minor refactor of code for adding PY3 to paths, resolves #880

nltk/compat.py

@@ -372,20 +372,22 @@ else:
                         "taggers/maxent_treebank_pos_tagger",
                         "tokenizers/punkt"]
 
+def add_py3_data(path):
+    if PY3:
+        for item in _PY3_DATA_UPDATES:
+            if item in str(path) and "/PY3" not in str(path):
+                pos = path.index(item) + len(item)
+                if path[pos:pos+4] == ".zip":
+                    pos += 4
+                path = path[:pos] + "/PY3" + path[pos:]
+                break
+    return path
+
 # for use in adding /PY3 to the second (filename) argument
 # of the file pointers in data.py
 def py3_data(init_func):
     def _decorator(*args, **kwargs):
-        if PY3:
-            path = args[1]
-            for item in _PY3_DATA_UPDATES:
-                if item in str(path) and "/PY3" not in str(path):
-                    pos = path.index(item) + len(item)
-                    if path[pos:pos+4] == ".zip":
-                        pos += 4
-                    path = path[:pos] + "/PY3" + path[pos:]
-                    args = (args[0], path) + args[2:]
-                    break
+        args = (args[0], add_py3_data(args[1])) + args[2:]
         return init_func(*args, **kwargs)
     return wraps(init_func)(_decorator)
 

nltk/data.py

@@ -56,7 +56,9 @@ except ImportError:
 # this import should be more specific:
 import nltk
 
-from nltk.compat import py3_data, text_type, string_types, BytesIO, urlopen, url2pathname
+from nltk.compat import py3_data, add_py3_data
+from nltk.compat import text_type, string_types, BytesIO, urlopen, url2pathname
+
 
 ######################################################################
 # Search Path
@@ -744,7 +746,8 @@ def load(resource_url, format='auto', cache=True, verbose=False,
     :type encoding: str
     :param encoding: the encoding of the input; only used for text formats.
     """
-    resource_url=normalize_resource_url(resource_url)
+    resource_url = normalize_resource_url(resource_url)
+    resource_url = add_py3_data(resource_url)
 
     # Determine the format of the resource.
     if format == 'auto':

nltk/tag/__init__.py

@@ -77,15 +77,10 @@ from nltk.tag.senna         import SennaTagger, SennaChunkTagger, SennaNERTagger
 from nltk.tag.mapping       import tagset_mapping, map_tag
 
 from nltk.data import load
-from nltk.compat import PY3
 
 
 # Standard treebank POS tagger
-
-if PY3:
-    _POS_TAGGER = 'taggers/maxent_treebank_pos_tagger/PY3/english.pickle'
-else:
-    _POS_TAGGER = 'taggers/maxent_treebank_pos_tagger/english.pickle'
+_POS_TAGGER = 'taggers/maxent_treebank_pos_tagger/english.pickle'
 
 def pos_tag(tokens):
     """