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 * @(#)diffreg.c 8.1 (Berkeley) 6/6/93 67 */ 68 69 #include <sys/cdefs.h> 70 __FBSDID("$FreeBSD$"); 71 72 #include <sys/capsicum.h> 73 #include <sys/stat.h> 74 75 #include <capsicum_helpers.h> 76 #include <ctype.h> 77 #include <err.h> 78 #include <errno.h> 79 #include <fcntl.h> 80 #include <paths.h> 81 #include <regex.h> 82 #include <stdbool.h> 83 #include <stddef.h> 84 #include <stdint.h> 85 #include <stdio.h> 86 #include <stdlib.h> 87 #include <string.h> 88 89 #include "pr.h" 90 #include "diff.h" 91 #include "xmalloc.h" 92 93 /* 94 * diff - compare two files. 95 */ 96 97 /* 98 * Uses an algorithm due to Harold Stone, which finds 99 * a pair of longest identical subsequences in the two 100 * files. 101 * 102 * The major goal is to generate the match vector J. 103 * J[i] is the index of the line in file1 corresponding 104 * to line i file0. J[i] = 0 if there is no 105 * such line in file1. 106 * 107 * Lines are hashed so as to work in core. All potential 108 * matches are located by sorting the lines of each file 109 * on the hash (called ``value''). In particular, this 110 * collects the equivalence classes in file1 together. 111 * Subroutine equiv replaces the value of each line in 112 * file0 by the index of the first element of its 113 * matching equivalence in (the reordered) file1. 114 * To save space equiv squeezes file1 into a single 115 * array member in which the equivalence classes 116 * are simply concatenated, except that their first 117 * members are flagged by changing sign. 118 * 119 * Next the indices that point into member are unsorted into 120 * array class according to the original order of file0. 121 * 122 * The cleverness lies in routine stone. This marches 123 * through the lines of file0, developing a vector klist 124 * of "k-candidates". At step i a k-candidate is a matched 125 * pair of lines x,y (x in file0 y in file1) such that 126 * there is a common subsequence of length k 127 * between the first i lines of file0 and the first y 128 * lines of file1, but there is no such subsequence for 129 * any smaller y. x is the earliest possible mate to y 130 * that occurs in such a subsequence. 131 * 132 * Whenever any of the members of the equivalence class of 133 * lines in file1 matable to a line in file0 has serial number 134 * less than the y of some k-candidate, that k-candidate 135 * with the smallest such y is replaced. The new 136 * k-candidate is chained (via pred) to the current 137 * k-1 candidate so that the actual subsequence can 138 * be recovered. When a member has serial number greater 139 * that the y of all k-candidates, the klist is extended. 140 * At the end, the longest subsequence is pulled out 141 * and placed in the array J by unravel 142 * 143 * With J in hand, the matches there recorded are 144 * check'ed against reality to assure that no spurious 145 * matches have crept in due to hashing. If they have, 146 * they are broken, and "jackpot" is recorded--a harmless 147 * matter except that a true match for a spuriously 148 * mated line may now be unnecessarily reported as a change. 149 * 150 * Much of the complexity of the program comes simply 151 * from trying to minimize core utilization and 152 * maximize the range of doable problems by dynamically 153 * allocating what is needed and reusing what is not. 154 * The core requirements for problems larger than somewhat 155 * are (in words) 2*length(file0) + length(file1) + 156 * 3*(number of k-candidates installed), typically about 157 * 6n words for files of length n. 158 */ 159 160 struct cand { 161 int x; 162 int y; 163 int pred; 164 }; 165 166 static struct line { 167 int serial; 168 int value; 169 } *file[2]; 170 171 /* 172 * The following struct is used to record change information when 173 * doing a "context" or "unified" diff. (see routine "change" to 174 * understand the highly mnemonic field names) 175 */ 176 struct context_vec { 177 int a; /* start line in old file */ 178 int b; /* end line in old file */ 179 int c; /* start line in new file */ 180 int d; /* end line in new file */ 181 }; 182 183 #define diff_output printf 184 #define MIN_PAD 1 185 static FILE *opentemp(const char *); 186 static void output(char *, FILE *, char *, FILE *, int); 187 static void check(FILE *, FILE *, int); 188 static void range(int, int, const char *); 189 static void uni_range(int, int); 190 static void dump_context_vec(FILE *, FILE *, int); 191 static void dump_unified_vec(FILE *, FILE *, int); 192 static void prepare(int, FILE *, size_t, int); 193 static void prune(void); 194 static void equiv(struct line *, int, struct line *, int, int *); 195 static void unravel(int); 196 static void unsort(struct line *, int, int *); 197 static void change(char *, FILE *, char *, FILE *, int, int, int, int, int *); 198 static void sort(struct line *, int); 199 static void print_header(const char *, const char *); 200 static void print_space(int, int, int); 201 static bool ignoreline_pattern(char *); 202 static bool ignoreline(char *, bool); 203 static int asciifile(FILE *); 204 static int fetch(long *, int, int, FILE *, int, int, int); 205 static int newcand(int, int, int); 206 static int search(int *, int, int); 207 static int skipline(FILE *); 208 static int isqrt(int); 209 static int stone(int *, int, int *, int *, int); 210 static int readhash(FILE *, int); 211 static int files_differ(FILE *, FILE *, int); 212 static char *match_function(const long *, int, FILE *); 213 static char *preadline(int, size_t, off_t); 214 215 static int *J; /* will be overlaid on class */ 216 static int *class; /* will be overlaid on file[0] */ 217 static int *klist; /* will be overlaid on file[0] after class */ 218 static int *member; /* will be overlaid on file[1] */ 219 static int clen; 220 static int inifdef; /* whether or not we are in a #ifdef block */ 221 static int len[2]; 222 static int pref, suff; /* length of prefix and suffix */ 223 static int slen[2]; 224 static int anychange; 225 static int hw, padding; /* half width and padding */ 226 static int edoffset; 227 static long *ixnew; /* will be overlaid on file[1] */ 228 static long *ixold; /* will be overlaid on klist */ 229 static struct cand *clist; /* merely a free storage pot for candidates */ 230 static int clistlen; /* the length of clist */ 231 static struct line *sfile[2]; /* shortened by pruning common prefix/suffix */ 232 static int (*chrtran)(int); /* translation table for case-folding */ 233 static struct context_vec *context_vec_start; 234 static struct context_vec *context_vec_end; 235 static struct context_vec *context_vec_ptr; 236 237 #define FUNCTION_CONTEXT_SIZE 55 238 static char lastbuf[FUNCTION_CONTEXT_SIZE]; 239 static int lastline; 240 static int lastmatchline; 241 242 static int 243 clow2low(int c) 244 { 245 246 return (c); 247 } 248 249 static int 250 cup2low(int c) 251 { 252 253 return tolower(c); 254 } 255 256 int 257 diffreg(char *file1, char *file2, int flags, int capsicum) 258 { 259 FILE *f1, *f2; 260 int i, rval; 261 struct pr *pr = NULL; 262 cap_rights_t rights_ro; 263 264 f1 = f2 = NULL; 265 rval = D_SAME; 266 anychange = 0; 267 lastline = 0; 268 lastmatchline = 0; 269 context_vec_ptr = context_vec_start - 1; 270 271 /* 272 * hw excludes padding and make sure when -t is not used, 273 * the second column always starts from the closest tab stop 274 */ 275 if (diff_format == D_SIDEBYSIDE) { 276 hw = width >> 1; 277 padding = tabsize - (hw % tabsize); 278 if ((flags & D_EXPANDTABS) != 0 || (padding % tabsize == 0)) 279 padding = MIN_PAD; 280 281 hw = (width >> 1) - 282 ((padding == MIN_PAD) ? (padding << 1) : padding) - 1; 283 } 284 285 286 if (flags & D_IGNORECASE) 287 chrtran = cup2low; 288 else 289 chrtran = clow2low; 290 if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode)) 291 return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2); 292 if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0) 293 goto closem; 294 295 if (flags & D_EMPTY1) 296 f1 = fopen(_PATH_DEVNULL, "r"); 297 else { 298 if (!S_ISREG(stb1.st_mode)) { 299 if ((f1 = opentemp(file1)) == NULL || 300 fstat(fileno(f1), &stb1) == -1) { 301 warn("%s", file1); 302 rval = D_ERROR; 303 status |= 2; 304 goto closem; 305 } 306 } else if (strcmp(file1, "-") == 0) 307 f1 = stdin; 308 else 309 f1 = fopen(file1, "r"); 310 } 311 if (f1 == NULL) { 312 warn("%s", file1); 313 rval = D_ERROR; 314 status |= 2; 315 goto closem; 316 } 317 318 if (flags & D_EMPTY2) 319 f2 = fopen(_PATH_DEVNULL, "r"); 320 else { 321 if (!S_ISREG(stb2.st_mode)) { 322 if ((f2 = opentemp(file2)) == NULL || 323 fstat(fileno(f2), &stb2) == -1) { 324 warn("%s", file2); 325 rval = D_ERROR; 326 status |= 2; 327 goto closem; 328 } 329 } else if (strcmp(file2, "-") == 0) 330 f2 = stdin; 331 else 332 f2 = fopen(file2, "r"); 333 } 334 if (f2 == NULL) { 335 warn("%s", file2); 336 rval = D_ERROR; 337 status |= 2; 338 goto closem; 339 } 340 341 if (lflag) 342 pr = start_pr(file1, file2); 343 344 if (capsicum) { 345 cap_rights_init(&rights_ro, CAP_READ, CAP_FSTAT, CAP_SEEK); 346 if (caph_rights_limit(fileno(f1), &rights_ro) < 0) 347 err(2, "unable to limit rights on: %s", file1); 348 if (caph_rights_limit(fileno(f2), &rights_ro) < 0) 349 err(2, "unable to limit rights on: %s", file2); 350 if (fileno(f1) == STDIN_FILENO || fileno(f2) == STDIN_FILENO) { 351 /* stding has already been limited */ 352 if (caph_limit_stderr() == -1) 353 err(2, "unable to limit stderr"); 354 if (caph_limit_stdout() == -1) 355 err(2, "unable to limit stdout"); 356 } else if (caph_limit_stdio() == -1) 357 err(2, "unable to limit stdio"); 358 359 caph_cache_catpages(); 360 caph_cache_tzdata(); 361 if (caph_enter() < 0) 362 err(2, "unable to enter capability mode"); 363 } 364 365 switch (files_differ(f1, f2, flags)) { 366 case 0: 367 goto closem; 368 case 1: 369 break; 370 default: 371 /* error */ 372 rval = D_ERROR; 373 status |= 2; 374 goto closem; 375 } 376 377 if (diff_format == D_BRIEF && ignore_pats == NULL && 378 (flags & (D_FOLDBLANKS|D_IGNOREBLANKS|D_IGNORECASE|D_STRIPCR)) == 0) 379 { 380 rval = D_DIFFER; 381 status |= 1; 382 goto closem; 383 } 384 if ((flags & D_FORCEASCII) == 0 && 385 (!asciifile(f1) || !asciifile(f2))) { 386 rval = D_BINARY; 387 status |= 1; 388 goto closem; 389 } 390 prepare(0, f1, stb1.st_size, flags); 391 prepare(1, f2, stb2.st_size, flags); 392 393 prune(); 394 sort(sfile[0], slen[0]); 395 sort(sfile[1], slen[1]); 396 397 member = (int *)file[1]; 398 equiv(sfile[0], slen[0], sfile[1], slen[1], member); 399 member = xreallocarray(member, slen[1] + 2, sizeof(*member)); 400 401 class = (int *)file[0]; 402 unsort(sfile[0], slen[0], class); 403 class = xreallocarray(class, slen[0] + 2, sizeof(*class)); 404 405 klist = xcalloc(slen[0] + 2, sizeof(*klist)); 406 clen = 0; 407 clistlen = 100; 408 clist = xcalloc(clistlen, sizeof(*clist)); 409 i = stone(class, slen[0], member, klist, flags); 410 free(member); 411 free(class); 412 413 J = xreallocarray(J, len[0] + 2, sizeof(*J)); 414 unravel(klist[i]); 415 free(clist); 416 free(klist); 417 418 ixold = xreallocarray(ixold, len[0] + 2, sizeof(*ixold)); 419 ixnew = xreallocarray(ixnew, len[1] + 2, sizeof(*ixnew)); 420 check(f1, f2, flags); 421 output(file1, f1, file2, f2, flags); 422 423 closem: 424 if (pr != NULL) 425 stop_pr(pr); 426 if (anychange) { 427 status |= 1; 428 if (rval == D_SAME) 429 rval = D_DIFFER; 430 } 431 if (f1 != NULL) 432 fclose(f1); 433 if (f2 != NULL) 434 fclose(f2); 435 436 return (rval); 437 } 438 439 /* 440 * Check to see if the given files differ. 441 * Returns 0 if they are the same, 1 if different, and -1 on error. 442 * XXX - could use code from cmp(1) [faster] 443 */ 444 static int 445 files_differ(FILE *f1, FILE *f2, int flags) 446 { 447 char buf1[BUFSIZ], buf2[BUFSIZ]; 448 size_t i, j; 449 450 if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size || 451 (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT)) 452 return (1); 453 for (;;) { 454 i = fread(buf1, 1, sizeof(buf1), f1); 455 j = fread(buf2, 1, sizeof(buf2), f2); 456 if ((!i && ferror(f1)) || (!j && ferror(f2))) 457 return (-1); 458 if (i != j) 459 return (1); 460 if (i == 0) 461 return (0); 462 if (memcmp(buf1, buf2, i) != 0) 463 return (1); 464 } 465 } 466 467 static FILE * 468 opentemp(const char *f) 469 { 470 char buf[BUFSIZ], tempfile[PATH_MAX]; 471 ssize_t nread; 472 int ifd, ofd; 473 474 if (strcmp(f, "-") == 0) 475 ifd = STDIN_FILENO; 476 else if ((ifd = open(f, O_RDONLY, 0644)) == -1) 477 return (NULL); 478 479 (void)strlcpy(tempfile, _PATH_TMP "/diff.XXXXXXXX", sizeof(tempfile)); 480 481 if ((ofd = mkstemp(tempfile)) == -1) { 482 close(ifd); 483 return (NULL); 484 } 485 unlink(tempfile); 486 while ((nread = read(ifd, buf, BUFSIZ)) > 0) { 487 if (write(ofd, buf, nread) != nread) { 488 close(ifd); 489 close(ofd); 490 return (NULL); 491 } 492 } 493 close(ifd); 494 lseek(ofd, (off_t)0, SEEK_SET); 495 return (fdopen(ofd, "r")); 496 } 497 498 char * 499 splice(char *dir, char *path) 500 { 501 char *tail, *buf; 502 size_t dirlen; 503 504 dirlen = strlen(dir); 505 while (dirlen != 0 && dir[dirlen - 1] == '/') 506 dirlen--; 507 if ((tail = strrchr(path, '/')) == NULL) 508 tail = path; 509 else 510 tail++; 511 xasprintf(&buf, "%.*s/%s", (int)dirlen, dir, tail); 512 return (buf); 513 } 514 515 static void 516 prepare(int i, FILE *fd, size_t filesize, int flags) 517 { 518 struct line *p; 519 int h; 520 size_t sz, j; 521 522 rewind(fd); 523 524 sz = MIN(filesize, SIZE_MAX) / 25; 525 if (sz < 100) 526 sz = 100; 527 528 p = xcalloc(sz + 3, sizeof(*p)); 529 for (j = 0; (h = readhash(fd, flags));) { 530 if (j == sz) { 531 sz = sz * 3 / 2; 532 p = xreallocarray(p, sz + 3, sizeof(*p)); 533 } 534 p[++j].value = h; 535 } 536 len[i] = j; 537 file[i] = p; 538 } 539 540 static void 541 prune(void) 542 { 543 int i, j; 544 545 for (pref = 0; pref < len[0] && pref < len[1] && 546 file[0][pref + 1].value == file[1][pref + 1].value; 547 pref++) 548 ; 549 for (suff = 0; suff < len[0] - pref && suff < len[1] - pref && 550 file[0][len[0] - suff].value == file[1][len[1] - suff].value; 551 suff++) 552 ; 553 for (j = 0; j < 2; j++) { 554 sfile[j] = file[j] + pref; 555 slen[j] = len[j] - pref - suff; 556 for (i = 0; i <= slen[j]; i++) 557 sfile[j][i].serial = i; 558 } 559 } 560 561 static void 562 equiv(struct line *a, int n, struct line *b, int m, int *c) 563 { 564 int i, j; 565 566 i = j = 1; 567 while (i <= n && j <= m) { 568 if (a[i].value < b[j].value) 569 a[i++].value = 0; 570 else if (a[i].value == b[j].value) 571 a[i++].value = j; 572 else 573 j++; 574 } 575 while (i <= n) 576 a[i++].value = 0; 577 b[m + 1].value = 0; 578 j = 0; 579 while (++j <= m) { 580 c[j] = -b[j].serial; 581 while (b[j + 1].value == b[j].value) { 582 j++; 583 c[j] = b[j].serial; 584 } 585 } 586 c[j] = -1; 587 } 588 589 /* Code taken from ping.c */ 590 static int 591 isqrt(int n) 592 { 593 int y, x = 1; 594 595 if (n == 0) 596 return (0); 597 598 do { /* newton was a stinker */ 599 y = x; 600 x = n / x; 601 x += y; 602 x /= 2; 603 } while ((x - y) > 1 || (x - y) < -1); 604 605 return (x); 606 } 607 608 static int 609 stone(int *a, int n, int *b, int *c, int flags) 610 { 611 int i, k, y, j, l; 612 int oldc, tc, oldl, sq; 613 u_int numtries, bound; 614 615 if (flags & D_MINIMAL) 616 bound = UINT_MAX; 617 else { 618 sq = isqrt(n); 619 bound = MAX(256, sq); 620 } 621 622 k = 0; 623 c[0] = newcand(0, 0, 0); 624 for (i = 1; i <= n; i++) { 625 j = a[i]; 626 if (j == 0) 627 continue; 628 y = -b[j]; 629 oldl = 0; 630 oldc = c[0]; 631 numtries = 0; 632 do { 633 if (y <= clist[oldc].y) 634 continue; 635 l = search(c, k, y); 636 if (l != oldl + 1) 637 oldc = c[l - 1]; 638 if (l <= k) { 639 if (clist[c[l]].y <= y) 640 continue; 641 tc = c[l]; 642 c[l] = newcand(i, y, oldc); 643 oldc = tc; 644 oldl = l; 645 numtries++; 646 } else { 647 c[l] = newcand(i, y, oldc); 648 k++; 649 break; 650 } 651 } while ((y = b[++j]) > 0 && numtries < bound); 652 } 653 return (k); 654 } 655 656 static int 657 newcand(int x, int y, int pred) 658 { 659 struct cand *q; 660 661 if (clen == clistlen) { 662 clistlen = clistlen * 11 / 10; 663 clist = xreallocarray(clist, clistlen, sizeof(*clist)); 664 } 665 q = clist + clen; 666 q->x = x; 667 q->y = y; 668 q->pred = pred; 669 return (clen++); 670 } 671 672 static int 673 search(int *c, int k, int y) 674 { 675 int i, j, l, t; 676 677 if (clist[c[k]].y < y) /* quick look for typical case */ 678 return (k + 1); 679 i = 0; 680 j = k + 1; 681 for (;;) { 682 l = (i + j) / 2; 683 if (l <= i) 684 break; 685 t = clist[c[l]].y; 686 if (t > y) 687 j = l; 688 else if (t < y) 689 i = l; 690 else 691 return (l); 692 } 693 return (l + 1); 694 } 695 696 static void 697 unravel(int p) 698 { 699 struct cand *q; 700 int i; 701 702 for (i = 0; i <= len[0]; i++) 703 J[i] = i <= pref ? i : 704 i > len[0] - suff ? i + len[1] - len[0] : 0; 705 for (q = clist + p; q->y != 0; q = clist + q->pred) 706 J[q->x + pref] = q->y + pref; 707 } 708 709 /* 710 * Check does double duty: 711 * 1. ferret out any fortuitous correspondences due 712 * to confounding by hashing (which result in "jackpot") 713 * 2. collect random access indexes to the two files 714 */ 715 static void 716 check(FILE *f1, FILE *f2, int flags) 717 { 718 int i, j, jackpot, c, d; 719 long ctold, ctnew; 720 721 rewind(f1); 722 rewind(f2); 723 j = 1; 724 ixold[0] = ixnew[0] = 0; 725 jackpot = 0; 726 ctold = ctnew = 0; 727 for (i = 1; i <= len[0]; i++) { 728 if (J[i] == 0) { 729 ixold[i] = ctold += skipline(f1); 730 continue; 731 } 732 while (j < J[i]) { 733 ixnew[j] = ctnew += skipline(f2); 734 j++; 735 } 736 if (flags & (D_FOLDBLANKS|D_IGNOREBLANKS|D_IGNORECASE|D_STRIPCR)) { 737 for (;;) { 738 c = getc(f1); 739 d = getc(f2); 740 /* 741 * GNU diff ignores a missing newline 742 * in one file for -b or -w. 743 */ 744 if (flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) { 745 if (c == EOF && d == '\n') { 746 ctnew++; 747 break; 748 } else if (c == '\n' && d == EOF) { 749 ctold++; 750 break; 751 } 752 } 753 ctold++; 754 ctnew++; 755 if (flags & D_STRIPCR && (c == '\r' || d == '\r')) { 756 if (c == '\r') { 757 if ((c = getc(f1)) == '\n') { 758 ctold++; 759 } else { 760 ungetc(c, f1); 761 } 762 } 763 if (d == '\r') { 764 if ((d = getc(f2)) == '\n') { 765 ctnew++; 766 } else { 767 ungetc(d, f2); 768 } 769 } 770 break; 771 } 772 if ((flags & D_FOLDBLANKS) && isspace(c) && 773 isspace(d)) { 774 do { 775 if (c == '\n') 776 break; 777 ctold++; 778 } while (isspace(c = getc(f1))); 779 do { 780 if (d == '\n') 781 break; 782 ctnew++; 783 } while (isspace(d = getc(f2))); 784 } else if ((flags & D_IGNOREBLANKS)) { 785 while (isspace(c) && c != '\n') { 786 c = getc(f1); 787 ctold++; 788 } 789 while (isspace(d) && d != '\n') { 790 d = getc(f2); 791 ctnew++; 792 } 793 } 794 if (chrtran(c) != chrtran(d)) { 795 jackpot++; 796 J[i] = 0; 797 if (c != '\n' && c != EOF) 798 ctold += skipline(f1); 799 if (d != '\n' && c != EOF) 800 ctnew += skipline(f2); 801 break; 802 } 803 if (c == '\n' || c == EOF) 804 break; 805 } 806 } else { 807 for (;;) { 808 ctold++; 809 ctnew++; 810 if ((c = getc(f1)) != (d = getc(f2))) { 811 /* jackpot++; */ 812 J[i] = 0; 813 if (c != '\n' && c != EOF) 814 ctold += skipline(f1); 815 if (d != '\n' && c != EOF) 816 ctnew += skipline(f2); 817 break; 818 } 819 if (c == '\n' || c == EOF) 820 break; 821 } 822 } 823 ixold[i] = ctold; 824 ixnew[j] = ctnew; 825 j++; 826 } 827 for (; j <= len[1]; j++) { 828 ixnew[j] = ctnew += skipline(f2); 829 } 830 /* 831 * if (jackpot) 832 * fprintf(stderr, "jackpot\n"); 833 */ 834 } 835 836 /* shellsort CACM #201 */ 837 static void 838 sort(struct line *a, int n) 839 { 840 struct line *ai, *aim, w; 841 int j, m = 0, k; 842 843 if (n == 0) 844 return; 845 for (j = 1; j <= n; j *= 2) 846 m = 2 * j - 1; 847 for (m /= 2; m != 0; m /= 2) { 848 k = n - m; 849 for (j = 1; j <= k; j++) { 850 for (ai = &a[j]; ai > a; ai -= m) { 851 aim = &ai[m]; 852 if (aim < ai) 853 break; /* wraparound */ 854 if (aim->value > ai[0].value || 855 (aim->value == ai[0].value && 856 aim->serial > ai[0].serial)) 857 break; 858 w.value = ai[0].value; 859 ai[0].value = aim->value; 860 aim->value = w.value; 861 w.serial = ai[0].serial; 862 ai[0].serial = aim->serial; 863 aim->serial = w.serial; 864 } 865 } 866 } 867 } 868 869 static void 870 unsort(struct line *f, int l, int *b) 871 { 872 int *a, i; 873 874 a = xcalloc(l + 1, sizeof(*a)); 875 for (i = 1; i <= l; i++) 876 a[f[i].serial] = f[i].value; 877 for (i = 1; i <= l; i++) 878 b[i] = a[i]; 879 free(a); 880 } 881 882 static int 883 skipline(FILE *f) 884 { 885 int i, c; 886 887 for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++) 888 continue; 889 return (i); 890 } 891 892 static void 893 output(char *file1, FILE *f1, char *file2, FILE *f2, int flags) 894 { 895 int i, j, m, i0, i1, j0, j1, nc; 896 897 rewind(f1); 898 rewind(f2); 899 m = len[0]; 900 J[0] = 0; 901 J[m + 1] = len[1] + 1; 902 if (diff_format != D_EDIT) { 903 for (i0 = 1; i0 <= m; i0 = i1 + 1) { 904 while (i0 <= m && J[i0] == J[i0 - 1] + 1){ 905 if (diff_format == D_SIDEBYSIDE && 906 suppress_common != 1) { 907 nc = fetch(ixold, i0, i0, f1, '\0', 908 1, flags); 909 print_space(nc, 910 (hw - nc) + (padding << 1) + 1, 911 flags); 912 fetch(ixnew, J[i0], J[i0], f2, '\0', 913 0, flags); 914 diff_output("\n"); 915 } 916 i0++; 917 } 918 j0 = J[i0 - 1] + 1; 919 i1 = i0 - 1; 920 while (i1 < m && J[i1 + 1] == 0) 921 i1++; 922 j1 = J[i1 + 1] - 1; 923 J[i1] = j1; 924 925 /* 926 * When using side-by-side, lines from both of the 927 * files are printed. The algorithm used by diff(1) 928 * identifies the ranges in which two files differ. 929 * See the change() function below. 930 * The for loop below consumes the shorter range, 931 * whereas one of the while loops deals with the 932 * longer one. 933 */ 934 if (diff_format == D_SIDEBYSIDE) { 935 for (i=i0, j=j0; i<=i1 && j<=j1; i++, j++) 936 change(file1, f1, file2, f2, i, i, 937 j, j, &flags); 938 939 while (i <= i1) { 940 change(file1, f1, file2, f2, 941 i, i, j+1, j, &flags); 942 i++; 943 } 944 945 while (j <= j1) { 946 change(file1, f1, file2, f2, 947 i+1, i, j, j, &flags); 948 j++; 949 } 950 } else 951 change(file1, f1, file2, f2, i0, i1, j0, 952 j1, &flags); 953 } 954 } else { 955 for (i0 = m; i0 >= 1; i0 = i1 - 1) { 956 while (i0 >= 1 && J[i0] == J[i0 + 1] - 1 && J[i0] != 0) 957 i0--; 958 j0 = J[i0 + 1] - 1; 959 i1 = i0 + 1; 960 while (i1 > 1 && J[i1 - 1] == 0) 961 i1--; 962 j1 = J[i1 - 1] + 1; 963 J[i1] = j1; 964 change(file1, f1, file2, f2, i1, i0, j1, j0, &flags); 965 } 966 } 967 if (m == 0) 968 change(file1, f1, file2, f2, 1, 0, 1, len[1], &flags); 969 if (diff_format == D_IFDEF || diff_format == D_GFORMAT) { 970 for (;;) { 971 #define c i0 972 if ((c = getc(f1)) == EOF) 973 return; 974 diff_output("%c", c); 975 } 976 #undef c 977 } 978 if (anychange != 0) { 979 if (diff_format == D_CONTEXT) 980 dump_context_vec(f1, f2, flags); 981 else if (diff_format == D_UNIFIED) 982 dump_unified_vec(f1, f2, flags); 983 } 984 } 985 986 static void 987 range(int a, int b, const char *separator) 988 { 989 diff_output("%d", a > b ? b : a); 990 if (a < b) 991 diff_output("%s%d", separator, b); 992 } 993 994 static void 995 uni_range(int a, int b) 996 { 997 if (a < b) 998 diff_output("%d,%d", a, b - a + 1); 999 else if (a == b) 1000 diff_output("%d", b); 1001 else 1002 diff_output("%d,0", b); 1003 } 1004 1005 static char * 1006 preadline(int fd, size_t rlen, off_t off) 1007 { 1008 char *line; 1009 ssize_t nr; 1010 1011 line = xmalloc(rlen + 1); 1012 if ((nr = pread(fd, line, rlen, off)) == -1) 1013 err(2, "preadline"); 1014 if (nr > 0 && line[nr-1] == '\n') 1015 nr--; 1016 line[nr] = '\0'; 1017 return (line); 1018 } 1019 1020 static bool 1021 ignoreline_pattern(char *line) 1022 { 1023 int ret; 1024 1025 ret = regexec(&ignore_re, line, 0, NULL, 0); 1026 free(line); 1027 return (ret == 0); /* if it matched, it should be ignored. */ 1028 } 1029 1030 static bool 1031 ignoreline(char *line, bool skip_blanks) 1032 { 1033 1034 if (ignore_pats != NULL && skip_blanks) 1035 return (ignoreline_pattern(line) || *line == '\0'); 1036 if (ignore_pats != NULL) 1037 return (ignoreline_pattern(line)); 1038 if (skip_blanks) 1039 return (*line == '\0'); 1040 /* No ignore criteria specified */ 1041 return (false); 1042 } 1043 1044 /* 1045 * Indicate that there is a difference between lines a and b of the from file 1046 * to get to lines c to d of the to file. If a is greater then b then there 1047 * are no lines in the from file involved and this means that there were 1048 * lines appended (beginning at b). If c is greater than d then there are 1049 * lines missing from the to file. 1050 */ 1051 static void 1052 change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d, 1053 int *pflags) 1054 { 1055 static size_t max_context = 64; 1056 long curpos; 1057 int i, nc; 1058 const char *walk; 1059 bool skip_blanks; 1060 1061 skip_blanks = (*pflags & D_SKIPBLANKLINES); 1062 restart: 1063 if ((diff_format != D_IFDEF || diff_format == D_GFORMAT) && 1064 a > b && c > d) 1065 return; 1066 if (ignore_pats != NULL || skip_blanks) { 1067 char *line; 1068 /* 1069 * All lines in the change, insert, or delete must 1070 * match an ignore pattern for the change to be 1071 * ignored. 1072 */ 1073 if (a <= b) { /* Changes and deletes. */ 1074 for (i = a; i <= b; i++) { 1075 line = preadline(fileno(f1), 1076 ixold[i] - ixold[i - 1], ixold[i - 1]); 1077 if (!ignoreline(line, skip_blanks)) 1078 goto proceed; 1079 } 1080 } 1081 if (a > b || c <= d) { /* Changes and inserts. */ 1082 for (i = c; i <= d; i++) { 1083 line = preadline(fileno(f2), 1084 ixnew[i] - ixnew[i - 1], ixnew[i - 1]); 1085 if (!ignoreline(line, skip_blanks)) 1086 goto proceed; 1087 } 1088 } 1089 return; 1090 } 1091 proceed: 1092 if (*pflags & D_HEADER && diff_format != D_BRIEF) { 1093 diff_output("%s %s %s\n", diffargs, file1, file2); 1094 *pflags &= ~D_HEADER; 1095 } 1096 if (diff_format == D_CONTEXT || diff_format == D_UNIFIED) { 1097 /* 1098 * Allocate change records as needed. 1099 */ 1100 if (context_vec_ptr == context_vec_end - 1) { 1101 ptrdiff_t offset = context_vec_ptr - context_vec_start; 1102 max_context <<= 1; 1103 context_vec_start = xreallocarray(context_vec_start, 1104 max_context, sizeof(*context_vec_start)); 1105 context_vec_end = context_vec_start + max_context; 1106 context_vec_ptr = context_vec_start + offset; 1107 } 1108 if (anychange == 0) { 1109 /* 1110 * Print the context/unidiff header first time through. 1111 */ 1112 print_header(file1, file2); 1113 anychange = 1; 1114 } else if (a > context_vec_ptr->b + (2 * diff_context) + 1 && 1115 c > context_vec_ptr->d + (2 * diff_context) + 1) { 1116 /* 1117 * If this change is more than 'diff_context' lines from the 1118 * previous change, dump the record and reset it. 1119 */ 1120 if (diff_format == D_CONTEXT) 1121 dump_context_vec(f1, f2, *pflags); 1122 else 1123 dump_unified_vec(f1, f2, *pflags); 1124 } 1125 context_vec_ptr++; 1126 context_vec_ptr->a = a; 1127 context_vec_ptr->b = b; 1128 context_vec_ptr->c = c; 1129 context_vec_ptr->d = d; 1130 return; 1131 } 1132 if (anychange == 0) 1133 anychange = 1; 1134 switch (diff_format) { 1135 case D_BRIEF: 1136 return; 1137 case D_NORMAL: 1138 case D_EDIT: 1139 range(a, b, ","); 1140 diff_output("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1141 if (diff_format == D_NORMAL) 1142 range(c, d, ","); 1143 diff_output("\n"); 1144 break; 1145 case D_REVERSE: 1146 diff_output("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1147 range(a, b, " "); 1148 diff_output("\n"); 1149 break; 1150 case D_NREVERSE: 1151 if (a > b) 1152 diff_output("a%d %d\n", b, d - c + 1); 1153 else { 1154 diff_output("d%d %d\n", a, b - a + 1); 1155 if (!(c > d)) 1156 /* add changed lines */ 1157 diff_output("a%d %d\n", b, d - c + 1); 1158 } 1159 break; 1160 } 1161 if (diff_format == D_GFORMAT) { 1162 curpos = ftell(f1); 1163 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1164 nc = ixold[a > b ? b : a - 1] - curpos; 1165 for (i = 0; i < nc; i++) 1166 diff_output("%c", getc(f1)); 1167 for (walk = group_format; *walk != '\0'; walk++) { 1168 if (*walk == '%') { 1169 walk++; 1170 switch (*walk) { 1171 case '<': 1172 fetch(ixold, a, b, f1, '<', 1, *pflags); 1173 break; 1174 case '>': 1175 fetch(ixnew, c, d, f2, '>', 0, *pflags); 1176 break; 1177 default: 1178 diff_output("%%%c", *walk); 1179 break; 1180 } 1181 continue; 1182 } 1183 diff_output("%c", *walk); 1184 } 1185 } 1186 if (diff_format == D_SIDEBYSIDE) { 1187 if (a > b) { 1188 print_space(0, hw + padding , *pflags); 1189 } else { 1190 nc = fetch(ixold, a, b, f1, '\0', 1, *pflags); 1191 print_space(nc, hw - nc + padding, *pflags); 1192 } 1193 diff_output("%c", (a>b)? '>' : ((c>d)? '<' : '|')); 1194 print_space(hw + padding + 1 , padding, *pflags); 1195 fetch(ixnew, c, d, f2, '\0', 0, *pflags); 1196 diff_output("\n"); 1197 } 1198 if (diff_format == D_NORMAL || diff_format == D_IFDEF) { 1199 fetch(ixold, a, b, f1, '<', 1, *pflags); 1200 if (a <= b && c <= d && diff_format == D_NORMAL) 1201 diff_output("---\n"); 1202 } 1203 if (diff_format != D_GFORMAT && diff_format != D_SIDEBYSIDE) 1204 fetch(ixnew, c, d, f2, diff_format == D_NORMAL ? '>' : '\0', 0, *pflags); 1205 if (edoffset != 0 && diff_format == D_EDIT) { 1206 /* 1207 * A non-zero edoffset value for D_EDIT indicates that the 1208 * last line printed was a bare dot (".") that has been 1209 * escaped as ".." to prevent ed(1) from misinterpreting 1210 * it. We have to add a substitute command to change this 1211 * back and restart where we left off. 1212 */ 1213 diff_output(".\n"); 1214 diff_output("%ds/.//\n", a + edoffset - 1); 1215 b = a + edoffset - 1; 1216 a = b + 1; 1217 c += edoffset; 1218 goto restart; 1219 } 1220 if ((diff_format == D_EDIT || diff_format == D_REVERSE) && c <= d) 1221 diff_output(".\n"); 1222 if (inifdef) { 1223 diff_output("#endif /* %s */\n", ifdefname); 1224 inifdef = 0; 1225 } 1226 } 1227 1228 static int 1229 fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile, int flags) 1230 { 1231 int i, j, c, lastc, col, nc, newcol; 1232 1233 edoffset = 0; 1234 nc = 0; 1235 /* 1236 * When doing #ifdef's, copy down to current line 1237 * if this is the first file, so that stuff makes it to output. 1238 */ 1239 if ((diff_format == D_IFDEF) && oldfile) { 1240 long curpos = ftell(lb); 1241 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1242 nc = f[a > b ? b : a - 1] - curpos; 1243 for (i = 0; i < nc; i++) 1244 diff_output("%c", getc(lb)); 1245 } 1246 if (a > b) 1247 return (0); 1248 if (diff_format == D_IFDEF) { 1249 if (inifdef) { 1250 diff_output("#else /* %s%s */\n", 1251 oldfile == 1 ? "!" : "", ifdefname); 1252 } else { 1253 if (oldfile) 1254 diff_output("#ifndef %s\n", ifdefname); 1255 else 1256 diff_output("#ifdef %s\n", ifdefname); 1257 } 1258 inifdef = 1 + oldfile; 1259 } 1260 for (i = a; i <= b; i++) { 1261 fseek(lb, f[i - 1], SEEK_SET); 1262 nc = (f[i] - f[i - 1]); 1263 if (diff_format == D_SIDEBYSIDE && hw < nc) 1264 nc = hw; 1265 if ((diff_format != D_IFDEF && diff_format != D_GFORMAT) && 1266 ch != '\0') { 1267 diff_output("%c", ch); 1268 if (Tflag && (diff_format == D_NORMAL || 1269 diff_format == D_CONTEXT || 1270 diff_format == D_UNIFIED)) 1271 diff_output("\t"); 1272 else if (diff_format != D_UNIFIED) 1273 diff_output(" "); 1274 } 1275 col = 0; 1276 for (j = 0, lastc = '\0'; j < nc; j++, lastc = c) { 1277 if ((c = getc(lb)) == EOF) { 1278 if (diff_format == D_EDIT || 1279 diff_format == D_REVERSE || 1280 diff_format == D_NREVERSE) 1281 warnx("No newline at end of file"); 1282 else 1283 diff_output("\n\\ No newline at end of " 1284 "file\n"); 1285 return col; 1286 } 1287 /* 1288 * when using --side-by-side, col needs to be increased 1289 * in any case to keep the columns aligned 1290 */ 1291 if (c == '\t') { 1292 if (flags & D_EXPANDTABS) { 1293 newcol = ((col/tabsize)+1)*tabsize; 1294 do { 1295 if (diff_format == D_SIDEBYSIDE) 1296 j++; 1297 diff_output(" "); 1298 } while (++col < newcol && j < nc); 1299 } else { 1300 if (diff_format == D_SIDEBYSIDE) { 1301 if ((j + tabsize) > nc) { 1302 diff_output("%*s", 1303 nc - j,""); 1304 j = col = nc; 1305 } else { 1306 diff_output("\t"); 1307 col += tabsize - 1; 1308 j += tabsize - 1; 1309 } 1310 } else { 1311 diff_output("\t"); 1312 col++; 1313 } 1314 } 1315 } else { 1316 if (diff_format == D_EDIT && j == 1 && c == '\n' 1317 && lastc == '.') { 1318 /* 1319 * Don't print a bare "." line 1320 * since that will confuse ed(1). 1321 * Print ".." instead and set the, 1322 * global variable edoffset to an 1323 * offset from which to restart. 1324 * The caller must check the value 1325 * of edoffset 1326 */ 1327 diff_output(".\n"); 1328 edoffset = i - a + 1; 1329 return edoffset; 1330 } 1331 /* when side-by-side, do not print a newline */ 1332 if (diff_format != D_SIDEBYSIDE || c != '\n') { 1333 diff_output("%c", c); 1334 col++; 1335 } 1336 } 1337 } 1338 } 1339 return col; 1340 } 1341 1342 /* 1343 * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578. 1344 */ 1345 static int 1346 readhash(FILE *f, int flags) 1347 { 1348 int i, t, space; 1349 int sum; 1350 1351 sum = 1; 1352 space = 0; 1353 if ((flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) == 0) { 1354 if (flags & D_IGNORECASE) 1355 for (i = 0; (t = getc(f)) != '\n'; i++) { 1356 if (flags & D_STRIPCR && t == '\r') { 1357 t = getc(f); 1358 if (t == '\n') 1359 break; 1360 ungetc(t, f); 1361 } 1362 if (t == EOF) { 1363 if (i == 0) 1364 return (0); 1365 break; 1366 } 1367 sum = sum * 127 + chrtran(t); 1368 } 1369 else 1370 for (i = 0; (t = getc(f)) != '\n'; i++) { 1371 if (flags & D_STRIPCR && t == '\r') { 1372 t = getc(f); 1373 if (t == '\n') 1374 break; 1375 ungetc(t, f); 1376 } 1377 if (t == EOF) { 1378 if (i == 0) 1379 return (0); 1380 break; 1381 } 1382 sum = sum * 127 + t; 1383 } 1384 } else { 1385 for (i = 0;;) { 1386 switch (t = getc(f)) { 1387 case '\r': 1388 case '\t': 1389 case '\v': 1390 case '\f': 1391 case ' ': 1392 space++; 1393 continue; 1394 default: 1395 if (space && (flags & D_IGNOREBLANKS) == 0) { 1396 i++; 1397 space = 0; 1398 } 1399 sum = sum * 127 + chrtran(t); 1400 i++; 1401 continue; 1402 case EOF: 1403 if (i == 0) 1404 return (0); 1405 /* FALLTHROUGH */ 1406 case '\n': 1407 break; 1408 } 1409 break; 1410 } 1411 } 1412 /* 1413 * There is a remote possibility that we end up with a zero sum. 1414 * Zero is used as an EOF marker, so return 1 instead. 1415 */ 1416 return (sum == 0 ? 1 : sum); 1417 } 1418 1419 static int 1420 asciifile(FILE *f) 1421 { 1422 unsigned char buf[BUFSIZ]; 1423 size_t cnt; 1424 1425 if (f == NULL) 1426 return (1); 1427 1428 rewind(f); 1429 cnt = fread(buf, 1, sizeof(buf), f); 1430 return (memchr(buf, '\0', cnt) == NULL); 1431 } 1432 1433 #define begins_with(s, pre) (strncmp(s, pre, sizeof(pre)-1) == 0) 1434 1435 static char * 1436 match_function(const long *f, int pos, FILE *fp) 1437 { 1438 unsigned char buf[FUNCTION_CONTEXT_SIZE]; 1439 size_t nc; 1440 int last = lastline; 1441 const char *state = NULL; 1442 1443 lastline = pos; 1444 while (pos > last) { 1445 fseek(fp, f[pos - 1], SEEK_SET); 1446 nc = f[pos] - f[pos - 1]; 1447 if (nc >= sizeof(buf)) 1448 nc = sizeof(buf) - 1; 1449 nc = fread(buf, 1, nc, fp); 1450 if (nc > 0) { 1451 buf[nc] = '\0'; 1452 buf[strcspn(buf, "\n")] = '\0'; 1453 if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$') { 1454 if (begins_with(buf, "private:")) { 1455 if (!state) 1456 state = " (private)"; 1457 } else if (begins_with(buf, "protected:")) { 1458 if (!state) 1459 state = " (protected)"; 1460 } else if (begins_with(buf, "public:")) { 1461 if (!state) 1462 state = " (public)"; 1463 } else { 1464 strlcpy(lastbuf, buf, sizeof lastbuf); 1465 if (state) 1466 strlcat(lastbuf, state, 1467 sizeof lastbuf); 1468 lastmatchline = pos; 1469 return lastbuf; 1470 } 1471 } 1472 } 1473 pos--; 1474 } 1475 return lastmatchline > 0 ? lastbuf : NULL; 1476 } 1477 1478 /* dump accumulated "context" diff changes */ 1479 static void 1480 dump_context_vec(FILE *f1, FILE *f2, int flags) 1481 { 1482 struct context_vec *cvp = context_vec_start; 1483 int lowa, upb, lowc, upd, do_output; 1484 int a, b, c, d; 1485 char ch, *f; 1486 1487 if (context_vec_start > context_vec_ptr) 1488 return; 1489 1490 b = d = 0; /* gcc */ 1491 lowa = MAX(1, cvp->a - diff_context); 1492 upb = MIN(len[0], context_vec_ptr->b + diff_context); 1493 lowc = MAX(1, cvp->c - diff_context); 1494 upd = MIN(len[1], context_vec_ptr->d + diff_context); 1495 1496 diff_output("***************"); 1497 if ((flags & D_PROTOTYPE)) { 1498 f = match_function(ixold, lowa-1, f1); 1499 if (f != NULL) 1500 diff_output(" %s", f); 1501 } 1502 diff_output("\n*** "); 1503 range(lowa, upb, ","); 1504 diff_output(" ****\n"); 1505 1506 /* 1507 * Output changes to the "old" file. The first loop suppresses 1508 * output if there were no changes to the "old" file (we'll see 1509 * the "old" lines as context in the "new" list). 1510 */ 1511 do_output = 0; 1512 for (; cvp <= context_vec_ptr; cvp++) 1513 if (cvp->a <= cvp->b) { 1514 cvp = context_vec_start; 1515 do_output++; 1516 break; 1517 } 1518 if (do_output) { 1519 while (cvp <= context_vec_ptr) { 1520 a = cvp->a; 1521 b = cvp->b; 1522 c = cvp->c; 1523 d = cvp->d; 1524 1525 if (a <= b && c <= d) 1526 ch = 'c'; 1527 else 1528 ch = (a <= b) ? 'd' : 'a'; 1529 1530 if (ch == 'a') 1531 fetch(ixold, lowa, b, f1, ' ', 0, flags); 1532 else { 1533 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1534 fetch(ixold, a, b, f1, 1535 ch == 'c' ? '!' : '-', 0, flags); 1536 } 1537 lowa = b + 1; 1538 cvp++; 1539 } 1540 fetch(ixold, b + 1, upb, f1, ' ', 0, flags); 1541 } 1542 /* output changes to the "new" file */ 1543 diff_output("--- "); 1544 range(lowc, upd, ","); 1545 diff_output(" ----\n"); 1546 1547 do_output = 0; 1548 for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++) 1549 if (cvp->c <= cvp->d) { 1550 cvp = context_vec_start; 1551 do_output++; 1552 break; 1553 } 1554 if (do_output) { 1555 while (cvp <= context_vec_ptr) { 1556 a = cvp->a; 1557 b = cvp->b; 1558 c = cvp->c; 1559 d = cvp->d; 1560 1561 if (a <= b && c <= d) 1562 ch = 'c'; 1563 else 1564 ch = (a <= b) ? 'd' : 'a'; 1565 1566 if (ch == 'd') 1567 fetch(ixnew, lowc, d, f2, ' ', 0, flags); 1568 else { 1569 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1570 fetch(ixnew, c, d, f2, 1571 ch == 'c' ? '!' : '+', 0, flags); 1572 } 1573 lowc = d + 1; 1574 cvp++; 1575 } 1576 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1577 } 1578 context_vec_ptr = context_vec_start - 1; 1579 } 1580 1581 /* dump accumulated "unified" diff changes */ 1582 static void 1583 dump_unified_vec(FILE *f1, FILE *f2, int flags) 1584 { 1585 struct context_vec *cvp = context_vec_start; 1586 int lowa, upb, lowc, upd; 1587 int a, b, c, d; 1588 char ch, *f; 1589 1590 if (context_vec_start > context_vec_ptr) 1591 return; 1592 1593 b = d = 0; /* gcc */ 1594 lowa = MAX(1, cvp->a - diff_context); 1595 upb = MIN(len[0], context_vec_ptr->b + diff_context); 1596 lowc = MAX(1, cvp->c - diff_context); 1597 upd = MIN(len[1], context_vec_ptr->d + diff_context); 1598 1599 diff_output("@@ -"); 1600 uni_range(lowa, upb); 1601 diff_output(" +"); 1602 uni_range(lowc, upd); 1603 diff_output(" @@"); 1604 if ((flags & D_PROTOTYPE)) { 1605 f = match_function(ixold, lowa-1, f1); 1606 if (f != NULL) 1607 diff_output(" %s", f); 1608 } 1609 diff_output("\n"); 1610 1611 /* 1612 * Output changes in "unified" diff format--the old and new lines 1613 * are printed together. 1614 */ 1615 for (; cvp <= context_vec_ptr; cvp++) { 1616 a = cvp->a; 1617 b = cvp->b; 1618 c = cvp->c; 1619 d = cvp->d; 1620 1621 /* 1622 * c: both new and old changes 1623 * d: only changes in the old file 1624 * a: only changes in the new file 1625 */ 1626 if (a <= b && c <= d) 1627 ch = 'c'; 1628 else 1629 ch = (a <= b) ? 'd' : 'a'; 1630 1631 switch (ch) { 1632 case 'c': 1633 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1634 fetch(ixold, a, b, f1, '-', 0, flags); 1635 fetch(ixnew, c, d, f2, '+', 0, flags); 1636 break; 1637 case 'd': 1638 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1639 fetch(ixold, a, b, f1, '-', 0, flags); 1640 break; 1641 case 'a': 1642 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1643 fetch(ixnew, c, d, f2, '+', 0, flags); 1644 break; 1645 } 1646 lowa = b + 1; 1647 lowc = d + 1; 1648 } 1649 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1650 1651 context_vec_ptr = context_vec_start - 1; 1652 } 1653 1654 static void 1655 print_header(const char *file1, const char *file2) 1656 { 1657 const char *time_format; 1658 char buf1[256]; 1659 char buf2[256]; 1660 char end1[10]; 1661 char end2[10]; 1662 struct tm tm1, tm2, *tm_ptr1, *tm_ptr2; 1663 int nsec1 = stb1.st_mtim.tv_nsec; 1664 int nsec2 = stb2.st_mtim.tv_nsec; 1665 1666 time_format = "%Y-%m-%d %H:%M:%S"; 1667 1668 if (cflag) 1669 time_format = "%c"; 1670 tm_ptr1 = localtime_r(&stb1.st_mtime, &tm1); 1671 tm_ptr2 = localtime_r(&stb2.st_mtime, &tm2); 1672 strftime(buf1, 256, time_format, tm_ptr1); 1673 strftime(buf2, 256, time_format, tm_ptr2); 1674 if (!cflag) { 1675 strftime(end1, 10, "%z", tm_ptr1); 1676 strftime(end2, 10, "%z", tm_ptr2); 1677 sprintf(buf1, "%s.%.9d %s", buf1, nsec1, end1); 1678 sprintf(buf2, "%s.%.9d %s", buf2, nsec2, end2); 1679 } 1680 if (label[0] != NULL) 1681 diff_output("%s %s\n", diff_format == D_CONTEXT ? "***" : "---", 1682 label[0]); 1683 else 1684 diff_output("%s %s\t%s\n", diff_format == D_CONTEXT ? "***" : "---", 1685 file1, buf1); 1686 if (label[1] != NULL) 1687 diff_output("%s %s\n", diff_format == D_CONTEXT ? "---" : "+++", 1688 label[1]); 1689 else 1690 diff_output("%s %s\t%s\n", diff_format == D_CONTEXT ? "---" : "+++", 1691 file2, buf2); 1692 } 1693 1694 /* 1695 * Prints n number of space characters either by using tab 1696 * or single space characters. 1697 * nc is the preceding number of characters 1698 */ 1699 static void 1700 print_space(int nc, int n, int flags) { 1701 int i, col; 1702 1703 col = n; 1704 if ((flags & D_EXPANDTABS) == 0) { 1705 /* first tabstop may be closer than tabsize */ 1706 i = tabsize - (nc % tabsize); 1707 while (col >= tabsize) { 1708 diff_output("\t"); 1709 col -= i; 1710 i = tabsize; 1711 } 1712 } 1713 diff_output("%*s", col, ""); 1714 } 1715