Annotation of src/usr.bin/cvs/diff.c, Revision 1.12
1.12 ! djm 1: /* $OpenBSD: diff.c,v 1.11 2004/12/07 17:10:56 tedu Exp $ */
1.1 jfb 2: /*
3: * Copyright (C) Caldera International Inc. 2001-2002.
4: * All rights reserved.
5: *
6: * Redistribution and use in source and binary forms, with or without
7: * modification, are permitted provided that the following conditions
8: * are met:
9: * 1. Redistributions of source code and documentation must retain the above
10: * copyright notice, this list of conditions and the following disclaimer.
11: * 2. Redistributions in binary form must reproduce the above copyright
12: * notice, this list of conditions and the following disclaimer in the
13: * documentation and/or other materials provided with the distribution.
14: * 3. All advertising materials mentioning features or use of this software
15: * must display the following acknowledgement:
16: * This product includes software developed or owned by Caldera
17: * International, Inc.
18: * 4. Neither the name of Caldera International, Inc. nor the names of other
19: * contributors may be used to endorse or promote products derived from
20: * this software without specific prior written permission.
21: *
22: * USE OF THE SOFTWARE PROVIDED FOR UNDER THIS LICENSE BY CALDERA
23: * INTERNATIONAL, INC. AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR
24: * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
25: * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
26: * IN NO EVENT SHALL CALDERA INTERNATIONAL, INC. BE LIABLE FOR ANY DIRECT,
27: * INDIRECT INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
28: * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
29: * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
31: * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
32: * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
33: * POSSIBILITY OF SUCH DAMAGE.
34: */
35: /*-
36: * Copyright (c) 1991, 1993
37: * The Regents of the University of California. All rights reserved.
38: * Copyright (c) 2004 Jean-Francois Brousseau. 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: * Uses an algorithm due to Harold Stone, which finds
68: * a pair of longest identical subsequences in the two
69: * files.
70: *
71: * The major goal is to generate the match vector J.
72: * J[i] is the index of the line in file1 corresponding
73: * to line i file0. J[i] = 0 if there is no
74: * such line in file1.
75: *
76: * Lines are hashed so as to work in core. All potential
77: * matches are located by sorting the lines of each file
78: * on the hash (called ``value''). In particular, this
79: * collects the equivalence classes in file1 together.
80: * Subroutine equiv replaces the value of each line in
81: * file0 by the index of the first element of its
82: * matching equivalence in (the reordered) file1.
83: * To save space equiv squeezes file1 into a single
84: * array member in which the equivalence classes
85: * are simply concatenated, except that their first
86: * members are flagged by changing sign.
87: *
88: * Next the indices that point into member are unsorted into
89: * array class according to the original order of file0.
90: *
91: * The cleverness lies in routine stone. This marches
92: * through the lines of file0, developing a vector klist
93: * of "k-candidates". At step i a k-candidate is a matched
94: * pair of lines x,y (x in file0 y in file1) such that
95: * there is a common subsequence of length k
96: * between the first i lines of file0 and the first y
97: * lines of file1, but there is no such subsequence for
98: * any smaller y. x is the earliest possible mate to y
99: * that occurs in such a subsequence.
100: *
101: * Whenever any of the members of the equivalence class of
102: * lines in file1 matable to a line in file0 has serial number
103: * less than the y of some k-candidate, that k-candidate
104: * with the smallest such y is replaced. The new
105: * k-candidate is chained (via pred) to the current
106: * k-1 candidate so that the actual subsequence can
107: * be recovered. When a member has serial number greater
108: * that the y of all k-candidates, the klist is extended.
109: * At the end, the longest subsequence is pulled out
110: * and placed in the array J by unravel
111: *
112: * With J in hand, the matches there recorded are
113: * check'ed against reality to assure that no spurious
114: * matches have crept in due to hashing. If they have,
115: * they are broken, and "jackpot" is recorded--a harmless
116: * matter except that a true match for a spuriously
117: * mated line may now be unnecessarily reported as a change.
118: *
119: * Much of the complexity of the program comes simply
120: * from trying to minimize core utilization and
121: * maximize the range of doable problems by dynamically
122: * allocating what is needed and reusing what is not.
123: * The core requirements for problems larger than somewhat
124: * are (in words) 2*length(file0) + length(file1) +
125: * 3*(number of k-candidates installed), typically about
126: * 6n words for files of length n.
127: */
128:
129: #include <sys/param.h>
130: #include <sys/stat.h>
131: #include <sys/wait.h>
132:
133: #include <errno.h>
134: #include <ctype.h>
135: #include <stdio.h>
136: #include <fcntl.h>
137: #include <paths.h>
138: #include <regex.h>
139: #include <dirent.h>
140: #include <stdlib.h>
141: #include <stddef.h>
142: #include <unistd.h>
143: #include <string.h>
144: #include <sysexits.h>
145:
146: #include "cvs.h"
147: #include "log.h"
148: #include "buf.h"
1.4 jfb 149: #include "proto.h"
1.1 jfb 150:
151:
152: #define CVS_DIFF_DEFCTX 3 /* default context length */
153:
154:
155: /*
156: * Output format options
157: */
158: #define D_NORMAL 0 /* Normal output */
159: #define D_CONTEXT 1 /* Diff with context */
160: #define D_UNIFIED 2 /* Unified context diff */
161: #define D_IFDEF 3 /* Diff with merged #ifdef's */
162: #define D_BRIEF 4 /* Say if the files differ */
163:
164: /*
165: * Status values for print_status() and diffreg() return values
166: */
167: #define D_SAME 0 /* Files are the same */
168: #define D_DIFFER 1 /* Files are different */
169: #define D_BINARY 2 /* Binary files are different */
170: #define D_COMMON 3 /* Subdirectory common to both dirs */
171: #define D_ONLY 4 /* Only exists in one directory */
172: #define D_MISMATCH1 5 /* path1 was a dir, path2 a file */
173: #define D_MISMATCH2 6 /* path1 was a file, path2 a dir */
174: #define D_ERROR 7 /* An error occurred */
175: #define D_SKIPPED1 8 /* path1 was a special file */
176: #define D_SKIPPED2 9 /* path2 was a special file */
177:
178: struct cand {
179: int x;
180: int y;
181: int pred;
182: } cand;
183:
184: struct line {
185: int serial;
186: int value;
187: } *file[2];
188:
189: /*
190: * The following struct is used to record change information when
191: * doing a "context" or "unified" diff. (see routine "change" to
192: * understand the highly mnemonic field names)
193: */
194: struct context_vec {
195: int a; /* start line in old file */
196: int b; /* end line in old file */
197: int c; /* start line in new file */
198: int d; /* end line in new file */
199: };
200:
1.4 jfb 201: struct diff_arg {
202: char *rev1;
203: char *rev2;
204: char *date1;
205: char *date2;
206: };
207:
1.1 jfb 208:
209: struct excludes {
210: char *pattern;
211: struct excludes *next;
212: };
213:
214:
215: char *splice(char *, char *);
216: int cvs_diffreg(const char *, const char *);
1.9 jfb 217: int cvs_diff_file(struct cvs_file *, void *);
218: int cvs_diff_sendflags(struct cvsroot *, struct diff_arg *);
1.1 jfb 219: static void output(const char *, FILE *, const char *, FILE *);
220: static void check(FILE *, FILE *);
221: static void range(int, int, char *);
222: static void uni_range(int, int);
223: static void dump_context_vec(FILE *, FILE *);
224: static void dump_unified_vec(FILE *, FILE *);
225: static void prepare(int, FILE *, off_t);
226: static void prune(void);
227: static void equiv(struct line *, int, struct line *, int, int *);
228: static void unravel(int);
229: static void unsort(struct line *, int, int *);
230: static void change(const char *, FILE *, const char *, FILE *, int, int, int, int);
231: static void sort(struct line *, int);
232: static int ignoreline(char *);
233: static int asciifile(FILE *);
234: static int fetch(long *, int, int, FILE *, int, int);
235: static int newcand(int, int, int);
236: static int search(int *, int, int);
237: static int skipline(FILE *);
238: static int isqrt(int);
239: static int stone(int *, int, int *, int *);
240: static int readhash(FILE *);
241: static int files_differ(FILE *, FILE *);
242: static char *preadline(int, size_t, off_t);
243:
244:
245: extern int cvs_client;
246:
1.8 jfb 247: static int aflag, bflag, dflag, iflag, Nflag, tflag, Tflag, wflag;
1.1 jfb 248: static int context, status;
249: static int format = D_NORMAL;
250: static struct stat stb1, stb2;
1.2 jfb 251: static char *ifdefname, *ignore_pats, diffargs[128];
252: static const char *diff_file;
1.1 jfb 253: regex_t ignore_re;
254:
255: static int *J; /* will be overlaid on class */
256: static int *class; /* will be overlaid on file[0] */
257: static int *klist; /* will be overlaid on file[0] after class */
258: static int *member; /* will be overlaid on file[1] */
259: static int clen;
260: static int inifdef; /* whether or not we are in a #ifdef block */
261: static int len[2];
262: static int pref, suff; /* length of prefix and suffix */
263: static int slen[2];
264: static int anychange;
265: static long *ixnew; /* will be overlaid on file[1] */
266: static long *ixold; /* will be overlaid on klist */
267: static struct cand *clist; /* merely a free storage pot for candidates */
268: static int clistlen; /* the length of clist */
269: static struct line *sfile[2]; /* shortened by pruning common prefix/suffix */
270: static u_char *chrtran; /* translation table for case-folding */
271: static struct context_vec *context_vec_start;
272: static struct context_vec *context_vec_end;
273: static struct context_vec *context_vec_ptr;
274:
275: #define FUNCTION_CONTEXT_SIZE 41
276: static int lastline;
277: static int lastmatchline;
278:
279:
280: /*
281: * chrtran points to one of 2 translation tables: cup2low if folding upper to
282: * lower case clow2low if not folding case
283: */
284: u_char clow2low[256] = {
285: 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a,
286: 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15,
287: 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
288: 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b,
289: 0x2c, 0x2d, 0x2e, 0x2f, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36,
290: 0x37, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x40, 0x41,
291: 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x4a, 0x4b, 0x4c,
292: 0x4d, 0x4e, 0x4f, 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,
293: 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, 0x60, 0x61, 0x62,
294: 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d,
295: 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78,
296: 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x80, 0x81, 0x82, 0x83,
297: 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e,
298: 0x8f, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99,
299: 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xa4,
300: 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,
301: 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba,
302: 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5,
303: 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf, 0xd0,
304: 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xdb,
305: 0xdc, 0xdd, 0xde, 0xdf, 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6,
306: 0xe7, 0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, 0xf0, 0xf1,
307: 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa, 0xfb, 0xfc,
308: 0xfd, 0xfe, 0xff
309: };
310:
311: u_char cup2low[256] = {
312: 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a,
313: 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15,
314: 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
315: 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b,
316: 0x2c, 0x2d, 0x2e, 0x2f, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36,
317: 0x37, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x60, 0x61,
318: 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c,
319: 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
320: 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x60, 0x61, 0x62,
321: 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d,
322: 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78,
323: 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x80, 0x81, 0x82, 0x83,
324: 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e,
325: 0x8f, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99,
326: 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xa4,
327: 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,
328: 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba,
329: 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5,
330: 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf, 0xd0,
331: 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xdb,
332: 0xdc, 0xdd, 0xde, 0xdf, 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6,
333: 0xe7, 0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, 0xf0, 0xf1,
334: 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa, 0xfb, 0xfc,
335: 0xfd, 0xfe, 0xff
336: };
337:
338:
339: /*
340: * cvs_diff()
341: *
342: * Handler for the `cvs diff' command.
343: *
344: * SYNOPSIS: cvs [args] diff [-clipu] [-D date] [-r rev]
345: */
346: int
347: cvs_diff(int argc, char **argv)
348: {
1.4 jfb 349: int ch, recurse, flags;
350: struct diff_arg darg;
351: struct cvsroot *root;
1.1 jfb 352:
353: context = CVS_DIFF_DEFCTX;
1.4 jfb 354: flags = CF_RECURSE|CF_IGNORE|CF_SORT|CF_KNOWN;
355: recurse = 1;
356:
357: memset(&darg, 0, sizeof(darg));
1.2 jfb 358: strlcpy(diffargs, argv[0], sizeof(diffargs));
1.1 jfb 359:
1.8 jfb 360: while ((ch = getopt(argc, argv, "cD:liN:r:u")) != -1) {
1.1 jfb 361: switch (ch) {
362: case 'c':
1.2 jfb 363: strlcat(diffargs, " -c", sizeof(diffargs));
1.1 jfb 364: format = D_CONTEXT;
365: break;
366: case 'D':
1.4 jfb 367: if (darg.date1 == NULL && darg.rev1 == NULL)
368: darg.date1 = optarg;
369: else if (darg.date2 == NULL && darg.rev2 == NULL)
370: darg.date2 = optarg;
1.1 jfb 371: else {
372: cvs_log(LP_ERR,
373: "no more than two revisions/dates can "
374: "be specified");
375: }
376: break;
377: case 'l':
1.2 jfb 378: strlcat(diffargs, " -l", sizeof(diffargs));
1.1 jfb 379: recurse = 0;
1.4 jfb 380: flags &= ~CF_RECURSE;
1.1 jfb 381: break;
382: case 'i':
1.2 jfb 383: strlcat(diffargs, " -i", sizeof(diffargs));
1.1 jfb 384: iflag = 1;
385: break;
1.8 jfb 386: case 'N':
387: strlcat(diffargs, " -N", sizeof(diffargs));
388: Nflag = 1;
389: break;
1.1 jfb 390: case 'r':
1.4 jfb 391: if ((darg.rev1 == NULL) && (darg.date1 == NULL))
392: darg.rev1 = optarg;
393: else if ((darg.rev2 == NULL) && (darg.date2 == NULL))
394: darg.rev2 = optarg;
1.1 jfb 395: else {
396: cvs_log(LP_ERR,
397: "no more than two revisions/dates can "
398: "be specified");
399: return (EX_USAGE);
400: }
401: break;
402: case 'u':
1.2 jfb 403: strlcat(diffargs, " -u", sizeof(diffargs));
1.1 jfb 404: format = D_UNIFIED;
405: break;
406: default:
407: return (EX_USAGE);
408: }
409: }
410:
411: argc -= optind;
412: argv += optind;
413:
414: if (argc == 0) {
1.6 jfb 415: cvs_files = cvs_file_get(".", flags);
1.10 deraadt 416: } else
1.6 jfb 417: cvs_files = cvs_file_getspec(argv, argc, 0);
1.7 jfb 418: if (cvs_files == NULL)
419: return (EX_DATAERR);
1.1 jfb 420:
1.6 jfb 421: cvs_file_examine(cvs_files, cvs_diff_file, &darg);
1.1 jfb 422:
1.6 jfb 423: root = cvs_files->cf_ddat->cd_root;
1.5 jfb 424: if (root->cr_method != CVS_METHOD_LOCAL) {
1.6 jfb 425: cvs_senddir(root, cvs_files);
1.4 jfb 426: cvs_sendreq(root, CVS_REQ_DIFF, NULL);
1.5 jfb 427: }
1.1 jfb 428:
1.4 jfb 429: return (0);
430: }
1.1 jfb 431:
432:
1.4 jfb 433: /*
434: * cvs_diff_sendflags()
435: *
436: */
437: int
438: cvs_diff_sendflags(struct cvsroot *root, struct diff_arg *dap)
439: {
440: /* send the flags */
441: if (format == D_CONTEXT)
442: cvs_sendarg(root, "-c", 0);
443: else if (format == D_UNIFIED)
444: cvs_sendarg(root, "-u", 0);
445:
446: if (dap->rev1 != NULL) {
447: cvs_sendarg(root, "-r", 0);
448: cvs_sendarg(root, dap->rev1, 1);
1.10 deraadt 449: } else if (dap->date1 != NULL) {
1.4 jfb 450: cvs_sendarg(root, "-D", 0);
451: cvs_sendarg(root, dap->date1, 1);
452: }
453: if (dap->rev2 != NULL) {
454: cvs_sendarg(root, "-r", 0);
455: cvs_sendarg(root, dap->rev2, 1);
1.10 deraadt 456: } else if (dap->date2 != NULL) {
1.4 jfb 457: cvs_sendarg(root, "-D", 0);
458: cvs_sendarg(root, dap->date2, 1);
1.1 jfb 459: }
460:
461: return (0);
462: }
463:
464:
465: /*
466: * cvs_diff_file()
467: *
468: * Diff a single file.
469: */
470: int
1.4 jfb 471: cvs_diff_file(struct cvs_file *cfp, void *arg)
1.1 jfb 472: {
1.8 jfb 473: char *dir, *repo, buf[64];
474: char fpath[MAXPATHLEN], dfpath[MAXPATHLEN], rcspath[MAXPATHLEN];
1.12 ! djm 475: char path_tmp1[MAXPATHLEN], path_tmp2[MAXPATHLEN];
1.1 jfb 476: BUF *b1, *b2;
477: RCSNUM *r1, *r2;
478: RCSFILE *rf;
1.4 jfb 479: struct diff_arg *dap;
1.1 jfb 480: struct cvs_ent *entp;
1.4 jfb 481: struct cvsroot *root;
1.1 jfb 482:
1.4 jfb 483: dap = (struct diff_arg *)arg;
1.1 jfb 484:
1.4 jfb 485: if (cfp->cf_type == DT_DIR) {
1.7 jfb 486: if (cfp->cf_cvstat == CVS_FST_UNKNOWN) {
487: root = cfp->cf_parent->cf_ddat->cd_root;
1.8 jfb 488: cvs_sendreq(root, CVS_REQ_QUESTIONABLE,
489: CVS_FILE_NAME(cfp));
1.10 deraadt 490: } else {
1.7 jfb 491: root = cfp->cf_ddat->cd_root;
492: if ((cfp->cf_parent == NULL) ||
493: (root != cfp->cf_parent->cf_ddat->cd_root)) {
494: cvs_connect(root);
495: cvs_diff_sendflags(root, dap);
496: }
497:
498: cvs_senddir(root, cfp);
1.4 jfb 499: }
1.1 jfb 500:
1.4 jfb 501: return (0);
1.1 jfb 502: }
503:
1.4 jfb 504: rf = NULL;
1.8 jfb 505: diff_file = cvs_file_getpath(cfp, fpath, sizeof(fpath));
1.7 jfb 506:
1.4 jfb 507: if (cfp->cf_parent != NULL) {
1.8 jfb 508: dir = cvs_file_getpath(cfp->cf_parent, dfpath, sizeof(dfpath));
1.7 jfb 509: root = cfp->cf_parent->cf_ddat->cd_root;
1.4 jfb 510: repo = cfp->cf_parent->cf_ddat->cd_repo;
1.10 deraadt 511: } else {
1.4 jfb 512: dir = ".";
1.7 jfb 513: root = NULL;
1.4 jfb 514: repo = NULL;
1.1 jfb 515: }
516:
1.4 jfb 517: if (cfp->cf_cvstat == CVS_FST_UNKNOWN) {
518: if (root->cr_method == CVS_METHOD_LOCAL)
519: cvs_log(LP_WARN, "I know nothing about %s", diff_file);
520: else
1.8 jfb 521: cvs_sendreq(root, CVS_REQ_QUESTIONABLE,
522: CVS_FILE_NAME(cfp));
1.4 jfb 523: return (0);
1.1 jfb 524: }
525:
1.4 jfb 526: entp = cvs_ent_getent(diff_file);
527: if (entp == NULL)
528: return (-1);
1.1 jfb 529:
1.4 jfb 530: if (root->cr_method != CVS_METHOD_LOCAL) {
531: if (cvs_sendentry(root, entp) < 0) {
532: cvs_ent_free(entp);
533: return (-1);
534: }
1.1 jfb 535: }
536:
1.4 jfb 537: if (cfp->cf_cvstat == CVS_FST_UPTODATE) {
538: if (root->cr_method != CVS_METHOD_LOCAL)
1.8 jfb 539: cvs_sendreq(root, CVS_REQ_UNCHANGED,
540: CVS_FILE_NAME(cfp));
1.4 jfb 541: cvs_ent_free(entp);
1.1 jfb 542: return (0);
543: }
544:
545: /* at this point, the file is modified */
1.4 jfb 546: if (root->cr_method != CVS_METHOD_LOCAL) {
1.8 jfb 547: cvs_sendreq(root, CVS_REQ_MODIFIED, CVS_FILE_NAME(cfp));
1.4 jfb 548: cvs_sendfile(root, diff_file);
1.10 deraadt 549: } else {
1.1 jfb 550: snprintf(rcspath, sizeof(rcspath), "%s/%s/%s%s",
1.4 jfb 551: root->cr_dir, repo, diff_file, RCS_FILE_EXT);
1.1 jfb 552:
553: rf = rcs_open(rcspath, RCS_MODE_READ);
1.4 jfb 554: if (rf == NULL) {
555: cvs_ent_free(entp);
1.1 jfb 556: return (-1);
1.4 jfb 557: }
1.1 jfb 558:
1.4 jfb 559: cvs_printf("Index: %s\n%s\nRCS file: %s\n", diff_file,
1.1 jfb 560: RCS_DIFF_DIV, rcspath);
561:
1.4 jfb 562: if (dap->rev1 == NULL)
1.1 jfb 563: r1 = entp->ce_rev;
564: else {
565: r1 = rcsnum_alloc();
1.4 jfb 566: rcsnum_aton(dap->rev1, NULL, r1);
1.1 jfb 567: }
568:
1.4 jfb 569: cvs_printf("retrieving revision %s\n",
1.1 jfb 570: rcsnum_tostr(r1, buf, sizeof(buf)));
571: b1 = rcs_getrev(rf, r1);
572:
1.4 jfb 573: if (dap->rev2 != NULL) {
574: cvs_printf("retrieving revision %s\n", dap->rev2);
1.1 jfb 575: r2 = rcsnum_alloc();
1.4 jfb 576: rcsnum_aton(dap->rev2, NULL, r2);
1.1 jfb 577: b2 = rcs_getrev(rf, r2);
1.10 deraadt 578: } else {
1.4 jfb 579: b2 = cvs_buf_load(diff_file, BUF_AUTOEXT);
1.1 jfb 580: }
581:
1.4 jfb 582: rcs_close(rf);
583:
1.2 jfb 584: printf("%s", diffargs);
585: printf(" -r%s", buf);
1.4 jfb 586: if (dap->rev2 != NULL)
587: printf(" -r%s", dap->rev2);
588: printf(" %s\n", diff_file);
1.12 ! djm 589: strlcpy(path_tmp1, "/tmp/diff1.XXXXXXXXXX", sizeof(path_tmp1));
! 590: if (cvs_buf_write_stmp(b1, path_tmp1, 0600) == -1)
! 591: return (-1);
! 592: strlcpy(path_tmp2, "/tmp/diff2.XXXXXXXXXX", sizeof(path_tmp1));
! 593: if (cvs_buf_write_stmp(b2, path_tmp2, 0600) == -1) {
! 594: (void)unlink(path_tmp1);
! 595: return (-1);
! 596: }
! 597: cvs_diffreg(path_tmp1, path_tmp2);
! 598: (void)unlink(path_tmp1);
! 599: (void)unlink(path_tmp2);
1.1 jfb 600: }
601:
1.4 jfb 602: cvs_ent_free(entp);
1.1 jfb 603: return (0);
604: }
605:
606:
607: int
608: cvs_diffreg(const char *file1, const char *file2)
609: {
610: FILE *f1, *f2;
611: int i, rval;
612:
613: f1 = f2 = NULL;
614: rval = D_SAME;
615: anychange = 0;
616: lastline = 0;
617: lastmatchline = 0;
618: context_vec_ptr = context_vec_start - 1;
619: chrtran = (iflag ? cup2low : clow2low);
620:
621: f1 = fopen(file1, "r");
622: if (f1 == NULL) {
623: cvs_log(LP_ERRNO, "%s", file1);
624: status |= 2;
625: goto closem;
626: }
627:
628: f2 = fopen(file2, "r");
629: if (f2 == NULL) {
630: cvs_log(LP_ERRNO, "%s", file2);
631: status |= 2;
632: goto closem;
633: }
634:
635: switch (files_differ(f1, f2)) {
636: case 0:
637: goto closem;
638: case 1:
639: break;
640: default:
641: /* error */
642: status |= 2;
643: goto closem;
644: }
645:
646: if (!asciifile(f1) || !asciifile(f2)) {
647: rval = D_BINARY;
648: status |= 1;
649: goto closem;
650: }
651: prepare(0, f1, stb1.st_size);
652: prepare(1, f2, stb2.st_size);
653: prune();
654: sort(sfile[0], slen[0]);
655: sort(sfile[1], slen[1]);
656:
657: member = (int *)file[1];
658: equiv(sfile[0], slen[0], sfile[1], slen[1], member);
659: member = realloc(member, (slen[1] + 2) * sizeof(int));
660:
661: class = (int *)file[0];
662: unsort(sfile[0], slen[0], class);
663: class = realloc(class, (slen[0] + 2) * sizeof(int));
664:
665: klist = malloc((slen[0] + 2) * sizeof(int));
666: clen = 0;
667: clistlen = 100;
668: clist = malloc(clistlen * sizeof(cand));
669: i = stone(class, slen[0], member, klist);
670: free(member);
671: free(class);
672:
673: J = realloc(J, (len[0] + 2) * sizeof(int));
674: unravel(klist[i]);
675: free(clist);
676: free(klist);
677:
678: ixold = realloc(ixold, (len[0] + 2) * sizeof(long));
679: ixnew = realloc(ixnew, (len[1] + 2) * sizeof(long));
680: check(f1, f2);
681: output(file1, f1, file2, f2);
682:
683: closem:
684: if (anychange) {
685: status |= 1;
686: if (rval == D_SAME)
687: rval = D_DIFFER;
688: }
689: if (f1 != NULL)
690: fclose(f1);
691: if (f2 != NULL)
692: fclose(f2);
693: return (rval);
694: }
695:
696: /*
697: * Check to see if the given files differ.
698: * Returns 0 if they are the same, 1 if different, and -1 on error.
699: * XXX - could use code from cmp(1) [faster]
700: */
701: static int
702: files_differ(FILE *f1, FILE *f2)
703: {
704: char buf1[BUFSIZ], buf2[BUFSIZ];
705: size_t i, j;
706:
707: if (stb1.st_size != stb2.st_size)
708: return (1);
709: for (;;) {
710: i = fread(buf1, 1, sizeof(buf1), f1);
711: j = fread(buf2, 1, sizeof(buf2), f2);
712: if (i != j)
713: return (1);
714: if (i == 0 && j == 0) {
715: if (ferror(f1) || ferror(f2))
716: return (1);
717: return (0);
718: }
719: if (memcmp(buf1, buf2, i) != 0)
720: return (1);
721: }
722: }
723:
724:
725: char *
726: splice(char *dir, char *file)
727: {
728: char *tail, *buf;
729:
730: if ((tail = strrchr(file, '/')) == NULL)
731: tail = file;
732: else
733: tail++;
734: asprintf(&buf, "%s/%s", dir, tail);
735: return (buf);
736: }
737:
738: static void
739: prepare(int i, FILE *fd, off_t filesize)
740: {
741: struct line *p;
742: int j, h;
743: size_t sz;
744:
745: rewind(fd);
746:
747: sz = (filesize <= SIZE_MAX ? filesize : SIZE_MAX) / 25;
748: if (sz < 100)
749: sz = 100;
750:
751: p = malloc((sz + 3) * sizeof(struct line));
752: for (j = 0; (h = readhash(fd));) {
753: if (j == (int)sz) {
754: sz = sz * 3 / 2;
755: p = realloc(p, (sz + 3) * sizeof(struct line));
756: }
757: p[++j].value = h;
758: }
759: len[i] = j;
760: file[i] = p;
761: }
762:
763: static void
764: prune(void)
765: {
766: int i, j;
767:
768: for (pref = 0; pref < len[0] && pref < len[1] &&
769: file[0][pref + 1].value == file[1][pref + 1].value;
770: pref++)
771: ;
772: for (suff = 0; suff < len[0] - pref && suff < len[1] - pref &&
773: file[0][len[0] - suff].value == file[1][len[1] - suff].value;
774: suff++)
775: ;
776: for (j = 0; j < 2; j++) {
777: sfile[j] = file[j] + pref;
778: slen[j] = len[j] - pref - suff;
779: for (i = 0; i <= slen[j]; i++)
780: sfile[j][i].serial = i;
781: }
782: }
783:
784: static void
785: equiv(struct line *a, int n, struct line *b, int m, int *c)
786: {
787: int i, j;
788:
789: i = j = 1;
790: while (i <= n && j <= m) {
791: if (a[i].value < b[j].value)
792: a[i++].value = 0;
793: else if (a[i].value == b[j].value)
794: a[i++].value = j;
795: else
796: j++;
797: }
798: while (i <= n)
799: a[i++].value = 0;
800: b[m + 1].value = 0;
801: j = 0;
802: while (++j <= m) {
803: c[j] = -b[j].serial;
804: while (b[j + 1].value == b[j].value) {
805: j++;
806: c[j] = b[j].serial;
807: }
808: }
809: c[j] = -1;
810: }
811:
812: /* Code taken from ping.c */
813: static int
814: isqrt(int n)
815: {
816: int y, x = 1;
817:
818: if (n == 0)
819: return(0);
820:
821: do { /* newton was a stinker */
822: y = x;
823: x = n / x;
824: x += y;
825: x /= 2;
826: } while ((x - y) > 1 || (x - y) < -1);
827:
828: return (x);
829: }
830:
831: static int
832: stone(int *a, int n, int *b, int *c)
833: {
834: int i, k, y, j, l;
835: int oldc, tc, oldl;
836: u_int numtries;
837:
1.4 jfb 838: const u_int bound = dflag ? UINT_MAX : MAX(256, isqrt(n));
1.1 jfb 839:
840: k = 0;
841: c[0] = newcand(0, 0, 0);
842: for (i = 1; i <= n; i++) {
843: j = a[i];
844: if (j == 0)
845: continue;
846: y = -b[j];
847: oldl = 0;
848: oldc = c[0];
849: numtries = 0;
850: do {
851: if (y <= clist[oldc].y)
852: continue;
853: l = search(c, k, y);
854: if (l != oldl + 1)
855: oldc = c[l - 1];
856: if (l <= k) {
857: if (clist[c[l]].y <= y)
858: continue;
859: tc = c[l];
860: c[l] = newcand(i, y, oldc);
861: oldc = tc;
862: oldl = l;
863: numtries++;
864: } else {
865: c[l] = newcand(i, y, oldc);
866: k++;
867: break;
868: }
869: } while ((y = b[++j]) > 0 && numtries < bound);
870: }
871: return (k);
872: }
873:
874: static int
875: newcand(int x, int y, int pred)
876: {
877: struct cand *q;
878:
879: if (clen == clistlen) {
880: clistlen = clistlen * 11 / 10;
881: clist = realloc(clist, clistlen * sizeof(cand));
882: }
883: q = clist + clen;
884: q->x = x;
885: q->y = y;
886: q->pred = pred;
887: return (clen++);
888: }
889:
890: static int
891: search(int *c, int k, int y)
892: {
893: int i, j, l, t;
894:
895: if (clist[c[k]].y < y) /* quick look for typical case */
896: return (k + 1);
897: i = 0;
898: j = k + 1;
899: while (1) {
900: l = i + j;
901: if ((l >>= 1) <= i)
902: break;
903: t = clist[c[l]].y;
904: if (t > y)
905: j = l;
906: else if (t < y)
907: i = l;
908: else
909: return (l);
910: }
911: return (l + 1);
912: }
913:
914: static void
915: unravel(int p)
916: {
917: struct cand *q;
918: int i;
919:
920: for (i = 0; i <= len[0]; i++)
921: J[i] = i <= pref ? i :
922: i > len[0] - suff ? i + len[1] - len[0] : 0;
923: for (q = clist + p; q->y != 0; q = clist + q->pred)
924: J[q->x + pref] = q->y + pref;
925: }
926:
927: /*
928: * Check does double duty:
929: * 1. ferret out any fortuitous correspondences due
930: * to confounding by hashing (which result in "jackpot")
931: * 2. collect random access indexes to the two files
932: */
933: static void
934: check(FILE *f1, FILE *f2)
935: {
936: int i, j, jackpot, c, d;
937: long ctold, ctnew;
938:
939: rewind(f1);
940: rewind(f2);
941: j = 1;
942: ixold[0] = ixnew[0] = 0;
943: jackpot = 0;
944: ctold = ctnew = 0;
945: for (i = 1; i <= len[0]; i++) {
946: if (J[i] == 0) {
947: ixold[i] = ctold += skipline(f1);
948: continue;
949: }
950: while (j < J[i]) {
951: ixnew[j] = ctnew += skipline(f2);
952: j++;
953: }
954: if (bflag || wflag || iflag) {
955: for (;;) {
956: c = getc(f1);
957: d = getc(f2);
958: /*
959: * GNU diff ignores a missing newline
960: * in one file if bflag || wflag.
961: */
962: if ((bflag || wflag) &&
963: ((c == EOF && d == '\n') ||
964: (c == '\n' && d == EOF))) {
965: break;
966: }
967: ctold++;
968: ctnew++;
969: if (bflag && isspace(c) && isspace(d)) {
970: do {
971: if (c == '\n')
972: break;
973: ctold++;
974: } while (isspace(c = getc(f1)));
975: do {
976: if (d == '\n')
977: break;
978: ctnew++;
979: } while (isspace(d = getc(f2)));
980: } else if (wflag) {
981: while (isspace(c) && c != '\n') {
982: c = getc(f1);
983: ctold++;
984: }
985: while (isspace(d) && d != '\n') {
986: d = getc(f2);
987: ctnew++;
988: }
989: }
990: if (chrtran[c] != chrtran[d]) {
991: jackpot++;
992: J[i] = 0;
993: if (c != '\n' && c != EOF)
994: ctold += skipline(f1);
995: if (d != '\n' && c != EOF)
996: ctnew += skipline(f2);
997: break;
998: }
999: if (c == '\n' || c == EOF)
1000: break;
1001: }
1002: } else {
1003: for (;;) {
1004: ctold++;
1005: ctnew++;
1006: if ((c = getc(f1)) != (d = getc(f2))) {
1007: /* jackpot++; */
1008: J[i] = 0;
1009: if (c != '\n' && c != EOF)
1010: ctold += skipline(f1);
1011: if (d != '\n' && c != EOF)
1012: ctnew += skipline(f2);
1013: break;
1014: }
1015: if (c == '\n' || c == EOF)
1016: break;
1017: }
1018: }
1019: ixold[i] = ctold;
1020: ixnew[j] = ctnew;
1021: j++;
1022: }
1023: for (; j <= len[1]; j++)
1024: ixnew[j] = ctnew += skipline(f2);
1025: /*
1026: * if (jackpot)
1027: * fprintf(stderr, "jackpot\n");
1028: */
1029: }
1030:
1031: /* shellsort CACM #201 */
1032: static void
1033: sort(struct line *a, int n)
1034: {
1035: struct line *ai, *aim, w;
1036: int j, m = 0, k;
1037:
1038: if (n == 0)
1039: return;
1040: for (j = 1; j <= n; j *= 2)
1041: m = 2 * j - 1;
1042: for (m /= 2; m != 0; m /= 2) {
1043: k = n - m;
1044: for (j = 1; j <= k; j++) {
1045: for (ai = &a[j]; ai > a; ai -= m) {
1046: aim = &ai[m];
1047: if (aim < ai)
1048: break; /* wraparound */
1049: if (aim->value > ai[0].value ||
1050: (aim->value == ai[0].value &&
1051: aim->serial > ai[0].serial))
1052: break;
1053: w.value = ai[0].value;
1054: ai[0].value = aim->value;
1055: aim->value = w.value;
1056: w.serial = ai[0].serial;
1057: ai[0].serial = aim->serial;
1058: aim->serial = w.serial;
1059: }
1060: }
1061: }
1062: }
1063:
1064: static void
1065: unsort(struct line *f, int l, int *b)
1066: {
1067: int *a, i;
1068:
1069: a = malloc((l + 1) * sizeof(int));
1070: for (i = 1; i <= l; i++)
1071: a[f[i].serial] = f[i].value;
1072: for (i = 1; i <= l; i++)
1073: b[i] = a[i];
1074: free(a);
1075: }
1076:
1077: static int
1078: skipline(FILE *f)
1079: {
1080: int i, c;
1081:
1082: for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++)
1083: continue;
1084: return (i);
1085: }
1086:
1087: static void
1088: output(const char *file1, FILE *f1, const char *file2, FILE *f2)
1089: {
1090: int m, i0, i1, j0, j1;
1091:
1092: rewind(f1);
1093: rewind(f2);
1094: m = len[0];
1095: J[0] = 0;
1096: J[m + 1] = len[1] + 1;
1097: for (i0 = 1; i0 <= m; i0 = i1 + 1) {
1098: while (i0 <= m && J[i0] == J[i0 - 1] + 1)
1099: i0++;
1100: j0 = J[i0 - 1] + 1;
1101: i1 = i0 - 1;
1102: while (i1 < m && J[i1 + 1] == 0)
1103: i1++;
1104: j1 = J[i1 + 1] - 1;
1105: J[i1] = j1;
1106: change(file1, f1, file2, f2, i0, i1, j0, j1);
1107: }
1108: if (m == 0)
1109: change(file1, f1, file2, f2, 1, 0, 1, len[1]);
1110: if (format == D_IFDEF) {
1111: for (;;) {
1112: #define c i0
1113: if ((c = getc(f1)) == EOF)
1114: return;
1115: putchar(c);
1116: }
1117: #undef c
1118: }
1119: if (anychange != 0) {
1120: if (format == D_CONTEXT)
1121: dump_context_vec(f1, f2);
1122: else if (format == D_UNIFIED)
1123: dump_unified_vec(f1, f2);
1124: }
1125: }
1126:
1127: static __inline void
1128: range(int a, int b, char *separator)
1129: {
1130: printf("%d", a > b ? b : a);
1131: if (a < b)
1132: printf("%s%d", separator, b);
1133: }
1134:
1135: static __inline void
1136: uni_range(int a, int b)
1137: {
1138: if (a < b)
1139: printf("%d,%d", a, b - a + 1);
1140: else if (a == b)
1141: printf("%d", b);
1142: else
1143: printf("%d,0", b);
1144: }
1145:
1146: static char *
1147: preadline(int fd, size_t len, off_t off)
1148: {
1149: char *line;
1150: ssize_t nr;
1151:
1152: line = malloc(len + 1);
1153: if (line == NULL) {
1154: cvs_log(LP_ERRNO, "failed to allocate line");
1155: return (NULL);
1156: }
1157: if ((nr = pread(fd, line, len, off)) < 0) {
1158: cvs_log(LP_ERRNO, "preadline failed");
1159: return (NULL);
1160: }
1161: line[nr] = '\0';
1162: return (line);
1163: }
1164:
1165: static int
1166: ignoreline(char *line)
1167: {
1168: int ret;
1169:
1170: ret = regexec(&ignore_re, line, 0, NULL, 0);
1171: free(line);
1172: return (ret == 0); /* if it matched, it should be ignored. */
1173: }
1174:
1175: /*
1176: * Indicate that there is a difference between lines a and b of the from file
1177: * to get to lines c to d of the to file. If a is greater then b then there
1178: * are no lines in the from file involved and this means that there were
1179: * lines appended (beginning at b). If c is greater than d then there are
1180: * lines missing from the to file.
1181: */
1182: static void
1183: change(const char *file1, FILE *f1, const char *file2, FILE *f2,
1184: int a, int b, int c, int d)
1185: {
1186: static size_t max_context = 64;
1187: int i;
1188:
1189: if (format != D_IFDEF && a > b && c > d)
1190: return;
1191: if (ignore_pats != NULL) {
1192: char *line;
1193: /*
1194: * All lines in the change, insert, or delete must
1195: * match an ignore pattern for the change to be
1196: * ignored.
1197: */
1198: if (a <= b) { /* Changes and deletes. */
1199: for (i = a; i <= b; i++) {
1200: line = preadline(fileno(f1),
1201: ixold[i] - ixold[i - 1], ixold[i - 1]);
1202: if (!ignoreline(line))
1203: goto proceed;
1204: }
1205: }
1206: if (a > b || c <= d) { /* Changes and inserts. */
1207: for (i = c; i <= d; i++) {
1208: line = preadline(fileno(f2),
1209: ixnew[i] - ixnew[i - 1], ixnew[i - 1]);
1210: if (!ignoreline(line))
1211: goto proceed;
1212: }
1213: }
1214: return;
1215: }
1216: proceed:
1217: if (format == D_CONTEXT || format == D_UNIFIED) {
1218: /*
1219: * Allocate change records as needed.
1220: */
1221: if (context_vec_ptr == context_vec_end - 1) {
1222: ptrdiff_t offset = context_vec_ptr - context_vec_start;
1223: max_context <<= 1;
1224: context_vec_start = realloc(context_vec_start,
1225: max_context * sizeof(struct context_vec));
1226: context_vec_end = context_vec_start + max_context;
1227: context_vec_ptr = context_vec_start + offset;
1228: }
1229: if (anychange == 0) {
1230: /*
1231: * Print the context/unidiff header first time through.
1232: */
1233: printf("%s %s %s",
1234: format == D_CONTEXT ? "***" : "---", diff_file,
1235: ctime(&stb1.st_mtime));
1236: printf("%s %s %s",
1237: format == D_CONTEXT ? "---" : "+++", diff_file,
1238: ctime(&stb2.st_mtime));
1239: anychange = 1;
1240: } else if (a > context_vec_ptr->b + (2 * context) + 1 &&
1241: c > context_vec_ptr->d + (2 * context) + 1) {
1242: /*
1243: * If this change is more than 'context' lines from the
1244: * previous change, dump the record and reset it.
1245: */
1246: if (format == D_CONTEXT)
1247: dump_context_vec(f1, f2);
1248: else
1249: dump_unified_vec(f1, f2);
1250: }
1251: context_vec_ptr++;
1252: context_vec_ptr->a = a;
1253: context_vec_ptr->b = b;
1254: context_vec_ptr->c = c;
1255: context_vec_ptr->d = d;
1256: return;
1257: }
1258: if (anychange == 0)
1259: anychange = 1;
1260: switch (format) {
1261: case D_BRIEF:
1262: return;
1263: case D_NORMAL:
1264: range(a, b, ",");
1265: putchar(a > b ? 'a' : c > d ? 'd' : 'c');
1266: if (format == D_NORMAL)
1267: range(c, d, ",");
1268: putchar('\n');
1269: break;
1270: }
1271: if (format == D_NORMAL || format == D_IFDEF) {
1272: fetch(ixold, a, b, f1, '<', 1);
1273: if (a <= b && c <= d && format == D_NORMAL)
1274: puts("---");
1275: }
1276: i = fetch(ixnew, c, d, f2, format == D_NORMAL ? '>' : '\0', 0);
1277: if (inifdef) {
1278: printf("#endif /* %s */\n", ifdefname);
1279: inifdef = 0;
1280: }
1281: }
1282:
1283: static int
1284: fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile)
1285: {
1286: int i, j, c, lastc, col, nc;
1287:
1288: /*
1289: * When doing #ifdef's, copy down to current line
1290: * if this is the first file, so that stuff makes it to output.
1291: */
1292: if (format == D_IFDEF && oldfile) {
1293: long curpos = ftell(lb);
1294: /* print through if append (a>b), else to (nb: 0 vs 1 orig) */
1295: nc = f[a > b ? b : a - 1] - curpos;
1296: for (i = 0; i < nc; i++)
1297: putchar(getc(lb));
1298: }
1299: if (a > b)
1300: return (0);
1301: if (format == D_IFDEF) {
1302: if (inifdef) {
1303: printf("#else /* %s%s */\n",
1304: oldfile == 1 ? "!" : "", ifdefname);
1305: } else {
1306: if (oldfile)
1307: printf("#ifndef %s\n", ifdefname);
1308: else
1309: printf("#ifdef %s\n", ifdefname);
1310: }
1311: inifdef = 1 + oldfile;
1312: }
1313: for (i = a; i <= b; i++) {
1314: fseek(lb, f[i - 1], SEEK_SET);
1315: nc = f[i] - f[i - 1];
1316: if (format != D_IFDEF && ch != '\0') {
1317: putchar(ch);
1318: if (Tflag && (format == D_NORMAL || format == D_CONTEXT
1319: || format == D_UNIFIED))
1320: putchar('\t');
1321: else if (format != D_UNIFIED)
1322: putchar(' ');
1323: }
1324: col = 0;
1325: for (j = 0, lastc = '\0'; j < nc; j++, lastc = c) {
1326: if ((c = getc(lb)) == EOF) {
1327: puts("\n\\ No newline at end of file");
1328: return (0);
1329: }
1330: if (c == '\t' && tflag) {
1331: do {
1332: putchar(' ');
1333: } while (++col & 7);
1334: } else {
1335: putchar(c);
1336: col++;
1337: }
1338: }
1339: }
1340: return (0);
1341: }
1342:
1343: /*
1344: * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578.
1345: */
1346: static int
1347: readhash(FILE *f)
1348: {
1349: int i, t, space;
1350: int sum;
1351:
1352: sum = 1;
1353: space = 0;
1354: if (!bflag && !wflag) {
1355: if (iflag)
1356: for (i = 0; (t = getc(f)) != '\n'; i++) {
1357: if (t == EOF) {
1358: if (i == 0)
1359: return (0);
1360: break;
1361: }
1362: sum = sum * 127 + chrtran[t];
1363: }
1364: else
1365: for (i = 0; (t = getc(f)) != '\n'; i++) {
1366: if (t == EOF) {
1367: if (i == 0)
1368: return (0);
1369: break;
1370: }
1371: sum = sum * 127 + t;
1372: }
1373: } else {
1374: for (i = 0;;) {
1375: switch (t = getc(f)) {
1376: case '\t':
1377: case ' ':
1378: space++;
1379: continue;
1380: default:
1381: if (space && !wflag) {
1382: i++;
1383: space = 0;
1384: }
1385: sum = sum * 127 + chrtran[t];
1386: i++;
1387: continue;
1388: case EOF:
1389: if (i == 0)
1390: return (0);
1391: /* FALLTHROUGH */
1392: case '\n':
1393: break;
1394: }
1395: break;
1396: }
1397: }
1398: /*
1399: * There is a remote possibility that we end up with a zero sum.
1400: * Zero is used as an EOF marker, so return 1 instead.
1401: */
1402: return (sum == 0 ? 1 : sum);
1403: }
1404:
1405: static int
1406: asciifile(FILE *f)
1407: {
1408: char buf[BUFSIZ];
1409: int i, cnt;
1410:
1411: if (aflag || f == NULL)
1412: return (1);
1413:
1414: rewind(f);
1415: cnt = fread(buf, 1, sizeof(buf), f);
1416: for (i = 0; i < cnt; i++)
1417: if (!isprint(buf[i]) && !isspace(buf[i]))
1418: return (0);
1419: return (1);
1420: }
1421:
1422:
1423: /* dump accumulated "context" diff changes */
1424: static void
1425: dump_context_vec(FILE *f1, FILE *f2)
1426: {
1427: struct context_vec *cvp = context_vec_start;
1428: int lowa, upb, lowc, upd, do_output;
1429: int a, b, c, d;
1.4 jfb 1430: char ch;
1.1 jfb 1431:
1432: if (context_vec_start > context_vec_ptr)
1433: return;
1434:
1435: b = d = 0; /* gcc */
1.4 jfb 1436: lowa = MAX(1, cvp->a - context);
1437: upb = MIN(len[0], context_vec_ptr->b + context);
1438: lowc = MAX(1, cvp->c - context);
1439: upd = MIN(len[1], context_vec_ptr->d + context);
1.1 jfb 1440:
1441: printf("***************");
1442: printf("\n*** ");
1443: range(lowa, upb, ",");
1444: printf(" ****\n");
1445:
1446: /*
1447: * Output changes to the "old" file. The first loop suppresses
1448: * output if there were no changes to the "old" file (we'll see
1449: * the "old" lines as context in the "new" list).
1450: */
1451: do_output = 0;
1452: for (; cvp <= context_vec_ptr; cvp++)
1453: if (cvp->a <= cvp->b) {
1454: cvp = context_vec_start;
1455: do_output++;
1456: break;
1457: }
1458: if (do_output) {
1459: while (cvp <= context_vec_ptr) {
1460: a = cvp->a;
1461: b = cvp->b;
1462: c = cvp->c;
1463: d = cvp->d;
1464:
1465: if (a <= b && c <= d)
1466: ch = 'c';
1467: else
1468: ch = (a <= b) ? 'd' : 'a';
1469:
1470: if (ch == 'a')
1471: fetch(ixold, lowa, b, f1, ' ', 0);
1472: else {
1473: fetch(ixold, lowa, a - 1, f1, ' ', 0);
1474: fetch(ixold, a, b, f1,
1475: ch == 'c' ? '!' : '-', 0);
1476: }
1477: lowa = b + 1;
1478: cvp++;
1479: }
1480: fetch(ixold, b + 1, upb, f1, ' ', 0);
1481: }
1482: /* output changes to the "new" file */
1483: printf("--- ");
1484: range(lowc, upd, ",");
1485: printf(" ----\n");
1486:
1487: do_output = 0;
1488: for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++)
1489: if (cvp->c <= cvp->d) {
1490: cvp = context_vec_start;
1491: do_output++;
1492: break;
1493: }
1494: if (do_output) {
1495: while (cvp <= context_vec_ptr) {
1496: a = cvp->a;
1497: b = cvp->b;
1498: c = cvp->c;
1499: d = cvp->d;
1500:
1501: if (a <= b && c <= d)
1502: ch = 'c';
1503: else
1504: ch = (a <= b) ? 'd' : 'a';
1505:
1506: if (ch == 'd')
1507: fetch(ixnew, lowc, d, f2, ' ', 0);
1508: else {
1509: fetch(ixnew, lowc, c - 1, f2, ' ', 0);
1510: fetch(ixnew, c, d, f2,
1511: ch == 'c' ? '!' : '+', 0);
1512: }
1513: lowc = d + 1;
1514: cvp++;
1515: }
1516: fetch(ixnew, d + 1, upd, f2, ' ', 0);
1517: }
1518: context_vec_ptr = context_vec_start - 1;
1519: }
1520:
1521: /* dump accumulated "unified" diff changes */
1522: static void
1523: dump_unified_vec(FILE *f1, FILE *f2)
1524: {
1525: struct context_vec *cvp = context_vec_start;
1526: int lowa, upb, lowc, upd;
1527: int a, b, c, d;
1.4 jfb 1528: char ch;
1.1 jfb 1529:
1530: if (context_vec_start > context_vec_ptr)
1531: return;
1532:
1533: b = d = 0; /* gcc */
1.4 jfb 1534: lowa = MAX(1, cvp->a - context);
1535: upb = MIN(len[0], context_vec_ptr->b + context);
1536: lowc = MAX(1, cvp->c - context);
1537: upd = MIN(len[1], context_vec_ptr->d + context);
1.1 jfb 1538:
1539: fputs("@@ -", stdout);
1540: uni_range(lowa, upb);
1541: fputs(" +", stdout);
1542: uni_range(lowc, upd);
1543: fputs(" @@", stdout);
1544: putchar('\n');
1545:
1546: /*
1547: * Output changes in "unified" diff format--the old and new lines
1548: * are printed together.
1549: */
1550: for (; cvp <= context_vec_ptr; cvp++) {
1551: a = cvp->a;
1552: b = cvp->b;
1553: c = cvp->c;
1554: d = cvp->d;
1555:
1556: /*
1557: * c: both new and old changes
1558: * d: only changes in the old file
1559: * a: only changes in the new file
1560: */
1561: if (a <= b && c <= d)
1562: ch = 'c';
1563: else
1564: ch = (a <= b) ? 'd' : 'a';
1565:
1566: switch (ch) {
1567: case 'c':
1568: fetch(ixold, lowa, a - 1, f1, ' ', 0);
1569: fetch(ixold, a, b, f1, '-', 0);
1570: fetch(ixnew, c, d, f2, '+', 0);
1571: break;
1572: case 'd':
1573: fetch(ixold, lowa, a - 1, f1, ' ', 0);
1574: fetch(ixold, a, b, f1, '-', 0);
1575: break;
1576: case 'a':
1577: fetch(ixnew, lowc, c - 1, f2, ' ', 0);
1578: fetch(ixnew, c, d, f2, '+', 0);
1579: break;
1580: }
1581: lowa = b + 1;
1582: lowc = d + 1;
1583: }
1584: fetch(ixnew, d + 1, upd, f2, ' ', 0);
1585:
1586: context_vec_ptr = context_vec_start - 1;
1587: }