public static void IndexTopics(CSETWebEntities entity, IndexWriter writer)
{
foreach (CATALOG_RECOMMENDATIONS_DATA data in entity.CATALOG_RECOMMENDATIONS_DATA)
{
Lucene.Net.Documents.Document lucDoc = new Lucene.Net.Documents.Document();
string text = "";
text += " " + data.Heading + " " + data.Requirement + " " + data.Supplemental_Guidance + " " + data.Enhancement;
lucDoc.Add(new Field(FieldNames.SHORT_NAME, data.Topic_Name, Field.Store.YES, Field.Index.ANALYZED));
lucDoc.Add(new Field(FieldNames.TEXT, text, Field.Store.YES, Field.Index.ANALYZED));
lucDoc.Add(new Field(FieldNames.RESOURCE_TYPE, ResourceTypeEnum.Catalog_Recommendation.ToString(), Field.Store.YES, Field.Index.NOT_ANALYZED));
lucDoc.Add(new Field(FieldNames.DOC_ID, data.Data_Id.ToString(), Field.Store.YES, Field.Index.NO));
writer.AddDocument(lucDoc);
}
foreach (PROCUREMENT_LANGUAGE_DATA data in entity.PROCUREMENT_LANGUAGE_DATA)
{
Lucene.Net.Documents.Document lucDoc = new Lucene.Net.Documents.Document();
string text = "";
text += " " + data.Basis + " " + data.Language_Guidance + " " + data.Procurement_Language + " " + data.Fatmeasures + " " + data.Satmeasures + " " + data.Maintenance_Guidance;
lucDoc.Add(new Field(FieldNames.SHORT_NAME, data.Topic_Name, Field.Store.YES, Field.Index.ANALYZED));
lucDoc.Add(new Field(FieldNames.TEXT, text, Field.Store.YES, Field.Index.ANALYZED));
lucDoc.Add(new Field(FieldNames.RESOURCE_TYPE, ResourceTypeEnum.Procurement_Language.ToString(), Field.Store.YES, Field.Index.NOT_ANALYZED)); lucDoc.Add(new Field(FieldNames.DOC_ID, data.Procurement_Id.ToString(), Field.Store.YES, Field.Index.NO));
writer.AddDocument(lucDoc);
}
}
public void Dispose()
{
if (_context != null)
{
_context.Dispose();
_context = null;
}
}
public void IndexDocs()
{
string solutionPath = System.IO.Directory.GetParent(AppDomain.CurrentDomain.BaseDirectory).Parent.Parent.Parent.Parent.FullName;
string documentDirectory = solutionPath + @"\CSETWeb_Api\CSETWeb_Api\Documents";
string fullDocumentDirectory = Path.GetFullPath(documentDirectory);
string luceneIndexDir = solutionPath + @"\CSETWeb_Api\CSETWeb_Api\LuceneIndex2";
string luceneIndexDestDir = solutionPath + @"\CSETWeb_Api\CSETWeb_Api\LuceneIndex";
CSETWebEntities entity = new CSETWebEntities();
Dictionary <int, GEN_FILE> dictionaryGenFiles = new Dictionary <int, GEN_FILE>();
List <REF_LIBRARY_PATH> listLibDocs = entity.REF_LIBRARY_PATH.ToList();
int count = 0;
Debug.WriteLine("Number of Resource Nodes: " + listLibDocs.Count);
foreach (REF_LIBRARY_PATH resnodes in listLibDocs)
{
foreach (GEN_FILE file in resnodes.GEN_FILE)
{
if (!dictionaryGenFiles.ContainsKey(file.Gen_File_Id))
{
dictionaryGenFiles.Add(file.Gen_File_Id, file);
}
}
}
if (System.IO.Directory.Exists(luceneIndexDir))
{
System.IO.Directory.Delete(luceneIndexDir, true);
}
DirectoryInfo di = System.IO.Directory.CreateDirectory(luceneIndexDir);
DirectoryInfo dir = new DirectoryInfo(luceneIndexDir);
FSDirectory fsdir = FSDirectory.Open(dir);
using (IndexWriter writer = new IndexWriter(fsdir, new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_30), true, IndexWriter.MaxFieldLength.UNLIMITED))
{
Debug.WriteLine("Documents Count: " + dictionaryGenFiles.Values.Count);
foreach (GEN_FILE doc in dictionaryGenFiles.Values)
{
//if (count > 30)
// break;
string filepath = fullDocumentDirectory + @"\" + doc.File_Name;
bool exists = File.Exists(filepath);
if (exists)
{
if (Path.GetExtension(filepath).ToLower() == ".pdf")
{
Document lucDoc = PDFParser.ParseDocument(filepath, doc);
writer.AddDocument(lucDoc);
Debug.WriteLine("Count: " + count + " Processed file: " + filepath);
count++;
}
else if (Path.GetExtension(filepath).ToLower() == ".docx")
{
Document lucDoc = WordParser.ParseDocument(filepath, doc);
writer.AddDocument(lucDoc);
Debug.WriteLine("Count: " + count + "Processed file: " + filepath);
count++;
}
else
{
Debug.Assert(false, "Can't read file because bad extension. Extension:" + filepath);
}
}
else
{
Debug.WriteLine("File doesn't exist:" + filepath);
}
}
TopicIndexer.IndexTopics(entity, writer);
writer.Optimize();
}
System.IO.Directory.Delete(luceneIndexDestDir, true);
System.IO.Directory.Move(luceneIndexDir, luceneIndexDestDir);
}
public FileRepository()
{
_context = new CSETWebEntities();
}