X-Git-Url: https://git.sommitrealweird.co.uk/rss2maildir.git/blobdiff_plain/7517d3b016968c154fea96215def1325a25b6bdc..aa4dd6a06ba1dd47704ac7dd6bcaa062673162d1:/rss2maildir.py

diff --git a/rss2maildir.py b/rss2maildir.py
index 84a9f09..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):
@@ -562,7 +572,7 @@ class HTML2Text(HTMLParser):
         try:
             entity = unichr(int(name))
         except:
-            if entity[0] == 'x':
+            if name[0] == 'x':
                 try:
                     entity = unichr(int('0%s' %(name,), 16))
                 except:
@@ -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]:
@@ -683,23 +700,30 @@ def parse_and_deliver(maildir, url, statedir):
         if item.has_key("content"):
             content = item["content"][0]["value"]
         else:
-            content = item["summary"]
+            if item.has_key("description"):
+                content = item["description"]
+            else:
+                content = u''
 
         md5sum = md5.md5(content.encode("utf-8")).hexdigest()
 
         prevmessageid = None
 
+        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"):
-            if db.has_key(url + "|" + item["guid"]):
-                data = db[url + "|" + item["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)
                 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]
@@ -723,8 +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() \
@@ -735,11 +759,13 @@ 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"].encode("utf-8")
+        title = item["title"]
         title = re.sub(u'<', u'&lt;', title)
         title = re.sub(u'>', u'&gt;', 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 +813,25 @@ 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 = []