diff options
Diffstat (limited to 'editors')
-rw-r--r-- | editors/Config.in | 39 | ||||
-rw-r--r-- | editors/Kbuild | 2 | ||||
-rw-r--r-- | editors/cmp.c | 145 | ||||
-rw-r--r-- | editors/diff.c | 1282 |
4 files changed, 1468 insertions, 0 deletions
diff --git a/editors/Config.in b/editors/Config.in index 936004c9b..3361d89f1 100644 --- a/editors/Config.in +++ b/editors/Config.in | |||
@@ -20,6 +20,45 @@ config FEATURE_AWK_MATH | |||
20 | Enable math functions of the Awk programming language. | 20 | Enable math functions of the Awk programming language. |
21 | NOTE: This will require libm to be present for linking. | 21 | NOTE: This will require libm to be present for linking. |
22 | 22 | ||
23 | config CMP | ||
24 | bool "cmp" | ||
25 | default n | ||
26 | help | ||
27 | cmp is used to compare two files and returns the result | ||
28 | to standard output. | ||
29 | |||
30 | config DIFF | ||
31 | bool "diff" | ||
32 | default n | ||
33 | help | ||
34 | diff compares two files or directories and outputs the | ||
35 | differences between them in a form that can be given to | ||
36 | the patch command. | ||
37 | |||
38 | config FEATURE_DIFF_BINARY | ||
39 | bool "Enable checks for binary files" | ||
40 | default y | ||
41 | depends on DIFF | ||
42 | help | ||
43 | This option enables support for checking for binary files | ||
44 | before a comparison is carried out. | ||
45 | |||
46 | config FEATURE_DIFF_DIR | ||
47 | bool "Enable directory support" | ||
48 | default y | ||
49 | depends on DIFF | ||
50 | help | ||
51 | This option enables support for directory and subdirectory | ||
52 | comparison. | ||
53 | |||
54 | config FEATURE_DIFF_MINIMAL | ||
55 | bool "Enable -d option to find smaller sets of changes" | ||
56 | default n | ||
57 | depends on DIFF | ||
58 | help | ||
59 | Enabling this option allows the use of -d to make diff | ||
60 | try hard to find the smallest possible set of changes. | ||
61 | |||
23 | config ED | 62 | config ED |
24 | bool "ed" | 63 | bool "ed" |
25 | default n | 64 | default n |
diff --git a/editors/Kbuild b/editors/Kbuild index d991e1faf..76302aa76 100644 --- a/editors/Kbuild +++ b/editors/Kbuild | |||
@@ -6,6 +6,8 @@ | |||
6 | 6 | ||
7 | lib-y:= | 7 | lib-y:= |
8 | lib-$(CONFIG_AWK) += awk.o | 8 | lib-$(CONFIG_AWK) += awk.o |
9 | lib-$(CONFIG_CMP) += cmp.o | ||
10 | lib-$(CONFIG_DIFF) += diff.o | ||
9 | lib-$(CONFIG_ED) += ed.o | 11 | lib-$(CONFIG_ED) += ed.o |
10 | lib-$(CONFIG_PATCH) += patch.o | 12 | lib-$(CONFIG_PATCH) += patch.o |
11 | lib-$(CONFIG_SED) += sed.o | 13 | lib-$(CONFIG_SED) += sed.o |
diff --git a/editors/cmp.c b/editors/cmp.c new file mode 100644 index 000000000..e5dda80ff --- /dev/null +++ b/editors/cmp.c | |||
@@ -0,0 +1,145 @@ | |||
1 | /* vi: set sw=4 ts=4: */ | ||
2 | /* | ||
3 | * Mini cmp implementation for busybox | ||
4 | * | ||
5 | * Copyright (C) 2000,2001 by Matt Kraai <kraai@alumni.carnegiemellon.edu> | ||
6 | * | ||
7 | * Licensed under GPLv2 or later, see file LICENSE in this tarball for details. | ||
8 | */ | ||
9 | |||
10 | /* BB_AUDIT SUSv3 (virtually) compliant -- uses nicer GNU format for -l. */ | ||
11 | /* http://www.opengroup.org/onlinepubs/007904975/utilities/cmp.html */ | ||
12 | |||
13 | /* Mar 16, 2003 Manuel Novoa III (mjn3@codepoet.org) | ||
14 | * | ||
15 | * Original version majorly reworked for SUSv3 compliance, bug fixes, and | ||
16 | * size optimizations. Changes include: | ||
17 | * 1) Now correctly distinguishes between errors and actual file differences. | ||
18 | * 2) Proper handling of '-' args. | ||
19 | * 3) Actual error checking of i/o. | ||
20 | * 4) Accept SUSv3 -l option. Note that we use the slightly nicer gnu format | ||
21 | * in the '-l' case. | ||
22 | */ | ||
23 | |||
24 | #include "libbb.h" | ||
25 | |||
26 | static FILE *cmp_xfopen_input(const char *filename) | ||
27 | { | ||
28 | FILE *fp; | ||
29 | |||
30 | fp = fopen_or_warn_stdin(filename); | ||
31 | if (fp) | ||
32 | return fp; | ||
33 | xfunc_die(); /* We already output an error message. */ | ||
34 | } | ||
35 | |||
36 | static const char fmt_eof[] = "cmp: EOF on %s\n"; | ||
37 | static const char fmt_differ[] = "%s %s differ: char %"OFF_FMT"d, line %d\n"; | ||
38 | // This fmt_l_opt uses gnu-isms. SUSv3 would be "%.0s%.0s%"OFF_FMT"d %o %o\n" | ||
39 | static const char fmt_l_opt[] = "%.0s%.0s%"OFF_FMT"d %3o %3o\n"; | ||
40 | |||
41 | static const char opt_chars[] = "sl"; | ||
42 | #define CMP_OPT_s (1<<0) | ||
43 | #define CMP_OPT_l (1<<1) | ||
44 | |||
45 | int cmp_main(int argc, char **argv); | ||
46 | int cmp_main(int argc, char **argv) | ||
47 | { | ||
48 | FILE *fp1, *fp2, *outfile = stdout; | ||
49 | const char *filename1, *filename2 = "-"; | ||
50 | USE_DESKTOP(off_t skip1 = 0, skip2 = 0;) | ||
51 | off_t char_pos = 0; | ||
52 | int line_pos = 1; /* Hopefully won't overflow... */ | ||
53 | const char *fmt; | ||
54 | int c1, c2; | ||
55 | unsigned opt; | ||
56 | int retval = 0; | ||
57 | |||
58 | xfunc_error_retval = 2; /* 1 is returned if files are different. */ | ||
59 | |||
60 | opt_complementary = "?:-1" | ||
61 | USE_DESKTOP(":?4") | ||
62 | SKIP_DESKTOP(":?2") | ||
63 | ":l--s:s--l"; | ||
64 | opt = getopt32(argc, argv, opt_chars); | ||
65 | argv += optind; | ||
66 | |||
67 | filename1 = *argv; | ||
68 | fp1 = cmp_xfopen_input(filename1); | ||
69 | |||
70 | if (*++argv) { | ||
71 | filename2 = *argv; | ||
72 | #if ENABLE_DESKTOP | ||
73 | if (*++argv) { | ||
74 | skip1 = XATOOFF(*argv); | ||
75 | if (*++argv) { | ||
76 | skip2 = XATOOFF(*argv); | ||
77 | } | ||
78 | } | ||
79 | #endif | ||
80 | } | ||
81 | |||
82 | fp2 = cmp_xfopen_input(filename2); | ||
83 | if (fp1 == fp2) { /* Paranoia check... stdin == stdin? */ | ||
84 | /* Note that we don't bother reading stdin. Neither does gnu wc. | ||
85 | * But perhaps we should, so that other apps down the chain don't | ||
86 | * get the input. Consider 'echo hello | (cmp - - && cat -)'. | ||
87 | */ | ||
88 | return 0; | ||
89 | } | ||
90 | |||
91 | if (opt & CMP_OPT_l) | ||
92 | fmt = fmt_l_opt; | ||
93 | else | ||
94 | fmt = fmt_differ; | ||
95 | |||
96 | #if ENABLE_DESKTOP | ||
97 | while (skip1) { getc(fp1); skip1--; } | ||
98 | while (skip2) { getc(fp2); skip2--; } | ||
99 | #endif | ||
100 | do { | ||
101 | c1 = getc(fp1); | ||
102 | c2 = getc(fp2); | ||
103 | ++char_pos; | ||
104 | if (c1 != c2) { /* Remember: a read error may have occurred. */ | ||
105 | retval = 1; /* But assume the files are different for now. */ | ||
106 | if (c2 == EOF) { | ||
107 | /* We know that fp1 isn't at EOF or in an error state. But to | ||
108 | * save space below, things are setup to expect an EOF in fp1 | ||
109 | * if an EOF occurred. So, swap things around. | ||
110 | */ | ||
111 | fp1 = fp2; | ||
112 | filename1 = filename2; | ||
113 | c1 = c2; | ||
114 | } | ||
115 | if (c1 == EOF) { | ||
116 | die_if_ferror(fp1, filename1); | ||
117 | fmt = fmt_eof; /* Well, no error, so it must really be EOF. */ | ||
118 | outfile = stderr; | ||
119 | /* There may have been output to stdout (option -l), so | ||
120 | * make sure we fflush before writing to stderr. */ | ||
121 | xfflush_stdout(); | ||
122 | } | ||
123 | if (!(opt & CMP_OPT_s)) { | ||
124 | if (opt & CMP_OPT_l) { | ||
125 | line_pos = c1; /* line_pos is unused in the -l case. */ | ||
126 | } | ||
127 | fprintf(outfile, fmt, filename1, filename2, char_pos, line_pos, c2); | ||
128 | if (opt) { /* This must be -l since not -s. */ | ||
129 | /* If we encountered an EOF, | ||
130 | * the while check will catch it. */ | ||
131 | continue; | ||
132 | } | ||
133 | } | ||
134 | break; | ||
135 | } | ||
136 | if (c1 == '\n') { | ||
137 | ++line_pos; | ||
138 | } | ||
139 | } while (c1 != EOF); | ||
140 | |||
141 | die_if_ferror(fp1, filename1); | ||
142 | die_if_ferror(fp2, filename2); | ||
143 | |||
144 | fflush_stdout_and_exit(retval); | ||
145 | } | ||
diff --git a/editors/diff.c b/editors/diff.c new file mode 100644 index 000000000..830c15ea6 --- /dev/null +++ b/editors/diff.c | |||
@@ -0,0 +1,1282 @@ | |||
1 | /* vi: set sw=4 ts=4: */ | ||
2 | /* | ||
3 | * Mini diff implementation for busybox, adapted from OpenBSD diff. | ||
4 | * | ||
5 | * Copyright (C) 2006 by Robert Sullivan <cogito.ergo.cogito@hotmail.com> | ||
6 | * Copyright (c) 2003 Todd C. Miller <Todd.Miller@courtesan.com> | ||
7 | * | ||
8 | * Sponsored in part by the Defense Advanced Research Projects | ||
9 | * Agency (DARPA) and Air Force Research Laboratory, Air Force | ||
10 | * Materiel Command, USAF, under agreement number F39502-99-1-0512. | ||
11 | * | ||
12 | * Licensed under GPLv2 or later, see file LICENSE in this tarball for details. | ||
13 | */ | ||
14 | |||
15 | #include "libbb.h" | ||
16 | |||
17 | #define FSIZE_MAX 32768 | ||
18 | |||
19 | /* | ||
20 | * Output flags | ||
21 | */ | ||
22 | #define D_HEADER 1 /* Print a header/footer between files */ | ||
23 | #define D_EMPTY1 2 /* Treat first file as empty (/dev/null) */ | ||
24 | #define D_EMPTY2 4 /* Treat second file as empty (/dev/null) */ | ||
25 | |||
26 | /* | ||
27 | * Status values for print_status() and diffreg() return values | ||
28 | * Guide: | ||
29 | * D_SAME - files are the same | ||
30 | * D_DIFFER - files differ | ||
31 | * D_BINARY - binary files differ | ||
32 | * D_COMMON - subdirectory common to both dirs | ||
33 | * D_ONLY - file only exists in one dir | ||
34 | * D_MISMATCH1 - path1 a dir, path2 a file | ||
35 | * D_MISMATCH2 - path1 a file, path2 a dir | ||
36 | * D_ERROR - error occurred | ||
37 | * D_SKIPPED1 - skipped path1 as it is a special file | ||
38 | * D_SKIPPED2 - skipped path2 as it is a special file | ||
39 | */ | ||
40 | |||
41 | #define D_SAME 0 | ||
42 | #define D_DIFFER (1<<0) | ||
43 | #define D_BINARY (1<<1) | ||
44 | #define D_COMMON (1<<2) | ||
45 | #define D_ONLY (1<<3) | ||
46 | #define D_MISMATCH1 (1<<4) | ||
47 | #define D_MISMATCH2 (1<<5) | ||
48 | #define D_ERROR (1<<6) | ||
49 | #define D_SKIPPED1 (1<<7) | ||
50 | #define D_SKIPPED2 (1<<8) | ||
51 | |||
52 | /* Command line options */ | ||
53 | #define FLAG_a (1<<0) | ||
54 | #define FLAG_b (1<<1) | ||
55 | #define FLAG_d (1<<2) | ||
56 | #define FLAG_i (1<<3) | ||
57 | #define FLAG_L (1<<4) | ||
58 | #define FLAG_N (1<<5) | ||
59 | #define FLAG_q (1<<6) | ||
60 | #define FLAG_r (1<<7) | ||
61 | #define FLAG_s (1<<8) | ||
62 | #define FLAG_S (1<<9) | ||
63 | #define FLAG_t (1<<10) | ||
64 | #define FLAG_T (1<<11) | ||
65 | #define FLAG_U (1<<12) | ||
66 | #define FLAG_w (1<<13) | ||
67 | |||
68 | struct cand { | ||
69 | int x; | ||
70 | int y; | ||
71 | int pred; | ||
72 | }; | ||
73 | |||
74 | struct line { | ||
75 | int serial; | ||
76 | int value; | ||
77 | }; | ||
78 | |||
79 | /* | ||
80 | * The following struct is used to record change information | ||
81 | * doing a "context" or "unified" diff. (see routine "change" to | ||
82 | * understand the highly mnemonic field names) | ||
83 | */ | ||
84 | struct context_vec { | ||
85 | int a; /* start line in old file */ | ||
86 | int b; /* end line in old file */ | ||
87 | int c; /* start line in new file */ | ||
88 | int d; /* end line in new file */ | ||
89 | }; | ||
90 | |||
91 | struct globals { | ||
92 | USE_FEATURE_DIFF_DIR(char **dl;) | ||
93 | USE_FEATURE_DIFF_DIR(int dl_count;) | ||
94 | /* This is the default number of lines of context. */ | ||
95 | int context; | ||
96 | size_t max_context; | ||
97 | int status; | ||
98 | char *start; | ||
99 | const char *label1; | ||
100 | const char *label2; | ||
101 | struct line *file[2]; | ||
102 | int *J; /* will be overlaid on class */ | ||
103 | int *class; /* will be overlaid on file[0] */ | ||
104 | int *klist; /* will be overlaid on file[0] after class */ | ||
105 | int *member; /* will be overlaid on file[1] */ | ||
106 | int clen; | ||
107 | int len[2]; | ||
108 | int pref, suff; /* length of prefix and suffix */ | ||
109 | int slen[2]; | ||
110 | bool anychange; | ||
111 | long *ixnew; /* will be overlaid on file[1] */ | ||
112 | long *ixold; /* will be overlaid on klist */ | ||
113 | struct cand *clist; /* merely a free storage pot for candidates */ | ||
114 | int clistlen; /* the length of clist */ | ||
115 | struct line *sfile[2]; /* shortened by pruning common prefix/suffix */ | ||
116 | struct context_vec *context_vec_start; | ||
117 | struct context_vec *context_vec_end; | ||
118 | struct context_vec *context_vec_ptr; | ||
119 | struct stat stb1, stb2; | ||
120 | }; | ||
121 | #define G (*ptr_to_globals) | ||
122 | #define dl (G.dl ) | ||
123 | #define dl_count (G.dl_count ) | ||
124 | #define context (G.context ) | ||
125 | #define max_context (G.max_context ) | ||
126 | #define status (G.status ) | ||
127 | #define start (G.start ) | ||
128 | #define label1 (G.label1 ) | ||
129 | #define label2 (G.label2 ) | ||
130 | #define file (G.file ) | ||
131 | #define J (G.J ) | ||
132 | #define class (G.class ) | ||
133 | #define klist (G.klist ) | ||
134 | #define member (G.member ) | ||
135 | #define clen (G.clen ) | ||
136 | #define len (G.len ) | ||
137 | #define pref (G.pref ) | ||
138 | #define suff (G.suff ) | ||
139 | #define slen (G.slen ) | ||
140 | #define anychange (G.anychange ) | ||
141 | #define ixnew (G.ixnew ) | ||
142 | #define ixold (G.ixold ) | ||
143 | #define clist (G.clist ) | ||
144 | #define clistlen (G.clistlen ) | ||
145 | #define sfile (G.sfile ) | ||
146 | #define context_vec_start (G.context_vec_start ) | ||
147 | #define context_vec_end (G.context_vec_end ) | ||
148 | #define context_vec_ptr (G.context_vec_ptr ) | ||
149 | #define stb1 (G.stb1 ) | ||
150 | #define stb2 (G.stb2 ) | ||
151 | #define INIT_G() do { \ | ||
152 | PTR_TO_GLOBALS = xzalloc(sizeof(G)); \ | ||
153 | context = 3; \ | ||
154 | max_context = 64; \ | ||
155 | } while (0) | ||
156 | |||
157 | |||
158 | static void print_only(const char *path, size_t dirlen, const char *entry) | ||
159 | { | ||
160 | if (dirlen > 1) | ||
161 | dirlen--; | ||
162 | printf("Only in %.*s: %s\n", (int) dirlen, path, entry); | ||
163 | } | ||
164 | |||
165 | static void print_status(int val, char *path1, char *path2, char *entry) | ||
166 | { | ||
167 | const char * const _entry = entry ? entry : ""; | ||
168 | char * const _path1 = entry ? concat_path_file(path1, _entry) : path1; | ||
169 | char * const _path2 = entry ? concat_path_file(path2, _entry) : path2; | ||
170 | |||
171 | switch (val) { | ||
172 | case D_ONLY: | ||
173 | print_only(path1, strlen(path1), entry); | ||
174 | break; | ||
175 | case D_COMMON: | ||
176 | printf("Common subdirectories: %s and %s\n", _path1, _path2); | ||
177 | break; | ||
178 | case D_BINARY: | ||
179 | printf("Binary files %s and %s differ\n", _path1, _path2); | ||
180 | break; | ||
181 | case D_DIFFER: | ||
182 | if (option_mask32 & FLAG_q) | ||
183 | printf("Files %s and %s differ\n", _path1, _path2); | ||
184 | break; | ||
185 | case D_SAME: | ||
186 | if (option_mask32 & FLAG_s) | ||
187 | printf("Files %s and %s are identical\n", _path1, _path2); | ||
188 | break; | ||
189 | case D_MISMATCH1: | ||
190 | printf("File %s is a %s while file %s is a %s\n", | ||
191 | _path1, "directory", _path2, "regular file"); | ||
192 | break; | ||
193 | case D_MISMATCH2: | ||
194 | printf("File %s is a %s while file %s is a %s\n", | ||
195 | _path1, "regular file", _path2, "directory"); | ||
196 | break; | ||
197 | case D_SKIPPED1: | ||
198 | printf("File %s is not a regular file or directory and was skipped\n", | ||
199 | _path1); | ||
200 | break; | ||
201 | case D_SKIPPED2: | ||
202 | printf("File %s is not a regular file or directory and was skipped\n", | ||
203 | _path2); | ||
204 | break; | ||
205 | } | ||
206 | if (entry) { | ||
207 | free(_path1); | ||
208 | free(_path2); | ||
209 | } | ||
210 | } | ||
211 | static void fiddle_sum(int *sum, int t) | ||
212 | { | ||
213 | *sum = (int)(*sum * 127 + t); | ||
214 | } | ||
215 | /* | ||
216 | * Hash function taken from Robert Sedgewick, Algorithms in C, 3d ed., p 578. | ||
217 | */ | ||
218 | static int readhash(FILE * f) | ||
219 | { | ||
220 | int i, t, space; | ||
221 | int sum; | ||
222 | |||
223 | sum = 1; | ||
224 | space = 0; | ||
225 | if (!(option_mask32 & (FLAG_b | FLAG_w))) { | ||
226 | for (i = 0; (t = getc(f)) != '\n'; i++) { | ||
227 | if (t == EOF) { | ||
228 | if (i == 0) | ||
229 | return 0; | ||
230 | break; | ||
231 | } | ||
232 | fiddle_sum(&sum, t); | ||
233 | } | ||
234 | } else { | ||
235 | for (i = 0;;) { | ||
236 | switch (t = getc(f)) { | ||
237 | case '\t': | ||
238 | case '\r': | ||
239 | case '\v': | ||
240 | case '\f': | ||
241 | case ' ': | ||
242 | space++; | ||
243 | continue; | ||
244 | default: | ||
245 | if (space && !(option_mask32 & FLAG_w)) { | ||
246 | i++; | ||
247 | space = 0; | ||
248 | } | ||
249 | fiddle_sum(&sum, t); | ||
250 | i++; | ||
251 | continue; | ||
252 | case EOF: | ||
253 | if (i == 0) | ||
254 | return 0; | ||
255 | /* FALLTHROUGH */ | ||
256 | case '\n': | ||
257 | break; | ||
258 | } | ||
259 | break; | ||
260 | } | ||
261 | } | ||
262 | /* | ||
263 | * There is a remote possibility that we end up with a zero sum. | ||
264 | * Zero is used as an EOF marker, so return 1 instead. | ||
265 | */ | ||
266 | return (sum == 0 ? 1 : sum); | ||
267 | } | ||
268 | |||
269 | |||
270 | /* | ||
271 | * Check to see if the given files differ. | ||
272 | * Returns 0 if they are the same, 1 if different, and -1 on error. | ||
273 | */ | ||
274 | static int files_differ(FILE * f1, FILE * f2, int flags) | ||
275 | { | ||
276 | size_t i, j; | ||
277 | |||
278 | if ((flags & (D_EMPTY1 | D_EMPTY2)) || stb1.st_size != stb2.st_size | ||
279 | || (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT) | ||
280 | ) { | ||
281 | return 1; | ||
282 | } | ||
283 | while (1) { | ||
284 | i = fread(bb_common_bufsiz1, 1, BUFSIZ/2, f1); | ||
285 | j = fread(bb_common_bufsiz1 + BUFSIZ/2, 1, BUFSIZ/2, f2); | ||
286 | if (i != j) | ||
287 | return 1; | ||
288 | if (i == 0) | ||
289 | return (ferror(f1) || ferror(f2)); | ||
290 | if (memcmp(bb_common_bufsiz1, | ||
291 | bb_common_bufsiz1 + BUFSIZ/2, i) != 0) | ||
292 | return 1; | ||
293 | } | ||
294 | } | ||
295 | |||
296 | |||
297 | static void prepare(int i, FILE * fd, off_t filesize) | ||
298 | { | ||
299 | struct line *p; | ||
300 | int h; | ||
301 | size_t j, sz; | ||
302 | |||
303 | rewind(fd); | ||
304 | |||
305 | sz = (filesize <= FSIZE_MAX ? filesize : FSIZE_MAX) / 25; | ||
306 | if (sz < 100) | ||
307 | sz = 100; | ||
308 | |||
309 | p = xmalloc((sz + 3) * sizeof(struct line)); | ||
310 | j = 0; | ||
311 | while ((h = readhash(fd))) { | ||
312 | if (j == sz) { | ||
313 | sz = sz * 3 / 2; | ||
314 | p = xrealloc(p, (sz + 3) * sizeof(struct line)); | ||
315 | } | ||
316 | p[++j].value = h; | ||
317 | } | ||
318 | len[i] = j; | ||
319 | file[i] = p; | ||
320 | } | ||
321 | |||
322 | |||
323 | static void prune(void) | ||
324 | { | ||
325 | int i, j; | ||
326 | |||
327 | for (pref = 0; pref < len[0] && pref < len[1] && | ||
328 | file[0][pref + 1].value == file[1][pref + 1].value; pref++) | ||
329 | ; | ||
330 | for (suff = 0; suff < len[0] - pref && suff < len[1] - pref && | ||
331 | file[0][len[0] - suff].value == file[1][len[1] - suff].value; | ||
332 | suff++) | ||
333 | ; | ||
334 | for (j = 0; j < 2; j++) { | ||
335 | sfile[j] = file[j] + pref; | ||
336 | slen[j] = len[j] - pref - suff; | ||
337 | for (i = 0; i <= slen[j]; i++) | ||
338 | sfile[j][i].serial = i; | ||
339 | } | ||
340 | } | ||
341 | |||
342 | |||
343 | static void equiv(struct line *a, int n, struct line *b, int m, int *c) | ||
344 | { | ||
345 | int i, j; | ||
346 | |||
347 | i = j = 1; | ||
348 | while (i <= n && j <= m) { | ||
349 | if (a[i].value < b[j].value) | ||
350 | a[i++].value = 0; | ||
351 | else if (a[i].value == b[j].value) | ||
352 | a[i++].value = j; | ||
353 | else | ||
354 | j++; | ||
355 | } | ||
356 | while (i <= n) | ||
357 | a[i++].value = 0; | ||
358 | b[m + 1].value = 0; | ||
359 | j = 0; | ||
360 | while (++j <= m) { | ||
361 | c[j] = -b[j].serial; | ||
362 | while (b[j + 1].value == b[j].value) { | ||
363 | j++; | ||
364 | c[j] = b[j].serial; | ||
365 | } | ||
366 | } | ||
367 | c[j] = -1; | ||
368 | } | ||
369 | |||
370 | |||
371 | static int isqrt(int n) | ||
372 | { | ||
373 | int y, x; | ||
374 | |||
375 | if (n == 0) | ||
376 | return 0; | ||
377 | x = 1; | ||
378 | do { | ||
379 | y = x; | ||
380 | x = n / x; | ||
381 | x += y; | ||
382 | x /= 2; | ||
383 | } while ((x - y) > 1 || (x - y) < -1); | ||
384 | |||
385 | return x; | ||
386 | } | ||
387 | |||
388 | |||
389 | static int newcand(int x, int y, int pred) | ||
390 | { | ||
391 | struct cand *q; | ||
392 | |||
393 | if (clen == clistlen) { | ||
394 | clistlen = clistlen * 11 / 10; | ||
395 | clist = xrealloc(clist, clistlen * sizeof(struct cand)); | ||
396 | } | ||
397 | q = clist + clen; | ||
398 | q->x = x; | ||
399 | q->y = y; | ||
400 | q->pred = pred; | ||
401 | return clen++; | ||
402 | } | ||
403 | |||
404 | |||
405 | static int search(int *c, int k, int y) | ||
406 | { | ||
407 | int i, j, l, t; | ||
408 | |||
409 | if (clist[c[k]].y < y) /* quick look for typical case */ | ||
410 | return k + 1; | ||
411 | i = 0; | ||
412 | j = k + 1; | ||
413 | while (1) { | ||
414 | l = i + j; | ||
415 | if ((l >>= 1) <= i) | ||
416 | break; | ||
417 | t = clist[c[l]].y; | ||
418 | if (t > y) | ||
419 | j = l; | ||
420 | else if (t < y) | ||
421 | i = l; | ||
422 | else | ||
423 | return l; | ||
424 | } | ||
425 | return l + 1; | ||
426 | } | ||
427 | |||
428 | |||
429 | static int stone(int *a, int n, int *b, int *c) | ||
430 | { | ||
431 | int i, k, y, j, l; | ||
432 | int oldc, tc, oldl; | ||
433 | unsigned int numtries; | ||
434 | |||
435 | #if ENABLE_FEATURE_DIFF_MINIMAL | ||
436 | const unsigned int bound = | ||
437 | (option_mask32 & FLAG_d) ? UINT_MAX : MAX(256, isqrt(n)); | ||
438 | #else | ||
439 | const unsigned int bound = MAX(256, isqrt(n)); | ||
440 | #endif | ||
441 | k = 0; | ||
442 | c[0] = newcand(0, 0, 0); | ||
443 | for (i = 1; i <= n; i++) { | ||
444 | j = a[i]; | ||
445 | if (j == 0) | ||
446 | continue; | ||
447 | y = -b[j]; | ||
448 | oldl = 0; | ||
449 | oldc = c[0]; | ||
450 | numtries = 0; | ||
451 | do { | ||
452 | if (y <= clist[oldc].y) | ||
453 | continue; | ||
454 | l = search(c, k, y); | ||
455 | if (l != oldl + 1) | ||
456 | oldc = c[l - 1]; | ||
457 | if (l <= k) { | ||
458 | if (clist[c[l]].y <= y) | ||
459 | continue; | ||
460 | tc = c[l]; | ||
461 | c[l] = newcand(i, y, oldc); | ||
462 | oldc = tc; | ||
463 | oldl = l; | ||
464 | numtries++; | ||
465 | } else { | ||
466 | c[l] = newcand(i, y, oldc); | ||
467 | k++; | ||
468 | break; | ||
469 | } | ||
470 | } while ((y = b[++j]) > 0 && numtries < bound); | ||
471 | } | ||
472 | return k; | ||
473 | } | ||
474 | |||
475 | |||
476 | static void unravel(int p) | ||
477 | { | ||
478 | struct cand *q; | ||
479 | int i; | ||
480 | |||
481 | for (i = 0; i <= len[0]; i++) | ||
482 | J[i] = i <= pref ? i : i > len[0] - suff ? i + len[1] - len[0] : 0; | ||
483 | for (q = clist + p; q->y != 0; q = clist + q->pred) | ||
484 | J[q->x + pref] = q->y + pref; | ||
485 | } | ||
486 | |||
487 | |||
488 | static void unsort(struct line *f, int l, int *b) | ||
489 | { | ||
490 | int *a, i; | ||
491 | |||
492 | a = xmalloc((l + 1) * sizeof(int)); | ||
493 | for (i = 1; i <= l; i++) | ||
494 | a[f[i].serial] = f[i].value; | ||
495 | for (i = 1; i <= l; i++) | ||
496 | b[i] = a[i]; | ||
497 | free(a); | ||
498 | } | ||
499 | |||
500 | |||
501 | static int skipline(FILE * f) | ||
502 | { | ||
503 | int i, c; | ||
504 | |||
505 | for (i = 1; (c = getc(f)) != '\n' && c != EOF; i++) | ||
506 | continue; | ||
507 | return i; | ||
508 | } | ||
509 | |||
510 | |||
511 | /* | ||
512 | * Check does double duty: | ||
513 | * 1. ferret out any fortuitous correspondences due | ||
514 | * to confounding by hashing (which result in "jackpot") | ||
515 | * 2. collect random access indexes to the two files | ||
516 | */ | ||
517 | static void check(FILE * f1, FILE * f2) | ||
518 | { | ||
519 | int i, j, jackpot, c, d; | ||
520 | long ctold, ctnew; | ||
521 | |||
522 | rewind(f1); | ||
523 | rewind(f2); | ||
524 | j = 1; | ||
525 | ixold[0] = ixnew[0] = 0; | ||
526 | jackpot = 0; | ||
527 | ctold = ctnew = 0; | ||
528 | for (i = 1; i <= len[0]; i++) { | ||
529 | if (J[i] == 0) { | ||
530 | ixold[i] = ctold += skipline(f1); | ||
531 | continue; | ||
532 | } | ||
533 | while (j < J[i]) { | ||
534 | ixnew[j] = ctnew += skipline(f2); | ||
535 | j++; | ||
536 | } | ||
537 | if ((option_mask32 & FLAG_b) || (option_mask32 & FLAG_w) | ||
538 | || (option_mask32 & FLAG_i)) { | ||
539 | while (1) { | ||
540 | c = getc(f1); | ||
541 | d = getc(f2); | ||
542 | /* | ||
543 | * GNU diff ignores a missing newline | ||
544 | * in one file if bflag || wflag. | ||
545 | */ | ||
546 | if (((option_mask32 & FLAG_b) || (option_mask32 & FLAG_w)) && | ||
547 | ((c == EOF && d == '\n') || (c == '\n' && d == EOF))) { | ||
548 | break; | ||
549 | } | ||
550 | ctold++; | ||
551 | ctnew++; | ||
552 | if ((option_mask32 & FLAG_b) && isspace(c) && isspace(d)) { | ||
553 | do { | ||
554 | if (c == '\n') | ||
555 | break; | ||
556 | ctold++; | ||
557 | } while (isspace(c = getc(f1))); | ||
558 | do { | ||
559 | if (d == '\n') | ||
560 | break; | ||
561 | ctnew++; | ||
562 | } while (isspace(d = getc(f2))); | ||
563 | } else if (option_mask32 & FLAG_w) { | ||
564 | while (isspace(c) && c != '\n') { | ||
565 | c = getc(f1); | ||
566 | ctold++; | ||
567 | } | ||
568 | while (isspace(d) && d != '\n') { | ||
569 | d = getc(f2); | ||
570 | ctnew++; | ||
571 | } | ||
572 | } | ||
573 | if (c != d) { | ||
574 | jackpot++; | ||
575 | J[i] = 0; | ||
576 | if (c != '\n' && c != EOF) | ||
577 | ctold += skipline(f1); | ||
578 | if (d != '\n' && c != EOF) | ||
579 | ctnew += skipline(f2); | ||
580 | break; | ||
581 | } | ||
582 | if (c == '\n' || c == EOF) | ||
583 | break; | ||
584 | } | ||
585 | } else { | ||
586 | while (1) { | ||
587 | ctold++; | ||
588 | ctnew++; | ||
589 | if ((c = getc(f1)) != (d = getc(f2))) { | ||
590 | J[i] = 0; | ||
591 | if (c != '\n' && c != EOF) | ||
592 | ctold += skipline(f1); | ||
593 | if (d != '\n' && c != EOF) | ||
594 | ctnew += skipline(f2); | ||
595 | break; | ||
596 | } | ||
597 | if (c == '\n' || c == EOF) | ||
598 | break; | ||
599 | } | ||
600 | } | ||
601 | ixold[i] = ctold; | ||
602 | ixnew[j] = ctnew; | ||
603 | j++; | ||
604 | } | ||
605 | for (; j <= len[1]; j++) | ||
606 | ixnew[j] = ctnew += skipline(f2); | ||
607 | } | ||
608 | |||
609 | |||
610 | /* shellsort CACM #201 */ | ||
611 | static void sort(struct line *a, int n) | ||
612 | { | ||
613 | struct line *ai, *aim, w; | ||
614 | int j, m = 0, k; | ||
615 | |||
616 | if (n == 0) | ||
617 | return; | ||
618 | for (j = 1; j <= n; j *= 2) | ||
619 | m = 2 * j - 1; | ||
620 | for (m /= 2; m != 0; m /= 2) { | ||
621 | k = n - m; | ||
622 | for (j = 1; j <= k; j++) { | ||
623 | for (ai = &a[j]; ai > a; ai -= m) { | ||
624 | aim = &ai[m]; | ||
625 | if (aim < ai) | ||
626 | break; /* wraparound */ | ||
627 | if (aim->value > ai[0].value || | ||
628 | (aim->value == ai[0].value && aim->serial > ai[0].serial)) | ||
629 | break; | ||
630 | w.value = ai[0].value; | ||
631 | ai[0].value = aim->value; | ||
632 | aim->value = w.value; | ||
633 | w.serial = ai[0].serial; | ||
634 | ai[0].serial = aim->serial; | ||
635 | aim->serial = w.serial; | ||
636 | } | ||
637 | } | ||
638 | } | ||
639 | } | ||
640 | |||
641 | |||
642 | static void uni_range(int a, int b) | ||
643 | { | ||
644 | if (a < b) | ||
645 | printf("%d,%d", a, b - a + 1); | ||
646 | else if (a == b) | ||
647 | printf("%d", b); | ||
648 | else | ||
649 | printf("%d,0", b); | ||
650 | } | ||
651 | |||
652 | |||
653 | static void fetch(long *f, int a, int b, FILE * lb, int ch) | ||
654 | { | ||
655 | int i, j, c, lastc, col, nc; | ||
656 | |||
657 | if (a > b) | ||
658 | return; | ||
659 | for (i = a; i <= b; i++) { | ||
660 | fseek(lb, f[i - 1], SEEK_SET); | ||
661 | nc = f[i] - f[i - 1]; | ||
662 | if (ch != '\0') { | ||
663 | putchar(ch); | ||
664 | if (option_mask32 & FLAG_T) | ||
665 | putchar('\t'); | ||
666 | } | ||
667 | col = 0; | ||
668 | for (j = 0, lastc = '\0'; j < nc; j++, lastc = c) { | ||
669 | if ((c = getc(lb)) == EOF) { | ||
670 | printf("\n\\ No newline at end of file\n"); | ||
671 | return; | ||
672 | } | ||
673 | if (c == '\t' && (option_mask32 & FLAG_t)) { | ||
674 | do { | ||
675 | putchar(' '); | ||
676 | } while (++col & 7); | ||
677 | } else { | ||
678 | putchar(c); | ||
679 | col++; | ||
680 | } | ||
681 | } | ||
682 | } | ||
683 | } | ||
684 | |||
685 | |||
686 | static int asciifile(FILE * f) | ||
687 | { | ||
688 | #if ENABLE_FEATURE_DIFF_BINARY | ||
689 | int i, cnt; | ||
690 | #endif | ||
691 | |||
692 | if ((option_mask32 & FLAG_a) || f == NULL) | ||
693 | return 1; | ||
694 | |||
695 | #if ENABLE_FEATURE_DIFF_BINARY | ||
696 | rewind(f); | ||
697 | cnt = fread(bb_common_bufsiz1, 1, BUFSIZ, f); | ||
698 | for (i = 0; i < cnt; i++) { | ||
699 | if (!isprint(bb_common_bufsiz1[i]) | ||
700 | && !isspace(bb_common_bufsiz1[i])) { | ||
701 | return 0; | ||
702 | } | ||
703 | } | ||
704 | #endif | ||
705 | return 1; | ||
706 | } | ||
707 | |||
708 | |||
709 | /* dump accumulated "unified" diff changes */ | ||
710 | static void dump_unified_vec(FILE * f1, FILE * f2) | ||
711 | { | ||
712 | struct context_vec *cvp = context_vec_start; | ||
713 | int lowa, upb, lowc, upd; | ||
714 | int a, b, c, d; | ||
715 | char ch; | ||
716 | |||
717 | if (context_vec_start > context_vec_ptr) | ||
718 | return; | ||
719 | |||
720 | b = d = 0; /* gcc */ | ||
721 | lowa = MAX(1, cvp->a - context); | ||
722 | upb = MIN(len[0], context_vec_ptr->b + context); | ||
723 | lowc = MAX(1, cvp->c - context); | ||
724 | upd = MIN(len[1], context_vec_ptr->d + context); | ||
725 | |||
726 | printf("@@ -"); | ||
727 | uni_range(lowa, upb); | ||
728 | printf(" +"); | ||
729 | uni_range(lowc, upd); | ||
730 | printf(" @@\n"); | ||
731 | |||
732 | /* | ||
733 | * Output changes in "unified" diff format--the old and new lines | ||
734 | * are printed together. | ||
735 | */ | ||
736 | for (; cvp <= context_vec_ptr; cvp++) { | ||
737 | a = cvp->a; | ||
738 | b = cvp->b; | ||
739 | c = cvp->c; | ||
740 | d = cvp->d; | ||
741 | |||
742 | /* | ||
743 | * c: both new and old changes | ||
744 | * d: only changes in the old file | ||
745 | * a: only changes in the new file | ||
746 | */ | ||
747 | if (a <= b && c <= d) | ||
748 | ch = 'c'; | ||
749 | else | ||
750 | ch = (a <= b) ? 'd' : 'a'; | ||
751 | #if 0 | ||
752 | switch (ch) { | ||
753 | case 'c': | ||
754 | fetch(ixold, lowa, a - 1, f1, ' '); | ||
755 | fetch(ixold, a, b, f1, '-'); | ||
756 | fetch(ixnew, c, d, f2, '+'); | ||
757 | break; | ||
758 | case 'd': | ||
759 | fetch(ixold, lowa, a - 1, f1, ' '); | ||
760 | fetch(ixold, a, b, f1, '-'); | ||
761 | break; | ||
762 | case 'a': | ||
763 | fetch(ixnew, lowc, c - 1, f2, ' '); | ||
764 | fetch(ixnew, c, d, f2, '+'); | ||
765 | break; | ||
766 | } | ||
767 | #else | ||
768 | if (ch == 'c' || ch == 'd') { | ||
769 | fetch(ixold, lowa, a - 1, f1, ' '); | ||
770 | fetch(ixold, a, b, f1, '-'); | ||
771 | } | ||
772 | if (ch == 'a') | ||
773 | fetch(ixnew, lowc, c - 1, f2, ' '); | ||
774 | if (ch == 'c' || ch == 'a') | ||
775 | fetch(ixnew, c, d, f2, '+'); | ||
776 | #endif | ||
777 | lowa = b + 1; | ||
778 | lowc = d + 1; | ||
779 | } | ||
780 | fetch(ixnew, d + 1, upd, f2, ' '); | ||
781 | |||
782 | context_vec_ptr = context_vec_start - 1; | ||
783 | } | ||
784 | |||
785 | |||
786 | static void print_header(const char *file1, const char *file2) | ||
787 | { | ||
788 | if (label1) | ||
789 | printf("--- %s\n", label1); | ||
790 | else | ||
791 | printf("--- %s\t%s", file1, ctime(&stb1.st_mtime)); | ||
792 | if (label2) | ||
793 | printf("+++ %s\n", label2); | ||
794 | else | ||
795 | printf("+++ %s\t%s", file2, ctime(&stb2.st_mtime)); | ||
796 | } | ||
797 | |||
798 | |||
799 | /* | ||
800 | * Indicate that there is a difference between lines a and b of the from file | ||
801 | * to get to lines c to d of the to file. If a is greater than b then there | ||
802 | * are no lines in the from file involved and this means that there were | ||
803 | * lines appended (beginning at b). If c is greater than d then there are | ||
804 | * lines missing from the to file. | ||
805 | */ | ||
806 | static void change(char *file1, FILE * f1, char *file2, FILE * f2, int a, | ||
807 | int b, int c, int d) | ||
808 | { | ||
809 | if ((a > b && c > d) || (option_mask32 & FLAG_q)) { | ||
810 | anychange = 1; | ||
811 | return; | ||
812 | } | ||
813 | |||
814 | /* | ||
815 | * Allocate change records as needed. | ||
816 | */ | ||
817 | if (context_vec_ptr == context_vec_end - 1) { | ||
818 | ptrdiff_t offset = context_vec_ptr - context_vec_start; | ||
819 | |||
820 | max_context <<= 1; | ||
821 | context_vec_start = xrealloc(context_vec_start, | ||
822 | max_context * sizeof(struct context_vec)); | ||
823 | context_vec_end = context_vec_start + max_context; | ||
824 | context_vec_ptr = context_vec_start + offset; | ||
825 | } | ||
826 | if (anychange == 0) { | ||
827 | /* | ||
828 | * Print the context/unidiff header first time through. | ||
829 | */ | ||
830 | print_header(file1, file2); | ||
831 | } else if (a > context_vec_ptr->b + (2 * context) + 1 && | ||
832 | c > context_vec_ptr->d + (2 * context) + 1) { | ||
833 | /* | ||
834 | * If this change is more than 'context' lines from the | ||
835 | * previous change, dump the record and reset it. | ||
836 | */ | ||
837 | dump_unified_vec(f1, f2); | ||
838 | } | ||
839 | context_vec_ptr++; | ||
840 | context_vec_ptr->a = a; | ||
841 | context_vec_ptr->b = b; | ||
842 | context_vec_ptr->c = c; | ||
843 | context_vec_ptr->d = d; | ||
844 | anychange = 1; | ||
845 | } | ||
846 | |||
847 | |||
848 | static void output(char *file1, FILE * f1, char *file2, FILE * f2) | ||
849 | { | ||
850 | /* Note that j0 and j1 can't be used as they are defined in math.h. | ||
851 | * This also allows the rather amusing variable 'j00'... */ | ||
852 | int m, i0, i1, j00, j01; | ||
853 | |||
854 | rewind(f1); | ||
855 | rewind(f2); | ||
856 | m = len[0]; | ||
857 | J[0] = 0; | ||
858 | J[m + 1] = len[1] + 1; | ||
859 | for (i0 = 1; i0 <= m; i0 = i1 + 1) { | ||
860 | while (i0 <= m && J[i0] == J[i0 - 1] + 1) | ||
861 | i0++; | ||
862 | j00 = J[i0 - 1] + 1; | ||
863 | i1 = i0 - 1; | ||
864 | while (i1 < m && J[i1 + 1] == 0) | ||
865 | i1++; | ||
866 | j01 = J[i1 + 1] - 1; | ||
867 | J[i1] = j01; | ||
868 | change(file1, f1, file2, f2, i0, i1, j00, j01); | ||
869 | } | ||
870 | if (m == 0) { | ||
871 | change(file1, f1, file2, f2, 1, 0, 1, len[1]); | ||
872 | } | ||
873 | if (anychange != 0 && !(option_mask32 & FLAG_q)) { | ||
874 | dump_unified_vec(f1, f2); | ||
875 | } | ||
876 | } | ||
877 | |||
878 | /* | ||
879 | * The following code uses an algorithm due to Harold Stone, | ||
880 | * which finds a pair of longest identical subsequences in | ||
881 | * the two files. | ||
882 | * | ||
883 | * The major goal is to generate the match vector J. | ||
884 | * J[i] is the index of the line in file1 corresponding | ||
885 | * to line i file0. J[i] = 0 if there is no | ||
886 | * such line in file1. | ||
887 | * | ||
888 | * Lines are hashed so as to work in core. All potential | ||
889 | * matches are located by sorting the lines of each file | ||
890 | * on the hash (called ``value''). In particular, this | ||
891 | * collects the equivalence classes in file1 together. | ||
892 | * Subroutine equiv replaces the value of each line in | ||
893 | * file0 by the index of the first element of its | ||
894 | * matching equivalence in (the reordered) file1. | ||
895 | * To save space equiv squeezes file1 into a single | ||
896 | * array member in which the equivalence classes | ||
897 | * are simply concatenated, except that their first | ||
898 | * members are flagged by changing sign. | ||
899 | * | ||
900 | * Next the indices that point into member are unsorted into | ||
901 | * array class according to the original order of file0. | ||
902 | * | ||
903 | * The cleverness lies in routine stone. This marches | ||
904 | * through the lines of file0, developing a vector klist | ||
905 | * of "k-candidates". At step i a k-candidate is a matched | ||
906 | * pair of lines x,y (x in file0 y in file1) such that | ||
907 | * there is a common subsequence of length k | ||
908 | * between the first i lines of file0 and the first y | ||
909 | * lines of file1, but there is no such subsequence for | ||
910 | * any smaller y. x is the earliest possible mate to y | ||
911 | * that occurs in such a subsequence. | ||
912 | * | ||
913 | * Whenever any of the members of the equivalence class of | ||
914 | * lines in file1 matable to a line in file0 has serial number | ||
915 | * less than the y of some k-candidate, that k-candidate | ||
916 | * with the smallest such y is replaced. The new | ||
917 | * k-candidate is chained (via pred) to the current | ||
918 | * k-1 candidate so that the actual subsequence can | ||
919 | * be recovered. When a member has serial number greater | ||
920 | * that the y of all k-candidates, the klist is extended. | ||
921 | * At the end, the longest subsequence is pulled out | ||
922 | * and placed in the array J by unravel | ||
923 | * | ||
924 | * With J in hand, the matches there recorded are | ||
925 | * checked against reality to assure that no spurious | ||
926 | * matches have crept in due to hashing. If they have, | ||
927 | * they are broken, and "jackpot" is recorded--a harmless | ||
928 | * matter except that a true match for a spuriously | ||
929 | * mated line may now be unnecessarily reported as a change. | ||
930 | * | ||
931 | * Much of the complexity of the program comes simply | ||
932 | * from trying to minimize core utilization and | ||
933 | * maximize the range of doable problems by dynamically | ||
934 | * allocating what is needed and reusing what is not. | ||
935 | * The core requirements for problems larger than somewhat | ||
936 | * are (in words) 2*length(file0) + length(file1) + | ||
937 | * 3*(number of k-candidates installed), typically about | ||
938 | * 6n words for files of length n. | ||
939 | */ | ||
940 | static unsigned diffreg(char * ofile1, char * ofile2, int flags) | ||
941 | { | ||
942 | char *file1 = ofile1; | ||
943 | char *file2 = ofile2; | ||
944 | FILE *f1 = stdin, *f2 = stdin; | ||
945 | unsigned rval; | ||
946 | int i; | ||
947 | |||
948 | anychange = 0; | ||
949 | context_vec_ptr = context_vec_start - 1; | ||
950 | |||
951 | if (S_ISDIR(stb1.st_mode) != S_ISDIR(stb2.st_mode)) | ||
952 | return (S_ISDIR(stb1.st_mode) ? D_MISMATCH1 : D_MISMATCH2); | ||
953 | |||
954 | rval = D_SAME; | ||
955 | |||
956 | if (LONE_DASH(file1) && LONE_DASH(file2)) | ||
957 | goto closem; | ||
958 | |||
959 | if (flags & D_EMPTY1) | ||
960 | f1 = xfopen(bb_dev_null, "r"); | ||
961 | else if (NOT_LONE_DASH(file1)) | ||
962 | f1 = xfopen(file1, "r"); | ||
963 | if (flags & D_EMPTY2) | ||
964 | f2 = xfopen(bb_dev_null, "r"); | ||
965 | else if (NOT_LONE_DASH(file2)) | ||
966 | f2 = xfopen(file2, "r"); | ||
967 | |||
968 | /* We can't diff non-seekable stream - we use rewind(), fseek(). | ||
969 | * This can be fixed (volunteers?). | ||
970 | * Meanwhile we should check it here by stat'ing input fds, | ||
971 | * but I am lazy and check that in main() instead. | ||
972 | * Check in main won't catch "diffing fifos buried in subdirectories" | ||
973 | * failure scenario - not very likely in real life... */ | ||
974 | |||
975 | i = files_differ(f1, f2, flags); | ||
976 | if (i == 0) | ||
977 | goto closem; | ||
978 | else if (i != 1) { /* 1 == ok */ | ||
979 | /* error */ | ||
980 | status |= 2; | ||
981 | goto closem; | ||
982 | } | ||
983 | |||
984 | if (!asciifile(f1) || !asciifile(f2)) { | ||
985 | rval = D_BINARY; | ||
986 | status |= 1; | ||
987 | goto closem; | ||
988 | } | ||
989 | |||
990 | prepare(0, f1, stb1.st_size); | ||
991 | prepare(1, f2, stb2.st_size); | ||
992 | prune(); | ||
993 | sort(sfile[0], slen[0]); | ||
994 | sort(sfile[1], slen[1]); | ||
995 | |||
996 | member = (int *) file[1]; | ||
997 | equiv(sfile[0], slen[0], sfile[1], slen[1], member); | ||
998 | member = xrealloc(member, (slen[1] + 2) * sizeof(int)); | ||
999 | |||
1000 | class = (int *) file[0]; | ||
1001 | unsort(sfile[0], slen[0], class); | ||
1002 | class = xrealloc(class, (slen[0] + 2) * sizeof(int)); | ||
1003 | |||
1004 | klist = xmalloc((slen[0] + 2) * sizeof(int)); | ||
1005 | clen = 0; | ||
1006 | clistlen = 100; | ||
1007 | clist = xmalloc(clistlen * sizeof(struct cand)); | ||
1008 | i = stone(class, slen[0], member, klist); | ||
1009 | free(member); | ||
1010 | free(class); | ||
1011 | |||
1012 | J = xrealloc(J, (len[0] + 2) * sizeof(int)); | ||
1013 | unravel(klist[i]); | ||
1014 | free(clist); | ||
1015 | free(klist); | ||
1016 | |||
1017 | ixold = xrealloc(ixold, (len[0] + 2) * sizeof(long)); | ||
1018 | ixnew = xrealloc(ixnew, (len[1] + 2) * sizeof(long)); | ||
1019 | check(f1, f2); | ||
1020 | output(file1, f1, file2, f2); | ||
1021 | |||
1022 | closem: | ||
1023 | if (anychange) { | ||
1024 | status |= 1; | ||
1025 | if (rval == D_SAME) | ||
1026 | rval = D_DIFFER; | ||
1027 | } | ||
1028 | fclose_if_not_stdin(f1); | ||
1029 | fclose_if_not_stdin(f2); | ||
1030 | if (file1 != ofile1) | ||
1031 | free(file1); | ||
1032 | if (file2 != ofile2) | ||
1033 | free(file2); | ||
1034 | return rval; | ||
1035 | } | ||
1036 | |||
1037 | |||
1038 | #if ENABLE_FEATURE_DIFF_DIR | ||
1039 | static void do_diff(char *dir1, char *path1, char *dir2, char *path2) | ||
1040 | { | ||
1041 | int flags = D_HEADER; | ||
1042 | int val; | ||
1043 | char *fullpath1 = NULL; /* if -N */ | ||
1044 | char *fullpath2 = NULL; | ||
1045 | |||
1046 | if (path1) | ||
1047 | fullpath1 = concat_path_file(dir1, path1); | ||
1048 | if (path2) | ||
1049 | fullpath2 = concat_path_file(dir2, path2); | ||
1050 | |||
1051 | if (!fullpath1 || stat(fullpath1, &stb1) != 0) { | ||
1052 | flags |= D_EMPTY1; | ||
1053 | memset(&stb1, 0, sizeof(stb1)); | ||
1054 | if (path2) { | ||
1055 | free(fullpath1); | ||
1056 | fullpath1 = concat_path_file(dir1, path2); | ||
1057 | } | ||
1058 | } | ||
1059 | if (!fullpath2 || stat(fullpath2, &stb2) != 0) { | ||
1060 | flags |= D_EMPTY2; | ||
1061 | memset(&stb2, 0, sizeof(stb2)); | ||
1062 | stb2.st_mode = stb1.st_mode; | ||
1063 | if (path1) { | ||
1064 | free(fullpath2); | ||
1065 | fullpath2 = concat_path_file(dir2, path1); | ||
1066 | } | ||
1067 | } | ||
1068 | |||
1069 | if (stb1.st_mode == 0) | ||
1070 | stb1.st_mode = stb2.st_mode; | ||
1071 | |||
1072 | if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) { | ||
1073 | printf("Common subdirectories: %s and %s\n", fullpath1, fullpath2); | ||
1074 | goto ret; | ||
1075 | } | ||
1076 | |||
1077 | if (!S_ISREG(stb1.st_mode) && !S_ISDIR(stb1.st_mode)) | ||
1078 | val = D_SKIPPED1; | ||
1079 | else if (!S_ISREG(stb2.st_mode) && !S_ISDIR(stb2.st_mode)) | ||
1080 | val = D_SKIPPED2; | ||
1081 | else | ||
1082 | val = diffreg(fullpath1, fullpath2, flags); | ||
1083 | |||
1084 | print_status(val, fullpath1, fullpath2, NULL); | ||
1085 | ret: | ||
1086 | free(fullpath1); | ||
1087 | free(fullpath2); | ||
1088 | } | ||
1089 | #endif | ||
1090 | |||
1091 | |||
1092 | #if ENABLE_FEATURE_DIFF_DIR | ||
1093 | static int dir_strcmp(const void *p1, const void *p2) | ||
1094 | { | ||
1095 | return strcmp(*(char *const *) p1, *(char *const *) p2); | ||
1096 | } | ||
1097 | |||
1098 | |||
1099 | /* This function adds a filename to dl, the directory listing. */ | ||
1100 | static int add_to_dirlist(const char *filename, | ||
1101 | struct stat ATTRIBUTE_UNUSED * sb, void *userdata, | ||
1102 | int depth ATTRIBUTE_UNUSED) | ||
1103 | { | ||
1104 | /* +2: with space for eventual trailing NULL */ | ||
1105 | dl = xrealloc(dl, (dl_count+2) * sizeof(dl[0])); | ||
1106 | dl[dl_count] = xstrdup(filename + (int)(ptrdiff_t)userdata); | ||
1107 | dl_count++; | ||
1108 | return TRUE; | ||
1109 | } | ||
1110 | |||
1111 | |||
1112 | /* This returns a sorted directory listing. */ | ||
1113 | static char **get_dir(char *path) | ||
1114 | { | ||
1115 | dl_count = 0; | ||
1116 | dl = xzalloc(sizeof(dl[0])); | ||
1117 | |||
1118 | /* If -r has been set, then the recursive_action function will be | ||
1119 | * used. Unfortunately, this outputs the root directory along with | ||
1120 | * the recursed paths, so use void *userdata to specify the string | ||
1121 | * length of the root directory - '(void*)(strlen(path)+)'. | ||
1122 | * add_to_dirlist then removes root dir prefix. */ | ||
1123 | |||
1124 | if (option_mask32 & FLAG_r) { | ||
1125 | recursive_action(path, ACTION_RECURSE|ACTION_FOLLOWLINKS, | ||
1126 | add_to_dirlist, NULL, | ||
1127 | (void*)(strlen(path)+1), 0); | ||
1128 | } else { | ||
1129 | DIR *dp; | ||
1130 | struct dirent *ep; | ||
1131 | |||
1132 | dp = warn_opendir(path); | ||
1133 | while ((ep = readdir(dp))) { | ||
1134 | if (!strcmp(ep->d_name, "..") || LONE_CHAR(ep->d_name, '.')) | ||
1135 | continue; | ||
1136 | add_to_dirlist(ep->d_name, NULL, (void*)(int)0, 0); | ||
1137 | } | ||
1138 | closedir(dp); | ||
1139 | } | ||
1140 | |||
1141 | /* Sort dl alphabetically. */ | ||
1142 | qsort(dl, dl_count, sizeof(char *), dir_strcmp); | ||
1143 | |||
1144 | dl[dl_count] = NULL; | ||
1145 | return dl; | ||
1146 | } | ||
1147 | |||
1148 | |||
1149 | static void diffdir(char *p1, char *p2) | ||
1150 | { | ||
1151 | char **dirlist1, **dirlist2; | ||
1152 | char *dp1, *dp2; | ||
1153 | int pos; | ||
1154 | |||
1155 | /* Check for trailing slashes. */ | ||
1156 | dp1 = last_char_is(p1, '/'); | ||
1157 | if (dp1 != NULL) | ||
1158 | *dp1 = '\0'; | ||
1159 | dp2 = last_char_is(p2, '/'); | ||
1160 | if (dp2 != NULL) | ||
1161 | *dp2 = '\0'; | ||
1162 | |||
1163 | /* Get directory listings for p1 and p2. */ | ||
1164 | |||
1165 | dirlist1 = get_dir(p1); | ||
1166 | dirlist2 = get_dir(p2); | ||
1167 | |||
1168 | /* If -S was set, find the starting point. */ | ||
1169 | if (start) { | ||
1170 | while (*dirlist1 != NULL && strcmp(*dirlist1, start) < 0) | ||
1171 | dirlist1++; | ||
1172 | while (*dirlist2 != NULL && strcmp(*dirlist2, start) < 0) | ||
1173 | dirlist2++; | ||
1174 | if ((*dirlist1 == NULL) || (*dirlist2 == NULL)) | ||
1175 | bb_error_msg(bb_msg_invalid_arg, "NULL", "-S"); | ||
1176 | } | ||
1177 | |||
1178 | /* Now that both dirlist1 and dirlist2 contain sorted directory | ||
1179 | * listings, we can start to go through dirlist1. If both listings | ||
1180 | * contain the same file, then do a normal diff. Otherwise, behaviour | ||
1181 | * is determined by whether the -N flag is set. */ | ||
1182 | while (*dirlist1 != NULL || *dirlist2 != NULL) { | ||
1183 | dp1 = *dirlist1; | ||
1184 | dp2 = *dirlist2; | ||
1185 | pos = dp1 == NULL ? 1 : dp2 == NULL ? -1 : strcmp(dp1, dp2); | ||
1186 | if (pos == 0) { | ||
1187 | do_diff(p1, dp1, p2, dp2); | ||
1188 | dirlist1++; | ||
1189 | dirlist2++; | ||
1190 | } else if (pos < 0) { | ||
1191 | if (option_mask32 & FLAG_N) | ||
1192 | do_diff(p1, dp1, p2, NULL); | ||
1193 | else | ||
1194 | print_only(p1, strlen(p1) + 1, dp1); | ||
1195 | dirlist1++; | ||
1196 | } else { | ||
1197 | if (option_mask32 & FLAG_N) | ||
1198 | do_diff(p1, NULL, p2, dp2); | ||
1199 | else | ||
1200 | print_only(p2, strlen(p2) + 1, dp2); | ||
1201 | dirlist2++; | ||
1202 | } | ||
1203 | } | ||
1204 | } | ||
1205 | #endif | ||
1206 | |||
1207 | |||
1208 | int diff_main(int argc, char **argv); | ||
1209 | int diff_main(int argc, char **argv) | ||
1210 | { | ||
1211 | bool gotstdin = 0; | ||
1212 | char *U_opt; | ||
1213 | char *f1, *f2; | ||
1214 | llist_t *L_arg = NULL; | ||
1215 | |||
1216 | INIT_G(); | ||
1217 | |||
1218 | /* exactly 2 params; collect multiple -L <label> */ | ||
1219 | opt_complementary = "=2:L::"; | ||
1220 | getopt32(argc, argv, "abdiL:NqrsS:tTU:wu" | ||
1221 | "p" /* ignored (for compatibility) */, | ||
1222 | &L_arg, &start, &U_opt); | ||
1223 | /*argc -= optind;*/ | ||
1224 | argv += optind; | ||
1225 | while (L_arg) { | ||
1226 | if (label1 && label2) | ||
1227 | bb_show_usage(); | ||
1228 | if (!label1) | ||
1229 | label1 = L_arg->data; | ||
1230 | else { /* then label2 is NULL */ | ||
1231 | label2 = label1; | ||
1232 | label1 = L_arg->data; | ||
1233 | } | ||
1234 | /* we leak L_arg here... */ | ||
1235 | L_arg = L_arg->link; | ||
1236 | } | ||
1237 | if (option_mask32 & FLAG_U) | ||
1238 | context = xatoi_u(U_opt); | ||
1239 | |||
1240 | /* | ||
1241 | * Do sanity checks, fill in stb1 and stb2 and call the appropriate | ||
1242 | * driver routine. Both drivers use the contents of stb1 and stb2. | ||
1243 | */ | ||
1244 | |||
1245 | f1 = argv[0]; | ||
1246 | f2 = argv[1]; | ||
1247 | if (LONE_DASH(f1)) { | ||
1248 | fstat(STDIN_FILENO, &stb1); | ||
1249 | gotstdin = 1; | ||
1250 | } else | ||
1251 | xstat(f1, &stb1); | ||
1252 | if (LONE_DASH(f2)) { | ||
1253 | fstat(STDIN_FILENO, &stb2); | ||
1254 | gotstdin = 1; | ||
1255 | } else | ||
1256 | xstat(f2, &stb2); | ||
1257 | if (gotstdin && (S_ISDIR(stb1.st_mode) || S_ISDIR(stb2.st_mode))) | ||
1258 | bb_error_msg_and_die("can't compare - to a directory"); | ||
1259 | if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) { | ||
1260 | #if ENABLE_FEATURE_DIFF_DIR | ||
1261 | diffdir(f1, f2); | ||
1262 | #else | ||
1263 | bb_error_msg_and_die("directory comparison not supported"); | ||
1264 | #endif | ||
1265 | } else { | ||
1266 | if (S_ISDIR(stb1.st_mode)) { | ||
1267 | f1 = concat_path_file(f1, f2); | ||
1268 | xstat(f1, &stb1); | ||
1269 | } | ||
1270 | if (S_ISDIR(stb2.st_mode)) { | ||
1271 | f2 = concat_path_file(f2, f1); | ||
1272 | xstat(f2, &stb2); | ||
1273 | } | ||
1274 | /* XXX: FIXME: */ | ||
1275 | /* We can't diff e.g. stdin supplied by a pipe - we use rewind(), fseek(). | ||
1276 | * This can be fixed (volunteers?) */ | ||
1277 | if (!S_ISREG(stb1.st_mode) || !S_ISREG(stb2.st_mode)) | ||
1278 | bb_error_msg_and_die("can't diff non-seekable stream"); | ||
1279 | print_status(diffreg(f1, f2, 0), f1, f2, NULL); | ||
1280 | } | ||
1281 | return status; | ||
1282 | } | ||