commit-graph: write Bloom filters to commit graph file
[git] / list-objects-filter-options.c
1 #include "cache.h"
2 #include "commit.h"
3 #include "config.h"
4 #include "revision.h"
5 #include "argv-array.h"
6 #include "list-objects.h"
7 #include "list-objects-filter.h"
8 #include "list-objects-filter-options.h"
9 #include "promisor-remote.h"
10 #include "trace.h"
11 #include "url.h"
12
13 static int parse_combine_filter(
14         struct list_objects_filter_options *filter_options,
15         const char *arg,
16         struct strbuf *errbuf);
17
18 /*
19  * Parse value of the argument to the "filter" keyword.
20  * On the command line this looks like:
21  *       --filter=<arg>
22  * and in the pack protocol as:
23  *       "filter" SP <arg>
24  *
25  * The filter keyword will be used by many commands.
26  * See Documentation/rev-list-options.txt for allowed values for <arg>.
27  *
28  * Capture the given arg as the "filter_spec".  This can be forwarded to
29  * subordinate commands when necessary (although it's better to pass it through
30  * expand_list_objects_filter_spec() first).  We also "intern" the arg for the
31  * convenience of the current command.
32  */
33 static int gently_parse_list_objects_filter(
34         struct list_objects_filter_options *filter_options,
35         const char *arg,
36         struct strbuf *errbuf)
37 {
38         const char *v0;
39
40         if (!arg)
41                 return 0;
42
43         if (filter_options->choice)
44                 BUG("filter_options already populated");
45
46         if (!strcmp(arg, "blob:none")) {
47                 filter_options->choice = LOFC_BLOB_NONE;
48                 return 0;
49
50         } else if (skip_prefix(arg, "blob:limit=", &v0)) {
51                 if (git_parse_ulong(v0, &filter_options->blob_limit_value)) {
52                         filter_options->choice = LOFC_BLOB_LIMIT;
53                         return 0;
54                 }
55
56         } else if (skip_prefix(arg, "tree:", &v0)) {
57                 if (!git_parse_ulong(v0, &filter_options->tree_exclude_depth)) {
58                         strbuf_addstr(errbuf, _("expected 'tree:<depth>'"));
59                         return 1;
60                 }
61                 filter_options->choice = LOFC_TREE_DEPTH;
62                 return 0;
63
64         } else if (skip_prefix(arg, "sparse:oid=", &v0)) {
65                 filter_options->sparse_oid_name = xstrdup(v0);
66                 filter_options->choice = LOFC_SPARSE_OID;
67                 return 0;
68
69         } else if (skip_prefix(arg, "sparse:path=", &v0)) {
70                 if (errbuf) {
71                         strbuf_addstr(
72                                 errbuf,
73                                 _("sparse:path filters support has been dropped"));
74                 }
75                 return 1;
76
77         } else if (skip_prefix(arg, "combine:", &v0)) {
78                 return parse_combine_filter(filter_options, v0, errbuf);
79
80         }
81         /*
82          * Please update _git_fetch() in git-completion.bash when you
83          * add new filters
84          */
85
86         strbuf_addf(errbuf, _("invalid filter-spec '%s'"), arg);
87
88         memset(filter_options, 0, sizeof(*filter_options));
89         return 1;
90 }
91
92 static const char *RESERVED_NON_WS = "~`!@#$^&*()[]{}\\;'\",<>?";
93
94 static int has_reserved_character(
95         struct strbuf *sub_spec, struct strbuf *errbuf)
96 {
97         const char *c = sub_spec->buf;
98         while (*c) {
99                 if (*c <= ' ' || strchr(RESERVED_NON_WS, *c)) {
100                         strbuf_addf(
101                                 errbuf,
102                                 _("must escape char in sub-filter-spec: '%c'"),
103                                 *c);
104                         return 1;
105                 }
106                 c++;
107         }
108
109         return 0;
110 }
111
112 static int parse_combine_subfilter(
113         struct list_objects_filter_options *filter_options,
114         struct strbuf *subspec,
115         struct strbuf *errbuf)
116 {
117         size_t new_index = filter_options->sub_nr;
118         char *decoded;
119         int result;
120
121         ALLOC_GROW_BY(filter_options->sub, filter_options->sub_nr, 1,
122                       filter_options->sub_alloc);
123
124         decoded = url_percent_decode(subspec->buf);
125
126         result = has_reserved_character(subspec, errbuf) ||
127                 gently_parse_list_objects_filter(
128                         &filter_options->sub[new_index], decoded, errbuf);
129
130         free(decoded);
131         return result;
132 }
133
134 static int parse_combine_filter(
135         struct list_objects_filter_options *filter_options,
136         const char *arg,
137         struct strbuf *errbuf)
138 {
139         struct strbuf **subspecs = strbuf_split_str(arg, '+', 0);
140         size_t sub;
141         int result = 0;
142
143         if (!subspecs[0]) {
144                 strbuf_addstr(errbuf, _("expected something after combine:"));
145                 result = 1;
146                 goto cleanup;
147         }
148
149         for (sub = 0; subspecs[sub] && !result; sub++) {
150                 if (subspecs[sub + 1]) {
151                         /*
152                          * This is not the last subspec. Remove trailing "+" so
153                          * we can parse it.
154                          */
155                         size_t last = subspecs[sub]->len - 1;
156                         assert(subspecs[sub]->buf[last] == '+');
157                         strbuf_remove(subspecs[sub], last, 1);
158                 }
159                 result = parse_combine_subfilter(
160                         filter_options, subspecs[sub], errbuf);
161         }
162
163         filter_options->choice = LOFC_COMBINE;
164
165 cleanup:
166         strbuf_list_free(subspecs);
167         if (result) {
168                 list_objects_filter_release(filter_options);
169                 memset(filter_options, 0, sizeof(*filter_options));
170         }
171         return result;
172 }
173
174 static int allow_unencoded(char ch)
175 {
176         if (ch <= ' ' || ch == '%' || ch == '+')
177                 return 0;
178         return !strchr(RESERVED_NON_WS, ch);
179 }
180
181 static void filter_spec_append_urlencode(
182         struct list_objects_filter_options *filter, const char *raw)
183 {
184         struct strbuf buf = STRBUF_INIT;
185         strbuf_addstr_urlencode(&buf, raw, allow_unencoded);
186         trace_printf("Add to combine filter-spec: %s\n", buf.buf);
187         string_list_append(&filter->filter_spec, strbuf_detach(&buf, NULL));
188 }
189
190 /*
191  * Changes filter_options into an equivalent LOFC_COMBINE filter options
192  * instance. Does not do anything if filter_options is already LOFC_COMBINE.
193  */
194 static void transform_to_combine_type(
195         struct list_objects_filter_options *filter_options)
196 {
197         assert(filter_options->choice);
198         if (filter_options->choice == LOFC_COMBINE)
199                 return;
200         {
201                 const int initial_sub_alloc = 2;
202                 struct list_objects_filter_options *sub_array =
203                         xcalloc(initial_sub_alloc, sizeof(*sub_array));
204                 sub_array[0] = *filter_options;
205                 memset(filter_options, 0, sizeof(*filter_options));
206                 filter_options->sub = sub_array;
207                 filter_options->sub_alloc = initial_sub_alloc;
208         }
209         filter_options->sub_nr = 1;
210         filter_options->choice = LOFC_COMBINE;
211         string_list_append(&filter_options->filter_spec, xstrdup("combine:"));
212         filter_spec_append_urlencode(
213                 filter_options,
214                 list_objects_filter_spec(&filter_options->sub[0]));
215         /*
216          * We don't need the filter_spec strings for subfilter specs, only the
217          * top level.
218          */
219         string_list_clear(&filter_options->sub[0].filter_spec, /*free_util=*/0);
220 }
221
222 void list_objects_filter_die_if_populated(
223         struct list_objects_filter_options *filter_options)
224 {
225         if (filter_options->choice)
226                 die(_("multiple filter-specs cannot be combined"));
227 }
228
229 void parse_list_objects_filter(
230         struct list_objects_filter_options *filter_options,
231         const char *arg)
232 {
233         struct strbuf errbuf = STRBUF_INIT;
234         int parse_error;
235
236         if (!filter_options->choice) {
237                 string_list_append(&filter_options->filter_spec, xstrdup(arg));
238
239                 parse_error = gently_parse_list_objects_filter(
240                         filter_options, arg, &errbuf);
241         } else {
242                 /*
243                  * Make filter_options an LOFC_COMBINE spec so we can trivially
244                  * add subspecs to it.
245                  */
246                 transform_to_combine_type(filter_options);
247
248                 string_list_append(&filter_options->filter_spec, xstrdup("+"));
249                 filter_spec_append_urlencode(filter_options, arg);
250                 ALLOC_GROW_BY(filter_options->sub, filter_options->sub_nr, 1,
251                               filter_options->sub_alloc);
252
253                 parse_error = gently_parse_list_objects_filter(
254                         &filter_options->sub[filter_options->sub_nr - 1], arg,
255                         &errbuf);
256         }
257         if (parse_error)
258                 die("%s", errbuf.buf);
259 }
260
261 int opt_parse_list_objects_filter(const struct option *opt,
262                                   const char *arg, int unset)
263 {
264         struct list_objects_filter_options *filter_options = opt->value;
265
266         if (unset || !arg)
267                 list_objects_filter_set_no_filter(filter_options);
268         else
269                 parse_list_objects_filter(filter_options, arg);
270         return 0;
271 }
272
273 const char *list_objects_filter_spec(struct list_objects_filter_options *filter)
274 {
275         if (!filter->filter_spec.nr)
276                 BUG("no filter_spec available for this filter");
277         if (filter->filter_spec.nr != 1) {
278                 struct strbuf concatted = STRBUF_INIT;
279                 strbuf_add_separated_string_list(
280                         &concatted, "", &filter->filter_spec);
281                 string_list_clear(&filter->filter_spec, /*free_util=*/0);
282                 string_list_append(
283                         &filter->filter_spec, strbuf_detach(&concatted, NULL));
284         }
285
286         return filter->filter_spec.items[0].string;
287 }
288
289 const char *expand_list_objects_filter_spec(
290         struct list_objects_filter_options *filter)
291 {
292         if (filter->choice == LOFC_BLOB_LIMIT) {
293                 struct strbuf expanded_spec = STRBUF_INIT;
294                 strbuf_addf(&expanded_spec, "blob:limit=%lu",
295                             filter->blob_limit_value);
296                 string_list_clear(&filter->filter_spec, /*free_util=*/0);
297                 string_list_append(
298                         &filter->filter_spec,
299                         strbuf_detach(&expanded_spec, NULL));
300         }
301
302         return list_objects_filter_spec(filter);
303 }
304
305 void list_objects_filter_release(
306         struct list_objects_filter_options *filter_options)
307 {
308         size_t sub;
309
310         if (!filter_options)
311                 return;
312         string_list_clear(&filter_options->filter_spec, /*free_util=*/0);
313         free(filter_options->sparse_oid_name);
314         for (sub = 0; sub < filter_options->sub_nr; sub++)
315                 list_objects_filter_release(&filter_options->sub[sub]);
316         free(filter_options->sub);
317         memset(filter_options, 0, sizeof(*filter_options));
318 }
319
320 void partial_clone_register(
321         const char *remote,
322         struct list_objects_filter_options *filter_options)
323 {
324         char *cfg_name;
325         char *filter_name;
326
327         /* Check if it is already registered */
328         if (!promisor_remote_find(remote)) {
329                 git_config_set("core.repositoryformatversion", "1");
330
331                 /* Add promisor config for the remote */
332                 cfg_name = xstrfmt("remote.%s.promisor", remote);
333                 git_config_set(cfg_name, "true");
334                 free(cfg_name);
335         }
336
337         /*
338          * Record the initial filter-spec in the config as
339          * the default for subsequent fetches from this remote.
340          */
341         filter_name = xstrfmt("remote.%s.partialclonefilter", remote);
342         /* NEEDSWORK: 'expand' result leaking??? */
343         git_config_set(filter_name,
344                        expand_list_objects_filter_spec(filter_options));
345         free(filter_name);
346
347         /* Make sure the config info are reset */
348         promisor_remote_reinit();
349 }
350
351 void partial_clone_get_default_filter_spec(
352         struct list_objects_filter_options *filter_options,
353         const char *remote)
354 {
355         struct promisor_remote *promisor = promisor_remote_find(remote);
356         struct strbuf errbuf = STRBUF_INIT;
357
358         /*
359          * Parse default value, but silently ignore it if it is invalid.
360          */
361         if (!promisor)
362                 return;
363
364         string_list_append(&filter_options->filter_spec,
365                            promisor->partial_clone_filter);
366         gently_parse_list_objects_filter(filter_options,
367                                          promisor->partial_clone_filter,
368                                          &errbuf);
369         strbuf_release(&errbuf);
370 }