Context Navigation

← Previous Revision
Next Revision →
Blame
Revision Log

gb_aci.cxx

Visit:

Last change on this file was 19346, checked in by westram, 3 years ago
reintegrates 'refactor' into 'trunk' polish `GBS_strstruct` adds: log:branches/refactor@19339:19345
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 64.4 KB

Line
1	// ============================================================ //
2	// //
3	// File : gb_aci.cxx //
4	// Purpose : ARB command interpreter (ACI) //
5	// //
6	// http://www.arb-home.de/ //
7	// //
8	// ============================================================ //
9
10	#include "gb_aci.h"
11	#include "gb_aci_impl.h"
12
13	#include <arb_strbuf.h>
14	#include <arb_match.h>
15
16	using namespace GBL_IMPL;
17
18	GBL_command_lookup_table::GBL_command_lookup_table(const GBL_command_definition *table, unsigned size) {
19	/*! create table to lookup ACI commands
20	* @param table command-table (has to exist as long as GBL_command_lookup_table exists; needs sentinel)
21	* @param size number of commands in 'table' ( = size of 'table' - 1)
22	*/
23
24	for (unsigned i = 0; i<size; ++i) {
25	const GBL_command_definition& cmd = table[i];
26	gb_assert(cmd.is_defined());
27
28	defined[cmd.identifier] = cmd.function;
29	}
30	gb_assert(table[size].is_sentinel());
31	}
32
33	namespace GBL_IMPL {
34	static const char search_matching_dquote(const char str) {
35	int c;
36	for (; (c=*str); str++) {
37	if (c=='\\') { // escaped characters
38	str++;
39	if (!(c=*str)) return NULp;
40	continue;
41	}
42	if (c=='"') return (char *)str;
43	}
44	return NULp;
45	}
46	inline char search_matching_dquote(char str) {
47	return const_cast<char>(search_matching_dquote(const_cast<const char>(str)));
48	}
49	const char search_matching_parenthesis(const char source) {
50	int c;
51	int deep = 0;
52	if (*source != '(') deep --; // first bracket
53	for (; (c=*source); source++) {
54	if (c=='\\') { // escaped characters
55	source++;
56	if (!*source) break;
57	continue;
58	}
59	if (c=='(') deep--;
60	else if (c==')') deep++;
61	if (!deep) return (char *)source;
62	if (c=='"') { // search the second "
63	source = search_matching_dquote(source);
64	if (!source) return NULp;
65	}
66	}
67	if (!c) return NULp;
68	return source;
69	}
70	static const char search_next_separator(const char source, const char *seps) {
71	// search the next separator
72	static char tab[256];
73	static bool init = false;
74
75	if (!init) {
76	memset(tab, 0, 256);
77	init = true;
78	}
79
80	for (const char p = seps; p; ++p) tab[safeCharIndex(*p)] = 1;
81
82	tab['('] = 2; // exclude () pairs
83	tab['"'] = 2; // exclude " pairs
84	tab['\\'] = 2; // exclude \-escaped chars
85
86	for (; *source; ++source) {
87	const char chType = tab[safeCharIndex(*source)];
88	if (chType == 0) continue; // accept char
89	if (chType == 1) break; // found separator
90
91	if (*source == '\\') {
92	++source; // -> skip over next char
93	if (!source[0]) break; // abort if end of string seen
94	}
95	else if (*source == '(') {
96	source = search_matching_parenthesis(source);
97	if (!source) break;
98	}
99	else if (*source == '\"') {
100	source = search_matching_dquote(source+1);
101	if (!source) break;
102	}
103	}
104	for (const char p = seps; p; ++p) tab[safeCharIndex(*p)] = 0; // clear tab
105	return source && source[0] ? source : NULp;
106	}
107	inline char search_next_separator(char source, const char *seps) {
108	return const_cast<char>(search_next_separator(const_cast<const char>(source), seps));
109	}
110	};
111
112	static void dumpStreams(const char *name, const GBL_streams& streams) {
113	if (streams.empty()) {
114	print_trace(GBS_global_string("%s [none]\n", name));
115	}
116	else {
117	int count = streams.size();
118	char *header = GBS_global_string_copy("%s", name);
119
120	print_trace(GBS_global_string("%s [0]='%s'", header, streams.get(0)));
121	if (count>1) {
122	LocallyModify<int> inc(traceIndent, traceIndent+strlen(header)+1);
123	for (int c = 1; c<count; c++) {
124	if (c == 10 \|\| c == 100 \|\| c == 1000) --traceIndent; // dec indentation
125	print_trace(GBS_global_string("[%i]='%s'\n", c, streams.get(c)));
126	}
127	}
128	free(header);
129	}
130	}
131
132	static const char shortenLongString(const char str, size_t wanted_len) {
133	// shortens the string 'str' to 'wanted_len' (appends '[..]' if string was shortened)
134
135	const char *result;
136	size_t len = strlen(str);
137
138	gb_assert(wanted_len>4);
139
140	if (len>wanted_len) {
141	static char *shortened_str;
142	static size_t short_len = 0;
143
144	if (short_len >= wanted_len) {
145	memcpy(shortened_str, str, wanted_len-4);
146	}
147	else {
148	freeset(shortened_str, ARB_strpartdup(str, str+wanted_len));
149	short_len = wanted_len;
150	}
151	strcpy(shortened_str+wanted_len-4, "[..]");
152	result = shortened_str;
153	}
154	else {
155	result = str;
156	}
157	return result;
158	}
159
160	static char apply_ACI(const char str, const char *commands, const GBL_call_env& callEnv) {
161	char *buffer = ARB_strdup(commands);
162
163	// ******************** remove all spaces and tabs *****************
164	{
165	const char *s1;
166	char *s2;
167	s1 = commands;
168	s2 = buffer;
169	{
170	int c;
171	for (; (c = *s1); s1++) {
172	if (c=='\\') {
173	*(s2++) = c;
174	if (!(c=*++s1)) { break; }
175	*(s2++) = c;
176	continue;
177	}
178
179	if (c=='"') { // search the second "
180	const char *hp = search_matching_dquote(s1+1);
181	if (!hp) {
182	GB_export_errorf("unbalanced '\"' in '%s'", commands);
183	free(buffer);
184	return NULp;
185	}
186	while (s1 <= hp) (s2++) = (s1++); // LOOP_VECTORIZED
187	s1--;
188	continue;
189	}
190	if (c!=' ' && c!='\t') *(s2++) = c;
191	}
192	}
193	*s2 = 0;
194	}
195
196	GBL_streams orig;
197
198	orig.insert(ARB_strdup(str));
199
200	GB_ERROR error = NULp;
201	GBL_streams out;
202	{
203	char s1, s2;
204	s1 = buffer;
205	if (*s1 == '\|') s1++;
206
207	// loop over all commands
208	for (; s1; s1 = s2) {
209	int separator;
210	GBL_COMMAND command;
211	s2 = search_next_separator(s1, "\|;,");
212	if (s2) {
213	separator = *(s2);
214	*(s2++) = 0;
215	}
216	else {
217	separator = 0;
218	}
219	// collect the parameters
220	GBL_streams argStreams;
221	if (*s1 == '"') { // copy "text" to out
222	char *end = search_matching_dquote(s1+1);
223	if (!end) {
224	UNCOVERED(); // seems unreachable (balancing is already ensured by search_next_separator)
225	error = "Missing second '\"'";
226	break;
227	}
228	*end = 0;
229
230	TRACE_ACI(GBS_global_string("copy '%s'\n", s1+1));
231	out.insert(ARB_strdup(s1+1));
232	}
233	else {
234	char *bracket = strchr(s1, '(');
235	if (bracket) { // I got the parameter list
236	*(bracket++) = 0;
237	int slen = strlen(bracket);
238	if (slen<1 \|\| bracket[slen-1] != ')') {
239	error = "Missing ')'";
240	}
241	else if (slen == 1) {
242	error = "Invalid empty parameter list '()'. To pass an empty argument use '(\"\")'";
243	}
244	else {
245	// go through the parameters
246	char p1, p2;
247	bracket[slen-1] = 0;
248	for (p1 = bracket; p1; p1 = p2) {
249	p2 = search_next_separator(p1, ";,");
250	if (p2) {
251	*(p2++) = 0;
252	}
253	if (p1[0] == '"') { // remove "" pairs
254	int len2;
255	p1++;
256	len2 = strlen(p1)-1;
257
258	if (p1[len2] != '\"') {
259	error = GBS_global_string("Invalid parameter syntax for '%s' (needs '\"' at begin AND end of parameter)", p1-1);
260	}
261	else {
262	p1[len2] = 0;
263	}
264	}
265	argStreams.insert(ARB_strdup(p1));
266	}
267	}
268	}
269	if (!error && (bracket \|\| *s1)) {
270	command = callEnv.get_env().lookup_command(s1);
271	if (!command) {
272	error = GBS_global_string("Unknown command '%s'", s1);
273	}
274	else {
275	GBL_command_arguments args(callEnv, s1, orig, argStreams, out);
276
277	TRACE_ACI(GBS_global_string("execute '%s':\n", args.get_cmdName()));
278	{
279	LocallyModify<int> inc(traceIndent, traceIndent+1);
280	if (traceACI) {
281	dumpStreams("ArgStreams", args.get_param_streams());
282	dumpStreams("InpStreams", args.input);
283	}
284
285	error = command(&args); // execute the command
286
287	if (!error && traceACI) dumpStreams("OutStreams", args.output);
288	}
289
290	if (error) {
291	char *dup_error = ARB_strdup(error);
292
293	#define MAX_PRINT_LEN 200
294
295	char *paramlist = NULp;
296	for (int j = 0; j<args.param_count(); ++j) {
297	const char *param = args.get_param(j);
298	const char *param_short = shortenLongString(param, MAX_PRINT_LEN);
299
300	if (!paramlist) paramlist = ARB_strdup(param_short);
301	else freeset(paramlist, GBS_global_string_copy("%s,%s", paramlist, param_short));
302	}
303	char *inputstreams = NULp;
304	for (int j = 0; j<args.input.size(); ++j) {
305	const char *input = args.input.get(j);
306	const char *input_short = shortenLongString(input, MAX_PRINT_LEN);
307
308	if (!inputstreams) inputstreams = ARB_strdup(input_short);
309	else freeset(inputstreams, GBS_global_string_copy("%s;%s", inputstreams, input_short));
310	}
311	#undef MAX_PRINT_LEN
312	if (paramlist) {
313	error = GBS_global_string("while applying '%s(%s)'\nto '%s':\n%s", s1, paramlist, inputstreams, dup_error);
314	}
315	else {
316	error = GBS_global_string("while applying '%s'\nto '%s':\n%s", s1, inputstreams, dup_error);
317	}
318
319	free(inputstreams);
320	free(paramlist);
321	free(dup_error);
322	}
323	}
324	}
325	}
326
327	if (error) break;
328
329	if (separator == '\|') { // out -> in pipe; clear in
330	out.swap(orig);
331	out.erase();
332	}
333	}
334	}
335
336	{
337	char *s1 = out.concatenated();
338	free(buffer);
339	if (!error) return s1;
340	free(s1);
341	}
342
343	GB_export_errorf("Command '%s' failed:\nReason: %s", commands, error);
344	return NULp;
345	}
346	// --------------------------------------------------------------------------------
347
348	char *GBL_streams::concatenated() const {
349	int count = size();
350	if (!count) return ARB_strdup("");
351	if (count == 1) return ARB_strdup(get(0));
352
353	GBS_strstruct buf(1000);
354	for (int i=0; i<count; i++) {
355	const char *s = get(i);
356	if (s) buf.cat(s);
357	}
358	return buf.release_memfriendly();
359	}
360
361	NOT4PERL char GB_command_interpreter_in_env(const char str, const char *commands, const GBL_call_env& callEnv) {
362	/* simple command interpreter returns NULp on error (which should be exported in that case)
363	* if first character is == ':' run string parser
364	* if first character is == '/' run regexpr
365	* else run ACI
366	*/
367
368	// @@@ most code here and code in apply_ACI could be moved into GBL_call_env::interpret_subcommand
369
370	LocallyModify<int> localT(traceACI); // localize effect of command 'trace'
371	SmartMallocPtr(char) heapstr;
372
373	if (!str) {
374	if (!callEnv.get_item_ref()) {
375	GB_export_error("ACI: no input streams found");
376	return NULp;
377	}
378
379	if (GB_read_type(callEnv.get_item_ref()) == GB_STRING) {
380	str = GB_read_char_pntr(callEnv.get_item_ref());
381	}
382	else {
383	char *asstr = GB_read_as_string(callEnv.get_item_ref());
384	if (!asstr) {
385	GB_export_error("Can't read this DB entry as string");
386	return NULp;
387	}
388
389	heapstr = asstr;
390	str = &*heapstr;
391	}
392	}
393
394	if (traceACI) {
395	print_trace(GBS_global_string("CI: command '%s' apply to '%s'\n", commands, str));
396	}
397	modify_trace_indent(+1);
398
399	char *result = NULp;
400
401	if (!commands \|\| !commands[0]) { // empty command -> do not modify string
402	result = ARB_strdup(str);
403	}
404	else if (commands[0] == ':') { // ':' -> string parser
405	result = GBS_string_eval_in_env(str, commands+1, callEnv);
406	}
407	else if (commands[0] == '/') { // regular expression
408	GB_ERROR err = NULp;
409	result = GBS_regreplace(str, commands, &err);
410
411	if (!result) {
412	if (strcmp(err, "Missing '/' between search and replace string") == 0) {
413	// if GBS_regreplace didn't find a third '/' -> silently use GBS_regmatch:
414	size_t matchlen;
415	err = NULp;
416
417	const char *matched = GBS_regmatch(str, commands, &matchlen, &err);
418
419	if (matched) result = ARB_strndup(matched, matchlen);
420	else if (!err) result = ARB_strdup("");
421	}
422
423	if (!result && err) GB_export_error(err);
424	}
425	}
426	else {
427	result = apply_ACI(str, commands, callEnv);
428	}
429
430	modify_trace_indent(-1);
431	if (traceACI) {
432	GBS_strstruct final_msg(1000);
433	if (result) {
434	final_msg.cat("CI: result ='");
435	final_msg.cat(result);
436	}
437	else {
438	final_msg.cat("CI: no result. error ='");
439	final_msg.cat(GB_get_error());
440	}
441	final_msg.put('\'');
442	final_msg.put('\n');
443	final_msg.nput('-', final_msg.get_position()-1);
444	final_msg.put('\n');
445
446	print_trace(final_msg.get_data());
447	}
448
449	gb_assert(contradicted(result, GB_have_error()));
450	return result;
451	}
452
453	char GB_command_interpreter(const char str, const char commands, GBDATA gb_main) {
454	//! @see GB_command_interpreter_in_env - this flavor runs in dummy environment
455	GBL_env env(gb_main, NULp);
456	GBL_call_env callEnv(NULp, env);
457
458	return GB_command_interpreter_in_env(str, commands, callEnv);
459	}
460
461	void GBL_custom_command_lookup_table::warn_about_overwritten_commands(const GBL_command_definition *custom_table, unsigned custom_size) const {
462	int errcount = 0;
463	for (unsigned i = 0; i<custom_size; ++i) {
464	const GBL_command_definition& cdef = custom_table[i];
465	gb_assert(cdef.is_defined());
466
467	const char *cmd = cdef.identifier;
468	if (base_table.lookup(cmd)) {
469	fprintf(stderr, "Warning: ACI-command '%s' is substituted w/o permission\n", cmd);
470	++errcount;
471	}
472	}
473	if (errcount>0) {
474	fprintf(stderr, "Warning: Detected probably unwanted substitution of %i ACI-commands\n", errcount);
475	gb_assert(0); // either use PERMIT_SUBSTITUTION or fix command names
476	}
477	}
478
479	// --------------------------------------------------------------------------------
480
481	#ifdef UNIT_TESTS
482	#include <test_unit.h>
483
484	#include <arb_defs.h>
485
486	#define TEST_CI__INTERNAL(input,cmd,expected,got,TEST_RESULT,callEnv) do { \
487	char *result; \
488	TEST_EXPECT_RESULT__NOERROREXPORTED(result = GB_command_interpreter_in_env(input, cmd, callEnv)); \
489	TEST_RESULT(result,expected,got); \
490	free(result); \
491	} while(0)
492
493	#define TEST_CI(input,cmd,expected) TEST_CI__INTERNAL(input, cmd, expected, narg, TEST_EXPECT_EQUAL__IGNARG, callEnv)
494	#define TEST_CI_WITH_ENV(input,env,cmd,expected) TEST_CI__INTERNAL(input, cmd, expected, narg, TEST_EXPECT_EQUAL__IGNARG, env)
495	#define TEST_CI__BROKEN(input,cmd,expected,regr) TEST_CI__INTERNAL(input, cmd, expected, regr, TEST_EXPECT_EQUAL__BROKEN, callEnv)
496	#define TEST_CI_NOOP(inandout,cmd) TEST_CI__INTERNAL(inandout, cmd, inandout, narg, TEST_EXPECT_EQUAL__IGNARG, callEnv)
497	#define TEST_CI_NOOP__BROKEN(inandout,regr,cmd) TEST_CI__INTERNAL(inandout, cmd, inandout, regr, TEST_EXPECT_EQUAL__BROKEN, callEnv)
498
499	#define TEST_CI_INVERSE(in,cmd,inv_cmd,out) do { \
500	TEST_CI(in, cmd, out); \
501	TEST_CI(out, inv_cmd, in); \
502	} while(0)
503
504	// @@@ rename errorpart_expected -> expected_errorpart
505
506	#define TEST_CI_ERROR_CONTAINS(input,cmd,errorpart_expected) \
507	TEST_EXPECT_NORESULT__ERROREXPORTED_CONTAINS(GB_command_interpreter_in_env(input, cmd, callEnv), errorpart_expected)
508
509	#define TEST_CI_ERROR_CONTAINS__BROKEN(input,cmd,errorpart_expected,unexpected_result) do{ \
510	char *result; \
511	TEST_EXPECT_NORESULT__ERROREXPORTED_CONTAINS__BROKEN(result = GB_command_interpreter_in_env(input, cmd, callEnv), errorpart_expected); \
512	TEST_EXPECT_EQUAL(result, unexpected_result); \
513	free(result); \
514	}while(0)
515
516	static GBDATA *RCI_gb_main = NULp;
517	static const char *RCI_input = NULp;
518	static const char *RCI_cmd = NULp;
519	static GBDATA *RCI_gbd = NULp;
520
521	inline void run_ci() {
522	GBL_env env(RCI_gb_main, NULp);
523	GBL_call_env callEnv(RCI_gbd, env);
524	GB_command_interpreter_in_env(RCI_input, RCI_cmd, callEnv);
525	}
526
527	#define TEST_CI_SEGFAULTS(input,cmd) do{ \
528	RCI_gb_main = gb_main; \
529	RCI_input = input; \
530	RCI_cmd = cmd; \
531	RCI_gbd = gb_data; \
532	TEST_EXPECT_SEGFAULT(run_ci); \
533	}while(0)
534
535	#define TEST_CI_SEGFAULTS__UNWANTED(input,cmd) do{ \
536	RCI_gb_main = gb_main; \
537	RCI_input = input; \
538	RCI_cmd = cmd; \
539	RCI_gbd = gb_data; \
540	TEST_EXPECT_SEGFAULT__UNWANTED(run_ci); \
541	}while(0)
542
543	#define ACI_SPLIT "\|split(\",\",0)"
544	#define ACI_MERGE "\|merge(\",\")"
545	#define WITH_SPLITTED(aci) ACI_SPLIT aci ACI_MERGE
546
547	static GB_ERROR gbx_custom(GBL_command_arguments *args) {
548	EXPECT_NO_PARAM(args);
549	for (int i=0; i<args->input.size(); ++i) {
550	args->output.insert(strdup("4711"));
551	}
552	return NULp;
553	}
554
555	class ACI_test_env : virtual Noncopyable {
556	GB_shell shell;
557	GBDATA *gb_main;
558	LocallyModify<int> traceMode;
559	GB_transaction ta;
560	GBDATA *gb_species;
561	public:
562	ACI_test_env() :
563	gb_main(GB_open("TEST_aci.arb", "rw")),
564	traceMode(traceACI, 0), // set to 1 to trace all ACI tests
565	ta(gb_main)
566	{
567	gb_assert(gb_main);
568	gb_species = GBT_find_species(gb_main, "LcbReu40"); // ../UNIT_TESTER/run/TEST_aci.arb@LcbReu40
569	}
570	~ACI_test_env() {
571	TEST_EXPECT_NO_ERROR(ta.close(NULp));
572	GB_close(gb_main);
573	}
574
575	GBDATA *gbmain() const { return gb_main; }
576	GBDATA *gbspecies() const { return gb_species; }
577	};
578
579	__ATTR__REDUCED_OPTIMIZE__NO_GCSE void TEST_GB_command_interpreter_1a() {
580	ACI_test_env E;
581	GBL_env base_env(E.gbmain(), NULp);
582
583	// execute ACI on species container (=GB_DB) in this section ------------------------------
584	GBDATA * const gb_data = E.gbspecies();
585	GBL_call_env callEnv(gb_data, base_env);
586
587	TEST_CI_NOOP("bla", "");
588
589	TEST_CI("bla", ":a=u", "blu"); // simple SRT
590
591	// GBS_REGREPLACE_TESTS:
592	TEST_CI("bla", "/a/u/", "blu"); // simple regExp replace
593
594	TEST_CI("test", "/_[0-9]+//", "test"); // simple regExp replace (failing, ie. no match -> no replace)
595
596	TEST_CI("blabla", "/l.*b/", "lab"); // simple regExp match
597	TEST_CI("blabla", "/b.b/", ""); // simple regExp match (failing)
598
599	TEST_CI("tx_01_2", "/_[0-9]+//", "tx"); // simple regExp replace (replace all occurrences)
600	TEST_CI("tx_01_2", "/_[0-9]+$//", "tx_01"); // simple regExp replace (replace one occurrence)
601
602	TEST_CI_ERROR_CONTAINS("xx_____", "/_//", "regular expression '_' matched an empty string"); // caused a deadlock until [16326]
603	TEST_CI("xx_____", "/_//", "xx"); // working removal of multiple '_'
604	TEST_CI("xx_____yy", "/(_+)([^_]\|$)/-=-\\2/", "xx-=-yy"); // replace multiple consecutive '_'
605	TEST_CI("xx_____", "/(_+)([^_]\|$)/-=-\\2/", "xx-=-"); // replace multiple consecutive '_'
606
607	TEST_CI("xx_____", "/_*$//", "xx"); // removal of multiple '_' from end of sequence
608	TEST_CI("xx", "/_*$//", "xx"); // removal of no '_' from end of sequence
609	TEST_CI("_____yy", "/^_*//", "yy"); // removal of multiple '_' from start of sequence
610	TEST_CI("yy", "/^_*//", "yy"); // removal of no '_' from start of sequence
611
612	TEST_CI("", "/^$/ABC/", "ABC"); // replacing a complete empty string should be possible
613	TEST_CI("XXX", "/^XXX$//", ""); // erase whole known text
614
615	TEST_CI("xx/yy/zz", "/\\//-/", "xx-yy-zz"); // search expression with an escaped slash
616	TEST_CI("xx-yy-zz", "/-/\\//", "xx/yy/zz"); // reverse (escaped slash in replace expression)
617
618	TEST_CI_ERROR_CONTAINS("xx", "\\///", "Unknown command");
619	TEST_CI ("x/x", "/\\//", "/");
620	TEST_CI_ERROR_CONTAINS("xx", "//\\/", "railing backslash"); // [Tt]railing (lib dependent?)
621
622	// escape / quote
623	TEST_CI_INVERSE("ac", "\|quote", "\|unquote", "\"ac\"");
624	TEST_CI_INVERSE("ac", "\|escape", "\|unescape", "ac");
625	TEST_CI_INVERSE("ac", "\|escape\|quote", "\|unquote\|unescape", "\"ac\"");
626	TEST_CI_INVERSE("ac", "\|quote\|escape", "\|unescape\|unquote", "\\\"ac\\\"");
627
628	TEST_CI_INVERSE("a\"b\\c", "\|quote", "\|unquote", "\"a\"b\\c\"");
629	TEST_CI_INVERSE("a\"b\\c", "\|escape", "\|unescape", "a\\\"b\\\\c");
630	TEST_CI_INVERSE("a\"b\\c", "\|escape\|quote", "\|unquote\|unescape", "\"a\\\"b\\\\c\"");
631	TEST_CI_INVERSE("a\"b\\c", "\|quote\|escape", "\|unescape\|unquote", "\\\"a\\\"b\\\\c\\\"");
632
633	TEST_CI_NOOP("ac", "\|unquote");
634	TEST_CI_NOOP("\"ac", "\|unquote");
635	TEST_CI_NOOP("ac\"", "\|unquote");
636
637	TEST_CI ("blabla", "\|coUNT(ab)", "4"); // simple ACI
638	TEST_CI ("l", "\|\"b\";dd;\"a\"\|dd", "bla"); // ACI with muliple streams
639	TEST_CI_ERROR_CONTAINS("bla", "\|count()", "Invalid empty parameter list"); // no longer interpret '()' as "1 empty arg" (force use of explicit form; see next line)
640	TEST_CI ("bla", "\|count(\"\")", "0"); // explicitly empty parameter (still strange, counts 0-byte before string-terminator; always 0)
641	TEST_CI ("b a", "\|count(\" \")", "1"); // space in quotes
642	TEST_CI ("b\\a", "\|count(\\a)", "2"); // count '\\' and 'a' (ok)
643	TEST_CI__BROKEN ("b\\a", "\|count(\"\\a\")", "1", "2"); // should only count 'a' (which is escaped in param)
644	TEST_CI ("b\\a", "\|count(\"\a\")", "0"); // does not contain '\a'
645	TEST_CI ("b\a", "\|count(\"\a\")", "1"); // counts '\a'
646
647	// escaping (behavior is unexpected or weird and not documented very well)
648	TEST_CI("b\\a\a", "\|count(\\a)", "2"); // counts '\\' and 'a' (but not '\a')
649	TEST_CI("b\\a", "\|contains(\"\\\\\")", "0"); // searches for 2 backslashes (not in input)
650	TEST_CI("b\\a", "\|contains(\"\")", "0"); // search for empty string never succeeds
651	TEST_CI("b\\a", "\|contains(\\)", "2"); // finds backslash at position 1
652	TEST_CI("b\\\\a", "\|contains(\"\\\\\")", "2"); // finds two backslashes at position 2
653
654	TEST_CI_ERROR_CONTAINS("b\\a", "\|contains(\"\\\")", "ARB ERROR: unbalanced '\"' in '\|contains(\"\\\")'"); // FIX: raises error (should search for 1 backslash)
655
656	// test binary ops
657	{
658	// LocallyModify<int> traceHere(traceACI, 1);
659	TEST_CI("", "\"5\";\"7\"\|minus", "-2");
660	TEST_CI("", "\"5\"\|minus(\"7\")", "-2");
661	// TEST_CI("", "minus(5,7)", "-2"); // @@@ this fails (stating command '5' fails). fix how?
662	TEST_CI("", "minus(\"\"5\"\",\"\"7\"\")", "-2"); // @@@ syntax needed here 'minus(""5"", ""7"")' should be documented more in-depth
663	TEST_CI("", "minus(\"\"5\";\"2\"\|mult\",\"\"7\"\")", "3"); // (5*2)-7
664	TEST_CI("", "minus(\"\"5\";\"2\\,\"\|mult\",\"\"7\"\")", "3"); // comma has to be escaped
665
666	TEST_CI_ERROR_CONTAINS("", "minus(\"\"5\";\"2,\"\|mult\",\"\"7\"\")", "Invalid parameter syntax for '\"\"5\";\"2'");
667	}
668
669	TEST_CI_NOOP("ab,bcb,abac", WITH_SPLITTED(""));
670	TEST_CI ("ab,bcb,abac", WITH_SPLITTED("\|len"), "2,3,4");
671	TEST_CI ("ab,bcb,abac", WITH_SPLITTED("\|count(a)"), "1,0,2");
672	TEST_CI ("ab,bcb,abac", WITH_SPLITTED("\|minus(len,count(a))"), "1,3,2");
673	TEST_CI ("ab,bcb,abac", WITH_SPLITTED("\|minus(\"\"5\"\",count(a))"), "4,5,3");
674
675	// test other recursive uses of GB_command_interpreter
676	TEST_CI("one", "\|dd;\"two\";dd\|command(\"dd;\"_\";dd;\"-\"\")", "one_one-two_two-one_one-");
677	TEST_CI("", "\|sequence\|command(\"/^([\\\\.-])[A-Z]./\\\\1/\")\|len", "9"); // count gaps at start of sequence
678	TEST_CI("one", "\|dd;dd\|eval(\"\"up\";\"per\"\|merge\")", "ONEONE");
679	TEST_CI("1,2,3", WITH_SPLITTED("\|select(\"\", \"\"one\"\", \"\"two\"\", \"\"three\"\")"), "one,two,three");
680	TEST_CI_ERROR_CONTAINS("1,4", WITH_SPLITTED("\|select(\"\", \"\"one\"\", \"\"two\"\", \"\"three\"\")"), "Illegal param number '4' (allowed [0..3])");
681
682	// test define and do
683	TEST_CI("ignored", "define(d4, \"dd;dd;dd;dd\")", "");
684	TEST_CI("ignored", "define(d16, \"do(d4)\|do(d4)\")", "");
685	TEST_CI("ignored", "define(d64, \"do(d4)\|do(d16)\")", "");
686	TEST_CI("ignored", "define(d4096, \"do(d64)\|do(d64)\")", "");
687
688	TEST_CI("x", "do(d4)", "xxxx");
689	TEST_CI("xy", "do(d4)", "xyxyxyxy");
690	TEST_CI("x", "do(d16)", "xxxxxxxxxxxxxxxx");
691	TEST_CI("x", "do(d64)\|len", "64");
692	TEST_CI("xy", "do(d4)\|len", "8");
693	TEST_CI("xy", "do(d4)\|len(\"\")", "8");
694	TEST_CI("xy", "do(d4)\|len(x)", "4");
695	TEST_CI("x", "do(d4096)\|len", "4096");
696
697	// create 4096 streams (disable trace; logs to much):
698	TEST_CI("x", "trace(0)\|dd;dd\|dd;dd\|dd;dd\|dd;dd\|dd;dd\|dd;dd\|dd;dd\|dd;dd\|dd;dd\|dd;dd\|dd;dd\|dd;dd\|streams", "4096");
699	}
700
701	__ATTR__REDUCED_OPTIMIZE__NO_GCSE void TEST_GB_command_interpreter_1b() {
702	ACI_test_env E;
703	GBL_env base_env(E.gbmain(), NULp);
704
705	// execute ACI on species container (=GB_DB) in this section ------------------------------
706	GBDATA * const gb_data = E.gbspecies();
707	GBL_call_env callEnv(gb_data, base_env);
708
709	// streams
710	TEST_CI("x", "dd;dd\|streams", "2");
711	TEST_CI("x", "dd;dd\|dd;dd\|streams", "4");
712	TEST_CI("x", "dd;dd\|dd;dd\|dd;dd\|streams", "8");
713	TEST_CI("x", "do(d4)\|streams", "1"); // stream is merged when do() returns
714
715	TEST_CI("", "ali_name", "ali_16s"); // ask for default-alignment name
716	TEST_CI("", "sequence_type", "rna"); // ask for sequence_type of default-alignment
717
718	// format
719	TEST_CI("acgt", "format", " acgt");
720	TEST_CI("acgt", "format(firsttab=1)", " acgt");
721	TEST_CI("acgt", "format(firsttab=1, width=2)",
722	" ac\n"
723	" gt");
724	TEST_CI("acgt", "format(firsttab=1,tab=1,width=2)",
725	" ac\n"
726	" gt");
727	TEST_CI("acgt", "format(firsttab=0,tab=0,width=2)",
728	"ac\n"
729	"gt");
730	TEST_CI("acgt", "format(firsttab=0,tab=0,width=1)",
731	"a\n"
732	"c\n"
733	"g\n"
734	"t");
735
736	TEST_CI_ERROR_CONTAINS("acgt", "format(gap=0)", "Unknown Parameter 'gap=0' in command 'format'");
737	TEST_CI_ERROR_CONTAINS("acgt", "format(numleft)", "Unknown Parameter 'numleft' in command 'format'");
738
739	// format_sequence
740	TEST_CI_ERROR_CONTAINS("acgt", "format_sequence(numright=5, numleft)", "You may only specify 'numleft' OR 'numright', not both");
741
742	TEST_CI("acgtacgtacgtacg", "format_sequence(firsttab=5,tab=5,width=4,numleft=1)",
743	"1 acgt\n"
744	"5 acgt\n"
745	"9 acgt\n"
746	"13 acg");
747
748	TEST_CI("acgtacgtacgtacg", "format_sequence(firsttab=5,tab=5,width=4,numright=9)", // test EMBL sequence formatting
749	" acgt 4\n"
750	" acgt 8\n"
751	" acgt 12\n"
752	" acg 15");
753
754	TEST_CI("acgtacgtacgtac", "format_sequence(firsttab=5,tab=5,width=4,gap=2,numright=-1)", // autodetect width for 'numright'
755	" ac gt 4\n"
756	" ac gt 8\n"
757	" ac gt 12\n"
758	" ac 14");
759
760	TEST_CI("acgt", "format_sequence(firsttab=0,tab=0,width=2,gap=1)",
761	"a c\n"
762	"g t");
763	TEST_CI("acgt", "format_sequence(firsttab=0,tab=0,width=4,gap=1)", "a c g t");
764	TEST_CI("acgt", "format_sequence(firsttab=0,tab=0,width=4,gap=2)", "ac gt");
765	TEST_CI("acgtacgt", "format_sequence(firsttab=0,width=10,gap=4)", "acgt acgt");
766	TEST_CI("acgtacgt", "format_sequence(firsttab=1,width=10,gap=4)", " acgt acgt");
767
768	TEST_CI("acgt", "format_sequence(firsttab=0,tab=0,gap=0)", "acgt");
769	TEST_CI("acgt", "format_sequence(firsttab=0,tab=0,gap=-1)", "acgt"); // no big alloc
770	TEST_CI("acgt", "format_sequence(firsttab=0,tab=-1,gap=-1)", "acgt"); // no big alloc
771	TEST_CI("acgt", "format(firsttab=0,tab=0,width=-1)", "acgt"); // no big alloc for(!)format
772
773	TEST_CI("acgt", "format(firsttab=-1,tab=0)", "acgt"); // did a 4Gb-alloc!
774	TEST_CI("acgt", "format(firsttab=-1,tab=-1)", "acgt"); // did a 4Gb-alloc!
775	TEST_CI("acgt", "format(firsttab=-1,tab=-1,width=-1)", "acgt"); // did a 4Gb-alloc!
776
777	TEST_CI("acgt", "format_sequence(firsttab=0,tab=0,gap=0,width=-1)", "acgt"); // did a 4Gb-alloc!
778	TEST_CI("acgt", "format_sequence(firsttab=-1,tab=0,gap=-1)", "acgt"); // did a 4Gb-alloc!
779	TEST_CI("acgt", "format_sequence(firsttab=-1,tab=-1,gap=-1)", "acgt"); // did a 4Gb-alloc!
780	TEST_CI("acgt", "format_sequence(firsttab=-1,tab=-1,gap=-1,width=-1)", "acgt"); // did a 4Gb-alloc!
781
782	TEST_CI_ERROR_CONTAINS("acgt", "format_sequence(nl=c)", "Unknown Parameter 'nl=c' in command 'format_sequence'");
783	TEST_CI_ERROR_CONTAINS("acgt", "format_sequence(forcenl=)", "Unknown Parameter 'forcenl=' in command 'format_sequence'");
784
785	TEST_CI_ERROR_CONTAINS("acgt", "format(width=0)", "Illegal zero width");
786	TEST_CI_ERROR_CONTAINS("acgt", "format_sequence(width=0)", "Illegal zero width");
787
788	// remove + keep
789	TEST_CI_NOOP("acgtacgt", "remove(-.)");
790	TEST_CI ("..acg--ta-cgt...", "remove(-.)", "acgtacgt");
791	TEST_CI ("..acg--ta-cgt...", "remove(acgt)", "..---...");
792
793	TEST_CI_NOOP("acgtacgt", "keep(acgt)");
794	TEST_CI ("..acg--ta-cgt...", "keep(-.)", "..---...");
795	TEST_CI ("..acg--ta-cgt...", "keep(acgt)", "acgtacgt");
796
797	// compare + icompare
798	TEST_CI("x,z,y,y,z,x,x,Z,y,Y,Z,x", WITH_SPLITTED("\|compare"), "-1,0,1,1,1,-1");
799	TEST_CI("x,z,y,y,z,x,x,Z,y,Y,Z,x", WITH_SPLITTED("\|icompare"), "-1,0,1,-1,0,1");
800
801	TEST_CI("x,y,z", WITH_SPLITTED("\|compare(\"y\")"), "-1,0,1");
802
803	// equals + iequals
804	TEST_CI("a,b,a,a,a,A", WITH_SPLITTED("\|equals"), "0,1,0");
805	TEST_CI("a,b,a,a,a,A", WITH_SPLITTED("\|iequals"), "0,1,1");
806
807	// contains + icontains
808	TEST_CI("abc,bcd,BCD", WITH_SPLITTED("\|contains(\"bc\")"), "2,1,0");
809	TEST_CI("abc,bcd,BCD", WITH_SPLITTED("\|icontains(\"bc\")"), "2,1,1");
810	TEST_CI("abc,bcd,BCD", WITH_SPLITTED("\|icontains(\"d\")"), "0,3,3");
811
812	// partof + ipartof
813	TEST_CI("abc,BCD,def,deg", WITH_SPLITTED("\|partof(\"abcdefg\")"), "1,0,4,0");
814	TEST_CI("abc,BCD,def,deg", WITH_SPLITTED("\|ipartof(\"abcdefg\")"), "1,2,4,0");
815
816	TEST_CI(", , ,x", WITH_SPLITTED("\|isempty"), "1,0,0,0");
817	TEST_CI(", , ,x", WITH_SPLITTED("\|crop(\" \")\|isempty"), "1,1,1,0");
818
819	// translate
820	TEST_CI("abcdefgh", "translate(abc,cba)", "cbadefgh");
821	TEST_CI("abcdefgh", "translate(cba,abc)", "cbadefgh");
822	TEST_CI("abcdefgh", "translate(hcba,abch,-)", "hcb----a");
823	TEST_CI("abcdefgh", "translate(aceg,aceg,-)", "a-c-e-g-");
824	TEST_CI("abbaabba", "translate(ab,ba,-)", "baabbaab");
825	TEST_CI("abbaabba", "translate(a,x,-)", "x--xx--x");
826	TEST_CI("abbaabba", "translate(,,-)", "--------");
827
828	// echo
829	TEST_CI("", "echo", "");
830	TEST_CI("", "echo(x,y,z)", "xyz");
831	TEST_CI("", "echo(x;y,z)", "xyz"); // check ';' as param-separator
832	TEST_CI("", "echo(x;y;z)", "xyz");
833	TEST_CI("", "echo(x,y,z)\|streams", "3");
834
835	// upper, lower + caps
836	TEST_CI("the QUICK brOwn Fox", "lower", "the quick brown fox");
837	TEST_CI("the QUICK brOwn Fox", "upper", "THE QUICK BROWN FOX");
838	TEST_CI("the QUICK brOwn FoX", "caps", "The Quick Brown Fox");
839	}
840
841	__ATTR__REDUCED_OPTIMIZE__NO_GCSE void TEST_GB_command_interpreter_2a() {
842	ACI_test_env E;
843	GBL_env base_env(E.gbmain(), NULp);
844
845	// execute ACI on species container (=GB_DB) in this section ------------------------------
846	GBDATA * const gb_data = E.gbspecies();
847	GBL_call_env callEnv(gb_data, base_env);
848
849	TEST_CI_ERROR_CONTAINS("a;b;c", "split(;)\|merge(-)", "Invalid separator (cannot be empty");
850	TEST_CI ("a;b;c", "split(\";\")\|merge(-)", "a-b-c");
851
852	// head, tail + mid/mid0
853	TEST_CI ("1234567890", "head(3)", "123");
854	TEST_CI ("1234567890", "head(9)", "123456789");
855	TEST_CI_NOOP("1234567890", "head(10)");
856	TEST_CI_NOOP("1234567890", "head(20)");
857
858	TEST_CI ("1234567890", "tail(4)", "7890");
859	TEST_CI ("1234567890", "tail(9)", "234567890");
860	TEST_CI_NOOP("1234567890", "tail(10)");
861	TEST_CI_NOOP("1234567890", "tail(20)");
862
863	TEST_CI("1234567890", "tail(0)", "");
864	TEST_CI("1234567890", "head(0)", "");
865	TEST_CI("1234567890", "tail(-2)", "");
866	TEST_CI("1234567890", "head(-2)", "");
867
868	TEST_CI("1234567890", "mid(3,5)", "345");
869	TEST_CI("1234567890", "mid(2,2)", "2");
870
871	TEST_CI("1234567890", "mid0(3,5)", "456");
872
873	TEST_CI("1234567890", "mid(9,20)", "90");
874	TEST_CI("1234567890", "mid(20,20)", "");
875
876	TEST_CI("1234567890", "tail(3)", "890"); // example from ../HELP_SOURCE/source/aci.hlp@mid0
877	TEST_CI("1234567890", "mid(-2,0)", "890");
878	TEST_CI("1234567890", "mid0(-3,-1)", "890");
879
880	// tab + pretab
881	TEST_CI("x,xx,xxx", WITH_SPLITTED("\|tab(2)"), "x ,xx,xxx");
882	TEST_CI("x,xx,xxx", WITH_SPLITTED("\|tab(3)"), "x ,xx ,xxx");
883	TEST_CI("x,xx,xxx", WITH_SPLITTED("\|tab(4)"), "x ,xx ,xxx ");
884	TEST_CI("x,xx,xxx", WITH_SPLITTED("\|pretab(2)"), " x,xx,xxx");
885	TEST_CI("x,xx,xxx", WITH_SPLITTED("\|pretab(3)"), " x, xx,xxx");
886	TEST_CI("x,xx,xxx", WITH_SPLITTED("\|pretab(4)"), " x, xx, xxx");
887
888	// crop
889	TEST_CI(" x x ", "crop(\" \")", "x x");
890	TEST_CI("\n \t x x \n \t", "crop(\"\t\n \")", "x x");
891
892	// cut, drop, dropempty and dropzero
893	TEST_CI("one,two,three,four,five,six", WITH_SPLITTED("\|cut(2,3,5)"), "two,three,five");
894	TEST_CI("one,two,three,four,five,six", WITH_SPLITTED("\|drop(2,3,5)"), "one,four,six");
895
896	TEST_CI_ERROR_CONTAINS("a", "drop(2)", "Illegal stream number '2' (allowed [1..1])");
897	TEST_CI_ERROR_CONTAINS("a", "drop(0)", "Illegal stream number '0' (allowed [1..1])");
898	TEST_CI_ERROR_CONTAINS("a", "drop", "syntax: drop(streamnumber[,streamnumber]+)");
899	TEST_CI_ERROR_CONTAINS("a", "cut(2)", "Illegal stream number '2' (allowed [1..1])");
900	TEST_CI_ERROR_CONTAINS("a", "cut(0)", "Illegal stream number '0' (allowed [1..1])");
901	TEST_CI_ERROR_CONTAINS("a", "cut", "syntax: cut(streamnumber[,streamnumber]+)");
902	TEST_CI_ERROR_CONTAINS("a", "cut()", "Invalid empty parameter list '()'");
903	TEST_CI_ERROR_CONTAINS("a", "cut(\"\")", "Illegal stream number '0' (allowed [1..1])"); // still strange (atoi("")->0)
904
905	TEST_CI("one,two,three,four,five,six", WITH_SPLITTED("\|dropempty\|streams"), "6");
906	TEST_CI("one,two,,,five,six", WITH_SPLITTED("\|dropempty\|streams"), "4");
907	TEST_CI(",,,,,", WITH_SPLITTED("\|dropempty"), "");
908	TEST_CI(",,,,,", WITH_SPLITTED("\|dropempty\|streams"), "0");
909
910	TEST_CI("1,0,0,2,3,0", WITH_SPLITTED("\|dropzero"), "1,2,3");
911	TEST_CI("0,0,0,0,0,0", WITH_SPLITTED("\|dropzero"), "");
912	TEST_CI("0,0,0,0,0,0", WITH_SPLITTED("\|dropzero\|streams"), "0");
913
914	TEST_CI("12345", "\|colsplit\|streams", "5");
915	TEST_CI("12345", "\|colsplit" ACI_MERGE, "1,2,3,4,5");
916	TEST_CI("12345", "\|colsplit(3)" ACI_MERGE, "123,45");
917	TEST_CI("12345,678,90", WITH_SPLITTED("\|colsplit(2)"), "12,34,5,67,8,90");
918	TEST_CI_NOOP("12345,678,90", WITH_SPLITTED("\|colsplit(5)"));
919
920	// swap
921	TEST_CI("1,2,3,four,five,six", WITH_SPLITTED("\|swap"), "1,2,3,four,six,five");
922	TEST_CI("1,2,3,four,five,six", WITH_SPLITTED("\|swap(2,3)"), "1,3,2,four,five,six");
923	TEST_CI("1,2,3,four,five,six", WITH_SPLITTED("\|swap(2,3)\|swap(4,3)"), "1,3,four,2,five,six");
924	TEST_CI_NOOP("1,2,3,four,five,six", WITH_SPLITTED("\|swap(3,3)"));
925	TEST_CI_NOOP("1,2,3,four,five,six", WITH_SPLITTED("\|swap(3,2)\|swap(2,3)"));
926	TEST_CI_NOOP("1,2,3,four,five,six", WITH_SPLITTED("\|swap(3,2)\|swap(3,1)\|swap(2,1)\|swap(1,3)"));
927
928	TEST_CI_ERROR_CONTAINS("a", "swap", "need at least two input streams");
929	TEST_CI_ERROR_CONTAINS("a,b", WITH_SPLITTED("\|swap(2,3)"), "Illegal stream number '3' (allowed [1..2])");
930	TEST_CI_ERROR_CONTAINS("a,b", WITH_SPLITTED("\|swap(3,2)"), "Illegal stream number '3' (allowed [1..2])");
931	TEST_CI_ERROR_CONTAINS("a,b", WITH_SPLITTED("\|swap(1)"), "syntax: swap[(streamnumber,streamnumber)]");
932	TEST_CI_ERROR_CONTAINS("a,b", WITH_SPLITTED("\|swap(1,2,3)"), "syntax: swap[(streamnumber,streamnumber)]");
933
934	// toback + tofront
935	TEST_CI ("front,mid,back", WITH_SPLITTED("\|toback(2)"), "front,back,mid");
936	TEST_CI ("front,mid,back", WITH_SPLITTED("\|tofront(2)"), "mid,front,back");
937	TEST_CI_NOOP("front,mid,back", WITH_SPLITTED("\|toback(3)"));
938	TEST_CI_NOOP("front,mid,back", WITH_SPLITTED("\|tofront(1)"));
939	TEST_CI_NOOP("a", WITH_SPLITTED("\|tofront(1)"));
940	TEST_CI_NOOP("a", WITH_SPLITTED("\|toback(1)"));
941
942	TEST_CI_ERROR_CONTAINS("a,b", WITH_SPLITTED("\|tofront(3)"), "Illegal stream number '3' (allowed [1..2])");
943	TEST_CI_ERROR_CONTAINS("a,b", WITH_SPLITTED("\|toback(3)"), "Illegal stream number '3' (allowed [1..2])");
944	TEST_CI_ERROR_CONTAINS("a,b", WITH_SPLITTED("\|tofront"), "syntax: tofront(streamnumber)");
945	TEST_CI_ERROR_CONTAINS("a,b", WITH_SPLITTED("\|toback(1,2)"), "syntax: toback(streamnumber)");
946	TEST_CI_ERROR_CONTAINS("a,b", WITH_SPLITTED("\|merge(1,2)"), "syntax: merge[(\"separator\")]");
947
948	// split
949	TEST_CI ("a\nb", "\|split" ACI_MERGE, "a,b");
950	TEST_CI ("a-b", "\|split(-)" ACI_MERGE, "a,b");
951	TEST_CI ("a-b", "\|split(-,0)" ACI_MERGE, "a,b");
952	TEST_CI ("a-b", "\|split(-,1)" ACI_MERGE, "a,-b");
953	TEST_CI ("a-b", "\|split(-,2)" ACI_MERGE, "a-,b");
954	TEST_CI_ERROR_CONTAINS("a-b", "\|split(-,3)" ACI_MERGE, "Illegal split mode '3' (valid: 0..2)");
955	TEST_CI_ERROR_CONTAINS("a\nb", "\|split(1,2,3)" ACI_MERGE, "syntax: split[(\"separator\"[,mode])]");
956
957	#define C0_9 "0123456789"
958	#define CA_Z "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
959	#define Ca_z "abcdefghijklmnopqrstuvwxyz"
960
961	// extract_words + extract_sequence
962	TEST_CI("1,2,3,four,five,six", "extract_words(\"" C0_9 "\",1)", "1 2 3");
963	TEST_CI("1,2,3,four,five,six", "extract_words(\"" Ca_z "\", 3)", "five four six");
964	TEST_CI("1,2,3,four,five,six", "extract_words(\"" CA_Z "\", 3)", ""); // extract words works case sensitive
965	TEST_CI("1,2,3,four,five,six", "extract_words(\"" Ca_z "\", 4)", "five four");
966	TEST_CI("1,2,3,four,five,six", "extract_words(\"" Ca_z "\", 5)", "");
967	TEST_CI("7 3b 12A 1 767 111 1 77", "extract_words(\"" C0_9 CA_Z Ca_z "\", 1)", "1 1 111 12A 3b 7 767 77"); // does sort a list of helix numbers
968
969	TEST_CI ("1,2,3,four,five,six", "extract_sequence(\"acgtu\", 1.0)", "");
970	TEST_CI ("1,2,3,four,five,six", "extract_sequence(\"acgtu\", 0.5)", "");
971	TEST_CI ("1,2,3,four,five,six", "extract_sequence(\"acgtu\", 0.0)", "four five six");
972	TEST_CI ("..acg--ta-cgt...", "extract_sequence(\"acgtu\", 1.0)", "");
973	TEST_CI_NOOP("..acg--ta-cgt...", "extract_sequence(\"acgtu-.\", 1.0)");
974	TEST_CI_NOOP("..acg--ta-ygt...", "extract_sequence(\"acgtu-.\", 0.7)");
975	TEST_CI ("70 ..acg--ta-cgt... 70", "extract_sequence(\"acgtu-.\", 1.0)", "..acg--ta-cgt...");
976
977	// checksum + gcgchecksum
978	TEST_CI("", "sequence\|checksum", "4C549A5F");
979	TEST_CI("", "sequence \| gcgchecksum", "4308");
980
981	// SRT
982	TEST_CI("The quick brown fox", "srt(\"quick=lazy:brown fox=dog\")", "The lazy dog"); // no need to escape spaces in quoted ACI parameter
983	TEST_CI("The quick brown fox", "srt(quick=lazy:brown\\ fox=dog)", "The lazy dog"); // spaces need to be escaped in unquoted ACI parameter
984	TEST_CI_ERROR_CONTAINS("x", "srt(x=y,z)", "SRT ERROR: no '=' found in command");
985	TEST_CI_ERROR_CONTAINS("x", "srt", "syntax: srt(expr[,expr]+)");
986
987	// REG-replace and -match
988	TEST_CI("starstostripes", "/\\*/--/", "stars--to--stripes");
989
990	TEST_CI_ERROR_CONTAINS("xxx", "//--", "Regular expression format is '/expr/' or '/expr/i', not '//--'");
991	TEST_CI_ERROR_CONTAINS("xxx", "/*/bla/",
992	#if defined(DARWIN)
993	// @@@ RESULT_MODIFIED_OSX: this test depends on library version
994	// should either test for one-of-several results or just test for any error
995	"repetition-operator operand invalid"
996	#else // !DARWIN
997	"Invalid preceding regular expression"
998	#endif
999	);
1000
1001	TEST_CI("sImILaRWIllBE,GonEEASIly", WITH_SPLITTED("\|command(/[A-Z]//)"), "small,only");
1002	TEST_CI("sthBIGinside,FATnotCAP", WITH_SPLITTED("\|command(/([A-Z])+/)"), "BIG,FAT"); // does only do match
1003
1004	// command-queue vs. command-pipe (vs. both as sub-commands)
1005	TEST_CI("a,bb,ccc", WITH_SPLITTED("\|\"[\";len;\"]\""), "[,1,2,3,]"); // queue
1006	TEST_CI("a,bb,ccc", WITH_SPLITTED("\|command(\"\"[\";len;\"]\"\")"), "[1],[2],[3]"); // queue as sub-command
1007
1008	TEST_CI("a,bb,ccc", WITH_SPLITTED("\|len\|minus(1)"), "0,1,2"); // pipe
1009	TEST_CI("a,bb,ccc", WITH_SPLITTED("\|command(\"len\|minus(1)\")"), "0,1,2"); // pipe as sub-command
1010
1011	TEST_CI( "a,bb,ccc,dd", WITH_SPLITTED("\|len\|minus"), "-1,1"); // pipe
1012	TEST_CI_ERROR_CONTAINS("a,bb,ccc,dd", WITH_SPLITTED("\|command(\"len\|minus\")"), "Expect an even number of input streams"); // pipe as sub-command FAILS
1013	}
1014
1015	__ATTR__REDUCED_OPTIMIZE__NO_GCSE void TEST_GB_command_interpreter_2b() {
1016	ACI_test_env E;
1017	GBL_env base_env(E.gbmain(), NULp);
1018
1019	// execute ACI on species container (=GB_DB) in this section ------------------------------
1020	GBDATA * const gb_data = E.gbspecies();
1021	GBL_call_env callEnv(gb_data, base_env);
1022
1023	// calculator
1024	TEST_CI("", "echo(9.9,3.9) \|plus;fplus" ACI_MERGE, "12,13.8");
1025	TEST_CI("", "echo(9.1,3.9) \|minus;fminus" ACI_MERGE, "6,5.2");
1026	TEST_CI("", "echo(9,3.5) \|mult;fmult" ACI_MERGE, "27,31.5");
1027	TEST_CI("", "echo(9,0.1) \|mult;fmult" ACI_MERGE, "0,0.9");
1028	TEST_CI("", "echo(9,3) \|div;fdiv" ACI_MERGE, "3,3");
1029	TEST_CI("", "echo(10,3) \|div;fdiv" ACI_MERGE, "3,3.33333");
1030
1031	TEST_CI("", "echo(9,3)\|rest", "0");
1032	TEST_CI("", "echo(9,5)\|rest", "4");
1033
1034	TEST_CI("", "echo(9,3) \|per_cent;fper_cent" ACI_MERGE, "300,300");
1035	TEST_CI("", "echo(3,9) \|per_cent;fper_cent" ACI_MERGE, "33,33.3333");
1036	TEST_CI("", "echo(1,8) \|per_cent;fper_cent" ACI_MERGE, "12,12.5");
1037	TEST_CI("", "echo(15,16)\|per_cent;fper_cent" ACI_MERGE, "93,93.75");
1038	TEST_CI("", "echo(1,8) \|fper_cent\|round(0)", "13");
1039	TEST_CI("", "echo(15,16)\|fper_cent\|round(0);round(1)" ACI_MERGE, "94,93.8");
1040
1041	TEST_CI("", "echo(1,2,3)\|plus(1)" ACI_MERGE, "2,3,4");
1042	TEST_CI("", "echo(1,2,3)\|minus(2)" ACI_MERGE, "-1,0,1");
1043	TEST_CI("", "echo(1,2,3)\|mult(42)" ACI_MERGE, "42,84,126");
1044	TEST_CI("", "echo(1,2,3)\|div(2)" ACI_MERGE, "0,1,1");
1045	TEST_CI("", "echo(1,2,3)\|rest(2)" ACI_MERGE, "1,0,1");
1046	TEST_CI("", "echo(1,2,3)\|per_cent(3)" ACI_MERGE, "33,66,100");
1047
1048	// rounding
1049	#define ROUND_FLOATS(dig) "echo(0.3826,0.50849,12.58,77.2,700.099,0.9472e-4,0.175e+7)\|round(" #dig ")" ACI_MERGE
1050
1051	TEST_CI("", ROUND_FLOATS(4), "0.3826,0.5085,12.58,77.2,700.099,0.0001,1.75e+06");
1052	TEST_CI("", ROUND_FLOATS(3), "0.383,0.508,12.58,77.2,700.099,0,1.75e+06");
1053	TEST_CI("", ROUND_FLOATS(2), "0.38,0.51,12.58,77.2,700.1,0,1.75e+06");
1054	TEST_CI("", ROUND_FLOATS(1), "0.4,0.5,12.6,77.2,700.1,0,1.75e+06");
1055	TEST_CI("", ROUND_FLOATS(0), "0,1,13,77,700,0,1.75e+06");
1056	TEST_CI("", ROUND_FLOATS(-1), "0,0,10,80,700,0,1.75e+06");
1057	TEST_CI("", ROUND_FLOATS(-2), "0,0,0,100,700,0,1.75e+06");
1058	TEST_CI("", ROUND_FLOATS(-3), "0,0,0,0,1000,0,1.75e+06");
1059	TEST_CI("", ROUND_FLOATS(-5), "0,0,0,0,0,0,1.8e+06");
1060	TEST_CI("", ROUND_FLOATS(-6), "0,0,0,0,0,0,2e+06");
1061
1062
1063	// compare (integers)
1064	TEST_CI("", "echo(9,3)\|isBelow;isAbove;isEqual", "010");
1065	TEST_CI("", "echo(3,9)\|isBelow;isAbove;isEqual", "100");
1066	TEST_CI("", "echo(5,5)\|isBelow;isAbove;isEqual", "001");
1067
1068	TEST_CI("", "echo(1,2,3)\|isBelow(2)", "100");
1069	TEST_CI("", "echo(1,2,3)\|isAbove(2)", "001");
1070	TEST_CI("", "echo(1,2,3)\|isEqual(2)", "010");
1071
1072	TEST_CI("", "echo(1,2,3,4,5)\|inRange(2,4)", "01110");
1073	TEST_CI("", "echo(-1,-2,-3,-4,-5)\|inRange(-2,-4)", "00000"); // empty range
1074	TEST_CI("", "echo(-1,-2,-3,-4,-5)\|inRange(-4,-2)", "01110");
1075
1076	// compare (floats)
1077	TEST_CI("", "echo(1.7,1.4) \|isBelow;isAbove;isEqual", "010");
1078	TEST_CI("", "echo(-0.7,0.1) \|isBelow;isAbove;isEqual", "100");
1079	TEST_CI("", "echo(5.10,5.1) \|isBelow;isAbove;isEqual", "001");
1080	TEST_CI("", "echo(0.10,.11) \|isBelow;isAbove;isEqual", "100");
1081	TEST_CI("", "echo(-7.1,-6.9)\|isBelow;isAbove;isEqual", "100");
1082	TEST_CI("", "echo(1e+5,1e+6)\|isBelow;isAbove;isEqual", "100");
1083	TEST_CI("", "echo(2e+5,1e+6)\|isBelow;isAbove;isEqual", "100");
1084	TEST_CI("", "echo(2e+5,1e-6)\|isBelow;isAbove;isEqual", "010");
1085	TEST_CI("", "echo(2e-5,1e+6)\|isBelow;isAbove;isEqual", "100");
1086
1087	TEST_CI("", "echo(.1,.2,.3,.4,.5) \|inRange(.2,.4)", "01110");
1088	TEST_CI("", "echo(.8,.9,1.0,1.1,1.2)\|inRange(.9,1.1)", "01110");
1089	TEST_CI("", "echo(-.2,-.1,0.0,.1,.2)\|inRange(-.1,.1)", "01110");
1090
1091	// boolean operators
1092	TEST_CI("0", "Not", "1");
1093	TEST_CI("1", "Not", "0");
1094
1095	TEST_CI("", "Not", "1");
1096	TEST_CI("text", "Not", "1");
1097
1098	TEST_CI("", "echo(0,1)\|Not", "10");
1099	TEST_CI("", "echo(0,0)\|Or;And", "00");
1100	TEST_CI("", "echo(0,1)\|Or;And", "10");
1101	TEST_CI("", "echo(1,0)\|Or;And", "10");
1102	TEST_CI("", "echo(1,1)\|Or;And", "11");
1103
1104	TEST_CI("", "command(echo(1\\,0)\|Or);command(echo(0\\,1)\|Or)\|And", "1");
1105
1106	// readdb
1107	TEST_CI("", "readdb(name)", "LcbReu40");
1108	TEST_CI("", "readdb(acc)", "X76328");
1109	TEST_CI("", "readdb(acc,name)", "X76328LcbReu40");
1110
1111	TEST_CI_ERROR_CONTAINS("", "readdb()", "Invalid empty parameter list '()'");
1112	TEST_CI_ERROR_CONTAINS("", "readdb", "syntax: readdb(fieldname[,fieldname]+)");
1113	TEST_CI ("", "readdb(\"\")", ""); // still weird (want field error?)
1114
1115	// taxonomy
1116	TEST_CI("", "taxonomy(1)", "No default tree");
1117	TEST_CI("", "taxonomy(tree_nuc, 1)", "group1");
1118	TEST_CI("", "taxonomy(tree_nuc, 5)", "lower-red/group1");
1119	}
1120
1121	__ATTR__REDUCED_OPTIMIZE__NO_GCSE void TEST_GB_command_interpreter_2c() {
1122	ACI_test_env E;
1123	GBL_env base_env(E.gbmain(), NULp);
1124
1125	GBDATA * const gb_data = E.gbspecies();
1126	GBL_call_env callEnv(gb_data, base_env);
1127
1128	GBL_env env_tree_nuc(E.gbmain(), "tree_nuc");
1129	GBL_call_env callEnv_tree_nuc(gb_data, env_tree_nuc);
1130
1131	TEST_CI_ERROR_CONTAINS("", "taxonomy", "syntax: taxonomy([tree_name,]count)");
1132	TEST_CI_ERROR_CONTAINS("", "taxonomy(1,2,3)", "syntax: taxonomy([tree_name,]count)");
1133	TEST_CI_WITH_ENV("", callEnv_tree_nuc, "taxonomy(1)", "group1");
1134
1135	// diff, filter + change
1136	TEST_CI("..acg--ta-cgt..." ","
1137	"..acg--ta-cgt...", WITH_SPLITTED("\|diff(pairwise=1)"),
1138	"................");
1139	TEST_CI("..acg--ta-cgt..." ","
1140	"..cgt--ta-acg...", WITH_SPLITTED("\|diff(pairwise=1,equal==)"),
1141	"==cgt=====acg===");
1142	TEST_CI("..acg--ta-cgt..." ","
1143	"..cgt--ta-acg...", WITH_SPLITTED("\|diff(pairwise=1,differ=X)"),
1144	"..XXX.....XXX...");
1145	TEST_CI("", "sequence\|diff(species=LcbFruct)\|checksum", "645E3107");
1146
1147	TEST_CI("..XXX.....XXX..." ","
1148	"..acg--ta-cgt...", WITH_SPLITTED("\|filter(pairwise=1,exclude=X)"),
1149	"..--ta-...");
1150	TEST_CI("..XXX.....XXX..." ","
1151	"..acg--ta-cgt...", WITH_SPLITTED("\|filter(pairwise=1,include=X)"),
1152	"acgcgt");
1153	TEST_CI("", "sequence\|filter(species=LcbFruct,include=.-)", "-----------T----T-------G----------C-----T----T...");
1154
1155	TEST_CI("...XXX....XXX..." ","
1156	"..acg--ta-cgt...", WITH_SPLITTED("\|change(pairwise=1,include=X,to=C,change=100)"),
1157	"..aCC--ta-CCC...");
1158	TEST_CI("...XXXXXXXXX...." ","
1159	"..acg--ta-cgt...", WITH_SPLITTED("\|change(pairwise=1,include=X,to=-,change=100)"),
1160	"..a---------t...");
1161
1162	// test environment forwarding
1163	TEST_CI("x", ":=,*(acc)", "x,X76328"); // test DB-item is forwarded to direct SRT-command
1164	TEST_CI("x", "srt(\"=,*(acc)\")", "x,X76328"); // test DB-item is forwarded to ACI-command 'srt'
1165	TEST_CI("x", ":=,*(acc\|dd;\",\";readdb(name))", "x,X76328,LcbReu40"); // test DB-item is forwarded to ACI-subexpression inside SRT-command
1166	TEST_CI("x", "srt(\"=,*(acc\|dd;\"\\,\";readdb(name))\")", "x,X76328,LcbReu40"); // test DB-item is forwarded to ACI-subexpression inside ACI-command 'srt'
1167	TEST_CI("x", "command(\"dd;\\\",\\\";readdb(name)\")", "x,LcbReu40"); // test DB-item is forwarded to ACI-subexpression inside ACI-command 'command'
1168
1169	// test treename is forwarded to sub-expressions
1170	TEST_CI_WITH_ENV("x", callEnv_tree_nuc, ":=,*(acc\|dd;\\\",\\\";taxonomy(1))", "x,X76328,group1");
1171	TEST_CI_WITH_ENV("", callEnv_tree_nuc, "taxonomy(5)\|srt(=\\,*(acc\|dd;\\\",\\\";taxonomy(1)))", "lower-red/group1,X76328,group1");
1172	TEST_CI_WITH_ENV("", callEnv_tree_nuc, "taxonomy(5)\|command(\"dd;\\\",\\\";taxonomy(1)\")", "lower-red/group1,group1");
1173
1174	// test database root is forwarded to sub-expressions (used by commands 'ali_name', 'sequence_type', ...)
1175	TEST_CI("x", ":=,*(acc\|dd;\\\",\\\";ali_name;\\\",\\\";sequence_type)", "x,X76328,ali_16s,rna");
1176	TEST_CI("x", "srt(\"=,*(acc\|dd;\\\",\\\";ali_name;\\\",\\\";sequence_type)\")", "x,X76328,ali_16s,rna");
1177	TEST_CI("x", "command(\"dd;\\\",\\\";ali_name;\\\",\\\";sequence_type\")", "x,ali_16s,rna");
1178
1179	// exec
1180	TEST_CI("c,b,c,b,a,a", WITH_SPLITTED("\|exec(\"(sort\|uniq)\")\|split\|dropempty"), "a,b,c");
1181	TEST_CI("a,aba,cac", WITH_SPLITTED("\|exec(\"perl\",-pe,s/([bc])/$1$1/g)\|split\|dropempty"), "a,abba,ccacc");
1182
1183	// error cases
1184	TEST_CI_ERROR_CONTAINS("", "nocmd", "Unknown command 'nocmd'");
1185	TEST_CI_ERROR_CONTAINS("", "\|nocmd", "Unknown command 'nocmd'");
1186	TEST_CI_ERROR_CONTAINS("", "caps(x)", "syntax: caps (no parameters)");
1187	TEST_CI_ERROR_CONTAINS("", "trace", "syntax: trace(0\|1)");
1188	TEST_CI_ERROR_CONTAINS("", "count", "syntax: count(\"characters to count\")");
1189	TEST_CI_ERROR_CONTAINS("", "count(a,b)", "syntax: count(\"characters to count\")");
1190	TEST_CI_ERROR_CONTAINS("", "len(a,b)", "syntax: len[(\"characters not to count\")]");
1191	TEST_CI_ERROR_CONTAINS("", "plus(a,b,c)", "syntax: plus[(Expr1[,Expr2])]");
1192	TEST_CI_ERROR_CONTAINS("", "count(a,b", "Reason: Missing ')'");
1193	TEST_CI_ERROR_CONTAINS("", "count(a,\"b)", "unbalanced '\"' in 'count(a,\"b)'");
1194	TEST_CI_ERROR_CONTAINS("", "count(a,\"b)\"", "Reason: Missing ')'");
1195	TEST_CI_ERROR_CONTAINS("", "dd;dd\|count", "syntax: count(\"characters to count\")");
1196	TEST_CI_ERROR_CONTAINS("", "\|count(\"a\"x)", "Invalid parameter syntax for '\"a\"x'");
1197	TEST_CI_ERROR_CONTAINS("", "\|count(\"a\"x\")", "unbalanced '\"' in '\|count(\"a\"x\")'");
1198	TEST_CI_ERROR_CONTAINS("", "\|count(\"a)", "unbalanced '\"' in '\|count(\"a)'");
1199
1200	TEST_CI_ERROR_CONTAINS__BROKEN("", "\|\"xx\"bla", "bla", "xx"); // @@@ should report some error referring to unseparated + unknown command 'bla'
1201
1202	TEST_CI_ERROR_CONTAINS("", "translate(a)", "syntax: translate(old,new[,other])");
1203	TEST_CI_ERROR_CONTAINS("", "translate(a,b,c,d)", "syntax: translate(old,new[,other])");
1204	TEST_CI_ERROR_CONTAINS("", "translate(a,b,xx)", "has to be one character");
1205	TEST_CI_ERROR_CONTAINS("", "translate(a,b,)", "has to be one character");
1206
1207	TEST_CI_ERROR_CONTAINS(NULp, "whatever", "ARB ERROR: Can't read this DB entry as string"); // here gb_data is the species container
1208
1209	TEST_CI("hello", ":??""=(?-?)", "(h-e)(l-l)o");
1210	TEST_CI("hello", ":??""=(?-?)?", "(h-e)?(l-l)?o");
1211	TEST_CI("hello", ":??""=(?-?0)?", "(h-e0)?(l-l0)?o");
1212	TEST_CI("hello", ":??""=(?-?3)?", "(h-?)e(l-?)lo");
1213
1214	// show linefeed is handled identical for encoded and escaped linefeeds:
1215	TEST_CI("abc", ":?=?\\n", "a\nb\nc\n");
1216	TEST_CI("abc", ":?=?\n", "a\nb\nc\n");
1217
1218	// same for string-terminator:
1219	TEST_CI("abc", ":?=?.\\0 ignored:b=d", "a.b.c.");
1220	TEST_CI("abc", ":?=?.\0 ignored:b=d", "a.b.c.");
1221
1222	TEST_CI("", ":=XY*(full_name\|len)", "XY21");
1223	TEST_CI("", ":=(full_name\\:reuteri=xxx)", "Lactobacillus xxx");
1224	TEST_CI("", ":=(abc\\:a=A)", ""); // non-existing field -> empty input -> empty output
1225	TEST_CI("hello world", ":* =(\\:=hi)-", "hi-world"); // srt subexpressions also work w/o key
1226	TEST_CI_ERROR_CONTAINS("", ":=(full_name\\:reuteri)", "no '=' found"); // test handling of errors from invalid srt-subexpression
1227
1228	TEST_CI("", ":=(acc#have no acc)", "X76328");
1229	TEST_CI("", ":=(abc#have no abc)", "have no abc");
1230	TEST_CI("", ":=(#no field)", "no field");
1231
1232	TEST_CI_ERROR_CONTAINS("", ":=(unbalanced", "Unbalanced parenthesis in '(unbalanced'");
1233	TEST_CI_ERROR_CONTAINS("", ":=(unba(lan)ced", "Unbalanced parenthesis in '(unba(lan)ced'");
1234	TEST_CI_ERROR_CONTAINS("", ":=(unba(lan)c)ed)", "Invalid char '(' in key 'unba(lan)c'"); // invalid key name
1235	TEST_CI ("", ":=(unbalanc)ed)", "ed)");
1236	}
1237
1238	__ATTR__REDUCED_OPTIMIZE void TEST_GB_command_interpreter_3() {
1239	ACI_test_env E;
1240	GBL_env base_env(E.gbmain(), NULp);
1241
1242	{
1243	// execute ACI on 'full_name' (=GB_STRING) in this section ------------------------------
1244	GBDATA * const gb_data = GB_entry(E.gbspecies(), "full_name");
1245	GBL_call_env callEnv(gb_data, base_env);
1246
1247	TEST_CI(NULp, "", "Lactobacillus reuteri"); // noop
1248	TEST_CI(NULp, "\|len", "21");
1249	TEST_CI(NULp, ":tobac=", "Lacillus reuteri");
1250	TEST_CI(NULp, "/ba.*us/B/", "LactoB reuteri");
1251	TEST_CI(NULp, ":::*=hello:::hell=heaven:::", "heaveno"); // test superfluous ':'s
1252	TEST_CI(NULp, ":* =;2,1;", ";reuteri,Lactobacillus;"); // tests multiple successful matches of ''
1253	TEST_CI(NULp, ":* ??=;?2,?1,2,1;", ";e,r,uteri,Lactobacillus;"); // tests multiple successful matches of '' and '?' (also tests working multi-wildcards "??" and "?*")
1254	TEST_CI(NULp, ":Lactoeutei=1", "Lactobacillus reuteri"); // test match failing after '*' (=> skips replace)
1255	TEST_CI(NULp, ":Lact?bac?lls=?1?2", "Lactobacillus reuteri"); // test match failing after 2nd '?' (=> skips replace)
1256	TEST_CI(NULp, ":*reuteri?=?1", "Lactobacillus reuteri"); // test match failing on '?' behind EOS (=> skips replace)
1257
1258	// tests for (unwanted) multi-wildcards:
1259	TEST_CI__BROKEN(NULp, ":Lacto?lus=(1,?1)", "(baci,l)", "Lactobacillus reuteri"); // @@@ diffcult to achieve (alternative: forbid "*?" and report error)
1260	TEST_CI__BROKEN("Lactobaci\4lus reuteri", ":Lacto?lus=(1,?1)", "<want error instead>", "(baci,?) reuteri"); // @@@ pathological case forcing a match for above situation (ASCII 4 is code for '?' wildcard)
1261	TEST_CI_ERROR_CONTAINS__BROKEN(NULp, ":Lacto*lus=(1,2)", "invalid", "Lactobacillus reuteri"); // @@@ impossible: (forbid "*" and report error)
1262	TEST_CI_ERROR_CONTAINS__BROKEN("Lactobac\3lus reuteri", ":Lacto*lus=(1,2)", "invalid", "(bac,) reuteri"); // @@@ pathological case forcing a match for above situation (ASCII 3 is code for '*' wildcard)
1263
1264	TEST_CI_ERROR_CONTAINS(NULp, ":=(\|wot)", "Unknown command 'wot'"); // provoke error in gbs_build_replace_string [coverage]
1265	TEST_CI_ERROR_CONTAINS("", ":=XY*(\|wot)", "Unknown command 'wot'"); // dito (other caller)
1266
1267	TEST_CI_ERROR_CONTAINS("", ":=XY*(full_name\|len)", "can't read key 'full_name' (DB item is no container)");
1268	TEST_CI ("", ":=XY*(../full_name\|len)", "XY21"); // searches entry via parent-entry (from non-container)
1269
1270	TEST_CI(NULp, "\|taxonomy(1)", "No default tree");
1271	TEST_CI_ERROR_CONTAINS(NULp, "\|taxonomy(tree_nuc,2)", "Container has neither 'name' nor 'group_name' entry - can't detect container type");
1272	}
1273	{
1274	// execute ACI on 'ARB_color' (=GB_INT) in this section ------------------------------
1275	GBDATA * const gb_data = GB_entry(E.gbspecies(), "ARB_color");
1276	GBL_call_env callEnv(gb_data, base_env);
1277
1278	TEST_CI(NULp, "", "1"); // noop
1279	TEST_CI("", "ali_name;\",\";sequence_type", "ali_16s,rna"); // test global database access works when specific database element is specified
1280	}
1281	{
1282	// execute ACI without database element in this section ------------------------------
1283	GBDATA * const gb_data = NULp;
1284	GBL_call_env callEnv(gb_data, base_env);
1285
1286	TEST_CI_ERROR_CONTAINS(NULp, "", "no input streams found");
1287	TEST_CI("", ":=\\tA1Z\t", "\tAZ\t"); // special case (match empty input using '*'); test TAB conversion
1288
1289	TEST_CI_ERROR_CONTAINS("", ":=XY*(\|wot)", "Unknown command 'wot'");
1290	TEST_CI_ERROR_CONTAINS("", ":=XY*(nokey\|len)", "can't read key 'nokey' (called w/o database item)");
1291	TEST_CI_ERROR_CONTAINS("", ":=XY*(nokey)", "can't read key 'nokey' (called w/o database item)");
1292
1293	// test global database access also works w/o specific database element
1294	TEST_CI("", "ali_name;\",\";sequence_type", "ali_16s,rna");
1295	TEST_CI("", "command(\"ali_name;\\\",\\\";sequence_type\")", "ali_16s,rna");
1296
1297	// empty+NULp commands:
1298	TEST_CI("in", NULp, "in");
1299	TEST_CI("in", "", "in");
1300	TEST_CI("in", ":", "in");
1301	TEST_CI("in", "::", "in");
1302
1303	// empty+NULp commands:
1304	TEST_CI("in", NULp, "in");
1305	TEST_CI("in", "", "in");
1306	TEST_CI("in", ":", "in");
1307	TEST_CI("in", "::", "in");
1308	}
1309
1310	// register custom ACI commands
1311	{
1312	const GBL_command_lookup_table& stdCmds = ACI_get_standard_commands();
1313
1314	GBL_command_definition custom_cmds[] = {
1315	{ "custom", gbx_custom }, // new command 'custom'
1316	{ "upper", stdCmds.lookup("lower") }, // change meaning of lower ..
1317	{ "lower", stdCmds.lookup("upper") }, // .. and upper
1318
1319	{NULp, NULp}
1320	};
1321
1322	GBL_custom_command_lookup_table custom(custom_cmds, ARRAY_ELEMS(custom_cmds)-1, stdCmds, PERMIT_SUBSTITUTION);
1323
1324	GBDATA * const gb_data = E.gbspecies();
1325
1326	GBL_env custom_env(E.gbmain(), NULp, custom);
1327	GBL_call_env customCallEnv(gb_data, custom_env);
1328	GBL_call_env callEnv(gb_data, base_env);
1329
1330	// lookup overwritten commands:
1331	TEST_EXPECT(custom.lookup("upper") == stdCmds.lookup("lower"));
1332	TEST_EXPECT(custom.lookup("lower") == stdCmds.lookup("upper"));
1333
1334	// test new commands:
1335	TEST_CI_WITH_ENV ("abc", customCallEnv, "dd;custom;dd", "abc4711abc");
1336	TEST_CI_ERROR_CONTAINS("abc", "dd;custom;dd", "Unknown command 'custom'"); // unknown in standard environment
1337
1338	// test overwritten commands:
1339	TEST_CI_WITH_ENV("abcDEF,", customCallEnv, "dd;lower;upper", "abcDEF,ABCDEF,abcdef,");
1340	TEST_CI ("abcDEF,", "dd;lower;upper", "abcDEF,abcdef,ABCDEF,");
1341	}
1342	}
1343
1344	void TEST_GB_command_interpreter_4() {
1345	ACI_test_env E;
1346	GBL_env base_env(E.gbmain(), NULp);
1347
1348	// execute ACI on species container (=GB_DB) in this section ------------------------------
1349	GBDATA * const gb_data = E.gbspecies();
1350	GBL_call_env callEnv(gb_data, base_env);
1351
1352	TEST_CI("LcbReu40", "findspec(\"readdb (acc)\")", "X76328");
1353	TEST_CI("LcbFruct", "findspec(\"readdb (acc)\")", "X76330");
1354	TEST_CI("", "readdb(name)\|findspec(\"readdb(acc)\")", "X76328");
1355
1356	TEST_CI("LcbReu40;lcbfruct", "split(\";\")\|findspec(\"readdb(acc)\")\|merge(\";\")", "X76328;X76330"); // usecase ("bring next-relatives info into name-independent format")
1357	TEST_CI("X76328;x76330", "split(\";\")\|findacc(\"readdb(name)\")\|merge(\";\")", "LcbReu40;LcbFruct"); // perform opposite (tests 'findacc')
1358
1359	TEST_CI ("", "findspec(\"invalid\")", ""); // does not execute command for unnamed item
1360	TEST_CI_ERROR_CONTAINS("LcbReu40", "findspec(\"invalid\")", "Unknown command 'invalid'");
1361	TEST_CI_ERROR_CONTAINS("unknown", "findspec(\"invalid\")", "No species with name 'unknown' found");
1362	TEST_CI_ERROR_CONTAINS("unknown", "findacc(\"invalid\")", "No species with acc 'unknown' found");
1363	}
1364
1365	#endif // UNIT_TESTS
1366

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

Context Navigation

source: branches/lib/ARBDB/gb_aci.cxx

Download in other formats: