/// <summary>
/// 得到结果
/// </summary>
/// <param name="searchQuery"></param>
/// <returns></returns>
protected virtual SearchResultInfo GetSearchResultByFind(SearchQueryInfo searchQuery)
{
var result = new SearchResultInfo {
Terms = Analyzer.Resolve(searchQuery.Key)
};
if (result.Terms == null)
{
return(result);
}
if (!StoreIndexs.ContainsKey(searchQuery.Name))
{
return(result);
}
var storeIndex = StoreIndexs[searchQuery.Name];
if (storeIndex == null)
{
return(result);
}
result.Feilds = storeIndex.StoreFields.Select(it => it.Name).ToList();
result.Words = new List <WordInfo>();
foreach (var term in result.Terms)
{
if (result.Words.Count(it => it.Name == term.Name) > 0)
{
continue;
}
var word = Worder.GetInfo(storeIndex, term.Name);
if (word != null)
{
result.Words.Add(word);
}
}
var documentIds = GetSearchDocumentIds(storeIndex, result, searchQuery);
if (documentIds != null)
{
result.DocumentCount = documentIds.Count;
AddSearchDocuments(storeIndex, searchQuery, result, documentIds);
}
if (result.Words.Count > 0)
{
result.DataCount = result.Words.Sum(it => it.DocumentCount) / result.Words.Count;
}
return(result);
}
/// <summary>
/// 得到文档向量
/// </summary>
/// <param name="result"></param>
/// <returns></returns>
protected virtual IDictionary <long, IList <double> > GetSeachDocumentVector(SearchResultInfo result)
{
var scDocs = new Dictionary <long, IList <double> >();
var i = 0;
foreach (var word in result.Words)
{
foreach (var inversion in word.Inversions)
{
if (!scDocs.ContainsKey(inversion.DocumentId))
{
var values = new double[result.Words.Count];
scDocs.Add(inversion.DocumentId, values);
}
scDocs[inversion.DocumentId][i] = inversion.Weight;
}
i++;
}
return(scDocs);
}
/// <summary>
/// 添加行
/// </summary>
/// <param name="storeIndex"></param>
/// <param name="searchQuery"></param>
/// <param name="result"></param>
/// <param name="documentIds"></param>
protected virtual void AddSearchDocuments(StoreIndexInfo storeIndex, SearchQueryInfo searchQuery, SearchResultInfo result, IList <long> documentIds)
{
if (searchQuery.PageSize > 0)
{
documentIds = documentIds.Skip(searchQuery.PageIndex * searchQuery.PageSize).Take(searchQuery.PageSize).ToList();
}
foreach (var documentId in documentIds)
{
var document = Documentor.GetInfo(storeIndex, documentId);
if (document != null)
{
result.Documents.Add(document);
}
}
}
/// <summary>
/// 查询得到结果
/// </summary>
/// <param name="storeIndex"></param>
/// <param name="result"></param>
/// <param name="searchQuery"></param>
/// <returns></returns>
protected virtual IList <long> GetSearchDocumentIds(StoreIndexInfo storeIndex, SearchResultInfo result, SearchQueryInfo searchQuery)
{
if (result.Words == null || result.Words.Count == 0)
{
return(null);
}
result.Documents = new List <DocumentInfo>();
if (result.Words.Count == 1)
{
return(result.Words[0].Inversions.Select(it => it.DocumentId).ToList());
}
var scQuery = new List <double>();
foreach (var word in result.Words)
{
var weight = word.DocumentCount == 0
? 0
: ((double)result.Terms.Count(it => it.Name == word.Name) / result.Terms.Count *
Math.Log(storeIndex.StoreDocument.DataCount / (word.DocumentCount / (double)storeIndex.StoreDocument.DataCount)));
scQuery.Add(weight);
}
var scDocuments = GetSeachDocumentVector(result);
return(GetOrderbySearchDocuments(scQuery, scDocuments));
}