kolen/obrygator.py

## obrygator.py
#!/usr/bin/env python
# -*- coding: utf-8 -*-

# Send RSS feed items to email as plaintext converted by lynx

__revision__ = '0.0.1'

import shelve, urllib, smtplib, popen2, os
import cgi
from xml.dom.minidom import parse
from email.MIMEText import MIMEText

import cgitb; cgitb.enable()

email_from = u'RSS Obrygator <kolen@darkyoshkarola.org>'
lynx_cmdline = 'lynx -force_html -image_links -dump -assume_local_charset=utf-8 -display_charset=utf-8 file:///dev/stdin'

class FeedDescr:
    def __init__(self, handle, url, email):
        self.url = url
        self.last_guid = ''
        self.title = ''
        self.link = ''
        self.handle = handle
        self.email_to = email

class FeedItem:
    def __repr__(self):
        return "<FeedItem '%s' '%s'>" %(self.title, self.guid)


class FeedFetcher:
    def __init__(self, descr):
        self.descr = descr
        self.items = []

    def fetch(self):
        f = urllib.urlopen(self.descr.url)
        doc = parse(f)

        self.descr.link = self.extractField(doc, 'link')
        self.descr.title = self.extractField(doc, 'title')

        new_last_guid = None
        rssitems = doc.getElementsByTagName('item')
        for item in rssitems:
            fi = FeedItem()

            fi.title = self.extractField(item, 'title')
            fi.link = self.extractField(item, 'link')
            fi.description = self.extractField(item, 'description')
            fi.creator = self.extractField(item, 'dc:creator')
            fi.guid = self.extractField(item, 'guid')

            if not new_last_guid:
                new_last_guid = fi.guid
                if not self.descr.last_guid:
                    self.items.append(fi)
                    break

            if self.descr.last_guid and fi.guid == self.descr.last_guid:
                break

            self.items.append(fi)

        self.descr.last_guid = new_last_guid

    def extractField(self, item, field):
        try:
            attrItems = item.getElementsByTagName(field)
            if not attrItems:
                return u''
            fc = attrItems[0].firstChild
            if fc == None:
                return u''
            return fc.data
        except:
            return "Unexpected error: %s" % (sys.exc_info()[0])


class FeedSender:
    def send(self, fetchedFeed, email):
        template = u'''
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
%s
%s
Лента: %s
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Автор: %s

%s
--
Люто и бешено преподнесено Obrygator/%s
'''

        html_template = u'''
        <html>
        <meta http-equiv="Content-type" value="text/html; charset=utf-8" />
        <body>
        %s
        </body>
        </html>
        '''


        for item in fetchedFeed.items:
            os.environ['LANG'] = 'en_EN'
            fin,fout = popen2.popen2(lynx_cmdline)
            fout.write((html_template % (item.description)).encode('utf-8'))
            fout.close()
            textualDescr = fin.read().decode('utf-8')

            mailbody = template % (item.title, item.link,
                                   fetchedFeed.descr.title,
                                   item.creator,
                                   textualDescr,
                                   __revision__)

            envelope = MIMEText(mailbody.encode('utf-8'), 'plain', 'utf-8')
            envelope['To'] = fetchedFeed.descr.email_to
            envelope['From'] = email_from
            envelope['Subject'] = "[%s] %s (%s)" % (fetchedFeed.descr.handle,
                                                    item.title,
                                                    fetchedFeed.descr.title)

            envelope.set_charset('utf-8')

            smtp = smtplib.SMTP()
            smtp.connect()
            smtp.sendmail(email_from, [fetchedFeed.descr.email_to.encode('utf-8')], envelope.as_string())


def ctype():
    print 'Content-type: text/html\n\n'

feeds = shelve.open('data/feeds', writeback=True)

fs = cgi.FieldStorage()

if fs['mode'].value == 'send':
    ctype()
    for key in feeds:
        feed = feeds[key]
        ff = FeedFetcher(feed)
        ff.fetch()

        fs = FeedSender()
        fs.send(ff, feed.email_to)

        #store back
        feeds[key] = feed

elif fs['mode'].value == 'add':
    if fs.has_key('add'):
        handle = fs['handle'].value
        url = fs['url'].value
        email_to = fs['email_to'].value

        fd = FeedDescr(handle, url, email_to)
        feeds[handle] = fd

        ctype()
        print "Okay"
    else:
        ctype()
        print "<html><body>"
        print '<form method="post">h:<input name="handle" type="text" />'
        print 'url:<input name="url" type="text" />'
        print 'email:<input name="email_to" type="text" />'
        print '<button type="submit">Add</button>'
        print '<input type="hidden" name="mode" value="add"/>'
        print '<input type="hidden" name="add" value="yes"/>'
        print '</form>'

        for key in feeds:
            print "<b>%s</b> %s %s %s %s<br />" % (key,
                                                   feeds[key].title.encode('utf-8'),
                                                   feeds[key].url.encode('utf-8'),
                                                   feeds[key].email_to.encode('utf-8'),
                                                   feeds[key].last_guid.encode('utf-8'))

        print "</body></html>"

elif fs['mode'].value == 'remove':
    del feeds[fs['handle'].value]
    ctype()

else:
    ctype()
    print 'error in args'


feeds.close()
	#!/usr/bin/env python
	# -- coding: utf-8 --

	# Send RSS feed items to email as plaintext converted by lynx

	__revision__ = '0.0.1'

	import shelve, urllib, smtplib, popen2, os
	import cgi
	from xml.dom.minidom import parse
	from email.MIMEText import MIMEText

	import cgitb; cgitb.enable()

	email_from = u'RSS Obrygator <kolen@darkyoshkarola.org>'
	lynx_cmdline = 'lynx -force_html -image_links -dump -assume_local_charset=utf-8 -display_charset=utf-8 file:///dev/stdin'

	class FeedDescr:
	def __init__(self, handle, url, email):
	self.url = url
	self.last_guid = ''
	self.title = ''
	self.link = ''
	self.handle = handle
	self.email_to = email

	class FeedItem:
	def __repr__(self):
	return "<FeedItem '%s' '%s'>" %(self.title, self.guid)


	class FeedFetcher:
	def __init__(self, descr):
	self.descr = descr
	self.items = []

	def fetch(self):
	f = urllib.urlopen(self.descr.url)
	doc = parse(f)

	self.descr.link = self.extractField(doc, 'link')
	self.descr.title = self.extractField(doc, 'title')

	new_last_guid = None
	rssitems = doc.getElementsByTagName('item')
	for item in rssitems:
	fi = FeedItem()

	fi.title = self.extractField(item, 'title')
	fi.link = self.extractField(item, 'link')
	fi.description = self.extractField(item, 'description')
	fi.creator = self.extractField(item, 'dc:creator')
	fi.guid = self.extractField(item, 'guid')

	if not new_last_guid:
	new_last_guid = fi.guid
	if not self.descr.last_guid:
	self.items.append(fi)
	break

	if self.descr.last_guid and fi.guid == self.descr.last_guid:
	break

	self.items.append(fi)

	self.descr.last_guid = new_last_guid

	def extractField(self, item, field):
	try:
	attrItems = item.getElementsByTagName(field)
	if not attrItems:
	return u''
	fc = attrItems[0].firstChild
	if fc == None:
	return u''
	return fc.data
	except:
	return "Unexpected error: %s" % (sys.exc_info()[0])


	class FeedSender:
	def send(self, fetchedFeed, email):
	template = u'''
	-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
	%s
	%s
	Лента: %s
	-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
	Автор: %s

	%s
	--
	Люто и бешено преподнесено Obrygator/%s
	'''

	html_template = u'''
	<html>
	<meta http-equiv="Content-type" value="text/html; charset=utf-8" />
	<body>
	%s
	</body>
	</html>
	'''


	for item in fetchedFeed.items:
	os.environ['LANG'] = 'en_EN'
	fin,fout = popen2.popen2(lynx_cmdline)
	fout.write((html_template % (item.description)).encode('utf-8'))
	fout.close()
	textualDescr = fin.read().decode('utf-8')

	mailbody = template % (item.title, item.link,
	fetchedFeed.descr.title,
	item.creator,
	textualDescr,
	__revision__)

	envelope = MIMEText(mailbody.encode('utf-8'), 'plain', 'utf-8')
	envelope['To'] = fetchedFeed.descr.email_to
	envelope['From'] = email_from
	envelope['Subject'] = "[%s] %s (%s)" % (fetchedFeed.descr.handle,
	item.title,
	fetchedFeed.descr.title)

	envelope.set_charset('utf-8')

	smtp = smtplib.SMTP()
	smtp.connect()
	smtp.sendmail(email_from, [fetchedFeed.descr.email_to.encode('utf-8')], envelope.as_string())


	def ctype():
	print 'Content-type: text/html\n\n'

	feeds = shelve.open('data/feeds', writeback=True)

	fs = cgi.FieldStorage()

	if fs['mode'].value == 'send':
	ctype()
	for key in feeds:
	feed = feeds[key]
	ff = FeedFetcher(feed)
	ff.fetch()

	fs = FeedSender()
	fs.send(ff, feed.email_to)

	#store back
	feeds[key] = feed

	elif fs['mode'].value == 'add':
	if fs.has_key('add'):
	handle = fs['handle'].value
	url = fs['url'].value
	email_to = fs['email_to'].value

	fd = FeedDescr(handle, url, email_to)
	feeds[handle] = fd

	ctype()
	print "Okay"
	else:
	ctype()
	print "<html><body>"
	print '<form method="post">h:<input name="handle" type="text" />'
	print 'url:<input name="url" type="text" />'
	print 'email:<input name="email_to" type="text" />'
	print '<button type="submit">Add</button>'
	print '<input type="hidden" name="mode" value="add"/>'
	print '<input type="hidden" name="add" value="yes"/>'
	print '</form>'

	for key in feeds:
	print "<b>%s</b> %s %s %s %s<br />" % (key,
	feeds[key].title.encode('utf-8'),
	feeds[key].url.encode('utf-8'),
	feeds[key].email_to.encode('utf-8'),
	feeds[key].last_guid.encode('utf-8'))

	print "</body></html>"

	elif fs['mode'].value == 'remove':
	del feeds[fs['handle'].value]
	ctype()

	else:
	ctype()
	print 'error in args'


	feeds.close()