Context Navigation

source: tags/arb-6.0/NTREE/NT_validNameParser.cxx

Visit:

Last change on this file was 11492, checked in by westram, 10 years ago
fix several doxygen issues (doxygen 1.6.3) force macro expansion for IF_ASSERTION_USED fixed changed param names named some anonymous (nested) structs remove/quote several '#' (triggering unknown commands)
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 11.9 KB

Line
1	/*
2	* Definition of all objects belonging to this version of
3	* the valid names text file
4	*
5	* 29. November 2002
6	*
7	* coded by Lothar Richter
8	*
9	* Copyright (C) 2002 Department of Microbiology (Technical University Munich)
10	*/
11
12	#if defined(DEVEL_LOTHAR)
13	#define DUMP
14	#endif // DEVEL_LOTHAR
15
16	#include "NT_validNameParser.h"
17	#include "NT_local.h"
18
19	#include <arb_assert.h>
20
21	#include <cstdlib>
22	#include <cstdlib>
23	#include <iostream>
24	#include <fstream>
25
26	using namespace std;
27
28	namespace validNames {
29
30
31	TokLPtr tokenize(const std::string& description, TokLPtr tokenLP) {
32	size_t tokenEnd = 0;
33	size_t tokenBegin = 0;
34
35	while (tokenEnd != description.size()) { // CC : warning: comparison between signed and unsigned (tokenEnd sollte nicht 'int' sondern 'unsigned' sein)
36	tokenEnd = description.find_first_of(' ', tokenBegin);
37	if (tokenEnd == string::npos) tokenEnd = description.size();
38	int tokLength = tokenEnd - tokenBegin;
39	if (tokLength != 0) {
40	tokenLP->push_back(description.substr(tokenBegin, tokenEnd - tokenBegin));
41	}
42	tokenBegin = tokenEnd + 1;
43
44	}
45	return tokenLP;
46	}
47
48
49
50
51
52	Desco determineType(const string& descriptionString)
53	{ // begin determineType
54
55	DESCT actType = NOTYPE;
56	TokLPtr tokenLP = new TokL;
57	tokenLP = tokenize(descriptionString, tokenLP);
58	// remove all tokens in parentheses
59	{
60	TokL::iterator it = tokenLP->begin();
61	while (it != tokenLP->end()) {
62	if (((it).at(0) == '(') && it != string("(corrig.)")) it = tokenLP->erase(it);
63	else ++it;
64	}
65	}
66
67	// check first word for upper case letters
68	string descNames[6]; // first the valid genus, species, subsp. then the other names
69	// stores occurrence of subsp. which is needed to retrieve the right tokens later on and status flags
70	int sspPos[2] = { 0, 0 }; // token subsp. occurs maximum twice
71	int ssp = 0;
72	bool isValid = true;
73	bool isRenamed = false;
74	bool isHetero = false;
75	bool isHomo = false;
76	bool isGenus = false;
77	// bool isSee = false;
78	bool isCorr = false;
79
80
81
82	for (TokL::iterator it = tokenLP->begin(); it != tokenLP->end(); ++it, ++ssp) {
83	if (isUpperCase(*it)) {
84	isGenus = true;
85	#if defined(DUMP)
86	std::cout << "genus detected" << std::endl;
87	#endif // DUMP
88	}
89
90
91	else { // begin operators
92	if (*it == string("->")) {
93	nt_assert(!isHetero);
94	nt_assert(!isHomo);
95	nt_assert(isValid); // only one operator per line allowed
96	isRenamed = true;
97	isValid = false;
98	#if defined(DUMP)
99	std::cout << "renaming detected" << std::endl;
100	#endif // DUMP
101	}
102	else {
103	if (*it == string("=>")) {
104	nt_assert(!isRenamed);
105	nt_assert(!isHomo);
106	nt_assert(isValid);
107	isHetero = true;
108	isValid = false;
109	#if defined(DUMP)
110	std::cout << "heteronym detected" << std::endl;
111	#endif // DUMP
112	}
113	else {
114	if (*it == string("=")) {
115	nt_assert(!isRenamed);
116	nt_assert(!isHetero);
117	nt_assert(isValid);
118	isHomo = true;
119	isValid = false;
120	#if defined(DUMP)
121	std::cout << "homonym detected" << std::endl;
122	#endif // DUMP
123	}
124	else {
125	if (*it == string("(corrig.)")) {
126	isCorr = true;
127	#if defined(DUMP)
128	std::cout << "correction" << std::endl;
129	#endif // DUMP
130	}
131	else {
132	if (*it == string("see:")) {
133	// isSee = true;
134	isValid = false;
135	#if defined(DUMP)
136	std::cout << "reference" << std::endl;
137	#endif // DUMP
138	}
139	else {
140	if (*it == string("subsp.")) {
141	#if defined(DUMP)
142	std::cout << "subspecies detected at position: >>>" << ssp << "<<<" << std::endl;
143	#endif // DUMP
144	ssp == 2 ? sspPos[0] = ssp : sspPos[1] = ssp;
145	// max. one subsp. on each operator side
146	#if defined(DUMP)
147	std::cout << "position of subsp.: " << sspPos[0] << "\tand: " << sspPos[1] << std::endl;
148	#endif // DUMP
149	}
150	}
151	}
152	}
153	}
154	}
155	}
156	}
157
158
159
160	if (isGenus) {
161	#if defined(DUMP)
162	std::cout << " GENUS description found " << std::endl;
163	#endif // DUMP
164	if (isValid) {
165	descNames[0] = (*tokenLP)[0];
166	actType = VALGEN;
167	#if defined(DUMP)
168	std::cout << "VALIDGEN type set to: " << actType << std::endl;
169	#endif// DUMP
170	}
171	else {
172	if (isHetero) {
173	descNames[0] = (*tokenLP)[2];
174	descNames[3] = (*tokenLP)[0];
175	actType = HETGEN;
176	#if defined(DUMP)
177	std::cout << "HETERONYMGEN type set to: " << actType << std::endl;
178	#endif // DUMP
179	}
180	else {
181	if (isHomo) {
182	descNames[0] = (*tokenLP)[2];
183	descNames[3] = (*tokenLP)[0];
184	actType = HOMGEN;
185	#if defined(DUMP)
186	std::cout << "HOMONYMGEN type set to: " << actType << std::endl;
187	#endif // DUMP
188
189	}
190	else {
191
192	if (isRenamed) {
193	descNames[0] = (*tokenLP)[2];
194	descNames[3] = (*tokenLP)[0];
195	actType = RENGEN;
196	#if defined(DUMP)
197	std::cout << "RENAMEDGEN type set to: " << actType << std::endl;
198	#endif // DUMP
199	}
200	else {
201	#if defined(DUMP)
202	std::cout << "no meaningful combination of conditions reached" << std::endl
203	<< "for line: " << descriptionString << std::endl;
204	std::cout << "description type is set to NOTYPE: " << NOTYPE << std::endl;
205	#endif // DUMP
206	isValid = false;
207	#if defined(DUMP)
208	std::cout << "isValid set to false " << std::endl;
209	#endif // DUMP
210	actType = NOTYPE;
211	}
212	}
213	}
214	}
215	}
216	else {
217
218	// just fancy experimental , maybe not 100% correct but looks good
219	if (!(((sspPos[0] == 0) \|\| (sspPos[0] == 2)) && (((sspPos[1] > 4)&&(sspPos[1]< 9))\|\|(sspPos[1]==0))))
220	{
221	#if defined(DUMP)
222	std::cout << "subsp. at strange position found in line:" << std::endl << descriptionString << endl;
223	std::cout << "description type is set to NOTYPE: " << NOTYPE << std::endl;
224	#endif // DUMP
225	isValid = false;
226	#if defined(DUMP)
227	std::cout << "isValid set to false " << std::endl;
228	#endif // DUMP
229	actType = NOTYPE;
230	}
231
232	if (isValid) {
233	descNames[0] = (*tokenLP)[0];
234	descNames[1] = (*tokenLP)[1];
235	if (sspPos[0] != 0) { descNames[2] = (*tokenLP)[sspPos[0]+1]; } // only if subsp. exists
236	actType = VALSPEC;
237	}
238	else { // begin else isHetero
239	if (isHetero) {
240	descNames[0] = (*tokenLP)[3 + sspPos[0]];
241	descNames[1] = (*tokenLP)[4 + sspPos[0]];
242	if (sspPos[1]!=0) { descNames[2]=(*tokenLP)[6 + sspPos[0]]; } // only if subsp. exists
243
244	descNames[3] = (*tokenLP)[0];
245	descNames[4] = (*tokenLP)[1];
246	if (sspPos[0]!=0) { descNames[5]=(*tokenLP)[sspPos[0]+1]; } // only if subsp. exists
247
248	actType = HETSPEC;
249	}
250	else {
251	if (isHomo) {
252	descNames[0] = (*tokenLP)[3 + sspPos[0]];
253	descNames[1] = (*tokenLP)[4 + sspPos[0]];
254	if (sspPos[1]!=0) { descNames[2]=(*tokenLP)[6 + sspPos[0]]; } // only if subsp. exists
255
256	descNames[3] = (*tokenLP)[0];
257	descNames[4] = (*tokenLP)[1];
258	if (sspPos[0]!=0) { descNames[5]=(*tokenLP)[sspPos[0]+1]; } // only if subsp. exists
259
260	actType = HOMSPEC;
261
262	}
263	else { // else branch isHomo
264	if (isRenamed) {
265	descNames[0] = (*tokenLP)[3 + sspPos[0]];
266	descNames[1] = (*tokenLP)[4 + sspPos[0]];
267	if (sspPos[1]!=0) { descNames[2]=(*tokenLP)[6 + sspPos[0]]; } // only if subsp. exists
268
269	descNames[3] = (*tokenLP)[0];
270	descNames[4] = (*tokenLP)[1];
271	if (sspPos[0]!=0) { descNames[5]=(*tokenLP)[sspPos[0]+1]; } // only if subsp. exists
272
273	actType = RENSPEC;
274
275	}
276	else { // species remaining cases
277	#if defined(DUMP)
278	std::cout << "not a valid description line detected" << std::endl;
279	std::cout << "isValid: " << isValid << std::endl;
280	std::cout << "isRenamed: " << isRenamed << std::endl;
281	std::cout << "isHetero: " << isHetero << std::endl;
282	std::cout << "isHomo: " << isHomo << std::endl;
283	std::cout << "isGenus: " << isGenus << std::endl;
284	std::cout << "isSee: " << isSee << std::endl;
285	std::cout << "isCorr: " << isCorr << std::endl;
286	std::cout << "sspPos: " << sspPos[0] << " and " << sspPos[1] << std::endl;
287	std::cout << descriptionString << std::endl;
288	#endif // DUMP
289	actType = NOTYPE;
290	}
291
292	}
293	}
294	}
295	}
296
297
298	#if defined(DUMP)
299	std::cout << descriptionString << std::endl;
300	std::cout << "classified as " << actType << std::endl;
301	#endif // DUMP
302
303	Desco actDesc(actType, isCorr, descNames[0], descNames[1], descNames[2], descNames[3], descNames[4], descNames[5]);
304	delete tokenLP;
305	return actDesc;
306	}
307
308
309	string Desco::getFirstName() {
310	string tmp = firstgen;
311	if (!firstspec.empty()) {
312	tmp = tmp + " " + firstspec;
313	if (!firstsub.empty()) {
314	tmp = tmp + " " + "subsp." + " " + firstsub;
315	}
316	}
317
318
319	return tmp;
320	}
321
322	string Desco::getSecondName() {
323	string tmp = secondgen;
324	if (!secondspec.empty()) {
325	tmp = tmp + " " + firstspec;
326	if (!secondsub.empty()) {
327	tmp = tmp + " " + "subsp." + " " + secondsub;
328	}
329	}
330	return tmp;
331	}
332
333
334	bool isUpperCase(const string& input)
335	{
336	for (size_t i=0; i<input.length(); ++i)
337	{
338	if (input[i]<'A' \|\| input[i]>'Z') { return false; }
339	}
340	return true;
341	}
342	}

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

Download in other formats: