Tweak extractor API

Only allow classes as extractors, and pass a config-dictionary to them. This allows for per-extension configuration.

README.rst

@@ -240,8 +240,8 @@ in ``setup.py``::
           """
           ...)
 
-Note - the registered extractor may be a class or an instance of
-an ``Extractor`` deriving object.
+Note - the registered extractor must be a class derived from the ``Extractor``
+base class.
 
 After installing ``mypackage`` the custom extractor gets executed.
 

setup.py

@@ -66,6 +66,7 @@ def run_tests(self):
       [console_scripts]
       polint = lingua.polint:main
       pot-create = lingua.extract:main
+
       [lingua.extractors]
       python = lingua.extractors.python:PythonExtractor
       xml = lingua.extractors.xml:XMLExtractor

src/lingua/extract.py

@@ -149,7 +149,9 @@ def read_config(filename):
                     file=sys.stderr)
             sys.exit(1)
         try:
-            EXTENSIONS[extension] = EXTRACTORS[plugin]
+            config = dict(config.items(section))
+            config.pop('plugin')
+            EXTENSIONS[extension] = EXTRACTORS[plugin](config)
         except KeyError:
             print('Unknown plugin %s. Check --list-plugins for available options' % plugin,
                     file=sys.stderr)

src/lingua/extractors/__init__.py

@@ -2,7 +2,6 @@
 from pkg_resources import working_set
 import abc
 import collections
-import inspect
 import os
 import re
 import sys
@@ -115,6 +114,8 @@ def update_keywords(keywords, specs):
 
 @add_metaclass(abc.ABCMeta)
 class Extractor(object):
+    def __init__(self, config=None):
+        self.config = config if config else {}
 
     @abc.abstractproperty
     def extensions(self):
@@ -128,11 +129,9 @@ def __call__(self, filename, options):
 def register_extractors():
     for entry_point in working_set.iter_entry_points('lingua.extractors'):
         extractor = entry_point.load(require=True)
-        if inspect.isclass(extractor):
-            extractor = extractor()
-        if not isinstance(extractor, Extractor):
+        if not issubclass(extractor, Extractor):
             raise ValueError(
                 u'Registered extractor must derive from ``Extractor``')
         EXTRACTORS[entry_point.name] = extractor
         for extension in extractor.extensions:
-            EXTENSIONS[extension] = extractor
+            EXTENSIONS[extension] = extractor()

src/lingua/extractors/xml.py

@@ -69,9 +69,6 @@ class XMLExtractor(Extractor, ElementProgram):
     extensions = ['.pt', '.zpt']
     DEFAULT_NAMESPACES = MacroProgram.DEFAULT_NAMESPACES
 
-    def __init__(self):
-        pass
-
     def __call__(self, filename, options):
         self.options = options
         self.filename = filename
@@ -86,7 +83,7 @@ def __call__(self, filename, options):
             print('Aborting due to parse error in %s: %s' %
                     (self.filename, e), file=sys.stderr)
             sys.exit(1)
-        super(XMLExtractor, self).__init__(source, filename=filename)
+        ElementProgram.__init__(self, source, filename=filename)
         return self.messages
 
     def visit(self, kind, args):