rss_feed.py: limit number of published items via "rss_max_items"

[webber.git] / plugins / rss_feed.py

diff --git a/plugins/rss_feed.py b/plugins/rss_feed.py
index 79fdae9fd5de6b22fddb31336c1c3d1f41afeea3..580d3040925ec0e07af77fd8af67eab82add60b4 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:
@@ -8,6 +8,7 @@ except ImportError:
        raise
 
 items = []
+max_age = 0
 
 
 @set_hook("checkconfig")
@@ -15,6 +16,12 @@ def checkconfig(params):
        if not cfg.has_key("rss_file"):
                log('no "rss_file:" configured, using "feed.rss":', 4)
                cfg.rss_file = "feed.rss"
+       if not cfg.has_key("rss_max_items"):
+               cfg.rss_max_items = 0
+       if cfg.has_key("rss_max_age_days"):
+               import time
+               global max_age
+               max_age = int(time.time()) - int(cfg.rss_max_age_days)*86400
 
 
 # Helper class needed for datetime.datetime to generate GMT timestamps
@@ -38,6 +45,8 @@ def sitemap_scan(params):
        global items
 
        file = params.file
+       if max_age and file["mtime"] < max_age:
+               return
        if not file.has_key("linktitle"):
                return
        if file.has_key("change"):
@@ -52,13 +61,26 @@ def sitemap_scan(params):
                link = full_url,
                guid = PyRSS2Gen.Guid("%s %s" % (full_url, file["mtime"]), isPermaLink=0),
                description = change,
-               pubDate = datetime.datetime.fromtimestamp(file["mtime"], utc),
+               pubDate = file["mtime"]
        )
        items.append(item)
 
 
 @set_hook("finish")
 def finish(params):
+       global items
+       # Sort items by pubDate, which still holds the mtime
+       items.sort(key=lambda i: i.pubDate, reverse=True)
+
+       # Limit to requested number
+       count = int(cfg.rss_max_items)
+       if count:
+               items = items[:count]
+
+       # convert mtime to real pupDate
+       for i in items:
+               i.pubDate = datetime.datetime.fromtimestamp(i.pubDate, utc)
+
        rss = PyRSS2Gen.RSS2(
                title = cfg.subtitle,
                link = "http://%s" % cfg.main_url,
@@ -76,8 +98,10 @@ def finish(params):
                pass
        f = open(os.path.join(cfg.out_dir, cfg.rss_file), "w")
        # Ugly XML beautification
-       s = rss.to_xml().replace("<", "\n<").replace("\n\n", "\n")[1:]
+       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)
+       f.write(s[1:])
        f.write("\n")