private static string XML_FILENAME; // word list
/**
* This main method reads a list of CSV words and POS tags and looks up against
* the NIHDB Lexicon for a corresponding entry. If found the baseform is written out into a XML
* file, which can be used in SimpleNLG or elsewhere.
*
* @param args : List of Arguments that this command line application must be provided with in order:
* <ol>
* <li>The full path to the NIHDB Lexicon database file e.g. C:\\NIHDB\\lexAccess2009</li>
* <li>The full path to the list of baseforms and POS tags to include in the written out XML Lexicon file</li>
* <li>The full path to the XML file that the XML Lexicon will be written out to.</li>
* </ol>
*
*<p>Example usage:
* java simplenlg.lexicon.util.NIHLexiconXMLDumpUtil C:\\NIHDB\\lexAccess2009 C:\\NIHDB\\wordlist.csv C:\\NIHDB\\default-lexicon.xml
*
* You will need to have the HSQLDB driver (org.hsqldb.jdbc.JDBCDriver) on your Java classpath before running this application.
*</p>
*/
public static void Main(string[] args)
{
Lexicon lex = null;
if (args.Length == 3)
{
DB_FILENAME = args[0];
WORDLIST_FILENAME = args[1];
XML_FILENAME = args[2];
// Check to see if the HSQLDB driver is available on the classpath:
bool dbDriverAvaliable = false;
try
{
// Type driverClass = Type.GetType("org.hsqldb.jdbc.JDBCDriver", false, typeof(NIHLexiconXMLDumpUtil).ClassLoader); //OUTCOMMENTED UNTIL SOLUTION FOUND FOR HSQLDB CONNECTION
Type driverClass = null;
if (null != driverClass)
{
dbDriverAvaliable = true;
}
}
catch (Exception)
{
Console.Error.WriteLine("*** Please add the HSQLDB JDBCDriver to your Java classpath and try again.");
}
if ((null != DB_FILENAME && DB_FILENAME.Length > 0) && (null != WORDLIST_FILENAME && WORDLIST_FILENAME.Length > 0) && (null != XML_FILENAME && XML_FILENAME.Length > 0) && dbDriverAvaliable)
{
lex = new NIHDBLexicon(DB_FILENAME, XMLRealiser.LexiconType.NIHDB_HSQL);
try
{
LineNumberReader wordListFile = new LineNumberReader(new System.IO.StreamReader(WORDLIST_FILENAME));
System.IO.StreamWriter xmlFile = new System.IO.StreamWriter(XML_FILENAME);
xmlFile.BaseStream.WriteByte(Convert.ToByte(string.Format("<lexicon>%n")));
string line = wordListFile.ReadLine();
while (!ReferenceEquals(line, null))
{
string[] cols = line.Split(',');
string @base = cols[0];
string cat = cols[1];
WordElement word = null;
if (cat.Equals("noun", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.NOUN));
}
else if (cat.Equals("verb", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.VERB));
}
else if (cat.Equals("adv", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.ADVERB));
}
else if (cat.Equals("adj", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.ADJECTIVE));
}
else if (cat.Equals("det", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.DETERMINER));
}
else if (cat.Equals("prep", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.PREPOSITION));
}
else if (cat.Equals("pron", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.PRONOUN));
}
else if (cat.Equals("conj", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.CONJUNCTION));
}
else if (cat.Equals("modal", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.MODAL));
}
else if (cat.Equals("interjection", StringComparison.OrdinalIgnoreCase))
{
word = lex.getWord(@base, new LexicalCategory(LexicalCategory.LexicalCategoryEnum.NOUN)); // Kilgarriff;s interjections are mostly nouns in the lexicon
}
if (word == null)
{
Console.WriteLine("*** The following baseform and POS tag is not found: " + @base + ":" + cat);
}
else
{
xmlFile.BaseStream.WriteByte(Convert.ToByte(word.toXML()));
}
line = wordListFile.ReadLine();
}
xmlFile.BaseStream.WriteByte(Convert.ToByte(string.Format("</lexicon>%n")));
wordListFile.Close();
xmlFile.Close();
lex.close();
Console.WriteLine("*** XML Lexicon Export Completed.");
}
catch (Exception e)
{
Console.Error.WriteLine("*** An Error occured during the export. The Exception message is below: ");
Console.Error.WriteLine(e.Message);
Console.Error.WriteLine("************************");
Console.Error.WriteLine("Please make sure you have the correct application arguments: ");
printArgumentsMessage();
}
}
else
{
printErrorArgumentMessage();
}
}
else
{
printErrorArgumentMessage();
}
}