Context Navigation

← Previous Revision
Next Revision →
Blame
Revision Log

source: branches/port5/ARBDB/adcolumns.c

Visit:

Last change on this file was 6020, checked in by westram, 16 years ago
splitted adtools.c into multiple files (no code change)
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 17.5 KB

Line
1	/* ============================================================ */
2	/* */
3	/* File : adcolumns.c */
4	/* Purpose : insert/delete columns */
5	/* */
6	/* Institute of Microbiology (Technical University Munich) */
7	/* www.arb-home.de */
8	/* */
9	/* ============================================================ */
10
11	#include <stdlib.h>
12	#include <string.h>
13
14	#include <adlocal.h>
15	#include <arbdbt.h>
16	#include <adGene.h>
17
18
19	/* ----------------------- */
20	/* insert/delete */
21
22	static char *insDelBuffer = 0;
23	static size_t insDelBuffer_size;
24
25	static void free_insDelBuffer() {
26	freeset(insDelBuffer, 0);
27	}
28
29	static const char gbt_insert_delete(const char source, long srclen, long destlen, long *newlenPtr, long pos, long nchar, long mod, char insert_what, char insert_tail, int extraByte) {
30	/* removes elems from or inserts elems into an array
31	*
32	* srclen len of source
33	* destlen if != 0, then cut or append characters to get this len, otherwise keep srclen
34	* newlenPtr the resulting len
35	* pos where to insert/delete
36	* nchar and how many items
37	* mod size of an item
38	* insert_what insert this character (mod times)
39	* insert_tail append this character (if destlen>srclen)
40	* extraByte 0 or 1. append extra zero byte at end? use 1 for strings!
41	*
42	* resulting array has destlen+nchar elements
43	*
44	* 1. array size is corrected to 'destlen' (by appending/cutting tail)
45	* 2. part is deleted inserted
46	*/
47
48	const char *result;
49
50	pos *= mod;
51	nchar *= mod;
52	srclen *= mod;
53	destlen *= mod;
54
55	if (!destlen) destlen = srclen; /* if no destlen is set then keep srclen */
56	if ((nchar<0) && (pos-nchar>destlen)) nchar = pos-destlen; /* clip maximum characters to delete at end of array */
57
58	if (destlen == srclen && (pos>srclen \|\| nchar == 0)) { /* length stays same and clip-range is empty or behind end of sequence */
59	/* before 26.2.09 the complete data was copied in this case - but nevertheless NULL(=failure) was returned.
60	* I guess this was some test accessing complete data w/o writing anything back to DB,
61	* but AFAIK it was not used anywhere --ralf
62	*/
63	result = NULL;
64	}
65	else {
66	long newlen = destlen+nchar; /* length of result (w/o trailing zero-byte) */
67	if (newlen == 0) {
68	result = "";
69	}
70	else {
71	size_t neededSpace = newlen+extraByte;
72
73	if (insDelBuffer && insDelBuffer_size<neededSpace) freeset(insDelBuffer, 0);
74	if (!insDelBuffer) {
75	insDelBuffer_size = neededSpace;
76	insDelBuffer = (char*)malloc(neededSpace);
77	}
78
79	char *dest = insDelBuffer;
80	gb_assert(dest);
81
82	if (pos>srclen) { /* insert/delete happens inside appended range */
83	insert_what = insert_tail;
84	pos = srclen; /* insert/delete directly after source, to avoid illegal access below */
85	}
86
87	gb_assert(pos >= 0);
88	if (pos>0) { /* copy part left of pos */
89	memcpy(dest, source, (size_t)pos);
90	dest += pos;
91	source += pos; srclen -= pos;
92	}
93
94	if (nchar>0) { /* insert */
95	memset(dest, insert_what, (size_t)nchar);
96	dest += nchar;
97	}
98	else if (nchar<0) { /* delete */
99	source += -nchar; srclen -= -nchar;
100	}
101
102	if (srclen>0) { /* copy rest of source */
103	memcpy(dest, source, (size_t)srclen);
104	dest += srclen;
105	source += srclen; srclen = 0;
106	}
107
108	long rest = newlen-(dest-insDelBuffer);
109	gb_assert(rest >= 0);
110
111	if (rest>0) { /* append tail */
112	memset(dest, insert_tail, rest);
113	dest += rest;
114	}
115
116	if (extraByte) dest[0] = 0; /* append zero byte (used for strings) */
117
118	result = insDelBuffer;
119	}
120	newlenPtr = newlen/mod; / report result length */
121	}
122	return result;
123	}
124
125	enum insDelTarget {
126	IDT_SPECIES = 0,
127	IDT_SAI,
128	IDT_SECSTRUCT,
129	};
130
131	static GB_CSTR targetType[] = {
132	"Species",
133	"SAI",
134	"SeceditStruct",
135	};
136
137	static GB_BOOL insdel_shall_be_applied_to(GBDATA *gb_data, enum insDelTarget target) {
138	GB_BOOL apply = GB_TRUE;
139	const char *key = GB_read_key_pntr(gb_data);
140
141	if (key[0] == '_') { // dont apply to keys starting with '_'
142	switch (target) {
143	case IDT_SECSTRUCT:
144	case IDT_SPECIES:
145	apply = GB_FALSE;
146	break;
147
148	case IDT_SAI:
149	if (strcmp(key, "_REF") != 0) { // despite key is _REF
150	apply = GB_FALSE;
151	}
152	break;
153	}
154	}
155
156	return apply;
157	}
158
159	struct insDel_params {
160	char *ali_name; // name of alignment
161	long ali_len; // wanted length of alignment
162	long pos; // start position of insert/delete
163	long nchar; // number of elements to insert/delete
164	const char *delete_chars; // characters allowed to delete (array with 256 entries, value == 0 means deletion allowed)
165	};
166
167
168
169	static GB_ERROR gbt_insert_character_gbd(GBDATA gb_data, enum insDelTarget target, const struct insDel_params params) {
170	GB_ERROR error = 0;
171	GB_TYPES type = GB_read_type(gb_data);
172
173	if (type == GB_DB) {
174	GBDATA *gb_child;
175	for (gb_child = GB_child(gb_data); gb_child && !error; gb_child = GB_nextChild(gb_child)) {
176	error = gbt_insert_character_gbd(gb_child, target, params);
177	}
178	}
179	else {
180	ad_assert(params->pos >= 0);
181	if (type >= GB_BITS && type != GB_LINK) {
182	long size = GB_read_count(gb_data);
183
184	if (params->ali_len != size \|\| params->nchar != 0) { /* nothing would change */
185	if (insdel_shall_be_applied_to(gb_data, target)) {
186	GB_CSTR source = 0;
187	long mod = sizeof(char);
188	char insert_what = 0;
189	char insert_tail = 0;
190	char extraByte = 0;
191	long pos = params->pos;
192	long nchar = params->nchar;
193
194	switch (type) {
195	case GB_STRING: {
196	source = GB_read_char_pntr(gb_data);
197	extraByte = 1;
198	insert_what = '-';
199	insert_tail = '.';
200
201	if (source) {
202	if (nchar > 0) { /* insert */
203	if (pos<size) { /* otherwise insert pos is behind (old and too short) sequence -> dots are inserted at tail */
204	if ((pos>0 && source[pos-1] == '.') \|\| source[pos] == '.') { /* dot at insert position? */
205	insert_what = '.'; /* insert dots */
206	}
207	}
208	}
209	else { /* delete */
210	long after = pos+(-nchar); /* position after deleted part */
211	long p;
212	GB_CSTR delete_chars = params->delete_chars;
213
214	if (after>size) after = size;
215	for (p = pos; p<after; p++){
216	if (delete_chars[((const unsigned char *)source)[p]]) {
217	error = GBS_global_string("You tried to delete '%c' at position %li -> Operation aborted", source[p], p);
218	}
219	}
220	}
221	}
222
223	break;
224	}
225	case GB_BITS: source = GB_read_bits_pntr(gb_data, '-', '+'); insert_what = '-'; insert_tail = '-'; break;
226	case GB_BYTES: source = GB_read_bytes_pntr(gb_data); break;
227	case GB_INTS: source = (GB_CSTR)GB_read_ints_pntr(gb_data); mod = sizeof(GB_UINT4); break;
228	case GB_FLOATS: source = (GB_CSTR)GB_read_floats_pntr(gb_data); mod = sizeof(float); break;
229
230	default :
231	error = GBS_global_string("Unhandled type '%i'", type);
232	GB_internal_error(error);
233	break;
234	}
235
236	if (!error) {
237	if (!source) error = GB_await_error();
238	else {
239	long modified_len;
240	GB_CSTR modified = gbt_insert_delete(source, size, params->ali_len, &modified_len, pos, nchar, mod, insert_what, insert_tail, extraByte);
241
242	if (modified) {
243	gb_assert(modified_len == (params->ali_len+params->nchar));
244
245	switch (type) {
246	case GB_STRING: error = GB_write_string(gb_data, modified); break;
247	case GB_BITS: error = GB_write_bits (gb_data, modified, modified_len, "-"); break;
248	case GB_BYTES: error = GB_write_bytes (gb_data, modified, modified_len); break;
249	case GB_INTS: error = GB_write_ints (gb_data, (GB_UINT4*)modified, modified_len); break;
250	case GB_FLOATS: error = GB_write_floats(gb_data, (float*)modified, modified_len); break;
251
252	default: gb_assert(0); break;
253	}
254	}
255	}
256	}
257	}
258	}
259	}
260	}
261
262	return error;
263	}
264
265	static GB_ERROR gbt_insert_character_item(GBDATA gb_item, enum insDelTarget item_type, const struct insDel_params params) {
266	GB_ERROR error = 0;
267	GBDATA *gb_ali = GB_entry(gb_item, params->ali_name);
268
269	if (gb_ali) {
270	error = gbt_insert_character_gbd(gb_ali, item_type, params);
271	if (error) {
272	const char *item_name = GBT_read_name(gb_item);
273	error = GBS_global_string("%s '%s': %s", targetType[item_type], item_name, error);
274	}
275	}
276
277	return error;
278	}
279
280	static GB_ERROR gbt_insert_character(GBDATA gb_item_data, const char item_field, enum insDelTarget item_type, const struct insDel_params *params) {
281	GBDATA *gb_item;
282	GB_ERROR error = 0;
283	long item_count = GB_number_of_subentries(gb_item_data);
284	long count = 0;
285
286	for (gb_item = GB_entry(gb_item_data, item_field);
287	gb_item && !error;
288	gb_item = GB_nextEntry(gb_item))
289	{
290	error = gbt_insert_character_item(gb_item, item_type, params);
291	count++;
292	GB_status((double)count/item_count);
293	}
294	return error;
295	}
296
297	static GB_ERROR gbt_insert_character_secstructs(GBDATA gb_secstructs, const struct insDel_params params) {
298	GB_ERROR error = 0;
299	GBDATA *gb_ali = GB_entry(gb_secstructs, params->ali_name);
300	if (gb_ali) {
301	long item_count = GB_number_of_subentries(gb_ali)-1;
302	long count = 0;
303	GBDATA *gb_item;
304
305	if (item_count<1) item_count = 1;
306
307	for (gb_item = GB_entry(gb_ali, "struct");
308	gb_item && !error;
309	gb_item = GB_nextEntry(gb_item))
310	{
311	GBDATA *gb_ref = GB_entry(gb_item, "ref");
312	if (gb_ref) {
313	error = gbt_insert_character_gbd(gb_ref, IDT_SECSTRUCT, params);
314	if (error) {
315	const char *item_name = GBT_read_name(gb_item);
316	error = GBS_global_string("%s '%s': %s", targetType[IDT_SECSTRUCT], item_name, error);
317	}
318	}
319	count++;
320	GB_status((double)count/item_count);
321	}
322	}
323	return error;
324	}
325
326	static GB_ERROR GBT_check_lengths(GBDATA Main, const char alignment_name) {
327	GB_ERROR error = 0;
328	GBDATA *gb_presets = GBT_find_or_create(Main,"presets",7);
329	GBDATA *gb_species_data = GBT_find_or_create(Main,"species_data",7);
330	GBDATA *gb_extended_data = GBT_find_or_create(Main,"extended_data",7);
331	GBDATA *gb_secstructs = GB_search(Main,"secedit/structs", GB_CREATE_CONTAINER);
332	GBDATA *gb_ali;
333
334	struct insDel_params params = { 0, 0, 0, 0, 0 };
335
336	for (gb_ali = GB_entry(gb_presets,"alignment");
337	gb_ali && !error;
338	gb_ali = GB_nextEntry(gb_ali))
339	{
340	GBDATA *gb_name = GB_find_string(gb_ali,"alignment_name",alignment_name,GB_IGNORE_CASE,down_level);
341
342	if (gb_name) {
343	GBDATA *gb_len = GB_entry(gb_ali,"alignment_len");
344
345	params.ali_name = GB_read_string(gb_name);
346	params.ali_len = GB_read_int(gb_len);
347
348	error = gbt_insert_character(gb_extended_data, "extended", IDT_SAI, &params);
349	if (!error) error = gbt_insert_character(gb_species_data, "species", IDT_SPECIES, &params);
350	if (!error) error = gbt_insert_character_secstructs(gb_secstructs, &params);
351
352	freeset(params.ali_name, 0);
353	}
354	}
355	free_insDelBuffer();
356	return error;
357	}
358
359	GB_ERROR GBT_format_alignment(GBDATA Main, const char alignment_name) {
360	GB_ERROR err = 0;
361
362	if (strcmp(alignment_name, GENOM_ALIGNMENT) != 0) { // NEVER EVER format 'ali_genom'
363	err = GBT_check_data(Main, alignment_name); // detect max. length
364	if (!err) err = GBT_check_lengths(Main, alignment_name); // format sequences in alignment
365	if (!err) err = GBT_check_data(Main, alignment_name); // sets state to "formatted"
366	}
367	else {
368	err = "It's forbidden to format '" GENOM_ALIGNMENT "'!";
369	}
370	return err;
371	}
372
373
374	GB_ERROR GBT_insert_character(GBDATA Main, char alignment_name, long pos, long count, char *char_delete)
375	{
376	/* if count > 0 insert 'count' characters at pos
377	* if count < 0 delete pos to pos+\|count\|
378	*
379	* Note: deleting is only performed, if found characters in deleted range are listed in 'char_delete'
380	* otherwise function returns with error
381	*
382	* This affects all species' and SAIs having data in given 'alignment_name' and
383	* modifies several data entries found there
384	* (see insdel_shall_be_applied_to for details which fields are affected).
385	*/
386
387	GB_ERROR error = 0;
388
389	if (pos<0) {
390	error = GB_export_error("Illegal sequence position");
391	}
392	else {
393	GBDATA *gb_ali;
394	GBDATA *gb_presets = GBT_find_or_create(Main,"presets",7);
395	GBDATA *gb_species_data = GBT_find_or_create(Main,"species_data",7);
396	GBDATA *gb_extended_data = GBT_find_or_create(Main,"extended_data",7);
397	GBDATA *gb_secstructs = GB_search(Main,"secedit/structs", GB_CREATE_CONTAINER);
398	char char_delete_list[256];
399
400	if (strchr(char_delete,'%') ) {
401	memset(char_delete_list,0,256);
402	}
403	else {
404	int ch;
405	for (ch = 0;ch<256; ch++) {
406	if (char_delete) {
407	if (strchr(char_delete,ch)) char_delete_list[ch] = 0;
408	else char_delete_list[ch] = 1;
409	}
410	else {
411	char_delete_list[ch] = 0;
412	}
413	}
414	}
415
416	for (gb_ali = GB_entry(gb_presets, "alignment");
417	gb_ali && !error;
418	gb_ali = GB_nextEntry(gb_ali))
419	{
420	GBDATA *gb_name = GB_find_string(gb_ali, "alignment_name", alignment_name, GB_IGNORE_CASE, down_level);
421
422	if (gb_name) {
423	GBDATA *gb_len = GB_entry(gb_ali, "alignment_len");
424	long len = GB_read_int(gb_len);
425	char *use = GB_read_string(gb_name);
426
427	if (pos > len) {
428	error = GBS_global_string("Can't insert at position %li (exceeds length %li of alignment '%s')", pos, len, use);
429	}
430	else {
431	if (count < 0 && pos-count > len) count = pos - len;
432	error = GB_write_int(gb_len, len + count);
433	}
434
435	if (!error) {
436	struct insDel_params params = { use, len, pos, count, char_delete_list };
437
438	error = gbt_insert_character(gb_species_data, "species", IDT_SPECIES, &params);
439	if (!error) error = gbt_insert_character(gb_extended_data, "extended", IDT_SAI, &params);
440	if (!error) error = gbt_insert_character_secstructs(gb_secstructs, &params);
441	}
442	free(use);
443	}
444	}
445
446	free_insDelBuffer();
447
448	if (!error) GB_disable_quicksave(Main,"a lot of sequences changed");
449	}
450	return error;
451	}

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

Download in other formats: