private static Query ParseQuery(SearchFilter searchFilter)
{
// 1. parse the query into field clauses and general terms
// We imagine that mostly, field clauses are meant to 'filter' results found searching for general terms.
// The resulting clause collections may be empty.
var queryParser = new NuGetQueryParser();
var clauses = queryParser.Parse(searchFilter.SearchTerm).Select(StandardizeSearchTerms).ToList();
var fieldSpecificTerms = clauses.Where(a => a.Field != null);
var generalTerms = clauses.Where(a => a.Field == null);
// Convert terms to appropriate Lucene Query objects
var analyzer = new PerFieldAnalyzer();
var fieldSpecificQueries = fieldSpecificTerms
.Select(c => AnalysisHelper.GetFieldQuery(analyzer, c.Field, c.TermOrPhrase))
.Where(q => !IsDegenerateQuery(q))
.ToList();
var generalQueries = generalTerms
.Select(c => AnalysisHelper.GetMultiFieldQuery(analyzer, Fields, c.TermOrPhrase))
.Where(q => !IsDegenerateQuery(q))
.ToList();
if (fieldSpecificQueries.Count == 0 &&
generalQueries.Count == 0)
{
return(new MatchAllDocsQuery());
}
// At this point we try to detect user intent...
// a) General search? [foo bar]
// b) Id-targeted search? [id:Foo bar]
// c) Other Field-targeted search? [author:Foo bar]
bool doExactId = !fieldSpecificQueries.Any();
Query generalQuery = BuildGeneralQuery(doExactId, searchFilter.SearchTerm, analyzer, generalTerms, generalQueries);
// IF field targeting is done, we should basically want to AND their field specific queries with all other query terms
if (fieldSpecificQueries.Any())
{
var combinedQuery = new BooleanQuery();
if (!IsDegenerateQuery(generalQuery))
{
combinedQuery.Add(generalQuery, Occur.MUST);
}
foreach (var fieldQuery in fieldSpecificQueries)
{
if (!IsDegenerateQuery(fieldQuery))
{
combinedQuery.Add(fieldQuery, Occur.MUST);
}
}
generalQuery = combinedQuery;
}
return(generalQuery);
}
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.Boost = 2.0f;
// Some terms in the multi-term query appear in at least one of the target fields.
var disjunctionQuery = new BooleanQuery();
disjunctionQuery.Boost = 0.1f;
// Suffix wildcard search e.g. jquer*
var wildCardQuery = new BooleanQuery();
wildCardQuery.Boost = 0.5f;
string escapedExactId = originalSearchText.ToLowerInvariant();
Query exactIdQuery = null;
Query wildCardIdQuery = null;
if (doExactId)
{
exactIdQuery = new TermQuery(new Term("Id-Exact", escapedExactId));
exactIdQuery.Boost = 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.Boost = 2.0f;
}
foreach (var termQuery in generalQueries)
{
conjuctionQuery.Add(termQuery, Occur.MUST);
disjunctionQuery.Add(termQuery, 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.Boost = 0.7f;
wildCardQuery.Add(wildCardTermQuery, 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);
}