private static Query ParseQuery(string searchTerm)
{
var fields = new Dictionary<string, float> { { "Id", 1.2f }, { "Title", 1.0f }, { "Tags", 0.8f }, { "Description", 0.1f },
{ "Author", 1.0f } };
var analyzer = new StandardAnalyzer(LuceneCommon.LuceneVersion);
var queryParser = new MultiFieldQueryParser(LuceneCommon.LuceneVersion, fields.Keys.ToArray(), analyzer, fields);
var conjuctionQuery = new BooleanQuery();
conjuctionQuery.SetBoost(2.0f);
var disjunctionQuery = new BooleanQuery();
disjunctionQuery.SetBoost(0.1f);
var wildCardQuery = new BooleanQuery();
wildCardQuery.SetBoost(0.5f);
// Escape the entire term we use for exact searches.
var escapedSearchTerm = Escape(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(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.Key, term + "*"));
wildCardTermQuery.SetBoost(0.7f * field.Value);
wildCardQuery.Add(wildCardTermQuery, BooleanClause.Occur.SHOULD);
}
}
var downloadCountBooster = new FieldScoreQuery("DownloadCount", FieldScoreQuery.Type.INT);
return new CustomScoreQuery(conjuctionQuery.Combine(new Query[] { exactIdQuery, wildCardIdQuery, conjuctionQuery, disjunctionQuery, wildCardQuery }),
downloadCountBooster);
}
private static Query BuildGeneralQuery(
bool doExactId,
string originalSearchText,
Analyzer analyzer,
IEnumerable<NuGetSearchTerm> generalTerms,
IEnumerable<Query> generalQueries)
{
// All terms in the multi-term query appear in at least one of the target fields.
var conjuctionQuery = new BooleanQuery();
conjuctionQuery.SetBoost(2.0f);
// Some terms in the multi-term query appear in at least one of the target fields.
var disjunctionQuery = new BooleanQuery();
disjunctionQuery.SetBoost(0.1f);
// Suffix wildcard search e.g. jquer*
var wildCardQuery = new BooleanQuery();
wildCardQuery.SetBoost(0.5f);
string escapedExactId = originalSearchText.ToLowerInvariant();
Query exactIdQuery = null;
Query wildCardIdQuery = null;
if (doExactId)
{
exactIdQuery = new TermQuery(new Term("Id-Exact", escapedExactId));
exactIdQuery.SetBoost(7.5f);
wildCardIdQuery = new WildcardQuery(new Term("Id-Exact", "*" + escapedExactId + "*"));
}
Query nearlyExactIdQuery = null;
if (generalTerms.Any())
{
string escapedApproximateId = string.Join(" ", generalTerms.Select(c => c.TermOrPhrase));
nearlyExactIdQuery = AnalysisHelper.GetFieldQuery(analyzer, "Id", escapedApproximateId);
nearlyExactIdQuery.SetBoost(2.0f);
}
foreach (var termQuery in generalQueries)
{
conjuctionQuery.Add(termQuery, BooleanClause.Occur.MUST);
disjunctionQuery.Add(termQuery, BooleanClause.Occur.SHOULD);
}
var sanitizedTerms = generalTerms.Select(c => c.TermOrPhrase.ToLowerInvariant());
foreach (var sanitizedTerm in sanitizedTerms)
{
foreach (var field in Fields)
{
var wildCardTermQuery = new WildcardQuery(new Term(field, sanitizedTerm + "*"));
wildCardTermQuery.SetBoost(0.7f);
wildCardQuery.Add(wildCardTermQuery, BooleanClause.Occur.SHOULD);
}
}
// OR of all the applicable queries
var queries = new Query[]
{
exactIdQuery, wildCardIdQuery, nearlyExactIdQuery, conjuctionQuery, disjunctionQuery, wildCardQuery
};
var queriesToCombine = queries.Where(q => !IsDegenerateQuery(q));
var query = conjuctionQuery.Combine(queriesToCombine.ToArray());
return query;
}
private static Query ParseQuery(SearchFilter searchFilter)
{
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 static Query ParseQuery(string searchTerm)
{
var fields = new Dictionary<string, float> { { "Id", 1.2f }, { "Title", 1.0f }, { "Tags", 1.0f}, { "Description", 0.8f }, { "Author", 0.6f } };
var analyzer = new StandardAnalyzer(LuceneCommon.LuceneVersion);
searchTerm = QueryParser.Escape(searchTerm).ToLowerInvariant();
var queryParser = new MultiFieldQueryParser(LuceneCommon.LuceneVersion, fields.Keys.ToArray(), analyzer, fields);
var conjuctionQuery = new BooleanQuery();
conjuctionQuery.SetBoost(1.5f);
var disjunctionQuery = new BooleanQuery();
var wildCardQuery = new BooleanQuery();
wildCardQuery.SetBoost(0.7f);
var exactIdQuery = new TermQuery(new Term("Id-Exact", searchTerm));
exactIdQuery.SetBoost(2.5f);
foreach(var term in searchTerm.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries))
{
conjuctionQuery.Add(queryParser.Parse(term), BooleanClause.Occur.MUST);
disjunctionQuery.Add(queryParser.Parse(term), BooleanClause.Occur.SHOULD);
foreach (var field in fields)
{
var wildCardTermQuery = new WildcardQuery(new Term(field.Key, term + "*"));
wildCardTermQuery.SetBoost(0.7f * field.Value);
wildCardQuery.Add(wildCardTermQuery, BooleanClause.Occur.SHOULD);
}
}
return conjuctionQuery.Combine(new Query[] { exactIdQuery, conjuctionQuery, disjunctionQuery, wildCardQuery });
}