"""
import collections as _collections
+from email.mime.message import MIMEMessage as _MIMEMessage
+from email.mime.multipart import MIMEMultipart as _MIMEMultipart
from email.utils import formataddr as _formataddr
import hashlib as _hashlib
import html.parser as _html_parser
from . import util as _util
-_feedparser.USER_AGENT = 'rss2email/{} +{}'.format(__version__, __url__)
+_USER_AGENT = 'rss2email/{} +{}'.format(__version__, __url__)
+_feedparser.USER_AGENT = _USER_AGENT
_urllib_request.install_opener(_urllib_request.build_opener())
_SOCKET_ERRORS = []
for e in ['error', 'herror', 'gaierror']:
# hints for value conversion
_boolean_attributes = [
+ 'digest',
'force_from',
'use_publisher_email',
'friendly_name',
_function_attributes = [
'post_process',
+ 'digest_post_process',
]
def __init__(self, name=None, url=None, to=None, config=None):
elif isinstance(exc, _sax.SAXParseException):
_LOG.error('sax parsing error: {}: {}'.format(exc, self))
warned = True
+ elif (parsed.bozo and
+ isinstance(exc, _feedparser.CharacterEncodingOverride)):
+ _LOG.warning(
+ 'incorrectly declared encoding: {}: {}'.format(exc, self))
+ warned = True
elif parsed.bozo or exc:
if exc is None:
exc = "can't process"
not version):
raise _error.ProcessingError(parsed=parsed, feed=feed)
+ def _html2text(self, html, baseurl=''):
+ self.config.setup_html2text(section=self.section)
+ return _html2text.html2text(html=html, baseurl=baseurl)
+
def _process_entry(self, parsed, entry):
id_ = self._get_entry_id(entry)
# If .trust_guid isn't set, we get back hashes of the content.
extra_headers = _collections.OrderedDict((
('Date', self._get_entry_date(entry)),
('Message-ID', '<{}@dev.null.invalid>'.format(_uuid.uuid4())),
- ('User-Agent', 'rss2email'),
+ ('User-Agent', _USER_AGENT),
('X-RSS-Feed', self.url),
('X-RSS-ID', id_),
('X-RSS-URL', self._get_entry_link(entry)),
if hasattr(entry, 'title_detail') and entry.title_detail:
title = entry.title_detail.value
if 'html' in entry.title_detail.type:
- title = _html2text.html2text(title)
+ title = self._html2text(title)
else:
content = self._get_entry_content(entry)
value = content['value']
if content['type'] in ('text/html', 'application/xhtml+xml'):
- value = _html2text.html2text(value)
+ value = self._html2text(value)
title = value[:70]
title = title.replace('\n', ' ').strip()
return title
lines.extend([
'</head>',
'<body>',
- '<div id="entry>',
+ '<div id="entry">',
'<h1 class="header"><a href="{}">{}</a></h1>'.format(
link, subject),
'<div id="body">',
else: # not self.html_mail
if content['type'] in ('text/html', 'application/xhtml+xml'):
try:
- lines = [_html2text.html2text(content['value'])]
+ lines = [self._html2text(content['value'])]
except _html_parser.HTMLParseError as e:
raise _error.ProcessingError(parsed=None, feed=self)
else:
if not self.to:
raise _error.NoToEmailAddress(feed=self)
parsed = self._fetch()
+
+ if self.digest:
+ digest = self._new_digest()
+ seen = []
+
for (guid, id_, sender, message) in self._process(parsed):
_LOG.debug('new message: {}'.format(message['Subject']))
- if send:
- self._send(sender=sender, message=message)
+ if self.digest:
+ seen.append((guid, id_))
+ self._append_to_digest(digest=digest, message=message)
+ else:
+ if send:
+ self._send(sender=sender, message=message)
+ if guid not in self.seen:
+ self.seen[guid] = {}
+ self.seen[guid]['id'] = id_
+
+ if self.digest and seen:
+ if self.digest_post_process:
+ digest = self.digest_post_process(
+ feed=self, parsed=parsed, seen=seen, message=digest)
+ if not digest:
+ return
+ self._send_digest(
+ digest=digest, seen=seen, sender=sender, send=send)
+
+ self.etag = parsed.get('etag', None)
+ self.modified = parsed.get('modified', None)
+
+ def _new_digest(self):
+ digest = _MIMEMultipart('digest')
+ digest['To'] = self.to # TODO: _Header(), _formataddr((recipient_name, recipient_addr))
+ digest['Subject'] = 'digest for {}'.format(self.name)
+ digest['Message-ID'] = '<{}@dev.null.invalid>'.format(_uuid.uuid4())
+ digest['User-Agent'] = _USER_AGENT
+ digest['X-RSS-Feed'] = self.url
+ return digest
+
+ def _append_to_digest(self, digest, message):
+ part = _MIMEMessage(message)
+ part.add_header('Content-Disposition', 'attachment')
+ digest.attach(part)
+
+ def _send_digest(self, digest, seen, sender, send=True):
+ """Send a digest message
+
+ The date is extracted from the last message in the digest
+ payload. We assume that this part exists. If you don't have
+ any messages in the digest, don't call this function.
+ """
+ digest['From'] = sender # TODO: _Header(), _formataddr()...
+ last_part = digest.get_payload()[-1]
+ last_message = last_part.get_payload()[0]
+ digest['Date'] = last_message['Date']
+
+ _LOG.debug('new digest for {}'.format(self))
+ if send:
+ self._send(sender=sender, message=digest)
+ for (guid, id_) in seen:
if guid not in self.seen:
self.seen[guid] = {}
self.seen[guid]['id'] = id_
- self.etag = parsed.get('etag', None)
- self.modified = parsed.get('modified', None)
projects / rss2email.git / blobdiff