1 /* $OpenBSD: diffreg.c,v 1.61 2004/12/09 18:56:10 millert Exp $ */ 2 3 /* 4 * Copyright (C) Caldera International Inc. 2001-2002. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code and documentation must retain the above 11 * copyright notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. All advertising materials mentioning features or use of this software 16 * must display the following acknowledgement: 17 * This product includes software developed or owned by Caldera 18 * International, Inc. 19 * 4. Neither the name of Caldera International, Inc. nor the names of other 20 * contributors may be used to endorse or promote products derived from 21 * this software without specific prior written permission. 22 * 23 * USE OF THE SOFTWARE PROVIDED FOR UNDER THIS LICENSE BY CALDERA 24 * INTERNATIONAL, INC. AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR 25 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 26 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 27 * IN NO EVENT SHALL CALDERA INTERNATIONAL, INC. BE LIABLE FOR ANY DIRECT, 28 * INDIRECT INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 29 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 30 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, 32 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING 33 * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 34 * POSSIBILITY OF SUCH DAMAGE. 35 */ 36 /*- 37 * Copyright (c) 1991, 1993 38 * The Regents of the University of California. All rights reserved. 39 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 3. Neither the name of the University nor the names of its contributors 49 * may be used to endorse or promote products derived from this software 50 * without specific prior written permission. 51 * 52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 55 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 62 * SUCH DAMAGE. 63 * 64 * @(#)diffreg.c 8.1 (Berkeley) 6/6/93 65 */ 66 67 #ifndef lint 68 static const char rcsid[] = "$OpenBSD: diffreg.c,v 1.61 2004/12/09 18:56:10 millert Exp $"; 69 #endif /* not lint */ 70 71 #include <sys/param.h> 72 #include <sys/stat.h> 73 #include <sys/wait.h> 74 75 #include <ctype.h> 76 #include <err.h> 77 #include <errno.h> 78 #include <fcntl.h> 79 #include <stddef.h> 80 #include <stdio.h> 81 #include <stdlib.h> 82 #include <string.h> 83 #include <unistd.h> 84 85 #include "diff.h" 86 #include "pathnames.h" 87 88 /* 89 * diff - compare two files. 90 */ 91 92 /* 93 * Uses an algorithm due to Harold Stone, which finds 94 * a pair of longest identical subsequences in the two 95 * files. 96 * 97 * The major goal is to generate the match vector J. 98 * J[i] is the index of the line in file1 corresponding 99 * to line i file0. J[i] = 0 if there is no 100 * such line in file1. 101 * 102 * Lines are hashed so as to work in core. All potential 103 * matches are located by sorting the lines of each file 104 * on the hash (called ``value''). In particular, this 105 * collects the equivalence classes in file1 together. 106 * Subroutine equiv replaces the value of each line in 107 * file0 by the index of the first element of its 108 * matching equivalence in (the reordered) file1. 109 * To save space equiv squeezes file1 into a single 110 * array member in which the equivalence classes 111 * are simply concatenated, except that their first 112 * members are flagged by changing sign. 113 * 114 * Next the indices that point into member are unsorted into 115 * array class according to the original order of file0. 116 * 117 * The cleverness lies in routine stone. This marches 118 * through the lines of file0, developing a vector klist 119 * of "k-candidates". At step i a k-candidate is a matched 120 * pair of lines x,y (x in file0 y in file1) such that 121 * there is a common subsequence of length k 122 * between the first i lines of file0 and the first y 123 * lines of file1, but there is no such subsequence for 124 * any smaller y. x is the earliest possible mate to y 125 * that occurs in such a subsequence. 126 * 127 * Whenever any of the members of the equivalence class of 128 * lines in file1 matable to a line in file0 has serial number 129 * less than the y of some k-candidate, that k-candidate 130 * with the smallest such y is replaced. The new 131 * k-candidate is chained (via pred) to the current 132 * k-1 candidate so that the actual subsequence can 133 * be recovered. When a member has serial number greater 134 * that the y of all k-candidates, the klist is extended. 135 * At the end, the longest subsequence is pulled out 136 * and placed in the array J by unravel 137 * 138 * With J in hand, the matches there recorded are 139 * check'ed against reality to assure that no spurious 140 * matches have crept in due to hashing. If they have, 141 * they are broken, and "jackpot" is recorded--a harmless 142 * matter except that a true match for a spuriously 143 * mated line may now be unnecessarily reported as a change. 144 * 145 * Much of the complexity of the program comes simply 146 * from trying to minimize core utilization and 147 * maximize the range of doable problems by dynamically 148 * allocating what is needed and reusing what is not. 149 * The core requirements for problems larger than somewhat 150 * are (in words) 2*length(file0) + length(file1) + 151 * 3*(number of k-candidates installed), typically about 152 * 6n words for files of length n. 153 */ 154 155 struct cand { 156 int x; 157 int y; 158 int pred; 159 }; 160 161 struct line { 162 int serial; 163 int value; 164 } *file[2]; 165 166 /* 167 * The following struct is used to record change information when 168 * doing a "context" or "unified" diff. (see routine "change" to 169 * understand the highly mnemonic field names) 170 */ 171 struct context_vec { 172 int a; /* start line in old file */ 173 int b; /* end line in old file */ 174 int c; /* start line in new file */ 175 int d; /* end line in new file */ 176 }; 177 178 static int *J; /* will be overlaid on class */ 179 static int *class; /* will be overlaid on file[0] */ 180 static int *klist; /* will be overlaid on file[0] after class */ 181 static int *member; /* will be overlaid on file[1] */ 182 static int clen; 183 static int inifdef; /* whether or not we are in a #ifdef block */ 184 static int len[2]; 185 static int pref, suff; /* length of prefix and suffix */ 186 static int slen[2]; 187 static int anychange; 188 static long *ixnew; /* will be overlaid on file[1] */ 189 static long *ixold; /* will be overlaid on klist */ 190 static struct cand *clist; /* merely a free storage pot for candidates */ 191 static int clistlen; /* the length of clist */ 192 static struct line *sfile[2]; /* shortened by pruning common prefix/suffix */ 193 static u_char *chrtran; /* translation table for case-folding */ 194 static struct context_vec *context_vec_start; 195 static struct context_vec *context_vec_end; 196 static struct context_vec *context_vec_ptr; 197 198 #define FUNCTION_CONTEXT_SIZE 41 199 static char lastbuf[FUNCTION_CONTEXT_SIZE]; 200 static int lastline; 201 static int lastmatchline; 202 203 static FILE *opentemp(const char *); 204 static void output(char *, FILE *, char *, FILE *); 205 static void check(char *, FILE *, char *, FILE *); 206 static void range(int, int, char *); 207 static void uni_range(int, int); 208 static void dump_context_vec(FILE *, FILE *); 209 static void dump_unified_vec(FILE *, FILE *); 210 static void prepare(int, FILE *, off_t); 211 static void prune(void); 212 static void equiv(struct line *, int, struct line *, int, int *); 213 static void unravel(int); 214 static void unsort(struct line *, int, int *); 215 static void change(char *, FILE *, char *, FILE *, int, int, int, int); 216 static void sort(struct line *, int); 217 static void print_header(const char *, const char *); 218 static int ignoreline(char *); 219 static int asciifile(FILE *); 220 static int fetch(long *, int, int, FILE *, int, int); 221 static int newcand(int, int, int); 222 static int search(int *, int, int); 223 static int skipline(FILE *); 224 static int isqrt(int); 225 static int stone(int *, int, int *, int *); 226 static int readhash(FILE *); 227 static int files_differ(FILE *, FILE *, int); 228 static __inline int min(int, int); 229 static __inline int max(int, int); 230 static char *match_function(const long *, int, FILE *); 231 static char *preadline(int, size_t, off_t); 232 233 234 /* 235 * chrtran points to one of 2 translation tables: cup2low if folding upper to 236 * lower case clow2low if not folding case 237 */ 238 u_char clow2low[256] = { 239 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a, 240 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 241 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20, 242 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b, 243 0x2c, 0x2d, 0x2e, 0x2f, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 244 0x37, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x40, 0x41, 245 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x4a, 0x4b, 0x4c, 246 0x4d, 0x4e, 0x4f, 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, 247 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, 0x60, 0x61, 0x62, 248 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 249 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 250 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x80, 0x81, 0x82, 0x83, 251 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 252 0x8f, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 253 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 254 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, 255 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 256 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 257 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf, 0xd0, 258 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xdb, 259 0xdc, 0xdd, 0xde, 0xdf, 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 260 0xe7, 0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, 0xf0, 0xf1, 261 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 262 0xfd, 0xfe, 0xff 263 }; 264 265 u_char cup2low[256] = { 266 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a, 267 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 268 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20, 269 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b, 270 0x2c, 0x2d, 0x2e, 0x2f, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 271 0x37, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x60, 0x61, 272 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 273 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 274 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x60, 0x61, 0x62, 275 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 276 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 277 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x80, 0x81, 0x82, 0x83, 278 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 279 0x8f, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 280 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 281 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, 282 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 283 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 284 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf, 0xd0, 285 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xdb, 286 0xdc, 0xdd, 0xde, 0xdf, 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 287 0xe7, 0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, 0xf0, 0xf1, 288 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 289 0xfd, 0xfe, 0xff 290 }; 291 292 int 293 diffreg(char *ofile1, char *ofile2, int flags) 294 { 295 char *file1 = ofile1; 296 char *file2 = ofile2; 297 FILE *f1 = NULL; 298 FILE *f2 = NULL; 299 int rval = D_SAME; 300 int i, ostdout = -1; 301 pid_t pid = -1; 302 303 anychange = 0; 304 lastline = 0; 305 lastmatchline = 0; 306 context_vec_ptr = context_vec_start - 1; 307 chrtran = (iflag ? cup2low : clow2low); 308 if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode)) 309 return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2); 310 if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0) 311 goto closem; 312 313 if (flags & D_EMPTY1) 314 f1 = fopen(_PATH_DEVNULL, "r"); 315 else { 316 if (!S_ISREG(stb1.st_mode)) { 317 if ((f1 = opentemp(file1)) == NULL || 318 fstat(fileno(f1), &stb1) < 0) { 319 warn("%s", file1); 320 status |= 2; 321 goto closem; 322 } 323 } else if (strcmp(file1, "-") == 0) 324 f1 = stdin; 325 else 326 f1 = fopen(file1, "r"); 327 } 328 if (f1 == NULL) { 329 warn("%s", file1); 330 status |= 2; 331 goto closem; 332 } 333 334 if (flags & D_EMPTY2) 335 f2 = fopen(_PATH_DEVNULL, "r"); 336 else { 337 if (!S_ISREG(stb2.st_mode)) { 338 if ((f2 = opentemp(file2)) == NULL || 339 fstat(fileno(f2), &stb2) < 0) { 340 warn("%s", file2); 341 status |= 2; 342 goto closem; 343 } 344 } else if (strcmp(file2, "-") == 0) 345 f2 = stdin; 346 else 347 f2 = fopen(file2, "r"); 348 } 349 if (f2 == NULL) { 350 warn("%s", file2); 351 status |= 2; 352 goto closem; 353 } 354 355 switch (files_differ(f1, f2, flags)) { 356 case 0: 357 goto closem; 358 case 1: 359 break; 360 default: 361 /* error */ 362 status |= 2; 363 goto closem; 364 } 365 366 if (!asciifile(f1) || !asciifile(f2)) { 367 rval = D_BINARY; 368 status |= 1; 369 goto closem; 370 } 371 if (lflag) { 372 /* redirect stdout to pr */ 373 int pfd[2]; 374 char *header; 375 char *prargv[] = { "pr", "-h", NULL, "-f", NULL }; 376 377 easprintf(&header, "%s %s %s", diffargs, file1, file2); 378 prargv[2] = header; 379 fflush(stdout); 380 rewind(stdout); 381 pipe(pfd); 382 switch ((pid = fork())) { 383 case -1: 384 warnx("No more processes"); 385 status |= 2; 386 free(header); 387 return (D_ERROR); 388 case 0: 389 /* child */ 390 if (pfd[0] != STDIN_FILENO) { 391 dup2(pfd[0], STDIN_FILENO); 392 close(pfd[0]); 393 } 394 close(pfd[1]); 395 execv(_PATH_PR, prargv); 396 _exit(127); 397 default: 398 /* parent */ 399 if (pfd[1] != STDOUT_FILENO) { 400 ostdout = dup(STDOUT_FILENO); 401 dup2(pfd[1], STDOUT_FILENO); 402 close(pfd[1]); 403 } 404 close(pfd[0]); 405 rewind(stdout); 406 free(header); 407 } 408 } else if (flags & D_HEADER) 409 printf("%s %s %s\n", diffargs, file1, file2); 410 prepare(0, f1, stb1.st_size); 411 prepare(1, f2, stb2.st_size); 412 prune(); 413 sort(sfile[0], slen[0]); 414 sort(sfile[1], slen[1]); 415 416 member = (int *)file[1]; 417 equiv(sfile[0], slen[0], sfile[1], slen[1], member); 418 member = erealloc(member, (slen[1] + 2) * sizeof(int)); 419 420 class = (int *)file[0]; 421 unsort(sfile[0], slen[0], class); 422 class = erealloc(class, (slen[0] + 2) * sizeof(int)); 423 424 klist = emalloc((slen[0] + 2) * sizeof(int)); 425 clen = 0; 426 clistlen = 100; 427 clist = emalloc(clistlen * sizeof(struct cand)); 428 i = stone(class, slen[0], member, klist); 429 free(member); 430 free(class); 431 432 J = erealloc(J, (len[0] + 2) * sizeof(int)); 433 unravel(klist[i]); 434 free(clist); 435 free(klist); 436 437 ixold = erealloc(ixold, (len[0] + 2) * sizeof(long)); 438 ixnew = erealloc(ixnew, (len[1] + 2) * sizeof(long)); 439 check(file1, f1, file2, f2); 440 output(file1, f1, file2, f2); 441 if (ostdout != -1) { 442 int wstatus; 443 444 /* close the pipe to pr and restore stdout */ 445 fflush(stdout); 446 rewind(stdout); 447 if (ostdout != STDOUT_FILENO) { 448 close(STDOUT_FILENO); 449 dup2(ostdout, STDOUT_FILENO); 450 close(ostdout); 451 } 452 waitpid(pid, &wstatus, 0); 453 } 454 closem: 455 if (anychange) { 456 status |= 1; 457 if (rval == D_SAME) 458 rval = D_DIFFER; 459 } 460 if (f1 != NULL) 461 fclose(f1); 462 if (f2 != NULL) 463 fclose(f2); 464 if (file1 != ofile1) 465 free(file1); 466 if (file2 != ofile2) 467 free(file2); 468 return (rval); 469 } 470 471 /* 472 * Check to see if the given files differ. 473 * Returns 0 if they are the same, 1 if different, and -1 on error. 474 * XXX - could use code from cmp(1) [faster] 475 */ 476 static int 477 files_differ(FILE *f1, FILE *f2, int flags) 478 { 479 char buf1[BUFSIZ], buf2[BUFSIZ]; 480 size_t i, j; 481 482 if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size || 483 (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT)) 484 return (1); 485 for (;;) { 486 i = fread(buf1, 1, sizeof(buf1), f1); 487 j = fread(buf2, 1, sizeof(buf2), f2); 488 if (i != j) 489 return (1); 490 if (i == 0 && j == 0) { 491 if (ferror(f1) || ferror(f2)) 492 return (1); 493 return (0); 494 } 495 if (memcmp(buf1, buf2, i) != 0) 496 return (1); 497 } 498 } 499 500 static FILE * 501 opentemp(const char *file) 502 { 503 char buf[BUFSIZ], *tempdir, tempfile[MAXPATHLEN]; 504 ssize_t nread; 505 int ifd, ofd; 506 507 if (strcmp(file, "-") == 0) 508 ifd = STDIN_FILENO; 509 else if ((ifd = open(file, O_RDONLY, 0644)) < 0) 510 return (NULL); 511 512 if ((tempdir = getenv("TMPDIR")) == NULL) 513 tempdir = _PATH_TMP; 514 if (snprintf(tempfile, sizeof(tempfile), "%s/diff.XXXXXXXX", 515 tempdir) >= sizeof(tempfile)) { 516 close(ifd); 517 errno = ENAMETOOLONG; 518 return (NULL); 519 } 520 521 if ((ofd = mkstemp(tempfile)) < 0) 522 return (NULL); 523 unlink(tempfile); 524 while ((nread = read(ifd, buf, BUFSIZ)) > 0) { 525 if (write(ofd, buf, nread) != nread) { 526 close(ifd); 527 close(ofd); 528 return (NULL); 529 } 530 } 531 close(ifd); 532 lseek(ofd, (off_t)0, SEEK_SET); 533 return (fdopen(ofd, "r")); 534 } 535 536 char * 537 splice(char *dir, char *file) 538 { 539 char *tail, *buf; 540 541 if ((tail = strrchr(file, '/')) == NULL) 542 tail = file; 543 else 544 tail++; 545 easprintf(&buf, "%s/%s", dir, tail); 546 return (buf); 547 } 548 549 static void 550 prepare(int i, FILE *fd, off_t filesize) 551 { 552 struct line *p; 553 int j, h; 554 size_t sz; 555 556 rewind(fd); 557 558 sz = (filesize <= SIZE_MAX ? filesize : SIZE_MAX) / 25; 559 if (sz < 100) 560 sz = 100; 561 562 p = emalloc((sz + 3) * sizeof(struct line)); 563 for (j = 0; (h = readhash(fd));) { 564 if (j == sz) { 565 sz = sz * 3 / 2; 566 p = erealloc(p, (sz + 3) * sizeof(struct line)); 567 } 568 p[++j].value = h; 569 } 570 len[i] = j; 571 file[i] = p; 572 } 573 574 static void 575 prune(void) 576 { 577 int i, j; 578 579 for (pref = 0; pref < len[0] && pref < len[1] && 580 file[0][pref + 1].value == file[1][pref + 1].value; 581 pref++) 582 ; 583 for (suff = 0; suff < len[0] - pref && suff < len[1] - pref && 584 file[0][len[0] - suff].value == file[1][len[1] - suff].value; 585 suff++) 586 ; 587 for (j = 0; j < 2; j++) { 588 sfile[j] = file[j] + pref; 589 slen[j] = len[j] - pref - suff; 590 for (i = 0; i <= slen[j]; i++) 591 sfile[j][i].serial = i; 592 } 593 } 594 595 static void 596 equiv(struct line *a, int n, struct line *b, int m, int *c) 597 { 598 int i, j; 599 600 i = j = 1; 601 while (i <= n && j <= m) { 602 if (a[i].value < b[j].value) 603 a[i++].value = 0; 604 else if (a[i].value == b[j].value) 605 a[i++].value = j; 606 else 607 j++; 608 } 609 while (i <= n) 610 a[i++].value = 0; 611 b[m + 1].value = 0; 612 j = 0; 613 while (++j <= m) { 614 c[j] = -b[j].serial; 615 while (b[j + 1].value == b[j].value) { 616 j++; 617 c[j] = b[j].serial; 618 } 619 } 620 c[j] = -1; 621 } 622 623 /* Code taken from ping.c */ 624 static int 625 isqrt(int n) 626 { 627 int y, x = 1; 628 629 if (n == 0) 630 return(0); 631 632 do { /* newton was a stinker */ 633 y = x; 634 x = n / x; 635 x += y; 636 x /= 2; 637 } while ((x - y) > 1 || (x - y) < -1); 638 639 return (x); 640 } 641 642 static int 643 stone(int *a, int n, int *b, int *c) 644 { 645 int i, k, y, j, l; 646 int oldc, tc, oldl; 647 u_int numtries; 648 649 const u_int bound = dflag ? UINT_MAX : max(256, isqrt(n)); 650 651 k = 0; 652 c[0] = newcand(0, 0, 0); 653 for (i = 1; i <= n; i++) { 654 j = a[i]; 655 if (j == 0) 656 continue; 657 y = -b[j]; 658 oldl = 0; 659 oldc = c[0]; 660 numtries = 0; 661 do { 662 if (y <= clist[oldc].y) 663 continue; 664 l = search(c, k, y); 665 if (l != oldl + 1) 666 oldc = c[l - 1]; 667 if (l <= k) { 668 if (clist[c[l]].y <= y) 669 continue; 670 tc = c[l]; 671 c[l] = newcand(i, y, oldc); 672 oldc = tc; 673 oldl = l; 674 numtries++; 675 } else { 676 c[l] = newcand(i, y, oldc); 677 k++; 678 break; 679 } 680 } while ((y = b[++j]) > 0 && numtries < bound); 681 } 682 return (k); 683 } 684 685 static int 686 newcand(int x, int y, int pred) 687 { 688 struct cand *q; 689 690 if (clen == clistlen) { 691 clistlen = clistlen * 11 / 10; 692 clist = erealloc(clist, clistlen * sizeof(struct cand)); 693 } 694 q = clist + clen; 695 q->x = x; 696 q->y = y; 697 q->pred = pred; 698 return (clen++); 699 } 700 701 static int 702 search(int *c, int k, int y) 703 { 704 int i, j, l, t; 705 706 if (clist[c[k]].y < y) /* quick look for typical case */ 707 return (k + 1); 708 i = 0; 709 j = k + 1; 710 while (1) { 711 l = i + j; 712 if ((l >>= 1) <= i) 713 break; 714 t = clist[c[l]].y; 715 if (t > y) 716 j = l; 717 else if (t < y) 718 i = l; 719 else 720 return (l); 721 } 722 return (l + 1); 723 } 724 725 static void 726 unravel(int p) 727 { 728 struct cand *q; 729 int i; 730 731 for (i = 0; i <= len[0]; i++) 732 J[i] = i <= pref ? i : 733 i > len[0] - suff ? i + len[1] - len[0] : 0; 734 for (q = clist + p; q->y != 0; q = clist + q->pred) 735 J[q->x + pref] = q->y + pref; 736 } 737 738 /* 739 * Check does double duty: 740 * 1. ferret out any fortuitous correspondences due 741 * to confounding by hashing (which result in "jackpot") 742 * 2. collect random access indexes to the two files 743 */ 744 static void 745 check(char *file1, FILE *f1, char *file2, FILE *f2) 746 { 747 int i, j, jackpot, c, d; 748 long ctold, ctnew; 749 750 rewind(f1); 751 rewind(f2); 752 j = 1; 753 ixold[0] = ixnew[0] = 0; 754 jackpot = 0; 755 ctold = ctnew = 0; 756 for (i = 1; i <= len[0]; i++) { 757 if (J[i] == 0) { 758 ixold[i] = ctold += skipline(f1); 759 continue; 760 } 761 while (j < J[i]) { 762 ixnew[j] = ctnew += skipline(f2); 763 j++; 764 } 765 if (bflag || wflag || iflag) { 766 for (;;) { 767 c = getc(f1); 768 d = getc(f2); 769 /* 770 * GNU diff ignores a missing newline 771 * in one file if bflag || wflag. 772 */ 773 if ((bflag || wflag) && 774 ((c == EOF && d == '\n') || 775 (c == '\n' && d == EOF))) { 776 break; 777 } 778 ctold++; 779 ctnew++; 780 if (bflag && isspace(c) && isspace(d)) { 781 do { 782 if (c == '\n') 783 break; 784 ctold++; 785 } while (isspace(c = getc(f1))); 786 do { 787 if (d == '\n') 788 break; 789 ctnew++; 790 } while (isspace(d = getc(f2))); 791 } else if (wflag) { 792 while (isspace(c) && c != '\n') { 793 c = getc(f1); 794 ctold++; 795 } 796 while (isspace(d) && d != '\n') { 797 d = getc(f2); 798 ctnew++; 799 } 800 } 801 if (chrtran[c] != chrtran[d]) { 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 } else { 814 for (;;) { 815 ctold++; 816 ctnew++; 817 if ((c = getc(f1)) != (d = getc(f2))) { 818 /* jackpot++; */ 819 J[i] = 0; 820 if (c != '\n' && c != EOF) 821 ctold += skipline(f1); 822 if (d != '\n' && c != EOF) 823 ctnew += skipline(f2); 824 break; 825 } 826 if (c == '\n' || c == EOF) 827 break; 828 } 829 } 830 ixold[i] = ctold; 831 ixnew[j] = ctnew; 832 j++; 833 } 834 for (; j <= len[1]; j++) 835 ixnew[j] = ctnew += skipline(f2); 836 /* 837 * if (jackpot) 838 * fprintf(stderr, "jackpot\n"); 839 */ 840 } 841 842 /* shellsort CACM #201 */ 843 static void 844 sort(struct line *a, int n) 845 { 846 struct line *ai, *aim, w; 847 int j, m = 0, k; 848 849 if (n == 0) 850 return; 851 for (j = 1; j <= n; j *= 2) 852 m = 2 * j - 1; 853 for (m /= 2; m != 0; m /= 2) { 854 k = n - m; 855 for (j = 1; j <= k; j++) { 856 for (ai = &a[j]; ai > a; ai -= m) { 857 aim = &ai[m]; 858 if (aim < ai) 859 break; /* wraparound */ 860 if (aim->value > ai[0].value || 861 (aim->value == ai[0].value && 862 aim->serial > ai[0].serial)) 863 break; 864 w.value = ai[0].value; 865 ai[0].value = aim->value; 866 aim->value = w.value; 867 w.serial = ai[0].serial; 868 ai[0].serial = aim->serial; 869 aim->serial = w.serial; 870 } 871 } 872 } 873 } 874 875 static void 876 unsort(struct line *f, int l, int *b) 877 { 878 int *a, i; 879 880 a = emalloc((l + 1) * sizeof(int)); 881 for (i = 1; i <= l; i++) 882 a[f[i].serial] = f[i].value; 883 for (i = 1; i <= l; i++) 884 b[i] = a[i]; 885 free(a); 886 } 887 888 static int 889 skipline(FILE *f) 890 { 891 int i, c; 892 893 for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++) 894 continue; 895 return (i); 896 } 897 898 static void 899 output(char *file1, FILE *f1, char *file2, FILE *f2) 900 { 901 int m, i0, i1, j0, j1; 902 903 rewind(f1); 904 rewind(f2); 905 m = len[0]; 906 J[0] = 0; 907 J[m + 1] = len[1] + 1; 908 if (format != D_EDIT) { 909 for (i0 = 1; i0 <= m; i0 = i1 + 1) { 910 while (i0 <= m && J[i0] == J[i0 - 1] + 1) 911 i0++; 912 j0 = J[i0 - 1] + 1; 913 i1 = i0 - 1; 914 while (i1 < m && J[i1 + 1] == 0) 915 i1++; 916 j1 = J[i1 + 1] - 1; 917 J[i1] = j1; 918 change(file1, f1, file2, f2, i0, i1, j0, j1); 919 } 920 } else { 921 for (i0 = m; i0 >= 1; i0 = i1 - 1) { 922 while (i0 >= 1 && J[i0] == J[i0 + 1] - 1 && J[i0] != 0) 923 i0--; 924 j0 = J[i0 + 1] - 1; 925 i1 = i0 + 1; 926 while (i1 > 1 && J[i1 - 1] == 0) 927 i1--; 928 j1 = J[i1 - 1] + 1; 929 J[i1] = j1; 930 change(file1, f1, file2, f2, i1, i0, j1, j0); 931 } 932 } 933 if (m == 0) 934 change(file1, f1, file2, f2, 1, 0, 1, len[1]); 935 if (format == D_IFDEF) { 936 for (;;) { 937 #define c i0 938 if ((c = getc(f1)) == EOF) 939 return; 940 putchar(c); 941 } 942 #undef c 943 } 944 if (anychange != 0) { 945 if (format == D_CONTEXT) 946 dump_context_vec(f1, f2); 947 else if (format == D_UNIFIED) 948 dump_unified_vec(f1, f2); 949 } 950 } 951 952 static __inline void 953 range(int a, int b, char *separator) 954 { 955 printf("%d", a > b ? b : a); 956 if (a < b) 957 printf("%s%d", separator, b); 958 } 959 960 static __inline void 961 uni_range(int a, int b) 962 { 963 if (a < b) 964 printf("%d,%d", a, b - a + 1); 965 else if (a == b) 966 printf("%d", b); 967 else 968 printf("%d,0", b); 969 } 970 971 static char * 972 preadline(int fd, size_t len, off_t off) 973 { 974 char *line; 975 ssize_t nr; 976 977 line = emalloc(len + 1); 978 if ((nr = pread(fd, line, len, off)) < 0) 979 err(1, "preadline"); 980 line[nr] = '\0'; 981 return (line); 982 } 983 984 static int 985 ignoreline(char *line) 986 { 987 int ret; 988 989 ret = regexec(&ignore_re, line, 0, NULL, 0); 990 free(line); 991 return (ret == 0); /* if it matched, it should be ignored. */ 992 } 993 994 /* 995 * Indicate that there is a difference between lines a and b of the from file 996 * to get to lines c to d of the to file. If a is greater then b then there 997 * are no lines in the from file involved and this means that there were 998 * lines appended (beginning at b). If c is greater than d then there are 999 * lines missing from the to file. 1000 */ 1001 static void 1002 change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d) 1003 { 1004 static size_t max_context = 64; 1005 int i; 1006 1007 restart: 1008 if (format != D_IFDEF && a > b && c > d) 1009 return; 1010 if (ignore_pats != NULL) { 1011 char *line; 1012 /* 1013 * All lines in the change, insert, or delete must 1014 * match an ignore pattern for the change to be 1015 * ignored. 1016 */ 1017 if (a <= b) { /* Changes and deletes. */ 1018 for (i = a; i <= b; i++) { 1019 line = preadline(fileno(f1), 1020 ixold[i] - ixold[i - 1], ixold[i - 1]); 1021 if (!ignoreline(line)) 1022 goto proceed; 1023 } 1024 } 1025 if (a > b || c <= d) { /* Changes and inserts. */ 1026 for (i = c; i <= d; i++) { 1027 line = preadline(fileno(f2), 1028 ixnew[i] - ixnew[i - 1], ixnew[i - 1]); 1029 if (!ignoreline(line)) 1030 goto proceed; 1031 } 1032 } 1033 return; 1034 } 1035 proceed: 1036 if (format == D_CONTEXT || format == D_UNIFIED) { 1037 /* 1038 * Allocate change records as needed. 1039 */ 1040 if (context_vec_ptr == context_vec_end - 1) { 1041 ptrdiff_t offset = context_vec_ptr - context_vec_start; 1042 max_context <<= 1; 1043 context_vec_start = erealloc(context_vec_start, 1044 max_context * sizeof(struct context_vec)); 1045 context_vec_end = context_vec_start + max_context; 1046 context_vec_ptr = context_vec_start + offset; 1047 } 1048 if (anychange == 0) { 1049 /* 1050 * Print the context/unidiff header first time through. 1051 */ 1052 print_header(file1, file2); 1053 anychange = 1; 1054 } else if (a > context_vec_ptr->b + (2 * context) + 1 && 1055 c > context_vec_ptr->d + (2 * context) + 1) { 1056 /* 1057 * If this change is more than 'context' lines from the 1058 * previous change, dump the record and reset it. 1059 */ 1060 if (format == D_CONTEXT) 1061 dump_context_vec(f1, f2); 1062 else 1063 dump_unified_vec(f1, f2); 1064 } 1065 context_vec_ptr++; 1066 context_vec_ptr->a = a; 1067 context_vec_ptr->b = b; 1068 context_vec_ptr->c = c; 1069 context_vec_ptr->d = d; 1070 return; 1071 } 1072 if (anychange == 0) 1073 anychange = 1; 1074 switch (format) { 1075 1076 case D_BRIEF: 1077 return; 1078 case D_NORMAL: 1079 case D_EDIT: 1080 range(a, b, ","); 1081 putchar(a > b ? 'a' : c > d ? 'd' : 'c'); 1082 if (format == D_NORMAL) 1083 range(c, d, ","); 1084 putchar('\n'); 1085 break; 1086 case D_REVERSE: 1087 putchar(a > b ? 'a' : c > d ? 'd' : 'c'); 1088 range(a, b, " "); 1089 putchar('\n'); 1090 break; 1091 case D_NREVERSE: 1092 if (a > b) 1093 printf("a%d %d\n", b, d - c + 1); 1094 else { 1095 printf("d%d %d\n", a, b - a + 1); 1096 if (!(c > d)) 1097 /* add changed lines */ 1098 printf("a%d %d\n", b, d - c + 1); 1099 } 1100 break; 1101 } 1102 if (format == D_NORMAL || format == D_IFDEF) { 1103 fetch(ixold, a, b, f1, '<', 1); 1104 if (a <= b && c <= d && format == D_NORMAL) 1105 puts("---"); 1106 } 1107 i = fetch(ixnew, c, d, f2, format == D_NORMAL ? '>' : '\0', 0); 1108 if (i != 0 && format == D_EDIT) { 1109 /* 1110 * A non-zero return value for D_EDIT indicates that the 1111 * last line printed was a bare dot (".") that has been 1112 * escaped as ".." to prevent ed(1) from misinterpreting 1113 * it. We have to add a substitute command to change this 1114 * back and restart where we left off. 1115 */ 1116 puts("."); 1117 printf("%ds/^\\.\\././\n", a); 1118 a += i; 1119 c += i; 1120 goto restart; 1121 } 1122 if ((format == D_EDIT || format == D_REVERSE) && c <= d) 1123 puts("."); 1124 if (inifdef) { 1125 printf("#endif /* %s */\n", ifdefname); 1126 inifdef = 0; 1127 } 1128 } 1129 1130 static int 1131 fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile) 1132 { 1133 int i, j, c, lastc, col, nc; 1134 1135 /* 1136 * When doing #ifdef's, copy down to current line 1137 * if this is the first file, so that stuff makes it to output. 1138 */ 1139 if (format == D_IFDEF && oldfile) { 1140 long curpos = ftell(lb); 1141 /* print through if append (a>b), else to (nb: 0 vs 1 orig) */ 1142 nc = f[a > b ? b : a - 1] - curpos; 1143 for (i = 0; i < nc; i++) 1144 putchar(getc(lb)); 1145 } 1146 if (a > b) 1147 return (0); 1148 if (format == D_IFDEF) { 1149 if (inifdef) { 1150 printf("#else /* %s%s */\n", 1151 oldfile == 1 ? "!" : "", ifdefname); 1152 } else { 1153 if (oldfile) 1154 printf("#ifndef %s\n", ifdefname); 1155 else 1156 printf("#ifdef %s\n", ifdefname); 1157 } 1158 inifdef = 1 + oldfile; 1159 } 1160 for (i = a; i <= b; i++) { 1161 fseek(lb, f[i - 1], SEEK_SET); 1162 nc = f[i] - f[i - 1]; 1163 if (format != D_IFDEF && ch != '\0') { 1164 putchar(ch); 1165 if (Tflag && (format == D_NORMAL || format == D_CONTEXT 1166 || format == D_UNIFIED)) 1167 putchar('\t'); 1168 else if (format != D_UNIFIED) 1169 putchar(' '); 1170 } 1171 col = 0; 1172 for (j = 0, lastc = '\0'; j < nc; j++, lastc = c) { 1173 if ((c = getc(lb)) == EOF) { 1174 if (format == D_EDIT || format == D_REVERSE || 1175 format == D_NREVERSE) 1176 warnx("No newline at end of file"); 1177 else 1178 puts("\n\\ No newline at end of file"); 1179 return (0); 1180 } 1181 if (c == '\t' && tflag) { 1182 do { 1183 putchar(' '); 1184 } while (++col & 7); 1185 } else { 1186 if (format == D_EDIT && j == 1 && c == '\n' 1187 && lastc == '.') { 1188 /* 1189 * Don't print a bare "." line 1190 * since that will confuse ed(1). 1191 * Print ".." instead and return, 1192 * giving the caller an offset 1193 * from which to restart. 1194 */ 1195 puts("."); 1196 return (i - a + 1); 1197 } 1198 putchar(c); 1199 col++; 1200 } 1201 } 1202 } 1203 return (0); 1204 } 1205 1206 /* 1207 * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578. 1208 */ 1209 static int 1210 readhash(FILE *f) 1211 { 1212 int i, t, space; 1213 int sum; 1214 1215 sum = 1; 1216 space = 0; 1217 if (!bflag && !wflag) { 1218 if (iflag) 1219 for (i = 0; (t = getc(f)) != '\n'; i++) { 1220 if (t == EOF) { 1221 if (i == 0) 1222 return (0); 1223 break; 1224 } 1225 sum = sum * 127 + chrtran[t]; 1226 } 1227 else 1228 for (i = 0; (t = getc(f)) != '\n'; i++) { 1229 if (t == EOF) { 1230 if (i == 0) 1231 return (0); 1232 break; 1233 } 1234 sum = sum * 127 + t; 1235 } 1236 } else { 1237 for (i = 0;;) { 1238 switch (t = getc(f)) { 1239 case '\t': 1240 case ' ': 1241 space++; 1242 continue; 1243 default: 1244 if (space && !wflag) { 1245 i++; 1246 space = 0; 1247 } 1248 sum = sum * 127 + chrtran[t]; 1249 i++; 1250 continue; 1251 case EOF: 1252 if (i == 0) 1253 return (0); 1254 /* FALLTHROUGH */ 1255 case '\n': 1256 break; 1257 } 1258 break; 1259 } 1260 } 1261 /* 1262 * There is a remote possibility that we end up with a zero sum. 1263 * Zero is used as an EOF marker, so return 1 instead. 1264 */ 1265 return (sum == 0 ? 1 : sum); 1266 } 1267 1268 static int 1269 asciifile(FILE *f) 1270 { 1271 unsigned char buf[BUFSIZ]; 1272 int i, cnt; 1273 1274 if (aflag || f == NULL) 1275 return (1); 1276 1277 rewind(f); 1278 cnt = fread(buf, 1, sizeof(buf), f); 1279 for (i = 0; i < cnt; i++) 1280 if (!isprint(buf[i]) && !isspace(buf[i])) 1281 return (0); 1282 return (1); 1283 } 1284 1285 static __inline int min(int a, int b) 1286 { 1287 return (a < b ? a : b); 1288 } 1289 1290 static __inline int max(int a, int b) 1291 { 1292 return (a > b ? a : b); 1293 } 1294 1295 static char * 1296 match_function(const long *f, int pos, FILE *file) 1297 { 1298 unsigned char buf[FUNCTION_CONTEXT_SIZE]; 1299 size_t nc; 1300 int last = lastline; 1301 char *p; 1302 1303 lastline = pos; 1304 while (pos > last) { 1305 fseek(file, f[pos - 1], SEEK_SET); 1306 nc = f[pos] - f[pos - 1]; 1307 if (nc >= sizeof(buf)) 1308 nc = sizeof(buf) - 1; 1309 nc = fread(buf, 1, nc, file); 1310 if (nc > 0) { 1311 buf[nc] = '\0'; 1312 p = strchr(buf, '\n'); 1313 if (p != NULL) 1314 *p = '\0'; 1315 if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$') { 1316 strlcpy(lastbuf, buf, sizeof lastbuf); 1317 lastmatchline = pos; 1318 return lastbuf; 1319 } 1320 } 1321 pos--; 1322 } 1323 return lastmatchline > 0 ? lastbuf : NULL; 1324 } 1325 1326 /* dump accumulated "context" diff changes */ 1327 static void 1328 dump_context_vec(FILE *f1, FILE *f2) 1329 { 1330 struct context_vec *cvp = context_vec_start; 1331 int lowa, upb, lowc, upd, do_output; 1332 int a, b, c, d; 1333 char ch, *f; 1334 1335 if (context_vec_start > context_vec_ptr) 1336 return; 1337 1338 b = d = 0; /* gcc */ 1339 lowa = max(1, cvp->a - context); 1340 upb = min(len[0], context_vec_ptr->b + context); 1341 lowc = max(1, cvp->c - context); 1342 upd = min(len[1], context_vec_ptr->d + context); 1343 1344 printf("***************"); 1345 if (pflag) { 1346 f = match_function(ixold, lowa-1, f1); 1347 if (f != NULL) { 1348 putchar(' '); 1349 fputs(f, stdout); 1350 } 1351 } 1352 printf("\n*** "); 1353 range(lowa, upb, ","); 1354 printf(" ****\n"); 1355 1356 /* 1357 * Output changes to the "old" file. The first loop suppresses 1358 * output if there were no changes to the "old" file (we'll see 1359 * the "old" lines as context in the "new" list). 1360 */ 1361 do_output = 0; 1362 for (; cvp <= context_vec_ptr; cvp++) 1363 if (cvp->a <= cvp->b) { 1364 cvp = context_vec_start; 1365 do_output++; 1366 break; 1367 } 1368 if (do_output) { 1369 while (cvp <= context_vec_ptr) { 1370 a = cvp->a; 1371 b = cvp->b; 1372 c = cvp->c; 1373 d = cvp->d; 1374 1375 if (a <= b && c <= d) 1376 ch = 'c'; 1377 else 1378 ch = (a <= b) ? 'd' : 'a'; 1379 1380 if (ch == 'a') 1381 fetch(ixold, lowa, b, f1, ' ', 0); 1382 else { 1383 fetch(ixold, lowa, a - 1, f1, ' ', 0); 1384 fetch(ixold, a, b, f1, 1385 ch == 'c' ? '!' : '-', 0); 1386 } 1387 lowa = b + 1; 1388 cvp++; 1389 } 1390 fetch(ixold, b + 1, upb, f1, ' ', 0); 1391 } 1392 /* output changes to the "new" file */ 1393 printf("--- "); 1394 range(lowc, upd, ","); 1395 printf(" ----\n"); 1396 1397 do_output = 0; 1398 for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++) 1399 if (cvp->c <= cvp->d) { 1400 cvp = context_vec_start; 1401 do_output++; 1402 break; 1403 } 1404 if (do_output) { 1405 while (cvp <= context_vec_ptr) { 1406 a = cvp->a; 1407 b = cvp->b; 1408 c = cvp->c; 1409 d = cvp->d; 1410 1411 if (a <= b && c <= d) 1412 ch = 'c'; 1413 else 1414 ch = (a <= b) ? 'd' : 'a'; 1415 1416 if (ch == 'd') 1417 fetch(ixnew, lowc, d, f2, ' ', 0); 1418 else { 1419 fetch(ixnew, lowc, c - 1, f2, ' ', 0); 1420 fetch(ixnew, c, d, f2, 1421 ch == 'c' ? '!' : '+', 0); 1422 } 1423 lowc = d + 1; 1424 cvp++; 1425 } 1426 fetch(ixnew, d + 1, upd, f2, ' ', 0); 1427 } 1428 context_vec_ptr = context_vec_start - 1; 1429 } 1430 1431 /* dump accumulated "unified" diff changes */ 1432 static void 1433 dump_unified_vec(FILE *f1, FILE *f2) 1434 { 1435 struct context_vec *cvp = context_vec_start; 1436 int lowa, upb, lowc, upd; 1437 int a, b, c, d; 1438 char ch, *f; 1439 1440 if (context_vec_start > context_vec_ptr) 1441 return; 1442 1443 b = d = 0; /* gcc */ 1444 lowa = max(1, cvp->a - context); 1445 upb = min(len[0], context_vec_ptr->b + context); 1446 lowc = max(1, cvp->c - context); 1447 upd = min(len[1], context_vec_ptr->d + context); 1448 1449 fputs("@@ -", stdout); 1450 uni_range(lowa, upb); 1451 fputs(" +", stdout); 1452 uni_range(lowc, upd); 1453 fputs(" @@", stdout); 1454 if (pflag) { 1455 f = match_function(ixold, lowa-1, f1); 1456 if (f != NULL) { 1457 putchar(' '); 1458 fputs(f, stdout); 1459 } 1460 } 1461 putchar('\n'); 1462 1463 /* 1464 * Output changes in "unified" diff format--the old and new lines 1465 * are printed together. 1466 */ 1467 for (; cvp <= context_vec_ptr; cvp++) { 1468 a = cvp->a; 1469 b = cvp->b; 1470 c = cvp->c; 1471 d = cvp->d; 1472 1473 /* 1474 * c: both new and old changes 1475 * d: only changes in the old file 1476 * a: only changes in the new file 1477 */ 1478 if (a <= b && c <= d) 1479 ch = 'c'; 1480 else 1481 ch = (a <= b) ? 'd' : 'a'; 1482 1483 switch (ch) { 1484 case 'c': 1485 fetch(ixold, lowa, a - 1, f1, ' ', 0); 1486 fetch(ixold, a, b, f1, '-', 0); 1487 fetch(ixnew, c, d, f2, '+', 0); 1488 break; 1489 case 'd': 1490 fetch(ixold, lowa, a - 1, f1, ' ', 0); 1491 fetch(ixold, a, b, f1, '-', 0); 1492 break; 1493 case 'a': 1494 fetch(ixnew, lowc, c - 1, f2, ' ', 0); 1495 fetch(ixnew, c, d, f2, '+', 0); 1496 break; 1497 } 1498 lowa = b + 1; 1499 lowc = d + 1; 1500 } 1501 fetch(ixnew, d + 1, upd, f2, ' ', 0); 1502 1503 context_vec_ptr = context_vec_start - 1; 1504 } 1505 1506 static void 1507 print_header(const char *file1, const char *file2) 1508 { 1509 if (label[0] != NULL) 1510 printf("%s %s\n", format == D_CONTEXT ? "***" : "---", 1511 label[0]); 1512 else 1513 printf("%s %s\t%s", format == D_CONTEXT ? "***" : "---", 1514 file1, ctime(&stb1.st_mtime)); 1515 if (label[1] != NULL) 1516 printf("%s %s\n", format == D_CONTEXT ? "---" : "+++", 1517 label[1]); 1518 else 1519 printf("%s %s\t%s", format == D_CONTEXT ? "---" : "+++", 1520 file2, ctime(&stb2.st_mtime)); 1521 } 1522