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 1dfba055 2020-10-07 stsp #include <arraylist.h>
33 1dfba055 2020-10-07 stsp #include <diff_main.h>
34 1dfba055 2020-10-07 stsp #include <diff_output.h>
35 8ad022d2 2020-05-05 neels
36 65a56b16 2020-10-12 neels enum diffreg_algo {
37 65a56b16 2020-10-12 neels DIFFREG_ALGO_MYERS_THEN_MYERS_DIVIDE = 0,
38 65a56b16 2020-10-12 neels DIFFREG_ALGO_MYERS_THEN_PATIENCE = 1,
39 65a56b16 2020-10-12 neels DIFFREG_ALGO_PATIENCE = 2,
40 65a56b16 2020-10-12 neels DIFFREG_ALGO_NONE = 3,
41 65a56b16 2020-10-12 neels };
42 65a56b16 2020-10-12 neels
43 3b0f3d61 2020-01-22 neels __dead void usage(void);
44 65a56b16 2020-10-12 neels int diffreg(char *, char *, enum diffreg_algo, bool, int, bool);
45 7a54ad3a 2020-09-20 stsp FILE * openfile(const char *, char **, struct stat *);
46 3b0f3d61 2020-01-22 neels
47 3b0f3d61 2020-01-22 neels __dead void
48 3b0f3d61 2020-01-22 neels usage(void)
49 3b0f3d61 2020-01-22 neels {
50 760fe30e 2020-05-05 neels fprintf(stderr,
51 65a56b16 2020-10-12 neels "usage: %s [-PQTwe] [-U n] file1 file2\n"
52 760fe30e 2020-05-05 neels "\n"
53 65a56b16 2020-10-12 neels " -P Use Patience Diff (slower but often nicer)\n"
54 65a56b16 2020-10-12 neels " -Q Use forward-Myers for small files, otherwise Patience\n"
55 65a56b16 2020-10-12 neels " -T Trivial algo: detect similar start and end only\n"
56 732e8ee0 2020-09-20 stsp " -w Ignore Whitespace\n"
57 0c9a7e9d 2020-10-07 stsp " -U n Number of Context Lines\n"
58 0c9a7e9d 2020-10-07 stsp " -e Produce ed script output\n"
59 760fe30e 2020-05-05 neels , getprogname());
60 3b0f3d61 2020-01-22 neels exit(1);
61 3b0f3d61 2020-01-22 neels }
62 3b0f3d61 2020-01-22 neels
63 3b0f3d61 2020-01-22 neels int
64 3b0f3d61 2020-01-22 neels main(int argc, char *argv[])
65 3b0f3d61 2020-01-22 neels {
66 3e6cba3a 2020-08-13 stsp int ch, rc;
67 65a56b16 2020-10-12 neels bool ignore_whitespace = false;
68 b7ba71f0 2020-10-07 stsp bool edscript = false;
69 527f2c8a 2020-09-20 stsp int context_lines = 3;
70 65a56b16 2020-10-12 neels enum diffreg_algo algo = DIFFREG_ALGO_MYERS_THEN_MYERS_DIVIDE;
71 3b0f3d61 2020-01-22 neels
72 65a56b16 2020-10-12 neels while ((ch = getopt(argc, argv, "PQTwU:e")) != -1) {
73 3b0f3d61 2020-01-22 neels switch (ch) {
74 65a56b16 2020-10-12 neels case 'P':
75 65a56b16 2020-10-12 neels algo = DIFFREG_ALGO_PATIENCE;
76 760fe30e 2020-05-05 neels break;
77 65a56b16 2020-10-12 neels case 'Q':
78 65a56b16 2020-10-12 neels algo = DIFFREG_ALGO_MYERS_THEN_PATIENCE;
79 65a56b16 2020-10-12 neels break;
80 65a56b16 2020-10-12 neels case 'T':
81 65a56b16 2020-10-12 neels algo = DIFFREG_ALGO_NONE;
82 65a56b16 2020-10-12 neels break;
83 732e8ee0 2020-09-20 stsp case 'w':
84 732e8ee0 2020-09-20 stsp ignore_whitespace = true;
85 732e8ee0 2020-09-20 stsp break;
86 0c9a7e9d 2020-10-07 stsp case 'U':
87 527f2c8a 2020-09-20 stsp context_lines = atoi(optarg);
88 527f2c8a 2020-09-20 stsp break;
89 b7ba71f0 2020-10-07 stsp case 'e':
90 b7ba71f0 2020-10-07 stsp edscript = true;
91 b7ba71f0 2020-10-07 stsp break;
92 3b0f3d61 2020-01-22 neels default:
93 3b0f3d61 2020-01-22 neels usage();
94 3b0f3d61 2020-01-22 neels }
95 3b0f3d61 2020-01-22 neels }
96 3b0f3d61 2020-01-22 neels
97 3b0f3d61 2020-01-22 neels argc -= optind;
98 3b0f3d61 2020-01-22 neels argv += optind;
99 3b0f3d61 2020-01-22 neels
100 3b0f3d61 2020-01-22 neels if (argc != 2)
101 3b0f3d61 2020-01-22 neels usage();
102 3b0f3d61 2020-01-22 neels
103 65a56b16 2020-10-12 neels rc = diffreg(argv[0], argv[1], algo, ignore_whitespace,
104 b7ba71f0 2020-10-07 stsp context_lines, edscript);
105 3e6cba3a 2020-08-13 stsp if (rc != DIFF_RC_OK) {
106 3e6cba3a 2020-08-13 stsp fprintf(stderr, "diff: %s\n", strerror(rc));
107 3e6cba3a 2020-08-13 stsp return 1;
108 3e6cba3a 2020-08-13 stsp }
109 3e6cba3a 2020-08-13 stsp return 0;
110 3b0f3d61 2020-01-22 neels }
111 3b0f3d61 2020-01-22 neels
112 0d27172a 2020-05-06 neels const struct diff_algo_config myers_then_patience;
113 0d27172a 2020-05-06 neels const struct diff_algo_config myers_then_myers_divide;
114 0d27172a 2020-05-06 neels const struct diff_algo_config patience;
115 0d27172a 2020-05-06 neels const struct diff_algo_config myers_divide;
116 3b0f3d61 2020-01-22 neels
117 760fe30e 2020-05-05 neels const struct diff_algo_config myers_then_patience = (struct diff_algo_config){
118 3b0f3d61 2020-01-22 neels .impl = diff_algo_myers,
119 9e668157 2020-01-27 neels .permitted_state_size = 1024 * 1024 * sizeof(int),
120 3b0f3d61 2020-01-22 neels .fallback_algo = &patience,
121 3b0f3d61 2020-01-22 neels };
122 3b0f3d61 2020-01-22 neels
123 0d27172a 2020-05-06 neels const struct diff_algo_config myers_then_myers_divide =
124 0d27172a 2020-05-06 neels (struct diff_algo_config){
125 760fe30e 2020-05-05 neels .impl = diff_algo_myers,
126 760fe30e 2020-05-05 neels .permitted_state_size = 1024 * 1024 * sizeof(int),
127 760fe30e 2020-05-05 neels .fallback_algo = &myers_divide,
128 760fe30e 2020-05-05 neels };
129 760fe30e 2020-05-05 neels
130 3b0f3d61 2020-01-22 neels const struct diff_algo_config patience = (struct diff_algo_config){
131 3b0f3d61 2020-01-22 neels .impl = diff_algo_patience,
132 0d27172a 2020-05-06 neels /* After subdivision, do Patience again: */
133 0d27172a 2020-05-06 neels .inner_algo = &patience,
134 0d27172a 2020-05-06 neels /* If subdivision failed, do Myers Divide et Impera: */
135 0d27172a 2020-05-06 neels .fallback_algo = &myers_then_myers_divide,
136 3b0f3d61 2020-01-22 neels };
137 3b0f3d61 2020-01-22 neels
138 3b0f3d61 2020-01-22 neels const struct diff_algo_config myers_divide = (struct diff_algo_config){
139 3b0f3d61 2020-01-22 neels .impl = diff_algo_myers_divide,
140 0d27172a 2020-05-06 neels /* When division succeeded, start from the top: */
141 0d27172a 2020-05-06 neels .inner_algo = &myers_then_myers_divide,
142 0d27172a 2020-05-06 neels /* (fallback_algo = NULL implies diff_algo_none). */
143 3b0f3d61 2020-01-22 neels };
144 3b0f3d61 2020-01-22 neels
145 65a56b16 2020-10-12 neels const struct diff_algo_config no_algo = (struct diff_algo_config){
146 65a56b16 2020-10-12 neels .impl = diff_algo_none,
147 65a56b16 2020-10-12 neels };
148 65a56b16 2020-10-12 neels
149 65a56b16 2020-10-12 neels /* If the state for a forward-Myers is small enough, use Myers, otherwise first
150 65a56b16 2020-10-12 neels * do a Myers-divide. */
151 65a56b16 2020-10-12 neels const struct diff_config diff_config_myers_then_myers_divide = {
152 3b0f3d61 2020-01-22 neels .atomize_func = diff_atomize_text_by_line,
153 760fe30e 2020-05-05 neels .algo = &myers_then_myers_divide,
154 3b0f3d61 2020-01-22 neels };
155 3b0f3d61 2020-01-22 neels
156 65a56b16 2020-10-12 neels /* If the state for a forward-Myers is small enough, use Myers, otherwise first
157 65a56b16 2020-10-12 neels * do a Patience. */
158 65a56b16 2020-10-12 neels const struct diff_config diff_config_myers_then_patience = {
159 760fe30e 2020-05-05 neels .atomize_func = diff_atomize_text_by_line,
160 760fe30e 2020-05-05 neels .algo = &myers_then_patience,
161 760fe30e 2020-05-05 neels };
162 760fe30e 2020-05-05 neels
163 65a56b16 2020-10-12 neels /* Directly force Patience as a first divider of the source file. */
164 65a56b16 2020-10-12 neels const struct diff_config diff_config_patience = {
165 65a56b16 2020-10-12 neels .atomize_func = diff_atomize_text_by_line,
166 65a56b16 2020-10-12 neels .algo = &patience,
167 65a56b16 2020-10-12 neels };
168 65a56b16 2020-10-12 neels
169 65a56b16 2020-10-12 neels /* Directly force Patience as a first divider of the source file. */
170 65a56b16 2020-10-12 neels const struct diff_config diff_config_no_algo = {
171 65a56b16 2020-10-12 neels .atomize_func = diff_atomize_text_by_line,
172 65a56b16 2020-10-12 neels };
173 65a56b16 2020-10-12 neels
174 3b0f3d61 2020-01-22 neels int
175 65a56b16 2020-10-12 neels diffreg(char *file1, char *file2, enum diffreg_algo algo, bool ignore_whitespace,
176 b7ba71f0 2020-10-07 stsp int context_lines, bool edscript)
177 3b0f3d61 2020-01-22 neels {
178 3b0f3d61 2020-01-22 neels char *str1, *str2;
179 7a54ad3a 2020-09-20 stsp FILE *f1, *f2;
180 3b0f3d61 2020-01-22 neels struct stat st1, st2;
181 3b0f3d61 2020-01-22 neels struct diff_input_info info = {
182 3b0f3d61 2020-01-22 neels .left_path = file1,
183 3b0f3d61 2020-01-22 neels .right_path = file2,
184 3b0f3d61 2020-01-22 neels };
185 f8cbb8fe 2020-05-05 neels struct diff_result *result;
186 3e6cba3a 2020-08-13 stsp int rc;
187 0d27172a 2020-05-06 neels const struct diff_config *cfg;
188 00d5652b 2020-09-22 stsp int diff_flags = 0;
189 0d27172a 2020-05-06 neels
190 65a56b16 2020-10-12 neels switch (algo) {
191 65a56b16 2020-10-12 neels default:
192 65a56b16 2020-10-12 neels case DIFFREG_ALGO_MYERS_THEN_MYERS_DIVIDE:
193 65a56b16 2020-10-12 neels cfg = &diff_config_myers_then_myers_divide;
194 65a56b16 2020-10-12 neels break;
195 65a56b16 2020-10-12 neels case DIFFREG_ALGO_MYERS_THEN_PATIENCE:
196 65a56b16 2020-10-12 neels cfg = &diff_config_myers_then_patience;
197 65a56b16 2020-10-12 neels break;
198 65a56b16 2020-10-12 neels case DIFFREG_ALGO_PATIENCE:
199 65a56b16 2020-10-12 neels cfg = &diff_config_patience;
200 65a56b16 2020-10-12 neels break;
201 65a56b16 2020-10-12 neels case DIFFREG_ALGO_NONE:
202 65a56b16 2020-10-12 neels cfg = &diff_config_no_algo;
203 65a56b16 2020-10-12 neels break;
204 65a56b16 2020-10-12 neels }
205 3b0f3d61 2020-01-22 neels
206 7a54ad3a 2020-09-20 stsp f1 = openfile(file1, &str1, &st1);
207 7a54ad3a 2020-09-20 stsp f2 = openfile(file2, &str2, &st2);
208 3b0f3d61 2020-01-22 neels
209 00d5652b 2020-09-22 stsp if (ignore_whitespace)
210 00d5652b 2020-09-22 stsp diff_flags |= DIFF_FLAG_IGNORE_WHITESPACE;
211 00d5652b 2020-09-22 stsp
212 7a54ad3a 2020-09-20 stsp result = diff_main(cfg, f1, str1, st1.st_size, f2, str2, st2.st_size,
213 00d5652b 2020-09-22 stsp diff_flags);
214 f8cbb8fe 2020-05-05 neels #if 0
215 f8cbb8fe 2020-05-05 neels rc = diff_output_plain(stdout, &info, result);
216 f8cbb8fe 2020-05-05 neels #else
217 b7ba71f0 2020-10-07 stsp if (edscript)
218 b7ba71f0 2020-10-07 stsp rc = diff_output_edscript(NULL, stdout, &info, result);
219 b7ba71f0 2020-10-07 stsp else {
220 b7ba71f0 2020-10-07 stsp rc = diff_output_unidiff(NULL, stdout, &info, result,
221 b7ba71f0 2020-10-07 stsp context_lines);
222 b7ba71f0 2020-10-07 stsp }
223 f8cbb8fe 2020-05-05 neels #endif
224 f8cbb8fe 2020-05-05 neels diff_result_free(result);
225 3b0f3d61 2020-01-22 neels
226 c6eecea3 2020-07-26 stsp if (str1)
227 c6eecea3 2020-07-26 stsp munmap(str1, st1.st_size);
228 c6eecea3 2020-07-26 stsp if (str2)
229 c6eecea3 2020-07-26 stsp munmap(str2, st2.st_size);
230 7a54ad3a 2020-09-20 stsp fclose(f1);
231 7a54ad3a 2020-09-20 stsp fclose(f2);
232 3b0f3d61 2020-01-22 neels
233 f8cbb8fe 2020-05-05 neels return rc;
234 3b0f3d61 2020-01-22 neels }
235 3b0f3d61 2020-01-22 neels
236 7a54ad3a 2020-09-20 stsp FILE *
237 c6eecea3 2020-07-26 stsp openfile(const char *path, char **p, struct stat *st)
238 3b0f3d61 2020-01-22 neels {
239 7a54ad3a 2020-09-20 stsp FILE *f = NULL;
240 3b0f3d61 2020-01-22 neels
241 7a54ad3a 2020-09-20 stsp f = fopen(path, "r");
242 7a54ad3a 2020-09-20 stsp if (f == NULL)
243 3b0f3d61 2020-01-22 neels err(2, "%s", path);
244 3b0f3d61 2020-01-22 neels
245 7a54ad3a 2020-09-20 stsp if (fstat(fileno(f), st) == -1)
246 3b0f3d61 2020-01-22 neels err(2, "%s", path);
247 3b0f3d61 2020-01-22 neels
248 c6eecea3 2020-07-26 stsp #ifndef DIFF_NO_MMAP
249 7a54ad3a 2020-09-20 stsp *p = mmap(NULL, st->st_size, PROT_READ, MAP_PRIVATE, fileno(f), 0);
250 c6eecea3 2020-07-26 stsp if (*p == MAP_FAILED)
251 c6eecea3 2020-07-26 stsp #endif
252 c6eecea3 2020-07-26 stsp *p = NULL; /* fall back on file I/O */
253 3b0f3d61 2020-01-22 neels
254 7a54ad3a 2020-09-20 stsp return f;
255 3b0f3d61 2020-01-22 neels }