/// <summary>
/// Use a multifield query parser
/// </summary>
/// <param name="fields"></param>
/// <param name="stringQuery"></param>
/// <param name="defaultOp"></param>
/// <returns></returns>
public QueryBuilder MatchParsedInput(IEnumerable <string> fields, string stringQuery, QueryParser.Operator defaultOp = null)
{
if (string.IsNullOrWhiteSpace(stringQuery))
{
return(this);
}
var parser = new MultiFieldQueryParser(Version, fields.ToArray(), _analyzer);
if (defaultOp != null)
{
parser.SetDefaultOperator(defaultOp);
}
Query query;
try
{
query = parser.Parse(stringQuery);
}
catch (ParseException)
{
query = parser.Parse(QueryParser.Escape(stringQuery.ToLower()));
}
return(query != null?AddSubQuery(query) : this);
}
/// <summary>
/// Search in the index using the following filters.
/// </summary>
/// <param name="queryString">Words to search for (ALL)</param>
/// <param name="start">Start record to retrieve (zero based)</param>
/// <param name="max">Amount of records to receive from the start index number.</param>
/// <param name="flavorId">Flavor to filter (can be null)</param>
/// <param name="createdByMe">Filter created by me records (can be null)</param>
/// <param name="sortField">Sort Field</param>
/// <param name="isDescending">Mark if sorting in descending order</param>
/// <returns>A SearchEngineResponse object containing the total of records and found items</returns>
public SearchEngineResponse Search(SearchParameters parameters, int start, int max, string sortField, bool isDescending)
{
try
{
if (parameters.Seasons == null)
{
parameters.Seasons = "fall winter spring summer".Split(' ');
}
var list = new List <SearchEngineResult>();
Query bodyQuery = null;
if (!String.IsNullOrEmpty(parameters.FreeText))
{
QueryParser parser = new MultiFieldQueryParser(Lucene.Net.Util.Version.LUCENE_29, new string[] { Categories, Silouhettes, Colors, Fabrics, Tags, EventTypes }, _analyzer);
parser.SetDefaultOperator(QueryParser.Operator.AND);
if (!string.IsNullOrEmpty(parameters.FreeText.Trim()))
{
bodyQuery = parser.Parse(parameters.FreeText);
}
}
return(PerformQuery(list, bodyQuery, start, max, -1, -1, parameters, sortField, isDescending));
}
catch (Lucene.Net.Store.NoSuchDirectoryException)
{
return(new SearchEngineResponse {
TotalCount = 0, Results = new List <SearchEngineResult>()
});
}
}
public void Search(string indexDir, string q, int pageSize, int pageIndex, out int recCount)
{
indexDir = HttpContext.Current.Server.MapPath("~/Search/");
string keywords = q;
var search = new IndexSearcher(indexDir);
q = GetKeyWordsSplitBySpace(q, new PanGuTokenizer());
string[] fields = { "title", "Category", "Desc" };
QueryParser qp = new MultiFieldQueryParser(fields, new PanGuAnalyzer(true));
qp.SetDefaultOperator(Lucene.Net.QueryParsers.QueryParser.OR_OPERATOR);
//var queryParser = new QueryParser(Lucene.Net.Util.Version.LUCENE_29, "Desc", new PanGuAnalyzer(true));
Query query = qp.Parse(q);
Hits hits = search.Search(query);
// 新的查询
TopDocs newHits = search.Search(query, 100);
ScoreDoc[] scoreDocs = newHits.ScoreDocs;
for (int i = 0; i < scoreDocs.Length; i++)
{
Document document = search.Doc(scoreDocs[i].doc);
document.GetField("id").StringValue();
}
recCount = hits.Length();
int j = (pageIndex - 1) * pageSize;
for (int i = 0; i <= hits.Length() - 1; i++)
{
hits.Doc(i).GetField("Desc").StringValue();
}
search.Close();
}
private Result GetSearchResults(string searchQuery, string[] searchFields, string sortField, bool sortDescending, int count, int skip, ApprovalState approvalState = ApprovalState.Any)
{
var reader = GetIndexReader();
var searcher = GetIndexSearcher(reader);
string sortFieldName;
if (string.IsNullOrWhiteSpace(sortField))
{
sortField = sortFieldName = CreatedField;
sortDescending = true;
}
else if (sortField == CreatedField)
{
sortFieldName = CreatedField;
}
else
{
sortFieldName = FieldNameForSorting(sortField);
}
Query query;
if (string.IsNullOrWhiteSpace(searchQuery) == false && searchFields != null && searchFields.Any())
{
//searchQuery = searchQuery.Replace("*", "").Replace(" ", "* ") + "*";
if (searchFields.Count() > 1)
{
var parser = new MultiFieldQueryParser(Version.LUCENE_29, searchFields, GetAnalyzer());
parser.SetDefaultOperator(QueryParser.Operator.AND);
Log.Info("searchQuery =" + searchQuery, null);
try
{
query = parser.Parse(searchQuery.Trim());
}
catch (ParseException ex)
{
Log.Error(ex, ex.Message, null);
query = parser.Parse(QueryParser.Escape(searchQuery.Trim()));
}
}
else
{
query = new QueryParser(Version.LUCENE_29, searchFields.FirstOrDefault(), GetAnalyzer()).Parse(searchQuery);
}
}
else
{
query = approvalState == ApprovalState.Any
? new MatchAllDocsQuery()
: (Query) new TermQuery(new Term(ApprovalField, approvalState.ToString()));
}
var docs = searcher.Search(
query,
null, reader.MaxDoc(),
new Sort(new SortField(sortFieldName, SortField.STRING, sortDescending))
);
var scoreDocs = docs.ScoreDocs;
var rows = new List <Row>();
for (var i = skip; i < (skip + count) && i < scoreDocs.Length; i++)
{
if (reader.IsDeleted(scoreDocs[i].doc))
{
continue;
}
var doc = searcher.Doc(scoreDocs[i].doc);
var row = ParseRow(doc);
rows.Add(row);
}
searcher.Close();
reader.Close();
return(new Result(scoreDocs.Count(), rows, sortField, sortDescending));
}
public static SearchResultDataSet <Article> Search(ArticleQuery query)
{
//触发事件
GlobalEvents.UserSearch(query.Title);
//索引文件不存在时，返回null
if (!GlobalSettings.CheckFileExist(PhysicalIndexDirectory))
{
return(new SearchResultDataSet <Article>());
}
DateTime startTime = DateTime.Now;
BooleanQuery currentQuery = new BooleanQuery();
//CategoryID
if (query.CategoryID.HasValue && query.CategoryID.Value != 0)
{
Term categoryIDTearm = new Term(NewsIndexField.CategoryID, query.CategoryID.ToString());
Query categoryIDQuery = new TermQuery(categoryIDTearm);
currentQuery.Add(categoryIDQuery, BooleanClause.Occur.MUST);
}
//KeyWord
if (!string.IsNullOrEmpty(query.Title))
{
query.Title = SearchHelper.LuceneKeywordsScrubber(query.Title);
if (!string.IsNullOrEmpty(query.Title))
{
string[] searchFieldsForKeyword = new string[4];
searchFieldsForKeyword[0] = NewsIndexField.Title;
searchFieldsForKeyword[1] = NewsIndexField.SubTitle;
searchFieldsForKeyword[2] = NewsIndexField.Abstract;
searchFieldsForKeyword[3] = NewsIndexField.Keywords;
MultiFieldQueryParser articleWordQueryParser = new MultiFieldQueryParser(searchFieldsForKeyword, SearchHelper.GetChineseAnalyzer());
articleWordQueryParser.SetLowercaseExpandedTerms(true);
articleWordQueryParser.SetDefaultOperator(QueryParser.OR_OPERATOR);
string keyWordsSplit = SearchHelper.SplitKeywordsBySpace(query.Title);
Query articleWordQuery = articleWordQueryParser.Parse(keyWordsSplit);
currentQuery.Add(articleWordQuery, BooleanClause.Occur.MUST);
}
}
//Search
IndexSearcher searcher = new IndexSearcher(PhysicalIndexDirectory);
Hits hits = searcher.Search(currentQuery);
SearchResultDataSet <Article> articles = new SearchResultDataSet <Article>();
int pageLowerBound = query.PageIndex * query.PageSize;
int pageUpperBound = pageLowerBound + query.PageSize;
if (pageUpperBound > hits.Length())
{
pageUpperBound = hits.Length();
}
//HighLight
PanGu.HighLight.Highlighter highlighter = null;
if (!string.IsNullOrEmpty(query.Title))
{
highlighter = new PanGu.HighLight.Highlighter(new PanGu.HighLight.SimpleHTMLFormatter("<font color=\"#c60a00\">", "</font>"), new PanGu.Segment());
highlighter.FragmentSize = 100;
}
for (int i = pageLowerBound; i < pageUpperBound; i++)
{
Article item = ConvertDocumentToArticle(hits.Doc(i));
if (!string.IsNullOrEmpty(query.Title))
{
string bestBody = null;
if (!string.IsNullOrEmpty(item.Abstract) && item.Abstract.Length > MaxNumFragmentsRequired)
{
bestBody = highlighter.GetBestFragment(query.Title, item.Abstract);
}
if (!string.IsNullOrEmpty(bestBody))
{
item.Abstract = bestBody;
}
else
{
item.Abstract = HtmlHelper.TrimHtml(item.Abstract, 100);
}
string bestSubject = null;
if (!string.IsNullOrEmpty(item.Title) && item.Title.Length > MaxNumFragmentsRequired)
{
bestSubject = highlighter.GetBestFragment(query.Title, item.Title);
}
if (!string.IsNullOrEmpty(bestSubject))
{
item.Title = bestSubject;
}
}
articles.Records.Add(item);
}
searcher.Close();
articles.TotalRecords = hits.Length();
DateTime endTime = DateTime.Now;
articles.SearchDuration = (endTime.Ticks - startTime.Ticks) / 1E7f;
articles.PageIndex = query.PageIndex;
articles.PageSize = query.PageSize;
return(articles);
}
/// <summary>
/// Searches the specified phrase in the specified search fields.
/// </summary>
/// <param name="wiki">The wiki.</param>
/// <param name="searchFields">The search fields.</param>
/// <param name="phrase">The phrase to search.</param>
/// <param name="searchOption">The search options.</param>
/// <returns>A list of <see cref="SearchResult"/> items.</returns>
public static List <SearchResult> Search(string wiki, SearchField[] searchFields, string phrase, SearchOptions searchOption)
{
IIndexDirectoryProviderV40 indexDirectoryProvider = Collectors.CollectorsBox.GetIndexDirectoryProvider(wiki);
Analyzer analyzer = new SimpleAnalyzer();
IndexSearcher searcher = new IndexSearcher(indexDirectoryProvider.GetDirectory(), false);
string[] searchFieldsAsString = (from f in searchFields select f.AsString()).ToArray();
MultiFieldQueryParser queryParser = new MultiFieldQueryParser(Lucene.Net.Util.Version.LUCENE_29, searchFieldsAsString, analyzer);
if (searchOption == SearchOptions.AllWords)
{
queryParser.SetDefaultOperator(QueryParser.Operator.AND);
}
if (searchOption == SearchOptions.AtLeastOneWord)
{
queryParser.SetDefaultOperator(QueryParser.Operator.OR);
}
try {
Query query = queryParser.Parse(phrase);
TopDocs topDocs = searcher.Search(query, 100);
Highlighter highlighter = new Highlighter(new SimpleHTMLFormatter("<b class=\"searchkeyword\">", "</b>"), new QueryScorer(query));
List <SearchResult> searchResults = new List <SearchResult>(topDocs.totalHits);
for (int i = 0; i < Math.Min(100, topDocs.totalHits); i++)
{
Document doc = searcher.Doc(topDocs.scoreDocs[i].doc);
SearchResult result = new SearchResult();
result.DocumentType = DocumentTypeFromString(doc.GetField(SearchField.DocumentType.AsString()).StringValue());
result.Relevance = topDocs.scoreDocs[i].score * 100;
switch (result.DocumentType)
{
case DocumentType.Page:
PageDocument page = new PageDocument();
page.Wiki = doc.GetField(SearchField.Wiki.AsString()).StringValue();
page.PageFullName = doc.GetField(SearchField.PageFullName.AsString()).StringValue();
page.Title = doc.GetField(SearchField.Title.AsString()).StringValue();
TokenStream tokenStream1 = analyzer.TokenStream(SearchField.Title.AsString(), new StringReader(page.Title));
page.HighlightedTitle = highlighter.GetBestFragments(tokenStream1, page.Title, 3, " [...] ");
page.Content = doc.GetField(SearchField.Content.AsString()).StringValue();
tokenStream1 = analyzer.TokenStream(SearchField.Content.AsString(), new StringReader(page.Content));
page.HighlightedContent = highlighter.GetBestFragments(tokenStream1, page.Content, 3, " [...] ");
result.Document = page;
break;
case DocumentType.Message:
MessageDocument message = new MessageDocument();
message.Wiki = doc.GetField(SearchField.Wiki.AsString()).StringValue();
message.PageFullName = doc.GetField(SearchField.PageFullName.AsString()).StringValue();
message.DateTime = DateTime.Parse(doc.GetField(SearchField.MessageDateTime.AsString()).StringValue());
message.Subject = doc.GetField(SearchField.Title.AsString()).StringValue();
message.Body = doc.GetField(SearchField.Content.AsString()).StringValue();
TokenStream tokenStream2 = analyzer.TokenStream(SearchField.Content.AsString(), new StringReader(message.Body));
message.HighlightedBody = highlighter.GetBestFragments(tokenStream2, message.Body, 3, " [...] ");
result.Document = message;
break;
case DocumentType.Attachment:
PageAttachmentDocument attachment = new PageAttachmentDocument();
attachment.Wiki = doc.GetField(SearchField.Wiki.AsString()).StringValue();
attachment.PageFullName = doc.GetField(SearchField.PageFullName.AsString()).StringValue();
attachment.FileName = doc.GetField(SearchField.FileName.AsString()).StringValue();
attachment.FileContent = doc.GetField(SearchField.FileContent.AsString()).StringValue();
TokenStream tokenStream3 = analyzer.TokenStream(SearchField.Content.AsString(), new StringReader(attachment.FileContent));
attachment.HighlightedFileContent = highlighter.GetBestFragments(tokenStream3, attachment.FileContent, 3, " [...] ");
result.Document = attachment;
break;
case DocumentType.File:
FileDocument file = new FileDocument();
file.Wiki = doc.GetField(SearchField.Wiki.AsString()).StringValue();
file.FileName = doc.GetField(SearchField.FileName.AsString()).StringValue();
file.FileContent = doc.GetField(SearchField.FileContent.AsString()).StringValue();
TokenStream tokenStream4 = analyzer.TokenStream(SearchField.Content.AsString(), new StringReader(file.FileContent));
file.HighlightedFileContent = highlighter.GetBestFragments(tokenStream4, file.FileContent, 3, " [...]");
result.Document = file;
break;
}
searchResults.Add(result);
}
searcher.Close();
return(searchResults);
}
catch (ParseException) {
return(new List <SearchResult>(0));
}
}
} // constructor
/// <summary>
/// Searches the keyword index using the keywordQuery.
///
/// See http://www.dotlucene.net/documentation/QuerySyntax.html for the format of the keywordQuery.
///
/// This function will return a fully-filled array of IndexableFileInfo objects.
/// </summary>
/// <param name="keywordQuery"></param>
/// <param name="queryForHighlighter"></param>
/// <returns></returns>
public IndexableFileInfo[] doSearch(string keywordQuery, string queryForHighlighter)
{
IndexSearcher searcher;
IndexReader indexReader;
try
{
FSDirectory indexDir = FSDirectory.GetDirectory(this.luceneIndexDir, false);
indexReader = IndexReader.Open(indexDir);
searcher = new IndexSearcher(indexReader);
}
catch
{
// if the luceneIndexDir does not contain index files (yet), IndexSearcher
// throws a nice Exception.
return(new IndexableFileInfo[0]);
}
List <IndexableFileInfo> arrayList = new List <IndexableFileInfo>();
try
{
string Query = keywordQuery;
if (Query == String.Empty)
{
return(new IndexableFileInfo[0]);
}
string HighlighterQuery = queryForHighlighter;
// -- weirdly enough, when the query is empty, an exception is thrown during the QueryParser.Parse
// this hack gets around that.
if (HighlighterQuery == String.Empty)
{
HighlighterQuery = Guid.NewGuid().ToString();
}
// parse the query, "text" is the default field to search
// note: use the StandardAnalyzer! (the SimpleAnalyzer doesn't work correctly when searching by fields that are integers!)
// MultiFieldQueryParser queryParser = new MultiFieldQueryParser(new string[] { "title", "contents" }, new hatWebPortalAnalyzer());
MultiFieldQueryParser queryParser = new MultiFieldQueryParser(new string[] { "title", "contents" }, new SimpleAnalyzer());
queryParser.SetDefaultOperator(QueryParser.AND_OPERATOR);
Query query = queryParser.Parse(Query);
QueryParser highlightQueryParser = new QueryParser("contents", new hatWebPortalAnalyzer());
Query highlighterQuery = highlightQueryParser.Parse(HighlighterQuery);
query = searcher.Rewrite(query); // is this needed?? " Expert: called to re-write queries into primitive queries."
// search
Hits hits = searcher.Search(query, Sort.RELEVANCE);
// create highlighter
Highlighter highlighter = new Highlighter(new SimpleHTMLFormatter("<strong>", "</strong>"), new QueryScorer(highlighterQuery));
// -- go through hits and return results
for (int i = 0; i < hits.Length(); i++)
{
Document d = hits.Doc(i);
string filename = d.Get("filename");
string plainText = d.Get("contents");
string title = d.Get("title");
string sectionName = d.Get("SectionName");
string filenameParams = d.Get("filenameParams");
bool contentIsPageSummary = Convert.ToBoolean(d.Get("contentIsPageSummary"));
double score = Convert.ToDouble(hits.Score(i));
DateTime lastModified = DateTools.StringToDate(d.Get("LastModified"));
TokenStream tokenStream = new hatWebPortalAnalyzer().TokenStream("contents", new StringReader(plainText));
string fragment = plainText;
if (!contentIsPageSummary)
{
fragment = highlighter.GetBestFragments(tokenStream, plainText, 2, "...");
}
IndexableFileInfo newHit = new IndexableFileInfo(filename, filenameParams, title, fragment, sectionName, lastModified, contentIsPageSummary, score);
arrayList.Add(newHit);
} // for
}
finally
{
searcher.Close();
indexReader.Close();
}
return(arrayList.ToArray());
} // SearchActiveDocument
private SearchEngineResponse PerformQuery(ICollection <SearchEngineResult> list, Query queryOrig, int start, int max, int blogId, int entryId, SearchParameters parameters, string sortField, bool isDescending)
{
var query = new BooleanQuery();
if (queryOrig != null)
{
query.Add(queryOrig, BooleanClause.Occur.MUST);
}
if (parameters != null)
{
if (parameters.CreatedByMe.HasValue)
{
query.Add(new TermQuery(new Term(CreatedByMe, Convert.ToInt32(parameters.CreatedByMe).ToString())), BooleanClause.Occur.MUST);
}
if (parameters.FlavorId.HasValue)
{
query.Add(GetIntRangeQuery(FlavorId, parameters.FlavorId.Value), BooleanClause.Occur.MUST);
}
if (parameters.IsRatedByMe.HasValue && parameters.IsRatedByMe.Value)
{
query.Add(GetIntRangeQuery(MyRating, 1, 5), BooleanClause.Occur.MUST);
}
if (parameters.IsRatedByFriends.HasValue && parameters.IsRatedByFriends.Value)
{
query.Add(GetIntRangeQuery(FriendRating, 1, 5), BooleanClause.Occur.MUST);
}
if (parameters.MyRating.HasValue)
{
query.Add(GetIntRangeQuery(MyRating, parameters.MyRating.Value), BooleanClause.Occur.MUST);
}
if (parameters.FriendRating.HasValue)
{
query.Add(GetIntRangeQuery(FriendRating, parameters.FriendRating.Value), BooleanClause.Occur.MUST);
}
if (parameters.IsUpToDate)
{
query.Add(new TermQuery(new Term(IsUpToDate, Convert.ToInt32(parameters.IsUpToDate).ToString())), BooleanClause.Occur.MUST);
}
if (parameters.TrendKeywords != null && parameters.TrendKeywords.Length > 0)
{
// Create a query to filter by multiple fields one phrase is required to appear.
QueryParser parser = new MultiFieldQueryParser(Lucene.Net.Util.Version.LUCENE_29, new string[] { Categories, Silouhettes, Colors, Fabrics, Seasons, EventTypes }, _analyzer);
parser.SetDefaultOperator(QueryParser.Operator.OR);
Query keywordsQuery = parser.Parse(string.Join(" ", parameters.TrendKeywords));
query.Add(keywordsQuery, BooleanClause.Occur.MUST);
}
if (parameters.EventTypes != null && parameters.EventTypes.Length > 0)
{
QueryParser parser = new QueryParser(Lucene.Net.Util.Version.LUCENE_29, EventTypes, _analyzer);
parser.SetDefaultOperator(QueryParser.Operator.AND);
string joined = string.Join(" ", parameters.EventTypes.Select(x => "\"" + x + "\"").ToArray());
query.Add(parser.Parse(joined), BooleanClause.Occur.MUST);
}
if (parameters.Seasons != null && parameters.Seasons.Length > 0)
{
QueryParser parser = new QueryParser(Lucene.Net.Util.Version.LUCENE_29, Seasons, _analyzer);
parser.SetDefaultOperator(QueryParser.Operator.OR);
query.Add(parser.Parse(string.Join(" ", parameters.Seasons)), BooleanClause.Occur.MUST);
}
//// Query to filter only by the colors field (maybe is faster or useful)
if (!String.IsNullOrEmpty(parameters.Color))
{
query.Add(new TermQuery(new Term(Colors, parameters.Color)), BooleanClause.Occur.MUST);
}
}
Hits hits;
if (sortField != null)
{
var sortTerm = new SortField(sortField, SortField.INT, isDescending);
var sort = new Sort(sortTerm);
hits = Searcher.Search(query, null, sort);
}
else
{
hits = Searcher.Search(query);
}
int length = hits.Length();
int resultsAdded = 0;
if (start < length)
{
for (int i = start; i < length && resultsAdded < max; i++)
{
SearchEngineResult result = CreateSearchResult(hits.Doc(i), 0);
list.Add(result);
resultsAdded++;
}
}
return(new SearchEngineResponse {
TotalCount = length, Results = list
});
}
