private static Query ParseQuery(SearchFilter searchFilter)
{
if (String.IsNullOrWhiteSpace(searchFilter.SearchTerm))
{
return(new MatchAllDocsQuery());
}
var fields = new[] { "Id", "Title", "Tags", "Description", "Author" };
var analyzer = new StandardAnalyzer(LuceneCommon.LuceneVersion);
var queryParser = new MultiFieldQueryParser(LuceneCommon.LuceneVersion, fields, analyzer);
// All terms in the multi-term query appear in at least one of the fields.
var conjuctionQuery = new BooleanQuery();
conjuctionQuery.SetBoost(2.0f);
// Some terms in the multi-term query appear in at least one of the fields.
var disjunctionQuery = new BooleanQuery();
disjunctionQuery.SetBoost(0.1f);
// Suffix wildcard search e.g. jquer*
var wildCardQuery = new BooleanQuery();
wildCardQuery.SetBoost(0.5f);
// Escape the entire term we use for exact searches.
var escapedSearchTerm = Escape(searchFilter.SearchTerm);
var exactIdQuery = new TermQuery(new Term("Id-Exact", escapedSearchTerm));
exactIdQuery.SetBoost(2.5f);
var wildCardIdQuery = new WildcardQuery(new Term("Id-Exact", "*" + escapedSearchTerm + "*"));
foreach (var term in GetSearchTerms(searchFilter.SearchTerm))
{
var termQuery = queryParser.Parse(term);
conjuctionQuery.Add(termQuery, BooleanClause.Occur.MUST);
disjunctionQuery.Add(termQuery, BooleanClause.Occur.SHOULD);
foreach (var field in fields)
{
var wildCardTermQuery = new WildcardQuery(new Term(field, term + "*"));
wildCardTermQuery.SetBoost(0.7f);
wildCardQuery.Add(wildCardTermQuery, BooleanClause.Occur.SHOULD);
}
}
// Create an OR of all the queries that we have
var combinedQuery =
conjuctionQuery.Combine(new Query[] { exactIdQuery, wildCardIdQuery, conjuctionQuery, disjunctionQuery, wildCardQuery });
if (searchFilter.SortProperty == SortProperty.Relevance)
{
// If searching by relevance, boost scores by download count.
var downloadCountBooster = new FieldScoreQuery("DownloadCount", FieldScoreQuery.Type.INT);
return(new CustomScoreQuery(combinedQuery, downloadCountBooster));
}
return(combinedQuery);
}
private SearchResults SearchCore(SearchFilter searchFilter)
{
// Get index timestamp
DateTime timestamp = File.GetLastWriteTimeUtc(LuceneCommon.GetIndexMetadataPath());
int numRecords = searchFilter.Skip + searchFilter.Take;
var searcher = new IndexSearcher(_directory, readOnly: true);
var query = ParseQuery(searchFilter.SearchTerm);
// IF searching by relevance, boost scores by download count.
if (searchFilter.SortOrder == SortOrder.Relevance)
{
var downloadCountBooster = new FieldScoreQuery("DownloadCount", FieldScoreQuery.Type.INT);
query = new CustomScoreQuery(query, downloadCountBooster);
}
string filterTerm;
if (SemVerLevelKey.ForSemVerLevel(searchFilter.SemVerLevel) == SemVerLevelKey.SemVer2)
{
filterTerm = searchFilter.IncludePrerelease ? "IsLatestSemVer2" : "IsLatestStableSemVer2";
}
else
{
filterTerm = searchFilter.IncludePrerelease ? "IsLatest" : "IsLatestStable";
}
Query filterQuery = new TermQuery(new Term(filterTerm, Boolean.TrueString));
if (searchFilter.CuratedFeed != null)
{
var feedFilterQuery = new TermQuery(new Term("CuratedFeedKey", searchFilter.CuratedFeed.Key.ToString(CultureInfo.InvariantCulture)));
BooleanQuery conjunctionQuery = new BooleanQuery();
conjunctionQuery.Add(filterQuery, Occur.MUST);
conjunctionQuery.Add(feedFilterQuery, Occur.MUST);
filterQuery = conjunctionQuery;
}
Filter filter = new QueryWrapperFilter(filterQuery);
var results = searcher.Search(query, filter: filter, n: numRecords, sort: new Sort(GetSortField(searchFilter)));
if (results.TotalHits == 0 || searchFilter.CountOnly)
{
return(new SearchResults(results.TotalHits, timestamp));
}
var packages = results.ScoreDocs
.Skip(searchFilter.Skip)
.Select(sd => PackageFromDoc(searcher.Doc(sd.Doc)))
.ToList();
return(new SearchResults(
results.TotalHits,
timestamp,
packages.AsQueryable()));
}
/// <summary>
/// Perform search
/// </summary>
/// <param name="query"></param>
/// <param name="startIndex"></param>
/// <param name="blockSize"></param>
/// <param name="indexDirEs"></param>
/// <param name="indexDirEn"></param>
/// <param name="sortBy"></param>
/// <returns></returns>
public List <IssueDocument> MedesSearch(Query query, int startIndex, int blockSize, Directory indexDirEs, Directory indexDirEn, Directory indexDirHe, string sortBy)
{
#if DEBUG
T.TraceMessage(string.Format("Begin search , query: '{0}'", query.ToString()));
#endif
List <IssueDocument> result = new List <IssueDocument>();
try
{
// build a multi searcher across the 2 indexes
MultiSearcher mSearcher = CombineSearchers(indexDirEs, indexDirEn, indexDirHe);
TopDocs tDocs = null;
int iterateLast = startIndex + blockSize;
string customScoreField = "article_id";
FieldScoreQuery dateBooster = new FieldScoreQuery(customScoreField, FieldScoreQuery.Type.FLOAT);
CustomScoreQuery customQuery = new CustomScoreQuery(query, dateBooster);
tDocs = mSearcher.Search(customQuery, 1000);
//ScoreDoc[] hits = tpDcs.scoreDocs;
if (startIndex + blockSize > tDocs.TotalHits)
{
iterateLast = tDocs.TotalHits;
}
for (int i = startIndex; i < iterateLast; i++)
{
// Document hitDoc = mSearcher.Doc(hits[i].doc);
Document hitDoc = mSearcher.Doc(i);
result.Add(new IssueDocument()
{
Id = Int32.Parse(hitDoc.Get("issue_id").ToString())
});
}
// close the searcher and indexes
mSearcher.Dispose();
indexDirEs.Dispose();
indexDirEn.Dispose();
indexDirHe.Dispose();
}
catch (Exception ex)
{
T.TraceError("Error MedesSearch, query '{0}'", query.ToString());
T.TraceError(ex);
throw ex;
}
return(result);
}
private IQueryable <Package> SearchCore(SearchFilter searchFilter, out int totalHits)
{
int numRecords = searchFilter.Skip + searchFilter.Take;
var searcher = new IndexSearcher(_directory, readOnly: true);
var query = ParseQuery(searchFilter);
// IF searching by relevance, boost scores by download count.
if (searchFilter.SortProperty == SortProperty.Relevance)
{
var downloadCountBooster = new FieldScoreQuery("DownloadCount", FieldScoreQuery.Type.INT);
query = new CustomScoreQuery(query, downloadCountBooster);
}
var filterTerm = searchFilter.IncludePrerelease ? "IsLatest" : "IsLatestStable";
Query filterQuery = new TermQuery(new Term(filterTerm, Boolean.TrueString));
if (searchFilter.CuratedFeedKey.HasValue)
{
var feedFilterQuery = new TermQuery(new Term("CuratedFeedKey", searchFilter.CuratedFeedKey.Value.ToString(CultureInfo.InvariantCulture)));
BooleanQuery conjunctionQuery = new BooleanQuery();
conjunctionQuery.Add(filterQuery, Occur.MUST);
conjunctionQuery.Add(feedFilterQuery, Occur.MUST);
filterQuery = conjunctionQuery;
}
Filter filter = new QueryWrapperFilter(filterQuery);
var results = searcher.Search(query, filter: filter, n: numRecords, sort: new Sort(GetSortField(searchFilter)));
totalHits = results.TotalHits;
if (results.TotalHits == 0 || searchFilter.CountOnly)
{
return(Enumerable.Empty <Package>().AsQueryable());
}
var packages = results.ScoreDocs
.Skip(searchFilter.Skip)
.Select(sd => PackageFromDoc(searcher.Doc(sd.Doc)))
.ToList();
return(packages.AsQueryable());
}
private SearchResults SearchCore(SearchFilter searchFilter)
{
// Get index timestamp
DateTime timestamp = File.GetLastWriteTimeUtc(LuceneCommon.IndexMetadataPath);
int numRecords = searchFilter.Skip + searchFilter.Take;
var searcher = new IndexSearcher(_directory, readOnly: true);
var query = ParseQuery(searchFilter);
// If searching by relevance, boost scores by download count.
if (searchFilter.SortProperty == SortProperty.Relevance)
{
var downloadCountBooster = new FieldScoreQuery("DownloadCount", FieldScoreQuery.Type.INT);
query = new CustomScoreQuery(query, downloadCountBooster);
}
var filterTerm = searchFilter.IncludePrerelease ? "IsLatest" : "IsLatestStable";
Query filterQuery = new TermQuery(new Term(filterTerm, Boolean.TrueString));
Filter filter = new QueryWrapperFilter(filterQuery);
var results = searcher.Search(query, filter: filter, n: numRecords, sort: new Sort(GetSortField(searchFilter)));
if (results.TotalHits == 0 || searchFilter.CountOnly)
{
return(new SearchResults(results.TotalHits, timestamp));
}
var packages = results.ScoreDocs
.Skip(searchFilter.Skip)
.Select(sd => PackageFromDoc(searcher.Doc(sd.Doc)))
.ToList();
return(new SearchResults(
results.TotalHits,
timestamp,
packages.AsQueryable()));
}
private static Query ParseQuery(SearchFilter searchFilter)
{
if (String.IsNullOrWhiteSpace(searchFilter.SearchTerm))
{
return(new MatchAllDocsQuery());
}
var fields = new[] { "Id", "Title", "Tags", "Description", "Author" };
var analyzer = new StandardAnalyzer(LuceneCommon.LuceneVersion);
var queryParser = new MultiFieldQueryParser(LuceneCommon.LuceneVersion, fields, analyzer);
// All terms in the multi-term query appear in at least one of the fields.
var conjuctionQuery = new BooleanQuery();
conjuctionQuery.Boost = 2.0f;
// Some terms in the multi-term query appear in at least one of the fields.
var disjunctionQuery = new BooleanQuery();
disjunctionQuery.Boost = 0.1f;
// Suffix wildcard search e.g. jquer*
var wildCardQuery = new BooleanQuery();
wildCardQuery.Boost = 0.5f;
// Escape the entire term we use for exact searches.
var escapedSearchTerm = Escape(searchFilter.SearchTerm).Replace("id\\:", string.Empty).Replace("author\\:", string.Empty).Replace("tag\\:", string.Empty);
bool searchLimiter = false;
bool onlySearchById = false;
bool onlySearchByAuthor = false;
bool onlySearchByTag = false;
var exactIdQuery = new TermQuery(new Term("Id-Exact", escapedSearchTerm));
exactIdQuery.Boost = 2.5f;
var wildCardIdQuery = new WildcardQuery(new Term("Id-Exact", "*" + escapedSearchTerm + "*"));
foreach (var term in GetSearchTerms(searchFilter.SearchTerm))
{
var localTerm = term.to_lower_invariant();
onlySearchById = localTerm.StartsWith("id\\:");
onlySearchByAuthor = localTerm.StartsWith("author\\:");
onlySearchByTag = localTerm.StartsWith("tag\\:");
if (onlySearchById || onlySearchByAuthor || onlySearchByTag)
{
searchLimiter = true;
}
localTerm = term.Replace("id\\:", string.Empty).Replace("author\\:", string.Empty).Replace("tag\\:", string.Empty);
var termQuery = queryParser.Parse(localTerm);
conjuctionQuery.Add(termQuery, Occur.MUST);
disjunctionQuery.Add(termQuery, Occur.SHOULD);
foreach (var field in fields)
{
if (onlySearchById && field != "Id")
{
continue;
}
if (onlySearchByAuthor && field != "Author")
{
continue;
}
if (onlySearchByTag && field != "Tags")
{
continue;
}
var wildCardTermQuery = new WildcardQuery(new Term(field, localTerm + "*"));
wildCardTermQuery.Boost = searchLimiter ? 2.5f : 0.7f;
wildCardQuery.Add(wildCardTermQuery, Occur.SHOULD);
}
}
// Create an OR of all the queries that we have
var combinedQuery = conjuctionQuery.Combine(new Query[] { exactIdQuery, wildCardIdQuery, conjuctionQuery, disjunctionQuery, wildCardQuery });
if (onlySearchById)
{
combinedQuery = conjuctionQuery.Combine(new Query[] { exactIdQuery, wildCardIdQuery, wildCardQuery });
}
else if (onlySearchByAuthor || onlySearchByTag)
{
combinedQuery = conjuctionQuery.Combine(new Query[] { wildCardQuery });
}
if (searchFilter.SortProperty == SortProperty.Relevance)
{
// If searching by relevance, boost scores by download count.
var downloadCountBooster = new FieldScoreQuery("DownloadCount", FieldScoreQuery.Type.INT);
return(new CustomScoreQuery(combinedQuery, downloadCountBooster));
}
return(combinedQuery);
}
private static void VisitQuery(FieldScoreQuery query, AzureQueryLogger.IndentedTextWriter writer)
{
}
public virtual Query VisitFieldScoreQuery(FieldScoreQuery fieldScoreq)
{
throw new NotImplementedException();
}
public virtual Query VisitFieldScoreQuery(FieldScoreQuery fieldScoreq)
{
throw new SnNotSupportedException();
}
