/**
* Main program.
*
* <p>\u0040param args Program parameters.
*/
public static void main(String[] args) {
// Initialize.
try {
if (!initialize(args)) {
System.exit(1);
}
// Process all files.
long startTime = System.currentTimeMillis();
int filesProcessed = processFiles(args);
long processingTime = (System.currentTimeMillis() - startTime + 999) / 1000;
// Terminate.
terminate(filesProcessed, processingTime);
} catch (Exception e) {
System.out.println(e.getMessage());
}
}
/**
* Process one file.
*
* <p>\u0040param xmlFileName Input file name to check for part of speech/lemma mismatches..
*/
protected static void processOneFile(String xmlFileName) {
try {
// Report document being processed.
System.out.println(
"(" + ++currentDocNumber + "/" + docsToProcess + ") " + "processing " + xmlFileName);
// Create filter to strip <w> and <c>
// elements.
XMLFilter filter = new StripAllWordElementsFilter(XMLReaderFactory.createXMLReader());
// Strip path from input file name.
String strippedFileName = FileNameUtils.stripPathName(xmlFileName);
strippedFileName = FileNameUtils.changeFileExtension(strippedFileName, "");
// Generate output file name.
String xmlOutputFileName =
new File(outputDirectoryName, strippedFileName + ".xml").getAbsolutePath();
// Make sure output directory exists.
FileUtils.createPathForFile(xmlOutputFileName);
// Copy input xml to output xml,
// stripping <w> and <c> elements.
new FilterAdornedFile(xmlFileName, xmlOutputFileName, filter);
// Read it back and fix spacing.
String fixedXML = FileUtils.readTextFile(xmlOutputFileName, "utf-8");
fixedXML = fixedXML.replaceAll("(\\s+)", " ");
fixedXML = fixedXML.replaceAll(" ([\\.?!,;:\\)])", "\u00241");
fixedXML = fixedXML.replaceAll("\\( ", "(");
fixedXML = fixedXML.replaceAll("\u00b6 ", "\u00b6");
fixedXML = fixedXML.replaceAll("__NS1:", "");
fixedXML = fixedXML.replaceAll("__NS1", "");
/*
fixedXML =
fixedXML.replaceAll
(
"</__NS1:" ,
""
);
*/
// Emit unadorned XML.
SAXBuilder builder = new SAXBuilder();
Document document = builder.build(new StringReader(fixedXML));
new AdornedXMLWriter(document, xmlOutputFileName);
} catch (Exception e) {
System.out.println(" Error: " + e.getMessage());
e.printStackTrace();
}
}
