Blame


1 3b0f3d61 2020-01-22 neels /* Commandline diff utility to test diff implementations. */
2 3b0f3d61 2020-01-22 neels /*
3 3b0f3d61 2020-01-22 neels * Copyright (c) 2018 Martin Pieuchot
4 3b0f3d61 2020-01-22 neels * Copyright (c) 2020 Neels Hofmeyr <neels@hofmeyr.de>
5 3b0f3d61 2020-01-22 neels *
6 3b0f3d61 2020-01-22 neels * Permission to use, copy, modify, and distribute this software for any
7 3b0f3d61 2020-01-22 neels * purpose with or without fee is hereby granted, provided that the above
8 3b0f3d61 2020-01-22 neels * copyright notice and this permission notice appear in all copies.
9 3b0f3d61 2020-01-22 neels *
10 3b0f3d61 2020-01-22 neels * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11 3b0f3d61 2020-01-22 neels * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 3b0f3d61 2020-01-22 neels * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13 3b0f3d61 2020-01-22 neels * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14 3b0f3d61 2020-01-22 neels * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15 3b0f3d61 2020-01-22 neels * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16 3b0f3d61 2020-01-22 neels * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 3b0f3d61 2020-01-22 neels */
18 3b0f3d61 2020-01-22 neels
19 3b0f3d61 2020-01-22 neels #include <sys/mman.h>
20 3b0f3d61 2020-01-22 neels #include <sys/stat.h>
21 18a9c7f8 2020-09-20 stsp #include <sys/types.h>
22 3b0f3d61 2020-01-22 neels
23 3b0f3d61 2020-01-22 neels #include <err.h>
24 3b0f3d61 2020-01-22 neels #include <fcntl.h>
25 3b0f3d61 2020-01-22 neels #include <inttypes.h>
26 3b0f3d61 2020-01-22 neels #include <stdio.h>
27 3b0f3d61 2020-01-22 neels #include <stdlib.h>
28 e10a628a 2020-09-16 stsp #include <stdbool.h>
29 3e6cba3a 2020-08-13 stsp #include <string.h>
30 3b0f3d61 2020-01-22 neels #include <unistd.h>
31 3b0f3d61 2020-01-22 neels
32 e10a628a 2020-09-16 stsp #include <diff/arraylist.h>
33 8ad022d2 2020-05-05 neels #include <diff/diff_main.h>
34 8ad022d2 2020-05-05 neels #include <diff/diff_output.h>
35 8ad022d2 2020-05-05 neels
36 3b0f3d61 2020-01-22 neels __dead void usage(void);
37 527f2c8a 2020-09-20 stsp int diffreg(char *, char *, bool, bool, int);
38 c6eecea3 2020-07-26 stsp int openfile(const char *, char **, struct stat *);
39 3b0f3d61 2020-01-22 neels
40 3b0f3d61 2020-01-22 neels __dead void
41 3b0f3d61 2020-01-22 neels usage(void)
42 3b0f3d61 2020-01-22 neels {
43 760fe30e 2020-05-05 neels fprintf(stderr,
44 527f2c8a 2020-09-20 stsp "usage: %s [-pw] [-C n] file1 file2\n"
45 760fe30e 2020-05-05 neels "\n"
46 760fe30e 2020-05-05 neels " -p Use Patience Diff (slower but often nicer)\n"
47 732e8ee0 2020-09-20 stsp " -w Ignore Whitespace\n"
48 527f2c8a 2020-09-20 stsp " -C n Number of Context Lines\n"
49 760fe30e 2020-05-05 neels , getprogname());
50 3b0f3d61 2020-01-22 neels exit(1);
51 3b0f3d61 2020-01-22 neels }
52 3b0f3d61 2020-01-22 neels
53 3b0f3d61 2020-01-22 neels int
54 3b0f3d61 2020-01-22 neels main(int argc, char *argv[])
55 3b0f3d61 2020-01-22 neels {
56 3e6cba3a 2020-08-13 stsp int ch, rc;
57 732e8ee0 2020-09-20 stsp bool do_patience = false, ignore_whitespace = false;
58 527f2c8a 2020-09-20 stsp int context_lines = 3;
59 3b0f3d61 2020-01-22 neels
60 527f2c8a 2020-09-20 stsp while ((ch = getopt(argc, argv, "pwC:")) != -1) {
61 3b0f3d61 2020-01-22 neels switch (ch) {
62 760fe30e 2020-05-05 neels case 'p':
63 760fe30e 2020-05-05 neels do_patience = true;
64 760fe30e 2020-05-05 neels break;
65 732e8ee0 2020-09-20 stsp case 'w':
66 732e8ee0 2020-09-20 stsp ignore_whitespace = true;
67 732e8ee0 2020-09-20 stsp break;
68 527f2c8a 2020-09-20 stsp case 'C':
69 527f2c8a 2020-09-20 stsp context_lines = atoi(optarg);
70 527f2c8a 2020-09-20 stsp break;
71 3b0f3d61 2020-01-22 neels default:
72 3b0f3d61 2020-01-22 neels usage();
73 3b0f3d61 2020-01-22 neels }
74 3b0f3d61 2020-01-22 neels }
75 3b0f3d61 2020-01-22 neels
76 3b0f3d61 2020-01-22 neels argc -= optind;
77 3b0f3d61 2020-01-22 neels argv += optind;
78 3b0f3d61 2020-01-22 neels
79 3b0f3d61 2020-01-22 neels if (argc != 2)
80 3b0f3d61 2020-01-22 neels usage();
81 3b0f3d61 2020-01-22 neels
82 527f2c8a 2020-09-20 stsp rc = diffreg(argv[0], argv[1], do_patience, ignore_whitespace,
83 527f2c8a 2020-09-20 stsp context_lines);
84 3e6cba3a 2020-08-13 stsp if (rc != DIFF_RC_OK) {
85 3e6cba3a 2020-08-13 stsp fprintf(stderr, "diff: %s\n", strerror(rc));
86 3e6cba3a 2020-08-13 stsp return 1;
87 3e6cba3a 2020-08-13 stsp }
88 3e6cba3a 2020-08-13 stsp return 0;
89 3b0f3d61 2020-01-22 neels }
90 3b0f3d61 2020-01-22 neels
91 0d27172a 2020-05-06 neels const struct diff_algo_config myers_then_patience;
92 0d27172a 2020-05-06 neels const struct diff_algo_config myers_then_myers_divide;
93 0d27172a 2020-05-06 neels const struct diff_algo_config patience;
94 0d27172a 2020-05-06 neels const struct diff_algo_config myers_divide;
95 3b0f3d61 2020-01-22 neels
96 760fe30e 2020-05-05 neels const struct diff_algo_config myers_then_patience = (struct diff_algo_config){
97 3b0f3d61 2020-01-22 neels .impl = diff_algo_myers,
98 9e668157 2020-01-27 neels .permitted_state_size = 1024 * 1024 * sizeof(int),
99 3b0f3d61 2020-01-22 neels .fallback_algo = &patience,
100 3b0f3d61 2020-01-22 neels };
101 3b0f3d61 2020-01-22 neels
102 0d27172a 2020-05-06 neels const struct diff_algo_config myers_then_myers_divide =
103 0d27172a 2020-05-06 neels (struct diff_algo_config){
104 760fe30e 2020-05-05 neels .impl = diff_algo_myers,
105 760fe30e 2020-05-05 neels .permitted_state_size = 1024 * 1024 * sizeof(int),
106 760fe30e 2020-05-05 neels .fallback_algo = &myers_divide,
107 760fe30e 2020-05-05 neels };
108 760fe30e 2020-05-05 neels
109 3b0f3d61 2020-01-22 neels const struct diff_algo_config patience = (struct diff_algo_config){
110 3b0f3d61 2020-01-22 neels .impl = diff_algo_patience,
111 0d27172a 2020-05-06 neels /* After subdivision, do Patience again: */
112 0d27172a 2020-05-06 neels .inner_algo = &patience,
113 0d27172a 2020-05-06 neels /* If subdivision failed, do Myers Divide et Impera: */
114 0d27172a 2020-05-06 neels .fallback_algo = &myers_then_myers_divide,
115 3b0f3d61 2020-01-22 neels };
116 3b0f3d61 2020-01-22 neels
117 3b0f3d61 2020-01-22 neels const struct diff_algo_config myers_divide = (struct diff_algo_config){
118 3b0f3d61 2020-01-22 neels .impl = diff_algo_myers_divide,
119 0d27172a 2020-05-06 neels /* When division succeeded, start from the top: */
120 0d27172a 2020-05-06 neels .inner_algo = &myers_then_myers_divide,
121 0d27172a 2020-05-06 neels /* (fallback_algo = NULL implies diff_algo_none). */
122 3b0f3d61 2020-01-22 neels };
123 3b0f3d61 2020-01-22 neels
124 3b0f3d61 2020-01-22 neels const struct diff_config diff_config = {
125 3b0f3d61 2020-01-22 neels .atomize_func = diff_atomize_text_by_line,
126 760fe30e 2020-05-05 neels .algo = &myers_then_myers_divide,
127 3b0f3d61 2020-01-22 neels };
128 3b0f3d61 2020-01-22 neels
129 760fe30e 2020-05-05 neels const struct diff_config diff_config_patience = {
130 760fe30e 2020-05-05 neels .atomize_func = diff_atomize_text_by_line,
131 760fe30e 2020-05-05 neels .algo = &myers_then_patience,
132 760fe30e 2020-05-05 neels };
133 760fe30e 2020-05-05 neels
134 3b0f3d61 2020-01-22 neels int
135 527f2c8a 2020-09-20 stsp diffreg(char *file1, char *file2, bool do_patience, bool ignore_whitespace,
136 527f2c8a 2020-09-20 stsp int context_lines)
137 3b0f3d61 2020-01-22 neels {
138 3b0f3d61 2020-01-22 neels char *str1, *str2;
139 c6eecea3 2020-07-26 stsp int fd1, fd2;
140 3b0f3d61 2020-01-22 neels struct stat st1, st2;
141 3b0f3d61 2020-01-22 neels struct diff_input_info info = {
142 3b0f3d61 2020-01-22 neels .left_path = file1,
143 3b0f3d61 2020-01-22 neels .right_path = file2,
144 3b0f3d61 2020-01-22 neels };
145 f8cbb8fe 2020-05-05 neels struct diff_result *result;
146 3e6cba3a 2020-08-13 stsp int rc;
147 0d27172a 2020-05-06 neels const struct diff_config *cfg;
148 0d27172a 2020-05-06 neels
149 0d27172a 2020-05-06 neels cfg = do_patience ? &diff_config_patience : &diff_config;
150 3b0f3d61 2020-01-22 neels
151 c6eecea3 2020-07-26 stsp fd1 = openfile(file1, &str1, &st1);
152 c6eecea3 2020-07-26 stsp fd2 = openfile(file2, &str2, &st2);
153 3b0f3d61 2020-01-22 neels
154 732e8ee0 2020-09-20 stsp result = diff_main(cfg, fd1, str1, st1.st_size, fd2, str2, st2.st_size,
155 732e8ee0 2020-09-20 stsp ignore_whitespace);
156 f8cbb8fe 2020-05-05 neels #if 0
157 f8cbb8fe 2020-05-05 neels rc = diff_output_plain(stdout, &info, result);
158 f8cbb8fe 2020-05-05 neels #else
159 527f2c8a 2020-09-20 stsp rc = diff_output_unidiff(stdout, &info, result, context_lines);
160 f8cbb8fe 2020-05-05 neels #endif
161 f8cbb8fe 2020-05-05 neels diff_result_free(result);
162 3b0f3d61 2020-01-22 neels
163 c6eecea3 2020-07-26 stsp if (str1)
164 c6eecea3 2020-07-26 stsp munmap(str1, st1.st_size);
165 c6eecea3 2020-07-26 stsp if (str2)
166 c6eecea3 2020-07-26 stsp munmap(str2, st2.st_size);
167 c6eecea3 2020-07-26 stsp close(fd1);
168 c6eecea3 2020-07-26 stsp close(fd2);
169 3b0f3d61 2020-01-22 neels
170 f8cbb8fe 2020-05-05 neels return rc;
171 3b0f3d61 2020-01-22 neels }
172 3b0f3d61 2020-01-22 neels
173 c6eecea3 2020-07-26 stsp int
174 c6eecea3 2020-07-26 stsp openfile(const char *path, char **p, struct stat *st)
175 3b0f3d61 2020-01-22 neels {
176 3b0f3d61 2020-01-22 neels int fd;
177 3b0f3d61 2020-01-22 neels
178 3b0f3d61 2020-01-22 neels fd = open(path, O_RDONLY);
179 3b0f3d61 2020-01-22 neels if (fd == -1)
180 3b0f3d61 2020-01-22 neels err(2, "%s", path);
181 3b0f3d61 2020-01-22 neels
182 3b0f3d61 2020-01-22 neels if (fstat(fd, st) == -1)
183 3b0f3d61 2020-01-22 neels err(2, "%s", path);
184 3b0f3d61 2020-01-22 neels
185 c6eecea3 2020-07-26 stsp #ifndef DIFF_NO_MMAP
186 c6eecea3 2020-07-26 stsp *p = mmap(NULL, st->st_size, PROT_READ, MAP_PRIVATE, fd, 0);
187 c6eecea3 2020-07-26 stsp if (*p == MAP_FAILED)
188 c6eecea3 2020-07-26 stsp #endif
189 c6eecea3 2020-07-26 stsp *p = NULL; /* fall back on file I/O */
190 3b0f3d61 2020-01-22 neels
191 c6eecea3 2020-07-26 stsp return fd;
192 3b0f3d61 2020-01-22 neels }