public void TestIndexDatabaseSerialization()
{
IndexRecord indexRecord1 = new IndexRecord(@"C:\DummyFile.txt",new List<int>() { 1,2,3,4,5 });
IndexRecord indexRecord2 = new IndexRecord(@"C:\DummyFile1.txt", new List<int>() { 1, 2, 3 });
List<IndexRecord> indexRecords = new List<IndexRecord>() { indexRecord1, indexRecord2 };
Dictionary<string,List<IndexRecord>> searchPatternToIndexRecords = new Dictionary<string,List<IndexRecord>>();
searchPatternToIndexRecords.Add("Test",indexRecords);
IndexDatabase indexDatabase = new IndexDatabase(searchPatternToIndexRecords,DateTime.Now);
//serialize the index database
IndexDatabaseWriter writer = new IndexDatabaseWriter();
writer.Persist(indexDatabase);
}
private void IndexInternal(List<string> files)
{
files.Sort(StringComparer.OrdinalIgnoreCase);
try
{
string fileText = string.Empty;
for (int i = 0; i < files.Count; i++)
{
//skip the file used for logging
if (0 == string.Compare(files[i], LogWriter.LogFileName, StringComparison.OrdinalIgnoreCase))
continue;
if (!Utilities.IsTextFile(files[i]))
continue;
//read the text in the text file
fileText = Utilities.ReadTextFromFiles(files[i]);
//we are not going to filter out any words as of the current release
//this may result in large search times, but we are OK with it for now.
List<string> wordsInFile = Utilities.ExtractWordsInText(fileText, null);
//send the words in the file to the file search engine as a pattern.
//since we need to identify all the possible positions of the current word
//we directly invoke the KMPEngine functionality on the current file.
foreach (string word in wordsInFile)
{
if (0 == string.Compare(" ", word) || 0 == string.Compare("\r", word)
|| 0 == string.Compare("\n", word) || 0 == string.Compare("\t", word)
|| 0 == string.Compare(Environment.NewLine, word))
continue;
KMPEngine kmpEngine = new KMPEngine(fileText, word);
List<int> occurences = kmpEngine.Find(true, true);
//since the word is already present in the file, we make an entry
//we are sure that there is ATLEAST ONE occurence.We go ahead and create
//index record. we add the record only if the number of occurences is
//greater than zero
if (occurences.Count == 0)
continue;
IndexRecord indexRecord = new IndexRecord(files[i], occurences);
if (!_indexDatabaseEntries.ContainsKey(word))
{
List<IndexRecord> indexRecords = new List<IndexRecord>();
indexRecords.Add(indexRecord);
_indexDatabaseEntries.Add(word, indexRecords);
LogWriter.WriteToLogFile(string.Format("Created first Index record for file {0} containing word {1}", files[i], word));
}
else
{
_indexDatabaseEntries[word].Add(indexRecord);
LogWriter.WriteToLogFile(string.Format("Appended Index record for file {0} containing word {1}", files[i], word));
}
}
}
LogWriter.WriteToLogFile(string.Format("Created index database with {0} index entries", _indexDatabaseEntries.Count));
}
catch (Exception ex)
{
LogWriter.WriteToLogFile(string.Format("Exception ex:{0}", ex.ToString()));
}
finally
{
LogWriter.Dispose();
}
}
