Morfeusz.cpp
10.9 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
/*
* File: Morfeusz.cpp
* Author: mlenart
*
* Created on November 13, 2013, 5:21 PM
*/
#include <string>
#include <iostream>
#include "fsa/fsa.hpp"
#include "utils.hpp"
#include "data/default_fsa.hpp"
#include "Morfeusz.hpp"
#include "MorphDeserializer.hpp"
#include "GeneratorDeserializer.hpp"
#include "InterpretedChunksDecoder.hpp"
#include "charset/CharsetConverter.hpp"
#include "charset/charset_utils.hpp"
#include "charset/CaseConverter.hpp"
#include "segrules/segrules.hpp"
#include "const.hpp"
// TODO - konstruktor kopiujący działający Tak-Jak-Trzeba
using namespace std;
static Deserializer<vector<InterpsGroup> >* initializeAnalyzerDeserializer() {
static Deserializer < vector < InterpsGroup > > *deserializer
= new MorphDeserializer();
return deserializer;
}
static MorfeuszOptions createDefaultOptions() {
MorfeuszOptions res;
res.caseSensitive = true;
res.encoding = UTF8;
return res;
}
static SegrulesFSA* getDefaultSegrulesFSA(const map<SegrulesOptions, SegrulesFSA*>& map) {
SegrulesOptions opts;
opts["aggl"] = "isolated";
opts["praet"] = "split";
return (*(map.find(opts))).second;
}
Morfeusz::Morfeusz()
: env(Tagset(DEFAULT_FSA), Tagset(DEFAULT_SYNTH_FSA), DEFAULT_MORFEUSZ_CHARSET),
analyzerPtr(DEFAULT_FSA),
analyzerFSA(FSAType::getFSA(analyzerPtr, *initializeAnalyzerDeserializer())),
segrulesFSAsMap(createSegrulesFSAsMap(analyzerPtr)),
currSegrulesFSA(getDefaultSegrulesFSA(segrulesFSAsMap)),
isAnalyzerFSAFromFile(false),
generatorPtr(DEFAULT_SYNTH_FSA),
isGeneratorFSAFromFile(false),
generator(generatorPtr, env),
options(createDefaultOptions()) {
}
static void deleteSegrulesFSAs(std::map<SegrulesOptions, SegrulesFSA*>& fsasMap) {
for (
std::map<SegrulesOptions, SegrulesFSA*>::iterator it = fsasMap.begin();
it != fsasMap.end();
++it) {
delete it->second;
}
fsasMap.clear();
}
void Morfeusz::setAnalyzerFile(const string& filename) {
if (this->isAnalyzerFSAFromFile) {
delete this->analyzerFSA;
deleteSegrulesFSAs(this->segrulesFSAsMap);
delete this->analyzerPtr;
}
this->analyzerPtr = readFile<unsigned char>(filename.c_str());
this->analyzerFSA = FSA< vector<InterpsGroup> > ::getFSA(analyzerPtr, *initializeAnalyzerDeserializer());
this->segrulesFSAsMap = createSegrulesFSAsMap(analyzerPtr);
this->isAnalyzerFSAFromFile = true;
}
void Morfeusz::setGeneratorFile(const string& filename) {
if (this->isGeneratorFSAFromFile) {
delete this->generatorPtr;
}
this->generatorPtr = readFile<unsigned char>(filename.c_str());
this->generator.setGeneratorPtr(generatorPtr);
}
Morfeusz::~Morfeusz() {
if (this->isAnalyzerFSAFromFile) {
delete this->analyzerFSA;
deleteSegrulesFSAs(this->segrulesFSAsMap);
delete this->analyzerPtr;
}
}
void Morfeusz::analyzeOneWord(
const char*& inputStart,
const char* inputEnd,
int startNodeNum,
std::vector<MorphInterpretation>& results) const {
while (inputStart != inputEnd
&& isEndOfWord(this->env.getCharsetConverter().peek(inputStart, inputEnd))) {
this->env.getCharsetConverter().next(inputStart, inputEnd);
}
vector<InterpretedChunk> accum;
FlexionGraph graph;
const char* currInput = inputStart;
SegrulesFSA* segrulesFSA = this->currSegrulesFSA;
doAnalyzeOneWord(currInput, inputEnd, accum, graph, segrulesFSA->initialState);
if (!graph.empty()) {
InterpretedChunksDecoder interpretedChunksDecoder(env);
int srcNode = startNodeNum;
for (unsigned int i = 0; i < graph.getTheGraph().size(); i++) {
vector<FlexionGraph::Edge>& edges = graph.getTheGraph()[i];
for (unsigned int j = 0; j < edges.size(); j++) {
FlexionGraph::Edge& e = edges[j];
int targetNode = startNodeNum + e.nextNode;
interpretedChunksDecoder.decode(srcNode, targetNode, e.chunk, back_inserter(results));
}
srcNode++;
}
// graph.getResults(*this->tagset, results);
}
else if (inputStart != inputEnd) {
this->appendIgnotiumToResults(string(inputStart, currInput), startNodeNum, results);
}
inputStart = currInput;
}
static inline void doShiftOrth(InterpretedChunk& from, InterpretedChunk& to) {
to.prefixChunks.insert(
to.prefixChunks.begin(),
from.prefixChunks.begin(),
from.prefixChunks.end());
to.prefixChunks.push_back(from);
from.orthWasShifted = true;
}
void Morfeusz::doAnalyzeOneWord(
const char*& inputData,
const char* inputEnd,
vector<InterpretedChunk>& accum,
FlexionGraph& graph,
SegrulesState segrulesState) const {
// cerr << "doAnalyzeOneWord " << inputData << endl;
bool endOfWord = inputData == inputEnd;
const char* currInput = inputData;
uint32_t codepoint = endOfWord ? 0 : this->env.getCharsetConverter().next(currInput, inputEnd);
// UnicodeChunk uchunk(*(this->charsetConverter), *(this->caseConverter));
vector<uint32_t> originalCodepoints;
vector<uint32_t> lowercaseCodepoints;
StateType state = this->analyzerFSA->getInitialState();
while (!isEndOfWord(codepoint)) {
uint32_t lowerCP = this->env.getCaseConverter().toLower(codepoint);
originalCodepoints.push_back(codepoint);
lowercaseCodepoints.push_back(lowerCP);
feedState(state, lowerCP, UTF8CharsetConverter());
codepoint = currInput == inputEnd ? 0 : this->env.getCharsetConverter().peek(currInput, inputEnd);
if (!isEndOfWord(codepoint)) {
if (state.isAccepting()) {
vector<InterpsGroup> val(state.getValue());
for (unsigned int i = 0; i < val.size(); i++) {
InterpsGroup& ig = val[i];
// newSegrulesState.proceedToNext(ig.type);
// this->currSegrulesFSA->proceedToNext(ig.type, segrulesStates, newSegrulesStates);
set<SegrulesState> newSegrulesStates;
currSegrulesFSA->proceedToNext(ig.type, segrulesState, newSegrulesStates);
for (
set<SegrulesState>::iterator it = newSegrulesStates.begin();
it != newSegrulesStates.end();
it++) {
SegrulesState newSegrulesState = *it;
// bool shiftOrth = newSegrulesState.getLastTransitionValue() == 1;
// bool shiftOrthSameType = newSegrulesState.getLastTransitionValue() == 2;
InterpretedChunk ic = {
inputData,
originalCodepoints,
lowercaseCodepoints,
ig,
newSegrulesState.shiftOrthFromPrevious,
false,
vector<InterpretedChunk>()
};
if (!accum.empty() && accum.back().shiftOrth) {
doShiftOrth(accum.back(), ic);
}
accum.push_back(ic);
const char* newCurrInput = currInput;
doAnalyzeOneWord(newCurrInput, inputEnd, accum, graph, newSegrulesState);
accum.pop_back();
}
}
}
this->env.getCharsetConverter().next(currInput, inputEnd);
}
}
// cerr << "end of word" << endl;
// we are at the end of word
if (state.isAccepting()) {
vector<InterpsGroup > val(state.getValue());
for (unsigned int i = 0; i < val.size(); i++) {
InterpsGroup& ig = val[i];
// cerr << "currInput=" << currInput << endl;
// cerr << "type=" << (int) ig.type << endl;
set<SegrulesState> newSegrulesStates;
currSegrulesFSA->proceedToNext(ig.type, segrulesState, newSegrulesStates);
for (
set<SegrulesState>::iterator it = newSegrulesStates.begin();
it != newSegrulesStates.end();
it++) {
SegrulesState newSegrulesState = *it;
if (newSegrulesState.accepting) {
InterpretedChunk ic = {
inputData,
originalCodepoints,
lowercaseCodepoints,
ig,
newSegrulesState.shiftOrthFromPrevious,
false,
vector<InterpretedChunk>()};
if (!accum.empty() && accum.back().shiftOrth) {
doShiftOrth(accum.back(), ic);
}
accum.push_back(ic);
graph.addPath(accum);
accum.pop_back();
}
}
}
}
inputData = currInput;
}
void Morfeusz::appendIgnotiumToResults(
const string& word,
int startNodeNum,
std::vector<MorphInterpretation>& results) const {
MorphInterpretation interp = MorphInterpretation::createIgn(startNodeNum, word, env.getAnalyzerTagset(), env.getCharsetConverter());
results.push_back(interp);
}
ResultsIterator Morfeusz::analyze(const string& text) const {
// const char* textStart = text.c_str();
// const char* textEnd = text.c_str() + text.length();
vector<MorphInterpretation> res;
this->analyze(text, res);
return ResultsIterator(res);
}
void Morfeusz::analyze(const string& text, vector<MorphInterpretation>& results) const {
const char* input = text.c_str();
const char* inputEnd = input + text.length();
while (input != inputEnd) {
int startNode = results.empty() ? 0 : results.back().getEndNode();
this->analyzeOneWord(input, inputEnd, startNode, results);
}
}
ResultsIterator Morfeusz::generate(const string& text) const {
// const char* textStart = text.c_str();
// const char* textEnd = text.c_str() + text.length();
vector<MorphInterpretation> res;
this->generate(text, res);
return ResultsIterator(res);
}
void Morfeusz::generate(const string& text, vector<MorphInterpretation>& results) const {
this->generator.generate(text, results);
}
void Morfeusz::setCharset(MorfeuszCharset charset) {
this->options.encoding = charset;
this->env.setCharset(charset);
}
ResultsIterator::ResultsIterator(vector<MorphInterpretation>& res) {
resultsBuffer.insert(resultsBuffer.begin(), res.begin(), res.end());
}
MorphInterpretation ResultsIterator::getNext() {
MorphInterpretation res = this->resultsBuffer.front();
this->resultsBuffer.pop_front();
return res;
}
bool ResultsIterator::hasNext() {
return !resultsBuffer.empty();
}