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
|
// $Id$
// 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 "XmlOption.h"
#include <vector>
#include <string>
#include <iostream>
#include "Util.h"
#include "StaticData.h"
#include "TranslationOption.h"
namespace {
std::string ParseXmlTagAttribute(const std::string& tag,const std::string& attributeName){
/*TODO deal with unescaping \"*/
string tagOpen = attributeName + "=\"";
size_t contentsStart = tag.find(tagOpen);
if (contentsStart == std::string::npos) return "";
contentsStart += tagOpen.size();
size_t contentsEnd = tag.find_first_of('"',contentsStart+1);
if (contentsEnd == std::string::npos) {
TRACE_ERR("Malformed XML attribute: "<< tag);
return "";
}
size_t possibleEnd;
while (tag.at(contentsEnd-1) == '\\' && (possibleEnd = tag.find_first_of('"',contentsEnd+1)) != std::string::npos) {
contentsEnd = possibleEnd;
}
return tag.substr(contentsStart,contentsEnd-contentsStart);
}
std::string TrimXml(const std::string& str) {
if (str.size() < 2) return str;
if (str[0] == '<' && str[str.size() - 1] == '>') {
return str.substr(1, str.size() - 2);
} else { return str; }
}
bool isXmlTag(const std::string& tag)
{
return tag[0] == '<';
}
inline std::vector<std::string> TokenizeXml(const std::string& str)
{
std::string lbrack = "<";
std::string rbrack = ">";
std::vector<std::string> tokens;
// Find first "non-delimiter".
std::string::size_type cpos = 0;
std::string::size_type lpos = 0;
std::string::size_type rpos = 0;
while (cpos != str.size()) {
lpos = str.find_first_of(lbrack, cpos);
if (lpos != std::string::npos) {
rpos = str.find_first_of(rbrack, lpos);
if (rpos == std::string::npos) {
TRACE_ERR("ERROR: malformed XML: " << str << endl);
return tokens;
}
} else {
tokens.push_back(str.substr(cpos));
break;
}
if (lpos - cpos > 0)
tokens.push_back(str.substr(cpos, lpos - cpos));
tokens.push_back(str.substr(lpos, rpos-lpos+1));
cpos = rpos + 1;
}
return tokens;
}
}
std::vector<TranslationOption*> ProcessAndStripXMLTags(std::string& line, const InputType &source) {
//parse XML markup in translation line
std::vector<TranslationOption*> res;
std::string rstr;
std::string linkedStr;
if (line.find_first_of('<') == std::string::npos) { return res; }
std::vector<std::string> xmlTokens = TokenizeXml(line);
std::string tagName = "";
std::string tagContents = "";
std::vector<std::string> altTexts;
std::vector<std::string> altProbs;
size_t tagStart=0;
size_t tagEnd=0;
size_t curWord=0;
int numUnary = 0;
bool doClose = false;
bool isLinked = false;
for (size_t xmlTokenPos = 0 ; xmlTokenPos < xmlTokens.size() ; xmlTokenPos++)
{
if(!isXmlTag(xmlTokens[xmlTokenPos]))
{
//phrase, not tag
rstr += xmlTokens[xmlTokenPos];
curWord = Tokenize(rstr).size();
}
else
{
//tag data
std::string tag = Trim(TrimXml(xmlTokens[xmlTokenPos]));
VERBOSE(3,"XML TAG IS: " << tag << std::endl);
std::string::size_type endOfName = tag.find_first_of(' ');
std::string nextTagName = tag;
bool isUnary = tag[tag.size() - 1] == '/';
bool isOpen = tag[0] != '/';
if (endOfName != std::string::npos) {
nextTagName = tag.substr(0,endOfName);
tagContents = tag.substr(endOfName+1);
}
if (nextTagName == "linked") {
isLinked = true;
linkedStr = "";
}
else if (nextTagName == "/linked") {
isLinked = false;
// recurse to process linked tags
std::vector<TranslationOption*> tOptions = ProcessAndStripXMLTags(linkedStr, source);
// link them together
std::vector<TranslationOption*>::const_iterator iterTransOpts1;
std::vector<TranslationOption*>::const_iterator iterTransOpts2;
for (iterTransOpts1 = tOptions.begin(); iterTransOpts1 != tOptions.end(); iterTransOpts1++) {
for (iterTransOpts2 = tOptions.begin(); iterTransOpts2 != tOptions.end(); iterTransOpts2++) {
if (iterTransOpts1 != iterTransOpts2) {
(**iterTransOpts1).AddLinkedTransOpt(*iterTransOpts2);
}
}
res.push_back(*iterTransOpts1);
}
}
else if (isLinked) {
linkedStr += xmlTokens[xmlTokenPos];
}
else if (isOpen)
{
//this is an open tag
tagName = nextTagName;
altTexts = TokenizeMultiCharSeparator(ParseXmlTagAttribute(tagContents,"english"), "||");
altProbs = TokenizeMultiCharSeparator(ParseXmlTagAttribute(tagContents,"prob"), "||");
std::string span = ParseXmlTagAttribute(tagContents,"span");
tagStart = curWord;
if (isUnary) {
numUnary++;
if (span.empty()) {
TRACE_ERR("ERROR: unary tags must have a span attribute: " << line << endl);
return res;
}
std::vector<std::string> ij = Tokenize(span, ",");
if (ij.size() != 2) {
TRACE_ERR("ERROR: span tag must be of the form \"i,j\": " << line << endl);
return res;
}
tagStart = atoi(ij[0].c_str());
tagEnd = atoi(ij[1].c_str());
if (tagEnd < tagStart) {
TRACE_ERR("ERROR: span tag " << span << " invalid" << endl);
return res;
}
doClose = true;
VERBOSE(3,"XML TAG IS UNARY" << endl);
}
VERBOSE(3,"XML TAG NAME IS: '" << tagName << "'" << endl);
VERBOSE(3,"XML TAG ENGLISH IS: '" << altTexts[0] << "'" << endl);
VERBOSE(3,"XML TAG PROB IS: '" << altProbs[0] << "'" << endl);
VERBOSE(3,"XML TAG STARTS AT WORD: " << tagStart << endl);
if (altTexts.size() != altProbs.size()) {
TRACE_ERR("ERROR: Unequal number of probabilities and translation alternatives: " << line << endl);
return res;
}
}
else if ((nextTagName.size() == 0) || (nextTagName.at(0) != '/') || (nextTagName.substr(1) != tagName))
{
//mismatched tag, abort!
TRACE_ERR("ERROR: tried to parse malformed XML with xml-input enabled: " << line << endl);
return res;
}
else {
doClose = true;
tagEnd = curWord-1; //size is inclusive
}
if (doClose) {
VERBOSE(3,"XML END TAG IS: " << nextTagName.substr(1) << endl);
VERBOSE(3,"XML TAG ENDS AT WORD: " << tagEnd << endl);
//store translation options into members
//TODO: deal with multiple XML options here
if (StaticData::Instance().GetXmlInputType() != XmlIgnore) {
const std::vector<FactorType> &outputFactorOrder = StaticData::Instance().GetOutputFactorOrder();
for (size_t i=0; i<altTexts.size(); ++i) {
//only store options if we aren't ignoring them
//set default probability
float probValue = 1;
if (altProbs[i] != "") probValue = Scan<float>(altProbs[i]);
//Convert from prob to log-prob
float scoreValue = FloorScore(TransformScore(probValue));
TargetPhrase targetPhrase(Output);
targetPhrase.CreateFromString(outputFactorOrder,altTexts[i],StaticData::Instance().GetFactorDelimiter());
targetPhrase.SetScore(scoreValue);
WordsRange range(tagStart,tagEnd);
TranslationOption *option = new TranslationOption(range,targetPhrase,source);
assert(option);
res.push_back(option);
}
}
tagName= "";
tagContents = "";
altTexts.clear();
altProbs.clear();
doClose = false;
}
}
}
line = rstr;
return res;
}
|
