private void FillList(TermIndexRecord rec, ref int entryCounter, Color backColor)
{
for (int i = 0; i < rec.DocsNumber; i++, entryCounter++)
{
Entry e = rec.GetEntryAt(i);
ListViewItem item = new ListViewItem();
item.BackColor = backColor;
item.Text = entryCounter.ToString();
item.SubItems.Add(e.DocIndex.ToString());
item.SubItems.Add(e.Offsets.Length.ToString());
if (hash.ContainsKey(e.DocIndex))
{
ListViewItem refItem = listEntries.Items[hash[e.DocIndex] - 1];
item.ForeColor = refItem.ForeColor = Color.Red;
ListViewItem.ListViewSubItem prev2 = refItem.SubItems[3];
prev2.Text = prev2.Text + ">>" + entryCounter.ToString();
item.SubItems.Add("<<" + hash[e.DocIndex].ToString());
}
else
{
item.SubItems.Add("");
}
hash[e.DocIndex] = entryCounter;
if (!Core.TextIndexManager.IsDocumentInIndex(e.DocIndex))
{
item.Font = delFont;
}
listEntries.Items.Add(item);
}
}
public static void Main(string[] Arguments)
{
TermIndexAccessor TermIndex = new TermIndexAccessor(Arguments[0]);
StreamWriter writer = new StreamWriter("lexdump.dat", false, System.Text.Encoding.Default);
int TotalDocEntries = 0, TotalInstances = 0;
Console.WriteLine("loading...");
TermIndex.Load();
Console.WriteLine("dumping...");
foreach (KeyPair pair in TermIndex.Keys)
{
TermIndexRecord record = TermIndex.GetRecordByHandle(pair._offset);
int InstancesCount = 0;
TotalDocEntries += record.DocsNumber;
for (int j = 0; j < record.DocsNumber; j++)
{
InstancesCount += record.GetEntryAt(j).Count;
TotalInstances += record.GetEntryAt(j).Count;
}
/*
* if( Arguments.Length == 1 || Arguments[ 1 ] == "bydoc" )
* writer.WriteLine( "{0,6} {1,8} {2}", record.DocsNumber, InstancesCount, record.Term );
* else
* writer.WriteLine( "{1,8} {0,6} {2}", record.DocsNumber, InstancesCount, record.Term );
*/
}
writer.WriteLine("--- 1 Terms number: " + TermIndex.TermsNumber);
writer.WriteLine("--- 2 Words number: " + TotalInstances);
writer.WriteLine("--- 3 Entries number: " + TotalDocEntries);
TermIndex.Close();
writer.Close();
Console.WriteLine("sorting...");
Process process = new Process();
process.StartInfo.FileName = "sort.exe";
process.StartInfo.Arguments = " /R /L C lexdump.dat /O lexdump.srt";
process.StartInfo.WorkingDirectory = ".";
process.StartInfo.CreateNoWindow = true;
process.StartInfo.UseShellExecute = false;
process.Start();
process.WaitForExit();
Console.WriteLine("done...");
}
protected void CrossIndexChecks(TermIndexAccessor termIndex)
{
foreach (KeyPair pair in termIndex.Keys)
{
TermIndexRecord termRecord = termIndex.GetRecordByHandle(pair._offset);
for (int j = 0; j < termRecord.DocsNumber; j++)
{
Entry entry = termRecord.GetEntryAt(j);
if (entry.DocIndex < -1)
{
throw new FormatException("DocIndex is negative in the TermIndex record entry");
}
if (entry.Count <= 0)
{
throw new FormatException("Number of term instances is negative in the TermIndex record");
}
}
}
}
private void buttonShowContent_Click(object sender, System.EventArgs e)
{
int termHC;
int entryCounter = 1;
listEntries.Items.Clear();
hash.Clear();
try
{
termHC = Int32.Parse(textTermID.Text);
}
catch (Exception)
{
termHC = Word.GetTermId(textTermID.Text.ToLower());
LexemeConstructor ctor = new LexemeConstructor(OMEnv.ScriptMorphoAnalyzer,
OMEnv.DictionaryServer);
string normForm = ctor.GetNormalizedToken(textTermID.Text);
labelNormForm.Text = "Normalized Form: " + normForm;
termHC = Word.GetTermId(normForm.ToLower());
}
if (termHC != -1)
{
labelHC.Text = termHC.ToString();
recMain = (TermIndexRecord)FullTextIndexer.Instance.GetTermRecordMain(termHC);
recMem = (TermIndexRecord)FullTextIndexer.Instance.GetTermRecordMem(termHC);
if (recMain != null)
{
FillList(recMain, ref entryCounter, Color.LightSkyBlue);
}
if (recMem != null)
{
FillList(recMem, ref entryCounter, Color.LightYellow);
}
}
else
{
MessageBox.Show("No such term in the index");
}
}
