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