diff.con commit [PATCH] Optimize diff-tree -[CM] --stdin (f0c6b2a)
   1/*
   2 * Copyright (C) 2005 Junio C Hamano
   3 */
   4#include <sys/types.h>
   5#include <sys/wait.h>
   6#include <signal.h>
   7#include "cache.h"
   8#include "diff.h"
   9#include "diffcore.h"
  10
  11static const char *diff_opts = "-pu";
  12static unsigned char null_sha1[20] = { 0, };
  13
  14static int reverse_diff;
  15static int use_size_cache;
  16
  17static const char *external_diff(void)
  18{
  19        static const char *external_diff_cmd = NULL;
  20        static int done_preparing = 0;
  21
  22        if (done_preparing)
  23                return external_diff_cmd;
  24
  25        /*
  26         * Default values above are meant to match the
  27         * Linux kernel development style.  Examples of
  28         * alternative styles you can specify via environment
  29         * variables are:
  30         *
  31         * GIT_DIFF_OPTS="-c";
  32         */
  33        if (gitenv("GIT_EXTERNAL_DIFF"))
  34                external_diff_cmd = gitenv("GIT_EXTERNAL_DIFF");
  35
  36        /* In case external diff fails... */
  37        diff_opts = gitenv("GIT_DIFF_OPTS") ? : diff_opts;
  38
  39        done_preparing = 1;
  40        return external_diff_cmd;
  41}
  42
  43/* Help to copy the thing properly quoted for the shell safety.
  44 * any single quote is replaced with '\'', and the caller is
  45 * expected to enclose the result within a single quote pair.
  46 *
  47 * E.g.
  48 *  original     sq_expand     result
  49 *  name     ==> name      ==> 'name'
  50 *  a b      ==> a b       ==> 'a b'
  51 *  a'b      ==> a'\''b    ==> 'a'\''b'
  52 */
  53static char *sq_expand(const char *src)
  54{
  55        static char *buf = NULL;
  56        int cnt, c;
  57        const char *cp;
  58        char *bp;
  59
  60        /* count bytes needed to store the quoted string. */
  61        for (cnt = 1, cp = src; *cp; cnt++, cp++)
  62                if (*cp == '\'')
  63                        cnt += 3;
  64
  65        buf = xmalloc(cnt);
  66        bp = buf;
  67        while ((c = *src++)) {
  68                if (c != '\'')
  69                        *bp++ = c;
  70                else {
  71                        bp = strcpy(bp, "'\\''");
  72                        bp += 4;
  73                }
  74        }
  75        *bp = 0;
  76        return buf;
  77}
  78
  79static struct diff_tempfile {
  80        const char *name; /* filename external diff should read from */
  81        char hex[41];
  82        char mode[10];
  83        char tmp_path[50];
  84} diff_temp[2];
  85
  86static void builtin_diff(const char *name_a,
  87                         const char *name_b,
  88                         struct diff_tempfile *temp,
  89                         const char *xfrm_msg)
  90{
  91        int i, next_at, cmd_size;
  92        const char *diff_cmd = "diff -L'%s%s' -L'%s%s'";
  93        const char *diff_arg  = "'%s' '%s'||:"; /* "||:" is to return 0 */
  94        const char *input_name_sq[2];
  95        const char *path0[2];
  96        const char *path1[2];
  97        const char *name_sq[2];
  98        char *cmd;
  99
 100        name_sq[0] = sq_expand(name_a);
 101        name_sq[1] = sq_expand(name_b);
 102
 103        /* diff_cmd and diff_arg have 6 %s in total which makes
 104         * the sum of these strings 12 bytes larger than required.
 105         * we use 2 spaces around diff-opts, and we need to count
 106         * terminating NUL, so we subtract 9 here.
 107         */
 108        cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
 109                        strlen(diff_arg) - 9);
 110        for (i = 0; i < 2; i++) {
 111                input_name_sq[i] = sq_expand(temp[i].name);
 112                if (!strcmp(temp[i].name, "/dev/null")) {
 113                        path0[i] = "/dev/null";
 114                        path1[i] = "";
 115                } else {
 116                        path0[i] = i ? "b/" : "a/";
 117                        path1[i] = name_sq[i];
 118                }
 119                cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
 120                             strlen(input_name_sq[i]));
 121        }
 122
 123        cmd = xmalloc(cmd_size);
 124
 125        next_at = 0;
 126        next_at += snprintf(cmd+next_at, cmd_size-next_at,
 127                            diff_cmd,
 128                            path0[0], path1[0], path0[1], path1[1]);
 129        next_at += snprintf(cmd+next_at, cmd_size-next_at,
 130                            " %s ", diff_opts);
 131        next_at += snprintf(cmd+next_at, cmd_size-next_at,
 132                            diff_arg, input_name_sq[0], input_name_sq[1]);
 133
 134        printf("diff --git a/%s b/%s\n", name_a, name_b);
 135        if (!path1[0][0])
 136                printf("new file mode %s\n", temp[1].mode);
 137        else if (!path1[1][0])
 138                printf("deleted file mode %s\n", temp[0].mode);
 139        else {
 140                if (strcmp(temp[0].mode, temp[1].mode)) {
 141                        printf("old mode %s\n", temp[0].mode);
 142                        printf("new mode %s\n", temp[1].mode);
 143                }
 144                if (xfrm_msg && xfrm_msg[0])
 145                        puts(xfrm_msg);
 146
 147                if (strncmp(temp[0].mode, temp[1].mode, 3))
 148                        /* we do not run diff between different kind
 149                         * of objects.
 150                         */
 151                        exit(0);
 152        }
 153        fflush(NULL);
 154        execlp("/bin/sh","sh", "-c", cmd, NULL);
 155}
 156
 157struct diff_filespec *alloc_filespec(const char *path)
 158{
 159        int namelen = strlen(path);
 160        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 161        spec->path = (char *)(spec + 1);
 162        strcpy(spec->path, path);
 163        spec->should_free = spec->should_munmap = 0;
 164        spec->xfrm_flags = 0;
 165        spec->size = 0;
 166        spec->data = NULL;
 167        spec->mode = 0;
 168        memset(spec->sha1, 0, 20);
 169        return spec;
 170}
 171
 172void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 173                   unsigned short mode)
 174{
 175        if (mode) {
 176                spec->mode = DIFF_FILE_CANON_MODE(mode);
 177                memcpy(spec->sha1, sha1, 20);
 178                spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 179        }
 180}
 181
 182/*
 183 * Given a name and sha1 pair, if the dircache tells us the file in
 184 * the work tree has that object contents, return true, so that
 185 * prepare_temp_file() does not have to inflate and extract.
 186 */
 187static int work_tree_matches(const char *name, const unsigned char *sha1)
 188{
 189        struct cache_entry *ce;
 190        struct stat st;
 191        int pos, len;
 192
 193        /* We do not read the cache ourselves here, because the
 194         * benchmark with my previous version that always reads cache
 195         * shows that it makes things worse for diff-tree comparing
 196         * two linux-2.6 kernel trees in an already checked out work
 197         * tree.  This is because most diff-tree comparisons deal with
 198         * only a small number of files, while reading the cache is
 199         * expensive for a large project, and its cost outweighs the
 200         * savings we get by not inflating the object to a temporary
 201         * file.  Practically, this code only helps when we are used
 202         * by diff-cache --cached, which does read the cache before
 203         * calling us.
 204         */
 205        if (!active_cache)
 206                return 0;
 207
 208        len = strlen(name);
 209        pos = cache_name_pos(name, len);
 210        if (pos < 0)
 211                return 0;
 212        ce = active_cache[pos];
 213        if ((lstat(name, &st) < 0) ||
 214            !S_ISREG(st.st_mode) || /* careful! */
 215            ce_match_stat(ce, &st) ||
 216            memcmp(sha1, ce->sha1, 20))
 217                return 0;
 218        /* we return 1 only when we can stat, it is a regular file,
 219         * stat information matches, and sha1 recorded in the cache
 220         * matches.  I.e. we know the file in the work tree really is
 221         * the same as the <name, sha1> pair.
 222         */
 223        return 1;
 224}
 225
 226static struct sha1_size_cache {
 227        unsigned char sha1[20];
 228        unsigned long size;
 229} **sha1_size_cache;
 230static int sha1_size_cache_nr, sha1_size_cache_alloc;
 231
 232static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 233                                                 unsigned long size)
 234{
 235        int first, last;
 236        struct sha1_size_cache *e;
 237
 238        first = 0;
 239        last = sha1_size_cache_nr;
 240        while (last > first) {
 241                int next = (last + first) >> 1;
 242                e = sha1_size_cache[next];
 243                int cmp = memcmp(e->sha1, sha1, 20);
 244                if (!cmp)
 245                        return e;
 246                if (cmp < 0) {
 247                        last = next;
 248                        continue;
 249                }
 250                first = next+1;
 251        }
 252        /* not found */
 253        if (size == UINT_MAX)
 254                return NULL;
 255        /* insert to make it at "first" */
 256        if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 257                sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 258                sha1_size_cache = xrealloc(sha1_size_cache,
 259                                           sha1_size_cache_alloc *
 260                                           sizeof(*sha1_size_cache));
 261        }
 262        sha1_size_cache_nr++;
 263        if (first < sha1_size_cache_nr)
 264                memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 265                        (sha1_size_cache_nr - first - 1) *
 266                        sizeof(*sha1_size_cache));
 267        e = xmalloc(sizeof(struct sha1_size_cache));
 268        sha1_size_cache[first] = e;
 269        memcpy(e->sha1, sha1, 20);
 270        e->size = size;
 271        return e;
 272}
 273
 274/*
 275 * While doing rename detection and pickaxe operation, we may need to
 276 * grab the data for the blob (or file) for our own in-core comparison.
 277 * diff_filespec has data and size fields for this purpose.
 278 */
 279int diff_populate_filespec(struct diff_filespec *s, int size_only)
 280{
 281        int err = 0;
 282        if (!DIFF_FILE_VALID(s))
 283                die("internal error: asking to populate invalid file.");
 284        if (S_ISDIR(s->mode))
 285                return -1;
 286
 287        if (!use_size_cache)
 288                size_only = 0;
 289
 290        if (s->data)
 291                return err;
 292        if (!s->sha1_valid ||
 293            work_tree_matches(s->path, s->sha1)) {
 294                struct stat st;
 295                int fd;
 296                if (lstat(s->path, &st) < 0) {
 297                        if (errno == ENOENT) {
 298                        err_empty:
 299                                err = -1;
 300                        empty:
 301                                s->data = "";
 302                                s->size = 0;
 303                                return err;
 304                        }
 305                }
 306                s->size = st.st_size;
 307                if (!s->size)
 308                        goto empty;
 309                if (size_only)
 310                        return 0;
 311                if (S_ISLNK(st.st_mode)) {
 312                        int ret;
 313                        s->data = xmalloc(s->size);
 314                        s->should_free = 1;
 315                        ret = readlink(s->path, s->data, s->size);
 316                        if (ret < 0) {
 317                                free(s->data);
 318                                goto err_empty;
 319                        }
 320                        return 0;
 321                }
 322                fd = open(s->path, O_RDONLY);
 323                if (fd < 0)
 324                        goto err_empty;
 325                s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 326                s->should_munmap = 1;
 327                close(fd);
 328        }
 329        else {
 330                /* We cannot do size only for SHA1 blobs */
 331                char type[20];
 332                struct sha1_size_cache *e;
 333
 334                if (size_only) {
 335                        e = locate_size_cache(s->sha1, UINT_MAX);
 336                        if (e) {
 337                                s->size = e->size;
 338                                return 0;
 339                        }
 340                }
 341                s->data = read_sha1_file(s->sha1, type, &s->size);
 342                s->should_free = 1;
 343                if (s->data && size_only)
 344                        locate_size_cache(s->sha1, s->size);
 345        }
 346        return 0;
 347}
 348
 349void diff_free_filespec_data(struct diff_filespec *s)
 350{
 351        if (s->should_free)
 352                free(s->data);
 353        else if (s->should_munmap)
 354                munmap(s->data, s->size);
 355        s->should_free = s->should_munmap = 0;
 356        s->data = NULL;
 357}
 358
 359static void prep_temp_blob(struct diff_tempfile *temp,
 360                           void *blob,
 361                           unsigned long size,
 362                           unsigned char *sha1,
 363                           int mode)
 364{
 365        int fd;
 366
 367        strcpy(temp->tmp_path, ".diff_XXXXXX");
 368        fd = mkstemp(temp->tmp_path);
 369        if (fd < 0)
 370                die("unable to create temp-file");
 371        if (write(fd, blob, size) != size)
 372                die("unable to write temp-file");
 373        close(fd);
 374        temp->name = temp->tmp_path;
 375        strcpy(temp->hex, sha1_to_hex(sha1));
 376        temp->hex[40] = 0;
 377        sprintf(temp->mode, "%06o", mode);
 378}
 379
 380static void prepare_temp_file(const char *name,
 381                              struct diff_tempfile *temp,
 382                              struct diff_filespec *one)
 383{
 384        if (!DIFF_FILE_VALID(one)) {
 385        not_a_valid_file:
 386                /* A '-' entry produces this for file-2, and
 387                 * a '+' entry produces this for file-1.
 388                 */
 389                temp->name = "/dev/null";
 390                strcpy(temp->hex, ".");
 391                strcpy(temp->mode, ".");
 392                return;
 393        }
 394
 395        if (!one->sha1_valid ||
 396            work_tree_matches(name, one->sha1)) {
 397                struct stat st;
 398                if (lstat(name, &st) < 0) {
 399                        if (errno == ENOENT)
 400                                goto not_a_valid_file;
 401                        die("stat(%s): %s", name, strerror(errno));
 402                }
 403                if (S_ISLNK(st.st_mode)) {
 404                        int ret;
 405                        char *buf, buf_[1024];
 406                        buf = ((sizeof(buf_) < st.st_size) ?
 407                               xmalloc(st.st_size) : buf_);
 408                        ret = readlink(name, buf, st.st_size);
 409                        if (ret < 0)
 410                                die("readlink(%s)", name);
 411                        prep_temp_blob(temp, buf, st.st_size,
 412                                       (one->sha1_valid ?
 413                                        one->sha1 : null_sha1),
 414                                       (one->sha1_valid ?
 415                                        one->mode : S_IFLNK));
 416                }
 417                else {
 418                        /* we can borrow from the file in the work tree */
 419                        temp->name = name;
 420                        if (!one->sha1_valid)
 421                                strcpy(temp->hex, sha1_to_hex(null_sha1));
 422                        else
 423                                strcpy(temp->hex, sha1_to_hex(one->sha1));
 424                        sprintf(temp->mode, "%06o",
 425                                S_IFREG |ce_permissions(st.st_mode));
 426                }
 427                return;
 428        }
 429        else {
 430                if (diff_populate_filespec(one, 0))
 431                        die("cannot read data blob for %s", one->path);
 432                prep_temp_blob(temp, one->data, one->size,
 433                               one->sha1, one->mode);
 434        }
 435}
 436
 437static void remove_tempfile(void)
 438{
 439        int i;
 440
 441        for (i = 0; i < 2; i++)
 442                if (diff_temp[i].name == diff_temp[i].tmp_path) {
 443                        unlink(diff_temp[i].name);
 444                        diff_temp[i].name = NULL;
 445                }
 446}
 447
 448static void remove_tempfile_on_signal(int signo)
 449{
 450        remove_tempfile();
 451}
 452
 453/* An external diff command takes:
 454 *
 455 * diff-cmd name infile1 infile1-sha1 infile1-mode \
 456 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 457 *
 458 */
 459static void run_external_diff(const char *pgm,
 460                              const char *name,
 461                              const char *other,
 462                              struct diff_filespec *one,
 463                              struct diff_filespec *two,
 464                              const char *xfrm_msg)
 465{
 466        struct diff_tempfile *temp = diff_temp;
 467        pid_t pid;
 468        int status;
 469        static int atexit_asked = 0;
 470
 471        if (one && two) {
 472                prepare_temp_file(name, &temp[0], one);
 473                prepare_temp_file(other ? : name, &temp[1], two);
 474                if (! atexit_asked &&
 475                    (temp[0].name == temp[0].tmp_path ||
 476                     temp[1].name == temp[1].tmp_path)) {
 477                        atexit_asked = 1;
 478                        atexit(remove_tempfile);
 479                }
 480                signal(SIGINT, remove_tempfile_on_signal);
 481        }
 482
 483        fflush(NULL);
 484        pid = fork();
 485        if (pid < 0)
 486                die("unable to fork");
 487        if (!pid) {
 488                if (pgm) {
 489                        if (one && two) {
 490                                const char *exec_arg[10];
 491                                const char **arg = &exec_arg[0];
 492                                *arg++ = pgm;
 493                                *arg++ = name;
 494                                *arg++ = temp[0].name;
 495                                *arg++ = temp[0].hex;
 496                                *arg++ = temp[0].mode;
 497                                *arg++ = temp[1].name;
 498                                *arg++ = temp[1].hex;
 499                                *arg++ = temp[1].mode;
 500                                if (other) {
 501                                        *arg++ = other;
 502                                        *arg++ = xfrm_msg;
 503                                }
 504                                *arg = NULL;
 505                                execvp(pgm, (char *const*) exec_arg);
 506                        }
 507                        else
 508                                execlp(pgm, pgm, name, NULL);
 509                }
 510                /*
 511                 * otherwise we use the built-in one.
 512                 */
 513                if (one && two)
 514                        builtin_diff(name, other ? : name, temp, xfrm_msg);
 515                else
 516                        printf("* Unmerged path %s\n", name);
 517                exit(0);
 518        }
 519        if (waitpid(pid, &status, 0) < 0 ||
 520            !WIFEXITED(status) || WEXITSTATUS(status)) {
 521                /* Earlier we did not check the exit status because
 522                 * diff exits non-zero if files are different, and
 523                 * we are not interested in knowing that.  It was a
 524                 * mistake which made it harder to quit a diff-*
 525                 * session that uses the git-apply-patch-script as
 526                 * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 527                 * should also exit non-zero only when it wants to
 528                 * abort the entire diff-* session.
 529                 */
 530                remove_tempfile();
 531                fprintf(stderr, "external diff died, stopping at %s.\n", name);
 532                exit(1);
 533        }
 534        remove_tempfile();
 535}
 536
 537static void run_diff(const char *name,
 538                     const char *other,
 539                     struct diff_filespec *one,
 540                     struct diff_filespec *two,
 541                     const char *xfrm_msg)
 542{
 543        const char *pgm = external_diff();
 544        if (!pgm &&
 545            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
 546            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
 547                /* a filepair that changes between file and symlink
 548                 * needs to be split into deletion and creation.
 549                 */
 550                struct diff_filespec *null = alloc_filespec(two->path);
 551                run_external_diff(NULL, name, other, one, null, xfrm_msg);
 552                free(null);
 553                null = alloc_filespec(one->path);
 554                run_external_diff(NULL, name, other, null, two, xfrm_msg);
 555                free(null);
 556        }
 557        else
 558                run_external_diff(pgm, name, other, one, two, xfrm_msg);
 559}
 560
 561void diff_setup(int flags)
 562{
 563        if (flags & DIFF_SETUP_REVERSE)
 564                reverse_diff = 1;
 565        if (flags & DIFF_SETUP_USE_CACHE) {
 566                if (!active_cache)
 567                        /* read-cache does not die even when it fails
 568                         * so it is safe for us to do this here.  Also
 569                         * it does not smudge active_cache or active_nr
 570                         * when it fails, so we do not have to worry about
 571                         * cleaning it up oufselves either.
 572                         */
 573                        read_cache();
 574        }
 575        if (flags & DIFF_SETUP_USE_SIZE_CACHE)
 576                use_size_cache = 1;
 577        
 578}
 579
 580struct diff_queue_struct diff_queued_diff;
 581
 582void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
 583{
 584        if (queue->alloc <= queue->nr) {
 585                queue->alloc = alloc_nr(queue->alloc);
 586                queue->queue = xrealloc(queue->queue,
 587                                        sizeof(dp) * queue->alloc);
 588        }
 589        queue->queue[queue->nr++] = dp;
 590}
 591
 592struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
 593                                 struct diff_filespec *one,
 594                                 struct diff_filespec *two)
 595{
 596        struct diff_filepair *dp = xmalloc(sizeof(*dp));
 597        dp->one = one;
 598        dp->two = two;
 599        dp->score = 0;
 600        dp->source_stays = 0;
 601        diff_q(queue, dp);
 602        return dp;
 603}
 604
 605void diff_free_filepair(struct diff_filepair *p)
 606{
 607        diff_free_filespec_data(p->one);
 608        diff_free_filespec_data(p->two);
 609        free(p);
 610}
 611
 612static void diff_flush_raw(struct diff_filepair *p,
 613                           int line_termination,
 614                           int inter_name_termination)
 615{
 616        int two_paths;
 617        char status[10];
 618
 619        if (line_termination) {
 620                const char *err = "path %s cannot be expressed without -z";
 621                if (strchr(p->one->path, line_termination) ||
 622                    strchr(p->one->path, inter_name_termination))
 623                        die(err, p->one->path);
 624                if (strchr(p->two->path, line_termination) ||
 625                    strchr(p->two->path, inter_name_termination))
 626                        die(err, p->two->path);
 627        }
 628
 629        switch (p->status) {
 630        case 'C': case 'R':
 631                two_paths = 1;
 632                sprintf(status, "%c%03d", p->status,
 633                        (int)(0.5 + p->score * 100.0/MAX_SCORE));
 634                break;
 635        default:
 636                two_paths = 0;
 637                status[0] = p->status;
 638                status[1] = 0;
 639                break;
 640        }
 641        printf(":%06o %06o %s ",
 642               p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
 643        printf("%s %s%c%s",
 644               sha1_to_hex(p->two->sha1),
 645               status,
 646               inter_name_termination,
 647               p->one->path);
 648        if (two_paths)
 649                printf("%c%s", inter_name_termination, p->two->path);
 650        putchar(line_termination);
 651}
 652
 653int diff_unmodified_pair(struct diff_filepair *p)
 654{
 655        /* This function is written stricter than necessary to support
 656         * the currently implemented transformers, but the idea is to
 657         * let transformers to produce diff_filepairs any way they want,
 658         * and filter and clean them up here before producing the output.
 659         */
 660        struct diff_filespec *one, *two;
 661
 662        if (DIFF_PAIR_UNMERGED(p))
 663                return 0; /* unmerged is interesting */
 664
 665        one = p->one;
 666        two = p->two;
 667
 668        /* deletion, addition, mode or type change
 669         * and rename are all interesting.
 670         */
 671        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
 672            DIFF_PAIR_MODE_CHANGED(p) ||
 673            strcmp(one->path, two->path))
 674                return 0;
 675
 676        /* both are valid and point at the same path.  that is, we are
 677         * dealing with a change.
 678         */
 679        if (one->sha1_valid && two->sha1_valid &&
 680            !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
 681                return 1; /* no change */
 682        if (!one->sha1_valid && !two->sha1_valid)
 683                return 1; /* both look at the same file on the filesystem. */
 684        return 0;
 685}
 686
 687static void diff_flush_patch(struct diff_filepair *p)
 688{
 689        const char *name, *other;
 690        char msg_[PATH_MAX*2+200], *msg;
 691
 692        if (diff_unmodified_pair(p))
 693                return;
 694
 695        name = p->one->path;
 696        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 697        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
 698            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
 699                return; /* no tree diffs in patch format */ 
 700
 701        switch (p->status) {
 702        case 'C':
 703                sprintf(msg_,
 704                        "similarity index %d%%\n"
 705                        "copy from %s\n"
 706                        "copy to %s",
 707                        (int)(0.5 + p->score * 100.0/MAX_SCORE),
 708                        p->one->path, p->two->path);
 709                msg = msg_;
 710                break;
 711        case 'R':
 712                sprintf(msg_,
 713                        "similarity index %d%%\n"
 714                        "rename old %s\n"
 715                        "rename new %s",
 716                        (int)(0.5 + p->score * 100.0/MAX_SCORE),
 717                        p->one->path, p->two->path);
 718                msg = msg_;
 719                break;
 720        default:
 721                msg = NULL;
 722        }
 723
 724        if (DIFF_PAIR_UNMERGED(p))
 725                run_diff(name, NULL, NULL, NULL, NULL);
 726        else
 727                run_diff(name, other, p->one, p->two, msg);
 728}
 729
 730int diff_queue_is_empty(void)
 731{
 732        struct diff_queue_struct *q = &diff_queued_diff;
 733        int i;
 734        for (i = 0; i < q->nr; i++)
 735                if (!diff_unmodified_pair(q->queue[i]))
 736                        return 0;
 737        return 1;
 738}
 739
 740#if DIFF_DEBUG
 741void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
 742{
 743        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
 744                x, one ? : "",
 745                s->path,
 746                DIFF_FILE_VALID(s) ? "valid" : "invalid",
 747                s->mode,
 748                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
 749        fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
 750                x, one ? : "",
 751                s->size, s->xfrm_flags);
 752}
 753
 754void diff_debug_filepair(const struct diff_filepair *p, int i)
 755{
 756        diff_debug_filespec(p->one, i, "one");
 757        diff_debug_filespec(p->two, i, "two");
 758        fprintf(stderr, "score %d, status %c source_stays %d\n",
 759                p->score, p->status ? : '?', p->source_stays);
 760}
 761
 762void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
 763{
 764        int i;
 765        if (msg)
 766                fprintf(stderr, "%s\n", msg);
 767        fprintf(stderr, "q->nr = %d\n", q->nr);
 768        for (i = 0; i < q->nr; i++) {
 769                struct diff_filepair *p = q->queue[i];
 770                diff_debug_filepair(p, i);
 771        }
 772}
 773#endif
 774
 775static void diff_resolve_rename_copy(void)
 776{
 777        int i, j;
 778        struct diff_filepair *p, *pp;
 779        struct diff_queue_struct *q = &diff_queued_diff;
 780
 781        diff_debug_queue("resolve-rename-copy", q);
 782
 783        for (i = 0; i < q->nr; i++) {
 784                p = q->queue[i];
 785                p->status = 0; /* undecided */
 786                if (DIFF_PAIR_UNMERGED(p))
 787                        p->status = 'U';
 788                else if (!DIFF_FILE_VALID(p->one))
 789                        p->status = 'N';
 790                else if (!DIFF_FILE_VALID(p->two)) {
 791                        /* Deleted entry may have been picked up by
 792                         * another rename-copy entry.  So we scan the
 793                         * queue and if we find one that uses us as the
 794                         * source we do not say delete for this entry.
 795                         */
 796                        for (j = 0; j < q->nr; j++) {
 797                                pp = q->queue[j];
 798                                if (!strcmp(p->one->path, pp->one->path) &&
 799                                    pp->score) {
 800                                        /* rename/copy are always valid
 801                                         * so we do not say DIFF_FILE_VALID()
 802                                         * on pp->one and pp->two.
 803                                         */
 804                                        p->status = 'X';
 805                                        break;
 806                                }
 807                        }
 808                        if (!p->status)
 809                                p->status = 'D';
 810                }
 811                else if (DIFF_PAIR_TYPE_CHANGED(p))
 812                        p->status = 'T';
 813
 814                /* from this point on, we are dealing with a pair
 815                 * whose both sides are valid and of the same type, i.e.
 816                 * either in-place edit or rename/copy edit.
 817                 */
 818                else if (p->score) {
 819                        if (p->source_stays) {
 820                                p->status = 'C';
 821                                continue;
 822                        }
 823                        /* See if there is some other filepair that
 824                         * copies from the same source as us.  If so
 825                         * we are a copy.  Otherwise we are a rename.
 826                         */
 827                        for (j = i + 1; j < q->nr; j++) {
 828                                pp = q->queue[j];
 829                                if (strcmp(pp->one->path, p->one->path))
 830                                        continue; /* not us */
 831                                if (!pp->score)
 832                                        continue; /* not a rename/copy */
 833                                /* pp is a rename/copy from the same source */
 834                                p->status = 'C';
 835                                break;
 836                        }
 837                        if (!p->status)
 838                                p->status = 'R';
 839                }
 840                else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
 841                         p->one->mode != p->two->mode)
 842                        p->status = 'M';
 843                else
 844                        /* this is a "no-change" entry.
 845                         * should not happen anymore.
 846                         * p->status = 'X';
 847                         */
 848                        die("internal error in diffcore: unmodified entry remains");
 849        }
 850        diff_debug_queue("resolve-rename-copy done", q);
 851}
 852
 853void diff_flush(int diff_output_style, int resolve_rename_copy)
 854{
 855        struct diff_queue_struct *q = &diff_queued_diff;
 856        int i;
 857        int line_termination = '\n';
 858        int inter_name_termination = '\t';
 859
 860        if (diff_output_style == DIFF_FORMAT_MACHINE)
 861                line_termination = inter_name_termination = 0;
 862        if (resolve_rename_copy)
 863                diff_resolve_rename_copy();
 864
 865        for (i = 0; i < q->nr; i++) {
 866                struct diff_filepair *p = q->queue[i];
 867                if ((diff_output_style == DIFF_FORMAT_NO_OUTPUT) ||
 868                    (p->status == 'X'))
 869                        continue;
 870                if (p->status == 0)
 871                        die("internal error in diff-resolve-rename-copy");
 872                switch (diff_output_style) {
 873                case DIFF_FORMAT_PATCH:
 874                        diff_flush_patch(p);
 875                        break;
 876                case DIFF_FORMAT_HUMAN:
 877                case DIFF_FORMAT_MACHINE:
 878                        diff_flush_raw(p, line_termination,
 879                                       inter_name_termination);
 880                        break;
 881                }
 882        }
 883        for (i = 0; i < q->nr; i++)
 884                diff_free_filepair(q->queue[i]);
 885        free(q->queue);
 886        q->queue = NULL;
 887        q->nr = q->alloc = 0;
 888}
 889
 890void diff_addremove(int addremove, unsigned mode,
 891                    const unsigned char *sha1,
 892                    const char *base, const char *path)
 893{
 894        char concatpath[PATH_MAX];
 895        struct diff_filespec *one, *two;
 896
 897        /* This may look odd, but it is a preparation for
 898         * feeding "there are unchanged files which should
 899         * not produce diffs, but when you are doing copy
 900         * detection you would need them, so here they are"
 901         * entries to the diff-core.  They will be prefixed
 902         * with something like '=' or '*' (I haven't decided
 903         * which but should not make any difference).
 904         * Feeding the same new and old to diff_change() 
 905         * also has the same effect.
 906         * Before the final output happens, they are pruned after
 907         * merged into rename/copy pairs as appropriate.
 908         */
 909        if (reverse_diff)
 910                addremove = (addremove == '+' ? '-' :
 911                             addremove == '-' ? '+' : addremove);
 912
 913        if (!path) path = "";
 914        sprintf(concatpath, "%s%s", base, path);
 915        one = alloc_filespec(concatpath);
 916        two = alloc_filespec(concatpath);
 917
 918        if (addremove != '+')
 919                fill_filespec(one, sha1, mode);
 920        if (addremove != '-')
 921                fill_filespec(two, sha1, mode);
 922
 923        diff_queue(&diff_queued_diff, one, two);
 924}
 925
 926void diff_helper_input(unsigned old_mode,
 927                       unsigned new_mode,
 928                       const unsigned char *old_sha1,
 929                       const unsigned char *new_sha1,
 930                       const char *old_path,
 931                       int status,
 932                       int score,
 933                       const char *new_path)
 934{
 935        struct diff_filespec *one, *two;
 936        struct diff_filepair *dp;
 937
 938        one = alloc_filespec(old_path);
 939        two = alloc_filespec(new_path);
 940        if (old_mode)
 941                fill_filespec(one, old_sha1, old_mode);
 942        if (new_mode)
 943                fill_filespec(two, new_sha1, new_mode);
 944        dp = diff_queue(&diff_queued_diff, one, two);
 945        dp->score = score * MAX_SCORE / 100;
 946        dp->status = status;
 947}
 948
 949void diff_change(unsigned old_mode, unsigned new_mode,
 950                 const unsigned char *old_sha1,
 951                 const unsigned char *new_sha1,
 952                 const char *base, const char *path) 
 953{
 954        char concatpath[PATH_MAX];
 955        struct diff_filespec *one, *two;
 956
 957        if (reverse_diff) {
 958                unsigned tmp;
 959                const unsigned char *tmp_c;
 960                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
 961                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
 962        }
 963        if (!path) path = "";
 964        sprintf(concatpath, "%s%s", base, path);
 965        one = alloc_filespec(concatpath);
 966        two = alloc_filespec(concatpath);
 967        fill_filespec(one, old_sha1, old_mode);
 968        fill_filespec(two, new_sha1, new_mode);
 969
 970        diff_queue(&diff_queued_diff, one, two);
 971}
 972
 973void diff_unmerge(const char *path)
 974{
 975        struct diff_filespec *one, *two;
 976        one = alloc_filespec(path);
 977        two = alloc_filespec(path);
 978        diff_queue(&diff_queued_diff, one, two);
 979}