public static byte[] Serialize(this DocumentTableRow document, Compression compression)
{
using (var stream = new MemoryStream())
{
document.Fields.Serialize(compression, stream);
return(stream.ToArray());
}
}
public IList <AnalyzedTerm> AnalyzeDocument(DocumentTableRow document)
{
var analyzedTerms = new List <AnalyzedTerm>();
foreach (var field in document.Fields.Values)
{
if (field.Analyze && field.Index)
{
var words = Analyze(field.Value);
var wordMatrix = new Dictionary <string, IList <int> >();
for (var index = 0; index < words.Count; index++)
{
var word = words[index];
IList <int> positions;
if (wordMatrix.TryGetValue(word, out positions))
{
positions.Add(index);
}
else
{
wordMatrix.Add(word, new List <int> {
index
});
}
//Log.DebugFormat("found term {0} at pos {1}", word, index);
}
foreach (var wordInfo in wordMatrix)
{
var postings = new List <int>(wordInfo.Value.Count);
foreach (var position in wordInfo.Value)
{
postings.Add(position);
}
analyzedTerms.Add(
new AnalyzedTerm(document.TableId, field.Key, wordInfo.Key, postings));
}
}
else if (field.Index)
{
var postings = new int[] { 0 };
analyzedTerms.Add(
new AnalyzedTerm(document.TableId, field.Key, field.Value, postings));
}
}
return(analyzedTerms);
}
