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
|
#include "Alignment.h"
#include <string>
#include <stdlib.h>
#include <cstring>
using namespace std;
void Alignment::Create( string fileName )
{
ifstream textFile;
char line[LINE_MAX_LENGTH];
// count the number of words first;
textFile.open(fileName.c_str());
istream *fileP = &textFile;
m_size = 0;
m_sentenceCount = 0;
while(!fileP->eof()) {
SAFE_GETLINE((*fileP), line, LINE_MAX_LENGTH, '\n');
if (fileP->eof()) break;
vector<string> alignmentSequence = Tokenize( line );
m_size += alignmentSequence.size();
m_sentenceCount++;
}
textFile.close();
cerr << m_size << " alignment points" << endl;
// allocate memory
m_array = (char*) calloc( sizeof( char ), m_size*2 );
m_sentenceEnd = (INDEX*) calloc( sizeof( INDEX ), m_sentenceCount );
// fill the array
int alignmentPointIndex = 0;
int sentenceId = 0;
textFile.open(fileName.c_str());
fileP = &textFile;
while(!fileP->eof()) {
SAFE_GETLINE((*fileP), line, LINE_MAX_LENGTH, '\n');
if (fileP->eof()) break;
vector<string> alignmentSequence = Tokenize( line );
for(int i=0; i<alignmentSequence.size(); i++) {
int s,t;
// cout << "scaning " << alignmentSequence[i].c_str() << endl;
if (! sscanf(alignmentSequence[i].c_str(), "%d-%d", &s, &t)) {
cerr << "WARNING: " << alignmentSequence[i] << " is a bad alignment point in sentence " << sentenceId << endl;
}
m_array[alignmentPointIndex++] = (char) s;
m_array[alignmentPointIndex++] = (char) t;
}
m_sentenceEnd[ sentenceId++ ] = alignmentPointIndex - 2;
}
textFile.close();
cerr << "done reading " << (alignmentPointIndex/2) << " alignment points, " << sentenceId << " sentences." << endl;
}
Alignment::~Alignment()
{
free(m_array);
free(m_sentenceEnd);
}
vector<string> Alignment::Tokenize( const char input[] )
{
vector< string > token;
bool betweenWords = true;
int start=0;
int i=0;
for(; input[i] != '\0'; i++) {
bool isSpace = (input[i] == ' ' || input[i] == '\t');
if (!isSpace && betweenWords) {
start = i;
betweenWords = false;
} else if (isSpace && !betweenWords) {
token.push_back( string( input+start, i-start ) );
betweenWords = true;
}
}
if (!betweenWords)
token.push_back( string( input+start, i-start ) );
return token;
}
bool Alignment::PhraseAlignment( INDEX sentence, char target_length,
char source_start, char source_end,
char &target_start, char &target_end,
char &pre_null, char &post_null )
{
vector< char > alignedTargetWords;
// get index for first alignment point
INDEX sentenceStart = 0;
if (sentence > 0) {
sentenceStart = m_sentenceEnd[ sentence-1 ] + 2;
}
// get target phrase boundaries
target_start = target_length;
target_end = 0;
for(INDEX ap = sentenceStart; ap <= m_sentenceEnd[ sentence ]; ap += 2 ) {
char source = m_array[ ap ];
if (source >= source_start && source <= source_end ) {
char target = m_array[ ap+1 ];
if (target < target_start) target_start = target;
if (target > target_end ) target_end = target;
}
}
if (target_start == target_length) {
return false; // done if no alignment points
}
// check consistency
for(INDEX ap = sentenceStart; ap <= m_sentenceEnd[ sentence ]; ap += 2 ) {
char target = m_array[ ap+1 ];
if (target >= target_start && target <= target_end ) {
char source = m_array[ ap ];
if (source < source_start || source > source_end) {
return false; // alignment point out of range
}
}
}
// create array for unaligned words
for( int i=0; i<target_length; i++ ) {
m_unaligned[i] = true;
}
for(INDEX ap = sentenceStart; ap <= m_sentenceEnd[ sentence ]; ap += 2 ) {
char target = m_array[ ap+1 ];
m_unaligned[ target ] = false;
}
// prior unaligned words
pre_null = 0;
for(char target = target_start-1; target >= 0 && m_unaligned[ target ]; target--) {
pre_null++;
}
// post unaligned words;
post_null = 0;
for(char target = target_end+1; target < target_length && m_unaligned[ target ]; target++) {
post_null++;
}
return true;
}
void Alignment::Save( string fileName )
{
FILE *pFile = fopen ( (fileName + ".align").c_str() , "w" );
fwrite( &m_size, sizeof(INDEX), 1, pFile );
fwrite( m_array, sizeof(char), m_size*2, pFile ); // corpus
fwrite( &m_sentenceCount, sizeof(INDEX), 1, pFile );
fwrite( m_sentenceEnd, sizeof(INDEX), m_sentenceCount, pFile); // sentence index
fclose( pFile );
}
void Alignment::Load( string fileName )
{
FILE *pFile = fopen ( (fileName + ".align").c_str() , "r" );
cerr << "loading from " << fileName << ".align" << endl;
fread( &m_size, sizeof(INDEX), 1, pFile );
cerr << "alignment points in corpus: " << m_size << endl;
m_array = (char*) calloc( sizeof(char), m_size*2 );
fread( m_array, sizeof(char), m_size*2, pFile ); // corpus
fread( &m_sentenceCount, sizeof(INDEX), 1, pFile );
cerr << "sentences in corpus: " << m_sentenceCount << endl;
m_sentenceEnd = (INDEX*) calloc( sizeof(INDEX), m_sentenceCount );
fread( m_sentenceEnd, sizeof(INDEX), m_sentenceCount, pFile); // sentence index
fclose( pFile );
cerr << "done loading\n";
}
|
