Context Navigation

source: branches/stable/ARBDB/adlang1.cxx

Visit:

Last change on this file was 18311, checked in by westram, 4 years ago
reintegrates 'saicalc' into 'trunk' implements #730 adds: log:branches/saicalc@18141:18310
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 95.3 KB

Line
1	// =============================================================== //
2	// //
3	// File : adlang1.cxx //
4	// Purpose : //
5	// //
6	// Institute of Microbiology (Technical University Munich) //
7	// http://www.arb-home.de/ //
8	// //
9	// =============================================================== //
10
11	#include "gb_aci_impl.h"
12	#include "gb_key.h"
13
14	#include "TreeNode.h"
15
16	#include <aw_awar_defs.hxx>
17
18	#include <adGene.h>
19	#include <ad_cb.h>
20
21	#include <arb_defs.h>
22	#include <arb_strbuf.h>
23	#include <arb_file.h>
24	#include <arb_strarray.h>
25	#include <arb_sort.h>
26
27	#include <cctype>
28	#include <cmath>
29	#include <algorithm>
30
31	// hook for 'export_sequence'
32
33	static gb_export_sequence_cb get_export_sequence = NULp;
34
35	NOT4PERL void GB_set_export_sequence_hook(gb_export_sequence_cb escb) {
36	gb_assert(!get_export_sequence \|\| !escb); // avoid unwanted overwrite
37	get_export_sequence = escb;
38	}
39
40	using namespace GBL_IMPL;
41
42	namespace GBL_IMPL {
43	// global ACI/SRT debug switch
44	int traceACI = 0;
45	int traceIndent = 0;
46
47	void print_trace(const char *text) {
48	FILE *DUMPOUT = stdout;
49
50	gb_assert(traceACI>0);
51	if (traceIndent>0) {
52	ConstStrArray line;
53	GBT_split_string(line, text, "\n", true);
54	for (unsigned L = 0; L<line.size(); ++L) {
55	for (int i = 0; i<traceIndent; ++i) {
56	fputc(' ', DUMPOUT);
57	}
58	fputs(line[L], DUMPOUT);
59	fputc('\n', DUMPOUT);
60	}
61	}
62	else {
63	fputs(text, DUMPOUT);
64	}
65	fflush(DUMPOUT);
66	}
67
68	GB_ERROR trace_params(const GBL_streams& param, gbl_param ppara, const char com) {
69	GB_ERROR error = NULp;
70	int i;
71
72	int argc = param.size();
73	for (i=0; i<argc; i++) {
74	gbl_param *para;
75	const char *argument = param.get(i);
76
77	for (para = ppara; para && !error; para = para->next) {
78	if (para->param_name) { // NULp means param is inactive (see PARAM_IF)
79	int len = strlen(para->param_name);
80
81	if (strncmp(argument, para->param_name, len) == 0) {
82	const char *value = argument+len; // set to start of value
83
84	if (para->type == GB_BIT) {
85	// GB_BIT is special cause param_name does NOT contain trailing '='
86
87	if (!value[0]) { // only 'name' -> handle like 'name=1'
88	;
89	}
90	else if (value[0] == '=') {
91	value++;
92	}
93	}
94
95	switch (para->type) {
96	case GB_STRING:
97	(const char *)para->varaddr = value;
98	break;
99
100	case GB_INT:
101	STATIC_ASSERT(sizeof(int) == sizeof(nat)); // assumed by GBL_PARAM_UINT
102	(int )para->varaddr = atoi(value);
103	break;
104
105	case GB_BIT:
106	// 'param=' is same as 'param' or 'param=1' (historical reason, don't change)
107	(int )para->varaddr = (value[0] ? atoi(value) : 1);
108	break;
109
110	case GB_BYTE:
111	(char )para->varaddr = *value; // this may use the terminal zero-byte (e.g. for p1 in 'p0=0,p1=,p2=2' )
112	if (value[0] && value[1]) { // found at least 2 chars
113	GB_warningf("Only one character expected in value '%s' of param '%s' - rest is ignored", value, para->param_name);
114	}
115	break;
116
117	default:
118	gb_assert(0);
119	error = GBS_global_string("Parameter '%s': Unknown type %i (internal error)", para->param_name, para->type);
120	break;
121	}
122	break; // accept parameter
123	}
124	}
125	}
126
127	if (!error && !para) { // no parameter found for argument
128	int pcount = 0;
129	gbl_param **params;
130	int k;
131	char *res;
132	GBS_strstruct *str = GBS_stropen(1000);
133	GB_ERROR err;
134
135
136	for (para = ppara; para; para = para->next) pcount++;
137	ARB_calloc(params, pcount);
138	for (k = 0, para = ppara; para; para = para->next) params[k++] = para;
139
140
141	for (pcount--; pcount>=0; pcount--) {
142	para = params[pcount];
143	if (para->param_name) {
144	GBS_strcat(str, " ");
145	GBS_strcat(str, para->param_name);
146	switch (para->type) {
147	case GB_STRING: GBS_strcat(str, "STRING"); break;
148	case GB_INT: GBS_strcat(str, "INT"); break;
149	case GB_FLOAT: GBS_strcat(str, "FLOAT"); break;
150	case GB_BYTE: GBS_strcat(str, "CHAR"); break;
151	case GB_BIT: GBS_strcat(str, " "); break;
152	default: gb_assert (0); GBS_strcat(str, "????"); break;
153	}
154	GBS_strcat(str, "\t\t;");
155	GBS_strcat(str, para->help_text);
156	GBS_strcat(str, "\n");
157	}
158	}
159	freenull(params);
160	res = GBS_strclose(str);
161	err = GB_export_errorf("Unknown Parameter '%s' in command '%s'\n PARAMETERS:\n%s", argument, com, res);
162	free(res);
163	return err;
164	}
165	}
166
167	return error;
168	}
169	};
170
171
172
173	// -------------------------
174	// String functions
175
176	static int gbl_stricmp(const char s1, const char s2) {
177	// case insensitive strcmp
178	int i;
179	for (i = 0; ; ++i) {
180	char c1 = tolower(s1[i]);
181	char c2 = tolower(s2[i]);
182
183	if (c1 == c2) {
184	if (!c1) break; // equal strings
185	}
186	else {
187	if (c1<c2) return -1;
188	return 1;
189	}
190	}
191	return 0;
192	}
193	static int gbl_strincmp(const char s1, const char s2, int size2) {
194	// case insensitive strcmp
195	int i;
196	for (i = 0; i<size2; ++i) {
197	char c1 = tolower(s1[i]);
198	char c2 = tolower(s2[i]);
199
200	if (c1 == c2) {
201	if (!c1) break; // equal strings
202	}
203	else {
204	if (c1<c2) return -1;
205	return 1;
206	}
207	}
208	return 0;
209	}
210	static const char gbl_stristr(const char haystack, const char *needle) {
211	// case insensitive strstr
212	const char *hp = haystack;
213	char c1 = toupper(needle[0]);
214	char c2 = tolower(c1);
215	int needle_size = strlen(needle);
216
217	if (c1 == c2) {
218	hp = strchr(hp, c1);
219	while (hp) {
220	if (gbl_strincmp(hp, needle, needle_size) == 0) return hp;
221	hp = strchr(hp+1, c1);
222	}
223	}
224	else {
225	while (hp) {
226	const char *h1 = strchr(hp, c1);
227	const char *h2 = strchr(hp, c2);
228
229	if (h1 && h2) {
230	if (h1<h2) {
231	if (gbl_strincmp(h1, needle, needle_size) == 0) return h1;
232	hp = h1+1;
233	}
234	else {
235	gb_assert(h1>h2);
236	if (gbl_strincmp(h2, needle, needle_size) == 0) return h2;
237	hp = h2+1;
238	}
239	}
240	else {
241	if (h1) { hp = h1; }
242	else if (h2) { hp = h2; c1 = c2; }
243	else { hp = NULp; }
244
245	while (hp) {
246	if (gbl_strincmp(hp, needle, needle_size) == 0) return hp;
247	hp = strchr(hp+1, c1);
248	}
249	}
250	}
251	}
252	return NULp;
253	}
254
255	inline int approve_pos(int pos, int len) { return pos<0 ? (-pos<len ? len+pos : 0) : pos; }
256
257	static GB_ERROR gbl_mid_streams(const GBL_streams& arg_input, GBL_streams& arg_out, int start, int end) {
258	// used as well to copy all streams (e.g. by 'dd')
259	for (int i=0; i<arg_input.size(); i++) {
260	const char *p = arg_input.get(i);
261	int len = strlen(p);
262
263	int s = approve_pos(start, len);
264	int e = approve_pos(end, len);
265
266	char *res;
267	if (s >= len \|\| e<s) {
268	res = ARB_strdup("");
269	}
270	else {
271	gb_assert(s >= 0);
272	res = ARB_strpartdup(p+s, p+e);
273	}
274	arg_out.insert(res);
275	}
276	return NULp;
277	}
278
279	static GB_ERROR gbl_trace(GBL_command_arguments *args) {
280	int tmp_trace;
281
282	EXPECT_PARAMS(args, 1, "0\|1");
283
284	tmp_trace = atoi(args->get_param(0));
285	if (tmp_trace<0 \|\| tmp_trace>1) return GBS_global_string("Illegal value %i to trace", tmp_trace);
286
287	if (tmp_trace != traceACI) {
288	traceACI = 1;
289	print_trace(GBS_global_string("%sctivated ACI trace\n", tmp_trace ? "A" : "De-a"));
290	traceACI = tmp_trace;
291	}
292
293	return gbl_mid_streams(args->input, args->output, 0, -1); // copy all streams
294	}
295
296	/* ---------------------------------------------------------------------------------------
297	* Binary operators work on pairs of values.
298	* Three different operational modes are implemented for all binary operators:
299	*
300	* 1. inputstreams\|operator
301	*
302	* The number of inputstreams has to be even and the operator will be
303	* applied to pairs of them.
304	*
305	* Example : a;b;c;d;e;f \| plus
306	* Result : a+b;c+d;e+f
307	*
308	* 2. inputstreams\|operator(x)
309	*
310	* The number of inputstreams has to be at least 1.
311	* The operator is applied to each inputstream.
312	*
313	* Example : a;b;c \| plus(d)
314	* Result : a+d;b+d;c+d
315	*
316	* 3. operator(x, y)
317	*
318	* @@@ this decription does not match behavior!
319	* @@@ check description in helpfile as well
320	*
321	* Inputstreams will be ignored and the operator is applied
322	* to the arguments
323	*
324	* Example : a;b \| plus(c,d)
325	* Result : c+d
326	*/
327
328	template <typename T>
329	GB_ERROR gbl_apply_binary_operator(GBL_command_arguments args, char (op)(const char , const char *, T), T client_data) {
330	GB_ERROR error = NULp;
331	switch (args->param_count()) {
332	case 0:
333	gb_assert(args->set_params_checked());
334	if (args->input.size() == 0) error = "Expect at least two input streams if called with 0 parameters";
335	else if (args->input.size()%2) error = "Expect an even number of input streams if called with 0 parameters";
336	else {
337	int inputpairs = args->input.size()/2;
338	int i;
339	for (i = 0; i<inputpairs; ++i) {
340	PASS_2_OUT(args, op(args->input.get(i2), args->input.get(i2+1), client_data));
341	}
342	}
343	break;
344
345	case 1:
346	gb_assert(args->set_params_checked());
347	if (args->input.size() == 0) error = "Expect at least one input stream if called with 1 parameter";
348	else {
349	int i;
350	const char *argument = args->get_param(0);
351	for (i = 0; i<args->input.size(); ++i) {
352	PASS_2_OUT(args, op(args->input.get(i), argument, client_data));
353	}
354	}
355	break;
356
357	case 2:
358	gb_assert(args->set_params_checked());
359	for (int i = 0; i<args->input.size(); ++i) {
360	char *result1 = args->get_callEnv().interpret_subcommand(args->input.get(i), args->get_param(0)); // @@@ EVALUATED_PARAM (#768)
361	if (!result1) error = GB_await_error();
362	else {
363	char *result2 = args->get_callEnv().interpret_subcommand(args->input.get(i), args->get_param(1)); // @@@ EVALUATED_PARAM (#768)
364	if (!result2) error = GB_await_error();
365	else {
366	PASS_2_OUT(args, op(result1, result2, client_data));
367	free(result2);
368	}
369	free(result1);
370	}
371	}
372	break;
373
374	default:
375	error = check_optional_parameters(args, 0, NULp, 2, "Expr1[,Expr2]", true, false);
376	break;
377	}
378
379	return error;
380	}
381
382	// --------------------------------
383	// escape/unescape strings
384
385	static char unEscapeString(const char escapedString) {
386	// replaces all \x by x
387	char *result = nulldup(escapedString);
388	char *to = result;
389	char *from = result;
390
391	while (1) {
392	char c = *from++;
393	if (!c) break;
394
395	if (c=='\\') {
396	to++ = from++;
397	}
398	else {
399	*to++ = c;
400	}
401	}
402	*to = 0;
403	return result;
404	}
405	static char escapeString(const char unescapedString) {
406	// replaces all '\' and '"' by '\\' and '\"'
407	int len = strlen(unescapedString);
408	char result = ARB_alloc<char>(2len+1);
409	char *to = result;
410	const char *from = unescapedString;
411
412	while (1) {
413	char c = *from++;
414	if (!c) break;
415
416	if (c=='\\' \|\| c == '\"') {
417	*to++ = '\\';
418	*to++ = c;
419	}
420	else {
421	*to++ = c;
422	}
423	}
424	*to = 0;
425	return result;
426	}
427
428	// ---------------------------------
429	// the commands themselves:
430
431	static GB_ERROR gbl_quote(GBL_command_arguments *args) {
432	EXPECT_NO_PARAM(args);
433
434	for (int i=0; i<args->input.size(); i++) {
435	FORMAT_2_OUT(args, "\"%s\"", args->input.get(i));
436	}
437	return NULp;
438	}
439	static GB_ERROR gbl_unquote(GBL_command_arguments *args) {
440	EXPECT_NO_PARAM(args);
441
442	for (int i=0; i<args->input.size(); i++) {
443	const char *str = args->input.get(i);
444	int len = strlen(str);
445
446	if (str[0] == '\"' && str[len-1] == '\"') {
447	PASS_2_OUT(args, ARB_strpartdup(str+1, str+len-2));
448	}
449	else {
450	IN_2_OUT(args, i);
451	}
452	}
453	return NULp;
454	}
455
456	static GB_ERROR gbl_escape(GBL_command_arguments *args) {
457	EXPECT_NO_PARAM(args);
458
459	for (int i=0; i<args->input.size(); i++) {
460	char *escaped = escapeString(args->input.get(i));
461	PASS_2_OUT(args, escaped);
462	}
463	return NULp;
464	}
465	static GB_ERROR gbl_unescape(GBL_command_arguments *args) {
466	EXPECT_NO_PARAM(args);
467
468	for (int i=0; i<args->input.size(); i++) {
469	char *unescaped = unEscapeString(args->input.get(i));
470	PASS_2_OUT(args, unescaped);
471	}
472	return NULp;
473	}
474
475	static GB_ERROR gbl_command(GBL_command_arguments *args) {
476	EXPECT_PARAMS(args, 1, "\"ACI command\"");
477
478	GB_ERROR error = NULp;
479	char *command = unEscapeString(args->get_param(0));
480
481	if (traceACI) {
482	print_trace(GBS_global_string("executing command '%s'\n", command));
483	}
484
485	for (int i=0; i<args->input.size() && !error; i++) {
486	char *result = args->get_callEnv().interpret_subcommand(args->input.get(i), command);
487	if (!result) error = GB_await_error();
488	else PASS_2_OUT(args, result);
489	}
490	free(command);
491	return error;
492	}
493
494	static GB_ERROR gbl_eval(GBL_command_arguments *args) {
495	EXPECT_PARAMS(args, 1, "\"expression evaluating to ACI command\"");
496
497	GB_ERROR error = NULp;
498	char *to_eval = unEscapeString(args->get_param(0));
499	TRACE_ACI(GBS_global_string("evaluating '%s'\n", to_eval));
500
501	char *command = args->get_callEnv().interpret_subcommand("", to_eval); // evaluate independent
502	if (!command) error = GB_await_error();
503	else {
504	TRACE_ACI(GBS_global_string("executing '%s'\n", command));
505
506	for (int i=0; i<args->input.size() && !error; i++) {
507	char *result = args->get_callEnv().interpret_subcommand(args->input.get(i), command);
508	if (!result) error = GB_await_error();
509	else PASS_2_OUT(args, result);
510	}
511	free(command);
512	}
513	free(to_eval);
514	return error;
515	}
516
517	class DefinedCommands : virtual Noncopyable {
518	GB_HASH *cmds;
519	public:
520	DefinedCommands() { cmds = GBS_create_dynaval_hash(100, GB_MIND_CASE, GBS_dynaval_free); }
521	~DefinedCommands() { GBS_free_hash(cmds); }
522
523	void set(const char name, char cmd) { GBS_dynaval_free(GBS_write_hash(cmds, name, (long)cmd)); } // takes ownership of 'cmd'!
524	const char get(const char name) const { return (const char *)GBS_read_hash(cmds, name); }
525	};
526
527	static DefinedCommands defined_commands;
528
529	static GB_ERROR gbl_define(GBL_command_arguments *args) {
530	COMMAND_DROPS_INPUT_STREAMS(args);
531	EXPECT_PARAMS(args, 2, "name, \"ACI command\"");
532
533	const char *name = args->get_param(0);
534	char *cmd = unEscapeString(args->get_param(1));
535
536	defined_commands.set(name, cmd);
537	TRACE_ACI(GBS_global_string("defining command '%s'='%s'\n", name, cmd));
538	return NULp;
539	}
540
541	static GB_ERROR gbl_do(GBL_command_arguments *args) {
542	EXPECT_PARAMS(args, 1, "definedCommandName");
543
544	GB_ERROR error = NULp;
545	const char *name = args->get_param(0);
546	const char *cmd = defined_commands.get(name);
547
548	if (!cmd) {
549	error = GBS_global_string("Can't do undefined command '%s' - use define(%s, ...) first", name, name);
550	}
551	else {
552	TRACE_ACI(GBS_global_string("executing defined command '%s'='%s' on %i streams\n", name, cmd, args->input.size()));
553
554	for (int i=0; i<args->input.size() && !error; i++) {
555	char *result = args->get_callEnv().interpret_subcommand(args->input.get(i), cmd);
556	if (!result) error = GB_await_error();
557	else PASS_2_OUT(args, result);
558	}
559	}
560	return error;
561	}
562
563	static GB_ERROR gbl_streams(GBL_command_arguments *args) {
564	EXPECT_NO_PARAM(args);
565
566	FORMAT_2_OUT(args, "%i", args->input.size());
567	return NULp;
568	}
569
570	static GB_ERROR expect_used_in_genome_db(GBL_command_arguments *args) {
571	if (GEN_is_genome_db(args->get_gb_main(), -1)) return NULp;
572	return GBS_global_string("ACI command '%s' can only be used in genome databases.", args->get_cmdName());
573	}
574
575	static GB_ERROR apply_to_origin(GBL_command_arguments *args, bool organism) {
576	EXPECT_PARAMS(args, 1, "\"ACI command\"");
577	EXPECT_ITEM_REFERENCED(args);
578
579	GB_ERROR error = expect_used_in_genome_db(args);
580	if (!error) {
581	if (!GEN_is_pseudo_gene_species(args->get_item_ref())) {
582	error = GBS_global_string("'%s' applies to gene-species only", args->get_cmdName());
583	}
584	else {
585	GBDATA *gb_origin = NULp;
586	if (organism) {
587	gb_origin = GEN_find_origin_organism(args->get_item_ref(), NULp);
588	}
589	else {
590	gb_origin = GEN_find_origin_gene(args->get_item_ref(), NULp);
591	}
592
593	if (!error && !gb_origin) error = GB_await_error();
594
595	if (!error) {
596	char *command = unEscapeString(args->get_param(0));
597	GBL_call_env callEnv(gb_origin, args->get_env()); // refer to gb_origin for subcommands
598	// Note: if calling env has a FieldTracker, field access from 'command' is not tracked.
599	// That access applies to different item.
600
601	for (int i=0; i<args->input.size() && !error; i++) {
602	char *result = callEnv.interpret_subcommand(args->input.get(i), command);
603	if (!result) error = GB_await_error();
604	else PASS_2_OUT(args, result);
605	}
606
607	free(command);
608	}
609	}
610	}
611	return error;
612	}
613
614	static GB_ERROR gbl_origin_gene(GBL_command_arguments *args) { return apply_to_origin(args, false); }
615	static GB_ERROR gbl_origin_organism(GBL_command_arguments *args) { return apply_to_origin(args, true); }
616
617
618	static GB_ERROR applyToItemFoundByKey(GBL_command_arguments args, const char itemname, GBDATA gb_item_data, const char key) {
619	GB_ERROR error = NULp;
620	char *command = unEscapeString(args->get_param(0));
621
622	for (int i=0; i<args->input.size() && !error; i++) {
623	const char *in = args->input.get(i);
624	if (in[0]) { // silently ignore empty input streams
625	GBDATA *gb_item = NULp;
626	{
627	GBDATA *gb_field = GB_find_string(gb_item_data, key, in, GB_IGNORE_CASE, SEARCH_GRANDCHILD);
628	if (gb_field) {
629	gb_item = GB_get_father(gb_field);
630	}
631	else {
632	error = GBS_global_string("No %s with %s '%s' found.", itemname, key, in);
633	}
634	}
635	if (gb_item) {
636	GBL_call_env callEnv(gb_item, args->get_env()); // refer to gb_item for subcommands
637	// Note: if calling env has a FieldTracker, field access from 'command' is not tracked.
638	// That access applies to different item.
639
640	char *result = callEnv.interpret_subcommand("", command);
641	if (!result) error = GB_await_error();
642	else PASS_2_OUT(args, result);
643	}
644	else {
645	if (!error) error = GB_await_error();
646	}
647	}
648	}
649
650	free(command);
651	return error;
652	}
653	static GB_ERROR gbl_findspec(GBL_command_arguments *args) {
654	EXPECT_PARAMS(args, 1, "\"ACI command\"");
655	return applyToItemFoundByKey(args, "species", GBT_get_species_data(args->get_gb_main()), "name");
656	}
657	static GB_ERROR gbl_findacc(GBL_command_arguments *args) {
658	EXPECT_PARAMS(args, 1, "\"ACI command\"");
659	return applyToItemFoundByKey(args, "species", GBT_get_species_data(args->get_gb_main()), "acc");
660	}
661
662	static GB_ERROR gbl_findgene(GBL_command_arguments *args) {
663	EXPECT_PARAMS(args, 1, "\"ACI command\"");
664	EXPECT_ITEM_REFERENCED(args);
665
666	GB_ERROR error = expect_used_in_genome_db(args);
667	if (!error) {
668	GBDATA *gb_item = args->get_item_ref();
669	if (GEN_is_organism(gb_item)) {
670	error = applyToItemFoundByKey(args, "gene", GEN_find_gene_data(gb_item), "name");
671	}
672	else if (strcmp(GB_read_key_pntr(gb_item), "gene") == 0) {
673	// if applied to gene -> find "brother" of gene
674	GBDATA *gb_organism = GB_get_grandfather(gb_item);
675	if (GEN_is_organism(gb_organism)) {
676	error = applyToItemFoundByKey(args, "gene", GEN_find_gene_data(gb_organism), "name");
677	}
678	else {
679	error = "'findgene' cannot be used here (was applied to 'gene', but could not find gene-owning organism)";
680	}
681	}
682	else {
683	error = GBS_global_string("'findgene' cannot be applied to '%s' (need an organism)",
684	GBT_get_name_or_description(gb_item));
685	}
686	}
687	return error;
688	}
689
690	class Tab {
691	bool tab[256];
692	public:
693	Tab(bool take, const char *invert) {
694	bool init = !take;
695	for (int i = 0; i<256; ++i) tab[i] = init;
696	for (int i = 0; invert[i]; ++i) tab[safeCharIndex(invert[i])] = take;
697	}
698	bool operator[](int i) const { return tab[i]; }
699	};
700
701	inline GB_ERROR count_by_tab(GBL_command_arguments *args, const Tab& tab) {
702	for (int i=0; i<args->input.size(); ++i) {
703	long sum = 0; // count frequencies
704	const char *p = args->input.get(i);
705
706	while (p) sum += tab[safeCharIndex((p++))];
707	FORMAT_2_OUT(args, "%li", sum);
708	}
709	return NULp;
710	}
711	inline GB_ERROR remove_by_tab(GBL_command_arguments *args, const Tab& tab) {
712	for (int i=0; i<args->input.size(); ++i) {
713	GBS_strstruct *strstruct = GBS_stropen(1000);
714	for (const char p = args->input.get(i); p; p++) {
715	if (!tab[(unsigned int)*p]) {
716	GBS_chrcat(strstruct, *p);
717	}
718	}
719	PASS_2_OUT(args, GBS_strclose(strstruct));
720	}
721	return NULp;
722	}
723
724	static GB_ERROR gbl_count(GBL_command_arguments *args) {
725	EXPECT_PARAMS(args, 1, "\"characters to count\"");
726	return count_by_tab(args, Tab(true, args->get_param(0)));
727	}
728	static GB_ERROR gbl_len(GBL_command_arguments *args) {
729	EXPECT_OPTIONAL_PARAMS(args, 0, NULp, 1, "\"characters not to count\"");
730	const char *exclude = args->get_optional_param(0, "");
731	return count_by_tab(args, Tab(false, exclude));
732	}
733	static GB_ERROR gbl_remove(GBL_command_arguments *args) {
734	EXPECT_PARAMS(args, 1, "\"characters to remove\"");
735	return remove_by_tab(args, Tab(true, args->get_param(0)));
736	}
737	static GB_ERROR gbl_keep(GBL_command_arguments *args) {
738	EXPECT_PARAMS(args, 1, "\"characters to keep\"");
739	return remove_by_tab(args, Tab(false, args->get_param(0)));
740	}
741
742
743	static char binop_compare(const char arg1, const char *arg2, bool case_sensitive) {
744	int result;
745
746	if (case_sensitive) result = strcmp(arg1, arg2);
747	else result = gbl_stricmp(arg1, arg2);
748
749	return GBS_global_string_copy("%i", result<0 ? -1 : (result>0 ? 1 : 0));
750	}
751	static char binop_equals(const char arg1, const char *arg2, bool case_sensitive) {
752	int result;
753
754	if (case_sensitive) result = strcmp(arg1, arg2);
755	else result = gbl_stricmp(arg1, arg2);
756
757	return GBS_global_string_copy("%i", result == 0 ? 1 : 0);
758	}
759	static char binop_contains(const char arg1, const char *arg2, bool case_sensitive) {
760	const char *found = NULp;
761
762	if (!arg2[0]) return strdup("0"); // do not report matches of empty string
763
764	if (case_sensitive) found = strstr(arg1, arg2);
765	else found = gbl_stristr(arg1, arg2);
766
767	return GBS_global_string_copy("%ti", found ? (found-arg1)+1 : 0);
768	}
769	static char binop_partof(const char arg1, const char *arg2, bool case_sensitive) {
770	return binop_contains(arg2, arg1, case_sensitive);
771	}
772
773	static GB_ERROR gbl_compare (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, binop_compare, true); }
774	static GB_ERROR gbl_icompare (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, binop_compare, false); }
775	static GB_ERROR gbl_equals (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, binop_equals, true); }
776	static GB_ERROR gbl_iequals (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, binop_equals, false); }
777	static GB_ERROR gbl_contains (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, binop_contains, true); }
778	static GB_ERROR gbl_icontains(GBL_command_arguments *args) { return gbl_apply_binary_operator(args, binop_contains, false); }
779	static GB_ERROR gbl_partof (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, binop_partof, true); }
780	static GB_ERROR gbl_ipartof (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, binop_partof, false); }
781
782	static GB_ERROR gbl_isEmpty(GBL_command_arguments *args) {
783	EXPECT_NO_PARAM(args);
784	for (int i=0; i<args->input.size(); i++) {
785	const char *str = args->input.get(i);
786	FORMAT_2_OUT(args, "%i", str[0] == 0);
787	}
788	return NULp;
789	}
790	static GB_ERROR gbl_inRange(GBL_command_arguments *args) {
791	EXPECT_PARAMS(args, 2, "low,high");
792
793	double low = strtod(args->get_param(0), NULp);
794	double high = strtod(args->get_param(1), NULp);
795
796	for (int i=0; i<args->input.size(); i++) {
797	double val = strtod(args->input.get(i), NULp);
798	bool inRange = low<=val && val<=high;
799	FORMAT_2_OUT(args, "%i", inRange);
800	}
801	return NULp;
802	}
803
804
805	static GB_ERROR gbl_translate(GBL_command_arguments *args) {
806	EXPECT_OPTIONAL_PARAMS(args, 2, "old,new", 1, "other");
807
808	const char *other = args->get_optional_param(2, NULp);
809	if (other && (other[0] == 0 \|\| other[1] != 0)) {
810	return "third parameter of translate has to be one character (i.e. \"-\")";
811	}
812	const char replace_other = other ? other[0] : 0;
813
814	// build translation table :
815	unsigned char tab[256];
816	{
817	const unsigned char o = (const unsigned char )args->get_param(0);
818	const unsigned char n = (const unsigned char )args->get_param(1);
819	char used[256];
820
821	if (strlen((const char )o) != strlen((const char )n)) {
822	return "arguments 1 and 2 of translate should be strings with identical length";
823	}
824
825	for (int i = 0; i<256; ++i) { // IRRELEVANT_LOOP
826	tab[i] = replace_other ? replace_other : i; // replace unused or identity translation
827	used[i] = 0;
828	}
829
830	for (int i = 0; o[i]; ++i) {
831	if (used[o[i]]) return GBS_global_string("character '%c' used twice in argument 1 of translate", o[i]);
832	used[o[i]] = 1;
833	tab[o[i]] = n[i]; // real translation
834	}
835	}
836
837	for (int i=0; i<args->input.size(); i++) {
838	GBS_strstruct *strstruct = GBS_stropen(1000);
839	for (const char p = args->input.get(i); p; p++) {
840	GBS_chrcat(strstruct, tab[(unsigned char)*p]);
841	}
842	PASS_2_OUT(args, GBS_strclose(strstruct));
843	}
844	return NULp;
845	}
846
847
848	static GB_ERROR gbl_echo(GBL_command_arguments *args) {
849	ACCEPT_ANY_PARAMS(args);
850	COMMAND_DROPS_INPUT_STREAMS(args);
851	for (int i=0; i<args->param_count(); i++) PARAM_2_OUT(args, i);
852	return NULp;
853	}
854
855	static GB_ERROR gbl_dd(GBL_command_arguments *args) {
856	EXPECT_NO_PARAM(args);
857	return gbl_mid_streams(args->input, args->output, 0, -1); // copy all streams
858	}
859
860	enum Case { UPPER, LOWER, CAPS };
861
862	static GB_ERROR convert_case(GBL_command_arguments *args, Case convTo) {
863	EXPECT_NO_PARAM(args);
864
865	for (int i=0; i<args->input.size(); i++) {
866	char *p = ARB_strdup(args->input.get(i));
867	bool last_was_alnum = false;
868
869	for (char *pp = p; pp[0]; ++pp) {
870	switch (convTo) {
871	case LOWER: pp[0] = tolower(pp[0]); break;
872	case UPPER: pp[0] = toupper(pp[0]); break;
873	case CAPS: {
874	bool alnum = isalnum(pp[0]);
875	if (alnum) pp[0] = (last_was_alnum ? tolower : toupper)(pp[0]);
876	last_was_alnum = alnum;
877	break;
878	}
879	default: gb_assert(0); break;
880	}
881	}
882
883	PASS_2_OUT(args, p);
884	}
885
886	return NULp;
887	}
888
889	static GB_ERROR gbl_caps (GBL_command_arguments *args) { return convert_case(args, CAPS); }
890	static GB_ERROR gbl_upper(GBL_command_arguments *args) { return convert_case(args, UPPER); }
891	static GB_ERROR gbl_lower(GBL_command_arguments *args) { return convert_case(args, LOWER); }
892
893	static GB_ERROR gbl_head(GBL_command_arguments *args) {
894	EXPECT_PARAMS(args, 1, "length_of_head");
895	int start = atoi(args->get_param(0));
896	if (start <= 0) return gbl_mid_streams(args->input, args->output, 1, 0); // empty all streams
897	return gbl_mid_streams(args->input, args->output, 0, start-1);
898	}
899	static GB_ERROR gbl_tail(GBL_command_arguments *args) {
900	EXPECT_PARAMS(args, 1, "length_of_tail");
901	int end = atoi(args->get_param(0));
902	if (end <= 0) return gbl_mid_streams(args->input, args->output, 1, 0); // empty all streams
903	return gbl_mid_streams(args->input, args->output, -end, -1);
904	}
905
906	inline GB_ERROR mid(GBL_command_arguments *args, int start_index) {
907	EXPECT_PARAMS(args, 2, "start,end");
908	return gbl_mid_streams(args->input, args->output, atoi(args->get_param(0))-start_index, atoi(args->get_param(1))-start_index);
909	}
910	static GB_ERROR gbl_mid0(GBL_command_arguments *args) { return mid(args, 0); }
911	static GB_ERROR gbl_mid (GBL_command_arguments *args) { return mid(args, 1); }
912
913	static GB_ERROR tab(GBL_command_arguments *args, bool pretab) {
914	EXPECT_PARAMS(args, 1, "tabstop");
915
916	int tab = atoi(args->get_param(0));
917	for (int i=0; i<args->input.size(); i++) {
918	int len = strlen(args->input.get(i));
919	if (len >= tab) IN_2_OUT(args, i);
920	else {
921	char *p = ARB_alloc<char>(tab+1);
922	if (pretab) {
923	int spaces = tab-len;
924	for (int j = 0; j<spaces; ++j) p[j] = ' ';
925	strcpy(p+spaces, args->input.get(i));
926	}
927	else {
928	strcpy(p, args->input.get(i));
929	for (int j=len; j<tab; j++) p[j] = ' ';
930	p[tab] = 0;
931	}
932	PASS_2_OUT(args, p);
933	}
934	}
935	return NULp;
936	}
937	static GB_ERROR gbl_tab (GBL_command_arguments *args) { return tab(args, false); }
938	static GB_ERROR gbl_pretab(GBL_command_arguments *args) { return tab(args, true); }
939
940	static GB_ERROR gbl_crop(GBL_command_arguments *args) {
941	EXPECT_PARAMS(args, 1, "\"chars_to_crop\"");
942
943	const char *chars_to_crop = args->get_param(0);
944	for (int i=0; i<args->input.size(); i++) {
945	const char *s = args->input.get(i);
946	while (s[0] && strchr(chars_to_crop, s[0])) s++; // crop at beg of line
947
948	int len = strlen(s);
949	char *p = ARB_alloc<char>(len+1);
950	strcpy(p, s);
951
952	{
953	char *pe = p+len-1;
954
955	while (pe >= p && strchr(chars_to_crop, pe[0])) { // crop at end of line
956	--pe;
957	}
958	gb_assert(pe >= (p-1));
959	pe[1] = 0;
960	}
961	PASS_2_OUT(args, p);
962	}
963	return NULp;
964	}
965
966
967
968	static GB_ERROR gbl_cut(GBL_command_arguments *args) {
969	EXPECT_PARAMS_PASSED(args, "streamnumber[,streamnumber]+");
970
971	for (int i=0; i<args->param_count(); i++) {
972	int stream = atoi(args->get_param(i));
973	EXPECT_LEGAL_STREAM_INDEX(args, stream);
974	IN_2_OUT(args, bio2info(stream));
975	}
976	return NULp;
977	}
978	static GB_ERROR gbl_drop(GBL_command_arguments *args) {
979	EXPECT_PARAMS_PASSED(args, "streamnumber[,streamnumber]+");
980
981	GB_ERROR error = NULp;
982	bool *dropped = ARB_alloc<bool>(args->input.size());
983
984	for (int i=0; i<args->input.size(); ++i) dropped[i] = false;
985
986	for (int i=0; i<args->param_count() && !error; ++i) {
987	int stream = atoi(args->get_param(i));
988	error = check_valid_stream_index(args, stream);
989	if (!error) dropped[bio2info(stream)] = true;
990	}
991
992	if (!error) {
993	for (int i=0; i<args->input.size(); ++i) {
994	if (!dropped[i]) IN_2_OUT(args, i);
995	}
996	}
997	free(dropped);
998
999	return error;
1000	}
1001
1002	static GB_ERROR gbl_dropempty(GBL_command_arguments *args) {
1003	EXPECT_NO_PARAM(args);
1004
1005	for (int i=0; i<args->input.size(); ++i) {
1006	if (args->input.get(i)[0]) { // if non-empty
1007	IN_2_OUT(args, i);
1008	}
1009	}
1010	return NULp;
1011	}
1012
1013	static GB_ERROR gbl_dropzero(GBL_command_arguments *args) {
1014	EXPECT_NO_PARAM(args);
1015
1016	for (int i=0; i<args->input.size(); ++i) {
1017	if (atoi(args->input.get(i))) { // if non-zero
1018	IN_2_OUT(args, i);
1019	}
1020	}
1021	return NULp;
1022	}
1023
1024	static GB_ERROR gbl_swap(GBL_command_arguments *args) {
1025	EXPECT_OPTIONAL_PARAMS(args, 0, NULp, 2, "streamnumber,streamnumber");
1026
1027	if (args->input.size()<2) return "need at least two input streams";
1028
1029	int swap1;
1030	int swap2;
1031	if (args->param_count() == 0) {
1032	swap1 = args->input.size()-1;
1033	swap2 = args->input.size()-2;
1034	}
1035	else {
1036	gb_assert(args->param_count() == 2);
1037
1038	swap1 = atoi(args->get_param(0));
1039	swap2 = atoi(args->get_param(1));
1040
1041	EXPECT_LEGAL_STREAM_INDEX(args, swap1);
1042	EXPECT_LEGAL_STREAM_INDEX(args, swap2);
1043
1044	swap1 = bio2info(swap1);
1045	swap2 = bio2info(swap2);
1046	}
1047
1048	for (int i = 0; i<args->input.size(); ++i) {
1049	int j = i == swap1 ? swap2 : (i == swap2 ? swap1 : i);
1050	IN_2_OUT(args, j);
1051	}
1052
1053	return NULp;
1054	}
1055
1056	static GB_ERROR backfront_stream(GBL_command_arguments *args, int toback) {
1057	EXPECT_PARAMS(args, 1, "streamnumber");
1058	if (args->input.size()<1) return "need at least one input stream";
1059
1060	int stream_to_move = atoi(args->get_param(0));
1061	EXPECT_LEGAL_STREAM_INDEX(args, stream_to_move);
1062	stream_to_move = bio2info(stream_to_move);
1063
1064	if (!toback) IN_2_OUT(args, stream_to_move);
1065	for (int i = 0; i<args->input.size(); ++i) {
1066	if (i != stream_to_move) IN_2_OUT(args, i);
1067	}
1068	if (toback) IN_2_OUT(args, stream_to_move);
1069
1070	return NULp;
1071	}
1072	static GB_ERROR gbl_toback (GBL_command_arguments *args) { return backfront_stream(args, 1); }
1073	static GB_ERROR gbl_tofront(GBL_command_arguments *args) { return backfront_stream(args, 0); }
1074
1075	static GB_ERROR gbl_merge(GBL_command_arguments *args) {
1076	EXPECT_OPTIONAL_PARAMS(args, 0, NULp, 1, "\"separator\"");
1077	const char *separator = args->get_optional_param(0, NULp);
1078
1079	if (args->input.size()) {
1080	GBS_strstruct *str = GBS_stropen(1000);
1081	GBS_strcat(str, args->input.get(0));
1082
1083	for (int i = 1; i<args->input.size(); ++i) {
1084	if (separator) GBS_strcat(str, separator);
1085	GBS_strcat(str, args->input.get(i));
1086	}
1087
1088	PASS_2_OUT(args, GBS_strclose(str));
1089	}
1090	return NULp;
1091	}
1092
1093	static GB_ERROR gbl_split(GBL_command_arguments *args) {
1094	EXPECT_OPTIONAL_PARAMS_CUSTOM(args, 0, NULp, 2, "\"separator\"[,mode]", true, false);
1095
1096	const char *separator = args->get_optional_param(0, "\n");
1097	int split_mode = atoi(args->get_optional_param(1, "0")); // 0: remove separator, 1: split before separator, 2: split behind separator
1098
1099	if (!separator[0]) {
1100	// e.g. happens if trying to specify character ';' or ','
1101	return "Invalid separator (cannot be empty; please try to quote the parameter)";
1102	}
1103
1104	if (split_mode<0 \|\| split_mode>2) return GBS_global_string("Illegal split mode '%i' (valid: 0..2)", split_mode);
1105
1106	{
1107	size_t sepLen = strlen(separator);
1108
1109	for (int i = 0; i<args->input.size(); ++i) {
1110	const char *in = args->input.get(i);
1111	const char *from = in; // search from here
1112
1113	while (in) {
1114	const char *splitAt = strstr(from, separator);
1115	if (splitAt) {
1116	size_t len;
1117	char *copy;
1118
1119	if (split_mode == 2) splitAt += sepLen; // split behind separator
1120
1121	len = splitAt-in;
1122	copy = ARB_strndup(in, len);
1123
1124	PASS_2_OUT(args, copy);
1125
1126	in = splitAt + (split_mode == 0 ? sepLen : 0);
1127	from = in+(split_mode == 1 ? sepLen : 0);
1128	}
1129	else {
1130	COPY_2_OUT(args, in); // last part
1131	in = NULp;
1132	}
1133	}
1134	}
1135	}
1136
1137	return NULp;
1138	}
1139
1140	static GB_ERROR gbl_colsplit(GBL_command_arguments *args) {
1141	EXPECT_OPTIONAL_PARAMS(args, 0, NULp, 1, "width");
1142
1143	int width = atoi(args->get_optional_param(0, "1"));
1144	if (width<1) return "Invalid width";
1145
1146	for (int i = 0; i<args->input.size(); ++i) {
1147	const char *in = args->input.get(i);
1148	int len = strlen(in);
1149
1150	while (len>0) {
1151	char *part = ARB_strpartdup(in, in+width-1);
1152	PASS_2_OUT(args, part);
1153
1154	in += width;
1155	len -= width;
1156	}
1157	}
1158
1159	return NULp;
1160	}
1161	// ----------------------------------
1162	// Extended string functions
1163
1164	static char do_extract_words(const char source, const char *chars, float minlen, bool sort_output) {
1165	/* extract all words in a text that:
1166	* if minlen < 1.0 -> contain more than minlen*len_of_text characters that also exists in chars
1167	* if minlen > 1.0 -> contain more than minlen characters that also exists in chars
1168	*/
1169
1170	char *s = ARB_strdup(source);
1171	char *ps = ARB_calloc<char>((strlen(source)>>1) + 1);
1172	GBS_strstruct *strstruct = GBS_stropen(1000);
1173	char *f = s;
1174	int count = 0;
1175	char *p; // @@@ fix locals
1176	char *h;
1177	int cnt;
1178	int len;
1179	int iminlen = (int) (minlen+.5);
1180
1181	while ((p = strtok(f, " \t,;:\|"))) {
1182	f = NULp;
1183	cnt = 0;
1184	len = strlen(p);
1185	for (h=p; *h; h++) {
1186	if (strchr(chars, *h)) cnt++;
1187	}
1188
1189	if (minlen == 1.0) {
1190	if (cnt != len) continue;
1191	}
1192	else if (minlen > 1.0) {
1193	if (cnt < iminlen) continue;
1194	}
1195	else {
1196	if (len < 3 \|\| cnt < minlen*len) continue;
1197	}
1198	ps[count] = p;
1199	count ++;
1200	}
1201	if (sort_output) {
1202	GB_sort((void **)ps, 0, count, GB_string_comparator, NULp);
1203	}
1204	for (cnt = 0; cnt<count; cnt++) {
1205	if (cnt) {
1206	GBS_chrcat(strstruct, ' ');
1207	}
1208	GBS_strcat(strstruct, ps[cnt]);
1209	}
1210
1211	free(ps);
1212	free(s);
1213	return GBS_strclose(strstruct);
1214	}
1215
1216	static GB_ERROR gbl_extract_words(GBL_command_arguments *args) {
1217	EXPECT_PARAMS(args, 2, "\"chars\", minchars");
1218
1219	float len = atof(args->get_param(1));
1220	for (int i=0; i<args->input.size(); i++) {
1221	char *res = do_extract_words(args->input.get(i), args->get_param(0), len, 1);
1222	gb_assert(res);
1223	PASS_2_OUT(args, res);
1224	}
1225	return NULp;
1226	}
1227
1228	static GB_ERROR gbl_extract_sequence(GBL_command_arguments *args) {
1229	EXPECT_PARAMS(args, 2, "\"chars\",minFrequency");
1230
1231	const char *chars = args->get_param(0);
1232	float minFreq = atof(args->get_param(1));
1233
1234	if (minFreq <0.0 \|\| minFreq > 1.0) return GBS_global_string("Illegal minFrequency=%f (allowed: ]0.0 .. 1.0[)", minFreq);
1235
1236	for (int i=0; i<args->input.size(); i++) {
1237	char *res = do_extract_words(args->input.get(i), chars, minFreq, 0);
1238	gb_assert(res);
1239	PASS_2_OUT(args, res);
1240	}
1241	return NULp;
1242	}
1243
1244	static GB_ERROR gbl_checksum(GBL_command_arguments *args) {
1245	GBL_BEGIN_PARAMS;
1246	GBL_PARAM_STRING(exclude, "exclude=", "", "Remove given characters before calculating");
1247	GBL_PARAM_BIT (upper, "toupper", 0, "Convert all characters to uppercase before calculating");
1248	GBL_TRACE_PARAMS(args);
1249	GBL_END_PARAMS;
1250
1251	for (int i=0; i<args->input.size(); i++) {
1252	long id = GBS_checksum(args->input.get(i), upper, exclude);
1253	FORMAT_2_OUT(args, "%lX", id);
1254	}
1255	return NULp;
1256	}
1257
1258	static GB_ERROR gbl_gcgchecksum(GBL_command_arguments *args) {
1259	EXPECT_NO_PARAM(args);
1260
1261	for (int i=0; i<args->input.size(); i++) {
1262	long id = GBS_gcgchecksum(args->input.get(i));
1263	FORMAT_2_OUT(args, "%li", id);
1264	}
1265	return NULp;
1266	}
1267
1268	// ------------
1269	// SRT
1270
1271	static GB_ERROR gbl_srt(GBL_command_arguments *args) {
1272	EXPECT_PARAMS_PASSED(args, "expr[,expr]+");
1273
1274	GB_ERROR error = NULp;
1275	for (int i=0; i<args->input.size() && !error; i++) {
1276	char *modsource = NULp;
1277
1278	for (int j=0; j<args->param_count() && !error; j++) {
1279	char *hs = GBS_string_eval_in_env(modsource ? modsource : args->input.get(i), args->get_param(j), args->get_callEnv());
1280
1281	if (hs) freeset(modsource, hs);
1282	else {
1283	error = GB_await_error();
1284	free(modsource);
1285	}
1286	}
1287
1288	if (!error) {
1289	if (modsource) PASS_2_OUT(args, modsource);
1290	else IN_2_OUT(args, i);
1291	}
1292	}
1293	return error;
1294	}
1295
1296	// -----------------------------
1297	// Calculator Functions
1298
1299	struct binop_pair {
1300	int (*INT) (int, int);
1301	double (*DOUBLE)(double, double);
1302	binop_pair(int (INT_)(int, int), double (DOUBLE_)(double, double)) : INT(INT_), DOUBLE(DOUBLE_) {}
1303	};
1304
1305	static char apply_numeric_binop(const char arg1, const char arg2, int (num_bin_op)(int,int)) {
1306	int v1 = atoi(arg1);
1307	int v2 = atoi(arg2);
1308	int result = num_bin_op(v1, v2);
1309
1310	return GBS_global_string_copy("%i", result);
1311	}
1312
1313	static char apply_double_binop(const char arg1, const char arg2, double (num_bin_op)(double,double)) {
1314	double v1 = strtod(arg1, NULp);
1315	double v2 = strtod(arg2, NULp);
1316	double result = num_bin_op(v1, v2);
1317
1318	return GBS_global_string_copy("%g", result);
1319	}
1320
1321	static char apply_auto_numeric_binop(const char arg1, const char *arg2, binop_pair multiop) {
1322	// argument type detection (int vs double)
1323	int i1 = atoi(arg1);
1324	int i2 = atoi(arg2);
1325	double d1 = strtod(arg1, NULp);
1326	double d2 = strtod(arg2, NULp);
1327
1328	if (double(i1) == d1 \|\| double(i2) == d2) {
1329	int result = multiop.INT(i1, i2);
1330	return GBS_global_string_copy("%i", result);
1331	}
1332	else {
1333	double result = multiop.DOUBLE(d1, d2);
1334	return GBS_global_string_copy("%g", result);
1335	}
1336	}
1337
1338
1339
1340	template <typename T> static T binop_plus (T v1, T v2) { return v1+v2; }
1341	template <typename T> static T binop_minus (T v1, T v2) { return v1-v2; }
1342	template <typename T> static T binop_mult (T v1, T v2) { return v1*v2; }
1343	template <typename T> static T binop_div (T v1, T v2) { return v2 ? v1/v2 : 0; }
1344	template <typename T> static T binop_per_cent(T v1, T v2) { return v2 ? (v1*100)/v2 : 0; }
1345
1346	static int binop_rest(int i1, int i2) { return i2 ? i1%i2 : 0; }
1347
1348
1349	static GB_ERROR gbl_plus (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_numeric_binop, binop_plus<int>); }
1350	static GB_ERROR gbl_fplus (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_double_binop, binop_plus<double>); }
1351	static GB_ERROR gbl_minus (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_numeric_binop, binop_minus<int>); }
1352	static GB_ERROR gbl_fminus (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_double_binop, binop_minus<double>); }
1353	static GB_ERROR gbl_mult (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_numeric_binop, binop_mult<int>); }
1354	static GB_ERROR gbl_fmult (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_double_binop, binop_mult<double>); }
1355	static GB_ERROR gbl_div (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_numeric_binop, binop_div<int>); }
1356	static GB_ERROR gbl_fdiv (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_double_binop, binop_div<double>); }
1357	static GB_ERROR gbl_rest (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_numeric_binop, binop_rest); }
1358	static GB_ERROR gbl_per_cent (GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_numeric_binop, binop_per_cent<int>); }
1359	static GB_ERROR gbl_fper_cent(GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_double_binop, binop_per_cent<double>); }
1360
1361	template <typename T> static T binop_isAbove(T i1, T i2) { return i1>i2; }
1362	template <typename T> static T binop_isBelow(T i1, T i2) { return i1<i2; }
1363	template <typename T> static T binop_isEqual(T i1, T i2) { return i1 == i2; }
1364
1365	static GB_ERROR gbl_isAbove(GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_auto_numeric_binop, binop_pair(binop_isAbove<int>, binop_isAbove<double>)); }
1366	static GB_ERROR gbl_isBelow(GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_auto_numeric_binop, binop_pair(binop_isBelow<int>, binop_isBelow<double>)); }
1367	static GB_ERROR gbl_isEqual(GBL_command_arguments *args) { return gbl_apply_binary_operator(args, apply_auto_numeric_binop, binop_pair(binop_isEqual<int>, binop_isEqual<double>)); }
1368
1369	inline double float_shift_factor(int digits) {
1370	if (digits<0) {
1371	return 1.0/float_shift_factor(-digits);
1372	}
1373	int factor = 1;
1374	while (digits>0) { // IRRELEVANT_LOOP (gcc 9.x refuses to optimize)
1375	factor *= 10;
1376	--digits;
1377	}
1378	return factor;
1379	}
1380
1381	static GB_ERROR gbl_round(GBL_command_arguments *args) {
1382	EXPECT_PARAMS(args, 1, "digits");
1383	int digits = atoi(args->get_param(0));
1384
1385	double factor = float_shift_factor(digits);
1386	for (int i=0; i<args->input.size(); ++i) {
1387	double val = strtod(args->input.get(i), NULp);
1388	val = round(val*factor)/factor;
1389	FORMAT_2_OUT(args, "%g", val);
1390	}
1391	return NULp;
1392	}
1393
1394
1395
1396	// boolean operators
1397
1398	static GB_ERROR gbl_not(GBL_command_arguments *args) {
1399	EXPECT_NO_PARAM(args);
1400
1401	for (int i=0; i<args->input.size(); ++i) {
1402	const char *s = args->input.get(i);
1403	int val = atoi(s);
1404	FORMAT_2_OUT(args, "%i", !val);
1405	}
1406	return NULp;
1407	}
1408
1409	static GB_ERROR gbl_and(GBL_command_arguments *args) {
1410	EXPECT_NO_PARAM(args);
1411	bool conjunction = true;
1412	for (int i=0; conjunction && i<args->input.size(); ++i) {
1413	const char *s = args->input.get(i);
1414	conjunction = conjunction && atoi(s);
1415	}
1416	FORMAT_2_OUT(args, "%i", conjunction);
1417	return NULp;
1418	}
1419	static GB_ERROR gbl_or(GBL_command_arguments *args) {
1420	EXPECT_NO_PARAM(args);
1421	bool disjunction = false;
1422	for (int i=0; !disjunction && i<args->input.size(); ++i) {
1423	const char *s = args->input.get(i);
1424	disjunction = disjunction \|\| atoi(s);
1425	}
1426	FORMAT_2_OUT(args, "%i", disjunction);
1427	return NULp;
1428	}
1429
1430	static GB_ERROR gbl_select(GBL_command_arguments *args) {
1431	ACCEPT_ANY_PARAMS(args);
1432
1433	GB_ERROR error = NULp;
1434	for (int i=0; i<args->input.size() && !error; i++) {
1435	int paraidx = atoi(args->input.get(i));
1436	error = check_valid_param_index(args, paraidx);
1437	if (!error) {
1438	char *result = args->get_callEnv().interpret_subcommand("", args->get_param(paraidx)); // @@@ EVALUATED_PARAM (#768)
1439	if (!result) error = GB_await_error();
1440	else PASS_2_OUT(args, result);
1441	}
1442	}
1443	return error;
1444	}
1445
1446	static GB_ERROR gbl_readdb(GBL_command_arguments *args) {
1447	EXPECT_PARAMS_PASSED(args, "fieldname[,fieldname]+");
1448	EXPECT_ITEM_REFERENCED(args);
1449	COMMAND_DROPS_INPUT_STREAMS(args);
1450
1451	GBS_strstruct *strstr = GBS_stropen(1024);
1452	for (int i=0; i<args->param_count(); i++) {
1453	char *val = GBT_read_as_string(args->get_item_ref(),
1454	args->track_field_access(args->get_param(i)));
1455	if (val) {
1456	GBS_strcat(strstr, val);
1457	free(val);
1458	}
1459	}
1460	PASS_2_OUT(args, GBS_strclose(strstr));
1461	return NULp;
1462	}
1463
1464
1465	enum GBT_ITEM_TYPE {
1466	GBT_ITEM_UNKNOWN,
1467	GBT_ITEM_SPECIES,
1468	GBT_ITEM_GENE
1469	};
1470
1471	static GBT_ITEM_TYPE identify_gb_item(GBDATA *gb_item) {
1472	/* returns: GBT_ITEM_UNKNOWN -> unknown database_item
1473	* GBT_ITEM_SPECIES -> /species_data/species
1474	* GBT_ITEM_GENE -> /species_data/species/gene_data/gene */
1475
1476	GBT_ITEM_TYPE res = GBT_ITEM_UNKNOWN;
1477	if (gb_item) {
1478	GBDATA *gb_father = GB_get_father(gb_item);
1479	if (gb_father) {
1480	const char *key = GB_KEY(gb_item);
1481
1482	if (strcmp(key, "species") == 0 &&
1483	strcmp(GB_KEY(gb_father), "species_data") == 0) {
1484	res = GBT_ITEM_SPECIES;
1485	}
1486	else if (strcmp(key, "gene") == 0 &&
1487	strcmp(GB_KEY(gb_father), "gene_data") == 0 &&
1488	identify_gb_item(GB_get_father(gb_father)) == GBT_ITEM_SPECIES) {
1489	res = GBT_ITEM_GENE;
1490	}
1491	}
1492	}
1493	return res;
1494	}
1495
1496	// --------------------------------------------------------------------------------
1497	// taxonomy caching
1498
1499	#if defined(DEBUG)
1500	// #define DUMP_TAXONOMY_CACHING
1501	#endif
1502
1503
1504	#define GROUP_COUNT_CHARS 6 // characters in taxonomy-key reserved for group-counter (hex number)
1505	#define BITS_PER_HEXCHAR 4
1506	#define MAX_GROUPS (1 << (GROUP_COUNT_CHARS*BITS_PER_HEXCHAR)) // resulting number of groups
1507
1508	struct cached_taxonomy {
1509	char *tree_name; // tree for which taxonomy is cached here
1510	int groups; // number of named groups in tree (at time of caching)
1511	GB_HASH taxonomy; / keys: "!species", ">XXXXgroup" and "<root>".
1512	* Species and groups contain their first parent (i.e. '>XXXXgroup' or '<root>').
1513	* Species not in hash are not members of tree.
1514	* The 'XXXX' in groupname is simply a counter to avoid multiple groups with same name.
1515	* The group-db-entries are stored in hash as pointers ('>>%p') and
1516	* point to their own group entry ('>XXXXgroup')
1517	*
1518	* Note: the number of 'X's in 'XXXX' above is defined by GROUP_COUNT_CHARS!
1519	*/
1520	};
1521
1522	static void free_cached_taxonomy(cached_taxonomy *ct) {
1523	free(ct->tree_name);
1524	GBS_free_hash(ct->taxonomy);
1525	free(ct);
1526	}
1527
1528	static void build_taxonomy_rek(TreeNode node, GB_HASH tax_hash, const char parent_group, int group_counter) {
1529	if (node->is_leaf()) {
1530	GBDATA *gb_species = node->gb_node;
1531	if (gb_species) { // not zombie
1532	const char *name = GBT_get_name(gb_species);
1533	if (name) GBS_write_hash(tax_hash, GBS_global_string("!%s", name), (long)ARB_strdup(parent_group));
1534	}
1535	}
1536	else {
1537	if (node->has_group_info()) { // node with name
1538	char *hash_entry;
1539	const char *hash_binary_entry;
1540	(*group_counter)++;
1541
1542	gb_assert((*group_counter)<MAX_GROUPS); // overflow - increase GROUP_COUNT_CHARS
1543
1544	TreeNode *keelTarget = node->keelTarget();
1545
1546	hash_entry = GBS_global_string_copy(">%0*x%s%s",
1547	GROUP_COUNT_CHARS, *group_counter,
1548	keelTarget ? "!" : "",
1549	node->name);
1550	GBS_write_hash(tax_hash, hash_entry, (long)ARB_strdup(parent_group));
1551
1552	hash_binary_entry = GBS_global_string(">>%p", node->gb_node);
1553	GBS_write_hash(tax_hash, hash_binary_entry, (long)ARB_strdup(hash_entry));
1554
1555	if (keelTarget) { // keeled group (projected to son)
1556	if (keelTarget->is_leftson()) {
1557	build_taxonomy_rek(node->get_leftson(), tax_hash, hash_entry, group_counter); // pass down hash_entry only to keelTarget
1558	build_taxonomy_rek(node->get_rightson(), tax_hash, parent_group, group_counter);
1559	}
1560	else {
1561	build_taxonomy_rek(node->get_leftson(), tax_hash, parent_group, group_counter);
1562	build_taxonomy_rek(node->get_rightson(), tax_hash, hash_entry, group_counter);
1563	}
1564	}
1565	else { // normal group
1566	build_taxonomy_rek(node->get_leftson(), tax_hash, hash_entry, group_counter); // pass down hash_entry to both sons
1567	build_taxonomy_rek(node->get_rightson(), tax_hash, hash_entry, group_counter);
1568	}
1569
1570	free(hash_entry);
1571	}
1572	else {
1573	build_taxonomy_rek(node->get_leftson(), tax_hash, parent_group, group_counter);
1574	build_taxonomy_rek(node->get_rightson(), tax_hash, parent_group, group_counter);
1575	}
1576	}
1577	}
1578
1579	static GB_HASH *cached_taxonomies = NULp;
1580
1581	static bool is_cached_taxonomy(const char /key/, long val, void cl_ct) {
1582	cached_taxonomy ct1 = (cached_taxonomy )val;
1583	cached_taxonomy ct2 = (cached_taxonomy )cl_ct;
1584
1585	return ct1 == ct2;
1586	}
1587
1588	static const char tree_of_cached_taxonomy(cached_taxonomy ct) {
1589	/* search the hash to find the correct cached taxonomy.
1590	* searching for tree name does not work, because the tree possibly already was deleted
1591	*/
1592	const char *tree = GBS_hash_next_element_that(cached_taxonomies, NULp, is_cached_taxonomy, ct);
1593	#ifdef DUMP_TAXONOMY_CACHING
1594	if (tree) printf("tree_of_cached_taxonomy: tree='%s' ct->tree_name='%s'\n", tree, ct->tree_name);
1595	#endif // DUMP_TAXONOMY_CACHING
1596	return tree;
1597	}
1598
1599	static void flush_taxonomy_cb(GBDATA gbd, cached_taxonomy ct) {
1600	/* this cb is bound all tree db members below "/tree_data/tree_xxx" which
1601	* may have an effect on the displayed taxonomy
1602	* it invalidates cached taxonomies for that tree (when changed or deleted)
1603	*/
1604
1605	GB_ERROR error = NULp;
1606	const char *found = tree_of_cached_taxonomy(ct);
1607
1608	if (found) {
1609	#ifdef DUMP_TAXONOMY_CACHING
1610	fprintf(stderr, "Deleting cached taxonomy ct=%p (tree='%s')\n", ct, found);
1611	#endif // DUMP_TAXONOMY_CACHING
1612	GBS_write_hash(cached_taxonomies, found, 0); // delete cached taxonomy from hash
1613	free_cached_taxonomy(ct);
1614	}
1615	#ifdef DUMP_TAXONOMY_CACHING
1616	else {
1617	fprintf(stderr, "No tree found for cached_taxonomies ct=%p (already deleted?)\n", ct);
1618	}
1619	#endif // DUMP_TAXONOMY_CACHING
1620
1621	if (!GB_inside_callback(gbd, GB_CB_DELETE)) {
1622	GB_remove_all_callbacks_to(gbd, GB_CB_CHANGED_OR_DELETED, CASTSIG(GB_CB, flush_taxonomy_cb));
1623	}
1624
1625	if (found && !error) {
1626	GBDATA *gb_main = GB_get_gb_main_during_cb();
1627	if (gb_main) {
1628	GBDATA *gb_tree_refresh = GB_search(gb_main, AWAR_TREE_REFRESH, GB_INT);
1629	if (!gb_tree_refresh) {
1630	error = GBS_global_string("%s (while trying to force refresh)", GB_await_error());
1631	}
1632	else {
1633	GB_touch(gb_tree_refresh); // Note : force tree update
1634	}
1635	}
1636	}
1637
1638	if (error) {
1639	fprintf(stderr, "Error in flush_taxonomy_cb: %s\n", error);
1640	}
1641	}
1642
1643	static void flush_taxonomy_if_new_group_cb(GBDATA gb_tree, cached_taxonomy ct) {
1644	// detects the creation of new groups and call flush_taxonomy_cb() manually
1645	#ifdef DUMP_TAXONOMY_CACHING
1646	fputs("flush_taxonomy_if_new_group_cb() has been called\n", stderr);
1647	#endif // DUMP_TAXONOMY_CACHING
1648
1649	const char *tree_name = tree_of_cached_taxonomy(ct);
1650	if (tree_name) {
1651	int groups = 0;
1652	GBDATA *gb_group_node;
1653
1654	for (gb_group_node = GB_entry(gb_tree, "node");
1655	gb_group_node;
1656	gb_group_node = GB_nextEntry(gb_group_node))
1657	{
1658	if (GB_entry(gb_group_node, "group_name")) {
1659	groups++; // count named groups only
1660	}
1661	}
1662
1663	#ifdef DUMP_TAXONOMY_CACHING
1664	fprintf(stderr, "cached_groups=%i counted_groups=%i\n", ct->groups, groups);
1665	#endif // DUMP_TAXONOMY_CACHING
1666	if (groups != ct->groups) {
1667	#ifdef DUMP_TAXONOMY_CACHING
1668	fprintf(stderr, "Number of groups changed -> invoking flush_taxonomy_cb() manually\n");
1669	#endif // DUMP_TAXONOMY_CACHING
1670	flush_taxonomy_cb(gb_tree, ct);
1671	}
1672	}
1673	#ifdef DUMP_TAXONOMY_CACHING
1674	else {
1675	fprintf(stderr, "cached taxonomy no longer valid.\n");
1676	}
1677	#endif // DUMP_TAXONOMY_CACHING
1678	}
1679
1680	static cached_taxonomy get_cached_taxonomy(GBDATA gb_main, const char tree_name, GB_ERROR error) {
1681	long cached;
1682	*error = NULp;
1683	if (!cached_taxonomies) {
1684	cached_taxonomies = GBS_create_hash(20, GB_IGNORE_CASE);
1685	}
1686	cached = GBS_read_hash(cached_taxonomies, tree_name);
1687	if (!cached) {
1688	TreeNode *tree = GBT_read_tree(gb_main, tree_name, new SimpleRoot);
1689	if (!tree) *error = GB_await_error();
1690	else *error = GBT_link_tree(tree, gb_main, false, NULp, NULp);
1691
1692	if (!*error) {
1693	GBDATA *gb_tree = GBT_find_tree(gb_main, tree_name);
1694	if (!gb_tree) {
1695	*error = GBS_global_string("Can't find tree '%s'", tree_name);
1696	}
1697	else {
1698	cached_taxonomy *ct = ARB_alloc<cached_taxonomy>(1);
1699	long nodes = GBT_count_leafs(tree);
1700	int group_counter = 0;
1701
1702	ct->tree_name = ARB_strdup(tree_name);
1703	ct->taxonomy = GBS_create_dynaval_hash(int(nodes), GB_IGNORE_CASE, GBS_dynaval_free);
1704	ct->groups = 0; // counted below
1705
1706	build_taxonomy_rek(tree, ct->taxonomy, "<root>", &group_counter);
1707	cached = (long)ct;
1708	GBS_write_hash(cached_taxonomies, tree_name, (long)ct);
1709
1710	GB_remove_all_callbacks_to(gb_tree, GB_CB_SON_CREATED, CASTSIG(GB_CB, flush_taxonomy_if_new_group_cb));
1711	GB_add_callback(gb_tree, GB_CB_SON_CREATED, makeDatabaseCallback(flush_taxonomy_if_new_group_cb, ct));
1712
1713	{
1714	GBDATA *gb_tree_entry = GB_entry(gb_tree, "tree");
1715	GBDATA *gb_group_node;
1716
1717	if (gb_tree_entry) {
1718	GB_remove_all_callbacks_to(gb_tree_entry, GB_CB_CHANGED_OR_DELETED, CASTSIG(GB_CB, flush_taxonomy_cb));
1719	GB_add_callback(gb_tree_entry, GB_CB_CHANGED_OR_DELETED, makeDatabaseCallback(flush_taxonomy_cb, ct));
1720	}
1721
1722	// add callbacks for all node/group_name subentries
1723	for (gb_group_node = GB_entry(gb_tree, "node");
1724	gb_group_node;
1725	gb_group_node = GB_nextEntry(gb_group_node))
1726	{
1727	GBDATA *gb_group_name = GB_entry(gb_group_node, "group_name");
1728	if (gb_group_name) { // group with id = 0 has no name
1729	GB_remove_all_callbacks_to(gb_group_name, GB_CB_CHANGED_OR_DELETED, CASTSIG(GB_CB, flush_taxonomy_cb));
1730	GB_add_callback(gb_group_name, GB_CB_CHANGED_OR_DELETED, makeDatabaseCallback(flush_taxonomy_cb, ct));
1731	ct->groups++;
1732	}
1733	}
1734	}
1735	#ifdef DUMP_TAXONOMY_CACHING
1736	fprintf(stderr, "Created taxonomy hash for '%s' (ct=%p)\n", tree_name, ct);
1737	#endif // DUMP_TAXONOMY_CACHING
1738	}
1739	}
1740
1741	destroy(tree);
1742	}
1743
1744	if (!*error) {
1745	cached_taxonomy ct = (cached_taxonomy)cached;
1746	gb_assert(ct);
1747	return ct;
1748	}
1749
1750	return NULp;
1751	}
1752
1753	static char get_taxonomy_string(GB_HASH tax_hash, const char group_key, int depth, GB_ERROR error) {
1754	long found;
1755	char *result = NULp;
1756
1757	gb_assert(depth>0);
1758	gb_assert(!(group_key[0] == '>' && group_key[1] == '>')); // internal group-pointers not allowed here!
1759
1760	found = GBS_read_hash(tax_hash, group_key);
1761	if (found) {
1762	const char parent_group_key = (const char )found;
1763	if (strcmp(parent_group_key, "<root>") == 0) { // root reached
1764	result = ARB_strdup(group_key+(GROUP_COUNT_CHARS+1)); // return own group name
1765	}
1766	else {
1767	if (depth>1) {
1768	char *parent_name = get_taxonomy_string(tax_hash, parent_group_key, depth-1, error);
1769	if (parent_name) {
1770	result = GBS_global_string_copy("%s/%s", parent_name, group_key+(GROUP_COUNT_CHARS+1));
1771	free(parent_name);
1772	}
1773	else {
1774	error = GBS_global_string("In get_taxonomy_string(%s): %s", group_key, error);
1775	result = NULp;
1776	}
1777	}
1778	else {
1779	result = ARB_strdup(group_key+(GROUP_COUNT_CHARS+1)); // return own group name
1780	}
1781	}
1782	}
1783	else {
1784	*error = GBS_global_string("Not in tax_hash: '%s'", group_key);
1785	}
1786	return result;
1787	}
1788
1789	static const char get_taxonomy(GBDATA gb_species_or_group, const char tree_name, bool is_current_tree, int depth, GB_ERROR error) {
1790	GBDATA *gb_main = GB_get_root(gb_species_or_group);
1791	cached_taxonomy *tax = get_cached_taxonomy(gb_main, tree_name, error);
1792	const char *result = NULp;
1793
1794	if (tax) {
1795	GBDATA *gb_name = GB_entry(gb_species_or_group, "name");
1796	GBDATA *gb_group_name = GB_entry(gb_species_or_group, "group_name");
1797
1798	if (gb_name && !gb_group_name) { // it's a species
1799	char *name = GB_read_string(gb_name);
1800	if (name) {
1801	GB_HASH *tax_hash = tax->taxonomy;
1802	long found = GBS_read_hash(tax_hash, GBS_global_string("!%s", name));
1803
1804	if (found) {
1805	const char parent_group = (const char )found;
1806
1807	if (strcmp(parent_group, "<root>") == 0) {
1808	result = ""; // not member of any group
1809	}
1810	else {
1811	static char *parent = NULp;
1812
1813	freeset(parent, get_taxonomy_string(tax_hash, parent_group, depth, error));
1814	result = parent;
1815	}
1816	}
1817	else {
1818	result = GBS_global_string("Species '%s' not in '%s'", name, tree_name);
1819	}
1820	free(name);
1821	}
1822	else {
1823	*error = GBS_global_string("Species without 'name' entry!");
1824	}
1825	}
1826	else if (gb_group_name && !gb_name) { // it's a group
1827	char *group_name = GB_read_string(gb_group_name);
1828	if (group_name) {
1829	if (is_current_tree) {
1830	GB_HASH *tax_hash = tax->taxonomy;
1831	long found = GBS_read_hash(tax_hash, GBS_global_string(">>%p", gb_species_or_group));
1832
1833	if (found) {
1834	static char *full_group = NULp;
1835	const char group_id = (const char )found;
1836
1837	freeset(full_group, get_taxonomy_string(tax_hash, group_id, depth, error));
1838	result = full_group;
1839	}
1840	else {
1841	result = GBS_global_string("Group '%s' not in '%s'", group_name, tree_name);
1842	}
1843	}
1844	else {
1845	*error = "It's not possible to specify the tree name in taxonomy() for groups";
1846	}
1847	free(group_name);
1848	}
1849	else {
1850	*error = "Group without 'group_name' entry";
1851	}
1852	}
1853	else if (gb_group_name) {
1854	*error = "Container has 'name' and 'group_name' entry - can't detect container type";
1855	}
1856	else {
1857	*error = "Container has neither 'name' nor 'group_name' entry - can't detect container type";
1858	}
1859	}
1860
1861	return result;
1862	}
1863
1864	static GB_ERROR gbl_taxonomy(GBL_command_arguments *args) {
1865	GB_ERROR error = check_optional_parameters(args, 1, "count", 1, "tree_name", false, true);
1866	if (!error) {
1867	EXPECT_ITEM_REFERENCED(args);
1868	COMMAND_DROPS_INPUT_STREAMS(args);
1869
1870	char *tree_name = NULp;
1871	bool is_current_tree = false;
1872	int depth = -1;
1873	char *result = NULp;
1874
1875	if (args->param_count() == 1) { // only 'depth'
1876	if (!args->get_treename()) {
1877	result = ARB_strdup("No default tree");
1878	}
1879	else {
1880	tree_name = ARB_strdup(args->get_treename());
1881	depth = atoi(args->get_param(0));
1882	is_current_tree = true;
1883	}
1884	}
1885	else { // 'tree_name', 'depth'
1886	tree_name = ARB_strdup(args->get_param(0));
1887	depth = atoi(args->get_param(1));
1888	}
1889
1890	if (!result) {
1891	if (depth<1) {
1892	error = GBS_global_string("Illegal depth '%i' (allowed 1..n)", depth);
1893	}
1894	if (!error) {
1895	const char *taxonomy_string = get_taxonomy(args->get_item_ref(), tree_name, is_current_tree, depth, &error);
1896	if (taxonomy_string) result = ARB_strdup(taxonomy_string);
1897	}
1898	}
1899
1900	gb_assert(contradicted(result, error));
1901	if (result) PASS_2_OUT(args, result);
1902	free(tree_name);
1903	}
1904	return error;
1905	}
1906
1907	static GB_ERROR gbl_sequence(GBL_command_arguments *args) {
1908	EXPECT_ITEM_REFERENCED(args);
1909	COMMAND_DROPS_INPUT_STREAMS(args);
1910
1911	GB_ERROR error = check_no_parameter(args);
1912	if (!error) {
1913	switch (identify_gb_item(args->get_item_ref())) {
1914	case GBT_ITEM_UNKNOWN: {
1915	error = "'sequence' used for unknown item";
1916	break;
1917	}
1918	case GBT_ITEM_SPECIES: {
1919	char *use = GBT_get_default_alignment(args->get_gb_main());
1920
1921	if (!use) {
1922	error = GB_have_error() ? GB_await_error() : "no default alignment defined";
1923	}
1924	else {
1925	GBDATA *gb_seq = GBT_find_sequence(args->get_item_ref(), use);
1926
1927	if (gb_seq) PASS_2_OUT(args, GB_read_string(gb_seq));
1928	else COPY_2_OUT(args, ""); // if current alignment does not exist -> return empty string
1929
1930	free(use);
1931	}
1932	break;
1933	}
1934	case GBT_ITEM_GENE: {
1935	char *seq = GBT_read_gene_sequence(args->get_item_ref(), true, 0);
1936
1937	if (!seq) error = GB_await_error();
1938	else PASS_2_OUT(args, seq);
1939
1940	break;
1941	}
1942	}
1943	}
1944	return error;
1945	}
1946
1947	static GB_ERROR gbl_export_sequence(GBL_command_arguments *args) {
1948	EXPECT_ITEM_REFERENCED(args);
1949	COMMAND_DROPS_INPUT_STREAMS(args);
1950
1951	GB_ERROR error = check_no_parameter(args);
1952	if (!error) {
1953	switch (identify_gb_item(args->get_item_ref())) {
1954	case GBT_ITEM_UNKNOWN: {
1955	error = "'export_sequence' used for unknown item";
1956	break;
1957	}
1958	case GBT_ITEM_SPECIES: {
1959	if (!get_export_sequence) {
1960	error = "No export-sequence-hook defined (can't use 'export_sequence' here)";
1961	}
1962	else {
1963	size_t len;
1964	const char *seq = get_export_sequence(args->get_item_ref(), &len, &error);
1965
1966	gb_assert(error \|\| seq);
1967
1968	if (seq) PASS_2_OUT(args, ARB_strduplen(seq, len));
1969	}
1970	break;
1971	}
1972	case GBT_ITEM_GENE: {
1973	error = "'export_sequence' cannot be used for gene";
1974	break;
1975	}
1976	}
1977	}
1978	return error;
1979	}
1980
1981	static GB_ERROR gbl_ali_name(GBL_command_arguments *args) {
1982	COMMAND_DROPS_INPUT_STREAMS(args);
1983
1984	GB_ERROR error = check_no_parameter(args);
1985	if (!error) {
1986	GBDATA *gb_main = args->get_gb_main();
1987	char *use = GBT_get_default_alignment(gb_main);
1988	PASS_2_OUT(args, use);
1989	}
1990	return error;
1991	}
1992
1993	static GB_ERROR gbl_sequence_type(GBL_command_arguments *args) {
1994	COMMAND_DROPS_INPUT_STREAMS(args);
1995
1996	GB_ERROR error = check_no_parameter(args);
1997	if (!error) {
1998	GBDATA *gb_main = args->get_gb_main();
1999	char *use = GBT_get_default_alignment(gb_main);
2000	PASS_2_OUT(args, GBT_get_alignment_type_string(gb_main, use));
2001	free(use);
2002	}
2003
2004	return error;
2005	}
2006
2007	static GB_ERROR format(GBL_command_arguments *args, bool simple_format) {
2008	// simple_format: true = "format", false="format_sequence"
2009
2010	GB_ERROR error = NULp;
2011	int ic;
2012
2013	GBL_BEGIN_PARAMS;
2014	GBL_PARAM_INT(firsttab, "firsttab=", 10, "Indent first line");
2015	GBL_PARAM_INT(tab, "tab=", 10, "Indent not first line");
2016	GBL_PARAM_INT(width, "width=", 50, "Sequence width (bases only)");
2017
2018	// "format_sequence"-only
2019	GBL_PARAM_BIT (numleft, PARAM_IF(!simple_format, "numleft"), 0, "Numbers left of sequence");
2020	GBL_PARAM_INT (numright, PARAM_IF(!simple_format, "numright="), 0, "Numbers right of sequence (specifies width; -1 -> auto-width)");
2021	GBL_PARAM_UINT(gap, PARAM_IF(!simple_format, "gap="), 10, "Insert ' ' every n sequence characters");
2022
2023	// "format"-only
2024	GBL_PARAM_STRING(nl, PARAM_IF(simple_format, "nl="), " ", "Break line at characters 'str' if wrapping needed");
2025	GBL_PARAM_STRING(forcenl, PARAM_IF(simple_format, "forcenl="), "\n", "Always break line at characters 'str'");
2026
2027	GBL_TRACE_PARAMS(args);
2028	GBL_END_PARAMS;
2029
2030	if (tab < 0) tab = 0;
2031	if (firsttab < 0) firsttab = 0;
2032
2033	if (width == 0) return "Illegal zero width";
2034	if (numleft && numright != 0) return "You may only specify 'numleft' OR 'numright', not both.";
2035
2036	if (gap<1) gap = UINT_MAX;
2037
2038	for (ic = 0; ic<args->input.size(); ++ic) {
2039	const char *src = args->input.get(ic);
2040	size_t data_size = strlen(src);
2041	size_t needed_size;
2042	size_t line_size;
2043	int numright_used = numright;
2044
2045	if (numright_used<0) {
2046	numright_used = calc_digits(data_size);
2047	}
2048
2049	{
2050	size_t lines;
2051
2052	if (simple_format) {
2053	lines = data_size/2 + 1; // worst case
2054	line_size = tab + (width>0 ? width : data_size) + 1;
2055	}
2056	else {
2057	size_t gapsPerLine = (width-1)/gap;
2058	lines = data_size/width+1;
2059	line_size = tab + width + gapsPerLine + 1;
2060
2061	if (numright_used) {
2062	// add space for numright
2063	line_size += numright_used+1; // plus space
2064	}
2065	}
2066
2067	needed_size = lines*line_size + firsttab + 1 + 10;
2068	}
2069
2070	char *result = ARB_alloc<char>(needed_size);
2071	if (!result) {
2072	error = GBS_global_string("Out of memory (tried to alloc %zu bytes)", needed_size);
2073	}
2074	else {
2075	char *dst = result;
2076	size_t rest_data = data_size;
2077
2078	if (simple_format) {
2079	/* format string w/o gaps or numleft
2080	* does word-wrapping at chars in nl
2081	*/
2082
2083	// build wrap table
2084	unsigned char isWrapChar[256];
2085	memset(isWrapChar, 0, sizeof(isWrapChar));
2086	for (int i = 0; nl[i]; ++i) isWrapChar[(unsigned char)nl[i]] = 1;
2087	for (int i = 0; forcenl[i]; ++i) isWrapChar[(unsigned char)forcenl[i]] = 2;
2088
2089	if (firsttab>0) {
2090	memset(dst, ' ', firsttab);
2091	dst += firsttab;
2092	}
2093
2094	while (width>0 && rest_data>unsigned(width)) {
2095	int take;
2096	int move;
2097	int took;
2098
2099	for (take = width; take > 0; --take) {
2100	if (isWrapChar[(unsigned char)src[take]]) break;
2101	}
2102	if (take <= 0) { // no wrap character found -> hard wrap at width
2103	take = move = width;
2104	}
2105	else { // soft wrap at last found wrap character
2106	move = take+1;
2107	}
2108
2109	for (took = 0; took<take; took++) {
2110	char c = src[took];
2111	if (isWrapChar[(unsigned char)c] == 2) { // forced newline
2112	take = took;
2113	move = take+1;
2114	break;
2115	}
2116	dst[took] = c;
2117	}
2118
2119	dst += take;
2120	src += move;
2121	rest_data -= move;
2122
2123	if (rest_data>0) {
2124	*dst++ = '\n';
2125	if (tab>0) {
2126	memset(dst, ' ', tab);
2127	dst += tab;
2128	}
2129	}
2130	}
2131
2132	if (rest_data>0) {
2133	size_t j, k;
2134	for (j = 0, k = 0; j<rest_data; ++j) {
2135	char c = src[j];
2136
2137	if (isWrapChar[(unsigned char)c] == 2) {
2138	dst[k++] = '\n';
2139	if (tab>0) {
2140	memset(dst+k, ' ', tab);
2141	k += tab;
2142	}
2143	}
2144	else {
2145	dst[k++] = c;
2146	}
2147	}
2148	src += j;
2149	dst += k;
2150	rest_data = 0;
2151	}
2152	}
2153	else {
2154	// "format_sequence" with gaps and numleft
2155	char *format = NULp;
2156	const char *src_start = src;
2157	const char *dst_linestart = dst;
2158
2159	if (numleft) {
2160	/* Warning: Be very careful, when you change format strings here!
2161	* currently all format strings result in '%u' or '%-##u' (where # are digits)
2162	*/
2163	if (firsttab>0) {
2164	char *firstFormat = GBS_global_string_copy("%%-%iu ", firsttab-1);
2165	dst += sprintf(dst, firstFormat, (unsigned)1);
2166	free(firstFormat);
2167	}
2168	else {
2169	dst += sprintf(dst, "%u ", (unsigned)1);
2170	}
2171	format = tab>0 ? GBS_global_string_copy("%%-%iu ", tab-1) : ARB_strdup("%u ");
2172	}
2173	else if (firsttab>0) {
2174	memset(dst, ' ', firsttab);
2175	dst += firsttab;
2176	}
2177
2178	while (rest_data>0) {
2179	size_t take = (width>0 && rest_data>unsigned(width)) ? width : rest_data;
2180
2181	rest_data -= take;
2182
2183	while (take>gap) {
2184	memcpy(dst, src, gap);
2185	dst += gap;
2186	src += gap;
2187	*dst++ = ' ';
2188	take -= gap;
2189	}
2190
2191	memcpy(dst, src, take);
2192	dst += take;
2193	src += take;
2194
2195	if (numright_used) {
2196	if (rest_data) *dst++ = ' ';
2197	else {
2198	// fill in missing spaces for proper alignment of numright
2199	size_t currSize = dst-dst_linestart;
2200	size_t wantSize = line_size-numright_used-1;
2201	if (currSize<wantSize) {
2202	size_t spaces = wantSize-currSize;
2203	memset(dst, ' ', spaces);
2204	dst += spaces;
2205	}
2206	}
2207	unsigned int num = (src-src_start);
2208	dst += sprintf(dst, "%*u", numright_used, num);
2209	}
2210
2211	if (rest_data>0) {
2212	*dst++ = '\n';
2213	dst_linestart = dst;
2214	if (numleft) {
2215	unsigned int num = (src-src_start)+1; // this goes to the '%u' (see comment above)
2216	dst += sprintf(dst, format, num);
2217	}
2218	else if (tab>0) {
2219	memset(dst, ' ', tab);
2220	dst += tab;
2221	}
2222	}
2223	}
2224
2225	free(format);
2226	}
2227
2228	*dst++ = 0; // close str
2229
2230	#if defined(DEBUG)
2231	{ // check for array overflow
2232	size_t used_size = dst-result;
2233	gb_assert(used_size <= needed_size);
2234	ARB_realloc(result, used_size);
2235	}
2236	#endif // DEBUG
2237	}
2238
2239	if (!error) PASS_2_OUT(args, result);
2240	else free(result);
2241	}
2242	return error;
2243	}
2244
2245	static GB_ERROR gbl_format (GBL_command_arguments *args) { return format(args, true); }
2246	static GB_ERROR gbl_format_sequence(GBL_command_arguments *args) { return format(args, false); }
2247
2248
2249	static char gbl_read_seq_sai_or_species(GBDATA gb_main, const char species, const char sai, const char ali, size_t seqLen) {
2250	/* Reads the alignment 'ali' of 'species' or 'sai'.
2251	* If 'ali' is NULp, use default alignment.
2252	* Returns NULp in case of error (which is exported then)
2253	*/
2254
2255	char *seq = NULp;
2256	GB_ERROR error = NULp;
2257
2258	int sources = !!species + !!sai;
2259	if (sources != 1) {
2260	error = "Either parameters 'species' or 'SAI' must be specified";
2261	}
2262	else {
2263	GBDATA *gb_item = NULp;
2264	const char *what = NULp;
2265	const char *name = NULp;
2266
2267	if (species) {
2268	gb_item = GBT_find_species(gb_main, species);
2269	what = "species";
2270	name = species;
2271	}
2272	else {
2273	gb_item = GBT_find_SAI(gb_main, sai);
2274	what = "SAI";
2275	name = sai;
2276	}
2277
2278	if (!gb_item) error = GBS_global_string("Can't find %s '%s'", what, name);
2279	else {
2280	char *freeMe = NULp;
2281
2282	if (!ali) {
2283	ali = freeMe = GBT_get_default_alignment(gb_main);
2284	if (!ali) error = "can't detect default alignment";
2285	}
2286
2287	if (ali) {
2288	GBDATA *gb_ali = GB_entry(gb_item, ali);
2289
2290	if (gb_ali) {
2291	GBDATA *gb_seq;
2292
2293	for (gb_seq = GB_child(gb_ali); gb_seq; gb_seq = GB_nextChild(gb_seq)) {
2294	long type = GB_read_type(gb_seq);
2295	if (type == GB_BITS) {
2296	seq = GB_read_bits(gb_seq, '-', '+');
2297	if (seqLen) *seqLen = GB_read_bits_count(gb_seq);
2298	break;
2299	}
2300	if (type == GB_STRING) {
2301	seq = GB_read_string(gb_seq);
2302	if (seqLen) *seqLen = GB_read_string_count(gb_seq);
2303	break;
2304	}
2305	}
2306	}
2307
2308	if (!seq) error = GBS_global_string("%s '%s' has no (usable) data in alignment '%s'", what, name, ali);
2309	}
2310	free(freeMe);
2311	}
2312	}
2313
2314	if (error) {
2315	gb_assert(!seq);
2316	GB_export_error(error);
2317	}
2318
2319	return seq;
2320	}
2321
2322	struct common_filter_params {
2323	const char *align;
2324	const char *sai;
2325	const char *species;
2326	int first;
2327	int pairwise;
2328	};
2329
2330	#define GBL_COMMON_FILTER_PARAMS \
2331	common_filter_params common_param; \
2332	GBL_STRUCT_PARAM_STRING(common_param, align, "align=", NULp, "alignment to use (defaults to default alignment)"); \
2333	GBL_STRUCT_PARAM_STRING(common_param, sai, "SAI=", NULp, "Use default sequence of given SAI as a filter"); \
2334	GBL_STRUCT_PARAM_STRING(common_param, species, "species=", NULp, "Use default sequence of given species as a filter"); \
2335	GBL_STRUCT_PARAM_BIT (common_param, first, "first=", 0, "Use 1st stream as filter for other streams"); \
2336	GBL_STRUCT_PARAM_BIT (common_param, pairwise, "pairwise=", 0, "Use 1st stream as filter for 2nd, 3rd for 4th, ...")
2337
2338	typedef char* (filter_fun)(const char seq, const char filter, size_t flen, void param);
2339	/* Note:
2340	* filter_fun has to return a heap copy of the filter-result.
2341	* if 'flen' != 0, it contains the length of 'filter'
2342	* 'param' may be any client data
2343	*/
2344
2345	static GB_ERROR apply_filters(GBL_command_arguments args, common_filter_params common, filter_fun filter_one, void *param) {
2346	GB_ERROR error = NULp;
2347
2348	if (args->input.size()==0) error = "No input stream";
2349	else {
2350	int methodCount = !!common->sai + !!common->species + !!common->pairwise + !!common->first;
2351
2352	if (methodCount != 1) error = "Need exactly one of the parameters 'SAI', 'species', 'pairwise' or 'first'";
2353	else {
2354	if (common->pairwise) {
2355	if (args->input.size() % 2) error = "Using 'pairwise' requires an even number of input streams";
2356	else {
2357	int i;
2358	for (i = 1; i<args->input.size(); i += 2) {
2359	PASS_2_OUT(args, filter_one(args->input.get(i), args->input.get(i-1), 0, param));
2360	}
2361	}
2362	}
2363	else {
2364	int i = 0;
2365	char *filter = NULp;
2366	size_t flen = 0;
2367
2368	if (common->first) {
2369	if (args->input.size()<2) error = "Using 'first' needs at least 2 input streams";
2370	else {
2371	const char *in = args->input.get(i++);
2372	gb_assert(in);
2373
2374	flen = strlen(in);
2375	filter = ARB_strduplen(in, flen);
2376	}
2377	}
2378	else {
2379	filter = gbl_read_seq_sai_or_species(args->get_gb_main(), common->species, common->sai, common->align, &flen);
2380	if (!filter) error = GB_await_error();
2381	}
2382
2383	gb_assert(filter \|\| error);
2384	if (filter) {
2385	for (; i<args->input.size(); ++i) {
2386	PASS_2_OUT(args, filter_one(args->input.get(i), filter, flen, param));
2387	}
2388	}
2389	free(filter);
2390	}
2391	}
2392	}
2393	return error;
2394	}
2395
2396	// -------------------------
2397	// calculate diff
2398
2399	struct diff_params {
2400	char equalC;
2401	char diffC;
2402	};
2403	static char calc_diff(const char seq, const char filter, size_t /flen/, void paramP) {
2404	// filters 'seq' through 'filter'
2405	// - replace all equal positions by 'equal_char' (if != 0)
2406	// - replace all differing positions by 'diff_char' (if != 0)
2407
2408	diff_params param = (diff_params)paramP;
2409	char equal_char = param->equalC;
2410	char diff_char = param->diffC;
2411
2412	char *result = ARB_strdup(seq);
2413	int p;
2414
2415	for (p = 0; result[p] && filter[p]; ++p) {
2416	if (result[p] == filter[p]) {
2417	if (equal_char) result[p] = equal_char;
2418	}
2419	else {
2420	if (diff_char) result[p] = diff_char;
2421	}
2422	}
2423
2424	// if 'seq' is longer than 'filter' and diff_char is given
2425	// -> fill rest of 'result' with 'diff_char'
2426	if (diff_char) {
2427	for (; result[p]; ++p) {
2428	result[p] = diff_char;
2429	}
2430	}
2431
2432	return result;
2433	}
2434	static GB_ERROR gbl_diff(GBL_command_arguments *args) {
2435	GBL_BEGIN_PARAMS;
2436	GBL_COMMON_FILTER_PARAMS;
2437
2438	diff_params param;
2439	GBL_STRUCT_PARAM_CHAR(param, equalC, "equal=", '.', "symbol for equal characters");
2440	GBL_STRUCT_PARAM_CHAR(param, diffC, "differ=", 0, "symbol for diff characters (default: use char from input stream)");
2441
2442	GBL_TRACE_PARAMS(args);
2443	GBL_END_PARAMS;
2444
2445	return apply_filters(args, &common_param, calc_diff, &param);
2446	}
2447
2448	// -------------------------
2449	// standard filter
2450
2451	enum filter_function { FP_FILTER, FP_MODIFY };
2452
2453	struct filter_params { // used by gbl_filter and gbl_change_gc
2454	filter_function function;
2455
2456	const char *include;
2457	const char *exclude;
2458
2459	// FP_MODIFY only:
2460	int change_pc;
2461	const char *change_to;
2462	};
2463
2464	static char filter_seq(const char seq, const char filter, size_t flen, void paramP) {
2465	filter_params param = (filter_params)paramP;
2466
2467	size_t slen = strlen(seq);
2468	if (!flen) flen = strlen(filter);
2469	size_t mlen = slen<flen ? slen : flen;
2470
2471	GBS_strstruct *out = GBS_stropen(mlen+1); // +1 to avoid invalid, zero-length buffer
2472
2473	const char *charset;
2474	int include;
2475
2476	if (param->include) {
2477	charset = param->include;
2478	include = 1;
2479	}
2480	else {
2481	gb_assert(param->exclude);
2482	charset = param->exclude;
2483	include = 0;
2484	}
2485
2486	size_t pos = 0;
2487	size_t rest = slen;
2488	size_t ctl = 0;
2489	if (param->function == FP_MODIFY) ctl = strlen(param->change_to);
2490
2491	int inset = 1; // 1 -> check chars in charset, 0 -> check chars NOT in charset
2492	while (rest) {
2493	size_t count;
2494	if (pos >= flen) { // behind filter
2495	// trigger last loop
2496	count = rest;
2497	inset = 0; // if 'include' -> 'applies' will get false, otherwise true
2498	// (meaning is: behind filter nothing can match 'include' or 'exclude')
2499	}
2500	else {
2501	count = (inset ? strspn : strcspn)(filter+pos, charset); // count how many chars are 'inset'
2502	}
2503	if (count) {
2504	int applies = !!include == !!inset; // true -> 'filter' matches 'include' or doesn't match 'exclude'
2505	if (count>rest) count = rest;
2506
2507	switch (param->function) {
2508	case FP_FILTER:
2509	if (applies) GBS_strncat(out, seq+pos, count);
2510	break;
2511
2512	case FP_MODIFY:
2513	if (applies) { // then modify
2514	size_t i;
2515	for (i = 0; i<count; i++) {
2516	char c = seq[pos+i];
2517	if (isalpha(c) && GB_random(100)<param->change_pc) c = param->change_to[GB_random(ctl)];
2518	GBS_chrcat(out, c);
2519	}
2520	}
2521	else { // otherwise simply copy
2522	GBS_strncat(out, seq+pos, count);
2523	}
2524	break;
2525	}
2526
2527	pos += count;
2528	rest -= count;
2529	}
2530	inset = 1-inset; // toggle
2531	}
2532	return GBS_strclose(out);
2533	}
2534
2535	static GB_ERROR gbl_filter(GBL_command_arguments *args) {
2536	GBL_BEGIN_PARAMS;
2537	GBL_COMMON_FILTER_PARAMS;
2538
2539	filter_params param;
2540	GBL_STRUCT_PARAM_STRING(param, exclude, "exclude=", NULp, "Exclude colums");
2541	GBL_STRUCT_PARAM_STRING(param, include, "include=", NULp, "Include colums");
2542	param.function = FP_FILTER;
2543
2544	GBL_TRACE_PARAMS(args);
2545	GBL_END_PARAMS;
2546
2547	GB_ERROR error = NULp;
2548	int inOrEx = !!param.include + !!param.exclude;
2549
2550	if (inOrEx != 1) error = "Need exactly one parameter of: 'include', 'exclude'";
2551	else error = apply_filters(args, &common_param, filter_seq, &param);
2552
2553	return error;
2554	}
2555
2556	static GB_ERROR gbl_change_gc(GBL_command_arguments *args) {
2557	GBL_BEGIN_PARAMS;
2558	GBL_COMMON_FILTER_PARAMS;
2559
2560	filter_params param;
2561	GBL_STRUCT_PARAM_STRING(param, exclude, "exclude=", NULp, "Exclude colums");
2562	GBL_STRUCT_PARAM_STRING(param, include, "include=", NULp, "Include colums");
2563	GBL_STRUCT_PARAM_INT (param, change_pc, "change=", 0, "percentage of changed columns (default: silently change nothing)");
2564	GBL_STRUCT_PARAM_STRING(param, change_to, "to=", "GC", "change to one of this");
2565	param.function = FP_MODIFY;
2566
2567	GBL_TRACE_PARAMS(args);
2568	GBL_END_PARAMS;
2569
2570	GB_ERROR error = NULp;
2571	int inOrEx = !!param.include + !!param.exclude;
2572
2573	if (inOrEx != 1) error = "Need exactly one parameter of: 'include', 'exclude'";
2574	else {
2575	error = apply_filters(args, &common_param, filter_seq, &param);
2576	}
2577
2578	return error;
2579	}
2580
2581	static GB_ERROR gbl_exec(GBL_command_arguments *args) {
2582	EXPECT_PARAMS_PASSED(args, "command[,arguments]+");
2583
2584	// write inputstreams to temp file:
2585	GB_ERROR error = NULp;
2586	char *inputname;
2587	int i;
2588	{
2589	char *filename = GB_unique_filename("arb_exec_input", "tmp");
2590	FILE *out = GB_fopen_tempfile(filename, "wt", &inputname);
2591
2592	if (!out) error = GB_await_error();
2593	else {
2594	for (i=0; i<args->input.size(); i++) {
2595	fprintf(out, "%s\n", args->input.get(i));
2596	}
2597	fclose(out);
2598	}
2599	free(filename);
2600	}
2601
2602	if (!error) {
2603	// build shell command to execute
2604	char *sys;
2605	{
2606	GBS_strstruct *str = GBS_stropen(1000);
2607
2608	GBS_strcat(str, args->get_param(0));
2609	for (i=1; i<args->param_count(); i++) {
2610	GBS_strcat(str, " \'");
2611	GBS_strcat(str, args->get_param(i)); // @@@ use GBK_singlequote here?
2612	GBS_chrcat(str, '\'');
2613	}
2614	GBS_strcat(str, " <");
2615	GBS_strcat(str, inputname);
2616
2617	sys = GBS_strclose(str);
2618	}
2619
2620	char *result = NULp;
2621	{
2622	FILE *in = popen(sys, "r");
2623	if (in) {
2624	GBS_strstruct *str = GBS_stropen(4096);
2625
2626	while ((i=getc(in)) != EOF) { GBS_chrcat(str, i); }
2627	result = GBS_strclose(str);
2628	pclose(in);
2629	}
2630	else {
2631	error = GBS_global_string("Cannot execute shell command '%s'", sys);
2632	}
2633	}
2634
2635	if (!error) {
2636	gb_assert(result);
2637	PASS_2_OUT(args, result);
2638	}
2639
2640	free(sys);
2641	}
2642
2643	gb_assert(GB_is_privatefile(inputname, false));
2644	GB_unlink_or_warn(inputname, &error);
2645	free(inputname);
2646
2647	return error;
2648	}
2649
2650
2651	static GBL_command_definition gbl_command_table[] = {
2652	{ "ali_name", gbl_ali_name },
2653	{ "caps", gbl_caps },
2654	{ "change", gbl_change_gc },
2655	{ "checksum", gbl_checksum },
2656	{ "command", gbl_command },
2657	{ "compare", gbl_compare },
2658	{ "colsplit", gbl_colsplit },
2659	{ "icompare", gbl_icompare },
2660	{ "contains", gbl_contains },
2661	{ "icontains", gbl_icontains },
2662	{ "count", gbl_count },
2663	{ "crop", gbl_crop },
2664	{ "cut", gbl_cut },
2665	{ "dd", gbl_dd },
2666	{ "define", gbl_define },
2667	{ "diff", gbl_diff },
2668	{ "div", gbl_div },
2669	{ "fdiv", gbl_fdiv },
2670	{ "do", gbl_do },
2671	{ "drop", gbl_drop },
2672	{ "dropempty", gbl_dropempty },
2673	{ "dropzero", gbl_dropzero },
2674	{ "echo", gbl_echo },
2675	{ "equals", gbl_equals },
2676	{ "iequals", gbl_iequals },
2677	{ "escape", gbl_escape },
2678	{ "unescape", gbl_unescape },
2679	{ "eval", gbl_eval },
2680	{ "exec", gbl_exec },
2681	{ "export_sequence", gbl_export_sequence },
2682	{ "extract_sequence", gbl_extract_sequence },
2683	{ "extract_words", gbl_extract_words },
2684	{ "filter", gbl_filter },
2685	{ "findspec", gbl_findspec },
2686	{ "findacc", gbl_findacc },
2687	{ "findgene", gbl_findgene },
2688	{ "format", gbl_format },
2689	{ "format_sequence", gbl_format_sequence },
2690	{ "gcgchecksum", gbl_gcgchecksum },
2691	{ "head", gbl_head },
2692	{ "inRange", gbl_inRange },
2693	{ "isAbove", gbl_isAbove },
2694	{ "isBelow", gbl_isBelow },
2695	{ "isEqual", gbl_isEqual },
2696	{ "isEmpty", gbl_isEmpty },
2697	{ "keep", gbl_keep },
2698	{ "left", gbl_head },
2699	{ "len", gbl_len },
2700	{ "lower", gbl_lower },
2701	{ "merge", gbl_merge },
2702	{ "mid", gbl_mid },
2703	{ "mid0", gbl_mid0 },
2704	{ "minus", gbl_minus },
2705	{ "fminus", gbl_fminus },
2706	{ "mult", gbl_mult },
2707	{ "fmult", gbl_fmult },
2708	{ "and", gbl_and },
2709	{ "or", gbl_or },
2710	{ "not", gbl_not },
2711	{ "origin_gene", gbl_origin_gene },
2712	{ "origin_organism", gbl_origin_organism },
2713	{ "partof", gbl_partof },
2714	{ "ipartof", gbl_ipartof },
2715	{ "per_cent", gbl_per_cent },
2716	{ "fper_cent", gbl_fper_cent },
2717	{ "plus", gbl_plus },
2718	{ "fplus", gbl_fplus },
2719	{ "pretab", gbl_pretab },
2720	{ "quote", gbl_quote },
2721	{ "unquote", gbl_unquote },
2722	{ "readdb", gbl_readdb },
2723	{ "remove", gbl_remove },
2724	{ "rest", gbl_rest },
2725	{ "right", gbl_tail },
2726	{ "round", gbl_round },
2727	{ "select", gbl_select },
2728	{ "sequence", gbl_sequence },
2729	{ "sequence_type", gbl_sequence_type },
2730	{ "split", gbl_split },
2731	{ "srt", gbl_srt },
2732	{ "streams", gbl_streams },
2733	{ "swap", gbl_swap },
2734	{ "tab", gbl_tab },
2735	{ "tail", gbl_tail },
2736	{ "taxonomy", gbl_taxonomy },
2737	{ "toback", gbl_toback },
2738	{ "tofront", gbl_tofront },
2739	{ "trace", gbl_trace },
2740	{ "translate", gbl_translate },
2741	{ "upper", gbl_upper },
2742
2743	{ NULp, NULp }
2744	};
2745
2746	const GBL_command_lookup_table& ACI_get_standard_commands() {
2747	static GBL_command_lookup_table clt(gbl_command_table, ARRAY_ELEMS(gbl_command_table)-1);
2748	return clt;
2749	}
2750

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

Download in other formats: