private void readConfigurationFile()
throws ConfigurationFileDuplicateKeyException, ConfigurationException, NumberFormatException,
TeEngineMlException, ParserRunException, NamedEntityRecognizerException,
TextPreprocessorException {
configurationFile = SystemInitialization.loadConfigurationFile(this.configurationFileName);
configurationFile.setExpandingEnvironmentVariables(true);
ConfigurationParams params =
configurationFile.getModuleConfiguration(RTE_PAIRS_PREPROCESS_MODULE_NAME);
instruments = new InstrumentsFactory().getDefaultInstruments(params);
if (params.containsKey(PREPROCESS_DO_NER)) {
doNer = params.getBoolean(PREPROCESS_DO_NER);
} else {
doNer = true;
}
if (params.containsKey(PREPROCESS_DO_TEXT_NORMALIZATION)) {
doTextNormalization = params.getBoolean(PREPROCESS_DO_TEXT_NORMALIZATION);
} else {
doTextNormalization = true;
}
}
/**
* Call this method once before starting to align sentence pairs.
*
* @param config a CommonConfig instance. The aligner retrieves the lexical resources
* configuration values.
* @throws LexicalResourceException if initialization of a resource failed
* @throws ConfigurationException if the configuration is invalid
*/
private void init(CommonConfig config) throws LexicalResourceException, ConfigurationException {
// Get the general parameters configuration section
NameValueTable paramsSection = null;
try {
paramsSection = config.getSection(GENERAL_PARAMS_CONF_SECTION);
} catch (ConfigurationException e) {
throw new ConfigurationException(e);
}
maxPhrase = paramsSection.getInteger(MAX_PHRASE_KEY);
// Get the Lexical Resources configuration section
NameValueTable lexicalResourcesSection = null;
try {
lexicalResourcesSection = config.getSection(LEXICAL_RESOURCES_CONF_SECTION);
} catch (ConfigurationException e) {
throw new ConfigurationException(e);
}
lexicalResources = new ArrayList<LexicalResource<? extends RuleInfo>>();
ConfigurationFile configFile = new ConfigurationFile(config);
// Get each resource and create it using the configuration section related to it
for (String resourceName : lexicalResourcesSection.keySet()) {
// Get the class name
String resourceClassName = lexicalResourcesSection.getString(resourceName);
// Get the configuration params
ConfigurationParams resourceParams = configFile.getModuleConfiguration(resourceName);
resourceParams.setExpandingEnvironmentVariables(true);
LexicalResource<? extends RuleInfo> lexicalResource =
createLexicalResource(resourceClassName, resourceParams);
if (lexicalResource != null) {
lexicalResources.add(lexicalResource);
PartOfSpeech leftSidePOS = null, rightSidePOS = null;
// Add the information about this resource
// Get the right and left side POS, in case it's mentioned
if (resourceParams.keySet().contains(LEFT_SIDE_POS_PARAM)) {
try {
leftSidePOS =
new ByCanonicalPartOfSpeech(resourceParams.getString(LEFT_SIDE_POS_PARAM));
} catch (UnsupportedPosTagStringException e) {
logger.warn(
"Could not load POS for left side: "
+ resourceParams.getString(LEFT_SIDE_POS_PARAM)
+ ". Alignment links of all POS will be retreived.");
}
}
if (resourceParams.keySet().contains(RIGHT_SIDE_POS_PARAM)) {
try {
rightSidePOS =
new ByCanonicalPartOfSpeech(resourceParams.getString(RIGHT_SIDE_POS_PARAM));
} catch (UnsupportedPosTagStringException e) {
logger.warn(
"Could not load POS for right side: "
+ resourceParams.getString(RIGHT_SIDE_POS_PARAM)
+ ". Alignment links of all POS will be retreived.");
}
}
lexicalResourcesInformation.put(
lexicalResource.getClass().getName(),
new LexicalResourceInformation(
resourceParams.getString(VERSION_PARAM),
resourceParams.getBoolean(USE_LEMMA_PARAM),
leftSidePOS,
rightSidePOS));
}
}
}