Erik, thank you for responsing. I will check the code to get some ideas for implementation.
I do need some cached ressources like the CharArraySet of protected words for a WordDelimiterFilter (for the MAX_LEN-parameter mentioned by Hoss) or a SynonymFilter . I think it would consume too much time to retrive these information again and again everytime I do a query. That's why Solr uses FilterFactories, since they are storing these information for queries. I will try to explain what I want to do. Perhaps my english explanation is not good, so if you feel so, please ask. I want to extend QueryParser to a) create the the MAX_LEN - param b) making multiWordSynonyms possible at query-time. I think b) is a little bit tricky. At query-time multi-word synonyms is not possible, - as I have understood - because of the way a querystring get's parsed by a QueryParser. However, what would be, if I modify the querystr at parsing time? Imagine what would be if I replace for example "my four word synonym" with "SOLRSYN_ID589"? I think this is possible, if I do the following: - creating a StringReader(querystr) - tokenize and filter the querystr (it is the same work like an analyzer does in Lucene) - put the result in the ToStringTokenFilter (the ToStringTokenFilter needs a field like "String newQuerystr") - retrive the newQueryStr (ToStringTokenFilter.newQuerystr) and go on working with it as you would normaly do Pros: - making MultiWordSynonyms at query-time possible Cons: - *every* field has to implement a SynonymFilter which retrives those synonyms which belong to - in my example - SOLRSYN_ID589. I *think* this makes querying against StringFields impossible, if there is a synonym. How does a synonym.txt has to looks like? index_synonyms.txt -> as normally given by the example (you can do whatever you want) query_synonyms.txt -> has to look different: ------------------------------------------- SOLRSYN_ID589 => my other example, a word, another word, three word example etc. Maybe one needs to implement another SynonymFilter with another synonyms.txt normal_synonyms.txt I think my example is bad, but I hope it explains what I mean: another word => theater, nightclub This is a really large overhead, however it makes multiword-synonyms possible, *if* I am right. If there are no cons for doing it this way, I will try to implement it next time. - Mitch -- View this message in context: http://n3.nabble.com/Minimum-Should-Match-the-other-way-round-tp694867p703353.html Sent from the Solr - User mailing list archive at Nabble.com.
