Re: Full text search on partial URLs

On 11/15/2013 07:40 PM, Zev Benjamin wrote:
>
> One problem that I've run into here is that I would also like to
> highlight matched text in my application.  For my existing search
> solution, I do this with ts_headline.  For partial matches, it's
> unfortunately not just a matter of searching for the text and adding the
> appropriate markup because my documents are HTML (the FTS lexer
> helpfully pulls out all the HTML tags so it hasn't been a problem so
> far) and we don't want to accidentally "highlight" some of the
> attributes of the markup.
>
> One way to solve this would be if there were a way to turn a tsvector
> and tsquery pair into a list of the offsets and lengths of the lexemes
> that match.  The highlighting could then be done at the application
> level rather than the database level while still leveraging Postgres's
> FTS functionality.

I've written C functions to implement this and attached them to this
email.  The support files necessary for making a module are available at
https://github.com/zbenjamin/tsearch_extras.  I'm new to the PostgreSQL
code base so any feedback or comments would be greatly appreciated.
Would these be appropriate to submit as patches to PostgreSQL?


Thanks,
Zev