usr.bin/patch/inp.c

   1 /*
   2  * $OpenBSD: inp.c,v 1.34 2006/03/11 19:41:30 otto Exp $
   3  * $DragonFly: src/usr.bin/patch/inp.c,v 1.6 2007/09/29 23:11:10 swildner Exp $
   4  * $NetBSD: inp.c,v 1.24 2015/07/24 18:56:00 christos Exp $
   5  */
   6
   7 /*
   8  * patch - a program to apply diffs to original files
   9  *
  10  * Copyright 1986, Larry Wall
  11  *
  12  * Redistribution and use in source and binary forms, with or without
  13  * modification, are permitted provided that the following condition is met:
  14  * 1. Redistributions of source code must retain the above copyright notice,
  15  * this condition and the following disclaimer.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND ANY
  18  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  19  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
  20  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
  21  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
  23  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
  24  * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  27  * SUCH DAMAGE.
  28  *
  29  * -C option added in 1998, original code by Marc Espie, based on FreeBSD
  30  * behaviour
  31  */
  32
  33 #include <sys/cdefs.h>
  34 __RCSID("$NetBSD: inp.c,v 1.24 2015/07/24 18:56:00 christos Exp $");
  35
  36 #include <sys/types.h>
  37 #include <sys/file.h>
  38 #include <sys/stat.h>
  39 #include <sys/mman.h>
  40 #include <sys/wait.h>
  41
  42 #include <ctype.h>
  43 #include <errno.h>
  44 #include <fcntl.h>
  45 #include <libgen.h>
  46 #include <limits.h>
  47 #include <stddef.h>
  48 #include <stdio.h>
  49 #include <stdlib.h>
  50 #include <string.h>
  51 #include <unistd.h>
  52
  53 #include "common.h"
  54 #include "util.h"
  55 #include "pch.h"
  56 #include "inp.h"
  57
  58
  59 /* Input-file-with-indexable-lines abstract type */
  60
  61 static off_t    i_size;         /* size of the input file */
  62 static char     *i_womp;        /* plan a buffer for entire file */
  63 static char     **i_ptr;        /* pointers to lines in i_womp */
  64 static char     empty_line[] = { '\0' };
  65
  66 static int      tifd = -1;      /* plan b virtual string array */
  67 static char     *tibuf[2];      /* plan b buffers */
  68 static LINENUM  tiline[2] = {-1, -1};   /* 1st line in each buffer */
  69 static LINENUM  lines_per_buf;  /* how many lines per buffer */
  70 static int      tireclen;       /* length of records in tmp file */
  71
  72 static bool     rev_in_string(const char *);
  73 static bool     reallocate_lines(size_t *);
  74
  75 /* returns false if insufficient memory */
  76 static bool     plan_a(const char *);
  77
  78 static void     plan_b(const char *);
  79
  80 /* New patch--prepare to edit another file. */
  81
  82 void
  83 re_input(void)
  84 {
  85         if (using_plan_a) {
  86                 i_size = 0;
  87                 free(i_ptr);
  88                 i_ptr = NULL;
  89                 if (i_womp != NULL) {
  90                         munmap(i_womp, i_size);
  91                         i_womp = NULL;
  92                 }
  93         } else {
  94                 using_plan_a = true;    /* maybe the next one is smaller */
  95                 close(tifd);
  96                 tifd = -1;
  97                 free(tibuf[0]);
  98                 free(tibuf[1]);
  99                 tibuf[0] = tibuf[1] = NULL;
 100                 tiline[0] = tiline[1] = -1;
 101                 tireclen = 0;
 102         }
 103 }
 104
 105 /* Construct the line index, somehow or other. */
 106
 107 void
 108 scan_input(const char *filename)
 109 {
 110         if (!plan_a(filename))
 111                 plan_b(filename);
 112         if (verbose) {
 113                 say("Patching file %s using Plan %s...\n", filename,
 114                     (using_plan_a ? "A" : "B"));
 115         }
 116 }
 117
 118 static bool
 119 reallocate_lines(size_t *lines_allocated)
 120 {
 121         char    **p;
 122         size_t  new_size;
 123
 124         new_size = *lines_allocated * 3 / 2;
 125         p = realloc(i_ptr, (new_size + 2) * sizeof(char *));
 126         if (p == NULL) {        /* shucks, it was a near thing */
 127                 munmap(i_womp, i_size);
 128                 i_womp = NULL;
 129                 free(i_ptr);
 130                 i_ptr = NULL;
 131                 *lines_allocated = 0;
 132                 return false;
 133         }
 134         *lines_allocated = new_size;
 135         i_ptr = p;
 136         return true;
 137 }
 138
 139 /* Try keeping everything in memory. */
 140
 141 static bool
 142 plan_a(const char *filename)
 143 {
 144         int             ifd, statfailed, devnull, pstat;
 145         char            *p, *s, lbuf[MAXLINELEN];
 146         struct stat     filestat;
 147         off_t           i;
 148         ptrdiff_t       sz;
 149         size_t          iline, lines_allocated;
 150         pid_t           pid;
 151         char            *argp[4] = {NULL};
 152
 153 #ifdef DEBUGGING
 154         if (debug & 8)
 155                 return false;
 156 #endif
 157
 158         if (filename == NULL || *filename == '\0')
 159                 return false;
 160
 161         statfailed = stat(filename, &filestat);
 162         if (statfailed && ok_to_create_file) {
 163                 if (verbose)
 164                         say("(Creating file %s...)\n", filename);
 165
 166                 /*
 167                  * in check_patch case, we still display `Creating file' even
 168                  * though we're not. The rule is that -C should be as similar
 169                  * to normal patch behavior as possible
 170                  */
 171                 if (check_only)
 172                         return true;
 173                 makedirs(filename, true);
 174                 close(creat(filename, 0666));
 175                 statfailed = stat(filename, &filestat);
 176         }
 177         if (statfailed && check_only)
 178                 fatal("%s not found, -C mode, can't probe further\n", filename);
 179         /* For nonexistent or read-only files, look for RCS versions.  */
 180         if (statfailed ||
 181             /* No one can write to it.  */
 182             (filestat.st_mode & 0222) == 0 ||
 183             /* I can't write to it.  */
 184             ((filestat.st_mode & 0022) == 0 && filestat.st_uid != getuid())) {
 185                 char    *filebase, *filedir;
 186                 struct stat     cstat;
 187                 char *tmp_filename1, *tmp_filename2;
 188
 189                 tmp_filename1 = strdup(filename);
 190                 tmp_filename2 = strdup(filename);
 191                 if (tmp_filename1 == NULL || tmp_filename2 == NULL)
 192                         fatal("strdupping filename");
 193
 194                 filebase = basename(tmp_filename1);
 195                 filedir = dirname(tmp_filename2);
 196
 197 #define try(f, a1, a2, a3) \
 198         (snprintf(lbuf, sizeof lbuf, f, a1, a2, a3), stat(lbuf, &cstat) == 0)
 199
 200                 /*
 201                  * else we can't write to it but it's not under a version
 202                  * control system, so just proceed.
 203                  */
 204                 if (try("%s/RCS/%s%s", filedir, filebase, RCSSUFFIX) ||
 205                     try("%s/RCS/%s%s", filedir, filebase, "") ||
 206                     try("%s/%s%s", filedir, filebase, RCSSUFFIX)) {
 207                         if (!statfailed) {
 208                                 if ((filestat.st_mode & 0222) != 0)
 209                                         /* The owner can write to it.  */
 210                                         fatal("file %s seems to be locked "
 211                                             "by somebody else under RCS\n",
 212                                             filename);
 213                                 /*
 214                                  * It might be checked out unlocked.  See if
 215                                  * it's safe to check out the default version
 216                                  * locked.
 217                                  */
 218                                 if (verbose)
 219                                         say("Comparing file %s to default "
 220                                             "RCS version...\n", filename);
 221
 222                                 switch (pid = fork()) {
 223                                 case -1:
 224                                         fatal("can't fork: %s\n",
 225                                             strerror(errno));
 226                                 case 0:
 227                                         devnull = open("/dev/null", O_RDONLY);
 228                                         if (devnull == -1) {
 229                                                 fatal("can't open /dev/null: %s",
 230                                                     strerror(errno));
 231                                         }
 232                                         (void)dup2(devnull, STDOUT_FILENO);
 233                                         argp[0] = __UNCONST(RCSDIFF);
 234                                         argp[1] = __UNCONST(filename);
 235                                         execv(RCSDIFF, argp);
 236                                         exit(127);
 237                                 }
 238                                 pid = waitpid(pid, &pstat, 0);
 239                                 if (pid == -1 || WEXITSTATUS(pstat) != 0) {
 240                                         fatal("can't check out file %s: "
 241                                             "differs from default RCS version\n",
 242                                             filename);
 243                                 }
 244                         }
 245
 246                         if (verbose)
 247                                 say("Checking out file %s from RCS...\n",
 248                                     filename);
 249
 250                         switch (pid = fork()) {
 251                         case -1:
 252                                 fatal("can't fork: %s\n", strerror(errno));
 253                         case 0:
 254                                 argp[0] = __UNCONST(CHECKOUT);
 255                                 argp[1] = __UNCONST("-l");
 256                                 argp[2] = __UNCONST(filename);
 257                                 execv(CHECKOUT, argp);
 258                                 exit(127);
 259                         }
 260                         pid = waitpid(pid, &pstat, 0);
 261                         if (pid == -1 || WEXITSTATUS(pstat) != 0 ||
 262                             stat(filename, &filestat)) {
 263                                 fatal("can't check out file %s from RCS\n",
 264                                     filename);
 265                         }
 266                 } else if (statfailed) {
 267                         fatal("can't find %s\n", filename);
 268                 }
 269                 free(tmp_filename1);
 270                 free(tmp_filename2);
 271         }
 272
 273         filemode = filestat.st_mode;
 274         if (!S_ISREG(filemode))
 275                 fatal("%s is not a normal file--can't patch\n", filename);
 276         i_size = filestat.st_size;
 277         if (out_of_mem) {
 278                 set_hunkmax();  /* make sure dynamic arrays are allocated */
 279                 out_of_mem = false;
 280                 return false;   /* force plan b because plan a bombed */
 281         }
 282         if ((uintmax_t)i_size > (uintmax_t)SIZE_MAX) {
 283                 say("block too large to mmap\n");
 284                 return false;
 285         }
 286         if ((ifd = open(filename, O_RDONLY)) < 0)
 287                 pfatal("can't open file %s", filename);
 288
 289         if (i_size) {
 290                 i_womp = mmap(NULL, i_size, PROT_READ, MAP_PRIVATE, ifd, 0);
 291                 if (i_womp == MAP_FAILED) {
 292                         perror("mmap failed");
 293                         i_womp = NULL;
 294                         close(ifd);
 295                         return false;
 296                 }
 297         } else {
 298                 i_womp = NULL;
 299         }
 300
 301         close(ifd);
 302 #if !defined(__minix)
 303         if (i_size)
 304                 madvise(i_womp, i_size, MADV_SEQUENTIAL);
 305 #endif /* !defined(__minix) */
 306
 307         /* estimate the number of lines */
 308         lines_allocated = i_size / 25;
 309         if (lines_allocated < 100)
 310                 lines_allocated = 100;
 311
 312         if (!reallocate_lines(&lines_allocated))
 313                 return false;
 314
 315         /* now scan the buffer and build pointer array */
 316         iline = 1;
 317         i_ptr[iline] = i_womp;
 318         /* test for NUL too, to maintain the behavior of the original code */
 319         for (s = i_womp, i = 0; i < i_size && *s != '\0'; s++, i++) {
 320                 if (*s == '\n') {
 321                         if (iline == lines_allocated) {
 322                                 if (!reallocate_lines(&lines_allocated))
 323                                         return false;
 324                         }
 325                         /* these are NOT NUL terminated */
 326                         i_ptr[++iline] = s + 1;
 327                 }
 328         }
 329         /* if the last line contains no EOL, append one */
 330         if (i_size > 0 && i_womp[i_size - 1] != '\n') {
 331                 last_line_missing_eol = true;
 332                 /* fix last line */
 333                 sz = s - i_ptr[iline];
 334                 p = malloc(sz + 1);
 335                 if (p == NULL) {
 336                         free(i_ptr);
 337                         i_ptr = NULL;
 338                         munmap(i_womp, i_size);
 339                         i_womp = NULL;
 340                         return false;
 341                 }
 342
 343                 memcpy(p, i_ptr[iline], sz);
 344                 p[sz] = '\n';
 345                 i_ptr[iline] = p;
 346                 /* count the extra line and make it point to some valid mem */
 347                 i_ptr[++iline] = empty_line;
 348         } else
 349                 last_line_missing_eol = false;
 350
 351         input_lines = iline - 1;
 352
 353         /* now check for revision, if any */
 354
 355         if (revision != NULL) {
 356                 if (!rev_in_string(i_womp)) {
 357                         if (force) {
 358                                 if (verbose)
 359                                         say("Warning: this file doesn't appear "
 360                                             "to be the %s version--patching anyway.\n",
 361                                             revision);
 362                         } else if (batch) {
 363                                 fatal("this file doesn't appear to be the "
 364                                     "%s version--aborting.\n",
 365                                     revision);
 366                         } else {
 367                                 ask("This file doesn't appear to be the "
 368                                     "%s version--patch anyway? [n] ",
 369                                     revision);
 370                                 if (*buf != 'y')
 371                                         fatal("aborted\n");
 372                         }
 373                 } else if (verbose)
 374                         say("Good.  This file appears to be the %s version.\n",
 375                             revision);
 376         }
 377         return true;            /* plan a will work */
 378 }
 379
 380 /* Keep (virtually) nothing in memory. */
 381
 382 static void
 383 plan_b(const char *filename)
 384 {
 385         FILE    *ifp;
 386         size_t  i = 0, j, maxlen = 1;
 387         char    *p;
 388         bool    found_revision = (revision == NULL);
 389
 390         using_plan_a = false;
 391         if ((ifp = fopen(filename, "r")) == NULL)
 392                 pfatal("can't open file %s", filename);
 393         unlink(TMPINNAME);
 394         if ((tifd = open(TMPINNAME, O_EXCL | O_CREAT | O_WRONLY, 0666)) < 0)
 395                 pfatal("can't open file %s", TMPINNAME);
 396         while (fgets(buf, buf_len, ifp) != NULL) {
 397                 if (revision != NULL && !found_revision && rev_in_string(buf))
 398                         found_revision = true;
 399                 if ((i = strlen(buf)) > maxlen)
 400                         maxlen = i;     /* find longest line */
 401         }
 402         last_line_missing_eol = i > 0 && buf[i - 1] != '\n';
 403         if (last_line_missing_eol && maxlen == i)
 404                 maxlen++;
 405
 406         if (revision != NULL) {
 407                 if (!found_revision) {
 408                         if (force) {
 409                                 if (verbose)
 410                                         say("Warning: this file doesn't appear "
 411                                             "to be the %s version--patching anyway.\n",
 412                                             revision);
 413                         } else if (batch) {
 414                                 fatal("this file doesn't appear to be the "
 415                                     "%s version--aborting.\n",
 416                                     revision);
 417                         } else {
 418                                 ask("This file doesn't appear to be the %s "
 419                                     "version--patch anyway? [n] ",
 420                                     revision);
 421                                 if (*buf != 'y')
 422                                         fatal("aborted\n");
 423                         }
 424                 } else if (verbose)
 425                         say("Good.  This file appears to be the %s version.\n",
 426                             revision);
 427         }
 428         fseek(ifp, 0L, SEEK_SET);       /* rewind file */
 429         lines_per_buf = BUFFERSIZE / maxlen;
 430         tireclen = maxlen;
 431         tibuf[0] = malloc(BUFFERSIZE + 1);
 432         if (tibuf[0] == NULL)
 433                 fatal("out of memory\n");
 434         tibuf[1] = malloc(BUFFERSIZE + 1);
 435         if (tibuf[1] == NULL)
 436                 fatal("out of memory\n");
 437         for (i = 1;; i++) {
 438                 p = tibuf[0] + maxlen * (i % lines_per_buf);
 439                 if (i % lines_per_buf == 0)     /* new block */
 440                         if (write(tifd, tibuf[0], BUFFERSIZE) < BUFFERSIZE)
 441                                 pfatal("can't write temp file");
 442                 if (fgets(p, maxlen + 1, ifp) == NULL) {
 443                         input_lines = i - 1;
 444                         if (i % lines_per_buf != 0)
 445                                 if (write(tifd, tibuf[0], BUFFERSIZE) < BUFFERSIZE)
 446                                         pfatal("can't write temp file");
 447                         break;
 448                 }
 449                 j = strlen(p);
 450                 /* These are '\n' terminated strings, so no need to add a NUL */
 451                 if (j == 0 || p[j - 1] != '\n')
 452                         p[j] = '\n';
 453         }
 454         fclose(ifp);
 455         close(tifd);
 456         if ((tifd = open(TMPINNAME, O_RDONLY)) < 0)
 457                 pfatal("can't reopen file %s", TMPINNAME);
 458 }
 459
 460 /*
 461  * Fetch a line from the input file, \n terminated, not necessarily \0.
 462  */
 463 char *
 464 ifetch(LINENUM line, int whichbuf)
 465 {
 466         if (line < 1 || line > input_lines) {
 467                 if (warn_on_invalid_line) {
 468                         say("No such line %ld in input file, ignoring\n", line);
 469                         warn_on_invalid_line = false;
 470                 }
 471                 return NULL;
 472         }
 473         if (using_plan_a)
 474                 return i_ptr[line];
 475         else {
 476                 LINENUM offline = line % lines_per_buf;
 477                 LINENUM baseline = line - offline;
 478
 479                 if (tiline[0] == baseline)
 480                         whichbuf = 0;
 481                 else if (tiline[1] == baseline)
 482                         whichbuf = 1;
 483                 else {
 484                         tiline[whichbuf] = baseline;
 485
 486                         if (lseek(tifd, (off_t) (baseline / lines_per_buf *
 487                             BUFFERSIZE), SEEK_SET) < 0)
 488                                 pfatal("cannot seek in the temporary input file");
 489
 490                         if (read(tifd, tibuf[whichbuf], BUFFERSIZE) < 0)
 491                                 pfatal("error reading tmp file %s", TMPINNAME);
 492                 }
 493                 return tibuf[whichbuf] + (tireclen * offline);
 494         }
 495 }
 496
 497 /*
 498  * True if the string argument contains the revision number we want.
 499  */
 500 static bool
 501 rev_in_string(const char *string)
 502 {
 503         const char      *s;
 504         size_t          patlen;
 505
 506         if (revision == NULL)
 507                 return true;
 508         patlen = strlen(revision);
 509         if (strnEQ(string, revision, patlen) && isspace((unsigned char)string[patlen]))
 510                 return true;
 511         for (s = string; *s; s++) {
 512                 if (isspace((unsigned char)*s) && strnEQ(s + 1, revision, patlen) &&
 513                     isspace((unsigned char)s[patlen + 1])) {
 514                         return true;
 515                 }
 516         }
 517         return false;
 518 }