Modified server to allow source features (#2109)

Author: anderleich

onmt/translate/translation_server.py

@@ -14,6 +14,8 @@
 
 from itertools import islice, zip_longest
 from copy import deepcopy
+from collections import defaultdict
+from argparse import Namespace
 
 from onmt.constants import DefaultTokens
 from onmt.utils.logging import init_logger
@@ -22,6 +24,7 @@
 from onmt.utils.alignment import to_word_align
 from onmt.utils.parse import ArgumentParser
 from onmt.translate.translator import build_translator
+from onmt.transforms.features import InferFeatsTransform
 
 
 def critical(func):
@@ -192,6 +195,7 @@ def start(self, config_file):
                                                       {}),
                       'ct2_translate_batch_args': conf.get(
                           'ct2_translate_batch_args', {}),
+                      'features_opt': conf.get('features', None)
                       }
             kwargs = {k: v for (k, v) in kwargs.items() if v is not None}
             model_id = conf.get("id", None)
@@ -299,14 +303,16 @@ class ServerModel(object):
     def __init__(self, opt, model_id, preprocess_opt=None, tokenizer_opt=None,
                  postprocess_opt=None, custom_opt=None, load=False, timeout=-1,
                  on_timeout="to_cpu", model_root="./", ct2_model=None,
-                 ct2_translator_args=None, ct2_translate_batch_args=None):
+                 ct2_translator_args=None, ct2_translate_batch_args=None,
+                 features_opt=None):
         self.model_root = model_root
         self.opt = self.parse_opt(opt)
         self.custom_opt = custom_opt
 
         self.model_id = model_id
         self.preprocess_opt = preprocess_opt
         self.tokenizers_opt = tokenizer_opt
+        self.features_opt = features_opt
         self.postprocess_opt = postprocess_opt
         self.timeout = timeout
         self.on_timeout = on_timeout
@@ -319,6 +325,7 @@ def __init__(self, opt, model_id, preprocess_opt=None, tokenizer_opt=None,
         self.unload_timer = None
         self.user_opt = opt
         self.tokenizers = None
+        self.feats_transform = None
 
         if len(self.opt.log_file) > 0:
             log_file = os.path.join(model_root, self.opt.log_file)
@@ -361,6 +368,10 @@ def __init__(self, opt, model_id, preprocess_opt=None, tokenizer_opt=None,
                     'tgt': tokenizer
                 }
 
+        if self.features_opt is not None:
+            self.feats_transform = InferFeatsTransform(
+                Namespace(**self.features_opt))
+
         if self.postprocess_opt is not None:
             self.logger.info("Loading postprocessor")
             self.postprocessor = []
@@ -497,21 +508,27 @@ def run(self, inputs):
             # every segment becomes a dict for flexibility purposes
             seg_dict = self.maybe_preprocess(inp)
             all_preprocessed.append(seg_dict)
-            for seg, ref in zip_longest(seg_dict["seg"], seg_dict["ref"]):
+            for seg, ref, feats in zip_longest(
+                    seg_dict["seg"], seg_dict["ref"],
+                    seg_dict["src_feats"]):
                 tok = self.maybe_tokenize(seg)
                 if ref is not None:
                     ref = self.maybe_tokenize(ref, side='tgt')
-                texts.append((tok, ref))
+                inferred_feats = self.transform_feats(seg, tok, feats)
+                texts.append((tok, ref, inferred_feats))
             tail_spaces.append(whitespaces_after)
 
         empty_indices = []
         texts_to_translate, texts_ref = [], []
-        for i, (tok, ref_tok) in enumerate(texts):
+        texts_features = defaultdict(list)
+        for i, (tok, ref_tok, feats) in enumerate(texts):
             if tok == "":
                 empty_indices.append(i)
             else:
                 texts_to_translate.append(tok)
                 texts_ref.append(ref_tok)
+                for feat_name, feat_values in feats.items():
+                    texts_features[feat_name].append(feat_values)
         if any([item is None for item in texts_ref]):
             texts_ref = None
 
@@ -522,6 +539,7 @@ def run(self, inputs):
             try:
                 scores, predictions = self.translator.translate(
                     texts_to_translate,
+                    src_feats=texts_features,
                     tgt=texts_ref,
                     batch_size=len(texts_to_translate)
                     if self.opt.batch_size == 0
@@ -682,6 +700,7 @@ def maybe_preprocess(self, sequence):
             sequence["seg"] = [sequence["src"].strip()]
             sequence.pop("src")
             sequence["ref"] = [sequence.get('ref', None)]
+            sequence["src_feats"] = [sequence.get('src_feats', {})]
             sequence["n_seg"] = 1
         if self.preprocess_opt is not None:
             return self.preprocess(sequence)
@@ -702,6 +721,23 @@ def preprocess(self, sequence):
             sequence = function(sequence, self)
         return sequence
 
+    def transform_feats(self, raw_src, tok_src, feats):
+        """Apply InferFeatsTransform to features"""
+        if self.feats_transform is None:
+            return feats
+        ex = {
+            "src": tok_src.split(' '),
+            "src_original": raw_src.split(' '),
+            "src_feats": {k:  v.split(' ') for k, v in feats.items()}
+        }
+        transformed_ex = self.feats_transform.apply(ex)
+        if not transformed_ex:
+            raise Exception("Error inferring feats")
+        transformed_feats = dict()
+        for feat_name, feat_values in transformed_ex["src_feats"].items():
+            transformed_feats[feat_name] = " ".join(feat_values)
+        return transformed_feats
+
     def build_tokenizer(self, tokenizer_opt):
         """Build tokenizer described by `tokenizer_opt`."""
         if "type" not in tokenizer_opt: