Remove charade in favour of chardet, the first is no longer maintained

misja · Oct 28, 2016 · 7ace593 · 7ace593
1 parent 0c447b0
commit 7ace593
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 5 deletions.
diff --git a/setup.py b/setup.py
@@ -40,7 +40,7 @@ def download_jars(datapath, version=boilerpipe_version):
     },
     install_requires=[
         'JPype1',
-        'charade',
+        'chardet',
     ],
     author='Misja Hoebe',
     author_email='misja.hoebe@gmail.com',

diff --git a/src/boilerpipe/extract/__init__.py b/src/boilerpipe/extract/__init__.py
@@ -4,7 +4,7 @@
 except ImportError:
     from urllib2 import Request, urlopen
 import socket
-import charade
+import chardet
 import threading
 
 socket.setdefaulttimeout(15)
@@ -40,7 +40,7 @@ def __init__(self, extractor='DefaultExtractor', **kwargs):
             self.data   = connection.read()
             encoding    = connection.headers['content-type'].lower().split('charset=')[-1]
             if encoding.lower() == 'text/html':
-                encoding = charade.detect(self.data)['encoding']
+                encoding = chardet.detect(self.data)['encoding']
             try:
                 self.data = unicode(self.data, encoding)
             except NameError:
@@ -49,10 +49,10 @@ def __init__(self, extractor='DefaultExtractor', **kwargs):
             self.data = kwargs['html']
             try:
                 if not isinstance(self.data, unicode):
-                    self.data = unicode(self.data, charade.detect(self.data)['encoding'])
+                    self.data = unicode(self.data, chardet.detect(self.data)['encoding'])
             except NameError:
                 if not isinstance(self.data, str):
-                    self.data = self.data.decode(charade.detect(self.data)['encoding'])        
+                    self.data = self.data.decode(chardet.detect(self.data)['encoding'])
         else:
             raise Exception('No text or url provided')