* make mail messages multipart/alternative messages with a text/plain and

  text/html part
* create a seen database that logs wether or not we've seen an item before by
  using the url of the item as the key, then check the md5sum to see if we need
  to see that item anyways

diff --git a/rss2maildir.py b/rss2maildir.py
index 968c647f6190a8daaaf8d39e7e8653141c7acf5d..60bbe906ed562b1767d6c716d9ed4741891fe3ea 100755 (executable)
--- a/rss2maildir.py
+++ b/rss2maildir.py
@@ -8,7 +8,8 @@ import urllib
 
 import feedparser
 
 
 import feedparser
 

-import email
+from email.MIMEMultipart import MIMEMultipart
+from email.MIMEText import MIMEText

 
 import datetime
 import random
 
 import datetime
 import random


@@ -20,25 +21,55 @@ from optparse import OptionParser
 from ConfigParser import SafeConfigParser
 
 from base64 import b64encode
 from ConfigParser import SafeConfigParser
 
 from base64 import b64encode

+import md5
+
+import cgi
+import dbm

 
 
 def parse_and_deliver(maildir, url, statedir):
     md = mailbox.Maildir(maildir)
     fp = feedparser.parse(url)
 
 
 def parse_and_deliver(maildir, url, statedir):
     md = mailbox.Maildir(maildir)
     fp = feedparser.parse(url)

+    db = dbm.open(os.path.join(statedir, "seen"), "c")

     for item in fp["items"]:
     for item in fp["items"]:

-        # things that we need in the message
-        msg = email.message_from_string("")
+        # have we seen it before?
+        # need to work out what the content is first...
+
+        if item.has_key("content"):
+            content = item["content"][0]["value"]
+        else:
+            content = item["summary"]
+
+        md5sum = md5.md5(content.encode("utf8")).hexdigest()
+
+        if db.has_key(item["link"]):
+            data = db[item["link"]]
+            data = cgi.parse_qs(data)
+            if data["contentmd5"][0] == md5sum:
+                continue
+
+        try:
+            author = item["author"]
+        except:
+            author = url
+
+        # create a basic email message
+        msg = MIMEMultipart("alternative")
+        messageid = "<" + datetime.datetime.now().strftime("%Y%m%d%H%M") + "." + "".join([random.choice(string.ascii_letters + string.digits) for a in range(0,6)]) + "@" + socket.gethostname() + ">"
+        msg.add_header("Message-ID", messageid)

         msg.set_unixfrom("\"%s\" <rss2maildir@localhost>" %(url))
         msg.set_unixfrom("\"%s\" <rss2maildir@localhost>" %(url))

-        msg.add_header("From", "\"%s\" <rss2maildir@localhost>" %(item["author"]))
+        msg.add_header("From", "\"%s\" <rss2maildir@localhost>" %(author))

         msg.add_header("To", "\"%s\" <rss2maildir@localhost>" %(url))
         msg.add_header("To", "\"%s\" <rss2maildir@localhost>" %(url))

-        msg.add_header("Date", datetime.datetime(*item["created_parsed"][0:6]).strftime("%a, %e %b %Y %T -0000"))
+        createddate = datetime.datetime(*item["updated_parsed"][0:6]).strftime("%a, %e %b %Y %T -0000")
+        msg.add_header("Date", createddate)

         msg.add_header("Subject", item["title"])
         msg.add_header("Subject", item["title"])

-        msg.set_charset("utf8")
-        msg.set_default_type(item["content"][0]["type"])
-        msg.set_type(item["content"][0]["type"])
-        msg.set_payload(b64encode(item["content"][0]["value"]))
+        msg.set_default_type("text/plain")

 
 

-        msg.add_header("Message-ID", "<" + datetime.datetime.now().strftime("%Y%m%d%H%M") + "." + "".join([random.choice(string.ascii_letters + string.digits) for a in range(0,6)]) + "@" + socket.gethostname() + ">")
+        htmlpart = MIMEText(content.encode("utf8"), "html", "utf8")
+        textpart = MIMEText(content.encode("utf8"), "plain", "utf8")
+
+        msg.attach(textpart)
+        msg.attach(htmlpart)

 
         # start by working out the filename we should be writting to, we do
         # this following the normal maildir style rules
 
         # start by working out the filename we should be writting to, we do
         # this following the normal maildir style rules


@@ -52,6 +83,12 @@ def parse_and_deliver(maildir, url, statedir):
         os.link(fn, newfn)
         os.unlink(fn)
 
         os.link(fn, newfn)
         os.unlink(fn)
 

+        # now add to the database about the item
+        data = urllib.urlencode((("message-id", messageid), ("created", createddate), ("contentmd5", md5sum)))
+        db[item["link"]] = data
+
+    db.close()
+

 # first off, parse the command line arguments
 
 oparser = OptionParser()
 # first off, parse the command line arguments
 
 oparser = OptionParser()


@@ -130,6 +167,18 @@ elif scp.has_option("general", "state_dir"):
         except:
             sys.stderr.write("Couldn't create state directory %s\n" %(new_state_dir))
             sys.exit(1)
         except:
             sys.stderr.write("Couldn't create state directory %s\n" %(new_state_dir))
             sys.exit(1)

+else:
+    try:
+        mode = os.stat(state_dir)[stat.ST_MODE]
+        if not stat.S_ISDIR(mode):
+            sys.stderr.write("State directory %s is not a directory\n" %(state_dir))
+            sys.exit(1)
+    except:
+        try:
+            os.mkdir(state_dir)
+        except:
+            sys.stderr.write("State directory %s could not be created\n" %(state_dir))
+            sys.exit(1)

 
 if scp.has_option("general", "maildir_root"):
     maildir_root = scp.get("general", "maildir_root")
 
 if scp.has_option("general", "maildir_root"):
     maildir_root = scp.get("general", "maildir_root")