Welcome to mirror list, hosted at ThFree Co, Russian Federation.

PropertiesConsolidator.cpp « phrase-extract - github.com/moses-smt/mosesdecoder.git - Unnamed repository; edit this file 'description' to name the repository.
summaryrefslogtreecommitdiff
blob: 71899e15d788040b78d730c947e0297abcbbadb4 (plain)
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
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
/***********************************************************************
  Moses - factored phrase-based language decoder
  Copyright (C) University of Edinburgh

  This library is free software; you can redistribute it and/or
  modify it under the terms of the GNU Lesser General Public
  License as published by the Free Software Foundation; either
  version 2.1 of the License, or (at your option) any later version.

  This library is distributed in the hope that it will be useful,
  but WITHOUT ANY WARRANTY; without even the implied warranty of
  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  Lesser General Public License for more details.

  You should have received a copy of the GNU Lesser General Public
  License along with this library; if not, write to the Free Software
  Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 ***********************************************************************/

#include "PropertiesConsolidator.h"

#include <sstream>
#include <limits>
#include <vector>

#include "moses/Util.h"
#include "phrase-extract/InputFileStream.h"
#include "phrase-extract/OutputFileStream.h"


namespace MosesTraining
{

void PropertiesConsolidator::ActivateSourceLabelsProcessing(const std::string &sourceLabelSetFile)
{
  Moses::InputFileStream inFile(sourceLabelSetFile);

  // read source label set
  m_sourceLabels.clear();
  std::string line;
  while (getline(inFile, line)) {
    std::istringstream tokenizer(line);
    std::string label;
    size_t index;
    try {
      tokenizer >> label >> index;
    } catch (const std::exception &e) {
      UTIL_THROW2("Error reading source label set file " << sourceLabelSetFile << " .");
    }
    std::pair< std::map<std::string,size_t>::iterator, bool > inserted = m_sourceLabels.insert( std::pair<std::string,size_t>(label,index) );
    UTIL_THROW_IF2(!inserted.second,"Source label set file " << sourceLabelSetFile << " should contain each syntactic label only once.");
  }

  inFile.Close();

  m_sourceLabelsFlag = true;
}


void PropertiesConsolidator::ActivatePartsOfSpeechProcessing(const std::string &partsOfSpeechFile)
{
  Moses::InputFileStream inFile(partsOfSpeechFile);

  // read parts-of-speech vocabulary
  m_partsOfSpeechVocabulary.clear();
  std::string line;
  while (getline(inFile, line)) {
    std::istringstream tokenizer(line);
    std::string label;
    size_t index;
    try {
      tokenizer >> label >> index;
    } catch (const std::exception &e) {
      UTIL_THROW2("Error reading part-of-speech vocabulary file " << partsOfSpeechFile << " .");
    }
    std::pair< std::map<std::string,size_t>::iterator, bool > inserted = m_partsOfSpeechVocabulary.insert( std::pair<std::string,size_t>(label,index) );
    UTIL_THROW_IF2(!inserted.second,"Part-of-speech vocabulary file " << partsOfSpeechFile << " should contain each POS tag only once.");
  }

  inFile.Close();

  m_partsOfSpeechFlag = true;
}


void PropertiesConsolidator::ActivateTargetSyntacticPreferencesProcessing(const std::string &targetSyntacticPreferencesLabelSetFile)
{
  Moses::InputFileStream inFile(targetSyntacticPreferencesLabelSetFile);

  // read target syntactic preferences label set
  m_targetSyntacticPreferencesLabels.clear();
  std::string line;
  while (getline(inFile, line)) {
    std::istringstream tokenizer(line);
    std::string label;
    size_t index;
    try {
      tokenizer >> label >> index;
    } catch (const std::exception &e) {
      UTIL_THROW2("Error reading target syntactic preferences label set file " << targetSyntacticPreferencesLabelSetFile << " .");
    }
    std::pair< std::map<std::string,size_t>::iterator, bool > inserted = m_targetSyntacticPreferencesLabels.insert( std::pair<std::string,size_t>(label,index) );
    UTIL_THROW_IF2(!inserted.second,"Target syntactic preferences label set file " << targetSyntacticPreferencesLabelSetFile << " should contain each syntactic label only once.");
  }

  inFile.Close();

  m_targetSyntacticPreferencesFlag = true;
}


void PropertiesConsolidator::ProcessPropertiesString(const std::string &propertiesString, Moses::OutputFileStream& out) const
{
  if ( propertiesString.empty() ) {
    return;
  }

  std::vector<std::string> toks;
  Moses::TokenizeMultiCharSeparator(toks, propertiesString, "{{");
  for (size_t i = 1; i < toks.size(); ++i) {
    std::string &tok = toks[i];
    if (tok.empty()) {
      continue;
    }
    size_t endPos = tok.rfind("}");
    tok = tok.substr(0, endPos - 1);
    std::vector<std::string> keyValue = Moses::TokenizeFirstOnly(tok, " ");
    assert(keyValue.size() == 2);

    if ( !keyValue[0].compare("SourceLabels") ) {

      if ( m_sourceLabelsFlag ) {

        // SourceLabels property: replace strings with vocabulary indices
        out << " {{" << keyValue[0];
        ProcessSourceLabelsPropertyValue(keyValue[1], out);
        out << "}}";

      } else { // don't process SourceLabels property
        out << " {{" << keyValue[0] << " " << keyValue[1] << "}}";
      }

    } else if ( !keyValue[0].compare("POS") ) {

      /* DO NOTHING (property is not registered in the decoder at the moment)
            if ( m_partsOfSpeechFlag ) {

              // POS property: replace strings with vocabulary indices
              out << " {{" << keyValue[0];
              ProcessPOSPropertyValue(keyValue[1], out);
              out << "}}";

            } else { // don't process POS property
              out << " {{" << keyValue[0] << " " << keyValue[1] << "}}";
            }
      */

    } else if ( !keyValue[0].compare("TargetPreferences") ) {

      if ( m_targetSyntacticPreferencesFlag ) {

        // TargetPreferences property: replace strings with vocabulary indices
        out << " {{" << keyValue[0];
        ProcessTargetSyntacticPreferencesPropertyValue(keyValue[1], out);
        out << "}}";

      } else { // don't process TargetPreferences property
        out << " {{" << keyValue[0] << " " << keyValue[1] << "}}";
      }

    } else {

      // output other property
      out << " {{" << keyValue[0] << " " << keyValue[1] << "}}";
    }
  }
}


void PropertiesConsolidator::ProcessSourceLabelsPropertyValue(const std::string &value, Moses::OutputFileStream& out) const
{
  // SourceLabels property: replace strings with vocabulary indices
  std::istringstream tokenizer(value);

  size_t nNTs;
  double totalCount;

  if (! (tokenizer >> nNTs)) { // first token: number of non-terminals (incl. left-hand side)
    UTIL_THROW2("Not able to read number of non-terminals from SourceLabels property. "
                << "Flawed SourceLabels property?");
  }
  assert( nNTs > 0 );
  out << " " << nNTs;

  if (! (tokenizer >> totalCount)) { // second token: overall rule count
    UTIL_THROW2("Not able to read overall rule count from SourceLabels property. "
                << "Flawed SourceLabels property?");
  }
  assert( totalCount > 0.0 );
  out << " " << totalCount;

  while (tokenizer.peek() != EOF) {
    try {

      size_t numberOfLHSsGivenRHS = std::numeric_limits<std::size_t>::max();

      std::string token;

      if (nNTs > 1) { // rule has right-hand side non-terminals, i.e. it's a hierarchical rule
        for (size_t i=0; i<nNTs-1; ++i) { // RHS source non-terminal labels
          tokenizer >> token; // RHS source non-terminal label
          std::map<std::string,size_t>::const_iterator found = m_sourceLabels.find(token);
          UTIL_THROW_IF2(found == m_sourceLabels.end(), "Label \"" << token << "\" from the phrase table not found in given label set.");
          out << " " << found->second;
        }

        tokenizer >> token; // sourceLabelsRHSCount
        out << " " << token;

        tokenizer >> numberOfLHSsGivenRHS;
        out << " " << numberOfLHSsGivenRHS;
      }

      for (size_t i=0; i<numberOfLHSsGivenRHS && tokenizer.peek()!=EOF; ++i) { // LHS source non-terminal labels seen with this RHS
        tokenizer >> token; // LHS source non-terminal label
        std::map<std::string,size_t>::const_iterator found = m_sourceLabels.find(token);
        UTIL_THROW_IF2(found == m_sourceLabels.end() ,"Label \"" << token << "\" from the phrase table not found in given label set.");
        out << " " << found->second;

        tokenizer >> token; // ruleSourceLabelledCount
        out << " " << token;
      }

    } catch (const std::exception &e) {
      UTIL_THROW2("Flawed item in SourceLabels property?");
    }
  }
}


void PropertiesConsolidator::ProcessPOSPropertyValue(const std::string &value, Moses::OutputFileStream& out) const
{
  std::istringstream tokenizer(value);
  while (tokenizer.peek() != EOF) {
    std::string token;
    tokenizer >> token;
    std::map<std::string,size_t>::const_iterator found = m_partsOfSpeechVocabulary.find(token);
    UTIL_THROW_IF2(found == m_partsOfSpeechVocabulary.end() ,"Part-of-speech \"" << token << "\" from the phrase table not found in given part-of-speech vocabulary.");
    out << " " << found->second;
  }
}


bool PropertiesConsolidator::GetPOSPropertyValueFromPropertiesString(const std::string &propertiesString, std::vector<std::string>& out) const
{
  out.clear();
  if ( propertiesString.empty() ) {
    return false;
  }

  std::vector<std::string> toks;
  Moses::TokenizeMultiCharSeparator(toks, propertiesString, "{{");
  for (size_t i = 1; i < toks.size(); ++i) {
    std::string &tok = toks[i];
    if (tok.empty()) {
      continue;
    }
    size_t endPos = tok.rfind("}");
    tok = tok.substr(0, endPos - 1);
    std::vector<std::string> keyValue = Moses::TokenizeFirstOnly(tok, " ");
    assert(keyValue.size() == 2);

    if ( !keyValue[0].compare("POS") ) {
      std::istringstream tokenizer(keyValue[1]);
      while (tokenizer.peek() != EOF) {
        std::string token;
        tokenizer >> token;
        out.push_back(token);
      }
      return true;
    }
  }

  return false;
}


void PropertiesConsolidator::ProcessTargetSyntacticPreferencesPropertyValue(const std::string &value, Moses::OutputFileStream& out) const
{
  // TargetPreferences property: replace strings with vocabulary indices
  std::istringstream tokenizer(value);

  size_t nNTs;
  double totalCount;

  if (! (tokenizer >> nNTs)) { // first token: number of non-terminals (incl. left-hand side)
    UTIL_THROW2("Not able to read number of non-terminals from TargetPreferences property. "
                << "Flawed TargetPreferences property?");
  }
  assert( nNTs > 0 );
  out << " " << nNTs;

  if (! (tokenizer >> totalCount)) { // second token: overall rule count
    UTIL_THROW2("Not able to read overall rule count from TargetPreferences property. "
                << "Flawed TargetPreferences property?");
  }
  assert( totalCount > 0.0 );
  out << " " << totalCount;

  while (tokenizer.peek() != EOF) {
    try {

      size_t numberOfLHSsGivenRHS = std::numeric_limits<std::size_t>::max();

      std::string token;

      if (nNTs > 1) { // rule has right-hand side non-terminals, i.e. it's a hierarchical rule
        for (size_t i=0; i<nNTs-1; ++i) { // RHS target preference non-terminal labels
          tokenizer >> token; // RHS target preference non-terminal label
          std::map<std::string,size_t>::const_iterator found = m_targetSyntacticPreferencesLabels.find(token);
          UTIL_THROW_IF2(found == m_targetSyntacticPreferencesLabels.end(), "Label \"" << token << "\" from the phrase table not found in given label set.");
          out << " " << found->second;
        }

        tokenizer >> token; // targetPreferenceRHSCount
        out << " " << token;

        tokenizer >> numberOfLHSsGivenRHS;
        out << " " << numberOfLHSsGivenRHS;
      }

      for (size_t i=0; i<numberOfLHSsGivenRHS && tokenizer.peek()!=EOF; ++i) { // LHS target preference non-terminal labels seen with this RHS
        tokenizer >> token; // LHS target preference non-terminal label
        std::map<std::string,size_t>::const_iterator found = m_targetSyntacticPreferencesLabels.find(token);
        UTIL_THROW_IF2(found == m_targetSyntacticPreferencesLabels.end() ,"Label \"" << token << "\" from the phrase table not found in given label set.");
        out << " " << found->second;

        tokenizer >> token; // ruleTargetPreferenceLabelledCount
        out << " " << token;
      }

    } catch (const std::exception &e) {
      UTIL_THROW2("Flawed item in TargetPreferences property?");
    }
  }
}


}  // namespace MosesTraining