Merge pull request #5 from gunthercox/chatterbot_352

Support saving unescaped unicode characters to the database

Author: gunthercox

jsondb/__init__.py

@@ -1,6 +1,6 @@
 from .db import Database
 
-__version__ = "0.1.2"
-__maintainer__ = "Gunther Cox"
-__email__ = "gunthercx@gmail.com"
+__version__ = '0.1.3'
+__maintainer__ = 'Gunther Cox'
+__email__ = 'gunthercx@gmail.com'
 

jsondb/compat.py

@@ -1,10 +1,10 @@
+import codecs
 import sys
+import io
 
-'''
-Use the faster cjson library if it is available
-'''
 
 try:
+    # Use the faster cjson library if it is available
     import cjson as json
 
     json_encode = json.encode
@@ -18,17 +18,19 @@
 
 
 def encode(value):
-    return json_encode(value)
+    value = json_encode(value, ensure_ascii=False)
+    if sys.version < '3':
+        return unicode(value)
+    return value
 
 
 def decode(value):
-    return json_decode(value)
+    return json_decode(value, encoding='utf-8')
 
 
 if sys.version < '3':
 
     # Python 2 and 3 unicode string compatability
-    import codecs
     def u(x):
         return codecs.unicode_escape_decode(x)[0]
 
@@ -42,3 +44,23 @@ def u(x):
     # Dictionary iteration compatibility
     def iteritems(dictionary):
         return dictionary.items()
+
+
+def open_file_for_reading(*args, **kwargs):
+    if sys.version < '3':
+        kwargs['mode'] = 'rb+'
+    else:
+        kwargs['encoding'] = 'utf-8'
+        kwargs['mode'] = 'r+'
+
+    return io.open(*args, **kwargs)
+
+
+def open_file_for_writing(*args, **kwargs):
+    if sys.version < '3':
+        kwargs['mode'] = 'w+'
+    else:
+        kwargs['encoding'] = 'utf-8'
+        kwargs['mode'] = 'w+'
+
+    return io.open(*args, **kwargs)

jsondb/file_writer.py

@@ -1,4 +1,4 @@
-from .compat import decode, encode
+from .compat import decode, encode, open_file_for_reading, open_file_for_writing
 
 
 def read_data(file_path):
@@ -9,7 +9,7 @@ def read_data(file_path):
     if not is_valid(file_path):
         write_data(file_path, {})
 
-    db = open(file_path, "r+")
+    db = open_file_for_reading(file_path)
     content = db.read()
 
     obj = decode(content)
@@ -22,7 +22,7 @@ def write_data(path, obj):
     """
     Writes to a file and returns the updated file content.
     """
-    with open(path, "w+") as db:
+    with open_file_for_writing(path) as db:
         db.write(encode(obj))
 
     return obj