private Searcher()
{
this.MongoDbServer = Constants.DefaultServerAddress;
this.DbName = Constants.DefaultDbName;
_segmentor = new ChineseSegmentor();
}
/// <summary>
/// 建索引主體
/// </summary>
/// <returns></returns>
public bool MakeIndex()
{
try
{
// 1. 解文
ParseSourceDoc();
// 2. 建索引
ChineseSegmentor segmentor = new ChineseSegmentor();
var server = MongoDbLib.GetServerConnection(_indexer.MongoDbServer);
var database = server.GetDatabase(_indexer.DbName);
var tblSourceText = database.GetCollection <SourceText>(Constants.TblSourceText);
// 斷詞，處理每個 Token
var sourcees = from s in tblSourceText.AsQueryable <SourceText>()
orderby s.DocId, s.ParaId
select s;
Dictionary <String, InvertedIndex> fullIndexes = new Dictionary <String, InvertedIndex>();
InvertedIndex aIndex = null;
foreach (var aSourceText in sourcees)
{
List <Pair <String, Int32> > result = segmentor.SegWords(aSourceText.Para);
foreach (var aToken in result)
{
if (fullIndexes.ContainsKey(aToken.First))
{
aIndex = fullIndexes[aToken.First];
}
else
{
aIndex = new InvertedIndex();
aIndex.Word = aToken.First;
}
aIndex.Indexes.Add(new IndexElement()
{
DocId = aSourceText.DocId,
ParaId = aSourceText.ParaId,
Offset = aToken.Second
});
fullIndexes[aToken.First] = aIndex;
}
}
// 在 Storage 存入 Word List
var wordListCollection = database.GetCollection(Constants.TblWordList);
List <BsonDocument> batch = new List <BsonDocument>();
List <String> wordList = fullIndexes.Keys.ToList();
for (int wordId = 0; wordId < fullIndexes.Count; wordId++)
{
aIndex = fullIndexes[wordList[wordId]];
aIndex.WordId = wordId;
batch.Add(new BsonDocument()
{
{ "Word", wordList[wordId] },
{ "WordId", wordId }
});
}
wordListCollection.InsertBatch(batch);
// 儲存全文索引
var tblFullText = database.GetCollection(Constants.TblFullText);
List <InvertedIndex> fullText = new List <InvertedIndex>();
tblFullText.InsertBatch <InvertedIndex>(fullIndexes.Values.ToList());
return(true);
}
catch (Exception e)
{
Console.WriteLine(e.StackTrace);
return(false);
}
}
