1 /* $OpenBSD: diffreg.c,v 1.91 2016/03/01 20:57:35 natano 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/procdesc.h> 74 #include <sys/stat.h> 75 #include <sys/types.h> 76 #include <sys/event.h> 77 #include <sys/wait.h> 78 79 #include <capsicum_helpers.h> 80 #include <ctype.h> 81 #include <err.h> 82 #include <errno.h> 83 #include <fcntl.h> 84 #include <paths.h> 85 #include <regex.h> 86 #include <stddef.h> 87 #include <stdint.h> 88 #include <stdio.h> 89 #include <stdlib.h> 90 #include <string.h> 91 #include <unistd.h> 92 #include <limits.h> 93 #include <signal.h> 94 95 #include "diff.h" 96 #include "xmalloc.h" 97 98 #define _PATH_PR "/usr/bin/pr" 99 100 /* 101 * diff - compare two files. 102 */ 103 104 /* 105 * Uses an algorithm due to Harold Stone, which finds 106 * a pair of longest identical subsequences in the two 107 * files. 108 * 109 * The major goal is to generate the match vector J. 110 * J[i] is the index of the line in file1 corresponding 111 * to line i file0. J[i] = 0 if there is no 112 * such line in file1. 113 * 114 * Lines are hashed so as to work in core. All potential 115 * matches are located by sorting the lines of each file 116 * on the hash (called ``value''). In particular, this 117 * collects the equivalence classes in file1 together. 118 * Subroutine equiv replaces the value of each line in 119 * file0 by the index of the first element of its 120 * matching equivalence in (the reordered) file1. 121 * To save space equiv squeezes file1 into a single 122 * array member in which the equivalence classes 123 * are simply concatenated, except that their first 124 * members are flagged by changing sign. 125 * 126 * Next the indices that point into member are unsorted into 127 * array class according to the original order of file0. 128 * 129 * The cleverness lies in routine stone. This marches 130 * through the lines of file0, developing a vector klist 131 * of "k-candidates". At step i a k-candidate is a matched 132 * pair of lines x,y (x in file0 y in file1) such that 133 * there is a common subsequence of length k 134 * between the first i lines of file0 and the first y 135 * lines of file1, but there is no such subsequence for 136 * any smaller y. x is the earliest possible mate to y 137 * that occurs in such a subsequence. 138 * 139 * Whenever any of the members of the equivalence class of 140 * lines in file1 matable to a line in file0 has serial number 141 * less than the y of some k-candidate, that k-candidate 142 * with the smallest such y is replaced. The new 143 * k-candidate is chained (via pred) to the current 144 * k-1 candidate so that the actual subsequence can 145 * be recovered. When a member has serial number greater 146 * that the y of all k-candidates, the klist is extended. 147 * At the end, the longest subsequence is pulled out 148 * and placed in the array J by unravel 149 * 150 * With J in hand, the matches there recorded are 151 * check'ed against reality to assure that no spurious 152 * matches have crept in due to hashing. If they have, 153 * they are broken, and "jackpot" is recorded--a harmless 154 * matter except that a true match for a spuriously 155 * mated line may now be unnecessarily reported as a change. 156 * 157 * Much of the complexity of the program comes simply 158 * from trying to minimize core utilization and 159 * maximize the range of doable problems by dynamically 160 * allocating what is needed and reusing what is not. 161 * The core requirements for problems larger than somewhat 162 * are (in words) 2*length(file0) + length(file1) + 163 * 3*(number of k-candidates installed), typically about 164 * 6n words for files of length n. 165 */ 166 167 struct cand { 168 int x; 169 int y; 170 int pred; 171 }; 172 173 static struct line { 174 int serial; 175 int value; 176 } *file[2]; 177 178 /* 179 * The following struct is used to record change information when 180 * doing a "context" or "unified" diff. (see routine "change" to 181 * understand the highly mnemonic field names) 182 */ 183 struct context_vec { 184 int a; /* start line in old file */ 185 int b; /* end line in old file */ 186 int c; /* start line in new file */ 187 int d; /* end line in new file */ 188 }; 189 190 #define diff_output printf 191 static FILE *opentemp(const char *); 192 static void output(char *, FILE *, char *, FILE *, int); 193 static void check(FILE *, FILE *, int); 194 static void range(int, int, const char *); 195 static void uni_range(int, int); 196 static void dump_context_vec(FILE *, FILE *, int); 197 static void dump_unified_vec(FILE *, FILE *, int); 198 static void prepare(int, FILE *, size_t, int); 199 static void prune(void); 200 static void equiv(struct line *, int, struct line *, int, int *); 201 static void unravel(int); 202 static void unsort(struct line *, int, int *); 203 static void change(char *, FILE *, char *, FILE *, int, int, int, int, int *); 204 static void sort(struct line *, int); 205 static void print_header(const char *, const char *); 206 static int ignoreline(char *); 207 static int asciifile(FILE *); 208 static int fetch(long *, int, int, FILE *, int, int, int); 209 static int newcand(int, int, int); 210 static int search(int *, int, int); 211 static int skipline(FILE *); 212 static int isqrt(int); 213 static int stone(int *, int, int *, int *, int); 214 static int readhash(FILE *, int); 215 static int files_differ(FILE *, FILE *, int); 216 static char *match_function(const long *, int, FILE *); 217 static char *preadline(int, size_t, off_t); 218 219 static int *J; /* will be overlaid on class */ 220 static int *class; /* will be overlaid on file[0] */ 221 static int *klist; /* will be overlaid on file[0] after class */ 222 static int *member; /* will be overlaid on file[1] */ 223 static int clen; 224 static int inifdef; /* whether or not we are in a #ifdef block */ 225 static int len[2]; 226 static int pref, suff; /* length of prefix and suffix */ 227 static int slen[2]; 228 static int anychange; 229 static long *ixnew; /* will be overlaid on file[1] */ 230 static long *ixold; /* will be overlaid on klist */ 231 static struct cand *clist; /* merely a free storage pot for candidates */ 232 static int clistlen; /* the length of clist */ 233 static struct line *sfile[2]; /* shortened by pruning common prefix/suffix */ 234 static int (*chrtran)(int); /* translation table for case-folding */ 235 static struct context_vec *context_vec_start; 236 static struct context_vec *context_vec_end; 237 static struct context_vec *context_vec_ptr; 238 239 #define FUNCTION_CONTEXT_SIZE 55 240 static char lastbuf[FUNCTION_CONTEXT_SIZE]; 241 static int lastline; 242 static int lastmatchline; 243 244 static int 245 clow2low(int c) 246 { 247 248 return (c); 249 } 250 251 static int 252 cup2low(int c) 253 { 254 255 return tolower(c); 256 } 257 258 int 259 diffreg(char *file1, char *file2, int flags, int capsicum) 260 { 261 FILE *f1, *f2; 262 int i, rval; 263 int ostdout = -1; 264 int pr_pd, kq; 265 struct kevent *e; 266 cap_rights_t rights_ro; 267 268 e = NULL; 269 kq = -1; 270 f1 = f2 = NULL; 271 rval = D_SAME; 272 anychange = 0; 273 lastline = 0; 274 lastmatchline = 0; 275 context_vec_ptr = context_vec_start - 1; 276 if (flags & D_IGNORECASE) 277 chrtran = cup2low; 278 else 279 chrtran = clow2low; 280 if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode)) 281 return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2); 282 if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0) 283 goto closem; 284 285 if (flags & D_EMPTY1) 286 f1 = fopen(_PATH_DEVNULL, "r"); 287 else { 288 if (!S_ISREG(stb1.st_mode)) { 289 if ((f1 = opentemp(file1)) == NULL || 290 fstat(fileno(f1), &stb1) < 0) { 291 warn("%s", file1); 292 status |= 2; 293 goto closem; 294 } 295 } else if (strcmp(file1, "-") == 0) 296 f1 = stdin; 297 else 298 f1 = fopen(file1, "r"); 299 } 300 if (f1 == NULL) { 301 warn("%s", file1); 302 status |= 2; 303 goto closem; 304 } 305 306 if (flags & D_EMPTY2) 307 f2 = fopen(_PATH_DEVNULL, "r"); 308 else { 309 if (!S_ISREG(stb2.st_mode)) { 310 if ((f2 = opentemp(file2)) == NULL || 311 fstat(fileno(f2), &stb2) < 0) { 312 warn("%s", file2); 313 status |= 2; 314 goto closem; 315 } 316 } else if (strcmp(file2, "-") == 0) 317 f2 = stdin; 318 else 319 f2 = fopen(file2, "r"); 320 } 321 if (f2 == NULL) { 322 warn("%s", file2); 323 status |= 2; 324 goto closem; 325 } 326 327 if (lflag) { 328 /* redirect stdout to pr */ 329 int pfd[2]; 330 pid_t pid; 331 char *header; 332 333 xasprintf(&header, "%s %s %s", diffargs, file1, file2); 334 signal(SIGPIPE, SIG_IGN); 335 fflush(stdout); 336 rewind(stdout); 337 pipe(pfd); 338 switch ((pid = pdfork(&pr_pd, PD_CLOEXEC))) { 339 case -1: 340 status |= 2; 341 free(header); 342 err(2, "No more processes"); 343 case 0: 344 /* child */ 345 if (pfd[0] != STDIN_FILENO) { 346 dup2(pfd[0], STDIN_FILENO); 347 close(pfd[0]); 348 } 349 close(pfd[1]); 350 execl(_PATH_PR, _PATH_PR, "-h", header, (char *)0); 351 _exit(127); 352 default: 353 354 /* parent */ 355 if (pfd[1] != STDOUT_FILENO) { 356 ostdout = dup(STDOUT_FILENO); 357 dup2(pfd[1], STDOUT_FILENO); 358 close(pfd[1]); 359 } 360 close(pfd[0]); 361 rewind(stdout); 362 free(header); 363 kq = kqueue(); 364 if (kq == -1) 365 err(2, "kqueue"); 366 e = xmalloc(sizeof(struct kevent)); 367 EV_SET(e, pr_pd, EVFILT_PROCDESC, EV_ADD, NOTE_EXIT, 0, 368 NULL); 369 if (kevent(kq, e, 1, NULL, 0, NULL) == -1) 370 err(2, "kevent"); 371 } 372 } 373 374 if (capsicum) { 375 cap_rights_init(&rights_ro, CAP_READ, CAP_FSTAT, CAP_SEEK); 376 if (cap_rights_limit(fileno(f1), &rights_ro) < 0 377 && errno != ENOSYS) 378 err(2, "unable to limit rights on: %s", file1); 379 if (cap_rights_limit(fileno(f2), &rights_ro) < 0 && 380 errno != ENOSYS) 381 err(2, "unable to limit rights on: %s", file2); 382 if (fileno(f1) == STDIN_FILENO || fileno(f2) == STDIN_FILENO) { 383 /* stding has already been limited */ 384 if (caph_limit_stderr() == -1) 385 err(2, "unable to limit stderr"); 386 if (caph_limit_stdout() == -1) 387 err(2, "unable to limit stdout"); 388 } else if (caph_limit_stdio() == -1) 389 err(2, "unable to limit stdio"); 390 391 caph_cache_catpages(); 392 caph_cache_tzdata(); 393 if (cap_enter() < 0 && errno != ENOSYS) 394 err(2, "unable to enter capability mode"); 395 } 396 397 switch (files_differ(f1, f2, flags)) { 398 case 0: 399 goto closem; 400 case 1: 401 break; 402 default: 403 /* error */ 404 status |= 2; 405 goto closem; 406 } 407 408 if ((flags & D_FORCEASCII) == 0 && 409 (!asciifile(f1) || !asciifile(f2))) { 410 rval = D_BINARY; 411 status |= 1; 412 goto closem; 413 } 414 prepare(0, f1, stb1.st_size, flags); 415 prepare(1, f2, stb2.st_size, flags); 416 417 prune(); 418 sort(sfile[0], slen[0]); 419 sort(sfile[1], slen[1]); 420 421 member = (int *)file[1]; 422 equiv(sfile[0], slen[0], sfile[1], slen[1], member); 423 member = xreallocarray(member, slen[1] + 2, sizeof(*member)); 424 425 class = (int *)file[0]; 426 unsort(sfile[0], slen[0], class); 427 class = xreallocarray(class, slen[0] + 2, sizeof(*class)); 428 429 klist = xcalloc(slen[0] + 2, sizeof(*klist)); 430 clen = 0; 431 clistlen = 100; 432 clist = xcalloc(clistlen, sizeof(*clist)); 433 i = stone(class, slen[0], member, klist, flags); 434 free(member); 435 free(class); 436 437 J = xreallocarray(J, len[0] + 2, sizeof(*J)); 438 unravel(klist[i]); 439 free(clist); 440 free(klist); 441 442 ixold = xreallocarray(ixold, len[0] + 2, sizeof(*ixold)); 443 ixnew = xreallocarray(ixnew, len[1] + 2, sizeof(*ixnew)); 444 check(f1, f2, flags); 445 output(file1, f1, file2, f2, flags); 446 if (ostdout != -1 && e != NULL) { 447 /* close the pipe to pr and restore stdout */ 448 int wstatus; 449 450 fflush(stdout); 451 if (ostdout != STDOUT_FILENO) { 452 close(STDOUT_FILENO); 453 dup2(ostdout, STDOUT_FILENO); 454 close(ostdout); 455 } 456 if (kevent(kq, NULL, 0, e, 1, NULL) == -1) 457 err(2, "kevent"); 458 wstatus = e[0].data; 459 close(kq); 460 if (WIFEXITED(wstatus) && WEXITSTATUS(wstatus) != 0) 461 errx(2, "pr exited abnormally"); 462 else if (WIFSIGNALED(wstatus)) 463 errx(2, "pr killed by signal %d", 464 WTERMSIG(wstatus)); 465 } 466 467 closem: 468 if (anychange) { 469 status |= 1; 470 if (rval == D_SAME) 471 rval = D_DIFFER; 472 } 473 if (f1 != NULL) 474 fclose(f1); 475 if (f2 != NULL) 476 fclose(f2); 477 478 return (rval); 479 } 480 481 /* 482 * Check to see if the given files differ. 483 * Returns 0 if they are the same, 1 if different, and -1 on error. 484 * XXX - could use code from cmp(1) [faster] 485 */ 486 static int 487 files_differ(FILE *f1, FILE *f2, int flags) 488 { 489 char buf1[BUFSIZ], buf2[BUFSIZ]; 490 size_t i, j; 491 492 if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size || 493 (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT)) 494 return (1); 495 for (;;) { 496 i = fread(buf1, 1, sizeof(buf1), f1); 497 j = fread(buf2, 1, sizeof(buf2), f2); 498 if ((!i && ferror(f1)) || (!j && ferror(f2))) 499 return (-1); 500 if (i != j) 501 return (1); 502 if (i == 0) 503 return (0); 504 if (memcmp(buf1, buf2, i) != 0) 505 return (1); 506 } 507 } 508 509 static FILE * 510 opentemp(const char *f) 511 { 512 char buf[BUFSIZ], tempfile[PATH_MAX]; 513 ssize_t nread; 514 int ifd, ofd; 515 516 if (strcmp(f, "-") == 0) 517 ifd = STDIN_FILENO; 518 else if ((ifd = open(f, O_RDONLY, 0644)) < 0) 519 return (NULL); 520 521 (void)strlcpy(tempfile, _PATH_TMP "/diff.XXXXXXXX", sizeof(tempfile)); 522 523 if ((ofd = mkstemp(tempfile)) < 0) { 524 close(ifd); 525 return (NULL); 526 } 527 unlink(tempfile); 528 while ((nread = read(ifd, buf, BUFSIZ)) > 0) { 529 if (write(ofd, buf, nread) != nread) { 530 close(ifd); 531 close(ofd); 532 return (NULL); 533 } 534 } 535 close(ifd); 536 lseek(ofd, (off_t)0, SEEK_SET); 537 return (fdopen(ofd, "r")); 538 } 539 540 char * 541 splice(char *dir, char *path) 542 { 543 char *tail, *buf; 544 size_t dirlen; 545 546 dirlen = strlen(dir); 547 while (dirlen != 0 && dir[dirlen - 1] == '/') 548 dirlen--; 549 if ((tail = strrchr(path, '/')) == NULL) 550 tail = path; 551 else 552 tail++; 553 xasprintf(&buf, "%.*s/%s", (int)dirlen, dir, tail); 554 return (buf); 555 } 556 557 static void 558 prepare(int i, FILE *fd, size_t filesize, int flags) 559 { 560 struct line *p; 561 int h; 562 size_t sz, j; 563 564 rewind(fd); 565 566 sz = MIN(filesize, SIZE_MAX) / 25; 567 if (sz < 100) 568 sz = 100; 569 570 p = xcalloc(sz + 3, sizeof(*p)); 571 for (j = 0; (h = readhash(fd, flags));) { 572 if (j == sz) { 573 sz = sz * 3 / 2; 574 p = xreallocarray(p, sz + 3, sizeof(*p)); 575 } 576 p[++j].value = h; 577 } 578 len[i] = j; 579 file[i] = p; 580 } 581 582 static void 583 prune(void) 584 { 585 int i, j; 586 587 for (pref = 0; pref < len[0] && pref < len[1] && 588 file[0][pref + 1].value == file[1][pref + 1].value; 589 pref++) 590 ; 591 for (suff = 0; suff < len[0] - pref && suff < len[1] - pref && 592 file[0][len[0] - suff].value == file[1][len[1] - suff].value; 593 suff++) 594 ; 595 for (j = 0; j < 2; j++) { 596 sfile[j] = file[j] + pref; 597 slen[j] = len[j] - pref - suff; 598 for (i = 0; i <= slen[j]; i++) 599 sfile[j][i].serial = i; 600 } 601 } 602 603 static void 604 equiv(struct line *a, int n, struct line *b, int m, int *c) 605 { 606 int i, j; 607 608 i = j = 1; 609 while (i <= n && j <= m) { 610 if (a[i].value < b[j].value) 611 a[i++].value = 0; 612 else if (a[i].value == b[j].value) 613 a[i++].value = j; 614 else 615 j++; 616 } 617 while (i <= n) 618 a[i++].value = 0; 619 b[m + 1].value = 0; 620 j = 0; 621 while (++j <= m) { 622 c[j] = -b[j].serial; 623 while (b[j + 1].value == b[j].value) { 624 j++; 625 c[j] = b[j].serial; 626 } 627 } 628 c[j] = -1; 629 } 630 631 /* Code taken from ping.c */ 632 static int 633 isqrt(int n) 634 { 635 int y, x = 1; 636 637 if (n == 0) 638 return (0); 639 640 do { /* newton was a stinker */ 641 y = x; 642 x = n / x; 643 x += y; 644 x /= 2; 645 } while ((x - y) > 1 || (x - y) < -1); 646 647 return (x); 648 } 649 650 static int 651 stone(int *a, int n, int *b, int *c, int flags) 652 { 653 int i, k, y, j, l; 654 int oldc, tc, oldl, sq; 655 u_int numtries, bound; 656 657 if (flags & D_MINIMAL) 658 bound = UINT_MAX; 659 else { 660 sq = isqrt(n); 661 bound = MAX(256, sq); 662 } 663 664 k = 0; 665 c[0] = newcand(0, 0, 0); 666 for (i = 1; i <= n; i++) { 667 j = a[i]; 668 if (j == 0) 669 continue; 670 y = -b[j]; 671 oldl = 0; 672 oldc = c[0]; 673 numtries = 0; 674 do { 675 if (y <= clist[oldc].y) 676 continue; 677 l = search(c, k, y); 678 if (l != oldl + 1) 679 oldc = c[l - 1]; 680 if (l <= k) { 681 if (clist[c[l]].y <= y) 682 continue; 683 tc = c[l]; 684 c[l] = newcand(i, y, oldc); 685 oldc = tc; 686 oldl = l; 687 numtries++; 688 } else { 689 c[l] = newcand(i, y, oldc); 690 k++; 691 break; 692 } 693 } while ((y = b[++j]) > 0 && numtries < bound); 694 } 695 return (k); 696 } 697 698 static int 699 newcand(int x, int y, int pred) 700 { 701 struct cand *q; 702 703 if (clen == clistlen) { 704 clistlen = clistlen * 11 / 10; 705 clist = xreallocarray(clist, clistlen, sizeof(*clist)); 706 } 707 q = clist + clen; 708 q->x = x; 709 q->y = y; 710 q->pred = pred; 711 return (clen++); 712 } 713 714 static int 715 search(int *c, int k, int y) 716 { 717 int i, j, l, t; 718 719 if (clist[c[k]].y < y) /* quick look for typical case */ 720 return (k + 1); 721 i = 0; 722 j = k + 1; 723 for (;;) { 724 l = (i + j) / 2; 725 if (l <= i) 726 break; 727 t = clist[c[l]].y; 728 if (t > y) 729 j = l; 730 else if (t < y) 731 i = l; 732 else 733 return (l); 734 } 735 return (l + 1); 736 } 737 738 static void 739 unravel(int p) 740 { 741 struct cand *q; 742 int i; 743 744 for (i = 0; i <= len[0]; i++) 745 J[i] = i <= pref ? i : 746 i > len[0] - suff ? i + len[1] - len[0] : 0; 747 for (q = clist + p; q->y != 0; q = clist + q->pred) 748 J[q->x + pref] = q->y + pref; 749 } 750 751 /* 752 * Check does double duty: 753 * 1. ferret out any fortuitous correspondences due 754 * to confounding by hashing (which result in "jackpot") 755 * 2. collect random access indexes to the two files 756 */ 757 static void 758 check(FILE *f1, FILE *f2, int flags) 759 { 760 int i, j, jackpot, c, d; 761 long ctold, ctnew; 762 763 rewind(f1); 764 rewind(f2); 765 j = 1; 766 ixold[0] = ixnew[0] = 0; 767 jackpot = 0; 768 ctold = ctnew = 0; 769 for (i = 1; i <= len[0]; i++) { 770 if (J[i] == 0) { 771 ixold[i] = ctold += skipline(f1); 772 continue; 773 } 774 while (j < J[i]) { 775 ixnew[j] = ctnew += skipline(f2); 776 j++; 777 } 778 if (flags & (D_FOLDBLANKS|D_IGNOREBLANKS|D_IGNORECASE|D_STRIPCR)) { 779 for (;;) { 780 c = getc(f1); 781 d = getc(f2); 782 /* 783 * GNU diff ignores a missing newline 784 * in one file for -b or -w. 785 */ 786 if (flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) { 787 if (c == EOF && d == '\n') { 788 ctnew++; 789 break; 790 } else if (c == '\n' && d == EOF) { 791 ctold++; 792 break; 793 } 794 } 795 ctold++; 796 ctnew++; 797 if (flags & D_STRIPCR) { 798 if (c == '\r') { 799 if ((c = getc(f1)) == '\n') { 800 ctnew++; 801 break; 802 } 803 } 804 if (d == '\r') { 805 if ((d = getc(f2)) == '\n') { 806 ctold++; 807 break; 808 } 809 } 810 } 811 if ((flags & D_FOLDBLANKS) && isspace(c) && 812 isspace(d)) { 813 do { 814 if (c == '\n') 815 break; 816 ctold++; 817 } while (isspace(c = getc(f1))); 818 do { 819 if (d == '\n') 820 break; 821 ctnew++; 822 } while (isspace(d = getc(f2))); 823 } else if ((flags & D_IGNOREBLANKS)) { 824 while (isspace(c) && c != '\n') { 825 c = getc(f1); 826 ctold++; 827 } 828 while (isspace(d) && d != '\n') { 829 d = getc(f2); 830 ctnew++; 831 } 832 } 833 if (chrtran(c) != chrtran(d)) { 834 jackpot++; 835 J[i] = 0; 836 if (c != '\n' && c != EOF) 837 ctold += skipline(f1); 838 if (d != '\n' && c != EOF) 839 ctnew += skipline(f2); 840 break; 841 } 842 if (c == '\n' || c == EOF) 843 break; 844 } 845 } else { 846 for (;;) { 847 ctold++; 848 ctnew++; 849 if ((c = getc(f1)) != (d = getc(f2))) { 850 /* jackpot++; */ 851 J[i] = 0; 852 if (c != '\n' && c != EOF) 853 ctold += skipline(f1); 854 if (d != '\n' && c != EOF) 855 ctnew += skipline(f2); 856 break; 857 } 858 if (c == '\n' || c == EOF) 859 break; 860 } 861 } 862 ixold[i] = ctold; 863 ixnew[j] = ctnew; 864 j++; 865 } 866 for (; j <= len[1]; j++) { 867 ixnew[j] = ctnew += skipline(f2); 868 } 869 /* 870 * if (jackpot) 871 * fprintf(stderr, "jackpot\n"); 872 */ 873 } 874 875 /* shellsort CACM #201 */ 876 static void 877 sort(struct line *a, int n) 878 { 879 struct line *ai, *aim, w; 880 int j, m = 0, k; 881 882 if (n == 0) 883 return; 884 for (j = 1; j <= n; j *= 2) 885 m = 2 * j - 1; 886 for (m /= 2; m != 0; m /= 2) { 887 k = n - m; 888 for (j = 1; j <= k; j++) { 889 for (ai = &a[j]; ai > a; ai -= m) { 890 aim = &ai[m]; 891 if (aim < ai) 892 break; /* wraparound */ 893 if (aim->value > ai[0].value || 894 (aim->value == ai[0].value && 895 aim->serial > ai[0].serial)) 896 break; 897 w.value = ai[0].value; 898 ai[0].value = aim->value; 899 aim->value = w.value; 900 w.serial = ai[0].serial; 901 ai[0].serial = aim->serial; 902 aim->serial = w.serial; 903 } 904 } 905 } 906 } 907 908 static void 909 unsort(struct line *f, int l, int *b) 910 { 911 int *a, i; 912 913 a = xcalloc(l + 1, sizeof(*a)); 914 for (i = 1; i <= l; i++) 915 a[f[i].serial] = f[i].value; 916 for (i = 1; i <= l; i++) 917 b[i] = a[i]; 918 free(a); 919 } 920 921 static int 922 skipline(FILE *f) 923 { 924 int i, c; 925 926 for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++) 927 continue; 928 return (i); 929 } 930 931 static void 932 output(char *file1, FILE *f1, char *file2, FILE *f2, int flags) 933 { 934 int m, i0, i1, j0, j1; 935 936 rewind(f1); 937 rewind(f2); 938 m = len[0]; 939 J[0] = 0; 940 J[m + 1] = len[1] + 1; 941 if (diff_format != D_EDIT) { 942 for (i0 = 1; i0 <= m; i0 = i1 + 1) { 943 while (i0 <= m && J[i0] == J[i0 - 1] + 1) 944 i0++; 945 j0 = J[i0 - 1] + 1; 946 i1 = i0 - 1; 947 while (i1 < m && J[i1 + 1] == 0) 948 i1++; 949 j1 = J[i1 + 1] - 1; 950 J[i1] = j1; 951 change(file1, f1, file2, f2, i0, i1, j0, j1, &flags); 952 } 953 } else { 954 for (i0 = m; i0 >= 1; i0 = i1 - 1) { 955 while (i0 >= 1 && J[i0] == J[i0 + 1] - 1 && J[i0] != 0) 956 i0--; 957 j0 = J[i0 + 1] - 1; 958 i1 = i0 + 1; 959 while (i1 > 1 && J[i1 - 1] == 0) 960 i1--; 961 j1 = J[i1 - 1] + 1; 962 J[i1] = j1; 963 change(file1, f1, file2, f2, i1, i0, j1, j0, &flags); 964 } 965 } 966 if (m == 0) 967 change(file1, f1, file2, f2, 1, 0, 1, len[1], &flags); 968 if (diff_format == D_IFDEF || diff_format == D_GFORMAT) { 969 for (;;) { 970 #define c i0 971 if ((c = getc(f1)) == EOF) 972 return; 973 diff_output("%c", c); 974 } 975 #undef c 976 } 977 if (anychange != 0) { 978 if (diff_format == D_CONTEXT) 979 dump_context_vec(f1, f2, flags); 980 else if (diff_format == D_UNIFIED) 981 dump_unified_vec(f1, f2, flags); 982 } 983 } 984 985 static void 986 range(int a, int b, const char *separator) 987 { 988 diff_output("%d", a > b ? b : a); 989 if (a < b) 990 diff_output("%s%d", separator, b); 991 } 992 993 static void 994 uni_range(int a, int b) 995 { 996 if (a < b) 997 diff_output("%d,%d", a, b - a + 1); 998 else if (a == b) 999 diff_output("%d", b); 1000 else 1001 diff_output("%d,0", b); 1002 } 1003 1004 static char * 1005 preadline(int fd, size_t rlen, off_t off) 1006 { 1007 char *line; 1008 ssize_t nr; 1009 1010 line = xmalloc(rlen + 1); 1011 if ((nr = pread(fd, line, rlen, off)) < 0) 1012 err(2, "preadline"); 1013 if (nr > 0 && line[nr-1] == '\n') 1014 nr--; 1015 line[nr] = '\0'; 1016 return (line); 1017 } 1018 1019 static int 1020 ignoreline(char *line) 1021 { 1022 int ret; 1023 1024 ret = regexec(&ignore_re, line, 0, NULL, 0); 1025 free(line); 1026 return (ret == 0); /* if it matched, it should be ignored. */ 1027 } 1028 1029 /* 1030 * Indicate that there is a difference between lines a and b of the from file 1031 * to get to lines c to d of the to file. If a is greater then b then there 1032 * are no lines in the from file involved and this means that there were 1033 * lines appended (beginning at b). If c is greater than d then there are 1034 * lines missing from the to file. 1035 */ 1036 static void 1037 change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d, 1038 int *pflags) 1039 { 1040 static size_t max_context = 64; 1041 long curpos; 1042 int i, nc, f; 1043 const char *walk; 1044 1045 restart: 1046 if ((diff_format != D_IFDEF || diff_format == D_GFORMAT) && 1047 a > b && c > d) 1048 return; 1049 if (ignore_pats != NULL) { 1050 char *line; 1051 /* 1052 * All lines in the change, insert, or delete must 1053 * match an ignore pattern for the change to be 1054 * ignored. 1055 */ 1056 if (a <= b) { /* Changes and deletes. */ 1057 for (i = a; i <= b; i++) { 1058 line = preadline(fileno(f1), 1059 ixold[i] - ixold[i - 1], ixold[i - 1]); 1060 if (!ignoreline(line)) 1061 goto proceed; 1062 } 1063 } 1064 if (a > b || c <= d) { /* Changes and inserts. */ 1065 for (i = c; i <= d; i++) { 1066 line = preadline(fileno(f2), 1067 ixnew[i] - ixnew[i - 1], ixnew[i - 1]); 1068 if (!ignoreline(line)) 1069 goto proceed; 1070 } 1071 } 1072 return; 1073 } 1074 proceed: 1075 if (*pflags & D_HEADER && diff_format != D_BRIEF) { 1076 diff_output("%s %s %s\n", diffargs, file1, file2); 1077 *pflags &= ~D_HEADER; 1078 } 1079 if (diff_format == D_CONTEXT || diff_format == D_UNIFIED) { 1080 /* 1081 * Allocate change records as needed. 1082 */ 1083 if (context_vec_ptr == context_vec_end - 1) { 1084 ptrdiff_t offset = context_vec_ptr - context_vec_start; 1085 max_context <<= 1; 1086 context_vec_start = xreallocarray(context_vec_start, 1087 max_context, sizeof(*context_vec_start)); 1088 context_vec_end = context_vec_start + max_context; 1089 context_vec_ptr = context_vec_start + offset; 1090 } 1091 if (anychange == 0) { 1092 /* 1093 * Print the context/unidiff header first time through. 1094 */ 1095 print_header(file1, file2); 1096 anychange = 1; 1097 } else if (a > context_vec_ptr->b + (2 * diff_context) + 1 && 1098 c > context_vec_ptr->d + (2 * diff_context) + 1) { 1099 /* 1100 * If this change is more than 'diff_context' lines from the 1101 * previous change, dump the record and reset it. 1102 */ 1103 if (diff_format == D_CONTEXT) 1104 dump_context_vec(f1, f2, *pflags); 1105 else 1106 dump_unified_vec(f1, f2, *pflags); 1107 } 1108 context_vec_ptr++; 1109 context_vec_ptr->a = a; 1110 context_vec_ptr->b = b; 1111 context_vec_ptr->c = c; 1112 context_vec_ptr->d = d; 1113 return; 1114 } 1115 if (anychange == 0) 1116 anychange = 1; 1117 switch (diff_format) { 1118 case D_BRIEF: 1119 return; 1120 case D_NORMAL: 1121 case D_EDIT: 1122 range(a, b, ","); 1123 diff_output("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1124 if (diff_format == D_NORMAL) 1125 range(c, d, ","); 1126 diff_output("\n"); 1127 break; 1128 case D_REVERSE: 1129 diff_output("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1130 range(a, b, " "); 1131 diff_output("\n"); 1132 break; 1133 case D_NREVERSE: 1134 if (a > b) 1135 diff_output("a%d %d\n", b, d - c + 1); 1136 else { 1137 diff_output("d%d %d\n", a, b - a + 1); 1138 if (!(c > d)) 1139 /* add changed lines */ 1140 diff_output("a%d %d\n", b, d - c + 1); 1141 } 1142 break; 1143 } 1144 if (diff_format == D_GFORMAT) { 1145 curpos = ftell(f1); 1146 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1147 nc = ixold[a > b ? b : a - 1] - curpos; 1148 for (i = 0; i < nc; i++) 1149 diff_output("%c", getc(f1)); 1150 for (walk = group_format; *walk != '\0'; walk++) { 1151 if (*walk == '%') { 1152 walk++; 1153 switch (*walk) { 1154 case '<': 1155 fetch(ixold, a, b, f1, '<', 1, *pflags); 1156 break; 1157 case '>': 1158 fetch(ixnew, c, d, f2, '>', 0, *pflags); 1159 break; 1160 default: 1161 diff_output("%%%c", *walk); 1162 break; 1163 } 1164 continue; 1165 } 1166 diff_output("%c", *walk); 1167 } 1168 } 1169 if (diff_format == D_NORMAL || diff_format == D_IFDEF) { 1170 fetch(ixold, a, b, f1, '<', 1, *pflags); 1171 if (a <= b && c <= d && diff_format == D_NORMAL) 1172 diff_output("---\n"); 1173 } 1174 f = 0; 1175 if (diff_format != D_GFORMAT) 1176 f = fetch(ixnew, c, d, f2, diff_format == D_NORMAL ? '>' : '\0', 0, *pflags); 1177 if (f != 0 && diff_format == D_EDIT) { 1178 /* 1179 * A non-zero return value for D_EDIT indicates that the 1180 * last line printed was a bare dot (".") that has been 1181 * escaped as ".." to prevent ed(1) from misinterpreting 1182 * it. We have to add a substitute command to change this 1183 * back and restart where we left off. 1184 */ 1185 diff_output(".\n"); 1186 diff_output("%ds/.//\n", a + f - 1); 1187 b = a + f - 1; 1188 a = b + 1; 1189 c += f; 1190 goto restart; 1191 } 1192 if ((diff_format == D_EDIT || diff_format == D_REVERSE) && c <= d) 1193 diff_output(".\n"); 1194 if (inifdef) { 1195 diff_output("#endif /* %s */\n", ifdefname); 1196 inifdef = 0; 1197 } 1198 } 1199 1200 static int 1201 fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile, int flags) 1202 { 1203 int i, j, c, lastc, col, nc; 1204 int newcol; 1205 1206 /* 1207 * When doing #ifdef's, copy down to current line 1208 * if this is the first file, so that stuff makes it to output. 1209 */ 1210 if ((diff_format == D_IFDEF) && oldfile) { 1211 long curpos = ftell(lb); 1212 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1213 nc = f[a > b ? b : a - 1] - curpos; 1214 for (i = 0; i < nc; i++) 1215 diff_output("%c", getc(lb)); 1216 } 1217 if (a > b) 1218 return (0); 1219 if (diff_format == D_IFDEF) { 1220 if (inifdef) { 1221 diff_output("#else /* %s%s */\n", 1222 oldfile == 1 ? "!" : "", ifdefname); 1223 } else { 1224 if (oldfile) 1225 diff_output("#ifndef %s\n", ifdefname); 1226 else 1227 diff_output("#ifdef %s\n", ifdefname); 1228 } 1229 inifdef = 1 + oldfile; 1230 } 1231 for (i = a; i <= b; i++) { 1232 fseek(lb, f[i - 1], SEEK_SET); 1233 nc = f[i] - f[i - 1]; 1234 if ((diff_format != D_IFDEF && diff_format != D_GFORMAT) && 1235 ch != '\0') { 1236 diff_output("%c", ch); 1237 if (Tflag && (diff_format == D_NORMAL || diff_format == D_CONTEXT 1238 || diff_format == D_UNIFIED)) 1239 diff_output("\t"); 1240 else if (diff_format != D_UNIFIED) 1241 diff_output(" "); 1242 } 1243 col = 0; 1244 for (j = 0, lastc = '\0'; j < nc; j++, lastc = c) { 1245 if ((c = getc(lb)) == EOF) { 1246 if (diff_format == D_EDIT || diff_format == D_REVERSE || 1247 diff_format == D_NREVERSE) 1248 warnx("No newline at end of file"); 1249 else 1250 diff_output("\n\\ No newline at end of " 1251 "file\n"); 1252 return (0); 1253 } 1254 if (c == '\t' && (flags & D_EXPANDTABS)) { 1255 newcol = ((col/tabsize)+1)*tabsize; 1256 do { 1257 diff_output(" "); 1258 } while (++col < newcol); 1259 } else { 1260 if (diff_format == D_EDIT && j == 1 && c == '\n' 1261 && lastc == '.') { 1262 /* 1263 * Don't print a bare "." line 1264 * since that will confuse ed(1). 1265 * Print ".." instead and return, 1266 * giving the caller an offset 1267 * from which to restart. 1268 */ 1269 diff_output(".\n"); 1270 return (i - a + 1); 1271 } 1272 diff_output("%c", c); 1273 col++; 1274 } 1275 } 1276 } 1277 return (0); 1278 } 1279 1280 /* 1281 * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578. 1282 */ 1283 static int 1284 readhash(FILE *f, int flags) 1285 { 1286 int i, t, space; 1287 int sum; 1288 1289 sum = 1; 1290 space = 0; 1291 if ((flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) == 0) { 1292 if (flags & D_IGNORECASE) 1293 for (i = 0; (t = getc(f)) != '\n'; i++) { 1294 if (flags & D_STRIPCR && t == '\r') { 1295 t = getc(f); 1296 if (t == '\n') 1297 break; 1298 ungetc(t, f); 1299 } 1300 if (t == EOF) { 1301 if (i == 0) 1302 return (0); 1303 break; 1304 } 1305 sum = sum * 127 + chrtran(t); 1306 } 1307 else 1308 for (i = 0; (t = getc(f)) != '\n'; i++) { 1309 if (flags & D_STRIPCR && t == '\r') { 1310 t = getc(f); 1311 if (t == '\n') 1312 break; 1313 ungetc(t, f); 1314 } 1315 if (t == EOF) { 1316 if (i == 0) 1317 return (0); 1318 break; 1319 } 1320 sum = sum * 127 + t; 1321 } 1322 } else { 1323 for (i = 0;;) { 1324 switch (t = getc(f)) { 1325 case '\r': 1326 case '\t': 1327 case '\v': 1328 case '\f': 1329 case ' ': 1330 space++; 1331 continue; 1332 default: 1333 if (space && (flags & D_IGNOREBLANKS) == 0) { 1334 i++; 1335 space = 0; 1336 } 1337 sum = sum * 127 + chrtran(t); 1338 i++; 1339 continue; 1340 case EOF: 1341 if (i == 0) 1342 return (0); 1343 /* FALLTHROUGH */ 1344 case '\n': 1345 break; 1346 } 1347 break; 1348 } 1349 } 1350 /* 1351 * There is a remote possibility that we end up with a zero sum. 1352 * Zero is used as an EOF marker, so return 1 instead. 1353 */ 1354 return (sum == 0 ? 1 : sum); 1355 } 1356 1357 static int 1358 asciifile(FILE *f) 1359 { 1360 unsigned char buf[BUFSIZ]; 1361 size_t cnt; 1362 1363 if (f == NULL) 1364 return (1); 1365 1366 rewind(f); 1367 cnt = fread(buf, 1, sizeof(buf), f); 1368 return (memchr(buf, '\0', cnt) == NULL); 1369 } 1370 1371 #define begins_with(s, pre) (strncmp(s, pre, sizeof(pre)-1) == 0) 1372 1373 static char * 1374 match_function(const long *f, int pos, FILE *fp) 1375 { 1376 unsigned char buf[FUNCTION_CONTEXT_SIZE]; 1377 size_t nc; 1378 int last = lastline; 1379 const char *state = NULL; 1380 1381 lastline = pos; 1382 while (pos > last) { 1383 fseek(fp, f[pos - 1], SEEK_SET); 1384 nc = f[pos] - f[pos - 1]; 1385 if (nc >= sizeof(buf)) 1386 nc = sizeof(buf) - 1; 1387 nc = fread(buf, 1, nc, fp); 1388 if (nc > 0) { 1389 buf[nc] = '\0'; 1390 buf[strcspn(buf, "\n")] = '\0'; 1391 if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$') { 1392 if (begins_with(buf, "private:")) { 1393 if (!state) 1394 state = " (private)"; 1395 } else if (begins_with(buf, "protected:")) { 1396 if (!state) 1397 state = " (protected)"; 1398 } else if (begins_with(buf, "public:")) { 1399 if (!state) 1400 state = " (public)"; 1401 } else { 1402 strlcpy(lastbuf, buf, sizeof lastbuf); 1403 if (state) 1404 strlcat(lastbuf, state, 1405 sizeof lastbuf); 1406 lastmatchline = pos; 1407 return lastbuf; 1408 } 1409 } 1410 } 1411 pos--; 1412 } 1413 return lastmatchline > 0 ? lastbuf : NULL; 1414 } 1415 1416 /* dump accumulated "context" diff changes */ 1417 static void 1418 dump_context_vec(FILE *f1, FILE *f2, int flags) 1419 { 1420 struct context_vec *cvp = context_vec_start; 1421 int lowa, upb, lowc, upd, do_output; 1422 int a, b, c, d; 1423 char ch, *f; 1424 1425 if (context_vec_start > context_vec_ptr) 1426 return; 1427 1428 b = d = 0; /* gcc */ 1429 lowa = MAX(1, cvp->a - diff_context); 1430 upb = MIN(len[0], context_vec_ptr->b + diff_context); 1431 lowc = MAX(1, cvp->c - diff_context); 1432 upd = MIN(len[1], context_vec_ptr->d + diff_context); 1433 1434 diff_output("***************"); 1435 if ((flags & D_PROTOTYPE)) { 1436 f = match_function(ixold, lowa-1, f1); 1437 if (f != NULL) 1438 diff_output(" %s", f); 1439 } 1440 diff_output("\n*** "); 1441 range(lowa, upb, ","); 1442 diff_output(" ****\n"); 1443 1444 /* 1445 * Output changes to the "old" file. The first loop suppresses 1446 * output if there were no changes to the "old" file (we'll see 1447 * the "old" lines as context in the "new" list). 1448 */ 1449 do_output = 0; 1450 for (; cvp <= context_vec_ptr; cvp++) 1451 if (cvp->a <= cvp->b) { 1452 cvp = context_vec_start; 1453 do_output++; 1454 break; 1455 } 1456 if (do_output) { 1457 while (cvp <= context_vec_ptr) { 1458 a = cvp->a; 1459 b = cvp->b; 1460 c = cvp->c; 1461 d = cvp->d; 1462 1463 if (a <= b && c <= d) 1464 ch = 'c'; 1465 else 1466 ch = (a <= b) ? 'd' : 'a'; 1467 1468 if (ch == 'a') 1469 fetch(ixold, lowa, b, f1, ' ', 0, flags); 1470 else { 1471 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1472 fetch(ixold, a, b, f1, 1473 ch == 'c' ? '!' : '-', 0, flags); 1474 } 1475 lowa = b + 1; 1476 cvp++; 1477 } 1478 fetch(ixold, b + 1, upb, f1, ' ', 0, flags); 1479 } 1480 /* output changes to the "new" file */ 1481 diff_output("--- "); 1482 range(lowc, upd, ","); 1483 diff_output(" ----\n"); 1484 1485 do_output = 0; 1486 for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++) 1487 if (cvp->c <= cvp->d) { 1488 cvp = context_vec_start; 1489 do_output++; 1490 break; 1491 } 1492 if (do_output) { 1493 while (cvp <= context_vec_ptr) { 1494 a = cvp->a; 1495 b = cvp->b; 1496 c = cvp->c; 1497 d = cvp->d; 1498 1499 if (a <= b && c <= d) 1500 ch = 'c'; 1501 else 1502 ch = (a <= b) ? 'd' : 'a'; 1503 1504 if (ch == 'd') 1505 fetch(ixnew, lowc, d, f2, ' ', 0, flags); 1506 else { 1507 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1508 fetch(ixnew, c, d, f2, 1509 ch == 'c' ? '!' : '+', 0, flags); 1510 } 1511 lowc = d + 1; 1512 cvp++; 1513 } 1514 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1515 } 1516 context_vec_ptr = context_vec_start - 1; 1517 } 1518 1519 /* dump accumulated "unified" diff changes */ 1520 static void 1521 dump_unified_vec(FILE *f1, FILE *f2, int flags) 1522 { 1523 struct context_vec *cvp = context_vec_start; 1524 int lowa, upb, lowc, upd; 1525 int a, b, c, d; 1526 char ch, *f; 1527 1528 if (context_vec_start > context_vec_ptr) 1529 return; 1530 1531 b = d = 0; /* gcc */ 1532 lowa = MAX(1, cvp->a - diff_context); 1533 upb = MIN(len[0], context_vec_ptr->b + diff_context); 1534 lowc = MAX(1, cvp->c - diff_context); 1535 upd = MIN(len[1], context_vec_ptr->d + diff_context); 1536 1537 diff_output("@@ -"); 1538 uni_range(lowa, upb); 1539 diff_output(" +"); 1540 uni_range(lowc, upd); 1541 diff_output(" @@"); 1542 if ((flags & D_PROTOTYPE)) { 1543 f = match_function(ixold, lowa-1, f1); 1544 if (f != NULL) 1545 diff_output(" %s", f); 1546 } 1547 diff_output("\n"); 1548 1549 /* 1550 * Output changes in "unified" diff format--the old and new lines 1551 * are printed together. 1552 */ 1553 for (; cvp <= context_vec_ptr; cvp++) { 1554 a = cvp->a; 1555 b = cvp->b; 1556 c = cvp->c; 1557 d = cvp->d; 1558 1559 /* 1560 * c: both new and old changes 1561 * d: only changes in the old file 1562 * a: only changes in the new file 1563 */ 1564 if (a <= b && c <= d) 1565 ch = 'c'; 1566 else 1567 ch = (a <= b) ? 'd' : 'a'; 1568 1569 switch (ch) { 1570 case 'c': 1571 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1572 fetch(ixold, a, b, f1, '-', 0, flags); 1573 fetch(ixnew, c, d, f2, '+', 0, flags); 1574 break; 1575 case 'd': 1576 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1577 fetch(ixold, a, b, f1, '-', 0, flags); 1578 break; 1579 case 'a': 1580 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1581 fetch(ixnew, c, d, f2, '+', 0, flags); 1582 break; 1583 } 1584 lowa = b + 1; 1585 lowc = d + 1; 1586 } 1587 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1588 1589 context_vec_ptr = context_vec_start - 1; 1590 } 1591 1592 static void 1593 print_header(const char *file1, const char *file2) 1594 { 1595 const char *time_format; 1596 char buf1[256]; 1597 char buf2[256]; 1598 char end1[10]; 1599 char end2[10]; 1600 struct tm tm1, tm2, *tm_ptr1, *tm_ptr2; 1601 int nsec1 = stb1.st_mtim.tv_nsec; 1602 int nsec2 = stb2.st_mtim.tv_nsec; 1603 1604 time_format = "%Y-%m-%d %H:%M:%S"; 1605 1606 if (cflag) 1607 time_format = "%c"; 1608 tm_ptr1 = localtime_r(&stb1.st_mtime, &tm1); 1609 tm_ptr2 = localtime_r(&stb2.st_mtime, &tm2); 1610 strftime(buf1, 256, time_format, tm_ptr1); 1611 strftime(buf2, 256, time_format, tm_ptr2); 1612 if (!cflag) { 1613 strftime(end1, 10, "%z", tm_ptr1); 1614 strftime(end2, 10, "%z", tm_ptr2); 1615 sprintf(buf1, "%s.%.9d %s", buf1, nsec1, end1); 1616 sprintf(buf2, "%s.%.9d %s", buf2, nsec2, end2); 1617 } 1618 if (label[0] != NULL) 1619 diff_output("%s %s\n", diff_format == D_CONTEXT ? "***" : "---", 1620 label[0]); 1621 else 1622 diff_output("%s %s\t%s\n", diff_format == D_CONTEXT ? "***" : "---", 1623 file1, buf1); 1624 if (label[1] != NULL) 1625 diff_output("%s %s\n", diff_format == D_CONTEXT ? "---" : "+++", 1626 label[1]); 1627 else 1628 diff_output("%s %s\t%s\n", diff_format == D_CONTEXT ? "---" : "+++", 1629 file2, buf2); 1630 } 1631