usr.bin/patch/inp.c

   1 /*-
   2  * Copyright 1986, Larry Wall
   3  *
   4  * Redistribution and use in source and binary forms, with or without
   5  * modification, are permitted provided that the following condition is met:
   6  * 1. Redistributions of source code must retain the above copyright notice,
   7  * this condition and the following disclaimer.
   8  *
   9  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND ANY
  10  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  11  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
  12  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
  13  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  14  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
  15  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
  16  * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  17  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  18  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  19  * SUCH DAMAGE.
  20  *
  21  * patch - a program to apply diffs to original files
  22  *
  23  * -C option added in 1998, original code by Marc Espie, based on FreeBSD
  24  * behaviour
  25  *
  26  * $OpenBSD: inp.c,v 1.36 2012/04/10 14:46:34 ajacoutot Exp $
  27  * $FreeBSD$
  28  */
  29
  30 #include <sys/types.h>
  31 #include <sys/file.h>
  32 #include <sys/stat.h>
  33 #include <sys/mman.h>
  34 #include <sys/wait.h>
  35
  36 #include <ctype.h>
  37 #include <errno.h>
  38 #include <libgen.h>
  39 #include <paths.h>
  40 #include <spawn.h>
  41 #include <stddef.h>
  42 #include <stdint.h>
  43 #include <stdio.h>
  44 #include <stdlib.h>
  45 #include <string.h>
  46 #include <unistd.h>
  47
  48 #include "common.h"
  49 #include "util.h"
  50 #include "pch.h"
  51 #include "inp.h"
  52
  53
  54 /* Input-file-with-indexable-lines abstract type */
  55
  56 static size_t   i_size;         /* size of the input file */
  57 static char     *i_womp;        /* plan a buffer for entire file */
  58 static char     **i_ptr;        /* pointers to lines in i_womp */
  59 static char     empty_line[] = { '\0' };
  60
  61 static int      tifd = -1;      /* plan b virtual string array */
  62 static char     *tibuf[2];      /* plan b buffers */
  63 static LINENUM  tiline[2] = {-1, -1};   /* 1st line in each buffer */
  64 static LINENUM  lines_per_buf;  /* how many lines per buffer */
  65 static int      tireclen;       /* length of records in tmp file */
  66
  67 static bool     rev_in_string(const char *);
  68 static bool     reallocate_lines(size_t *);
  69
  70 /* returns false if insufficient memory */
  71 static bool     plan_a(const char *);
  72
  73 static void     plan_b(const char *);
  74
  75 /* New patch--prepare to edit another file. */
  76
  77 void
  78 re_input(void)
  79 {
  80         if (using_plan_a) {
  81                 free(i_ptr);
  82                 i_ptr = NULL;
  83                 if (i_womp != NULL) {
  84                         munmap(i_womp, i_size);
  85                         i_womp = NULL;
  86                 }
  87                 i_size = 0;
  88         } else {
  89                 using_plan_a = true;    /* maybe the next one is smaller */
  90                 close(tifd);
  91                 tifd = -1;
  92                 free(tibuf[0]);
  93                 free(tibuf[1]);
  94                 tibuf[0] = tibuf[1] = NULL;
  95                 tiline[0] = tiline[1] = -1;
  96                 tireclen = 0;
  97         }
  98 }
  99
 100 /* Construct the line index, somehow or other. */
 101
 102 void
 103 scan_input(const char *filename)
 104 {
 105         if (!plan_a(filename))
 106                 plan_b(filename);
 107         if (verbose) {
 108                 say("Patching file %s using Plan %s...\n", filename,
 109                     (using_plan_a ? "A" : "B"));
 110         }
 111 }
 112
 113 static bool
 114 reallocate_lines(size_t *lines_allocated)
 115 {
 116         char    **p;
 117         size_t  new_size;
 118
 119         new_size = *lines_allocated * 3 / 2;
 120         p = realloc(i_ptr, (new_size + 2) * sizeof(char *));
 121         if (p == NULL) {        /* shucks, it was a near thing */
 122                 munmap(i_womp, i_size);
 123                 i_womp = NULL;
 124                 free(i_ptr);
 125                 i_ptr = NULL;
 126                 *lines_allocated = 0;
 127                 return false;
 128         }
 129         *lines_allocated = new_size;
 130         i_ptr = p;
 131         return true;
 132 }
 133
 134 /* Try keeping everything in memory. */
 135
 136 static bool
 137 plan_a(const char *filename)
 138 {
 139         int             ifd, statfailed, pstat;
 140         char            *p, *s, lbuf[INITLINELEN];
 141         struct stat     filestat;
 142         ptrdiff_t       sz;
 143         size_t          i;
 144         size_t          iline, lines_allocated;
 145         pid_t           pid;
 146
 147 #ifdef DEBUGGING
 148         if (debug & 8)
 149                 return false;
 150 #endif
 151
 152         if (filename == NULL || *filename == '\0')
 153                 return false;
 154
 155         statfailed = stat(filename, &filestat);
 156         if (statfailed && ok_to_create_file) {
 157                 if (verbose)
 158                         say("(Creating file %s...)\n", filename);
 159
 160                 /*
 161                  * in check_patch case, we still display `Creating file' even
 162                  * though we're not. The rule is that -C should be as similar
 163                  * to normal patch behavior as possible
 164                  */
 165                 if (check_only)
 166                         return true;
 167                 makedirs(filename, true);
 168                 close(creat(filename, 0666));
 169                 statfailed = stat(filename, &filestat);
 170         }
 171         if (statfailed && check_only)
 172                 fatal("%s not found, -C mode, can't probe further\n", filename);
 173         /* For nonexistent or read-only files, look for RCS versions.  */
 174
 175         if (statfailed ||
 176             /* No one can write to it.  */
 177             (filestat.st_mode & 0222) == 0 ||
 178             /* I can't write to it.  */
 179             ((filestat.st_mode & 0022) == 0 && filestat.st_uid != getuid())) {
 180                 char    *filebase, *filedir;
 181                 struct stat     cstat;
 182                 char    *tmp_filename1, *tmp_filename2;
 183                 char    *argp[4] = { NULL };
 184                 posix_spawn_file_actions_t file_actions;
 185
 186                 tmp_filename1 = strdup(filename);
 187                 tmp_filename2 = strdup(filename);
 188                 if (tmp_filename1 == NULL || tmp_filename2 == NULL)
 189                         fatal("strdupping filename");
 190
 191                 filebase = basename(tmp_filename1);
 192                 filedir = dirname(tmp_filename2);
 193
 194                 memset(argp, 0, sizeof(argp));
 195
 196 #define try(f, a1, a2, a3) \
 197         (snprintf(lbuf, sizeof(lbuf), f, a1, a2, a3), stat(lbuf, &cstat) == 0)
 198
 199                 /*
 200                  * else we can't write to it but it's not under a version
 201                  * control system, so just proceed.
 202                  */
 203                 if (try("%s/RCS/%s%s", filedir, filebase, RCSSUFFIX) ||
 204                     try("%s/RCS/%s%s", filedir, filebase, "") ||
 205                     try("%s/%s%s", filedir, filebase, RCSSUFFIX)) {
 206                         if (!statfailed) {
 207                                 if ((filestat.st_mode & 0222) != 0)
 208                                         /* The owner can write to it.  */
 209                                         fatal("file %s seems to be locked "
 210                                             "by somebody else under RCS\n",
 211                                             filename);
 212                                 /*
 213                                  * It might be checked out unlocked.  See if
 214                                  * it's safe to check out the default version
 215                                  * locked.
 216                                  */
 217                                 if (verbose)
 218                                         say("Comparing file %s to default "
 219                                             "RCS version...\n", filename);
 220
 221                                 argp[0] = __DECONST(char *, RCSDIFF);
 222                                 argp[1] = __DECONST(char *, filename);
 223                                 posix_spawn_file_actions_init(&file_actions);
 224                                 posix_spawn_file_actions_addopen(&file_actions,
 225                                     STDOUT_FILENO, _PATH_DEVNULL, O_WRONLY, 0);
 226                                 if (posix_spawn(&pid, RCSDIFF, &file_actions,
 227                                     NULL, argp, NULL) == 0) {
 228                                         pid = waitpid(pid, &pstat, 0);
 229                                         if (pid == -1 || WEXITSTATUS(pstat) != 0)
 230                                                 fatal("can't check out file %s: "
 231                                                     "differs from default RCS version\n",
 232                                                     filename);
 233                                 } else
 234                                         fatal("posix_spawn: %s\n", strerror(errno));
 235                                 posix_spawn_file_actions_destroy(&file_actions);
 236                         }
 237
 238                         if (verbose)
 239                                 say("Checking out file %s from RCS...\n",
 240                                     filename);
 241
 242                         argp[0] = __DECONST(char *, CHECKOUT);
 243                         argp[1] = __DECONST(char *, "-l");
 244                         argp[2] = __DECONST(char *, filename);
 245                         if (posix_spawn(&pid, CHECKOUT, NULL, NULL, argp,
 246                             NULL) == 0) {
 247                                 pid = waitpid(pid, &pstat, 0);
 248                                 if (pid == -1 || WEXITSTATUS(pstat) != 0 ||
 249                                     stat(filename, &filestat))
 250                                         fatal("can't check out file %s from RCS\n",
 251                                             filename);
 252                         } else
 253                                 fatal("posix_spawn: %s\n", strerror(errno));
 254                 } else if (statfailed) {
 255                         fatal("can't find %s\n", filename);
 256                 }
 257                 free(tmp_filename1);
 258                 free(tmp_filename2);
 259         }
 260
 261         filemode = filestat.st_mode;
 262         if (!S_ISREG(filemode))
 263                 fatal("%s is not a normal file--can't patch\n", filename);
 264         if ((uint64_t)filestat.st_size > SIZE_MAX) {
 265                 say("block too large to mmap\n");
 266                 return false;
 267         }
 268         i_size = (size_t)filestat.st_size;
 269         if (out_of_mem) {
 270                 set_hunkmax();  /* make sure dynamic arrays are allocated */
 271                 out_of_mem = false;
 272                 return false;   /* force plan b because plan a bombed */
 273         }
 274         if ((ifd = open(filename, O_RDONLY)) < 0)
 275                 pfatal("can't open file %s", filename);
 276
 277         if (i_size) {
 278                 i_womp = mmap(NULL, i_size, PROT_READ, MAP_PRIVATE, ifd, 0);
 279                 if (i_womp == MAP_FAILED) {
 280                         perror("mmap failed");
 281                         i_womp = NULL;
 282                         close(ifd);
 283                         return false;
 284                 }
 285         } else {
 286                 i_womp = NULL;
 287         }
 288
 289         close(ifd);
 290         if (i_size)
 291                 madvise(i_womp, i_size, MADV_SEQUENTIAL);
 292
 293         /* estimate the number of lines */
 294         lines_allocated = i_size / 25;
 295         if (lines_allocated < 100)
 296                 lines_allocated = 100;
 297
 298         if (!reallocate_lines(&lines_allocated))
 299                 return false;
 300
 301         /* now scan the buffer and build pointer array */
 302         iline = 1;
 303         i_ptr[iline] = i_womp;
 304         /* test for NUL too, to maintain the behavior of the original code */
 305         for (s = i_womp, i = 0; i < i_size && *s != '\0'; s++, i++) {
 306                 if (*s == '\n') {
 307                         if (iline == lines_allocated) {
 308                                 if (!reallocate_lines(&lines_allocated))
 309                                         return false;
 310                         }
 311                         /* these are NOT NUL terminated */
 312                         i_ptr[++iline] = s + 1;
 313                 }
 314         }
 315         /* if the last line contains no EOL, append one */
 316         if (i_size > 0 && i_womp[i_size - 1] != '\n') {
 317                 last_line_missing_eol = true;
 318                 /* fix last line */
 319                 sz = s - i_ptr[iline];
 320                 p = malloc(sz + 1);
 321                 if (p == NULL) {
 322                         free(i_ptr);
 323                         i_ptr = NULL;
 324                         munmap(i_womp, i_size);
 325                         i_womp = NULL;
 326                         return false;
 327                 }
 328
 329                 memcpy(p, i_ptr[iline], sz);
 330                 p[sz] = '\n';
 331                 i_ptr[iline] = p;
 332                 /* count the extra line and make it point to some valid mem */
 333                 i_ptr[++iline] = empty_line;
 334         } else
 335                 last_line_missing_eol = false;
 336
 337         input_lines = iline - 1;
 338
 339         /* now check for revision, if any */
 340
 341         if (revision != NULL) {
 342                 if (!rev_in_string(i_womp)) {
 343                         if (force) {
 344                                 if (verbose)
 345                                         say("Warning: this file doesn't appear "
 346                                             "to be the %s version--patching anyway.\n",
 347                                             revision);
 348                         } else if (batch) {
 349                                 fatal("this file doesn't appear to be the "
 350                                     "%s version--aborting.\n",
 351                                     revision);
 352                         } else {
 353                                 ask("This file doesn't appear to be the "
 354                                     "%s version--patch anyway? [n] ",
 355                                     revision);
 356                                 if (*buf != 'y')
 357                                         fatal("aborted\n");
 358                         }
 359                 } else if (verbose)
 360                         say("Good.  This file appears to be the %s version.\n",
 361                             revision);
 362         }
 363         return true;            /* plan a will work */
 364 }
 365
 366 /* Keep (virtually) nothing in memory. */
 367
 368 static void
 369 plan_b(const char *filename)
 370 {
 371         FILE    *ifp;
 372         size_t  i = 0, j, maxlen = 1;
 373         char    *p;
 374         bool    found_revision = (revision == NULL);
 375
 376         using_plan_a = false;
 377         if ((ifp = fopen(filename, "r")) == NULL)
 378                 pfatal("can't open file %s", filename);
 379         unlink(TMPINNAME);
 380         if ((tifd = open(TMPINNAME, O_EXCL | O_CREAT | O_WRONLY, 0666)) < 0)
 381                 pfatal("can't open file %s", TMPINNAME);
 382         while (fgets(buf, buf_size, ifp) != NULL) {
 383                 if (revision != NULL && !found_revision && rev_in_string(buf))
 384                         found_revision = true;
 385                 if ((i = strlen(buf)) > maxlen)
 386                         maxlen = i;     /* find longest line */
 387         }
 388         last_line_missing_eol = i > 0 && buf[i - 1] != '\n';
 389         if (last_line_missing_eol && maxlen == i)
 390                 maxlen++;
 391
 392         if (revision != NULL) {
 393                 if (!found_revision) {
 394                         if (force) {
 395                                 if (verbose)
 396                                         say("Warning: this file doesn't appear "
 397                                             "to be the %s version--patching anyway.\n",
 398                                             revision);
 399                         } else if (batch) {
 400                                 fatal("this file doesn't appear to be the "
 401                                     "%s version--aborting.\n",
 402                                     revision);
 403                         } else {
 404                                 ask("This file doesn't appear to be the %s "
 405                                     "version--patch anyway? [n] ",
 406                                     revision);
 407                                 if (*buf != 'y')
 408                                         fatal("aborted\n");
 409                         }
 410                 } else if (verbose)
 411                         say("Good.  This file appears to be the %s version.\n",
 412                             revision);
 413         }
 414         fseek(ifp, 0L, SEEK_SET);       /* rewind file */
 415         lines_per_buf = BUFFERSIZE / maxlen;
 416         tireclen = maxlen;
 417         tibuf[0] = malloc(BUFFERSIZE + 1);
 418         if (tibuf[0] == NULL)
 419                 fatal("out of memory\n");
 420         tibuf[1] = malloc(BUFFERSIZE + 1);
 421         if (tibuf[1] == NULL)
 422                 fatal("out of memory\n");
 423         for (i = 1;; i++) {
 424                 p = tibuf[0] + maxlen * (i % lines_per_buf);
 425                 if (i % lines_per_buf == 0)     /* new block */
 426                         if (write(tifd, tibuf[0], BUFFERSIZE) < BUFFERSIZE)
 427                                 pfatal("can't write temp file");
 428                 if (fgets(p, maxlen + 1, ifp) == NULL) {
 429                         input_lines = i - 1;
 430                         if (i % lines_per_buf != 0)
 431                                 if (write(tifd, tibuf[0], BUFFERSIZE) < BUFFERSIZE)
 432                                         pfatal("can't write temp file");
 433                         break;
 434                 }
 435                 j = strlen(p);
 436                 /* These are '\n' terminated strings, so no need to add a NUL */
 437                 if (j == 0 || p[j - 1] != '\n')
 438                         p[j] = '\n';
 439         }
 440         fclose(ifp);
 441         close(tifd);
 442         if ((tifd = open(TMPINNAME, O_RDONLY)) < 0)
 443                 pfatal("can't reopen file %s", TMPINNAME);
 444 }
 445
 446 /*
 447  * Fetch a line from the input file, \n terminated, not necessarily \0.
 448  */
 449 char *
 450 ifetch(LINENUM line, int whichbuf)
 451 {
 452         if (line < 1 || line > input_lines) {
 453                 if (warn_on_invalid_line) {
 454                         say("No such line %ld in input file, ignoring\n", line);
 455                         warn_on_invalid_line = false;
 456                 }
 457                 return NULL;
 458         }
 459         if (using_plan_a)
 460                 return i_ptr[line];
 461         else {
 462                 LINENUM offline = line % lines_per_buf;
 463                 LINENUM baseline = line - offline;
 464
 465                 if (tiline[0] == baseline)
 466                         whichbuf = 0;
 467                 else if (tiline[1] == baseline)
 468                         whichbuf = 1;
 469                 else {
 470                         tiline[whichbuf] = baseline;
 471
 472                         if (lseek(tifd, (off_t) (baseline / lines_per_buf *
 473                             BUFFERSIZE), SEEK_SET) < 0)
 474                                 pfatal("cannot seek in the temporary input file");
 475
 476                         if (read(tifd, tibuf[whichbuf], BUFFERSIZE) < 0)
 477                                 pfatal("error reading tmp file %s", TMPINNAME);
 478                 }
 479                 return tibuf[whichbuf] + (tireclen * offline);
 480         }
 481 }
 482
 483 /*
 484  * True if the string argument contains the revision number we want.
 485  */
 486 static bool
 487 rev_in_string(const char *string)
 488 {
 489         const char      *s;
 490         size_t          patlen;
 491
 492         if (revision == NULL)
 493                 return true;
 494         patlen = strlen(revision);
 495         if (strnEQ(string, revision, patlen) && isspace((unsigned char)string[patlen]))
 496                 return true;
 497         for (s = string; *s; s++) {
 498                 if (isspace((unsigned char)*s) && strnEQ(s + 1, revision, patlen) &&
 499                     isspace((unsigned char)s[patlen + 1])) {
 500                         return true;
 501                 }
 502         }
 503         return false;
 504 }