Context Navigation

source: tags/svn.1.5.4/NTREE/NT_validNameParser.cxx

Visit:

Last change on this file was 8261, checked in by westram, 14 years ago
removed unused(-but-set) variables/params
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 12.0 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.hxx"
17
18	#include <arb_assert.h>
19
20	#include <cstdlib>
21	#include <cstdlib>
22	#include <iostream>
23	#include <fstream>
24
25	using namespace std;
26
27	#define nt_assert(bed) arb_assert(bed)
28
29	namespace validNames {
30
31
32	TokLPtr tokenize(const string& description, TokLPtr tokenLP)
33	{
34	size_t tokenEnd = 0;
35	size_t tokenBegin = 0;
36
37	while (tokenEnd != description.size()) { // CC : warning: comparison between signed and unsigned (tokenEnd sollte nicht 'int' sondern 'unsigned' sein)
38	tokenEnd = description.find_first_of(' ', tokenBegin);
39	if (tokenEnd == string::npos) tokenEnd = description.size();
40	int tokLength = tokenEnd - tokenBegin;
41	if (tokLength != 0) {
42	tokenLP->push_back(description.substr(tokenBegin, tokenEnd - tokenBegin));
43	}
44	tokenBegin = tokenEnd + 1;
45
46	}
47	return tokenLP;
48	}
49
50
51
52
53
54	Desco determineType(const string& descriptionString)
55	{ // begin determineType
56
57	DESCT actType = NOTYPE;
58	TokLPtr tokenLP = new TokL;
59	tokenLP = tokenize(descriptionString, tokenLP);
60	// remove all tokens in parentheses
61	{
62	TokL::iterator it = tokenLP->begin();
63	while (it != tokenLP->end()) {
64	if (((it).at(0) == '(') && it != string("(corrig.)")) it = tokenLP->erase(it);
65	else ++it;
66	}
67	}
68
69	// check first word for upper case letters
70	string descNames[6]; // first the valid genus, species, subsp. then the other names
71	// stores occurrence of subsp. which is needed to retrieve the right tokens later on and status flags
72	int sspPos[2] = { 0, 0 }; // token subsp. occurs maximum twice
73	int ssp = 0;
74	bool isValid = true;
75	bool isRenamed = false;
76	bool isHetero = false;
77	bool isHomo = false;
78	bool isGenus = false;
79	// bool isSee = false;
80	bool isCorr = false;
81
82
83
84	for (TokL::iterator it = tokenLP->begin(); it != tokenLP->end(); it ++, ssp++)
85	{
86	if (isUpperCase(*it)) {
87	isGenus = true;
88	#if defined(DUMP)
89	std::cout << "genus detected" << std::endl;
90	#endif // DUMP
91	}
92
93
94	else { // begin operators
95	if (*it == string("->")) {
96	nt_assert(!isHetero);
97	nt_assert(!isHomo);
98	nt_assert(isValid); // only one operator per line allowed
99	isRenamed = true;
100	isValid = false;
101	#if defined(DUMP)
102	std::cout << "renaming detected" << std::endl;
103	#endif // DUMP
104	}
105	else {
106	if (*it == string("=>")) {
107	nt_assert(!isRenamed);
108	nt_assert(!isHomo);
109	nt_assert(isValid);
110	isHetero = true;
111	isValid = false;
112	#if defined(DUMP)
113	std::cout << "heteronym detected" << std::endl;
114	#endif // DUMP
115	}
116	else {
117	if (*it == string("=")) {
118	nt_assert(!isRenamed);
119	nt_assert(!isHetero);
120	nt_assert(isValid);
121	isHomo = true;
122	isValid = false;
123	#if defined(DUMP)
124	std::cout << "homonym detected" << std::endl;
125	#endif // DUMP
126	}
127	else {
128	if (*it == string("(corrig.)")) {
129	isCorr = true;
130	#if defined(DUMP)
131	std::cout << "correction" << std::endl;
132	#endif // DUMP
133	}
134	else {
135	if (*it == string("see:")) {
136	// isSee = true;
137	isValid = false;
138	#if defined(DUMP)
139	std::cout << "reference" << std::endl;
140	#endif // DUMP
141	}
142	else {
143	if (*it == string("subsp.")) {
144	#if defined(DUMP)
145	std::cout << "subspecies detected at position: >>>" << ssp << "<<<" << std::endl;
146	#endif // DUMP
147	ssp == 2 ? sspPos[0] = ssp : sspPos[1] = ssp;
148	// max. one subsp. on each operator side
149	#if defined(DUMP)
150	std::cout << "position of subsp.: " << sspPos[0] << "\tand: " << sspPos[1] << std::endl;
151	#endif // DUMP
152	}
153	}
154	}
155	}
156	}
157	}
158	}
159	}
160
161
162
163	if (isGenus) {
164	#if defined(DUMP)
165	std::cout << " GENUS description found " << std::endl;
166	#endif // DUMP
167	if (isValid) {
168	descNames[0] = (*tokenLP)[0];
169	actType = VALGEN;
170	#if defined(DUMP)
171	std::cout << "VALIDGEN type set to: " << actType << std::endl;
172	#endif// DUMP
173	}
174	else {
175	if (isHetero) {
176	descNames[0] = (*tokenLP)[2];
177	descNames[3] = (*tokenLP)[0];
178	actType = HETGEN;
179	#if defined(DUMP)
180	std::cout << "HETERONYMGEN type set to: " << actType << std::endl;
181	#endif // DUMP
182	}
183	else {
184	if (isHomo) {
185	descNames[0] = (*tokenLP)[2];
186	descNames[3] = (*tokenLP)[0];
187	actType = HOMGEN;
188	#if defined(DUMP)
189	std::cout << "HOMONYMGEN type set to: " << actType << std::endl;
190	#endif // DUMP
191
192	}
193	else {
194
195	if (isRenamed) {
196	descNames[0] = (*tokenLP)[2];
197	descNames[3] = (*tokenLP)[0];
198	actType = RENGEN;
199	#if defined(DUMP)
200	std::cout << "RENAMEDGEN type set to: " << actType << std::endl;
201	#endif // DUMP
202	}
203	else {
204	#if defined(DUMP)
205	std::cout << "no meaningful combination of conditions reached" << std::endl
206	<< "for line: " << descriptionString << std::endl;
207	std::cout << "description type is set to NOTYPE: " << NOTYPE << std::endl;
208	#endif // DUMP
209	isValid = false;
210	#if defined(DUMP)
211	std::cout << "isValid set to false " << std::endl;
212	#endif // DUMP
213	actType = NOTYPE;
214	}
215	}
216	}
217	}
218	}
219	else {
220
221	// just fancy experimental , maybe not 100% correct but looks good
222	if (!(((sspPos[0] == 0) \|\| (sspPos[0] == 2)) && (((sspPos[1] > 4)&&(sspPos[1]< 9))\|\|(sspPos[1]==0))))
223	{
224	#if defined(DUMP)
225	std::cout << "subsp. at strange position found in line:" << std::endl << descriptionString << endl;
226	std::cout << "description type is set to NOTYPE: " << NOTYPE << std::endl;
227	#endif // DUMP
228	isValid = false;
229	#if defined(DUMP)
230	std::cout << "isValid set to false " << std::endl;
231	#endif // DUMP
232	actType = NOTYPE;
233	}
234
235	if (isValid) {
236	descNames[0] = (*tokenLP)[0];
237	descNames[1] = (*tokenLP)[1];
238	if (sspPos[0] != 0) { descNames[2] = (*tokenLP)[sspPos[0]+1]; } // only if subsp. exists
239	actType = VALSPEC;
240	}
241	else { // begin else isHetero
242	if (isHetero) {
243	descNames[0] = (*tokenLP)[3 + sspPos[0]];
244	descNames[1] = (*tokenLP)[4 + sspPos[0]];
245	if (sspPos[1]!=0) { descNames[2]=(*tokenLP)[6 + sspPos[0]]; } // only if subsp. exists
246
247	descNames[3] = (*tokenLP)[0];
248	descNames[4] = (*tokenLP)[1];
249	if (sspPos[0]!=0) { descNames[5]=(*tokenLP)[sspPos[0]+1]; } // only if subsp. exists
250
251	actType = HETSPEC;
252	}
253	else {
254	if (isHomo) {
255	descNames[0] = (*tokenLP)[3 + sspPos[0]];
256	descNames[1] = (*tokenLP)[4 + sspPos[0]];
257	if (sspPos[1]!=0) { descNames[2]=(*tokenLP)[6 + sspPos[0]]; } // only if subsp. exists
258
259	descNames[3] = (*tokenLP)[0];
260	descNames[4] = (*tokenLP)[1];
261	if (sspPos[0]!=0) { descNames[5]=(*tokenLP)[sspPos[0]+1]; } // only if subsp. exists
262
263	actType = HOMSPEC;
264
265	}
266	else { // else branch isHomo
267	if (isRenamed) {
268	descNames[0] = (*tokenLP)[3 + sspPos[0]];
269	descNames[1] = (*tokenLP)[4 + sspPos[0]];
270	if (sspPos[1]!=0) { descNames[2]=(*tokenLP)[6 + sspPos[0]]; } // only if subsp. exists
271
272	descNames[3] = (*tokenLP)[0];
273	descNames[4] = (*tokenLP)[1];
274	if (sspPos[0]!=0) { descNames[5]=(*tokenLP)[sspPos[0]+1]; } // only if subsp. exists
275
276	actType = RENSPEC;
277
278	}
279	else { // species remaining cases
280	#if defined(DUMP)
281	std::cout << "not a valid description line detected" << std::endl;
282	std::cout << "isValid: " << isValid << std::endl;
283	std::cout << "isRenamed: " << isRenamed << std::endl;
284	std::cout << "isHetero: " << isHetero << std::endl;
285	std::cout << "isHomo: " << isHomo << std::endl;
286	std::cout << "isGenus: " << isGenus << std::endl;
287	std::cout << "isSee: " << isSee << std::endl;
288	std::cout << "isCorr: " << isCorr << std::endl;
289	std::cout << "sspPos: " << sspPos[0] << " and " << sspPos[1] << std::endl;
290	std::cout << descriptionString << std::endl;
291	#endif // DUMP
292	actType = NOTYPE;
293	}
294
295	}
296	}
297	}
298	}
299
300
301	#if defined(DUMP)
302	std::cout << descriptionString << std::endl;
303	std::cout << "classified as " << actType << std::endl;
304	#endif // DUMP
305
306	Desco actDesc(actType, isCorr, descNames[0], descNames[1], descNames[2], descNames[3], descNames[4], descNames[5]);
307	delete tokenLP;
308	return actDesc;
309	}
310
311
312	string Desco::getFirstName() {
313	string tmp = firstgen;
314	if (!firstspec.empty()) {
315	tmp = tmp + " " + firstspec;
316	if (!firstsub.empty()) {
317	tmp = tmp + " " + "subsp." + " " + firstsub;
318	}
319	}
320
321
322	return tmp;
323	}
324
325	string Desco::getSecondName() {
326	string tmp = secondgen;
327	if (!secondspec.empty()) {
328	tmp = tmp + " " + firstspec;
329	if (!secondsub.empty()) {
330	tmp = tmp + " " + "subsp." + " " + secondsub;
331	}
332	}
333	return tmp;
334	}
335
336
337	bool isUpperCase(const string& input)
338	{
339	for (size_t i=0; i<input.length(); ++i)
340	{
341	if (input[i]<'A' \|\| input[i]>'Z') { return false; }
342	}
343	return true;
344	}
345	}

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

Download in other formats: