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 | static int result_counter = 0; |
---|
35 | static int ignore_more_results = false; |
---|
36 | |
---|
37 | const char *ED4_SearchPositionTypeId[SEARCH_PATTERNS+1] = |
---|
38 | { |
---|
39 | "User1", "User2", |
---|
40 | "Probe", |
---|
41 | "Primer (local)", "Primer (region)", "Primer (global)", |
---|
42 | "Signature (local)", "Signature (region)", "Signature (global)", |
---|
43 | "Any" |
---|
44 | }; |
---|
45 | |
---|
46 | // -------------------------------------------------------------------------------- |
---|
47 | |
---|
48 | struct SearchAwarList { // contains names of awars |
---|
49 | const char *pattern, |
---|
50 | *min_mismatches, |
---|
51 | *max_mismatches, |
---|
52 | *case_sensitive, |
---|
53 | *tu, |
---|
54 | *pat_gaps, |
---|
55 | *seq_gaps, |
---|
56 | *reverse, |
---|
57 | *complement, |
---|
58 | *exact, |
---|
59 | *show, |
---|
60 | *openFolded, |
---|
61 | *autoJump; |
---|
62 | }; |
---|
63 | |
---|
64 | class SearchSettings : virtual Noncopyable { |
---|
65 | char *pattern; |
---|
66 | int min_mismatches; |
---|
67 | int max_mismatches; |
---|
68 | ED4_SEARCH_CASE case_sensitive; |
---|
69 | ED4_SEARCH_TU tu; |
---|
70 | ED4_SEARCH_GAPS pat_gaps; |
---|
71 | ED4_SEARCH_GAPS seq_gaps; |
---|
72 | int reverse; |
---|
73 | int complement; |
---|
74 | int exact; |
---|
75 | int open_folded; |
---|
76 | int autoJump; |
---|
77 | |
---|
78 | public: |
---|
79 | |
---|
80 | void update(SearchAwarList *awarList) { |
---|
81 | AW_root *root = ED4_ROOT->aw_root; |
---|
82 | |
---|
83 | freeset(pattern, root->awar(awarList->pattern)->read_string()); |
---|
84 | |
---|
85 | min_mismatches = root->awar(awarList->min_mismatches)->read_int(); |
---|
86 | max_mismatches = root->awar(awarList->max_mismatches)->read_int(); |
---|
87 | case_sensitive = ED4_SEARCH_CASE(root->awar(awarList->case_sensitive)->read_int()); |
---|
88 | tu = ED4_SEARCH_TU(root->awar(awarList->tu)->read_int()); |
---|
89 | pat_gaps = ED4_SEARCH_GAPS(root->awar(awarList->pat_gaps)->read_int()); |
---|
90 | seq_gaps = ED4_SEARCH_GAPS(root->awar(awarList->seq_gaps)->read_int()); |
---|
91 | open_folded = root->awar(awarList->openFolded)->read_int(); |
---|
92 | autoJump = root->awar(awarList->autoJump)->read_int(); |
---|
93 | reverse = root->awar(awarList->reverse)->read_int(); |
---|
94 | complement = root->awar(awarList->complement)->read_int(); |
---|
95 | exact = root->awar(awarList->exact)->read_int(); |
---|
96 | |
---|
97 | if (complement) { |
---|
98 | if (IS_AMINO()) { |
---|
99 | complement = 0; |
---|
100 | root->awar(awarList->complement)->write_int(0); |
---|
101 | aw_message("Search for complement is not supported for this alignment type (has been disabled)"); |
---|
102 | } |
---|
103 | } |
---|
104 | } |
---|
105 | |
---|
106 | SearchSettings(SearchAwarList *awarList) { pattern = 0; update(awarList); } |
---|
107 | ~SearchSettings() { delete pattern; } |
---|
108 | |
---|
109 | GB_CSTR get_pattern() const { return pattern; } |
---|
110 | int get_min_mismatches() const { return min_mismatches; } |
---|
111 | int get_max_mismatches() const { return max_mismatches; } |
---|
112 | ED4_SEARCH_CASE get_case_sensitive() const { return case_sensitive; } |
---|
113 | ED4_SEARCH_TU get_tu() const { return tu; } |
---|
114 | ED4_SEARCH_GAPS get_pat_gaps() const { return pat_gaps; } |
---|
115 | ED4_SEARCH_GAPS get_seq_gaps() const { return seq_gaps; } |
---|
116 | int get_open_folded() const { return open_folded; } |
---|
117 | int get_autoJump() const { return autoJump; } |
---|
118 | int get_reverse() const { return reverse; } |
---|
119 | int get_complement() const { return complement; } |
---|
120 | int get_exact() const { return exact; } |
---|
121 | }; |
---|
122 | |
---|
123 | // -------------------------------------------------------------------------------- |
---|
124 | |
---|
125 | class SearchTree; |
---|
126 | typedef void (*reportMatch)(int start, int end, GB_CSTR comment, int mismatches[MAX_MISMATCHES]); |
---|
127 | |
---|
128 | class SearchTreeNode : virtual Noncopyable { |
---|
129 | char c; // character |
---|
130 | SearchTreeNode *son; // son != 0 (exception: FOUND) |
---|
131 | SearchTreeNode *brother; |
---|
132 | char *comment; // 0 or comment given in search pattern |
---|
133 | |
---|
134 | static SearchTreeNode FOUND; |
---|
135 | static int start_offset; |
---|
136 | static reportMatch report; |
---|
137 | static int min_mismatches; |
---|
138 | static int max_mismatches; |
---|
139 | static int *uni2real; // transform unified positions to real sequence positions |
---|
140 | |
---|
141 | public: |
---|
142 | |
---|
143 | SearchTreeNode(GB_CSTR pattern, GB_CSTR comment); |
---|
144 | ~SearchTreeNode(); |
---|
145 | |
---|
146 | SearchTreeNode *insert_unified_pattern(GB_CSTR pattern, GB_CSTR pattern_comment); |
---|
147 | void findMatches(int off, GB_CSTR seq, int len, int mismatches, int mismatch_list[MAX_MISMATCHES]); |
---|
148 | |
---|
149 | // you must call the following functions before calling findMatches(): |
---|
150 | |
---|
151 | static void set_start_offset(int off) { start_offset = off; } |
---|
152 | static void set_report(reportMatch r, int *u2r) { report = r; uni2real = u2r; } |
---|
153 | static void set_mismatches(int minMis, int maxMis) |
---|
154 | { |
---|
155 | e4_assert(maxMis <= MAX_MISMATCHES); |
---|
156 | e4_assert(minMis <= maxMis); |
---|
157 | min_mismatches = minMis; |
---|
158 | max_mismatches = maxMis; |
---|
159 | } |
---|
160 | }; |
---|
161 | |
---|
162 | // -------------------------------------------------------------------------------- |
---|
163 | |
---|
164 | SearchTreeNode SearchTreeNode::FOUND(0, 0); |
---|
165 | int SearchTreeNode::start_offset; |
---|
166 | reportMatch SearchTreeNode::report; |
---|
167 | int SearchTreeNode::min_mismatches; |
---|
168 | int SearchTreeNode::max_mismatches; |
---|
169 | int *SearchTreeNode::uni2real; |
---|
170 | |
---|
171 | SearchTreeNode::SearchTreeNode(GB_CSTR pattern, GB_CSTR pattern_comment) |
---|
172 | { |
---|
173 | comment = 0; |
---|
174 | if (pattern) { |
---|
175 | e4_assert(pattern[0]); |
---|
176 | c = pattern[0]; |
---|
177 | if (pattern[1]) { |
---|
178 | son = new SearchTreeNode(pattern+1, pattern_comment); |
---|
179 | } |
---|
180 | else { |
---|
181 | son = &FOUND; |
---|
182 | comment = nulldup(pattern_comment); |
---|
183 | } |
---|
184 | } |
---|
185 | else { |
---|
186 | e4_assert(this==&FOUND); |
---|
187 | c = 0; |
---|
188 | son = 0; |
---|
189 | comment = nulldup(pattern_comment); |
---|
190 | } |
---|
191 | brother = 0; |
---|
192 | } |
---|
193 | |
---|
194 | SearchTreeNode::~SearchTreeNode() |
---|
195 | { |
---|
196 | if (brother!=&FOUND) delete brother; |
---|
197 | if (son!=&FOUND) delete son; |
---|
198 | delete comment; |
---|
199 | } |
---|
200 | |
---|
201 | |
---|
202 | SearchTreeNode *SearchTreeNode::insert_unified_pattern(GB_CSTR pattern, GB_CSTR pattern_comment) |
---|
203 | { |
---|
204 | if (!this) { |
---|
205 | if (pattern[0]) { |
---|
206 | return new SearchTreeNode(pattern, pattern_comment); |
---|
207 | } |
---|
208 | |
---|
209 | return &FOUND; |
---|
210 | } |
---|
211 | |
---|
212 | if (this==&FOUND) { |
---|
213 | if (pattern[0]) { |
---|
214 | SearchTreeNode *neu = new SearchTreeNode(pattern, pattern_comment); |
---|
215 | |
---|
216 | neu->brother = &FOUND; |
---|
217 | return neu; |
---|
218 | } |
---|
219 | return &FOUND; |
---|
220 | } |
---|
221 | |
---|
222 | e4_assert(c); |
---|
223 | |
---|
224 | if (pattern[0]) { // pattern contains sth. |
---|
225 | if (c==pattern[0]) { |
---|
226 | e4_assert(son); |
---|
227 | son = son->insert_unified_pattern(pattern+1, pattern_comment); |
---|
228 | } |
---|
229 | else { |
---|
230 | if (brother) { |
---|
231 | brother = brother->insert_unified_pattern(pattern, pattern_comment); |
---|
232 | } |
---|
233 | else { |
---|
234 | brother = new SearchTreeNode(pattern, pattern_comment); |
---|
235 | } |
---|
236 | } |
---|
237 | } |
---|
238 | else { // empty pattern -> insert FOUND |
---|
239 | if (brother) { |
---|
240 | if (brother!=&FOUND) { |
---|
241 | brother = brother->insert_unified_pattern(pattern, pattern_comment); |
---|
242 | } |
---|
243 | } |
---|
244 | else { |
---|
245 | brother = &FOUND; |
---|
246 | } |
---|
247 | } |
---|
248 | |
---|
249 | return this; |
---|
250 | } |
---|
251 | |
---|
252 | void SearchTreeNode::findMatches(int off, GB_CSTR seq, int len, int mismatches, int mismatch_list[MAX_MISMATCHES]) |
---|
253 | { |
---|
254 | e4_assert(mismatches <= MAX_MISMATCHES); |
---|
255 | if (len) { |
---|
256 | int matches = c=='?' || c==seq[0]; |
---|
257 | int use_mismatch = 0; |
---|
258 | |
---|
259 | if (!matches && mismatches<max_mismatches) { |
---|
260 | int c_is_gap = c=='-' || c=='.'; |
---|
261 | int seq_is_gap = seq[0]=='-' || seq[0]=='.'; |
---|
262 | |
---|
263 | if (c_is_gap==seq_is_gap) { |
---|
264 | mismatch_list[mismatches] = uni2real[off]; |
---|
265 | mismatches++; |
---|
266 | use_mismatch = 1; |
---|
267 | matches = 1; |
---|
268 | } |
---|
269 | } |
---|
270 | |
---|
271 | if (matches) { |
---|
272 | if (son==&FOUND) { |
---|
273 | if (mismatches>=min_mismatches) { |
---|
274 | report(uni2real[start_offset], uni2real[off], comment, mismatch_list); |
---|
275 | } |
---|
276 | } |
---|
277 | else { |
---|
278 | son->findMatches(off+1, seq+1, len-1, mismatches, mismatch_list); |
---|
279 | } |
---|
280 | |
---|
281 | if (use_mismatch) { |
---|
282 | mismatches--; |
---|
283 | mismatch_list[mismatches] = -1; |
---|
284 | } |
---|
285 | } |
---|
286 | } |
---|
287 | |
---|
288 | if (brother==&FOUND) { |
---|
289 | if (mismatches>=min_mismatches) { |
---|
290 | report(uni2real[start_offset], uni2real[off-1], comment, mismatch_list); |
---|
291 | } |
---|
292 | } |
---|
293 | else if (brother) { |
---|
294 | brother->findMatches(off, seq, len, mismatches, mismatch_list); |
---|
295 | } |
---|
296 | } |
---|
297 | |
---|
298 | // -------------------------------------------------------------------------------- |
---|
299 | |
---|
300 | class SearchTree : virtual Noncopyable { |
---|
301 | const SearchSettings *sett; |
---|
302 | SearchTreeNode *root; |
---|
303 | unsigned char unified[256]; |
---|
304 | int shortestPattern; |
---|
305 | |
---|
306 | static char unify_char(char c, int case_sensitive, int T_equal_U); |
---|
307 | |
---|
308 | char *unify_str(const char *data, int len, ED4_SEARCH_GAPS gaps, int *new_len, int **uni2real); |
---|
309 | char *unify_pattern(const char *pattern, int *new_len); |
---|
310 | char *unify_sequence(const char *sequence, int len, int *new_len, int **uni2real); |
---|
311 | |
---|
312 | SearchTree(const SearchTree &); // forbidden |
---|
313 | |
---|
314 | public: |
---|
315 | |
---|
316 | SearchTree(const SearchSettings *s); |
---|
317 | ~SearchTree(); |
---|
318 | |
---|
319 | void findMatches(const char *seq, int len, reportMatch report); |
---|
320 | int get_shortestPattern() const { return shortestPattern; } |
---|
321 | }; |
---|
322 | |
---|
323 | // -------------------------------------------------------------------------------- |
---|
324 | |
---|
325 | static char *shortenString(char *s) |
---|
326 | { |
---|
327 | char *end = strchr(s, '\0'); |
---|
328 | |
---|
329 | while (end>s && isspace(end[-1])) { |
---|
330 | *--end = 0; |
---|
331 | } |
---|
332 | while (isspace(s[0])) { |
---|
333 | s++; |
---|
334 | } |
---|
335 | |
---|
336 | return s; |
---|
337 | } |
---|
338 | |
---|
339 | static void splitTokComment(char **tok, char **commentP) |
---|
340 | { |
---|
341 | char *num = strchr(*tok, '#'); |
---|
342 | |
---|
343 | if (num) { |
---|
344 | num[0] = 0; |
---|
345 | *commentP = shortenString(num+1); |
---|
346 | } |
---|
347 | else { |
---|
348 | *commentP = 0; |
---|
349 | } |
---|
350 | |
---|
351 | *tok = shortenString(*tok); |
---|
352 | } |
---|
353 | |
---|
354 | static char *appendComment(const char *s1, int l1, const char *s2) { |
---|
355 | if (s1) { |
---|
356 | int l2 = strlen(s2); |
---|
357 | char *s = (char*)malloc(l1+1+l2+1); |
---|
358 | |
---|
359 | sprintf(s, "%s %s", s1, s2); |
---|
360 | return s; |
---|
361 | } |
---|
362 | |
---|
363 | return 0; |
---|
364 | } |
---|
365 | |
---|
366 | SearchTree::SearchTree(const SearchSettings *s) |
---|
367 | { |
---|
368 | sett = s; |
---|
369 | root = 0; |
---|
370 | shortestPattern = INT_MAX; |
---|
371 | |
---|
372 | { |
---|
373 | int i; |
---|
374 | int case_sensitive = (sett->get_case_sensitive()==ED4_SC_CASE_SENSITIVE); |
---|
375 | int T_equal_U = (sett->get_tu()==ED4_ST_T_EQUAL_U); |
---|
376 | |
---|
377 | for (i=0; i<256; i++) { |
---|
378 | unified[i] = unify_char(i, case_sensitive, T_equal_U); |
---|
379 | } |
---|
380 | } |
---|
381 | |
---|
382 | #define INSERT_ROOT_PATTERN(tok, com) \ |
---|
383 | do { \ |
---|
384 | if (root) { \ |
---|
385 | root = root->insert_unified_pattern(tok, com); \ |
---|
386 | } \ |
---|
387 | else { \ |
---|
388 | root = new SearchTreeNode(tok, com); \ |
---|
389 | } \ |
---|
390 | } while (0) |
---|
391 | |
---|
392 | |
---|
393 | { |
---|
394 | char *pattern = strdup(sett->get_pattern()); |
---|
395 | const char *trenner = "\n,"; |
---|
396 | char *tok = strtok(pattern, trenner); |
---|
397 | char *comment; |
---|
398 | char T_or_U; |
---|
399 | GB_ERROR T_or_U_error = GBT_determine_T_or_U(ED4_ROOT->alignment_type, &T_or_U, "complement"); |
---|
400 | bool show_T_or_U_error = false; |
---|
401 | |
---|
402 | while (tok) { |
---|
403 | splitTokComment(&tok, &comment); |
---|
404 | int uni_tok_len; |
---|
405 | char *uni_tok = unify_pattern(tok, &uni_tok_len); |
---|
406 | |
---|
407 | if (uni_tok[0]) { |
---|
408 | int s_exact = sett->get_exact(); |
---|
409 | int s_reverse = sett->get_reverse(); |
---|
410 | int s_complement = sett->get_complement(); |
---|
411 | |
---|
412 | if (uni_tok_len<shortestPattern) { |
---|
413 | shortestPattern = uni_tok_len; |
---|
414 | } |
---|
415 | |
---|
416 | if (!s_exact || (!s_reverse && !s_complement)) { |
---|
417 | // insert original search pattern if all patterns shall be used (!s_exact) |
---|
418 | // or if neither s_reverse nor s_complement |
---|
419 | INSERT_ROOT_PATTERN(uni_tok, comment); |
---|
420 | } |
---|
421 | int commentLen = comment ? strlen(comment) : 0; |
---|
422 | |
---|
423 | if (s_reverse) { |
---|
424 | char *reverse = GBT_reverseNucSequence(uni_tok, uni_tok_len); |
---|
425 | |
---|
426 | if (!s_exact || !s_complement) { |
---|
427 | char *reverseComment = appendComment(comment, commentLen, "(reverse)"); |
---|
428 | |
---|
429 | INSERT_ROOT_PATTERN(reverse, reverseComment); // insert reverse pattern |
---|
430 | free(reverseComment); |
---|
431 | } |
---|
432 | if (s_complement) { |
---|
433 | e4_assert(IS_NUCLEOTIDE()); |
---|
434 | if (T_or_U) { |
---|
435 | char *revcomp = GBT_complementNucSequence(reverse, uni_tok_len, T_or_U); |
---|
436 | char *revcompComment = appendComment(comment, commentLen, "(reverse complement)"); |
---|
437 | char *uni_revcomp = unify_pattern(revcomp, 0); |
---|
438 | |
---|
439 | INSERT_ROOT_PATTERN(uni_revcomp, revcompComment); // insert reverse complement pattern |
---|
440 | |
---|
441 | free(uni_revcomp); |
---|
442 | free(revcompComment); |
---|
443 | free(revcomp); |
---|
444 | } |
---|
445 | else { |
---|
446 | show_T_or_U_error = true; // only show error if it matters |
---|
447 | } |
---|
448 | } |
---|
449 | |
---|
450 | free(reverse); |
---|
451 | } |
---|
452 | |
---|
453 | if (s_complement) { |
---|
454 | GB_transaction ta(GLOBAL_gb_main); |
---|
455 | |
---|
456 | if (T_or_U) { |
---|
457 | if (!s_exact || !s_reverse) { |
---|
458 | char *complement = GBT_complementNucSequence(uni_tok, uni_tok_len, T_or_U); |
---|
459 | char *complementComment = appendComment(comment, commentLen, "(complement)"); |
---|
460 | char *uni_complement = unify_pattern(complement, 0); |
---|
461 | |
---|
462 | INSERT_ROOT_PATTERN(uni_complement, complementComment); // insert complement pattern |
---|
463 | |
---|
464 | free(uni_complement); |
---|
465 | free(complementComment); |
---|
466 | free(complement); |
---|
467 | } |
---|
468 | } |
---|
469 | else { |
---|
470 | show_T_or_U_error = true; // only show error if it matters |
---|
471 | } |
---|
472 | } |
---|
473 | |
---|
474 | } |
---|
475 | |
---|
476 | tok = strtok(0, trenner); |
---|
477 | free(uni_tok); |
---|
478 | } |
---|
479 | free(pattern); |
---|
480 | |
---|
481 | if (show_T_or_U_error && T_or_U_error) aw_message(T_or_U_error); |
---|
482 | } |
---|
483 | |
---|
484 | #undef INSERT_ROOT_PATTERN |
---|
485 | } |
---|
486 | |
---|
487 | SearchTree::~SearchTree() |
---|
488 | { |
---|
489 | delete root; |
---|
490 | } |
---|
491 | |
---|
492 | char SearchTree::unify_char(char c, int case_sensitive, int T_equal_U) |
---|
493 | { |
---|
494 | if (!case_sensitive) { |
---|
495 | c = tolower(c); |
---|
496 | } |
---|
497 | |
---|
498 | if (T_equal_U) { |
---|
499 | if (c=='t') { |
---|
500 | c = 'u'; |
---|
501 | } |
---|
502 | else if (c=='T') { |
---|
503 | c = 'U'; |
---|
504 | } |
---|
505 | } |
---|
506 | |
---|
507 | if (c=='.') { |
---|
508 | c = '-'; |
---|
509 | } |
---|
510 | |
---|
511 | return c; |
---|
512 | } |
---|
513 | |
---|
514 | char *SearchTree::unify_str(const char *data, int len, ED4_SEARCH_GAPS gaps, int *new_len, int **uni2real) |
---|
515 | { |
---|
516 | char *p = (char*)malloc(len+1); |
---|
517 | |
---|
518 | if (!p) { |
---|
519 | return 0; |
---|
520 | } |
---|
521 | |
---|
522 | char *pp = p; |
---|
523 | int nlen = 0; |
---|
524 | int realPos = 0; |
---|
525 | |
---|
526 | if (uni2real) { |
---|
527 | int *u2r = *uni2real; |
---|
528 | |
---|
529 | if (gaps==ED4_SG_CONSIDER_GAPS) { |
---|
530 | while (realPos<len) { |
---|
531 | *pp++ = unified[(unsigned char)data[realPos]]; |
---|
532 | u2r[nlen++] = realPos++; |
---|
533 | } |
---|
534 | } |
---|
535 | else { |
---|
536 | while (realPos<len) { |
---|
537 | unsigned char c = data[realPos]; |
---|
538 | |
---|
539 | if (c!='-' && c!='.') { |
---|
540 | *pp++ = unified[c]; |
---|
541 | u2r[nlen++] = realPos; |
---|
542 | } |
---|
543 | realPos++; |
---|
544 | } |
---|
545 | } |
---|
546 | } |
---|
547 | else { |
---|
548 | if (gaps==ED4_SG_CONSIDER_GAPS) { |
---|
549 | while (realPos<len) { |
---|
550 | *pp++ = unified[(unsigned char)data[realPos++]]; |
---|
551 | nlen++; |
---|
552 | } |
---|
553 | } |
---|
554 | else { |
---|
555 | while (realPos<len) { |
---|
556 | unsigned char c = data[realPos++]; |
---|
557 | |
---|
558 | if (c!='-' && c!='.') { |
---|
559 | *pp++ = unified[c]; |
---|
560 | nlen++; |
---|
561 | } |
---|
562 | } |
---|
563 | } |
---|
564 | } |
---|
565 | |
---|
566 | // --------------- |
---|
567 | |
---|
568 | pp[0] = 0; |
---|
569 | if (new_len) { |
---|
570 | *new_len = nlen; |
---|
571 | } |
---|
572 | return p; |
---|
573 | } |
---|
574 | |
---|
575 | char *SearchTree::unify_pattern(const char *pattern, int *new_len) |
---|
576 | { |
---|
577 | int len = strlen(pattern); |
---|
578 | return unify_str(pattern, len, sett->get_pat_gaps(), new_len, 0); |
---|
579 | } |
---|
580 | |
---|
581 | char *SearchTree::unify_sequence(const char *sequence, int len, int *new_len, int **uni2real) |
---|
582 | { |
---|
583 | return unify_str(sequence, len, sett->get_seq_gaps(), new_len, uni2real); |
---|
584 | } |
---|
585 | |
---|
586 | void SearchTree::findMatches(const char *seq, int len, reportMatch report) |
---|
587 | { |
---|
588 | if (root) { |
---|
589 | int new_len; |
---|
590 | int *uni2real = (int*)malloc(len*sizeof(int)); |
---|
591 | char *uni_seq = uni2real ? unify_sequence(seq, len, &new_len, &uni2real) : NULL; |
---|
592 | |
---|
593 | if (uni_seq) { |
---|
594 | int off; |
---|
595 | char *useq = uni_seq; |
---|
596 | int mismatch_list[MAX_MISMATCHES]; |
---|
597 | |
---|
598 | for (off=0; off<MAX_MISMATCHES; off++) { |
---|
599 | mismatch_list[off] = -1; |
---|
600 | } |
---|
601 | |
---|
602 | SearchTreeNode::set_report(report, uni2real); |
---|
603 | SearchTreeNode::set_mismatches(sett->get_min_mismatches(), sett->get_max_mismatches()); |
---|
604 | |
---|
605 | for (off=0; off<new_len && !ignore_more_results; off++, useq++) { |
---|
606 | SearchTreeNode::set_start_offset(off); |
---|
607 | root->findMatches(off, useq, new_len-off, 0, mismatch_list); |
---|
608 | } |
---|
609 | |
---|
610 | free(uni_seq); |
---|
611 | free(uni2real); |
---|
612 | } |
---|
613 | else { |
---|
614 | aw_message("Out of swapspace?"); |
---|
615 | if (uni2real) free(uni2real); |
---|
616 | } |
---|
617 | } |
---|
618 | } |
---|
619 | |
---|
620 | // -------------------------------------------------------------------------------- |
---|
621 | |
---|
622 | #define AWAR_NAME(t, s) ED4_AWAR_##t##_SEARCH_##s |
---|
623 | |
---|
624 | #define AWAR_LIST(t) \ |
---|
625 | AWAR_NAME(t, PATTERN), \ |
---|
626 | AWAR_NAME(t, MIN_MISMATCHES), \ |
---|
627 | AWAR_NAME(t, MAX_MISMATCHES), \ |
---|
628 | AWAR_NAME(t, CASE), \ |
---|
629 | AWAR_NAME(t, TU), \ |
---|
630 | AWAR_NAME(t, PAT_GAPS), \ |
---|
631 | AWAR_NAME(t, SEQ_GAPS), \ |
---|
632 | AWAR_NAME(t, REVERSE), \ |
---|
633 | AWAR_NAME(t, COMPLEMENT), \ |
---|
634 | AWAR_NAME(t, EXACT), \ |
---|
635 | AWAR_NAME(t, SHOW), \ |
---|
636 | AWAR_NAME(t, OPEN_FOLDED), \ |
---|
637 | AWAR_NAME(t, AUTO_JUMP) |
---|
638 | |
---|
639 | static struct SearchAwarList awar_list[SEARCH_PATTERNS] = { |
---|
640 | { AWAR_LIST(USER1) }, |
---|
641 | { AWAR_LIST(USER2) }, |
---|
642 | { AWAR_LIST(PROBE) }, |
---|
643 | { AWAR_LIST(PRIMER1) }, |
---|
644 | { AWAR_LIST(PRIMER2) }, |
---|
645 | { AWAR_LIST(PRIMER3) }, |
---|
646 | { AWAR_LIST(SIG1) }, |
---|
647 | { AWAR_LIST(SIG2) }, |
---|
648 | { AWAR_LIST(SIG3) }, |
---|
649 | }; |
---|
650 | |
---|
651 | static inline int resultsAreShown(ED4_SearchPositionType type) { |
---|
652 | return ED4_ROOT->aw_root->awar(awar_list[type].show)->read_int(); |
---|
653 | } |
---|
654 | |
---|
655 | enum search_params_changed_action { |
---|
656 | REFRESH_IF_SHOWN = 1, |
---|
657 | REFRESH_ALWAYS = 2, |
---|
658 | RECALC_SEARCH_TREE = 4, |
---|
659 | TEST_MIN_MISMATCH = 8, |
---|
660 | TEST_MAX_MISMATCH = 16, |
---|
661 | DO_AUTO_JUMP = 32 |
---|
662 | }; |
---|
663 | |
---|
664 | // -------------------------------------------------------------------------------- |
---|
665 | |
---|
666 | static SearchSettings *settings[SEARCH_PATTERNS]; // last searched settings for each type |
---|
667 | static SearchTree *tree[SEARCH_PATTERNS]; // Search trees for each type |
---|
668 | |
---|
669 | // -------------------------------------------------------------------------------- |
---|
670 | |
---|
671 | static void searchParamsChanged(AW_root *root, AW_CL cl_type, AW_CL cl_action) |
---|
672 | { |
---|
673 | ED4_SearchPositionType type = ED4_SearchPositionType(cl_type); |
---|
674 | enum search_params_changed_action action = (enum search_params_changed_action)cl_action; |
---|
675 | |
---|
676 | result_counter = 0; |
---|
677 | ignore_more_results = false; |
---|
678 | |
---|
679 | // check awar values |
---|
680 | |
---|
681 | if (action & (TEST_MIN_MISMATCH|TEST_MAX_MISMATCH)) { |
---|
682 | int mimi = root->awar(awar_list[type].min_mismatches)->read_int(); |
---|
683 | int mami = root->awar(awar_list[type].max_mismatches)->read_int(); |
---|
684 | |
---|
685 | if (mimi>mami) { |
---|
686 | if (action & TEST_MIN_MISMATCH) { // max has changed |
---|
687 | root->awar(awar_list[type].min_mismatches)->write_int(mami); |
---|
688 | } |
---|
689 | if (action & TEST_MAX_MISMATCH) { // min has changed |
---|
690 | root->awar(awar_list[type].max_mismatches)->write_int(mimi); |
---|
691 | } |
---|
692 | } |
---|
693 | } |
---|
694 | |
---|
695 | // init new search |
---|
696 | |
---|
697 | recalc : |
---|
698 | ED4_SearchResults::setNewSearch(type); |
---|
699 | if (!settings[type]) return; |
---|
700 | settings[type]->update(&awar_list[type]); |
---|
701 | |
---|
702 | if (action & RECALC_SEARCH_TREE) { |
---|
703 | delete tree[type]; |
---|
704 | tree[type] = new SearchTree(settings[type]); |
---|
705 | } |
---|
706 | |
---|
707 | if (action & (RECALC_SEARCH_TREE|TEST_MIN_MISMATCH|TEST_MAX_MISMATCH)) { |
---|
708 | int patLen = tree[type]->get_shortestPattern(); |
---|
709 | int maxMis = root->awar(awar_list[type].max_mismatches)->read_int(); |
---|
710 | |
---|
711 | if (patLen < 3*maxMis) { |
---|
712 | int maxMaxMis = tree[type]->get_shortestPattern()/3; |
---|
713 | |
---|
714 | aw_message(GBS_global_string("Too many mismatches (patterns of length=%i allow only %i max. mismatches)", patLen, maxMaxMis)); |
---|
715 | |
---|
716 | root->awar(awar_list[type].max_mismatches)->write_int(maxMaxMis); |
---|
717 | if (root->awar(awar_list[type].min_mismatches)->read_int() > maxMaxMis) { |
---|
718 | root->awar(awar_list[type].min_mismatches)->write_int(maxMaxMis); |
---|
719 | } |
---|
720 | goto recalc; |
---|
721 | } |
---|
722 | } |
---|
723 | |
---|
724 | if (action & REFRESH_IF_SHOWN) { |
---|
725 | if (resultsAreShown(type)) { |
---|
726 | action = (enum search_params_changed_action)(action|REFRESH_ALWAYS); |
---|
727 | } |
---|
728 | } |
---|
729 | |
---|
730 | if (settings[type]->get_autoJump() && (action & DO_AUTO_JUMP)) { // auto jump |
---|
731 | for (ED4_window *win = ED4_ROOT->first_window; win; win = win->next) { |
---|
732 | ED4_LocalWinContext uses(win); |
---|
733 | |
---|
734 | ED4_cursor *cursor = ¤t_cursor(); |
---|
735 | bool jumped = false; |
---|
736 | |
---|
737 | if (cursor->owner_of_cursor && cursor->owner_of_cursor->is_sequence_terminal()) { |
---|
738 | int pos = cursor->get_sequence_pos(); |
---|
739 | ED4_sequence_terminal *seq_term = cursor->owner_of_cursor->to_sequence_terminal(); |
---|
740 | ED4_SearchResults *result = &seq_term->results(); |
---|
741 | |
---|
742 | result->search(seq_term); |
---|
743 | ED4_SearchPosition *found = result->get_last_starting_before(type, pos+1, 0); |
---|
744 | int bestPos = -1; |
---|
745 | |
---|
746 | if (found) { |
---|
747 | bestPos = found->get_start_pos(); |
---|
748 | } |
---|
749 | |
---|
750 | if (pos>=1) { |
---|
751 | found = result->get_first_starting_after(type, pos-1, 0); |
---|
752 | if (found) { |
---|
753 | int next_pos = found->get_start_pos(); |
---|
754 | |
---|
755 | if (abs(pos-next_pos)<abs(pos-bestPos)) { |
---|
756 | bestPos = next_pos; |
---|
757 | } |
---|
758 | } |
---|
759 | } |
---|
760 | |
---|
761 | if (bestPos!=-1) { |
---|
762 | if (bestPos == pos) { |
---|
763 | jumped = true; // already there |
---|
764 | } |
---|
765 | else { |
---|
766 | jumped = seq_term->setCursorTo(cursor, bestPos, 1, ED4_JUMP_KEEP_POSITION); |
---|
767 | } |
---|
768 | } |
---|
769 | } |
---|
770 | |
---|
771 | if (!jumped) { |
---|
772 | ED4_search_cb(0, ED4_encodeSearchDescriptor(+1, type), (AW_CL)current_ed4w()); |
---|
773 | } |
---|
774 | } |
---|
775 | } |
---|
776 | |
---|
777 | if (action & REFRESH_ALWAYS) ED4_ROOT->request_refresh_for_sequence_terminals(); |
---|
778 | root->awar(ED4_AWAR_SEARCH_RESULT_CHANGED)->touch(); // trigger refresh in SECEDIT |
---|
779 | } |
---|
780 | |
---|
781 | void ED4_create_search_awars(AW_root *root) |
---|
782 | { |
---|
783 | #define cb(action) add_callback(searchParamsChanged, AW_CL(i), AW_CL(action)) |
---|
784 | |
---|
785 | int i; |
---|
786 | for (i=0; i<SEARCH_PATTERNS; i++) { |
---|
787 | root->awar_string(awar_list[i].pattern, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN | RECALC_SEARCH_TREE | DO_AUTO_JUMP); |
---|
788 | 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); |
---|
789 | 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); |
---|
790 | 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); |
---|
791 | root->awar_int(awar_list[i].reverse, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN | RECALC_SEARCH_TREE | DO_AUTO_JUMP); |
---|
792 | root->awar_int(awar_list[i].complement, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN | RECALC_SEARCH_TREE | DO_AUTO_JUMP); |
---|
793 | root->awar_int(awar_list[i].exact, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN | RECALC_SEARCH_TREE | DO_AUTO_JUMP); |
---|
794 | root->awar_int(awar_list[i].min_mismatches, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN | TEST_MAX_MISMATCH | DO_AUTO_JUMP); |
---|
795 | root->awar_int(awar_list[i].max_mismatches, 0, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN | TEST_MIN_MISMATCH | DO_AUTO_JUMP); |
---|
796 | root->awar_int(awar_list[i].seq_gaps, ED4_SG_IGNORE_GAPS, GLOBAL_gb_main) ->cb(REFRESH_IF_SHOWN | DO_AUTO_JUMP); |
---|
797 | root->awar_int(awar_list[i].show, 1, GLOBAL_gb_main) ->cb(REFRESH_ALWAYS); |
---|
798 | root->awar_int(awar_list[i].openFolded, 1, GLOBAL_gb_main) ->cb(0); |
---|
799 | root->awar_int(awar_list[i].autoJump, 1, GLOBAL_gb_main) ->cb(DO_AUTO_JUMP); |
---|
800 | |
---|
801 | settings[i] = new SearchSettings(&awar_list[i]); |
---|
802 | tree[i] = new SearchTree(settings[i]); |
---|
803 | } |
---|
804 | |
---|
805 | root->awar_int(ED4_AWAR_SEARCH_RESULT_CHANGED, 0, GLOBAL_gb_main); |
---|
806 | |
---|
807 | #undef cb |
---|
808 | |
---|
809 | // awars to save/load search parameters: |
---|
810 | { |
---|
811 | char *dir = strdup(GB_path_in_arbprop("search_settings")); |
---|
812 | AW_create_fileselection_awars(root, ED4_SEARCH_SAVE_BASE, dir, ".asp", "noname.asp"); |
---|
813 | root->awar(ED4_SEARCH_SAVE_BASE"/directory")->write_string(dir); |
---|
814 | free(dir); |
---|
815 | } |
---|
816 | } |
---|
817 | |
---|
818 | // -------------------------------------------------------------------------------- |
---|
819 | |
---|
820 | char *ED4_SearchPosition::lastShownComment = 0; |
---|
821 | |
---|
822 | ED4_SearchPosition::ED4_SearchPosition(int sp, int ep, ED4_SearchPositionType wf, GB_CSTR found_comment, int mismatches[MAX_MISMATCHES]) |
---|
823 | { |
---|
824 | e4_assert(sp<=ep && sp>=0); |
---|
825 | start_pos = sp; |
---|
826 | end_pos = ep; |
---|
827 | whatsFound = wf; |
---|
828 | next = 0; |
---|
829 | comment = found_comment; |
---|
830 | memcpy(mismatch, mismatches, sizeof(*mismatch)*MAX_MISMATCHES); |
---|
831 | } |
---|
832 | ED4_SearchPosition::ED4_SearchPosition(const ED4_SearchPosition& other) { |
---|
833 | start_pos = other.start_pos; |
---|
834 | end_pos = other.end_pos; |
---|
835 | whatsFound = other.whatsFound; |
---|
836 | next = 0; |
---|
837 | comment = strdup(other.comment); |
---|
838 | memcpy(mismatch, other.mismatch, sizeof(mismatch[0])*MAX_MISMATCHES); |
---|
839 | } |
---|
840 | |
---|
841 | ED4_SearchPosition *ED4_SearchPosition::insert(ED4_SearchPosition *toAdd) |
---|
842 | { |
---|
843 | ED4_SearchPosition *head = this; |
---|
844 | ED4_SearchPosition *self = this; |
---|
845 | ED4_SearchPosition *last = 0; |
---|
846 | |
---|
847 | while (1) { |
---|
848 | if (toAdd->cmp(*self)<=0) { // add before self |
---|
849 | toAdd->next = self; |
---|
850 | if (last) last->next = toAdd; |
---|
851 | else head = toAdd; |
---|
852 | break; |
---|
853 | } |
---|
854 | if (!self->next) { // add to end of list |
---|
855 | self->next = toAdd; |
---|
856 | break; |
---|
857 | } |
---|
858 | last = self; |
---|
859 | self = self->next; |
---|
860 | } |
---|
861 | |
---|
862 | return head; |
---|
863 | } |
---|
864 | ED4_SearchPosition *ED4_SearchPosition::remove(ED4_SearchPositionType typeToRemove) |
---|
865 | { |
---|
866 | ED4_SearchPosition *head = this; |
---|
867 | ED4_SearchPosition *self = this; |
---|
868 | ED4_SearchPosition *last = 0; |
---|
869 | |
---|
870 | while (self) { |
---|
871 | if (self->whatsFound == typeToRemove) { // remove self |
---|
872 | ED4_SearchPosition **ptrToMe = last ? &(last->next) : &head; |
---|
873 | |
---|
874 | *ptrToMe = self->next; |
---|
875 | self->next = 0; |
---|
876 | delete self; |
---|
877 | self = *ptrToMe; |
---|
878 | } |
---|
879 | else { // do not remove |
---|
880 | last = self; |
---|
881 | self = self->next; |
---|
882 | } |
---|
883 | } |
---|
884 | |
---|
885 | return head; |
---|
886 | } |
---|
887 | |
---|
888 | #ifdef TEST_SEARCH_POSITION |
---|
889 | int ED4_SearchPosition::ok() const |
---|
890 | { |
---|
891 | #ifndef NDEBUG |
---|
892 | if (next) { |
---|
893 | int c = cmp(*next); |
---|
894 | |
---|
895 | if (c>0) { |
---|
896 | printf("ED4_SearchPosition: list not sorted\n"); |
---|
897 | return 0; |
---|
898 | } |
---|
899 | |
---|
900 | return next->ok(); |
---|
901 | } |
---|
902 | #endif |
---|
903 | return 1; |
---|
904 | } |
---|
905 | #endif |
---|
906 | |
---|
907 | GB_CSTR ED4_SearchPosition::get_comment() const |
---|
908 | { |
---|
909 | if (!comment) return 0; |
---|
910 | if (lastShownComment && strcmp(lastShownComment, comment)==0) return 0; // do not show comment twice |
---|
911 | |
---|
912 | delete lastShownComment; |
---|
913 | lastShownComment = strdup(comment); |
---|
914 | return lastShownComment; |
---|
915 | } |
---|
916 | |
---|
917 | ED4_SearchPosition *ED4_SearchPosition::get_next_at(int pos) const |
---|
918 | { |
---|
919 | ED4_SearchPosition *self = this->next; |
---|
920 | |
---|
921 | while (self) { |
---|
922 | if (self->start_pos > pos) break; // none found |
---|
923 | if (self->containsPos(pos)) return self; |
---|
924 | self = self->next; |
---|
925 | } |
---|
926 | return 0; |
---|
927 | } |
---|
928 | |
---|
929 | // -------------------------------------------------------------------------------- |
---|
930 | |
---|
931 | int ED4_SearchResults::initialized = 0; |
---|
932 | int ED4_SearchResults::timeOfLastSearch[SEARCH_PATTERNS]; |
---|
933 | int ED4_SearchResults::timeOfNextSearch[SEARCH_PATTERNS]; |
---|
934 | int ED4_SearchResults::shown[SEARCH_PATTERNS]; |
---|
935 | int ED4_SearchResults::bufferSize; |
---|
936 | char *ED4_SearchResults::buffer; |
---|
937 | |
---|
938 | ED4_SearchResults::ED4_SearchResults() |
---|
939 | { |
---|
940 | if (!initialized) { |
---|
941 | int i; |
---|
942 | |
---|
943 | for (i=0; i<SEARCH_PATTERNS; i++) { |
---|
944 | timeOfLastSearch[i] = 0; |
---|
945 | timeOfNextSearch[i] = 1; |
---|
946 | shown[i] = resultsAreShown(ED4_SearchPositionType(i)); |
---|
947 | } |
---|
948 | |
---|
949 | bufferSize = 100; |
---|
950 | buffer = (char*)GB_calloc(bufferSize, sizeof(char)); |
---|
951 | |
---|
952 | initialized = 1; |
---|
953 | } |
---|
954 | |
---|
955 | arraySize = 0; // list-format |
---|
956 | array = 0; |
---|
957 | first = 0; |
---|
958 | int i; |
---|
959 | for (i=0; i<SEARCH_PATTERNS; i++) { |
---|
960 | timeOf[i] = 0; |
---|
961 | } |
---|
962 | } |
---|
963 | |
---|
964 | ED4_SearchResults::~ED4_SearchResults() { |
---|
965 | delete first; |
---|
966 | free(array); |
---|
967 | } |
---|
968 | |
---|
969 | // -------------------------------------------------------------------------------- |
---|
970 | |
---|
971 | static ED4_SearchResults *reportToResult = 0; |
---|
972 | static ED4_SearchPositionType reportType; |
---|
973 | |
---|
974 | static void reportSearchPosition(int start, int end, GB_CSTR comment, int mismatches[MAX_MISMATCHES]) |
---|
975 | { |
---|
976 | ED4_SearchPosition *pos = new ED4_SearchPosition(start, end, reportType, comment, mismatches); |
---|
977 | reportToResult->addSearchPosition(pos); |
---|
978 | } |
---|
979 | |
---|
980 | // -------------------------------------------------------------------------------- |
---|
981 | |
---|
982 | void ED4_SearchResults::addSearchPosition(ED4_SearchPosition *pos) { |
---|
983 | static int max_allowed_results = 100000; |
---|
984 | |
---|
985 | if (ignore_more_results) return; |
---|
986 | |
---|
987 | if (is_array()) { |
---|
988 | to_list(); |
---|
989 | } |
---|
990 | |
---|
991 | ++result_counter; |
---|
992 | if (result_counter >= max_allowed_results) { |
---|
993 | if (aw_question("many_search_results", |
---|
994 | GBS_global_string("More than %i results found!", result_counter), "Allow more,That's enough") == 0) { |
---|
995 | max_allowed_results = max_allowed_results*2; |
---|
996 | } |
---|
997 | else { |
---|
998 | ignore_more_results = true; |
---|
999 | return; |
---|
1000 | } |
---|
1001 | } |
---|
1002 | |
---|
1003 | if (first) { |
---|
1004 | first = first->insert(pos); |
---|
1005 | #ifdef TEST_SEARCH_POSITION |
---|
1006 | e4_assert(first->ok()); |
---|
1007 | #endif |
---|
1008 | } |
---|
1009 | else { |
---|
1010 | first = pos; |
---|
1011 | } |
---|
1012 | } |
---|
1013 | |
---|
1014 | void ED4_SearchResults::search(const ED4_sequence_terminal *seq_terminal) |
---|
1015 | { |
---|
1016 | int i; |
---|
1017 | int needed[SEARCH_PATTERNS]; |
---|
1018 | int st_needed = 0; |
---|
1019 | |
---|
1020 | for (i=0; i<SEARCH_PATTERNS; i++) { |
---|
1021 | if (timeOf[i]<timeOfNextSearch[i]) { |
---|
1022 | timeOf[i] = timeOfNextSearch[i]; |
---|
1023 | timeOfLastSearch[i] = timeOfNextSearch[i]; |
---|
1024 | needed[i] = 1; |
---|
1025 | st_needed = 1; |
---|
1026 | if (first) { |
---|
1027 | if (is_array()) { |
---|
1028 | to_list(); |
---|
1029 | } |
---|
1030 | first = first->remove(ED4_SearchPositionType(i)); |
---|
1031 | #if defined TEST_SEARCH_POSITION |
---|
1032 | e4_assert(!first || first->ok()); |
---|
1033 | #endif |
---|
1034 | } |
---|
1035 | } |
---|
1036 | else { |
---|
1037 | needed[i] = 0; |
---|
1038 | } |
---|
1039 | } |
---|
1040 | |
---|
1041 | if (st_needed) { |
---|
1042 | int len; |
---|
1043 | char *seq = seq_terminal->resolve_pointer_to_string_copy(&len); |
---|
1044 | |
---|
1045 | if (seq) { |
---|
1046 | reportToResult = this; |
---|
1047 | for (i=0; i<SEARCH_PATTERNS; i++) { |
---|
1048 | reportType = ED4_SearchPositionType(i); |
---|
1049 | if (needed[i]) { |
---|
1050 | if (is_array()) { |
---|
1051 | to_list(); |
---|
1052 | } |
---|
1053 | tree[i]->findMatches(seq, len, reportSearchPosition); |
---|
1054 | } |
---|
1055 | } |
---|
1056 | reportToResult = 0; |
---|
1057 | } |
---|
1058 | free(seq); |
---|
1059 | } |
---|
1060 | } |
---|
1061 | |
---|
1062 | ED4_SearchPosition *ED4_SearchResults::get_first_at(ED4_SearchPositionType type, int start, int end) const { |
---|
1063 | if (is_list()) { |
---|
1064 | const_cast<ED4_SearchResults*>(this)->to_array(); |
---|
1065 | } |
---|
1066 | |
---|
1067 | int l = 0; |
---|
1068 | int h = arraySize-1; |
---|
1069 | |
---|
1070 | ED4_SearchPosition *pos = 0; |
---|
1071 | int m = 0; |
---|
1072 | |
---|
1073 | while (l<=h) { |
---|
1074 | m = (l+h)/2; |
---|
1075 | pos = array[m]; |
---|
1076 | if (pos->get_end_pos()<start) { |
---|
1077 | l = m+1; |
---|
1078 | } |
---|
1079 | else if (pos->get_start_pos()>end) { |
---|
1080 | h = m-1; |
---|
1081 | } |
---|
1082 | else { |
---|
1083 | e4_assert(pos->get_end_pos()>=start && pos->get_start_pos()<=end); |
---|
1084 | break; |
---|
1085 | } |
---|
1086 | } |
---|
1087 | |
---|
1088 | if (l>h) { |
---|
1089 | return 0; |
---|
1090 | } |
---|
1091 | |
---|
1092 | if (pos) { |
---|
1093 | int best_m = m; |
---|
1094 | |
---|
1095 | while (m>0) { |
---|
1096 | m--; |
---|
1097 | pos = array[m]; |
---|
1098 | if (pos->get_end_pos()>=start && pos->get_start_pos()<=end) { |
---|
1099 | best_m = m; |
---|
1100 | } |
---|
1101 | } |
---|
1102 | pos = array[best_m]; |
---|
1103 | |
---|
1104 | while (pos) { |
---|
1105 | if (type==ED4_ANY_PATTERN || pos->get_whatsFound()==type) { |
---|
1106 | break; |
---|
1107 | } |
---|
1108 | pos = pos->get_next(); |
---|
1109 | } |
---|
1110 | |
---|
1111 | return pos; |
---|
1112 | } |
---|
1113 | |
---|
1114 | return 0; |
---|
1115 | } |
---|
1116 | |
---|
1117 | ED4_SearchPosition *ED4_SearchResults::get_first_starting_after(ED4_SearchPositionType type, int pos, int mustBeShown) const |
---|
1118 | { |
---|
1119 | ED4_SearchPosition *sp = first; |
---|
1120 | |
---|
1121 | while (sp) { |
---|
1122 | if (type==ED4_ANY_PATTERN || sp->get_whatsFound()==type) { |
---|
1123 | if (sp->get_start_pos()>pos && (!mustBeShown || resultsAreShown(sp->get_whatsFound()))) { |
---|
1124 | break; |
---|
1125 | } |
---|
1126 | } |
---|
1127 | sp = sp->get_next(); |
---|
1128 | } |
---|
1129 | |
---|
1130 | return sp; |
---|
1131 | } |
---|
1132 | |
---|
1133 | ED4_SearchPosition *ED4_SearchResults::get_last_starting_before(ED4_SearchPositionType type, int pos, int mustBeShown) const |
---|
1134 | { |
---|
1135 | ED4_SearchPosition *sp = first, |
---|
1136 | *best = 0; |
---|
1137 | |
---|
1138 | while (sp) { |
---|
1139 | if (type==ED4_ANY_PATTERN || sp->get_whatsFound()==type) { |
---|
1140 | if (sp->get_start_pos()<pos && (!mustBeShown || resultsAreShown(sp->get_whatsFound()))) { |
---|
1141 | best = sp; |
---|
1142 | } |
---|
1143 | else { |
---|
1144 | break; |
---|
1145 | } |
---|
1146 | } |
---|
1147 | sp = sp->get_next(); |
---|
1148 | } |
---|
1149 | |
---|
1150 | return best; |
---|
1151 | } |
---|
1152 | |
---|
1153 | void ED4_SearchResults::setNewSearch(ED4_SearchPositionType type) |
---|
1154 | { |
---|
1155 | if (type==ED4_ANY_PATTERN) { |
---|
1156 | int i; |
---|
1157 | for (i=0; i<SEARCH_PATTERNS; i++) { |
---|
1158 | setNewSearch(ED4_SearchPositionType(i)); |
---|
1159 | } |
---|
1160 | } |
---|
1161 | else { |
---|
1162 | int next_unused_stamp = timeOfLastSearch[type] + 1; |
---|
1163 | |
---|
1164 | shown[type] = resultsAreShown(type); |
---|
1165 | if (timeOfNextSearch[type]!=next_unused_stamp) { |
---|
1166 | timeOfNextSearch[type] = next_unused_stamp; |
---|
1167 | } |
---|
1168 | } |
---|
1169 | } |
---|
1170 | |
---|
1171 | void ED4_SearchResults::searchAgain() |
---|
1172 | { |
---|
1173 | int i; |
---|
1174 | |
---|
1175 | for (i=0; i<SEARCH_PATTERNS; i++) { |
---|
1176 | timeOf[i] = 0; |
---|
1177 | } |
---|
1178 | } |
---|
1179 | |
---|
1180 | char *ED4_SearchResults::buildColorString(const ED4_sequence_terminal *seq_terminal, int start, int end) |
---|
1181 | // builds a char buffer (access is only valid from result[start] till result[end]) |
---|
1182 | { |
---|
1183 | int i; |
---|
1184 | int st_shown = 0; |
---|
1185 | |
---|
1186 | e4_assert(start<=end); // confirming the condition |
---|
1187 | for (i=0; i<SEARCH_PATTERNS; i++) { |
---|
1188 | if (shown[i]) { |
---|
1189 | st_shown = 1; |
---|
1190 | break; |
---|
1191 | } |
---|
1192 | } |
---|
1193 | if (!st_shown) { |
---|
1194 | return 0; // nothing shown |
---|
1195 | } |
---|
1196 | |
---|
1197 | search(seq_terminal); |
---|
1198 | if (!get_first()) { |
---|
1199 | return 0; // nothing found |
---|
1200 | } |
---|
1201 | |
---|
1202 | int needed_size = end-start+1; |
---|
1203 | if (needed_size>bufferSize) { |
---|
1204 | free(buffer); |
---|
1205 | bufferSize = needed_size; |
---|
1206 | buffer = (char*)GB_calloc(bufferSize, sizeof(char)); |
---|
1207 | } |
---|
1208 | else { |
---|
1209 | memset(buffer, 0, sizeof(char)*needed_size); |
---|
1210 | } |
---|
1211 | |
---|
1212 | // search first search-result that goes in between start-end |
---|
1213 | |
---|
1214 | ED4_SearchPosition *pos = get_first_at(ED4_ANY_PATTERN, start, end); |
---|
1215 | e4_assert(!pos || (pos->get_start_pos()<=end && pos->get_end_pos()>=start)); |
---|
1216 | |
---|
1217 | int correct_neg_values = 0; |
---|
1218 | |
---|
1219 | while (pos && pos->get_start_pos()<=end) { |
---|
1220 | int what = int(pos->get_whatsFound()); |
---|
1221 | |
---|
1222 | if (shown[what]) { |
---|
1223 | int color = ED4_G_SBACK_0 + what; |
---|
1224 | int s = std::max(pos->get_start_pos(), start)-start; |
---|
1225 | int e = std::min(pos->get_end_pos(), end)-start; |
---|
1226 | |
---|
1227 | for (i=s; i<=e; i++) { |
---|
1228 | if (buffer[i]==0 || abs(buffer[i])>abs(color)) { |
---|
1229 | buffer[i] = color; |
---|
1230 | } |
---|
1231 | } |
---|
1232 | |
---|
1233 | const int *mismatches = pos->getMismatches(); |
---|
1234 | |
---|
1235 | for (i=0; i<5 && mismatches[i]>=0; i++) { |
---|
1236 | int mpos = mismatches[i]; |
---|
1237 | |
---|
1238 | if (mpos>=start && mpos<=end) { |
---|
1239 | int rpos = mpos-start; |
---|
1240 | if (buffer[rpos]==color) { |
---|
1241 | buffer[rpos] = -buffer[rpos]; |
---|
1242 | correct_neg_values = 1; |
---|
1243 | } |
---|
1244 | } |
---|
1245 | } |
---|
1246 | } |
---|
1247 | pos = pos->get_next(); |
---|
1248 | } |
---|
1249 | |
---|
1250 | if (correct_neg_values) { |
---|
1251 | for (i=end-start; i>=0; i--) { |
---|
1252 | if (buffer[i]<0) { |
---|
1253 | buffer[i] = ED4_G_MBACK; |
---|
1254 | } |
---|
1255 | } |
---|
1256 | } |
---|
1257 | |
---|
1258 | return buffer-start; |
---|
1259 | } |
---|
1260 | |
---|
1261 | ED4_SearchPosition *ED4_SearchResults::get_shown_at(int pos) const |
---|
1262 | // used by ED4_cursor::updateAwars to get search-comment |
---|
1263 | { |
---|
1264 | ED4_SearchPosition *p = get_last_starting_before(ED4_ANY_PATTERN, pos, 0); // @@@ tofix: should find last_ending before |
---|
1265 | ED4_SearchPosition *best = 0; |
---|
1266 | |
---|
1267 | if (!p) { |
---|
1268 | p = get_first(); |
---|
1269 | } |
---|
1270 | |
---|
1271 | if (p && !p->containsPos(pos)) { |
---|
1272 | p = p->get_next_at(pos); |
---|
1273 | } |
---|
1274 | |
---|
1275 | while (p) { |
---|
1276 | e4_assert(p->containsPos(pos)); |
---|
1277 | ED4_SearchPositionType type = p->get_whatsFound(); |
---|
1278 | if (shown[type]) { |
---|
1279 | if (!best || type<best->get_whatsFound()) { |
---|
1280 | best = p; |
---|
1281 | } |
---|
1282 | } |
---|
1283 | p = p->get_next_at(pos); |
---|
1284 | } |
---|
1285 | |
---|
1286 | return best; |
---|
1287 | } |
---|
1288 | |
---|
1289 | void ED4_SearchResults::to_array() { |
---|
1290 | e4_assert(is_list()); |
---|
1291 | |
---|
1292 | ED4_SearchPosition *pos = first; |
---|
1293 | |
---|
1294 | { |
---|
1295 | int a_arraySize = 0; |
---|
1296 | |
---|
1297 | while (pos) { |
---|
1298 | a_arraySize++; |
---|
1299 | pos = pos->get_next(); |
---|
1300 | } |
---|
1301 | arraySize = a_arraySize; |
---|
1302 | } |
---|
1303 | |
---|
1304 | ED4_SearchPosition **a_array = (ED4_SearchPosition**)malloc(sizeof(ED4_SearchPosition*)*arraySize); |
---|
1305 | |
---|
1306 | pos = first; |
---|
1307 | for (int e=0; e<arraySize; e++) { |
---|
1308 | e4_assert(pos); |
---|
1309 | a_array[e] = pos; |
---|
1310 | pos = pos->get_next(); |
---|
1311 | } |
---|
1312 | |
---|
1313 | array = a_array; |
---|
1314 | } |
---|
1315 | |
---|
1316 | void ED4_SearchResults::to_list() { |
---|
1317 | e4_assert(is_array()); |
---|
1318 | |
---|
1319 | freenull(array); |
---|
1320 | arraySize = 0; |
---|
1321 | } |
---|
1322 | |
---|
1323 | |
---|
1324 | // -------------------------------------------------------------------------------- |
---|
1325 | |
---|
1326 | inline void decodeSearchDescriptor(int descriptor, int *direction, ED4_SearchPositionType *pattern) |
---|
1327 | { |
---|
1328 | *direction = descriptor&1 ? 1 : -1; |
---|
1329 | *pattern = ED4_SearchPositionType(descriptor/2); |
---|
1330 | } |
---|
1331 | |
---|
1332 | static AW_CL last_searchDescriptor = -1; |
---|
1333 | |
---|
1334 | GB_ERROR ED4_repeat_last_search(ED4_window *ed4w) { |
---|
1335 | if (int(last_searchDescriptor)==-1) { |
---|
1336 | return GBS_global_string("You have to search first, before you can repeat a search."); |
---|
1337 | } |
---|
1338 | |
---|
1339 | ED4_search_cb(0, last_searchDescriptor, (AW_CL)ed4w); |
---|
1340 | return 0; |
---|
1341 | } |
---|
1342 | |
---|
1343 | void ED4_search_cb(AW_window *, AW_CL searchDescriptor, AW_CL cl_ed4w) { |
---|
1344 | ED4_window *ed4w = (ED4_window*)cl_ed4w; |
---|
1345 | e4_assert(ed4w); |
---|
1346 | |
---|
1347 | ED4_LocalWinContext uses(ed4w); |
---|
1348 | |
---|
1349 | last_searchDescriptor = searchDescriptor; |
---|
1350 | |
---|
1351 | int direction; |
---|
1352 | ED4_SearchPositionType pattern; |
---|
1353 | decodeSearchDescriptor(searchDescriptor, &direction, &pattern); |
---|
1354 | |
---|
1355 | int searchOnlyForShownPatterns = pattern==ED4_ANY_PATTERN; |
---|
1356 | |
---|
1357 | // detect position where to start searching |
---|
1358 | |
---|
1359 | ED4_terminal *terminal = 0; // start search at this terminal |
---|
1360 | int pos; // ... and at this position |
---|
1361 | |
---|
1362 | ED4_cursor *cursor = ¤t_cursor(); |
---|
1363 | if (cursor->owner_of_cursor) { // if cursor is shown -> use cursor position |
---|
1364 | terminal = cursor->owner_of_cursor; |
---|
1365 | pos = cursor->get_sequence_pos(); |
---|
1366 | } |
---|
1367 | else { // start at end or beginning |
---|
1368 | if (direction==1) { |
---|
1369 | terminal = ED4_ROOT->root_group_man->get_first_terminal(); |
---|
1370 | pos = INT_MIN; |
---|
1371 | } |
---|
1372 | else { |
---|
1373 | terminal = ED4_ROOT->root_group_man->get_last_terminal(); |
---|
1374 | pos = INT_MAX; |
---|
1375 | } |
---|
1376 | } |
---|
1377 | |
---|
1378 | ED4_terminal *start_terminal = terminal; |
---|
1379 | int start_pos = pos; |
---|
1380 | int last_loop = 0; |
---|
1381 | |
---|
1382 | while (terminal) { |
---|
1383 | if (terminal->is_sequence_terminal()) { |
---|
1384 | ED4_sequence_terminal *seq_terminal = terminal->to_sequence_terminal(); |
---|
1385 | ED4_SearchResults& results = seq_terminal->results(); |
---|
1386 | ED4_SearchPosition *found = 0; |
---|
1387 | |
---|
1388 | if (pattern==ED4_ANY_PATTERN || settings[pattern]->get_open_folded() || !terminal->is_in_folded_group()) { |
---|
1389 | results.search(seq_terminal); |
---|
1390 | |
---|
1391 | if (direction==1) { |
---|
1392 | found = results.get_first_starting_after(pattern, pos, searchOnlyForShownPatterns); |
---|
1393 | } |
---|
1394 | else { |
---|
1395 | found = results.get_last_starting_before(pattern, pos, searchOnlyForShownPatterns); |
---|
1396 | } |
---|
1397 | |
---|
1398 | if (found) { |
---|
1399 | pos = found->get_start_pos(); |
---|
1400 | if (terminal==start_terminal && pos==start_pos) { |
---|
1401 | if (searchOnlyForShownPatterns) { |
---|
1402 | aw_message("There are no other shown patterns"); |
---|
1403 | } |
---|
1404 | else { |
---|
1405 | aw_message("This is the only occurrence of the search pattern"); |
---|
1406 | } |
---|
1407 | } |
---|
1408 | else { |
---|
1409 | seq_terminal->setCursorTo(¤t_cursor(), pos, true, ED4_JUMP_KEEP_POSITION); |
---|
1410 | } |
---|
1411 | break; |
---|
1412 | } |
---|
1413 | else if (last_loop) { |
---|
1414 | if (searchOnlyForShownPatterns) { |
---|
1415 | aw_message("There are no shown patterns"); |
---|
1416 | } |
---|
1417 | else { |
---|
1418 | aw_message("Search pattern was not found in any sequence"); |
---|
1419 | } |
---|
1420 | break; |
---|
1421 | } |
---|
1422 | } |
---|
1423 | } |
---|
1424 | |
---|
1425 | if (direction==1) { |
---|
1426 | terminal = terminal->get_next_terminal(); |
---|
1427 | if (!terminal) terminal = ED4_ROOT->root_group_man->get_first_terminal(); |
---|
1428 | pos = INT_MIN; |
---|
1429 | } |
---|
1430 | else { |
---|
1431 | terminal = terminal->get_prev_terminal(); |
---|
1432 | if (!terminal) terminal = ED4_ROOT->root_group_man->get_last_terminal(); |
---|
1433 | pos = INT_MAX; |
---|
1434 | } |
---|
1435 | |
---|
1436 | if (terminal==start_terminal) { |
---|
1437 | last_loop = 1; |
---|
1438 | } |
---|
1439 | } |
---|
1440 | } |
---|
1441 | |
---|
1442 | static void ED4_mark_matching_species(AW_window * /* aww */, AW_CL cl_pattern) { |
---|
1443 | ED4_SearchPositionType pattern = ED4_SearchPositionType(cl_pattern); |
---|
1444 | ED4_terminal *terminal = ED4_ROOT->root_group_man->get_first_terminal(); |
---|
1445 | GB_transaction ta(GLOBAL_gb_main); |
---|
1446 | |
---|
1447 | while (terminal) { |
---|
1448 | if (terminal->is_sequence_terminal()) { |
---|
1449 | ED4_sequence_terminal *seq_terminal = terminal->to_sequence_terminal(); |
---|
1450 | ED4_SearchResults& results = seq_terminal->results(); |
---|
1451 | |
---|
1452 | results.search(seq_terminal); |
---|
1453 | ED4_SearchPosition *found = results.get_first_starting_after(pattern, INT_MIN, false); |
---|
1454 | |
---|
1455 | if (found) { |
---|
1456 | ED4_species_manager *species_man = seq_terminal->get_parent(ED4_L_SPECIES)->to_species_manager(); |
---|
1457 | if (species_man->is_species_seq_manager()) { |
---|
1458 | GBDATA *gbd = species_man->get_species_pointer(); |
---|
1459 | e4_assert(gbd); |
---|
1460 | GB_write_flag(gbd, 1); |
---|
1461 | } |
---|
1462 | } |
---|
1463 | } |
---|
1464 | |
---|
1465 | terminal = terminal->get_next_terminal(); |
---|
1466 | } |
---|
1467 | } |
---|
1468 | |
---|
1469 | #define ESC '\\' |
---|
1470 | |
---|
1471 | static char *pattern2str(GB_CSTR p) { |
---|
1472 | char *s = GB_give_buffer(strlen(p)*2); |
---|
1473 | char *s1 = s; |
---|
1474 | GB_CSTR p1 = p; |
---|
1475 | |
---|
1476 | while (1) { |
---|
1477 | char c = *p1++; |
---|
1478 | |
---|
1479 | if (!c) { |
---|
1480 | break; |
---|
1481 | } |
---|
1482 | else if (c==ESC) { |
---|
1483 | *s1++ = ESC; |
---|
1484 | *s1++ = ESC; |
---|
1485 | } |
---|
1486 | else if (c=='\n') { |
---|
1487 | *s1++ = ESC; |
---|
1488 | *s1++ = 'n'; |
---|
1489 | } |
---|
1490 | else { |
---|
1491 | *s1++ = c; |
---|
1492 | } |
---|
1493 | } |
---|
1494 | |
---|
1495 | *s1 = 0; |
---|
1496 | return strdup(s); |
---|
1497 | } |
---|
1498 | |
---|
1499 | static void str2pattern(char *s) { // works on string |
---|
1500 | char *a = s; |
---|
1501 | char *n = s; |
---|
1502 | |
---|
1503 | while (1) { |
---|
1504 | char c = *a++; |
---|
1505 | |
---|
1506 | if (c==ESC) { |
---|
1507 | c = *a++; |
---|
1508 | if (c==ESC) { |
---|
1509 | *n++ = ESC; |
---|
1510 | } |
---|
1511 | else if (c=='n') { |
---|
1512 | *n++ = '\n'; |
---|
1513 | } |
---|
1514 | else { // should not occur |
---|
1515 | *n++ = ESC; |
---|
1516 | *n++ = c; |
---|
1517 | } |
---|
1518 | } |
---|
1519 | else { |
---|
1520 | *n++ = c; |
---|
1521 | if (!c) { |
---|
1522 | break; |
---|
1523 | } |
---|
1524 | } |
---|
1525 | } |
---|
1526 | } |
---|
1527 | |
---|
1528 | #undef ESC |
---|
1529 | |
---|
1530 | static void save_search_paras_to_file(AW_window *aw, AW_CL cl_type) { |
---|
1531 | GB_ERROR error = NULL; |
---|
1532 | AW_root *root = ED4_ROOT->aw_root; |
---|
1533 | char *filename = root->awar(ED4_SEARCH_SAVE_BASE"/file_name")->read_string(); |
---|
1534 | |
---|
1535 | { |
---|
1536 | FILE *in = fopen(filename, "rt"); |
---|
1537 | if (in) { |
---|
1538 | fclose(in); |
---|
1539 | |
---|
1540 | char *question = GBS_global_string_copy("'%s' already exists", filename); |
---|
1541 | if (aw_question("overwrite_search_params", question, "Overwrite,Cancel") != 0) { |
---|
1542 | error = "Wont overwrite existing file"; |
---|
1543 | } |
---|
1544 | } |
---|
1545 | } |
---|
1546 | |
---|
1547 | if (!error) { |
---|
1548 | FILE *out = fopen(filename, "wt"); |
---|
1549 | |
---|
1550 | if (!out) { |
---|
1551 | error = GBS_global_string("Can't write file '%s' (%s)", filename, strerror(errno)); |
---|
1552 | } |
---|
1553 | else { |
---|
1554 | ED4_SearchPositionType type = ED4_SearchPositionType(cl_type); |
---|
1555 | SearchSettings *s = settings[type]; |
---|
1556 | |
---|
1557 | char *fpat = pattern2str(s->get_pattern()); |
---|
1558 | |
---|
1559 | fprintf(out, |
---|
1560 | "pattern=%s\n" |
---|
1561 | "minmis=%i\n" |
---|
1562 | "maxmis=%i\n" |
---|
1563 | "case=%i\n" |
---|
1564 | "tu=%i\n" |
---|
1565 | "patgaps=%i\n" |
---|
1566 | "seqgaps=%i\n" |
---|
1567 | "openfolded=%i\n" |
---|
1568 | "autojump=%i\n" |
---|
1569 | "reverse=%i\n" |
---|
1570 | "complement=%i\n" |
---|
1571 | "exact=%i\n", |
---|
1572 | fpat, |
---|
1573 | s->get_min_mismatches(), |
---|
1574 | s->get_max_mismatches(), |
---|
1575 | s->get_case_sensitive(), |
---|
1576 | s->get_tu(), |
---|
1577 | s->get_pat_gaps(), |
---|
1578 | s->get_seq_gaps(), |
---|
1579 | s->get_open_folded(), |
---|
1580 | s->get_autoJump(), |
---|
1581 | s->get_reverse(), |
---|
1582 | s->get_complement(), |
---|
1583 | s->get_exact()); |
---|
1584 | |
---|
1585 | free(fpat); |
---|
1586 | fclose(out); |
---|
1587 | } |
---|
1588 | } |
---|
1589 | |
---|
1590 | if (error) aw_message(error); |
---|
1591 | else AW_POPDOWN(aw); |
---|
1592 | |
---|
1593 | free(filename); |
---|
1594 | } |
---|
1595 | |
---|
1596 | static void load_search_paras_from_file(AW_window *aw, AW_CL cl_type) { |
---|
1597 | GB_CSTR error = NULL; |
---|
1598 | AW_root *root = ED4_ROOT->aw_root; |
---|
1599 | char *filename = root->awar(ED4_SEARCH_SAVE_BASE"/file_name")->read_string(); |
---|
1600 | FILE *in = fopen(filename, "rt"); |
---|
1601 | |
---|
1602 | if (!in) { |
---|
1603 | error = GBS_global_string("File '%s' not found", filename); |
---|
1604 | } |
---|
1605 | else { |
---|
1606 | ED4_SearchPositionType type = ED4_SearchPositionType(cl_type); |
---|
1607 | SearchAwarList *al = &awar_list[type]; |
---|
1608 | |
---|
1609 | while (1) { |
---|
1610 | const int BUFFERSIZE = 10000; |
---|
1611 | char buffer[BUFFERSIZE]; |
---|
1612 | if (!fgets(buffer, BUFFERSIZE, in)) break; |
---|
1613 | |
---|
1614 | char *content = strchr(buffer, '='); |
---|
1615 | if (!content) { |
---|
1616 | aw_message(GBS_global_string("Format error in '%s' - ignored!", filename)); |
---|
1617 | } |
---|
1618 | else { |
---|
1619 | *content++ = 0; |
---|
1620 | |
---|
1621 | if (strcmp(buffer, "pattern")==0) { |
---|
1622 | str2pattern(content); |
---|
1623 | root->awar(al->pattern)->write_string(content); |
---|
1624 | } |
---|
1625 | else { |
---|
1626 | int value = atoi(content); |
---|
1627 | |
---|
1628 | if (strcmp(buffer, "minmis")==0) root->awar(al->min_mismatches)->write_int(value); |
---|
1629 | else if (strcmp(buffer, "maxmis")==0) root->awar(al->max_mismatches)->write_int(value); |
---|
1630 | else if (strcmp(buffer, "case")==0) root->awar(al->case_sensitive)->write_int(value); |
---|
1631 | else if (strcmp(buffer, "tu")==0) root->awar(al->tu)->write_int(value); |
---|
1632 | else if (strcmp(buffer, "patgaps")==0) root->awar(al->pat_gaps)->write_int(value); |
---|
1633 | else if (strcmp(buffer, "seqgaps")==0) root->awar(al->seq_gaps)->write_int(value); |
---|
1634 | else if (strcmp(buffer, "openfolded")==0) root->awar(al->openFolded)->write_int(value); |
---|
1635 | else if (strcmp(buffer, "autojump")==0) root->awar(al->autoJump)->write_int(value); |
---|
1636 | else if (strcmp(buffer, "reverse")==0) root->awar(al->reverse)->write_int(value); |
---|
1637 | else if (strcmp(buffer, "complement")==0) root->awar(al->complement)->write_int(value); |
---|
1638 | else if (strcmp(buffer, "exact")==0) root->awar(al->exact)->write_int(value); |
---|
1639 | else { |
---|
1640 | aw_message(GBS_global_string("Unknown tag '%s' in '%s' - ignored!", buffer, filename)); |
---|
1641 | } |
---|
1642 | } |
---|
1643 | } |
---|
1644 | } |
---|
1645 | |
---|
1646 | fclose(in); |
---|
1647 | } |
---|
1648 | |
---|
1649 | if (error) aw_message(error); |
---|
1650 | else AW_POPDOWN(aw); |
---|
1651 | |
---|
1652 | free(filename); |
---|
1653 | } |
---|
1654 | |
---|
1655 | 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) { |
---|
1656 | char *window_title = GBS_global_string_copy(title_format, typeId); |
---|
1657 | char *window_id_src = GBS_global_string_copy(id_format, typeId, winNum); |
---|
1658 | char *window_id = GBS_string_2_key(window_id_src); |
---|
1659 | |
---|
1660 | aws->init(root, window_id, window_title); |
---|
1661 | |
---|
1662 | free(window_id); |
---|
1663 | free(window_id_src); |
---|
1664 | free(window_title); |
---|
1665 | } |
---|
1666 | |
---|
1667 | struct LoadSaveSearchParam { |
---|
1668 | ED4_SearchPositionType type; |
---|
1669 | int winNum; |
---|
1670 | |
---|
1671 | LoadSaveSearchParam(ED4_SearchPositionType type_, int winNum_) |
---|
1672 | : type(type_), |
---|
1673 | winNum(winNum_) |
---|
1674 | {} |
---|
1675 | }; |
---|
1676 | |
---|
1677 | static AW_window *loadsave_search_parameters(AW_root *root, const LoadSaveSearchParam& param, bool load) { |
---|
1678 | AW_window_simple *aws = new AW_window_simple; |
---|
1679 | |
---|
1680 | if (load) { |
---|
1681 | aws_init_localized(root, aws, "load_%s_search_para_%i", "Load %s Search Parameters", ED4_SearchPositionTypeId[param.type], param.winNum); |
---|
1682 | } |
---|
1683 | else { |
---|
1684 | aws_init_localized(root, aws, "save_%s_search_para_%i", "Save %s Search Parameters", ED4_SearchPositionTypeId[param.type], param.winNum); |
---|
1685 | } |
---|
1686 | |
---|
1687 | aws->load_xfig("edit4/save_search.fig"); |
---|
1688 | |
---|
1689 | aws->at("close"); aws->callback((AW_CB0)AW_POPDOWN); |
---|
1690 | aws->create_button("CLOSE", "CLOSE", "C"); |
---|
1691 | |
---|
1692 | aws->callback(makeHelpCallback("search_parameters.hlp")); |
---|
1693 | aws->at("help"); |
---|
1694 | aws->create_button("HELP", "HELP", "H"); |
---|
1695 | |
---|
1696 | AW_create_standard_fileselection(aws, ED4_SEARCH_SAVE_BASE); |
---|
1697 | |
---|
1698 | aws->callback((AW_CB0)AW_POPDOWN); |
---|
1699 | aws->at("cancel"); |
---|
1700 | aws->create_button("CANCEL", "CANCEL", "C"); |
---|
1701 | |
---|
1702 | aws->at("save"); |
---|
1703 | if (load) { |
---|
1704 | aws->callback(load_search_paras_from_file, (AW_CL)param.type); |
---|
1705 | aws->create_button("LOAD", "LOAD", "L"); |
---|
1706 | } |
---|
1707 | else { |
---|
1708 | aws->callback(save_search_paras_to_file, (AW_CL)param.type); |
---|
1709 | aws->create_button("SAVE", "SAVE", "S"); |
---|
1710 | } |
---|
1711 | |
---|
1712 | return aws; |
---|
1713 | } |
---|
1714 | |
---|
1715 | static AW_window *load_search_parameters(AW_root *root, AW_CL cl_param) { |
---|
1716 | LoadSaveSearchParam *param = (LoadSaveSearchParam*)cl_param; |
---|
1717 | return loadsave_search_parameters(root, *param, true); |
---|
1718 | } |
---|
1719 | |
---|
1720 | static AW_window *save_search_parameters(AW_root *root, AW_CL cl_param) { |
---|
1721 | LoadSaveSearchParam *param = (LoadSaveSearchParam*)cl_param; |
---|
1722 | return loadsave_search_parameters(root, *param, false); |
---|
1723 | } |
---|
1724 | |
---|
1725 | |
---|
1726 | static void search_init_config(AWT_config_definition& cdef, int search_type) { |
---|
1727 | SearchAwarList *awarList = &awar_list[search_type]; |
---|
1728 | |
---|
1729 | cdef.add(awarList->show, "show"); |
---|
1730 | cdef.add(awarList->openFolded, "openFolded"); |
---|
1731 | cdef.add(awarList->autoJump, "autoJump"); |
---|
1732 | cdef.add(awarList->pattern, "pattern"); |
---|
1733 | cdef.add(awarList->min_mismatches, "min_mismatches"); |
---|
1734 | cdef.add(awarList->max_mismatches, "max_mismatches"); |
---|
1735 | cdef.add(awarList->seq_gaps, "seq_gaps"); |
---|
1736 | cdef.add(awarList->pat_gaps, "pat_gaps"); |
---|
1737 | cdef.add(awarList->tu, "tu"); |
---|
1738 | cdef.add(awarList->case_sensitive, "case_sensitive"); |
---|
1739 | cdef.add(awarList->reverse, "reverse"); |
---|
1740 | cdef.add(awarList->complement, "complement"); |
---|
1741 | cdef.add(awarList->exact, "exact"); |
---|
1742 | } |
---|
1743 | |
---|
1744 | static char *search_store_config(AW_window *aww, AW_CL cl_search_type, AW_CL) { |
---|
1745 | AWT_config_definition cdef(aww->get_root()); |
---|
1746 | search_init_config(cdef, int(cl_search_type)); |
---|
1747 | return cdef.read(); |
---|
1748 | } |
---|
1749 | |
---|
1750 | static void search_restore_config(AW_window *aww, const char *stored_string, AW_CL cl_search_type, AW_CL) { |
---|
1751 | AWT_config_definition cdef(aww->get_root()); |
---|
1752 | search_init_config(cdef, int(cl_search_type)); |
---|
1753 | cdef.write(stored_string); |
---|
1754 | } |
---|
1755 | |
---|
1756 | struct search_windows : public Noncopyable { |
---|
1757 | AW_window_simple *windows[SEARCH_PATTERNS]; |
---|
1758 | search_windows() { for (int i = 0; i<SEARCH_PATTERNS; ++i) windows[i] = NULL; } |
---|
1759 | }; |
---|
1760 | |
---|
1761 | typedef std::map<ED4_window*, SmartPtr<search_windows> > search_window_map; |
---|
1762 | |
---|
1763 | void ED4_popup_search_window(AW_window *aww, AW_CL cl_search_type) { |
---|
1764 | ED4_SearchPositionType type = (ED4_SearchPositionType)cl_search_type; |
---|
1765 | |
---|
1766 | ED4_WinContext uses(aww); |
---|
1767 | ED4_window *ed4w = uses.get_ed4w(); |
---|
1768 | |
---|
1769 | static search_window_map swm; |
---|
1770 | { // create search windows pointer array for current ed4w |
---|
1771 | search_window_map::iterator sw4win = swm.find(ed4w); |
---|
1772 | if (sw4win == swm.end()) swm[ed4w] = new search_windows; |
---|
1773 | } |
---|
1774 | |
---|
1775 | SmartPtr<search_windows> sw = swm[ed4w]; |
---|
1776 | AW_window_simple *aws = sw->windows[type]; |
---|
1777 | if (!aws) { |
---|
1778 | SearchAwarList *awarList = &awar_list[type]; |
---|
1779 | |
---|
1780 | sw->windows[type] = aws = new AW_window_simple; |
---|
1781 | |
---|
1782 | aws_init_localized(aww->get_root(), aws, "%s_search_%i", "%s Search", ED4_SearchPositionTypeId[type], ed4w->id); |
---|
1783 | aws->load_xfig("edit4/search.fig"); |
---|
1784 | |
---|
1785 | aws->at("close"); |
---|
1786 | aws->callback((AW_CB0)AW_POPDOWN); |
---|
1787 | aws->create_button("CLOSE", "CLOSE", "C"); |
---|
1788 | |
---|
1789 | aws->at("help"); |
---|
1790 | aws->callback(makeHelpCallback("e4_search.hlp")); |
---|
1791 | aws->create_button("HELP", "HELP", "H"); |
---|
1792 | |
---|
1793 | LoadSaveSearchParam *param = new LoadSaveSearchParam(type, ed4w->id); // bound to callbacks (dont free) |
---|
1794 | |
---|
1795 | aws->at("load"); |
---|
1796 | aws->callback(AW_POPUP, (AW_CL)load_search_parameters, (AW_CL)param); |
---|
1797 | aws->create_button("LOAD", "LOAD", "L"); |
---|
1798 | |
---|
1799 | aws->at("save"); |
---|
1800 | aws->callback(AW_POPUP, (AW_CL)save_search_parameters, (AW_CL)param); |
---|
1801 | aws->create_button("SAVE", "SAVE", "S"); |
---|
1802 | |
---|
1803 | aws->at("next"); |
---|
1804 | aws->callback(ED4_search_cb, (AW_CL)ED4_encodeSearchDescriptor(+1, type), (AW_CL)ed4w); |
---|
1805 | aws->create_button("SEARCH_NEXT", "#edit/next.xpm", "N"); |
---|
1806 | |
---|
1807 | aws->at("previous"); |
---|
1808 | aws->callback(ED4_search_cb, (AW_CL)ED4_encodeSearchDescriptor(-1, type), (AW_CL)ed4w); |
---|
1809 | aws->create_button("SEARCH_LAST", "#edit/last.xpm", "L"); |
---|
1810 | |
---|
1811 | aws->at("mark"); |
---|
1812 | aws->callback(ED4_mark_matching_species, (AW_CL)type); |
---|
1813 | aws->create_autosize_button("MARK_SPECIES", "Mark species with matches", "M"); |
---|
1814 | |
---|
1815 | aws->at("show"); |
---|
1816 | aws->create_toggle(awarList->show); |
---|
1817 | |
---|
1818 | aws->at("open"); |
---|
1819 | aws->create_toggle(awarList->openFolded); |
---|
1820 | |
---|
1821 | aws->at("jump"); |
---|
1822 | aws->create_toggle(awarList->autoJump); |
---|
1823 | |
---|
1824 | aws->at("pattern"); |
---|
1825 | aws->create_text_field(awarList->pattern, 28, 17); |
---|
1826 | |
---|
1827 | aws->at("minmis"); |
---|
1828 | aws->create_toggle_field(awarList->min_mismatches, 1); |
---|
1829 | aws->insert_default_toggle("0", "0", 0); |
---|
1830 | aws->insert_toggle("1", "1", 1); |
---|
1831 | aws->insert_toggle("2", "2", 2); |
---|
1832 | aws->update_toggle_field(); |
---|
1833 | |
---|
1834 | aws->at("maxmis"); |
---|
1835 | aws->create_toggle_field(awarList->max_mismatches, 1); |
---|
1836 | aws->insert_default_toggle("0", "0", 0); |
---|
1837 | aws->insert_toggle("1", "1", 1); |
---|
1838 | aws->insert_toggle("2", "2", 2); |
---|
1839 | aws->insert_toggle("3", "3", 3); |
---|
1840 | aws->insert_toggle("4", "4", 4); |
---|
1841 | aws->insert_toggle("5", "5", 5); |
---|
1842 | aws->update_toggle_field(); |
---|
1843 | |
---|
1844 | aws->at("seq_gaps"); |
---|
1845 | aws->create_toggle(awarList->seq_gaps); |
---|
1846 | aws->at("pat_gaps"); |
---|
1847 | aws->create_toggle(awarList->pat_gaps); |
---|
1848 | aws->at("tu"); |
---|
1849 | aws->create_toggle(awarList->tu); |
---|
1850 | aws->at("case"); |
---|
1851 | aws->create_toggle(awarList->case_sensitive); |
---|
1852 | aws->at("reverse"); |
---|
1853 | aws->create_toggle(awarList->reverse); |
---|
1854 | aws->at("complement"); |
---|
1855 | aws->create_toggle(awarList->complement); |
---|
1856 | aws->at("exact"); |
---|
1857 | aws->create_toggle(awarList->exact); |
---|
1858 | |
---|
1859 | aws->at("config"); |
---|
1860 | AWT_insert_config_manager(aws, AW_ROOT_DEFAULT, "search", search_store_config, search_restore_config, (AW_CL)type, 0); |
---|
1861 | |
---|
1862 | } |
---|
1863 | |
---|
1864 | aws->activate(); |
---|
1865 | } |
---|
1866 | |
---|
1867 | |
---|
1868 | |
---|