1 /* $OpenBSD: diffreg.c,v 1.93 2019/06/28 13:35:00 deraadt Exp $ */ 2 3 /*- 4 * SPDX-License-Identifier: BSD-4-Clause 5 * 6 * Copyright (C) Caldera International Inc. 2001-2002. 7 * All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code and documentation must retain the above 13 * copyright notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. All advertising materials mentioning features or use of this software 18 * must display the following acknowledgement: 19 * This product includes software developed or owned by Caldera 20 * International, Inc. 21 * 4. Neither the name of Caldera International, Inc. nor the names of other 22 * contributors may be used to endorse or promote products derived from 23 * this software without specific prior written permission. 24 * 25 * USE OF THE SOFTWARE PROVIDED FOR UNDER THIS LICENSE BY CALDERA 26 * INTERNATIONAL, INC. AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR 27 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 28 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 29 * IN NO EVENT SHALL CALDERA INTERNATIONAL, INC. BE LIABLE FOR ANY DIRECT, 30 * INDIRECT INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 31 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 32 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, 34 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING 35 * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 36 * POSSIBILITY OF SUCH DAMAGE. 37 */ 38 /*- 39 * Copyright (c) 1991, 1993 40 * The Regents of the University of California. All rights reserved. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 3. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 */ 66 67 #include <sys/capsicum.h> 68 #include <sys/stat.h> 69 70 #include <capsicum_helpers.h> 71 #include <ctype.h> 72 #include <err.h> 73 #include <errno.h> 74 #include <fcntl.h> 75 #include <limits.h> 76 #include <math.h> 77 #include <paths.h> 78 #include <regex.h> 79 #include <stdbool.h> 80 #include <stdckdint.h> 81 #include <stddef.h> 82 #include <stdint.h> 83 #include <stdio.h> 84 #include <stdlib.h> 85 #include <string.h> 86 87 #include "pr.h" 88 #include "diff.h" 89 #include "xmalloc.h" 90 91 /* 92 * diff - compare two files. 93 */ 94 95 /* 96 * Uses an algorithm due to Harold Stone, which finds a pair of longest 97 * identical subsequences in the two files. 98 * 99 * The major goal is to generate the match vector J. J[i] is the index of 100 * the line in file1 corresponding to line i file0. J[i] = 0 if there is no 101 * such line in file1. 102 * 103 * Lines are hashed so as to work in core. All potential matches are 104 * located by sorting the lines of each file on the hash (called 105 * ``value''). In particular, this collects the equivalence classes in 106 * file1 together. Subroutine equiv replaces the value of each line in 107 * file0 by the index of the first element of its matching equivalence in 108 * (the reordered) file1. To save space equiv squeezes file1 into a single 109 * array member in which the equivalence classes are simply concatenated, 110 * except that their first members are flagged by changing sign. 111 * 112 * Next the indices that point into member are unsorted into array class 113 * according to the original order of file0. 114 * 115 * The cleverness lies in routine stone. This marches through the lines of 116 * file0, developing a vector klist of "k-candidates". At step i 117 * a k-candidate is a matched pair of lines x,y (x in file0 y in file1) 118 * such that there is a common subsequence of length k between the first 119 * i lines of file0 and the first y lines of file1, but there is no such 120 * subsequence for any smaller y. x is the earliest possible mate to y that 121 * occurs in such a subsequence. 122 * 123 * Whenever any of the members of the equivalence class of lines in file1 124 * matable to a line in file0 has serial number less than the y of some 125 * k-candidate, that k-candidate with the smallest such y is replaced. The 126 * new k-candidate is chained (via pred) to the current k-1 candidate so 127 * that the actual subsequence can be recovered. When a member has serial 128 * number greater that the y of all k-candidates, the klist is extended. At 129 * the end, the longest subsequence is pulled out and placed in the array J 130 * by unravel. 131 * 132 * With J in hand, the matches there recorded are check'ed against reality 133 * to assure that no spurious matches have crept in due to hashing. If they 134 * have, they are broken, and "jackpot" is recorded -- a harmless matter 135 * except that a true match for a spuriously mated line may now be 136 * unnecessarily reported as a change. 137 * 138 * Much of the complexity of the program comes simply from trying to 139 * minimize core utilization and maximize the range of doable problems by 140 * dynamically allocating what is needed and reusing what is not. The core 141 * requirements for problems larger than somewhat are (in words) 142 * 2*length(file0) + length(file1) + 3*(number of k-candidates installed), 143 * typically about 6n words for files of length n. 144 */ 145 146 struct cand { 147 int x; 148 int y; 149 int pred; 150 }; 151 152 static struct line { 153 int serial; 154 int value; 155 } *file[2]; 156 157 /* 158 * The following struct is used to record change information when 159 * doing a "context" or "unified" diff. (see routine "change" to 160 * understand the highly mnemonic field names) 161 */ 162 struct context_vec { 163 int a; /* start line in old file */ 164 int b; /* end line in old file */ 165 int c; /* start line in new file */ 166 int d; /* end line in new file */ 167 }; 168 169 enum readhash { RH_BINARY, RH_OK, RH_EOF }; 170 171 static int diffreg_stone(char *, char *, int, int); 172 static FILE *opentemp(const char *); 173 static void output(char *, FILE *, char *, FILE *, int); 174 static void check(FILE *, FILE *, int); 175 static void range(int, int, const char *); 176 static void uni_range(int, int); 177 static void dump_context_vec(FILE *, FILE *, int); 178 static void dump_unified_vec(FILE *, FILE *, int); 179 static bool prepare(int, FILE *, size_t, int); 180 static void prune(void); 181 static void equiv(struct line *, int, struct line *, int, int *); 182 static void unravel(int); 183 static void unsort(struct line *, int, int *); 184 static void change(char *, FILE *, char *, FILE *, int, int, int, int, int *); 185 static void sort(struct line *, int); 186 static void print_header(const char *, const char *); 187 static void print_space(int, int, int); 188 static bool ignoreline_pattern(char *); 189 static bool ignoreline(char *, bool); 190 static int asciifile(FILE *); 191 static int fetch(long *, int, int, FILE *, int, int, int); 192 static int newcand(int, int, int); 193 static int search(int *, int, int); 194 static int skipline(FILE *); 195 static int stone(int *, int, int *, int *, int); 196 static enum readhash readhash(FILE *, int, unsigned *); 197 static int files_differ(FILE *, FILE *, int); 198 static char *match_function(const long *, int, FILE *); 199 static char *preadline(int, size_t, off_t); 200 201 static int *J; /* will be overlaid on class */ 202 static int *class; /* will be overlaid on file[0] */ 203 static int *klist; /* will be overlaid on file[0] after class */ 204 static int *member; /* will be overlaid on file[1] */ 205 static int clen; 206 static int inifdef; /* whether or not we are in a #ifdef block */ 207 static size_t len[2]; /* lengths of files in lines */ 208 static size_t pref, suff; /* lengths of prefix and suffix */ 209 static size_t slen[2]; /* lengths of files minus pref / suff */ 210 static int anychange; 211 static int hw, lpad,rpad; /* half width and padding */ 212 static int edoffset; 213 static long *ixnew; /* will be overlaid on file[1] */ 214 static long *ixold; /* will be overlaid on klist */ 215 static struct cand *clist; /* merely a free storage pot for candidates */ 216 static int clistlen; /* the length of clist */ 217 static struct line *sfile[2]; /* shortened by pruning common prefix/suffix */ 218 static int (*chrtran)(int); /* translation table for case-folding */ 219 static struct context_vec *context_vec_start; 220 static struct context_vec *context_vec_end; 221 static struct context_vec *context_vec_ptr; 222 223 #define FUNCTION_CONTEXT_SIZE 55 224 static char lastbuf[FUNCTION_CONTEXT_SIZE]; 225 static int lastline; 226 static int lastmatchline; 227 228 int 229 diffreg(char *file1, char *file2, int flags, int capsicum) 230 { 231 /* 232 * If we have set the algorithm with -A or --algorithm use that if we 233 * can and if not print an error. 234 */ 235 if (diff_algorithm_set) { 236 if (diff_algorithm == D_DIFFMYERS || 237 diff_algorithm == D_DIFFPATIENCE) { 238 if (can_libdiff(flags)) 239 return diffreg_new(file1, file2, flags, capsicum); 240 else 241 errx(2, "cannot use Myers algorithm with selected options"); 242 } else { 243 /* Fallback to using stone. */ 244 return diffreg_stone(file1, file2, flags, capsicum); 245 } 246 } else { 247 if (can_libdiff(flags)) 248 return diffreg_new(file1, file2, flags, capsicum); 249 else 250 return diffreg_stone(file1, file2, flags, capsicum); 251 } 252 } 253 254 static int 255 clow2low(int c) 256 { 257 258 return (c); 259 } 260 261 static int 262 cup2low(int c) 263 { 264 265 return (tolower(c)); 266 } 267 268 int 269 diffreg_stone(char *file1, char *file2, int flags, int capsicum) 270 { 271 FILE *f1, *f2; 272 int i, rval; 273 struct pr *pr = NULL; 274 cap_rights_t rights_ro; 275 276 f1 = f2 = NULL; 277 rval = D_SAME; 278 anychange = 0; 279 lastline = 0; 280 lastmatchline = 0; 281 282 /* 283 * In side-by-side mode, we need to print the left column, a 284 * change marker surrounded by padding, and the right column. 285 * 286 * If expanding tabs, we don't care about alignment, so we simply 287 * subtract 3 from the width and divide by two. 288 * 289 * If not expanding tabs, we need to ensure that the right column 290 * is aligned to a tab stop. We start with the same formula, then 291 * decrement until we reach a size that lets us tab-align the 292 * right column. We then adjust the width down if necessary for 293 * the padding calculation to work. 294 * 295 * Left padding is half the space left over, rounded down; right 296 * padding is whatever is needed to match the width. 297 */ 298 if (diff_format == D_SIDEBYSIDE) { 299 if (flags & D_EXPANDTABS) { 300 if (width > 3) { 301 hw = (width - 3) / 2; 302 } else { 303 /* not enough space */ 304 hw = 0; 305 } 306 } else if (width <= 3 || width <= tabsize) { 307 /* not enough space */ 308 hw = 0; 309 } else { 310 hw = (width - 3) / 2; 311 while (hw > 0 && roundup(hw + 3, tabsize) + hw > width) 312 hw--; 313 if (width - (roundup(hw + 3, tabsize) + hw) < tabsize) 314 width = roundup(hw + 3, tabsize) + hw; 315 } 316 lpad = (width - hw * 2 - 1) / 2; 317 rpad = (width - hw * 2 - 1) - lpad; 318 } 319 320 if (flags & D_IGNORECASE) 321 chrtran = cup2low; 322 else 323 chrtran = clow2low; 324 if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode)) 325 return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2); 326 if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0) 327 goto closem; 328 329 if (flags & D_EMPTY1) 330 f1 = fopen(_PATH_DEVNULL, "r"); 331 else { 332 if (!S_ISREG(stb1.st_mode)) { 333 if ((f1 = opentemp(file1)) == NULL || 334 fstat(fileno(f1), &stb1) == -1) { 335 warn("%s", file1); 336 rval = D_ERROR; 337 status |= 2; 338 goto closem; 339 } 340 } else if (strcmp(file1, "-") == 0) 341 f1 = stdin; 342 else 343 f1 = fopen(file1, "r"); 344 } 345 if (f1 == NULL) { 346 warn("%s", file1); 347 rval = D_ERROR; 348 status |= 2; 349 goto closem; 350 } 351 352 if (flags & D_EMPTY2) 353 f2 = fopen(_PATH_DEVNULL, "r"); 354 else { 355 if (!S_ISREG(stb2.st_mode)) { 356 if ((f2 = opentemp(file2)) == NULL || 357 fstat(fileno(f2), &stb2) == -1) { 358 warn("%s", file2); 359 rval = D_ERROR; 360 status |= 2; 361 goto closem; 362 } 363 } else if (strcmp(file2, "-") == 0) 364 f2 = stdin; 365 else 366 f2 = fopen(file2, "r"); 367 } 368 if (f2 == NULL) { 369 warn("%s", file2); 370 rval = D_ERROR; 371 status |= 2; 372 goto closem; 373 } 374 375 if (stb1.st_dev == stb2.st_dev && stb1.st_ino == stb2.st_ino) 376 goto closem; 377 378 if (lflag) 379 pr = start_pr(file1, file2); 380 381 if (capsicum) { 382 cap_rights_init(&rights_ro, CAP_READ, CAP_FSTAT, CAP_SEEK); 383 if (caph_rights_limit(fileno(f1), &rights_ro) < 0) 384 err(2, "unable to limit rights on: %s", file1); 385 if (caph_rights_limit(fileno(f2), &rights_ro) < 0) 386 err(2, "unable to limit rights on: %s", file2); 387 if (fileno(f1) == STDIN_FILENO || fileno(f2) == STDIN_FILENO) { 388 /* stdin has already been limited */ 389 if (caph_limit_stderr() == -1) 390 err(2, "unable to limit stderr"); 391 if (caph_limit_stdout() == -1) 392 err(2, "unable to limit stdout"); 393 } else if (caph_limit_stdio() == -1) 394 err(2, "unable to limit stdio"); 395 396 caph_cache_catpages(); 397 caph_cache_tzdata(); 398 if (caph_enter() < 0) 399 err(2, "unable to enter capability mode"); 400 } 401 402 switch (files_differ(f1, f2, flags)) { 403 case 0: 404 goto closem; 405 case 1: 406 break; 407 default: 408 /* error */ 409 if (ferror(f1)) 410 warn("%s", file1); 411 if (ferror(f2)) 412 warn("%s", file2); 413 rval = D_ERROR; 414 status |= 2; 415 goto closem; 416 } 417 418 if (diff_format == D_BRIEF && ignore_pats == NULL && 419 (flags & (D_FOLDBLANKS|D_IGNOREBLANKS|D_IGNORECASE| 420 D_SKIPBLANKLINES|D_STRIPCR)) == 0) 421 { 422 rval = D_DIFFER; 423 status |= 1; 424 goto closem; 425 } 426 if ((flags & D_FORCEASCII) != 0) { 427 (void)prepare(0, f1, stb1.st_size, flags); 428 (void)prepare(1, f2, stb2.st_size, flags); 429 } else if (!asciifile(f1) || !asciifile(f2) || 430 !prepare(0, f1, stb1.st_size, flags) || 431 !prepare(1, f2, stb2.st_size, flags)) { 432 rval = D_BINARY; 433 status |= 1; 434 goto closem; 435 } 436 if (len[0] > INT_MAX - 2) 437 errc(1, EFBIG, "%s", file1); 438 if (len[1] > INT_MAX - 2) 439 errc(1, EFBIG, "%s", file2); 440 441 prune(); 442 sort(sfile[0], slen[0]); 443 sort(sfile[1], slen[1]); 444 445 member = (int *)file[1]; 446 equiv(sfile[0], slen[0], sfile[1], slen[1], member); 447 member = xreallocarray(member, slen[1] + 2, sizeof(*member)); 448 449 class = (int *)file[0]; 450 unsort(sfile[0], slen[0], class); 451 class = xreallocarray(class, slen[0] + 2, sizeof(*class)); 452 453 klist = xcalloc(slen[0] + 2, sizeof(*klist)); 454 clen = 0; 455 clistlen = 100; 456 clist = xcalloc(clistlen, sizeof(*clist)); 457 i = stone(class, slen[0], member, klist, flags); 458 free(member); 459 free(class); 460 461 J = xreallocarray(J, len[0] + 2, sizeof(*J)); 462 unravel(klist[i]); 463 free(clist); 464 free(klist); 465 466 ixold = xreallocarray(ixold, len[0] + 2, sizeof(*ixold)); 467 ixnew = xreallocarray(ixnew, len[1] + 2, sizeof(*ixnew)); 468 check(f1, f2, flags); 469 output(file1, f1, file2, f2, flags); 470 471 closem: 472 if (pr != NULL) 473 stop_pr(pr); 474 if (anychange) { 475 status |= 1; 476 if (rval == D_SAME) 477 rval = D_DIFFER; 478 } 479 if (f1 != NULL) 480 fclose(f1); 481 if (f2 != NULL) 482 fclose(f2); 483 484 return (rval); 485 } 486 487 /* 488 * Check to see if the given files differ. 489 * Returns 0 if they are the same, 1 if different, and -1 on error. 490 * XXX - could use code from cmp(1) [faster] 491 */ 492 static int 493 files_differ(FILE *f1, FILE *f2, int flags) 494 { 495 char buf1[BUFSIZ], buf2[BUFSIZ]; 496 size_t i, j; 497 498 if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size || 499 (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT)) 500 return (1); 501 502 if (stb1.st_dev == stb2.st_dev && stb1.st_ino == stb2.st_ino) 503 return (0); 504 505 for (;;) { 506 if ((i = fread(buf1, 1, sizeof(buf1), f1)) == 0 && ferror(f1)) 507 return (-1); 508 if ((j = fread(buf2, 1, sizeof(buf2), f2)) == 0 && ferror(f2)) 509 return (-1); 510 if (i != j) 511 return (1); 512 if (i == 0) 513 return (0); 514 if (memcmp(buf1, buf2, i) != 0) 515 return (1); 516 } 517 } 518 519 static FILE * 520 opentemp(const char *f) 521 { 522 char buf[BUFSIZ], tempfile[PATH_MAX]; 523 ssize_t nread; 524 int ifd, ofd; 525 526 if (strcmp(f, "-") == 0) 527 ifd = STDIN_FILENO; 528 else if ((ifd = open(f, O_RDONLY, 0644)) == -1) 529 return (NULL); 530 531 (void)strlcpy(tempfile, _PATH_TMP "/diff.XXXXXXXX", sizeof(tempfile)); 532 533 if ((ofd = mkstemp(tempfile)) == -1) { 534 close(ifd); 535 return (NULL); 536 } 537 unlink(tempfile); 538 while ((nread = read(ifd, buf, BUFSIZ)) > 0) { 539 if (write(ofd, buf, nread) != nread) { 540 close(ifd); 541 close(ofd); 542 return (NULL); 543 } 544 } 545 close(ifd); 546 lseek(ofd, (off_t)0, SEEK_SET); 547 return (fdopen(ofd, "r")); 548 } 549 550 static bool 551 prepare(int i, FILE *fd, size_t filesize, int flags) 552 { 553 struct line *p; 554 unsigned h; 555 size_t sz, j = 0; 556 enum readhash r; 557 558 rewind(fd); 559 560 sz = MIN(filesize, SIZE_MAX) / 25; 561 if (sz < 100) 562 sz = 100; 563 564 p = xcalloc(sz + 3, sizeof(*p)); 565 while ((r = readhash(fd, flags, &h)) != RH_EOF) { 566 if (r == RH_BINARY) 567 return (false); 568 if (j == SIZE_MAX) 569 break; 570 if (j == sz) { 571 sz = sz * 3 / 2; 572 p = xreallocarray(p, sz + 3, sizeof(*p)); 573 } 574 p[++j].value = h; 575 } 576 577 len[i] = j; 578 file[i] = p; 579 580 return (true); 581 } 582 583 static void 584 prune(void) 585 { 586 size_t i, j; 587 588 for (pref = 0; pref < len[0] && pref < len[1] && 589 file[0][pref + 1].value == file[1][pref + 1].value; 590 pref++) 591 ; 592 for (suff = 0; suff < len[0] - pref && suff < len[1] - pref && 593 file[0][len[0] - suff].value == file[1][len[1] - suff].value; 594 suff++) 595 ; 596 for (j = 0; j < 2; j++) { 597 sfile[j] = file[j] + pref; 598 slen[j] = len[j] - pref - suff; 599 for (i = 0; i <= slen[j]; i++) 600 sfile[j][i].serial = i; 601 } 602 } 603 604 static void 605 equiv(struct line *a, int n, struct line *b, int m, int *c) 606 { 607 int i, j; 608 609 i = j = 1; 610 while (i <= n && j <= m) { 611 if (a[i].value < b[j].value) 612 a[i++].value = 0; 613 else if (a[i].value == b[j].value) 614 a[i++].value = j; 615 else 616 j++; 617 } 618 while (i <= n) 619 a[i++].value = 0; 620 b[m + 1].value = 0; 621 j = 0; 622 while (++j <= m) { 623 c[j] = -b[j].serial; 624 while (b[j + 1].value == b[j].value) { 625 j++; 626 c[j] = b[j].serial; 627 } 628 } 629 c[j] = -1; 630 } 631 632 static int 633 stone(int *a, int n, int *b, int *c, int flags) 634 { 635 int i, k, y, j, l; 636 int oldc, tc, oldl, sq; 637 unsigned numtries, bound; 638 639 if (flags & D_MINIMAL) 640 bound = UINT_MAX; 641 else { 642 sq = sqrt(n); 643 bound = MAX(256, sq); 644 } 645 646 k = 0; 647 c[0] = newcand(0, 0, 0); 648 for (i = 1; i <= n; i++) { 649 j = a[i]; 650 if (j == 0) 651 continue; 652 y = -b[j]; 653 oldl = 0; 654 oldc = c[0]; 655 numtries = 0; 656 do { 657 if (y <= clist[oldc].y) 658 continue; 659 l = search(c, k, y); 660 if (l != oldl + 1) 661 oldc = c[l - 1]; 662 if (l <= k) { 663 if (clist[c[l]].y <= y) 664 continue; 665 tc = c[l]; 666 c[l] = newcand(i, y, oldc); 667 oldc = tc; 668 oldl = l; 669 numtries++; 670 } else { 671 c[l] = newcand(i, y, oldc); 672 k++; 673 break; 674 } 675 } while ((y = b[++j]) > 0 && numtries < bound); 676 } 677 return (k); 678 } 679 680 static int 681 newcand(int x, int y, int pred) 682 { 683 struct cand *q; 684 685 if (clen == clistlen) { 686 clistlen = clistlen * 11 / 10; 687 clist = xreallocarray(clist, clistlen, sizeof(*clist)); 688 } 689 q = clist + clen; 690 q->x = x; 691 q->y = y; 692 q->pred = pred; 693 return (clen++); 694 } 695 696 static int 697 search(int *c, int k, int y) 698 { 699 int i, j, l, t; 700 701 if (clist[c[k]].y < y) /* quick look for typical case */ 702 return (k + 1); 703 i = 0; 704 j = k + 1; 705 for (;;) { 706 l = (i + j) / 2; 707 if (l <= i) 708 break; 709 t = clist[c[l]].y; 710 if (t > y) 711 j = l; 712 else if (t < y) 713 i = l; 714 else 715 return (l); 716 } 717 return (l + 1); 718 } 719 720 static void 721 unravel(int p) 722 { 723 struct cand *q; 724 size_t i; 725 726 for (i = 0; i <= len[0]; i++) 727 J[i] = i <= pref ? i : 728 i > len[0] - suff ? i + len[1] - len[0] : 0; 729 for (q = clist + p; q->y != 0; q = clist + q->pred) 730 J[q->x + pref] = q->y + pref; 731 } 732 733 /* 734 * Check does double duty: 735 * 1. ferret out any fortuitous correspondences due to confounding by 736 * hashing (which result in "jackpot") 737 * 2. collect random access indexes to the two files 738 */ 739 static void 740 check(FILE *f1, FILE *f2, int flags) 741 { 742 int i, j, /* jackpot, */ c, d; 743 long ctold, ctnew; 744 745 rewind(f1); 746 rewind(f2); 747 j = 1; 748 ixold[0] = ixnew[0] = 0; 749 /* jackpot = 0; */ 750 ctold = ctnew = 0; 751 for (i = 1; i <= (int)len[0]; i++) { 752 if (J[i] == 0) { 753 ixold[i] = ctold += skipline(f1); 754 continue; 755 } 756 while (j < J[i]) { 757 ixnew[j] = ctnew += skipline(f2); 758 j++; 759 } 760 if (flags & (D_FOLDBLANKS | D_IGNOREBLANKS | D_IGNORECASE | D_STRIPCR)) { 761 for (;;) { 762 c = getc(f1); 763 d = getc(f2); 764 /* 765 * GNU diff ignores a missing newline 766 * in one file for -b or -w. 767 */ 768 if (flags & (D_FOLDBLANKS | D_IGNOREBLANKS)) { 769 if (c == EOF && isspace(d)) { 770 ctnew++; 771 break; 772 } else if (isspace(c) && d == EOF) { 773 ctold++; 774 break; 775 } 776 } 777 ctold++; 778 ctnew++; 779 if (flags & D_STRIPCR && (c == '\r' || d == '\r')) { 780 if (c == '\r') { 781 if ((c = getc(f1)) == '\n') { 782 ctold++; 783 } else { 784 ungetc(c, f1); 785 } 786 } 787 if (d == '\r') { 788 if ((d = getc(f2)) == '\n') { 789 ctnew++; 790 } else { 791 ungetc(d, f2); 792 } 793 } 794 break; 795 } 796 if ((flags & D_FOLDBLANKS) && isspace(c) && 797 isspace(d)) { 798 do { 799 if (c == '\n') 800 break; 801 ctold++; 802 } while (isspace(c = getc(f1))); 803 do { 804 if (d == '\n') 805 break; 806 ctnew++; 807 } while (isspace(d = getc(f2))); 808 } else if (flags & D_IGNOREBLANKS) { 809 while (isspace(c) && c != '\n') { 810 c = getc(f1); 811 ctold++; 812 } 813 while (isspace(d) && d != '\n') { 814 d = getc(f2); 815 ctnew++; 816 } 817 } 818 if (chrtran(c) != chrtran(d)) { 819 /* jackpot++; */ 820 J[i] = 0; 821 if (c != '\n' && c != EOF) 822 ctold += skipline(f1); 823 if (d != '\n' && c != EOF) 824 ctnew += skipline(f2); 825 break; 826 } 827 if (c == '\n' || c == EOF) 828 break; 829 } 830 } else { 831 for (;;) { 832 ctold++; 833 ctnew++; 834 if ((c = getc(f1)) != (d = getc(f2))) { 835 /* jackpot++; */ 836 J[i] = 0; 837 if (c != '\n' && c != EOF) 838 ctold += skipline(f1); 839 if (d != '\n' && c != EOF) 840 ctnew += skipline(f2); 841 break; 842 } 843 if (c == '\n' || c == EOF) 844 break; 845 } 846 } 847 ixold[i] = ctold; 848 ixnew[j] = ctnew; 849 j++; 850 } 851 for (; j <= (int)len[1]; j++) { 852 ixnew[j] = ctnew += skipline(f2); 853 } 854 /* 855 * if (jackpot) 856 * fprintf(stderr, "jackpot\n"); 857 */ 858 } 859 860 /* shellsort CACM #201 */ 861 static void 862 sort(struct line *a, int n) 863 { 864 struct line *ai, *aim, w; 865 int j, m = 0, k; 866 867 if (n == 0) 868 return; 869 for (j = 1; j <= n; j *= 2) 870 m = 2 * j - 1; 871 for (m /= 2; m != 0; m /= 2) { 872 k = n - m; 873 for (j = 1; j <= k; j++) { 874 for (ai = &a[j]; ai > a; ai -= m) { 875 aim = &ai[m]; 876 if (aim < ai) 877 break; /* wraparound */ 878 if (aim->value > ai[0].value || 879 (aim->value == ai[0].value && 880 aim->serial > ai[0].serial)) 881 break; 882 w.value = ai[0].value; 883 ai[0].value = aim->value; 884 aim->value = w.value; 885 w.serial = ai[0].serial; 886 ai[0].serial = aim->serial; 887 aim->serial = w.serial; 888 } 889 } 890 } 891 } 892 893 static void 894 unsort(struct line *f, int l, int *b) 895 { 896 int *a, i; 897 898 a = xcalloc(l + 1, sizeof(*a)); 899 for (i = 1; i <= l; i++) 900 a[f[i].serial] = f[i].value; 901 for (i = 1; i <= l; i++) 902 b[i] = a[i]; 903 free(a); 904 } 905 906 static int 907 skipline(FILE *f) 908 { 909 int i, c; 910 911 for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++) 912 continue; 913 return (i); 914 } 915 916 static void 917 output(char *file1, FILE *f1, char *file2, FILE *f2, int flags) 918 { 919 int i, j, m, i0, i1, j0, j1, nc; 920 921 rewind(f1); 922 rewind(f2); 923 m = len[0]; 924 J[0] = 0; 925 J[m + 1] = len[1] + 1; 926 if (diff_format != D_EDIT) { 927 for (i0 = 1; i0 <= m; i0 = i1 + 1) { 928 while (i0 <= m && J[i0] == J[i0 - 1] + 1) { 929 if (diff_format == D_SIDEBYSIDE && suppress_common != 1) { 930 nc = fetch(ixold, i0, i0, f1, '\0', 1, flags); 931 print_space(nc, hw - nc + lpad + 1 + rpad, flags); 932 fetch(ixnew, J[i0], J[i0], f2, '\0', 0, flags); 933 printf("\n"); 934 } 935 i0++; 936 } 937 j0 = J[i0 - 1] + 1; 938 i1 = i0 - 1; 939 while (i1 < m && J[i1 + 1] == 0) 940 i1++; 941 j1 = J[i1 + 1] - 1; 942 J[i1] = j1; 943 944 /* 945 * When using side-by-side, lines from both of the files are 946 * printed. The algorithm used by diff(1) identifies the ranges 947 * in which two files differ. 948 * See the change() function below. 949 * The for loop below consumes the shorter range, whereas one of 950 * the while loops deals with the longer one. 951 */ 952 if (diff_format == D_SIDEBYSIDE) { 953 for (i = i0, j = j0; i <= i1 && j <= j1; i++, j++) 954 change(file1, f1, file2, f2, i, i, j, j, &flags); 955 956 while (i <= i1) { 957 change(file1, f1, file2, f2, i, i, j + 1, j, &flags); 958 i++; 959 } 960 961 while (j <= j1) { 962 change(file1, f1, file2, f2, i + 1, i, j, j, &flags); 963 j++; 964 } 965 } else 966 change(file1, f1, file2, f2, i0, i1, j0, j1, &flags); 967 } 968 } else { 969 for (i0 = m; i0 >= 1; i0 = i1 - 1) { 970 while (i0 >= 1 && J[i0] == J[i0 + 1] - 1 && J[i0] != 0) 971 i0--; 972 j0 = J[i0 + 1] - 1; 973 i1 = i0 + 1; 974 while (i1 > 1 && J[i1 - 1] == 0) 975 i1--; 976 j1 = J[i1 - 1] + 1; 977 J[i1] = j1; 978 change(file1, f1, file2, f2, i1, i0, j1, j0, &flags); 979 } 980 } 981 if (m == 0) 982 change(file1, f1, file2, f2, 1, 0, 1, len[1], &flags); 983 if (diff_format == D_IFDEF || diff_format == D_GFORMAT) { 984 for (;;) { 985 #define c i0 986 if ((c = getc(f1)) == EOF) 987 return; 988 printf("%c", c); 989 } 990 #undef c 991 } 992 if (anychange != 0) { 993 if (diff_format == D_CONTEXT) 994 dump_context_vec(f1, f2, flags); 995 else if (diff_format == D_UNIFIED) 996 dump_unified_vec(f1, f2, flags); 997 } 998 } 999 1000 static void 1001 range(int a, int b, const char *separator) 1002 { 1003 printf("%d", a > b ? b : a); 1004 if (a < b) 1005 printf("%s%d", separator, b); 1006 } 1007 1008 static void 1009 uni_range(int a, int b) 1010 { 1011 if (a < b) 1012 printf("%d,%d", a, b - a + 1); 1013 else if (a == b) 1014 printf("%d", b); 1015 else 1016 printf("%d,0", b); 1017 } 1018 1019 static char * 1020 preadline(int fd, size_t rlen, off_t off) 1021 { 1022 char *line; 1023 ssize_t nr; 1024 1025 line = xmalloc(rlen + 1); 1026 if ((nr = pread(fd, line, rlen, off)) == -1) 1027 err(2, "preadline"); 1028 if (nr > 0 && line[nr-1] == '\n') 1029 nr--; 1030 line[nr] = '\0'; 1031 return (line); 1032 } 1033 1034 static bool 1035 ignoreline_pattern(char *line) 1036 { 1037 int ret; 1038 1039 ret = regexec(&ignore_re, line, 0, NULL, 0); 1040 return (ret == 0); /* if it matched, it should be ignored. */ 1041 } 1042 1043 static bool 1044 ignoreline(char *line, bool skip_blanks) 1045 { 1046 1047 if (skip_blanks && *line == '\0') 1048 return (true); 1049 if (ignore_pats != NULL && ignoreline_pattern(line)) 1050 return (true); 1051 return (false); 1052 } 1053 1054 /* 1055 * Indicate that there is a difference between lines a and b of the from file 1056 * to get to lines c to d of the to file. If a is greater then b then there 1057 * are no lines in the from file involved and this means that there were 1058 * lines appended (beginning at b). If c is greater than d then there are 1059 * lines missing from the to file. 1060 */ 1061 static void 1062 change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d, 1063 int *pflags) 1064 { 1065 static size_t max_context = 64; 1066 long curpos; 1067 int dist, i, nc; 1068 const char *walk; 1069 bool skip_blanks, ignore; 1070 1071 skip_blanks = (*pflags & D_SKIPBLANKLINES); 1072 restart: 1073 if ((diff_format != D_IFDEF || diff_format == D_GFORMAT) && 1074 a > b && c > d) 1075 return; 1076 if (ignore_pats != NULL || skip_blanks) { 1077 char *line; 1078 /* 1079 * All lines in the change, insert, or delete must match an ignore 1080 * pattern for the change to be ignored. 1081 */ 1082 if (a <= b) { /* Changes and deletes. */ 1083 for (i = a; i <= b; i++) { 1084 line = preadline(fileno(f1), 1085 ixold[i] - ixold[i - 1], ixold[i - 1]); 1086 ignore = ignoreline(line, skip_blanks); 1087 free(line); 1088 if (!ignore) 1089 goto proceed; 1090 } 1091 } 1092 if (a > b || c <= d) { /* Changes and inserts. */ 1093 for (i = c; i <= d; i++) { 1094 line = preadline(fileno(f2), 1095 ixnew[i] - ixnew[i - 1], ixnew[i - 1]); 1096 ignore = ignoreline(line, skip_blanks); 1097 free(line); 1098 if (!ignore) 1099 goto proceed; 1100 } 1101 } 1102 return; 1103 } 1104 proceed: 1105 if (*pflags & D_HEADER && diff_format != D_BRIEF) { 1106 printf("%s %s %s\n", diffargs, file1, file2); 1107 *pflags &= ~D_HEADER; 1108 } 1109 if (diff_format == D_CONTEXT || diff_format == D_UNIFIED) { 1110 /* 1111 * Allocate change records as needed. 1112 */ 1113 if (context_vec_start == NULL || 1114 context_vec_ptr == context_vec_end - 1) { 1115 ptrdiff_t offset = -1; 1116 1117 if (context_vec_start != NULL) 1118 offset = context_vec_ptr - context_vec_start; 1119 max_context <<= 1; 1120 context_vec_start = xreallocarray(context_vec_start, 1121 max_context, sizeof(*context_vec_start)); 1122 context_vec_end = context_vec_start + max_context; 1123 context_vec_ptr = context_vec_start + offset; 1124 } 1125 if (anychange == 0) { 1126 /* 1127 * Print the context/unidiff header first time through. 1128 */ 1129 print_header(file1, file2); 1130 anychange = 1; 1131 } else if (!ckd_add(&dist, diff_context, diff_context) && 1132 a - context_vec_ptr->b - 1 > dist && 1133 c - context_vec_ptr->d - 1 > dist) { 1134 /* 1135 * If this change is more than 'diff_context' lines from the 1136 * previous change, dump the record and reset it. 1137 */ 1138 if (diff_format == D_CONTEXT) 1139 dump_context_vec(f1, f2, *pflags); 1140 else 1141 dump_unified_vec(f1, f2, *pflags); 1142 } 1143 context_vec_ptr++; 1144 context_vec_ptr->a = a; 1145 context_vec_ptr->b = b; 1146 context_vec_ptr->c = c; 1147 context_vec_ptr->d = d; 1148 return; 1149 } 1150 if (anychange == 0) 1151 anychange = 1; 1152 switch (diff_format) { 1153 case D_BRIEF: 1154 return; 1155 case D_NORMAL: 1156 case D_EDIT: 1157 range(a, b, ","); 1158 printf("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1159 if (diff_format == D_NORMAL) 1160 range(c, d, ","); 1161 printf("\n"); 1162 break; 1163 case D_REVERSE: 1164 printf("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1165 range(a, b, " "); 1166 printf("\n"); 1167 break; 1168 case D_NREVERSE: 1169 if (a > b) 1170 printf("a%d %d\n", b, d - c + 1); 1171 else { 1172 printf("d%d %d\n", a, b - a + 1); 1173 if (!(c > d)) 1174 /* add changed lines */ 1175 printf("a%d %d\n", b, d - c + 1); 1176 } 1177 break; 1178 } 1179 if (diff_format == D_GFORMAT) { 1180 curpos = ftell(f1); 1181 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1182 nc = ixold[a > b ? b : a - 1] - curpos; 1183 for (i = 0; i < nc; i++) 1184 printf("%c", getc(f1)); 1185 for (walk = group_format; *walk != '\0'; walk++) { 1186 if (*walk == '%') { 1187 walk++; 1188 switch (*walk) { 1189 case '<': 1190 fetch(ixold, a, b, f1, '<', 1, *pflags); 1191 break; 1192 case '>': 1193 fetch(ixnew, c, d, f2, '>', 0, *pflags); 1194 break; 1195 default: 1196 printf("%%%c", *walk); 1197 break; 1198 } 1199 continue; 1200 } 1201 printf("%c", *walk); 1202 } 1203 } 1204 if (diff_format == D_SIDEBYSIDE) { 1205 if (color && a > b) 1206 printf("\033[%sm", add_code); 1207 else if (color && c > d) 1208 printf("\033[%sm", del_code); 1209 if (a > b) { 1210 print_space(0, hw + lpad, *pflags); 1211 } else { 1212 nc = fetch(ixold, a, b, f1, '\0', 1, *pflags); 1213 print_space(nc, hw - nc + lpad, *pflags); 1214 } 1215 if (color && a > b) 1216 printf("\033[%sm", add_code); 1217 else if (color && c > d) 1218 printf("\033[%sm", del_code); 1219 printf("%c", (a > b) ? '>' : ((c > d) ? '<' : '|')); 1220 if (color && c > d) 1221 printf("\033[m"); 1222 print_space(hw + lpad + 1, rpad, *pflags); 1223 fetch(ixnew, c, d, f2, '\0', 0, *pflags); 1224 printf("\n"); 1225 } 1226 if (diff_format == D_NORMAL || diff_format == D_IFDEF) { 1227 fetch(ixold, a, b, f1, '<', 1, *pflags); 1228 if (a <= b && c <= d && diff_format == D_NORMAL) 1229 printf("---\n"); 1230 } 1231 if (diff_format != D_GFORMAT && diff_format != D_SIDEBYSIDE) 1232 fetch(ixnew, c, d, f2, diff_format == D_NORMAL ? '>' : '\0', 0, *pflags); 1233 if (edoffset != 0 && diff_format == D_EDIT) { 1234 /* 1235 * A non-zero edoffset value for D_EDIT indicates that the last line 1236 * printed was a bare dot (".") that has been escaped as ".." to 1237 * prevent ed(1) from misinterpreting it. We have to add a 1238 * substitute command to change this back and restart where we left 1239 * off. 1240 */ 1241 printf(".\n"); 1242 printf("%ds/.//\n", a + edoffset - 1); 1243 b = a + edoffset - 1; 1244 a = b + 1; 1245 c += edoffset; 1246 goto restart; 1247 } 1248 if ((diff_format == D_EDIT || diff_format == D_REVERSE) && c <= d) 1249 printf(".\n"); 1250 if (inifdef) { 1251 printf("#endif /* %s */\n", ifdefname); 1252 inifdef = 0; 1253 } 1254 } 1255 1256 static int 1257 fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile, int flags) 1258 { 1259 int i, j, c, lastc, col, nc, newcol; 1260 1261 edoffset = 0; 1262 nc = 0; 1263 col = 0; 1264 /* 1265 * When doing #ifdef's, copy down to current line 1266 * if this is the first file, so that stuff makes it to output. 1267 */ 1268 if ((diff_format == D_IFDEF) && oldfile) { 1269 long curpos = ftell(lb); 1270 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1271 nc = f[a > b ? b : a - 1] - curpos; 1272 for (i = 0; i < nc; i++) 1273 printf("%c", getc(lb)); 1274 } 1275 if (a > b) 1276 return (0); 1277 if (diff_format == D_IFDEF) { 1278 if (inifdef) { 1279 printf("#else /* %s%s */\n", 1280 oldfile == 1 ? "!" : "", ifdefname); 1281 } else { 1282 if (oldfile) 1283 printf("#ifndef %s\n", ifdefname); 1284 else 1285 printf("#ifdef %s\n", ifdefname); 1286 } 1287 inifdef = 1 + oldfile; 1288 } 1289 for (i = a; i <= b; i++) { 1290 fseek(lb, f[i - 1], SEEK_SET); 1291 nc = f[i] - f[i - 1]; 1292 if (diff_format == D_SIDEBYSIDE && hw < nc) 1293 nc = hw; 1294 if (diff_format != D_IFDEF && diff_format != D_GFORMAT && 1295 ch != '\0') { 1296 if (color && (ch == '>' || ch == '+')) 1297 printf("\033[%sm", add_code); 1298 else if (color && (ch == '<' || ch == '-')) 1299 printf("\033[%sm", del_code); 1300 printf("%c", ch); 1301 if (Tflag && (diff_format == D_NORMAL || 1302 diff_format == D_CONTEXT || 1303 diff_format == D_UNIFIED)) 1304 printf("\t"); 1305 else if (diff_format != D_UNIFIED) 1306 printf(" "); 1307 } 1308 col = j = 0; 1309 lastc = '\0'; 1310 while (j < nc && (hw == 0 || col < hw)) { 1311 c = getc(lb); 1312 if (flags & D_STRIPCR && c == '\r') { 1313 if ((c = getc(lb)) == '\n') 1314 j++; 1315 else { 1316 ungetc(c, lb); 1317 c = '\r'; 1318 } 1319 } 1320 if (c == EOF) { 1321 if (diff_format == D_EDIT || 1322 diff_format == D_REVERSE || 1323 diff_format == D_NREVERSE) 1324 warnx("No newline at end of file"); 1325 else 1326 printf("\n\\ No newline at end of file\n"); 1327 return (col); 1328 } 1329 /* 1330 * when using --side-by-side, col needs to be increased 1331 * in any case to keep the columns aligned 1332 */ 1333 if (c == '\t') { 1334 /* 1335 * Calculate where the tab would bring us. 1336 * If it would take us to the end of the 1337 * column, either clip it (if expanding 1338 * tabs) or return right away (if not). 1339 */ 1340 newcol = roundup(col + 1, tabsize); 1341 if ((flags & D_EXPANDTABS) == 0) { 1342 if (hw > 0 && newcol >= hw) 1343 return (col); 1344 printf("\t"); 1345 } else { 1346 if (hw > 0 && newcol > hw) 1347 newcol = hw; 1348 printf("%*s", newcol - col, ""); 1349 } 1350 col = newcol; 1351 } else { 1352 if (diff_format == D_EDIT && j == 1 && c == '\n' && 1353 lastc == '.') { 1354 /* 1355 * Don't print a bare "." line since that will confuse 1356 * ed(1). Print ".." instead and set the, global variable 1357 * edoffset to an offset from which to restart. The 1358 * caller must check the value of edoffset 1359 */ 1360 printf(".\n"); 1361 edoffset = i - a + 1; 1362 return (edoffset); 1363 } 1364 /* when side-by-side, do not print a newline */ 1365 if (diff_format != D_SIDEBYSIDE || c != '\n') { 1366 if (color && c == '\n') 1367 printf("\033[m%c", c); 1368 else 1369 printf("%c", c); 1370 col++; 1371 } 1372 } 1373 1374 j++; 1375 lastc = c; 1376 } 1377 } 1378 if (color && diff_format == D_SIDEBYSIDE) 1379 printf("\033[m"); 1380 return (col); 1381 } 1382 1383 /* 1384 * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578. 1385 */ 1386 static enum readhash 1387 readhash(FILE *f, int flags, unsigned *hash) 1388 { 1389 int i, t, space; 1390 unsigned sum; 1391 1392 sum = 1; 1393 space = 0; 1394 for (i = 0;;) { 1395 switch (t = getc(f)) { 1396 case '\0': 1397 if ((flags & D_FORCEASCII) == 0) 1398 return (RH_BINARY); 1399 goto hashchar; 1400 case '\r': 1401 if (flags & D_STRIPCR) { 1402 t = getc(f); 1403 if (t == '\n') 1404 break; 1405 ungetc(t, f); 1406 } 1407 /* FALLTHROUGH */ 1408 case '\t': 1409 case '\v': 1410 case '\f': 1411 case ' ': 1412 if ((flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) != 0) { 1413 space++; 1414 continue; 1415 } 1416 /* FALLTHROUGH */ 1417 default: 1418 hashchar: 1419 if (space && (flags & D_IGNOREBLANKS) == 0) { 1420 i++; 1421 space = 0; 1422 } 1423 sum = sum * 127 + chrtran(t); 1424 i++; 1425 continue; 1426 case EOF: 1427 if (i == 0) 1428 return (RH_EOF); 1429 /* FALLTHROUGH */ 1430 case '\n': 1431 break; 1432 } 1433 break; 1434 } 1435 *hash = sum; 1436 return (RH_OK); 1437 } 1438 1439 static int 1440 asciifile(FILE *f) 1441 { 1442 unsigned char buf[BUFSIZ]; 1443 size_t cnt; 1444 1445 if (f == NULL) 1446 return (1); 1447 1448 rewind(f); 1449 cnt = fread(buf, 1, sizeof(buf), f); 1450 return (memchr(buf, '\0', cnt) == NULL); 1451 } 1452 1453 #define begins_with(s, pre) (strncmp(s, pre, sizeof(pre) - 1) == 0) 1454 1455 static char * 1456 match_function(const long *f, int pos, FILE *fp) 1457 { 1458 unsigned char buf[FUNCTION_CONTEXT_SIZE]; 1459 size_t nc; 1460 int last = lastline; 1461 const char *state = NULL; 1462 1463 lastline = pos; 1464 for (; pos > last; pos--) { 1465 fseek(fp, f[pos - 1], SEEK_SET); 1466 nc = f[pos] - f[pos - 1]; 1467 if (nc >= sizeof(buf)) 1468 nc = sizeof(buf) - 1; 1469 nc = fread(buf, 1, nc, fp); 1470 if (nc == 0) 1471 continue; 1472 buf[nc] = '\0'; 1473 buf[strcspn(buf, "\n")] = '\0'; 1474 if (most_recent_pat != NULL) { 1475 int ret = regexec(&most_recent_re, buf, 0, NULL, 0); 1476 1477 if (ret != 0) 1478 continue; 1479 strlcpy(lastbuf, buf, sizeof(lastbuf)); 1480 lastmatchline = pos; 1481 return (lastbuf); 1482 } else if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$' 1483 || buf[0] == '-' || buf[0] == '+') { 1484 if (begins_with(buf, "private:")) { 1485 if (!state) 1486 state = " (private)"; 1487 } else if (begins_with(buf, "protected:")) { 1488 if (!state) 1489 state = " (protected)"; 1490 } else if (begins_with(buf, "public:")) { 1491 if (!state) 1492 state = " (public)"; 1493 } else { 1494 strlcpy(lastbuf, buf, sizeof(lastbuf)); 1495 if (state) 1496 strlcat(lastbuf, state, sizeof(lastbuf)); 1497 lastmatchline = pos; 1498 return (lastbuf); 1499 } 1500 } 1501 } 1502 return (lastmatchline > 0 ? lastbuf : NULL); 1503 } 1504 1505 /* dump accumulated "context" diff changes */ 1506 static void 1507 dump_context_vec(FILE *f1, FILE *f2, int flags) 1508 { 1509 struct context_vec *cvp = context_vec_start; 1510 int lowa, upb, lowc, upd, do_output; 1511 int a, b, c, d; 1512 char ch, *f; 1513 1514 if (context_vec_start > context_vec_ptr) 1515 return; 1516 1517 b = d = 0; /* gcc */ 1518 if (ckd_sub(&lowa, cvp->a, diff_context) || lowa < 1) 1519 lowa = 1; 1520 if (ckd_add(&upb, context_vec_ptr->b, diff_context) || upb > (int)len[0]) 1521 upb = (int)len[0]; 1522 if (ckd_sub(&lowc, cvp->c, diff_context) || lowc < 1) 1523 lowc = 1; 1524 if (ckd_add(&upd, context_vec_ptr->d, diff_context) || upd > (int)len[1]) 1525 upd = (int)len[1]; 1526 1527 printf("***************"); 1528 if (flags & (D_PROTOTYPE | D_MATCHLAST)) { 1529 f = match_function(ixold, cvp->a - 1, f1); 1530 if (f != NULL) 1531 printf(" %s", f); 1532 } 1533 printf("\n*** "); 1534 range(lowa, upb, ","); 1535 printf(" ****\n"); 1536 1537 /* 1538 * Output changes to the "old" file. The first loop suppresses 1539 * output if there were no changes to the "old" file (we'll see 1540 * the "old" lines as context in the "new" list). 1541 */ 1542 do_output = 0; 1543 for (; cvp <= context_vec_ptr; cvp++) 1544 if (cvp->a <= cvp->b) { 1545 cvp = context_vec_start; 1546 do_output++; 1547 break; 1548 } 1549 if (do_output) { 1550 while (cvp <= context_vec_ptr) { 1551 a = cvp->a; 1552 b = cvp->b; 1553 c = cvp->c; 1554 d = cvp->d; 1555 1556 if (a <= b && c <= d) 1557 ch = 'c'; 1558 else 1559 ch = (a <= b) ? 'd' : 'a'; 1560 1561 if (ch == 'a') 1562 fetch(ixold, lowa, b, f1, ' ', 0, flags); 1563 else { 1564 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1565 fetch(ixold, a, b, f1, 1566 ch == 'c' ? '!' : '-', 0, flags); 1567 } 1568 lowa = b + 1; 1569 cvp++; 1570 } 1571 fetch(ixold, b + 1, upb, f1, ' ', 0, flags); 1572 } 1573 /* output changes to the "new" file */ 1574 printf("--- "); 1575 range(lowc, upd, ","); 1576 printf(" ----\n"); 1577 1578 do_output = 0; 1579 for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++) 1580 if (cvp->c <= cvp->d) { 1581 cvp = context_vec_start; 1582 do_output++; 1583 break; 1584 } 1585 if (do_output) { 1586 while (cvp <= context_vec_ptr) { 1587 a = cvp->a; 1588 b = cvp->b; 1589 c = cvp->c; 1590 d = cvp->d; 1591 1592 if (a <= b && c <= d) 1593 ch = 'c'; 1594 else 1595 ch = (a <= b) ? 'd' : 'a'; 1596 1597 if (ch == 'd') 1598 fetch(ixnew, lowc, d, f2, ' ', 0, flags); 1599 else { 1600 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1601 fetch(ixnew, c, d, f2, 1602 ch == 'c' ? '!' : '+', 0, flags); 1603 } 1604 lowc = d + 1; 1605 cvp++; 1606 } 1607 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1608 } 1609 context_vec_ptr = context_vec_start - 1; 1610 } 1611 1612 /* dump accumulated "unified" diff changes */ 1613 static void 1614 dump_unified_vec(FILE *f1, FILE *f2, int flags) 1615 { 1616 struct context_vec *cvp = context_vec_start; 1617 int lowa, upb, lowc, upd; 1618 int a, b, c, d; 1619 char ch, *f; 1620 1621 if (context_vec_start > context_vec_ptr) 1622 return; 1623 1624 b = d = 0; /* gcc */ 1625 if (ckd_sub(&lowa, cvp->a, diff_context) || lowa < 1) 1626 lowa = 1; 1627 if (ckd_add(&upb, context_vec_ptr->b, diff_context) || upb > (int)len[0]) 1628 upb = (int)len[0]; 1629 if (ckd_sub(&lowc, cvp->c, diff_context) || lowc < 1) 1630 lowc = 1; 1631 if (ckd_add(&upd, context_vec_ptr->d, diff_context) || upd > (int)len[1]) 1632 upd = (int)len[1]; 1633 1634 printf("@@ -"); 1635 uni_range(lowa, upb); 1636 printf(" +"); 1637 uni_range(lowc, upd); 1638 printf(" @@"); 1639 if (flags & (D_PROTOTYPE | D_MATCHLAST)) { 1640 f = match_function(ixold, cvp->a - 1, f1); 1641 if (f != NULL) 1642 printf(" %s", f); 1643 } 1644 printf("\n"); 1645 1646 /* 1647 * Output changes in "unified" diff format--the old and new lines 1648 * are printed together. 1649 */ 1650 for (; cvp <= context_vec_ptr; cvp++) { 1651 a = cvp->a; 1652 b = cvp->b; 1653 c = cvp->c; 1654 d = cvp->d; 1655 1656 /* 1657 * c: both new and old changes 1658 * d: only changes in the old file 1659 * a: only changes in the new file 1660 */ 1661 if (a <= b && c <= d) 1662 ch = 'c'; 1663 else 1664 ch = (a <= b) ? 'd' : 'a'; 1665 1666 switch (ch) { 1667 case 'c': 1668 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1669 fetch(ixold, a, b, f1, '-', 0, flags); 1670 fetch(ixnew, c, d, f2, '+', 0, flags); 1671 break; 1672 case 'd': 1673 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1674 fetch(ixold, a, b, f1, '-', 0, flags); 1675 break; 1676 case 'a': 1677 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1678 fetch(ixnew, c, d, f2, '+', 0, flags); 1679 break; 1680 } 1681 lowa = b + 1; 1682 lowc = d + 1; 1683 } 1684 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1685 1686 context_vec_ptr = context_vec_start - 1; 1687 } 1688 1689 static void 1690 print_header(const char *file1, const char *file2) 1691 { 1692 const char *time_format; 1693 char buf[256]; 1694 struct tm tm1, tm2, *tm_ptr1, *tm_ptr2; 1695 int nsec1 = stb1.st_mtim.tv_nsec; 1696 int nsec2 = stb2.st_mtim.tv_nsec; 1697 1698 time_format = "%Y-%m-%d %H:%M:%S"; 1699 1700 if (cflag) 1701 time_format = "%c"; 1702 tm_ptr1 = localtime_r(&stb1.st_mtime, &tm1); 1703 tm_ptr2 = localtime_r(&stb2.st_mtime, &tm2); 1704 if (label[0] != NULL) 1705 printf("%s %s\n", diff_format == D_CONTEXT ? "***" : "---", 1706 label[0]); 1707 else { 1708 strftime(buf, sizeof(buf), time_format, tm_ptr1); 1709 printf("%s %s\t%s", diff_format == D_CONTEXT ? "***" : "---", 1710 file1, buf); 1711 if (!cflag) { 1712 strftime(buf, sizeof(buf), "%z", tm_ptr1); 1713 printf(".%.9d %s", nsec1, buf); 1714 } 1715 printf("\n"); 1716 } 1717 if (label[1] != NULL) 1718 printf("%s %s\n", diff_format == D_CONTEXT ? "---" : "+++", 1719 label[1]); 1720 else { 1721 strftime(buf, sizeof(buf), time_format, tm_ptr2); 1722 printf("%s %s\t%s", diff_format == D_CONTEXT ? "---" : "+++", 1723 file2, buf); 1724 if (!cflag) { 1725 strftime(buf, sizeof(buf), "%z", tm_ptr2); 1726 printf(".%.9d %s", nsec2, buf); 1727 } 1728 printf("\n"); 1729 } 1730 } 1731 1732 /* 1733 * Prints n number of space characters either by using tab 1734 * or single space characters. 1735 * nc is the preceding number of characters 1736 */ 1737 static void 1738 print_space(int nc, int n, int flags) 1739 { 1740 int col, newcol, tabstop; 1741 1742 col = nc; 1743 newcol = nc + n; 1744 /* first, use tabs if allowed */ 1745 if ((flags & D_EXPANDTABS) == 0) { 1746 while ((tabstop = roundup(col + 1, tabsize)) <= newcol) { 1747 printf("\t"); 1748 col = tabstop; 1749 } 1750 } 1751 /* finish with spaces */ 1752 printf("%*s", newcol - col, ""); 1753 } 1754