public static Query BuildQuery(string query, IndexQuery indexQuery, RavenPerFieldAnalyzerWrapper analyzer)
{
var originalQuery = query;
try
{
var queryParser = new RangeQueryParser(Version.LUCENE_29, indexQuery.DefaultField ?? string.Empty, analyzer)
{
DefaultOperator = indexQuery.DefaultOperator == QueryOperator.Or
? QueryParser.Operator.OR
: QueryParser.Operator.AND,
AllowLeadingWildcard = true
};
query = PreProcessComments(query);
query = PreProcessMixedInclusiveExclusiveRangeQueries(query);
query = PreProcessUntokenizedTerms(query, queryParser);
query = PreProcessSearchTerms(query);
query = PreProcessDateTerms(query, queryParser);
var generatedQuery = queryParser.Parse(query);
generatedQuery = HandleMethods(generatedQuery, analyzer);
return(generatedQuery);
}
catch (ParseException pe)
{
if (originalQuery == query)
{
throw new ParseException("Could not parse: '" + query + "'", pe);
}
throw new ParseException("Could not parse modified query: '" + query + "' original was: '" + originalQuery + "'", pe);
}
}
public static Query BuildQuery(string query, IndexQuery indexQuery, RavenPerFieldAnalyzerWrapper analyzer)
{
using (CultureHelper.EnsureInvariantCulture())
{
if (UseLuceneASTParser)
{
try
{
var parser = new LuceneQueryParser();
parser.IsDefaultOperatorAnd = indexQuery.DefaultOperator == QueryOperator.And;
parser.Parse(query);
var res = parser.LuceneAST.ToQuery(
new LuceneASTQueryConfiguration
{
Analayzer = analyzer,
DefaultOperator = indexQuery.DefaultOperator,
FieldName = indexQuery.DefaultField ?? string.Empty
});
// The parser already throws parse exception if there is a syntax error.
// We now return null in the case of a term query that has been fully analyzed, so we need to return a valid query.
if (res == null)
{
return(new BooleanQuery());
}
return(res);
}
catch (ParseException pe)
{
throw new ParseException("Could not parse: '" + query + "'", pe);
}
}
var originalQuery = query;
try
{
var queryParser = new RangeQueryParser(Version.LUCENE_29, indexQuery.DefaultField ?? string.Empty, analyzer)
{
DefaultOperator = indexQuery.DefaultOperator == QueryOperator.Or
? QueryParser.Operator.OR
: QueryParser.Operator.AND,
AllowLeadingWildcard = true
};
query = PreProcessComments(query);
query = PreProcessMixedInclusiveExclusiveRangeQueries(query);
query = PreProcessUntokenizedTerms(query, queryParser);
query = PreProcessSearchTerms(query);
query = PreProcessDateTerms(query, queryParser);
var generatedQuery = queryParser.Parse(query);
generatedQuery = HandleMethods(generatedQuery, analyzer);
return(generatedQuery);
}
catch (ParseException pe)
{
if (originalQuery == query)
{
throw new ParseException("Could not parse: '" + query + "'", pe);
}
throw new ParseException("Could not parse modified query: '" + query + "' original was: '" + originalQuery + "'", pe);
}
}
}
private static string PreProcessDateTerms(string query, RangeQueryParser queryParser)
{
var searchMatches = dateQuery.Matches(query);
if (searchMatches.Count < 1)
{
return(query);
}
var queryStringBuilder = new StringBuilder(query);
for (var i = searchMatches.Count - 1; i >= 0; i--) // reversing the scan so we won't affect positions of later items
{
var searchMatch = searchMatches[i];
var field = searchMatch.Groups[1].Value;
var termReplacement = searchMatch.Groups[2].Value;
var replaceToken = queryParser.ReplaceToken(field, termReplacement);
queryStringBuilder.Remove(searchMatch.Index, searchMatch.Length);
queryStringBuilder
.Insert(searchMatch.Index, field)
.Insert(searchMatch.Index + field.Length, ":")
.Insert(searchMatch.Index + field.Length + 1, replaceToken);
}
return(queryStringBuilder.ToString());
}
public static Query BuildQuery(string query, IndexQuery indexQuery, PerFieldAnalyzerWrapper analyzer)
{
var originalQuery = query;
Analyzer keywordAnalyzer = new KeywordAnalyzer();
try
{
var queryParser = new RangeQueryParser(Version.LUCENE_29, indexQuery.DefaultField ?? string.Empty, analyzer)
{
DefaultOperator = indexQuery.DefaultOperator == QueryOperator.Or
? QueryParser.Operator.OR
: QueryParser.Operator.AND,
AllowLeadingWildcard = true
};
query = PreProcessUntokenizedTerms(query, queryParser);
query = PreProcessSearchTerms(query);
query = PreProcessDateTerms(query, queryParser);
var generatedQuery = queryParser.Parse(query);
generatedQuery = HandleMethods(generatedQuery);
return generatedQuery;
}
catch (ParseException pe)
{
if (originalQuery == query)
throw new ParseException("Could not parse: '" + query + "'", pe);
throw new ParseException("Could not parse modified query: '" + query + "' original was: '" + originalQuery + "'", pe);
}
finally
{
keywordAnalyzer.Close();
}
}
public static Query BuildQuery(string query, IndexQuery indexQuery, PerFieldAnalyzerWrapper analyzer)
{
var originalQuery = query;
Analyzer keywordAnalyzer = new KeywordAnalyzer();
try
{
var queryParser = new RangeQueryParser(Version.LUCENE_29, indexQuery.DefaultField ?? string.Empty, analyzer)
{
DefaultOperator = indexQuery.DefaultOperator == QueryOperator.Or
? QueryParser.Operator.OR
: QueryParser.Operator.AND,
AllowLeadingWildcard = true
};
query = PreProcessUntokenizedTerms(query, queryParser);
query = PreProcessSearchTerms(query);
query = PreProcessDateTerms(query, queryParser);
return(queryParser.Parse(query));
}
catch (ParseException pe)
{
if (originalQuery == query)
{
throw new ParseException("Could not parse: '" + query + "'", pe);
}
throw new ParseException("Could not parse modified query: '" + query + "' original was: '" + originalQuery + "'", pe);
}
finally
{
keywordAnalyzer.Close();
}
}
/// <summary>
/// Detects untokenized fields and sets as NotAnalyzed in analyzer
/// </summary>
private static string PreProcessUntokenizedTerms(string query, RangeQueryParser queryParser)
{
var untokenizedMatches = untokenizedQuery.Matches(query);
if (untokenizedMatches.Count < 1)
{
return(query);
}
var sb = new StringBuilder(query);
MatchCollection fieldMatches = null;
// process in reverse order to leverage match string indexes
for (var i = untokenizedMatches.Count; i > 0; i--)
{
var match = untokenizedMatches[i - 1];
// specify that term for this field should not be tokenized
var value = match.Groups[2].Value;
var term = match.Groups[2];
string name = match.Groups[1].Value;
if (string.IsNullOrEmpty(value))
{
value = match.Groups[3].Value;
term = match.Groups[3];
if (fieldMatches == null)
{
fieldMatches = fieldQuery.Matches(query);
}
var lastField = fieldMatches.Cast <Match>().LastOrDefault(x => x.Index <= term.Index);
if (lastField != null)
{
name = lastField.Groups[1].Value;
}
}
var rawTerm = value.Substring(2, value.Length - 4);
queryParser.SetUntokenized(name, Unescape(rawTerm));
// introduce " " around the term
var startIndex = term.Index;
var length = term.Length - 2;
if (sb[startIndex + length - 1] != '"')
{
sb.Insert(startIndex + length, '"');
length += 1;
}
if (sb[startIndex + 2] != '"')
{
sb.Insert(startIndex + 2, '"');
length += 1;
}
// remove enclosing "[[" "]]" from term value (again in reverse order)
sb.Remove(startIndex + length, 2);
sb.Remove(startIndex, 2);
}
return(sb.ToString());
}
public void MrsJones()
{
using (var dir = new RAMDirectory())
using (var analyzer = new LowerCaseKeywordAnalyzer())
{
using (var writer = new IndexWriter(dir, analyzer, true, IndexWriter.MaxFieldLength.UNLIMITED))
{
var document = new Lucene.Net.Documents.Document();
document.Add(new Field("Name", "MRS. SHABA", Field.Store.NO, Field.Index.ANALYZED_NO_NORMS));
writer.AddDocument(document);
}
var searcher = new IndexSearcher(dir, true);
var termEnum = searcher.IndexReader.Terms();
while (termEnum.Next())
{
var buffer = termEnum.Term.Text;
Console.WriteLine(buffer);
}
var queryParser = new RangeQueryParser(Version.LUCENE_29, "", analyzer);
var query = queryParser.Parse("Name:\"MRS. S*\"");
Console.WriteLine(query);
var result = searcher.Search(query, 10);
Assert.NotEqual(0, result.TotalHits);
}
}
public static Query BuildQuery(string query, IndexQuery indexQuery, RavenPerFieldAnalyzerWrapper analyzer)
{
using (CultureHelper.EnsureInvariantCulture())
{
if (UseLuceneASTParser)
{
try
{
var parser = new LuceneQueryParser();
parser.Parse(query);
var res = parser.LuceneAST.ToQuery(
new LuceneASTQueryConfiguration
{
Analayzer = analyzer,
DefaultOperator = indexQuery.DefaultOperator,
FieldName = indexQuery.DefaultField ?? string.Empty
});
// The parser should throw ParseException in this case.
if (res == null)
{
throw new GeoAPI.IO.ParseException("Could not parse query");
}
return(res);
}
catch (ParseException pe)
{
throw new ParseException("Could not parse: '" + query + "'", pe);
}
}
var originalQuery = query;
try
{
var queryParser = new RangeQueryParser(Version.LUCENE_29, indexQuery.DefaultField ?? string.Empty, analyzer)
{
DefaultOperator = indexQuery.DefaultOperator == QueryOperator.Or
? QueryParser.Operator.OR
: QueryParser.Operator.AND,
AllowLeadingWildcard = true
};
query = PreProcessComments(query);
query = PreProcessMixedInclusiveExclusiveRangeQueries(query);
query = PreProcessUntokenizedTerms(query, queryParser);
query = PreProcessSearchTerms(query);
query = PreProcessDateTerms(query, queryParser);
var generatedQuery = queryParser.Parse(query);
generatedQuery = HandleMethods(generatedQuery, analyzer);
return(generatedQuery);
}
catch (ParseException pe)
{
if (originalQuery == query)
{
throw new ParseException("Could not parse: '" + query + "'", pe);
}
throw new ParseException("Could not parse modified query: '" + query + "' original was: '" + originalQuery + "'", pe);
}
}
}
private static string PreProcessDateTerms(string query, RangeQueryParser queryParser)
{
var searchMatches = dateQuery.Matches(query);
if (searchMatches.Count > 0)
{
query = TokenReplace(query, searchMatches, queryParser.ReplaceToken);
}
searchMatches = inDatesQuery.Matches(query);
if (searchMatches.Count == 0)
{
return(query);
}
return(TokenReplace(query, searchMatches, queryParser.ReplaceDateTimeTokensInMethod));
}
public static Query BuildQuery(string query, PerFieldAnalyzerWrapper analyzer)
{
var keywordAnalyzer = new KeywordAnalyzer();
try
{
query = PreProcessUntokenizedTerms(analyzer, query, keywordAnalyzer);
var queryParser = new RangeQueryParser(Version.LUCENE_29, "", analyzer);
queryParser.SetAllowLeadingWildcard(true);
return queryParser.Parse(query);;
}
finally
{
keywordAnalyzer.Close();
}
}
public static Query BuildQuery(string query, PerFieldAnalyzerWrapper analyzer)
{
Analyzer keywordAnalyzer = null;
try
{
query = PreProcessUntokenizedTerms(analyzer, query, ref keywordAnalyzer);
var queryParser = new RangeQueryParser(Version.LUCENE_29, string.Empty, analyzer);
queryParser.SetAllowLeadingWildcard(true); // not the recommended approach, should rather use ReverseFilter
return queryParser.Parse(query);
}
finally
{
if (keywordAnalyzer != null)
keywordAnalyzer.Close();
}
}
public static Query BuildQuery(string query, PerFieldAnalyzerWrapper analyzer)
{
var keywordAnalyzer = new KeywordAnalyzer();
try
{
query = PreProcessUntokenizedTerms(analyzer, query, keywordAnalyzer);
var queryParser = new RangeQueryParser(Version.LUCENE_29, "", analyzer);
queryParser.SetAllowLeadingWildcard(true);
return(queryParser.Parse(query));;
}
finally
{
keywordAnalyzer.Close();
}
}
public static Query BuildQuery(string query, string defaultField, PerFieldAnalyzerWrapper analyzer)
{
Analyzer keywordAnalyzer = new KeywordAnalyzer();
try
{
var queryParser = new RangeQueryParser(Version.LUCENE_29, defaultField ?? string.Empty, analyzer);
query = PreProcessUntokenizedTerms(query, queryParser);
query = PreProcessSearchTerms(query);
query = PreProcessDateTerms(query, queryParser);
queryParser.SetAllowLeadingWildcard(true); // not the recommended approach, should rather use ReverseFilter
return queryParser.Parse(query);
}
finally
{
keywordAnalyzer.Close();
}
}
public static Query BuildQuery(string query, PerFieldAnalyzerWrapper analyzer)
{
Analyzer keywordAnalyzer = new KeywordAnalyzer();
try
{
var queryParser = new RangeQueryParser(Version.LUCENE_29, string.Empty, analyzer);
query = PreProcessUntokenizedTerms(query, queryParser);
query = PreProcessSearchTerms(query);
queryParser.SetAllowLeadingWildcard(true); // not the recommended approach, should rather use ReverseFilter
return(queryParser.Parse(query));
}
finally
{
keywordAnalyzer.Close();
}
}
/// <summary>
/// Detects untokenized fields and sets as NotAnalyzed in analyzer
/// </summary>
private static string PreProcessUntokenizedTerms(string query, RangeQueryParser queryParser)
{
var untokenizedMatches = untokenizedQuery.Matches(query);
if (untokenizedMatches.Count < 1)
{
return(query);
}
var sb = new StringBuilder(query);
// process in reverse order to leverage match string indexes
for (var i = untokenizedMatches.Count; i > 0; i--)
{
var match = untokenizedMatches[i - 1];
// specify that term for this field should not be tokenized
var value = match.Groups[2].Value;
var rawTerm = value.Substring(2, value.Length - 4);
queryParser.SetUntokenized(match.Groups[1].Value, Unescape(rawTerm));
var term = match.Groups[2];
// introduce " " around the term
var startIndex = term.Index;
var length = term.Length - 2;
if (sb[startIndex + length - 1] != '"')
{
sb.Insert(startIndex + length, '"');
length += 1;
}
if (sb[startIndex + 2] != '"')
{
sb.Insert(startIndex + 2, '"');
length += 1;
}
// remove enclosing "[[" "]]" from term value (again in reverse order)
sb.Remove(startIndex + length, 2);
sb.Remove(startIndex, 2);
}
return(sb.ToString());
}
internal static string PreProcessDateTerms(string query, RangeQueryParser queryParser)
{
// First we should check if this query might match the regex because regex are expenssive...
if (!MightMatchDateTerms(query))
{
return(query);
}
var searchMatches = dateQuery.Matches(query);
if (searchMatches.Count > 0)
{
query = TokenReplace(query, searchMatches, queryParser.ReplaceToken);
}
searchMatches = inDatesQuery.Matches(query);
if (searchMatches.Count == 0)
{
return(query);
}
return(TokenReplace(query, searchMatches, queryParser.ReplaceDateTimeTokensInMethod));
}
/// <summary>
/// Detects untokenized fields and sets as NotAnalyzed in analyzer
/// </summary>
private static string PreProcessUntokenizedTerms(string query, RangeQueryParser queryParser)
{
var untokenizedMatches = untokenizedQuery.Matches(query);
if (untokenizedMatches.Count < 1)
return query;
var sb = new StringBuilder(query);
// process in reverse order to leverage match string indexes
for (var i = untokenizedMatches.Count; i > 0; i--)
{
var match = untokenizedMatches[i - 1];
// specify that term for this field should not be tokenized
var value = match.Groups[2].Value;
var rawTerm = value.Substring(2, value.Length-4);
queryParser.SetUntokenized(match.Groups[1].Value, Unescape(rawTerm));
var term = match.Groups[2];
// introduce " " around the term
var startIndex = term.Index;
var length = term.Length - 2;
if (sb[startIndex + length - 1] != '"')
{
sb.Insert(startIndex + length, '"');
length += 1;
}
if (sb[startIndex + 2] != '"')
{
sb.Insert(startIndex + 2, '"');
length += 1;
}
// remove enclosing "[[" "]]" from term value (again in reverse order)
sb.Remove(startIndex + length, 2);
sb.Remove(startIndex, 2);
}
return sb.ToString();
}
private static string PreProcessDateTerms(string query, RangeQueryParser queryParser)
{
var searchMatches = dateQuery.Matches(query);
if (searchMatches.Count < 1)
return query;
var queryStringBuilder = new StringBuilder(query);
for (var i = searchMatches.Count-1; i >= 0; i--) // reversing the scan so we won't affect positions of later items
{
var searchMatch = searchMatches[i];
var field = searchMatch.Groups[1].Value;
var termReplacement = searchMatch.Groups[2].Value;
var replaceToken = queryParser.ReplaceToken(field, termReplacement);
queryStringBuilder.Remove(searchMatch.Index, searchMatch.Length);
queryStringBuilder
.Insert(searchMatch.Index, field)
.Insert(searchMatch.Index + field.Length, ":")
.Insert(searchMatch.Index + field.Length + 1, replaceToken);
}
return queryStringBuilder.ToString();
}
internal static string PreProcessDateTerms(string query, RangeQueryParser queryParser)
{
// First we should check if this query might match the regex because regex are expenssive...
if (!MightMatchDateTerms(query)) return query;
var searchMatches = dateQuery.Matches(query);
if (searchMatches.Count > 0)
{
query = TokenReplace(query, searchMatches,queryParser.ReplaceToken);
}
searchMatches = inDatesQuery.Matches(query);
if (searchMatches.Count == 0)
return query;
return TokenReplace(query, searchMatches,queryParser.ReplaceDateTimeTokensInMethod);
}
/// <summary>
/// Detects untokenized fields and sets as NotAnalyzed in analyzer
/// </summary>
private static string PreProcessUntokenizedTerms(string query, RangeQueryParser queryParser)
{
var untokenizedMatches = untokenizedQuery.Matches(query);
if (untokenizedMatches.Count < 1)
return query;
var sb = new StringBuilder(query);
MatchCollection fieldMatches = null;
// process in reverse order to leverage match string indexes
for (var i = untokenizedMatches.Count; i > 0; i--)
{
var match = untokenizedMatches[i - 1];
// specify that term for this field should not be tokenized
var value = match.Groups[2].Value;
var term = match.Groups[2];
string name = match.Groups[1].Value;
if (string.IsNullOrEmpty(value))
{
value = match.Groups[3].Value;
term = match.Groups[3];
if(fieldMatches == null)
fieldMatches = fieldQuery.Matches(query);
var lastField = fieldMatches.Cast<Match>().LastOrDefault(x => x.Index <= term.Index);
if (lastField != null)
{
name = lastField.Groups[1].Value;
}
}
var rawTerm = value.Substring(2, value.Length - 4);
queryParser.SetUntokenized(name, Unescape(rawTerm));
// introduce " " around the term
var startIndex = term.Index;
var length = term.Length - 2;
if (sb[startIndex + length - 1] != '"')
{
sb.Insert(startIndex + length, '"');
length += 1;
}
if (sb[startIndex + 2] != '"')
{
sb.Insert(startIndex + 2, '"');
length += 1;
}
// remove enclosing "[[" "]]" from term value (again in reverse order)
sb.Remove(startIndex + length, 2);
sb.Remove(startIndex, 2);
}
return sb.ToString();
}
private static string PreProcessDateTerms(string query, RangeQueryParser queryParser)
{
var searchMatches = dateQuery.Matches(query);
if (searchMatches.Count > 0)
{
query = TokenReplace(query, searchMatches,queryParser.ReplaceToken);
}
searchMatches = inDatesQuery.Matches(query);
if (searchMatches.Count == 0)
return query;
return TokenReplace(query, searchMatches,queryParser.ReplaceDateTimeTokensInMethod);
}