add microformats2.html_to_activities(), use it in MF2/AS demo

for #31

app.py

@@ -73,13 +73,15 @@ def get(self):
     url = util.get_required_param(self, 'url')
     logging.info('Fetching %s', url)
     resp = urllib2.urlopen(url, timeout=appengine_config.HTTP_TIMEOUT)
+    if url != resp.geturl():
+      logging.info('Redirected to %s', resp.geturl())
     body = resp.read()
 
     # decode data
     if input == 'activitystreams':
       activities = json.loads(body)
-    # elif input == 'html':
-    #   activities = json.loads(body)
+    elif input == 'html':
+      activities = microformats2.html_to_activities(body, resp.geturl())
     elif input == 'json-mf2':
       activities = [microformats2.json_to_object(item)
                     for item in json.loads(body).get('items', [])]

granary/microformats2.py

@@ -9,6 +9,7 @@
 import string
 import xml.sax.saxutils
 
+from mf2py import parser
 from oauth_dropins.webutil import util
 import source
 
@@ -272,6 +273,19 @@ def json_to_object(mf2):
   return util.trim_nulls(obj)
 
 
+def html_to_activities(html, url=None):
+  """Converts a microformats2 HTML h-feed to ActivityStreams activities.
+
+  Args:
+    html: string HTML
+    url: optional string URL that HTML came from
+
+  Returns: list of ActivityStreams activity dicts
+  """
+  parsed = parser.Parser(doc=html, url=None).to_dict()
+  return [{'object': json_to_object(item)} for item in parsed.get('items', [])]
+
+
 def activities_to_html(activities):
   """Converts ActivityStreams activities to a microformats2 HTML h-feed.
 
@@ -291,7 +305,7 @@ def activities_to_html(activities):
 %s
 </body>
 </html>
-""" % '\n'.join(object_to_html(a.get('object') or a, a.get('context', {}))
+  """ % '\n'.join(object_to_html(a.get('object') or a, a.get('context', {}))
                 for a in activities)
 
 

test_app.py

@@ -71,9 +71,19 @@
 
 </body>
 </html>
+
 """
 
-ATOM = """\
+ATOM_CONTENT = """\
+  <content type="xhtml">
+  <div xmlns="http://www.w3.org/1999/xhtml">
+
+<br />
+<br />
+  %s<br />
+  
+  </div>
+  </content>
 """
 
 
@@ -97,10 +107,12 @@ def test_url_json_mf2_to_html(self):
     self.assert_equals(200, resp.status_int)
     self.assert_equals(HTML, resp.body)
 
-  # def test_url_html_to_atom(self):
-  #   self.expect_urlopen('http://my/posts.json', json.dumps(ACTIVITIES))
-  #   self.mox.ReplayAll()
-  #   resp = app.application.get_response(
-  #     '/url?url=http://my/posts.json&input=activitystreams&output=json-mf2')
-  #   self.assert_equals(200, resp.status_int)
-  #   self.assert_equals(MF2_JSON, json.loads(resp.body))
+  def test_url_html_to_atom(self):
+    self.expect_urlopen('http://my/posts.html', HTML)
+    self.mox.ReplayAll()
+
+    resp = app.application.get_response(
+      '/url?url=http://my/posts.html&input=html&output=atom')
+    self.assert_equals(200, resp.status_int)
+    self.assertIn(ATOM_CONTENT % 'foo bar', resp.body)
+    self.assertIn(ATOM_CONTENT % 'baz baj', resp.body)