1 /* $OpenBSD: diffreg.c,v 1.93 2019/06/28 13:35:00 deraadt Exp $ */ 2 3 /*- 4 * SPDX-License-Identifier: BSD-4-Clause 5 * 6 * Copyright (C) Caldera International Inc. 2001-2002. 7 * All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code and documentation must retain the above 13 * copyright notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. All advertising materials mentioning features or use of this software 18 * must display the following acknowledgement: 19 * This product includes software developed or owned by Caldera 20 * International, Inc. 21 * 4. Neither the name of Caldera International, Inc. nor the names of other 22 * contributors may be used to endorse or promote products derived from 23 * this software without specific prior written permission. 24 * 25 * USE OF THE SOFTWARE PROVIDED FOR UNDER THIS LICENSE BY CALDERA 26 * INTERNATIONAL, INC. AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR 27 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 28 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 29 * IN NO EVENT SHALL CALDERA INTERNATIONAL, INC. BE LIABLE FOR ANY DIRECT, 30 * INDIRECT INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 31 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 32 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, 34 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING 35 * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 36 * POSSIBILITY OF SUCH DAMAGE. 37 */ 38 /*- 39 * Copyright (c) 1991, 1993 40 * The Regents of the University of California. All rights reserved. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 3. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 */ 66 67 #include <sys/capsicum.h> 68 #include <sys/stat.h> 69 70 #include <capsicum_helpers.h> 71 #include <ctype.h> 72 #include <err.h> 73 #include <errno.h> 74 #include <fcntl.h> 75 #include <limits.h> 76 #include <math.h> 77 #include <paths.h> 78 #include <regex.h> 79 #include <stdbool.h> 80 #include <stdckdint.h> 81 #include <stddef.h> 82 #include <stdint.h> 83 #include <stdio.h> 84 #include <stdlib.h> 85 #include <string.h> 86 87 #include "pr.h" 88 #include "diff.h" 89 #include "xmalloc.h" 90 91 /* 92 * diff - compare two files. 93 */ 94 95 /* 96 * Uses an algorithm due to Harold Stone, which finds a pair of longest 97 * identical subsequences in the two files. 98 * 99 * The major goal is to generate the match vector J. J[i] is the index of 100 * the line in file1 corresponding to line i file0. J[i] = 0 if there is no 101 * such line in file1. 102 * 103 * Lines are hashed so as to work in core. All potential matches are 104 * located by sorting the lines of each file on the hash (called 105 * ``value''). In particular, this collects the equivalence classes in 106 * file1 together. Subroutine equiv replaces the value of each line in 107 * file0 by the index of the first element of its matching equivalence in 108 * (the reordered) file1. To save space equiv squeezes file1 into a single 109 * array member in which the equivalence classes are simply concatenated, 110 * except that their first members are flagged by changing sign. 111 * 112 * Next the indices that point into member are unsorted into array class 113 * according to the original order of file0. 114 * 115 * The cleverness lies in routine stone. This marches through the lines of 116 * file0, developing a vector klist of "k-candidates". At step i 117 * a k-candidate is a matched pair of lines x,y (x in file0 y in file1) 118 * such that there is a common subsequence of length k between the first 119 * i lines of file0 and the first y lines of file1, but there is no such 120 * subsequence for any smaller y. x is the earliest possible mate to y that 121 * occurs in such a subsequence. 122 * 123 * Whenever any of the members of the equivalence class of lines in file1 124 * matable to a line in file0 has serial number less than the y of some 125 * k-candidate, that k-candidate with the smallest such y is replaced. The 126 * new k-candidate is chained (via pred) to the current k-1 candidate so 127 * that the actual subsequence can be recovered. When a member has serial 128 * number greater that the y of all k-candidates, the klist is extended. At 129 * the end, the longest subsequence is pulled out and placed in the array J 130 * by unravel. 131 * 132 * With J in hand, the matches there recorded are check'ed against reality 133 * to assure that no spurious matches have crept in due to hashing. If they 134 * have, they are broken, and "jackpot" is recorded -- a harmless matter 135 * except that a true match for a spuriously mated line may now be 136 * unnecessarily reported as a change. 137 * 138 * Much of the complexity of the program comes simply from trying to 139 * minimize core utilization and maximize the range of doable problems by 140 * dynamically allocating what is needed and reusing what is not. The core 141 * requirements for problems larger than somewhat are (in words) 142 * 2*length(file0) + length(file1) + 3*(number of k-candidates installed), 143 * typically about 6n words for files of length n. 144 */ 145 146 struct cand { 147 int x; 148 int y; 149 int pred; 150 }; 151 152 static struct line { 153 int serial; 154 int value; 155 } *file[2]; 156 157 /* 158 * The following struct is used to record change information when 159 * doing a "context" or "unified" diff. (see routine "change" to 160 * understand the highly mnemonic field names) 161 */ 162 struct context_vec { 163 int a; /* start line in old file */ 164 int b; /* end line in old file */ 165 int c; /* start line in new file */ 166 int d; /* end line in new file */ 167 }; 168 169 enum readhash { RH_BINARY, RH_OK, RH_EOF }; 170 171 static int diffreg_stone(char *, char *, int, int); 172 static FILE *opentemp(const char *); 173 static void output(char *, FILE *, char *, FILE *, int); 174 static void check(FILE *, FILE *, int); 175 static void range(int, int, const char *); 176 static void uni_range(int, int); 177 static void dump_context_vec(FILE *, FILE *, int); 178 static void dump_unified_vec(FILE *, FILE *, int); 179 static bool prepare(int, FILE *, size_t, int); 180 static void prune(void); 181 static void equiv(struct line *, int, struct line *, int, int *); 182 static void unravel(int); 183 static void unsort(struct line *, int, int *); 184 static void change(char *, FILE *, char *, FILE *, int, int, int, int, int *); 185 static void sort(struct line *, int); 186 static void print_header(const char *, const char *); 187 static void print_space(int, int, int); 188 static bool ignoreline_pattern(char *); 189 static bool ignoreline(char *, bool); 190 static int asciifile(FILE *); 191 static int fetch(long *, int, int, FILE *, int, int, int); 192 static int newcand(int, int, int); 193 static int search(int *, int, int); 194 static int skipline(FILE *); 195 static int stone(int *, int, int *, int *, int); 196 static enum readhash readhash(FILE *, int, unsigned *); 197 static int files_differ(FILE *, FILE *, int); 198 static char *match_function(const long *, int, FILE *); 199 static char *preadline(int, size_t, off_t); 200 201 static int *J; /* will be overlaid on class */ 202 static int *class; /* will be overlaid on file[0] */ 203 static int *klist; /* will be overlaid on file[0] after class */ 204 static int *member; /* will be overlaid on file[1] */ 205 static int clen; 206 static int inifdef; /* whether or not we are in a #ifdef block */ 207 static size_t len[2]; /* lengths of files in lines */ 208 static size_t pref, suff; /* lengths of prefix and suffix */ 209 static size_t slen[2]; /* lengths of files minus pref / suff */ 210 static int anychange; 211 static int hw, lpad,rpad; /* half width and padding */ 212 static int edoffset; 213 static long *ixnew; /* will be overlaid on file[1] */ 214 static long *ixold; /* will be overlaid on klist */ 215 static struct cand *clist; /* merely a free storage pot for candidates */ 216 static int clistlen; /* the length of clist */ 217 static struct line *sfile[2]; /* shortened by pruning common prefix/suffix */ 218 static int (*chrtran)(int); /* translation table for case-folding */ 219 static struct context_vec *context_vec_start; 220 static struct context_vec *context_vec_end; 221 static struct context_vec *context_vec_ptr; 222 223 #define FUNCTION_CONTEXT_SIZE 55 224 static char lastbuf[FUNCTION_CONTEXT_SIZE]; 225 static int lastline; 226 static int lastmatchline; 227 228 int 229 diffreg(char *file1, char *file2, int flags, int capsicum) 230 { 231 /* 232 * If we have set the algorithm with -A or --algorithm use that if we 233 * can and if not print an error. 234 */ 235 if (diff_algorithm_set) { 236 if (diff_algorithm == D_DIFFMYERS || 237 diff_algorithm == D_DIFFPATIENCE) { 238 if (can_libdiff(flags)) 239 return diffreg_new(file1, file2, flags, capsicum); 240 else 241 errx(2, "cannot use Myers algorithm with selected options"); 242 } else { 243 /* Fallback to using stone. */ 244 return diffreg_stone(file1, file2, flags, capsicum); 245 } 246 } else { 247 if (can_libdiff(flags)) 248 return diffreg_new(file1, file2, flags, capsicum); 249 else 250 return diffreg_stone(file1, file2, flags, capsicum); 251 } 252 } 253 254 static int 255 clow2low(int c) 256 { 257 258 return (c); 259 } 260 261 static int 262 cup2low(int c) 263 { 264 265 return (tolower(c)); 266 } 267 268 int 269 diffreg_stone(char *file1, char *file2, int flags, int capsicum) 270 { 271 FILE *f1, *f2; 272 int i, rval; 273 struct pr *pr = NULL; 274 cap_rights_t rights_ro; 275 276 f1 = f2 = NULL; 277 rval = D_SAME; 278 anychange = 0; 279 lastline = 0; 280 lastmatchline = 0; 281 282 /* 283 * In side-by-side mode, we need to print the left column, a 284 * change marker surrounded by padding, and the right column. 285 * 286 * If expanding tabs, we don't care about alignment, so we simply 287 * subtract 3 from the width and divide by two. 288 * 289 * If not expanding tabs, we need to ensure that the right column 290 * is aligned to a tab stop. We start with the same formula, then 291 * decrement until we reach a size that lets us tab-align the 292 * right column. We then adjust the width down if necessary for 293 * the padding calculation to work. 294 * 295 * Left padding is half the space left over, rounded down; right 296 * padding is whatever is needed to match the width. 297 */ 298 if (diff_format == D_SIDEBYSIDE) { 299 if (flags & D_EXPANDTABS) { 300 if (width > 3) { 301 hw = (width - 3) / 2; 302 } else { 303 /* not enough space */ 304 hw = 0; 305 } 306 } else if (width <= 3 || width <= tabsize) { 307 /* not enough space */ 308 hw = 0; 309 } else { 310 hw = (width - 3) / 2; 311 while (hw > 0 && roundup(hw + 3, tabsize) + hw > width) 312 hw--; 313 if (width - (roundup(hw + 3, tabsize) + hw) < tabsize) 314 width = roundup(hw + 3, tabsize) + hw; 315 } 316 lpad = (width - hw * 2 - 1) / 2; 317 rpad = (width - hw * 2 - 1) - lpad; 318 } 319 320 if (flags & D_IGNORECASE) 321 chrtran = cup2low; 322 else 323 chrtran = clow2low; 324 if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode)) 325 return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2); 326 if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0) 327 goto closem; 328 329 if (flags & D_EMPTY1) 330 f1 = fopen(_PATH_DEVNULL, "r"); 331 else { 332 if (!S_ISREG(stb1.st_mode)) { 333 if ((f1 = opentemp(file1)) == NULL || 334 fstat(fileno(f1), &stb1) == -1) { 335 warn("%s", file1); 336 rval = D_ERROR; 337 status |= 2; 338 goto closem; 339 } 340 } else if (strcmp(file1, "-") == 0) 341 f1 = stdin; 342 else 343 f1 = fopen(file1, "r"); 344 } 345 if (f1 == NULL) { 346 warn("%s", file1); 347 rval = D_ERROR; 348 status |= 2; 349 goto closem; 350 } 351 352 if (flags & D_EMPTY2) 353 f2 = fopen(_PATH_DEVNULL, "r"); 354 else { 355 if (!S_ISREG(stb2.st_mode)) { 356 if ((f2 = opentemp(file2)) == NULL || 357 fstat(fileno(f2), &stb2) == -1) { 358 warn("%s", file2); 359 rval = D_ERROR; 360 status |= 2; 361 goto closem; 362 } 363 } else if (strcmp(file2, "-") == 0) 364 f2 = stdin; 365 else 366 f2 = fopen(file2, "r"); 367 } 368 if (f2 == NULL) { 369 warn("%s", file2); 370 rval = D_ERROR; 371 status |= 2; 372 goto closem; 373 } 374 375 if (stb1.st_dev == stb2.st_dev && stb1.st_ino == stb2.st_ino) 376 goto closem; 377 378 if (lflag) 379 pr = start_pr(file1, file2); 380 381 if (capsicum) { 382 cap_rights_init(&rights_ro, CAP_READ, CAP_FSTAT, CAP_SEEK); 383 if (caph_rights_limit(fileno(f1), &rights_ro) < 0) 384 err(2, "unable to limit rights on: %s", file1); 385 if (caph_rights_limit(fileno(f2), &rights_ro) < 0) 386 err(2, "unable to limit rights on: %s", file2); 387 if (fileno(f1) == STDIN_FILENO || fileno(f2) == STDIN_FILENO) { 388 /* stdin has already been limited */ 389 if (caph_limit_stderr() == -1) 390 err(2, "unable to limit stderr"); 391 if (caph_limit_stdout() == -1) 392 err(2, "unable to limit stdout"); 393 } else if (caph_limit_stdio() == -1) 394 err(2, "unable to limit stdio"); 395 396 caph_cache_catpages(); 397 caph_cache_tzdata(); 398 if (caph_enter() < 0) 399 err(2, "unable to enter capability mode"); 400 } 401 402 switch (files_differ(f1, f2, flags)) { 403 case 0: 404 goto closem; 405 case 1: 406 break; 407 default: 408 /* error */ 409 if (ferror(f1)) 410 warn("%s", file1); 411 if (ferror(f2)) 412 warn("%s", file2); 413 rval = D_ERROR; 414 status |= 2; 415 goto closem; 416 } 417 418 if ((diff_format == D_BRIEF || diff_format == D_QUIET) && 419 ignore_pats == NULL && 420 (flags & (D_FOLDBLANKS|D_IGNOREBLANKS|D_IGNORECASE| 421 D_SKIPBLANKLINES|D_STRIPCR)) == 0) 422 { 423 rval = D_DIFFER; 424 status |= 1; 425 goto closem; 426 } 427 if ((flags & D_FORCEASCII) != 0) { 428 (void)prepare(0, f1, stb1.st_size, flags); 429 (void)prepare(1, f2, stb2.st_size, flags); 430 } else if (!asciifile(f1) || !asciifile(f2) || 431 !prepare(0, f1, stb1.st_size, flags) || 432 !prepare(1, f2, stb2.st_size, flags)) { 433 rval = D_BINARY; 434 status |= 1; 435 goto closem; 436 } 437 if (len[0] > INT_MAX - 2) 438 errc(1, EFBIG, "%s", file1); 439 if (len[1] > INT_MAX - 2) 440 errc(1, EFBIG, "%s", file2); 441 442 prune(); 443 sort(sfile[0], slen[0]); 444 sort(sfile[1], slen[1]); 445 446 member = (int *)file[1]; 447 equiv(sfile[0], slen[0], sfile[1], slen[1], member); 448 member = xreallocarray(member, slen[1] + 2, sizeof(*member)); 449 450 class = (int *)file[0]; 451 unsort(sfile[0], slen[0], class); 452 class = xreallocarray(class, slen[0] + 2, sizeof(*class)); 453 454 klist = xcalloc(slen[0] + 2, sizeof(*klist)); 455 clen = 0; 456 clistlen = 100; 457 clist = xcalloc(clistlen, sizeof(*clist)); 458 i = stone(class, slen[0], member, klist, flags); 459 free(member); 460 free(class); 461 462 J = xreallocarray(J, len[0] + 2, sizeof(*J)); 463 unravel(klist[i]); 464 free(clist); 465 free(klist); 466 467 ixold = xreallocarray(ixold, len[0] + 2, sizeof(*ixold)); 468 ixnew = xreallocarray(ixnew, len[1] + 2, sizeof(*ixnew)); 469 check(f1, f2, flags); 470 output(file1, f1, file2, f2, flags); 471 472 closem: 473 if (pr != NULL) 474 stop_pr(pr); 475 if (anychange) { 476 status |= 1; 477 if (rval == D_SAME) 478 rval = D_DIFFER; 479 } 480 if (f1 != NULL) 481 fclose(f1); 482 if (f2 != NULL) 483 fclose(f2); 484 485 return (rval); 486 } 487 488 /* 489 * Check to see if the given files differ. 490 * Returns 0 if they are the same, 1 if different, and -1 on error. 491 * XXX - could use code from cmp(1) [faster] 492 */ 493 static int 494 files_differ(FILE *f1, FILE *f2, int flags) 495 { 496 char buf1[BUFSIZ], buf2[BUFSIZ]; 497 size_t i, j; 498 499 if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size || 500 (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT)) 501 return (1); 502 503 if (stb1.st_dev == stb2.st_dev && stb1.st_ino == stb2.st_ino) 504 return (0); 505 506 for (;;) { 507 if ((i = fread(buf1, 1, sizeof(buf1), f1)) == 0 && ferror(f1)) 508 return (-1); 509 if ((j = fread(buf2, 1, sizeof(buf2), f2)) == 0 && ferror(f2)) 510 return (-1); 511 if (i != j) 512 return (1); 513 if (i == 0) 514 return (0); 515 if (memcmp(buf1, buf2, i) != 0) 516 return (1); 517 } 518 } 519 520 static FILE * 521 opentemp(const char *f) 522 { 523 char buf[BUFSIZ], tempfile[PATH_MAX]; 524 ssize_t nread; 525 int ifd, ofd; 526 527 if (strcmp(f, "-") == 0) 528 ifd = STDIN_FILENO; 529 else if ((ifd = open(f, O_RDONLY, 0644)) == -1) 530 return (NULL); 531 532 (void)strlcpy(tempfile, _PATH_TMP "/diff.XXXXXXXX", sizeof(tempfile)); 533 534 if ((ofd = mkstemp(tempfile)) == -1) { 535 close(ifd); 536 return (NULL); 537 } 538 unlink(tempfile); 539 while ((nread = read(ifd, buf, BUFSIZ)) > 0) { 540 if (write(ofd, buf, nread) != nread) { 541 close(ifd); 542 close(ofd); 543 return (NULL); 544 } 545 } 546 if (nread == -1) { 547 close(ifd); 548 close(ofd); 549 return (NULL); 550 } 551 close(ifd); 552 lseek(ofd, (off_t)0, SEEK_SET); 553 return (fdopen(ofd, "r")); 554 } 555 556 static bool 557 prepare(int i, FILE *fd, size_t filesize, int flags) 558 { 559 struct line *p; 560 unsigned h; 561 size_t sz, j = 0; 562 enum readhash r; 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 while ((r = readhash(fd, flags, &h)) != RH_EOF) { 572 if (r == RH_BINARY) 573 return (false); 574 if (j == SIZE_MAX) 575 break; 576 if (j == sz) { 577 sz = sz * 3 / 2; 578 p = xreallocarray(p, sz + 3, sizeof(*p)); 579 } 580 p[++j].value = h; 581 } 582 583 len[i] = j; 584 file[i] = p; 585 586 return (true); 587 } 588 589 static void 590 prune(void) 591 { 592 size_t i, j; 593 594 for (pref = 0; pref < len[0] && pref < len[1] && 595 file[0][pref + 1].value == file[1][pref + 1].value; 596 pref++) 597 ; 598 for (suff = 0; suff < len[0] - pref && suff < len[1] - pref && 599 file[0][len[0] - suff].value == file[1][len[1] - suff].value; 600 suff++) 601 ; 602 for (j = 0; j < 2; j++) { 603 sfile[j] = file[j] + pref; 604 slen[j] = len[j] - pref - suff; 605 for (i = 0; i <= slen[j]; i++) 606 sfile[j][i].serial = i; 607 } 608 } 609 610 static void 611 equiv(struct line *a, int n, struct line *b, int m, int *c) 612 { 613 int i, j; 614 615 i = j = 1; 616 while (i <= n && j <= m) { 617 if (a[i].value < b[j].value) 618 a[i++].value = 0; 619 else if (a[i].value == b[j].value) 620 a[i++].value = j; 621 else 622 j++; 623 } 624 while (i <= n) 625 a[i++].value = 0; 626 b[m + 1].value = 0; 627 j = 0; 628 while (++j <= m) { 629 c[j] = -b[j].serial; 630 while (b[j + 1].value == b[j].value) { 631 j++; 632 c[j] = b[j].serial; 633 } 634 } 635 c[j] = -1; 636 } 637 638 static int 639 stone(int *a, int n, int *b, int *c, int flags) 640 { 641 int i, k, y, j, l; 642 int oldc, tc, oldl, sq; 643 unsigned numtries, bound; 644 645 if (flags & D_MINIMAL) 646 bound = UINT_MAX; 647 else { 648 sq = sqrt(n); 649 bound = MAX(256, sq); 650 } 651 652 k = 0; 653 c[0] = newcand(0, 0, 0); 654 for (i = 1; i <= n; i++) { 655 j = a[i]; 656 if (j == 0) 657 continue; 658 y = -b[j]; 659 oldl = 0; 660 oldc = c[0]; 661 numtries = 0; 662 do { 663 if (y <= clist[oldc].y) 664 continue; 665 l = search(c, k, y); 666 if (l != oldl + 1) 667 oldc = c[l - 1]; 668 if (l <= k) { 669 if (clist[c[l]].y <= y) 670 continue; 671 tc = c[l]; 672 c[l] = newcand(i, y, oldc); 673 oldc = tc; 674 oldl = l; 675 numtries++; 676 } else { 677 c[l] = newcand(i, y, oldc); 678 k++; 679 break; 680 } 681 } while ((y = b[++j]) > 0 && numtries < bound); 682 } 683 return (k); 684 } 685 686 static int 687 newcand(int x, int y, int pred) 688 { 689 struct cand *q; 690 691 if (clen == clistlen) { 692 clistlen = clistlen * 11 / 10; 693 clist = xreallocarray(clist, clistlen, sizeof(*clist)); 694 } 695 q = clist + clen; 696 q->x = x; 697 q->y = y; 698 q->pred = pred; 699 return (clen++); 700 } 701 702 static int 703 search(int *c, int k, int y) 704 { 705 int i, j, l, t; 706 707 if (clist[c[k]].y < y) /* quick look for typical case */ 708 return (k + 1); 709 i = 0; 710 j = k + 1; 711 for (;;) { 712 l = (i + j) / 2; 713 if (l <= i) 714 break; 715 t = clist[c[l]].y; 716 if (t > y) 717 j = l; 718 else if (t < y) 719 i = l; 720 else 721 return (l); 722 } 723 return (l + 1); 724 } 725 726 static void 727 unravel(int p) 728 { 729 struct cand *q; 730 size_t i; 731 732 for (i = 0; i <= len[0]; i++) 733 J[i] = i <= pref ? i : 734 i > len[0] - suff ? i + len[1] - len[0] : 0; 735 for (q = clist + p; q->y != 0; q = clist + q->pred) 736 J[q->x + pref] = q->y + pref; 737 } 738 739 /* 740 * Check does double duty: 741 * 1. ferret out any fortuitous correspondences due to confounding by 742 * hashing (which result in "jackpot") 743 * 2. collect random access indexes to the two files 744 */ 745 static void 746 check(FILE *f1, FILE *f2, int flags) 747 { 748 int i, j, /* jackpot, */ c, d; 749 long ctold, ctnew; 750 751 rewind(f1); 752 rewind(f2); 753 j = 1; 754 ixold[0] = ixnew[0] = 0; 755 /* jackpot = 0; */ 756 ctold = ctnew = 0; 757 for (i = 1; i <= (int)len[0]; i++) { 758 if (J[i] == 0) { 759 ixold[i] = ctold += skipline(f1); 760 continue; 761 } 762 while (j < J[i]) { 763 ixnew[j] = ctnew += skipline(f2); 764 j++; 765 } 766 if (flags & (D_FOLDBLANKS | D_IGNOREBLANKS | D_IGNORECASE | D_STRIPCR)) { 767 for (;;) { 768 c = getc(f1); 769 d = getc(f2); 770 /* 771 * GNU diff ignores a missing newline 772 * in one file for -b or -w. 773 */ 774 if (flags & (D_FOLDBLANKS | D_IGNOREBLANKS)) { 775 if (c == EOF && isspace(d)) { 776 ctnew++; 777 break; 778 } else if (isspace(c) && d == EOF) { 779 ctold++; 780 break; 781 } 782 } 783 ctold++; 784 ctnew++; 785 if (flags & D_STRIPCR && (c == '\r' || d == '\r')) { 786 if (c == '\r') { 787 if ((c = getc(f1)) == '\n') { 788 ctold++; 789 } else { 790 ungetc(c, f1); 791 } 792 } 793 if (d == '\r') { 794 if ((d = getc(f2)) == '\n') { 795 ctnew++; 796 } else { 797 ungetc(d, f2); 798 } 799 } 800 break; 801 } 802 if ((flags & D_FOLDBLANKS) && isspace(c) && 803 isspace(d)) { 804 do { 805 if (c == '\n') 806 break; 807 ctold++; 808 } while (isspace(c = getc(f1))); 809 do { 810 if (d == '\n') 811 break; 812 ctnew++; 813 } while (isspace(d = getc(f2))); 814 } else if (flags & D_IGNOREBLANKS) { 815 while (isspace(c) && c != '\n') { 816 c = getc(f1); 817 ctold++; 818 } 819 while (isspace(d) && d != '\n') { 820 d = getc(f2); 821 ctnew++; 822 } 823 } 824 if (chrtran(c) != chrtran(d)) { 825 /* jackpot++; */ 826 J[i] = 0; 827 if (c != '\n' && c != EOF) 828 ctold += skipline(f1); 829 if (d != '\n' && c != EOF) 830 ctnew += skipline(f2); 831 break; 832 } 833 if (c == '\n' || c == EOF) 834 break; 835 } 836 } else { 837 for (;;) { 838 ctold++; 839 ctnew++; 840 if ((c = getc(f1)) != (d = getc(f2))) { 841 /* jackpot++; */ 842 J[i] = 0; 843 if (c != '\n' && c != EOF) 844 ctold += skipline(f1); 845 if (d != '\n' && c != EOF) 846 ctnew += skipline(f2); 847 break; 848 } 849 if (c == '\n' || c == EOF) 850 break; 851 } 852 } 853 ixold[i] = ctold; 854 ixnew[j] = ctnew; 855 j++; 856 } 857 for (; j <= (int)len[1]; j++) { 858 ixnew[j] = ctnew += skipline(f2); 859 } 860 /* 861 * if (jackpot) 862 * fprintf(stderr, "jackpot\n"); 863 */ 864 } 865 866 /* shellsort CACM #201 */ 867 static void 868 sort(struct line *a, int n) 869 { 870 struct line *ai, *aim, w; 871 int j, m = 0, k; 872 873 if (n == 0) 874 return; 875 for (j = 1; j <= n; j *= 2) 876 m = 2 * j - 1; 877 for (m /= 2; m != 0; m /= 2) { 878 k = n - m; 879 for (j = 1; j <= k; j++) { 880 for (ai = &a[j]; ai > a; ai -= m) { 881 aim = &ai[m]; 882 if (aim < ai) 883 break; /* wraparound */ 884 if (aim->value > ai[0].value || 885 (aim->value == ai[0].value && 886 aim->serial > ai[0].serial)) 887 break; 888 w.value = ai[0].value; 889 ai[0].value = aim->value; 890 aim->value = w.value; 891 w.serial = ai[0].serial; 892 ai[0].serial = aim->serial; 893 aim->serial = w.serial; 894 } 895 } 896 } 897 } 898 899 static void 900 unsort(struct line *f, int l, int *b) 901 { 902 int *a, i; 903 904 a = xcalloc(l + 1, sizeof(*a)); 905 for (i = 1; i <= l; i++) 906 a[f[i].serial] = f[i].value; 907 for (i = 1; i <= l; i++) 908 b[i] = a[i]; 909 free(a); 910 } 911 912 static int 913 skipline(FILE *f) 914 { 915 int i, c; 916 917 for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++) 918 continue; 919 return (i); 920 } 921 922 static void 923 output(char *file1, FILE *f1, char *file2, FILE *f2, int flags) 924 { 925 int i, j, m, i0, i1, j0, j1, nc; 926 927 rewind(f1); 928 rewind(f2); 929 m = len[0]; 930 J[0] = 0; 931 J[m + 1] = len[1] + 1; 932 if (diff_format != D_EDIT) { 933 for (i0 = 1; i0 <= m; i0 = i1 + 1) { 934 while (i0 <= m && J[i0] == J[i0 - 1] + 1) { 935 if (diff_format == D_SIDEBYSIDE && suppress_common != 1) { 936 nc = fetch(ixold, i0, i0, f1, '\0', 1, flags); 937 print_space(nc, hw - nc + lpad + 1 + rpad, flags); 938 fetch(ixnew, J[i0], J[i0], f2, '\0', 0, flags); 939 printf("\n"); 940 } 941 i0++; 942 } 943 j0 = J[i0 - 1] + 1; 944 i1 = i0 - 1; 945 while (i1 < m && J[i1 + 1] == 0) 946 i1++; 947 j1 = J[i1 + 1] - 1; 948 J[i1] = j1; 949 950 /* 951 * When using side-by-side, lines from both of the files are 952 * printed. The algorithm used by diff(1) identifies the ranges 953 * in which two files differ. 954 * See the change() function below. 955 * The for loop below consumes the shorter range, whereas one of 956 * the while loops deals with the longer one. 957 */ 958 if (diff_format == D_SIDEBYSIDE) { 959 for (i = i0, j = j0; i <= i1 && j <= j1; i++, j++) 960 change(file1, f1, file2, f2, i, i, j, j, &flags); 961 962 while (i <= i1) { 963 change(file1, f1, file2, f2, i, i, j + 1, j, &flags); 964 i++; 965 } 966 967 while (j <= j1) { 968 change(file1, f1, file2, f2, i + 1, i, j, j, &flags); 969 j++; 970 } 971 } else 972 change(file1, f1, file2, f2, i0, i1, j0, j1, &flags); 973 } 974 } else { 975 for (i0 = m; i0 >= 1; i0 = i1 - 1) { 976 while (i0 >= 1 && J[i0] == J[i0 + 1] - 1 && J[i0] != 0) 977 i0--; 978 j0 = J[i0 + 1] - 1; 979 i1 = i0 + 1; 980 while (i1 > 1 && J[i1 - 1] == 0) 981 i1--; 982 j1 = J[i1 - 1] + 1; 983 J[i1] = j1; 984 change(file1, f1, file2, f2, i1, i0, j1, j0, &flags); 985 } 986 } 987 if (m == 0) 988 change(file1, f1, file2, f2, 1, 0, 1, len[1], &flags); 989 if (diff_format == D_IFDEF || diff_format == D_GFORMAT) { 990 for (;;) { 991 #define c i0 992 if ((c = getc(f1)) == EOF) 993 return; 994 printf("%c", c); 995 } 996 #undef c 997 } 998 if (anychange != 0) { 999 if (diff_format == D_CONTEXT) 1000 dump_context_vec(f1, f2, flags); 1001 else if (diff_format == D_UNIFIED) 1002 dump_unified_vec(f1, f2, flags); 1003 } 1004 } 1005 1006 static void 1007 range(int a, int b, const char *separator) 1008 { 1009 printf("%d", a > b ? b : a); 1010 if (a < b) 1011 printf("%s%d", separator, b); 1012 } 1013 1014 static void 1015 uni_range(int a, int b) 1016 { 1017 if (a < b) 1018 printf("%d,%d", a, b - a + 1); 1019 else if (a == b) 1020 printf("%d", b); 1021 else 1022 printf("%d,0", b); 1023 } 1024 1025 static char * 1026 preadline(int fd, size_t rlen, off_t off) 1027 { 1028 char *line; 1029 ssize_t nr; 1030 1031 line = xmalloc(rlen + 1); 1032 if ((nr = pread(fd, line, rlen, off)) == -1) 1033 err(2, "preadline"); 1034 if (nr > 0 && line[nr-1] == '\n') 1035 nr--; 1036 line[nr] = '\0'; 1037 return (line); 1038 } 1039 1040 static bool 1041 ignoreline_pattern(char *line) 1042 { 1043 int ret; 1044 1045 ret = regexec(&ignore_re, line, 0, NULL, 0); 1046 return (ret == 0); /* if it matched, it should be ignored. */ 1047 } 1048 1049 static bool 1050 ignoreline(char *line, bool skip_blanks) 1051 { 1052 1053 if (skip_blanks && *line == '\0') 1054 return (true); 1055 if (ignore_pats != NULL && ignoreline_pattern(line)) 1056 return (true); 1057 return (false); 1058 } 1059 1060 /* 1061 * Indicate that there is a difference between lines a and b of the from file 1062 * to get to lines c to d of the to file. If a is greater then b then there 1063 * are no lines in the from file involved and this means that there were 1064 * lines appended (beginning at b). If c is greater than d then there are 1065 * lines missing from the to file. 1066 */ 1067 static void 1068 change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d, 1069 int *pflags) 1070 { 1071 static size_t max_context = 64; 1072 long curpos; 1073 int dist, i, nc; 1074 const char *walk; 1075 bool skip_blanks, ignore; 1076 1077 skip_blanks = (*pflags & D_SKIPBLANKLINES); 1078 restart: 1079 if ((diff_format != D_IFDEF || diff_format == D_GFORMAT) && 1080 a > b && c > d) 1081 return; 1082 if (ignore_pats != NULL || skip_blanks) { 1083 char *line; 1084 /* 1085 * All lines in the change, insert, or delete must match an ignore 1086 * pattern for the change to be ignored. 1087 */ 1088 if (a <= b) { /* Changes and deletes. */ 1089 for (i = a; i <= b; i++) { 1090 line = preadline(fileno(f1), 1091 ixold[i] - ixold[i - 1], ixold[i - 1]); 1092 ignore = ignoreline(line, skip_blanks); 1093 free(line); 1094 if (!ignore) 1095 goto proceed; 1096 } 1097 } 1098 if (a > b || c <= d) { /* Changes and inserts. */ 1099 for (i = c; i <= d; i++) { 1100 line = preadline(fileno(f2), 1101 ixnew[i] - ixnew[i - 1], ixnew[i - 1]); 1102 ignore = ignoreline(line, skip_blanks); 1103 free(line); 1104 if (!ignore) 1105 goto proceed; 1106 } 1107 } 1108 return; 1109 } 1110 proceed: 1111 if ((*pflags & D_HEADER) != 0 && 1112 diff_format != D_BRIEF && diff_format != D_QUIET) { 1113 printf("%s %s %s\n", diffargs, file1, file2); 1114 *pflags &= ~D_HEADER; 1115 } 1116 if (diff_format == D_CONTEXT || diff_format == D_UNIFIED) { 1117 /* 1118 * Allocate change records as needed. 1119 */ 1120 if (context_vec_start == NULL || 1121 context_vec_ptr == context_vec_end - 1) { 1122 ptrdiff_t offset = -1; 1123 1124 if (context_vec_start != NULL) 1125 offset = context_vec_ptr - context_vec_start; 1126 max_context <<= 1; 1127 context_vec_start = xreallocarray(context_vec_start, 1128 max_context, sizeof(*context_vec_start)); 1129 context_vec_end = context_vec_start + max_context; 1130 context_vec_ptr = context_vec_start + offset; 1131 } 1132 if (anychange == 0) { 1133 /* 1134 * Print the context/unidiff header first time through. 1135 */ 1136 print_header(file1, file2); 1137 anychange = 1; 1138 } else if (!ckd_add(&dist, diff_context, diff_context) && 1139 a - context_vec_ptr->b - 1 > dist && 1140 c - context_vec_ptr->d - 1 > dist) { 1141 /* 1142 * If this change is more than 'diff_context' lines from the 1143 * previous change, dump the record and reset it. 1144 */ 1145 if (diff_format == D_CONTEXT) 1146 dump_context_vec(f1, f2, *pflags); 1147 else 1148 dump_unified_vec(f1, f2, *pflags); 1149 } 1150 context_vec_ptr++; 1151 context_vec_ptr->a = a; 1152 context_vec_ptr->b = b; 1153 context_vec_ptr->c = c; 1154 context_vec_ptr->d = d; 1155 return; 1156 } 1157 if (anychange == 0) 1158 anychange = 1; 1159 switch (diff_format) { 1160 case D_BRIEF: 1161 case D_QUIET: 1162 return; 1163 case D_NORMAL: 1164 case D_EDIT: 1165 range(a, b, ","); 1166 printf("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1167 if (diff_format == D_NORMAL) 1168 range(c, d, ","); 1169 printf("\n"); 1170 break; 1171 case D_REVERSE: 1172 printf("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1173 range(a, b, " "); 1174 printf("\n"); 1175 break; 1176 case D_NREVERSE: 1177 if (a > b) 1178 printf("a%d %d\n", b, d - c + 1); 1179 else { 1180 printf("d%d %d\n", a, b - a + 1); 1181 if (!(c > d)) 1182 /* add changed lines */ 1183 printf("a%d %d\n", b, d - c + 1); 1184 } 1185 break; 1186 } 1187 if (diff_format == D_GFORMAT) { 1188 curpos = ftell(f1); 1189 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1190 nc = ixold[a > b ? b : a - 1] - curpos; 1191 for (i = 0; i < nc; i++) 1192 printf("%c", getc(f1)); 1193 for (walk = group_format; *walk != '\0'; walk++) { 1194 if (*walk == '%') { 1195 walk++; 1196 switch (*walk) { 1197 case '<': 1198 fetch(ixold, a, b, f1, '<', 1, *pflags); 1199 break; 1200 case '>': 1201 fetch(ixnew, c, d, f2, '>', 0, *pflags); 1202 break; 1203 default: 1204 printf("%%%c", *walk); 1205 break; 1206 } 1207 continue; 1208 } 1209 printf("%c", *walk); 1210 } 1211 } 1212 if (diff_format == D_SIDEBYSIDE) { 1213 if (color && a > b) 1214 printf("\033[%sm", add_code); 1215 else if (color && c > d) 1216 printf("\033[%sm", del_code); 1217 if (a > b) { 1218 print_space(0, hw + lpad, *pflags); 1219 } else { 1220 nc = fetch(ixold, a, b, f1, '\0', 1, *pflags); 1221 print_space(nc, hw - nc + lpad, *pflags); 1222 } 1223 if (color && a > b) 1224 printf("\033[%sm", add_code); 1225 else if (color && c > d) 1226 printf("\033[%sm", del_code); 1227 printf("%c", (a > b) ? '>' : ((c > d) ? '<' : '|')); 1228 if (color && c > d) 1229 printf("\033[m"); 1230 print_space(hw + lpad + 1, rpad, *pflags); 1231 fetch(ixnew, c, d, f2, '\0', 0, *pflags); 1232 printf("\n"); 1233 } 1234 if (diff_format == D_NORMAL || diff_format == D_IFDEF) { 1235 fetch(ixold, a, b, f1, '<', 1, *pflags); 1236 if (a <= b && c <= d && diff_format == D_NORMAL) 1237 printf("---\n"); 1238 } 1239 if (diff_format != D_GFORMAT && diff_format != D_SIDEBYSIDE) 1240 fetch(ixnew, c, d, f2, diff_format == D_NORMAL ? '>' : '\0', 0, *pflags); 1241 if (edoffset != 0 && diff_format == D_EDIT) { 1242 /* 1243 * A non-zero edoffset value for D_EDIT indicates that the last line 1244 * printed was a bare dot (".") that has been escaped as ".." to 1245 * prevent ed(1) from misinterpreting it. We have to add a 1246 * substitute command to change this back and restart where we left 1247 * off. 1248 */ 1249 printf(".\n"); 1250 printf("%ds/.//\n", a + edoffset - 1); 1251 b = a + edoffset - 1; 1252 a = b + 1; 1253 c += edoffset; 1254 goto restart; 1255 } 1256 if ((diff_format == D_EDIT || diff_format == D_REVERSE) && c <= d) 1257 printf(".\n"); 1258 if (inifdef) { 1259 printf("#endif /* %s */\n", ifdefname); 1260 inifdef = 0; 1261 } 1262 } 1263 1264 static int 1265 fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile, int flags) 1266 { 1267 int i, j, c, lastc, col, nc, newcol; 1268 1269 edoffset = 0; 1270 nc = 0; 1271 col = 0; 1272 /* 1273 * When doing #ifdef's, copy down to current line 1274 * if this is the first file, so that stuff makes it to output. 1275 */ 1276 if ((diff_format == D_IFDEF) && oldfile) { 1277 long curpos = ftell(lb); 1278 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1279 nc = f[a > b ? b : a - 1] - curpos; 1280 for (i = 0; i < nc; i++) 1281 printf("%c", getc(lb)); 1282 } 1283 if (a > b) 1284 return (0); 1285 if (diff_format == D_IFDEF) { 1286 if (inifdef) { 1287 printf("#else /* %s%s */\n", 1288 oldfile == 1 ? "!" : "", ifdefname); 1289 } else { 1290 if (oldfile) 1291 printf("#ifndef %s\n", ifdefname); 1292 else 1293 printf("#ifdef %s\n", ifdefname); 1294 } 1295 inifdef = 1 + oldfile; 1296 } 1297 for (i = a; i <= b; i++) { 1298 fseek(lb, f[i - 1], SEEK_SET); 1299 nc = f[i] - f[i - 1]; 1300 if (diff_format == D_SIDEBYSIDE && hw < nc) 1301 nc = hw; 1302 if (diff_format != D_IFDEF && diff_format != D_GFORMAT && 1303 ch != '\0') { 1304 if (color && (ch == '>' || ch == '+')) 1305 printf("\033[%sm", add_code); 1306 else if (color && (ch == '<' || ch == '-')) 1307 printf("\033[%sm", del_code); 1308 printf("%c", ch); 1309 if (Tflag && (diff_format == D_NORMAL || 1310 diff_format == D_CONTEXT || 1311 diff_format == D_UNIFIED)) 1312 printf("\t"); 1313 else if (diff_format != D_UNIFIED) 1314 printf(" "); 1315 } 1316 col = j = 0; 1317 lastc = '\0'; 1318 while (j < nc && (hw == 0 || col < hw)) { 1319 c = getc(lb); 1320 if (flags & D_STRIPCR && c == '\r') { 1321 if ((c = getc(lb)) == '\n') 1322 j++; 1323 else { 1324 ungetc(c, lb); 1325 c = '\r'; 1326 } 1327 } 1328 if (c == EOF) { 1329 if (diff_format == D_EDIT || 1330 diff_format == D_REVERSE || 1331 diff_format == D_NREVERSE) 1332 warnx("No newline at end of file"); 1333 else 1334 printf("\n\\ No newline at end of file\n"); 1335 return (col); 1336 } 1337 /* 1338 * when using --side-by-side, col needs to be increased 1339 * in any case to keep the columns aligned 1340 */ 1341 if (c == '\t') { 1342 /* 1343 * Calculate where the tab would bring us. 1344 * If it would take us to the end of the 1345 * column, either clip it (if expanding 1346 * tabs) or return right away (if not). 1347 */ 1348 newcol = roundup(col + 1, tabsize); 1349 if ((flags & D_EXPANDTABS) == 0) { 1350 if (hw > 0 && newcol >= hw) 1351 return (col); 1352 printf("\t"); 1353 } else { 1354 if (hw > 0 && newcol > hw) 1355 newcol = hw; 1356 printf("%*s", newcol - col, ""); 1357 } 1358 col = newcol; 1359 } else { 1360 if (diff_format == D_EDIT && j == 1 && c == '\n' && 1361 lastc == '.') { 1362 /* 1363 * Don't print a bare "." line since that will confuse 1364 * ed(1). Print ".." instead and set the, global variable 1365 * edoffset to an offset from which to restart. The 1366 * caller must check the value of edoffset 1367 */ 1368 printf(".\n"); 1369 edoffset = i - a + 1; 1370 return (edoffset); 1371 } 1372 /* when side-by-side, do not print a newline */ 1373 if (diff_format != D_SIDEBYSIDE || c != '\n') { 1374 if (color && c == '\n') 1375 printf("\033[m%c", c); 1376 else 1377 printf("%c", c); 1378 col++; 1379 } 1380 } 1381 1382 j++; 1383 lastc = c; 1384 } 1385 } 1386 if (color && diff_format == D_SIDEBYSIDE) 1387 printf("\033[m"); 1388 return (col); 1389 } 1390 1391 /* 1392 * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578. 1393 */ 1394 static enum readhash 1395 readhash(FILE *f, int flags, unsigned *hash) 1396 { 1397 int i, t, space; 1398 unsigned sum; 1399 1400 sum = 1; 1401 space = 0; 1402 for (i = 0;;) { 1403 switch (t = getc(f)) { 1404 case '\0': 1405 if ((flags & D_FORCEASCII) == 0) 1406 return (RH_BINARY); 1407 goto hashchar; 1408 case '\r': 1409 if (flags & D_STRIPCR) { 1410 t = getc(f); 1411 if (t == '\n') 1412 break; 1413 ungetc(t, f); 1414 } 1415 /* FALLTHROUGH */ 1416 case '\t': 1417 case '\v': 1418 case '\f': 1419 case ' ': 1420 if ((flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) != 0) { 1421 space++; 1422 continue; 1423 } 1424 /* FALLTHROUGH */ 1425 default: 1426 hashchar: 1427 if (space && (flags & D_IGNOREBLANKS) == 0) { 1428 i++; 1429 space = 0; 1430 } 1431 sum = sum * 127 + chrtran(t); 1432 i++; 1433 continue; 1434 case EOF: 1435 if (i == 0) 1436 return (RH_EOF); 1437 /* FALLTHROUGH */ 1438 case '\n': 1439 break; 1440 } 1441 break; 1442 } 1443 *hash = sum; 1444 return (RH_OK); 1445 } 1446 1447 static int 1448 asciifile(FILE *f) 1449 { 1450 unsigned char buf[BUFSIZ]; 1451 size_t cnt; 1452 1453 if (f == NULL) 1454 return (1); 1455 1456 rewind(f); 1457 cnt = fread(buf, 1, sizeof(buf), f); 1458 return (memchr(buf, '\0', cnt) == NULL); 1459 } 1460 1461 #define begins_with(s, pre) (strncmp(s, pre, sizeof(pre) - 1) == 0) 1462 1463 static char * 1464 match_function(const long *f, int pos, FILE *fp) 1465 { 1466 unsigned char buf[FUNCTION_CONTEXT_SIZE]; 1467 size_t nc; 1468 int last = lastline; 1469 const char *state = NULL; 1470 1471 lastline = pos; 1472 for (; pos > last; pos--) { 1473 fseek(fp, f[pos - 1], SEEK_SET); 1474 nc = f[pos] - f[pos - 1]; 1475 if (nc >= sizeof(buf)) 1476 nc = sizeof(buf) - 1; 1477 nc = fread(buf, 1, nc, fp); 1478 if (nc == 0) 1479 continue; 1480 buf[nc] = '\0'; 1481 buf[strcspn(buf, "\n")] = '\0'; 1482 if (most_recent_pat != NULL) { 1483 int ret = regexec(&most_recent_re, buf, 0, NULL, 0); 1484 1485 if (ret != 0) 1486 continue; 1487 strlcpy(lastbuf, buf, sizeof(lastbuf)); 1488 lastmatchline = pos; 1489 return (lastbuf); 1490 } else if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$' 1491 || buf[0] == '-' || buf[0] == '+') { 1492 if (begins_with(buf, "private:")) { 1493 if (!state) 1494 state = " (private)"; 1495 } else if (begins_with(buf, "protected:")) { 1496 if (!state) 1497 state = " (protected)"; 1498 } else if (begins_with(buf, "public:")) { 1499 if (!state) 1500 state = " (public)"; 1501 } else { 1502 strlcpy(lastbuf, buf, sizeof(lastbuf)); 1503 if (state) 1504 strlcat(lastbuf, state, sizeof(lastbuf)); 1505 lastmatchline = pos; 1506 return (lastbuf); 1507 } 1508 } 1509 } 1510 return (lastmatchline > 0 ? lastbuf : NULL); 1511 } 1512 1513 /* dump accumulated "context" diff changes */ 1514 static void 1515 dump_context_vec(FILE *f1, FILE *f2, int flags) 1516 { 1517 struct context_vec *cvp = context_vec_start; 1518 int lowa, upb, lowc, upd, do_output; 1519 int a, b, c, d; 1520 char ch, *f; 1521 1522 if (context_vec_start > context_vec_ptr) 1523 return; 1524 1525 b = d = 0; /* gcc */ 1526 if (ckd_sub(&lowa, cvp->a, diff_context) || lowa < 1) 1527 lowa = 1; 1528 if (ckd_add(&upb, context_vec_ptr->b, diff_context) || upb > (int)len[0]) 1529 upb = (int)len[0]; 1530 if (ckd_sub(&lowc, cvp->c, diff_context) || lowc < 1) 1531 lowc = 1; 1532 if (ckd_add(&upd, context_vec_ptr->d, diff_context) || upd > (int)len[1]) 1533 upd = (int)len[1]; 1534 1535 printf("***************"); 1536 if (flags & (D_PROTOTYPE | D_MATCHLAST)) { 1537 f = match_function(ixold, cvp->a - 1, f1); 1538 if (f != NULL) 1539 printf(" %s", f); 1540 } 1541 printf("\n*** "); 1542 range(lowa, upb, ","); 1543 printf(" ****\n"); 1544 1545 /* 1546 * Output changes to the "old" file. The first loop suppresses 1547 * output if there were no changes to the "old" file (we'll see 1548 * the "old" lines as context in the "new" list). 1549 */ 1550 do_output = 0; 1551 for (; cvp <= context_vec_ptr; cvp++) 1552 if (cvp->a <= cvp->b) { 1553 cvp = context_vec_start; 1554 do_output++; 1555 break; 1556 } 1557 if (do_output) { 1558 while (cvp <= context_vec_ptr) { 1559 a = cvp->a; 1560 b = cvp->b; 1561 c = cvp->c; 1562 d = cvp->d; 1563 1564 if (a <= b && c <= d) 1565 ch = 'c'; 1566 else 1567 ch = (a <= b) ? 'd' : 'a'; 1568 1569 if (ch == 'a') 1570 fetch(ixold, lowa, b, f1, ' ', 0, flags); 1571 else { 1572 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1573 fetch(ixold, a, b, f1, 1574 ch == 'c' ? '!' : '-', 0, flags); 1575 } 1576 lowa = b + 1; 1577 cvp++; 1578 } 1579 fetch(ixold, b + 1, upb, f1, ' ', 0, flags); 1580 } 1581 /* output changes to the "new" file */ 1582 printf("--- "); 1583 range(lowc, upd, ","); 1584 printf(" ----\n"); 1585 1586 do_output = 0; 1587 for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++) 1588 if (cvp->c <= cvp->d) { 1589 cvp = context_vec_start; 1590 do_output++; 1591 break; 1592 } 1593 if (do_output) { 1594 while (cvp <= context_vec_ptr) { 1595 a = cvp->a; 1596 b = cvp->b; 1597 c = cvp->c; 1598 d = cvp->d; 1599 1600 if (a <= b && c <= d) 1601 ch = 'c'; 1602 else 1603 ch = (a <= b) ? 'd' : 'a'; 1604 1605 if (ch == 'd') 1606 fetch(ixnew, lowc, d, f2, ' ', 0, flags); 1607 else { 1608 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1609 fetch(ixnew, c, d, f2, 1610 ch == 'c' ? '!' : '+', 0, flags); 1611 } 1612 lowc = d + 1; 1613 cvp++; 1614 } 1615 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1616 } 1617 context_vec_ptr = context_vec_start - 1; 1618 } 1619 1620 /* dump accumulated "unified" diff changes */ 1621 static void 1622 dump_unified_vec(FILE *f1, FILE *f2, int flags) 1623 { 1624 struct context_vec *cvp = context_vec_start; 1625 int lowa, upb, lowc, upd; 1626 int a, b, c, d; 1627 char ch, *f; 1628 1629 if (context_vec_start > context_vec_ptr) 1630 return; 1631 1632 b = d = 0; /* gcc */ 1633 if (ckd_sub(&lowa, cvp->a, diff_context) || lowa < 1) 1634 lowa = 1; 1635 if (ckd_add(&upb, context_vec_ptr->b, diff_context) || upb > (int)len[0]) 1636 upb = (int)len[0]; 1637 if (ckd_sub(&lowc, cvp->c, diff_context) || lowc < 1) 1638 lowc = 1; 1639 if (ckd_add(&upd, context_vec_ptr->d, diff_context) || upd > (int)len[1]) 1640 upd = (int)len[1]; 1641 1642 printf("@@ -"); 1643 uni_range(lowa, upb); 1644 printf(" +"); 1645 uni_range(lowc, upd); 1646 printf(" @@"); 1647 if (flags & (D_PROTOTYPE | D_MATCHLAST)) { 1648 f = match_function(ixold, cvp->a - 1, f1); 1649 if (f != NULL) 1650 printf(" %s", f); 1651 } 1652 printf("\n"); 1653 1654 /* 1655 * Output changes in "unified" diff format--the old and new lines 1656 * are printed together. 1657 */ 1658 for (; cvp <= context_vec_ptr; cvp++) { 1659 a = cvp->a; 1660 b = cvp->b; 1661 c = cvp->c; 1662 d = cvp->d; 1663 1664 /* 1665 * c: both new and old changes 1666 * d: only changes in the old file 1667 * a: only changes in the new file 1668 */ 1669 if (a <= b && c <= d) 1670 ch = 'c'; 1671 else 1672 ch = (a <= b) ? 'd' : 'a'; 1673 1674 switch (ch) { 1675 case 'c': 1676 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1677 fetch(ixold, a, b, f1, '-', 0, flags); 1678 fetch(ixnew, c, d, f2, '+', 0, flags); 1679 break; 1680 case 'd': 1681 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1682 fetch(ixold, a, b, f1, '-', 0, flags); 1683 break; 1684 case 'a': 1685 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1686 fetch(ixnew, c, d, f2, '+', 0, flags); 1687 break; 1688 } 1689 lowa = b + 1; 1690 lowc = d + 1; 1691 } 1692 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1693 1694 context_vec_ptr = context_vec_start - 1; 1695 } 1696 1697 static void 1698 print_header(const char *file1, const char *file2) 1699 { 1700 const char *time_format; 1701 char buf[256]; 1702 struct tm tm1, tm2, *tm_ptr1, *tm_ptr2; 1703 int nsec1 = stb1.st_mtim.tv_nsec; 1704 int nsec2 = stb2.st_mtim.tv_nsec; 1705 1706 time_format = "%Y-%m-%d %H:%M:%S"; 1707 1708 if (cflag) 1709 time_format = "%c"; 1710 tm_ptr1 = localtime_r(&stb1.st_mtime, &tm1); 1711 tm_ptr2 = localtime_r(&stb2.st_mtime, &tm2); 1712 if (label[0] != NULL) 1713 printf("%s %s\n", diff_format == D_CONTEXT ? "***" : "---", 1714 label[0]); 1715 else { 1716 strftime(buf, sizeof(buf), time_format, tm_ptr1); 1717 printf("%s %s\t%s", diff_format == D_CONTEXT ? "***" : "---", 1718 file1, buf); 1719 if (!cflag) { 1720 strftime(buf, sizeof(buf), "%z", tm_ptr1); 1721 printf(".%.9d %s", nsec1, buf); 1722 } 1723 printf("\n"); 1724 } 1725 if (label[1] != NULL) 1726 printf("%s %s\n", diff_format == D_CONTEXT ? "---" : "+++", 1727 label[1]); 1728 else { 1729 strftime(buf, sizeof(buf), time_format, tm_ptr2); 1730 printf("%s %s\t%s", diff_format == D_CONTEXT ? "---" : "+++", 1731 file2, buf); 1732 if (!cflag) { 1733 strftime(buf, sizeof(buf), "%z", tm_ptr2); 1734 printf(".%.9d %s", nsec2, buf); 1735 } 1736 printf("\n"); 1737 } 1738 } 1739 1740 /* 1741 * Prints n number of space characters either by using tab 1742 * or single space characters. 1743 * nc is the preceding number of characters 1744 */ 1745 static void 1746 print_space(int nc, int n, int flags) 1747 { 1748 int col, newcol, tabstop; 1749 1750 col = nc; 1751 newcol = nc + n; 1752 /* first, use tabs if allowed */ 1753 if ((flags & D_EXPANDTABS) == 0) { 1754 while ((tabstop = roundup(col + 1, tabsize)) <= newcol) { 1755 printf("\t"); 1756 col = tabstop; 1757 } 1758 } 1759 /* finish with spaces */ 1760 printf("%*s", newcol - col, ""); 1761 } 1762