From 0e69a6f3d3b095e89eff414cd951b9702fe48452 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Andr=C3=A9=20Luiz=20de=20Oliveira=20Ramos?=
 <andrel.ramos97@gmail.com>
Date: Sat, 17 Sep 2016 19:20:12 -0300
Subject: [PATCH] Extracting topics from external blogs

---
 .gitignore |  1 +
 admin.py   | 66 ++++++++++++++++++++++++++++++++++++++++++++++++------
 models.py  | 11 ++++-----
 utils.py   |  4 ++--
 4 files changed, 68 insertions(+), 14 deletions(-)

diff --git a/.gitignore b/.gitignore
index 0d20b64..3179668 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1 +1,2 @@
 *.pyc
+.idea/
\ No newline at end of file
diff --git a/admin.py b/admin.py
index dc23b56..7a0f41e 100644
--- a/admin.py
+++ b/admin.py
@@ -1,24 +1,76 @@
 # coding : utf -8
+import feedparser
+from flask_admin.actions import action
 from quokka import admin
 from quokka.utils.translation import _, _l
 from quokka.core.admin.models import ModelAdmin
 from .models import ExternalBlogs, AggregatedTopic
+from quokka.modules.posts.models import Post
 
 
 class ExternalBlogsAdmin(ModelAdmin):
-    roles_accepted = ('admin', 'editor')
+    roles_accepted = ('admin')
     column_filters = ('name', 'root_url')
     column_searchable_list = ('name', 'root_url', 'feeds_url')
-    column_list = ('name', 'root_url', 'feeds_url')
-    form_columns = ('name', 'root_url', 'feeds_url')
+    column_list = ('name', 'root_url', 'feeds_url', 'channel')
+    form_columns = ('name', 'root_url', 'feeds_url', 'channel')
+
+    __content_format_dict = {
+        'text/plain': 'plaintext',
+        'plaintext': 'plaintext',
+        'text/html': 'html',
+        'html': 'html',
+        'text/markdown': 'markdown',
+        'markdown': 'markdown',
+    }
+
+    @action('get_external_posts', _l('Get external topics'))
+    def get_external_posts(self, ids):
+        blogs = ExternalBlogs.objects(id__in=ids)
+
+        for blog in blogs:
+            feed = feedparser.parse(blog.feeds_url)
+
+            for entry in feed['entries']:
+
+                # If already exist a topic with this url
+                if AggregatedTopic.objects(original_url=entry['feedburner_origlink']).first():
+                    continue # Go to next iteration
+
+                body = '{sumary}... <a href={link}>Continue reading</a>'.format(
+                    sumary=entry['summary'].encode('ascii', 'ignore'),
+                    link=entry['feedburner_origlink'].encode('ascii', 'ignore')
+                )
+                content_format = self.__content_format_dict[entry['content'][0]['type']]
+
+                post = {
+                    "title": entry['title'].encode('ascii', 'ignore'),
+                    "slug": entry['title'].replace(' ', '-').encode('ascii', 'ignore'),
+                    "summary": entry['summary'],
+                    "content_format": content_format,
+                    "body": body,
+                    "channel": blog.channel,
+                    "tags": [tag['term'].encode('ascii', 'ignore') for tag in entry['tags']],
+                    "published": True
+                }
+                post, _ = Post.objects.get_or_create(**post)
+
+                AggregatedTopic(
+                    original_url=entry['feedburner_origlink'].encode('ascii', 'ignore'),
+                    blog=blog,
+                    post=post
+                ).save()
 
 
 class AggregatedTopicAdmin(ModelAdmin):
     roles_accepted = ('admin', 'editor')
-    column_filters = ('title', 'date', 'blog')
-    column_searchable_list = ('title', 'topic_url')
-    column_list = ('title', 'date', 'blog', 'topic_url', 'description')
-    form_columns = ('title', 'date', 'blog', 'topic_url', 'description')
+    column_filters = ('date_added', 'original_url', 'blog', 'post')
+    column_searchable_list = ['original_url']
+    column_list = ('date_added', 'blog', 'original_url', 'post')
+    form_columns = ('date_added', 'blog', 'original_url', 'post')
+
+    def on_model_delete(self, topic):
+        topic.post.delete()
 
 # Register RSSaggregator models to quokka admin page
 admin.register(ExternalBlogs, ExternalBlogsAdmin, category=_("RSSaggregator"), name=_l("External Blogs"))
diff --git a/models.py b/models.py
index 8cdf304..ccc2a1c 100644
--- a/models.py
+++ b/models.py
@@ -1,3 +1,4 @@
+from datetime import datetime
 from quokka.core.db import db
 
 
@@ -8,6 +9,7 @@ class ExternalBlogs(db.Document):
     name = db.StringField(max_length=255, required=True)
     root_url = db.StringField(default='')
     feeds_url = db.StringField(required=True)
+    channel = db.ReferenceField('Channel', required=True, reverse_delete_rule=db.CASCADE)
 
     def __str__(self):
         return self.name
@@ -18,11 +20,10 @@ class AggregatedTopic(db.Document):
     Store topics from external blogs
     """
 
-    title = db.StringField(max_length=255, required=True)
-    description = db.StringField(required=False)
-    topic_url = db.StringField(required=True)
-    date = db.DateTimeField(required=False)
+    original_url = db.StringField(required=True)
+    date_added = db.DateTimeField(default=datetime.now())
     blog = db.ReferenceField('ExternalBlogs', required=True, reverse_delete_rule=db.CASCADE)
+    post = db.ReferenceField('Post', required=True, reverse_delete_rule=db.CASCADE)
 
     def __str__(self):
-        return self.title
+        return self.post.title
diff --git a/utils.py b/utils.py
index 6d93bad..b17cedb 100644
--- a/utils.py
+++ b/utils.py
@@ -3,10 +3,10 @@
 
 
 def get_external_topic(**kwargs):
-    '''
+    """
     :param kwargs: This function allows kwargs to give filter power for user
     :return: AggregatedTopic List
-    '''
+    """
     topics = AggregatedTopic.objects(**kwargs)
 
     return topics.order_by('date')