Issue #6: deal with relative RSS feed links (see Hacker News).

passiomatic · Mar 14, 2015 · 2a10aa9 · 2a10aa9
1 parent ad885c8
commit 2a10aa9
Show file tree

Hide file tree

Showing 4 changed files with 13 additions and 2 deletions.
diff --git a/coldsweat/frontend.py b/coldsweat/frontend.py
@@ -362,7 +362,7 @@ def feed_add_1(self):
             #return self.respond_with_template('_feed_add_wizard_1.html', locals())
 
         if not sniff_feed(response.text):
-            links = find_feed_links(response.text)
+            links = find_feed_links(response.text, base_url=self_link)
             return self.respond_with_template('_feed_add_wizard_2.html', locals())
 
         # It's a feed

diff --git a/coldsweat/markup.py b/coldsweat/markup.py
@@ -165,6 +165,7 @@ def start_base(self, attrs):
         d = dict(attrs)
         if 'href' not in d:
             return
+        # Override passed base URL
         self.base_url = d['href']
 
     def start_link(self, attrs):                
@@ -179,7 +180,6 @@ def start_link(self, attrs):
         if 'href' not in d:
             return
 
-        #@@TODO: check if it's relative URL before join
         url, title = urlparse.urljoin(self.base_url, d['href']), d['title'] if 'title' in d else u''
         self.links.append((url, title))
 

diff --git a/coldsweat/tests/discovery.py b/coldsweat/tests/discovery.py
@@ -25,6 +25,7 @@ def run_tests():
         ('discovery/html5-xhtml.html', 'http://example.com/feed'),
         ('discovery/xhtml.html', 'http://somedomain.com/articles.xml' ),
         ('discovery/html4-base.html', 'http://somedomain.com/articles.xml' ),
+        ('discovery/html4-no-base.html', 'http://example.com/articles.xml' ),
     ]
 
     for filename, expected_url in test_files:

diff --git a/coldsweat/tests/discovery/html4-no-base.html b/coldsweat/tests/discovery/html4-no-base.html
@@ -0,0 +1,10 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
+<html>
+    <head>
+        <title>Minimal HTML4 w/out BASE document</title>
+        <link rel="Alternate" type="application/rss+xml" title="RSS" href="articles.xml">
+    </head>
+    <body>
+        Content here.
+    </body>
+</html>