy(file, "rename", p, line_prefix);
4049                break;
4050        default:
4051                if (p->score) {
4052                        fprintf(file, "%s rewrite ", line_prefix);
4053                        write_name_quoted(p->two->path, file, ' ');
4054                        fprintf(file, "(%d%%)\n", similarity_index(p));
4055                }
4056                show_mode_change(file, p, !p->score, line_prefix);
4057                break;
4058        }
4059}
4060
4061struct patch_id_t {
4062        git_SHA_CTX *ctx;
4063        int patchlen;
4064};
4065
4066static int remove_space(char *line, int len)
4067{
4068        int i;
4069        char *dst = line;
4070        unsigned char c;
4071
4072        for (i = 0; i < len; i++)
4073                if (!isspace((c = line[i])))
4074                        *dst++ = c;
4075
4076        return dst - line;
4077}
4078
4079static void patch_id_consume(void *priv, char *line, unsigned long len)
4080{
4081        struct patch_id_t *data = priv;
4082        int new_len;
4083
4084        /* Ignore line numbers when computing the SHA1 of the patch */
4085        if (!prefixcmp(line, "@@ -"))
4086                return;
4087
4088        new_len = remove_space(line, len);
4089
4090        git_SHA1_Update(data->ctx, line, new_len);
4091        data->patchlen += new_len;
4092}
4093
4094/* returns 0 upon success, and writes result into sha1 */
4095static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4096{
4097        struct diff_queue_struct *q = &diff_queued_diff;
4098        int i;
4099        git_SHA_CTX ctx;
4100        struct patch_id_t data;
4101        char buffer[PATH_MAX * 4 + 20];
4102
4103        git_SHA1_Init(&ctx);
4104        memset(&data, 0, sizeof(struct patch_id_t));
4105        data.ctx = &ctx;
4106
4107        for (i = 0; i < q->nr; i++) {
4108                xpparam_t xpp;
4109                xdemitconf_t xecfg;
4110                mmfile_t mf1, mf2;
4111                struct diff_filepair *p = q->queue[i];
4112                int len1, len2;
4113
4114                memset(&xpp, 0, sizeof(xpp));
4115                memset(&xecfg, 0, sizeof(xecfg));
4116                if (p->status == 0)
4117                        return error("internal diff status error");
4118                if (p->status == DIFF_STATUS_UNKNOWN)
4119                        continue;
4120                if (diff_unmodified_pair(p))
4121                        continue;
4122                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4123                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4124                        continue;
4125                if (DIFF_PAIR_UNMERGED(p))
4126                        continue;
4127
4128                diff_fill_sha1_info(p->one);
4129                diff_fill_sha1_info(p->two);
4130                if (fill_mmfile(&mf1, p->one) < 0 ||
4131                                fill_mmfile(&mf2, p->two) < 0)
4132                        return error("unable to read files to diff");
4133
4134                len1 = remove_space(p->one->path, strlen(p->one->path));
4135                len2 = remove_space(p->two->path, strlen(p->two->path));
4136                if (p->one->mode == 0)
4137                        len1 = snprintf(buffer, sizeof(buffer),
4138                                        "diff--gita/%.*sb/%.*s"
4139                                        "newfilemode%06o"
4140                                        "---/dev/null"
4141                                        "+++b/%.*s",
4142                                        len1, p->one->path,
4143                                        len2, p->two->path,
4144                                        p->two->mode,
4145                                        len2, p->two->path);
4146                else if (p->two->mode == 0)
4147                        len1 = snprintf(buffer, sizeof(buffer),
4148                                        "diff--gita/%.*sb/%.*s"
4149                                        "deletedfilemode%06o"
4150                                        "---a/%.*s"
4151                                        "+++/dev/null",
4152                                        len1, p->one->path,
4153                                        len2, p->two->path,
4154                                        p->one->mode,
4155                                        len1, p->one->path);
4156                else
4157                        len1 = snprintf(buffer, sizeof(buffer),
4158                                        "diff--gita/%.*sb/%.*s"
4159                                        "---a/%.*s"
4160                                        "+++b/%.*s",
4161                                        len1, p->one->path,
4162                                        len2, p->two->path,
4163                                        len1, p->one->path,
4164                                        len2, p->two->path);
4165                git_SHA1_Update(&ctx, buffer, len1);
4166
4167                if (diff_filespec_is_binary(p->one) ||
4168                    diff_filespec_is_binary(p->two)) {
4169                        git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4170                        git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4171                        continue;
4172                }
4173
4174                xpp.flags = 0;
4175                xecfg.ctxlen = 3;
4176                xecfg.flags = 0;
4177                xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4178                              &xpp, &xecfg);
4179        }
4180
4181        git_SHA1_Final(sha1, &ctx);
4182        return 0;
4183}
4184
4185int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4186{
4187        struct diff_queue_struct *q = &diff_queued_diff;
4188        int i;
4189        int result = diff_get_patch_id(options, sha1);
4190
4191        for (i = 0; i < q->nr; i++)
4192                diff_free_filepair(q->queue[i]);
4193
4194        free(q->queue);
4195        DIFF_QUEUE_CLEAR(q);
4196
4197        return result;
4198}
4199
4200static int is_summary_empty(const struct diff_queue_struct *q)
4201{
4202        int i;
4203
4204        for (i = 0; i < q->nr; i++) {
4205                const struct diff_filepair *p = q->queue[i];
4206
4207                switch (p->status) {
4208                case DIFF_STATUS_DELETED:
4209                case DIFF_STATUS_ADDED:
4210                case DIFF_STATUS_COPIED:
4211                case DIFF_STATUS_RENAMED:
4212                        return 0;
4213                default:
4214                        if (p->score)
4215                                return 0;
4216                        if (p->one->mode && p->two->mode &&
4217                            p->one->mode != p->two->mode)
4218                                return 0;
4219                        break;
4220                }
4221        }
4222        return 1;
4223}
4224
4225static const char rename_limit_warning[] =
4226"inexact rename detection was skipped due to too many files.";
4227
4228static const char degrade_cc_to_c_warning[] =
4229"only found copies from modified paths due to too many files.";
4230
4231static const char rename_limit_advice[] =
4232"you may want to set your %s variable to at least "
4233"%d and retry the command.";
4234
4235void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4236{
4237        if (degraded_cc)
4238                warning(degrade_cc_to_c_warning);
4239        else if (needed)
4240                warning(rename_limit_warning);
4241        else
4242                return;
4243        if (0 < needed && needed < 32767)
4244                warning(rename_limit_advice, varname, needed);
4245}
4246
4247void diff_flush(struct diff_options *options)
4248{
4249        struct diff_queue_struct *q = &diff_queued_diff;
4250        int i, output_format = options->output_format;
4251        int separator = 0;
4252        int dirstat_by_line = 0;
4253
4254        /*
4255         * Order: raw, stat, summary, patch
4256         * or:    name/name-status/checkdiff (other bits clear)
4257         */
4258        if (!q->nr)
4259                goto free_queue;
4260
4261        if (output_format & (DIFF_FORMAT_RAW |
4262                             DIFF_FORMAT_NAME |
4263                             DIFF_FORMAT_NAME_STATUS |
4264                             DIFF_FORMAT_CHECKDIFF)) {
4265                for (i = 0; i < q->nr; i++) {
4266                        struct diff_filepair *p = q->queue[i];
4267                        if (check_pair_status(p))
4268                                flush_one_pair(p, options);
4269                }
4270                separator++;
4271        }
4272
4273        if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4274                dirstat_by_line = 1;
4275
4276        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4277            dirstat_by_line) {
4278                struct diffstat_t diffstat;
4279
4280                memset(&diffstat, 0, sizeof(struct diffstat_t));
4281                for (i = 0; i < q->nr; i++) {
4282                        struct diff_filepair *p = q->queue[i];
4283                        if (check_pair_status(p))
4284                                diff_flush_stat(p, options, &diffstat);
4285                }
4286                if (output_format & DIFF_FORMAT_NUMSTAT)
4287                        show_numstat(&diffstat, options);
4288                if (output_format & DIFF_FORMAT_DIFFSTAT)
4289                        show_stats(&diffstat, options);
4290                if (output_format & DIFF_FORMAT_SHORTSTAT)
4291                        show_shortstats(&diffstat, options);
4292                if (output_format & DIFF_FORMAT_DIRSTAT)
4293                        show_dirstat_by_line(&diffstat, options);
4294                free_diffstat_info(&diffstat);
4295                separator++;
4296        }
4297        if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4298                show_dirstat(options);
4299
4300        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4301                for (i = 0; i < q->nr; i++) {
4302                        diff_summary(options, q->queue[i]);
4303                }
4304                separator++;
4305        }
4306
4307        if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4308            DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4309            DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4310                /*
4311                 * run diff_flush_patch for the exit status. setting
4312                 * options->file to /dev/null should be safe, becaue we
4313                 * aren't supposed to produce any output anyway.
4314                 */
4315                if (options->close_file)
4316                        fclose(options->file);
4317                options->file = fopen("/dev/null", "w");
4318                if (!options->file)
4319                        die_errno("Could not open /dev/null");
4320                options->close_file = 1;
4321                for (i = 0; i < q->nr; i++) {
4322                        struct diff_filepair *p = q->queue[i];
4323                        if (check_pair_status(p))
4324                                diff_flush_patch(p, options);
4325                        if (options->found_changes)
4326                                break;
4327                }
4328        }
4329
4330        if (output_format & DIFF_FORMAT_PATCH) {
4331                if (separator) {
4332                        putc(options->line_termination, options->file);
4333                        if (options->stat_sep) {
4334                                /* attach patch instead of inline */
4335                                fputs(options->stat_sep, options->file);
4336                        }
4337                }
4338
4339                for (i = 0; i < q->nr; i++) {
4340                        struct diff_filepair *p = q->queue[i];
4341                        if (check_pair_status(p))
4342                                diff_flush_patch(p, options);
4343                }
4344        }
4345
4346        if (output_format & DIFF_FORMAT_CALLBACK)
4347                options->format_callback(q, options, options->format_callback_data);
4348
4349        for (i = 0; i < q->nr; i++)
4350                diff_free_filepair(q->queue[i]);
4351free_queue:
4352        free(q->queue);
4353        DIFF_QUEUE_CLEAR(q);
4354        if (options->close_file)
4355                fclose(options->file);
4356
4357        /*
4358         * Report the content-level differences with HAS_CHANGES;
4359         * diff_addremove/diff_change does not set the bit when
4360         * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4361         */
4362        if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4363                if (options->found_changes)
4364                        DIFF_OPT_SET(options, HAS_CHANGES);
4365                else
4366                        DIFF_OPT_CLR(options, HAS_CHANGES);
4367        }
4368}
4369
4370static void diffcore_apply_filter(const char *filter)
4371{
4372        int i;
4373        struct diff_queue_struct *q = &diff_queued_diff;
4374        struct diff_queue_struct outq;
4375        DIFF_QUEUE_CLEAR(&outq);
4376
4377        if (!filter)
4378                return;
4379
4380        if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4381                int found;
4382                for (i = found = 0; !found && i < q->nr; i++) {
4383                        struct diff_filepair *p = q->queue[i];
4384                        if (((p->status == DIFF_STATUS_MODIFIED) &&
4385                             ((p->score &&
4386                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4387                              (!p->score &&
4388                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4389                            ((p->status != DIFF_STATUS_MODIFIED) &&
4390                             strchr(filter, p->status)))
4391                                found++;
4392                }
4393                if (found)
4394                        return;
4395
4396                /* otherwise we will clear the whole queue
4397                 * by copying the empty outq at the end of this
4398                 * function, but first clear the current entries
4399                 * in the queue.
4400                 */
4401                for (i = 0; i < q->nr; i++)
4402                        diff_free_filepair(q->queue[i]);
4403        }
4404        else {
4405                /* Only the matching ones */
4406                for (i = 0; i < q->nr; i++) {
4407                        struct diff_filepair *p = q->queue[i];
4408
4409                        if (((p->status == DIFF_STATUS_MODIFIED) &&
4410                             ((p->score &&
4411                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4412                              (!p->score &&
4413                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4414                            ((p->status != DIFF_STATUS_MODIFIED) &&
4415                             strchr(filter, p->status)))
4416                                diff_q(&outq, p);
4417                        else
4418                                diff_free_filepair(p);
4419                }
4420        }
4421        free(q->queue);
4422        *q = outq;
4423}
4424
4425/* Check whether two filespecs with the same mode and size are identical */
4426static int diff_filespec_is_identical(struct diff_filespec *one,
4427                                      struct diff_filespec *two)
4428{
4429        if (S_ISGITLINK(one->mode))
4430                return 0;
4431        if (diff_populate_filespec(one, 0))
4432                return 0;
4433        if (diff_populate_filespec(two, 0))
4434                return 0;
4435        return !memcmp(one->data, two->data, one->size);
4436}
4437
4438static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4439{
4440        int i;
4441        struct diff_queue_struct *q = &diff_queued_diff;
4442        struct diff_queue_struct outq;
4443        DIFF_QUEUE_CLEAR(&outq);
4444
4445        for (i = 0; i < q->nr; i++) {
4446                struct diff_filepair *p = q->queue[i];
4447
4448                /*
4449                 * 1. Entries that come from stat info dirtiness
4450                 *    always have both sides (iow, not create/delete),
4451                 *    one side of the object name is unknown, with
4452                 *    the same mode and size.  Keep the ones that
4453                 *    do not match these criteria.  They have real
4454                 *    differences.
4455                 *
4456                 * 2. At this point, the file is known to be modified,
4457                 *    with the same mode and size, and the object
4458                 *    name of one side is unknown.  Need to inspect
4459                 *    the identical contents.
4460                 */
4461                if (!DIFF_FILE_VALID(p->one) || /* (1) */
4462                    !DIFF_FILE_VALID(p->two) ||
4463                    (p->one->sha1_valid && p->two->sha1_valid) ||
4464                    (p->one->mode != p->two->mode) ||
4465                    diff_populate_filespec(p->one, 1) ||
4466                    diff_populate_filespec(p->two, 1) ||
4467                    (p->one->size != p->two->size) ||
4468                    !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4469                        diff_q(&outq, p);
4470                else {
4471                        /*
4472                         * The caller can subtract 1 from skip_stat_unmatch
4473                         * to determine how many paths were dirty only
4474                         * due to stat info mismatch.
4475                         */
4476                        if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4477                                diffopt->skip_stat_unmatch++;
4478                        diff_free_filepair(p);
4479                }
4480        }
4481        free(q->queue);
4482        *q = outq;
4483}
4484
4485static int diffnamecmp(const void *a_, const void *b_)
4486{
4487        const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4488        const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4489        const char *name_a, *name_b;
4490
4491        name_a = a->one ? a->one->path : a->two->path;
4492        name_b = b->one ? b->one->path : b->two->path;
4493        return strcmp(name_a, name_b);
4494}
4495
4496void diffcore_fix_diff_index(struct diff_options *options)
4497{
4498        struct diff_queue_struct *q = &diff_queued_diff;
4499        qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4500}
4501
4502void diffcore_std(struct diff_options *options)
4503{
4504        if (options->skip_stat_unmatch)
4505                diffcore_skip_stat_unmatch(options);
4506        if (!options->found_follow) {
4507                /* See try_to_follow_renames() in tree-diff.c */
4508                if (options->break_opt != -1)
4509                        diffcore_break(options->break_opt);
4510                if (options->detect_rename)
4511                        diffcore_rename(options);
4512                if (options->break_opt != -1)
4513                        diffcore_merge_broken();
4514        }
4515        if (options->pickaxe)
4516                diffcore_pickaxe(options);
4517        if (options->orderfile)
4518                diffcore_order(options->orderfile);
4519        if (!options->found_follow)
4520                /* See try_to_follow_renames() in tree-diff.c */
4521                diff_resolve_rename_copy();
4522        diffcore_apply_filter(options->filter);
4523
4524        if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4525                DIFF_OPT_SET(options, HAS_CHANGES);
4526        else
4527                DIFF_OPT_CLR(options, HAS_CHANGES);
4528
4529        options->found_follow = 0;
4530}
4531
4532int diff_result_code(struct diff_options *opt, int status)
4533{
4534        int result = 0;
4535
4536        diff_warn_rename_limit("diff.renamelimit",
4537                               opt->needed_rename_limit,
4538                               opt->degraded_cc_to_c);
4539        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4540            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4541                return status;
4542        if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4543            DIFF_OPT_TST(opt, HAS_CHANGES))
4544                result |= 01;
4545        if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4546            DIFF_OPT_TST(opt, CHECK_FAILED))
4547                result |= 02;
4548        return result;
4549}
4550
4551int diff_can_quit_early(struct diff_options *opt)
4552{
4553        return (DIFF_OPT_TST(opt, QUICK) &&
4554                !opt->filter &&
4555                DIFF_OPT_TST(opt, HAS_CHANGES));
4556}
4557
4558/*
4559 * Shall changes to this submodule be ignored?
4560 *
4561 * Submodule changes can be configured to be ignored separately for each path,
4562 * but that configuration can be overridden from the command line.
4563 */
4564static int is_submodule_ignored(const char *path, struct diff_options *options)
4565{
4566        int ignored = 0;
4567        unsigned orig_flags = options->flags;
4568        if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4569                set_diffopt_flags_from_submodule_config(options, path);
4570        if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4571                ignored = 1;
4572        options->flags = orig_flags;
4573        return ignored;
4574}
4575
4576void diff_addremove(struct diff_options *options,
4577                    int addremove, unsigned mode,
4578                    const unsigned char *sha1,
4579                    const char *concatpath, unsigned dirty_submodule)
4580{
4581        struct diff_filespec *one, *two;
4582
4583        if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4584                return;
4585
4586        /* This may look odd, but it is a preparation for
4587         * feeding "there are unchanged files which should
4588         * not produce diffs, but when you are doing copy
4589         * detection you would need them, so here they are"
4590         * entries to the diff-core.  They will be prefixed
4591         * with something like '=' or '*' (I haven't decided
4592         * which but should not make any difference).
4593         * Feeding the same new and old to diff_change()
4594         * also has the same effect.
4595         * Before the final output happens, they are pruned after
4596         * merged into rename/copy pairs as appropriate.
4597         */
4598        if (DIFF_OPT_TST(options, REVERSE_DIFF))
4599                addremove = (addremove == '+' ? '-' :
4600                             addremove == '-' ? '+' : addremove);
4601
4602        if (options->prefix &&
4603            strncmp(concatpath, options->prefix, options->prefix_length))
4604                return;
4605
4606        one = alloc_filespec(concatpath);
4607        two = alloc_filespec(concatpath);
4608
4609        if (addremove != '+')
4610                fill_filespec(one, sha1, mode);
4611        if (addremove != '-') {
4612                fill_filespec(two, sha1, mode);
4613                two->dirty_submodule = dirty_submodule;
4614        }
4615
4616        diff_queue(&diff_queued_diff, one, two);
4617        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4618                DIFF_OPT_SET(options, HAS_CHANGES);
4619}
4620
4621void diff_change(struct diff_options *options,
4622                 unsigned old_mode, unsigned new_mode,
4623                 const unsigned char *old_sha1,
4624                 const unsigned char *new_sha1,
4625                 const char *concatpath,
4626                 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4627{
4628        struct diff_filespec *one, *two;
4629
4630        if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4631            is_submodule_ignored(concatpath, options))
4632                return;
4633
4634        if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4635                unsigned tmp;
4636                const unsigned char *tmp_c;
4637                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4638                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4639                tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4640                        new_dirty_submodule = tmp;
4641        }
4642
4643        if (options->prefix &&
4644            strncmp(concatpath, options->prefix, options->prefix_length))
4645                return;
4646
4647        one = alloc_filespec(concatpath);
4648        two = alloc_filespec(concatpath);
4649        fill_filespec(one, old_sha1, old_mode);
4650        fill_filespec(two, new_sha1, new_mode);
4651        one->dirty_submodule = old_dirty_submodule;
4652        two->dirty_submodule = new_dirty_submodule;
4653
4654        diff_queue(&diff_queued_diff, one, two);
4655        if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4656                DIFF_OPT_SET(options, HAS_CHANGES);
4657}
4658
4659struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4660{
4661        struct diff_filepair *pair;
4662        struct diff_filespec *one, *two;
4663
4664        if (options->prefix &&
4665            strncmp(path, options->prefix, options->prefix_length))
4666                return NULL;
4667
4668        one = alloc_filespec(path);
4669        two = alloc_filespec(path);
4670        pair = diff_queue(&diff_queued_diff, one, two);
4671        pair->is_unmerged = 1;
4672        return pair;
4673}
4674
4675static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4676                size_t *outsize)
4677{
4678        struct diff_tempfile *temp;
4679        const char *argv[3];
4680        const char **arg = argv;
4681        struct child_process child;
4682        struct strbuf buf = STRBUF_INIT;
4683        int err = 0;
4684
4685        temp = prepare_temp_file(spec->path, spec);
4686        *arg++ = pgm;
4687        *arg++ = temp->name;
4688        *arg = NULL;
4689
4690        memset(&child, 0, sizeof(child));
4691        child.use_shell = 1;
4692        child.argv = argv;
4693        child.out = -1;
4694        if (start_command(&child)) {
4695                remove_tempfile();
4696                return NULL;
4697        }
4698
4699        if (strbuf_read(&buf, child.out, 0) < 0)
4700                err = error("error reading from textconv command '%s'", pgm);
4701        close(child.out);
4702
4703        if (finish_command(&child) || err) {
4704                strbuf_release(&buf);
4705                remove_tempfile();
4706                return NULL;
4707        }
4708        remove_tempfile();
4709
4710        return strbuf_detach(&buf, outsize);
4711}
4712
4713size_t fill_textconv(struct userdiff_driver *driver,
4714                     struct diff_filespec *df,
4715                     char **outbuf)
4716{
4717        size_t size;
4718
4719        if (!driver || !driver->textconv) {
4720                if (!DIFF_FILE_VALID(df)) {
4721                        *outbuf = "";
4722                        return 0;
4723                }
4724                if (diff_populate_filespec(df, 0))
4725                        die("unable to read files to diff");
4726                *outbuf = df->data;
4727                return df->size;
4728        }
4729
4730        if (driver->textconv_cache && df->sha1_valid) {
4731                *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4732                                          &size);
4733                if (*outbuf)
4734                        return size;
4735        }
4736
4737        *outbuf = run_textconv(driver->textconv, df, &size);
4738        if (!*outbuf)
4739                die("unable to read files to diff");
4740
4741        if (driver->textconv_cache && df->sha1_valid) {
4742                /* ignore errors, as we might be in a readonly repository */
4743                notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4744                                size);
4745                /*
4746                 * we could save up changes and flush them all at the end,
4747                 * but we would need an extra call after all diffing is done.
4748                 * Since generating a cache entry is the slow path anyway,
4749                 * this extra overhead probably isn't a big deal.
4750                 */
4751                notes_cache_write(driver->textconv_cache);
4752        }
4753
4754        return size;
4755}