def discover_salmon_endpoint(self):
"""Discovers and returns the Salmon endpoint URL for this salmon.
It'd be nice to use an XRD/LRDD library here, but I haven't found much.
github.com/jcarbaugh/python-xrd is only for generating, not reading.
pydataportability.net looks powerful but also crazy heavyweight; it
requires Zope and strongly recommends running inside virtualenv. No thanks.
Returns: string URL or None
"""
url = json.loads(self.vars)["in_reply_to"]
logging.debug("Discovering salmon endpoint for %r", url)
body = util.urlfetch(url)
# first look in the document itself
endpoint = django_salmon.discover_salmon_endpoint(body)
if endpoint:
logging.debug("Found in original document: %r", endpoint)
return endpoint
# next, look in its feed, if any
#
# background on feed autodiscovery:
# http://blog.whatwg.org/feed-autodiscovery
parsed = feedparser.parse(body)
for link in parsed.feed.get("links", []):
rels = link.get("rel").split()
href = link.get("href")
if href and ("feed" in rels or "alternate" in rels):
endpoint = django_salmon.discover_salmon_endpoint(util.urlfetch(href))
if endpoint:
logging.debug("Found in feed: %r", endpoint)
return endpoint
# next, look in /.well-known/host-meta
host_meta_url = "http://%s/.well-known/host-meta" % util.domain_from_link(url)
endpoint = django_salmon.discover_salmon_endpoint(util.urlfetch(host_meta_url))
if endpoint:
logging.debug("Found in host-meta: %r", endpoint)
return endpoint
logging.debug("No salmon endpoint found!")
return None
def test_endpoint_discovery():
for fixture in fixtures:
endpoint = django_salmon.discover_salmon_endpoint(fixture)
assert endpoint == 'http://testserver/salmon/endpoint'
def _try_salmon(self, resp):
"""
Args:
resp: Response
"""
# fetch target HTML page, extract Atom rel-alternate link
target = resp.target()
if not self.target_resp:
self.target_resp = common.requests_get(target)
parsed = util.parse_html(self.target_resp)
atom_url = parsed.find('link',
rel='alternate',
type=common.CONTENT_TYPE_ATOM)
if not atom_url or not atom_url.get('href'):
self.error('Target post %s has no Atom link' % resp.target(),
status=400)
# fetch Atom target post, extract and inject id into source object
base_url = ''
base = parsed.find('base')
if base and base.get('href'):
base_url = base['href']
atom_link = parsed.find('link',
rel='alternate',
type=common.CONTENT_TYPE_ATOM)
atom_url = urllib.parse.urljoin(
resp.target(), urllib.parse.urljoin(base_url, atom_link['href']))
feed = common.requests_get(atom_url).text
parsed = feedparser.parse(feed)
logging.info('Parsed: %s', json_dumps(parsed, indent=2))
entry = parsed.entries[0]
target_id = entry.id
in_reply_to = self.source_obj.get('inReplyTo')
source_obj_obj = self.source_obj.get('object')
if in_reply_to:
for elem in in_reply_to:
if elem.get('url') == target:
elem['id'] = target_id
elif isinstance(source_obj_obj, dict):
source_obj_obj['id'] = target_id
# Mastodon (and maybe others?) require a rel-mentioned link to the
# original post's author to make it show up as a reply:
# app/services/process_interaction_service.rb
# ...so add them as a tag, which atom renders as a rel-mention link.
authors = entry.get('authors', None)
if authors:
url = entry.authors[0].get('href')
if url:
self.source_obj.setdefault('tags', []).append({'url': url})
# extract and discover salmon endpoint
logging.info('Discovering Salmon endpoint in %s', atom_url)
endpoint = django_salmon.discover_salmon_endpoint(feed)
if not endpoint:
# try webfinger
parsed = urllib.parse.urlparse(resp.target())
# TODO: test missing email
author = entry.get('author_detail', {})
email = author.get('email') or '@'.join(
(author.get('name', ''), parsed.netloc))
try:
# TODO: always https?
profile = common.requests_get(
'%s://%s/.well-known/webfinger?resource=acct:%s' %
(parsed.scheme, parsed.netloc, email),
verify=False)
endpoint = django_salmon.get_salmon_replies_link(
profile.json())
except requests.HTTPError as e:
pass
if not endpoint:
self.error('No salmon endpoint found!', status=400)
logging.info('Discovered Salmon endpoint %s', endpoint)
# construct reply Atom object
self.source_url = resp.source()
activity = self.source_obj
if self.source_obj.get('verb') not in source.VERBS_WITH_OBJECT:
activity = {'object': self.source_obj}
entry = atom.activity_to_atom(activity, xml_base=self.source_url)
logging.info('Converted %s to Atom:\n%s', self.source_url, entry)
# sign reply and wrap in magic envelope
domain = urllib.parse.urlparse(self.source_url).netloc
key = MagicKey.get_or_create(domain)
logging.info('Using key for %s: %s', domain, key)
magic_envelope = magicsigs.magic_envelope(entry,
common.CONTENT_TYPE_ATOM,
key).decode()
logging.info('Sending Salmon slap to %s', endpoint)
common.requests_post(
endpoint,
data=common.XML_UTF8 + magic_envelope,
headers={'Content-Type': common.CONTENT_TYPE_MAGIC_ENVELOPE})
return True
def send_salmon(self, source_obj, target_resp=None):
self.resp.protocol = 'ostatus'
# fetch target HTML page, extract Atom rel-alternate link
if not target_resp:
target_resp = common.requests_get(self.resp.target())
parsed = BeautifulSoup(target_resp.content, from_encoding=target_resp.encoding)
atom_url = parsed.find('link', rel='alternate', type=common.CONTENT_TYPE_ATOM)
if not atom_url or not atom_url.get('href'):
common.error(self, 'Target post %s has no Atom link' % self.resp.target(),
status=400)
# fetch Atom target post, extract and inject id into source object
feed = common.requests_get(atom_url['href']).text
parsed = feedparser.parse(feed)
logging.info('Parsed: %s', json.dumps(parsed, indent=2,
default=lambda key: '-'))
entry = parsed.entries[0]
target_id = entry.id
in_reply_to = source_obj.get('inReplyTo')
source_obj_obj = source_obj.get('object')
if in_reply_to:
in_reply_to[0]['id'] = target_id
elif isinstance(source_obj_obj, dict):
source_obj_obj['id'] = target_id
# Mastodon (and maybe others?) require a rel-mentioned link to the
# original post's author to make it show up as a reply:
# app/services/process_interaction_service.rb
# ...so add them as a tag, which atom renders as a rel-mention link.
authors = entry.get('authors', None)
if authors:
url = entry.authors[0].get('href')
if url:
source_obj.setdefault('tags', []).append({'url': url})
# extract and discover salmon endpoint
logging.info('Discovering Salmon endpoint in %s', atom_url['href'])
endpoint = django_salmon.discover_salmon_endpoint(feed)
if not endpoint:
# try webfinger
parsed = urlparse.urlparse(self.resp.target())
# TODO: test missing email
email = entry.author_detail.get('email') or '@'.join(
(entry.author_detail.name, parsed.netloc))
try:
# TODO: always https?
resp = common.requests_get(
'%s://%s/.well-known/webfinger?resource=acct:%s' %
(parsed.scheme, parsed.netloc, email), verify=False)
endpoint = django_salmon.get_salmon_replies_link(resp.json())
except requests.HTTPError as e:
pass
if not endpoint:
common.error(self, 'No salmon endpoint found!', status=400)
logging.info('Discovered Salmon endpoint %s', endpoint)
# construct reply Atom object
source_url = self.resp.source()
activity = (source_obj if source_obj.get('verb') in source.VERBS_WITH_OBJECT
else {'object': source_obj})
entry = atom.activity_to_atom(activity, xml_base=source_url)
logging.info('Converted %s to Atom:\n%s', source_url, entry)
# sign reply and wrap in magic envelope
domain = urlparse.urlparse(source_url).netloc
key = MagicKey.get_or_create(domain)
logging.info('Using key for %s: %s', domain, key)
magic_envelope = magicsigs.magic_envelope(
entry, common.CONTENT_TYPE_ATOM, key)
logging.info('Sending Salmon slap to %s', endpoint)
common.requests_post(
endpoint, data=common.XML_UTF8 + magic_envelope,
headers={'Content-Type': common.CONTENT_TYPE_MAGIC_ENVELOPE})
def test_endpoint_discovery():
for fixture in fixtures:
endpoint = django_salmon.discover_salmon_endpoint(fixture)
assert endpoint == "http://testserver/salmon/endpoint"
