Annotation of src/usr.bin/cvs/diff.c, Revision 1.27
1.27 ! joris 1: /* $OpenBSD: diff.c,v 1.26 2005/04/11 17:56:27 joris 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:
145: #include "cvs.h"
146: #include "log.h"
147: #include "buf.h"
1.4 jfb 148: #include "proto.h"
1.1 jfb 149:
150:
151: #define CVS_DIFF_DEFCTX 3 /* default context length */
152:
153:
154: /*
155: * Output format options
156: */
157: #define D_NORMAL 0 /* Normal output */
158: #define D_CONTEXT 1 /* Diff with context */
159: #define D_UNIFIED 2 /* Unified context diff */
160: #define D_IFDEF 3 /* Diff with merged #ifdef's */
161: #define D_BRIEF 4 /* Say if the files differ */
162:
163: /*
164: * Status values for print_status() and diffreg() return values
165: */
166: #define D_SAME 0 /* Files are the same */
167: #define D_DIFFER 1 /* Files are different */
168: #define D_BINARY 2 /* Binary files are different */
169: #define D_COMMON 3 /* Subdirectory common to both dirs */
170: #define D_ONLY 4 /* Only exists in one directory */
171: #define D_MISMATCH1 5 /* path1 was a dir, path2 a file */
172: #define D_MISMATCH2 6 /* path1 was a file, path2 a dir */
173: #define D_ERROR 7 /* An error occurred */
174: #define D_SKIPPED1 8 /* path1 was a special file */
175: #define D_SKIPPED2 9 /* path2 was a special file */
176:
177: struct cand {
178: int x;
179: int y;
180: int pred;
181: } cand;
182:
183: struct line {
184: int serial;
185: int value;
186: } *file[2];
187:
188: /*
189: * The following struct is used to record change information when
190: * doing a "context" or "unified" diff. (see routine "change" to
191: * understand the highly mnemonic field names)
192: */
193: struct context_vec {
194: int a; /* start line in old file */
195: int b; /* end line in old file */
196: int c; /* start line in new file */
197: int d; /* end line in new file */
198: };
199:
1.4 jfb 200: struct diff_arg {
201: char *rev1;
202: char *rev2;
203: char *date1;
204: char *date2;
205: };
206:
1.1 jfb 207:
208: struct excludes {
209: char *pattern;
210: struct excludes *next;
211: };
212:
213:
214: char *splice(char *, char *);
1.23 joris 215: int cvs_diff_options(char *, int, char **, int *);
1.1 jfb 216: int cvs_diffreg(const char *, const char *);
1.9 jfb 217: int cvs_diff_file(struct cvs_file *, void *);
1.23 joris 218: int cvs_diff_sendflags(struct cvsroot *);
219: int cvs_diff_cleanup(void);
220:
1.1 jfb 221: static void output(const char *, FILE *, const char *, FILE *);
222: static void check(FILE *, FILE *);
223: static void range(int, int, char *);
224: static void uni_range(int, int);
225: static void dump_context_vec(FILE *, FILE *);
226: static void dump_unified_vec(FILE *, FILE *);
1.20 jfb 227: static int prepare(int, FILE *, off_t);
1.1 jfb 228: static void prune(void);
229: static void equiv(struct line *, int, struct line *, int, int *);
230: static void unravel(int);
231: static void unsort(struct line *, int, int *);
232: static void change(const char *, FILE *, const char *, FILE *, int, int, int, int);
233: static void sort(struct line *, int);
234: static int ignoreline(char *);
235: static int asciifile(FILE *);
236: static int fetch(long *, int, int, FILE *, int, int);
237: static int newcand(int, int, int);
238: static int search(int *, int, int);
239: static int skipline(FILE *);
240: static int isqrt(int);
241: static int stone(int *, int, int *, int *);
242: static int readhash(FILE *);
243: static int files_differ(FILE *, FILE *);
1.17 jfb 244: static char *match_function(const long *, int, FILE *);
1.1 jfb 245: static char *preadline(int, size_t, off_t);
246:
247:
1.17 jfb 248: static int aflag, bflag, dflag, iflag, Nflag, pflag, tflag, Tflag, wflag;
1.1 jfb 249: static int context, status;
250: static int format = D_NORMAL;
251: static struct stat stb1, stb2;
1.2 jfb 252: static char *ifdefname, *ignore_pats, diffargs[128];
253: static const char *diff_file;
1.1 jfb 254: regex_t ignore_re;
255:
256: static int *J; /* will be overlaid on class */
257: static int *class; /* will be overlaid on file[0] */
258: static int *klist; /* will be overlaid on file[0] after class */
259: static int *member; /* will be overlaid on file[1] */
260: static int clen;
261: static int inifdef; /* whether or not we are in a #ifdef block */
262: static int len[2];
263: static int pref, suff; /* length of prefix and suffix */
264: static int slen[2];
265: static int anychange;
266: static long *ixnew; /* will be overlaid on file[1] */
267: static long *ixold; /* will be overlaid on klist */
268: static struct cand *clist; /* merely a free storage pot for candidates */
269: static int clistlen; /* the length of clist */
270: static struct line *sfile[2]; /* shortened by pruning common prefix/suffix */
271: static u_char *chrtran; /* translation table for case-folding */
272: static struct context_vec *context_vec_start;
273: static struct context_vec *context_vec_end;
274: static struct context_vec *context_vec_ptr;
275:
276: #define FUNCTION_CONTEXT_SIZE 41
1.17 jfb 277: static char lastbuf[FUNCTION_CONTEXT_SIZE];
278: static int lastline;
279: static int lastmatchline;
1.1 jfb 280:
281:
282: /*
283: * chrtran points to one of 2 translation tables: cup2low if folding upper to
284: * lower case clow2low if not folding case
285: */
286: u_char clow2low[256] = {
287: 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a,
288: 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15,
289: 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
290: 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b,
291: 0x2c, 0x2d, 0x2e, 0x2f, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36,
292: 0x37, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x40, 0x41,
293: 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x4a, 0x4b, 0x4c,
294: 0x4d, 0x4e, 0x4f, 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,
295: 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, 0x60, 0x61, 0x62,
296: 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d,
297: 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78,
298: 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x80, 0x81, 0x82, 0x83,
299: 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e,
300: 0x8f, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99,
301: 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xa4,
302: 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,
303: 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba,
304: 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5,
305: 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf, 0xd0,
306: 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xdb,
307: 0xdc, 0xdd, 0xde, 0xdf, 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6,
308: 0xe7, 0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, 0xf0, 0xf1,
309: 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa, 0xfb, 0xfc,
310: 0xfd, 0xfe, 0xff
311: };
312:
313: u_char cup2low[256] = {
314: 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a,
315: 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15,
316: 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
317: 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b,
318: 0x2c, 0x2d, 0x2e, 0x2f, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36,
319: 0x37, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x60, 0x61,
320: 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c,
321: 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
322: 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x60, 0x61, 0x62,
323: 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d,
324: 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78,
325: 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x80, 0x81, 0x82, 0x83,
326: 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e,
327: 0x8f, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99,
328: 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xa4,
329: 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,
330: 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba,
331: 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5,
332: 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf, 0xd0,
333: 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xdb,
334: 0xdc, 0xdd, 0xde, 0xdf, 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6,
335: 0xe7, 0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, 0xf0, 0xf1,
336: 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa, 0xfb, 0xfc,
337: 0xfd, 0xfe, 0xff
338: };
339:
1.23 joris 340: struct cvs_cmd_info cvs_diff = {
341: cvs_diff_options,
342: cvs_diff_sendflags,
343: cvs_diff_file,
344: cvs_diff_cleanup,
345: NULL,
346: CF_RECURSE | CF_IGNORE | CF_SORT | CF_KNOWN,
347: CVS_REQ_DIFF,
348: CVS_CMD_SENDDIR
349: };
350:
351: static struct diff_arg *dap = NULL;
352: static int recurse;
1.1 jfb 353:
354: int
1.23 joris 355: cvs_diff_options(char *opt, int argc, char **argv, int *arg)
1.1 jfb 356: {
1.23 joris 357: int ch;
1.1 jfb 358:
1.23 joris 359: dap = (struct diff_arg *)malloc(sizeof(*dap));
360: if (dap == NULL)
1.25 xsa 361: return (-1);
1.23 joris 362: dap->date1 = dap->date2 = dap->rev1 = dap->rev2 = NULL;
1.2 jfb 363: strlcpy(diffargs, argv[0], sizeof(diffargs));
1.1 jfb 364:
1.23 joris 365: while ((ch = getopt(argc, argv, opt)) != -1) {
1.1 jfb 366: switch (ch) {
367: case 'c':
1.2 jfb 368: strlcat(diffargs, " -c", sizeof(diffargs));
1.1 jfb 369: format = D_CONTEXT;
370: break;
371: case 'D':
1.23 joris 372: if (dap->date1 == NULL && dap->rev1 == NULL) {
373: dap->date1 = optarg;
374: } else if (dap->date2 == NULL && dap->rev2 == NULL) {
375: dap->date2 = optarg;
376: } else {
1.1 jfb 377: cvs_log(LP_ERR,
378: "no more than two revisions/dates can "
379: "be specified");
380: }
381: break;
382: case 'l':
1.2 jfb 383: strlcat(diffargs, " -l", sizeof(diffargs));
1.1 jfb 384: recurse = 0;
1.23 joris 385: cvs_diff.file_flags &= ~CF_RECURSE;
1.1 jfb 386: break;
387: case 'i':
1.2 jfb 388: strlcat(diffargs, " -i", sizeof(diffargs));
1.1 jfb 389: iflag = 1;
390: break;
1.8 jfb 391: case 'N':
392: strlcat(diffargs, " -N", sizeof(diffargs));
393: Nflag = 1;
394: break;
1.17 jfb 395: case 'p':
396: strlcat(diffargs, " -p", sizeof(diffargs));
397: pflag = 1;
398: break;
1.1 jfb 399: case 'r':
1.23 joris 400: if ((dap->rev1 == NULL) && (dap->date1 == NULL)) {
401: dap->rev1 = optarg;
402: } else if ((dap->rev2 == NULL) &&
403: (dap->date2 == NULL)) {
404: dap->rev2 = optarg;
405: } else {
1.1 jfb 406: cvs_log(LP_ERR,
407: "no more than two revisions/dates can "
408: "be specified");
1.26 joris 409: return (1);
1.1 jfb 410: }
411: break;
412: case 'u':
1.2 jfb 413: strlcat(diffargs, " -u", sizeof(diffargs));
1.1 jfb 414: format = D_UNIFIED;
415: break;
416: default:
1.26 joris 417: return (1);
1.1 jfb 418: }
419: }
420:
1.23 joris 421: *arg = optind;
422: return (0);
423: }
1.1 jfb 424:
1.23 joris 425: int
426: cvs_diff_cleanup(void)
427: {
428: if (dap != NULL)
429: free(dap);
1.4 jfb 430: return (0);
431: }
1.1 jfb 432:
1.4 jfb 433: /*
434: * cvs_diff_sendflags()
435: *
436: */
437: int
1.23 joris 438: cvs_diff_sendflags(struct cvsroot *root)
1.4 jfb 439: {
440: /* send the flags */
1.17 jfb 441: if (Nflag && (cvs_sendarg(root, "-N", 0) < 0))
442: return (-1);
443: if (pflag && (cvs_sendarg(root, "-p", 0) < 0))
444: return (-1);
445:
1.4 jfb 446: if (format == D_CONTEXT)
447: cvs_sendarg(root, "-c", 0);
448: else if (format == D_UNIFIED)
449: cvs_sendarg(root, "-u", 0);
450:
451: if (dap->rev1 != NULL) {
452: cvs_sendarg(root, "-r", 0);
1.14 jfb 453: cvs_sendarg(root, dap->rev1, 0);
1.10 deraadt 454: } else if (dap->date1 != NULL) {
1.4 jfb 455: cvs_sendarg(root, "-D", 0);
1.14 jfb 456: cvs_sendarg(root, dap->date1, 0);
1.4 jfb 457: }
458: if (dap->rev2 != NULL) {
459: cvs_sendarg(root, "-r", 0);
1.14 jfb 460: cvs_sendarg(root, dap->rev2, 0);
1.10 deraadt 461: } else if (dap->date2 != NULL) {
1.4 jfb 462: cvs_sendarg(root, "-D", 0);
1.14 jfb 463: cvs_sendarg(root, dap->date2, 0);
1.1 jfb 464: }
465:
466: return (0);
467: }
468:
469:
470: /*
471: * cvs_diff_file()
472: *
473: * Diff a single file.
474: */
475: int
1.4 jfb 476: cvs_diff_file(struct cvs_file *cfp, void *arg)
1.1 jfb 477: {
1.8 jfb 478: char *dir, *repo, buf[64];
479: char fpath[MAXPATHLEN], dfpath[MAXPATHLEN], rcspath[MAXPATHLEN];
1.12 djm 480: char path_tmp1[MAXPATHLEN], path_tmp2[MAXPATHLEN];
1.1 jfb 481: BUF *b1, *b2;
482: RCSNUM *r1, *r2;
483: RCSFILE *rf;
484: struct cvs_ent *entp;
1.4 jfb 485: struct cvsroot *root;
1.1 jfb 486:
1.4 jfb 487: if (cfp->cf_type == DT_DIR) {
1.7 jfb 488: if (cfp->cf_cvstat == CVS_FST_UNKNOWN) {
489: root = cfp->cf_parent->cf_ddat->cd_root;
1.8 jfb 490: cvs_sendreq(root, CVS_REQ_QUESTIONABLE,
491: CVS_FILE_NAME(cfp));
1.10 deraadt 492: } else {
1.7 jfb 493: root = cfp->cf_ddat->cd_root;
1.23 joris 494: #if 0
1.7 jfb 495: if ((cfp->cf_parent == NULL) ||
496: (root != cfp->cf_parent->cf_ddat->cd_root)) {
497: cvs_connect(root);
1.23 joris 498: cvs_diff_sendflags(root);
1.7 jfb 499: }
1.23 joris 500: #endif
1.7 jfb 501:
502: cvs_senddir(root, cfp);
1.4 jfb 503: }
1.1 jfb 504:
1.13 jfb 505: return (0);
506: }
507:
508: if (cfp->cf_cvstat == CVS_FST_LOST) {
509: cvs_log(LP_WARN, "cannot find file %s", CVS_FILE_NAME(cfp));
1.4 jfb 510: return (0);
1.1 jfb 511: }
512:
1.4 jfb 513: rf = NULL;
1.8 jfb 514: diff_file = cvs_file_getpath(cfp, fpath, sizeof(fpath));
1.7 jfb 515:
1.4 jfb 516: if (cfp->cf_parent != NULL) {
1.8 jfb 517: dir = cvs_file_getpath(cfp->cf_parent, dfpath, sizeof(dfpath));
1.7 jfb 518: root = cfp->cf_parent->cf_ddat->cd_root;
1.4 jfb 519: repo = cfp->cf_parent->cf_ddat->cd_repo;
1.10 deraadt 520: } else {
1.4 jfb 521: dir = ".";
1.7 jfb 522: root = NULL;
1.4 jfb 523: repo = NULL;
1.1 jfb 524: }
525:
1.4 jfb 526: if (cfp->cf_cvstat == CVS_FST_UNKNOWN) {
527: if (root->cr_method == CVS_METHOD_LOCAL)
528: cvs_log(LP_WARN, "I know nothing about %s", diff_file);
529: else
1.8 jfb 530: cvs_sendreq(root, CVS_REQ_QUESTIONABLE,
531: CVS_FILE_NAME(cfp));
1.4 jfb 532: return (0);
1.1 jfb 533: }
534:
1.4 jfb 535: entp = cvs_ent_getent(diff_file);
536: if (entp == NULL)
537: return (-1);
1.1 jfb 538:
1.4 jfb 539: if (root->cr_method != CVS_METHOD_LOCAL) {
540: if (cvs_sendentry(root, entp) < 0) {
541: cvs_ent_free(entp);
542: return (-1);
543: }
1.1 jfb 544: }
545:
1.4 jfb 546: if (cfp->cf_cvstat == CVS_FST_UPTODATE) {
547: if (root->cr_method != CVS_METHOD_LOCAL)
1.8 jfb 548: cvs_sendreq(root, CVS_REQ_UNCHANGED,
549: CVS_FILE_NAME(cfp));
1.4 jfb 550: cvs_ent_free(entp);
1.1 jfb 551: return (0);
552: }
553:
554: /* at this point, the file is modified */
1.4 jfb 555: if (root->cr_method != CVS_METHOD_LOCAL) {
1.8 jfb 556: cvs_sendreq(root, CVS_REQ_MODIFIED, CVS_FILE_NAME(cfp));
1.4 jfb 557: cvs_sendfile(root, diff_file);
1.10 deraadt 558: } else {
1.1 jfb 559: snprintf(rcspath, sizeof(rcspath), "%s/%s/%s%s",
1.4 jfb 560: root->cr_dir, repo, diff_file, RCS_FILE_EXT);
1.1 jfb 561:
1.22 jfb 562: rf = rcs_open(rcspath, RCS_READ);
1.4 jfb 563: if (rf == NULL) {
564: cvs_ent_free(entp);
1.1 jfb 565: return (-1);
1.4 jfb 566: }
1.1 jfb 567:
1.4 jfb 568: cvs_printf("Index: %s\n%s\nRCS file: %s\n", diff_file,
1.1 jfb 569: RCS_DIFF_DIV, rcspath);
570:
1.4 jfb 571: if (dap->rev1 == NULL)
1.1 jfb 572: r1 = entp->ce_rev;
573: else {
1.21 jfb 574: if ((r1 = rcsnum_parse(dap->rev1)) == NULL) {
1.20 jfb 575: cvs_ent_free(entp);
576: return (-1);
577: }
1.1 jfb 578: }
579:
1.4 jfb 580: cvs_printf("retrieving revision %s\n",
1.1 jfb 581: rcsnum_tostr(r1, buf, sizeof(buf)));
582: b1 = rcs_getrev(rf, r1);
583:
1.20 jfb 584: if (r1 != entp->ce_rev)
585: rcsnum_free(r1);
586:
1.4 jfb 587: if (dap->rev2 != NULL) {
588: cvs_printf("retrieving revision %s\n", dap->rev2);
1.21 jfb 589: if ((r2 = rcsnum_parse(dap->rev2)) == NULL) {
1.20 jfb 590: cvs_ent_free(entp);
591: return (-1);
592: }
1.1 jfb 593: b2 = rcs_getrev(rf, r2);
1.20 jfb 594: rcsnum_free(r2);
1.10 deraadt 595: } else {
1.4 jfb 596: b2 = cvs_buf_load(diff_file, BUF_AUTOEXT);
1.1 jfb 597: }
598:
1.4 jfb 599: rcs_close(rf);
600:
1.2 jfb 601: printf("%s", diffargs);
602: printf(" -r%s", buf);
1.4 jfb 603: if (dap->rev2 != NULL)
604: printf(" -r%s", dap->rev2);
605: printf(" %s\n", diff_file);
1.12 djm 606: strlcpy(path_tmp1, "/tmp/diff1.XXXXXXXXXX", sizeof(path_tmp1));
1.20 jfb 607: if (cvs_buf_write_stmp(b1, path_tmp1, 0600) == -1) {
608: cvs_buf_free(b1);
609: cvs_buf_free(b2);
1.12 djm 610: return (-1);
1.20 jfb 611: }
612: cvs_buf_free(b1);
613:
1.19 jfb 614: strlcpy(path_tmp2, "/tmp/diff2.XXXXXXXXXX", sizeof(path_tmp2));
1.12 djm 615: if (cvs_buf_write_stmp(b2, path_tmp2, 0600) == -1) {
1.20 jfb 616: cvs_buf_free(b2);
1.12 djm 617: (void)unlink(path_tmp1);
618: return (-1);
619: }
1.20 jfb 620: cvs_buf_free(b2);
621:
1.12 djm 622: cvs_diffreg(path_tmp1, path_tmp2);
623: (void)unlink(path_tmp1);
624: (void)unlink(path_tmp2);
1.1 jfb 625: }
626:
1.4 jfb 627: cvs_ent_free(entp);
1.1 jfb 628: return (0);
629: }
630:
631:
632: int
633: cvs_diffreg(const char *file1, const char *file2)
634: {
635: FILE *f1, *f2;
636: int i, rval;
1.20 jfb 637: void *tmp;
1.1 jfb 638:
639: f1 = f2 = NULL;
640: rval = D_SAME;
641: anychange = 0;
642: lastline = 0;
643: lastmatchline = 0;
644: context_vec_ptr = context_vec_start - 1;
645: chrtran = (iflag ? cup2low : clow2low);
646:
647: f1 = fopen(file1, "r");
648: if (f1 == NULL) {
649: cvs_log(LP_ERRNO, "%s", file1);
650: status |= 2;
651: goto closem;
652: }
653:
654: f2 = fopen(file2, "r");
655: if (f2 == NULL) {
656: cvs_log(LP_ERRNO, "%s", file2);
657: status |= 2;
658: goto closem;
659: }
660:
661: switch (files_differ(f1, f2)) {
662: case 0:
663: goto closem;
664: case 1:
665: break;
666: default:
667: /* error */
668: status |= 2;
669: goto closem;
670: }
671:
672: if (!asciifile(f1) || !asciifile(f2)) {
673: rval = D_BINARY;
674: status |= 1;
675: goto closem;
676: }
1.20 jfb 677: if ((prepare(0, f1, stb1.st_size) < 0) ||
678: (prepare(1, f2, stb2.st_size) < 0)) {
679: status |= 2;
680: goto closem;
681: }
1.1 jfb 682: prune();
683: sort(sfile[0], slen[0]);
684: sort(sfile[1], slen[1]);
685:
686: member = (int *)file[1];
687: equiv(sfile[0], slen[0], sfile[1], slen[1], member);
1.20 jfb 688: if ((tmp = realloc(member, (slen[1] + 2) * sizeof(int))) == NULL) {
689: status |= 2;
690: goto closem;
691: }
692: member = (int *)tmp;
1.1 jfb 693:
694: class = (int *)file[0];
695: unsort(sfile[0], slen[0], class);
1.20 jfb 696: if ((tmp = realloc(class, (slen[0] + 2) * sizeof(int))) == NULL) {
697: status |= 2;
698: goto closem;
699: }
700: class = (int *)tmp;
1.1 jfb 701:
1.20 jfb 702: if ((klist = malloc((slen[0] + 2) * sizeof(int))) == NULL) {
703: cvs_log(LP_ERRNO, "failed to allocate klist");
704: status |= 2;
705: goto closem;
706: }
1.1 jfb 707: clen = 0;
708: clistlen = 100;
1.20 jfb 709: if ((clist = malloc(clistlen * sizeof(cand))) == NULL) {
710: cvs_log(LP_ERRNO, "failed to allocate clist");
711: status |= 2;
712: goto closem;
713: }
1.1 jfb 714: i = stone(class, slen[0], member, klist);
715: free(member);
716: free(class);
717:
718: J = realloc(J, (len[0] + 2) * sizeof(int));
719: unravel(klist[i]);
720: free(clist);
721: free(klist);
722:
723: ixold = realloc(ixold, (len[0] + 2) * sizeof(long));
724: ixnew = realloc(ixnew, (len[1] + 2) * sizeof(long));
725: check(f1, f2);
726: output(file1, f1, file2, f2);
727:
728: closem:
729: if (anychange) {
730: status |= 1;
731: if (rval == D_SAME)
732: rval = D_DIFFER;
733: }
734: if (f1 != NULL)
735: fclose(f1);
736: if (f2 != NULL)
737: fclose(f2);
1.20 jfb 738:
1.1 jfb 739: return (rval);
740: }
741:
742: /*
743: * Check to see if the given files differ.
744: * Returns 0 if they are the same, 1 if different, and -1 on error.
745: * XXX - could use code from cmp(1) [faster]
746: */
747: static int
748: files_differ(FILE *f1, FILE *f2)
749: {
750: char buf1[BUFSIZ], buf2[BUFSIZ];
751: size_t i, j;
752:
753: if (stb1.st_size != stb2.st_size)
754: return (1);
755: for (;;) {
756: i = fread(buf1, 1, sizeof(buf1), f1);
757: j = fread(buf2, 1, sizeof(buf2), f2);
758: if (i != j)
759: return (1);
760: if (i == 0 && j == 0) {
761: if (ferror(f1) || ferror(f2))
762: return (1);
763: return (0);
764: }
765: if (memcmp(buf1, buf2, i) != 0)
766: return (1);
767: }
768: }
769:
770:
771: char *
1.15 jfb 772: splice(char *dir, char *filename)
1.1 jfb 773: {
774: char *tail, *buf;
775:
1.15 jfb 776: if ((tail = strrchr(filename, '/')) == NULL)
777: tail = filename;
1.1 jfb 778: else
779: tail++;
780: asprintf(&buf, "%s/%s", dir, tail);
781: return (buf);
782: }
783:
1.20 jfb 784: static int
1.1 jfb 785: prepare(int i, FILE *fd, off_t filesize)
786: {
1.20 jfb 787: void *tmp;
1.1 jfb 788: struct line *p;
789: int j, h;
790: size_t sz;
791:
792: rewind(fd);
793:
1.16 xsa 794: sz = ((size_t)filesize <= SIZE_MAX ? (size_t)filesize : SIZE_MAX) / 25;
1.1 jfb 795: if (sz < 100)
796: sz = 100;
797:
1.20 jfb 798: p = (struct line *)malloc((sz + 3) * sizeof(struct line));
799: if (p == NULL) {
800: cvs_log(LP_ERRNO, "failed to prepare line array");
801: return (-1);
802: }
1.1 jfb 803: for (j = 0; (h = readhash(fd));) {
804: if (j == (int)sz) {
805: sz = sz * 3 / 2;
1.20 jfb 806: tmp = realloc(p, (sz + 3) * sizeof(struct line));
807: if (tmp == NULL) {
808: cvs_log(LP_ERRNO, "failed to grow line array");
809: free(p);
810: return (-1);
811: }
812: p = (struct line *)tmp;
1.1 jfb 813: }
814: p[++j].value = h;
815: }
816: len[i] = j;
817: file[i] = p;
1.20 jfb 818:
819: return (0);
1.1 jfb 820: }
821:
822: static void
823: prune(void)
824: {
825: int i, j;
826:
827: for (pref = 0; pref < len[0] && pref < len[1] &&
828: file[0][pref + 1].value == file[1][pref + 1].value;
829: pref++)
830: ;
831: for (suff = 0; suff < len[0] - pref && suff < len[1] - pref &&
832: file[0][len[0] - suff].value == file[1][len[1] - suff].value;
833: suff++)
834: ;
835: for (j = 0; j < 2; j++) {
836: sfile[j] = file[j] + pref;
837: slen[j] = len[j] - pref - suff;
838: for (i = 0; i <= slen[j]; i++)
839: sfile[j][i].serial = i;
840: }
841: }
842:
843: static void
844: equiv(struct line *a, int n, struct line *b, int m, int *c)
845: {
846: int i, j;
847:
848: i = j = 1;
849: while (i <= n && j <= m) {
850: if (a[i].value < b[j].value)
851: a[i++].value = 0;
852: else if (a[i].value == b[j].value)
853: a[i++].value = j;
854: else
855: j++;
856: }
857: while (i <= n)
858: a[i++].value = 0;
859: b[m + 1].value = 0;
860: j = 0;
861: while (++j <= m) {
862: c[j] = -b[j].serial;
863: while (b[j + 1].value == b[j].value) {
864: j++;
865: c[j] = b[j].serial;
866: }
867: }
868: c[j] = -1;
869: }
870:
871: /* Code taken from ping.c */
872: static int
873: isqrt(int n)
874: {
875: int y, x = 1;
876:
877: if (n == 0)
1.18 jfb 878: return (0);
1.1 jfb 879:
880: do { /* newton was a stinker */
881: y = x;
882: x = n / x;
883: x += y;
884: x /= 2;
885: } while ((x - y) > 1 || (x - y) < -1);
886:
887: return (x);
888: }
889:
890: static int
891: stone(int *a, int n, int *b, int *c)
892: {
893: int i, k, y, j, l;
894: int oldc, tc, oldl;
895: u_int numtries;
896:
1.18 jfb 897: /* XXX move the isqrt() out of the macro to avoid multiple calls */
898: const u_int bound = dflag ? UINT_MAX : MAX(256, (u_int)isqrt(n));
1.1 jfb 899:
900: k = 0;
901: c[0] = newcand(0, 0, 0);
902: for (i = 1; i <= n; i++) {
903: j = a[i];
904: if (j == 0)
905: continue;
906: y = -b[j];
907: oldl = 0;
908: oldc = c[0];
909: numtries = 0;
910: do {
911: if (y <= clist[oldc].y)
912: continue;
913: l = search(c, k, y);
914: if (l != oldl + 1)
915: oldc = c[l - 1];
916: if (l <= k) {
917: if (clist[c[l]].y <= y)
918: continue;
919: tc = c[l];
920: c[l] = newcand(i, y, oldc);
921: oldc = tc;
922: oldl = l;
923: numtries++;
924: } else {
925: c[l] = newcand(i, y, oldc);
926: k++;
927: break;
928: }
929: } while ((y = b[++j]) > 0 && numtries < bound);
930: }
931: return (k);
932: }
933:
934: static int
935: newcand(int x, int y, int pred)
936: {
937: struct cand *q;
938:
939: if (clen == clistlen) {
940: clistlen = clistlen * 11 / 10;
941: clist = realloc(clist, clistlen * sizeof(cand));
1.20 jfb 942: if (clist == NULL) {
943: cvs_log(LP_ERRNO, "failed to resize clist");
944: return (-1);
945: }
1.1 jfb 946: }
947: q = clist + clen;
948: q->x = x;
949: q->y = y;
950: q->pred = pred;
951: return (clen++);
952: }
953:
954: static int
955: search(int *c, int k, int y)
956: {
957: int i, j, l, t;
958:
959: if (clist[c[k]].y < y) /* quick look for typical case */
960: return (k + 1);
961: i = 0;
962: j = k + 1;
963: while (1) {
964: l = i + j;
965: if ((l >>= 1) <= i)
966: break;
967: t = clist[c[l]].y;
968: if (t > y)
969: j = l;
970: else if (t < y)
971: i = l;
972: else
973: return (l);
974: }
975: return (l + 1);
976: }
977:
978: static void
979: unravel(int p)
980: {
981: struct cand *q;
982: int i;
983:
984: for (i = 0; i <= len[0]; i++)
985: J[i] = i <= pref ? i :
986: i > len[0] - suff ? i + len[1] - len[0] : 0;
987: for (q = clist + p; q->y != 0; q = clist + q->pred)
988: J[q->x + pref] = q->y + pref;
989: }
990:
991: /*
992: * Check does double duty:
993: * 1. ferret out any fortuitous correspondences due
994: * to confounding by hashing (which result in "jackpot")
995: * 2. collect random access indexes to the two files
996: */
997: static void
998: check(FILE *f1, FILE *f2)
999: {
1000: int i, j, jackpot, c, d;
1001: long ctold, ctnew;
1002:
1003: rewind(f1);
1004: rewind(f2);
1005: j = 1;
1006: ixold[0] = ixnew[0] = 0;
1007: jackpot = 0;
1008: ctold = ctnew = 0;
1009: for (i = 1; i <= len[0]; i++) {
1010: if (J[i] == 0) {
1011: ixold[i] = ctold += skipline(f1);
1012: continue;
1013: }
1014: while (j < J[i]) {
1015: ixnew[j] = ctnew += skipline(f2);
1016: j++;
1017: }
1018: if (bflag || wflag || iflag) {
1019: for (;;) {
1020: c = getc(f1);
1021: d = getc(f2);
1022: /*
1023: * GNU diff ignores a missing newline
1024: * in one file if bflag || wflag.
1025: */
1026: if ((bflag || wflag) &&
1027: ((c == EOF && d == '\n') ||
1028: (c == '\n' && d == EOF))) {
1029: break;
1030: }
1031: ctold++;
1032: ctnew++;
1033: if (bflag && isspace(c) && isspace(d)) {
1034: do {
1035: if (c == '\n')
1036: break;
1037: ctold++;
1038: } while (isspace(c = getc(f1)));
1039: do {
1040: if (d == '\n')
1041: break;
1042: ctnew++;
1043: } while (isspace(d = getc(f2)));
1044: } else if (wflag) {
1045: while (isspace(c) && c != '\n') {
1046: c = getc(f1);
1047: ctold++;
1048: }
1049: while (isspace(d) && d != '\n') {
1050: d = getc(f2);
1051: ctnew++;
1052: }
1053: }
1054: if (chrtran[c] != chrtran[d]) {
1055: jackpot++;
1056: J[i] = 0;
1057: if (c != '\n' && c != EOF)
1058: ctold += skipline(f1);
1059: if (d != '\n' && c != EOF)
1060: ctnew += skipline(f2);
1061: break;
1062: }
1063: if (c == '\n' || c == EOF)
1064: break;
1065: }
1066: } else {
1067: for (;;) {
1068: ctold++;
1069: ctnew++;
1070: if ((c = getc(f1)) != (d = getc(f2))) {
1071: /* jackpot++; */
1072: J[i] = 0;
1073: if (c != '\n' && c != EOF)
1074: ctold += skipline(f1);
1075: if (d != '\n' && c != EOF)
1076: ctnew += skipline(f2);
1077: break;
1078: }
1079: if (c == '\n' || c == EOF)
1080: break;
1081: }
1082: }
1083: ixold[i] = ctold;
1084: ixnew[j] = ctnew;
1085: j++;
1086: }
1087: for (; j <= len[1]; j++)
1088: ixnew[j] = ctnew += skipline(f2);
1089: /*
1090: * if (jackpot)
1091: * fprintf(stderr, "jackpot\n");
1092: */
1093: }
1094:
1095: /* shellsort CACM #201 */
1096: static void
1097: sort(struct line *a, int n)
1098: {
1099: struct line *ai, *aim, w;
1100: int j, m = 0, k;
1101:
1102: if (n == 0)
1103: return;
1104: for (j = 1; j <= n; j *= 2)
1105: m = 2 * j - 1;
1106: for (m /= 2; m != 0; m /= 2) {
1107: k = n - m;
1108: for (j = 1; j <= k; j++) {
1109: for (ai = &a[j]; ai > a; ai -= m) {
1110: aim = &ai[m];
1111: if (aim < ai)
1112: break; /* wraparound */
1113: if (aim->value > ai[0].value ||
1114: (aim->value == ai[0].value &&
1115: aim->serial > ai[0].serial))
1116: break;
1117: w.value = ai[0].value;
1118: ai[0].value = aim->value;
1119: aim->value = w.value;
1120: w.serial = ai[0].serial;
1121: ai[0].serial = aim->serial;
1122: aim->serial = w.serial;
1123: }
1124: }
1125: }
1126: }
1127:
1128: static void
1129: unsort(struct line *f, int l, int *b)
1130: {
1131: int *a, i;
1132:
1.20 jfb 1133: if ((a = (int *)malloc((l + 1) * sizeof(int))) == NULL) {
1134: cvs_log(LP_ERRNO, "failed to allocate sort array");
1135: return;
1136: }
1.1 jfb 1137: for (i = 1; i <= l; i++)
1138: a[f[i].serial] = f[i].value;
1139: for (i = 1; i <= l; i++)
1140: b[i] = a[i];
1141: free(a);
1142: }
1143:
1144: static int
1145: skipline(FILE *f)
1146: {
1147: int i, c;
1148:
1149: for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++)
1150: continue;
1151: return (i);
1152: }
1153:
1154: static void
1155: output(const char *file1, FILE *f1, const char *file2, FILE *f2)
1156: {
1157: int m, i0, i1, j0, j1;
1158:
1159: rewind(f1);
1160: rewind(f2);
1161: m = len[0];
1162: J[0] = 0;
1163: J[m + 1] = len[1] + 1;
1164: for (i0 = 1; i0 <= m; i0 = i1 + 1) {
1165: while (i0 <= m && J[i0] == J[i0 - 1] + 1)
1166: i0++;
1167: j0 = J[i0 - 1] + 1;
1168: i1 = i0 - 1;
1169: while (i1 < m && J[i1 + 1] == 0)
1170: i1++;
1171: j1 = J[i1 + 1] - 1;
1172: J[i1] = j1;
1173: change(file1, f1, file2, f2, i0, i1, j0, j1);
1174: }
1175: if (m == 0)
1176: change(file1, f1, file2, f2, 1, 0, 1, len[1]);
1177: if (format == D_IFDEF) {
1178: for (;;) {
1179: #define c i0
1180: if ((c = getc(f1)) == EOF)
1181: return;
1182: putchar(c);
1183: }
1184: #undef c
1185: }
1186: if (anychange != 0) {
1187: if (format == D_CONTEXT)
1188: dump_context_vec(f1, f2);
1189: else if (format == D_UNIFIED)
1190: dump_unified_vec(f1, f2);
1191: }
1192: }
1193:
1194: static __inline void
1195: range(int a, int b, char *separator)
1196: {
1197: printf("%d", a > b ? b : a);
1198: if (a < b)
1199: printf("%s%d", separator, b);
1200: }
1201:
1202: static __inline void
1203: uni_range(int a, int b)
1204: {
1205: if (a < b)
1206: printf("%d,%d", a, b - a + 1);
1207: else if (a == b)
1208: printf("%d", b);
1209: else
1210: printf("%d,0", b);
1211: }
1212:
1213: static char *
1.15 jfb 1214: preadline(int fd, size_t rlen, off_t off)
1.1 jfb 1215: {
1216: char *line;
1217: ssize_t nr;
1218:
1.15 jfb 1219: line = malloc(rlen + 1);
1.1 jfb 1220: if (line == NULL) {
1221: cvs_log(LP_ERRNO, "failed to allocate line");
1222: return (NULL);
1223: }
1.15 jfb 1224: if ((nr = pread(fd, line, rlen, off)) < 0) {
1.1 jfb 1225: cvs_log(LP_ERRNO, "preadline failed");
1226: return (NULL);
1227: }
1228: line[nr] = '\0';
1229: return (line);
1230: }
1231:
1232: static int
1233: ignoreline(char *line)
1234: {
1235: int ret;
1236:
1237: ret = regexec(&ignore_re, line, 0, NULL, 0);
1238: free(line);
1239: return (ret == 0); /* if it matched, it should be ignored. */
1240: }
1241:
1242: /*
1243: * Indicate that there is a difference between lines a and b of the from file
1244: * to get to lines c to d of the to file. If a is greater then b then there
1245: * are no lines in the from file involved and this means that there were
1246: * lines appended (beginning at b). If c is greater than d then there are
1247: * lines missing from the to file.
1248: */
1249: static void
1250: change(const char *file1, FILE *f1, const char *file2, FILE *f2,
1251: int a, int b, int c, int d)
1252: {
1253: static size_t max_context = 64;
1254: int i;
1255:
1256: if (format != D_IFDEF && a > b && c > d)
1257: return;
1258: if (ignore_pats != NULL) {
1259: char *line;
1260: /*
1261: * All lines in the change, insert, or delete must
1262: * match an ignore pattern for the change to be
1263: * ignored.
1264: */
1265: if (a <= b) { /* Changes and deletes. */
1266: for (i = a; i <= b; i++) {
1267: line = preadline(fileno(f1),
1268: ixold[i] - ixold[i - 1], ixold[i - 1]);
1269: if (!ignoreline(line))
1270: goto proceed;
1271: }
1272: }
1273: if (a > b || c <= d) { /* Changes and inserts. */
1274: for (i = c; i <= d; i++) {
1275: line = preadline(fileno(f2),
1276: ixnew[i] - ixnew[i - 1], ixnew[i - 1]);
1277: if (!ignoreline(line))
1278: goto proceed;
1279: }
1280: }
1281: return;
1282: }
1283: proceed:
1284: if (format == D_CONTEXT || format == D_UNIFIED) {
1285: /*
1286: * Allocate change records as needed.
1287: */
1288: if (context_vec_ptr == context_vec_end - 1) {
1289: ptrdiff_t offset = context_vec_ptr - context_vec_start;
1290: max_context <<= 1;
1291: context_vec_start = realloc(context_vec_start,
1292: max_context * sizeof(struct context_vec));
1293: context_vec_end = context_vec_start + max_context;
1294: context_vec_ptr = context_vec_start + offset;
1295: }
1296: if (anychange == 0) {
1297: /*
1298: * Print the context/unidiff header first time through.
1299: */
1300: printf("%s %s %s",
1301: format == D_CONTEXT ? "***" : "---", diff_file,
1302: ctime(&stb1.st_mtime));
1303: printf("%s %s %s",
1304: format == D_CONTEXT ? "---" : "+++", diff_file,
1305: ctime(&stb2.st_mtime));
1306: anychange = 1;
1307: } else if (a > context_vec_ptr->b + (2 * context) + 1 &&
1308: c > context_vec_ptr->d + (2 * context) + 1) {
1309: /*
1310: * If this change is more than 'context' lines from the
1311: * previous change, dump the record and reset it.
1312: */
1313: if (format == D_CONTEXT)
1314: dump_context_vec(f1, f2);
1315: else
1316: dump_unified_vec(f1, f2);
1317: }
1318: context_vec_ptr++;
1319: context_vec_ptr->a = a;
1320: context_vec_ptr->b = b;
1321: context_vec_ptr->c = c;
1322: context_vec_ptr->d = d;
1323: return;
1324: }
1325: if (anychange == 0)
1326: anychange = 1;
1327: switch (format) {
1328: case D_BRIEF:
1329: return;
1330: case D_NORMAL:
1331: range(a, b, ",");
1332: putchar(a > b ? 'a' : c > d ? 'd' : 'c');
1333: if (format == D_NORMAL)
1334: range(c, d, ",");
1335: putchar('\n');
1336: break;
1337: }
1338: if (format == D_NORMAL || format == D_IFDEF) {
1339: fetch(ixold, a, b, f1, '<', 1);
1340: if (a <= b && c <= d && format == D_NORMAL)
1341: puts("---");
1342: }
1343: i = fetch(ixnew, c, d, f2, format == D_NORMAL ? '>' : '\0', 0);
1344: if (inifdef) {
1345: printf("#endif /* %s */\n", ifdefname);
1346: inifdef = 0;
1347: }
1348: }
1349:
1350: static int
1351: fetch(long *f, int a, int b, FILE *lb, int ch, int oldfile)
1352: {
1353: int i, j, c, lastc, col, nc;
1354:
1355: /*
1356: * When doing #ifdef's, copy down to current line
1357: * if this is the first file, so that stuff makes it to output.
1358: */
1359: if (format == D_IFDEF && oldfile) {
1360: long curpos = ftell(lb);
1361: /* print through if append (a>b), else to (nb: 0 vs 1 orig) */
1362: nc = f[a > b ? b : a - 1] - curpos;
1363: for (i = 0; i < nc; i++)
1364: putchar(getc(lb));
1365: }
1366: if (a > b)
1367: return (0);
1368: if (format == D_IFDEF) {
1369: if (inifdef) {
1370: printf("#else /* %s%s */\n",
1371: oldfile == 1 ? "!" : "", ifdefname);
1372: } else {
1373: if (oldfile)
1374: printf("#ifndef %s\n", ifdefname);
1375: else
1376: printf("#ifdef %s\n", ifdefname);
1377: }
1378: inifdef = 1 + oldfile;
1379: }
1380: for (i = a; i <= b; i++) {
1381: fseek(lb, f[i - 1], SEEK_SET);
1382: nc = f[i] - f[i - 1];
1383: if (format != D_IFDEF && ch != '\0') {
1384: putchar(ch);
1385: if (Tflag && (format == D_NORMAL || format == D_CONTEXT
1386: || format == D_UNIFIED))
1387: putchar('\t');
1388: else if (format != D_UNIFIED)
1389: putchar(' ');
1390: }
1391: col = 0;
1392: for (j = 0, lastc = '\0'; j < nc; j++, lastc = c) {
1393: if ((c = getc(lb)) == EOF) {
1394: puts("\n\\ No newline at end of file");
1395: return (0);
1396: }
1397: if (c == '\t' && tflag) {
1398: do {
1399: putchar(' ');
1400: } while (++col & 7);
1401: } else {
1402: putchar(c);
1403: col++;
1404: }
1405: }
1406: }
1407: return (0);
1408: }
1409:
1410: /*
1411: * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578.
1412: */
1413: static int
1414: readhash(FILE *f)
1415: {
1416: int i, t, space;
1417: int sum;
1418:
1419: sum = 1;
1420: space = 0;
1421: if (!bflag && !wflag) {
1422: if (iflag)
1423: for (i = 0; (t = getc(f)) != '\n'; i++) {
1424: if (t == EOF) {
1425: if (i == 0)
1426: return (0);
1427: break;
1428: }
1429: sum = sum * 127 + chrtran[t];
1430: }
1431: else
1432: for (i = 0; (t = getc(f)) != '\n'; i++) {
1433: if (t == EOF) {
1434: if (i == 0)
1435: return (0);
1436: break;
1437: }
1438: sum = sum * 127 + t;
1439: }
1440: } else {
1441: for (i = 0;;) {
1442: switch (t = getc(f)) {
1443: case '\t':
1444: case ' ':
1445: space++;
1446: continue;
1447: default:
1448: if (space && !wflag) {
1449: i++;
1450: space = 0;
1451: }
1452: sum = sum * 127 + chrtran[t];
1453: i++;
1454: continue;
1455: case EOF:
1456: if (i == 0)
1457: return (0);
1458: /* FALLTHROUGH */
1459: case '\n':
1460: break;
1461: }
1462: break;
1463: }
1464: }
1465: /*
1466: * There is a remote possibility that we end up with a zero sum.
1467: * Zero is used as an EOF marker, so return 1 instead.
1468: */
1469: return (sum == 0 ? 1 : sum);
1470: }
1471:
1472: static int
1473: asciifile(FILE *f)
1474: {
1475: char buf[BUFSIZ];
1476: int i, cnt;
1477:
1478: if (aflag || f == NULL)
1479: return (1);
1480:
1481: rewind(f);
1482: cnt = fread(buf, 1, sizeof(buf), f);
1483: for (i = 0; i < cnt; i++)
1484: if (!isprint(buf[i]) && !isspace(buf[i]))
1485: return (0);
1486: return (1);
1487: }
1488:
1.17 jfb 1489: static char*
1490: match_function(const long *f, int pos, FILE *fp)
1491: {
1492: unsigned char buf[FUNCTION_CONTEXT_SIZE];
1493: size_t nc;
1494: int last = lastline;
1495: char *p;
1496:
1497: lastline = pos;
1498: while (pos > last) {
1499: fseek(fp, f[pos - 1], SEEK_SET);
1500: nc = f[pos] - f[pos - 1];
1501: if (nc >= sizeof(buf))
1502: nc = sizeof(buf) - 1;
1503: nc = fread(buf, 1, nc, fp);
1504: if (nc > 0) {
1505: buf[nc] = '\0';
1506: p = strchr(buf, '\n');
1507: if (p != NULL)
1508: *p = '\0';
1509: if (isalpha(buf[0]) || buf[0] == '_' || buf[0] == '$') {
1510: strlcpy(lastbuf, buf, sizeof lastbuf);
1511: lastmatchline = pos;
1512: return lastbuf;
1513: }
1514: }
1515: pos--;
1516: }
1517: return (lastmatchline > 0) ? lastbuf : NULL;
1518: }
1519:
1.1 jfb 1520:
1521: /* dump accumulated "context" diff changes */
1522: static void
1523: dump_context_vec(FILE *f1, FILE *f2)
1524: {
1525: struct context_vec *cvp = context_vec_start;
1526: int lowa, upb, lowc, upd, do_output;
1527: int a, b, c, d;
1.17 jfb 1528: char ch, *f;
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: printf("***************");
1.17 jfb 1540: if (pflag) {
1541: f = match_function(ixold, lowa - 1, f1);
1542: if (f != NULL) {
1543: putchar(' ');
1544: fputs(f, stdout);
1545: }
1546: }
1.1 jfb 1547: printf("\n*** ");
1548: range(lowa, upb, ",");
1549: printf(" ****\n");
1550:
1551: /*
1552: * Output changes to the "old" file. The first loop suppresses
1553: * output if there were no changes to the "old" file (we'll see
1554: * the "old" lines as context in the "new" list).
1555: */
1556: do_output = 0;
1557: for (; cvp <= context_vec_ptr; cvp++)
1558: if (cvp->a <= cvp->b) {
1559: cvp = context_vec_start;
1560: do_output++;
1561: break;
1562: }
1563: if (do_output) {
1564: while (cvp <= context_vec_ptr) {
1565: a = cvp->a;
1566: b = cvp->b;
1567: c = cvp->c;
1568: d = cvp->d;
1569:
1570: if (a <= b && c <= d)
1571: ch = 'c';
1572: else
1573: ch = (a <= b) ? 'd' : 'a';
1574:
1575: if (ch == 'a')
1576: fetch(ixold, lowa, b, f1, ' ', 0);
1577: else {
1578: fetch(ixold, lowa, a - 1, f1, ' ', 0);
1579: fetch(ixold, a, b, f1,
1580: ch == 'c' ? '!' : '-', 0);
1581: }
1582: lowa = b + 1;
1583: cvp++;
1584: }
1585: fetch(ixold, b + 1, upb, f1, ' ', 0);
1586: }
1587: /* output changes to the "new" file */
1588: printf("--- ");
1589: range(lowc, upd, ",");
1590: printf(" ----\n");
1591:
1592: do_output = 0;
1593: for (cvp = context_vec_start; cvp <= context_vec_ptr; cvp++)
1594: if (cvp->c <= cvp->d) {
1595: cvp = context_vec_start;
1596: do_output++;
1597: break;
1598: }
1599: if (do_output) {
1600: while (cvp <= context_vec_ptr) {
1601: a = cvp->a;
1602: b = cvp->b;
1603: c = cvp->c;
1604: d = cvp->d;
1605:
1606: if (a <= b && c <= d)
1607: ch = 'c';
1608: else
1609: ch = (a <= b) ? 'd' : 'a';
1610:
1611: if (ch == 'd')
1612: fetch(ixnew, lowc, d, f2, ' ', 0);
1613: else {
1614: fetch(ixnew, lowc, c - 1, f2, ' ', 0);
1615: fetch(ixnew, c, d, f2,
1616: ch == 'c' ? '!' : '+', 0);
1617: }
1618: lowc = d + 1;
1619: cvp++;
1620: }
1621: fetch(ixnew, d + 1, upd, f2, ' ', 0);
1622: }
1623: context_vec_ptr = context_vec_start - 1;
1624: }
1625:
1626: /* dump accumulated "unified" diff changes */
1627: static void
1628: dump_unified_vec(FILE *f1, FILE *f2)
1629: {
1630: struct context_vec *cvp = context_vec_start;
1631: int lowa, upb, lowc, upd;
1632: int a, b, c, d;
1.17 jfb 1633: char ch, *f;
1.1 jfb 1634:
1635: if (context_vec_start > context_vec_ptr)
1636: return;
1637:
1638: b = d = 0; /* gcc */
1.4 jfb 1639: lowa = MAX(1, cvp->a - context);
1640: upb = MIN(len[0], context_vec_ptr->b + context);
1641: lowc = MAX(1, cvp->c - context);
1642: upd = MIN(len[1], context_vec_ptr->d + context);
1.1 jfb 1643:
1644: fputs("@@ -", stdout);
1645: uni_range(lowa, upb);
1646: fputs(" +", stdout);
1647: uni_range(lowc, upd);
1648: fputs(" @@", stdout);
1.17 jfb 1649: if (pflag) {
1650: f = match_function(ixold, lowa - 1, f1);
1651: if (f != NULL) {
1652: putchar(' ');
1653: fputs(f, stdout);
1654: }
1655: }
1.1 jfb 1656: putchar('\n');
1657:
1658: /*
1659: * Output changes in "unified" diff format--the old and new lines
1660: * are printed together.
1661: */
1662: for (; cvp <= context_vec_ptr; cvp++) {
1663: a = cvp->a;
1664: b = cvp->b;
1665: c = cvp->c;
1666: d = cvp->d;
1667:
1668: /*
1669: * c: both new and old changes
1670: * d: only changes in the old file
1671: * a: only changes in the new file
1672: */
1673: if (a <= b && c <= d)
1674: ch = 'c';
1675: else
1676: ch = (a <= b) ? 'd' : 'a';
1677:
1678: switch (ch) {
1679: case 'c':
1680: fetch(ixold, lowa, a - 1, f1, ' ', 0);
1681: fetch(ixold, a, b, f1, '-', 0);
1682: fetch(ixnew, c, d, f2, '+', 0);
1683: break;
1684: case 'd':
1685: fetch(ixold, lowa, a - 1, f1, ' ', 0);
1686: fetch(ixold, a, b, f1, '-', 0);
1687: break;
1688: case 'a':
1689: fetch(ixnew, lowc, c - 1, f2, ' ', 0);
1690: fetch(ixnew, c, d, f2, '+', 0);
1691: break;
1692: }
1693: lowa = b + 1;
1694: lowc = d + 1;
1695: }
1696: fetch(ixnew, d + 1, upd, f2, ' ', 0);
1697:
1698: context_vec_ptr = context_vec_start - 1;
1699: }