Context Navigation

← Previous Revision
Next Revision →
Blame
Revision Log

AW_rename.cxx

Visit:

Last change on this file was 19206, checked in by westram, 2 years ago
reintegrates 'ali' into 'trunk' adds: log:branches/ali@19183:19205
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 24.8 KB

Line
1	// =============================================================== //
2	// //
3	// File : AW_rename.cxx //
4	// Purpose : //
5	// //
6	// Institute of Microbiology (Technical University Munich) //
7	// http://www.arb-home.de/ //
8	// //
9	// =============================================================== //
10
11	#include "AW_rename.hxx"
12
13	#include <aw_awars.hxx>
14	#include <aw_window.hxx>
15	#include <aw_root.hxx>
16	#include <aw_question.hxx>
17	#include <aw_msg.hxx>
18	#include <arb_progress.h>
19
20	#include <names_client.h>
21	#include <servercntrl.h>
22	#include <client.h>
23
24	#include <cctype>
25	#include <ctime>
26
27	static const char get_addid(GBDATA gb_main) {
28	GB_transaction ta(gb_main);
29	GBDATA *gb_addid = GB_search(gb_main, AWAR_NAMESERVER_ADDID, GB_FIND);
30	return gb_addid ? GB_read_char_pntr(gb_addid) : NULp;
31	}
32
33	static GB_ERROR set_addid(GBDATA gb_main, const char addid) {
34	GB_ERROR error = GB_push_transaction(gb_main);
35	if (!error) error = GBT_write_string(gb_main, AWAR_NAMESERVER_ADDID, null2empty(addid));
36	return GB_end_transaction(gb_main, error);
37	}
38
39	const char AW_get_nameserver_addid(GBDATA gb_main) {
40	// return the additional field used for nameserver connection
41	const char *addid = get_addid(gb_main);
42
43	aw_assert(addid); // the current DB has no entry AWAR_NAMESERVER_ADDID! (programmers error)
44	if (!addid) addid = ""; // NDEBUG fallback
45
46	return addid;
47	}
48
49	GB_ERROR AW_select_nameserver(GBDATA gb_main, GBDATA gb_other_main) {
50	// if entry AWAR_NAMESERVER_ADDID isn't defined yet, try to detect a reasonable value
51	// from arb_tcp.dat. Ask user if multiple servers are defined.
52	//
53	// if gb_other_main is defined try to use value from there.
54
55	const char *addid = get_addid(gb_main);
56	GB_ERROR error = NULp;
57
58	if (!addid && gb_other_main && gb_other_main != gb_main) {
59	// look whether main DB has a defined addid
60	addid = get_addid(gb_other_main);
61	set_addid(gb_main, addid);
62	}
63
64	if (!addid) {
65	const char * const nameservers = GBS_get_arb_tcp_entries("ARB_NAME_SERVER");
66
67	if (!nameservers) error = GB_await_error();
68	else {
69	int serverCount = 0;
70
71	for (int c = 0; nameservers[c]; c++) serverCount++;
72
73	if (serverCount == 0) {
74	error = GBS_global_string("No nameserver defined.");
75	}
76	else {
77	char *fieldNames = ARB_alloc<char>(serverCount);
78	for (int c = 0; c<serverCount; c++) {
79	const char *ipport = GBS_read_arb_tcp(nameservers[c]);
80	if (!ipport) {
81	error = GB_await_error();
82	fieldNames[c] = NULp;
83	}
84	else {
85	fieldNames[c] = nulldup(GBS_scan_arb_tcp_param(ipport, "-f")); // may return 0
86
87	// parameter -f contains default value (e.g. '-fstart=1')
88	if (fieldNames[c]) {
89	char *equal = strchr(fieldNames[c], '=');
90	if (equal) equal[0] = 0;
91	}
92	}
93	}
94
95	if (!error) {
96	if (serverCount == 1) { // exactly 1 server defined -> don't ask
97	error = set_addid(gb_main, fieldNames[0]);
98	}
99	else { // let the user select which nameserver to use
100	aw_assert(serverCount>1);
101
102	int len = serverCount; // commas+0term
103	const char *nofield = "None (only 'acc')";
104
105	for (int c = 0; c<serverCount; c++) {
106	if (fieldNames[c]) len += strlen(fieldNames[c]);
107	else len += strlen(nofield);
108	}
109
110	char *buttons = ARB_alloc<char>(len);
111	buttons[0] = 0;
112	for (int c = 0; c<serverCount; c++) {
113	if (c) strcat(buttons, ",");
114	strcat(buttons, fieldNames[c] ? fieldNames[c] : nofield);
115	}
116
117	int answer = aw_question("nameserv_select",
118	"Select if and which additional DB field you want to use",
119	buttons, false, "namesadmin.hlp");
120
121	error = set_addid(gb_main, fieldNames[answer]);
122
123	free(buttons);
124	}
125	}
126
127	for (int c = 0; c<serverCount; c++) free(fieldNames[c]);
128	free(fieldNames);
129	}
130	}
131	}
132
133	return error;
134	}
135
136
137	// ------------------------------------
138	// class NameServerConnection
139
140	class NameServerConnection : virtual Noncopyable {
141	aisc_com *link;
142	T_AN_LOCAL locs;
143	T_AN_MAIN com;
144	int persistent; // if true -> connection will not be closed
145	time_t linktime; // time, when link has been established
146
147	int init_local_com_names()
148	{
149	if (!link) return 1; //!* create and init local com structure **
150	if (aisc_create(link, AN_MAIN, com,
151	MAIN_LOCAL, AN_LOCAL, locs,
152	LOCAL_WHOAMI, "i bin der arb_tree",
153	NULp)) {
154	return 1;
155	}
156	return 0;
157	}
158
159	GB_ERROR reconnect(GBDATA *gb_main) { // reconnect ignoring consistency
160	int old_persistent = persistent;
161
162	printf("Reconnecting name server\n");
163
164	persistent = 0; // otherwise disconnect() won't disconnect
165	disconnect();
166	persistent = old_persistent; // restore previous persistence
167
168	return connect(gb_main);
169	}
170
171	char *fieldUsedByServer(GB_ERROR& err) {
172	char *field = NULp;
173	if (aisc_get(link, AN_MAIN, com,
174	MAIN_ADD_FIELD, &field,
175	NULp)) {
176	err = "Connection Problems with the NAME_SERVER";
177	aw_assert(!field);
178	}
179	return field;
180	}
181
182	GB_ERROR expectServerUsesField(const char *expected_field) {
183	GB_ERROR err = NULp;
184	char *server_field = fieldUsedByServer(err);
185
186	if (!err && strcmp(expected_field, server_field) != 0) {
187	err = GBS_global_string("Additional field doesn't match (expected='%s', server uses='%s')", expected_field, server_field);
188	}
189	free(server_field);
190	return err;
191	}
192
193	public:
194
195	NameServerConnection() {
196	link = NULp;
197	locs.clear();
198	com.clear();
199	persistent = 0;
200	}
201	virtual ~NameServerConnection() {
202	aw_assert(persistent == 0); // forgot to remove persistence ?
203	disconnect();
204	}
205
206	GB_ERROR connect(GBDATA *gb_main) {
207	aw_assert(!GB_have_error());
208
209	arb_progress::show_comment("Connecting to name server");
210
211	GB_ERROR err = NULp;
212	if (!link) {
213	const char *add_field = AW_get_nameserver_addid(gb_main);
214	const char *server_id = GBS_nameserver_tag(add_field);
215
216	err = arb_look_and_start_server(AISC_MAGIC_NUMBER, server_id);
217
218	if (!err) {
219	const char *ipport = GBS_read_arb_tcp(server_id);
220	if (!ipport) err = GB_await_error();
221	else {
222	link = aisc_open(ipport, com, AISC_MAGIC_NUMBER, &err);
223	linktime = time(NULp);
224
225	if (!err) {
226	if (init_local_com_names()) {
227	err = GBS_global_string("Can't connect %s %s", server_id, ipport);
228	}
229	else {
230	err = expectServerUsesField(add_field);
231	}
232	}
233	}
234	}
235	}
236	else {
237	long linkAge = int(time(NULp)-linktime);
238	bool doReconnect = false;
239
240	#if defined(DEBUG) && 0
241	// print information about name-server link age
242	static long lastage = -1;
243	if (linkAge != lastage) {
244	printf("Age of NameServerConnection: %li\n", linkAge);
245	lastage = linkAge;
246	}
247	#endif // DEBUG
248
249	if (linkAge > (5*60)) { // perform a reconnect after 5 minutes
250	// Reason : The pipe to the name server breaks after some time
251	doReconnect = true;
252	}
253	else {
254	const char *add_field = AW_get_nameserver_addid(gb_main);
255	GB_ERROR error = expectServerUsesField(add_field);
256
257	if (error) {
258	printf("Error: %s\n", error);
259	doReconnect = true;
260	}
261	}
262
263	if (doReconnect) {
264	err = reconnect(gb_main);
265	}
266	}
267	aw_assert(!GB_have_error());
268	return err;
269	}
270
271	void disconnect() {
272	if (persistent == 0) {
273	if (link) {
274	aisc_close(link, com);
275	locs.clear();
276	com.clear();
277	}
278	link = NULp;
279	}
280	}
281
282	void persistence(bool persist) {
283	if (persist) {
284	++persistent;
285	}
286	else {
287	--persistent;
288	if (persistent <= 0) {
289	persistent = 0;
290	disconnect();
291	}
292	}
293	}
294
295
296	aisc_com *getLink() { return link; }
297	const T_AN_LOCAL& getLocs() const { return locs; }
298	};
299
300	static NameServerConnection name_server;
301
302	PersistentNameServerConnection::PersistentNameServerConnection() {
303	name_server.persistence(true);
304	}
305	PersistentNameServerConnection::~PersistentNameServerConnection() {
306	name_server.persistence(false);
307	}
308
309	// --------------------------------------------------------------------------------
310
311	GB_ERROR AW_test_nameserver(GBDATA *gb_main) {
312	return name_server.connect(gb_main);
313	}
314
315	// --------------------------------------------------------------------------------
316
317	GB_ERROR AWTC_generate_one_name(GBDATA gb_main, const char full_name, const char acc, const char addid, char*& new_name) {
318	// create a unique short name for 'full_name'
319	// the result is written into 'new_name' (as malloc-copy)
320	// if fails: GB_ERROR!=0 && new_name==0
321	// acc and addid may be 0
322
323	new_name = NULp;
324	if (!acc) acc = "";
325
326	arb_progress progress("Generating species ID");
327
328	GB_ERROR err = name_server.connect(gb_main);
329	if (err) return err;
330
331	static char *shrt = NULp;
332	if (strlen(full_name)) {
333	if (aisc_nput(name_server.getLink(), AN_LOCAL, name_server.getLocs(),
334	LOCAL_FULL_NAME, full_name,
335	LOCAL_ACCESSION, acc,
336	LOCAL_ADDID, null2empty(addid),
337	LOCAL_ADVICE, "",
338	NULp)) {
339	err = "Connection Problems with the NAME_SERVER";
340	}
341	if (aisc_get(name_server.getLink(), AN_LOCAL, name_server.getLocs(),
342	LOCAL_GET_SHORT, &shrt,
343	NULp)) {
344	err = "Connection Problems with the NAME_SERVER";
345	}
346	}
347
348	if (err) {
349	free(shrt);
350	}
351	else {
352	if (shrt) {
353	new_name = shrt;
354	shrt = NULp;
355	}
356	else {
357	err = GB_export_errorf("Failed to generate species ID (shortname) for '%s'", full_name);
358	}
359	}
360
361	name_server.disconnect();
362
363	return err;
364	}
365
366	GB_ERROR AWTC_recreate_name(GBDATA *gb_species) {
367	GBDATA *gb_main = GB_get_root(gb_species);
368	arb_progress progress("Recreating species ID");
369	GB_ERROR error = name_server.connect(gb_main);
370
371	if (!error) {
372	const char *add_field = AW_get_nameserver_addid(gb_main);
373	char *ali_name = GBT_get_default_alignment(gb_main);
374
375	if (!ali_name) {
376	error = GB_await_error();
377	}
378	else {
379	GBDATA *gb_name = GB_entry(gb_species, "name");
380	GBDATA *gb_full_name = GB_entry(gb_species, "full_name");
381	GBDATA *gb_acc = GBT_gen_accession_number(gb_species, ali_name);
382	GBDATA *gb_addfield = add_field[0] ? GB_entry(gb_species, add_field) : NULp;
383
384	char *name = gb_name ? GB_read_string (gb_name) : strdup("");
385	char *full_name = gb_full_name ? GB_read_string (gb_full_name) : strdup("");
386	char *acc = gb_acc ? GB_read_string (gb_acc) : strdup("");
387	char *addid = gb_addfield ? GB_read_as_string(gb_addfield) : strdup("");
388
389	long deleted = 0;
390	char *shrt = NULp;
391
392	if (aisc_nput(name_server.getLink(), AN_LOCAL, name_server.getLocs(),
393	LOCAL_FULL_NAME, full_name,
394	LOCAL_ACCESSION, acc,
395	LOCAL_ADDID, addid,
396	LOCAL_ADVICE, "",
397	NULp) != 0 \|\|
398	aisc_get(name_server.getLink(), AN_LOCAL, name_server.getLocs(),
399	LOCAL_DEL_SHORT, &deleted,
400	NULp) != 0 \|\|
401	aisc_get(name_server.getLink(), AN_LOCAL, name_server.getLocs(),
402	LOCAL_GET_SHORT, &shrt,
403	NULp) != 0)
404	{
405	error = "Connection Problems with the NAME_SERVER";
406	}
407	name_server.disconnect();
408
409	if (!error) {
410	GBT_begin_rename_session(gb_main, 0);
411	error = GBT_rename_species(name, shrt, true);
412	if (error) {
413	if (GBT_find_species(gb_main, shrt)) { // it was a rename error
414	int done = 0;
415	error = NULp;
416	for (int count = 2; !done && !error && count<10; count++) {
417	const char *other_short = GBS_global_string("%s.%i", shrt, count);
418	if (!GBT_find_species(gb_main, other_short)) {
419	error = GBT_rename_species(name, other_short, true);
420	if (!error) done = 1;
421	}
422	}
423
424	if (!done && !error) {
425	error = "Failed to regenerate species ID. Please use 'Species/Synchronize IDs'";
426	}
427	}
428	}
429
430	if (error) GBT_abort_rename_session();
431	else error = GBT_commit_rename_session();
432	}
433
434	free(shrt);
435	free(addid);
436	free(acc);
437	free(full_name);
438	free(name);
439	free(ali_name);
440	}
441	}
442
443	return error;
444	}
445
446	char AWTC_create_numbered_suffix(GB_HASH species_name_hash, const char *shortname, GB_ERROR& warning) {
447	char *newshort = NULp;
448	if (GBS_read_hash(species_name_hash, shortname)) {
449	int i;
450	ARB_alloc(newshort, strlen(shortname)+20);
451	for (i = 1; ; i++) {
452	sprintf(newshort, "%s.%i", shortname, i);
453	if (!GBS_read_hash(species_name_hash, newshort))break;
454	}
455
456	warning =
457	"There are duplicated species!\n"
458	"The IDs of these species ('name') contain a '.' character followed by a number.\n"
459	"We strongly recommend you try understand and solve this problem\n"
460	"(see HELP in 'Species/Synchronize IDs' window)";
461	}
462	return newshort;
463	}
464
465	GB_ERROR AWTC_pars_names(GBDATA gb_main, bool isWarningPtr) {
466	// rename species according to name_server
467	// 'isWarning' is set to true, in case of duplicates-warning
468
469	arb_progress gen_progress("Generating new shortnames (IDs)");
470	GB_ERROR err = name_server.connect(gb_main);
471	bool isWarning = false;
472
473	if (!err) {
474	err = GBT_begin_rename_session(gb_main, 1);
475	if (!err) {
476	char *ali_name = GBT_get_default_alignment(gb_main);
477	GB_ERROR warning = NULp;
478
479	if (!ali_name) {
480	err = GB_await_error();
481	}
482	else {
483	long spcount = GBT_get_species_count(gb_main);
484	GB_HASH *hash = GBS_create_hash(spcount, GB_IGNORE_CASE);
485
486	if (spcount) {
487	arb_progress progress("Renaming species", spcount);
488	const char *add_field = AW_get_nameserver_addid(gb_main);
489
490	for (GBDATA *gb_species = GBT_first_species(gb_main);
491	gb_species && !err;
492	gb_species = GBT_next_species(gb_species))
493	{
494	GBDATA *gb_name = GB_entry(gb_species, "name");
495	GBDATA *gb_full_name = GB_entry(gb_species, "full_name");
496	GBDATA *gb_acc = GBT_gen_accession_number(gb_species, ali_name);
497	GBDATA *gb_addfield = add_field[0] ? GB_entry(gb_species, add_field) : NULp;
498
499	char *name = gb_name ? GB_read_string (gb_name) : strdup("");
500	char *full_name = gb_full_name ? GB_read_string (gb_full_name) : strdup("");
501	char *acc = gb_acc ? GB_read_string (gb_acc) : strdup("");
502	char *addid = gb_addfield ? GB_read_as_string(gb_addfield) : strdup(""); // empty value will be set to default by nameserver
503
504	char *shrt = NULp;
505
506	if (full_name[0] \|\| acc[0] \|\| addid[0]) {
507	if (aisc_nput(name_server.getLink(), AN_LOCAL, name_server.getLocs(),
508	LOCAL_FULL_NAME, full_name,
509	LOCAL_ACCESSION, acc,
510	LOCAL_ADDID, addid,
511	LOCAL_ADVICE, name,
512	NULp)) {
513	err = "Connection Problems with the NAME_SERVER";
514	}
515	if (aisc_get(name_server.getLink(), AN_LOCAL, name_server.getLocs(),
516	LOCAL_GET_SHORT, &shrt,
517	NULp)) {
518	err = "Connection Problems with the NAME_SERVER";
519	}
520	}
521	else {
522	shrt = strdup(name);
523	}
524	if (!err) {
525	char *newshrt = AWTC_create_numbered_suffix(hash, shrt, warning);
526	if (newshrt) freeset(shrt, newshrt);
527
528	GBS_incr_hash(hash, shrt);
529	err = GBT_rename_species(name, shrt, true);
530	}
531
532	free(shrt);
533	free(addid);
534	free(acc);
535	free(full_name);
536	free(name);
537
538	progress.inc_and_check_user_abort(err);
539	}
540	}
541	else {
542	gen_progress.sub_progress_skipped(); // trigger skipped subcounter
543	}
544	GBS_free_hash(hash);
545	free(ali_name);
546	}
547
548	if (err) GBT_abort_rename_session();
549	else err = GBT_commit_rename_session();
550
551	if (!err) {
552	err = warning;
553	if (warning) isWarning = true;
554	}
555	}
556	name_server.disconnect();
557	}
558
559	if (isWarningPtr) *isWarningPtr = isWarning;
560	gen_progress.done(); // needed if err
561
562	return err;
563	}
564
565
566	static void awt_rename_cb(AW_window aww, GBDATA gb_main) {
567	GB_ERROR error = AWTC_pars_names(gb_main, NULp);
568	if (error) aw_message(error);
569	aww->get_root()->awar(AWAR_TREE_REFRESH)->touch();
570	}
571
572
573	AW_window AWTC_create_rename_window(AW_root root, GBDATA *gb_main) {
574	AW_window_simple *aws = new AW_window_simple;
575	aws->init(root, "AUTORENAME_SPECIES", "Synchronize species IDs");
576
577	aws->load_xfig("awtc/autoren.fig");
578
579	aws->at("close");
580	aws->callback(AW_POPDOWN);
581	aws->create_button("CLOSE", "CLOSE", "C");
582
583	aws->at("help");
584	aws->callback(makeHelpCallback("rename.hlp"));
585	aws->create_button("HELP", "HELP", "H");
586
587	aws->at("config");
588	aws->callback(makeWindowCallback(AW_popup_namesadmin_window, gb_main));
589	aws->create_autosize_button("CONFIG_NAMESERVER", "Configure nameserver", "n");
590
591	aws->at("go");
592	aws->highlight();
593	aws->callback(makeWindowCallback(awt_rename_cb, gb_main));
594	aws->create_button("GO", "GO", "G");
595
596	return aws;
597	}
598
599	UniqueNameDetector::UniqueNameDetector(GBDATA *gb_item_data, long additionalEntries) {
600	hash = GBS_create_hash(GB_number_of_subentries(gb_item_data)+additionalEntries, GB_IGNORE_CASE);
601
602	for (GBDATA *gb_item = GB_child(gb_item_data); gb_item; gb_item = GB_nextChild(gb_item)) {
603	GBDATA *gb_name = GB_entry(gb_item, "name");
604	if (gb_name) { // item has name -> insert to hash
605	GBS_write_hash(hash, GB_read_char_pntr(gb_name), 1);
606	}
607	}
608	}
609
610	UniqueNameDetector::~UniqueNameDetector() { GBS_free_hash(hash); }
611
612	static char makeUniqueShortName(const char prefix, UniqueNameDetector& existing) {
613	// generates a non-existing short-name (name starts with prefix)
614	//
615	// returns NULp if it fails
616
617	char *result = NULp;
618	int prefix_len = strlen(prefix);
619
620	aw_assert(prefix_len<8); // prefix has to be shorter than 8 chars!
621	if (prefix_len<8) {
622	const int max_nums[8] = { 100000000, 10000000, 1000000, 100000, 10000, 1000, 100, 10 };
623	static int next_try[8] = { 0, 0, 0, 0, 0, 0, 0, 0 };
624
625	int max_num = max_nums[prefix_len];
626	char short_name[9];
627	strcpy(short_name, prefix);
628
629	char *dig_pos = short_name+prefix_len;
630	int num = next_try[prefix_len];
631	int stop = num ? num-1 : max_num;
632
633	while (num != stop) {
634	sprintf(dig_pos, "%i", num);
635	++num;
636	if (!existing.name_known(short_name)) {
637	result = strdup(short_name);
638	break;
639	}
640	if (num == max_num && stop != max_num) num = 0;
641	}
642	if (num == max_num) num = 0;
643	next_try[prefix_len] = num;
644	}
645	return result;
646	}
647
648	char AWTC_makeUniqueShortName(const char prefix, UniqueNameDetector& existingNames) {
649	// generates a unique species name from prefix
650	// (prefix will be fillup with zero digits and then shortened down to first char)
651	//
652	// returns NULp if failed (and exports error)
653
654	int len = strlen(prefix);
655	char p[9];
656	strncpy(p, prefix, 8);
657
658	if (len>8) len = 8;
659	else {
660	if (len == 0) p[len++] = 'x'; // don't use digit as first character
661	while (len<8) p[len++] = '0';
662	}
663
664	p[len] = 0;
665
666	char *result = NULp;
667
668	for (int l = len-1; l>0 && !result; --l) {
669	p[l] = 0;
670	result = makeUniqueShortName(p, existingNames);
671	}
672
673	aw_assert(!result \|\| strlen(result) <= 8);
674	if (!result) GB_export_errorf("Failed to create unique species ID (prefix='%s')", prefix);
675
676	return result;
677	}
678
679	char *AWTC_generate_random_name(UniqueNameDetector& existingNames) {
680	char *new_species_name = NULp;
681	char short_name[9];
682	int count = 10000;
683
684	short_name[8] = 0;
685	while (count--) {
686	short_name[0] = 'a'+GB_random(26); // first character has to be alpha
687
688	for (int x=1; x<8; ++x) {
689	int r = GB_random(36); // rest may be alphanumeric
690	short_name[x] = r<10 ? ('0'+r) : ('a'+r-10);
691	}
692
693	if (!existingNames.name_known(short_name)) {
694	new_species_name = strdup(short_name);
695	break;
696	}
697	}
698
699	if (!new_species_name) {
700	aw_message("Failed to generate a random name - retrying (this might hang forever)");
701	return AWTC_generate_random_name(existingNames);
702	}
703
704	return new_species_name;
705	}
706
707	int AWTC_name_quality(const char *short_name) {
708	// result 0 = ok for external tools
709	// 1 = ok for ARB
710	// 2 = not ok
711
712	int len = -1;
713	int alnum_count = 0;
714	int ascii_count = 0;
715
716	while (char c = short_name[++len]) {
717	alnum_count += (isalnum(c) != 0);
718	ascii_count += (c > 32 && c < 127);
719	}
720
721	if (len>0) {
722	if (len <= 8) {
723	if (len == alnum_count) return 0; // ok for external programs
724	}
725	if (len == ascii_count) return 1; // ok for ARB
726	}
727	return 2; // not ok
728	}

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

Context Navigation

source: branches/ali/SL/AW_NAME/AW_rename.cxx

Download in other formats: