1 /* $OpenBSD: diffreg.c,v 1.93 2019/06/28 13:35:00 deraadt Exp $ */ 2 3 /*- 4 * SPDX-License-Identifier: BSD-4-Clause 5 * 6 * Copyright (C) Caldera International Inc. 2001-2002. 7 * All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code and documentation must retain the above 13 * copyright notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. All advertising materials mentioning features or use of this software 18 * must display the following acknowledgement: 19 * This product includes software developed or owned by Caldera 20 * International, Inc. 21 * 4. Neither the name of Caldera International, Inc. nor the names of other 22 * contributors may be used to endorse or promote products derived from 23 * this software without specific prior written permission. 24 * 25 * USE OF THE SOFTWARE PROVIDED FOR UNDER THIS LICENSE BY CALDERA 26 * INTERNATIONAL, INC. AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR 27 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 28 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 29 * IN NO EVENT SHALL CALDERA INTERNATIONAL, INC. BE LIABLE FOR ANY DIRECT, 30 * INDIRECT INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 31 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 32 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, 34 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING 35 * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 36 * POSSIBILITY OF SUCH DAMAGE. 37 */ 38 /*- 39 * Copyright (c) 1991, 1993 40 * The Regents of the University of California. All rights reserved. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 3. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 * 66 * @(#)diffreg.c 8.1 (Berkeley) 6/6/93 67 */ 68 69 #include <sys/cdefs.h> 70 #include <sys/capsicum.h> 71 #include <sys/stat.h> 72 73 #include <capsicum_helpers.h> 74 #include <ctype.h> 75 #include <err.h> 76 #include <errno.h> 77 #include <fcntl.h> 78 #include <math.h> 79 #include <paths.h> 80 #include <regex.h> 81 #include <stdbool.h> 82 #include <stddef.h> 83 #include <stdint.h> 84 #include <stdio.h> 85 #include <stdlib.h> 86 #include <string.h> 87 88 #include "pr.h" 89 #include "diff.h" 90 #include "xmalloc.h" 91 92 /* 93 * diff - compare two files. 94 */ 95 96 /* 97 * Uses an algorithm due to Harold Stone, which finds a pair of longest 98 * identical subsequences in the two files. 99 * 100 * The major goal is to generate the match vector J. J[i] is the index of 101 * the line in file1 corresponding to line i file0. J[i] = 0 if there is no 102 * such line in file1. 103 * 104 * Lines are hashed so as to work in core. All potential matches are 105 * located by sorting the lines of each file on the hash (called 106 * ``value''). In particular, this collects the equivalence classes in 107 * file1 together. Subroutine equiv replaces the value of each line in 108 * file0 by the index of the first element of its matching equivalence in 109 * (the reordered) file1. To save space equiv squeezes file1 into a single 110 * array member in which the equivalence classes are simply concatenated, 111 * except that their first members are flagged by changing sign. 112 * 113 * Next the indices that point into member are unsorted into array class 114 * according to the original order of file0. 115 * 116 * The cleverness lies in routine stone. This marches through the lines of 117 * file0, developing a vector klist of "k-candidates". At step i 118 * a k-candidate is a matched pair of lines x,y (x in file0 y in file1) 119 * such that there is a common subsequence of length k between the first 120 * i lines of file0 and the first y lines of file1, but there is no such 121 * subsequence for any smaller y. x is the earliest possible mate to y that 122 * occurs in such a subsequence. 123 * 124 * Whenever any of the members of the equivalence class of lines in file1 125 * matable to a line in file0 has serial number less than the y of some 126 * k-candidate, that k-candidate with the smallest such y is replaced. The 127 * new k-candidate is chained (via pred) to the current k-1 candidate so 128 * that the actual subsequence can be recovered. When a member has serial 129 * number greater that the y of all k-candidates, the klist is extended. At 130 * the end, the longest subsequence is pulled out and placed in the array J 131 * by unravel. 132 * 133 * With J in hand, the matches there recorded are check'ed against reality 134 * to assure that no spurious matches have crept in due to hashing. If they 135 * have, they are broken, and "jackpot" is recorded -- a harmless matter 136 * except that a true match for a spuriously mated line may now be 137 * unnecessarily reported as a change. 138 * 139 * Much of the complexity of the program comes simply from trying to 140 * minimize core utilization and maximize the range of doable problems by 141 * dynamically allocating what is needed and reusing what is not. The core 142 * requirements for problems larger than somewhat are (in words) 143 * 2*length(file0) + length(file1) + 3*(number of k-candidates installed), 144 * typically about 6n words for files of length n. 145 */ 146 147 struct cand { 148 int x; 149 int y; 150 int pred; 151 }; 152 153 static struct line { 154 int serial; 155 int value; 156 } *file[2]; 157 158 /* 159 * The following struct is used to record change information when 160 * doing a "context" or "unified" diff. (see routine "change" to 161 * understand the highly mnemonic field names) 162 */ 163 struct context_vec { 164 int a; /* start line in old file */ 165 int b; /* end line in old file */ 166 int c; /* start line in new file */ 167 int d; /* end line in new file */ 168 }; 169 170 enum readhash { RH_BINARY, RH_OK, RH_EOF }; 171 172 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 static int 229 clow2low(int c) 230 { 231 232 return (c); 233 } 234 235 static int 236 cup2low(int c) 237 { 238 239 return (tolower(c)); 240 } 241 242 int 243 diffreg(char *file1, char *file2, int flags, int capsicum) 244 { 245 FILE *f1, *f2; 246 int i, rval; 247 struct pr *pr = NULL; 248 cap_rights_t rights_ro; 249 250 f1 = f2 = NULL; 251 rval = D_SAME; 252 anychange = 0; 253 lastline = 0; 254 lastmatchline = 0; 255 256 /* 257 * In side-by-side mode, we need to print the left column, a 258 * change marker surrounded by padding, and the right column. 259 * 260 * If expanding tabs, we don't care about alignment, so we simply 261 * subtract 3 from the width and divide by two. 262 * 263 * If not expanding tabs, we need to ensure that the right column 264 * is aligned to a tab stop. We start with the same formula, then 265 * decrement until we reach a size that lets us tab-align the 266 * right column. We then adjust the width down if necessary for 267 * the padding calculation to work. 268 * 269 * Left padding is half the space left over, rounded down; right 270 * padding is whatever is needed to match the width. 271 */ 272 if (diff_format == D_SIDEBYSIDE) { 273 if (flags & D_EXPANDTABS) { 274 if (width > 3) { 275 hw = (width - 3) / 2; 276 } else { 277 /* not enough space */ 278 hw = 0; 279 } 280 } else if (width <= 3 || width <= tabsize) { 281 /* not enough space */ 282 hw = 0; 283 } else { 284 hw = (width - 3) / 2; 285 while (hw > 0 && roundup(hw + 3, tabsize) + hw > width) 286 hw--; 287 if (width - (roundup(hw + 3, tabsize) + hw) < tabsize) 288 width = roundup(hw + 3, tabsize) + hw; 289 } 290 lpad = (width - hw * 2 - 1) / 2; 291 rpad = (width - hw * 2 - 1) - lpad; 292 } 293 294 if (flags & D_IGNORECASE) 295 chrtran = cup2low; 296 else 297 chrtran = clow2low; 298 if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode)) 299 return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2); 300 if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0) 301 goto closem; 302 303 if (flags & D_EMPTY1) 304 f1 = fopen(_PATH_DEVNULL, "r"); 305 else { 306 if (!S_ISREG(stb1.st_mode)) { 307 if ((f1 = opentemp(file1)) == NULL || 308 fstat(fileno(f1), &stb1) == -1) { 309 warn("%s", file1); 310 rval = D_ERROR; 311 status |= 2; 312 goto closem; 313 } 314 } else if (strcmp(file1, "-") == 0) 315 f1 = stdin; 316 else 317 f1 = fopen(file1, "r"); 318 } 319 if (f1 == NULL) { 320 warn("%s", file1); 321 rval = D_ERROR; 322 status |= 2; 323 goto closem; 324 } 325 326 if (flags & D_EMPTY2) 327 f2 = fopen(_PATH_DEVNULL, "r"); 328 else { 329 if (!S_ISREG(stb2.st_mode)) { 330 if ((f2 = opentemp(file2)) == NULL || 331 fstat(fileno(f2), &stb2) == -1) { 332 warn("%s", file2); 333 rval = D_ERROR; 334 status |= 2; 335 goto closem; 336 } 337 } else if (strcmp(file2, "-") == 0) 338 f2 = stdin; 339 else 340 f2 = fopen(file2, "r"); 341 } 342 if (f2 == NULL) { 343 warn("%s", file2); 344 rval = D_ERROR; 345 status |= 2; 346 goto closem; 347 } 348 349 if (lflag) 350 pr = start_pr(file1, file2); 351 352 if (capsicum) { 353 cap_rights_init(&rights_ro, CAP_READ, CAP_FSTAT, CAP_SEEK); 354 if (caph_rights_limit(fileno(f1), &rights_ro) < 0) 355 err(2, "unable to limit rights on: %s", file1); 356 if (caph_rights_limit(fileno(f2), &rights_ro) < 0) 357 err(2, "unable to limit rights on: %s", file2); 358 if (fileno(f1) == STDIN_FILENO || fileno(f2) == STDIN_FILENO) { 359 /* stdin has already been limited */ 360 if (caph_limit_stderr() == -1) 361 err(2, "unable to limit stderr"); 362 if (caph_limit_stdout() == -1) 363 err(2, "unable to limit stdout"); 364 } else if (caph_limit_stdio() == -1) 365 err(2, "unable to limit stdio"); 366 367 caph_cache_catpages(); 368 caph_cache_tzdata(); 369 if (caph_enter() < 0) 370 err(2, "unable to enter capability mode"); 371 } 372 373 switch (files_differ(f1, f2, flags)) { 374 case 0: 375 goto closem; 376 case 1: 377 break; 378 default: 379 /* error */ 380 rval = D_ERROR; 381 status |= 2; 382 goto closem; 383 } 384 385 if (diff_format == D_BRIEF && ignore_pats == NULL && 386 (flags & (D_FOLDBLANKS|D_IGNOREBLANKS|D_IGNORECASE| 387 D_SKIPBLANKLINES|D_STRIPCR)) == 0) 388 { 389 rval = D_DIFFER; 390 status |= 1; 391 goto closem; 392 } 393 if ((flags & D_FORCEASCII) != 0) { 394 (void)prepare(0, f1, stb1.st_size, flags); 395 (void)prepare(1, f2, stb2.st_size, flags); 396 } else if (!asciifile(f1) || !asciifile(f2) || 397 !prepare(0, f1, stb1.st_size, flags) || 398 !prepare(1, f2, stb2.st_size, flags)) { 399 rval = D_BINARY; 400 status |= 1; 401 goto closem; 402 } 403 if (len[0] > INT_MAX - 2) 404 errc(1, EFBIG, "%s", file1); 405 if (len[1] > INT_MAX - 2) 406 errc(1, EFBIG, "%s", file2); 407 408 prune(); 409 sort(sfile[0], slen[0]); 410 sort(sfile[1], slen[1]); 411 412 member = (int *)file[1]; 413 equiv(sfile[0], slen[0], sfile[1], slen[1], member); 414 member = xreallocarray(member, slen[1] + 2, sizeof(*member)); 415 416 class = (int *)file[0]; 417 unsort(sfile[0], slen[0], class); 418 class = xreallocarray(class, slen[0] + 2, sizeof(*class)); 419 420 klist = xcalloc(slen[0] + 2, sizeof(*klist)); 421 clen = 0; 422 clistlen = 100; 423 clist = xcalloc(clistlen, sizeof(*clist)); 424 i = stone(class, slen[0], member, klist, flags); 425 free(member); 426 free(class); 427 428 J = xreallocarray(J, len[0] + 2, sizeof(*J)); 429 unravel(klist[i]); 430 free(clist); 431 free(klist); 432 433 ixold = xreallocarray(ixold, len[0] + 2, sizeof(*ixold)); 434 ixnew = xreallocarray(ixnew, len[1] + 2, sizeof(*ixnew)); 435 check(f1, f2, flags); 436 output(file1, f1, file2, f2, flags); 437 438 closem: 439 if (pr != NULL) 440 stop_pr(pr); 441 if (anychange) { 442 status |= 1; 443 if (rval == D_SAME) 444 rval = D_DIFFER; 445 } 446 if (f1 != NULL) 447 fclose(f1); 448 if (f2 != NULL) 449 fclose(f2); 450 451 return (rval); 452 } 453 454 /* 455 * Check to see if the given files differ. 456 * Returns 0 if they are the same, 1 if different, and -1 on error. 457 * XXX - could use code from cmp(1) [faster] 458 */ 459 static int 460 files_differ(FILE *f1, FILE *f2, int flags) 461 { 462 char buf1[BUFSIZ], buf2[BUFSIZ]; 463 size_t i, j; 464 465 if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size || 466 (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT)) 467 return (1); 468 469 if (stb1.st_dev == stb2.st_dev && stb1.st_ino == stb2.st_ino) 470 return (0); 471 472 for (;;) { 473 i = fread(buf1, 1, sizeof(buf1), f1); 474 j = fread(buf2, 1, sizeof(buf2), f2); 475 if ((!i && ferror(f1)) || (!j && ferror(f2))) 476 return (-1); 477 if (i != j) 478 return (1); 479 if (i == 0) 480 return (0); 481 if (memcmp(buf1, buf2, i) != 0) 482 return (1); 483 } 484 } 485 486 static FILE * 487 opentemp(const char *f) 488 { 489 char buf[BUFSIZ], tempfile[PATH_MAX]; 490 ssize_t nread; 491 int ifd, ofd; 492 493 if (strcmp(f, "-") == 0) 494 ifd = STDIN_FILENO; 495 else if ((ifd = open(f, O_RDONLY, 0644)) == -1) 496 return (NULL); 497 498 (void)strlcpy(tempfile, _PATH_TMP "/diff.XXXXXXXX", sizeof(tempfile)); 499 500 if ((ofd = mkstemp(tempfile)) == -1) { 501 close(ifd); 502 return (NULL); 503 } 504 unlink(tempfile); 505 while ((nread = read(ifd, buf, BUFSIZ)) > 0) { 506 if (write(ofd, buf, nread) != nread) { 507 close(ifd); 508 close(ofd); 509 return (NULL); 510 } 511 } 512 close(ifd); 513 lseek(ofd, (off_t)0, SEEK_SET); 514 return (fdopen(ofd, "r")); 515 } 516 517 static bool 518 prepare(int i, FILE *fd, size_t filesize, int flags) 519 { 520 struct line *p; 521 unsigned h; 522 size_t sz, j = 0; 523 enum readhash r; 524 525 rewind(fd); 526 527 sz = MIN(filesize, SIZE_MAX) / 25; 528 if (sz < 100) 529 sz = 100; 530 531 p = xcalloc(sz + 3, sizeof(*p)); 532 while ((r = readhash(fd, flags, &h)) != RH_EOF) { 533 if (r == RH_BINARY) 534 return (false); 535 if (j == SIZE_MAX) 536 break; 537 if (j == sz) { 538 sz = sz * 3 / 2; 539 p = xreallocarray(p, sz + 3, sizeof(*p)); 540 } 541 p[++j].value = h; 542 } 543 544 len[i] = j; 545 file[i] = p; 546 547 return (true); 548 } 549 550 static void 551 prune(void) 552 { 553 size_t i, j; 554 555 for (pref = 0; pref < len[0] && pref < len[1] && 556 file[0][pref + 1].value == file[1][pref + 1].value; 557 pref++) 558 ; 559 for (suff = 0; suff < len[0] - pref && suff < len[1] - pref && 560 file[0][len[0] - suff].value == file[1][len[1] - suff].value; 561 suff++) 562 ; 563 for (j = 0; j < 2; j++) { 564 sfile[j] = file[j] + pref; 565 slen[j] = len[j] - pref - suff; 566 for (i = 0; i <= slen[j]; i++) 567 sfile[j][i].serial = i; 568 } 569 } 570 571 static void 572 equiv(struct line *a, int n, struct line *b, int m, int *c) 573 { 574 int i, j; 575 576 i = j = 1; 577 while (i <= n && j <= m) { 578 if (a[i].value < b[j].value) 579 a[i++].value = 0; 580 else if (a[i].value == b[j].value) 581 a[i++].value = j; 582 else 583 j++; 584 } 585 while (i <= n) 586 a[i++].value = 0; 587 b[m + 1].value = 0; 588 j = 0; 589 while (++j <= m) { 590 c[j] = -b[j].serial; 591 while (b[j + 1].value == b[j].value) { 592 j++; 593 c[j] = b[j].serial; 594 } 595 } 596 c[j] = -1; 597 } 598 599 static int 600 stone(int *a, int n, int *b, int *c, int flags) 601 { 602 int i, k, y, j, l; 603 int oldc, tc, oldl, sq; 604 unsigned numtries, bound; 605 606 if (flags & D_MINIMAL) 607 bound = UINT_MAX; 608 else { 609 sq = sqrt(n); 610 bound = MAX(256, sq); 611 } 612 613 k = 0; 614 c[0] = newcand(0, 0, 0); 615 for (i = 1; i <= n; i++) { 616 j = a[i]; 617 if (j == 0) 618 continue; 619 y = -b[j]; 620 oldl = 0; 621 oldc = c[0]; 622 numtries = 0; 623 do { 624 if (y <= clist[oldc].y) 625 continue; 626 l = search(c, k, y); 627 if (l != oldl + 1) 628 oldc = c[l - 1]; 629 if (l <= k) { 630 if (clist[c[l]].y <= y) 631 continue; 632 tc = c[l]; 633 c[l] = newcand(i, y, oldc); 634 oldc = tc; 635 oldl = l; 636 numtries++; 637 } else { 638 c[l] = newcand(i, y, oldc); 639 k++; 640 break; 641 } 642 } while ((y = b[++j]) > 0 && numtries < bound); 643 } 644 return (k); 645 } 646 647 static int 648 newcand(int x, int y, int pred) 649 { 650 struct cand *q; 651 652 if (clen == clistlen) { 653 clistlen = clistlen * 11 / 10; 654 clist = xreallocarray(clist, clistlen, sizeof(*clist)); 655 } 656 q = clist + clen; 657 q->x = x; 658 q->y = y; 659 q->pred = pred; 660 return (clen++); 661 } 662 663 static int 664 search(int *c, int k, int y) 665 { 666 int i, j, l, t; 667 668 if (clist[c[k]].y < y) /* quick look for typical case */ 669 return (k + 1); 670 i = 0; 671 j = k + 1; 672 for (;;) { 673 l = (i + j) / 2; 674 if (l <= i) 675 break; 676 t = clist[c[l]].y; 677 if (t > y) 678 j = l; 679 else if (t < y) 680 i = l; 681 else 682 return (l); 683 } 684 return (l + 1); 685 } 686 687 static void 688 unravel(int p) 689 { 690 struct cand *q; 691 size_t i; 692 693 for (i = 0; i <= len[0]; i++) 694 J[i] = i <= pref ? i : 695 i > len[0] - suff ? i + len[1] - len[0] : 0; 696 for (q = clist + p; q->y != 0; q = clist + q->pred) 697 J[q->x + pref] = q->y + pref; 698 } 699 700 /* 701 * Check does double duty: 702 * 1. ferret out any fortuitous correspondences due to confounding by 703 * hashing (which result in "jackpot") 704 * 2. collect random access indexes to the two files 705 */ 706 static void 707 check(FILE *f1, FILE *f2, int flags) 708 { 709 int i, j, /* jackpot, */ c, d; 710 long ctold, ctnew; 711 712 rewind(f1); 713 rewind(f2); 714 j = 1; 715 ixold[0] = ixnew[0] = 0; 716 /* jackpot = 0; */ 717 ctold = ctnew = 0; 718 for (i = 1; i <= (int)len[0]; i++) { 719 if (J[i] == 0) { 720 ixold[i] = ctold += skipline(f1); 721 continue; 722 } 723 while (j < J[i]) { 724 ixnew[j] = ctnew += skipline(f2); 725 j++; 726 } 727 if (flags & (D_FOLDBLANKS | D_IGNOREBLANKS | D_IGNORECASE | D_STRIPCR)) { 728 for (;;) { 729 c = getc(f1); 730 d = getc(f2); 731 /* 732 * GNU diff ignores a missing newline 733 * in one file for -b or -w. 734 */ 735 if (flags & (D_FOLDBLANKS | D_IGNOREBLANKS)) { 736 if (c == EOF && d == '\n') { 737 ctnew++; 738 break; 739 } else if (c == '\n' && d == EOF) { 740 ctold++; 741 break; 742 } 743 } 744 ctold++; 745 ctnew++; 746 if (flags & D_STRIPCR && (c == '\r' || d == '\r')) { 747 if (c == '\r') { 748 if ((c = getc(f1)) == '\n') { 749 ctold++; 750 } else { 751 ungetc(c, f1); 752 } 753 } 754 if (d == '\r') { 755 if ((d = getc(f2)) == '\n') { 756 ctnew++; 757 } else { 758 ungetc(d, f2); 759 } 760 } 761 break; 762 } 763 if ((flags & D_FOLDBLANKS) && isspace(c) && 764 isspace(d)) { 765 do { 766 if (c == '\n') 767 break; 768 ctold++; 769 } while (isspace(c = getc(f1))); 770 do { 771 if (d == '\n') 772 break; 773 ctnew++; 774 } while (isspace(d = getc(f2))); 775 } else if (flags & D_IGNOREBLANKS) { 776 while (isspace(c) && c != '\n') { 777 c = getc(f1); 778 ctold++; 779 } 780 while (isspace(d) && d != '\n') { 781 d = getc(f2); 782 ctnew++; 783 } 784 } 785 if (chrtran(c) != chrtran(d)) { 786 /* jackpot++; */ 787 J[i] = 0; 788 if (c != '\n' && c != EOF) 789 ctold += skipline(f1); 790 if (d != '\n' && c != EOF) 791 ctnew += skipline(f2); 792 break; 793 } 794 if (c == '\n' || c == EOF) 795 break; 796 } 797 } else { 798 for (;;) { 799 ctold++; 800 ctnew++; 801 if ((c = getc(f1)) != (d = getc(f2))) { 802 /* jackpot++; */ 803 J[i] = 0; 804 if (c != '\n' && c != EOF) 805 ctold += skipline(f1); 806 if (d != '\n' && c != EOF) 807 ctnew += skipline(f2); 808 break; 809 } 810 if (c == '\n' || c == EOF) 811 break; 812 } 813 } 814 ixold[i] = ctold; 815 ixnew[j] = ctnew; 816 j++; 817 } 818 for (; j <= (int)len[1]; j++) { 819 ixnew[j] = ctnew += skipline(f2); 820 } 821 /* 822 * if (jackpot) 823 * fprintf(stderr, "jackpot\n"); 824 */ 825 } 826 827 /* shellsort CACM #201 */ 828 static void 829 sort(struct line *a, int n) 830 { 831 struct line *ai, *aim, w; 832 int j, m = 0, k; 833 834 if (n == 0) 835 return; 836 for (j = 1; j <= n; j *= 2) 837 m = 2 * j - 1; 838 for (m /= 2; m != 0; m /= 2) { 839 k = n - m; 840 for (j = 1; j <= k; j++) { 841 for (ai = &a[j]; ai > a; ai -= m) { 842 aim = &ai[m]; 843 if (aim < ai) 844 break; /* wraparound */ 845 if (aim->value > ai[0].value || 846 (aim->value == ai[0].value && 847 aim->serial > ai[0].serial)) 848 break; 849 w.value = ai[0].value; 850 ai[0].value = aim->value; 851 aim->value = w.value; 852 w.serial = ai[0].serial; 853 ai[0].serial = aim->serial; 854 aim->serial = w.serial; 855 } 856 } 857 } 858 } 859 860 static void 861 unsort(struct line *f, int l, int *b) 862 { 863 int *a, i; 864 865 a = xcalloc(l + 1, sizeof(*a)); 866 for (i = 1; i <= l; i++) 867 a[f[i].serial] = f[i].value; 868 for (i = 1; i <= l; i++) 869 b[i] = a[i]; 870 free(a); 871 } 872 873 static int 874 skipline(FILE *f) 875 { 876 int i, c; 877 878 for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++) 879 continue; 880 return (i); 881 } 882 883 static void 884 output(char *file1, FILE *f1, char *file2, FILE *f2, int flags) 885 { 886 int i, j, m, i0, i1, j0, j1, nc; 887 888 rewind(f1); 889 rewind(f2); 890 m = len[0]; 891 J[0] = 0; 892 J[m + 1] = len[1] + 1; 893 if (diff_format != D_EDIT) { 894 for (i0 = 1; i0 <= m; i0 = i1 + 1) { 895 while (i0 <= m && J[i0] == J[i0 - 1] + 1) { 896 if (diff_format == D_SIDEBYSIDE && suppress_common != 1) { 897 nc = fetch(ixold, i0, i0, f1, '\0', 1, flags); 898 print_space(nc, hw - nc + lpad + 1 + rpad, flags); 899 fetch(ixnew, J[i0], J[i0], f2, '\0', 0, flags); 900 printf("\n"); 901 } 902 i0++; 903 } 904 j0 = J[i0 - 1] + 1; 905 i1 = i0 - 1; 906 while (i1 < m && J[i1 + 1] == 0) 907 i1++; 908 j1 = J[i1 + 1] - 1; 909 J[i1] = j1; 910 911 /* 912 * When using side-by-side, lines from both of the files are 913 * printed. The algorithm used by diff(1) identifies the ranges 914 * in which two files differ. 915 * See the change() function below. 916 * The for loop below consumes the shorter range, whereas one of 917 * the while loops deals with the longer one. 918 */ 919 if (diff_format == D_SIDEBYSIDE) { 920 for (i = i0, j = j0; i <= i1 && j <= j1; i++, j++) 921 change(file1, f1, file2, f2, i, i, j, j, &flags); 922 923 while (i <= i1) { 924 change(file1, f1, file2, f2, i, i, j + 1, j, &flags); 925 i++; 926 } 927 928 while (j <= j1) { 929 change(file1, f1, file2, f2, i + 1, i, j, j, &flags); 930 j++; 931 } 932 } else 933 change(file1, f1, file2, f2, i0, i1, j0, j1, &flags); 934 } 935 } else { 936 for (i0 = m; i0 >= 1; i0 = i1 - 1) { 937 while (i0 >= 1 && J[i0] == J[i0 + 1] - 1 && J[i0] != 0) 938 i0--; 939 j0 = J[i0 + 1] - 1; 940 i1 = i0 + 1; 941 while (i1 > 1 && J[i1 - 1] == 0) 942 i1--; 943 j1 = J[i1 - 1] + 1; 944 J[i1] = j1; 945 change(file1, f1, file2, f2, i1, i0, j1, j0, &flags); 946 } 947 } 948 if (m == 0) 949 change(file1, f1, file2, f2, 1, 0, 1, len[1], &flags); 950 if (diff_format == D_IFDEF || diff_format == D_GFORMAT) { 951 for (;;) { 952 #define c i0 953 if ((c = getc(f1)) == EOF) 954 return; 955 printf("%c", c); 956 } 957 #undef c 958 } 959 if (anychange != 0) { 960 if (diff_format == D_CONTEXT) 961 dump_context_vec(f1, f2, flags); 962 else if (diff_format == D_UNIFIED) 963 dump_unified_vec(f1, f2, flags); 964 } 965 } 966 967 static void 968 range(int a, int b, const char *separator) 969 { 970 printf("%d", a > b ? b : a); 971 if (a < b) 972 printf("%s%d", separator, b); 973 } 974 975 static void 976 uni_range(int a, int b) 977 { 978 if (a < b) 979 printf("%d,%d", a, b - a + 1); 980 else if (a == b) 981 printf("%d", b); 982 else 983 printf("%d,0", b); 984 } 985 986 static char * 987 preadline(int fd, size_t rlen, off_t off) 988 { 989 char *line; 990 ssize_t nr; 991 992 line = xmalloc(rlen + 1); 993 if ((nr = pread(fd, line, rlen, off)) == -1) 994 err(2, "preadline"); 995 if (nr > 0 && line[nr-1] == '\n') 996 nr--; 997 line[nr] = '\0'; 998 return (line); 999 } 1000 1001 static bool 1002 ignoreline_pattern(char *line) 1003 { 1004 int ret; 1005 1006 ret = regexec(&ignore_re, line, 0, NULL, 0); 1007 return (ret == 0); /* if it matched, it should be ignored. */ 1008 } 1009 1010 static bool 1011 ignoreline(char *line, bool skip_blanks) 1012 { 1013 1014 if (skip_blanks && *line == '\0') 1015 return (true); 1016 if (ignore_pats != NULL && ignoreline_pattern(line)) 1017 return (true); 1018 return (false); 1019 } 1020 1021 /* 1022 * Indicate that there is a difference between lines a and b of the from file 1023 * to get to lines c to d of the to file. If a is greater then b then there 1024 * are no lines in the from file involved and this means that there were 1025 * lines appended (beginning at b). If c is greater than d then there are 1026 * lines missing from the to file. 1027 */ 1028 static void 1029 change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d, 1030 int *pflags) 1031 { 1032 static size_t max_context = 64; 1033 long curpos; 1034 int i, nc; 1035 const char *walk; 1036 bool skip_blanks, ignore; 1037 1038 skip_blanks = (*pflags & D_SKIPBLANKLINES); 1039 restart: 1040 if ((diff_format != D_IFDEF || diff_format == D_GFORMAT) && 1041 a > b && c > d) 1042 return; 1043 if (ignore_pats != NULL || skip_blanks) { 1044 char *line; 1045 /* 1046 * All lines in the change, insert, or delete must match an ignore 1047 * pattern for the change to be ignored. 1048 */ 1049 if (a <= b) { /* Changes and deletes. */ 1050 for (i = a; i <= b; i++) { 1051 line = preadline(fileno(f1), 1052 ixold[i] - ixold[i - 1], ixold[i - 1]); 1053 ignore = ignoreline(line, skip_blanks); 1054 free(line); 1055 if (!ignore) 1056 goto proceed; 1057 } 1058 } 1059 if (a > b || c <= d) { /* Changes and inserts. */ 1060 for (i = c; i <= d; i++) { 1061 line = preadline(fileno(f2), 1062 ixnew[i] - ixnew[i - 1], ixnew[i - 1]); 1063 ignore = ignoreline(line, skip_blanks); 1064 free(line); 1065 if (!ignore) 1066 goto proceed; 1067 } 1068 } 1069 return; 1070 } 1071 proceed: 1072 if (*pflags & D_HEADER && diff_format != D_BRIEF) { 1073 printf("%s %s %s\n", diffargs, file1, file2); 1074 *pflags &= ~D_HEADER; 1075 } 1076 if (diff_format == D_CONTEXT || diff_format == D_UNIFIED) { 1077 /* 1078 * Allocate change records as needed. 1079 */ 1080 if (context_vec_start == NULL || 1081 context_vec_ptr == context_vec_end - 1) { 1082 ptrdiff_t offset = -1; 1083 1084 if (context_vec_start != NULL) 1085 offset = context_vec_ptr - context_vec_start; 1086 max_context <<= 1; 1087 context_vec_start = xreallocarray(context_vec_start, 1088 max_context, sizeof(*context_vec_start)); 1089 context_vec_end = context_vec_start + max_context; 1090 context_vec_ptr = context_vec_start + offset; 1091 } 1092 if (anychange == 0) { 1093 /* 1094 * Print the context/unidiff header first time through. 1095 */ 1096 print_header(file1, file2); 1097 anychange = 1; 1098 } else if (a > context_vec_ptr->b + (2 * diff_context) + 1 && 1099 c > context_vec_ptr->d + (2 * diff_context) + 1) { 1100 /* 1101 * If this change is more than 'diff_context' lines from the 1102 * previous change, dump the record and reset it. 1103 */ 1104 if (diff_format == D_CONTEXT) 1105 dump_context_vec(f1, f2, *pflags); 1106 else 1107 dump_unified_vec(f1, f2, *pflags); 1108 } 1109 context_vec_ptr++; 1110 context_vec_ptr->a = a; 1111 context_vec_ptr->b = b; 1112 context_vec_ptr->c = c; 1113 context_vec_ptr->d = d; 1114 return; 1115 } 1116 if (anychange == 0) 1117 anychange = 1; 1118 switch (diff_format) { 1119 case D_BRIEF: 1120 return; 1121 case D_NORMAL: 1122 case D_EDIT: 1123 range(a, b, ","); 1124 printf("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1125 if (diff_format == D_NORMAL) 1126 range(c, d, ","); 1127 printf("\n"); 1128 break; 1129 case D_REVERSE: 1130 printf("%c", a > b ? 'a' : c > d ? 'd' : 'c'); 1131 range(a, b, " "); 1132 printf("\n"); 1133 break; 1134 case D_NREVERSE: 1135 if (a > b) 1136 printf("a%d %d\n", b, d - c + 1); 1137 else { 1138 printf("d%d %d\n", a, b - a + 1); 1139 if (!(c > d)) 1140 /* add changed lines */ 1141 printf("a%d %d\n", b, d - c + 1); 1142 } 1143 break; 1144 } 1145 if (diff_format == D_GFORMAT) { 1146 curpos = ftell(f1); 1147 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1148 nc = ixold[a > b ? b : a - 1] - curpos; 1149 for (i = 0; i < nc; i++) 1150 printf("%c", getc(f1)); 1151 for (walk = group_format; *walk != '\0'; walk++) { 1152 if (*walk == '%') { 1153 walk++; 1154 switch (*walk) { 1155 case '<': 1156 fetch(ixold, a, b, f1, '<', 1, *pflags); 1157 break; 1158 case '>': 1159 fetch(ixnew, c, d, f2, '>', 0, *pflags); 1160 break; 1161 default: 1162 printf("%%%c", *walk); 1163 break; 1164 } 1165 continue; 1166 } 1167 printf("%c", *walk); 1168 } 1169 } 1170 if (diff_format == D_SIDEBYSIDE) { 1171 if (color && a > b) 1172 printf("\033[%sm", add_code); 1173 else if (color && c > d) 1174 printf("\033[%sm", del_code); 1175 if (a > b) { 1176 print_space(0, hw + lpad, *pflags); 1177 } else { 1178 nc = fetch(ixold, a, b, f1, '\0', 1, *pflags); 1179 print_space(nc, hw - nc + lpad, *pflags); 1180 } 1181 if (color && a > b) 1182 printf("\033[%sm", add_code); 1183 else if (color && c > d) 1184 printf("\033[%sm", del_code); 1185 printf("%c", (a > b) ? '>' : ((c > d) ? '<' : '|')); 1186 if (color && c > d) 1187 printf("\033[m"); 1188 print_space(hw + lpad + 1, rpad, *pflags); 1189 fetch(ixnew, c, d, f2, '\0', 0, *pflags); 1190 printf("\n"); 1191 } 1192 if (diff_format == D_NORMAL || diff_format == D_IFDEF) { 1193 fetch(ixold, a, b, f1, '<', 1, *pflags); 1194 if (a <= b && c <= d && diff_format == D_NORMAL) 1195 printf("---\n"); 1196 } 1197 if (diff_format != D_GFORMAT && diff_format != D_SIDEBYSIDE) 1198 fetch(ixnew, c, d, f2, diff_format == D_NORMAL ? '>' : '\0', 0, *pflags); 1199 if (edoffset != 0 && diff_format == D_EDIT) { 1200 /* 1201 * A non-zero edoffset value for D_EDIT indicates that the last line 1202 * printed was a bare dot (".") that has been escaped as ".." to 1203 * prevent ed(1) from misinterpreting it. We have to add a 1204 * substitute command to change this back and restart where we left 1205 * off. 1206 */ 1207 printf(".\n"); 1208 printf("%ds/.//\n", a + edoffset - 1); 1209 b = a + edoffset - 1; 1210 a = b + 1; 1211 c += edoffset; 1212 goto restart; 1213 } 1214 if ((diff_format == D_EDIT || diff_format == D_REVERSE) && c <= d) 1215 printf(".\n"); 1216 if (inifdef) { 1217 printf("#endif /* %s */\n", ifdefname); 1218 inifdef = 0; 1219 } 1220 } 1221 1222 static int 1223 fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile, int flags) 1224 { 1225 int i, j, c, lastc, col, nc, newcol; 1226 1227 edoffset = 0; 1228 nc = 0; 1229 /* 1230 * When doing #ifdef's, copy down to current line 1231 * if this is the first file, so that stuff makes it to output. 1232 */ 1233 if ((diff_format == D_IFDEF) && oldfile) { 1234 long curpos = ftell(lb); 1235 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1236 nc = f[a > b ? b : a - 1] - curpos; 1237 for (i = 0; i < nc; i++) 1238 printf("%c", getc(lb)); 1239 } 1240 if (a > b) 1241 return (0); 1242 if (diff_format == D_IFDEF) { 1243 if (inifdef) { 1244 printf("#else /* %s%s */\n", 1245 oldfile == 1 ? "!" : "", ifdefname); 1246 } else { 1247 if (oldfile) 1248 printf("#ifndef %s\n", ifdefname); 1249 else 1250 printf("#ifdef %s\n", ifdefname); 1251 } 1252 inifdef = 1 + oldfile; 1253 } 1254 for (i = a; i <= b; i++) { 1255 fseek(lb, f[i - 1], SEEK_SET); 1256 nc = f[i] - f[i - 1]; 1257 if (diff_format == D_SIDEBYSIDE && hw < nc) 1258 nc = hw; 1259 if (diff_format != D_IFDEF && diff_format != D_GFORMAT && 1260 ch != '\0') { 1261 if (color && (ch == '>' || ch == '+')) 1262 printf("\033[%sm", add_code); 1263 else if (color && (ch == '<' || ch == '-')) 1264 printf("\033[%sm", del_code); 1265 printf("%c", ch); 1266 if (Tflag && (diff_format == D_NORMAL || 1267 diff_format == D_CONTEXT || 1268 diff_format == D_UNIFIED)) 1269 printf("\t"); 1270 else if (diff_format != D_UNIFIED) 1271 printf(" "); 1272 } 1273 col = j = 0; 1274 lastc = '\0'; 1275 while (j < nc && (hw == 0 || col < hw)) { 1276 c = getc(lb); 1277 if (flags & D_STRIPCR && c == '\r') { 1278 if ((c = getc(lb)) == '\n') 1279 j++; 1280 else { 1281 ungetc(c, lb); 1282 c = '\r'; 1283 } 1284 } 1285 if (c == EOF) { 1286 if (diff_format == D_EDIT || 1287 diff_format == D_REVERSE || 1288 diff_format == D_NREVERSE) 1289 warnx("No newline at end of file"); 1290 else 1291 printf("\n\\ No newline at end of file\n"); 1292 return (col); 1293 } 1294 if (c == '\t') { 1295 /* 1296 * Calculate where the tab would bring us. 1297 * If it would take us to the end of the 1298 * column, either clip it (if expanding 1299 * tabs) or return right away (if not). 1300 */ 1301 newcol = roundup(col + 1, tabsize); 1302 if ((flags & D_EXPANDTABS) == 0) { 1303 if (hw > 0 && newcol >= hw) 1304 return (col); 1305 printf("\t"); 1306 } else { 1307 if (hw > 0 && newcol > hw) 1308 newcol = hw; 1309 printf("%*s", newcol - col, ""); 1310 } 1311 col = newcol; 1312 } else { 1313 if (diff_format == D_EDIT && j == 1 && c == '\n' && 1314 lastc == '.') { 1315 /* 1316 * Don't print a bare "." line since that will confuse 1317 * ed(1). Print ".." instead and set the, global variable 1318 * edoffset to an offset from which to restart. The 1319 * caller must check the value of edoffset 1320 */ 1321 printf(".\n"); 1322 edoffset = i - a + 1; 1323 return (edoffset); 1324 } 1325 /* when side-by-side, do not print a newline */ 1326 if (diff_format != D_SIDEBYSIDE || c != '\n') { 1327 if (color && c == '\n') 1328 printf("\033[m%c", c); 1329 else 1330 printf("%c", c); 1331 col++; 1332 } 1333 } 1334 1335 j++; 1336 lastc = c; 1337 } 1338 } 1339 if (color && diff_format == D_SIDEBYSIDE) 1340 printf("\033[m"); 1341 return (col); 1342 } 1343 1344 /* 1345 * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578. 1346 */ 1347 static enum readhash 1348 readhash(FILE *f, int flags, unsigned *hash) 1349 { 1350 int i, t, space; 1351 unsigned sum; 1352 1353 sum = 1; 1354 space = 0; 1355 for (i = 0;;) { 1356 switch (t = getc(f)) { 1357 case '\0': 1358 if ((flags & D_FORCEASCII) == 0) 1359 return (RH_BINARY); 1360 goto hashchar; 1361 case '\r': 1362 if (flags & D_STRIPCR) { 1363 t = getc(f); 1364 if (t == '\n') 1365 break; 1366 ungetc(t, f); 1367 } 1368 /* FALLTHROUGH */ 1369 case '\t': 1370 case '\v': 1371 case '\f': 1372 case ' ': 1373 if ((flags & (D_FOLDBLANKS|D_IGNOREBLANKS)) != 0) { 1374 space++; 1375 continue; 1376 } 1377 /* FALLTHROUGH */ 1378 default: 1379 hashchar: 1380 if (space && (flags & D_IGNOREBLANKS) == 0) { 1381 i++; 1382 space = 0; 1383 } 1384 sum = sum * 127 + chrtran(t); 1385 i++; 1386 continue; 1387 case EOF: 1388 if (i == 0) 1389 return (RH_EOF); 1390 /* FALLTHROUGH */ 1391 case '\n': 1392 break; 1393 } 1394 break; 1395 } 1396 *hash = sum; 1397 return (RH_OK); 1398 } 1399 1400 static int 1401 asciifile(FILE *f) 1402 { 1403 unsigned char buf[BUFSIZ]; 1404 size_t cnt; 1405 1406 if (f == NULL) 1407 return (1); 1408 1409 rewind(f); 1410 cnt = fread(buf, 1, sizeof(buf), f); 1411 return (memchr(buf, '\0', cnt) == NULL); 1412 } 1413 1414 #define begins_with(s, pre) (strncmp(s, pre, sizeof(pre) - 1) == 0) 1415 1416 static char * 1417 match_function(const long *f, int pos, FILE *fp) 1418 { 1419 unsigned char buf[FUNCTION_CONTEXT_SIZE]; 1420 size_t nc; 1421 int last = lastline; 1422 const char *state = NULL; 1423 1424 lastline = pos; 1425 for (; pos > last; pos--) { 1426 fseek(fp, f[pos - 1], SEEK_SET); 1427 nc = f[pos] - f[pos - 1]; 1428 if (nc >= sizeof(buf)) 1429 nc = sizeof(buf) - 1; 1430 nc = fread(buf, 1, nc, fp); 1431 if (nc == 0) 1432 continue; 1433 buf[nc] = '\0'; 1434 buf[strcspn(buf, "\n")] = '\0'; 1435 if (most_recent_pat != NULL) { 1436 int ret = regexec(&most_recent_re, buf, 0, NULL, 0); 1437 1438 if (ret != 0) 1439 continue; 1440 strlcpy(lastbuf, buf, sizeof(lastbuf)); 1441 lastmatchline = pos; 1442 return (lastbuf); 1443 } else if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$' 1444 || buf[0] == '-' || buf[0] == '+') { 1445 if (begins_with(buf, "private:")) { 1446 if (!state) 1447 state = " (private)"; 1448 } else if (begins_with(buf, "protected:")) { 1449 if (!state) 1450 state = " (protected)"; 1451 } else if (begins_with(buf, "public:")) { 1452 if (!state) 1453 state = " (public)"; 1454 } else { 1455 strlcpy(lastbuf, buf, sizeof(lastbuf)); 1456 if (state) 1457 strlcat(lastbuf, state, sizeof(lastbuf)); 1458 lastmatchline = pos; 1459 return (lastbuf); 1460 } 1461 } 1462 } 1463 return (lastmatchline > 0 ? lastbuf : NULL); 1464 } 1465 1466 /* dump accumulated "context" diff changes */ 1467 static void 1468 dump_context_vec(FILE *f1, FILE *f2, int flags) 1469 { 1470 struct context_vec *cvp = context_vec_start; 1471 int lowa, upb, lowc, upd, do_output; 1472 int a, b, c, d; 1473 char ch, *f; 1474 1475 if (context_vec_start > context_vec_ptr) 1476 return; 1477 1478 b = d = 0; /* gcc */ 1479 lowa = MAX(1, cvp->a - diff_context); 1480 upb = MIN((int)len[0], context_vec_ptr->b + diff_context); 1481 lowc = MAX(1, cvp->c - diff_context); 1482 upd = MIN((int)len[1], context_vec_ptr->d + diff_context); 1483 1484 printf("***************"); 1485 if (flags & (D_PROTOTYPE | D_MATCHLAST)) { 1486 f = match_function(ixold, cvp->a - 1, f1); 1487 if (f != NULL) 1488 printf(" %s", f); 1489 } 1490 printf("\n*** "); 1491 range(lowa, upb, ","); 1492 printf(" ****\n"); 1493 1494 /* 1495 * Output changes to the "old" file. The first loop suppresses 1496 * output if there were no changes to the "old" file (we'll see 1497 * the "old" lines as context in the "new" list). 1498 */ 1499 do_output = 0; 1500 for (; cvp <= context_vec_ptr; cvp++) 1501 if (cvp->a <= cvp->b) { 1502 cvp = context_vec_start; 1503 do_output++; 1504 break; 1505 } 1506 if (do_output) { 1507 while (cvp <= context_vec_ptr) { 1508 a = cvp->a; 1509 b = cvp->b; 1510 c = cvp->c; 1511 d = cvp->d; 1512 1513 if (a <= b && c <= d) 1514 ch = 'c'; 1515 else 1516 ch = (a <= b) ? 'd' : 'a'; 1517 1518 if (ch == 'a') 1519 fetch(ixold, lowa, b, f1, ' ', 0, flags); 1520 else { 1521 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1522 fetch(ixold, a, b, f1, 1523 ch == 'c' ? '!' : '-', 0, flags); 1524 } 1525 lowa = b + 1; 1526 cvp++; 1527 } 1528 fetch(ixold, b + 1, upb, f1, ' ', 0, flags); 1529 } 1530 /* output changes to the "new" file */ 1531 printf("--- "); 1532 range(lowc, upd, ","); 1533 printf(" ----\n"); 1534 1535 do_output = 0; 1536 for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++) 1537 if (cvp->c <= cvp->d) { 1538 cvp = context_vec_start; 1539 do_output++; 1540 break; 1541 } 1542 if (do_output) { 1543 while (cvp <= context_vec_ptr) { 1544 a = cvp->a; 1545 b = cvp->b; 1546 c = cvp->c; 1547 d = cvp->d; 1548 1549 if (a <= b && c <= d) 1550 ch = 'c'; 1551 else 1552 ch = (a <= b) ? 'd' : 'a'; 1553 1554 if (ch == 'd') 1555 fetch(ixnew, lowc, d, f2, ' ', 0, flags); 1556 else { 1557 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1558 fetch(ixnew, c, d, f2, 1559 ch == 'c' ? '!' : '+', 0, flags); 1560 } 1561 lowc = d + 1; 1562 cvp++; 1563 } 1564 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1565 } 1566 context_vec_ptr = context_vec_start - 1; 1567 } 1568 1569 /* dump accumulated "unified" diff changes */ 1570 static void 1571 dump_unified_vec(FILE *f1, FILE *f2, int flags) 1572 { 1573 struct context_vec *cvp = context_vec_start; 1574 int lowa, upb, lowc, upd; 1575 int a, b, c, d; 1576 char ch, *f; 1577 1578 if (context_vec_start > context_vec_ptr) 1579 return; 1580 1581 b = d = 0; /* gcc */ 1582 lowa = MAX(1, cvp->a - diff_context); 1583 upb = MIN((int)len[0], context_vec_ptr->b + diff_context); 1584 lowc = MAX(1, cvp->c - diff_context); 1585 upd = MIN((int)len[1], context_vec_ptr->d + diff_context); 1586 1587 printf("@@ -"); 1588 uni_range(lowa, upb); 1589 printf(" +"); 1590 uni_range(lowc, upd); 1591 printf(" @@"); 1592 if (flags & (D_PROTOTYPE | D_MATCHLAST)) { 1593 f = match_function(ixold, cvp->a - 1, f1); 1594 if (f != NULL) 1595 printf(" %s", f); 1596 } 1597 printf("\n"); 1598 1599 /* 1600 * Output changes in "unified" diff format--the old and new lines 1601 * are printed together. 1602 */ 1603 for (; cvp <= context_vec_ptr; cvp++) { 1604 a = cvp->a; 1605 b = cvp->b; 1606 c = cvp->c; 1607 d = cvp->d; 1608 1609 /* 1610 * c: both new and old changes 1611 * d: only changes in the old file 1612 * a: only changes in the new file 1613 */ 1614 if (a <= b && c <= d) 1615 ch = 'c'; 1616 else 1617 ch = (a <= b) ? 'd' : 'a'; 1618 1619 switch (ch) { 1620 case 'c': 1621 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1622 fetch(ixold, a, b, f1, '-', 0, flags); 1623 fetch(ixnew, c, d, f2, '+', 0, flags); 1624 break; 1625 case 'd': 1626 fetch(ixold, lowa, a - 1, f1, ' ', 0, flags); 1627 fetch(ixold, a, b, f1, '-', 0, flags); 1628 break; 1629 case 'a': 1630 fetch(ixnew, lowc, c - 1, f2, ' ', 0, flags); 1631 fetch(ixnew, c, d, f2, '+', 0, flags); 1632 break; 1633 } 1634 lowa = b + 1; 1635 lowc = d + 1; 1636 } 1637 fetch(ixnew, d + 1, upd, f2, ' ', 0, flags); 1638 1639 context_vec_ptr = context_vec_start - 1; 1640 } 1641 1642 static void 1643 print_header(const char *file1, const char *file2) 1644 { 1645 const char *time_format; 1646 char buf[256]; 1647 struct tm tm1, tm2, *tm_ptr1, *tm_ptr2; 1648 int nsec1 = stb1.st_mtim.tv_nsec; 1649 int nsec2 = stb2.st_mtim.tv_nsec; 1650 1651 time_format = "%Y-%m-%d %H:%M:%S"; 1652 1653 if (cflag) 1654 time_format = "%c"; 1655 tm_ptr1 = localtime_r(&stb1.st_mtime, &tm1); 1656 tm_ptr2 = localtime_r(&stb2.st_mtime, &tm2); 1657 if (label[0] != NULL) 1658 printf("%s %s\n", diff_format == D_CONTEXT ? "***" : "---", 1659 label[0]); 1660 else { 1661 strftime(buf, sizeof(buf), time_format, tm_ptr1); 1662 printf("%s %s\t%s", diff_format == D_CONTEXT ? "***" : "---", 1663 file1, buf); 1664 if (!cflag) { 1665 strftime(buf, sizeof(buf), "%z", tm_ptr1); 1666 printf(".%.9d %s", nsec1, buf); 1667 } 1668 printf("\n"); 1669 } 1670 if (label[1] != NULL) 1671 printf("%s %s\n", diff_format == D_CONTEXT ? "---" : "+++", 1672 label[1]); 1673 else { 1674 strftime(buf, sizeof(buf), time_format, tm_ptr2); 1675 printf("%s %s\t%s", diff_format == D_CONTEXT ? "---" : "+++", 1676 file2, buf); 1677 if (!cflag) { 1678 strftime(buf, sizeof(buf), "%z", tm_ptr2); 1679 printf(".%.9d %s", nsec2, buf); 1680 } 1681 printf("\n"); 1682 } 1683 } 1684 1685 /* 1686 * Prints n number of space characters either by using tab 1687 * or single space characters. 1688 * nc is the preceding number of characters 1689 */ 1690 static void 1691 print_space(int nc, int n, int flags) 1692 { 1693 int col, newcol, tabstop; 1694 1695 col = nc; 1696 newcol = nc + n; 1697 /* first, use tabs if allowed */ 1698 if ((flags & D_EXPANDTABS) == 0) { 1699 while ((tabstop = roundup(col + 1, tabsize)) <= newcol) { 1700 printf("\t"); 1701 col = tabstop; 1702 } 1703 } 1704 /* finish with spaces */ 1705 printf("%*s", newcol - col, ""); 1706 } 1707