X-Git-Url: https://git.sommitrealweird.co.uk/rss2maildir.git/blobdiff_plain/b22a8446e1882b310b9eed73b0346c09a0eeb181..8488a0a338597575257a677f34b4db2754c33927:/rss2maildir.py diff --git a/rss2maildir.py b/rss2maildir.py index bbd9b1b..533e34d 100755 --- a/rss2maildir.py +++ b/rss2maildir.py @@ -683,23 +683,29 @@ def parse_and_deliver(maildir, url, statedir): if item.has_key("content"): content = item["content"][0]["value"] else: - content = item["description"] + if item.has_key("description"): + content = item["description"] + else: + content = u'' md5sum = md5.md5(content.encode("utf-8")).hexdigest() prevmessageid = None + db_guid_key = (url + u'|' + item["guid"]).encode("utf-8") + db_link_key = (url + u'|' + item["link"]).encode("utf-8") + # check if there's a guid too - if that exists and we match the md5, # return if item.has_key("guid"): - if db.has_key(url + "|" + item["guid"]): - data = db[url + "|" + item["guid"]] + if db.has_key(db_guid_key): + data = db[db_guid_key] data = cgi.parse_qs(data) if data["contentmd5"][0] == md5sum: continue - if db.has_key(url + "|" + item["link"]): - data = db[url + "|" + item["link"]] + if db.has_key(db_link_key): + data = db[db_link_key] data = cgi.parse_qs(data) if data.has_key("message-id"): prevmessageid = data["message-id"][0] @@ -736,10 +742,10 @@ def parse_and_deliver(maildir, url, statedir): pass msg.add_header("Date", createddate) subj_gen = HTML2Text() - title = item["title"].encode("utf-8") + title = item["title"] title = re.sub(u'<', u'<', title) title = re.sub(u'>', u'>', title) - subj_gen.feed(title) + subj_gen.feed(title.encode("utf-8")) msg.add_header("Subject", subj_gen.gettext()) msg.set_default_type("text/plain") @@ -787,25 +793,25 @@ def parse_and_deliver(maildir, url, statedir): ("created", createddate), \ ("contentmd5", md5sum) \ )) - db[url + "|" + item["guid"]] = data + db[db_guid_key] = data try: - data = db[url + "|" + item["link"]] + data = db[db_link_key] data = cgi.parse_qs(data) newdata = urllib.urlencode(( \ ("message-id", messageid), \ ("created", data["created"][0]), \ ("contentmd5", data["contentmd5"][0]) \ )) - db[url + "|" + item["link"]] = newdata + db[db_link_key] = newdata except: - db[url + "|" + item["link"]] = data + db[db_link_key] = data else: data = urllib.urlencode(( \ ("message-id", messageid), \ ("created", createddate), \ ("contentmd5", md5sum) \ )) - db[url + "|" + item["link"]] = data + db[db_link_key] = data if headers: data = []