Context Navigation

source: tags/testbuild/EDIT4/ED4_search.cxx

Visit:

Last change on this file was 12848, checked in by westram, 10 years ago
reintegrates 'fix' into 'trunk' removes calls to `aw_string_selection` and `aw_string_selection2awar` (implementing another step for #179) adds: log:branches/fix@12839:12847
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 56.8 KB

Line
1	// =============================================================== //
2	// //
3	// File : ED4_search.cxx //
4	// Purpose : //
5	// //
6	// Coded by Ralf Westram (coder@reallysoft.de) //
7	// Institute of Microbiology (Technical University Munich) //
8	// http://www.arb-home.de/ //
9	// //
10	// =============================================================== //
11
12	#include <ed4_extern.hxx>
13	#include "ed4_awars.hxx"
14	#include "ed4_class.hxx"
15	#include "ed4_tools.hxx"
16
17	#include <fast_aligner.hxx>
18
19	#include <awt_config_manager.hxx>
20
21	#include <aw_awars.hxx>
22	#include <aw_file.hxx>
23	#include <aw_msg.hxx>
24	#include <aw_root.hxx>
25	#include <aw_question.hxx>
26
27	#include <arbdbt.h>
28
29	#include <climits>
30	#include <cctype>
31	#include <cerrno>
32	#include <map>
33
34	const char *ED4_SearchPositionTypeId[SEARCH_PATTERNS+1] =
35	{
36	"User1", "User2",
37	"Probe",
38	"Primer (local)", "Primer (region)", "Primer (global)",
39	"Signature (local)", "Signature (region)", "Signature (global)",
40	"Any"
41	};
42
43	// --------------------------------------------------------------------------------
44
45	struct SearchAwarList { // contains names of awars
46	const char *pattern,
47	*min_mismatches,
48	*max_mismatches,
49	*case_sensitive,
50	*tu,
51	*pat_gaps,
52	*seq_gaps,
53	*reverse,
54	*complement,
55	*exact,
56	*show,
57	*openFolded,
58	*autoJump;
59	};
60
61	class SearchSettings : virtual Noncopyable {
62	char *pattern;
63	int min_mismatches;
64	int max_mismatches;
65	ED4_SEARCH_CASE case_sensitive;
66	ED4_SEARCH_TU tu;
67	ED4_SEARCH_GAPS pat_gaps;
68	ED4_SEARCH_GAPS seq_gaps;
69	int reverse;
70	int complement;
71	int exact;
72	int open_folded;
73	int autoJump;
74
75	public:
76
77	void update(SearchAwarList *awarList) {
78	AW_root *root = ED4_ROOT->aw_root;
79
80	freeset(pattern, root->awar(awarList->pattern)->read_string());
81
82	min_mismatches = root->awar(awarList->min_mismatches)->read_int();
83	max_mismatches = root->awar(awarList->max_mismatches)->read_int();
84	case_sensitive = ED4_SEARCH_CASE(root->awar(awarList->case_sensitive)->read_int());
85	tu = ED4_SEARCH_TU(root->awar(awarList->tu)->read_int());
86	pat_gaps = ED4_SEARCH_GAPS(root->awar(awarList->pat_gaps)->read_int());
87	seq_gaps = ED4_SEARCH_GAPS(root->awar(awarList->seq_gaps)->read_int());
88	open_folded = root->awar(awarList->openFolded)->read_int();
89	autoJump = root->awar(awarList->autoJump)->read_int();
90	reverse = root->awar(awarList->reverse)->read_int();
91	complement = root->awar(awarList->complement)->read_int();
92	exact = root->awar(awarList->exact)->read_int();
93
94	if (complement) {
95	if (IS_AMINO()) {
96	complement = 0;
97	root->awar(awarList->complement)->write_int(0);
98	aw_message("Search for complement is not supported for this alignment type (has been disabled)");
99	}
100	}
101	}
102
103	SearchSettings(SearchAwarList *awarList) { pattern = 0; update(awarList); }
104	~SearchSettings() { delete pattern; }
105
106	GB_CSTR get_pattern() const { return pattern; }
107	int get_min_mismatches() const { return min_mismatches; }
108	int get_max_mismatches() const { return max_mismatches; }
109	ED4_SEARCH_CASE get_case_sensitive() const { return case_sensitive; }
110	ED4_SEARCH_TU get_tu() const { return tu; }
111	ED4_SEARCH_GAPS get_pat_gaps() const { return pat_gaps; }
112	ED4_SEARCH_GAPS get_seq_gaps() const { return seq_gaps; }
113	int get_open_folded() const { return open_folded; }
114	int get_autoJump() const { return autoJump; }
115	int get_reverse() const { return reverse; }
116	int get_complement() const { return complement; }
117	int get_exact() const { return exact; }
118	};
119
120	// --------------------------------------------------------------------------------
121
122	class SearchTree;
123	typedef void (*reportMatch)(int start, int end, GB_CSTR comment, int mismatches[MAX_MISMATCHES]);
124
125	class SearchTreeNode : virtual Noncopyable {
126	char c; // character
127	SearchTreeNode *son; // son != 0 (exception: FOUND)
128	SearchTreeNode *brother;
129	char *comment; // 0 or comment given in search pattern
130
131	static SearchTreeNode FOUND;
132	static int start_offset;
133	static reportMatch report;
134	static int min_mismatches;
135	static int max_mismatches;
136	static int *uni2real; // transform unified positions to real sequence positions
137
138	public:
139
140	SearchTreeNode(GB_CSTR pattern, GB_CSTR comment);
141	~SearchTreeNode();
142
143	SearchTreeNode *insert_unified_pattern(GB_CSTR pattern, GB_CSTR pattern_comment);
144	void findMatches(int off, GB_CSTR seq, int len, int mismatches, int mismatch_list[MAX_MISMATCHES]);
145
146	// you must call the following functions before calling findMatches():
147
148	static void set_start_offset(int off) { start_offset = off; }
149	static void set_report(reportMatch r, int *u2r) { report = r; uni2real = u2r; }
150	static void set_mismatches(int minMis, int maxMis)
151	{
152	e4_assert(maxMis <= MAX_MISMATCHES);
153	e4_assert(minMis <= maxMis);
154	min_mismatches = minMis;
155	max_mismatches = maxMis;
156	}
157	};
158
159	// --------------------------------------------------------------------------------
160
161	SearchTreeNode SearchTreeNode::FOUND(0, 0);
162	int SearchTreeNode::start_offset;
163	reportMatch SearchTreeNode::report;
164	int SearchTreeNode::min_mismatches;
165	int SearchTreeNode::max_mismatches;
166	int *SearchTreeNode::uni2real;
167
168	SearchTreeNode::SearchTreeNode(GB_CSTR pattern, GB_CSTR pattern_comment)
169	{
170	comment = 0;
171	if (pattern) {
172	e4_assert(pattern[0]);
173	c = pattern[0];
174	if (pattern[1]) {
175	son = new SearchTreeNode(pattern+1, pattern_comment);
176	}
177	else {
178	son = &FOUND;
179	comment = nulldup(pattern_comment);
180	}
181	}
182	else {
183	e4_assert(this==&FOUND);
184	c = 0;
185	son = 0;
186	comment = nulldup(pattern_comment);
187	}
188	brother = 0;
189	}
190
191	SearchTreeNode::~SearchTreeNode()
192	{
193	if (brother!=&FOUND) delete brother;
194	if (son!=&FOUND) delete son;
195	delete comment;
196	}
197
198
199	SearchTreeNode *SearchTreeNode::insert_unified_pattern(GB_CSTR pattern, GB_CSTR pattern_comment)
200	{
201	if (!this) {
202	if (pattern[0]) {
203	return new SearchTreeNode(pattern, pattern_comment);
204	}
205
206	return &FOUND;
207	}
208
209	if (this==&FOUND) {
210	if (pattern[0]) {
211	SearchTreeNode *neu = new SearchTreeNode(pattern, pattern_comment);
212
213	neu->brother = &FOUND;
214	return neu;
215	}
216	return &FOUND;
217	}
218
219	e4_assert(c);
220
221	if (pattern[0]) { // pattern contains sth.
222	if (c==pattern[0]) {
223	e4_assert(son);
224	son = son->insert_unified_pattern(pattern+1, pattern_comment);
225	}
226	else {
227	if (brother) {
228	brother = brother->insert_unified_pattern(pattern, pattern_comment);
229	}
230	else {
231	brother = new SearchTreeNode(pattern, pattern_comment);
232	}
233	}
234	}
235	else { // empty pattern -> insert FOUND
236	if (brother) {
237	if (brother!=&FOUND) {
238	brother = brother->insert_unified_pattern(pattern, pattern_comment);
239	}
240	}
241	else {
242	brother = &FOUND;
243	}
244	}
245
246	return this;
247	}
248
249	void SearchTreeNode::findMatches(int off, GB_CSTR seq, int len, int mismatches, int mismatch_list[MAX_MISMATCHES])
250	{
251	e4_assert(mismatches <= MAX_MISMATCHES);
252	if (len) {
253	int matches = c=='?' \|\| c==seq[0];
254	int use_mismatch = 0;
255
256	if (!matches && mismatches<max_mismatches) {
257	int c_is_gap = c=='-' \|\| c=='.';
258	int seq_is_gap = seq[0]=='-' \|\| seq[0]=='.';
259
260	if (c_is_gap==seq_is_gap) {
261	mismatch_list[mismatches] = uni2real[off];
262	mismatches++;
263	use_mismatch = 1;
264	matches = 1;
265	}
266	}
267
268	if (matches) {
269	if (son==&FOUND) {
270	if (mismatches>=min_mismatches) {
271	report(uni2real[start_offset], uni2real[off], comment, mismatch_list);
272	}
273	}
274	else {
275	son->findMatches(off+1, seq+1, len-1, mismatches, mismatch_list);
276	}
277
278	if (use_mismatch) {
279	mismatches--;
280	mismatch_list[mismatches] = -1;
281	}
282	}
283	}
284
285	if (brother==&FOUND) {
286	if (mismatches>=min_mismatches) {
287	report(uni2real[start_offset], uni2real[off-1], comment, mismatch_list);
288	}
289	}
290	else if (brother) {
291	brother->findMatches(off, seq, len, mismatches, mismatch_list);
292	}
293	}
294
295	// --------------------------------------------------------------------------------
296
297	class SearchTree : virtual Noncopyable {
298	const SearchSettings *sett;
299	SearchTreeNode *root;
300	unsigned char unified[256];
301	int shortestPattern;
302
303	static char unify_char(char c, int case_sensitive, int T_equal_U);
304
305	char unify_str(const char data, int len, ED4_SEARCH_GAPS gaps, int new_len, int *uni2real);
306	char unify_pattern(const char pattern, int *new_len);
307	char unify_sequence(const char sequence, int len, int new_len, int *uni2real);
308
309	SearchTree(const SearchTree &); // forbidden
310
311	public:
312
313	SearchTree(const SearchSettings *s);
314	~SearchTree();
315
316	void findMatches(const char *seq, int len, reportMatch report);
317	int get_shortestPattern() const { return shortestPattern; }
318	};
319
320	// --------------------------------------------------------------------------------
321
322	static char shortenString(char s)
323	{
324	char *end = strchr(s, '\0');
325
326	while (end>s && isspace(end[-1])) {
327	*--end = 0;
328	}
329	while (isspace(s[0])) {
330	s++;
331	}
332
333	return s;
334	}
335
336	static void splitTokComment(char tok, char commentP)
337	{
338	char num = strchr(tok, '#');
339
340	if (num) {
341	num[0] = 0;
342	*commentP = shortenString(num+1);
343	}
344	else {
345	*commentP = 0;
346	}
347
348	tok = shortenString(tok);
349	}
350
351	static char appendComment(const char s1, int l1, const char *s2) {
352	if (s1) {
353	int l2 = strlen(s2);
354	char s = (char)malloc(l1+1+l2+1);
355
356	sprintf(s, "%s %s", s1, s2);
357	return s;
358	}
359
360	return 0;
361	}
362
363	SearchTree::SearchTree(const SearchSettings *s)
364	{
365	sett = s;
366	root = 0;
367	shortestPattern = INT_MAX;
368
369	{
370	int i;
371	int case_sensitive = (sett->get_case_sensitive()==ED4_SC_CASE_SENSITIVE);
372	int T_equal_U = (sett->get_tu()==ED4_ST_T_EQUAL_U);
373
374	for (i=0; i<256; i++) {
375	unified[i] = unify_char(i, case_sensitive, T_equal_U);
376	}
377	}
378
379	#define INSERT_ROOT_PATTERN(tok, com) \
380	do { \
381	if (root) { \
382	root = root->insert_unified_pattern(tok, com); \
383	} \
384	else { \
385	root = new SearchTreeNode(tok, com); \
386	} \
387	} while (0)
388
389
390	{
391	char *pattern = strdup(sett->get_pattern());
392	const char *trenner = "\n,";
393	char *tok = strtok(pattern, trenner);
394	char *comment;
395	char T_or_U;
396	GB_ERROR T_or_U_error = GBT_determine_T_or_U(ED4_ROOT->alignment_type, &T_or_U, "complement");
397	bool show_T_or_U_error = false;
398
399	while (tok) {
400	splitTokComment(&tok, &comment);
401	int uni_tok_len;
402	char *uni_tok = unify_pattern(tok, &uni_tok_len);
403
404	if (uni_tok[0]) {
405	int s_exact = sett->get_exact();
406	int s_reverse = sett->get_reverse();
407	int s_complement = sett->get_complement();
408
409	if (uni_tok_len<shortestPattern) {
410	shortestPattern = uni_tok_len;
411	}
412
413	if (!s_exact \|\| (!s_reverse && !s_complement)) {
414	// insert original search pattern if all patterns shall be used (!s_exact)
415	// or if neither s_reverse nor s_complement
416	INSERT_ROOT_PATTERN(uni_tok, comment);
417	}
418	int commentLen = comment ? strlen(comment) : 0;
419
420	if (s_reverse) {
421	char *reverse = GBT_reverseNucSequence(uni_tok, uni_tok_len);
422
423	if (!s_exact \|\| !s_complement) {
424	char *reverseComment = appendComment(comment, commentLen, "(reverse)");
425
426	INSERT_ROOT_PATTERN(reverse, reverseComment); // insert reverse pattern
427	free(reverseComment);
428	}
429	if (s_complement) {
430	e4_assert(IS_NUCLEOTIDE());
431	if (T_or_U) {
432	char *revcomp = GBT_complementNucSequence(reverse, uni_tok_len, T_or_U);
433	char *revcompComment = appendComment(comment, commentLen, "(reverse complement)");
434	char *uni_revcomp = unify_pattern(revcomp, 0);
435
436	INSERT_ROOT_PATTERN(uni_revcomp, revcompComment); // insert reverse complement pattern
437
438	free(uni_revcomp);
439	free(revcompComment);
440	free(revcomp);
441	}
442	else {
443	show_T_or_U_error = true; // only show error if it matters
444	}
445	}
446
447	free(reverse);
448	}
449
450	if (s_complement) {
451	GB_transaction ta(GLOBAL_gb_main);
452
453	if (T_or_U) {
454	if (!s_exact \|\| !s_reverse) {
455	char *complement = GBT_complementNucSequence(uni_tok, uni_tok_len, T_or_U);
456	char *complementComment = appendComment(comment, commentLen, "(complement)");
457	char *uni_complement = unify_pattern(complement, 0);
458
459	INSERT_ROOT_PATTERN(uni_complement, complementComment); // insert complement pattern
460
461	free(uni_complement);
462	free(complementComment);
463	free(complement);
464	}
465	}
466	else {
467	show_T_or_U_error = true; // only show error if it matters
468	}
469	}
470
471	}
472
473	tok = strtok(0, trenner);
474	free(uni_tok);
475	}
476	free(pattern);
477
478	if (show_T_or_U_error && T_or_U_error) aw_message(T_or_U_error);
479	}
480
481	#undef INSERT_ROOT_PATTERN
482	}
483
484	SearchTree::~SearchTree()
485	{
486	delete root;
487	}
488
489	char SearchTree::unify_char(char c, int case_sensitive, int T_equal_U)
490	{
491	if (!case_sensitive) {
492	c = tolower(c);
493	}
494
495	if (T_equal_U) {
496	if (c=='t') {
497	c = 'u';
498	}
499	else if (c=='T') {
500	c = 'U';
501	}
502	}
503
504	if (c=='.') {
505	c = '-';
506	}
507
508	return c;
509	}
510
511	char SearchTree::unify_str(const char data, int len, ED4_SEARCH_GAPS gaps, int new_len, int *uni2real)
512	{
513	char p = (char)malloc(len+1);
514
515	if (!p) {
516	return 0;
517	}
518
519	char *pp = p;
520	int nlen = 0;
521	int realPos = 0;
522
523	if (uni2real) {
524	int u2r = uni2real;
525
526	if (gaps==ED4_SG_CONSIDER_GAPS) {
527	while (realPos<len) {
528	*pp++ = unified[(unsigned char)data[realPos]];
529	u2r[nlen++] = realPos++;
530	}
531	}
532	else {
533	while (realPos<len) {
534	unsigned char c = data[realPos];
535
536	if (c!='-' && c!='.') {
537	*pp++ = unified[c];
538	u2r[nlen++] = realPos;
539	}
540	realPos++;
541	}
542	}
543	}
544	else {
545	if (gaps==ED4_SG_CONSIDER_GAPS) {
546	while (realPos<len) {
547	*pp++ = unified[(unsigned char)data[realPos++]];
548	nlen++;
549	}
550	}
551	else {
552	while (realPos<len) {
553	unsigned char c = data[realPos++];
554
555	if (c!='-' && c!='.') {
556	*pp++ = unified[c];
557	nlen++;
558	}
559	}
560	}
561	}
562
563	// ---------------
564
565	pp[0] = 0;
566	if (new_len) {
567	*new_len = nlen;
568	}
569	return p;
570	}
571
572	char SearchTree::unify_pattern(const char pattern, int *new_len)
573	{
574	int len = strlen(pattern);
575	return unify_str(pattern, len, sett->get_pat_gaps(), new_len, 0);
576	}
577
578	char SearchTree::unify_sequence(const char sequence, int len, int new_len, int *uni2real)
579	{
580	return unify_str(sequence, len, sett->get_seq_gaps(), new_len, uni2real);
581	}
582
583	void SearchTree::findMatches(const char *seq, int len, reportMatch report)
584	{
585	if (root) {
586	int new_len;
587	int uni2real = (int)malloc(len*sizeof(int));
588	char *uni_seq = uni2real ? unify_sequence(seq, len, &new_len, &uni2real) : NULL;
589
590	if (uni_seq) {
591	int off;
592	char *useq = uni_seq;
593	int mismatch_list[MAX_MISMATCHES];
594
595	for (off=0; off<MAX_MISMATCHES; off++) {
596	mismatch_list[off] = -1;
597	}
598
599	SearchTreeNode::set_report(report, uni2real);
600	SearchTreeNode::set_mismatches(sett->get_min_mismatches(), sett->get_max_mismatches());
601
602	for (off=0; off<new_len; off++, useq++) {
603	SearchTreeNode::set_start_offset(off);
604	root->findMatches(off, useq, new_len-off, 0, mismatch_list);
605	}
606
607	free(uni_seq);
608	free(uni2real);
609	}
610	else {
611	aw_message("Out of swapspace?");
612	if (uni2real) free(uni2real);
613	}
614	}
615	}
616
617	// --------------------------------------------------------------------------------
618
619	#define AWAR_NAME(t, s) ED4_AWAR_##t##_SEARCH_##s
620
621	#define AWAR_LIST(t) \
622	AWAR_NAME(t, PATTERN), \
623	AWAR_NAME(t, MIN_MISMATCHES), \
624	AWAR_NAME(t, MAX_MISMATCHES), \
625	AWAR_NAME(t, CASE), \
626	AWAR_NAME(t, TU), \
627	AWAR_NAME(t, PAT_GAPS), \
628	AWAR_NAME(t, SEQ_GAPS), \
629	AWAR_NAME(t, REVERSE), \
630	AWAR_NAME(t, COMPLEMENT), \
631	AWAR_NAME(t, EXACT), \
632	AWAR_NAME(t, SHOW), \
633	AWAR_NAME(t, OPEN_FOLDED), \
634	AWAR_NAME(t, AUTO_JUMP)
635
636	static struct SearchAwarList awar_list[SEARCH_PATTERNS] = {
637	{ AWAR_LIST(USER1) },
638	{ AWAR_LIST(USER2) },
639	{ AWAR_LIST(PROBE) },
640	{ AWAR_LIST(PRIMER1) },
641	{ AWAR_LIST(PRIMER2) },
642	{ AWAR_LIST(PRIMER3) },
643	{ AWAR_LIST(SIG1) },
644	{ AWAR_LIST(SIG2) },
645	{ AWAR_LIST(SIG3) },
646	};
647
648	static inline int resultsAreShown(ED4_SearchPositionType type) {
649	return ED4_ROOT->aw_root->awar(awar_list[type].show)->read_int();
650	}
651
652	enum search_params_changed_action {
653	REFRESH_IF_SHOWN = 1,
654	REFRESH_ALWAYS = 2,
655	RECALC_SEARCH_TREE = 4,
656	TEST_MIN_MISMATCH = 8,
657	TEST_MAX_MISMATCH = 16,
658	DO_AUTO_JUMP = 32
659	};
660
661	// --------------------------------------------------------------------------------
662
663	static SearchSettings *settings[SEARCH_PATTERNS]; // last searched settings for each type
664	static SearchTree *tree[SEARCH_PATTERNS]; // Search trees for each type
665
666	// --------------------------------------------------------------------------------
667
668	static void searchParamsChanged(AW_root *root, AW_CL cl_type, AW_CL cl_action)
669	{
670	ED4_SearchPositionType type = ED4_SearchPositionType(cl_type);
671	enum search_params_changed_action action = (enum search_params_changed_action)cl_action;
672
673	// check awar values
674
675	if (action & (TEST_MIN_MISMATCH\|TEST_MAX_MISMATCH)) {
676	int mimi = root->awar(awar_list[type].min_mismatches)->read_int();
677	int mami = root->awar(awar_list[type].max_mismatches)->read_int();
678
679	if (mimi>mami) {
680	if (action & TEST_MIN_MISMATCH) { // max has changed
681	root->awar(awar_list[type].min_mismatches)->write_int(mami);
682	}
683	if (action & TEST_MAX_MISMATCH) { // min has changed
684	root->awar(awar_list[type].max_mismatches)->write_int(mimi);
685	}
686	}
687	}
688
689	// init new search
690
691	recalc :
692	ED4_SearchResults::setNewSearch(type);
693	if (!settings[type]) return;
694	settings[type]->update(&awar_list[type]);
695
696	if (action & RECALC_SEARCH_TREE) {
697	delete tree[type];
698	tree[type] = new SearchTree(settings[type]);
699	}
700
701	if (action & (RECALC_SEARCH_TREE\|TEST_MIN_MISMATCH\|TEST_MAX_MISMATCH)) {
702	int patLen = tree[type]->get_shortestPattern();
703	int maxMis = root->awar(awar_list[type].max_mismatches)->read_int();
704
705	if (patLen < 3*maxMis) {
706	int maxMaxMis = tree[type]->get_shortestPattern()/3;
707
708	aw_message(GBS_global_string("Too many mismatches (patterns of length=%i allow only %i max. mismatches)", patLen, maxMaxMis));
709
710	root->awar(awar_list[type].max_mismatches)->write_int(maxMaxMis);
711	if (root->awar(awar_list[type].min_mismatches)->read_int() > maxMaxMis) {
712	root->awar(awar_list[type].min_mismatches)->write_int(maxMaxMis);
713	}
714	goto recalc;
715	}
716	}
717
718	if (action & REFRESH_IF_SHOWN) {
719	if (resultsAreShown(type)) {
720	action = (enum search_params_changed_action)(action\|REFRESH_ALWAYS);
721	}
722	}
723
724	if (settings[type]->get_autoJump() && (action & DO_AUTO_JUMP)) { // auto jump
725	for (ED4_window *win = ED4_ROOT->first_window; win; win = win->next) {
726	ED4_LocalWinContext uses(win);
727
728	ED4_cursor *cursor = &current_cursor();
729	bool jumped = false;
730
731	if (cursor->owner_of_cursor && cursor->owner_of_cursor->is_sequence_terminal()) {
732	int pos = cursor->get_sequence_pos();
733	ED4_sequence_terminal *seq_term = cursor->owner_of_cursor->to_sequence_terminal();
734	ED4_SearchResults *result = &seq_term->results();
735
736	result->search(seq_term);
737	ED4_SearchPosition *found = result->get_last_starting_before(type, pos+1, 0);
738	int bestPos = -1;
739
740	if (found) {
741	bestPos = found->get_start_pos();
742	}
743
744	if (pos>=1) {
745	found = result->get_first_starting_after(type, pos-1, 0);
746	if (found) {
747	int next_pos = found->get_start_pos();
748
749	if (abs(pos-next_pos)<abs(pos-bestPos)) {
750	bestPos = next_pos;
751	}
752	}
753	}
754
755	if (bestPos!=-1) {
756	if (bestPos == pos) {
757	jumped = true; // already there
758	}
759	else {
760	jumped = seq_term->setCursorTo(cursor, bestPos, 1, ED4_JUMP_KEEP_POSITION);
761	}
762	}
763	}
764
765	if (!jumped) {
766	ED4_search_cb(0, ED4_encodeSearchDescriptor(+1, type), (AW_CL)current_ed4w());
767	}
768	}
769	}
770
771	if (action & REFRESH_ALWAYS) ED4_ROOT->request_refresh_for_sequence_terminals();
772	root->awar(ED4_AWAR_SEARCH_RESULT_CHANGED)->touch(); // trigger refresh in SECEDIT
773	}
774
775	void ED4_create_search_awars(AW_root *root)
776	{
777	#define cb(action) add_callback(searchParamsChanged, AW_CL(i), AW_CL(action))
778
779	int i;
780	for (i=0; i<SEARCH_PATTERNS; i++) {
781	root->awar_string(awar_list[i].pattern, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| RECALC_SEARCH_TREE \| DO_AUTO_JUMP);
782	root->awar_int(awar_list[i].case_sensitive, ED4_SC_CASE_INSENSITIVE, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| RECALC_SEARCH_TREE \| DO_AUTO_JUMP);
783	root->awar_int(awar_list[i].tu, ED4_ST_T_EQUAL_U, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| RECALC_SEARCH_TREE \| DO_AUTO_JUMP);
784	root->awar_int(awar_list[i].pat_gaps, ED4_SG_IGNORE_GAPS, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| RECALC_SEARCH_TREE \| DO_AUTO_JUMP);
785	root->awar_int(awar_list[i].reverse, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| RECALC_SEARCH_TREE \| DO_AUTO_JUMP);
786	root->awar_int(awar_list[i].complement, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| RECALC_SEARCH_TREE \| DO_AUTO_JUMP);
787	root->awar_int(awar_list[i].exact, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| RECALC_SEARCH_TREE \| DO_AUTO_JUMP);
788	root->awar_int(awar_list[i].min_mismatches, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| TEST_MAX_MISMATCH \| DO_AUTO_JUMP);
789	root->awar_int(awar_list[i].max_mismatches, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| TEST_MIN_MISMATCH \| DO_AUTO_JUMP);
790	root->awar_int(awar_list[i].seq_gaps, ED4_SG_IGNORE_GAPS, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN \| DO_AUTO_JUMP);
791	root->awar_int(awar_list[i].show, 1, GLOBAL_gb_main) ->cb(REFRESH_ALWAYS);
792	root->awar_int(awar_list[i].openFolded, 1, GLOBAL_gb_main) ->cb(0);
793	root->awar_int(awar_list[i].autoJump, 1, GLOBAL_gb_main) ->cb(DO_AUTO_JUMP);
794
795	settings[i] = new SearchSettings(&awar_list[i]);
796	tree[i] = new SearchTree(settings[i]);
797	}
798
799	root->awar_int(ED4_AWAR_SEARCH_RESULT_CHANGED, 0, GLOBAL_gb_main);
800
801	#undef cb
802
803	// awars to save/load search parameters:
804	{
805	char *dir = strdup(GB_path_in_arbprop("search_settings"));
806	AW_create_fileselection_awars(root, ED4_SEARCH_SAVE_BASE, dir, ".asp", "noname.asp");
807	root->awar(ED4_SEARCH_SAVE_BASE"/directory")->write_string(dir);
808	free(dir);
809	}
810	}
811
812	// --------------------------------------------------------------------------------
813
814	char *ED4_SearchPosition::lastShownComment = 0;
815
816	ED4_SearchPosition::ED4_SearchPosition(int sp, int ep, ED4_SearchPositionType wf, GB_CSTR found_comment, int mismatches[MAX_MISMATCHES])
817	{
818	e4_assert(sp<=ep && sp>=0);
819	start_pos = sp;
820	end_pos = ep;
821	whatsFound = wf;
822	next = 0;
823	comment = found_comment;
824	memcpy(mismatch, mismatches, sizeof(mismatch)MAX_MISMATCHES);
825	}
826	ED4_SearchPosition::ED4_SearchPosition(const ED4_SearchPosition& other) {
827	start_pos = other.start_pos;
828	end_pos = other.end_pos;
829	whatsFound = other.whatsFound;
830	next = 0;
831	comment = strdup(other.comment);
832	memcpy(mismatch, other.mismatch, sizeof(mismatch[0])*MAX_MISMATCHES);
833	}
834
835	ED4_SearchPosition ED4_SearchPosition::insert(ED4_SearchPosition toAdd)
836	{
837	ED4_SearchPosition *head = this;
838	ED4_SearchPosition *self = this;
839	ED4_SearchPosition *last = 0;
840
841	while (1) {
842	if (toAdd->cmp(*self)<=0) { // add before self
843	toAdd->next = self;
844	if (last) last->next = toAdd;
845	else head = toAdd;
846	break;
847	}
848	if (!self->next) { // add to end of list
849	self->next = toAdd;
850	break;
851	}
852	last = self;
853	self = self->next;
854	}
855
856	return head;
857	}
858	ED4_SearchPosition *ED4_SearchPosition::remove(ED4_SearchPositionType typeToRemove)
859	{
860	ED4_SearchPosition *head = this;
861	ED4_SearchPosition *self = this;
862	ED4_SearchPosition *last = 0;
863
864	while (self) {
865	if (self->whatsFound == typeToRemove) { // remove self
866	ED4_SearchPosition **ptrToMe = last ? &(last->next) : &head;
867
868	*ptrToMe = self->next;
869	self->next = 0;
870	delete self;
871	self = *ptrToMe;
872	}
873	else { // do not remove
874	last = self;
875	self = self->next;
876	}
877	}
878
879	return head;
880	}
881
882	#ifdef TEST_SEARCH_POSITION
883	int ED4_SearchPosition::ok() const
884	{
885	#ifndef NDEBUG
886	if (next) {
887	int c = cmp(*next);
888
889	if (c>0) {
890	printf("ED4_SearchPosition: list not sorted\n");
891	return 0;
892	}
893
894	return next->ok();
895	}
896	#endif
897	return 1;
898	}
899	#endif
900
901	GB_CSTR ED4_SearchPosition::get_comment() const
902	{
903	if (!comment) return 0;
904	if (lastShownComment && strcmp(lastShownComment, comment)==0) return 0; // do not show comment twice
905
906	delete lastShownComment;
907	lastShownComment = strdup(comment);
908	return lastShownComment;
909	}
910
911	ED4_SearchPosition *ED4_SearchPosition::get_next_at(int pos) const
912	{
913	ED4_SearchPosition *self = this->next;
914
915	while (self) {
916	if (self->start_pos > pos) break; // none found
917	if (self->containsPos(pos)) return self;
918	self = self->next;
919	}
920	return 0;
921	}
922
923	// --------------------------------------------------------------------------------
924
925	int ED4_SearchResults::initialized = 0;
926	int ED4_SearchResults::timeOfLastSearch[SEARCH_PATTERNS];
927	int ED4_SearchResults::timeOfNextSearch[SEARCH_PATTERNS];
928	int ED4_SearchResults::shown[SEARCH_PATTERNS];
929	int ED4_SearchResults::bufferSize;
930	char *ED4_SearchResults::buffer;
931
932	ED4_SearchResults::ED4_SearchResults()
933	{
934	if (!initialized) {
935	int i;
936
937	for (i=0; i<SEARCH_PATTERNS; i++) {
938	timeOfLastSearch[i] = 0;
939	timeOfNextSearch[i] = 1;
940	shown[i] = resultsAreShown(ED4_SearchPositionType(i));
941	}
942
943	bufferSize = 100;
944	buffer = (char*)GB_calloc(bufferSize, sizeof(char));
945
946	initialized = 1;
947	}
948
949	arraySize = 0; // list-format
950	array = 0;
951	first = 0;
952	int i;
953	for (i=0; i<SEARCH_PATTERNS; i++) {
954	timeOf[i] = 0;
955	}
956	}
957
958	ED4_SearchResults::~ED4_SearchResults() {
959	delete first;
960	free(array);
961	}
962
963	// --------------------------------------------------------------------------------
964
965	static ED4_SearchResults *reportToResult = 0;
966	static ED4_SearchPositionType reportType;
967
968	static void reportSearchPosition(int start, int end, GB_CSTR comment, int mismatches[MAX_MISMATCHES])
969	{
970	ED4_SearchPosition *pos = new ED4_SearchPosition(start, end, reportType, comment, mismatches);
971	reportToResult->addSearchPosition(pos);
972	}
973
974	// --------------------------------------------------------------------------------
975
976	void ED4_SearchResults::addSearchPosition(ED4_SearchPosition *pos) {
977	if (is_array()) {
978	to_list();
979	}
980
981	if (first) {
982	first = first->insert(pos);
983	#ifdef TEST_SEARCH_POSITION
984	e4_assert(first->ok());
985	#endif
986	}
987	else {
988	first = pos;
989	}
990	}
991
992	void ED4_SearchResults::search(const ED4_sequence_terminal *seq_terminal)
993	{
994	int i;
995	int needed[SEARCH_PATTERNS];
996	int st_needed = 0;
997
998	for (i=0; i<SEARCH_PATTERNS; i++) {
999	if (timeOf[i]<timeOfNextSearch[i]) {
1000	timeOf[i] = timeOfNextSearch[i];
1001	timeOfLastSearch[i] = timeOfNextSearch[i];
1002	needed[i] = 1;
1003	st_needed = 1;
1004	if (first) {
1005	if (is_array()) {
1006	to_list();
1007	}
1008	first = first->remove(ED4_SearchPositionType(i));
1009	#if defined TEST_SEARCH_POSITION
1010	e4_assert(!first \|\| first->ok());
1011	#endif
1012	}
1013	}
1014	else {
1015	needed[i] = 0;
1016	}
1017	}
1018
1019	if (st_needed) {
1020	int len;
1021	char *seq = seq_terminal->resolve_pointer_to_string_copy(&len);
1022
1023	if (seq) {
1024	reportToResult = this;
1025	for (i=0; i<SEARCH_PATTERNS; i++) {
1026	reportType = ED4_SearchPositionType(i);
1027	if (needed[i]) {
1028	if (is_array()) {
1029	to_list();
1030	}
1031	tree[i]->findMatches(seq, len, reportSearchPosition);
1032	}
1033	}
1034	reportToResult = 0;
1035	}
1036	free(seq);
1037	}
1038	}
1039
1040	ED4_SearchPosition *ED4_SearchResults::get_first_at(ED4_SearchPositionType type, int start, int end) const {
1041	if (is_list()) {
1042	const_cast<ED4_SearchResults*>(this)->to_array();
1043	}
1044
1045	int l = 0;
1046	int h = arraySize-1;
1047
1048	ED4_SearchPosition *pos = 0;
1049	int m = 0;
1050
1051	while (l<=h) {
1052	m = (l+h)/2;
1053	pos = array[m];
1054	if (pos->get_end_pos()<start) {
1055	l = m+1;
1056	}
1057	else if (pos->get_start_pos()>end) {
1058	h = m-1;
1059	}
1060	else {
1061	e4_assert(pos->get_end_pos()>=start && pos->get_start_pos()<=end);
1062	break;
1063	}
1064	}
1065
1066	if (l>h) {
1067	return 0;
1068	}
1069
1070	if (pos) {
1071	int best_m = m;
1072
1073	while (m>0) {
1074	m--;
1075	pos = array[m];
1076	if (pos->get_end_pos()>=start && pos->get_start_pos()<=end) {
1077	best_m = m;
1078	}
1079	}
1080	pos = array[best_m];
1081
1082	while (pos) {
1083	if (type==ED4_ANY_PATTERN \|\| pos->get_whatsFound()==type) {
1084	break;
1085	}
1086	pos = pos->get_next();
1087	}
1088
1089	return pos;
1090	}
1091
1092	return 0;
1093	}
1094
1095	ED4_SearchPosition *ED4_SearchResults::get_first_starting_after(ED4_SearchPositionType type, int pos, int mustBeShown) const
1096	{
1097	ED4_SearchPosition *sp = first;
1098
1099	while (sp) {
1100	if (type==ED4_ANY_PATTERN \|\| sp->get_whatsFound()==type) {
1101	if (sp->get_start_pos()>pos && (!mustBeShown \|\| resultsAreShown(sp->get_whatsFound()))) {
1102	break;
1103	}
1104	}
1105	sp = sp->get_next();
1106	}
1107
1108	return sp;
1109	}
1110
1111	ED4_SearchPosition *ED4_SearchResults::get_last_starting_before(ED4_SearchPositionType type, int pos, int mustBeShown) const
1112	{
1113	ED4_SearchPosition *sp = first,
1114	*best = 0;
1115
1116	while (sp) {
1117	if (type==ED4_ANY_PATTERN \|\| sp->get_whatsFound()==type) {
1118	if (sp->get_start_pos()<pos && (!mustBeShown \|\| resultsAreShown(sp->get_whatsFound()))) {
1119	best = sp;
1120	}
1121	else {
1122	break;
1123	}
1124	}
1125	sp = sp->get_next();
1126	}
1127
1128	return best;
1129	}
1130
1131	void ED4_SearchResults::setNewSearch(ED4_SearchPositionType type)
1132	{
1133	if (type==ED4_ANY_PATTERN) {
1134	int i;
1135	for (i=0; i<SEARCH_PATTERNS; i++) {
1136	setNewSearch(ED4_SearchPositionType(i));
1137	}
1138	}
1139	else {
1140	int next_unused_stamp = timeOfLastSearch[type] + 1;
1141
1142	shown[type] = resultsAreShown(type);
1143	if (timeOfNextSearch[type]!=next_unused_stamp) {
1144	timeOfNextSearch[type] = next_unused_stamp;
1145	}
1146	}
1147	}
1148
1149	void ED4_SearchResults::searchAgain()
1150	{
1151	int i;
1152
1153	for (i=0; i<SEARCH_PATTERNS; i++) {
1154	timeOf[i] = 0;
1155	}
1156	}
1157
1158	char ED4_SearchResults::buildColorString(const ED4_sequence_terminal seq_terminal, int start, int end)
1159	// builds a char buffer (access is only valid from result[start] till result[end])
1160	{
1161	int i;
1162	int st_shown = 0;
1163
1164	e4_assert(start<=end); // confirming the condition
1165	for (i=0; i<SEARCH_PATTERNS; i++) {
1166	if (shown[i]) {
1167	st_shown = 1;
1168	break;
1169	}
1170	}
1171	if (!st_shown) {
1172	return 0; // nothing shown
1173	}
1174
1175	search(seq_terminal);
1176	if (!get_first()) {
1177	return 0; // nothing found
1178	}
1179
1180	int needed_size = end-start+1;
1181	if (needed_size>bufferSize) {
1182	free(buffer);
1183	bufferSize = needed_size;
1184	buffer = (char*)GB_calloc(bufferSize, sizeof(char));
1185	}
1186	else {
1187	memset(buffer, 0, sizeof(char)*needed_size);
1188	}
1189
1190	// search first search-result that goes in between start-end
1191
1192	ED4_SearchPosition *pos = get_first_at(ED4_ANY_PATTERN, start, end);
1193	e4_assert(!pos \|\| (pos->get_start_pos()<=end && pos->get_end_pos()>=start));
1194
1195	int correct_neg_values = 0;
1196
1197	while (pos && pos->get_start_pos()<=end) {
1198	int what = int(pos->get_whatsFound());
1199
1200	if (shown[what]) {
1201	int color = ED4_G_SBACK_0 + what;
1202	int s = std::max(pos->get_start_pos(), start)-start;
1203	int e = std::min(pos->get_end_pos(), end)-start;
1204
1205	for (i=s; i<=e; i++) {
1206	if (buffer[i]==0 \|\| abs(buffer[i])>abs(color)) {
1207	buffer[i] = color;
1208	}
1209	}
1210
1211	const int *mismatches = pos->getMismatches();
1212
1213	for (i=0; i<5 && mismatches[i]>=0; i++) {
1214	int mpos = mismatches[i];
1215
1216	if (mpos>=start && mpos<=end) {
1217	int rpos = mpos-start;
1218	if (buffer[rpos]==color) {
1219	buffer[rpos] = -buffer[rpos];
1220	correct_neg_values = 1;
1221	}
1222	}
1223	}
1224	}
1225	pos = pos->get_next();
1226	}
1227
1228	if (correct_neg_values) {
1229	for (i=end-start; i>=0; i--) {
1230	if (buffer[i]<0) {
1231	buffer[i] = ED4_G_MBACK;
1232	}
1233	}
1234	}
1235
1236	return buffer-start;
1237	}
1238
1239	ED4_SearchPosition *ED4_SearchResults::get_shown_at(int pos) const
1240	// used by ED4_cursor::updateAwars to get search-comment
1241	{
1242	ED4_SearchPosition *p = get_last_starting_before(ED4_ANY_PATTERN, pos, 0); // @@@ tofix: should find last_ending before
1243	ED4_SearchPosition *best = 0;
1244
1245	if (!p) {
1246	p = get_first();
1247	}
1248
1249	if (p && !p->containsPos(pos)) {
1250	p = p->get_next_at(pos);
1251	}
1252
1253	while (p) {
1254	e4_assert(p->containsPos(pos));
1255	ED4_SearchPositionType type = p->get_whatsFound();
1256	if (shown[type]) {
1257	if (!best \|\| type<best->get_whatsFound()) {
1258	best = p;
1259	}
1260	}
1261	p = p->get_next_at(pos);
1262	}
1263
1264	return best;
1265	}
1266
1267	void ED4_SearchResults::to_array() {
1268	e4_assert(is_list());
1269
1270	ED4_SearchPosition *pos = first;
1271
1272	{
1273	int a_arraySize = 0;
1274
1275	while (pos) {
1276	a_arraySize++;
1277	pos = pos->get_next();
1278	}
1279	arraySize = a_arraySize;
1280	}
1281
1282	ED4_SearchPosition a_array = (ED4_SearchPosition)malloc(sizeof(ED4_SearchPosition)arraySize);
1283
1284	pos = first;
1285	for (int e=0; e<arraySize; e++) {
1286	e4_assert(pos);
1287	a_array[e] = pos;
1288	pos = pos->get_next();
1289	}
1290
1291	array = a_array;
1292	}
1293
1294	void ED4_SearchResults::to_list() {
1295	e4_assert(is_array());
1296
1297	freenull(array);
1298	arraySize = 0;
1299	}
1300
1301
1302	// --------------------------------------------------------------------------------
1303
1304	inline void decodeSearchDescriptor(int descriptor, int direction, ED4_SearchPositionType pattern)
1305	{
1306	*direction = descriptor&1 ? 1 : -1;
1307	*pattern = ED4_SearchPositionType(descriptor/2);
1308	}
1309
1310	static AW_CL last_searchDescriptor = -1;
1311
1312	GB_ERROR ED4_repeat_last_search(ED4_window *ed4w) {
1313	if (int(last_searchDescriptor)==-1) {
1314	return GBS_global_string("You have to search first, before you can repeat a search.");
1315	}
1316
1317	ED4_search_cb(0, last_searchDescriptor, (AW_CL)ed4w);
1318	return 0;
1319	}
1320
1321	void ED4_search_cb(AW_window *, AW_CL searchDescriptor, AW_CL cl_ed4w) {
1322	ED4_window ed4w = (ED4_window)cl_ed4w;
1323	e4_assert(ed4w);
1324
1325	ED4_LocalWinContext uses(ed4w);
1326
1327	last_searchDescriptor = searchDescriptor;
1328
1329	int direction;
1330	ED4_SearchPositionType pattern;
1331	decodeSearchDescriptor(searchDescriptor, &direction, &pattern);
1332
1333	int searchOnlyForShownPatterns = pattern==ED4_ANY_PATTERN;
1334
1335	// detect position where to start searching
1336
1337	ED4_terminal *terminal = 0; // start search at this terminal
1338	int pos; // ... and at this position
1339
1340	ED4_cursor *cursor = &current_cursor();
1341	if (cursor->owner_of_cursor) { // if cursor is shown -> use cursor position
1342	terminal = cursor->owner_of_cursor;
1343	pos = cursor->get_sequence_pos();
1344	}
1345	else { // start at end or beginning
1346	if (direction==1) {
1347	terminal = ED4_ROOT->root_group_man->get_first_terminal();
1348	pos = INT_MIN;
1349	}
1350	else {
1351	terminal = ED4_ROOT->root_group_man->get_last_terminal();
1352	pos = INT_MAX;
1353	}
1354	}
1355
1356	ED4_terminal *start_terminal = terminal;
1357	int start_pos = pos;
1358	int last_loop = 0;
1359
1360	while (terminal) {
1361	if (terminal->is_sequence_terminal()) {
1362	ED4_sequence_terminal *seq_terminal = terminal->to_sequence_terminal();
1363	ED4_SearchResults& results = seq_terminal->results();
1364	ED4_SearchPosition *found = 0;
1365
1366	if (pattern==ED4_ANY_PATTERN \|\| settings[pattern]->get_open_folded() \|\| !terminal->is_in_folded_group()) {
1367	results.search(seq_terminal);
1368
1369	if (direction==1) {
1370	found = results.get_first_starting_after(pattern, pos, searchOnlyForShownPatterns);
1371	}
1372	else {
1373	found = results.get_last_starting_before(pattern, pos, searchOnlyForShownPatterns);
1374	}
1375
1376	if (found) {
1377	pos = found->get_start_pos();
1378	if (terminal==start_terminal && pos==start_pos) {
1379	if (searchOnlyForShownPatterns) {
1380	aw_message("There are no other shown patterns");
1381	}
1382	else {
1383	aw_message("This is the only occurrence of the search pattern");
1384	}
1385	}
1386	else {
1387	seq_terminal->setCursorTo(&current_cursor(), pos, true, ED4_JUMP_KEEP_POSITION);
1388	}
1389	break;
1390	}
1391	else if (last_loop) {
1392	if (searchOnlyForShownPatterns) {
1393	aw_message("There are no shown patterns");
1394	}
1395	else {
1396	aw_message("Search pattern was not found in any sequence");
1397	}
1398	break;
1399	}
1400	}
1401	}
1402
1403	if (direction==1) {
1404	terminal = terminal->get_next_terminal();
1405	if (!terminal) terminal = ED4_ROOT->root_group_man->get_first_terminal();
1406	pos = INT_MIN;
1407	}
1408	else {
1409	terminal = terminal->get_prev_terminal();
1410	if (!terminal) terminal = ED4_ROOT->root_group_man->get_last_terminal();
1411	pos = INT_MAX;
1412	}
1413
1414	if (terminal==start_terminal) {
1415	last_loop = 1;
1416	}
1417	}
1418	}
1419
1420	static void ED4_mark_matching_species(AW_window * /* aww */, AW_CL cl_pattern) {
1421	ED4_SearchPositionType pattern = ED4_SearchPositionType(cl_pattern);
1422	ED4_terminal *terminal = ED4_ROOT->root_group_man->get_first_terminal();
1423	GB_transaction ta(GLOBAL_gb_main);
1424
1425	while (terminal) {
1426	if (terminal->is_sequence_terminal()) {
1427	ED4_sequence_terminal *seq_terminal = terminal->to_sequence_terminal();
1428	ED4_SearchResults& results = seq_terminal->results();
1429
1430	results.search(seq_terminal);
1431	ED4_SearchPosition *found = results.get_first_starting_after(pattern, INT_MIN, false);
1432
1433	if (found) {
1434	ED4_species_manager *species_man = seq_terminal->get_parent(ED4_L_SPECIES)->to_species_manager();
1435	if (species_man->is_species_seq_manager()) {
1436	GBDATA *gbd = species_man->get_species_pointer();
1437	e4_assert(gbd);
1438	GB_write_flag(gbd, 1);
1439	}
1440	}
1441	}
1442
1443	terminal = terminal->get_next_terminal();
1444	}
1445	}
1446
1447	#define ESC '\\'
1448
1449	static char *pattern2str(GB_CSTR p) {
1450	char s = GB_give_buffer(strlen(p)2);
1451	char *s1 = s;
1452	GB_CSTR p1 = p;
1453
1454	while (1) {
1455	char c = *p1++;
1456
1457	if (!c) {
1458	break;
1459	}
1460	else if (c==ESC) {
1461	*s1++ = ESC;
1462	*s1++ = ESC;
1463	}
1464	else if (c=='\n') {
1465	*s1++ = ESC;
1466	*s1++ = 'n';
1467	}
1468	else {
1469	*s1++ = c;
1470	}
1471	}
1472
1473	*s1 = 0;
1474	return strdup(s);
1475	}
1476
1477	static void str2pattern(char *s) { // works on string
1478	char *a = s;
1479	char *n = s;
1480
1481	while (1) {
1482	char c = *a++;
1483
1484	if (c==ESC) {
1485	c = *a++;
1486	if (c==ESC) {
1487	*n++ = ESC;
1488	}
1489	else if (c=='n') {
1490	*n++ = '\n';
1491	}
1492	else { // should not occur
1493	*n++ = ESC;
1494	*n++ = c;
1495	}
1496	}
1497	else {
1498	*n++ = c;
1499	if (!c) {
1500	break;
1501	}
1502	}
1503	}
1504	}
1505
1506	#undef ESC
1507
1508	static void save_search_paras_to_file(AW_window *aw, AW_CL cl_type) {
1509	GB_ERROR error = NULL;
1510	AW_root *root = ED4_ROOT->aw_root;
1511	char *filename = root->awar(ED4_SEARCH_SAVE_BASE"/file_name")->read_string();
1512
1513	{
1514	FILE *in = fopen(filename, "rt");
1515	if (in) {
1516	fclose(in);
1517
1518	char *question = GBS_global_string_copy("'%s' already exists", filename);
1519	if (aw_question("overwrite_search_params", question, "Overwrite,Cancel") != 0) {
1520	error = "Wont overwrite existing file";
1521	}
1522	}
1523	}
1524
1525	if (!error) {
1526	FILE *out = fopen(filename, "wt");
1527
1528	if (!out) {
1529	error = GBS_global_string("Can't write file '%s' (%s)", filename, strerror(errno));
1530	}
1531	else {
1532	ED4_SearchPositionType type = ED4_SearchPositionType(cl_type);
1533	SearchSettings *s = settings[type];
1534
1535	char *fpat = pattern2str(s->get_pattern());
1536
1537	fprintf(out,
1538	"pattern=%s\n"
1539	"minmis=%i\n"
1540	"maxmis=%i\n"
1541	"case=%i\n"
1542	"tu=%i\n"
1543	"patgaps=%i\n"
1544	"seqgaps=%i\n"
1545	"openfolded=%i\n"
1546	"autojump=%i\n"
1547	"reverse=%i\n"
1548	"complement=%i\n"
1549	"exact=%i\n",
1550	fpat,
1551	s->get_min_mismatches(),
1552	s->get_max_mismatches(),
1553	s->get_case_sensitive(),
1554	s->get_tu(),
1555	s->get_pat_gaps(),
1556	s->get_seq_gaps(),
1557	s->get_open_folded(),
1558	s->get_autoJump(),
1559	s->get_reverse(),
1560	s->get_complement(),
1561	s->get_exact());
1562
1563	free(fpat);
1564	fclose(out);
1565	}
1566	}
1567
1568	if (error) aw_message(error);
1569	else AW_POPDOWN(aw);
1570
1571	free(filename);
1572	}
1573
1574	static void load_search_paras_from_file(AW_window *aw, AW_CL cl_type) {
1575	GB_CSTR error = NULL;
1576	AW_root *root = ED4_ROOT->aw_root;
1577	char *filename = root->awar(ED4_SEARCH_SAVE_BASE"/file_name")->read_string();
1578	FILE *in = fopen(filename, "rt");
1579
1580	if (!in) {
1581	error = GBS_global_string("File '%s' not found", filename);
1582	}
1583	else {
1584	ED4_SearchPositionType type = ED4_SearchPositionType(cl_type);
1585	SearchAwarList *al = &awar_list[type];
1586
1587	while (1) {
1588	const int BUFFERSIZE = 10000;
1589	char buffer[BUFFERSIZE];
1590	if (!fgets(buffer, BUFFERSIZE, in)) break;
1591
1592	char *content = strchr(buffer, '=');
1593	if (!content) {
1594	aw_message(GBS_global_string("Format error in '%s' - ignored!", filename));
1595	}
1596	else {
1597	*content++ = 0;
1598
1599	if (strcmp(buffer, "pattern")==0) {
1600	str2pattern(content);
1601	root->awar(al->pattern)->write_string(content);
1602	}
1603	else {
1604	int value = atoi(content);
1605
1606	if (strcmp(buffer, "minmis")==0) root->awar(al->min_mismatches)->write_int(value);
1607	else if (strcmp(buffer, "maxmis")==0) root->awar(al->max_mismatches)->write_int(value);
1608	else if (strcmp(buffer, "case")==0) root->awar(al->case_sensitive)->write_int(value);
1609	else if (strcmp(buffer, "tu")==0) root->awar(al->tu)->write_int(value);
1610	else if (strcmp(buffer, "patgaps")==0) root->awar(al->pat_gaps)->write_int(value);
1611	else if (strcmp(buffer, "seqgaps")==0) root->awar(al->seq_gaps)->write_int(value);
1612	else if (strcmp(buffer, "openfolded")==0) root->awar(al->openFolded)->write_int(value);
1613	else if (strcmp(buffer, "autojump")==0) root->awar(al->autoJump)->write_int(value);
1614	else if (strcmp(buffer, "reverse")==0) root->awar(al->reverse)->write_int(value);
1615	else if (strcmp(buffer, "complement")==0) root->awar(al->complement)->write_int(value);
1616	else if (strcmp(buffer, "exact")==0) root->awar(al->exact)->write_int(value);
1617	else {
1618	aw_message(GBS_global_string("Unknown tag '%s' in '%s' - ignored!", buffer, filename));
1619	}
1620	}
1621	}
1622	}
1623
1624	fclose(in);
1625	}
1626
1627	if (error) aw_message(error);
1628	else AW_POPDOWN(aw);
1629
1630	free(filename);
1631	}
1632
1633	static void aws_init_localized(AW_root root, AW_window_simple aws, GB_CSTR id_format, GB_CSTR title_format, GB_CSTR typeId, int winNum) {
1634	char *window_title = GBS_global_string_copy(title_format, typeId);
1635	char *window_id_src = GBS_global_string_copy(id_format, typeId, winNum);
1636	char *window_id = GBS_string_2_key(window_id_src);
1637
1638	aws->init(root, window_id, window_title);
1639
1640	free(window_id);
1641	free(window_id_src);
1642	free(window_title);
1643	}
1644
1645	struct LoadSaveSearchParam {
1646	ED4_SearchPositionType type;
1647	int winNum;
1648
1649	LoadSaveSearchParam(ED4_SearchPositionType type_, int winNum_)
1650	: type(type_),
1651	winNum(winNum_)
1652	{}
1653	};
1654
1655	static AW_window loadsave_search_parameters(AW_root root, const LoadSaveSearchParam& param, bool load) {
1656	AW_window_simple *aws = new AW_window_simple;
1657
1658	if (load) {
1659	aws_init_localized(root, aws, "load_%s_search_para_%i", "Load %s Search Parameters", ED4_SearchPositionTypeId[param.type], param.winNum);
1660	}
1661	else {
1662	aws_init_localized(root, aws, "save_%s_search_para_%i", "Save %s Search Parameters", ED4_SearchPositionTypeId[param.type], param.winNum);
1663	}
1664
1665	aws->load_xfig("edit4/save_search.fig");
1666
1667	aws->at("close"); aws->callback((AW_CB0)AW_POPDOWN);
1668	aws->create_button("CLOSE", "CLOSE", "C");
1669
1670	aws->callback(makeHelpCallback("search_parameters.hlp"));
1671	aws->at("help");
1672	aws->create_button("HELP", "HELP", "H");
1673
1674	AW_create_standard_fileselection(aws, ED4_SEARCH_SAVE_BASE);
1675
1676	aws->callback((AW_CB0)AW_POPDOWN);
1677	aws->at("cancel");
1678	aws->create_button("CANCEL", "CANCEL", "C");
1679
1680	aws->at("save");
1681	if (load) {
1682	aws->callback(load_search_paras_from_file, (AW_CL)param.type);
1683	aws->create_button("LOAD", "LOAD", "L");
1684	}
1685	else {
1686	aws->callback(save_search_paras_to_file, (AW_CL)param.type);
1687	aws->create_button("SAVE", "SAVE", "S");
1688	}
1689
1690	return aws;
1691	}
1692
1693	static AW_window load_search_parameters(AW_root root, AW_CL cl_param) {
1694	LoadSaveSearchParam param = (LoadSaveSearchParam)cl_param;
1695	return loadsave_search_parameters(root, *param, true);
1696	}
1697
1698	static AW_window save_search_parameters(AW_root root, AW_CL cl_param) {
1699	LoadSaveSearchParam param = (LoadSaveSearchParam)cl_param;
1700	return loadsave_search_parameters(root, *param, false);
1701	}
1702
1703
1704	static void search_init_config(AWT_config_definition& cdef, int search_type) {
1705	SearchAwarList *awarList = &awar_list[search_type];
1706
1707	cdef.add(awarList->show, "show");
1708	cdef.add(awarList->openFolded, "openFolded");
1709	cdef.add(awarList->autoJump, "autoJump");
1710	cdef.add(awarList->pattern, "pattern");
1711	cdef.add(awarList->min_mismatches, "min_mismatches");
1712	cdef.add(awarList->max_mismatches, "max_mismatches");
1713	cdef.add(awarList->seq_gaps, "seq_gaps");
1714	cdef.add(awarList->pat_gaps, "pat_gaps");
1715	cdef.add(awarList->tu, "tu");
1716	cdef.add(awarList->case_sensitive, "case_sensitive");
1717	cdef.add(awarList->reverse, "reverse");
1718	cdef.add(awarList->complement, "complement");
1719	cdef.add(awarList->exact, "exact");
1720	}
1721
1722	static char *search_store_config(AW_CL cl_search_type, AW_CL) {
1723	AWT_config_definition cdef;
1724	search_init_config(cdef, int(cl_search_type));
1725	return cdef.read();
1726	}
1727
1728	static void search_restore_config(const char *stored_string, AW_CL cl_search_type, AW_CL) {
1729	AWT_config_definition cdef;
1730	search_init_config(cdef, int(cl_search_type));
1731	cdef.write(stored_string);
1732	}
1733
1734	struct search_windows : public Noncopyable {
1735	AW_window_simple *windows[SEARCH_PATTERNS];
1736	search_windows() { for (int i = 0; i<SEARCH_PATTERNS; ++i) windows[i] = NULL; }
1737	};
1738
1739	typedef std::map<ED4_window*, SmartPtr<search_windows> > search_window_map;
1740
1741	void ED4_popup_search_window(AW_window *aww, AW_CL cl_search_type) {
1742	ED4_SearchPositionType type = (ED4_SearchPositionType)cl_search_type;
1743
1744	ED4_WinContext uses(aww);
1745	ED4_window *ed4w = uses.get_ed4w();
1746
1747	static search_window_map swm;
1748	{ // create search windows pointer array for current ed4w
1749	search_window_map::iterator sw4win = swm.find(ed4w);
1750	if (sw4win == swm.end()) swm[ed4w] = new search_windows;
1751	}
1752
1753	SmartPtr<search_windows> sw = swm[ed4w];
1754	AW_window_simple *aws = sw->windows[type];
1755	if (!aws) {
1756	SearchAwarList *awarList = &awar_list[type];
1757
1758	sw->windows[type] = aws = new AW_window_simple;
1759
1760	aws_init_localized(aww->get_root(), aws, "%s_search_%i", "%s Search", ED4_SearchPositionTypeId[type], ed4w->id);
1761	aws->load_xfig("edit4/search.fig");
1762
1763	aws->at("close");
1764	aws->callback((AW_CB0)AW_POPDOWN);
1765	aws->create_button("CLOSE", "CLOSE", "C");
1766
1767	aws->at("help");
1768	aws->callback(makeHelpCallback("e4_search.hlp"));
1769	aws->create_button("HELP", "HELP", "H");
1770
1771	LoadSaveSearchParam *param = new LoadSaveSearchParam(type, ed4w->id); // bound to callbacks (dont free)
1772
1773	aws->at("load");
1774	aws->callback(AW_POPUP, (AW_CL)load_search_parameters, (AW_CL)param);
1775	aws->create_button("LOAD", "LOAD", "L");
1776
1777	aws->at("save");
1778	aws->callback(AW_POPUP, (AW_CL)save_search_parameters, (AW_CL)param);
1779	aws->create_button("SAVE", "SAVE", "S");
1780
1781	aws->at("next");
1782	aws->callback(ED4_search_cb, (AW_CL)ED4_encodeSearchDescriptor(+1, type), (AW_CL)ed4w);
1783	aws->create_button("SEARCH_NEXT", "#edit/next.xpm", "N");
1784
1785	aws->at("previous");
1786	aws->callback(ED4_search_cb, (AW_CL)ED4_encodeSearchDescriptor(-1, type), (AW_CL)ed4w);
1787	aws->create_button("SEARCH_LAST", "#edit/last.xpm", "L");
1788
1789	aws->at("mark");
1790	aws->callback(ED4_mark_matching_species, (AW_CL)type);
1791	aws->create_autosize_button("MARK_SPECIES", "Mark species with matches", "M");
1792
1793	aws->at("show");
1794	aws->create_toggle(awarList->show);
1795
1796	aws->at("open");
1797	aws->create_toggle(awarList->openFolded);
1798
1799	aws->at("jump");
1800	aws->create_toggle(awarList->autoJump);
1801
1802	aws->at("pattern");
1803	aws->create_text_field(awarList->pattern, 28, 17);
1804
1805	aws->at("minmis");
1806	aws->create_toggle_field(awarList->min_mismatches, 1);
1807	aws->insert_default_toggle("0", "0", 0);
1808	aws->insert_toggle("1", "1", 1);
1809	aws->insert_toggle("2", "2", 2);
1810	aws->update_toggle_field();
1811
1812	aws->at("maxmis");
1813	aws->create_toggle_field(awarList->max_mismatches, 1);
1814	aws->insert_default_toggle("0", "0", 0);
1815	aws->insert_toggle("1", "1", 1);
1816	aws->insert_toggle("2", "2", 2);
1817	aws->insert_toggle("3", "3", 3);
1818	aws->insert_toggle("4", "4", 4);
1819	aws->insert_toggle("5", "5", 5);
1820	aws->update_toggle_field();
1821
1822	aws->at("seq_gaps");
1823	aws->create_toggle(awarList->seq_gaps);
1824	aws->at("pat_gaps");
1825	aws->create_toggle(awarList->pat_gaps);
1826	aws->at("tu");
1827	aws->create_toggle(awarList->tu);
1828	aws->at("case");
1829	aws->create_toggle(awarList->case_sensitive);
1830	aws->at("reverse");
1831	aws->create_toggle(awarList->reverse);
1832	aws->at("complement");
1833	aws->create_toggle(awarList->complement);
1834	aws->at("exact");
1835	aws->create_toggle(awarList->exact);
1836
1837	aws->at("config");
1838	AWT_insert_config_manager(aws, AW_ROOT_DEFAULT, "search", search_store_config, search_restore_config, (AW_CL)type, 0);
1839
1840	}
1841
1842	aws->activate();
1843	}
1844
1845
1846

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

Download in other formats: