dstl/baleen

Pipeline initialization fails with MongoHistory and language.OpenNLP

jbaker-dstl opened this issue · 3 comments

Trying to create the following pipeline causes the pipeline initialization to fail on language.OpenNLP.

history:
   class: uk.gov.dstl.baleen.history.mongo.MongoHistory

collectionreader:
  class: FolderReader
  folders:
  - corpus

annotators:
- language.OpenNLP
- class: gazetteer.Mongo
  collection: person_gazetteer
  valueField: name
  type: Person
- class: stats.OpenNLP
  model: en-ner-person.bin
  type: Person

consumers:
- Mongo

Problem isn't specific to language.OpenNLP - MongoHistory appears to fail on the first annotator regardless.

Problem is a change made in 2.2.0 where we cast the parameters of some configuration information into Strings, and thereby lose the resource that MongoHistory is dependent on (i.e. SharedMongoResource). Looking into the best way to fix this, and it will be addressed in the next release.

Fixed in 2.3.0-SNAPSHOT