/**
* Performs some basic processing of the input source and target files. The input files are also
* copied to the /input folder. This is necessary because the some tools (e.g., the MADA
* morphological analyser) produce their 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
*/
protected void preprocessing() {
String sourceInputFolder = input + File.separator + sourceLang;
String targetInputFolder = input + File.separator + targetLang;
File origSourceFile = new File(getSourceFile());
File inputSourceFile = new File(sourceInputFolder + File.separator + origSourceFile.getName());
System.out.println("source input:" + getSourceFile());
System.out.println("target input:" + getTargetFile());
File origTargetFile = new File(getTargetFile());
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);
setSourceFile(inputSourceFile.getPath());
setTargetFile(inputTargetFile.getPath());
} catch (Exception e) {
e.printStackTrace();
}
}
/**
* makes a copy of one file into another
*
* @param sourceFile the source file path
* @param destFile the destination file path
* @throws IOException
*/
private static void copyFile(File sourceFile, File destFile) throws IOException {
if (sourceFile.equals(destFile)) {
System.out.println("source=dest");
return;
}
if (!destFile.exists()) {
destFile.createNewFile();
}
java.nio.channels.FileChannel source = null;
java.nio.channels.FileChannel destination = null;
try {
source = new FileInputStream(sourceFile).getChannel();
destination = new FileOutputStream(destFile).getChannel();
destination.transferFrom(source, 0, source.size());
} finally {
if (source != null) {
source.close();
}
if (destination != null) {
destination.close();
}
}
}
/**
* constructs the folders required by the application. These are, typically: <br>
*
* <ul>
* <li>/input and subfolders
* <ul>
* <li>/input/<i>sourceLang</i>, /input/<i>targetLang</i> (for storing the results of
* processing the input files with various tools, such as pos tagger, transliterator,
* morphological analyser),<br>
* <li>/input/systems/<i>systemName</i> (for storing system specific resources - for
* example, the compiled and processed word lattices in the case of the IBM system
* </ul>
* <li>/output (for storing the resulting feature files),
* </ul>
*
* Subclasses can override this method to construct, for example, subfolders of the systems folder
* for each available MT system, or temporary folders
*/
public void constructFolders() {
File f = new File(input);
if (!f.exists()) {
f.mkdir();
System.out.println("folder created " + f.getPath());
}
f = new File(input + File.separator + sourceLang);
if (!f.exists()) {
f.mkdir();
System.out.println("folder created " + f.getPath());
}
f = new File(input + File.separator + targetLang);
if (!f.exists()) {
f.mkdir();
System.out.println("folder created " + f.getPath());
}
f = new File(input + File.separator + targetLang + File.separator + "temp");
if (!f.exists()) {
f.mkdir();
System.out.println("folder created " + f.getPath());
}
f = new File(input + File.separator + "systems");
if (!f.exists()) {
f.mkdir();
System.out.println("folder created " + f.getPath());
}
String output = resourceManager.getString("output");
f = new File(output);
if (!f.exists()) {
f.mkdir();
System.out.println("folder created " + f.getPath());
}
}