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 char *crlf_to_git(const char *path, const char *src, unsigned long *sizep, int action)
  84 {
  85         char *buffer, *dst;
  86         unsigned long size, nsize;
  87         struct text_stat stats;
  88
  89         if ((action == CRLF_BINARY) || (action == CRLF_GUESS && !auto_crlf))
  90                 return NULL;
  91
  92         size = *sizep;
  93         if (!size)
  94                 return NULL;
  95
  96         gather_stats(src, size, &stats);
  97
  98         /* No CR? Nothing to convert, regardless. */
  99         if (!stats.cr)
 100                 return NULL;
 101
 102         if (action == CRLF_GUESS) {
 103                 /*
 104                  * We're currently not going to even try to convert stuff
 105                  * that has bare CR characters. Does anybody do that crazy
 106                  * stuff?
 107                  */
 108                 if (stats.cr != stats.crlf)
 109                         return NULL;
 110
 111                 /*
 112                  * And add some heuristics for binary vs text, of course...
 113                  */
 114                 if (is_binary(size, &stats))
 115                         return NULL;
 116         }
 117
 118         /*
 119          * Ok, allocate a new buffer, fill it in, and return it
 120          * to let the caller know that we switched buffers.
 121          */
 122         nsize = size - stats.crlf;
 123         buffer = xmalloc(nsize);
 124         *sizep = nsize;
 125
 126         dst = buffer;
 127         if (action == CRLF_GUESS) {
 128                 /*
 129                  * If we guessed, we already know we rejected a file with
 130                  * lone CR, and we can strip a CR without looking at what
 131                  * follow it.
 132                  */
 133                 do {
 134                         unsigned char c = *src++;
 135                         if (c != '\r')
 136                                 *dst++ = c;
 137                 } while (--size);
 138         } else {
 139                 do {
 140                         unsigned char c = *src++;
 141                         if (! (c == '\r' && (1 < size && *src == '\n')))
 142                                 *dst++ = c;
 143                 } while (--size);
 144         }
 145
 146         return buffer;
 147 }
 148
 149 static char *crlf_to_worktree(const char *path, const char *src, unsigned long *sizep, int action)
 150 {
 151         char *buffer, *dst;
 152         unsigned long size, nsize;
 153         struct text_stat stats;
 154         unsigned char last;
 155
 156         if ((action == CRLF_BINARY) || (action == CRLF_INPUT) ||
 157             (action == CRLF_GUESS && auto_crlf <= 0))
 158                 return NULL;
 159
 160         size = *sizep;
 161         if (!size)
 162                 return NULL;
 163
 164         gather_stats(src, size, &stats);
 165
 166         /* No LF? Nothing to convert, regardless. */
 167         if (!stats.lf)
 168                 return NULL;
 169
 170         /* Was it already in CRLF format? */
 171         if (stats.lf == stats.crlf)
 172                 return NULL;
 173
 174         if (action == CRLF_GUESS) {
 175                 /* If we have any bare CR characters, we're not going to touch it */
 176                 if (stats.cr != stats.crlf)
 177                         return NULL;
 178
 179                 if (is_binary(size, &stats))
 180                         return NULL;
 181         }
 182
 183         /*
 184          * Ok, allocate a new buffer, fill it in, and return it
 185          * to let the caller know that we switched buffers.
 186          */
 187         nsize = size + stats.lf - stats.crlf;
 188         buffer = xmalloc(nsize);
 189         *sizep = nsize;
 190         last = 0;
 191
 192         dst = buffer;
 193         do {
 194                 unsigned char c = *src++;
 195                 if (c == '\n' && last != '\r')
 196                         *dst++ = '\r';
 197                 *dst++ = c;
 198                 last = c;
 199         } while (--size);
 200
 201         return buffer;
 202 }
 203
 204 static int filter_buffer(const char *path, const char *src,
 205                          unsigned long size, const char *cmd)
 206 {
 207         /*
 208          * Spawn cmd and feed the buffer contents through its stdin.
 209          */
 210         struct child_process child_process;
 211         int pipe_feed[2];
 212         int write_err, status;
 213
 214         memset(&child_process, 0, sizeof(child_process));
 215
 216         if (pipe(pipe_feed) < 0) {
 217                 error("cannot create pipe to run external filter %s", cmd);
 218                 return 1;
 219         }
 220
 221         child_process.pid = fork();
 222         if (child_process.pid < 0) {
 223                 error("cannot fork to run external filter %s", cmd);
 224                 close(pipe_feed[0]);
 225                 close(pipe_feed[1]);
 226                 return 1;
 227         }
 228         if (!child_process.pid) {
 229                 dup2(pipe_feed[0], 0);
 230                 close(pipe_feed[0]);
 231                 close(pipe_feed[1]);
 232                 execlp("sh", "sh", "-c", cmd, NULL);
 233                 return 1;
 234         }
 235         close(pipe_feed[0]);
 236
 237         write_err = (write_in_full(pipe_feed[1], src, size) < 0);
 238         if (close(pipe_feed[1]))
 239                 write_err = 1;
 240         if (write_err)
 241                 error("cannot feed the input to external filter %s", cmd);
 242
 243         status = finish_command(&child_process);
 244         if (status)
 245                 error("external filter %s failed %d", cmd, -status);
 246         return (write_err || status);
 247 }
 248
 249 static char *apply_filter(const char *path, const char *src,
 250                           unsigned long *sizep, const char *cmd)
 251 {
 252         /*
 253          * Create a pipeline to have the command filter the buffer's
 254          * contents.
 255          *
 256          * (child --> cmd) --> us
 257          */
 258         const int SLOP = 4096;
 259         int pipe_feed[2];
 260         int status;
 261         char *dst;
 262         unsigned long dstsize, dstalloc;
 263         struct child_process child_process;
 264
 265         if (!cmd)
 266                 return NULL;
 267
 268         memset(&child_process, 0, sizeof(child_process));
 269
 270         if (pipe(pipe_feed) < 0) {
 271                 error("cannot create pipe to run external filter %s", cmd);
 272                 return NULL;
 273         }
 274
 275         fflush(NULL);
 276         child_process.pid = fork();
 277         if (child_process.pid < 0) {
 278                 error("cannot fork to run external filter %s", cmd);
 279                 close(pipe_feed[0]);
 280                 close(pipe_feed[1]);
 281                 return NULL;
 282         }
 283         if (!child_process.pid) {
 284                 dup2(pipe_feed[1], 1);
 285                 close(pipe_feed[0]);
 286                 close(pipe_feed[1]);
 287                 exit(filter_buffer(path, src, *sizep, cmd));
 288         }
 289         close(pipe_feed[1]);
 290
 291         dstalloc = *sizep;
 292         dst = xmalloc(dstalloc);
 293         dstsize = 0;
 294
 295         while (1) {
 296                 ssize_t numread = xread(pipe_feed[0], dst + dstsize,
 297                                         dstalloc - dstsize);
 298
 299                 if (numread <= 0) {
 300                         if (!numread)
 301                                 break;
 302                         error("read from external filter %s failed", cmd);
 303                         free(dst);
 304                         dst = NULL;
 305                         break;
 306                 }
 307                 dstsize += numread;
 308                 if (dstalloc <= dstsize + SLOP) {
 309                         dstalloc = dstsize + SLOP;
 310                         dst = xrealloc(dst, dstalloc);
 311                 }
 312         }
 313         if (close(pipe_feed[0])) {
 314                 error("read from external filter %s failed", cmd);
 315                 free(dst);
 316                 dst = NULL;
 317         }
 318
 319         status = finish_command(&child_process);
 320         if (status) {
 321                 error("external filter %s failed %d", cmd, -status);
 322                 free(dst);
 323                 dst = NULL;
 324         }
 325
 326         if (dst)
 327                 *sizep = dstsize;
 328         return dst;
 329 }
 330
 331 static struct convert_driver {
 332         const char *name;
 333         struct convert_driver *next;
 334         char *smudge;
 335         char *clean;
 336 } *user_convert, **user_convert_tail;
 337
 338 static int read_convert_config(const char *var, const char *value)
 339 {
 340         const char *ep, *name;
 341         int namelen;
 342         struct convert_driver *drv;
 343
 344         /*
 345          * External conversion drivers are configured using
 346          * "filter.<name>.variable".
 347          */
 348         if (prefixcmp(var, "filter.") || (ep = strrchr(var, '.')) == var + 6)
 349                 return 0;
 350         name = var + 7;
 351         namelen = ep - name;
 352         for (drv = user_convert; drv; drv = drv->next)
 353                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
 354                         break;
 355         if (!drv) {
 356                 char *namebuf;
 357                 drv = xcalloc(1, sizeof(struct convert_driver));
 358                 namebuf = xmalloc(namelen + 1);
 359                 memcpy(namebuf, name, namelen);
 360                 namebuf[namelen] = 0;
 361                 drv->name = namebuf;
 362                 drv->next = NULL;
 363                 *user_convert_tail = drv;
 364                 user_convert_tail = &(drv->next);
 365         }
 366
 367         ep++;
 368
 369         /*
 370          * filter.<name>.smudge and filter.<name>.clean specifies
 371          * the command line:
 372          *
 373          *      command-line
 374          *
 375          * The command-line will not be interpolated in any way.
 376          */
 377
 378         if (!strcmp("smudge", ep)) {
 379                 if (!value)
 380                         return error("%s: lacks value", var);
 381                 drv->smudge = strdup(value);
 382                 return 0;
 383         }
 384
 385         if (!strcmp("clean", ep)) {
 386                 if (!value)
 387                         return error("%s: lacks value", var);
 388                 drv->clean = strdup(value);
 389                 return 0;
 390         }
 391         return 0;
 392 }
 393
 394 static void setup_convert_check(struct git_attr_check *check)
 395 {
 396         static struct git_attr *attr_crlf;
 397         static struct git_attr *attr_ident;
 398         static struct git_attr *attr_filter;
 399
 400         if (!attr_crlf) {
 401                 attr_crlf = git_attr("crlf", 4);
 402                 attr_ident = git_attr("ident", 5);
 403                 attr_filter = git_attr("filter", 6);
 404                 user_convert_tail = &user_convert;
 405                 git_config(read_convert_config);
 406         }
 407         check[0].attr = attr_crlf;
 408         check[1].attr = attr_ident;
 409         check[2].attr = attr_filter;
 410 }
 411
 412 static int count_ident(const char *cp, unsigned long size)
 413 {
 414         /*
 415          * "$ident: 0000000000000000000000000000000000000000 $" <=> "$ident$"
 416          */
 417         int cnt = 0;
 418         char ch;
 419
 420         while (size) {
 421                 ch = *cp++;
 422                 size--;
 423                 if (ch != '$')
 424                         continue;
 425                 if (size < 6)
 426                         break;
 427                 if (memcmp("ident", cp, 5))
 428                         continue;
 429                 ch = cp[5];
 430                 cp += 6;
 431                 size -= 6;
 432                 if (ch == '$')
 433                         cnt++; /* $ident$ */
 434                 if (ch != ':')
 435                         continue;
 436
 437                 /*
 438                  * "$ident: ... "; scan up to the closing dollar sign and discard.
 439                  */
 440                 while (size) {
 441                         ch = *cp++;
 442                         size--;
 443                         if (ch == '$') {
 444                                 cnt++;
 445                                 break;
 446                         }
 447                 }
 448         }
 449         return cnt;
 450 }
 451
 452 static char *ident_to_git(const char *path, const char *src, unsigned long *sizep, int ident)
 453 {
 454         int cnt;
 455         unsigned long size;
 456         char *dst, *buf;
 457
 458         if (!ident)
 459                 return NULL;
 460         size = *sizep;
 461         cnt = count_ident(src, size);
 462         if (!cnt)
 463                 return NULL;
 464         buf = xmalloc(size);
 465
 466         for (dst = buf; size; size--) {
 467                 char ch = *src++;
 468                 *dst++ = ch;
 469                 if ((ch == '$') && (6 <= size) &&
 470                     !memcmp("ident:", src, 6)) {
 471                         unsigned long rem = size - 6;
 472                         const char *cp = src + 6;
 473                         do {
 474                                 ch = *cp++;
 475                                 if (ch == '$')
 476                                         break;
 477                                 rem--;
 478                         } while (rem);
 479                         if (!rem)
 480                                 continue;
 481                         memcpy(dst, "ident$", 6);
 482                         dst += 6;
 483                         size -= (cp - src);
 484                         src = cp;
 485                 }
 486         }
 487
 488         *sizep = dst - buf;
 489         return buf;
 490 }
 491
 492 static char *ident_to_worktree(const char *path, const char *src, unsigned long *sizep, int ident)
 493 {
 494         int cnt;
 495         unsigned long size;
 496         char *dst, *buf;
 497         unsigned char sha1[20];
 498
 499         if (!ident)
 500                 return NULL;
 501
 502         size = *sizep;
 503         cnt = count_ident(src, size);
 504         if (!cnt)
 505                 return NULL;
 506
 507         hash_sha1_file(src, size, "blob", sha1);
 508         buf = xmalloc(size + cnt * 43);
 509
 510         for (dst = buf; size; size--) {
 511                 const char *cp;
 512                 char ch = *src++;
 513                 *dst++ = ch;
 514                 if ((ch != '$') || (size < 6) || memcmp("ident", src, 5))
 515                         continue;
 516
 517                 if (src[5] == ':') {
 518                         /* discard up to but not including the closing $ */
 519                         unsigned long rem = size - 6;
 520                         cp = src + 6;
 521                         do {
 522                                 ch = *cp++;
 523                                 if (ch == '$')
 524                                         break;
 525                                 rem--;
 526                         } while (rem);
 527                         if (!rem)
 528                                 continue;
 529                         size -= (cp - src);
 530                 } else if (src[5] == '$')
 531                         cp = src + 5;
 532                 else
 533                         continue;
 534
 535                 memcpy(dst, "ident: ", 7);
 536                 dst += 7;
 537                 memcpy(dst, sha1_to_hex(sha1), 40);
 538                 dst += 40;
 539                 *dst++ = ' ';
 540                 size -= (cp - src);
 541                 src = cp;
 542                 *dst++ = *src++;
 543                 size--;
 544         }
 545
 546         *sizep = dst - buf;
 547         return buf;
 548 }
 549
 550 static int git_path_check_crlf(const char *path, struct git_attr_check *check)
 551 {
 552         const char *value = check->value;
 553
 554         if (ATTR_TRUE(value))
 555                 return CRLF_TEXT;
 556         else if (ATTR_FALSE(value))
 557                 return CRLF_BINARY;
 558         else if (ATTR_UNSET(value))
 559                 ;
 560         else if (!strcmp(value, "input"))
 561                 return CRLF_INPUT;
 562         return CRLF_GUESS;
 563 }
 564
 565 static struct convert_driver *git_path_check_convert(const char *path,
 566                                              struct git_attr_check *check)
 567 {
 568         const char *value = check->value;
 569         struct convert_driver *drv;
 570
 571         if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
 572                 return NULL;
 573         for (drv = user_convert; drv; drv = drv->next)
 574                 if (!strcmp(value, drv->name))
 575                         return drv;
 576         return NULL;
 577 }
 578
 579 static int git_path_check_ident(const char *path, struct git_attr_check *check)
 580 {
 581         const char *value = check->value;
 582
 583         return !!ATTR_TRUE(value);
 584 }
 585
 586 char *convert_to_git(const char *path, const char *src, unsigned long *sizep)
 587 {
 588         struct git_attr_check check[3];
 589         int crlf = CRLF_GUESS;
 590         int ident = 0;
 591         char *filter = NULL;
 592         char *buf, *buf2;
 593
 594         setup_convert_check(check);
 595         if (!git_checkattr(path, ARRAY_SIZE(check), check)) {
 596                 struct convert_driver *drv;
 597                 crlf = git_path_check_crlf(path, check + 0);
 598                 ident = git_path_check_ident(path, check + 1);
 599                 drv = git_path_check_convert(path, check + 2);
 600                 if (drv && drv->clean)
 601                         filter = drv->clean;
 602         }
 603
 604         buf = apply_filter(path, src, sizep, filter);
 605
 606         buf2 = crlf_to_git(path, buf ? buf : src, sizep, crlf);
 607         if (buf2) {
 608                 free(buf);
 609                 buf = buf2;
 610         }
 611
 612         buf2 = ident_to_git(path, buf ? buf : src, sizep, ident);
 613         if (buf2) {
 614                 free(buf);
 615                 buf = buf2;
 616         }
 617
 618         return buf;
 619 }
 620
 621 char *convert_to_working_tree(const char *path, const char *src, unsigned long *sizep)
 622 {
 623         struct git_attr_check check[3];
 624         int crlf = CRLF_GUESS;
 625         int ident = 0;
 626         char *filter = NULL;
 627         char *buf, *buf2;
 628
 629         setup_convert_check(check);
 630         if (!git_checkattr(path, ARRAY_SIZE(check), check)) {
 631                 struct convert_driver *drv;
 632                 crlf = git_path_check_crlf(path, check + 0);
 633                 ident = git_path_check_ident(path, check + 1);
 634                 drv = git_path_check_convert(path, check + 2);
 635                 if (drv && drv->smudge)
 636                         filter = drv->smudge;
 637         }
 638
 639         buf = ident_to_worktree(path, src, sizep, ident);
 640
 641         buf2 = crlf_to_worktree(path, buf ? buf : src, sizep, crlf);
 642         if (buf2) {
 643                 free(buf);
 644                 buf = buf2;
 645         }
 646
 647         buf2 = apply_filter(path, buf ? buf : src, sizep, filter);
 648         if (buf2) {
 649                 free(buf);
 650                 buf = buf2;
 651         }
 652
 653         return buf;
 654 }