From 6761835fa53b69d0197993a1bb757883f0971aed Mon Sep 17 00:00:00 2001
From: Kyle Mahan <kyle.mahan@gmail.com>
Date: Wed, 22 Apr 2015 22:07:25 +0000
Subject: [PATCH] use mf2py parsed 'alternates'

---
 woodwind/tasks.py | 22 +++++++++++++++-------
 woodwind/views.py | 10 ++++------
 2 files changed, 19 insertions(+), 13 deletions(-)

diff --git a/woodwind/tasks.py b/woodwind/tasks.py
index 6db2e61..0ead8a1 100644
--- a/woodwind/tasks.py
+++ b/woodwind/tasks.py
@@ -116,13 +116,17 @@ def update_feed(feed_id, content=None, is_polling=True):
                 if not old:
                     # set a default value for published if none is provided
                     entry.published = entry.published or now
-                    new_ids.append(entry.id)
-                    for irt in entry.get_property('in-reply-to', []):
-                        reply_pairs.append((entry.id, irt))
+                    in_reply_tos = entry.get_property('in-reply-to', [])
                     feed.entries.append(entry)
                     db.session.commit()
+
+                    new_ids.append(entry.id)
+                    for irt in in_reply_tos:
+                        reply_pairs.append((entry.id, irt))
+
                 elif not is_content_equal(old, entry):
                     entry.published = entry.published or old.published
+                    in_reply_tos = entry.get_property('in-reply-to', [])
                     # we're updating an old entriy, use the original
                     # retrieved time
                     entry.retrieved = old.retrieved
@@ -130,11 +134,13 @@ def update_feed(feed_id, content=None, is_polling=True):
                     # punt on deleting for now, learn about cascade
                     # and stuff later
                     # session.delete(old)
-                    updated_ids.append(entry.id)
-                    for irt in entry.get_property('in-reply-to', []):
-                        reply_pairs.append((entry.id, irt))
                     feed.entries.append(entry)
                     db.session.commit()
+
+                    updated_ids.append(entry.id)
+                    for irt in in_reply_tos:
+                        reply_pairs.append((entry.id, irt))
+
                 else:
                     current_app.logger.debug(
                         'skipping previously seen post %s', old.permalink)
@@ -225,7 +231,9 @@ def notify_feed_updated(app, feed, entry_ids):
     """
     from flask import render_template
     import flask.ext.login as flask_login
-
+    current_app.logger.debug(
+        'notifying feed updated for entries %r', entry_ids)
+    
     entries = Entry.query\
                    .filter(Entry.id.in_(entry_ids))\
                    .order_by(Entry.retrieved.desc(),
diff --git a/woodwind/views.py b/woodwind/views.py
index 8a24943..a392a3e 100644
--- a/woodwind/views.py
+++ b/woodwind/views.py
@@ -371,9 +371,9 @@ def find_possible_feeds(origin):
         })
 
     elif content_type == 'text/html':
+        parsed = mf2py.parse(doc=resp.text, url=origin)
         # if text/html, then parse and look for h-entries
-        hfeed = mf2util.interpret_feed(
-            mf2py.Parser(doc=resp.text).to_dict(), origin)
+        hfeed = mf2util.interpret_feed(parsed, origin)
         if hfeed.get('entries'):
             feeds.append({
                 'origin': origin,
@@ -382,13 +382,11 @@ def find_possible_feeds(origin):
             })
 
         # then look for link rel="alternate"
-        soup = bs4.BeautifulSoup(resp.text)
-        for link in soup.find_all('link', {'rel': 'alternate'}):
+        for link in parsed.get('alternates', []):
             if link.get('type') in xml_feed_types:
-                feed_url = urllib.parse.urljoin(origin, link.get('href'))
                 feeds.append({
                     'origin': origin,
-                    'feed': feed_url,
+                    'feed': link.get('url'),
                     'type': 'xml',
                 })
     return feeds