1 | // =============================================================== // |
---|
2 | // // |
---|
3 | // File : adname.cxx // |
---|
4 | // Purpose : species names // |
---|
5 | // // |
---|
6 | // Institute of Microbiology (Technical University Munich) // |
---|
7 | // http://www.arb-home.de/ // |
---|
8 | // // |
---|
9 | // =============================================================== // |
---|
10 | |
---|
11 | #include "gb_local.h" |
---|
12 | |
---|
13 | #include <ad_config.h> |
---|
14 | #include <arbdbt.h> |
---|
15 | |
---|
16 | #include <arb_progress.h> |
---|
17 | #include <arb_strbuf.h> |
---|
18 | #include <arb_strarray.h> |
---|
19 | #include <arb_file.h> |
---|
20 | #include <arb_diff.h> |
---|
21 | |
---|
22 | #include <cctype> |
---|
23 | |
---|
24 | struct gbt_renamed { |
---|
25 | int used_by; |
---|
26 | char data[1]; |
---|
27 | }; |
---|
28 | |
---|
29 | static struct { |
---|
30 | GB_HASH *renamed_hash; |
---|
31 | GB_HASH *old_species_hash; |
---|
32 | GBDATA *gb_main; |
---|
33 | GBDATA *gb_species_data; |
---|
34 | int all_flag; |
---|
35 | } NameSession; |
---|
36 | |
---|
37 | #if defined(WARN_TODO) |
---|
38 | #warning change all_flag into estimated number of renames ( == 0 shall mean all) |
---|
39 | #endif |
---|
40 | |
---|
41 | GB_ERROR GBT_begin_rename_session(GBDATA *gb_main, int all_flag) { |
---|
42 | /* Starts a rename session (to rename one or many species) |
---|
43 | * all_flag == 1 -> rename all species in DB |
---|
44 | * Call GBT_abort_rename_session() or GBT_commit_rename_session() to close the session. |
---|
45 | */ |
---|
46 | |
---|
47 | GB_ERROR error = GB_push_transaction(gb_main); |
---|
48 | if (!error) { |
---|
49 | NameSession.gb_main = gb_main; |
---|
50 | NameSession.gb_species_data = GBT_get_species_data(gb_main); |
---|
51 | |
---|
52 | if (!all_flag) { // this is meant to be used for single or few species |
---|
53 | int hash_size = 128; |
---|
54 | |
---|
55 | NameSession.renamed_hash = GBS_create_dynaval_hash(hash_size, GB_MIND_CASE, GBS_dynaval_free); |
---|
56 | NameSession.old_species_hash = 0; |
---|
57 | } |
---|
58 | else { |
---|
59 | NameSession.renamed_hash = GBS_create_dynaval_hash(GBT_get_species_count(gb_main), GB_MIND_CASE, GBS_dynaval_free); |
---|
60 | NameSession.old_species_hash = GBT_create_species_hash(gb_main); |
---|
61 | } |
---|
62 | NameSession.all_flag = all_flag; |
---|
63 | } |
---|
64 | return error; |
---|
65 | } |
---|
66 | |
---|
67 | GB_ERROR GBT_rename_species(const char *oldname, const char *newname, bool ignore_protection) |
---|
68 | { |
---|
69 | GBDATA *gb_species; |
---|
70 | GBDATA *gb_name; |
---|
71 | GB_ERROR error; |
---|
72 | |
---|
73 | if (strcmp(oldname, newname) == 0) |
---|
74 | return 0; |
---|
75 | |
---|
76 | #if defined(DEBUG) && 1 |
---|
77 | if (isdigit(oldname[0])) { |
---|
78 | printf("oldname='%s' newname='%s'\n", oldname, newname); |
---|
79 | } |
---|
80 | #endif |
---|
81 | |
---|
82 | if (NameSession.all_flag) { |
---|
83 | gb_assert(NameSession.old_species_hash); |
---|
84 | gb_species = (GBDATA *)GBS_read_hash(NameSession.old_species_hash, oldname); |
---|
85 | } |
---|
86 | else { |
---|
87 | GBDATA *gb_found_species; |
---|
88 | |
---|
89 | gb_assert(NameSession.old_species_hash == 0); |
---|
90 | gb_found_species = GBT_find_species_rel_species_data(NameSession.gb_species_data, newname); |
---|
91 | gb_species = GBT_find_species_rel_species_data(NameSession.gb_species_data, oldname); |
---|
92 | |
---|
93 | if (gb_found_species && gb_species != gb_found_species) { |
---|
94 | return GB_export_errorf("A species named '%s' already exists.", newname); |
---|
95 | } |
---|
96 | } |
---|
97 | |
---|
98 | if (!gb_species) { |
---|
99 | return GB_export_errorf("Expected that a species named '%s' exists (maybe there are duplicate species, database might be corrupt)", oldname); |
---|
100 | } |
---|
101 | |
---|
102 | gb_name = GB_entry(gb_species, "name"); |
---|
103 | if (ignore_protection) GB_push_my_security(NameSession.gb_main); |
---|
104 | error = GB_write_string(gb_name, newname); |
---|
105 | if (ignore_protection) GB_pop_my_security(NameSession.gb_main); |
---|
106 | |
---|
107 | if (!error) { |
---|
108 | if (NameSession.old_species_hash) { |
---|
109 | GBS_write_hash(NameSession.old_species_hash, oldname, 0); |
---|
110 | } |
---|
111 | gbt_renamed *rns = (gbt_renamed *)GB_calloc(strlen(newname) + sizeof (gbt_renamed), sizeof(char)); |
---|
112 | strcpy(&rns->data[0], newname); |
---|
113 | GBS_write_hash(NameSession.renamed_hash, oldname, (long)rns); |
---|
114 | } |
---|
115 | return error; |
---|
116 | } |
---|
117 | |
---|
118 | static void gbt_free_rename_session_data() { |
---|
119 | if (NameSession.renamed_hash) { |
---|
120 | GBS_free_hash(NameSession.renamed_hash); |
---|
121 | NameSession.renamed_hash = 0; |
---|
122 | } |
---|
123 | if (NameSession.old_species_hash) { |
---|
124 | GBS_free_hash(NameSession.old_species_hash); |
---|
125 | NameSession.old_species_hash = 0; |
---|
126 | } |
---|
127 | } |
---|
128 | |
---|
129 | GB_ERROR GBT_abort_rename_session() { |
---|
130 | gbt_free_rename_session_data(); |
---|
131 | return GB_abort_transaction(NameSession.gb_main); |
---|
132 | } |
---|
133 | |
---|
134 | static const char *currentTreeName = 0; |
---|
135 | |
---|
136 | static GB_ERROR gbt_rename_tree_rek(GBT_TREE *tree, int tree_index) { |
---|
137 | if (tree) { |
---|
138 | if (tree->is_leaf) { |
---|
139 | if (tree->name) { |
---|
140 | gbt_renamed *rns = (gbt_renamed *)GBS_read_hash(NameSession.renamed_hash, tree->name); |
---|
141 | if (rns) { |
---|
142 | char *newname; |
---|
143 | if (rns->used_by == tree_index) { // species more than once in the tree |
---|
144 | static int counter = 0; |
---|
145 | char buffer[256]; |
---|
146 | |
---|
147 | sprintf(buffer, "%s_%i", rns->data, counter++); |
---|
148 | GB_warningf("Species '%s' more than once in '%s', creating zombie '%s'", |
---|
149 | tree->name, currentTreeName, buffer); |
---|
150 | newname = buffer; |
---|
151 | } |
---|
152 | else { |
---|
153 | newname = &rns->data[0]; |
---|
154 | } |
---|
155 | freedup(tree->name, newname); |
---|
156 | rns->used_by = tree_index; |
---|
157 | } |
---|
158 | } |
---|
159 | } |
---|
160 | else { |
---|
161 | gbt_rename_tree_rek(tree->leftson, tree_index); |
---|
162 | gbt_rename_tree_rek(tree->rightson, tree_index); |
---|
163 | } |
---|
164 | } |
---|
165 | return NULL; |
---|
166 | } |
---|
167 | |
---|
168 | GB_ERROR GBT_commit_rename_session() { // goes to header: __ATTR__USERESULT |
---|
169 | arb_progress commit_progress("Renaming name references", 3); |
---|
170 | commit_progress.allow_title_reuse(); |
---|
171 | |
---|
172 | GB_ERROR error = 0; |
---|
173 | |
---|
174 | // rename species in trees |
---|
175 | { |
---|
176 | ConstStrArray tree_names; |
---|
177 | GBT_get_tree_names(tree_names, NameSession.gb_main, false); |
---|
178 | |
---|
179 | if (!tree_names.empty()) { |
---|
180 | int tree_count = tree_names.size(); |
---|
181 | arb_progress progress(GBS_global_string("Renaming species in %i tree%c", tree_count, "s"[tree_count<2]), |
---|
182 | tree_count*3); |
---|
183 | |
---|
184 | for (int count = 0; count<tree_count && !error; ++count) { |
---|
185 | const char *tname = tree_names[count]; |
---|
186 | GBT_TREE *tree = GBT_read_tree(NameSession.gb_main, tname, GBT_TREE_NodeFactory()); |
---|
187 | ++progress; |
---|
188 | |
---|
189 | if (tree) { |
---|
190 | currentTreeName = tname; // provide tree name (used for error message) |
---|
191 | gbt_rename_tree_rek(tree, count+1); |
---|
192 | currentTreeName = 0; |
---|
193 | |
---|
194 | ++progress; |
---|
195 | |
---|
196 | GBT_write_tree(NameSession.gb_main, tname, tree); |
---|
197 | delete tree; |
---|
198 | |
---|
199 | progress.inc_and_check_user_abort(error); |
---|
200 | } |
---|
201 | else { |
---|
202 | GBT_message(NameSession.gb_main, GBS_global_string("Warning: failed to read '%s' (Reason: %s)", tname, GB_await_error())); |
---|
203 | ++progress; |
---|
204 | ++progress; |
---|
205 | } |
---|
206 | } |
---|
207 | } |
---|
208 | commit_progress.inc_and_check_user_abort(error); |
---|
209 | } |
---|
210 | // rename configurations |
---|
211 | if (!error) { |
---|
212 | ConstStrArray config_names; |
---|
213 | GBT_get_configuration_names(config_names, NameSession.gb_main); |
---|
214 | |
---|
215 | if (!config_names.empty()) { |
---|
216 | int config_count = config_names.size(); |
---|
217 | arb_progress progress(GBS_global_string("Renaming species in %i config%c", config_count, "s"[config_count<2]), config_count); |
---|
218 | |
---|
219 | for (int count = 0; !error && count<config_count; ++count) { |
---|
220 | GBT_config *config = GBT_load_configuration_data(NameSession.gb_main, config_names[count], &error); |
---|
221 | if (!error) { |
---|
222 | int need_save = 0; |
---|
223 | int mode; |
---|
224 | |
---|
225 | for (mode = 0; !error && mode<2; ++mode) { |
---|
226 | char **configStrPtr = (mode == 0 ? &config->top_area : &config->middle_area); |
---|
227 | GBT_config_parser *parser = GBT_start_config_parser(*configStrPtr); |
---|
228 | GBT_config_item *item = GBT_create_config_item(); |
---|
229 | GBS_strstruct *strstruct = GBS_stropen(1000); |
---|
230 | |
---|
231 | error = GBT_parse_next_config_item(parser, item); |
---|
232 | while (!error && item->type != CI_END_OF_CONFIG) { |
---|
233 | if (item->type == CI_SPECIES) { |
---|
234 | gbt_renamed *rns = (gbt_renamed *)GBS_read_hash(NameSession.renamed_hash, item->name); |
---|
235 | if (rns) { // species was renamed |
---|
236 | freedup(item->name, rns->data); |
---|
237 | need_save = 1; |
---|
238 | } |
---|
239 | } |
---|
240 | GBT_append_to_config_string(item, strstruct); |
---|
241 | error = GBT_parse_next_config_item(parser, item); |
---|
242 | } |
---|
243 | |
---|
244 | if (!error) freeset(*configStrPtr, GBS_strclose(strstruct)); |
---|
245 | else { |
---|
246 | error = GBS_global_string("Failed to parse configuration '%s' (Reason: %s)", config_names[count], error); |
---|
247 | } |
---|
248 | |
---|
249 | GBT_free_config_item(item); |
---|
250 | GBT_free_config_parser(parser); |
---|
251 | } |
---|
252 | |
---|
253 | if (!error && need_save) { |
---|
254 | error = GBT_save_configuration_data(config, NameSession.gb_main, config_names[count]); |
---|
255 | } |
---|
256 | GBT_free_configuration_data(config); |
---|
257 | } |
---|
258 | progress.inc_and_check_user_abort(error); |
---|
259 | } |
---|
260 | } |
---|
261 | } |
---|
262 | commit_progress.inc_and_check_user_abort(error); |
---|
263 | |
---|
264 | // rename links in pseudo-species |
---|
265 | if (!error && GEN_is_genome_db(NameSession.gb_main, -1)) { |
---|
266 | GBDATA *gb_pseudo; |
---|
267 | for (gb_pseudo = GEN_first_pseudo_species(NameSession.gb_main); |
---|
268 | gb_pseudo && !error; |
---|
269 | gb_pseudo = GEN_next_pseudo_species(gb_pseudo)) |
---|
270 | { |
---|
271 | GBDATA *gb_origin_organism = GB_entry(gb_pseudo, "ARB_origin_species"); |
---|
272 | if (gb_origin_organism) { |
---|
273 | const char *origin = GB_read_char_pntr(gb_origin_organism); |
---|
274 | gbt_renamed *rns = (gbt_renamed *)GBS_read_hash(NameSession.renamed_hash, origin); |
---|
275 | if (rns) { // species was renamed |
---|
276 | const char *newname = &rns->data[0]; |
---|
277 | error = GB_write_string(gb_origin_organism, newname); |
---|
278 | } |
---|
279 | } |
---|
280 | } |
---|
281 | } |
---|
282 | commit_progress.inc_and_check_user_abort(error); |
---|
283 | |
---|
284 | gbt_free_rename_session_data(); |
---|
285 | |
---|
286 | error = GB_end_transaction(NameSession.gb_main, error); |
---|
287 | return error; |
---|
288 | } |
---|
289 | |
---|
290 | // -------------------------------------------------------------------------------- |
---|
291 | |
---|
292 | #ifdef UNIT_TESTS |
---|
293 | #ifndef TEST_UNIT_H |
---|
294 | #include <test_unit.h> |
---|
295 | #endif |
---|
296 | |
---|
297 | // #define TEST_AUTO_UPDATE // uncomment to auto-update test result db |
---|
298 | |
---|
299 | void TEST_SLOW_rename_session() { |
---|
300 | const char *inputname = "TEST_opti_ascii_in.arb"; |
---|
301 | const char *outputname = "TEST_opti_ascii_renamed.arb"; |
---|
302 | const char *expectedname = "TEST_opti_ascii_renamed_expected.arb"; |
---|
303 | |
---|
304 | { |
---|
305 | GB_shell shell; |
---|
306 | GBDATA *gb_main; |
---|
307 | TEST_EXPECT_RESULT__NOERROREXPORTED(gb_main = GB_open(inputname, "rw")); |
---|
308 | |
---|
309 | for (int session = 1; session <= 2; ++session) { |
---|
310 | TEST_ANNOTATE(GBS_global_string("session=%i", session)); |
---|
311 | |
---|
312 | TEST_EXPECT_NO_ERROR(GBT_begin_rename_session(gb_main, 0)); |
---|
313 | if (session == 2) { // session 1 tests renaming nothing |
---|
314 | // only in config 'some': |
---|
315 | TEST_EXPECT_NO_ERROR(GBT_rename_species("FrnPhilo", "olihPnrF", true)); |
---|
316 | TEST_EXPECT_NO_ERROR(GBT_rename_species("DsfDesul", "luseDfsD", true)); |
---|
317 | // also in config 'other': |
---|
318 | TEST_EXPECT_NO_ERROR(GBT_rename_species("CalSacch", "hccaSlaC", true)); |
---|
319 | TEST_EXPECT_NO_ERROR(GBT_rename_species("LacReute", "etueRcaL", true)); |
---|
320 | } |
---|
321 | TEST_EXPECT_NO_ERROR(GBT_commit_rename_session()); |
---|
322 | } |
---|
323 | |
---|
324 | TEST_EXPECT_NO_ERROR(GB_save_as(gb_main, outputname, "a")); |
---|
325 | GB_close(gb_main); |
---|
326 | } |
---|
327 | |
---|
328 | #if defined(TEST_AUTO_UPDATE) |
---|
329 | TEST_COPY_FILE(outputname, expectedname); |
---|
330 | #endif |
---|
331 | TEST_EXPECT_TEXTFILE_DIFFLINES(outputname, expectedname, 0); |
---|
332 | TEST_EXPECT_ZERO_OR_SHOW_ERRNO(GB_unlink(outputname)); |
---|
333 | } |
---|
334 | TEST_PUBLISH(TEST_SLOW_rename_session); |
---|
335 | |
---|
336 | #endif // UNIT_TESTS |
---|
337 | |
---|
338 | // -------------------------------------------------------------------------------- |
---|