in morfologik-addon/src/main/java/opennlp/morfologik/cmdline/builder/XMLDictionaryToTableTool.java [51:105]
public void run(String[] args) {
Params params = validateAndParseParams(args, Params.class);
File dictInFile = params.getInputFile();
File dictOutFile = params.getOutputFile();
Charset encoding = params.getEncoding();
SEPARATOR = params.getSeparator();
CmdLineUtil.checkInputFile("dictionary input file", dictInFile);
CmdLineUtil.checkOutputFile("dictionary output file", dictOutFile);
POSDictionary tagDictionary;
try {
tagDictionary = POSDictionary.create(new FileInputStream(dictInFile));
} catch (IOException e) {
throw new TerminateToolException(-1,
"Error while loading XML POS Dictionary: " + e.getMessage(), e);
}
Iterator<String> iterator = tagDictionary.iterator();
try (BufferedWriter writer = Files.newBufferedWriter(dictOutFile.toPath(),
encoding)) {
while (iterator.hasNext()) {
String word = iterator.next();
for (String tag : tagDictionary.getTags(word)) {
if(valid(word,tag)) {
String entry = createEntry(word, tag);
writer.write(entry);
writer.newLine();
}
}
}
writer.close();
System.out.println("Created dictionary: " + dictOutFile.toPath());
} catch (IOException e) {
throw new TerminateToolException(-1, "Error while writing output: "
+ e.getMessage(), e);
}
Properties info = new Properties();
info.setProperty("fsa.dict.separator", SEPARATOR);
info.setProperty("fsa.dict.encoding", params.getEncoding().name());
info.setProperty("fsa.dict.encoder", params.getEncoder());
Path metaPath = DictionaryMetadata.getExpectedMetadataLocation(dictOutFile.toPath());
try {
info.store(Files.newOutputStream(metaPath), "Info file for FSA Morfologik dictionary.");
} catch (IOException e) {
throw new TerminateToolException(-1, "Error while writing metadata output: "
+ e.getMessage(), e);
}
System.out.println("Created metadata: " + dictOutFile.toPath());
}