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