ft:mark

On Thu, Jul 2, 2015 at 6:13 PM, Lars Johnsen <yoonsen at gmail.com> wrote: > The full text function ft:mark() puts a mark around each of the words that > occur in a match, starting from the first matching word to the last, > including stop words, except for punctuation characters. Is it possible to > check for the kind of characters (or strings) that ft:mark() will skip when > marking matches? Or, would it be possible to ask ft:mark() to put one marker > for the whole match? > > The case I am using it for is to get the sequence of matching words within a > match, and sometimes, for long strings, there may be several sequences. A > contiguous sequence of marked elements maybe assumed to make up a match, > while non contiguous do not. One marker around the match solves the problem, > as would detecting characters that are never marked. > > Regards > Lars G Johnsen > National Library of Norway

> Hi Lars, > > I guess it was due to the internal XQFT representation of match > positions that adjacent matches are marked one by one, and it's > probably difficult to change this without too much additional effort. > > However, you can use XQuery for transforming your result; see e.g. the > attached XQuery code. I must confess I didn't spend too much time on > it (right now, it's simply too hot around here...), so I'm looking for > revised versions ;) > > Christian > > > On Thu, Jul 2, 2015 at 6:13 PM, Lars Johnsen <yoonsen at gmail.com> wrote: > > The full text function ft:mark() puts a mark around each of the words > that > > occur in a match, starting from the first matching word to the last, > > including stop words, except for punctuation characters. Is it possible > to > > check for the kind of characters (or strings) that ft:mark() will skip > when > > marking matches? Or, would it be possible to ask ft:mark() to put one > marker > > for the whole match? > > > > The case I am using it for is to get the sequence of matching words > within a > > match, and sometimes, for long strings, there may be several sequences. A > > contiguous sequence of marked elements maybe assumed to make up a match, > > while non contiguous do not. One marker around the match solves the > problem, > > as would detecting characters that are never marked. > > > > Regards > > Lars G Johnsen > > National Library of Norway >