rss_feed.py: changed XML beatification a but

[webber.git] / plugins / rss_feed.py

diff --git a/plugins/rss_feed.py b/plugins/rss_feed.py
index 71c54111bd786660143ed6d5ca0c98603ced7671..a8adbec8bb068ce3e7ddd29bdd1e02a14d1d0aad 100644 (file)
--- a/plugins/rss_feed.py
+++ b/plugins/rss_feed.py
@@ -1,6 +1,6 @@
 # -*- coding: iso-8859-1 -*-
 from webber import *
-import os, datetime
+import os, datetime, re
 try:
        import PyRSS2Gen
 except ImportError:
@@ -17,8 +17,8 @@ def checkconfig(params):
                cfg.rss_file = "feed.rss"
 
 
+# Helper class needed for datetime.datetime to generate GMT timestamps
 ZERO = datetime.timedelta(0)
-
 class UTC(datetime.tzinfo):
     """UTC"""
 
@@ -50,25 +50,36 @@ def sitemap_scan(params):
        item = PyRSS2Gen.RSSItem(
                title = file["title"],
                link = full_url,
-               guid = PyRSS2Gen.Guid("%s %s" % (full_url, file["mtime"])),
+               guid = PyRSS2Gen.Guid("%s %s" % (full_url, file["mtime"]), isPermaLink=0),
                description = change,
                pubDate = datetime.datetime.fromtimestamp(file["mtime"], utc),
        )
        items.append(item)
 
 
-
 @set_hook("finish")
 def finish(params):
        rss = PyRSS2Gen.RSS2(
                title = cfg.subtitle,
                link = "http://%s" % cfg.main_url,
                description = cfg.subtitle,
-               lastBuildDate = datetime.datetime.now(),
+               lastBuildDate = datetime.datetime.now(utc),
                items = items,
        )
+       # Step one of self-reference
+       # (see http://feedvalidator.org/docs/warning/MissingAtomSelfLink.html)
+       rss.rss_attrs["xmlns:atom"] = "http://www.w3.org/2005/Atom"
+
        try:
                os.makedirs(cfg.out_dir)
        except:
                pass
-       rss.write_xml( open(os.path.join(cfg.out_dir, cfg.rss_file), "w"))
+       f = open(os.path.join(cfg.out_dir, cfg.rss_file), "w")
+       # Ugly XML beautification
+       s = rss.to_xml()
+       s = re.sub("<(?!/)", "\n<", s)
+       s = s.replace("\n\n", "\n")
+       # Step two of self-reference
+       s = s.replace('<channel>', '<channel>\n<atom:link href="http://%s/%s" rel="self" type="application/rss+xml" />' % (cfg.main_url, cfg.rss_file))
+       f.write(s[1:])
+       f.write("\n")