git.oblomov.eu Git - git/blob - 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 #include "utf8.h"
  11
  12 /*
  13  * convert.c - convert a file when checking it out and checking it in.
  14  *
  15  * This should use the pathname to decide on whether it wants to do some
  16  * more interesting conversions (automatic gzip/unzip, general format
  17  * conversions etc etc), but by default it just does automatic CRLF<->LF
  18  * translation when the "text" attribute or "auto_crlf" option is set.
  19  */
  20
  21 /* Stat bits: When BIN is set, the txt bits are unset */
  22 #define CONVERT_STAT_BITS_TXT_LF    0x1
  23 #define CONVERT_STAT_BITS_TXT_CRLF  0x2
  24 #define CONVERT_STAT_BITS_BIN       0x4
  25
  26 enum crlf_action {
  27         CRLF_UNDEFINED,
  28         CRLF_BINARY,
  29         CRLF_TEXT,
  30         CRLF_TEXT_INPUT,
  31         CRLF_TEXT_CRLF,
  32         CRLF_AUTO,
  33         CRLF_AUTO_INPUT,
  34         CRLF_AUTO_CRLF
  35 };
  36
  37 struct text_stat {
  38         /* NUL, CR, LF and CRLF counts */
  39         unsigned nul, lonecr, lonelf, crlf;
  40
  41         /* These are just approximations! */
  42         unsigned printable, nonprintable;
  43 };
  44
  45 static void gather_stats(const char *buf, unsigned long size, struct text_stat *stats)
  46 {
  47         unsigned long i;
  48
  49         memset(stats, 0, sizeof(*stats));
  50
  51         for (i = 0; i < size; i++) {
  52                 unsigned char c = buf[i];
  53                 if (c == '\r') {
  54                         if (i+1 < size && buf[i+1] == '\n') {
  55                                 stats->crlf++;
  56                                 i++;
  57                         } else
  58                                 stats->lonecr++;
  59                         continue;
  60                 }
  61                 if (c == '\n') {
  62                         stats->lonelf++;
  63                         continue;
  64                 }
  65                 if (c == 127)
  66                         /* DEL */
  67                         stats->nonprintable++;
  68                 else if (c < 32) {
  69                         switch (c) {
  70                                 /* BS, HT, ESC and FF */
  71                         case '\b': case '\t': case '\033': case '\014':
  72                                 stats->printable++;
  73                                 break;
  74                         case 0:
  75                                 stats->nul++;
  76                                 /* fall through */
  77                         default:
  78                                 stats->nonprintable++;
  79                         }
  80                 }
  81                 else
  82                         stats->printable++;
  83         }
  84
  85         /* If file ends with EOF then don't count this EOF as non-printable. */
  86         if (size >= 1 && buf[size-1] == '\032')
  87                 stats->nonprintable--;
  88 }
  89
  90 /*
  91  * The same heuristics as diff.c::mmfile_is_binary()
  92  * We treat files with bare CR as binary
  93  */
  94 static int convert_is_binary(unsigned long size, const struct text_stat *stats)
  95 {
  96         if (stats->lonecr)
  97                 return 1;
  98         if (stats->nul)
  99                 return 1;
 100         if ((stats->printable >> 7) < stats->nonprintable)
 101                 return 1;
 102         return 0;
 103 }
 104
 105 static unsigned int gather_convert_stats(const char *data, unsigned long size)
 106 {
 107         struct text_stat stats;
 108         int ret = 0;
 109         if (!data || !size)
 110                 return 0;
 111         gather_stats(data, size, &stats);
 112         if (convert_is_binary(size, &stats))
 113                 ret |= CONVERT_STAT_BITS_BIN;
 114         if (stats.crlf)
 115                 ret |= CONVERT_STAT_BITS_TXT_CRLF;
 116         if (stats.lonelf)
 117                 ret |=  CONVERT_STAT_BITS_TXT_LF;
 118
 119         return ret;
 120 }
 121
 122 static const char *gather_convert_stats_ascii(const char *data, unsigned long size)
 123 {
 124         unsigned int convert_stats = gather_convert_stats(data, size);
 125
 126         if (convert_stats & CONVERT_STAT_BITS_BIN)
 127                 return "-text";
 128         switch (convert_stats) {
 129         case CONVERT_STAT_BITS_TXT_LF:
 130                 return "lf";
 131         case CONVERT_STAT_BITS_TXT_CRLF:
 132                 return "crlf";
 133         case CONVERT_STAT_BITS_TXT_LF | CONVERT_STAT_BITS_TXT_CRLF:
 134                 return "mixed";
 135         default:
 136                 return "none";
 137         }
 138 }
 139
 140 const char *get_cached_convert_stats_ascii(const struct index_state *istate,
 141                                            const char *path)
 142 {
 143         const char *ret;
 144         unsigned long sz;
 145         void *data = read_blob_data_from_index(istate, path, &sz);
 146         ret = gather_convert_stats_ascii(data, sz);
 147         free(data);
 148         return ret;
 149 }
 150
 151 const char *get_wt_convert_stats_ascii(const char *path)
 152 {
 153         const char *ret = "";
 154         struct strbuf sb = STRBUF_INIT;
 155         if (strbuf_read_file(&sb, path, 0) >= 0)
 156                 ret = gather_convert_stats_ascii(sb.buf, sb.len);
 157         strbuf_release(&sb);
 158         return ret;
 159 }
 160
 161 static int text_eol_is_crlf(void)
 162 {
 163         if (auto_crlf == AUTO_CRLF_TRUE)
 164                 return 1;
 165         else if (auto_crlf == AUTO_CRLF_INPUT)
 166                 return 0;
 167         if (core_eol == EOL_CRLF)
 168                 return 1;
 169         if (core_eol == EOL_UNSET && EOL_NATIVE == EOL_CRLF)
 170                 return 1;
 171         return 0;
 172 }
 173
 174 static enum eol output_eol(enum crlf_action crlf_action)
 175 {
 176         switch (crlf_action) {
 177         case CRLF_BINARY:
 178                 return EOL_UNSET;
 179         case CRLF_TEXT_CRLF:
 180                 return EOL_CRLF;
 181         case CRLF_TEXT_INPUT:
 182                 return EOL_LF;
 183         case CRLF_UNDEFINED:
 184         case CRLF_AUTO_CRLF:
 185                 return EOL_CRLF;
 186         case CRLF_AUTO_INPUT:
 187                 return EOL_LF;
 188         case CRLF_TEXT:
 189         case CRLF_AUTO:
 190                 /* fall through */
 191                 return text_eol_is_crlf() ? EOL_CRLF : EOL_LF;
 192         }
 193         warning("Illegal crlf_action %d\n", (int)crlf_action);
 194         return core_eol;
 195 }
 196
 197 static void check_global_conv_flags_eol(const char *path, enum crlf_action crlf_action,
 198                             struct text_stat *old_stats, struct text_stat *new_stats,
 199                             int conv_flags)
 200 {
 201         if (old_stats->crlf && !new_stats->crlf ) {
 202                 /*
 203                  * CRLFs would not be restored by checkout
 204                  */
 205                 if (conv_flags & CONV_EOL_RNDTRP_DIE)
 206                         die(_("CRLF would be replaced by LF in %s."), path);
 207                 else if (conv_flags & CONV_EOL_RNDTRP_WARN)
 208                         warning(_("CRLF will be replaced by LF in %s.\n"
 209                                   "The file will have its original line"
 210                                   " endings in your working directory."), path);
 211         } else if (old_stats->lonelf && !new_stats->lonelf ) {
 212                 /*
 213                  * CRLFs would be added by checkout
 214                  */
 215                 if (conv_flags & CONV_EOL_RNDTRP_DIE)
 216                         die(_("LF would be replaced by CRLF in %s"), path);
 217                 else if (conv_flags & CONV_EOL_RNDTRP_WARN)
 218                         warning(_("LF will be replaced by CRLF in %s.\n"
 219                                   "The file will have its original line"
 220                                   " endings in your working directory."), path);
 221         }
 222 }
 223
 224 static int has_crlf_in_index(const struct index_state *istate, const char *path)
 225 {
 226         unsigned long sz;
 227         void *data;
 228         const char *crp;
 229         int has_crlf = 0;
 230
 231         data = read_blob_data_from_index(istate, path, &sz);
 232         if (!data)
 233                 return 0;
 234
 235         crp = memchr(data, '\r', sz);
 236         if (crp) {
 237                 unsigned int ret_stats;
 238                 ret_stats = gather_convert_stats(data, sz);
 239                 if (!(ret_stats & CONVERT_STAT_BITS_BIN) &&
 240                     (ret_stats & CONVERT_STAT_BITS_TXT_CRLF))
 241                         has_crlf = 1;
 242         }
 243         free(data);
 244         return has_crlf;
 245 }
 246
 247 static int will_convert_lf_to_crlf(size_t len, struct text_stat *stats,
 248                                    enum crlf_action crlf_action)
 249 {
 250         if (output_eol(crlf_action) != EOL_CRLF)
 251                 return 0;
 252         /* No "naked" LF? Nothing to convert, regardless. */
 253         if (!stats->lonelf)
 254                 return 0;
 255
 256         if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
 257                 /* If we have any CR or CRLF line endings, we do not touch it */
 258                 /* This is the new safer autocrlf-handling */
 259                 if (stats->lonecr || stats->crlf)
 260                         return 0;
 261
 262                 if (convert_is_binary(len, stats))
 263                         return 0;
 264         }
 265         return 1;
 266
 267 }
 268
 269 static const char *default_encoding = "UTF-8";
 270
 271 static int encode_to_git(const char *path, const char *src, size_t src_len,
 272                          struct strbuf *buf, const char *enc, int conv_flags)
 273 {
 274         char *dst;
 275         int dst_len;
 276         int die_on_error = conv_flags & CONV_WRITE_OBJECT;
 277
 278         /*
 279          * No encoding is specified or there is nothing to encode.
 280          * Tell the caller that the content was not modified.
 281          */
 282         if (!enc || (src && !src_len))
 283                 return 0;
 284
 285         /*
 286          * Looks like we got called from "would_convert_to_git()".
 287          * This means Git wants to know if it would encode (= modify!)
 288          * the content. Let's answer with "yes", since an encoding was
 289          * specified.
 290          */
 291         if (!buf && !src)
 292                 return 1;
 293
 294         dst = reencode_string_len(src, src_len, default_encoding, enc,
 295                                   &dst_len);
 296         if (!dst) {
 297                 /*
 298                  * We could add the blob "as-is" to Git. However, on checkout
 299                  * we would try to reencode to the original encoding. This
 300                  * would fail and we would leave the user with a messed-up
 301                  * working tree. Let's try to avoid this by screaming loud.
 302                  */
 303                 const char* msg = _("failed to encode '%s' from %s to %s");
 304                 if (die_on_error)
 305                         die(msg, path, enc, default_encoding);
 306                 else {
 307                         error(msg, path, enc, default_encoding);
 308                         return 0;
 309                 }
 310         }
 311
 312         strbuf_attach(buf, dst, dst_len, dst_len + 1);
 313         return 1;
 314 }
 315
 316 static int encode_to_worktree(const char *path, const char *src, size_t src_len,
 317                               struct strbuf *buf, const char *enc)
 318 {
 319         char *dst;
 320         int dst_len;
 321
 322         /*
 323          * No encoding is specified or there is nothing to encode.
 324          * Tell the caller that the content was not modified.
 325          */
 326         if (!enc || (src && !src_len))
 327                 return 0;
 328
 329         dst = reencode_string_len(src, src_len, enc, default_encoding,
 330                                   &dst_len);
 331         if (!dst) {
 332                 error("failed to encode '%s' from %s to %s",
 333                         path, default_encoding, enc);
 334                 return 0;
 335         }
 336
 337         strbuf_attach(buf, dst, dst_len, dst_len + 1);
 338         return 1;
 339 }
 340
 341 static int crlf_to_git(const struct index_state *istate,
 342                        const char *path, const char *src, size_t len,
 343                        struct strbuf *buf,
 344                        enum crlf_action crlf_action, int conv_flags)
 345 {
 346         struct text_stat stats;
 347         char *dst;
 348         int convert_crlf_into_lf;
 349
 350         if (crlf_action == CRLF_BINARY ||
 351             (src && !len))
 352                 return 0;
 353
 354         /*
 355          * If we are doing a dry-run and have no source buffer, there is
 356          * nothing to analyze; we must assume we would convert.
 357          */
 358         if (!buf && !src)
 359                 return 1;
 360
 361         gather_stats(src, len, &stats);
 362         /* Optimization: No CRLF? Nothing to convert, regardless. */
 363         convert_crlf_into_lf = !!stats.crlf;
 364
 365         if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
 366                 if (convert_is_binary(len, &stats))
 367                         return 0;
 368                 /*
 369                  * If the file in the index has any CR in it, do not
 370                  * convert.  This is the new safer autocrlf handling,
 371                  * unless we want to renormalize in a merge or
 372                  * cherry-pick.
 373                  */
 374                 if ((!(conv_flags & CONV_EOL_RENORMALIZE)) &&
 375                     has_crlf_in_index(istate, path))
 376                         convert_crlf_into_lf = 0;
 377         }
 378         if (((conv_flags & CONV_EOL_RNDTRP_WARN) ||
 379              ((conv_flags & CONV_EOL_RNDTRP_DIE) && len))) {
 380                 struct text_stat new_stats;
 381                 memcpy(&new_stats, &stats, sizeof(new_stats));
 382                 /* simulate "git add" */
 383                 if (convert_crlf_into_lf) {
 384                         new_stats.lonelf += new_stats.crlf;
 385                         new_stats.crlf = 0;
 386                 }
 387                 /* simulate "git checkout" */
 388                 if (will_convert_lf_to_crlf(len, &new_stats, crlf_action)) {
 389                         new_stats.crlf += new_stats.lonelf;
 390                         new_stats.lonelf = 0;
 391                 }
 392                 check_global_conv_flags_eol(path, crlf_action, &stats, &new_stats, conv_flags);
 393         }
 394         if (!convert_crlf_into_lf)
 395                 return 0;
 396
 397         /*
 398          * At this point all of our source analysis is done, and we are sure we
 399          * would convert. If we are in dry-run mode, we can give an answer.
 400          */
 401         if (!buf)
 402                 return 1;
 403
 404         /* only grow if not in place */
 405         if (strbuf_avail(buf) + buf->len < len)
 406                 strbuf_grow(buf, len - buf->len);
 407         dst = buf->buf;
 408         if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
 409                 /*
 410                  * If we guessed, we already know we rejected a file with
 411                  * lone CR, and we can strip a CR without looking at what
 412                  * follow it.
 413                  */
 414                 do {
 415                         unsigned char c = *src++;
 416                         if (c != '\r')
 417                                 *dst++ = c;
 418                 } while (--len);
 419         } else {
 420                 do {
 421                         unsigned char c = *src++;
 422                         if (! (c == '\r' && (1 < len && *src == '\n')))
 423                                 *dst++ = c;
 424                 } while (--len);
 425         }
 426         strbuf_setlen(buf, dst - buf->buf);
 427         return 1;
 428 }
 429
 430 static int crlf_to_worktree(const char *path, const char *src, size_t len,
 431                             struct strbuf *buf, enum crlf_action crlf_action)
 432 {
 433         char *to_free = NULL;
 434         struct text_stat stats;
 435
 436         if (!len || output_eol(crlf_action) != EOL_CRLF)
 437                 return 0;
 438
 439         gather_stats(src, len, &stats);
 440         if (!will_convert_lf_to_crlf(len, &stats, crlf_action))
 441                 return 0;
 442
 443         /* are we "faking" in place editing ? */
 444         if (src == buf->buf)
 445                 to_free = strbuf_detach(buf, NULL);
 446
 447         strbuf_grow(buf, len + stats.lonelf);
 448         for (;;) {
 449                 const char *nl = memchr(src, '\n', len);
 450                 if (!nl)
 451                         break;
 452                 if (nl > src && nl[-1] == '\r') {
 453                         strbuf_add(buf, src, nl + 1 - src);
 454                 } else {
 455                         strbuf_add(buf, src, nl - src);
 456                         strbuf_addstr(buf, "\r\n");
 457                 }
 458                 len -= nl + 1 - src;
 459                 src  = nl + 1;
 460         }
 461         strbuf_add(buf, src, len);
 462
 463         free(to_free);
 464         return 1;
 465 }
 466
 467 struct filter_params {
 468         const char *src;
 469         unsigned long size;
 470         int fd;
 471         const char *cmd;
 472         const char *path;
 473 };
 474
 475 static int filter_buffer_or_fd(int in, int out, void *data)
 476 {
 477         /*
 478          * Spawn cmd and feed the buffer contents through its stdin.
 479          */
 480         struct child_process child_process = CHILD_PROCESS_INIT;
 481         struct filter_params *params = (struct filter_params *)data;
 482         int write_err, status;
 483         const char *argv[] = { NULL, NULL };
 484
 485         /* apply % substitution to cmd */
 486         struct strbuf cmd = STRBUF_INIT;
 487         struct strbuf path = STRBUF_INIT;
 488         struct strbuf_expand_dict_entry dict[] = {
 489                 { "f", NULL, },
 490                 { NULL, NULL, },
 491         };
 492
 493         /* quote the path to preserve spaces, etc. */
 494         sq_quote_buf(&path, params->path);
 495         dict[0].value = path.buf;
 496
 497         /* expand all %f with the quoted path */
 498         strbuf_expand(&cmd, params->cmd, strbuf_expand_dict_cb, &dict);
 499         strbuf_release(&path);
 500
 501         argv[0] = cmd.buf;
 502
 503         child_process.argv = argv;
 504         child_process.use_shell = 1;
 505         child_process.in = -1;
 506         child_process.out = out;
 507
 508         if (start_command(&child_process)) {
 509                 strbuf_release(&cmd);
 510                 return error("cannot fork to run external filter '%s'", params->cmd);
 511         }
 512
 513         sigchain_push(SIGPIPE, SIG_IGN);
 514
 515         if (params->src) {
 516                 write_err = (write_in_full(child_process.in,
 517                                            params->src, params->size) < 0);
 518                 if (errno == EPIPE)
 519                         write_err = 0;
 520         } else {
 521                 write_err = copy_fd(params->fd, child_process.in);
 522                 if (write_err == COPY_WRITE_ERROR && errno == EPIPE)
 523                         write_err = 0;
 524         }
 525
 526         if (close(child_process.in))
 527                 write_err = 1;
 528         if (write_err)
 529                 error("cannot feed the input to external filter '%s'", params->cmd);
 530
 531         sigchain_pop(SIGPIPE);
 532
 533         status = finish_command(&child_process);
 534         if (status)
 535                 error("external filter '%s' failed %d", params->cmd, status);
 536
 537         strbuf_release(&cmd);
 538         return (write_err || status);
 539 }
 540
 541 static int apply_single_file_filter(const char *path, const char *src, size_t len, int fd,
 542                         struct strbuf *dst, const char *cmd)
 543 {
 544         /*
 545          * Create a pipeline to have the command filter the buffer's
 546          * contents.
 547          *
 548          * (child --> cmd) --> us
 549          */
 550         int err = 0;
 551         struct strbuf nbuf = STRBUF_INIT;
 552         struct async async;
 553         struct filter_params params;
 554
 555         memset(&async, 0, sizeof(async));
 556         async.proc = filter_buffer_or_fd;
 557         async.data = &params;
 558         async.out = -1;
 559         params.src = src;
 560         params.size = len;
 561         params.fd = fd;
 562         params.cmd = cmd;
 563         params.path = path;
 564
 565         fflush(NULL);
 566         if (start_async(&async))
 567                 return 0;       /* error was already reported */
 568
 569         if (strbuf_read(&nbuf, async.out, len) < 0) {
 570                 err = error("read from external filter '%s' failed", cmd);
 571         }
 572         if (close(async.out)) {
 573                 err = error("read from external filter '%s' failed", cmd);
 574         }
 575         if (finish_async(&async)) {
 576                 err = error("external filter '%s' failed", cmd);
 577         }
 578
 579         if (!err) {
 580                 strbuf_swap(dst, &nbuf);
 581         }
 582         strbuf_release(&nbuf);
 583         return !err;
 584 }
 585
 586 #define CAP_CLEAN    (1u<<0)
 587 #define CAP_SMUDGE   (1u<<1)
 588 #define CAP_DELAY    (1u<<2)
 589
 590 struct cmd2process {
 591         struct subprocess_entry subprocess; /* must be the first member! */
 592         unsigned int supported_capabilities;
 593 };
 594
 595 static int subprocess_map_initialized;
 596 static struct hashmap subprocess_map;
 597
 598 static int start_multi_file_filter_fn(struct subprocess_entry *subprocess)
 599 {
 600         static int versions[] = {2, 0};
 601         static struct subprocess_capability capabilities[] = {
 602                 { "clean",  CAP_CLEAN  },
 603                 { "smudge", CAP_SMUDGE },
 604                 { "delay",  CAP_DELAY  },
 605                 { NULL, 0 }
 606         };
 607         struct cmd2process *entry = (struct cmd2process *)subprocess;
 608         return subprocess_handshake(subprocess, "git-filter", versions, NULL,
 609                                     capabilities,
 610                                     &entry->supported_capabilities);
 611 }
 612
 613 static void handle_filter_error(const struct strbuf *filter_status,
 614                                 struct cmd2process *entry,
 615                                 const unsigned int wanted_capability) {
 616         if (!strcmp(filter_status->buf, "error"))
 617                 ; /* The filter signaled a problem with the file. */
 618         else if (!strcmp(filter_status->buf, "abort") && wanted_capability) {
 619                 /*
 620                  * The filter signaled a permanent problem. Don't try to filter
 621                  * files with the same command for the lifetime of the current
 622                  * Git process.
 623                  */
 624                  entry->supported_capabilities &= ~wanted_capability;
 625         } else {
 626                 /*
 627                  * Something went wrong with the protocol filter.
 628                  * Force shutdown and restart if another blob requires filtering.
 629                  */
 630                 error("external filter '%s' failed", entry->subprocess.cmd);
 631                 subprocess_stop(&subprocess_map, &entry->subprocess);
 632                 free(entry);
 633         }
 634 }
 635
 636 static int apply_multi_file_filter(const char *path, const char *src, size_t len,
 637                                    int fd, struct strbuf *dst, const char *cmd,
 638                                    const unsigned int wanted_capability,
 639                                    struct delayed_checkout *dco)
 640 {
 641         int err;
 642         int can_delay = 0;
 643         struct cmd2process *entry;
 644         struct child_process *process;
 645         struct strbuf nbuf = STRBUF_INIT;
 646         struct strbuf filter_status = STRBUF_INIT;
 647         const char *filter_type;
 648
 649         if (!subprocess_map_initialized) {
 650                 subprocess_map_initialized = 1;
 651                 hashmap_init(&subprocess_map, cmd2process_cmp, NULL, 0);
 652                 entry = NULL;
 653         } else {
 654                 entry = (struct cmd2process *)subprocess_find_entry(&subprocess_map, cmd);
 655         }
 656
 657         fflush(NULL);
 658
 659         if (!entry) {
 660                 entry = xmalloc(sizeof(*entry));
 661                 entry->supported_capabilities = 0;
 662
 663                 if (subprocess_start(&subprocess_map, &entry->subprocess, cmd, start_multi_file_filter_fn)) {
 664                         free(entry);
 665                         return 0;
 666                 }
 667         }
 668         process = &entry->subprocess.process;
 669
 670         if (!(entry->supported_capabilities & wanted_capability))
 671                 return 0;
 672
 673         if (wanted_capability & CAP_CLEAN)
 674                 filter_type = "clean";
 675         else if (wanted_capability & CAP_SMUDGE)
 676                 filter_type = "smudge";
 677         else
 678                 die("unexpected filter type");
 679
 680         sigchain_push(SIGPIPE, SIG_IGN);
 681
 682         assert(strlen(filter_type) < LARGE_PACKET_DATA_MAX - strlen("command=\n"));
 683         err = packet_write_fmt_gently(process->in, "command=%s\n", filter_type);
 684         if (err)
 685                 goto done;
 686
 687         err = strlen(path) > LARGE_PACKET_DATA_MAX - strlen("pathname=\n");
 688         if (err) {
 689                 error("path name too long for external filter");
 690                 goto done;
 691         }
 692
 693         err = packet_write_fmt_gently(process->in, "pathname=%s\n", path);
 694         if (err)
 695                 goto done;
 696
 697         if ((entry->supported_capabilities & CAP_DELAY) &&
 698             dco && dco->state == CE_CAN_DELAY) {
 699                 can_delay = 1;
 700                 err = packet_write_fmt_gently(process->in, "can-delay=1\n");
 701                 if (err)
 702                         goto done;
 703         }
 704
 705         err = packet_flush_gently(process->in);
 706         if (err)
 707                 goto done;
 708
 709         if (fd >= 0)
 710                 err = write_packetized_from_fd(fd, process->in);
 711         else
 712                 err = write_packetized_from_buf(src, len, process->in);
 713         if (err)
 714                 goto done;
 715
 716         err = subprocess_read_status(process->out, &filter_status);
 717         if (err)
 718                 goto done;
 719
 720         if (can_delay && !strcmp(filter_status.buf, "delayed")) {
 721                 string_list_insert(&dco->filters, cmd);
 722                 string_list_insert(&dco->paths, path);
 723         } else {
 724                 /* The filter got the blob and wants to send us a response. */
 725                 err = strcmp(filter_status.buf, "success");
 726                 if (err)
 727                         goto done;
 728
 729                 err = read_packetized_to_strbuf(process->out, &nbuf) < 0;
 730                 if (err)
 731                         goto done;
 732
 733                 err = subprocess_read_status(process->out, &filter_status);
 734                 if (err)
 735                         goto done;
 736
 737                 err = strcmp(filter_status.buf, "success");
 738         }
 739
 740 done:
 741         sigchain_pop(SIGPIPE);
 742
 743         if (err)
 744                 handle_filter_error(&filter_status, entry, wanted_capability);
 745         else
 746                 strbuf_swap(dst, &nbuf);
 747         strbuf_release(&nbuf);
 748         return !err;
 749 }
 750
 751
 752 int async_query_available_blobs(const char *cmd, struct string_list *available_paths)
 753 {
 754         int err;
 755         char *line;
 756         struct cmd2process *entry;
 757         struct child_process *process;
 758         struct strbuf filter_status = STRBUF_INIT;
 759
 760         assert(subprocess_map_initialized);
 761         entry = (struct cmd2process *)subprocess_find_entry(&subprocess_map, cmd);
 762         if (!entry) {
 763                 error("external filter '%s' is not available anymore although "
 764                       "not all paths have been filtered", cmd);
 765                 return 0;
 766         }
 767         process = &entry->subprocess.process;
 768         sigchain_push(SIGPIPE, SIG_IGN);
 769
 770         err = packet_write_fmt_gently(
 771                 process->in, "command=list_available_blobs\n");
 772         if (err)
 773                 goto done;
 774
 775         err = packet_flush_gently(process->in);
 776         if (err)
 777                 goto done;
 778
 779         while ((line = packet_read_line(process->out, NULL))) {
 780                 const char *path;
 781                 if (skip_prefix(line, "pathname=", &path))
 782                         string_list_insert(available_paths, xstrdup(path));
 783                 else
 784                         ; /* ignore unknown keys */
 785         }
 786
 787         err = subprocess_read_status(process->out, &filter_status);
 788         if (err)
 789                 goto done;
 790
 791         err = strcmp(filter_status.buf, "success");
 792
 793 done:
 794         sigchain_pop(SIGPIPE);
 795
 796         if (err)
 797                 handle_filter_error(&filter_status, entry, 0);
 798         return !err;
 799 }
 800
 801 static struct convert_driver {
 802         const char *name;
 803         struct convert_driver *next;
 804         const char *smudge;
 805         const char *clean;
 806         const char *process;
 807         int required;
 808 } *user_convert, **user_convert_tail;
 809
 810 static int apply_filter(const char *path, const char *src, size_t len,
 811                         int fd, struct strbuf *dst, struct convert_driver *drv,
 812                         const unsigned int wanted_capability,
 813                         struct delayed_checkout *dco)
 814 {
 815         const char *cmd = NULL;
 816
 817         if (!drv)
 818                 return 0;
 819
 820         if (!dst)
 821                 return 1;
 822
 823         if ((wanted_capability & CAP_CLEAN) && !drv->process && drv->clean)
 824                 cmd = drv->clean;
 825         else if ((wanted_capability & CAP_SMUDGE) && !drv->process && drv->smudge)
 826                 cmd = drv->smudge;
 827
 828         if (cmd && *cmd)
 829                 return apply_single_file_filter(path, src, len, fd, dst, cmd);
 830         else if (drv->process && *drv->process)
 831                 return apply_multi_file_filter(path, src, len, fd, dst,
 832                         drv->process, wanted_capability, dco);
 833
 834         return 0;
 835 }
 836
 837 static int read_convert_config(const char *var, const char *value, void *cb)
 838 {
 839         const char *key, *name;
 840         int namelen;
 841         struct convert_driver *drv;
 842
 843         /*
 844          * External conversion drivers are configured using
 845          * "filter.<name>.variable".
 846          */
 847         if (parse_config_key(var, "filter", &name, &namelen, &key) < 0 || !name)
 848                 return 0;
 849         for (drv = user_convert; drv; drv = drv->next)
 850                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
 851                         break;
 852         if (!drv) {
 853                 drv = xcalloc(1, sizeof(struct convert_driver));
 854                 drv->name = xmemdupz(name, namelen);
 855                 *user_convert_tail = drv;
 856                 user_convert_tail = &(drv->next);
 857         }
 858
 859         /*
 860          * filter.<name>.smudge and filter.<name>.clean specifies
 861          * the command line:
 862          *
 863          *      command-line
 864          *
 865          * The command-line will not be interpolated in any way.
 866          */
 867
 868         if (!strcmp("smudge", key))
 869                 return git_config_string(&drv->smudge, var, value);
 870
 871         if (!strcmp("clean", key))
 872                 return git_config_string(&drv->clean, var, value);
 873
 874         if (!strcmp("process", key))
 875                 return git_config_string(&drv->process, var, value);
 876
 877         if (!strcmp("required", key)) {
 878                 drv->required = git_config_bool(var, value);
 879                 return 0;
 880         }
 881
 882         return 0;
 883 }
 884
 885 static int count_ident(const char *cp, unsigned long size)
 886 {
 887         /*
 888          * "$Id: 0000000000000000000000000000000000000000 $" <=> "$Id$"
 889          */
 890         int cnt = 0;
 891         char ch;
 892
 893         while (size) {
 894                 ch = *cp++;
 895                 size--;
 896                 if (ch != '$')
 897                         continue;
 898                 if (size < 3)
 899                         break;
 900                 if (memcmp("Id", cp, 2))
 901                         continue;
 902                 ch = cp[2];
 903                 cp += 3;
 904                 size -= 3;
 905                 if (ch == '$')
 906                         cnt++; /* $Id$ */
 907                 if (ch != ':')
 908                         continue;
 909
 910                 /*
 911                  * "$Id: ... "; scan up to the closing dollar sign and discard.
 912                  */
 913                 while (size) {
 914                         ch = *cp++;
 915                         size--;
 916                         if (ch == '$') {
 917                                 cnt++;
 918                                 break;
 919                         }
 920                         if (ch == '\n')
 921                                 break;
 922                 }
 923         }
 924         return cnt;
 925 }
 926
 927 static int ident_to_git(const char *path, const char *src, size_t len,
 928                         struct strbuf *buf, int ident)
 929 {
 930         char *dst, *dollar;
 931
 932         if (!ident || (src && !count_ident(src, len)))
 933                 return 0;
 934
 935         if (!buf)
 936                 return 1;
 937
 938         /* only grow if not in place */
 939         if (strbuf_avail(buf) + buf->len < len)
 940                 strbuf_grow(buf, len - buf->len);
 941         dst = buf->buf;
 942         for (;;) {
 943                 dollar = memchr(src, '$', len);
 944                 if (!dollar)
 945                         break;
 946                 memmove(dst, src, dollar + 1 - src);
 947                 dst += dollar + 1 - src;
 948                 len -= dollar + 1 - src;
 949                 src  = dollar + 1;
 950
 951                 if (len > 3 && !memcmp(src, "Id:", 3)) {
 952                         dollar = memchr(src + 3, '$', len - 3);
 953                         if (!dollar)
 954                                 break;
 955                         if (memchr(src + 3, '\n', dollar - src - 3)) {
 956                                 /* Line break before the next dollar. */
 957                                 continue;
 958                         }
 959
 960                         memcpy(dst, "Id$", 3);
 961                         dst += 3;
 962                         len -= dollar + 1 - src;
 963                         src  = dollar + 1;
 964                 }
 965         }
 966         memmove(dst, src, len);
 967         strbuf_setlen(buf, dst + len - buf->buf);
 968         return 1;
 969 }
 970
 971 static int ident_to_worktree(const char *path, const char *src, size_t len,
 972                              struct strbuf *buf, int ident)
 973 {
 974         unsigned char sha1[20];
 975         char *to_free = NULL, *dollar, *spc;
 976         int cnt;
 977
 978         if (!ident)
 979                 return 0;
 980
 981         cnt = count_ident(src, len);
 982         if (!cnt)
 983                 return 0;
 984
 985         /* are we "faking" in place editing ? */
 986         if (src == buf->buf)
 987                 to_free = strbuf_detach(buf, NULL);
 988         hash_sha1_file(src, len, "blob", sha1);
 989
 990         strbuf_grow(buf, len + cnt * 43);
 991         for (;;) {
 992                 /* step 1: run to the next '$' */
 993                 dollar = memchr(src, '$', len);
 994                 if (!dollar)
 995                         break;
 996                 strbuf_add(buf, src, dollar + 1 - src);
 997                 len -= dollar + 1 - src;
 998                 src  = dollar + 1;
 999
1000                 /* step 2: does it looks like a bit like Id:xxx$ or Id$ ? */
1001                 if (len < 3 || memcmp("Id", src, 2))
1002                         continue;
1003
1004                 /* step 3: skip over Id$ or Id:xxxxx$ */
1005                 if (src[2] == '$') {
1006                         src += 3;
1007                         len -= 3;
1008                 } else if (src[2] == ':') {
1009                         /*
1010                          * It's possible that an expanded Id has crept its way into the
1011                          * repository, we cope with that by stripping the expansion out.
1012                          * This is probably not a good idea, since it will cause changes
1013                          * on checkout, which won't go away by stash, but let's keep it
1014                          * for git-style ids.
1015                          */
1016                         dollar = memchr(src + 3, '$', len - 3);
1017                         if (!dollar) {
1018                                 /* incomplete keyword, no more '$', so just quit the loop */
1019                                 break;
1020                         }
1021
1022                         if (memchr(src + 3, '\n', dollar - src - 3)) {
1023                                 /* Line break before the next dollar. */
1024                                 continue;
1025                         }
1026
1027                         spc = memchr(src + 4, ' ', dollar - src - 4);
1028                         if (spc && spc < dollar-1) {
1029                                 /* There are spaces in unexpected places.
1030                                  * This is probably an id from some other
1031                                  * versioning system. Keep it for now.
1032                                  */
1033                                 continue;
1034                         }
1035
1036                         len -= dollar + 1 - src;
1037                         src  = dollar + 1;
1038                 } else {
1039                         /* it wasn't a "Id$" or "Id:xxxx$" */
1040                         continue;
1041                 }
1042
1043                 /* step 4: substitute */
1044                 strbuf_addstr(buf, "Id: ");
1045                 strbuf_add(buf, sha1_to_hex(sha1), 40);
1046                 strbuf_addstr(buf, " $");
1047         }
1048         strbuf_add(buf, src, len);
1049
1050         free(to_free);
1051         return 1;
1052 }
1053
1054 static const char *git_path_check_encoding(struct attr_check_item *check)
1055 {
1056         const char *value = check->value;
1057
1058         if (ATTR_UNSET(value) || !strlen(value))
1059                 return NULL;
1060
1061         if (ATTR_TRUE(value) || ATTR_FALSE(value)) {
1062                 die(_("true/false are no valid working-tree-encodings"));
1063         }
1064
1065         /* Don't encode to the default encoding */
1066         if (same_encoding(value, default_encoding))
1067                 return NULL;
1068
1069         return value;
1070 }
1071
1072 static enum crlf_action git_path_check_crlf(struct attr_check_item *check)
1073 {
1074         const char *value = check->value;
1075
1076         if (ATTR_TRUE(value))
1077                 return CRLF_TEXT;
1078         else if (ATTR_FALSE(value))
1079                 return CRLF_BINARY;
1080         else if (ATTR_UNSET(value))
1081                 ;
1082         else if (!strcmp(value, "input"))
1083                 return CRLF_TEXT_INPUT;
1084         else if (!strcmp(value, "auto"))
1085                 return CRLF_AUTO;
1086         return CRLF_UNDEFINED;
1087 }
1088
1089 static enum eol git_path_check_eol(struct attr_check_item *check)
1090 {
1091         const char *value = check->value;
1092
1093         if (ATTR_UNSET(value))
1094                 ;
1095         else if (!strcmp(value, "lf"))
1096                 return EOL_LF;
1097         else if (!strcmp(value, "crlf"))
1098                 return EOL_CRLF;
1099         return EOL_UNSET;
1100 }
1101
1102 static struct convert_driver *git_path_check_convert(struct attr_check_item *check)
1103 {
1104         const char *value = check->value;
1105         struct convert_driver *drv;
1106
1107         if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1108                 return NULL;
1109         for (drv = user_convert; drv; drv = drv->next)
1110                 if (!strcmp(value, drv->name))
1111                         return drv;
1112         return NULL;
1113 }
1114
1115 static int git_path_check_ident(struct attr_check_item *check)
1116 {
1117         const char *value = check->value;
1118
1119         return !!ATTR_TRUE(value);
1120 }
1121
1122 struct conv_attrs {
1123         struct convert_driver *drv;
1124         enum crlf_action attr_action; /* What attr says */
1125         enum crlf_action crlf_action; /* When no attr is set, use core.autocrlf */
1126         int ident;
1127         const char *working_tree_encoding; /* Supported encoding or default encoding if NULL */
1128 };
1129
1130 static void convert_attrs(struct conv_attrs *ca, const char *path)
1131 {
1132         static struct attr_check *check;
1133
1134         if (!check) {
1135                 check = attr_check_initl("crlf", "ident", "filter",
1136                                          "eol", "text", "working-tree-encoding",
1137                                          NULL);
1138                 user_convert_tail = &user_convert;
1139                 git_config(read_convert_config, NULL);
1140         }
1141
1142         if (!git_check_attr(path, check)) {
1143                 struct attr_check_item *ccheck = check->items;
1144                 ca->crlf_action = git_path_check_crlf(ccheck + 4);
1145                 if (ca->crlf_action == CRLF_UNDEFINED)
1146                         ca->crlf_action = git_path_check_crlf(ccheck + 0);
1147                 ca->ident = git_path_check_ident(ccheck + 1);
1148                 ca->drv = git_path_check_convert(ccheck + 2);
1149                 if (ca->crlf_action != CRLF_BINARY) {
1150                         enum eol eol_attr = git_path_check_eol(ccheck + 3);
1151                         if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_LF)
1152                                 ca->crlf_action = CRLF_AUTO_INPUT;
1153                         else if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_CRLF)
1154                                 ca->crlf_action = CRLF_AUTO_CRLF;
1155                         else if (eol_attr == EOL_LF)
1156                                 ca->crlf_action = CRLF_TEXT_INPUT;
1157                         else if (eol_attr == EOL_CRLF)
1158                                 ca->crlf_action = CRLF_TEXT_CRLF;
1159                 }
1160                 ca->working_tree_encoding = git_path_check_encoding(ccheck + 5);
1161         } else {
1162                 ca->drv = NULL;
1163                 ca->crlf_action = CRLF_UNDEFINED;
1164                 ca->ident = 0;
1165         }
1166
1167         /* Save attr and make a decision for action */
1168         ca->attr_action = ca->crlf_action;
1169         if (ca->crlf_action == CRLF_TEXT)
1170                 ca->crlf_action = text_eol_is_crlf() ? CRLF_TEXT_CRLF : CRLF_TEXT_INPUT;
1171         if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_FALSE)
1172                 ca->crlf_action = CRLF_BINARY;
1173         if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_TRUE)
1174                 ca->crlf_action = CRLF_AUTO_CRLF;
1175         if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_INPUT)
1176                 ca->crlf_action = CRLF_AUTO_INPUT;
1177 }
1178
1179 int would_convert_to_git_filter_fd(const char *path)
1180 {
1181         struct conv_attrs ca;
1182
1183         convert_attrs(&ca, path);
1184         if (!ca.drv)
1185                 return 0;
1186
1187         /*
1188          * Apply a filter to an fd only if the filter is required to succeed.
1189          * We must die if the filter fails, because the original data before
1190          * filtering is not available.
1191          */
1192         if (!ca.drv->required)
1193                 return 0;
1194
1195         return apply_filter(path, NULL, 0, -1, NULL, ca.drv, CAP_CLEAN, NULL);
1196 }
1197
1198 const char *get_convert_attr_ascii(const char *path)
1199 {
1200         struct conv_attrs ca;
1201
1202         convert_attrs(&ca, path);
1203         switch (ca.attr_action) {
1204         case CRLF_UNDEFINED:
1205                 return "";
1206         case CRLF_BINARY:
1207                 return "-text";
1208         case CRLF_TEXT:
1209                 return "text";
1210         case CRLF_TEXT_INPUT:
1211                 return "text eol=lf";
1212         case CRLF_TEXT_CRLF:
1213                 return "text eol=crlf";
1214         case CRLF_AUTO:
1215                 return "text=auto";
1216         case CRLF_AUTO_CRLF:
1217                 return "text=auto eol=crlf";
1218         case CRLF_AUTO_INPUT:
1219                 return "text=auto eol=lf";
1220         }
1221         return "";
1222 }
1223
1224 int convert_to_git(const struct index_state *istate,
1225                    const char *path, const char *src, size_t len,
1226                    struct strbuf *dst, int conv_flags)
1227 {
1228         int ret = 0;
1229         struct conv_attrs ca;
1230
1231         convert_attrs(&ca, path);
1232
1233         ret |= apply_filter(path, src, len, -1, dst, ca.drv, CAP_CLEAN, NULL);
1234         if (!ret && ca.drv && ca.drv->required)
1235                 die("%s: clean filter '%s' failed", path, ca.drv->name);
1236
1237         if (ret && dst) {
1238                 src = dst->buf;
1239                 len = dst->len;
1240         }
1241
1242         ret |= encode_to_git(path, src, len, dst, ca.working_tree_encoding, conv_flags);
1243         if (ret && dst) {
1244                 src = dst->buf;
1245                 len = dst->len;
1246         }
1247
1248         if (!(conv_flags & CONV_EOL_KEEP_CRLF)) {
1249                 ret |= crlf_to_git(istate, path, src, len, dst, ca.crlf_action, conv_flags);
1250                 if (ret && dst) {
1251                         src = dst->buf;
1252                         len = dst->len;
1253                 }
1254         }
1255         return ret | ident_to_git(path, src, len, dst, ca.ident);
1256 }
1257
1258 void convert_to_git_filter_fd(const struct index_state *istate,
1259                               const char *path, int fd, struct strbuf *dst,
1260                               int conv_flags)
1261 {
1262         struct conv_attrs ca;
1263         convert_attrs(&ca, path);
1264
1265         assert(ca.drv);
1266         assert(ca.drv->clean || ca.drv->process);
1267
1268         if (!apply_filter(path, NULL, 0, fd, dst, ca.drv, CAP_CLEAN, NULL))
1269                 die("%s: clean filter '%s' failed", path, ca.drv->name);
1270
1271         encode_to_git(path, dst->buf, dst->len, dst, ca.working_tree_encoding, conv_flags);
1272         crlf_to_git(istate, path, dst->buf, dst->len, dst, ca.crlf_action, conv_flags);
1273         ident_to_git(path, dst->buf, dst->len, dst, ca.ident);
1274 }
1275
1276 static int convert_to_working_tree_internal(const char *path, const char *src,
1277                                             size_t len, struct strbuf *dst,
1278                                             int normalizing, struct delayed_checkout *dco)
1279 {
1280         int ret = 0, ret_filter = 0;
1281         struct conv_attrs ca;
1282
1283         convert_attrs(&ca, path);
1284
1285         ret |= ident_to_worktree(path, src, len, dst, ca.ident);
1286         if (ret) {
1287                 src = dst->buf;
1288                 len = dst->len;
1289         }
1290         /*
1291          * CRLF conversion can be skipped if normalizing, unless there
1292          * is a smudge or process filter (even if the process filter doesn't
1293          * support smudge).  The filters might expect CRLFs.
1294          */
1295         if ((ca.drv && (ca.drv->smudge || ca.drv->process)) || !normalizing) {
1296                 ret |= crlf_to_worktree(path, src, len, dst, ca.crlf_action);
1297                 if (ret) {
1298                         src = dst->buf;
1299                         len = dst->len;
1300                 }
1301         }
1302
1303         ret |= encode_to_worktree(path, src, len, dst, ca.working_tree_encoding);
1304         if (ret) {
1305                 src = dst->buf;
1306                 len = dst->len;
1307         }
1308
1309         ret_filter = apply_filter(
1310                 path, src, len, -1, dst, ca.drv, CAP_SMUDGE, dco);
1311         if (!ret_filter && ca.drv && ca.drv->required)
1312                 die("%s: smudge filter %s failed", path, ca.drv->name);
1313
1314         return ret | ret_filter;
1315 }
1316
1317 int async_convert_to_working_tree(const char *path, const char *src,
1318                                   size_t len, struct strbuf *dst,
1319                                   void *dco)
1320 {
1321         return convert_to_working_tree_internal(path, src, len, dst, 0, dco);
1322 }
1323
1324 int convert_to_working_tree(const char *path, const char *src, size_t len, struct strbuf *dst)
1325 {
1326         return convert_to_working_tree_internal(path, src, len, dst, 0, NULL);
1327 }
1328
1329 int renormalize_buffer(const struct index_state *istate, const char *path,
1330                        const char *src, size_t len, struct strbuf *dst)
1331 {
1332         int ret = convert_to_working_tree_internal(path, src, len, dst, 1, NULL);
1333         if (ret) {
1334                 src = dst->buf;
1335                 len = dst->len;
1336         }
1337         return ret | convert_to_git(istate, path, src, len, dst, CONV_EOL_RENORMALIZE);
1338 }
1339
1340 /*****************************************************************
1341  *
1342  * Streaming conversion support
1343  *
1344  *****************************************************************/
1345
1346 typedef int (*filter_fn)(struct stream_filter *,
1347                          const char *input, size_t *isize_p,
1348                          char *output, size_t *osize_p);
1349 typedef void (*free_fn)(struct stream_filter *);
1350
1351 struct stream_filter_vtbl {
1352         filter_fn filter;
1353         free_fn free;
1354 };
1355
1356 struct stream_filter {
1357         struct stream_filter_vtbl *vtbl;
1358 };
1359
1360 static int null_filter_fn(struct stream_filter *filter,
1361                           const char *input, size_t *isize_p,
1362                           char *output, size_t *osize_p)
1363 {
1364         size_t count;
1365
1366         if (!input)
1367                 return 0; /* we do not keep any states */
1368         count = *isize_p;
1369         if (*osize_p < count)
1370                 count = *osize_p;
1371         if (count) {
1372                 memmove(output, input, count);
1373                 *isize_p -= count;
1374                 *osize_p -= count;
1375         }
1376         return 0;
1377 }
1378
1379 static void null_free_fn(struct stream_filter *filter)
1380 {
1381         ; /* nothing -- null instances are shared */
1382 }
1383
1384 static struct stream_filter_vtbl null_vtbl = {
1385         null_filter_fn,
1386         null_free_fn,
1387 };
1388
1389 static struct stream_filter null_filter_singleton = {
1390         &null_vtbl,
1391 };
1392
1393 int is_null_stream_filter(struct stream_filter *filter)
1394 {
1395         return filter == &null_filter_singleton;
1396 }
1397
1398
1399 /*
1400  * LF-to-CRLF filter
1401  */
1402
1403 struct lf_to_crlf_filter {
1404         struct stream_filter filter;
1405         unsigned has_held:1;
1406         char held;
1407 };
1408
1409 static int lf_to_crlf_filter_fn(struct stream_filter *filter,
1410                                 const char *input, size_t *isize_p,
1411                                 char *output, size_t *osize_p)
1412 {
1413         size_t count, o = 0;
1414         struct lf_to_crlf_filter *lf_to_crlf = (struct lf_to_crlf_filter *)filter;
1415
1416         /*
1417          * We may be holding onto the CR to see if it is followed by a
1418          * LF, in which case we would need to go to the main loop.
1419          * Otherwise, just emit it to the output stream.
1420          */
1421         if (lf_to_crlf->has_held && (lf_to_crlf->held != '\r' || !input)) {
1422                 output[o++] = lf_to_crlf->held;
1423                 lf_to_crlf->has_held = 0;
1424         }
1425
1426         /* We are told to drain */
1427         if (!input) {
1428                 *osize_p -= o;
1429                 return 0;
1430         }
1431
1432         count = *isize_p;
1433         if (count || lf_to_crlf->has_held) {
1434                 size_t i;
1435                 int was_cr = 0;
1436
1437                 if (lf_to_crlf->has_held) {
1438                         was_cr = 1;
1439                         lf_to_crlf->has_held = 0;
1440                 }
1441
1442                 for (i = 0; o < *osize_p && i < count; i++) {
1443                         char ch = input[i];
1444
1445                         if (ch == '\n') {
1446                                 output[o++] = '\r';
1447                         } else if (was_cr) {
1448                                 /*
1449                                  * Previous round saw CR and it is not followed
1450                                  * by a LF; emit the CR before processing the
1451                                  * current character.
1452                                  */
1453                                 output[o++] = '\r';
1454                         }
1455
1456                         /*
1457                          * We may have consumed the last output slot,
1458                          * in which case we need to break out of this
1459                          * loop; hold the current character before
1460                          * returning.
1461                          */
1462                         if (*osize_p <= o) {
1463                                 lf_to_crlf->has_held = 1;
1464                                 lf_to_crlf->held = ch;
1465                                 continue; /* break but increment i */
1466                         }
1467
1468                         if (ch == '\r') {
1469                                 was_cr = 1;
1470                                 continue;
1471                         }
1472
1473                         was_cr = 0;
1474                         output[o++] = ch;
1475                 }
1476
1477                 *osize_p -= o;
1478                 *isize_p -= i;
1479
1480                 if (!lf_to_crlf->has_held && was_cr) {
1481                         lf_to_crlf->has_held = 1;
1482                         lf_to_crlf->held = '\r';
1483                 }
1484         }
1485         return 0;
1486 }
1487
1488 static void lf_to_crlf_free_fn(struct stream_filter *filter)
1489 {
1490         free(filter);
1491 }
1492
1493 static struct stream_filter_vtbl lf_to_crlf_vtbl = {
1494         lf_to_crlf_filter_fn,
1495         lf_to_crlf_free_fn,
1496 };
1497
1498 static struct stream_filter *lf_to_crlf_filter(void)
1499 {
1500         struct lf_to_crlf_filter *lf_to_crlf = xcalloc(1, sizeof(*lf_to_crlf));
1501
1502         lf_to_crlf->filter.vtbl = &lf_to_crlf_vtbl;
1503         return (struct stream_filter *)lf_to_crlf;
1504 }
1505
1506 /*
1507  * Cascade filter
1508  */
1509 #define FILTER_BUFFER 1024
1510 struct cascade_filter {
1511         struct stream_filter filter;
1512         struct stream_filter *one;
1513         struct stream_filter *two;
1514         char buf[FILTER_BUFFER];
1515         int end, ptr;
1516 };
1517
1518 static int cascade_filter_fn(struct stream_filter *filter,
1519                              const char *input, size_t *isize_p,
1520                              char *output, size_t *osize_p)
1521 {
1522         struct cascade_filter *cas = (struct cascade_filter *) filter;
1523         size_t filled = 0;
1524         size_t sz = *osize_p;
1525         size_t to_feed, remaining;
1526
1527         /*
1528          * input -- (one) --> buf -- (two) --> output
1529          */
1530         while (filled < sz) {
1531                 remaining = sz - filled;
1532
1533                 /* do we already have something to feed two with? */
1534                 if (cas->ptr < cas->end) {
1535                         to_feed = cas->end - cas->ptr;
1536                         if (stream_filter(cas->two,
1537                                           cas->buf + cas->ptr, &to_feed,
1538                                           output + filled, &remaining))
1539                                 return -1;
1540                         cas->ptr += (cas->end - cas->ptr) - to_feed;
1541                         filled = sz - remaining;
1542                         continue;
1543                 }
1544
1545                 /* feed one from upstream and have it emit into our buffer */
1546                 to_feed = input ? *isize_p : 0;
1547                 if (input && !to_feed)
1548                         break;
1549                 remaining = sizeof(cas->buf);
1550                 if (stream_filter(cas->one,
1551                                   input, &to_feed,
1552                                   cas->buf, &remaining))
1553                         return -1;
1554                 cas->end = sizeof(cas->buf) - remaining;
1555                 cas->ptr = 0;
1556                 if (input) {
1557                         size_t fed = *isize_p - to_feed;
1558                         *isize_p -= fed;
1559                         input += fed;
1560                 }
1561
1562                 /* do we know that we drained one completely? */
1563                 if (input || cas->end)
1564                         continue;
1565
1566                 /* tell two to drain; we have nothing more to give it */
1567                 to_feed = 0;
1568                 remaining = sz - filled;
1569                 if (stream_filter(cas->two,
1570                                   NULL, &to_feed,
1571                                   output + filled, &remaining))
1572                         return -1;
1573                 if (remaining == (sz - filled))
1574                         break; /* completely drained two */
1575                 filled = sz - remaining;
1576         }
1577         *osize_p -= filled;
1578         return 0;
1579 }
1580
1581 static void cascade_free_fn(struct stream_filter *filter)
1582 {
1583         struct cascade_filter *cas = (struct cascade_filter *)filter;
1584         free_stream_filter(cas->one);
1585         free_stream_filter(cas->two);
1586         free(filter);
1587 }
1588
1589 static struct stream_filter_vtbl cascade_vtbl = {
1590         cascade_filter_fn,
1591         cascade_free_fn,
1592 };
1593
1594 static struct stream_filter *cascade_filter(struct stream_filter *one,
1595                                             struct stream_filter *two)
1596 {
1597         struct cascade_filter *cascade;
1598
1599         if (!one || is_null_stream_filter(one))
1600                 return two;
1601         if (!two || is_null_stream_filter(two))
1602                 return one;
1603
1604         cascade = xmalloc(sizeof(*cascade));
1605         cascade->one = one;
1606         cascade->two = two;
1607         cascade->end = cascade->ptr = 0;
1608         cascade->filter.vtbl = &cascade_vtbl;
1609         return (struct stream_filter *)cascade;
1610 }
1611
1612 /*
1613  * ident filter
1614  */
1615 #define IDENT_DRAINING (-1)
1616 #define IDENT_SKIPPING (-2)
1617 struct ident_filter {
1618         struct stream_filter filter;
1619         struct strbuf left;
1620         int state;
1621         char ident[45]; /* ": x40 $" */
1622 };
1623
1624 static int is_foreign_ident(const char *str)
1625 {
1626         int i;
1627
1628         if (!skip_prefix(str, "$Id: ", &str))
1629                 return 0;
1630         for (i = 0; str[i]; i++) {
1631                 if (isspace(str[i]) && str[i+1] != '$')
1632                         return 1;
1633         }
1634         return 0;
1635 }
1636
1637 static void ident_drain(struct ident_filter *ident, char **output_p, size_t *osize_p)
1638 {
1639         size_t to_drain = ident->left.len;
1640
1641         if (*osize_p < to_drain)
1642                 to_drain = *osize_p;
1643         if (to_drain) {
1644                 memcpy(*output_p, ident->left.buf, to_drain);
1645                 strbuf_remove(&ident->left, 0, to_drain);
1646                 *output_p += to_drain;
1647                 *osize_p -= to_drain;
1648         }
1649         if (!ident->left.len)
1650                 ident->state = 0;
1651 }
1652
1653 static int ident_filter_fn(struct stream_filter *filter,
1654                            const char *input, size_t *isize_p,
1655                            char *output, size_t *osize_p)
1656 {
1657         struct ident_filter *ident = (struct ident_filter *)filter;
1658         static const char head[] = "$Id";
1659
1660         if (!input) {
1661                 /* drain upon eof */
1662                 switch (ident->state) {
1663                 default:
1664                         strbuf_add(&ident->left, head, ident->state);
1665                         /* fallthrough */
1666                 case IDENT_SKIPPING:
1667                         /* fallthrough */
1668                 case IDENT_DRAINING:
1669                         ident_drain(ident, &output, osize_p);
1670                 }
1671                 return 0;
1672         }
1673
1674         while (*isize_p || (ident->state == IDENT_DRAINING)) {
1675                 int ch;
1676
1677                 if (ident->state == IDENT_DRAINING) {
1678                         ident_drain(ident, &output, osize_p);
1679                         if (!*osize_p)
1680                                 break;
1681                         continue;
1682                 }
1683
1684                 ch = *(input++);
1685                 (*isize_p)--;
1686
1687                 if (ident->state == IDENT_SKIPPING) {
1688                         /*
1689                          * Skipping until '$' or LF, but keeping them
1690                          * in case it is a foreign ident.
1691                          */
1692                         strbuf_addch(&ident->left, ch);
1693                         if (ch != '\n' && ch != '$')
1694                                 continue;
1695                         if (ch == '$' && !is_foreign_ident(ident->left.buf)) {
1696                                 strbuf_setlen(&ident->left, sizeof(head) - 1);
1697                                 strbuf_addstr(&ident->left, ident->ident);
1698                         }
1699                         ident->state = IDENT_DRAINING;
1700                         continue;
1701                 }
1702
1703                 if (ident->state < sizeof(head) &&
1704                     head[ident->state] == ch) {
1705                         ident->state++;
1706                         continue;
1707                 }
1708
1709                 if (ident->state)
1710                         strbuf_add(&ident->left, head, ident->state);
1711                 if (ident->state == sizeof(head) - 1) {
1712                         if (ch != ':' && ch != '$') {
1713                                 strbuf_addch(&ident->left, ch);
1714                                 ident->state = 0;
1715                                 continue;
1716                         }
1717
1718                         if (ch == ':') {
1719                                 strbuf_addch(&ident->left, ch);
1720                                 ident->state = IDENT_SKIPPING;
1721                         } else {
1722                                 strbuf_addstr(&ident->left, ident->ident);
1723                                 ident->state = IDENT_DRAINING;
1724                         }
1725                         continue;
1726                 }
1727
1728                 strbuf_addch(&ident->left, ch);
1729                 ident->state = IDENT_DRAINING;
1730         }
1731         return 0;
1732 }
1733
1734 static void ident_free_fn(struct stream_filter *filter)
1735 {
1736         struct ident_filter *ident = (struct ident_filter *)filter;
1737         strbuf_release(&ident->left);
1738         free(filter);
1739 }
1740
1741 static struct stream_filter_vtbl ident_vtbl = {
1742         ident_filter_fn,
1743         ident_free_fn,
1744 };
1745
1746 static struct stream_filter *ident_filter(const unsigned char *sha1)
1747 {
1748         struct ident_filter *ident = xmalloc(sizeof(*ident));
1749
1750         xsnprintf(ident->ident, sizeof(ident->ident),
1751                   ": %s $", sha1_to_hex(sha1));
1752         strbuf_init(&ident->left, 0);
1753         ident->filter.vtbl = &ident_vtbl;
1754         ident->state = 0;
1755         return (struct stream_filter *)ident;
1756 }
1757
1758 /*
1759  * Return an appropriately constructed filter for the path, or NULL if
1760  * the contents cannot be filtered without reading the whole thing
1761  * in-core.
1762  *
1763  * Note that you would be crazy to set CRLF, smuge/clean or ident to a
1764  * large binary blob you would want us not to slurp into the memory!
1765  */
1766 struct stream_filter *get_stream_filter(const char *path, const unsigned char *sha1)
1767 {
1768         struct conv_attrs ca;
1769         struct stream_filter *filter = NULL;
1770
1771         convert_attrs(&ca, path);
1772         if (ca.drv && (ca.drv->process || ca.drv->smudge || ca.drv->clean))
1773                 return NULL;
1774
1775         if (ca.working_tree_encoding)
1776                 return NULL;
1777
1778         if (ca.crlf_action == CRLF_AUTO || ca.crlf_action == CRLF_AUTO_CRLF)
1779                 return NULL;
1780
1781         if (ca.ident)
1782                 filter = ident_filter(sha1);
1783
1784         if (output_eol(ca.crlf_action) == EOL_CRLF)
1785                 filter = cascade_filter(filter, lf_to_crlf_filter());
1786         else
1787                 filter = cascade_filter(filter, &null_filter_singleton);
1788
1789         return filter;
1790 }
1791
1792 void free_stream_filter(struct stream_filter *filter)
1793 {
1794         filter->vtbl->free(filter);
1795 }
1796
1797 int stream_filter(struct stream_filter *filter,
1798                   const char *input, size_t *isize_p,
1799                   char *output, size_t *osize_p)
1800 {
1801         return filter->vtbl->filter(filter, input, isize_p, output, osize_p);
1802 }