freshlybakedca.ke
Git fork
1/*
2 * Copyright (C) 2005 Junio C Hamano
3 * Copyright (C) 2010 Google Inc.
4 */
5
6#define DISABLE_SIGN_COMPARE_WARNINGS
7
8#include "git-compat-util.h"
9#include "diff.h"
10#include "diffcore.h"
11#include "xdiff-interface.h"
12#include "kwset.h"
13#include "oidset.h"
14#include "pretty.h"
15#include "quote.h"
16
17typedef int (*pickaxe_fn)(mmfile_t *one, mmfile_t *two,
18 struct diff_options *o,
19 regex_t *regexp, kwset_t kws);
20
21struct diffgrep_cb {
22 regex_t *regexp;
23 int hit;
24};
25
26static int diffgrep_consume(void *priv, char *line, unsigned long len)
27{
28 struct diffgrep_cb *data = priv;
29 regmatch_t regmatch;
30
31 if (line[0] != '+' && line[0] != '-')
32 return 0;
33 if (data->hit)
34 BUG("Already matched in diffgrep_consume! Broken xdiff_emit_line_fn?");
35 if (!regexec_buf(data->regexp, line + 1, len - 1, 1,
36 ®match, 0)) {
37 data->hit = 1;
38 return 1;
39 }
40 return 0;
41}
42
43static int diff_grep(mmfile_t *one, mmfile_t *two,
44 struct diff_options *o,
45 regex_t *regexp, kwset_t kws UNUSED)
46{
47 struct diffgrep_cb ecbdata;
48 xpparam_t xpp;
49 xdemitconf_t xecfg;
50 int ret;
51
52 /*
53 * We have both sides; need to run textual diff and see if
54 * the pattern appears on added/deleted lines.
55 */
56 memset(&xpp, 0, sizeof(xpp));
57 memset(&xecfg, 0, sizeof(xecfg));
58 ecbdata.regexp = regexp;
59 ecbdata.hit = 0;
60 xecfg.flags = XDL_EMIT_NO_HUNK_HDR;
61 xecfg.ctxlen = o->context;
62 xecfg.interhunkctxlen = o->interhunkcontext;
63
64 /*
65 * An xdiff error might be our "data->hit" from above. See the
66 * comment for xdiff_emit_line_fn in xdiff-interface.h
67 */
68 ret = xdi_diff_outf(one, two, NULL, diffgrep_consume,
69 &ecbdata, &xpp, &xecfg);
70 if (ecbdata.hit)
71 return 1;
72 if (ret)
73 return ret;
74 return 0;
75}
76
77static unsigned int contains(mmfile_t *mf, regex_t *regexp, kwset_t kws,
78 unsigned int limit)
79{
80 unsigned int cnt = 0;
81 unsigned long sz = mf->size;
82 const char *data = mf->ptr;
83
84 if (regexp) {
85 regmatch_t regmatch;
86 int flags = 0;
87
88 while (sz &&
89 !regexec_buf(regexp, data, sz, 1, ®match, flags)) {
90 flags |= REG_NOTBOL;
91 data += regmatch.rm_eo;
92 sz -= regmatch.rm_eo;
93 if (sz && regmatch.rm_so == regmatch.rm_eo) {
94 data++;
95 sz--;
96 }
97 cnt++;
98
99 if (limit && cnt == limit)
100 return cnt;
101 }
102
103 } else { /* Classic exact string match */
104 while (sz) {
105 struct kwsmatch kwsm;
106 size_t offset = kwsexec(kws, data, sz, &kwsm);
107 if (offset == -1)
108 break;
109 sz -= offset + kwsm.size[0];
110 data += offset + kwsm.size[0];
111 cnt++;
112
113 if (limit && cnt == limit)
114 return cnt;
115 }
116 }
117 return cnt;
118}
119
120static int has_changes(mmfile_t *one, mmfile_t *two,
121 struct diff_options *o UNUSED,
122 regex_t *regexp, kwset_t kws)
123{
124 unsigned int c1 = one ? contains(one, regexp, kws, 0) : 0;
125 unsigned int c2 = two ? contains(two, regexp, kws, c1 + 1) : 0;
126 return c1 != c2;
127}
128
129static int pickaxe_match(struct diff_filepair *p, struct diff_options *o,
130 regex_t *regexp, kwset_t kws, pickaxe_fn fn)
131{
132 struct userdiff_driver *textconv_one = NULL;
133 struct userdiff_driver *textconv_two = NULL;
134 mmfile_t mf1, mf2;
135 int ret;
136
137 /* ignore unmerged */
138 if (!DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two))
139 return 0;
140
141 if (o->objfind) {
142 return (DIFF_FILE_VALID(p->one) &&
143 oidset_contains(o->objfind, &p->one->oid)) ||
144 (DIFF_FILE_VALID(p->two) &&
145 oidset_contains(o->objfind, &p->two->oid));
146 }
147
148 if (o->flags.allow_textconv) {
149 textconv_one = get_textconv(o->repo, p->one);
150 textconv_two = get_textconv(o->repo, p->two);
151 }
152
153 /*
154 * If we have an unmodified pair, we know that the count will be the
155 * same and don't even have to load the blobs. Unless textconv is in
156 * play, _and_ we are using two different textconv filters (e.g.,
157 * because a pair is an exact rename with different textconv attributes
158 * for each side, which might generate different content).
159 */
160 if (textconv_one == textconv_two && diff_unmodified_pair(p))
161 return 0;
162
163 if ((o->pickaxe_opts & DIFF_PICKAXE_KIND_G) &&
164 !o->flags.text &&
165 ((!textconv_one && diff_filespec_is_binary(o->repo, p->one)) ||
166 (!textconv_two && diff_filespec_is_binary(o->repo, p->two))))
167 return 0;
168
169 mf1.size = fill_textconv(o->repo, textconv_one, p->one, &mf1.ptr);
170 mf2.size = fill_textconv(o->repo, textconv_two, p->two, &mf2.ptr);
171
172 ret = fn(&mf1, &mf2, o, regexp, kws);
173
174 if (textconv_one)
175 free(mf1.ptr);
176 if (textconv_two)
177 free(mf2.ptr);
178 diff_free_filespec_data(p->one);
179 diff_free_filespec_data(p->two);
180
181 return ret;
182}
183
184static void pickaxe(struct diff_queue_struct *q, struct diff_options *o,
185 regex_t *regexp, kwset_t kws, pickaxe_fn fn)
186{
187 int i;
188 struct diff_queue_struct outq = DIFF_QUEUE_INIT;
189
190 if (o->pickaxe_opts & DIFF_PICKAXE_ALL) {
191 /* Showing the whole changeset if needle exists */
192 for (i = 0; i < q->nr; i++) {
193 struct diff_filepair *p = q->queue[i];
194 if (pickaxe_match(p, o, regexp, kws, fn))
195 return; /* do not munge the queue */
196 }
197
198 /*
199 * Otherwise we will clear the whole queue by copying
200 * the empty outq at the end of this function, but
201 * first clear the current entries in the queue.
202 */
203 for (i = 0; i < q->nr; i++)
204 diff_free_filepair(q->queue[i]);
205 } else {
206 /* Showing only the filepairs that has the needle */
207 for (i = 0; i < q->nr; i++) {
208 struct diff_filepair *p = q->queue[i];
209 if (pickaxe_match(p, o, regexp, kws, fn))
210 diff_q(&outq, p);
211 else
212 diff_free_filepair(p);
213 }
214 }
215
216 free(q->queue);
217 *q = outq;
218}
219
220static void regcomp_or_die(regex_t *regex, const char *needle, int cflags)
221{
222 int err = regcomp(regex, needle, cflags);
223 if (err) {
224 /* The POSIX.2 people are surely sick */
225 char errbuf[1024];
226 regerror(err, regex, errbuf, 1024);
227 die("invalid regex: %s", errbuf);
228 }
229}
230
231void diffcore_pickaxe(struct diff_options *o)
232{
233 const char *needle = o->pickaxe;
234 int opts = o->pickaxe_opts;
235 regex_t regex, *regexp = NULL;
236 kwset_t kws = NULL;
237 pickaxe_fn fn;
238
239 if (opts & ~DIFF_PICKAXE_KIND_OBJFIND &&
240 (!needle || !*needle))
241 BUG("should have needle under -G or -S");
242 if (opts & (DIFF_PICKAXE_REGEX | DIFF_PICKAXE_KIND_G)) {
243 int cflags = REG_EXTENDED | REG_NEWLINE;
244 if (o->pickaxe_opts & DIFF_PICKAXE_IGNORE_CASE)
245 cflags |= REG_ICASE;
246 regcomp_or_die(®ex, needle, cflags);
247 regexp = ®ex;
248
249 if (opts & DIFF_PICKAXE_KIND_G)
250 fn = diff_grep;
251 else if (opts & DIFF_PICKAXE_REGEX)
252 fn = has_changes;
253 else
254 /*
255 * We don't need to check the combination of
256 * -G and --pickaxe-regex, by the time we get
257 * here diff.c has already died if they're
258 * combined. See the usage tests in
259 * t4209-log-pickaxe.sh.
260 */
261 BUG("unreachable");
262 } else if (opts & DIFF_PICKAXE_KIND_S) {
263 if (o->pickaxe_opts & DIFF_PICKAXE_IGNORE_CASE &&
264 has_non_ascii(needle)) {
265 struct strbuf sb = STRBUF_INIT;
266 int cflags = REG_NEWLINE | REG_ICASE;
267
268 basic_regex_quote_buf(&sb, needle);
269 regcomp_or_die(®ex, sb.buf, cflags);
270 strbuf_release(&sb);
271 regexp = ®ex;
272 } else {
273 kws = kwsalloc(o->pickaxe_opts & DIFF_PICKAXE_IGNORE_CASE
274 ? tolower_trans_tbl : NULL);
275 kwsincr(kws, needle, strlen(needle));
276 kwsprep(kws);
277 }
278 fn = has_changes;
279 } else if (opts & DIFF_PICKAXE_KIND_OBJFIND) {
280 fn = NULL;
281 } else {
282 BUG("unknown pickaxe_opts flag");
283 }
284
285 pickaxe(&diff_queued_diff, o, regexp, kws, fn);
286
287 if (regexp)
288 regfree(regexp);
289 if (kws)
290 kwsfree(kws);
291 return;
292}