nutch field problem

Question

I was using something like:

Field notdirectory = new Field("notdirectory","1", Field.Store.NO, Field.Index.UN_TOKENIZED);

and queries like "notdirectory:1" can be processed quite well all the time.

But recently I've changed the "Field.Store.NO, Field.Index.UN_TOKENIZED" to index a non-numeric string:

Field stateField = new Field("state","irn_" + state, Field.Store.NO, Field.Index.UN_TOKENIZED);

and queries like "state:irn_CA" can never fetch any results any more,even though I watch through hadoop logs that "irn_CA" is added to "state" field in fact.

So I doubt for Fields that satisfy "Field.Store.NO, Field.Index.UN_TOKENIZED",only numeric Fields can searchable,but I didn't see any documents about that.

So what's the true reason for this?

Answer 1

I think, you are using StandardAnalyzer for parsing the input query, which will tokenize your input query "irn_CA" into two tokens - "irn" and "CA". Since the index has "irn_CA" as single token, it won't match.

Try using KeywordAnalyzer for while searching. It will generate single token for the query string and match the indexed token correctly.