From a6d9fb5bb65066887e5a7e5d44b078e722b2b002 Mon Sep 17 00:00:00 2001
From: Vimos Tan <vimostan@gmail.com>
Date: Wed, 30 Aug 2017 14:49:14 +0800
Subject: [PATCH] fix issue #1292

---
 .../tokenizer/test_customized_tokenizer.py    | 46 +++++++++++++++++++
 spacy/tokenizer.pyx                           |  3 +-
 2 files changed, 48 insertions(+), 1 deletion(-)
 create mode 100644 spacy/tests/tokenizer/test_customized_tokenizer.py

diff --git a/spacy/tests/tokenizer/test_customized_tokenizer.py b/spacy/tests/tokenizer/test_customized_tokenizer.py
new file mode 100644
index 000000000..97a7db64c
--- /dev/null
+++ b/spacy/tests/tokenizer/test_customized_tokenizer.py
@@ -0,0 +1,46 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+from ... import load
+from ...tokenizer import Tokenizer
+from ... import util
+
+import pytest
+
+
+def test_customized_tokenizer_handles_infixes():
+    def custom_tokenizer(nlp_model):
+        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)
+
+        # infix_re = re.compile(ur'[\[\]!&:,()]')
+
+        tokenizer = Tokenizer(nlp_model.vocab,
+                              nlp_model.Defaults.tokenizer_exceptions,
+                              prefix_re.search,
+                              suffix_re.search,
+                              infix_re.finditer,
+                              token_match=None)
+        return lambda text: tokenizer(text)
+
+    nlp = load('en', create_make_doc=custom_tokenizer)
+
+    sentence = "The 8 and 10-county definitions are not used for the greater Southern California Megaregion."
+    context = [word.text for word in nlp(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 nlp(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'.']
diff --git a/spacy/tokenizer.pyx b/spacy/tokenizer.pyx
index 276f0ef20..799e4bdaa 100644
--- a/spacy/tokenizer.pyx
+++ b/spacy/tokenizer.pyx
@@ -312,7 +312,8 @@ cdef class Tokenizer:
 
                         start = infix_end
                     span = string[start:]
-                    tokens.push_back(self.vocab.get(tokens.mem, span), False)
+                    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():
             lexeme = deref(it)