From 0ea5af88b66f27cf905c108d8587911930051562 Mon Sep 17 00:00:00 2001
From: Matthew Honnibal <honnibal@gmail.com>
Date: Tue, 7 Apr 2015 03:45:40 +0200
Subject: [PATCH] * Add multi-word expression RegexMatcher

---
 spacy/multi_words.py | 8 ++++++++
 1 file changed, 8 insertions(+)
 create mode 100644 spacy/multi_words.py

diff --git a/spacy/multi_words.py b/spacy/multi_words.py
new file mode 100644
index 000000000..8f8cda2b9
--- /dev/null
+++ b/spacy/multi_words.py
@@ -0,0 +1,8 @@
+class RegexMerger(object):
+    def __init__(self, regexes):
+        self.regexes = regexes
+
+    def __call__(self, tokens):
+        for tag, entity_type, regex in self.regexes:
+            for m in regex.finditer(unicode(tokens)):
+                tokens.merge(m.start(), m.end(), tag, m.group(), entity_type)