Prevent KeywordTokenizer from creating multiple key-value pairs

Question

I use the Lucene java QueryParser with KeywordAnalyzer. A query like topic:(hello world) is broken up in to multiple parts by the KeywordTokenizer so the resulting Query object looks like this topic:(hello) topic:(world) i.e. Instead of one, I now have two key-value pairs. I would like the QueryParser to interpret hello world as one value, without using double quotes. What is the best way to do so?

Parsing topic:("hello world") results in a single key value combination but, using double quotes is not an option.

I am not using the Lucene search engine. I am using Lucene's QueryParser just for parsing the query, not for searching. The text Hello World is entered at runtime, by the user so that can change. I would like KeywordTokenizer to treat Hello World as one Token instead of parsing splitting it in to two Tokens.

Answer 1

You'll need to use a BooleanQuery. Here's a code snippet using the .NET port of Lucene. This should work with both the KeywordAnalyzer and the StandardAnalyzer.

var luceneAnalyzer = new KeywordAnalyzer();

var query1 = new QueryParser("Topic", luceneAnalyzer).Parse("hello");
var query2 = new QueryParser("Topic", luceneAnalyzer).Parse("world");

BooleanQuery filterQuery = new BooleanQuery();
filterQuery.Add(query1, BooleanClause.Occur.MUST);
filterQuery.Add(query1, BooleanClause.Occur.MUST);

TopDocs results = searcher.Search(filterQuery);

Answer 2

You can construct the query directly as follows. This preserves the space.

Query query = new TermQuery(new Term("field", "value has space"));

If you print query as System.out.println(query); you will see the following result.

field:value has space