X-Git-Url: https://git.sommitrealweird.co.uk/rss2maildir.git/blobdiff_plain/8488a0a338597575257a677f34b4db2754c33927..aa4dd6a06ba1dd47704ac7dd6bcaa062673162d1:/rss2maildir.py

diff --git a/rss2maildir.py b/rss2maildir.py
index 533e34d..a9abe4d 100755
--- a/rss2maildir.py
+++ b/rss2maildir.py
@@ -39,7 +39,11 @@ from optparse import OptionParser
 from ConfigParser import SafeConfigParser
 
 from base64 import b64encode
-import md5
+
+if sys.version_info[0] == 2 and sys.version_info[1] >= 6:
+    import hashlib as md5
+else:
+    import md5
 
 import cgi
 import dbm
@@ -303,9 +307,15 @@ class HTML2Text(HTMLParser):
         url = u''
         for attr in attrs:
             if attr[0] == 'alt':
-                alt = attr[1].decode('utf-8')
+                if isinstance(attr[1], str):
+                    alt = u'%s' %(attr[1].decode("utf-8"))
+                else:
+                    alt = attr[1]
             elif attr[0] == 'src':
-                url = attr[1].decode('utf-8')
+                if isinstance(attr[1], str):
+                    url = u'%s' %(attr[1].decode("utf-8"))
+                else:
+                    url = attr[1]
         if url:
             if alt:
                 if self.images.has_key(alt):
@@ -608,10 +618,17 @@ def open_url(method, url):
         (type, rest) = urllib.splittype(url)
         (host, path) = urllib.splithost(rest)
         (host, port) = urllib.splitport(host)
-        if port == None:
+        if type == "https":
+            if port == None:
+                port = 443
+        elif port == None:
             port = 80
         try:
-            conn = httplib.HTTPConnection("%s:%s" %(host, port))
+            conn = None
+            if type == "http":
+                conn = httplib.HTTPConnection("%s:%s" %(host, port))
+            else:
+                conn = httplib.HTTPSConnection("%s:%s" %(host, port))
             conn.request(method, path)
             response = conn.getresponse()
             if response.status in [301, 302, 303, 307]:
@@ -692,12 +709,13 @@ def parse_and_deliver(maildir, url, statedir):
 
         prevmessageid = None
 
-        db_guid_key = (url + u'|' + item["guid"]).encode("utf-8")
+        db_guid_key = None
         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"):
+            db_guid_key = (url + u'|' + item["guid"]).encode("utf-8")
             if db.has_key(db_guid_key):
                 data = db[db_guid_key]
                 data = cgi.parse_qs(data)
@@ -729,8 +747,8 @@ def parse_and_deliver(maildir, url, statedir):
                 ]) + "@" + socket.gethostname() + ">"
         msg.add_header("Message-ID", messageid)
         msg.set_unixfrom("\"%s\" <rss2maildir@localhost>" %(url))
-        msg.add_header("From", "\"%s\" <rss2maildir@localhost>" %(author))
-        msg.add_header("To", "\"%s\" <rss2maildir@localhost>" %(url))
+        msg.add_header("From", "\"%s\" <rss2maildir@localhost>" %(author.encode("utf-8")))
+        msg.add_header("To", "\"%s\" <rss2maildir@localhost>" %(url.encode("utf-8")))
         if prevmessageid:
             msg.add_header("References", prevmessageid)
         createddate = datetime.datetime.now() \
@@ -741,6 +759,8 @@ def parse_and_deliver(maildir, url, statedir):
         except:
             pass
         msg.add_header("Date", createddate)
+        msg.add_header("X-rss2maildir-rundate", datetime.datetime.now() \
+            .strftime("%a, %e %b %Y %T -0000"))
         subj_gen = HTML2Text()
         title = item["title"]
         title = re.sub(u'<', u'&lt;', title)