convert.c

   1 #define NO_THE_INDEX_COMPATIBILITY_MACROS
   2 #include "cache.h"
   3 #include "config.h"
   4 #include "attr.h"
   5 #include "run-command.h"
   6 #include "quote.h"
   7 #include "sigchain.h"
   8 #include "pkt-line.h"
   9 #include "sub-process.h"
  10
  11 /*
  12  * convert.c - convert a file when checking it out and checking it in.
  13  *
  14  * This should use the pathname to decide on whether it wants to do some
  15  * more interesting conversions (automatic gzip/unzip, general format
  16  * conversions etc etc), but by default it just does automatic CRLF<->LF
  17  * translation when the "text" attribute or "auto_crlf" option is set.
  18  */
  19
  20 /* Stat bits: When BIN is set, the txt bits are unset */
  21 #define CONVERT_STAT_BITS_TXT_LF    0x1
  22 #define CONVERT_STAT_BITS_TXT_CRLF  0x2
  23 #define CONVERT_STAT_BITS_BIN       0x4
  24
  25 enum crlf_action {
  26         CRLF_UNDEFINED,
  27         CRLF_BINARY,
  28         CRLF_TEXT,
  29         CRLF_TEXT_INPUT,
  30         CRLF_TEXT_CRLF,
  31         CRLF_AUTO,
  32         CRLF_AUTO_INPUT,
  33         CRLF_AUTO_CRLF
  34 };
  35
  36 struct text_stat {
  37         /* NUL, CR, LF and CRLF counts */
  38         unsigned nul, lonecr, lonelf, crlf;
  39
  40         /* These are just approximations! */
  41         unsigned printable, nonprintable;
  42 };
  43
  44 static void gather_stats(const char *buf, unsigned long size, struct text_stat *stats)
  45 {
  46         unsigned long i;
  47
  48         memset(stats, 0, sizeof(*stats));
  49
  50         for (i = 0; i < size; i++) {
  51                 unsigned char c = buf[i];
  52                 if (c == '\r') {
  53                         if (i+1 < size && buf[i+1] == '\n') {
  54                                 stats->crlf++;
  55                                 i++;
  56                         } else
  57                                 stats->lonecr++;
  58                         continue;
  59                 }
  60                 if (c == '\n') {
  61                         stats->lonelf++;
  62                         continue;
  63                 }
  64                 if (c == 127)
  65                         /* DEL */
  66                         stats->nonprintable++;
  67                 else if (c < 32) {
  68                         switch (c) {
  69                                 /* BS, HT, ESC and FF */
  70                         case '\b': case '\t': case '\033': case '\014':
  71                                 stats->printable++;
  72                                 break;
  73                         case 0:
  74                                 stats->nul++;
  75                                 /* fall through */
  76                         default:
  77                                 stats->nonprintable++;
  78                         }
  79                 }
  80                 else
  81                         stats->printable++;
  82         }
  83
  84         /* If file ends with EOF then don't count this EOF as non-printable. */
  85         if (size >= 1 && buf[size-1] == '\032')
  86                 stats->nonprintable--;
  87 }
  88
  89 /*
  90  * The same heuristics as diff.c::mmfile_is_binary()
  91  * We treat files with bare CR as binary
  92  */
  93 static int convert_is_binary(unsigned long size, const struct text_stat *stats)
  94 {
  95         if (stats->lonecr)
  96                 return 1;
  97         if (stats->nul)
  98                 return 1;
  99         if ((stats->printable >> 7) < stats->nonprintable)
 100                 return 1;
 101         return 0;
 102 }
 103
 104 static unsigned int gather_convert_stats(const char *data, unsigned long size)
 105 {
 106         struct text_stat stats;
 107         int ret = 0;
 108         if (!data || !size)
 109                 return 0;
 110         gather_stats(data, size, &stats);
 111         if (convert_is_binary(size, &stats))
 112                 ret |= CONVERT_STAT_BITS_BIN;
 113         if (stats.crlf)
 114                 ret |= CONVERT_STAT_BITS_TXT_CRLF;
 115         if (stats.lonelf)
 116                 ret |=  CONVERT_STAT_BITS_TXT_LF;
 117
 118         return ret;
 119 }
 120
 121 static const char *gather_convert_stats_ascii(const char *data, unsigned long size)
 122 {
 123         unsigned int convert_stats = gather_convert_stats(data, size);
 124
 125         if (convert_stats & CONVERT_STAT_BITS_BIN)
 126                 return "-text";
 127         switch (convert_stats) {
 128         case CONVERT_STAT_BITS_TXT_LF:
 129                 return "lf";
 130         case CONVERT_STAT_BITS_TXT_CRLF:
 131                 return "crlf";
 132         case CONVERT_STAT_BITS_TXT_LF | CONVERT_STAT_BITS_TXT_CRLF:
 133                 return "mixed";
 134         default:
 135                 return "none";
 136         }
 137 }
 138
 139 const char *get_cached_convert_stats_ascii(const struct index_state *istate,
 140                                            const char *path)
 141 {
 142         const char *ret;
 143         unsigned long sz;
 144         void *data = read_blob_data_from_index(istate, path, &sz);
 145         ret = gather_convert_stats_ascii(data, sz);
 146         free(data);
 147         return ret;
 148 }
 149
 150 const char *get_wt_convert_stats_ascii(const char *path)
 151 {
 152         const char *ret = "";
 153         struct strbuf sb = STRBUF_INIT;
 154         if (strbuf_read_file(&sb, path, 0) >= 0)
 155                 ret = gather_convert_stats_ascii(sb.buf, sb.len);
 156         strbuf_release(&sb);
 157         return ret;
 158 }
 159
 160 static int text_eol_is_crlf(void)
 161 {
 162         if (auto_crlf == AUTO_CRLF_TRUE)
 163                 return 1;
 164         else if (auto_crlf == AUTO_CRLF_INPUT)
 165                 return 0;
 166         if (core_eol == EOL_CRLF)
 167                 return 1;
 168         if (core_eol == EOL_UNSET && EOL_NATIVE == EOL_CRLF)
 169                 return 1;
 170         return 0;
 171 }
 172
 173 static enum eol output_eol(enum crlf_action crlf_action)
 174 {
 175         switch (crlf_action) {
 176         case CRLF_BINARY:
 177                 return EOL_UNSET;
 178         case CRLF_TEXT_CRLF:
 179                 return EOL_CRLF;
 180         case CRLF_TEXT_INPUT:
 181                 return EOL_LF;
 182         case CRLF_UNDEFINED:
 183         case CRLF_AUTO_CRLF:
 184                 return EOL_CRLF;
 185         case CRLF_AUTO_INPUT:
 186                 return EOL_LF;
 187         case CRLF_TEXT:
 188         case CRLF_AUTO:
 189                 /* fall through */
 190                 return text_eol_is_crlf() ? EOL_CRLF : EOL_LF;
 191         }
 192         warning("Illegal crlf_action %d\n", (int)crlf_action);
 193         return core_eol;
 194 }
 195
 196 static void check_safe_crlf(const char *path, enum crlf_action crlf_action,
 197                             struct text_stat *old_stats, struct text_stat *new_stats,
 198                             enum safe_crlf checksafe)
 199 {
 200         if (old_stats->crlf && !new_stats->crlf ) {
 201                 /*
 202                  * CRLFs would not be restored by checkout
 203                  */
 204                 if (checksafe == SAFE_CRLF_WARN)
 205                         warning(_("CRLF will be replaced by LF in %s.\n"
 206                                   "The file will have its original line"
 207                                   " endings in your working directory."), path);
 208                 else /* i.e. SAFE_CRLF_FAIL */
 209                         die(_("CRLF would be replaced by LF in %s."), path);
 210         } else if (old_stats->lonelf && !new_stats->lonelf ) {
 211                 /*
 212                  * CRLFs would be added by checkout
 213                  */
 214                 if (checksafe == SAFE_CRLF_WARN)
 215                         warning(_("LF will be replaced by CRLF in %s.\n"
 216                                   "The file will have its original line"
 217                                   " endings in your working directory."), path);
 218                 else /* i.e. SAFE_CRLF_FAIL */
 219                         die(_("LF would be replaced by CRLF in %s"), path);
 220         }
 221 }
 222
 223 static int has_cr_in_index(const struct index_state *istate, const char *path)
 224 {
 225         unsigned long sz;
 226         void *data;
 227         int has_cr;
 228
 229         data = read_blob_data_from_index(istate, path, &sz);
 230         if (!data)
 231                 return 0;
 232         has_cr = memchr(data, '\r', sz) != NULL;
 233         free(data);
 234         return has_cr;
 235 }
 236
 237 static int will_convert_lf_to_crlf(size_t len, struct text_stat *stats,
 238                                    enum crlf_action crlf_action)
 239 {
 240         if (output_eol(crlf_action) != EOL_CRLF)
 241                 return 0;
 242         /* No "naked" LF? Nothing to convert, regardless. */
 243         if (!stats->lonelf)
 244                 return 0;
 245
 246         if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
 247                 /* If we have any CR or CRLF line endings, we do not touch it */
 248                 /* This is the new safer autocrlf-handling */
 249                 if (stats->lonecr || stats->crlf)
 250                         return 0;
 251
 252                 if (convert_is_binary(len, stats))
 253                         return 0;
 254         }
 255         return 1;
 256
 257 }
 258
 259 static int crlf_to_git(const struct index_state *istate,
 260                        const char *path, const char *src, size_t len,
 261                        struct strbuf *buf,
 262                        enum crlf_action crlf_action, enum safe_crlf checksafe)
 263 {
 264         struct text_stat stats;
 265         char *dst;
 266         int convert_crlf_into_lf;
 267
 268         if (crlf_action == CRLF_BINARY ||
 269             (src && !len))
 270                 return 0;
 271
 272         /*
 273          * If we are doing a dry-run and have no source buffer, there is
 274          * nothing to analyze; we must assume we would convert.
 275          */
 276         if (!buf && !src)
 277                 return 1;
 278
 279         gather_stats(src, len, &stats);
 280         /* Optimization: No CRLF? Nothing to convert, regardless. */
 281         convert_crlf_into_lf = !!stats.crlf;
 282
 283         if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
 284                 if (convert_is_binary(len, &stats))
 285                         return 0;
 286                 /*
 287                  * If the file in the index has any CR in it, do not
 288                  * convert.  This is the new safer autocrlf handling,
 289                  * unless we want to renormalize in a merge or
 290                  * cherry-pick.
 291                  */
 292                 if ((checksafe != SAFE_CRLF_RENORMALIZE) &&
 293                     has_cr_in_index(istate, path))
 294                         convert_crlf_into_lf = 0;
 295         }
 296         if ((checksafe == SAFE_CRLF_WARN ||
 297             (checksafe == SAFE_CRLF_FAIL)) && len) {
 298                 struct text_stat new_stats;
 299                 memcpy(&new_stats, &stats, sizeof(new_stats));
 300                 /* simulate "git add" */
 301                 if (convert_crlf_into_lf) {
 302                         new_stats.lonelf += new_stats.crlf;
 303                         new_stats.crlf = 0;
 304                 }
 305                 /* simulate "git checkout" */
 306                 if (will_convert_lf_to_crlf(len, &new_stats, crlf_action)) {
 307                         new_stats.crlf += new_stats.lonelf;
 308                         new_stats.lonelf = 0;
 309                 }
 310                 check_safe_crlf(path, crlf_action, &stats, &new_stats, checksafe);
 311         }
 312         if (!convert_crlf_into_lf)
 313                 return 0;
 314
 315         /*
 316          * At this point all of our source analysis is done, and we are sure we
 317          * would convert. If we are in dry-run mode, we can give an answer.
 318          */
 319         if (!buf)
 320                 return 1;
 321
 322         /* only grow if not in place */
 323         if (strbuf_avail(buf) + buf->len < len)
 324                 strbuf_grow(buf, len - buf->len);
 325         dst = buf->buf;
 326         if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
 327                 /*
 328                  * If we guessed, we already know we rejected a file with
 329                  * lone CR, and we can strip a CR without looking at what
 330                  * follow it.
 331                  */
 332                 do {
 333                         unsigned char c = *src++;
 334                         if (c != '\r')
 335                                 *dst++ = c;
 336                 } while (--len);
 337         } else {
 338                 do {
 339                         unsigned char c = *src++;
 340                         if (! (c == '\r' && (1 < len && *src == '\n')))
 341                                 *dst++ = c;
 342                 } while (--len);
 343         }
 344         strbuf_setlen(buf, dst - buf->buf);
 345         return 1;
 346 }
 347
 348 static int crlf_to_worktree(const char *path, const char *src, size_t len,
 349                             struct strbuf *buf, enum crlf_action crlf_action)
 350 {
 351         char *to_free = NULL;
 352         struct text_stat stats;
 353
 354         if (!len || output_eol(crlf_action) != EOL_CRLF)
 355                 return 0;
 356
 357         gather_stats(src, len, &stats);
 358         if (!will_convert_lf_to_crlf(len, &stats, crlf_action))
 359                 return 0;
 360
 361         /* are we "faking" in place editing ? */
 362         if (src == buf->buf)
 363                 to_free = strbuf_detach(buf, NULL);
 364
 365         strbuf_grow(buf, len + stats.lonelf);
 366         for (;;) {
 367                 const char *nl = memchr(src, '\n', len);
 368                 if (!nl)
 369                         break;
 370                 if (nl > src && nl[-1] == '\r') {
 371                         strbuf_add(buf, src, nl + 1 - src);
 372                 } else {
 373                         strbuf_add(buf, src, nl - src);
 374                         strbuf_addstr(buf, "\r\n");
 375                 }
 376                 len -= nl + 1 - src;
 377                 src  = nl + 1;
 378         }
 379         strbuf_add(buf, src, len);
 380
 381         free(to_free);
 382         return 1;
 383 }
 384
 385 struct filter_params {
 386         const char *src;
 387         unsigned long size;
 388         int fd;
 389         const char *cmd;
 390         const char *path;
 391 };
 392
 393 static int filter_buffer_or_fd(int in, int out, void *data)
 394 {
 395         /*
 396          * Spawn cmd and feed the buffer contents through its stdin.
 397          */
 398         struct child_process child_process = CHILD_PROCESS_INIT;
 399         struct filter_params *params = (struct filter_params *)data;
 400         int write_err, status;
 401         const char *argv[] = { NULL, NULL };
 402
 403         /* apply % substitution to cmd */
 404         struct strbuf cmd = STRBUF_INIT;
 405         struct strbuf path = STRBUF_INIT;
 406         struct strbuf_expand_dict_entry dict[] = {
 407                 { "f", NULL, },
 408                 { NULL, NULL, },
 409         };
 410
 411         /* quote the path to preserve spaces, etc. */
 412         sq_quote_buf(&path, params->path);
 413         dict[0].value = path.buf;
 414
 415         /* expand all %f with the quoted path */
 416         strbuf_expand(&cmd, params->cmd, strbuf_expand_dict_cb, &dict);
 417         strbuf_release(&path);
 418
 419         argv[0] = cmd.buf;
 420
 421         child_process.argv = argv;
 422         child_process.use_shell = 1;
 423         child_process.in = -1;
 424         child_process.out = out;
 425
 426         if (start_command(&child_process))
 427                 return error("cannot fork to run external filter '%s'", params->cmd);
 428
 429         sigchain_push(SIGPIPE, SIG_IGN);
 430
 431         if (params->src) {
 432                 write_err = (write_in_full(child_process.in,
 433                                            params->src, params->size) < 0);
 434                 if (errno == EPIPE)
 435                         write_err = 0;
 436         } else {
 437                 write_err = copy_fd(params->fd, child_process.in);
 438                 if (write_err == COPY_WRITE_ERROR && errno == EPIPE)
 439                         write_err = 0;
 440         }
 441
 442         if (close(child_process.in))
 443                 write_err = 1;
 444         if (write_err)
 445                 error("cannot feed the input to external filter '%s'", params->cmd);
 446
 447         sigchain_pop(SIGPIPE);
 448
 449         status = finish_command(&child_process);
 450         if (status)
 451                 error("external filter '%s' failed %d", params->cmd, status);
 452
 453         strbuf_release(&cmd);
 454         return (write_err || status);
 455 }
 456
 457 static int apply_single_file_filter(const char *path, const char *src, size_t len, int fd,
 458                         struct strbuf *dst, const char *cmd)
 459 {
 460         /*
 461          * Create a pipeline to have the command filter the buffer's
 462          * contents.
 463          *
 464          * (child --> cmd) --> us
 465          */
 466         int err = 0;
 467         struct strbuf nbuf = STRBUF_INIT;
 468         struct async async;
 469         struct filter_params params;
 470
 471         memset(&async, 0, sizeof(async));
 472         async.proc = filter_buffer_or_fd;
 473         async.data = &params;
 474         async.out = -1;
 475         params.src = src;
 476         params.size = len;
 477         params.fd = fd;
 478         params.cmd = cmd;
 479         params.path = path;
 480
 481         fflush(NULL);
 482         if (start_async(&async))
 483                 return 0;       /* error was already reported */
 484
 485         if (strbuf_read(&nbuf, async.out, len) < 0) {
 486                 err = error("read from external filter '%s' failed", cmd);
 487         }
 488         if (close(async.out)) {
 489                 err = error("read from external filter '%s' failed", cmd);
 490         }
 491         if (finish_async(&async)) {
 492                 err = error("external filter '%s' failed", cmd);
 493         }
 494
 495         if (!err) {
 496                 strbuf_swap(dst, &nbuf);
 497         }
 498         strbuf_release(&nbuf);
 499         return !err;
 500 }
 501
 502 #define CAP_CLEAN    (1u<<0)
 503 #define CAP_SMUDGE   (1u<<1)
 504 #define CAP_DELAY    (1u<<2)
 505
 506 struct cmd2process {
 507         struct subprocess_entry subprocess; /* must be the first member! */
 508         unsigned int supported_capabilities;
 509 };
 510
 511 static int subprocess_map_initialized;
 512 static struct hashmap subprocess_map;
 513
 514 static int start_multi_file_filter_fn(struct subprocess_entry *subprocess)
 515 {
 516         int err, i;
 517         struct cmd2process *entry = (struct cmd2process *)subprocess;
 518         struct string_list cap_list = STRING_LIST_INIT_NODUP;
 519         char *cap_buf;
 520         const char *cap_name;
 521         struct child_process *process = &subprocess->process;
 522         const char *cmd = subprocess->cmd;
 523
 524         static const struct {
 525                 const char *name;
 526                 unsigned int cap;
 527         } known_caps[] = {
 528                 { "clean",  CAP_CLEAN  },
 529                 { "smudge", CAP_SMUDGE },
 530                 { "delay",  CAP_DELAY  },
 531         };
 532
 533         sigchain_push(SIGPIPE, SIG_IGN);
 534
 535         err = packet_writel(process->in, "git-filter-client", "version=2", NULL);
 536         if (err)
 537                 goto done;
 538
 539         err = strcmp(packet_read_line(process->out, NULL), "git-filter-server");
 540         if (err) {
 541                 error("external filter '%s' does not support filter protocol version 2", cmd);
 542                 goto done;
 543         }
 544         err = strcmp(packet_read_line(process->out, NULL), "version=2");
 545         if (err)
 546                 goto done;
 547         err = packet_read_line(process->out, NULL) != NULL;
 548         if (err)
 549                 goto done;
 550
 551         for (i = 0; i < ARRAY_SIZE(known_caps); ++i) {
 552                 err = packet_write_fmt_gently(
 553                         process->in, "capability=%s\n", known_caps[i].name);
 554                 if (err)
 555                         goto done;
 556         }
 557         err = packet_flush_gently(process->in);
 558         if (err)
 559                 goto done;
 560
 561         for (;;) {
 562                 cap_buf = packet_read_line(process->out, NULL);
 563                 if (!cap_buf)
 564                         break;
 565                 string_list_split_in_place(&cap_list, cap_buf, '=', 1);
 566
 567                 if (cap_list.nr != 2 || strcmp(cap_list.items[0].string, "capability"))
 568                         continue;
 569
 570                 cap_name = cap_list.items[1].string;
 571                 i = ARRAY_SIZE(known_caps) - 1;
 572                 while (i >= 0 && strcmp(cap_name, known_caps[i].name))
 573                         i--;
 574
 575                 if (i >= 0)
 576                         entry->supported_capabilities |= known_caps[i].cap;
 577                 else
 578                         warning("external filter '%s' requested unsupported filter capability '%s'",
 579                         cmd, cap_name);
 580
 581                 string_list_clear(&cap_list, 0);
 582         }
 583
 584 done:
 585         sigchain_pop(SIGPIPE);
 586
 587         return err;
 588 }
 589
 590 static void handle_filter_error(const struct strbuf *filter_status,
 591                                 struct cmd2process *entry,
 592                                 const unsigned int wanted_capability) {
 593         if (!strcmp(filter_status->buf, "error"))
 594                 ; /* The filter signaled a problem with the file. */
 595         else if (!strcmp(filter_status->buf, "abort") && wanted_capability) {
 596                 /*
 597                  * The filter signaled a permanent problem. Don't try to filter
 598                  * files with the same command for the lifetime of the current
 599                  * Git process.
 600                  */
 601                  entry->supported_capabilities &= ~wanted_capability;
 602         } else {
 603                 /*
 604                  * Something went wrong with the protocol filter.
 605                  * Force shutdown and restart if another blob requires filtering.
 606                  */
 607                 error("external filter '%s' failed", entry->subprocess.cmd);
 608                 subprocess_stop(&subprocess_map, &entry->subprocess);
 609                 free(entry);
 610         }
 611 }
 612
 613 static int apply_multi_file_filter(const char *path, const char *src, size_t len,
 614                                    int fd, struct strbuf *dst, const char *cmd,
 615                                    const unsigned int wanted_capability,
 616                                    struct delayed_checkout *dco)
 617 {
 618         int err;
 619         int can_delay = 0;
 620         struct cmd2process *entry;
 621         struct child_process *process;
 622         struct strbuf nbuf = STRBUF_INIT;
 623         struct strbuf filter_status = STRBUF_INIT;
 624         const char *filter_type;
 625
 626         if (!subprocess_map_initialized) {
 627                 subprocess_map_initialized = 1;
 628                 hashmap_init(&subprocess_map, (hashmap_cmp_fn) cmd2process_cmp,
 629                              NULL, 0);
 630                 entry = NULL;
 631         } else {
 632                 entry = (struct cmd2process *)subprocess_find_entry(&subprocess_map, cmd);
 633         }
 634
 635         fflush(NULL);
 636
 637         if (!entry) {
 638                 entry = xmalloc(sizeof(*entry));
 639                 entry->supported_capabilities = 0;
 640
 641                 if (subprocess_start(&subprocess_map, &entry->subprocess, cmd, start_multi_file_filter_fn)) {
 642                         free(entry);
 643                         return 0;
 644                 }
 645         }
 646         process = &entry->subprocess.process;
 647
 648         if (!(entry->supported_capabilities & wanted_capability))
 649                 return 0;
 650
 651         if (wanted_capability & CAP_CLEAN)
 652                 filter_type = "clean";
 653         else if (wanted_capability & CAP_SMUDGE)
 654                 filter_type = "smudge";
 655         else
 656                 die("unexpected filter type");
 657
 658         sigchain_push(SIGPIPE, SIG_IGN);
 659
 660         assert(strlen(filter_type) < LARGE_PACKET_DATA_MAX - strlen("command=\n"));
 661         err = packet_write_fmt_gently(process->in, "command=%s\n", filter_type);
 662         if (err)
 663                 goto done;
 664
 665         err = strlen(path) > LARGE_PACKET_DATA_MAX - strlen("pathname=\n");
 666         if (err) {
 667                 error("path name too long for external filter");
 668                 goto done;
 669         }
 670
 671         err = packet_write_fmt_gently(process->in, "pathname=%s\n", path);
 672         if (err)
 673                 goto done;
 674
 675         if ((entry->supported_capabilities & CAP_DELAY) &&
 676             dco && dco->state == CE_CAN_DELAY) {
 677                 can_delay = 1;
 678                 err = packet_write_fmt_gently(process->in, "can-delay=1\n");
 679                 if (err)
 680                         goto done;
 681         }
 682
 683         err = packet_flush_gently(process->in);
 684         if (err)
 685                 goto done;
 686
 687         if (fd >= 0)
 688                 err = write_packetized_from_fd(fd, process->in);
 689         else
 690                 err = write_packetized_from_buf(src, len, process->in);
 691         if (err)
 692                 goto done;
 693
 694         err = subprocess_read_status(process->out, &filter_status);
 695         if (err)
 696                 goto done;
 697
 698         if (can_delay && !strcmp(filter_status.buf, "delayed")) {
 699                 string_list_insert(&dco->filters, cmd);
 700                 string_list_insert(&dco->paths, path);
 701         } else {
 702                 /* The filter got the blob and wants to send us a response. */
 703                 err = strcmp(filter_status.buf, "success");
 704                 if (err)
 705                         goto done;
 706
 707                 err = read_packetized_to_strbuf(process->out, &nbuf) < 0;
 708                 if (err)
 709                         goto done;
 710
 711                 err = subprocess_read_status(process->out, &filter_status);
 712                 if (err)
 713                         goto done;
 714
 715                 err = strcmp(filter_status.buf, "success");
 716         }
 717
 718 done:
 719         sigchain_pop(SIGPIPE);
 720
 721         if (err)
 722                 handle_filter_error(&filter_status, entry, wanted_capability);
 723         else
 724                 strbuf_swap(dst, &nbuf);
 725         strbuf_release(&nbuf);
 726         return !err;
 727 }
 728
 729
 730 int async_query_available_blobs(const char *cmd, struct string_list *available_paths)
 731 {
 732         int err;
 733         char *line;
 734         struct cmd2process *entry;
 735         struct child_process *process;
 736         struct strbuf filter_status = STRBUF_INIT;
 737
 738         assert(subprocess_map_initialized);
 739         entry = (struct cmd2process *)subprocess_find_entry(&subprocess_map, cmd);
 740         if (!entry) {
 741                 error("external filter '%s' is not available anymore although "
 742                       "not all paths have been filtered", cmd);
 743                 return 0;
 744         }
 745         process = &entry->subprocess.process;
 746         sigchain_push(SIGPIPE, SIG_IGN);
 747
 748         err = packet_write_fmt_gently(
 749                 process->in, "command=list_available_blobs\n");
 750         if (err)
 751                 goto done;
 752
 753         err = packet_flush_gently(process->in);
 754         if (err)
 755                 goto done;
 756
 757         while ((line = packet_read_line(process->out, NULL))) {
 758                 const char *path;
 759                 if (skip_prefix(line, "pathname=", &path))
 760                         string_list_insert(available_paths, xstrdup(path));
 761                 else
 762                         ; /* ignore unknown keys */
 763         }
 764
 765         err = subprocess_read_status(process->out, &filter_status);
 766         if (err)
 767                 goto done;
 768
 769         err = strcmp(filter_status.buf, "success");
 770
 771 done:
 772         sigchain_pop(SIGPIPE);
 773
 774         if (err)
 775                 handle_filter_error(&filter_status, entry, 0);
 776         return !err;
 777 }
 778
 779 static struct convert_driver {
 780         const char *name;
 781         struct convert_driver *next;
 782         const char *smudge;
 783         const char *clean;
 784         const char *process;
 785         int required;
 786 } *user_convert, **user_convert_tail;
 787
 788 static int apply_filter(const char *path, const char *src, size_t len,
 789                         int fd, struct strbuf *dst, struct convert_driver *drv,
 790                         const unsigned int wanted_capability,
 791                         struct delayed_checkout *dco)
 792 {
 793         const char *cmd = NULL;
 794
 795         if (!drv)
 796                 return 0;
 797
 798         if (!dst)
 799                 return 1;
 800
 801         if ((wanted_capability & CAP_CLEAN) && !drv->process && drv->clean)
 802                 cmd = drv->clean;
 803         else if ((wanted_capability & CAP_SMUDGE) && !drv->process && drv->smudge)
 804                 cmd = drv->smudge;
 805
 806         if (cmd && *cmd)
 807                 return apply_single_file_filter(path, src, len, fd, dst, cmd);
 808         else if (drv->process && *drv->process)
 809                 return apply_multi_file_filter(path, src, len, fd, dst,
 810                         drv->process, wanted_capability, dco);
 811
 812         return 0;
 813 }
 814
 815 static int read_convert_config(const char *var, const char *value, void *cb)
 816 {
 817         const char *key, *name;
 818         int namelen;
 819         struct convert_driver *drv;
 820
 821         /*
 822          * External conversion drivers are configured using
 823          * "filter.<name>.variable".
 824          */
 825         if (parse_config_key(var, "filter", &name, &namelen, &key) < 0 || !name)
 826                 return 0;
 827         for (drv = user_convert; drv; drv = drv->next)
 828                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
 829                         break;
 830         if (!drv) {
 831                 drv = xcalloc(1, sizeof(struct convert_driver));
 832                 drv->name = xmemdupz(name, namelen);
 833                 *user_convert_tail = drv;
 834                 user_convert_tail = &(drv->next);
 835         }
 836
 837         /*
 838          * filter.<name>.smudge and filter.<name>.clean specifies
 839          * the command line:
 840          *
 841          *      command-line
 842          *
 843          * The command-line will not be interpolated in any way.
 844          */
 845
 846         if (!strcmp("smudge", key))
 847                 return git_config_string(&drv->smudge, var, value);
 848
 849         if (!strcmp("clean", key))
 850                 return git_config_string(&drv->clean, var, value);
 851
 852         if (!strcmp("process", key))
 853                 return git_config_string(&drv->process, var, value);
 854
 855         if (!strcmp("required", key)) {
 856                 drv->required = git_config_bool(var, value);
 857                 return 0;
 858         }
 859
 860         return 0;
 861 }
 862
 863 static int count_ident(const char *cp, unsigned long size)
 864 {
 865         /*
 866          * "$Id: 0000000000000000000000000000000000000000 $" <=> "$Id$"
 867          */
 868         int cnt = 0;
 869         char ch;
 870
 871         while (size) {
 872                 ch = *cp++;
 873                 size--;
 874                 if (ch != '$')
 875                         continue;
 876                 if (size < 3)
 877                         break;
 878                 if (memcmp("Id", cp, 2))
 879                         continue;
 880                 ch = cp[2];
 881                 cp += 3;
 882                 size -= 3;
 883                 if (ch == '$')
 884                         cnt++; /* $Id$ */
 885                 if (ch != ':')
 886                         continue;
 887
 888                 /*
 889                  * "$Id: ... "; scan up to the closing dollar sign and discard.
 890                  */
 891                 while (size) {
 892                         ch = *cp++;
 893                         size--;
 894                         if (ch == '$') {
 895                                 cnt++;
 896                                 break;
 897                         }
 898                         if (ch == '\n')
 899                                 break;
 900                 }
 901         }
 902         return cnt;
 903 }
 904
 905 static int ident_to_git(const char *path, const char *src, size_t len,
 906                         struct strbuf *buf, int ident)
 907 {
 908         char *dst, *dollar;
 909
 910         if (!ident || (src && !count_ident(src, len)))
 911                 return 0;
 912
 913         if (!buf)
 914                 return 1;
 915
 916         /* only grow if not in place */
 917         if (strbuf_avail(buf) + buf->len < len)
 918                 strbuf_grow(buf, len - buf->len);
 919         dst = buf->buf;
 920         for (;;) {
 921                 dollar = memchr(src, '$', len);
 922                 if (!dollar)
 923                         break;
 924                 memmove(dst, src, dollar + 1 - src);
 925                 dst += dollar + 1 - src;
 926                 len -= dollar + 1 - src;
 927                 src  = dollar + 1;
 928
 929                 if (len > 3 && !memcmp(src, "Id:", 3)) {
 930                         dollar = memchr(src + 3, '$', len - 3);
 931                         if (!dollar)
 932                                 break;
 933                         if (memchr(src + 3, '\n', dollar - src - 3)) {
 934                                 /* Line break before the next dollar. */
 935                                 continue;
 936                         }
 937
 938                         memcpy(dst, "Id$", 3);
 939                         dst += 3;
 940                         len -= dollar + 1 - src;
 941                         src  = dollar + 1;
 942                 }
 943         }
 944         memmove(dst, src, len);
 945         strbuf_setlen(buf, dst + len - buf->buf);
 946         return 1;
 947 }
 948
 949 static int ident_to_worktree(const char *path, const char *src, size_t len,
 950                              struct strbuf *buf, int ident)
 951 {
 952         unsigned char sha1[20];
 953         char *to_free = NULL, *dollar, *spc;
 954         int cnt;
 955
 956         if (!ident)
 957                 return 0;
 958
 959         cnt = count_ident(src, len);
 960         if (!cnt)
 961                 return 0;
 962
 963         /* are we "faking" in place editing ? */
 964         if (src == buf->buf)
 965                 to_free = strbuf_detach(buf, NULL);
 966         hash_sha1_file(src, len, "blob", sha1);
 967
 968         strbuf_grow(buf, len + cnt * 43);
 969         for (;;) {
 970                 /* step 1: run to the next '$' */
 971                 dollar = memchr(src, '$', len);
 972                 if (!dollar)
 973                         break;
 974                 strbuf_add(buf, src, dollar + 1 - src);
 975                 len -= dollar + 1 - src;
 976                 src  = dollar + 1;
 977
 978                 /* step 2: does it looks like a bit like Id:xxx$ or Id$ ? */
 979                 if (len < 3 || memcmp("Id", src, 2))
 980                         continue;
 981
 982                 /* step 3: skip over Id$ or Id:xxxxx$ */
 983                 if (src[2] == '$') {
 984                         src += 3;
 985                         len -= 3;
 986                 } else if (src[2] == ':') {
 987                         /*
 988                          * It's possible that an expanded Id has crept its way into the
 989                          * repository, we cope with that by stripping the expansion out.
 990                          * This is probably not a good idea, since it will cause changes
 991                          * on checkout, which won't go away by stash, but let's keep it
 992                          * for git-style ids.
 993                          */
 994                         dollar = memchr(src + 3, '$', len - 3);
 995                         if (!dollar) {
 996                                 /* incomplete keyword, no more '$', so just quit the loop */
 997                                 break;
 998                         }
 999
1000                         if (memchr(src + 3, '\n', dollar - src - 3)) {
1001                                 /* Line break before the next dollar. */
1002                                 continue;
1003                         }
1004
1005                         spc = memchr(src + 4, ' ', dollar - src - 4);
1006                         if (spc && spc < dollar-1) {
1007                                 /* There are spaces in unexpected places.
1008                                  * This is probably an id from some other
1009                                  * versioning system. Keep it for now.
1010                                  */
1011                                 continue;
1012                         }
1013
1014                         len -= dollar + 1 - src;
1015                         src  = dollar + 1;
1016                 } else {
1017                         /* it wasn't a "Id$" or "Id:xxxx$" */
1018                         continue;
1019                 }
1020
1021                 /* step 4: substitute */
1022                 strbuf_addstr(buf, "Id: ");
1023                 strbuf_add(buf, sha1_to_hex(sha1), 40);
1024                 strbuf_addstr(buf, " $");
1025         }
1026         strbuf_add(buf, src, len);
1027
1028         free(to_free);
1029         return 1;
1030 }
1031
1032 static enum crlf_action git_path_check_crlf(struct attr_check_item *check)
1033 {
1034         const char *value = check->value;
1035
1036         if (ATTR_TRUE(value))
1037                 return CRLF_TEXT;
1038         else if (ATTR_FALSE(value))
1039                 return CRLF_BINARY;
1040         else if (ATTR_UNSET(value))
1041                 ;
1042         else if (!strcmp(value, "input"))
1043                 return CRLF_TEXT_INPUT;
1044         else if (!strcmp(value, "auto"))
1045                 return CRLF_AUTO;
1046         return CRLF_UNDEFINED;
1047 }
1048
1049 static enum eol git_path_check_eol(struct attr_check_item *check)
1050 {
1051         const char *value = check->value;
1052
1053         if (ATTR_UNSET(value))
1054                 ;
1055         else if (!strcmp(value, "lf"))
1056                 return EOL_LF;
1057         else if (!strcmp(value, "crlf"))
1058                 return EOL_CRLF;
1059         return EOL_UNSET;
1060 }
1061
1062 static struct convert_driver *git_path_check_convert(struct attr_check_item *check)
1063 {
1064         const char *value = check->value;
1065         struct convert_driver *drv;
1066
1067         if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1068                 return NULL;
1069         for (drv = user_convert; drv; drv = drv->next)
1070                 if (!strcmp(value, drv->name))
1071                         return drv;
1072         return NULL;
1073 }
1074
1075 static int git_path_check_ident(struct attr_check_item *check)
1076 {
1077         const char *value = check->value;
1078
1079         return !!ATTR_TRUE(value);
1080 }
1081
1082 struct conv_attrs {
1083         struct convert_driver *drv;
1084         enum crlf_action attr_action; /* What attr says */
1085         enum crlf_action crlf_action; /* When no attr is set, use core.autocrlf */
1086         int ident;
1087 };
1088
1089 static void convert_attrs(struct conv_attrs *ca, const char *path)
1090 {
1091         static struct attr_check *check;
1092
1093         if (!check) {
1094                 check = attr_check_initl("crlf", "ident", "filter",
1095                                          "eol", "text", NULL);
1096                 user_convert_tail = &user_convert;
1097                 git_config(read_convert_config, NULL);
1098         }
1099
1100         if (!git_check_attr(path, check)) {
1101                 struct attr_check_item *ccheck = check->items;
1102                 ca->crlf_action = git_path_check_crlf(ccheck + 4);
1103                 if (ca->crlf_action == CRLF_UNDEFINED)
1104                         ca->crlf_action = git_path_check_crlf(ccheck + 0);
1105                 ca->attr_action = ca->crlf_action;
1106                 ca->ident = git_path_check_ident(ccheck + 1);
1107                 ca->drv = git_path_check_convert(ccheck + 2);
1108                 if (ca->crlf_action != CRLF_BINARY) {
1109                         enum eol eol_attr = git_path_check_eol(ccheck + 3);
1110                         if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_LF)
1111                                 ca->crlf_action = CRLF_AUTO_INPUT;
1112                         else if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_CRLF)
1113                                 ca->crlf_action = CRLF_AUTO_CRLF;
1114                         else if (eol_attr == EOL_LF)
1115                                 ca->crlf_action = CRLF_TEXT_INPUT;
1116                         else if (eol_attr == EOL_CRLF)
1117                                 ca->crlf_action = CRLF_TEXT_CRLF;
1118                 }
1119                 ca->attr_action = ca->crlf_action;
1120         } else {
1121                 ca->drv = NULL;
1122                 ca->crlf_action = CRLF_UNDEFINED;
1123                 ca->ident = 0;
1124         }
1125         if (ca->crlf_action == CRLF_TEXT)
1126                 ca->crlf_action = text_eol_is_crlf() ? CRLF_TEXT_CRLF : CRLF_TEXT_INPUT;
1127         if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_FALSE)
1128                 ca->crlf_action = CRLF_BINARY;
1129         if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_TRUE)
1130                 ca->crlf_action = CRLF_AUTO_CRLF;
1131         if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_INPUT)
1132                 ca->crlf_action = CRLF_AUTO_INPUT;
1133 }
1134
1135 int would_convert_to_git_filter_fd(const char *path)
1136 {
1137         struct conv_attrs ca;
1138
1139         convert_attrs(&ca, path);
1140         if (!ca.drv)
1141                 return 0;
1142
1143         /*
1144          * Apply a filter to an fd only if the filter is required to succeed.
1145          * We must die if the filter fails, because the original data before
1146          * filtering is not available.
1147          */
1148         if (!ca.drv->required)
1149                 return 0;
1150
1151         return apply_filter(path, NULL, 0, -1, NULL, ca.drv, CAP_CLEAN, NULL);
1152 }
1153
1154 const char *get_convert_attr_ascii(const char *path)
1155 {
1156         struct conv_attrs ca;
1157
1158         convert_attrs(&ca, path);
1159         switch (ca.attr_action) {
1160         case CRLF_UNDEFINED:
1161                 return "";
1162         case CRLF_BINARY:
1163                 return "-text";
1164         case CRLF_TEXT:
1165                 return "text";
1166         case CRLF_TEXT_INPUT:
1167                 return "text eol=lf";
1168         case CRLF_TEXT_CRLF:
1169                 return "text eol=crlf";
1170         case CRLF_AUTO:
1171                 return "text=auto";
1172         case CRLF_AUTO_CRLF:
1173                 return "text=auto eol=crlf";
1174         case CRLF_AUTO_INPUT:
1175                 return "text=auto eol=lf";
1176         }
1177         return "";
1178 }
1179
1180 int convert_to_git(const struct index_state *istate,
1181                    const char *path, const char *src, size_t len,
1182                    struct strbuf *dst, enum safe_crlf checksafe)
1183 {
1184         int ret = 0;
1185         struct conv_attrs ca;
1186
1187         convert_attrs(&ca, path);
1188
1189         ret |= apply_filter(path, src, len, -1, dst, ca.drv, CAP_CLEAN, NULL);
1190         if (!ret && ca.drv && ca.drv->required)
1191                 die("%s: clean filter '%s' failed", path, ca.drv->name);
1192
1193         if (ret && dst) {
1194                 src = dst->buf;
1195                 len = dst->len;
1196         }
1197         ret |= crlf_to_git(istate, path, src, len, dst, ca.crlf_action, checksafe);
1198         if (ret && dst) {
1199                 src = dst->buf;
1200                 len = dst->len;
1201         }
1202         return ret | ident_to_git(path, src, len, dst, ca.ident);
1203 }
1204
1205 void convert_to_git_filter_fd(const struct index_state *istate,
1206                               const char *path, int fd, struct strbuf *dst,
1207                               enum safe_crlf checksafe)
1208 {
1209         struct conv_attrs ca;
1210         convert_attrs(&ca, path);
1211
1212         assert(ca.drv);
1213         assert(ca.drv->clean || ca.drv->process);
1214
1215         if (!apply_filter(path, NULL, 0, fd, dst, ca.drv, CAP_CLEAN, NULL))
1216                 die("%s: clean filter '%s' failed", path, ca.drv->name);
1217
1218         crlf_to_git(istate, path, dst->buf, dst->len, dst, ca.crlf_action, checksafe);
1219         ident_to_git(path, dst->buf, dst->len, dst, ca.ident);
1220 }
1221
1222 static int convert_to_working_tree_internal(const char *path, const char *src,
1223                                             size_t len, struct strbuf *dst,
1224                                             int normalizing, struct delayed_checkout *dco)
1225 {
1226         int ret = 0, ret_filter = 0;
1227         struct conv_attrs ca;
1228
1229         convert_attrs(&ca, path);
1230
1231         ret |= ident_to_worktree(path, src, len, dst, ca.ident);
1232         if (ret) {
1233                 src = dst->buf;
1234                 len = dst->len;
1235         }
1236         /*
1237          * CRLF conversion can be skipped if normalizing, unless there
1238          * is a smudge or process filter (even if the process filter doesn't
1239          * support smudge).  The filters might expect CRLFs.
1240          */
1241         if ((ca.drv && (ca.drv->smudge || ca.drv->process)) || !normalizing) {
1242                 ret |= crlf_to_worktree(path, src, len, dst, ca.crlf_action);
1243                 if (ret) {
1244                         src = dst->buf;
1245                         len = dst->len;
1246                 }
1247         }
1248
1249         ret_filter = apply_filter(
1250                 path, src, len, -1, dst, ca.drv, CAP_SMUDGE, dco);
1251         if (!ret_filter && ca.drv && ca.drv->required)
1252                 die("%s: smudge filter %s failed", path, ca.drv->name);
1253
1254         return ret | ret_filter;
1255 }
1256
1257 int async_convert_to_working_tree(const char *path, const char *src,
1258                                   size_t len, struct strbuf *dst,
1259                                   void *dco)
1260 {
1261         return convert_to_working_tree_internal(path, src, len, dst, 0, dco);
1262 }
1263
1264 int convert_to_working_tree(const char *path, const char *src, size_t len, struct strbuf *dst)
1265 {
1266         return convert_to_working_tree_internal(path, src, len, dst, 0, NULL);
1267 }
1268
1269 int renormalize_buffer(const struct index_state *istate, const char *path,
1270                        const char *src, size_t len, struct strbuf *dst)
1271 {
1272         int ret = convert_to_working_tree_internal(path, src, len, dst, 1, NULL);
1273         if (ret) {
1274                 src = dst->buf;
1275                 len = dst->len;
1276         }
1277         return ret | convert_to_git(istate, path, src, len, dst, SAFE_CRLF_RENORMALIZE);
1278 }
1279
1280 /*****************************************************************
1281  *
1282  * Streaming conversion support
1283  *
1284  *****************************************************************/
1285
1286 typedef int (*filter_fn)(struct stream_filter *,
1287                          const char *input, size_t *isize_p,
1288                          char *output, size_t *osize_p);
1289 typedef void (*free_fn)(struct stream_filter *);
1290
1291 struct stream_filter_vtbl {
1292         filter_fn filter;
1293         free_fn free;
1294 };
1295
1296 struct stream_filter {
1297         struct stream_filter_vtbl *vtbl;
1298 };
1299
1300 static int null_filter_fn(struct stream_filter *filter,
1301                           const char *input, size_t *isize_p,
1302                           char *output, size_t *osize_p)
1303 {
1304         size_t count;
1305
1306         if (!input)
1307                 return 0; /* we do not keep any states */
1308         count = *isize_p;
1309         if (*osize_p < count)
1310                 count = *osize_p;
1311         if (count) {
1312                 memmove(output, input, count);
1313                 *isize_p -= count;
1314                 *osize_p -= count;
1315         }
1316         return 0;
1317 }
1318
1319 static void null_free_fn(struct stream_filter *filter)
1320 {
1321         ; /* nothing -- null instances are shared */
1322 }
1323
1324 static struct stream_filter_vtbl null_vtbl = {
1325         null_filter_fn,
1326         null_free_fn,
1327 };
1328
1329 static struct stream_filter null_filter_singleton = {
1330         &null_vtbl,
1331 };
1332
1333 int is_null_stream_filter(struct stream_filter *filter)
1334 {
1335         return filter == &null_filter_singleton;
1336 }
1337
1338
1339 /*
1340  * LF-to-CRLF filter
1341  */
1342
1343 struct lf_to_crlf_filter {
1344         struct stream_filter filter;
1345         unsigned has_held:1;
1346         char held;
1347 };
1348
1349 static int lf_to_crlf_filter_fn(struct stream_filter *filter,
1350                                 const char *input, size_t *isize_p,
1351                                 char *output, size_t *osize_p)
1352 {
1353         size_t count, o = 0;
1354         struct lf_to_crlf_filter *lf_to_crlf = (struct lf_to_crlf_filter *)filter;
1355
1356         /*
1357          * We may be holding onto the CR to see if it is followed by a
1358          * LF, in which case we would need to go to the main loop.
1359          * Otherwise, just emit it to the output stream.
1360          */
1361         if (lf_to_crlf->has_held && (lf_to_crlf->held != '\r' || !input)) {
1362                 output[o++] = lf_to_crlf->held;
1363                 lf_to_crlf->has_held = 0;
1364         }
1365
1366         /* We are told to drain */
1367         if (!input) {
1368                 *osize_p -= o;
1369                 return 0;
1370         }
1371
1372         count = *isize_p;
1373         if (count || lf_to_crlf->has_held) {
1374                 size_t i;
1375                 int was_cr = 0;
1376
1377                 if (lf_to_crlf->has_held) {
1378                         was_cr = 1;
1379                         lf_to_crlf->has_held = 0;
1380                 }
1381
1382                 for (i = 0; o < *osize_p && i < count; i++) {
1383                         char ch = input[i];
1384
1385                         if (ch == '\n') {
1386                                 output[o++] = '\r';
1387                         } else if (was_cr) {
1388                                 /*
1389                                  * Previous round saw CR and it is not followed
1390                                  * by a LF; emit the CR before processing the
1391                                  * current character.
1392                                  */
1393                                 output[o++] = '\r';
1394                         }
1395
1396                         /*
1397                          * We may have consumed the last output slot,
1398                          * in which case we need to break out of this
1399                          * loop; hold the current character before
1400                          * returning.
1401                          */
1402                         if (*osize_p <= o) {
1403                                 lf_to_crlf->has_held = 1;
1404                                 lf_to_crlf->held = ch;
1405                                 continue; /* break but increment i */
1406                         }
1407
1408                         if (ch == '\r') {
1409                                 was_cr = 1;
1410                                 continue;
1411                         }
1412
1413                         was_cr = 0;
1414                         output[o++] = ch;
1415                 }
1416
1417                 *osize_p -= o;
1418                 *isize_p -= i;
1419
1420                 if (!lf_to_crlf->has_held && was_cr) {
1421                         lf_to_crlf->has_held = 1;
1422                         lf_to_crlf->held = '\r';
1423                 }
1424         }
1425         return 0;
1426 }
1427
1428 static void lf_to_crlf_free_fn(struct stream_filter *filter)
1429 {
1430         free(filter);
1431 }
1432
1433 static struct stream_filter_vtbl lf_to_crlf_vtbl = {
1434         lf_to_crlf_filter_fn,
1435         lf_to_crlf_free_fn,
1436 };
1437
1438 static struct stream_filter *lf_to_crlf_filter(void)
1439 {
1440         struct lf_to_crlf_filter *lf_to_crlf = xcalloc(1, sizeof(*lf_to_crlf));
1441
1442         lf_to_crlf->filter.vtbl = &lf_to_crlf_vtbl;
1443         return (struct stream_filter *)lf_to_crlf;
1444 }
1445
1446 /*
1447  * Cascade filter
1448  */
1449 #define FILTER_BUFFER 1024
1450 struct cascade_filter {
1451         struct stream_filter filter;
1452         struct stream_filter *one;
1453         struct stream_filter *two;
1454         char buf[FILTER_BUFFER];
1455         int end, ptr;
1456 };
1457
1458 static int cascade_filter_fn(struct stream_filter *filter,
1459                              const char *input, size_t *isize_p,
1460                              char *output, size_t *osize_p)
1461 {
1462         struct cascade_filter *cas = (struct cascade_filter *) filter;
1463         size_t filled = 0;
1464         size_t sz = *osize_p;
1465         size_t to_feed, remaining;
1466
1467         /*
1468          * input -- (one) --> buf -- (two) --> output
1469          */
1470         while (filled < sz) {
1471                 remaining = sz - filled;
1472
1473                 /* do we already have something to feed two with? */
1474                 if (cas->ptr < cas->end) {
1475                         to_feed = cas->end - cas->ptr;
1476                         if (stream_filter(cas->two,
1477                                           cas->buf + cas->ptr, &to_feed,
1478                                           output + filled, &remaining))
1479                                 return -1;
1480                         cas->ptr += (cas->end - cas->ptr) - to_feed;
1481                         filled = sz - remaining;
1482                         continue;
1483                 }
1484
1485                 /* feed one from upstream and have it emit into our buffer */
1486                 to_feed = input ? *isize_p : 0;
1487                 if (input && !to_feed)
1488                         break;
1489                 remaining = sizeof(cas->buf);
1490                 if (stream_filter(cas->one,
1491                                   input, &to_feed,
1492                                   cas->buf, &remaining))
1493                         return -1;
1494                 cas->end = sizeof(cas->buf) - remaining;
1495                 cas->ptr = 0;
1496                 if (input) {
1497                         size_t fed = *isize_p - to_feed;
1498                         *isize_p -= fed;
1499                         input += fed;
1500                 }
1501
1502                 /* do we know that we drained one completely? */
1503                 if (input || cas->end)
1504                         continue;
1505
1506                 /* tell two to drain; we have nothing more to give it */
1507                 to_feed = 0;
1508                 remaining = sz - filled;
1509                 if (stream_filter(cas->two,
1510                                   NULL, &to_feed,
1511                                   output + filled, &remaining))
1512                         return -1;
1513                 if (remaining == (sz - filled))
1514                         break; /* completely drained two */
1515                 filled = sz - remaining;
1516         }
1517         *osize_p -= filled;
1518         return 0;
1519 }
1520
1521 static void cascade_free_fn(struct stream_filter *filter)
1522 {
1523         struct cascade_filter *cas = (struct cascade_filter *)filter;
1524         free_stream_filter(cas->one);
1525         free_stream_filter(cas->two);
1526         free(filter);
1527 }
1528
1529 static struct stream_filter_vtbl cascade_vtbl = {
1530         cascade_filter_fn,
1531         cascade_free_fn,
1532 };
1533
1534 static struct stream_filter *cascade_filter(struct stream_filter *one,
1535                                             struct stream_filter *two)
1536 {
1537         struct cascade_filter *cascade;
1538
1539         if (!one || is_null_stream_filter(one))
1540                 return two;
1541         if (!two || is_null_stream_filter(two))
1542                 return one;
1543
1544         cascade = xmalloc(sizeof(*cascade));
1545         cascade->one = one;
1546         cascade->two = two;
1547         cascade->end = cascade->ptr = 0;
1548         cascade->filter.vtbl = &cascade_vtbl;
1549         return (struct stream_filter *)cascade;
1550 }
1551
1552 /*
1553  * ident filter
1554  */
1555 #define IDENT_DRAINING (-1)
1556 #define IDENT_SKIPPING (-2)
1557 struct ident_filter {
1558         struct stream_filter filter;
1559         struct strbuf left;
1560         int state;
1561         char ident[45]; /* ": x40 $" */
1562 };
1563
1564 static int is_foreign_ident(const char *str)
1565 {
1566         int i;
1567
1568         if (!skip_prefix(str, "$Id: ", &str))
1569                 return 0;
1570         for (i = 0; str[i]; i++) {
1571                 if (isspace(str[i]) && str[i+1] != '$')
1572                         return 1;
1573         }
1574         return 0;
1575 }
1576
1577 static void ident_drain(struct ident_filter *ident, char **output_p, size_t *osize_p)
1578 {
1579         size_t to_drain = ident->left.len;
1580
1581         if (*osize_p < to_drain)
1582                 to_drain = *osize_p;
1583         if (to_drain) {
1584                 memcpy(*output_p, ident->left.buf, to_drain);
1585                 strbuf_remove(&ident->left, 0, to_drain);
1586                 *output_p += to_drain;
1587                 *osize_p -= to_drain;
1588         }
1589         if (!ident->left.len)
1590                 ident->state = 0;
1591 }
1592
1593 static int ident_filter_fn(struct stream_filter *filter,
1594                            const char *input, size_t *isize_p,
1595                            char *output, size_t *osize_p)
1596 {
1597         struct ident_filter *ident = (struct ident_filter *)filter;
1598         static const char head[] = "$Id";
1599
1600         if (!input) {
1601                 /* drain upon eof */
1602                 switch (ident->state) {
1603                 default:
1604                         strbuf_add(&ident->left, head, ident->state);
1605                 case IDENT_SKIPPING:
1606                         /* fallthru */
1607                 case IDENT_DRAINING:
1608                         ident_drain(ident, &output, osize_p);
1609                 }
1610                 return 0;
1611         }
1612
1613         while (*isize_p || (ident->state == IDENT_DRAINING)) {
1614                 int ch;
1615
1616                 if (ident->state == IDENT_DRAINING) {
1617                         ident_drain(ident, &output, osize_p);
1618                         if (!*osize_p)
1619                                 break;
1620                         continue;
1621                 }
1622
1623                 ch = *(input++);
1624                 (*isize_p)--;
1625
1626                 if (ident->state == IDENT_SKIPPING) {
1627                         /*
1628                          * Skipping until '$' or LF, but keeping them
1629                          * in case it is a foreign ident.
1630                          */
1631                         strbuf_addch(&ident->left, ch);
1632                         if (ch != '\n' && ch != '$')
1633                                 continue;
1634                         if (ch == '$' && !is_foreign_ident(ident->left.buf)) {
1635                                 strbuf_setlen(&ident->left, sizeof(head) - 1);
1636                                 strbuf_addstr(&ident->left, ident->ident);
1637                         }
1638                         ident->state = IDENT_DRAINING;
1639                         continue;
1640                 }
1641
1642                 if (ident->state < sizeof(head) &&
1643                     head[ident->state] == ch) {
1644                         ident->state++;
1645                         continue;
1646                 }
1647
1648                 if (ident->state)
1649                         strbuf_add(&ident->left, head, ident->state);
1650                 if (ident->state == sizeof(head) - 1) {
1651                         if (ch != ':' && ch != '$') {
1652                                 strbuf_addch(&ident->left, ch);
1653                                 ident->state = 0;
1654                                 continue;
1655                         }
1656
1657                         if (ch == ':') {
1658                                 strbuf_addch(&ident->left, ch);
1659                                 ident->state = IDENT_SKIPPING;
1660                         } else {
1661                                 strbuf_addstr(&ident->left, ident->ident);
1662                                 ident->state = IDENT_DRAINING;
1663                         }
1664                         continue;
1665                 }
1666
1667                 strbuf_addch(&ident->left, ch);
1668                 ident->state = IDENT_DRAINING;
1669         }
1670         return 0;
1671 }
1672
1673 static void ident_free_fn(struct stream_filter *filter)
1674 {
1675         struct ident_filter *ident = (struct ident_filter *)filter;
1676         strbuf_release(&ident->left);
1677         free(filter);
1678 }
1679
1680 static struct stream_filter_vtbl ident_vtbl = {
1681         ident_filter_fn,
1682         ident_free_fn,
1683 };
1684
1685 static struct stream_filter *ident_filter(const unsigned char *sha1)
1686 {
1687         struct ident_filter *ident = xmalloc(sizeof(*ident));
1688
1689         xsnprintf(ident->ident, sizeof(ident->ident),
1690                   ": %s $", sha1_to_hex(sha1));
1691         strbuf_init(&ident->left, 0);
1692         ident->filter.vtbl = &ident_vtbl;
1693         ident->state = 0;
1694         return (struct stream_filter *)ident;
1695 }
1696
1697 /*
1698  * Return an appropriately constructed filter for the path, or NULL if
1699  * the contents cannot be filtered without reading the whole thing
1700  * in-core.
1701  *
1702  * Note that you would be crazy to set CRLF, smuge/clean or ident to a
1703  * large binary blob you would want us not to slurp into the memory!
1704  */
1705 struct stream_filter *get_stream_filter(const char *path, const unsigned char *sha1)
1706 {
1707         struct conv_attrs ca;
1708         struct stream_filter *filter = NULL;
1709
1710         convert_attrs(&ca, path);
1711         if (ca.drv && (ca.drv->process || ca.drv->smudge || ca.drv->clean))
1712                 return NULL;
1713
1714         if (ca.crlf_action == CRLF_AUTO || ca.crlf_action == CRLF_AUTO_CRLF)
1715                 return NULL;
1716
1717         if (ca.ident)
1718                 filter = ident_filter(sha1);
1719
1720         if (output_eol(ca.crlf_action) == EOL_CRLF)
1721                 filter = cascade_filter(filter, lf_to_crlf_filter());
1722         else
1723                 filter = cascade_filter(filter, &null_filter_singleton);
1724
1725         return filter;
1726 }
1727
1728 void free_stream_filter(struct stream_filter *filter)
1729 {
1730         filter->vtbl->free(filter);
1731 }
1732
1733 int stream_filter(struct stream_filter *filter,
1734                   const char *input, size_t *isize_p,
1735                   char *output, size_t *osize_p)
1736 {
1737         return filter->vtbl->filter(filter, input, isize_p, output, osize_p);
1738 }