Marco Schmidt writes: > I'm trying to find out whether Lucene is an option for a project of > mine. I have texts which also have a date and a list of numbers > associated with each of them. These numbers are ID values which connect > the article to certain categories. So a particular article X might > belong to categories 17, 49 and 112. A search for all articles > containing "foo bar" and belonging to categories 100 to 140 should > return X (because it also contains "foo bar"). Is it possible to do this > with Lucene and if it is, how? I've read about the concept of fields in > Lucene, but it seems to me that you can only store text in them, not > integers, let alone list of integers. None of the tutorials I've seen > deals with more complex queries like that. Basically what I want to > accomplish could be done nicely with databases with full text search > capability, if that full text search wasn't so awful. > Where's the problem? 100 is a text as well as an integer (one has to keep in mind, that treating it as text changes sort order, which may require leading 0 to compensate). Lucene does not understand the "words" you index anyway.
So if a document has a field `category' with content '017 049 112' and some `text' field with content 'bla fasel foo bar' and you do a range query 100 - 140 on category (search all documents containing any word, that is alphanumerically sorted between 100 and 140) and a apropriate query on text it will find, what you want. There are some caveats like choosing an apropriate analyzer or considering the maximum number of terms the range query covers, but in principle there is no difference between a text field containing words and a category field containing categories. Morus --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
