Context Navigation

source: branches/profile/NTREE/NT_dbrepair.cxx

Visit:

Last change on this file was 10271, checked in by westram, 11 years ago
fix #335 activate test (written+working in [7813]; now succeeding again)
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 40.9 KB

Line
1	// =============================================================== //
2	// //
3	// File : NT_dbrepair.cxx //
4	// Purpose : repair database bugs //
5	// //
6	// Coded by Ralf Westram (coder@reallysoft.de) in May 2008 //
7	// Institute of Microbiology (Technical University Munich) //
8	// http://www.arb-home.de/ //
9	// //
10	// =============================================================== //
11
12	#include "NT_local.h"
13
14	#include <arbdbt.h>
15	#include <adGene.h>
16
17	#include <items.h>
18	#include <GEN.hxx>
19	#include <EXP.hxx>
20	#include <aw_color_groups.hxx>
21	#include <aw_msg.hxx>
22	#include <arb_progress.h>
23	#include <aw_question.hxx>
24
25	#include <arb_str.h>
26	#include <arb_strarray.h>
27
28	#include <map>
29	#include <set>
30	#include <string>
31	#include <vector>
32
33	using namespace std;
34
35	#if defined(WARN_TODO)
36	#warning the whole fix mechanism should be part of some lower-level-library
37	// meanwhile DB checks are only performed by ARB_NTREE
38	// ItemSelector should go to same library as this module
39	#endif
40
41	// --------------------------------------------------------------------------------
42	// CheckedConsistencies provides an easy way to automatically correct flues in the database
43	// by calling a check routine exactly once.
44	//
45	// For an example see nt_check_database_consistency()
46	//
47	// Note: this makes problems if DB is loaded with older ARB version and some already
48	// fixed flues a put into DB again.
49	// see http://bugs.arb-home.de/ticket/143
50
51	typedef GB_ERROR (item_check_fun)(GBDATA gb_item, ItemSelector& sel);
52
53	typedef map<string, item_check_fun> item_check_map;
54	typedef item_check_map::const_iterator item_check_iter;
55
56	class CheckedConsistencies : virtual Noncopyable {
57	GBDATA *gb_main;
58	size_t species_count;
59	size_t sai_count;
60	set<string> consistencies;
61	item_check_map item_checks;
62
63	GB_ERROR perform_selected_item_checks(ItemSelector& sel);
64
65	public:
66
67	CheckedConsistencies(GBDATA *gb_main_) : gb_main(gb_main_) {
68	GB_transaction ta(gb_main);
69	GBDATA *gb_checks = GB_search(gb_main, "checks", GB_CREATE_CONTAINER);
70
71	for (GBDATA *gb_check = GB_entry(gb_checks, "check"); gb_check; gb_check = GB_nextEntry(gb_check)) {
72	consistencies.insert(GB_read_char_pntr(gb_check));
73	}
74
75	species_count = GBT_get_species_count(gb_main);
76	sai_count = GBT_get_SAI_count(gb_main);
77	}
78
79	bool was_performed(const string& check_name) const {
80	return consistencies.find(check_name) != consistencies.end();
81	}
82
83	GB_ERROR register_as_performed(const string& check_name) {
84	GB_ERROR error = 0;
85	if (was_performed(check_name)) {
86	printf("check '%s' already has been registered before. Duplicated check name?\n", check_name.c_str());
87	}
88	else {
89	GB_transaction ta(gb_main);
90
91	GBDATA *gb_checks = GB_search(gb_main, "checks", GB_CREATE_CONTAINER);
92	GBDATA *gb_check = GB_create(gb_checks, "check", GB_STRING);
93
94	if (!gb_check) error = GB_await_error();
95	else error = GB_write_string(gb_check, check_name.c_str());
96
97	if (!error) consistencies.insert(check_name);
98	}
99	return error;
100	}
101
102	void perform_check(const string& check_name,
103	GB_ERROR (do_check)(GBDATA gb_main, size_t species, size_t sais),
104	GB_ERROR& error)
105	{
106	if (!error && !was_performed(check_name)) {
107	arb_progress progress(check_name.c_str());
108	error = do_check(gb_main, species_count, sai_count);
109	if (!error) register_as_performed(check_name);
110	}
111	}
112
113	void register_item_check(const string& check_name, item_check_fun item_check) {
114	if (!was_performed(check_name)) {
115	item_checks[check_name] = item_check;
116	}
117	}
118
119	void perform_item_checks(GB_ERROR& error);
120
121	GB_ERROR forgetDoneChecks() {
122	GB_ERROR error = 0;
123	GB_transaction ta(gb_main);
124
125	GBDATA *gb_checks = GB_search(gb_main, "checks", GB_CREATE_CONTAINER);
126	for (GBDATA *gb_check = GB_entry(gb_checks, "check"); gb_check && !error; gb_check = GB_nextEntry(gb_check)) {
127	char *check_name = GB_read_string(gb_check);
128
129	#if defined(DEBUG)
130	printf("Deleting check '%s'\n", check_name);
131	#endif // DEBUG
132	error = GB_delete(gb_check);
133	consistencies.erase(check_name);
134	free(check_name);
135	}
136	return error;
137	}
138	};
139
140	GB_ERROR CheckedConsistencies::perform_selected_item_checks(ItemSelector& sel) {
141	GB_ERROR error = NULL;
142	item_check_iter end = item_checks.end();
143
144	for (GBDATA *gb_cont = sel.get_first_item_container(gb_main, NULL, QUERY_ALL_ITEMS);
145	gb_cont && !error;
146	gb_cont = sel.get_next_item_container(gb_cont, QUERY_ALL_ITEMS))
147	{
148	for (GBDATA *gb_item = sel.get_first_item(gb_cont, QUERY_ALL_ITEMS);
149	gb_item && !error;
150	gb_item = sel.get_next_item(gb_item, QUERY_ALL_ITEMS))
151	{
152	for (item_check_iter chk = item_checks.begin(); chk != end && !error; ++chk) {
153	error = chk->second(gb_item, sel);
154	}
155	}
156	}
157
158	return error;
159	}
160
161	void CheckedConsistencies::perform_item_checks(GB_ERROR& error) {
162	if (!item_checks.empty()) {
163	if (!error) {
164	GB_transaction ta(gb_main);
165	bool is_genome_db = GEN_is_genome_db(gb_main, -1);
166
167	error = perform_selected_item_checks(SPECIES_get_selector());
168	if (!error && is_genome_db) {
169	error = perform_selected_item_checks(GEN_get_selector());
170	if (!error) error = perform_selected_item_checks(EXP_get_selector());
171	}
172
173	error = ta.close(error);
174	}
175
176	if (!error) {
177	item_check_iter end = item_checks.end();
178	for (item_check_iter chk = item_checks.begin(); chk != end && !error; ++chk) {
179	error = register_as_performed(chk->first);
180	}
181
182	if (!error) item_checks.clear();
183	}
184	}
185	}
186
187	// --------------------------------------------------------------------------------
188
189	static GB_ERROR NT_fix_gene_data(GBDATA gb_main, size_t species_count, size_t / sai_count */) {
190	GB_transaction ta(gb_main);
191	arb_progress progress(species_count);
192
193	size_t deleted_gene_datas = 0;
194	size_t generated_gene_datas = 0;
195	GB_ERROR error = 0;
196
197	for (GBDATA *gb_species = GBT_first_species(gb_main);
198	gb_species && !error;
199	gb_species = GBT_next_species(gb_species))
200	{
201	bool is_organism = (GB_entry(gb_species, GENOM_ALIGNMENT) != 0); // same test as GEN_is_organism, but w/o genome-db-assertion
202	GBDATA *gb_gene_data = GEN_find_gene_data(gb_species);
203
204	if (is_organism && !gb_gene_data) {
205	gb_gene_data = GEN_findOrCreate_gene_data(gb_species); // @@@ check result & handle error
206	generated_gene_datas++;
207	}
208	else if (!is_organism && gb_gene_data) {
209	GBDATA *gb_child = GB_child(gb_gene_data);
210	if (!gb_child) {
211	error = GB_delete(gb_gene_data);
212	if (!error) deleted_gene_datas++;
213	}
214	else {
215	error = GBS_global_string("Non-empty 'gene_data' found for species '%s',\n"
216	"which has no alignment '" GENOM_ALIGNMENT "',\n"
217	"i.e. which is not regarded as full-genome organism.\n"
218	"This causes problems - please fix!",
219	GBT_read_name(gb_species));
220	}
221	}
222
223	progress.inc_and_check_user_abort(error);
224	}
225
226	if (!error) {
227	if (deleted_gene_datas) {
228	aw_message(GBS_global_string("Deleted %zu useless empty 'gene_data' entries.", deleted_gene_datas));
229	}
230	if (generated_gene_datas) {
231	aw_message(GBS_global_string("Re-created %zu missing 'gene_data' entries.\nThese organisms have no genes yet!", generated_gene_datas));
232	}
233	}
234	return ta.close(error);
235	}
236
237	// --------------------------------------------------------------------------------
238
239	static GBDATA expectField(GBDATA gb_gene, const char *field, GB_ERROR& data_error) {
240	GBDATA *gb_field = 0;
241	if (!data_error) {
242	gb_field = GB_entry(gb_gene, field);
243	if (!gb_field) data_error = GBS_global_string("Expected field '%s' missing", field);
244	}
245	return gb_field;
246	}
247
248	static GBDATA disexpectField(GBDATA gb_gene, const char *field, GB_ERROR& data_error) {
249	GBDATA *gb_field = 0;
250	if (!data_error) {
251	gb_field = GB_entry(gb_gene, field);
252	if (gb_field) data_error = GBS_global_string("Unexpected field '%s' exists (wrong value in pos_joined?)", field);
253	}
254	GBS_reuse_buffer(field);
255	return gb_field;
256	}
257
258	static GB_ERROR NT_convert_gene_locations(GBDATA gb_main, size_t species_count, size_t / sai_count */) {
259	GB_transaction ta(gb_main);
260	GB_ERROR error = 0;
261	long fixed_genes = 0;
262	long skipped_genes = 0;
263	long genes = 0;
264
265	typedef vector<GBDATA*> GBvec;
266	GBvec toDelete;
267
268	arb_progress progress(species_count);
269
270	for (GBDATA *gb_organism = GEN_first_organism(gb_main);
271	gb_organism && !error;
272	gb_organism = GEN_next_organism(gb_organism))
273	{
274	GBDATA *gb_gene_data = GEN_find_gene_data(gb_organism);
275	nt_assert(gb_gene_data);
276	if (gb_gene_data) {
277	for (GBDATA *gb_gene = GEN_first_gene_rel_gene_data(gb_gene_data);
278	gb_gene && !error;
279	gb_gene = GEN_next_gene(gb_gene))
280	{
281	genes++;
282
283	int parts = 1;
284	{
285	GBDATA *gb_pos_joined = GB_entry(gb_gene, "pos_joined");
286	if (gb_pos_joined) parts = GB_read_int(gb_pos_joined); // its a joined gene
287	}
288
289	GBDATA *gb_pos_start = GB_entry(gb_gene, "pos_start"); // test for new format
290	if (!gb_pos_start) {
291	GBDATA *gb_pos_begin = GB_entry(gb_gene, "pos_begin"); // test for old format
292	if (!gb_pos_begin) {
293	error = "Neither 'pos_begin' nor 'pos_start' found - format of gene location is unknown";
294	}
295	}
296
297	if (!gb_pos_start && !error) { // assume old format
298	// parts<-1 would be valid in new format, but here we have old format
299	if (parts<1) error = GBS_global_string("Illegal value in 'pos_joined' (%i)", parts);
300
301	GB_ERROR data_error = 0; // error in this gene -> don't convert
302	GEN_position *pos = GEN_new_position(parts, false); // all were joinable (no information about it was stored)
303
304	// parse old gene information into 'pos'
305	//
306	// old-format was:
307	// Start-Positions: pos_begin, pos_begin2, pos_begin3, ...
308	// End-Positions: pos_end, pos_end2, pos_end3, ...
309	// Joined?: pos_joined (always >= 1)
310	// Complement: complement (one entry for all parts)
311	// Certainty: pos_uncertain (maybe pos_uncertain1 etc.)
312
313	int complement = 0;
314	{
315	GBDATA *gb_complement = GB_entry(gb_gene, "complement");
316	if (gb_complement) {
317	complement = GB_read_byte(gb_complement);
318	toDelete.push_back(gb_complement);
319	}
320	}
321
322	bool has_uncertain_fields = false;
323	for (int p = 1; p <= parts && !error && !data_error; ++p) {
324	GBDATA *gb_pos_begin = 0;
325	GBDATA *gb_pos_end = 0;
326	const char *pos_uncertain_field = 0;
327
328	if (p == 1) {
329	gb_pos_begin = expectField(gb_gene, "pos_begin", data_error);
330	gb_pos_end = expectField(gb_gene, "pos_end", data_error);
331
332	pos_uncertain_field = "pos_uncertain";
333	}
334	else {
335	const char *pos_begin_field = GBS_global_string("pos_begin%i", p);
336	const char *pos_end_field = GBS_global_string("pos_end%i", p);
337
338	gb_pos_begin = expectField(gb_gene, pos_begin_field, data_error);
339	gb_pos_end = expectField(gb_gene, pos_end_field, data_error);
340
341	GBS_reuse_buffer(pos_end_field);
342	GBS_reuse_buffer(pos_begin_field);
343
344	if (!data_error) pos_uncertain_field = GBS_global_string("pos_uncertain%i", p);
345	}
346
347	int pospos = complement ? (parts-p) : (p-1);
348
349	if (!data_error) {
350	GBDATA *gb_pos_uncertain = GB_entry(gb_gene, pos_uncertain_field);
351
352	if (!gb_pos_uncertain) {
353	if (has_uncertain_fields) data_error = GBS_global_string("Expected field '%s' missing", pos_uncertain_field);
354	}
355	else {
356	if (p == 1) has_uncertain_fields = true;
357	else {
358	if (!has_uncertain_fields) {
359	data_error = GBS_global_string("Found '%s' as first certainty-information", pos_uncertain_field);
360	}
361	}
362	}
363
364	if (!data_error) {
365	int begin = GB_read_int(gb_pos_begin);
366	int end = GB_read_int(gb_pos_end);
367
368	pos->start_pos[pospos] = begin;
369	pos->stop_pos[pospos] = end;
370	pos->complement[pospos] = complement; // set all complement entries to same value (old format only had one complement entry)
371
372	if (gb_pos_uncertain) {
373	const char *uncertain = GB_read_char_pntr(gb_pos_uncertain);
374
375	if (!uncertain) error = GB_await_error();
376	else {
377	if (!pos->start_uncertain) GEN_use_uncertainties(pos);
378
379	if (strlen(uncertain) != 2) {
380	data_error = "wrong length";
381	}
382	else {
383	for (int up = 0; up<2; up++) {
384	if (strchr("<=>", uncertain[up]) == 0) {
385	data_error = GBS_global_string("illegal character '%c'", uncertain[up]);
386	}
387	else {
388	(up == 0 ? pos->start_uncertain[pospos] : pos->stop_uncertain[pospos]) = uncertain[up];
389	}
390	}
391	}
392
393
394	toDelete.push_back(gb_pos_uncertain);
395	}
396	}
397
398	toDelete.push_back(gb_pos_begin);
399	toDelete.push_back(gb_pos_end);
400	}
401	}
402	}
403
404	for (int p = parts+1; p <= parts+4 && !error && !data_error; ++p) {
405	disexpectField(gb_gene, GBS_global_string("pos_begin%i", p), data_error);
406	disexpectField(gb_gene, GBS_global_string("pos_end%i", p), data_error);
407	disexpectField(gb_gene, GBS_global_string("complement%i", p), data_error);
408	disexpectField(gb_gene, GBS_global_string("pos_uncertain%i", p), data_error);
409	}
410
411	// now save new position data
412
413	if (data_error) {
414	skipped_genes++;
415	}
416	else if (!error) {
417	error = GEN_write_position(gb_gene, pos, 0);
418
419	if (!error) {
420	// delete old-format entries
421	GBvec::const_iterator end = toDelete.end();
422	for (GBvec::const_iterator i = toDelete.begin(); i != end && !error; ++i) {
423	GBDATA gb_del = i;
424	error = GB_delete(gb_del);
425	}
426
427	if (!error) fixed_genes++;
428	}
429	}
430
431	toDelete.clear();
432	GEN_free_position(pos);
433
434	if (data_error \|\| error) {
435	char *gene_id = GEN_global_gene_identifier(gb_gene, gb_organism);
436	if (error) {
437	error = GBS_global_string("Gene '%s': %s", gene_id, error);
438	}
439	else {
440	aw_message(GBS_global_string("Gene '%s' was not converted, fix data manually!\nReason: %s", gene_id, data_error));
441	}
442	free(gene_id);
443	}
444	}
445	}
446	}
447
448	progress.inc_and_check_user_abort(error);
449	}
450
451	if (!error) {
452	if (fixed_genes>0) aw_message(GBS_global_string("Fixed location entries of %li genes.", fixed_genes));
453	if (skipped_genes>0) {
454	aw_message(GBS_global_string("Didn't fix location entries of %li genes (see warnings).", skipped_genes));
455	error = "Not all gene locations were fixed.\nFix manually, save DB and restart ARB with that DB.\nMake sure you have a backup copy of the original DB!";
456	}
457
458	if (fixed_genes \|\| skipped_genes) {
459	long already_fixed_genes = genes-(fixed_genes+skipped_genes);
460	if (already_fixed_genes>0) aw_message(GBS_global_string("Location entries of %li genes already were in new format.", already_fixed_genes));
461	}
462	}
463
464	return error;
465	}
466
467
468	// --------------------------------------------------------------------------------
469
470	static GB_ERROR NT_del_mark_move_REF(GBDATA *gb_main, size_t species_count, size_t sai_count) {
471	GB_transaction ta(gb_main);
472	GB_ERROR error = 0;
473	size_t all = species_count+sai_count;
474	size_t removed = 0;
475
476	// delete 'mark' entries from all alignments of species/SAIs
477
478	arb_progress progress(all);
479	ConstStrArray ali_names;
480	GBT_get_alignment_names(ali_names, gb_main);
481
482	for (int pass = 0; pass < 2 && !error; ++pass) {
483	for (GBDATA *gb_item = (pass == 0) ? GBT_first_species(gb_main) : GBT_first_SAI(gb_main);
484	gb_item && !error;
485	gb_item = (pass == 0) ? GBT_next_species(gb_item) : GBT_next_SAI(gb_item))
486	{
487	for (int ali = 0; ali_names[ali] && !error; ++ali) {
488	GBDATA *gb_ali = GB_entry(gb_item, ali_names[ali]);
489	if (gb_ali) {
490	GBDATA *gb_mark = GB_entry(gb_ali, "mark");
491	if (gb_mark) {
492	error = GB_delete(gb_mark);
493	removed++;
494	}
495	}
496	}
497
498	progress.inc_and_check_user_abort(error);
499	}
500	}
501
502	{
503	char *helix_name = GBT_get_default_helix(gb_main);
504	GBDATA *gb_helix = GBT_find_SAI(gb_main, helix_name);
505
506	if (gb_helix) {
507	for (int ali = 0; ali_names[ali] && !error; ++ali) {
508	GBDATA *gb_ali = GB_entry(gb_helix, ali_names[ali]);
509	GBDATA *gb_old_ref = GB_entry(gb_ali, "REF");
510	GBDATA *gb_new_ref = GB_entry(gb_ali, "_REF");
511
512	if (gb_old_ref) {
513	if (gb_new_ref) {
514	error = GBS_global_string("SAI:%s has 'REF' and '_REF' in '%s' (data corrupt?!)",
515	helix_name, ali_names[ali]);
516	}
517	else { // move info from REF -> _REF
518	char *content = GB_read_string(gb_old_ref);
519	if (!content) error = GB_await_error();
520	else {
521	gb_new_ref = GB_create(gb_ali, "_REF", GB_STRING);
522	if (!gb_new_ref) error = GB_await_error();
523	else {
524	error = GB_write_string(gb_new_ref, content);
525	if (!error) error = GB_delete(gb_old_ref);
526	}
527	free(content);
528	}
529	}
530	}
531	}
532	}
533
534	free(helix_name);
535	}
536
537	if (!error) {
538	if (removed) {
539	aw_message(GBS_global_string("Deleted %zu useless 'mark' entries.", removed));
540	}
541	}
542
543	return ta.close(error);
544	}
545
546	// --------------------------------------------------------------------------------
547
548	static bool testDictionaryCompression(GBDATA *gbd, GBQUARK key_quark, bool testUse) {
549	// returns true, if
550	// testUse == true and ANY entries below 'gbd' with quark 'key_quark' uses dictionary compression
551	// testUse == false and ALL entries below 'gbd' with quark 'key_quark' can be decompressed w/o errors
552
553	nt_assert(GB_read_type(gbd) == GB_DB);
554
555	for (GBDATA *gb_sub = GB_child(gbd); gb_sub; gb_sub = GB_nextChild(gb_sub)) {
556	switch (GB_read_type(gb_sub)) {
557	case GB_DB:
558	// return false if any compression failed or return true if any uses dict-compression
559	if (testDictionaryCompression(gb_sub, key_quark, testUse) == testUse) return testUse;
560	break;
561
562	case GB_STRING:
563	case GB_LINK:
564	if (GB_get_quark(gb_sub) == key_quark && GB_is_dictionary_compressed(gb_sub)) {
565	if (testUse) return true;
566
567	const char *decompressed = GB_read_char_pntr(gb_sub);
568	if (!decompressed) return false;
569	}
570	break;
571
572	default:
573	break;
574	}
575	}
576
577	return !testUse;
578	}
579
580	class Dict;
581	typedef SmartPtr<Dict> DictPtr;
582
583
584	class KeyInfo : virtual Noncopyable {
585	string name; // keyname
586	DictPtr original;
587
588	bool compressionTested;
589	bool compressed;
590
591	void init() {
592	compressionTested = false;
593	compressed = false;
594	}
595
596	public:
597	KeyInfo(const char *Name) : name(Name) { init(); }
598	KeyInfo(const char *Name, DictPtr originalDict) : name(Name), original(originalDict) { init(); }
599
600	void testCompressed(GBDATA *gb_main) {
601	nt_assert(!compressionTested);
602	compressed = testDictionaryCompression(gb_main, GB_find_or_create_quark(gb_main, name.c_str()), true);
603	compressionTested = true;
604	}
605
606	const string& getName() const { return name; }
607
608	bool isCompressed() const {
609	nt_assert(compressionTested);
610	return compressed;
611	}
612	};
613
614
615	class Dict : virtual Noncopyable {
616	string group; // lowercase keyname
617	string orgkey;
618	DictData *data;
619
620	map<string, bool> decompressWorks; // key -> bool
621
622	public:
623	static GBDATA *gb_main;
624
625	Dict(const char Group, const char OrgKey, DictData *Data) : group(Group), orgkey(OrgKey), data(Data) {}
626
627	const string& getGroup() const { return group; }
628	const string& getOriginalKey() const { return orgkey; }
629
630	bool mayBeUsedWith(const string& key) const { return strcasecmp(group.c_str(), key.c_str()) == 0; }
631
632	GB_ERROR assignToKey(const string& key) const { return GB_set_dictionary(gb_main, key.c_str(), data); }
633	GB_ERROR unassignFromKey(const string& key) const { return GB_set_dictionary(gb_main, key.c_str(), NULL); }
634
635	bool canDecompress(const string& key) {
636	nt_assert(mayBeUsedWith(key));
637	if (decompressWorks.find(key) == decompressWorks.end()) {
638	bool works = false;
639	GB_ERROR error = assignToKey(key);
640
641	if (!error) works = testDictionaryCompression(gb_main, GB_find_or_create_quark(gb_main, key.c_str()), false);
642	decompressWorks[key] = works;
643
644	GB_ERROR err2 = unassignFromKey(key);
645	if (err2) {
646	aw_message(GBS_global_string("Error while removing @dictionary from key '%s': %s", key.c_str(), err2));
647	}
648	}
649	return decompressWorks[key];
650	}
651	};
652	GBDATA *Dict::gb_main = NULL;
653
654
655	typedef map<string, int> KeyCounter; // groupname -> occur count
656	typedef SmartPtr<KeyInfo> KeyInfoPtr;
657	typedef map<string, KeyInfoPtr> Keys; // keyname -> info
658	typedef map<string, DictPtr> DictMap;
659	typedef vector<DictPtr> Dicts;
660	typedef set<string> StringSet;
661
662	#define STATUS_PREFIX "Dictionary: "
663
664	template<typename CONT, typename KEY>
665	bool contains(const CONT& container, const KEY& key) {
666	return container.find(key) != container.end();
667	}
668
669	static GB_ERROR findAffectedKeys(GBDATA *gb_key_data, KeyCounter& kcount, Keys& keys, Dicts& dicts) {
670	GB_ERROR error = 0;
671	GBDATA *gb_main = GB_get_root(gb_key_data);
672
673	for (int pass = 1; pass <= 2; ++pass) {
674	for (GBDATA *gb_key = GB_entry(gb_key_data, "@key"); !error && gb_key; gb_key = GB_nextEntry(gb_key)) {
675	GBDATA *gb_name = GB_entry(gb_key, "@name");
676	const char *keyName = GB_read_char_pntr(gb_name);
677
678	if (!keyName) {
679	error = GBS_global_string("@key w/o @name (%s)", GB_await_error());
680	}
681	else {
682	char *keyGroup = strdup(keyName);
683	ARB_strlower(keyGroup);
684
685	switch (pass) {
686	case 1:
687	kcount[keyGroup]++;
688	break;
689	case 2:
690	if (kcount[keyGroup]>1) {
691	GBDATA *gb_dictionary = GB_entry(gb_key, "@dictionary");
692	if (gb_dictionary) {
693	DictPtr dict = new Dict(keyGroup, keyName, GB_get_dictionary(gb_main, keyName));
694	keys[keyName] = new KeyInfo(keyName, dict);
695	dicts.push_back(dict);
696	}
697	else keys[keyName] = new KeyInfo(keyName);
698	}
699	else kcount.erase(keyGroup);
700	break;
701	}
702	free(keyGroup);
703	}
704	}
705	}
706	return error;
707	}
708
709	static GB_ERROR deleteDataOfKey(GBDATA *gbd, GBQUARK key_quark, StringSet& deletedData, long& deleted, long& notDeleted) {
710	GB_ERROR error = 0;
711	for (GBDATA *gb_sub = GB_child(gbd); gb_sub; gb_sub = GB_nextChild(gb_sub)) {
712	switch (GB_read_type(gb_sub)) {
713	case GB_DB:
714	error = deleteDataOfKey(gb_sub, key_quark, deletedData, deleted, notDeleted);
715	break;
716
717	case GB_STRING:
718	case GB_LINK:
719	if (GB_get_quark(gb_sub) == key_quark) {
720	if (GB_is_dictionary_compressed(gb_sub)) {
721	string path(GB_get_db_path(gb_sub));
722	error = GB_delete(gb_sub);
723	if (!error) {
724	deletedData.insert(path);
725	deleted++;
726	}
727	}
728	else {
729	notDeleted++;
730	}
731	}
732	break;
733	default:
734	break;
735	}
736	}
737	return error;
738	}
739
740	static char readFirstCompressedDataOf(GBDATA gbd, GBQUARK key_quark) {
741	char *data = 0;
742	for (GBDATA *gb_sub = GB_child(gbd); !data && gb_sub; gb_sub = GB_nextChild(gb_sub)) {
743	switch (GB_read_type(gb_sub)) {
744	case GB_DB:
745	data = readFirstCompressedDataOf(gb_sub, key_quark);
746	break;
747
748	case GB_STRING:
749	case GB_LINK:
750	if (GB_get_quark(gb_sub) == key_quark) {
751	if (GB_is_dictionary_compressed(gb_sub)) {
752	data = GB_read_as_string(gb_sub);
753	}
754	}
755	break;
756	default:
757	break;
758	}
759	}
760	return data;
761	}
762
763
764	static GB_ERROR NT_fix_dict_compress(GBDATA *gb_main, size_t, size_t) {
765	GB_transaction ta(gb_main);
766	GBDATA *gb_key_data = GB_search(gb_main, GB_SYSTEM_FOLDER "/" GB_SYSTEM_KEY_DATA, GB_FIND);
767	GB_ERROR error = 0;
768
769	Dict::gb_main = gb_main;
770
771	if (!gb_key_data) {
772	error = "No " GB_SYSTEM_KEY_DATA " found.. DB corrupted?";
773	}
774	else {
775	KeyCounter kcount; // strlwr(keyname) -> count
776	Keys keys;
777	Dicts dicts;
778
779	error = findAffectedKeys(gb_key_data, kcount, keys, dicts);
780
781	// count affected keys
782	int affectedKeys = 0;
783	for (KeyCounter::iterator kci = kcount.begin(); kci != kcount.end(); ++kci) {
784	affectedKeys += kci->second;
785	}
786
787	if (!error && affectedKeys>0) {
788	// check which keys are compressed
789
790	{
791	arb_progress progress(STATUS_PREFIX "search compressed data", affectedKeys);
792
793	for (Keys::iterator ki = keys.begin(); ki != keys.end(); ++ki) {
794	KeyInfoPtr k = ki->second;
795	k->testCompressed(gb_main);
796	++progress;
797	}
798	}
799
800	// test which key/dict combinations work
801	int combinations = 0; // possible key/dict combinations
802
803	DictMap use; // keyname -> dictionary (which dictionary to use)
804	StringSet multiDecompressible; // keys which can be decompressed with multiple dictionaries
805
806	for (int pass = 1; pass <= 2; ++pass) {
807	arb_progress *progress = NULL;
808	if (pass == 2 && combinations) progress = new arb_progress(STATUS_PREFIX "test compression", combinations);
809
810	for (Dicts::iterator di = dicts.begin(); di != dicts.end(); ++di) {
811	DictPtr d = *di;
812
813	for (Keys::iterator ki = keys.begin(); ki != keys.end(); ++ki) {
814	KeyInfoPtr k = ki->second;
815	const string& keyname = k->getName();
816
817	if (k->isCompressed() && d->mayBeUsedWith(keyname)) {
818	switch (pass) {
819	case 1:
820	combinations++;
821	break;
822	case 2:
823	if (d->canDecompress(keyname)) {
824	if (!contains(use, keyname)) { // first dictionary working with keyname
825	use[keyname] = d;
826	}
827	else { // already have another dictionary working with keyname
828	multiDecompressible.insert(keyname);
829	}
830	}
831	++(*progress);
832	break;
833	}
834	}
835	}
836	}
837	delete progress;
838	}
839
840	StringSet notDecompressible; // keys which can be decompressed with none of the dictionaries
841	for (Keys::iterator ki = keys.begin(); ki != keys.end(); ++ki) {
842	KeyInfoPtr k = ki->second;
843	const string& keyname = k->getName();
844
845	if (k->isCompressed()) {
846	if (!contains(use, keyname)) notDecompressible.insert(keyname);
847	if (contains(multiDecompressible, keyname)) use.erase(keyname);
848	}
849	}
850
851	bool dataLost = false;
852	int reassigned = 0;
853
854	if (!notDecompressible.empty()) {
855	// bad .. found undecompressible data
856	int nd_count = notDecompressible.size();
857	aw_message(GBS_global_string("Detected corrupted dictionary compression\n"
858	"Data of %i DB-keys is lost and will be deleted", nd_count));
859
860	arb_progress progress(STATUS_PREFIX "deleting corrupt data", nd_count);
861
862	StringSet deletedData;
863	long deleted = 0;
864	long notDeleted = 0;
865
866	for (StringSet::iterator ki = notDecompressible.begin(); !error && ki != notDecompressible.end(); ++ki) {
867	const string& keyname = *ki;
868
869	error = deleteDataOfKey(gb_main, GB_find_or_create_quark(gb_main, keyname.c_str()), deletedData, deleted, notDeleted);
870	++progress;
871	}
872
873	if (!error) {
874	nt_assert(deleted); // at least 1 db-entry should have been deleted
875
876	aw_message(GBS_global_string("Deleted %li of %li affected DB-entries", deleted, deleted+notDeleted));
877	aw_message("see console for a list of affected keys");
878
879	printf("Deleted keys:\n");
880	for (StringSet::iterator di = deletedData.begin(); di != deletedData.end(); ++di) {
881	printf("* %s\n", di->c_str());
882	}
883	}
884	}
885
886	if (!error && !multiDecompressible.empty()) {
887	for (StringSet::iterator ki = multiDecompressible.begin(); !error && ki != multiDecompressible.end(); ++ki) {
888	const string& keyname = *ki;
889	int possible = 0;
890	vector<DictPtr> possibleDicts;
891
892	printf("--------------------------------------------------------------------------------\n");
893
894	for (Dicts::iterator di = dicts.begin(); !error && di != dicts.end(); ++di) {
895	DictPtr d = *di;
896	if (d->mayBeUsedWith(keyname) && d->canDecompress(keyname)) {
897	error = d->assignToKey(keyname);
898	if (!error) {
899	char *data = readFirstCompressedDataOf(gb_main, GB_find_or_create_quark(gb_main, keyname.c_str()));
900
901	nt_assert(data);
902	possible++;
903	printf("possibility %i = '%s'\n", possible, data);
904	free(data);
905
906	possibleDicts.push_back(d);
907
908	error = d->unassignFromKey(keyname);
909	}
910	}
911	}
912
913	if (!error) {
914	nt_assert(possible>0);
915
916	int selected;
917	if (possible>1) {
918	char *question = GBS_global_string_copy("%i possibilities to decompress field '%s' have been detected\n"
919	"and example data was dumped to the console.\n"
920	"Please examine output and decide which is the correct possibility!",
921	possible, keyname.c_str());
922
923	const char *buttons = "Abort";
924	for (int p = 1; p <= possible; ++p) buttons = GBS_global_string("%s,%i", buttons, p);
925	selected = aw_question("dict_decompress_bug", question, buttons, false, NULL);
926	free(question);
927	}
928	else {
929	selected = 1;
930	}
931
932	if (!selected) {
933	error = "Aborted by user";
934	}
935	else {
936	use[keyname] = possibleDicts[selected-1];
937	}
938	}
939	}
940	}
941
942	// now all redundancies should be eliminated and we can assign dictionaries to affected keys
943	if (!error) {
944	for (Keys::iterator ki = keys.begin(); !error && ki != keys.end(); ++ki) {
945	KeyInfoPtr k = ki->second;
946	const string& keyname = k->getName();
947
948	if (k->isCompressed()) {
949	if (!contains(use, keyname)) {
950	error = GBS_global_string("No dictionary detected for key '%s'", keyname.c_str());
951	}
952	else {
953	DictPtr d = use[keyname];
954
955	if (d->getOriginalKey() != keyname) {
956	d->assignToKey(keyname); // set the dictionary
957	aw_message(GBS_global_string("Assigning '%s'-dictionary to '%s'",
958	d->getOriginalKey().c_str(), keyname.c_str()));
959	reassigned++;
960	}
961	}
962	}
963	}
964	}
965
966	if (dataLost\|\|reassigned) {
967	aw_message(dataLost
968	? "We apologize for the data-loss."
969	: "No conflicts detected in compressed data.");
970	aw_message("Dictionaries fixed.\n"
971	"Please save your database with a new name.");
972	}
973	}
974	}
975
976	Dict::gb_main = NULL;
977	return ta.close(error);
978	}
979
980	// --------------------------------------------------------------------------------
981
982	static GB_ERROR remove_dup_colors(GBDATA *gb_item, ItemSelector& IF_DEBUG(sel)) {
983	// Databases out there may contain multiple 'ARB_color' entries.
984	// Due to some already fixed bug - maybe introduced in r5309 and fixed in r5825
985
986	GBDATA *gb_color = GB_entry(gb_item, AW_COLOR_GROUP_ENTRY);
987	GB_ERROR error = NULL;
988
989	#if defined(DEBUG)
990	int del_count = 0;
991	#endif // DEBUG
992
993	if (gb_color) {
994	GB_push_my_security(gb_color);
995	while (!error) {
996	GBDATA *gb_next_color = GB_nextEntry(gb_color);
997	if (!gb_next_color) break;
998
999	error = GB_delete(gb_next_color);
1000	#if defined(DEBUG)
1001	if (!error) del_count++;
1002	#endif // DEBUG
1003	}
1004	GB_pop_my_security(gb_color);
1005	}
1006
1007	#if defined(DEBUG)
1008	if (del_count) fprintf(stderr,
1009	"- deleted %i duplicated '" AW_COLOR_GROUP_ENTRY "' from %s '%s'\n",
1010	del_count,
1011	sel.item_name,
1012	sel.generate_item_id(GB_get_root(gb_item), gb_item));
1013	#endif // DEBUG
1014
1015	return error;
1016	}
1017
1018	// --------------------------------------------------------------------------------
1019
1020	GB_ERROR NT_repair_DB(GBDATA *gb_main) {
1021	// status is already open and will be closed by caller!
1022
1023	CheckedConsistencies check(gb_main);
1024	GB_ERROR err = 0;
1025	bool is_genome_db;
1026	{
1027	GB_transaction ta(gb_main);
1028	is_genome_db = GEN_is_genome_db(gb_main, -1);
1029	}
1030
1031	check.perform_check("fix gene_data", NT_fix_gene_data, err);
1032	check.perform_check("fix_dict_compress", NT_fix_dict_compress, err); // do this before NT_del_mark_move_REF (cause 'REF' is affected)
1033	check.perform_check("del_mark_move_REF", NT_del_mark_move_REF, err);
1034
1035	if (is_genome_db) {
1036	check.perform_check("convert_gene_locations", NT_convert_gene_locations, err);
1037	}
1038
1039	check.register_item_check("duplicated_item_colors", remove_dup_colors);
1040	check.perform_item_checks(err);
1041
1042	return err;
1043	}
1044
1045	void NT_rerepair_DB(AW_window*, AW_CL cl_gbmain, AW_CL) {
1046	// re-perform all DB checks
1047	GBDATA gb_main = reinterpret_cast<GBDATA>(cl_gbmain);
1048	GB_ERROR err = 0;
1049	{
1050	CheckedConsistencies check(gb_main);
1051	err = check.forgetDoneChecks();
1052	}
1053	if (!err) {
1054	arb_progress progress("DB-Repair");
1055	err = NT_repair_DB(gb_main);
1056	}
1057
1058	if (err) aw_message(err);
1059	}
1060
1061

Note: See TracBrowser for help on using the repository browser.

Download in other formats: