private static Map<Integer, Alignment> readAlignments(String fileName) {
Map<Integer,Alignment> alignments = new HashMap<Integer, Alignment>();
try {
BufferedReader in = new BufferedReader(new FileReader(fileName));
while (in.ready()) {
String line = in.readLine();
String[] words = line.split("\\s+");
if (words.length != 4)
throw new RuntimeException("Bad alignment file "+fileName+", bad line was "+line);
Integer sentenceID = Integer.parseInt(words[0]);
Integer englishPosition = Integer.parseInt(words[1])-1;
Integer frenchPosition = Integer.parseInt(words[2])-1;
String type = words[3];
Alignment alignment = alignments.get(sentenceID);
if (alignment == null) {
alignment = new Alignment();
alignments.put(sentenceID, alignment);
}
alignment.addAlignment(englishPosition, frenchPosition, type.equals("S"));
}
} catch (IOException e) {
throw new RuntimeException(e);
}
return alignments;
}
private static List<SentencePair> readSentencePairs(String baseFileName) {
List<SentencePair> sentencePairs = new ArrayList<SentencePair>();
String englishFileName = baseFileName + "." + ENGLISH_EXTENSION;
String frenchFileName = baseFileName + "." + FRENCH_EXTENSION;
try {
BufferedReader englishIn = new BufferedReader(new FileReader(englishFileName));
//BufferedReader frenchIn = new BufferedReader(new FileReader(frenchFileName));
BufferedReader frenchIn = new BufferedReader(new InputStreamReader(
new FileInputStream(frenchFileName), StandardCharsets.ISO_8859_1));
while (englishIn.ready() && frenchIn.ready()) {
String englishLine = englishIn.readLine();
String frenchLine = frenchIn.readLine();
Pair<Integer,List<String>> englishSentenceAndID = readSentence(englishLine);
Pair<Integer,List<String>> frenchSentenceAndID = readSentence(frenchLine);
if (! englishSentenceAndID.getFirst().equals(frenchSentenceAndID.getFirst()))
throw new RuntimeException("Sentence ID confusion in file "+baseFileName+", lines were:\n\t"+englishLine+"\n\t"+frenchLine);
sentencePairs.add(new SentencePair(englishSentenceAndID.getFirst(), baseFileName, englishSentenceAndID.getSecond(), frenchSentenceAndID.getSecond()));
}
} catch (IOException e) {
throw new RuntimeException(e);
}
return sentencePairs;
}
