From 9cd50d1bdff631d03e295587c118ca593ce7f7f5 Mon Sep 17 00:00:00 2001
From: cmdln <cmdln@thecommandline.net>
Date: Sat, 25 Sep 2010 17:34:05 -0400
Subject: [PATCH] New script to find and re-link enclosures to files hosted at
 the Archive.

---
 relink.py | 63 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
 create mode 100755 relink.py

diff --git a/relink.py b/relink.py
new file mode 100755
index 0000000..2483ea6
--- /dev/null
+++ b/relink.py
@@ -0,0 +1,63 @@
+#!/usr/bin/python
+from BeautifulSoup import BeautifulStoneSoup
+import re
+import urllib2
+from urllib2 import HTTPError, URLError
+import os.path
+import logging
+
+
+
+def __repoint():
+    logging.basicConfig(level=logging.INFO,
+            format='%(message)s')
+    f = open("cmdln_m4a.xml")
+    try:
+        soup = BeautifulStoneSoup(f)
+        enclosures = soup.findAll('enclosure')
+        for enclosure in enclosures:
+            url = enclosure['url']
+            if url.find('archive.org') != -1:
+                continue
+            title = enclosure.findPreviousSibling('title')
+            rewritten = 'http://www.archive.org/download/%s/%s' % (__archive_slug(title.string),
+                os.path.basename(url))
+            (mime_type, length) =  __url_info(rewritten)
+            if mime_type is None:
+                print 'Could not find media, %s.' % rewritten
+                break
+            enclosure['url'] = rewritten
+            enclosure['type'] = mime_type
+            enclosure['length'] = length
+        print soup
+    finally:
+        f.close()
+
+
+def __archive_slug(title):
+    paren = title.find('(')
+    slug = title[:paren -1]
+    slug = re.sub('\([^0-9]\)-\([^0-9]\)', '\1\2', slug)
+    slug = re.sub('[^A-Za-z0-9\-]', ' ', slug)
+    slug = re.sub(' {2,}', ' ', slug)
+    tokens = slug.split(' ')
+    tokens = [t.capitalize() for t in tokens]
+    slug = ''.join(tokens)
+    return slug
+
+def __url_info(url):
+    try:
+        usock = urllib2.urlopen(url)
+        if usock.info() is None:
+            return (None, None)
+        return (usock.info().type, usock.info().get('Content-Length'))
+    except HTTPError, e:
+        logging.error('Failed with HTTP status code %d' % e.code)
+        return (None, None)
+    except URLError, e:
+        logging.error('Failed to connect with network.')
+        logging.debug('Network failure reason, %s.' % e.reason)
+        return (None, None)
+
+if __name__ == "__main__":
+    __repoint()