]>
Commit | Line | Data |
---|---|---|
52e95789 JH |
1 | /* |
2 | * Copyright (C) 2005 Junio C Hamano | |
f506b8e8 | 3 | * Copyright (C) 2010 Google Inc. |
52e95789 JH |
4 | */ |
5 | #include "cache.h" | |
6 | #include "diff.h" | |
7 | #include "diffcore.h" | |
f506b8e8 | 8 | #include "xdiff-interface.h" |
b95c5ada | 9 | #include "kwset.h" |
f506b8e8 | 10 | |
61690bf4 JK |
11 | typedef int (*pickaxe_fn)(mmfile_t *one, mmfile_t *two, |
12 | struct diff_options *o, | |
13 | regex_t *regexp, kwset_t kws); | |
14 | ||
15 | static int pickaxe_match(struct diff_filepair *p, struct diff_options *o, | |
16 | regex_t *regexp, kwset_t kws, pickaxe_fn fn); | |
8a94151d RS |
17 | |
18 | static void pickaxe(struct diff_queue_struct *q, struct diff_options *o, | |
19 | regex_t *regexp, kwset_t kws, pickaxe_fn fn) | |
20 | { | |
21 | int i; | |
22 | struct diff_queue_struct outq; | |
23 | ||
24 | DIFF_QUEUE_CLEAR(&outq); | |
25 | ||
26 | if (o->pickaxe_opts & DIFF_PICKAXE_ALL) { | |
27 | /* Showing the whole changeset if needle exists */ | |
28 | for (i = 0; i < q->nr; i++) { | |
29 | struct diff_filepair *p = q->queue[i]; | |
61690bf4 | 30 | if (pickaxe_match(p, o, regexp, kws, fn)) |
8a94151d RS |
31 | return; /* do not munge the queue */ |
32 | } | |
33 | ||
34 | /* | |
35 | * Otherwise we will clear the whole queue by copying | |
36 | * the empty outq at the end of this function, but | |
37 | * first clear the current entries in the queue. | |
38 | */ | |
39 | for (i = 0; i < q->nr; i++) | |
40 | diff_free_filepair(q->queue[i]); | |
41 | } else { | |
42 | /* Showing only the filepairs that has the needle */ | |
43 | for (i = 0; i < q->nr; i++) { | |
44 | struct diff_filepair *p = q->queue[i]; | |
61690bf4 | 45 | if (pickaxe_match(p, o, regexp, kws, fn)) |
8a94151d RS |
46 | diff_q(&outq, p); |
47 | else | |
48 | diff_free_filepair(p); | |
49 | } | |
50 | } | |
51 | ||
52 | free(q->queue); | |
53 | *q = outq; | |
54 | } | |
55 | ||
f506b8e8 JH |
56 | struct diffgrep_cb { |
57 | regex_t *regexp; | |
58 | int hit; | |
59 | }; | |
60 | ||
61 | static void diffgrep_consume(void *priv, char *line, unsigned long len) | |
62 | { | |
63 | struct diffgrep_cb *data = priv; | |
64 | regmatch_t regmatch; | |
65 | int hold; | |
66 | ||
67 | if (line[0] != '+' && line[0] != '-') | |
68 | return; | |
69 | if (data->hit) | |
70 | /* | |
71 | * NEEDSWORK: we should have a way to terminate the | |
72 | * caller early. | |
73 | */ | |
74 | return; | |
75 | /* Yuck -- line ought to be "const char *"! */ | |
76 | hold = line[len]; | |
77 | line[len] = '\0'; | |
78 | data->hit = !regexec(data->regexp, line + 1, 1, ®match, 0); | |
79 | line[len] = hold; | |
80 | } | |
81 | ||
61690bf4 JK |
82 | static int diff_grep(mmfile_t *one, mmfile_t *two, |
83 | struct diff_options *o, | |
db99cb70 | 84 | regex_t *regexp, kwset_t kws) |
f506b8e8 JH |
85 | { |
86 | regmatch_t regmatch; | |
61690bf4 JK |
87 | struct diffgrep_cb ecbdata; |
88 | xpparam_t xpp; | |
89 | xdemitconf_t xecfg; | |
bc615898 | 90 | |
61690bf4 JK |
91 | if (!one) |
92 | return !regexec(regexp, two->ptr, 1, ®match, 0); | |
93 | if (!two) | |
94 | return !regexec(regexp, one->ptr, 1, ®match, 0); | |
f506b8e8 | 95 | |
61690bf4 JK |
96 | /* |
97 | * We have both sides; need to run textual diff and see if | |
98 | * the pattern appears on added/deleted lines. | |
99 | */ | |
100 | memset(&xpp, 0, sizeof(xpp)); | |
101 | memset(&xecfg, 0, sizeof(xecfg)); | |
102 | ecbdata.regexp = regexp; | |
103 | ecbdata.hit = 0; | |
104 | xecfg.ctxlen = o->context; | |
105 | xecfg.interhunkctxlen = o->interhunkcontext; | |
106 | xdi_diff_outf(one, two, diffgrep_consume, &ecbdata, | |
107 | &xpp, &xecfg); | |
108 | return ecbdata.hit; | |
f506b8e8 JH |
109 | } |
110 | ||
111 | static void diffcore_pickaxe_grep(struct diff_options *o) | |
112 | { | |
8a94151d | 113 | int err; |
f506b8e8 | 114 | regex_t regex; |
accccde4 | 115 | int cflags = REG_EXTENDED | REG_NEWLINE; |
f506b8e8 | 116 | |
accccde4 JH |
117 | if (DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE)) |
118 | cflags |= REG_ICASE; | |
119 | ||
120 | err = regcomp(®ex, o->pickaxe, cflags); | |
f506b8e8 JH |
121 | if (err) { |
122 | char errbuf[1024]; | |
123 | regerror(err, ®ex, errbuf, 1024); | |
124 | regfree(®ex); | |
276b22d3 | 125 | die("invalid regex: %s", errbuf); |
f506b8e8 JH |
126 | } |
127 | ||
8a94151d | 128 | pickaxe(&diff_queued_diff, o, ®ex, NULL, diff_grep); |
f506b8e8 | 129 | |
2b5f07f1 | 130 | regfree(®ex); |
f506b8e8 JH |
131 | return; |
132 | } | |
52e95789 | 133 | |
3bdb5b9f | 134 | static unsigned int contains(mmfile_t *mf, regex_t *regexp, kwset_t kws) |
52e95789 | 135 | { |
2002eed6 | 136 | unsigned int cnt; |
ce163c79 | 137 | unsigned long sz; |
52e95789 | 138 | const char *data; |
2002eed6 | 139 | |
ef90ab66 JK |
140 | sz = mf->size; |
141 | data = mf->ptr; | |
2002eed6 JH |
142 | cnt = 0; |
143 | ||
d01d8c67 PB |
144 | if (regexp) { |
145 | regmatch_t regmatch; | |
146 | int flags = 0; | |
147 | ||
50fd6997 | 148 | assert(data[sz] == '\0'); |
d01d8c67 PB |
149 | while (*data && !regexec(regexp, data, 1, ®match, flags)) { |
150 | flags |= REG_NOTBOL; | |
50fd6997 RS |
151 | data += regmatch.rm_eo; |
152 | if (*data && regmatch.rm_so == regmatch.rm_eo) | |
153 | data++; | |
2002eed6 JH |
154 | cnt++; |
155 | } | |
d01d8c67 PB |
156 | |
157 | } else { /* Classic exact string match */ | |
ce163c79 | 158 | while (sz) { |
5d176fb6 RS |
159 | struct kwsmatch kwsm; |
160 | size_t offset = kwsexec(kws, data, sz, &kwsm); | |
b95c5ada FK |
161 | const char *found; |
162 | if (offset == -1) | |
ce163c79 | 163 | break; |
b95c5ada FK |
164 | else |
165 | found = data + offset; | |
5d176fb6 RS |
166 | sz -= found - data + kwsm.size[0]; |
167 | data = found + kwsm.size[0]; | |
ce163c79 | 168 | cnt++; |
d01d8c67 | 169 | } |
2002eed6 JH |
170 | } |
171 | return cnt; | |
52e95789 JH |
172 | } |
173 | ||
61690bf4 JK |
174 | static int has_changes(mmfile_t *one, mmfile_t *two, |
175 | struct diff_options *o, | |
5d176fb6 | 176 | regex_t *regexp, kwset_t kws) |
61690bf4 | 177 | { |
3bdb5b9f RS |
178 | unsigned int one_contains = one ? contains(one, regexp, kws) : 0; |
179 | unsigned int two_contains = two ? contains(two, regexp, kws) : 0; | |
180 | return one_contains != two_contains; | |
61690bf4 JK |
181 | } |
182 | ||
183 | static int pickaxe_match(struct diff_filepair *p, struct diff_options *o, | |
184 | regex_t *regexp, kwset_t kws, pickaxe_fn fn) | |
15dafaf8 | 185 | { |
bc615898 SR |
186 | struct userdiff_driver *textconv_one = NULL; |
187 | struct userdiff_driver *textconv_two = NULL; | |
ef90ab66 JK |
188 | mmfile_t mf1, mf2; |
189 | int ret; | |
190 | ||
8fa4b09f JK |
191 | if (!o->pickaxe[0]) |
192 | return 0; | |
193 | ||
61690bf4 JK |
194 | /* ignore unmerged */ |
195 | if (!DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) | |
196 | return 0; | |
197 | ||
a8f61094 SR |
198 | if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) { |
199 | textconv_one = get_textconv(p->one); | |
200 | textconv_two = get_textconv(p->two); | |
201 | } | |
bc615898 | 202 | |
ef90ab66 JK |
203 | /* |
204 | * If we have an unmodified pair, we know that the count will be the | |
205 | * same and don't even have to load the blobs. Unless textconv is in | |
206 | * play, _and_ we are using two different textconv filters (e.g., | |
207 | * because a pair is an exact rename with different textconv attributes | |
208 | * for each side, which might generate different content). | |
209 | */ | |
210 | if (textconv_one == textconv_two && diff_unmodified_pair(p)) | |
211 | return 0; | |
212 | ||
7cdb9b42 JK |
213 | mf1.size = fill_textconv(textconv_one, p->one, &mf1.ptr); |
214 | mf2.size = fill_textconv(textconv_two, p->two, &mf2.ptr); | |
ef90ab66 | 215 | |
61690bf4 JK |
216 | ret = fn(DIFF_FILE_VALID(p->one) ? &mf1 : NULL, |
217 | DIFF_FILE_VALID(p->two) ? &mf2 : NULL, | |
218 | o, regexp, kws); | |
ef90ab66 JK |
219 | |
220 | if (textconv_one) | |
221 | free(mf1.ptr); | |
222 | if (textconv_two) | |
223 | free(mf2.ptr); | |
224 | diff_free_filespec_data(p->one); | |
225 | diff_free_filespec_data(p->two); | |
226 | ||
227 | return ret; | |
15dafaf8 RS |
228 | } |
229 | ||
f506b8e8 | 230 | static void diffcore_pickaxe_count(struct diff_options *o) |
52e95789 | 231 | { |
382f013b JH |
232 | const char *needle = o->pickaxe; |
233 | int opts = o->pickaxe_opts; | |
52e95789 | 234 | unsigned long len = strlen(needle); |
d01d8c67 | 235 | regex_t regex, *regexp = NULL; |
b95c5ada | 236 | kwset_t kws = NULL; |
52e95789 | 237 | |
d01d8c67 PB |
238 | if (opts & DIFF_PICKAXE_REGEX) { |
239 | int err; | |
240 | err = regcomp(®ex, needle, REG_EXTENDED | REG_NEWLINE); | |
241 | if (err) { | |
242 | /* The POSIX.2 people are surely sick */ | |
243 | char errbuf[1024]; | |
244 | regerror(err, ®ex, errbuf, 1024); | |
245 | regfree(®ex); | |
276b22d3 | 246 | die("invalid regex: %s", errbuf); |
d01d8c67 PB |
247 | } |
248 | regexp = ®ex; | |
b95c5ada | 249 | } else { |
accccde4 JH |
250 | kws = kwsalloc(DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE) |
251 | ? tolower_trans_tbl : NULL); | |
b95c5ada FK |
252 | kwsincr(kws, needle, len); |
253 | kwsprep(kws); | |
d01d8c67 PB |
254 | } |
255 | ||
8a94151d | 256 | pickaxe(&diff_queued_diff, o, regexp, kws, has_changes); |
8e854b00 | 257 | |
95ae69b9 | 258 | if (opts & DIFF_PICKAXE_REGEX) |
d01d8c67 | 259 | regfree(®ex); |
b95c5ada FK |
260 | else |
261 | kwsfree(kws); | |
52e95789 JH |
262 | return; |
263 | } | |
f506b8e8 JH |
264 | |
265 | void diffcore_pickaxe(struct diff_options *o) | |
266 | { | |
267 | /* Might want to warn when both S and G are on; I don't care... */ | |
268 | if (o->pickaxe_opts & DIFF_PICKAXE_KIND_G) | |
8520913c | 269 | diffcore_pickaxe_grep(o); |
f506b8e8 | 270 | else |
8520913c | 271 | diffcore_pickaxe_count(o); |
f506b8e8 | 272 | } |