private void ParseDocumentCoNLLUP(string fileName, LegalTextParserFactory parserFactory)
{
//Convert to an XML representation
CoNLLDocument sourceDoc = null;
//Parse the same way as XML
ParseDocumentXml(sourceDoc, parserFactory);
}
public ParsedDocument(string fileName, LegalTextParserFactory parserFactory, bool isXml = true)
{
if (isXml)
{
ParseDocumentXml(fileName, parserFactory);
}
else
{
ParseDocumentCoNLLUP(fileName, parserFactory);
}
}
public CorpusIndexManager(IndexingMode indexingMode, string rootDirectory, string indexDirectory, string tbxIateLocation = null, string logLocation = null, int?parallelIndexingLimit = null, bool readOnly = true)
{
m_readOnly = readOnly;
m_indexingMode = indexingMode;
if (!Directory.Exists(rootDirectory))
{
throw new ArgumentException("The source directory must exist!");
}
/*if (Directory.Exists(indexDirectory) && !Directory.EnumerateFileSystemEntries(indexDirectory).GetEnumerator().MoveNext())
* {
* throw new ArgumentException("The indexing directory must be empty!");
* }*/
m_rootSourceDirectory = rootDirectory;
if (logLocation != null)
{
m_logLocation = logLocation;
if (!m_logLocation.EndsWith(Path.DirectorySeparatorChar.ToString()))
{
m_logLocation += Path.DirectorySeparatorChar;
}
if (!Directory.Exists(m_logLocation))
{
Directory.CreateDirectory(m_logLocation);
}
}
m_rootIndexDirectory = indexDirectory;
if (!m_rootIndexDirectory.EndsWith(Path.DirectorySeparatorChar.ToString()))
{
m_rootIndexDirectory += Path.DirectorySeparatorChar;
}
if (!Directory.Exists(m_rootIndexDirectory))
{
Directory.CreateDirectory(m_rootIndexDirectory);
}
if (tbxIateLocation != null)
{
Tbx sourceTbx = null;
var serializerTbx = new System.Xml.Serialization.XmlSerializer(typeof(Tbx));
using (var xmlReader = XmlReader.Create(tbxIateLocation))
{
sourceTbx = (Tbx)serializerTbx.Deserialize(xmlReader);
}
m_parserFactory = new LegalTextParserFactory(sourceTbx);
}
SetupWriters();
}
private void ParseDocumentXml(CoNLLDocument sourceDoc, LegalTextParserFactory parserFactory)
{
if (sourceDoc == null || sourceDoc.doc == null)
{
throw new InvalidOperationException("Unsupported document found in corpus!");
}
var parser = parserFactory.CreateParser(sourceDoc.doc.language);
var pass1 = parser.ParsePass1(sourceDoc);
parser.ParsePass2(pass1);
m_marcellDocument = pass1;
}
private void ParseDocumentXml(string fileName, LegalTextParserFactory parserFactory)
{
CoNLLDocument sourceDoc = null;
using (var fileReader = new SanitizedStreamReader(fileName))
{
using (var xmlReader = XmlReader.Create(fileReader))
{
sourceDoc = (CoNLLDocument)SerializerCoNLLUP.Deserialize(xmlReader);
}
}
if (sourceDoc == null || sourceDoc.doc == null)
{
throw new InvalidOperationException("Unsupported document found in corpus!");
}
//Preprocess the document to fix common problems in xml files
Preparse(sourceDoc, fileName);
ParseDocumentXml(sourceDoc, parserFactory);
}
