fixed kendra start date

[iron-blogger] / scan-feeds.py

diff --git a/scan-feeds.py b/scan-feeds.py
index bf60aefbaeec5a3b4e6b02bda4bffe89ce2f99cc..0cf58f1ea2530001c2cea1edfccc6ae4d501c964 100755 (executable)
--- a/scan-feeds.py
+++ b/scan-feeds.py
@@ -15,7 +15,7 @@ try:
 except IOError:
     log = {}
 
 except IOError:
     log = {}
 

-START = datetime.datetime(2009, 12, 21, 6)
+START = datetime.datetime(2011, 10, 24, 6)

 
 def parse_published(pub):
     return parse(pub).astimezone(tz.tzlocal()).replace(tzinfo=None)
 
 def parse_published(pub):
     return parse(pub).astimezone(tz.tzlocal()).replace(tzinfo=None)


@@ -30,6 +30,9 @@ def get_link(post):
 
 def parse_feeds(weeks, uri):
     feed = feedparser.parse(uri)
 
 def parse_feeds(weeks, uri):
     feed = feedparser.parse(uri)

+
+    print >>sys.stderr, "Parsing: %s" % uri
+

     if not feed.entries:
         print >>sys.stderr, "WARN: no entries for ", uri
     for post in feed.entries:
     if not feed.entries:
         print >>sys.stderr, "WARN: no entries for ", uri
     for post in feed.entries:


@@ -48,10 +51,16 @@ def parse_feeds(weeks, uri):
         if post['url'] not in [p['url'] for p in weeks[wn]]:
             weeks[wn].append(post)
 
         if post['url'] not in [p['url'] for p in weeks[wn]]:
             weeks[wn].append(post)
 

-for (username, u) in users.items():
-    weeks = log.setdefault(username, [])
-    for l in u['links']:
-        parse_feeds(weeks, l[2])
+if len(sys.argv) > 1:
+    for username in sys.argv[1:]:
+        weeks = log.setdefault(username, [])
+        for l in users[username]['links']:
+            parse_feeds(weeks, l[2])
+else:
+    for (username, u) in users.items():
+        weeks = log.setdefault(username, [])
+        for l in u['links']:
+            parse_feeds(weeks, l[2])

 
 with open('out/report.yml', 'w') as f:
     yaml.safe_dump(log, f)
 
 with open('out/report.yml', 'w') as f:
     yaml.safe_dump(log, f)