X-Git-Url: https://git.sommitrealweird.co.uk/rss2maildir.git/blobdiff_plain/e3114c366ffcca31fedc33388dd887043e7e0af7..41170825eb09fcf3b3c69c9812073727ee8a2b4e:/rss2maildir.py diff --git a/rss2maildir.py b/rss2maildir.py index 268d192..25ddb3c 100755 --- a/rss2maildir.py +++ b/rss2maildir.py @@ -88,23 +88,7 @@ class HTML2Text(HTMLParser): elif tag.lower() == "a": self.inlink = True elif tag.lower() == "br": - if self.inparagraph: - self.text = self.text \ - + u'\n'.join( \ - textwrap.wrap(self.currentparagraph, 70)) \ - + u'\n' - self.currentparagraph = "" - elif self.inblockquote: - self.text = self.text \ - + u'\n> ' \ - + u'\n> '.join( \ - [a.strip() \ - for a in textwrap.wrap(self.blockquote, 68) \ - ]) \ - + u'\n' - self.blockquote = u'' - else: - self.text = self.text + u'\n' + self.handle_br() elif tag.lower() == "blockquote": self.inblockquote = True self.text = self.text + u'\n' @@ -139,6 +123,9 @@ class HTML2Text(HTMLParser): def handle_startendtag(self, tag, attrs): if tag.lower() == "br": + self.handle_br() + + def handle_br(self): if self.inparagraph: self.text = self.text \ + u'\n'.join( \ @@ -278,9 +265,13 @@ def parse_and_deliver(maildir, url, statedir): md5sum = md5.md5(content.encode("utf-8")).hexdigest() + prevmessageid = None + if db.has_key(url + "|" + item["link"]): data = db[url + "|" + item["link"]] data = cgi.parse_qs(data) + if data.has_key("message-id"): + prevmessageid = data["message-id"][0] if data["contentmd5"][0] == md5sum: continue @@ -303,6 +294,8 @@ def parse_and_deliver(maildir, url, statedir): msg.set_unixfrom("\"%s\" " %(url)) msg.add_header("From", "\"%s\" " %(author)) msg.add_header("To", "\"%s\" " %(url)) + if prevmessageid: + msg.add_header("References", prevmessageid) createddate = datetime.datetime(*item["updated_parsed"][0:6]) \ .strftime("%a, %e %b %Y %T -0000") msg.add_header("Date", createddate) @@ -337,6 +330,8 @@ def parse_and_deliver(maildir, url, statedir): os.unlink(fn) # now add to the database about the item + if prevmessageid: + messageid = prevmessageid + " " + messageid data = urllib.urlencode(( ("message-id", messageid), \ ("created", createddate), \