diff options
| author | Victor Chahuneau <vchahune@cs.cmu.edu> | 2012-09-06 17:46:41 +0100 |
|---|---|---|
| committer | Victor Chahuneau <vchahune@cs.cmu.edu> | 2012-09-06 17:46:41 +0100 |
| commit | 8249f6445ed28c3dc902f0eb10b1f6283058c553 (patch) | |
| tree | 50a1735defa16b4af7e1de5e0e6f12e4d7c4a5c6 /python/pkg/cdec/sa/extractor.py | |
| parent | 28194c2d099b9ea039b60ac35393626ce26d326c (diff) | |
[cdec.sa] Allow sentence annotation and initial configuration
Diffstat (limited to 'python/pkg/cdec/sa/extractor.py')
| -rw-r--r-- | python/pkg/cdec/sa/extractor.py | 8 |
1 files changed, 6 insertions, 2 deletions
diff --git a/python/pkg/cdec/sa/extractor.py b/python/pkg/cdec/sa/extractor.py index 94392c30..a5ce8a68 100644 --- a/python/pkg/cdec/sa/extractor.py +++ b/python/pkg/cdec/sa/extractor.py @@ -71,10 +71,14 @@ class GrammarExtractor: sampler = cdec.sa.Sampler(300, fsarray) self.factory.configure(fsarray, edarray, sampler, scorer) + # Initialize feature definitions with configuration + for fn in cdec.sa._SA_CONFIGURE: + fn(config) def grammar(self, sentence): if isinstance(sentence, unicode): sentence = sentence.encode('utf8') - words = chain(('<s>',), sentence.split(), ('</s>',)) + words = tuple(chain(('<s>',), sentence.split(), ('</s>',))) + meta = cdec.sa.annotate(words) cnet = cdec.sa.make_lattice(words) - return self.factory.input(cnet) + return self.factory.input(cnet, meta) |