if we have training data, make sure it gets indexed by the active learner

[fgregg]

Start addressing #757, #753

[victorpoluceno]

I still see this error happening in a random way. Sometimes it works sometimes it does not. But in my case I'm using the RecordLink. This is the stack trace:

Traceback (most recent call last):
  File "/Users/victorpoluceno/go/src/github.com/FindHotel/sovitin/scripts/candidates/build/dedupe/dedupe/predicates.py", line 199, in __call__
    centers = [self.index._doc_to_id[doc]]
AttributeError: 'NoneType' object has no attribute '_doc_to_id'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "train.py", line 82, in <module>
    main(args)
  File "train.py", line 70, in main
    args.provider_code, args.model_version)
  File "train.py", line 39, in train
    sample_size=15000)
  File "/Users/victorpoluceno/go/src/github.com/FindHotel/sovitin/scripts/candidates/build/dedupe/dedupe/api.py", line 901, in sample
    self.active_learner.mark(examples, y)
  File "/Users/victorpoluceno/go/src/github.com/FindHotel/sovitin/scripts/candidates/build/dedupe/dedupe/labeler.py", line 348, in mark
    learner.fit_transform(self.pairs, self.y)
  File "/Users/victorpoluceno/go/src/github.com/FindHotel/sovitin/scripts/candidates/build/dedupe/dedupe/labeler.py", line 198, in fit_transform
    recall=1.0)
  File "/Users/victorpoluceno/go/src/github.com/FindHotel/sovitin/scripts/candidates/build/dedupe/dedupe/training.py", line 31, in learn
    dupe_cover = Cover(self.blocker.predicates, matches)
  File "/Users/victorpoluceno/go/src/github.com/FindHotel/sovitin/scripts/candidates/build/dedupe/dedupe/training.py", line 383, in __init__
    self._cover(predicates, pairs)
  File "/Users/victorpoluceno/go/src/github.com/FindHotel/sovitin/scripts/candidates/build/dedupe/dedupe/training.py", line 391, in _cover
    in enumerate(pairs)
  File "/Users/victorpoluceno/go/src/github.com/FindHotel/sovitin/scripts/candidates/build/dedupe/dedupe/training.py", line 393, in <setcomp>
    set(predicate(record_2, target=True)))}
  File "/Users/victorpoluceno/go/src/github.com/FindHotel/sovitin/scripts/candidates/build/dedupe/dedupe/predicates.py", line 203, in __call__
    raise AttributeError("Attempting to block with an index "
AttributeError: Attempting to block with an index predicate without indexing records

[victorpoluceno]

@fgregg after your latest changes, I can confirm this now works for RecordLink. 👍

Thanks for fixing this!

[quillan86]

Hi - have these changes been uploaded to pip?