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
|
#include "PhrasePairCollection.h"
#include <string>
#include <stdlib.h>
#include <cstring>
#include <algorithm>
using namespace std;
PhrasePairCollection::PhrasePairCollection( SuffixArray *sa, TargetCorpus *tc, Alignment *a )
:m_suffixArray(sa)
,m_targetCorpus(tc)
,m_alignment(a)
,m_size(0)
,m_max_lookup(10000)
,m_max_pp_target(50)
,m_max_pp(50)
{}
PhrasePairCollection::~PhrasePairCollection()
{}
bool PhrasePairCollection::GetCollection( const vector< string > sourceString )
{
INDEX first_match, last_match;
if (! m_suffixArray->FindMatches( sourceString, first_match, last_match )) {
return false;
}
cerr << "\tfirst match " << first_match << endl;
cerr << "\tlast match " << last_match << endl;
INDEX found = last_match - first_match +1;
map< vector< WORD_ID >, INDEX > index;
for( INDEX i=first_match; i<=last_match; i++ ) {
int position = m_suffixArray->GetPosition( i );
int source_start = m_suffixArray->GetWordInSentence( position );
int source_end = source_start + sourceString.size()-1;
INDEX sentence_id = m_suffixArray->GetSentence( position );
int sentence_length = m_suffixArray->GetSentenceLength( sentence_id );
int target_length = m_targetCorpus->GetSentenceLength( sentence_id );
cerr << "match " << (i-first_match)
<< " in sentence " << sentence_id
<< ", starting at word " << source_start
<< " of " << sentence_length
<< ". target sentence has " << target_length << " words.";
char target_start, target_end, pre_null, post_null;
if (m_alignment->PhraseAlignment( sentence_id, target_length, source_start, source_end, target_start, target_end, pre_null, post_null)) {
cerr << " aligned to [" << (int)target_start << "," << (int)target_end << "]";
cerr << " +(" << (int)pre_null << "," << (int)post_null << ")";
bool null_boundary_words = false;
for( char pre = 0; pre <= pre_null && (pre==0||null_boundary_words); pre++ ) {
for( char post = 0; post <= post_null && (post==0||null_boundary_words); post++ ) {
vector< WORD_ID > targetString;
cerr << "; ";
for( char target = target_start-pre; target <= target_end+post; target++ ) {
targetString.push_back( m_targetCorpus->GetWordId( sentence_id, target) );
cerr << m_targetCorpus->GetWord( sentence_id, target) << " ";
}
PhrasePair *phrasePair = new PhrasePair( m_suffixArray, m_targetCorpus, m_alignment, sentence_id, target_length, position, source_start, source_end, target_start-pre, target_end+post, pre, post, pre_null-pre, post_null-post);
// matchCollection.Add( sentence_id, )
if (index.find( targetString ) == index.end()) {
index[targetString] = m_collection.size();
vector< PhrasePair* > emptyVector;
m_collection.push_back( emptyVector );
}
m_collection[ index[targetString] ].push_back( phrasePair );
m_size++;
}
}
}
else {
cerr << "mismatch " << (i-first_match)
<< " in sentence " << sentence_id
<< ", starting at word " << source_start
<< " of " << sentence_length
<< ". target sentence has " << target_length << " words.";
Mismatch *mismatch = new Mismatch( m_suffixArray, m_targetCorpus, m_alignment, sentence_id, position, sentence_length, target_length, source_start, source_end );
if (mismatch->Unaligned())
m_unaligned.push_back( mismatch );
else
m_mismatch.push_back( mismatch );
}
cerr << endl;
if (found > m_max_lookup) {
i += found/m_max_lookup-1;
}
}
sort(m_collection.begin(), m_collection.end(), CompareBySize());
}
void PhrasePairCollection::Print()
{
vector< vector<PhrasePair*> >::iterator ppWithSameTarget;
for( ppWithSameTarget = m_collection.begin(); ppWithSameTarget != m_collection.end(); ppWithSameTarget++ ) {
(*(ppWithSameTarget->begin()))->PrintTarget( &cout );
int count = ppWithSameTarget->size();
cout << "(" << count << ")" << endl;
vector< PhrasePair* >::iterator p;
for(p = ppWithSameTarget->begin(); p != ppWithSameTarget->end(); p++ ) {
(*p)->Print( &cout, 100 );
}
}
}
void PhrasePairCollection::PrintHTML()
{
int pp_target = 0;
bool singleton = false;
// loop over all translations
vector< vector<PhrasePair*> >::iterator ppWithSameTarget;
for( ppWithSameTarget = m_collection.begin(); ppWithSameTarget != m_collection.end() && pp_target<m_max_pp_target; ppWithSameTarget++, pp_target++ ) {
int count = ppWithSameTarget->size();
if (!singleton) {
if (count == 1) {
singleton = true;
cout << "<p class=\"pp_singleton_header\">singleton"
<< (m_collection.end() - ppWithSameTarget==1?"":"s") << " ("
<< (m_collection.end() - ppWithSameTarget)
<< "/" << m_size << ")</p>";
}
else {
cout << "<p class=\"pp_target_header\">";
(*(ppWithSameTarget->begin()))->PrintTarget( &cout );
cout << " (" << count << "/" << m_size << ")" << endl;
cout << "<p><div id=\"pp_" << pp_target << "\">";
}
cout << "<table align=\"center\">";
}
vector< PhrasePair* >::iterator p;
// loop over all sentences where translation occurs
int pp=0;
int i=0;
for(p = ppWithSameTarget->begin(); i<10 && pp<count && p != ppWithSameTarget->end(); p++, pp++, i++ ) {
(*p)->PrintClippedHTML( &cout, 160 );
if (count > m_max_pp) {
p += count/m_max_pp-1;
pp += count/m_max_pp-1;
}
}
if (i == 10 && pp < count) {
// extended table
cout << "<tr><td colspan=7 align=center class=\"pp_more\" onclick=\"javascript:document.getElementById('pp_" << pp_target << "').style.display = 'none'; document.getElementById('pp_ext_" << pp_target << "').style.display = 'block';\">(more)</td></tr></table></div>";
cout << "<div id=\"pp_ext_" << pp_target << "\" style=\"display:none;\";\">";
cout << "<table align=\"center\">";
for(i=0, pp=0, p = ppWithSameTarget->begin(); i<m_max_pp && pp<count && p != ppWithSameTarget->end(); p++, pp++, i++ ) {
(*p)->PrintClippedHTML( &cout, 160 );
if (count > m_max_pp) {
p += count/m_max_pp-1;
pp += count/m_max_pp-1;
}
}
}
if (!singleton) cout << "</table></div>\n";
if (!singleton && pp_target == 9) {
cout << "<div id=\"pp_toggle\" onclick=\"javascript:document.getElementById('pp_toggle').style.display = 'none'; document.getElementById('pp_additional').style.display = 'block';\">";
cout << "<p class=\"pp_target_header\">(more)</p></div>";
cout << "<div id=\"pp_additional\" style=\"display:none;\";\">";
}
}
if (singleton) cout << "</table></div>\n";
else if (pp_target > 9) cout << "</div>";
int max_mismatch = m_max_pp/3;
// unaligned phrases
if (m_unaligned.size() > 0) {
cout << "<p class=\"pp_singleton_header\">unaligned"
<< " (" << (m_unaligned.size()) << ")</p>";
cout << "<table align=\"center\">";
int step_size = 1;
if (m_unaligned.size() > max_mismatch)
step_size = (m_unaligned.size()+max_mismatch-1) / max_mismatch;
for(int i=0;i<m_unaligned.size();i+=step_size)
m_unaligned[i]->PrintClippedHTML( &cout, 160 );
cout << "</table>";
}
// mismatched phrases
if (m_mismatch.size() > 0) {
cout << "<p class=\"pp_singleton_header\">mismatched"
<< " (" << (m_mismatch.size()) << ")</p>";
cout << "<table align=\"center\">";
int step_size = 1;
if (m_mismatch.size() > max_mismatch)
step_size = (m_mismatch.size()+max_mismatch-1) / max_mismatch;
for(int i=0;i<m_mismatch.size();i+=step_size)
m_mismatch[i]->PrintClippedHTML( &cout, 160 );
cout << "</table>";
}
}
|
