Merge remote-tracking branch 'origin/combo-lambo' into develop

# Conflicts:
#	combo/main.py

combo/main.py

@@ -86,8 +86,10 @@ flags.DEFINE_integer(name="batch_size", default=1,
 flags.DEFINE_boolean(name="silent", default=True,
                      help="Silent prediction to file (without printing to console).")
 flags.DEFINE_enum(name="predictor_name", default="combo-spacy",
-                  enum_values=["combo", "combo-spacy"],
-                  help="Use predictor with whitespace or spacy tokenizer.")
+                  enum_values=["combo", "combo-spacy", "combo-lambo"],
+                  help="Use predictor with whitespace, spacy or LAMBO tokenizer.")
+#flags.DEFINE_string(name="lambo_model_name", default="en",
+#                    help="LAMBO model name (if LAMBO used for segmentation).")
 flags.DEFINE_string(name="save_relation_distribution_path", default=None,
                      help="Save relation distribution to file.")
 
@@ -179,7 +181,7 @@ def _get_predictor() -> predictors.Predictor:
     )
 
     return predictors.Predictor.from_archive(
-        archive, FLAGS.predictor_name
+        archive, FLAGS.predictor_name#, extra_args={"lambo_model_name" : FLAGS.lambo_model_name}
     )
 
 

combo/predict.py

@@ -12,13 +12,14 @@ from overrides import overrides
 
 from combo import data
 from combo.data import sentence2conllu, tokens2conllu, conllu2sentence
-from combo.utils import download, graph
+from combo.utils import download, graph, lambo
 
 logger = logging.getLogger(__name__)
 
 
 @predictor.Predictor.register("combo")
 @predictor.Predictor.register("combo-spacy", constructor="with_spacy_tokenizer")
+@predictor.Predictor.register("combo-lambo", constructor="with_lambo_tokenizer")
 class COMBO(predictor.Predictor):
 
     def __init__(self,
@@ -58,7 +59,11 @@ class COMBO(predictor.Predictor):
 
     def predict(self, sentence: Union[str, List[str], List[List[str]], List[data.Sentence]]):
         if isinstance(sentence, str):
-            return self.predict_json({"sentence": sentence})
+            if isinstance(self._tokenizer,lambo.LamboTokenizer):
+                segmented = self._tokenizer.segment(sentence)
+                return self.predict(segmented)
+            else:
+                return self.predict_json({"sentence": sentence})
         elif isinstance(sentence, list):
             if len(sentence) == 0:
                 return []
@@ -239,6 +244,11 @@ class COMBO(predictor.Predictor):
     def with_spacy_tokenizer(cls, model: models.Model,
                              dataset_reader: allen_data.DatasetReader):
         return cls(model, dataset_reader, tokenizers.SpacyTokenizer())
+    
+    @classmethod
+    def with_lambo_tokenizer(cls, model: models.Model,
+                             dataset_reader: allen_data.DatasetReader, lambo_model_name : str = 'en'):
+        return cls(model, dataset_reader, lambo.LamboTokenizer(lambo_model_name))
 
     @classmethod
     def from_pretrained(cls, path: str, tokenizer=tokenizers.SpacyTokenizer(),

combo/utils/lambo.py

+from typing import List
+
+from allennlp.data.tokenizers.tokenizer import Tokenizer
+from allennlp.data.tokenizers.token_class import Token
+from lambo.segmenter.lambo import Lambo
+
+class LamboTokenizer(Tokenizer):
+
+    def __init__(self, model: str) -> None:
+        self.lambo=Lambo.get(model)
+
+    # Simple tokenisation: ignoring sentence split
+    def tokenize(self, text: str) -> List[Token]:
+        result=[]
+        document = self.lambo.segment(text)
+        for turn in document.turns:
+            for sentence in turn.sentences:
+                for token in sentence.tokens:
+                    result.append(Token(token.text))
+        return result
+    
+    # Full segmentation: divide into sentences and tokens
+    def segment(self, text: str) -> List[List[str]]:
+        result = []
+        document = self.lambo.segment(text)
+        for turn in document.turns:
+            for sentence in turn.sentences:
+                resultS=[]
+                for token in sentence.tokens:
+                    resultS.append(token.text)
+                result.append(resultS)
+        return result
\ No newline at end of file