zs

commit f98e35d1336421d251ad29089a23f77125aabc53
parent 931341e7b92f25527dc5894fafb8a6ff4c645ff5
Author: Christoph Lohmann <20h@r-36.net>
Date:   Thu, 27 Mar 2014 17:24:40 +0100

Changes:

* Add a way to handle the Enclosure from a group file entry too.
* Support the decoding of HTML entities in headers.
* Handle some more error.

Diffstat:
feed.py
 | 
7
+++++++
feedemail.py
 | 
4
++++
zs.py
 | 
2
++

3 files changed, 13 insertions(+), 0 deletions(-)
diff --git a/feed.py b/feed.py
@@ -130,6 +130,13 @@ def parse(astr):
 				else:
 					article["file"] = str(entry.enclosure)
 
+			if hasattr(entry, "group") and \
+					hasattr(entry.group, "content"):
+				if "url" in entry.group.content:
+					article["file"] = \
+						str(entry.group.content.\
+						attrib["file"])
+
 			# updated
 			if hasattr(entry, "updated"):
 				article["updated"] = parseiso(entry.updated)
diff --git a/feedemail.py b/feedemail.py
@@ -11,6 +11,7 @@ from email.mime.multipart import MIMEMultipart
 from email.utils import formataddr, formatdate, parseaddr
 from email.header import Header
 import time
+import lxml.html.soupparser
 
 import html2text as h2t
 h2t.UNICODE_SNOB = 1 
@@ -20,6 +21,9 @@ h2t.INLINE_LINKS = 0
 html2text = h2t.html2text
 
 def normalizeheader(hstr):
+	hstr = lxml.html.soupparser.tostring(hstr.replace("\n", " ").strip(),\
+			encoding="utf-8")
+
 	return hstr.replace("\n", " ").strip()
 
 def send(feed, to, smtphost="localhost", smtpport=None, ssl="False", \
diff --git a/zs.py b/zs.py
@@ -37,6 +37,8 @@ def run(db, selfeed=None, dryrun=False):
 				retries += 1
 		except socket.gaierror:
 			continue
+		except urllib.error.URLError:
+			continue
 		except TimeoutError:
 			continue
 		except ConnectionResetError: