Merge pull request #1 from donspaulding/master

Refactor the loading of the JSON files

Author: emilssolmanis

python/referer_parser/__init__.py

@@ -1,44 +1,49 @@
 import os
 import json
-from urlparse import urlparse, parse_qsl
 
+try:
+    from urlparse import urlparse, parse_qsl
+    iteritems = lambda dikt: dikt.iteritems()
+    text_type = unicode
+except ImportError:  # urlparse was renamed urllib.parse in Python 3
+    from urllib.parse import urlparse, parse_qsl
+    iteritems = lambda dikt: dikt.items()
+    text_type = str
 
-JSON_FILE = os.path.join(os.path.dirname(__file__), 'data', 'referers.json')
-REFERERS = {}
-
-with open(JSON_FILE) as json_content:
-    for medium, conf_list in json.load(json_content).iteritems():
-        for ref, config in conf_list.iteritems():
-            make_ref = None
-            if 'parameters' in config:
-                def make_ref_params(config_dict):
-                    return {
-                        'name': ref,
-                        'params': map(unicode.lower, config_dict['parameters']),
-                        'medium': medium,
+def load_referers(json_file):
+    referers_dict = {}
+    with open(json_file) as json_content:
+        for medium, conf_list in iteritems(json.load(json_content)):
+            for referer_name, config in iteritems(conf_list):
+                params = None
+                if 'parameters' in config:
+                    params = list(map(text_type.lower, config['parameters']))
+                for domain in config['domains']:
+                    referers_dict[domain] = {
+                        'name': referer_name,
+                        'medium': medium
                     }
-                make_ref = make_ref_params
-            else:
-                make_ref = lambda _: {'name': ref, 'medium': medium}
-            for domain in config['domains']:
-                REFERERS[domain] = make_ref(config)
+                    if params is not None:
+                        referers_dict[domain]['params'] = params
+    return referers_dict
 
+JSON_FILE = os.path.join(os.path.dirname(__file__), 'data', 'referers.json')
+REFERERS = load_referers(JSON_FILE)
 
 class Referer(object):
-    def __init__(self, ref_url, curr_url=None):
+    def __init__(self, ref_url, curr_url=None, referers=REFERERS):
         self.known = False
         self.referer = None
         self.medium = 'unknown'
         self.search_parameter = None
         self.search_term = None
+        self.referers = referers
 
         ref_uri = urlparse(ref_url)
         ref_host = ref_uri.hostname
         self.known = ref_uri.scheme in {'http', 'https'}
         self.uri = ref_uri
 
-        # print "Scheme: %s" % ref_uri.scheme
-
         if not self.known:
             return
 
@@ -49,48 +54,47 @@ def __init__(self, ref_url, curr_url=None):
                 self.medium = 'internal'
                 return
 
-        # print "Getting referer with path"
-        referer = self.__lookup_referer(ref_host, ref_uri.path, True)
-        # print "Got %s" % referer
+        referer = self._lookup_referer(ref_host, ref_uri.path, True)
         if not referer:
-            # print "Getting referer without path"
-            referer = self.__lookup_referer(ref_host, ref_uri.path, False)
-            # print "Got %s" % referer
+            referer = self._lookup_referer(ref_host, ref_uri.path, False)
             if not referer:
                 self.medium = 'unknown'
                 return
 
-        # print "Assigning name %s" % referer['name']
         self.referer = referer['name']
         self.medium = referer['medium']
 
         if referer['medium'] == 'search':
             if 'params' not in referer or not referer['params']:
-                # print "Returning"
                 return
             for param, val in parse_qsl(ref_uri.query):
                 if param.lower() in referer['params']:
                     self.search_parameter = param
                     self.search_term = val
 
-    def __lookup_referer(self, ref_host, ref_path, include_path):
+    def _lookup_referer(self, ref_host, ref_path, include_path):
         referer = None
         try:
-            referer = REFERERS[ref_host + ref_path] if include_path else REFERERS[ref_host]
+            if include_path:
+                referer = self.referers[ref_host + ref_path]
+            else:
+                referer = self.referers[ref_host]
         except KeyError:
             if include_path:
                 path_parts = ref_path.split('/')
                 if len(path_parts) > 1:
                     try:
-                        referer = REFERERS[ref_host + '/' + path_parts[1]]
+                        referer = self.referers[ref_host + '/' + path_parts[1]]
                     except KeyError:
                         pass
         if not referer:
             try:
                 idx = ref_host.index('.')
-                return self.__lookup_referer(ref_host[idx + 1:], ref_path, include_path)
+                return self._lookup_referer(
+                    ref_host[idx + 1:],
+                    ref_path, include_path
+                )
             except ValueError:
                 return None
         else:
             return referer
-        

python/test_referer_parser.py renamed to python/referer_parser/test/__init__.py

@@ -0,0 +1,13 @@
+# Tox (http://tox.testrun.org/) is a tool for running tests
+# in multiple virtualenvs. This configuration file will run the
+# test suite on all supported python versions. To use it, "pip install tox"
+# and then run "tox" from this directory.
+
+[tox]
+#envlist = py27, py33, pypy, jython
+envlist = py27
+
+[testenv]
+commands = nosetests referer_parser.test
+deps =
+    nose