cleanit 0.2

HISTORY.rst

@@ -1,6 +1,11 @@
 Changelog
 ---------
 
+0.2
+^^^^^
+**release date:** 2016-02-27
+* Removing chardet and python-magic dependencies. Either encoding is specified or it should be guessed by pysrt
+
 0.1
 ^^^^^
 **release date:** 2015-10-16

cleanit/__init__.py

@@ -1,3 +1,3 @@
 __title__ = 'cleanit'
 __author__ = 'Rato'
-__version__ = '0.1.1'
+__version__ = '0.2'

cleanit/cli.py

@@ -38,14 +38,17 @@ def cleanit(config, force, test, debug, verbose, path):
         click.echo('Discarded %s' % discarded_paths, color='red')
 
     click.echo('Collected %d subtitles' % len(collected_subtitles), color='green')
-    for sub in collected_subtitles:
+    for i in reversed(range(len(collected_subtitles))):
+        sub = collected_subtitles[i]
         modified = api.clean_subtitle(sub, cfg.rules)
         if (modified or force) and not test:
             click.echo("Saving '%s'" % sub.path, color='green')
             api.save_subtitle(sub)
             click.echo("Saved '%s'" % sub.path, color='green')
         elif verbose > 0:
             click.echo("No modification for '%s'" % sub.path, color='green')
+        # to free up memory
+        del collected_subtitles[i]
 
 
 def scan(path, collected, discarded):

cleanit/subtitle.py

@@ -1,7 +1,5 @@
 # -*- coding: utf-8 -*-
-import chardet
 import logging
-import magic
 import pysrt
 
 
@@ -19,46 +17,11 @@ def __repr__(self):
         return '<%s [%s]>' % (self.__class__.__name__, self.path)
 
     def load(self):
-        if not self.encoding:
-            self.encoding = self.guess_encoding()
         self.subtitle = pysrt.open(self.path, encoding=self.encoding)
 
     def save(self, path=None, encoding=None):
         self.subtitle.save(path=path if path else self.path, encoding=encoding if encoding else self.encoding)
 
-    def guess_encoding(self):
-        filename = self.path.decode('utf-8')
-
-        # Unfortunately there are several magic modules...
-        encoding = None
-
-        # Debian/Ubuntu python-magic
-        if hasattr(magic.Magic, 'file'):
-            m = magic.open(magic.MAGIC_MIME_ENCODING)
-            m.load()
-            encoding = m.file(filename)
-            logger.debug("Guessed encoding '%s' for '%s' using debian's python-magic" % (encoding, self.path))
-            m.close()
-
-        # https://pypi.python.org/pypi/python-magic
-        elif hasattr(magic.Magic, 'from_file'):
-            m = magic.Magic(mime_encoding=True)
-            encoding = m.from_file(filename)
-            logger.debug("Guessed encoding '%s' for '%s' using pypi's python-magic" % (encoding, self.path))
-
-        # https://pypi.python.org/pypi/filemagic
-        elif hasattr(magic.Magic, 'id_filename'):
-            with magic.Magic(flags=magic.MAGIC_MIME_ENCODING) as m:
-                encoding = m.id_filename(filename)
-            logger.debug("Guessed encoding '%s' for '%s' using pypi's filemagic" % (encoding, self.path))
-
-        if not encoding or encoding in ['unknown-8bit']:
-            with open(filename, 'rb') as f:
-                encoding = chardet.detect(f.read())['encoding']
-            logger.debug("Guessed encoding '%s' for '%s' using chardet" % (encoding, self.path))
-
-        return encoding
-
     def clean(self, rules, clean_indexes=True):
         self.load()
 

setup.py

@@ -27,8 +27,7 @@ def run_tests(self):
         sys.exit(errno)
 
 # requirements
-install_requirements = ['appdirs>=1.4.0', 'chardet>=2.3.0', 'click>=4.0', 'jsonschema>=2.5.1', 'pysrt>=1.0.1',
-                        'python-magic>=0.4.6', 'pyyaml>=3.11']
+install_requirements = ['appdirs>=1.4.0', 'click>=4.0', 'jsonschema>=2.5.1', 'pysrt>=1.0.1', 'pyyaml>=3.11']
 
 test_requirements = ['pytest', 'pytest-pep8', 'pytest-flakes', 'pytest-cov']