public void initialize(PropertiesManager propertiesManager, FeatureManager featureManager) {
// load monolingual terminology, if it is given
this.monoTerms = null;
String monoTerminologyPath = propertiesManager.getProperty("terminology.monolingual-mwu");
if (monoTerminologyPath != null) {
this.monoTerms = new Terminology(monoTerminologyPath);
}
// load bilingual terminology, if it is given
this.biTerms = null;
String biTerminologyPath = propertiesManager.getProperty("terminology.bilingual-mwu");
if (biTerminologyPath != null) {
this.biTerms = new Terminology(biTerminologyPath);
}
}
/**
* Performs some basic processing of the input source and target files For English, this consists
* of converting the input to lower case and tokenizing For Arabic, this consists of
* transliteration and tokenization. Please note that the current tools used for tokenizing Arabic
* also perform POS tagging and morphological analysis Although we could separate the tokenization
* process from the more in-depth text analysis performed by these tools, for efficiency reasons
* this is not desirable The input files are also copied to the /input folder. This is necessary
* because the MADA analyser produces its output in the same folder as the input file, which may
* cause problems if the right access rights are not available for that particular folder
*/
private static void preprocessing() {
String sourceInputFolder = input + File.separator + sourceLang;
String targetInputFolder = input + File.separator + targetLang;
File origSourceFile = new File(sourceFile);
File inputSourceFile = new File(sourceInputFolder + File.separator + origSourceFile.getName());
System.out.println("source input:" + sourceFile);
System.out.println("target input:" + targetFile);
File origTargetFile = new File(targetFile);
File inputTargetFile = new File(targetInputFolder + File.separator + origTargetFile.getName());
try {
System.out.println("copying input to " + inputSourceFile.getPath());
copyFile(origSourceFile, inputSourceFile);
System.out.println("copying input to " + inputTargetFile.getPath());
copyFile(origTargetFile, inputTargetFile);
} catch (Exception e) {
e.printStackTrace();
}
// run tokenizer for source (English)
System.out.println("running tokenizer");
String src_abbr = "";
if (sourceLang.equalsIgnoreCase("english")) src_abbr = "en";
else if (sourceLang.equalsIgnoreCase("spanish")) src_abbr = "es";
else if (sourceLang.equalsIgnoreCase("french")) src_abbr = "fr";
else if (sourceLang.equalsIgnoreCase("german")) src_abbr = "de";
else if (targetLang.equalsIgnoreCase("dutch")) src_abbr = "nl";
else if (targetLang.equalsIgnoreCase("portuguese")) src_abbr = "pt";
else if (targetLang.equalsIgnoreCase("czech")) tgt_abbr = "cs";
else System.out.println("Don't recognise the source language");
String tgt_abbr = "";
if (targetLang.equalsIgnoreCase("english")) tgt_abbr = "en";
else if (targetLang.equalsIgnoreCase("spanish")) tgt_abbr = "es";
else if (targetLang.equalsIgnoreCase("french")) tgt_abbr = "fr";
else if (targetLang.equalsIgnoreCase("german")) tgt_abbr = "de";
else if (targetLang.equalsIgnoreCase("dutch")) tgt_abbr = "nl";
else if (targetLang.equalsIgnoreCase("portuguese")) tgt_abbr = "pt";
else if (targetLang.equalsIgnoreCase("czech")) tgt_abbr = "cs";
else System.out.println("Don't recognise the target language");
String truecasePath = "";
if (null != resourceManager.getProperty(sourceLang + ".lowercase")) {
truecasePath = resourceManager.getProperty(sourceLang + ".lowercase") + " -q ";
} else {
truecasePath =
resourceManager.getString(sourceLang + ".truecase")
+ " --model "
+ resourceManager.getString(sourceLang + ".truecase.model");
}
Tokenizer enTok =
new Tokenizer(
inputSourceFile.getPath(),
inputSourceFile.getPath() + ".tok",
truecasePath,
resourceManager.getString(sourceLang + ".tokenizer"),
src_abbr,
forceRun);
// Tokenizer enTok = new Tokenizer(inputSourceFile.getPath(), inputSourceFile.getPath() +
// ".tok", resourceManager.getString("english.lowercase"),
// resourceManager.getString("english.tokenizer"), "en", forceRun);
enTok.run();
sourceFile = enTok.getTok();
System.out.println(sourceFile);
// run tokenizer for target (Spanish)
System.out.println("running tokenizer");
// Tokenizer esTok = new Tokenizer(inputTargetFile.getPath(), inputTargetFile.getPath() +
// ".tok", resourceManager.getString("spanish.lowercase"),
// resourceManager.getString("spanish.tokenizer"), "es", forceRun);
if (null != resourceManager.getProperty(targetLang + ".lowercase")) {
truecasePath = resourceManager.getProperty(targetLang + ".lowercase") + " -q ";
} else {
truecasePath =
resourceManager.getString(targetLang + ".truecase")
+ " --model "
+ resourceManager.getString(targetLang + ".truecase.model");
}
Tokenizer esTok =
new Tokenizer(
inputTargetFile.getPath(),
inputTargetFile.getPath() + ".tok",
truecasePath,
resourceManager.getString(targetLang + ".tokenizer"),
tgt_abbr,
forceRun);
esTok.run();
targetFile = esTok.getTok();
System.out.println(targetFile);
// Normalize files to avoid strange characters in UTF-8 that may break the PoS tagger
// normalize_utf8();
}
