projects
/
rss2maildir.git
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
Entity handling fixes
[rss2maildir.git]
/
rss2maildir.py
diff --git
a/rss2maildir.py
b/rss2maildir.py
index df7236a2bcd65421c6ecf5f69a7b429421ceb773..0dfd0f04b0f035224a900ab94d73e400aa2d82e9 100755
(executable)
--- a/
rss2maildir.py
+++ b/
rss2maildir.py
@@
-520,14
+520,15
@@
class HTML2Text(HTMLParser):
def handle_entityref(self, name):
entity = name
def handle_entityref(self, name):
entity = name
- if HTML2Text.entities.has_key(name
.lower()
):
- entity = HTML2Text.entities[name
.lower()
]
+ if HTML2Text.entities.has_key(name):
+ entity = HTML2Text.entities[name]
elif name[0] == "#":
entity = unichr(int(name[1:]))
else:
entity = "&" + name + ";"
elif name[0] == "#":
entity = unichr(int(name[1:]))
else:
entity = "&" + name + ";"
- self.curdata = self.curdata + unicode(entity, "utf-8")
+ self.curdata = self.curdata + unicode(entity.encode('utf-8'), \
+ "utf-8")
def gettext(self):
self.handle_curdata()
def gettext(self):
self.handle_curdata()
@@
-682,7
+683,9
@@
def parse_and_deliver(maildir, url, statedir):
except:
pass
msg.add_header("Date", createddate)
except:
pass
msg.add_header("Date", createddate)
- msg.add_header("Subject", item["title"])
+ subj_gen = HTML2Text()
+ subj_gen.feed(item["title"].encod("utf-8"))
+ msg.add_header("Subject", subj_gen.gettext())
msg.set_default_type("text/plain")
htmlcontent = content.encode("utf-8")
msg.set_default_type("text/plain")
htmlcontent = content.encode("utf-8")