htmlOffsetAdjust and matchText enabled gets StringIndexOutOfBoundsException about solrtexttagger HOT 7 CLOSED

Thanks for reporting... I didn't get a chance to look today but will hopefully look closer tomorrow (Friday).

from solrtexttagger.

Thanks, i guess ignoring wrong offset correction might also one solution

              if(offsetPair[0]>=0 && offsetPair[1]>=0 ) {
            	  startOffset = offsetPair[0];
              	  endOffset = offsetPair[1];
              }

from solrtexttagger.

I'd rather not put band-aids on the consumption end; I'd rather have reasonable offsets at the source end. Thus -- fix the bug. Thanks for sample data... I ran it through a debugger and it appears we need to call startTag.isSyntacticalEmptyElementTag(). It's not completely clear if testing this actually requires more than one tag; if not you can leave assertXmlTag alone. Can you work on a fix with test?

from solrtexttagger.

Thanks applied fix. change to asertXmlTag is not required, i was trying to add testcase according to exact scenario i had encountered but this fix does not need it.
updated pull request.

from solrtexttagger.

is there any reason, html offset corrector have not been migrated to apache solr? I was planning to move everything solr 7.4 so trying to find if it has any missing feature.

from solrtexttagger.

I wasn't quite comfortable with adding yet another dependency for an optional feature of a fringe feature of everyone using Solr. And I have doubts about the overall approach... it works, isn't bad but it'd be really nice if (somehow) the HtmlStripCharFilterFactory (or derivative) analysis component handled this sort of thing instead of the tagger itself needing to have such a feature, which feels a bit bolted on. I don't have the bandwidth to tackle redoing it though.
#28 (comment)

from solrtexttagger.

BTW a TODO in this project is to extend the hander shipping in Solr 7.4 to provide the htmlOffsetAdjust option. I had this in mind when I ported it to Solr so that this feature could be added separately. Perhaps that would be a separate branch which only provides that feature, tests, that, and does nothing else.

from solrtexttagger.