solr.**PatternReplaceCharFilterFactor**y is a brilliant idea, thanks so much :)
On Wed, Jul 4, 2012 at 2:46 PM, Jack Krupansky <j...@basetechnology.com>wrote: > That's a perfectly reasonable request. But, WDF doesn't have such a > feature. > > Maybe what is needed is a distinct "ampersand filter" that runs before WDF > and detects ampersands that are likely shorthands for "and" and expands > them. It would also need to be able to detect "AT&T" (capital letter before > the &) and not expand it (and you can set up a character type table for WDF > that treats "&" as a letter. A single "&" could also be expanded to "and" - > that could also be done with the synonym filter, but that would not help > you with the embedded "&" of "Apples&Oranges". > > Maybe a simple character filter that always expands "&" to " and " would > be good enough for a lot of common cases, as a rough approximation. > > Maybe solr.**PatternReplaceCharFilterFactor**y could be used to > accomplish that. Match "&" and replace with " and ". > > -- Jack Krupansky > > -----Original Message----- From: Stephen Lacy > Sent: Wednesday, July 04, 2012 8:16 AM > To: solr-user@lucene.apache.org > Subject: WordDelimiterFilter removes ampersands > > > If a user writes a query "Apples & Oranges" the word delimiter filter > factory will change this into "Apples Oranges" > Which isn't very useful for me as I'd prefer especially when the phrase is > wrapped in quotes that the original is preserved. > However I still want to be able to separate Apples&Oranges into Apples & > Oranges so preserveOriginal isn't really useful. > What I really would like to be able to do is tell WordDelimeterFilter to > treat it like it's neither alpha nor numeric, however > that doesn't mean that you remove it completely. > > Thanks for your help > Stephen >
