convert.c

   1 #include "cache.h"
   2 #include "attr.h"
   3 #include "run-command.h"
   4
   5 /*
   6  * convert.c - convert a file when checking it out and checking it in.
   7  *
   8  * This should use the pathname to decide on whether it wants to do some
   9  * more interesting conversions (automatic gzip/unzip, general format
  10  * conversions etc etc), but by default it just does automatic CRLF<->LF
  11  * translation when the "auto_crlf" option is set.
  12  */
  13
  14 #define CRLF_GUESS      (-1)
  15 #define CRLF_BINARY     0
  16 #define CRLF_TEXT       1
  17 #define CRLF_INPUT      2
  18
  19 struct text_stat {
  20         /* CR, LF and CRLF counts */
  21         unsigned cr, lf, crlf;
  22
  23         /* These are just approximations! */
  24         unsigned printable, nonprintable;
  25 };
  26
  27 static void gather_stats(const char *buf, unsigned long size, struct text_stat *stats)
  28 {
  29         unsigned long i;
  30
  31         memset(stats, 0, sizeof(*stats));
  32
  33         for (i = 0; i < size; i++) {
  34                 unsigned char c = buf[i];
  35                 if (c == '\r') {
  36                         stats->cr++;
  37                         if (i+1 < size && buf[i+1] == '\n')
  38                                 stats->crlf++;
  39                         continue;
  40                 }
  41                 if (c == '\n') {
  42                         stats->lf++;
  43                         continue;
  44                 }
  45                 if (c == 127)
  46                         /* DEL */
  47                         stats->nonprintable++;
  48                 else if (c < 32) {
  49                         switch (c) {
  50                                 /* BS, HT, ESC and FF */
  51                         case '\b': case '\t': case '\033': case '\014':
  52                                 stats->printable++;
  53                                 break;
  54                         default:
  55                                 stats->nonprintable++;
  56                         }
  57                 }
  58                 else
  59                         stats->printable++;
  60         }
  61 }
  62
  63 /*
  64  * The same heuristics as diff.c::mmfile_is_binary()
  65  */
  66 static int is_binary(unsigned long size, struct text_stat *stats)
  67 {
  68
  69         if ((stats->printable >> 7) < stats->nonprintable)
  70                 return 1;
  71         /*
  72          * Other heuristics? Average line length might be relevant,
  73          * as might LF vs CR vs CRLF counts..
  74          *
  75          * NOTE! It might be normal to have a low ratio of CRLF to LF
  76          * (somebody starts with a LF-only file and edits it with an editor
  77          * that adds CRLF only to lines that are added..). But do  we
  78          * want to support CR-only? Probably not.
  79          */
  80         return 0;
  81 }
  82
  83 static int crlf_to_git(const char *path, const char *src, size_t len,
  84                        struct strbuf *buf, int action)
  85 {
  86         struct text_stat stats;
  87         char *dst;
  88
  89         if ((action == CRLF_BINARY) || !auto_crlf || !len)
  90                 return 0;
  91
  92         gather_stats(src, len, &stats);
  93         /* No CR? Nothing to convert, regardless. */
  94         if (!stats.cr)
  95                 return 0;
  96
  97         if (action == CRLF_GUESS) {
  98                 /*
  99                  * We're currently not going to even try to convert stuff
 100                  * that has bare CR characters. Does anybody do that crazy
 101                  * stuff?
 102                  */
 103                 if (stats.cr != stats.crlf)
 104                         return 0;
 105
 106                 /*
 107                  * And add some heuristics for binary vs text, of course...
 108                  */
 109                 if (is_binary(len, &stats))
 110                         return 0;
 111         }
 112
 113         /* only grow if not in place */
 114         if (strbuf_avail(buf) + buf->len < len)
 115                 strbuf_grow(buf, len - buf->len);
 116         dst = buf->buf;
 117         if (action == CRLF_GUESS) {
 118                 /*
 119                  * If we guessed, we already know we rejected a file with
 120                  * lone CR, and we can strip a CR without looking at what
 121                  * follow it.
 122                  */
 123                 do {
 124                         unsigned char c = *src++;
 125                         if (c != '\r')
 126                                 *dst++ = c;
 127                 } while (--len);
 128         } else {
 129                 do {
 130                         unsigned char c = *src++;
 131                         if (! (c == '\r' && (1 < len && *src == '\n')))
 132                                 *dst++ = c;
 133                 } while (--len);
 134         }
 135         strbuf_setlen(buf, dst - buf->buf);
 136         return 1;
 137 }
 138
 139 static int crlf_to_worktree(const char *path, const char *src, size_t len,
 140                             struct strbuf *buf, int action)
 141 {
 142         char *to_free = NULL;
 143         struct text_stat stats;
 144
 145         if ((action == CRLF_BINARY) || (action == CRLF_INPUT) ||
 146             auto_crlf <= 0)
 147                 return 0;
 148
 149         if (!len)
 150                 return 0;
 151
 152         gather_stats(src, len, &stats);
 153
 154         /* No LF? Nothing to convert, regardless. */
 155         if (!stats.lf)
 156                 return 0;
 157
 158         /* Was it already in CRLF format? */
 159         if (stats.lf == stats.crlf)
 160                 return 0;
 161
 162         if (action == CRLF_GUESS) {
 163                 /* If we have any bare CR characters, we're not going to touch it */
 164                 if (stats.cr != stats.crlf)
 165                         return 0;
 166
 167                 if (is_binary(len, &stats))
 168                         return 0;
 169         }
 170
 171         /* are we "faking" in place editing ? */
 172         if (src == buf->buf)
 173                 to_free = strbuf_detach(buf, NULL);
 174
 175         strbuf_grow(buf, len + stats.lf - stats.crlf);
 176         for (;;) {
 177                 const char *nl = memchr(src, '\n', len);
 178                 if (!nl)
 179                         break;
 180                 if (nl > src && nl[-1] == '\r') {
 181                         strbuf_add(buf, src, nl + 1 - src);
 182                 } else {
 183                         strbuf_add(buf, src, nl - src);
 184                         strbuf_addstr(buf, "\r\n");
 185                 }
 186                 len -= nl + 1 - src;
 187                 src  = nl + 1;
 188         }
 189         strbuf_add(buf, src, len);
 190
 191         free(to_free);
 192         return 1;
 193 }
 194
 195 static int filter_buffer(const char *path, const char *src,
 196                          unsigned long size, const char *cmd)
 197 {
 198         /*
 199          * Spawn cmd and feed the buffer contents through its stdin.
 200          */
 201         struct child_process child_process;
 202         int pipe_feed[2];
 203         int write_err, status;
 204
 205         memset(&child_process, 0, sizeof(child_process));
 206
 207         if (pipe(pipe_feed) < 0) {
 208                 error("cannot create pipe to run external filter %s", cmd);
 209                 return 1;
 210         }
 211
 212         child_process.pid = fork();
 213         if (child_process.pid < 0) {
 214                 error("cannot fork to run external filter %s", cmd);
 215                 close(pipe_feed[0]);
 216                 close(pipe_feed[1]);
 217                 return 1;
 218         }
 219         if (!child_process.pid) {
 220                 dup2(pipe_feed[0], 0);
 221                 close(pipe_feed[0]);
 222                 close(pipe_feed[1]);
 223                 execlp("sh", "sh", "-c", cmd, NULL);
 224                 return 1;
 225         }
 226         close(pipe_feed[0]);
 227
 228         write_err = (write_in_full(pipe_feed[1], src, size) < 0);
 229         if (close(pipe_feed[1]))
 230                 write_err = 1;
 231         if (write_err)
 232                 error("cannot feed the input to external filter %s", cmd);
 233
 234         status = finish_command(&child_process);
 235         if (status)
 236                 error("external filter %s failed %d", cmd, -status);
 237         return (write_err || status);
 238 }
 239
 240 static int apply_filter(const char *path, const char *src, size_t len,
 241                         struct strbuf *dst, const char *cmd)
 242 {
 243         /*
 244          * Create a pipeline to have the command filter the buffer's
 245          * contents.
 246          *
 247          * (child --> cmd) --> us
 248          */
 249         int pipe_feed[2];
 250         int status, ret = 1;
 251         struct child_process child_process;
 252         struct strbuf nbuf;
 253
 254         if (!cmd)
 255                 return 0;
 256
 257         memset(&child_process, 0, sizeof(child_process));
 258
 259         if (pipe(pipe_feed) < 0) {
 260                 error("cannot create pipe to run external filter %s", cmd);
 261                 return 0;
 262         }
 263
 264         fflush(NULL);
 265         child_process.pid = fork();
 266         if (child_process.pid < 0) {
 267                 error("cannot fork to run external filter %s", cmd);
 268                 close(pipe_feed[0]);
 269                 close(pipe_feed[1]);
 270                 return 0;
 271         }
 272         if (!child_process.pid) {
 273                 dup2(pipe_feed[1], 1);
 274                 close(pipe_feed[0]);
 275                 close(pipe_feed[1]);
 276                 exit(filter_buffer(path, src, len, cmd));
 277         }
 278         close(pipe_feed[1]);
 279
 280         strbuf_init(&nbuf, 0);
 281         if (strbuf_read(&nbuf, pipe_feed[0], len) < 0) {
 282                 error("read from external filter %s failed", cmd);
 283                 ret = 0;
 284         }
 285         if (close(pipe_feed[0])) {
 286                 error("read from external filter %s failed", cmd);
 287                 ret = 0;
 288         }
 289         status = finish_command(&child_process);
 290         if (status) {
 291                 error("external filter %s failed %d", cmd, -status);
 292                 ret = 0;
 293         }
 294
 295         if (ret) {
 296                 strbuf_swap(dst, &nbuf);
 297         }
 298         strbuf_release(&nbuf);
 299         return ret;
 300 }
 301
 302 static struct convert_driver {
 303         const char *name;
 304         struct convert_driver *next;
 305         char *smudge;
 306         char *clean;
 307 } *user_convert, **user_convert_tail;
 308
 309 static int read_convert_config(const char *var, const char *value)
 310 {
 311         const char *ep, *name;
 312         int namelen;
 313         struct convert_driver *drv;
 314
 315         /*
 316          * External conversion drivers are configured using
 317          * "filter.<name>.variable".
 318          */
 319         if (prefixcmp(var, "filter.") || (ep = strrchr(var, '.')) == var + 6)
 320                 return 0;
 321         name = var + 7;
 322         namelen = ep - name;
 323         for (drv = user_convert; drv; drv = drv->next)
 324                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
 325                         break;
 326         if (!drv) {
 327                 drv = xcalloc(1, sizeof(struct convert_driver));
 328                 drv->name = xmemdupz(name, namelen);
 329                 *user_convert_tail = drv;
 330                 user_convert_tail = &(drv->next);
 331         }
 332
 333         ep++;
 334
 335         /*
 336          * filter.<name>.smudge and filter.<name>.clean specifies
 337          * the command line:
 338          *
 339          *      command-line
 340          *
 341          * The command-line will not be interpolated in any way.
 342          */
 343
 344         if (!strcmp("smudge", ep)) {
 345                 if (!value)
 346                         return error("%s: lacks value", var);
 347                 drv->smudge = strdup(value);
 348                 return 0;
 349         }
 350
 351         if (!strcmp("clean", ep)) {
 352                 if (!value)
 353                         return error("%s: lacks value", var);
 354                 drv->clean = strdup(value);
 355                 return 0;
 356         }
 357         return 0;
 358 }
 359
 360 static void setup_convert_check(struct git_attr_check *check)
 361 {
 362         static struct git_attr *attr_crlf;
 363         static struct git_attr *attr_ident;
 364         static struct git_attr *attr_filter;
 365
 366         if (!attr_crlf) {
 367                 attr_crlf = git_attr("crlf", 4);
 368                 attr_ident = git_attr("ident", 5);
 369                 attr_filter = git_attr("filter", 6);
 370                 user_convert_tail = &user_convert;
 371                 git_config(read_convert_config);
 372         }
 373         check[0].attr = attr_crlf;
 374         check[1].attr = attr_ident;
 375         check[2].attr = attr_filter;
 376 }
 377
 378 static int count_ident(const char *cp, unsigned long size)
 379 {
 380         /*
 381          * "$Id: 0000000000000000000000000000000000000000 $" <=> "$Id$"
 382          */
 383         int cnt = 0;
 384         char ch;
 385
 386         while (size) {
 387                 ch = *cp++;
 388                 size--;
 389                 if (ch != '$')
 390                         continue;
 391                 if (size < 3)
 392                         break;
 393                 if (memcmp("Id", cp, 2))
 394                         continue;
 395                 ch = cp[2];
 396                 cp += 3;
 397                 size -= 3;
 398                 if (ch == '$')
 399                         cnt++; /* $Id$ */
 400                 if (ch != ':')
 401                         continue;
 402
 403                 /*
 404                  * "$Id: ... "; scan up to the closing dollar sign and discard.
 405                  */
 406                 while (size) {
 407                         ch = *cp++;
 408                         size--;
 409                         if (ch == '$') {
 410                                 cnt++;
 411                                 break;
 412                         }
 413                 }
 414         }
 415         return cnt;
 416 }
 417
 418 static int ident_to_git(const char *path, const char *src, size_t len,
 419                         struct strbuf *buf, int ident)
 420 {
 421         char *dst, *dollar;
 422
 423         if (!ident || !count_ident(src, len))
 424                 return 0;
 425
 426         /* only grow if not in place */
 427         if (strbuf_avail(buf) + buf->len < len)
 428                 strbuf_grow(buf, len - buf->len);
 429         dst = buf->buf;
 430         for (;;) {
 431                 dollar = memchr(src, '$', len);
 432                 if (!dollar)
 433                         break;
 434                 memcpy(dst, src, dollar + 1 - src);
 435                 dst += dollar + 1 - src;
 436                 len -= dollar + 1 - src;
 437                 src  = dollar + 1;
 438
 439                 if (len > 3 && !memcmp(src, "Id:", 3)) {
 440                         dollar = memchr(src + 3, '$', len - 3);
 441                         if (!dollar)
 442                                 break;
 443                         memcpy(dst, "Id$", 3);
 444                         dst += 3;
 445                         len -= dollar + 1 - src;
 446                         src  = dollar + 1;
 447                 }
 448         }
 449         memcpy(dst, src, len);
 450         strbuf_setlen(buf, dst + len - buf->buf);
 451         return 1;
 452 }
 453
 454 static int ident_to_worktree(const char *path, const char *src, size_t len,
 455                              struct strbuf *buf, int ident)
 456 {
 457         unsigned char sha1[20];
 458         char *to_free = NULL, *dollar;
 459         int cnt;
 460
 461         if (!ident)
 462                 return 0;
 463
 464         cnt = count_ident(src, len);
 465         if (!cnt)
 466                 return 0;
 467
 468         /* are we "faking" in place editing ? */
 469         if (src == buf->buf)
 470                 to_free = strbuf_detach(buf, NULL);
 471         hash_sha1_file(src, len, "blob", sha1);
 472
 473         strbuf_grow(buf, len + cnt * 43);
 474         for (;;) {
 475                 /* step 1: run to the next '$' */
 476                 dollar = memchr(src, '$', len);
 477                 if (!dollar)
 478                         break;
 479                 strbuf_add(buf, src, dollar + 1 - src);
 480                 len -= dollar + 1 - src;
 481                 src  = dollar + 1;
 482
 483                 /* step 2: does it looks like a bit like Id:xxx$ or Id$ ? */
 484                 if (len < 3 || memcmp("Id", src, 2))
 485                         continue;
 486
 487                 /* step 3: skip over Id$ or Id:xxxxx$ */
 488                 if (src[2] == '$') {
 489                         src += 3;
 490                         len -= 3;
 491                 } else if (src[2] == ':') {
 492                         /*
 493                          * It's possible that an expanded Id has crept its way into the
 494                          * repository, we cope with that by stripping the expansion out
 495                          */
 496                         dollar = memchr(src + 3, '$', len - 3);
 497                         if (!dollar) {
 498                                 /* incomplete keyword, no more '$', so just quit the loop */
 499                                 break;
 500                         }
 501
 502                         len -= dollar + 1 - src;
 503                         src  = dollar + 1;
 504                 } else {
 505                         /* it wasn't a "Id$" or "Id:xxxx$" */
 506                         continue;
 507                 }
 508
 509                 /* step 4: substitute */
 510                 strbuf_addstr(buf, "Id: ");
 511                 strbuf_add(buf, sha1_to_hex(sha1), 40);
 512                 strbuf_addstr(buf, " $");
 513         }
 514         strbuf_add(buf, src, len);
 515
 516         free(to_free);
 517         return 1;
 518 }
 519
 520 static int git_path_check_crlf(const char *path, struct git_attr_check *check)
 521 {
 522         const char *value = check->value;
 523
 524         if (ATTR_TRUE(value))
 525                 return CRLF_TEXT;
 526         else if (ATTR_FALSE(value))
 527                 return CRLF_BINARY;
 528         else if (ATTR_UNSET(value))
 529                 ;
 530         else if (!strcmp(value, "input"))
 531                 return CRLF_INPUT;
 532         return CRLF_GUESS;
 533 }
 534
 535 static struct convert_driver *git_path_check_convert(const char *path,
 536                                              struct git_attr_check *check)
 537 {
 538         const char *value = check->value;
 539         struct convert_driver *drv;
 540
 541         if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
 542                 return NULL;
 543         for (drv = user_convert; drv; drv = drv->next)
 544                 if (!strcmp(value, drv->name))
 545                         return drv;
 546         return NULL;
 547 }
 548
 549 static int git_path_check_ident(const char *path, struct git_attr_check *check)
 550 {
 551         const char *value = check->value;
 552
 553         return !!ATTR_TRUE(value);
 554 }
 555
 556 int convert_to_git(const char *path, const char *src, size_t len, struct strbuf *dst)
 557 {
 558         struct git_attr_check check[3];
 559         int crlf = CRLF_GUESS;
 560         int ident = 0, ret = 0;
 561         char *filter = NULL;
 562
 563         setup_convert_check(check);
 564         if (!git_checkattr(path, ARRAY_SIZE(check), check)) {
 565                 struct convert_driver *drv;
 566                 crlf = git_path_check_crlf(path, check + 0);
 567                 ident = git_path_check_ident(path, check + 1);
 568                 drv = git_path_check_convert(path, check + 2);
 569                 if (drv && drv->clean)
 570                         filter = drv->clean;
 571         }
 572
 573         ret |= apply_filter(path, src, len, dst, filter);
 574         if (ret) {
 575                 src = dst->buf;
 576                 len = dst->len;
 577         }
 578         ret |= crlf_to_git(path, src, len, dst, crlf);
 579         if (ret) {
 580                 src = dst->buf;
 581                 len = dst->len;
 582         }
 583         return ret | ident_to_git(path, src, len, dst, ident);
 584 }
 585
 586 int convert_to_working_tree(const char *path, const char *src, size_t len, struct strbuf *dst)
 587 {
 588         struct git_attr_check check[3];
 589         int crlf = CRLF_GUESS;
 590         int ident = 0, ret = 0;
 591         char *filter = NULL;
 592
 593         setup_convert_check(check);
 594         if (!git_checkattr(path, ARRAY_SIZE(check), check)) {
 595                 struct convert_driver *drv;
 596                 crlf = git_path_check_crlf(path, check + 0);
 597                 ident = git_path_check_ident(path, check + 1);
 598                 drv = git_path_check_convert(path, check + 2);
 599                 if (drv && drv->smudge)
 600                         filter = drv->smudge;
 601         }
 602
 603         ret |= ident_to_worktree(path, src, len, dst, ident);
 604         if (ret) {
 605                 src = dst->buf;
 606                 len = dst->len;
 607         }
 608         ret |= crlf_to_worktree(path, src, len, dst, crlf);
 609         if (ret) {
 610                 src = dst->buf;
 611                 len = dst->len;
 612         }
 613         return ret | apply_filter(path, src, len, dst, filter);
 614 }