Draft, halfway patch.

New patch. Almost done except test cases.

New patch. I added test case. Will commit tonight.

trunk: Committed revision 1170616.
3x: Committed revision 1170620.

Bulk close after 3.5 is released

Hi, I just installed trunk, and I'm still seeing this problem. Not sure how to provide a STR, but in snippets I'm seeing words cut in the middle rather than at their boundaries.

If it's useful, the query I'm sending is:

INFO: [] webapp=/solr path=/select/ params={
sort=score+asc&
fl=id,absolute_url,court_id,local_path,source,download_url,status,dateFiled&
hl.fl=text,caseName,westCite,docketNumber,lexisCite,court_citation_string&
f.text.hl.snippets=5&
hl=true&
q=willingness&
fq=dateFiled:{TO}&
fq=

status_exact"Non-Precedential"OR"Relating-to+orders"OR"Precedential"OR"Errata")&f.docketNumber.hl.alternateField=docketNumber&
f.docketNumber.hl.fragListBuilder=single&
f.lexisCite.hl.fragListBuilder=single&
f.caseName.hl.fragListBuilder=single&
f.westCite.hl.fragListBuilder=single&
f.court_citation_string.hl.fragListBuilder=single&

f.text.hl.alternateField=text&
f.caseName.hl.alternateField=caseName&
f.court_citation_string.hl.alternateField=court_citation_string
f.lexisCite.hl.alternateField=lexisCite&
f.westCite.hl.alternateField=westCite&
f.text.hl.maxAlternateFieldLength=500&
}

And I'm getting a snippet that contains:

...g and willingness to read with care.” Rosenau v. Unifund Corp., 539 F.3d 218, 221 (3d Cir. 2008) (internal...

You can see the first and last word are both cut off.

Are you sure that you have boundaryScanner tag in your solrconfig.xml?

<boundaryScanner name="default"
                 default="true"
                 class="solr.highlight.SimpleBoundaryScanner">
  <lst name="defaults">
    <str name="hl.bs.maxScan">10</str>
    <str name="hl.bs.chars">.,!? &#9;&#10;&#13;</str>
  </lst>
</boundaryScanner>

Oops. I suppose that would do it, huh? Is there a reason why this isn't in the config by default? Seems like one more place for a newbie to Solr (like myself) to miss a useful feature.

I'll update the wiki as well to make it clear what the BoundaryScanners are for and what a valid entry for them looks like.