Merged in jaesivsm/pyaggr3g470r (pull request #11)

misc improvement in http crawler

13 files changed, 71 insertions, 49 deletions

diff --git a/pyaggr3g470r/controllers/abstract.py b/pyaggr3g470r/controllers/abstract.py
index 9a9004af..95f9e211 100644
--- a/pyaggr3g470r/controllers/abstract.py
+++ b/pyaggr3g470r/controllers/abstract.py
@@ -70,7 +70,9 @@ class AbstractController(object):
     def create(self, **attrs):
         assert self._user_id_key in attrs or self.user_id is not None, \
                 "You must provide user_id one way or another"
-        attrs[self._user_id_key] = self.user_id or attrs.get(self._user_id_key)
+
+        if self._user_id_key not in attrs:
+            attrs[self._user_id_key] = self.user_id
         obj = self._db_cls(**attrs)
         db.session.add(obj)
         db.session.commit()
diff --git a/pyaggr3g470r/controllers/article.py b/pyaggr3g470r/controllers/article.py
index 0ec53a2f..bcd73e99 100644
--- a/pyaggr3g470r/controllers/article.py
+++ b/pyaggr3g470r/controllers/article.py
@@ -1,7 +1,6 @@
 from sqlalchemy import func
 
 from bootstrap import db
-import conf
 from .abstract import AbstractController
 from pyaggr3g470r.models import Article
 
@@ -15,13 +14,6 @@ class ArticleController(AbstractController):
             self.update({'id': article.id}, {'readed': True})
         return article
 
-    def delete(self, obj_id):
-        obj = super(ArticleController, self).delete(obj_id)
-        if not conf.ON_HEROKU:
-            import pyaggr3g470r.search as fastsearch
-            fastsearch.delete_article(self.user_id, obj.feed_id, obj_id)
-        return obj
-
     def challenge(self, ids):
         """Will return each id that wasn't found in the database."""
         for id_ in ids:
diff --git a/pyaggr3g470r/controllers/feed.py b/pyaggr3g470r/controllers/feed.py
index 8db279ae..82714e39 100644
--- a/pyaggr3g470r/controllers/feed.py
+++ b/pyaggr3g470r/controllers/feed.py
@@ -27,8 +27,9 @@ from .abstract import AbstractController
 from pyaggr3g470r.models import Feed
 
 logger = logging.getLogger(__name__)
-DEFAULT_MAX_ERROR = conf.DEFAULT_MAX_ERROR
 DEFAULT_LIMIT = 5
+DEFAULT_REFRESH_RATE = 60
+DEFAULT_MAX_ERROR = conf.DEFAULT_MAX_ERROR
 
 
 class FeedController(AbstractController):
@@ -42,11 +43,10 @@ class FeedController(AbstractController):
                                 .order_by('Feed.last_retrieved')
                                 .limit(limit)]
 
-    def list_fetchable(self, max_error=DEFAULT_MAX_ERROR, limit=DEFAULT_LIMIT):
-        from pyaggr3g470r.controllers import UserController
+    def list_fetchable(self, max_error=DEFAULT_MAX_ERROR, limit=DEFAULT_LIMIT,
+                       refresh_rate=DEFAULT_REFRESH_RATE):
         now = datetime.now()
-        user = UserController(self.user_id).get(id=self.user_id)
-        max_last = now - timedelta(minutes=user.refresh_rate or 60)
+        max_last = now - timedelta(minutes=refresh_rate)
         feeds = self.list_late(max_last, max_error, limit)
         if feeds:
             self.update({'id__in': [feed.id for feed in feeds]},
diff --git a/pyaggr3g470r/lib/crawler.py b/pyaggr3g470r/lib/crawler.py
index 1cb61973..339c4b12 100644
--- a/pyaggr3g470r/lib/crawler.py
+++ b/pyaggr3g470r/lib/crawler.py
@@ -16,7 +16,6 @@ import time
 import conf
 import json
 import logging
-import requests
 import feedparser
 import dateutil.parser
 from hashlib import md5
@@ -97,6 +96,7 @@ class AbstractCrawler:
     @classmethod
     def get_counter_callback(cls):
         cls.__counter__ += 1
+
         def debump(*args, **kwargs):
             cls.__counter__ -= 1
         return debump
@@ -157,6 +157,7 @@ class PyAggUpdater(AbstractCrawler):
             content = entry['summary']
 
         return {'feed_id': self.feed['id'],
+                'user_id': self.feed['user_id'],
                 'entry_id': extract_id(entry).get('entry_id', None),
                 'link': entry.get('link', self.feed['site_link']),
                 'title': entry.get('title', 'No title'),
@@ -176,11 +177,11 @@ class PyAggUpdater(AbstractCrawler):
         for id_to_create in results:
             entry = self.to_article(
                     self.entries[tuple(sorted(id_to_create.items()))])
-            logger.warn('%r %r - creating %r - %r', self.feed['id'],
-                        self.feed['title'], entry['title'], id_to_create)
+            logger.warn('%r %r - creating %r for %r - %r', self.feed['id'],
+                        self.feed['title'], entry['title'], entry['user_id'],
+                        id_to_create)
             self.query_pyagg('post', 'article', entry)
 
-        now = datetime.now()
         logger.debug('%r %r - updating feed etag %r last_mod %r',
                      self.feed['id'], self.feed['title'],
                      self.headers.get('etag', ''),
@@ -264,8 +265,10 @@ class FeedCrawler(AbstractCrawler):
         ids, entries = [], {}
         parsed_response = feedparser.parse(response.text)
         for entry in parsed_response['entries']:
-            entries[tuple(sorted(extract_id(entry).items()))] = entry
-            ids.append(extract_id(entry))
+            entry_ids = extract_id(entry)
+            entry_ids['feed_id'] = self.feed['id']
+            entries[tuple(sorted(entry_ids.items()))] = entry
+            ids.append(entry_ids)
         logger.debug('%r %r - found %d entries %r',
                      self.feed['id'], self.feed['title'], len(ids), ids)
         future = self.query_pyagg('get', 'articles/challenge', {'ids': ids})
diff --git a/pyaggr3g470r/models/feed.py b/pyaggr3g470r/models/feed.py
index aff11460..e43045f1 100644
--- a/pyaggr3g470r/models/feed.py
+++ b/pyaggr3g470r/models/feed.py
@@ -63,6 +63,7 @@ class Feed(db.Model):
 
     def dump(self):
         return {"id": self.id,
+                "user_id": self.user_id,
                 "title": self.title,
                 "description": self.description,
                 "link": self.link,
diff --git a/pyaggr3g470r/templates/layout.html b/pyaggr3g470r/templates/layout.html
index fba29ae3..1be20f35 100644
--- a/pyaggr3g470r/templates/layout.html
+++ b/pyaggr3g470r/templates/layout.html
@@ -6,7 +6,7 @@
     <meta name="viewport" content="width=device-width, initial-scale=1.0" />
     <meta name="description" content="pyAggr3g470r is a web-based news aggregator." />
     <meta name="author" content="" />
-    <title>pyAggr3g470r{% if head_title %} - {{ head_title }}{% endif %}</title>
+    <title>pyAggr3g470r{% if head_titles %} - {{ ' - '.join(head_titles) }}{% endif %}</title>
     <link rel="shortcut icon" href="{{ url_for('static', filename='img/favicon.png') }}" />
     <!-- Bootstrap core CSS -->
     <link href="{{ url_for('static', filename='css/bootstrap.css') }}" rel="stylesheet" media="screen" />
@@ -26,7 +26,12 @@
             <span class="icon-bar"></span>
           </button>
           <a class="navbar-brand" href="{{ url_for("home") }}">pyAggr3g470r</a>
-          <span class="navbar-brand">{% if head_title %} - {{ head_title }}{% endif %}</span>
+          {% if head_titles %}
+            {% for head_title in head_titles %}
+              <span class="navbar-brand"> - </span>
+              <span class="navbar-brand">{{ head_title }}</span>
+            {% endfor %}
+          {% endif %}
         </div>
 
         <!-- Collect the nav links, forms, and other content for toggling -->
diff --git a/pyaggr3g470r/utils.py b/pyaggr3g470r/utils.py
index ea8a87bf..628703d2 100755
--- a/pyaggr3g470r/utils.py
+++ b/pyaggr3g470r/utils.py
@@ -172,7 +172,7 @@ def import_json(email, json_content):
                         description="",
                         link=feed["link"],
                         site_link=feed["site_link"],
-                        created_date=datetime.datetime.\
+                        created_date=datetime.datetime.
                             fromtimestamp(int(feed["created_date"])),
                         enabled=feed["enabled"])
         user.feeds.append(new_feed)
@@ -191,10 +191,10 @@ def import_json(email, json_content):
                                 title=article["title"],
                                 content=article["content"],
                                 readed=article["readed"],
-                                like=article["like"], \
-                                retrieved_date=datetime.datetime.\
+                                like=article["like"],
+                                retrieved_date=datetime.datetime.
                                     fromtimestamp(int(article["retrieved_date"])),
-                                date=datetime.datetime.\
+                                date=datetime.datetime.
                                     fromtimestamp(int(article["date"])),
                                 user_id=user.id,
                                 feed_id=user_feed.id)
@@ -228,7 +228,7 @@ def open_url(url):
     if conf.HTTP_PROXY == "":
         proxy = {}
     else:
-        proxy = {"http" : conf.HTTP_PROXY}
+        proxy = {"http": conf.HTTP_PROXY}
     opener = urllib.request.FancyURLopener(proxy)
     try:
         opener = urllib.request.build_opener()
@@ -236,27 +236,28 @@ def open_url(url):
         return (True, opener.open(url))
     except urllib.error.HTTPError as e:
         # server couldn't fulfill the request
-        error = (url, e.code, \
-                        http.server.BaseHTTPRequestHandler.responses[e.code][1])
+        error = (url, e.code,
+                 http.server.BaseHTTPRequestHandler.responses[e.code][1])
         return (False, error)
     except urllib.error.URLError as e:
         # failed to reach the server
         if type(e.reason) == str:
             error = (url, e.reason, e.reason)
-            #pyaggr3g470r_log.error(url + " " + e.reason)
         else:
             error = (url, e.reason.errno, e.reason.strerror)
         return (False, error)
 
+
 def clear_string(data):
     """
     Clear a string by removing HTML tags, HTML special caracters
     and consecutive white spaces (more that one).
     """
-    p = re.compile('<[^>]+>') # HTML tags
-    q = re.compile('\s') # consecutive white spaces
+    p = re.compile('<[^>]+>')  # HTML tags
+    q = re.compile('\s')  # consecutive white spaces
     return p.sub('', q.sub(' ', data))
 
+
 def load_stop_words():
     """
     Load the stop words and return them in a list.
diff --git a/pyaggr3g470r/views/api/article.py b/pyaggr3g470r/views/api/article.py
index c3ec2d34..516eef8f 100644
--- a/pyaggr3g470r/views/api/article.py
+++ b/pyaggr3g470r/views/api/article.py
@@ -11,7 +11,8 @@ from pyaggr3g470r.views.api.common import PyAggAbstractResource,\
                                           PyAggResourceMulti
 
 
-ARTICLE_ATTRS = {'feed_id': {'type': str},
+ARTICLE_ATTRS = {'user_id': {'type': int},
+                 'feed_id': {'type': int},
                  'entry_id': {'type': str},
                  'link': {'type': str},
                  'title': {'type': str},
diff --git a/pyaggr3g470r/views/api/common.py b/pyaggr3g470r/views/api/common.py
index b8477d4b..ca344c04 100644
--- a/pyaggr3g470r/views/api/common.py
+++ b/pyaggr3g470r/views/api/common.py
@@ -51,7 +51,8 @@ def authenticate(func):
             # authentication via HTTP only
             auth = request.authorization
             if auth is not None:
-                user = User.query.filter(User.nickname == auth.username).first()
+                user = User.query.filter(
+                        User.nickname == auth.username).first()
                 if user and user.check_password(auth.password) \
                         and user.activation_key == "":
                     g.user = user
@@ -61,6 +62,7 @@ def authenticate(func):
         raise Unauthorized({'WWWAuthenticate': 'Basic realm="Login Required"'})
     return wrapper
 
+
 def to_response(func):
     """Will cast results of func as a result, and try to extract
     a status_code for the Response object"""
@@ -158,7 +160,8 @@ class PyAggResourceMulti(PyAggAbstractResource):
             return [res for res in self.controller.read().limit(limit)]
         if not limit:
             return [res for res in self.controller.read(**request.json).all()]
-        return [res for res in self.controller.read(**request.json).limit(limit)]
+        return [res
+                for res in self.controller.read(**request.json).limit(limit)]
 
     def post(self):
         """creating several objects. payload should be a list of dict.
diff --git a/pyaggr3g470r/views/api/feed.py b/pyaggr3g470r/views/api/feed.py
index 7d0e2862..ad185de9 100644
--- a/pyaggr3g470r/views/api/feed.py
+++ b/pyaggr3g470r/views/api/feed.py
@@ -3,8 +3,10 @@
 
 from flask import g
 
-from pyaggr3g470r.controllers.feed import FeedController, \
-                                          DEFAULT_MAX_ERROR, DEFAULT_LIMIT
+from pyaggr3g470r.controllers.feed import (FeedController,
+                                           DEFAULT_MAX_ERROR,
+                                           DEFAULT_LIMIT,
+                                           DEFAULT_REFRESH_RATE)
 
 from pyaggr3g470r.views.api.common import PyAggAbstractResource, \
                                           PyAggResourceNew, \
@@ -41,11 +43,20 @@ class FetchableFeedAPI(PyAggAbstractResource):
     controller_cls = FeedController
     to_date = ['date', 'last_retrieved']
     attrs = {'max_error': {'type': int, 'default': DEFAULT_MAX_ERROR},
-             'limit': {'type': int, 'default': DEFAULT_LIMIT}}
+             'limit': {'type': int, 'default': DEFAULT_LIMIT},
+             'refresh_rate': {'type': int, 'default': DEFAULT_REFRESH_RATE},
+             'retreive_all': {'type': bool, 'default': False}}
 
     def get(self):
-        return [feed for feed in self.controller.list_fetchable(
-                                                    **self.reqparse_args())]
+        args = self.reqparse_args()
+        if g.user.refresh_rate:
+            args['refresh_rate'] = g.user.refresh_rate
+
+        dont_filter_by_user = args.pop('retreive_all') and g.user.is_admin()
+
+        contr = self.controller_cls() if dont_filter_by_user \
+                else self.controller
+        return [feed for feed in contr.list_fetchable(**args)]
 
 g.api.add_resource(FeedNewAPI, '/feed', endpoint='feed_new.json')
 g.api.add_resource(FeedAPI, '/feed/<int:obj_id>', endpoint='feed.json')
diff --git a/pyaggr3g470r/views/article.py b/pyaggr3g470r/views/article.py
index 08c92686..75360485 100644
--- a/pyaggr3g470r/views/article.py
+++ b/pyaggr3g470r/views/article.py
@@ -49,7 +49,7 @@ def article(article_id=None):
         next_article = article.source.articles[-1]
 
     return render_template('article.html',
-                           head_title=utils.clear_string(article.title),
+                           head_titles=[utils.clear_string(article.title)],
                            article=article,
                            previous_article=previous_article,
                            next_article=next_article)
diff --git a/pyaggr3g470r/views/feed.py b/pyaggr3g470r/views/feed.py
index 159dce64..bc75aa29 100644
--- a/pyaggr3g470r/views/feed.py
+++ b/pyaggr3g470r/views/feed.py
@@ -53,7 +53,7 @@ def feed(feed_id=None):
     elapsed = today - last_article
 
     return render_template('feed.html',
-                           head_title=utils.clear_string(feed.title),
+                           head_titles=[utils.clear_string(feed.title)],
                            feed=feed, tag_cloud=tag_cloud,
                            first_post_date=first_article,
                            end_post_date=last_article,
diff --git a/pyaggr3g470r/views/views.py b/pyaggr3g470r/views/views.py
index 9f4ef0b7..f344ad61 100644
--- a/pyaggr3g470r/views/views.py
+++ b/pyaggr3g470r/views/views.py
@@ -228,15 +228,19 @@ def signup():
             flash(gettext('Problem while sending activation email') + ': ' + str(e), 'danger')
             return redirect(url_for('home'))
 
-        flash(gettext('Your account has been created. Check your mail to confirm it.'), 'success')
+        flash(gettext('Your account has been created. '
+                      'Check your mail to confirm it.'), 'success')
         return redirect(url_for('home'))
 
     return render_template('signup.html', form=form)
 
 
-def render_home(filters=None, head_title='', page_to_render='home', **kwargs):
+def render_home(filters=None, head_titles=None,
+                page_to_render='home', **kwargs):
     if filters is None:
         filters = {}
+    if head_titles is None:
+        head_titles = []
     feed_contr = FeedController(g.user.id)
     arti_contr = ArticleController(g.user.id)
     feeds = {feed.id: feed.title for feed in feed_contr.read()}
@@ -255,8 +259,7 @@ def render_home(filters=None, head_title='', page_to_render='home', **kwargs):
         filters['readed'] = filter_ == 'read'
     if feed_id:
         filters['feed_id'] = feed_id
-        head_title = "%s%s" % (feed_contr.get(id=feed_id).title,
-                               (' - %s' % head_title) if head_title else '')
+        head_titles.append(feed_contr.get(id=feed_id).title)
 
     sort_param = {"feed": Article.title.desc(),
                   "date": Article.date.desc(),
@@ -286,7 +289,7 @@ def render_home(filters=None, head_title='', page_to_render='home', **kwargs):
     return render_template('home.html', gen_url=gen_url, feed_id=feed_id,
                            filter_=filter_, limit=limit, feeds=feeds,
                            unread=unread, articles=articles, in_error=in_error,
-                           head_title=head_title, sort_=sort_, **kwargs)
+                           head_titles=head_titles, sort_=sort_, **kwargs)
 
 
 @app.route('/')
@@ -299,7 +302,7 @@ def home():
 @app.route('/favorites')
 @login_required
 def favorites():
-    return render_home({'like': True}, gettext('Favorites'), 'favorites')
+    return render_home({'like': True}, [gettext('Favorites')], 'favorites')
 
 
 @app.route('/search', methods=['GET'])
@@ -319,7 +322,7 @@ def search():
         filters['content__like'] = "%%%s%%" % query
     if len(filters) > 1:
         filters = {"__or__": filters}
-    return render_home(filters, "%s %s" % (gettext('Search:'), query),
+    return render_home(filters, ["%s %s" % (gettext('Search:'), query)],
                        'search', search_query=query, search_title=search_title,
                        search_content=search_content)