field validation fails for regex queries

[joelb-git]

This works fine:

$ ./lqt -i ~/.m2/repository/bt/coref/tac-2012-kb-seeded-index-22/LOCATION_ORGANIZATION_PERSON \
-q bad-field-name:value
Exception in thread "main" java.lang.RuntimeException: Invalid field names: [bad-field-name]
    at com.basistech.gcoref.utils.LuceneQueryTool.runQuery(LuceneQueryTool.java:313)
    at com.basistech.gcoref.utils.LuceneQueryTool.run(LuceneQueryTool.java:245)
    at com.basistech.gcoref.utils.LuceneQueryTool.main(LuceneQueryTool.java:597)

But this does not:

$ ./lqt -i ~/.m2/repository/bt/coref/tac-2012-kb-seeded-index-22/LOCATION_ORGANIZATION_PERSON \
-q bad-field-name:/value/
# no exception

I think this is because query.extractTerms() returns no terms when the
underlying query is regex. One possible solution is to detect this and
then try to parse out the field name ourselves. That could be tricky,
e.g. if --default-field is used, or if there are multiple boolean
clauses in the query, on different fields, etc.

[joelb-git]

This might be relevant:

http://www.gossamer-threads.com/lists/lucene/java-user/186647