Finished mp3 addition.

append.py

@@ -1,7 +1,9 @@
 import sys
 import feedparser
+import urllib2
 from urllib2 import HTTPError, URLError
 import logging
+import re
 
 def __fetch_feed(url):
     try:
@@ -16,36 +18,86 @@ def __fetch_feed(url):
         logging.debug('Network failure reason, %s.' % e.reason)
         return None
 
-def __append_mp3(entry):
+def __append(entry, suffix, append_fn):
-    latest = __fetch_feed('cmdln_mp3_in.xml')
+    latest = __fetch_feed('cmdln_%s.xml' % suffix)
     if entry.title.find(latest.title) != -1:
         logging.info('Up to date.')
         return
-    f = open('cmdln_mp3_in.xml')
+    f = open('cmdln_%s.xml' % suffix)
-    o = open('cmdln_mp3_out.xml', 'w')
+    o = open('cmdln_%s_out.xml' % suffix, 'w')
     first = False
     try:
         for line in f:
             if line.find('<item>') != -1 and not first:
-                logging.info('Insert new item.')
+                append_fn(entry, o)
                 first = True
-            else:
-                logging.info('Preserve existing item.')
             o.write(line)
     finally:
         f.close()
 
+
+def __append_mp3(entry, output):
+    (url, mime_type, size) = __enclosure(entry.enclosures, 'http://cmdln.evenflow.nl/mp3', 'mp3')
+    output.write("""        <item>
+            <title>%(title)s (Comment Line 240-949-2638)</title>
+            <link>%(link)s</link>
+            <description><![CDATA[%(description)s]]></description>
+            <pubDate>%(pubDate)s</pubDate>
+            <enclosure url="%(url)s" length="%(size)s" type="%(mime_type)s"/>
+            <guid isPermaLink="false">%(permalink)s</guid>
+        </item>
+""" % { 'title': entry.title,
+        'link': entry.link,
+        'description': __description(entry.content),
+        'pubDate' : entry.date,
+        'permalink' : __permalink(entry.title),
+        'url' : url,
+        'mime_type' : mime_type,
+        'size' : size })
+    logging.info('Insert new MP3 item.')
+
+
+def __permalink(title):
+    permalink = title.lower()
+    permalink = re.sub('-', '', permalink)
+    permalink = re.sub('[^a-z0-9]', '-', permalink)
+    permalink = re.sub('-{2,}', '-', permalink)
+    if len(permalink) > 48:
+        permalink = permalink[:48]
+    return permalink
+
+
+def __description(content):
+    description = content[0].value
+    description = re.sub('<p></p>\n', '', description)
+    description = re.sub(re.compile('License</a>.</p>.*$', re.M | re.S), 'License</a>.</p>', description)
+    description = re.sub('</p>\n', '</p>\n\n', description)
+    return re.sub('<p>View the <a', '<p>More news, commentary, and alternate feeds available at http://thecommandline.net/.  View the <a', description)
+
+
+def __enclosure(enclosures, base_url, suffix):
+    m = re.search('cmdln.net_[0-9]{4}-[0-9]{2}-[0-9]{2}', enclosures[0].href)
+    url = '%s/%s.%s' % (base_url, m.group(), suffix)
+    usock = urllib2.urlopen(url)
+    mime_type = usock.info().type
+    size =  usock.info().get('Content-Length')
+    if size is None:
+        size = 0
+    return (url, mime_type, size)
+
+
 def main():
     logging.basicConfig(level=logging.INFO,
             format='%(message)s')
-    entry = __fetch_feed('http://thecommandline.net/category/podcast/feed/')
+    #entry = __fetch_feed('http://thecommandline.net/category/podcast/feed/')
+    entry = __fetch_feed('cmdln_site.xml')
     if entry is None:
         logging.error('Failed to fetch feed.')
         sys.exit(1)
 
-    __append_mp3(entry)
+    __append(entry, 'mp3', __append_mp3)
 
-    print entry.title
 
 if __name__ == "__main__":
+    
     main()

cmdln_mp3.xml

@@ -22,24 +22,6 @@
             <description>Logo thanks to Jared Axelrod of Free Planet X.</description>
         </image>
         <atom:link href="http://idisk.me.com/cmdln/Public/cmdln_mp3.xml" rel="self" type="application/rss+xml"/>
-        <item>
-            <title>TCLP 2010-09-15 Interview: Randy Chertkow of Beatnik Turtle (Comment Line 240-949-2638)</title>
-            <link>http://thecommandline.net/2010/09/15/beatnik_turtle4/</link>
-            <description><![CDATA[<p>This is a feature cast, an episode of The Command Line Podcast.</p>
-
-<p>There is no new hacker word of the week this week.</p>
-
-<p>The feature this week is the interview I recorded with Randy Chertkow of <a href="http://beatnikturtle.com/">Beatnik Turtle</a>. If you are interested, here are the <a href="http://thecommandline.net/2007/09/05/beatnik_turtle/">2007</a>, <a href="http://thecommandline.net/2008/09/10/beatnik_turtle_2/">2008</a>, and <a href="http://thecommandline.net/2009/09/09/beatnik_turtle3/">2009</a> interviews as well. In the interview, we mention <a href="http://www.indieguide.com/">the Indie Guide web site</a>, <a href="http://emusician.com/searchresults/?ord=d&amp;terms=chertkow&amp;x=0&amp;y=0">Randy and Jason's columns at Electronic Musician</a>, and <a href="http://trackmydisc.com/">TrackMyDisc.com</a>.</p>
-
-<p>More news and commentary and alternate feeds available at http://thecommandline.net/.  View the <a href="http://thecommandline.net/wiki/2010_09_15">detailed show notes online</a>. You can grab the <a href="http://www.archive.org/details/Tclp2010-09-15InterviewRandyChertkowOfBeatnikTurtle">flac encoded audio</a> from the Internet Archive.</p>
-
-<p><a rel="license" href="http://creativecommons.org/licenses/by-sa/3.0/us/"><img style="border-width: 0px;" src="http://thecommandline.net/wp-content/uploads/cc-by-sa.png" alt="Creative Commons License" /></a></p>
-
-<p>This work is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/3.0/us/">Creative Commons Attribution-Share Alike 3.0 United States License</a>.</p>]]></description>
-            <pubDate>Wed, 15 Sep 2010 21:30:14 -0400</pubDate>
-            <enclosure url="http://cmdln.evenflow.nl/mp3/cmdln.net_2010-09-15.mp3" length="56084190" type="audio/mpeg"/>
-            <guid isPermaLink="false">tclp-20100915-interview-randy-chertkow-of-beatn</guid>
-        </item>
         <item>
             <title>TCLP 2010-09-08 Creative Commons, Legal Issue Panel from Dragon*Con 2010 (Comment Line 240-949-2638)</title>
             <link>http://thecommandline.net/2010/09/08/dragoncon_cc_4/</link>
@@ -4335,4 +4317,4 @@
             <guid isPermaLink="false">tclp-20080319-hacking-101-parallel-computing-p</guid>
         </item>
     </channel>
-</rss>
+</rss>