From f8de403483f587e39ddd7148807d305d762b7736 Mon Sep 17 00:00:00 2001
From: Matthew Honnibal <honnibal@gmail.com>
Date: Mon, 12 Oct 2015 17:00:01 +1100
Subject: [PATCH] * Work on pickling Vocab instances. The current
 implementation is not correct, but it may serve to see whether this approach
 is workable. Pickling is necessary to address Issue #125

---
 spacy/vocab.pyx           | 12 +++++++++---
 tests/vocab/test_vocab.py | 11 +++++++++++
 2 files changed, 20 insertions(+), 3 deletions(-)

diff --git a/spacy/vocab.pyx b/spacy/vocab.pyx
index 7f07a64ba..dd6792104 100644
--- a/spacy/vocab.pyx
+++ b/spacy/vocab.pyx
@@ -99,7 +99,7 @@ cdef class Vocab:
         return self.length
 
     def __reduce__(self):
-        tmp_dir = tempfile.mkdtmp()
+        tmp_dir = tempfile.mkdtemp()
         lex_loc = path.join(tmp_dir, 'lexemes.bin')
         str_loc = path.join(tmp_dir, 'strings.txt')
         map_loc = path.join(tmp_dir, 'tag_map.json')
@@ -108,7 +108,7 @@ cdef class Vocab:
         self.strings.dump(str_loc)
         json.dump(self.morphology.tag_map, open(map_loc, 'w'))
 
-        return (Vocab.from_dir, (tmp_dir, self.get_lex_attr), None, None)
+        return (unpickle_vocab, (tmp_dir,), None, None)
 
     cdef const LexemeC* get(self, Pool mem, unicode string) except NULL:
         '''Get a pointer to a LexemeC from the lexicon, creating a new Lexeme
@@ -353,7 +353,13 @@ cdef class Vocab:
         return vec_len
 
 
-copy_reg.constructor(Vocab.from_dir)
+def unpickle_vocab(data_dir):
+    # TODO: This needs fixing --- the trouble is, we can't pickle staticmethods,
+    # so we need to fiddle with the design of Language a little bit.
+    from .language import Language
+    return Vocab.from_dir(data_dir, Language.default_lex_attrs())
+
+copy_reg.constructor(unpickle_vocab)
 
 
 def write_binary_vectors(in_loc, out_loc):
diff --git a/tests/vocab/test_vocab.py b/tests/vocab/test_vocab.py
index 153e0d546..1ab3746f3 100644
--- a/tests/vocab/test_vocab.py
+++ b/tests/vocab/test_vocab.py
@@ -1,10 +1,13 @@
 from __future__ import unicode_literals
 import pytest
+import StringIO
+import pickle
 
 from spacy.attrs import LEMMA, ORTH, PROB, IS_ALPHA
 from spacy.parts_of_speech import NOUN, VERB
 
 
+
 def test_neq(en_vocab):
     addr = en_vocab['Hello']
     assert en_vocab['bye'].orth != addr.orth
@@ -38,3 +41,11 @@ def test_symbols(en_vocab):
     assert en_vocab.strings['ORTH'] == ORTH
     assert en_vocab.strings['PROB'] == PROB
     
+
+def test_pickle_vocab(en_vocab):
+    file_ = StringIO.StringIO()
+    pickle.dump(en_vocab, file_)
+
+    file_.seek(0)
+
+    loaded = pickle.load(file_)