Move SSL redirection into util.py

Author: adoxography

neam.py

@@ -1,10 +1,9 @@
 #!/usr/bin/python3
 from neam.python.neam import main as run_neam
+from neam.python.util import unverified_https_context
 
 import nltk
 
-from contextlib import contextmanager
-import ssl
 import sys
 
 NEEDED_COLLECTIONS = {
@@ -18,28 +17,21 @@ def main():
     Makes sure the NLTK dependencies have been downloaded, then passes off to
     the real NEAM script
     """
-    ensure_collections_exist(NEEDED_COLLECTIONS)
+    ensure_nltk_collections_exist(NEEDED_COLLECTIONS)
     run_neam()
 
 
-def ensure_collections_exist(collections):
+def ensure_nltk_collections_exist(collections):
     """
-    Ensures that all required collections exist on the system, and downloads
-    them if they are not present
+    Ensures that all required NLTK collections exist on the system, and
+    downloads them if they are not present
 
     :param collections: The collections that must be present - a dict of
                         folder/name pairs, corresponding to where the
                         collections should exist in the filesystem
     :see: http://www.nltk.org/_modules/nltk/downloader.html
     """
-    missing_collections = []
-
-    for collection_type, collection_names in collections.items():
-        for collection in collection_names:
-            try:
-                nltk.data.find(f"{collection_type}/{collection}")
-            except LookupError:
-                missing_collections.append(collection)
+    missing_collections = find_missing_nltk_collections(collections)
 
     if missing_collections:
         print("Downloading NLTK models...", file=sys.stderr)
@@ -49,23 +41,27 @@ def ensure_collections_exist(collections):
                 nltk.download(collection)
 
 
-@contextmanager
-def unverified_https_context():
+def find_missing_nltk_collections(collections):
     """
-    Turns off the SSL check for the duration of the function
+    Checks if a dict of NLTK collections exist on the system and returns a
+    list of collections that are missing
+
+    :param collections: The collections that must be present - a dict of
+                        folder/name pairs, corresponding to where the
+                        collections should exist in the filesystem
+    :return: A list of collections not present on the system
     """
-    try:
-        create_unverified_https_context = ssl._create_unverified_context
-        create_default_https_context = ssl.create_default_https_context
-    except AttributeError:
-        pass
-    else:
-        ssl._create_default_https_context = create_unverified_https_context
+    missing_collections = []
+
+    for collection_type, collection_names in collections.items():
+        for collection in collection_names:
+            try:
+                nltk.data.find(f"{collection_type}/{collection}")
+            except LookupError:
+                missing_collections.append(collection)
 
-    try:
-        yield
-    finally:
-        ssl._create_default_https_context = create_default_https_context
+    return missing_collections
 
 
-main()
+if __name__ == '__main__':
+    main()

neam/python/util.py

@@ -1,6 +1,40 @@
+"""
+Miscellaneous helper methods - should be refactored should this file get too
+large.
+"""
+from contextlib import contextmanager
 import re
+import ssl
+
 
 def multi_sub(correspondences, text):
-    pattern = '({})'.format('|'.join(re.escape(key) for key in correspondences.keys()))
+    """
+    Substitutes multiple keys from a dict with their values within a string
+
+    :param correspondences: A dict where the key is the string to look up and
+                            the value is the string to replace it with
+    :param text: The string on which to conduct the replacements
+    :return: The substituted string
+    """
+    options = '|'.join(re.escape(key) for key in correspondences.keys())
+    pattern = '(' + options + ')'
     return re.sub(pattern, lambda match: correspondences[match.group(0)], text)
 
+
+@contextmanager
+def unverified_https_context():
+    """
+    Turns off the SSL check for the duration of the function
+    """
+    try:
+        create_unverified_https_context = ssl._create_unverified_context
+        create_default_https_context = ssl.create_default_https_context
+    except AttributeError:
+        pass
+    else:
+        ssl._create_default_https_context = create_unverified_https_context
+
+    try:
+        yield
+    finally:
+        ssl._create_default_https_context = create_default_https_context