public static string GetText(string filename, string extension)
{
try
{
IDocumentHandler docHandler = null;
if (extension.ToLower().StartsWith("htm") || extension.ToLower().StartsWith(".htm"))
{
docHandler = new NTidyHTMLHandler();
}
else if (extension.ToLower().StartsWith("txt") || extension.ToLower().StartsWith(".txt"))
{
docHandler = new TextHandler();
}
else if (extension.ToLower().StartsWith("pdf") || extension.ToLower().StartsWith(".pdf"))
{
docHandler = new PdfHandler();
}
if (docHandler == null)
{
return("");
}
else
{
return(docHandler.GetText(filename));
}
}
catch (Exception ex)
{
GXLogging.Error(log, "GetText error", ex);
return("");
}
}
public static string HtmlPreview(Object obj, string query, string textType, string preTag, string postTag, int fragmentSize, int maxNumFragments)
{
string text;
GxSilentTrnSdt silent = obj as GxSilentTrnSdt;
GxFile file = obj as GxFile;
if (silent != null)
{
text = (silent).Transaction.ToString();
}
else if (file != null)
{
text = DocumentHandler.GetText(file.GetAbsoluteName(), System.IO.Path.GetExtension(file.GetAbsoluteName()));
}
else if (textType.ToLower().StartsWith("htm"))
{
text = new NTidyHTMLHandler().GetTextFromString(obj.ToString());
}
else
{
text = obj.ToString();
}
if (!string.IsNullOrEmpty(query) && !string.IsNullOrEmpty(text))
{
if (qp == null)
{
qp = new QueryParser(Lucene.Net.Util.Version.LUCENE_24, IndexRecord.CONTENTFIELD, Indexer.CreateAnalyzer());
qp.DefaultOperator = QueryParser.Operator.AND;
qp.MultiTermRewriteMethod = MultiTermQuery.SCORING_BOOLEAN_QUERY_REWRITE;
}
Query unReWrittenQuery = qp.Parse(query);
Query q = unReWrittenQuery;
try
{
if (reader == null)
{
reader = Indexer.Reader;
}
if (!queries.TryGetValue(query, out q))
{
q = unReWrittenQuery.Rewrite(reader);//required to expand search terms (for the usage of highlighting with wildcards)
if (queries.Count == int.MaxValue)
{
queries.Clear();
}
queries[query] = q;
}
}
catch (Exception ex)
{
GXLogging.Error(log, "HTMLPreview error", ex);
}
QueryScorer scorer = new QueryScorer(q);
SimpleHTMLFormatter formatter = new SimpleHTMLFormatter(preTag, postTag);
Highlighter highlighter = new Highlighter(formatter, scorer);
IFragmenter fragmenter = new SimpleFragmenter(fragmentSize);
highlighter.TextFragmenter = fragmenter;
TokenStream tokenStream = new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_24).TokenStream("Content", new StringReader(text));
String result = highlighter.GetBestFragments(tokenStream, text, maxNumFragments, "...");
return(result);
}
else
{
return(text);
}
}