From c42aa9a1e2ec1a9afc0f6a3fbbc402a4d05ec4f7 Mon Sep 17 00:00:00 2001
From: Cédric Bonhomme <kimble.mandel@gmail.com>
Date: Sun, 11 Aug 2013 14:13:31 +0200
Subject: It is now possible to fetch articles with feedparser through a HTTP
 proxy, for example privoxy/tor. The address of the proxy is specified in the
 configuration file.

---
 source/cfg/pyAggr3g470r.cfg-sample |  1 +
 source/conf.py                     |  1 +
 source/feedgetter.py               | 11 ++++++++---
 3 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/source/cfg/pyAggr3g470r.cfg-sample b/source/cfg/pyAggr3g470r.cfg-sample
index bc314ebe..97993533 100755
--- a/source/cfg/pyAggr3g470r.cfg-sample
+++ b/source/cfg/pyAggr3g470r.cfg-sample
@@ -14,3 +14,4 @@ password = your_password
 [misc]
 diaspora_pod = joindiaspora.com
 feed_list = ./var/feed.lst
+http_proxy = 127.0.0.1:8118
diff --git a/source/conf.py b/source/conf.py
index 0dc42170..98aa0d48 100644
--- a/source/conf.py
+++ b/source/conf.py
@@ -52,3 +52,4 @@ password =  config.get('mail','password')
 
 DIASPORA_POD = config.get('misc', 'diaspora_pod')
 FEED_LIST = config.get('misc', 'feed_list')
+HTTP_PROXY = config.get('misc', 'http_proxy')
diff --git a/source/feedgetter.py b/source/feedgetter.py
index 44fd7daa..cec74938 100755
--- a/source/feedgetter.py
+++ b/source/feedgetter.py
@@ -20,14 +20,15 @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>
 
 __author__ = "Cedric Bonhomme"
-__version__ = "$Revision: 1.6 $"
+__version__ = "$Revision: 1.7 $"
 __date__ = "$Date: 2010/09/02 $"
-__revision__ = "$Date: 2013/06/10 $"
+__revision__ = "$Date: 2013/08/11 $"
 __copyright__ = "Copyright (c) Cedric Bonhomme"
 __license__ = "GPLv3"
 
 import hashlib
 import threading
+import urllib.request
 import feedparser
 from bs4 import BeautifulSoup
 from datetime import datetime
@@ -109,7 +110,11 @@ class FeedGetter(object):
         """
         Add the articles of the feed 'a_feed' in the SQLite base.
         """
-        a_feed = feedparser.parse(feed_link)
+        if conf.HTTP_PROXY == "":
+            proxy = urllib.request.ProxyHandler({})
+        else:
+            proxy = urllib.request.ProxyHandler({"http":conf.HTTP_PROXY})
+        a_feed = feedparser.parse(feed_link, handlers = [proxy])
         if a_feed['entries'] == []:
             return
         try:
-- 
cgit