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
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
|
// $Id: XmlOption.cpp 1960 2008-12-15 12:52:38Z phkoehn $
// vim:tabstop=2
/***********************************************************************
Moses - factored phrase-based language decoder
Copyright (C) 2006 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 <cassert>
#include <vector>
#include <string>
#include <set>
#include <iostream>
#include <cstdlib>
#include <sstream>
#include "SyntaxTree.h"
#include "XmlException.h"
using namespace std;
namespace MosesTraining
{
inline std::vector<std::string> Tokenize(const std::string& str,
const std::string& delimiters = " \t")
{
std::vector<std::string> tokens;
// Skip delimiters at beginning.
std::string::size_type lastPos = str.find_first_not_of(delimiters, 0);
// Find first "non-delimiter".
std::string::size_type pos = str.find_first_of(delimiters, lastPos);
while (std::string::npos != pos || std::string::npos != lastPos) {
// Found a token, add it to the vector.
tokens.push_back(str.substr(lastPos, pos - lastPos));
// Skip delimiters. Note the "not_of"
lastPos = str.find_first_not_of(delimiters, pos);
// Find next "non-delimiter"
pos = str.find_first_of(delimiters, lastPos);
}
return tokens;
}
std::string Trim(const std::string& str, const std::string dropChars = " \t\n\r")
{
std::string res = str;
res.erase(str.find_last_not_of(dropChars)+1);
return res.erase(0, res.find_first_not_of(dropChars));
}
string ParseXmlTagAttribute(const string& tag,const string& attributeName)
{
/*TODO deal with unescaping \"*/
string tagOpen = attributeName + "=\"";
size_t contentsStart = tag.find(tagOpen);
if (contentsStart == string::npos) return "";
contentsStart += tagOpen.size();
size_t contentsEnd = tag.find_first_of('"',contentsStart+1);
if (contentsEnd == string::npos) {
cerr << "Malformed XML attribute: "<< tag;
return "";
}
size_t possibleEnd;
while (tag.at(contentsEnd-1) == '\\' && (possibleEnd = tag.find_first_of('"',contentsEnd+1)) != string::npos) {
contentsEnd = possibleEnd;
}
return tag.substr(contentsStart,contentsEnd-contentsStart);
}
/**
* Remove "<" and ">" from XML tag
*
* \param str xml token to be stripped
*/
string TrimXml(const string& str)
{
// too short to be xml token -> do nothing
if (str.size() < 2) return str;
// strip first and last character
if (str[0] == '<' && str[str.size() - 1] == '>') {
return str.substr(1, str.size() - 2);
}
// not an xml token -> do nothing
else {
return str;
}
}
/**
* Check if the token is an XML tag, i.e. starts with "<"
*
* \param tag token to be checked
*/
bool isXmlTag(const string& tag)
{
return tag[0] == '<';
}
/**
* Unescape XML special characters.
*/
string unescape(const string& str)
{
string s;
s.reserve(str.size());
string::size_type n;
string::size_type start = 0;
while ((n = str.find('&', start)) != string::npos) {
s += str.substr(start, n-start);
string::size_type end = str.find(';', n);
assert(n != string::npos);
string name = str.substr(n+1, end-n-1);
if (name == "lt") {
s += string("<");
} else if (name == "gt") {
s += string(">");
} else if (name == "#91") {
s += string("[");
} else if (name == "#93") {
s += string("]");
} else if (name == "bra") {
s += string("[");
} else if (name == "ket") {
s += string("]");
} else if (name == "bar" || name == "#124") {
s += string("|");
} else if (name == "amp") {
s += string("&");
} else if (name == "apos") {
s += string("'");
} else if (name == "quot") {
s += string("\"");
} else {
// Currently only handles the following five XML escape sequences:
// < <
// > >
// & &
// ' '
// " "
// Numeric character references (like ö) are not supported.
std::ostringstream msg;
msg << "unsupported XML escape sequence: &" << name << ";";
throw XmlException(msg.str());
}
if (end == str.size()-1) {
return s;
}
start = end + 1;
}
s += str.substr(start);
return s;
}
/**
* Split up the input character string into tokens made up of
* either XML tags or text.
* example: this <b> is a </b> test .
* => (this ), (<b>), ( is a ), (</b>), ( test .)
*
* \param str input string
*/
vector<string> TokenizeXml(const string& str)
{
string lbrack = "<";
string rbrack = ">";
vector<string> tokens; // vector of tokens to be returned
string::size_type cpos = 0; // current position in string
string::size_type lpos = 0; // left start of xml tag
string::size_type rpos = 0; // right end of xml tag
// walk thorugh the string (loop vver cpos)
while (cpos != str.size()) {
// find the next opening "<" of an xml tag
lpos = str.find_first_of(lbrack, cpos);
if (lpos != string::npos) {
// find the end of the xml tag
rpos = str.find_first_of(rbrack, lpos);
// sanity check: there has to be closing ">"
if (rpos == string::npos) {
cerr << "ERROR: malformed XML: " << str << endl;
return tokens;
}
} else { // no more tags found
// add the rest as token
tokens.push_back(str.substr(cpos));
break;
}
// add stuff before xml tag as token, if there is any
if (lpos - cpos > 0)
tokens.push_back(str.substr(cpos, lpos - cpos));
// add xml tag as token
tokens.push_back(str.substr(lpos, rpos-lpos+1));
cpos = rpos + 1;
}
return tokens;
}
/**
* Process a sentence with xml annotation
* Xml tags may specifiy additional/replacing translation options
* and reordering constraints
*
* \param line in: sentence, out: sentence without the xml
* \param res vector with translation options specified by xml
* \param reorderingConstraint reordering constraint zones specified by xml
* \param walls reordering constraint walls specified by xml
*/
/*TODO: we'd only have to return a vector of XML options if we dropped linking. 2-d vector
is so we can link things up afterwards. We can't create TranslationOptions as we
parse because we don't have the completed source parsed until after this function
removes all the markup from it (CreateFromString in Sentence::Read).
*/
bool ProcessAndStripXMLTags(string &line, SyntaxTree &tree, set< string > &labelCollection, map< string, int > &topLabelCollection, bool unescapeSpecialChars )
{
//parse XML markup in translation line
// no xml tag? we're done.
if (line.find_first_of('<') == string::npos) {
return true;
}
// break up input into a vector of xml tags and text
// example: (this), (<b>), (is a), (</b>), (test .)
vector<string> xmlTokens = TokenizeXml(line);
// we need to store opened tags, until they are closed
// tags are stored as tripled (tagname, startpos, contents)
typedef pair< string, pair< size_t, string > > OpenedTag;
vector< OpenedTag > tagStack; // stack that contains active opened tags
string cleanLine; // return string (text without xml)
size_t wordPos = 0; // position in sentence (in terms of number of words)
// loop through the tokens
for (size_t xmlTokenPos = 0 ; xmlTokenPos < xmlTokens.size() ; xmlTokenPos++) {
// not a xml tag, but regular text (may contain many words)
if(!isXmlTag(xmlTokens[xmlTokenPos])) {
// add a space at boundary, if necessary
if (cleanLine.size()>0 &&
cleanLine[cleanLine.size() - 1] != ' ' &&
xmlTokens[xmlTokenPos][0] != ' ') {
cleanLine += " ";
}
// add words to output
if (unescapeSpecialChars) {
cleanLine += unescape(xmlTokens[xmlTokenPos]);
} else {
cleanLine += xmlTokens[xmlTokenPos];
}
wordPos = Tokenize(cleanLine).size(); // count all the words
}
// process xml tag
else {
// *** get essential information about tag ***
// strip extra boundary spaces and "<" and ">"
string tag = Trim(TrimXml(xmlTokens[xmlTokenPos]));
// cerr << "XML TAG IS: " << tag << std::endl;
if (tag.size() == 0) {
cerr << "ERROR: empty tag name: " << line << endl;
return false;
}
// check if unary (e.g., "<wall/>")
bool isUnary = ( tag[tag.size() - 1] == '/' );
// check if opening tag (e.g. "<a>", not "</a>")g
bool isClosed = ( tag[0] == '/' );
bool isOpen = !isClosed;
if (isClosed && isUnary) {
cerr << "ERROR: can't have both closed and unary tag <" << tag << ">: " << line << endl;
return false;
}
if (isClosed)
tag = tag.substr(1); // remove "/" at the beginning
if (isUnary)
tag = tag.substr(0,tag.size()-1); // remove "/" at the end
// find the tag name and contents
string::size_type endOfName = tag.find_first_of(' ');
string tagName = tag;
string tagContent = "";
if (endOfName != string::npos) {
tagName = tag.substr(0,endOfName);
tagContent = tag.substr(endOfName+1);
}
// *** process new tag ***
if (isOpen || isUnary) {
// put the tag on the tag stack
OpenedTag openedTag = make_pair( tagName, make_pair( wordPos, tagContent ) );
tagStack.push_back( openedTag );
// cerr << "XML TAG " << tagName << " (" << tagContent << ") added to stack, now size " << tagStack.size() << endl;
}
// *** process completed tag ***
if (isClosed || isUnary) {
// pop last opened tag from stack;
if (tagStack.size() == 0) {
cerr << "ERROR: tag " << tagName << " closed, but not opened" << ":" << line << endl;
return false;
}
OpenedTag openedTag = tagStack.back();
tagStack.pop_back();
// tag names have to match
if (openedTag.first != tagName) {
cerr << "ERROR: tag " << openedTag.first << " closed by tag " << tagName << ": " << line << endl;
return false;
}
// assemble remaining information about tag
size_t startPos = openedTag.second.first;
string tagContent = openedTag.second.second;
size_t endPos = wordPos;
// span attribute overwrites position
string span = ParseXmlTagAttribute(tagContent,"span");
if (! span.empty()) {
vector<string> ij = Tokenize(span, "-");
if (ij.size() != 1 && ij.size() != 2) {
cerr << "ERROR: span attribute must be of the form \"i-j\" or \"i\": " << line << endl;
return false;
}
startPos = atoi(ij[0].c_str());
if (ij.size() == 1) endPos = startPos + 1;
else endPos = atoi(ij[1].c_str()) + 1;
}
// cerr << "XML TAG " << tagName << " (" << tagContent << ") spanning " << startPos << " to " << (endPos-1) << " complete, commence processing" << endl;
if (startPos > endPos) {
cerr << "ERROR: tag " << tagName << " startPos is bigger than endPos (" << startPos << "-" << endPos << "): " << line << endl;
return false;
} else if (startPos == endPos) {
cerr << "WARNING: tag " << tagName << ". Ignoring 0 span (" << startPos << "-" << endPos << "): " << line << endl;
continue;
}
string label = ParseXmlTagAttribute(tagContent,"label");
labelCollection.insert( label );
string pcfgString = ParseXmlTagAttribute(tagContent,"pcfg");
float pcfgScore = pcfgString == "" ? 0.0f
: std::atof(pcfgString.c_str());
// report what we have processed so far
if (0) {
cerr << "XML TAG NAME IS: '" << tagName << "'" << endl;
cerr << "XML TAG LABEL IS: '" << label << "'" << endl;
cerr << "XML SPAN IS: " << startPos << "-" << (endPos-1) << endl;
}
SyntaxNode *node = tree.AddNode( startPos, endPos-1, label );
node->SetPcfgScore(pcfgScore);
}
}
}
// we are done. check if there are tags that are still open
if (tagStack.size() > 0) {
cerr << "ERROR: some opened tags were never closed: " << line << endl;
return false;
}
// collect top labels
const vector< SyntaxNode* >& topNodes = tree.GetNodes( 0, wordPos-1 );
for( vector< SyntaxNode* >::const_iterator node = topNodes.begin(); node != topNodes.end(); node++ ) {
SyntaxNode *n = *node;
const string &label = n->GetLabel();
if (topLabelCollection.find( label ) == topLabelCollection.end())
topLabelCollection[ label ] = 0;
topLabelCollection[ label ]++;
}
// return de-xml'ed sentence in line
line = cleanLine;
return true;
}
}
|