X-Git-Url: https://git.sommitrealweird.co.uk/rss2maildir.git/blobdiff_plain/7dc8a4550f4f01b8c41273576de1b9618827fb96..9a4c55083ef741cd562529f93b331f4c583e61b7:/rss2maildir.py diff --git a/rss2maildir.py b/rss2maildir.py index aa3b0cc..3e4ed60 100755 --- a/rss2maildir.py +++ b/rss2maildir.py @@ -46,18 +46,17 @@ import dbm from HTMLParser import HTMLParser -entities = { - "amp": "&", - "lt": "<", - "gt": ">", - "pound": "£", - "copy": "©", - "apos": "'", - "quot": "\"", - "nbsp": " ", - } - class HTML2Text(HTMLParser): + entities = { + "amp": "&", + "lt": "<", + "gt": ">", + "pound": "£", + "copy": "©", + "apos": "'", + "quot": "\"", + "nbsp": " ", + } def __init__(self): self.inheadingone = False @@ -110,7 +109,7 @@ class HTML2Text(HTMLParser): self.item = u'' self.inul = True self.text = self.text + "\n" - elif tag.lower() == "li" and self.inul: + elif tag.lower() == "li": if not self.initem: self.initem = True self.item = u'' @@ -121,6 +120,7 @@ class HTML2Text(HTMLParser): textwrap.wrap(self.item, 67)]) \ + u'\n' self.item = u'' + self.initem = True def handle_startendtag(self, tag, attrs): if tag.lower() == "br": @@ -199,7 +199,7 @@ class HTML2Text(HTMLParser): self.inpre = False elif tag.lower() == "li": self.initem = False - if self.item != "": + if self.item != u'': self.text = self.text \ + u' * ' \ + u'\n '.join( \ @@ -218,21 +218,23 @@ class HTML2Text(HTMLParser): self.blockquote = self.blockquote \ + unicode(data, "utf-8").strip() \ + u' ' + elif self.initem: + self.item = self.item + unicode(data, "utf-8") elif self.inparagraph: self.currentparagraph = self.currentparagraph \ + unicode(data, "utf-8").strip() \ + u' ' - elif self.inul and self.initem: - self.item = self.item + unicode(data, "utf-8") elif self.inpre: self.text = self.text + unicode(data, "utf-8") else: - self.text = self.text + unicode(data, "utf-8").strip() + u' ' + isallwhitespace = data.strip() + if isallwhitespace != "" and self.text[-1] == "\n": + self.text = self.text + unicode(data, "utf-8").strip() + u' ' def handle_entityref(self, name): entity = name - if entities.has_key(name.lower()): - entity = entities[name.lower()] + if HTML2Text.entities.has_key(name.lower()): + entity = HTML2Text.entities[name.lower()] elif name[0] == "#": entity = unichr(int(name[1:])) else: @@ -250,6 +252,8 @@ class HTML2Text(HTMLParser): data = self.text if self.inparagraph: data = data + "\n".join(textwrap.wrap(self.currentparagraph, 70)) + if data[-1] != '\n': + data = data + '\n' return data def open_url(method, url): @@ -339,6 +343,15 @@ def parse_and_deliver(maildir, url, statedir): prevmessageid = None + # 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"]] + data = cgi.parse_qs(data) + if data["contentmd5"][0] == md5sum: + continue + if db.has_key(url + "|" + item["link"]): data = db[url + "|" + item["link"]] data = cgi.parse_qs(data) @@ -379,10 +392,18 @@ def parse_and_deliver(maildir, url, statedir): msg.add_header("Subject", item["title"]) msg.set_default_type("text/plain") - htmlpart = MIMEText(content.encode("utf-8"), "html", "utf-8") + htmlcontent = content.encode("utf-8") + htmlcontent = "%s\n\n

Item URL: %s

" %( \ + content, \ + item["link"], \ + item["link"] ) + htmlpart = MIMEText(htmlcontent.encode("utf-8"), "html", "utf-8") textparser = HTML2Text() textparser.feed(content.encode("utf-8")) textcontent = textparser.gettext() + textcontent = "%s\n\nItem URL: %s" %( \ + textcontent, \ + item["link"] ) textpart = MIMEText(textcontent.encode("utf-8"), "plain", "utf-8") msg.attach(textpart) msg.attach(htmlpart) @@ -409,12 +430,31 @@ def parse_and_deliver(maildir, url, statedir): # now add to the database about the item if prevmessageid: messageid = prevmessageid + " " + messageid - data = urllib.urlencode(( - ("message-id", messageid), \ - ("created", createddate), \ - ("contentmd5", md5sum) \ - )) - db[url + "|" + item["link"]] = data + if item.has_key("guid") and item["guid"] != item["link"]: + data = urllib.urlencode(( \ + ("message-id", messageid), \ + ("created", createddate), \ + ("contentmd5", md5sum) \ + )) + db[url + "|" + item["guid"]] = data + try: + data = db[url + "|" + item["link"]] + data = cgi.parse_qs(data) + newdata = urllib.urlencode(( \ + ("message-id", messageid), \ + ("created", data["created"][0]), \ + ("contentmd5", data["contentmd5"][0]) \ + )) + db[url + "|" + item["link"]] = newdata + except: + db[url + "|" + item["link"]] = data + else: + data = urllib.urlencode(( \ + ("message-id", messageid), \ + ("created", createddate), \ + ("contentmd5", md5sum) \ + )) + db[url + "|" + item["link"]] = data if headers: data = [] @@ -428,178 +468,182 @@ def parse_and_deliver(maildir, url, statedir): db.close() feeddb.close() -# first off, parse the command line arguments +if __name__ == "__main__": + # This only gets executed if we really called the program + # first off, parse the command line arguments -oparser = OptionParser() -oparser.add_option( - "-c", "--conf", dest="conf", - help="location of config file" - ) -oparser.add_option( - "-s", "--statedir", dest="statedir", - help="location of directory to store state in" - ) + oparser = OptionParser() + oparser.add_option( + "-c", "--conf", dest="conf", + help="location of config file" + ) + oparser.add_option( + "-s", "--statedir", dest="statedir", + help="location of directory to store state in" + ) -(options, args) = oparser.parse_args() + (options, args) = oparser.parse_args() -# check for the configfile + # check for the configfile -configfile = None + configfile = None -if options.conf != None: - # does the file exist? - try: - os.stat(options.conf) - configfile = options.conf - except: - # should exit here as the specified file doesn't exist - sys.stderr.write( \ - "Config file %s does not exist. Exiting.\n" %(options.conf,)) - sys.exit(2) -else: - # check through the default locations - try: - os.stat("%s/.rss2maildir.conf" %(os.environ["HOME"],)) - configfile = "%s/.rss2maildir.conf" %(os.environ["HOME"],) - except: + if options.conf != None: + # does the file exist? try: - os.stat("/etc/rss2maildir.conf") - configfile = "/etc/rss2maildir.conf" + os.stat(options.conf) + configfile = options.conf except: - sys.stderr.write("No config file found. Exiting.\n") + # should exit here as the specified file doesn't exist + sys.stderr.write( \ + "Config file %s does not exist. Exiting.\n" %(options.conf,)) sys.exit(2) + else: + # check through the default locations + try: + os.stat("%s/.rss2maildir.conf" %(os.environ["HOME"],)) + configfile = "%s/.rss2maildir.conf" %(os.environ["HOME"],) + except: + try: + os.stat("/etc/rss2maildir.conf") + configfile = "/etc/rss2maildir.conf" + except: + sys.stderr.write("No config file found. Exiting.\n") + sys.exit(2) -# Right - if we've got this far, we've got a config file, now for the hard -# bits... + # Right - if we've got this far, we've got a config file, now for the hard + # bits... -scp = SafeConfigParser() -scp.read(configfile) + scp = SafeConfigParser() + scp.read(configfile) -maildir_root = "RSSMaildir" -state_dir = "state" + maildir_root = "RSSMaildir" + state_dir = "state" -if options.statedir != None: - state_dir = options.statedir - 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 to make the directory + if options.statedir != None: + state_dir = options.statedir try: - os.mkdir(state_dir) + 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: - sys.stderr.write("Couldn't create statedir %s" %(state_dir)) - sys.exit(1) -elif scp.has_option("general", "state_dir"): - new_state_dir = scp.get("general", "state_dir") - 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 to create it + # try to make the directory + try: + os.mkdir(state_dir) + except: + sys.stderr.write("Couldn't create statedir %s" %(state_dir)) + sys.exit(1) + elif scp.has_option("general", "state_dir"): + new_state_dir = scp.get("general", "state_dir") try: - os.mkdir(new_state_dir) - state_dir = new_state_dir + 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: - sys.stderr.write( \ - "Couldn't create state directory %s\n" %(new_state_dir)) - sys.exit(1) -else: + # try to create it + try: + os.mkdir(new_state_dir) + state_dir = new_state_dir + 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") + try: - mode = os.stat(state_dir)[stat.ST_MODE] + mode = os.stat(maildir_root)[stat.ST_MODE] if not stat.S_ISDIR(mode): sys.stderr.write( \ - "State directory %s is not a directory\n" %(state_dir)) + "Maildir Root %s is not a directory\n" \ + %(maildir_root)) sys.exit(1) except: try: - os.mkdir(state_dir) + os.mkdir(maildir_root) except: - sys.stderr.write( \ - "State directory %s could not be created\n" %(state_dir)) + sys.stderr.write("Couldn't create Maildir Root %s\n" \ + %(maildir_root)) sys.exit(1) -if scp.has_option("general", "maildir_root"): - maildir_root = scp.get("general", "maildir_root") - -try: - mode = os.stat(maildir_root)[stat.ST_MODE] - if not stat.S_ISDIR(mode): - sys.stderr.write( \ - "Maildir Root %s is not a directory\n" \ - %(maildir_root)) - sys.exit(1) -except: - try: - os.mkdir(maildir_root) - except: - sys.stderr.write("Couldn't create Maildir Root %s\n" %(maildir_root)) - sys.exit(1) - -feeds = scp.sections() -try: - feeds.remove("general") -except: - pass - -for section in feeds: - # check if the directory exists - maildir = None + feeds = scp.sections() try: - maildir = scp.get(section, "maildir") + feeds.remove("general") except: - maildir = section + pass + + for section in feeds: + # check if the directory exists + maildir = None + try: + maildir = scp.get(section, "maildir") + except: + maildir = section - maildir = urllib.urlencode(((section, maildir),)).split("=")[1] - maildir = os.path.join(maildir_root, maildir) + maildir = urllib.urlencode(((section, maildir),)).split("=")[1] + maildir = os.path.join(maildir_root, maildir) - try: - exists = os.stat(maildir) - if stat.S_ISDIR(exists[stat.ST_MODE]): - # check if there's a new, cur and tmp directory + try: + exists = os.stat(maildir) + if stat.S_ISDIR(exists[stat.ST_MODE]): + # check if there's a new, cur and tmp directory + try: + mode = os.stat(os.path.join(maildir, "cur"))[stat.ST_MODE] + except: + os.mkdir(os.path.join(maildir, "cur")) + if not stat.S_ISDIR(mode): + sys.stderr.write("Broken maildir: %s\n" %(maildir)) + try: + mode = os.stat(os.path.join(maildir, "tmp"))[stat.ST_MODE] + except: + os.mkdir(os.path.join(maildir, "tmp")) + if not stat.S_ISDIR(mode): + sys.stderr.write("Broken maildir: %s\n" %(maildir)) + try: + mode = os.stat(os.path.join(maildir, "new"))[stat.ST_MODE] + if not stat.S_ISDIR(mode): + sys.stderr.write("Broken maildir: %s\n" %(maildir)) + except: + os.mkdir(os.path.join(maildir, "new")) + else: + sys.stderr.write("Broken maildir: %s\n" %(maildir)) + except: try: - mode = os.stat(os.path.join(maildir, "cur"))[stat.ST_MODE] + os.mkdir(maildir) except: - os.mkdir(os.path.join(maildir, "cur")) - if not stat.S_ISDIR(mode): - sys.stderr.write("Broken maildir: %s\n" %(maildir)) + sys.stderr.write("Couldn't create root maildir %s\n" \ + %(maildir)) + sys.exit(1) try: - mode = os.stat(os.path.join(maildir, "tmp"))[stat.ST_MODE] - except: + os.mkdir(os.path.join(maildir, "new")) + os.mkdir(os.path.join(maildir, "cur")) os.mkdir(os.path.join(maildir, "tmp")) - if not stat.S_ISDIR(mode): - sys.stderr.write("Broken maildir: %s\n" %(maildir)) - try: - mode = os.stat(os.path.join(maildir, "new"))[stat.ST_MODE] - if not stat.S_ISDIR(mode): - sys.stderr.write("Broken maildir: %s\n" %(maildir)) except: - os.mkdir(os.path.join(maildir, "new")) - else: - sys.stderr.write("Broken maildir: %s\n" %(maildir)) - except: - try: - os.mkdir(maildir) - except: - sys.stderr.write("Couldn't create root maildir %s\n" %(maildir)) - sys.exit(1) - try: - os.mkdir(os.path.join(maildir, "new")) - os.mkdir(os.path.join(maildir, "cur")) - os.mkdir(os.path.join(maildir, "tmp")) - except: - sys.stderr.write( \ - "Couldn't create required maildir directories for %s\n" \ - %(section,)) - sys.exit(1) + sys.stderr.write( \ + "Couldn't create required maildir directories for %s\n" \ + %(section,)) + sys.exit(1) - # right - we've got the directories, we've got the section, we know the - # url... lets play! + # right - we've got the directories, we've got the section, we know the + # url... lets play! - parse_and_deliver(maildir, section, state_dir) + parse_and_deliver(maildir, section, state_dir)