Merge pull request #1294 from Vimos/master

Fix issue #1292 and add test case for the Assertion Error

spacy/tests/tokenizer/test_customized_tokenizer.py

@@ -0,0 +1,40 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+from ...en import English
+from ...tokenizer import Tokenizer
+from ... import util
+
+import pytest
+
+@pytest.fixture
+def tokenizer(en_vocab):
+    prefix_re = util.compile_prefix_regex(nlp_model.Defaults.prefixes)
+    suffix_re = util.compile_suffix_regex(nlp_model.Defaults.suffixes)
+    custom_infixes = ['\.\.\.+',
+                      '(?<=[0-9])-(?=[0-9])',
+                      # '(?<=[0-9]+),(?=[0-9]+)',
+                      '[0-9]+(,[0-9]+)+',
+                      u'[\[\]!&:,()\*—–\/-]']
+
+    infix_re = util.compile_infix_regex(custom_infixes)
+    return Tokenizer(en_vocab,
+                     English.Defaults.tokenizer_exceptions,
+                     prefix_re.search,
+                     suffix_re.search,
+                     infix_re.finditer,
+                     token_match=None)
+
+def test_customized_tokenizer_handles_infixes(tokenizer):
+    sentence = "The 8 and 10-county definitions are not used for the greater Southern California Megaregion."
+    context = [word.text for word in tokenizer(sentence)]
+    assert context == [u'The', u'8', u'and', u'10', u'-', u'county', u'definitions', u'are', u'not', u'used',
+                       u'for',
+                       u'the', u'greater', u'Southern', u'California', u'Megaregion', u'.']
+
+    # the trailing '-' may cause Assertion Error
+    sentence = "The 8- and 10-county definitions are not used for the greater Southern California Megaregion."
+    context = [word.text for word in tokenizer(sentence)]
+    assert context == [u'The', u'8', u'-', u'and', u'10', u'-', u'county', u'definitions', u'are', u'not', u'used',
+                       u'for',
+                       u'the', u'greater', u'Southern', u'California', u'Megaregion', u'.']

spacy/tokenizer.pyx

@@ -312,6 +312,7 @@ cdef class Tokenizer:
 
                         start = infix_end
                     span = string[start:]
+                    if span:
                         tokens.push_back(self.vocab.get(tokens.mem, span), False)
         cdef vector[const LexemeC*].reverse_iterator it = suffixes.rbegin()
         while it != suffixes.rend():