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: head/usr.bin/patch/inp.c 246091 2013-01-29 20:05:16Z delphij $
  28  */
  29
  30 #include <sys/types.h>
  31 #include <sys/file.h>
  32 #include <sys/stat.h>
  33 #include <sys/mman.h>
  34
  35 #include <ctype.h>
  36 #include <libgen.h>
  37 #include <limits.h>
  38 #include <stddef.h>
  39 #include <stdio.h>
  40 #include <stdlib.h>
  41 #include <string.h>
  42 #include <unistd.h>
  43
  44 #include "common.h"
  45 #include "util.h"
  46 #include "pch.h"
  47 #include "inp.h"
  48
  49
  50 /* Input-file-with-indexable-lines abstract type */
  51
  52 static size_t   i_size;         /* size of the input file */
  53 static char     *i_womp;        /* plan a buffer for entire file */
  54 static char     **i_ptr;        /* pointers to lines in i_womp */
  55 static char     empty_line[] = { '\0' };
  56
  57 static int      tifd = -1;      /* plan b virtual string array */
  58 static char     *tibuf[2];      /* plan b buffers */
  59 static LINENUM  tiline[2] = {-1, -1};   /* 1st line in each buffer */
  60 static LINENUM  lines_per_buf;  /* how many lines per buffer */
  61 static int      tireclen;       /* length of records in tmp file */
  62
  63 static bool     rev_in_string(const char *);
  64 static bool     reallocate_lines(size_t *);
  65
  66 /* returns false if insufficient memory */
  67 static bool     plan_a(const char *);
  68
  69 static void     plan_b(const char *);
  70
  71 /* New patch--prepare to edit another file. */
  72
  73 void
  74 re_input(void)
  75 {
  76         if (using_plan_a) {
  77                 free(i_ptr);
  78                 i_ptr = NULL;
  79                 if (i_womp != NULL) {
  80                         munmap(i_womp, i_size);
  81                         i_womp = NULL;
  82                 }
  83                 i_size = 0;
  84         } else {
  85                 using_plan_a = true;    /* maybe the next one is smaller */
  86                 close(tifd);
  87                 tifd = -1;
  88                 free(tibuf[0]);
  89                 free(tibuf[1]);
  90                 tibuf[0] = tibuf[1] = NULL;
  91                 tiline[0] = tiline[1] = -1;
  92                 tireclen = 0;
  93         }
  94 }
  95
  96 /* Construct the line index, somehow or other. */
  97
  98 void
  99 scan_input(const char *filename)
 100 {
 101         if (!plan_a(filename))
 102                 plan_b(filename);
 103         if (verbose) {
 104                 say("Patching file %s using Plan %s...\n", filename,
 105                     (using_plan_a ? "A" : "B"));
 106         }
 107 }
 108
 109 static bool
 110 reallocate_lines(size_t *lines_allocated)
 111 {
 112         char    **p;
 113         size_t  new_size;
 114
 115         new_size = *lines_allocated * 3 / 2;
 116         p = realloc(i_ptr, (new_size + 2) * sizeof(char *));
 117         if (p == NULL) {        /* shucks, it was a near thing */
 118                 munmap(i_womp, i_size);
 119                 i_womp = NULL;
 120                 free(i_ptr);
 121                 i_ptr = NULL;
 122                 *lines_allocated = 0;
 123                 return false;
 124         }
 125         *lines_allocated = new_size;
 126         i_ptr = p;
 127         return true;
 128 }
 129
 130 /* Try keeping everything in memory. */
 131
 132 static bool
 133 plan_a(const char *filename)
 134 {
 135         int             ifd, statfailed;
 136         char            *p, *s;
 137         struct stat     filestat;
 138         ptrdiff_t       sz;
 139         size_t          i;
 140         size_t          iline, lines_allocated;
 141
 142 #ifdef DEBUGGING
 143         if (debug & 8)
 144                 return false;
 145 #endif
 146
 147         if (filename == NULL || *filename == '\0')
 148                 return false;
 149
 150         statfailed = stat(filename, &filestat);
 151         if (statfailed && ok_to_create_file) {
 152                 if (verbose)
 153                         say("(Creating file %s...)\n", filename);
 154
 155                 /*
 156                  * in check_patch case, we still display `Creating file' even
 157                  * though we're not. The rule is that -C should be as similar
 158                  * to normal patch behavior as possible
 159                  */
 160                 if (check_only)
 161                         return true;
 162                 makedirs(filename, true);
 163                 close(creat(filename, 0666));
 164                 statfailed = stat(filename, &filestat);
 165         }
 166         if (statfailed && check_only)
 167                 fatal("%s not found, -C mode, can't probe further\n", filename);
 168         /*
 169          * For nonexistent or read-only files we no longer attempt to locate
 170          * or checkout the file via RCS or SCCS, or do any call to system()
 171          * whatsoever.  CVE-2015-1416.
 172          */
 173         if (statfailed ||
 174             /* No one can write to it.  */
 175             (filestat.st_mode & 0222) == 0 ||
 176             /* I can't write to it.  */
 177             ((filestat.st_mode & 0022) == 0 && filestat.st_uid != getuid())) {
 178                 if (statfailed)
 179                         fatal("can't find %s\n", filename);
 180         }
 181         filemode = filestat.st_mode;
 182         if (!S_ISREG(filemode))
 183                 fatal("%s is not a normal file--can't patch\n", filename);
 184         if ((uint64_t)filestat.st_size > SIZE_MAX) {
 185                 say("block too large to mmap\n");
 186                 return false;
 187         }
 188         i_size = (size_t)filestat.st_size;
 189         if (out_of_mem) {
 190                 set_hunkmax();  /* make sure dynamic arrays are allocated */
 191                 out_of_mem = false;
 192                 return false;   /* force plan b because plan a bombed */
 193         }
 194         if ((ifd = open(filename, O_RDONLY)) < 0)
 195                 pfatal("can't open file %s", filename);
 196
 197         if (i_size) {
 198                 i_womp = mmap(NULL, i_size, PROT_READ, MAP_PRIVATE, ifd, 0);
 199                 if (i_womp == MAP_FAILED) {
 200                         perror("mmap failed");
 201                         i_womp = NULL;
 202                         close(ifd);
 203                         return false;
 204                 }
 205         } else {
 206                 i_womp = NULL;
 207         }
 208
 209         close(ifd);
 210         if (i_size)
 211                 madvise(i_womp, i_size, MADV_SEQUENTIAL);
 212
 213         /* estimate the number of lines */
 214         lines_allocated = i_size / 25;
 215         if (lines_allocated < 100)
 216                 lines_allocated = 100;
 217
 218         if (!reallocate_lines(&lines_allocated))
 219                 return false;
 220
 221         /* now scan the buffer and build pointer array */
 222         iline = 1;
 223         i_ptr[iline] = i_womp;
 224         /* test for NUL too, to maintain the behavior of the original code */
 225         for (s = i_womp, i = 0; i < i_size && *s != '\0'; s++, i++) {
 226                 if (*s == '\n') {
 227                         if (iline == lines_allocated) {
 228                                 if (!reallocate_lines(&lines_allocated))
 229                                         return false;
 230                         }
 231                         /* these are NOT NUL terminated */
 232                         i_ptr[++iline] = s + 1;
 233                 }
 234         }
 235         /* if the last line contains no EOL, append one */
 236         if (i_size > 0 && i_womp[i_size - 1] != '\n') {
 237                 last_line_missing_eol = true;
 238                 /* fix last line */
 239                 sz = s - i_ptr[iline];
 240                 p = malloc(sz + 1);
 241                 if (p == NULL) {
 242                         free(i_ptr);
 243                         i_ptr = NULL;
 244                         munmap(i_womp, i_size);
 245                         i_womp = NULL;
 246                         return false;
 247                 }
 248
 249                 memcpy(p, i_ptr[iline], sz);
 250                 p[sz] = '\n';
 251                 i_ptr[iline] = p;
 252                 /* count the extra line and make it point to some valid mem */
 253                 i_ptr[++iline] = empty_line;
 254         } else
 255                 last_line_missing_eol = false;
 256
 257         input_lines = iline - 1;
 258
 259         /* now check for revision, if any */
 260
 261         if (revision != NULL) {
 262                 if (!rev_in_string(i_womp)) {
 263                         if (force) {
 264                                 if (verbose)
 265                                         say("Warning: this file doesn't appear "
 266                                             "to be the %s version--patching anyway.\n",
 267                                             revision);
 268                         } else if (batch) {
 269                                 fatal("this file doesn't appear to be the "
 270                                     "%s version--aborting.\n",
 271                                     revision);
 272                         } else {
 273                                 ask("This file doesn't appear to be the "
 274                                     "%s version--patch anyway? [n] ",
 275                                     revision);
 276                                 if (*buf != 'y')
 277                                         fatal("aborted\n");
 278                         }
 279                 } else if (verbose)
 280                         say("Good.  This file appears to be the %s version.\n",
 281                             revision);
 282         }
 283         return true;            /* plan a will work */
 284 }
 285
 286 /* Keep (virtually) nothing in memory. */
 287
 288 static void
 289 plan_b(const char *filename)
 290 {
 291         FILE    *ifp;
 292         size_t  i = 0, j, maxlen = 1;
 293         char    *p;
 294         bool    found_revision = (revision == NULL);
 295
 296         using_plan_a = false;
 297         if ((ifp = fopen(filename, "r")) == NULL)
 298                 pfatal("can't open file %s", filename);
 299         unlink(TMPINNAME);
 300         if ((tifd = open(TMPINNAME, O_EXCL | O_CREAT | O_WRONLY, 0666)) < 0)
 301                 pfatal("can't open file %s", TMPINNAME);
 302         while (fgets(buf, buf_size, ifp) != NULL) {
 303                 if (revision != NULL && !found_revision && rev_in_string(buf))
 304                         found_revision = true;
 305                 if ((i = strlen(buf)) > maxlen)
 306                         maxlen = i;     /* find longest line */
 307         }
 308         last_line_missing_eol = i > 0 && buf[i - 1] != '\n';
 309         if (last_line_missing_eol && maxlen == i)
 310                 maxlen++;
 311
 312         if (revision != NULL) {
 313                 if (!found_revision) {
 314                         if (force) {
 315                                 if (verbose)
 316                                         say("Warning: this file doesn't appear "
 317                                             "to be the %s version--patching anyway.\n",
 318                                             revision);
 319                         } else if (batch) {
 320                                 fatal("this file doesn't appear to be the "
 321                                     "%s version--aborting.\n",
 322                                     revision);
 323                         } else {
 324                                 ask("This file doesn't appear to be the %s "
 325                                     "version--patch anyway? [n] ",
 326                                     revision);
 327                                 if (*buf != 'y')
 328                                         fatal("aborted\n");
 329                         }
 330                 } else if (verbose)
 331                         say("Good.  This file appears to be the %s version.\n",
 332                             revision);
 333         }
 334         fseek(ifp, 0L, SEEK_SET);       /* rewind file */
 335         lines_per_buf = BUFFERSIZE / maxlen;
 336         tireclen = maxlen;
 337         tibuf[0] = malloc(BUFFERSIZE + 1);
 338         if (tibuf[0] == NULL)
 339                 fatal("out of memory\n");
 340         tibuf[1] = malloc(BUFFERSIZE + 1);
 341         if (tibuf[1] == NULL)
 342                 fatal("out of memory\n");
 343         for (i = 1;; i++) {
 344                 p = tibuf[0] + maxlen * (i % lines_per_buf);
 345                 if (i % lines_per_buf == 0)     /* new block */
 346                         if (write(tifd, tibuf[0], BUFFERSIZE) < BUFFERSIZE)
 347                                 pfatal("can't write temp file");
 348                 if (fgets(p, maxlen + 1, ifp) == NULL) {
 349                         input_lines = i - 1;
 350                         if (i % lines_per_buf != 0)
 351                                 if (write(tifd, tibuf[0], BUFFERSIZE) < BUFFERSIZE)
 352                                         pfatal("can't write temp file");
 353                         break;
 354                 }
 355                 j = strlen(p);
 356                 /* These are '\n' terminated strings, so no need to add a NUL */
 357                 if (j == 0 || p[j - 1] != '\n')
 358                         p[j] = '\n';
 359         }
 360         fclose(ifp);
 361         close(tifd);
 362         if ((tifd = open(TMPINNAME, O_RDONLY)) < 0)
 363                 pfatal("can't reopen file %s", TMPINNAME);
 364 }
 365
 366 /*
 367  * Fetch a line from the input file, \n terminated, not necessarily \0.
 368  */
 369 char *
 370 ifetch(LINENUM line, int whichbuf)
 371 {
 372         if (line < 1 || line > input_lines) {
 373                 if (warn_on_invalid_line) {
 374                         say("No such line %ld in input file, ignoring\n", line);
 375                         warn_on_invalid_line = false;
 376                 }
 377                 return NULL;
 378         }
 379         if (using_plan_a)
 380                 return i_ptr[line];
 381         else {
 382                 LINENUM offline = line % lines_per_buf;
 383                 LINENUM baseline = line - offline;
 384
 385                 if (tiline[0] == baseline)
 386                         whichbuf = 0;
 387                 else if (tiline[1] == baseline)
 388                         whichbuf = 1;
 389                 else {
 390                         tiline[whichbuf] = baseline;
 391
 392                         if (lseek(tifd, (off_t) (baseline / lines_per_buf *
 393                             BUFFERSIZE), SEEK_SET) < 0)
 394                                 pfatal("cannot seek in the temporary input file");
 395
 396                         if (read(tifd, tibuf[whichbuf], BUFFERSIZE) < 0)
 397                                 pfatal("error reading tmp file %s", TMPINNAME);
 398                 }
 399                 return tibuf[whichbuf] + (tireclen * offline);
 400         }
 401 }
 402
 403 /*
 404  * True if the string argument contains the revision number we want.
 405  */
 406 static bool
 407 rev_in_string(const char *string)
 408 {
 409         const char      *s;
 410         size_t          patlen;
 411
 412         if (revision == NULL)
 413                 return true;
 414         patlen = strlen(revision);
 415         if (strnEQ(string, revision, patlen) && isspace((unsigned char)string[patlen]))
 416                 return true;
 417         for (s = string; *s; s++) {
 418                 if (isspace((unsigned char)*s) && strnEQ(s + 1, revision, patlen) &&
 419                     isspace((unsigned char)s[patlen + 1])) {
 420                         return true;
 421                 }
 422         }
 423         return false;
 424 }