def find_mention_item(self, items):
"""Returns the mf2 item that mentions (or replies to, likes, etc) the target.
May modify the items arg, e.g. may set or replace content.html or
content.value.
Args:
items: sequence of mf2 item dicts
Returns:
mf2 item dict or None
"""
# find target URL in source
for item in items:
props = item.setdefault('properties', {})
# find first non-empty content element
content = props.setdefault('content', [{}])[0]
text = content.get('html') or content.get('value')
for type in 'in-reply-to', 'like', 'like-of', 'repost', 'repost-of':
urls = [urllib.parse.urldefrag(u)[0] for u in
microformats2.get_string_urls(props.get(type, []))]
if self.any_target_in(urls):
break
else:
if text and self.any_target_in(text):
type = 'post'
url = get_first(props, 'url') or self.source_url
name = get_first(props, 'name') or get_first(props, 'summary')
text = content['html'] = f'mentioned this in {util.pretty_link(url, text=name, max_length=280)}.'
else:
type = None
if type:
# found the target!
rsvp = get_first(props, 'rsvp')
if rsvp:
self.entity.type = 'rsvp'
if not text:
content['value'] = f'RSVPed {rsvp}.'
else:
self.entity.type = {'in-reply-to': 'comment',
'like-of': 'like',
'repost-of': 'repost',
}.get(type, type)
if not text:
content['value'] = {'comment': 'replied to this.',
'like': 'liked this.',
'repost': 'reposted this.',
}[self.entity.type]
return item
# check children in case this is eg an h-feed
found = self.find_mention_item(item.get('children', []))
if found:
return found
return None
def _single_target(self):
"""
Returns: string URL, the source's inReplyTo or object (if appropriate)
"""
target = util.get_first(self.source_obj, 'inReplyTo')
if target:
return util.get_url(target)
if self.source_obj.get('verb') in source.VERBS_WITH_OBJECT:
return util.get_url(util.get_first(self.source_obj, 'object'))
def _render_attachments(attachments, obj):
"""Renders ActivityStreams attachments (or tags etc) as HTML.
Note that the returned HTML is included in Atom as well as HTML documents,
so it *must* be HTML4 / XHTML, not HTML5! All tags must be closed, etc.
Args:
attachments: sequence of decoded JSON ActivityStreams objects
obj: top-level decoded JSON ActivityStreams object
Returns:
string, rendered HTML
"""
content = ''
for att in attachments:
name = att.get('displayName', '')
stream = get_first(att, 'stream', {}).get('url') or ''
image = get_first(att, 'image', {}).get('url') or ''
open_a_tag = False
content += '\n<p>'
type = att.get('objectType')
if type == 'video':
if stream:
content += vid(stream, poster=image)
elif type == 'audio':
if stream:
content += aud(stream)
else:
url = att.get('url') or obj.get('url')
if url:
content += '\n<a class="link" href="%s">' % url
open_a_tag = True
if image:
content += '\n' + img(image, name)
if name and type != 'image':
content += '\n<span class="name">%s</span>' % name
if open_a_tag:
content += '\n</a>'
summary = att.get('summary')
if summary and summary != name:
content += '\n<span class="summary">%s</span>' % summary
content += '\n</p>'
return content
def _render_attachments(attachments, obj):
"""Renders ActivityStreams attachments (or tags etc) as HTML.
Note that the returned HTML is included in Atom as well as HTML documents,
so it *must* be HTML4 / XHTML, not HTML5! All tags must be closed, etc.
Args:
attachments: sequence of decoded JSON ActivityStreams objects
obj: top-level decoded JSON ActivityStreams object
Returns:
string, rendered HTML
"""
content = ''
for att in attachments:
name = att.get('displayName', '')
stream = get_first(att, 'stream', {}).get('url') or ''
image = get_first(att, 'image', {}).get('url') or ''
open_a_tag = False
content += '\n<p>'
if att.get('objectType') == 'video':
if stream:
content += vid(stream, poster=image)
elif att.get('objectType') == 'audio':
if stream:
content += aud(stream)
else:
url = att.get('url') or obj.get('url')
if url:
content += '\n<a class="link" href="%s">' % url
open_a_tag = True
if image:
content += '\n' + img(image, name)
if name:
content += '\n<span class="name">%s</span>' % name
if open_a_tag:
content += '\n</a>'
summary = att.get('summary')
if summary and summary != name:
content += '\n<span class="summary">%s</span>' % summary
content += '\n</p>'
return content
def first_props(props):
"""Converts a multiply-valued dict to singly valued.
Args:
props: dict of properties, where each value is a sequence
Returns: corresponding dict with just the first value of each sequence, or ''
if the sequence is empty
"""
return {k: util.get_first(props, k, '') for k in props} if props else {}
def _prepare_actor(actor):
"""Preprocesses an AS1 actor to prepare it to be rendered as Atom.
Modifies actor in place.
Args:
actor: ActivityStreams 1 actor dict
"""
if actor:
actor['image'] = util.get_first(actor, 'image')
def first_props(props):
"""Converts a multiply-valued dict to singly valued.
Args:
props: dict of properties, where each value is a sequence
Returns:
corresponding dict with just the first value of each sequence, or ''
if the sequence is empty
"""
return {k: util.get_first(props, k, '') for k in props} if props else {}
def activities_to_atom(activities,
actor,
title=None,
request_url=None,
host_url=None,
xml_base=None,
rels=None,
reader=True):
"""Converts ActivityStreams 1 activities to an Atom feed.
Args:
activities: list of ActivityStreams activity dicts
actor: ActivityStreams actor dict, the author of the feed
title: string, the feed <title> element. Defaults to 'User feed for [NAME]'
request_url: the URL of this Atom feed, if any. Used in a link rel="self".
host_url: the home URL for this Atom feed, if any. Used in the top-level
feed <id> element.
xml_base: the base URL, if any. Used in the top-level xml:base attribute.
rels: rel links to include. dict mapping string rel value to string URL.
reader: boolean, whether the output will be rendered in a feed reader.
Currently just includes location if True, not otherwise.
Returns:
unicode string with Atom XML
"""
# Strip query params from URLs so that we don't include access tokens, etc
host_url = (_remove_query_params(host_url)
if host_url else 'https://github.com/snarfed/granary')
if request_url is None:
request_url = host_url
for a in activities:
_prepare_activity(a, reader=reader)
updated = (util.get_first(activities[0], 'object', default={}).get(
'published', '') if activities else '')
if actor is None:
actor = {}
return jinja_env.get_template(FEED_TEMPLATE).render(
actor=Defaulter(actor),
host_url=host_url,
items=[Defaulter(a) for a in activities],
mimetypes=mimetypes,
rels=rels or {},
request_url=request_url,
title=title or 'User feed for ' + source.Source.actor_name(actor),
updated=updated,
VERBS_WITH_OBJECT=source.VERBS_WITH_OBJECT,
xml_base=xml_base,
)
def activities_to_atom(activities, actor, title=None, request_url=None,
host_url=None, xml_base=None, rels=None, reader=True):
"""Converts ActivityStreams 1 activities to an Atom feed.
Args:
activities: list of ActivityStreams activity dicts
actor: ActivityStreams actor dict, the author of the feed
title: string, the feed <title> element. Defaults to 'User feed for [NAME]'
request_url: the URL of this Atom feed, if any. Used in a link rel="self".
host_url: the home URL for this Atom feed, if any. Used in the top-level
feed <id> element.
xml_base: the base URL, if any. Used in the top-level xml:base attribute.
rels: rel links to include. dict mapping string rel value to string URL.
reader: boolean, whether the output will be rendered in a feed reader.
Currently just includes location if True, not otherwise.
Returns:
unicode string with Atom XML
"""
# Strip query params from URLs so that we don't include access tokens, etc
host_url = (_remove_query_params(host_url) if host_url
else 'https://github.com/snarfed/granary')
if request_url is None:
request_url = host_url
for a in activities:
_prepare_activity(a, reader=reader)
updated = (util.get_first(activities[0], 'object', default={}).get('published', '')
if activities else '')
if actor is None:
actor = {}
return jinja_env.get_template(FEED_TEMPLATE).render(
actor=Defaulter(actor),
host_url=host_url,
items=[Defaulter(a) for a in activities],
mimetypes=mimetypes,
rels=rels or {},
request_url=request_url,
title=title or 'User feed for ' + source.Source.actor_name(actor),
updated=updated,
VERBS_WITH_OBJECT=source.VERBS_WITH_OBJECT,
xml_base=xml_base,
)
def _create(self, obj, preview, include_link=False, ignore_formatting=False):
"""Creates or previews creating for the previous two methods.
https://www.flickr.com/services/api/upload.api.html
https://www.flickr.com/services/api/flickr.photos.comments.addComment.html
https://www.flickr.com/services/api/flickr.favorites.add.html
https://www.flickr.com/services/api/flickr.photos.people.add.html
Args:
obj: ActivityStreams object
preview: boolean
include_link: boolean
Return:
a CreationResult
"""
# photo, comment, or like
type = source.object_type(obj)
logging.debug('publishing object type %s to Flickr', type)
link_text = '(Originally published at: %s)' % obj.get('url')
image_url = util.get_first(obj, 'image', {}).get('url')
video_url = util.get_first(obj, 'stream', {}).get('url')
content = self._content_for_create(obj, ignore_formatting=ignore_formatting,
strip_first_video_tag=bool(video_url))
if (video_url or image_url) and type in ('note', 'article'):
name = obj.get('displayName')
people = self._get_person_tags(obj)
hashtags = [t.get('displayName') for t in obj.get('tags', [])
if t.get('objectType') == 'hashtag' and t.get('displayName')]
lat = obj.get('location', {}).get('latitude')
lng = obj.get('location', {}).get('longitude')
# if name does not represent an explicit title, then we'll just
# use it as the title and wipe out the content
if name and content and not mf2util.is_name_a_title(name, content):
name = content
content = None
# add original post link
if include_link:
content = ((content + '\n\n') if content else '') + link_text
if preview:
preview_content = ''
if name:
preview_content += '<h4>%s</h4>' % name
if content:
preview_content += '<div>%s</div>' % content
if hashtags:
preview_content += '<div> %s</div>' % ' '.join('#' + t for t in hashtags)
if people:
preview_content += '<div> with %s</div>' % ', '.join(
('<a href="%s">%s</a>' % (
p.get('url'), p.get('displayName') or 'User %s' % p.get('id'))
for p in people))
if lat and lng:
preview_content += '<div> at <a href="https://maps.google.com/maps?q=%s,%s">%s, %s</a></div>' % (lat, lng, lat, lng)
if video_url:
preview_content += ('<video controls src="%s"><a href="%s">this video'
'</a></video>' % (video_url, video_url))
else:
preview_content += '<img src="%s" />' % image_url
return source.creation_result(content=preview_content, description='post')
params = []
if name:
params.append(('title', name))
if content:
params.append(('description', content))
if hashtags:
params.append(
('tags', ','.join('"%s"' % t if ' ' in t else t for t in hashtags)))
file = util.urlopen(video_url or image_url)
resp = self.upload(params, file)
photo_id = resp.get('id')
resp.update({
'type': 'post',
'url': self.photo_url(self.path_alias() or self.user_id(), photo_id),
})
if video_url:
resp['granary_message'] = \
"Note that videos take time to process before they're visible."
# add person tags
for person_id in sorted(p.get('id') for p in people):
self.call_api_method('flickr.photos.people.add', {
'photo_id': photo_id,
'user_id': person_id,
})
# add location
if lat and lng:
self.call_api_method('flickr.photos.geo.setLocation', {
'photo_id': photo_id,
'lat': lat,
'lon': lng,
})
return source.creation_result(resp)
base_obj = self.base_object(obj)
base_id = base_obj.get('id')
base_url = base_obj.get('url')
# maybe a comment on a flickr photo?
if type == 'comment' or obj.get('inReplyTo'):
if not base_id:
return source.creation_result(
abort=True,
error_plain='Could not find a photo to comment on.',
error_html='Could not find a photo to <a href="http://indiewebcamp.com/reply">comment on</a>. '
'Check that your post has an <a href="http://indiewebcamp.com/comment">in-reply-to</a> '
'link to a Flickr photo or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Flickr.')
if include_link:
content += '\n\n' + link_text
if preview:
return source.creation_result(
content=content,
description='comment on <a href="%s">this photo</a>.' % base_url)
resp = self.call_api_method('flickr.photos.comments.addComment', {
'photo_id': base_id,
'comment_text': content,
})
resp = resp.get('comment', {})
resp.update({
'type': 'comment',
'url': resp.get('permalink'),
})
return source.creation_result(resp)
if type == 'like':
if not base_id:
return source.creation_result(
abort=True,
error_plain='Could not find a photo to favorite.',
error_html='Could not find a photo to <a href="http://indiewebcamp.com/like">favorite</a>. '
'Check that your post has an <a href="http://indiewebcamp.com/like">like-of</a> '
'link to a Flickr photo or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Flickr.')
if preview:
return source.creation_result(
description='favorite <a href="%s">this photo</a>.' % base_url)
# this method doesn't return any data
self.call_api_method('flickr.favorites.add', {
'photo_id': base_id,
})
# TODO should we canonicalize the base_url (e.g. removing trailing path
# info like "/in/contacts/")
return source.creation_result({
'type': 'like',
'url': '%s#favorited-by-%s' % (base_url, self.user_id()),
})
return source.creation_result(
abort=False,
error_plain='Cannot publish type=%s to Flickr.' % type,
error_html='Cannot publish type=%s to Flickr.' % type)
def json_to_object(mf2, actor=None, fetch_mf2=False):
"""Converts a single microformats2 JSON item to an ActivityStreams object.
Supports h-entry, h-event, h-card, and other single item times. Does *not* yet
support h-feed.
Args:
mf2: dict, decoded JSON microformats2 object
actor: optional author AS actor object. usually comes from a rel="author"
link. if mf2 has its own author, that will override this.
fetch_mf2: boolean, whether to fetch additional pages via HTTP if necessary,
e.g. to determine authorship: https://indieweb.org/authorship
Returns:
dict, ActivityStreams object
"""
if not mf2 or not isinstance(mf2, dict):
return {}
mf2 = copy.copy(mf2)
props = mf2.setdefault('properties', {})
prop = first_props(props)
rsvp = prop.get('rsvp')
# convert author
mf2_author = prop.get('author')
if mf2_author and isinstance(mf2_author, dict):
author = json_to_object(mf2_author)
else:
# the author h-card may be on another page. run full authorship algorithm:
# https://indieweb.org/authorship
def fetch(url):
return mf2py.parse(util.requests_get(url).text, url=url, img_with_alt=True)
author = mf2util.find_author(
{'items': [mf2]}, hentry=mf2, fetch_mf2_func=fetch if fetch_mf2 else None)
if author:
author = {
'objectType': 'person',
'url': author.get('url'),
'displayName': author.get('name'),
'image': [{'url': author.get('photo')}],
}
if not author:
author = actor
mf2_types = mf2.get('type') or []
if 'h-geo' in mf2_types or 'p-location' in mf2_types:
mf2_type = 'location'
elif 'tag-of' in props:
# TODO: remove once this is in mf2util
# https://github.com/kylewm/mf2util/issues/18
mf2_type = 'tag'
elif 'follow-of' in props: # ditto
mf2_type = 'follow'
else:
# mf2 'photo' type is a note or article *with* a photo, but AS 'photo' type
# *is* a photo. so, special case photo type to fall through to underlying
# mf2 type without photo.
# https://github.com/snarfed/bridgy/issues/702
without_photo = copy.deepcopy(mf2)
without_photo.get('properties', {}).pop('photo', None)
mf2_type = mf2util.post_type_discovery(without_photo)
as_type, as_verb = MF2_TO_AS_TYPE_VERB.get(mf2_type, (None, None))
if rsvp:
as_verb = 'rsvp-%s' % rsvp
# special case GitHub issues that are in-reply-to the repo or its issues URL
in_reply_tos = get_string_urls(props.get('in-reply-to', []))
for url in in_reply_tos:
if re.match(r'^https?://github.com/[^/]+/[^/]+(/issues)?/?$', url):
as_type = 'issue'
def is_absolute(url):
"""Filter out relative and invalid URLs (mf2py gives absolute urls)."""
return urllib.parse.urlparse(url).netloc
urls = props.get('url') and get_string_urls(props.get('url'))
# quotations: https://indieweb.org/quotation#How_to_markup
attachments = [
json_to_object(quote)
for quote in mf2.get('children', []) + props.get('quotation-of', [])
if isinstance(quote, dict) and 'h-cite' in set(quote.get('type', []))]
# audio and video
for type in 'audio', 'video':
attachments.extend({'objectType': type, 'stream': {'url': url}}
for url in get_string_urls(props.get(type, [])))
obj = {
'id': prop.get('uid'),
'objectType': as_type,
'verb': as_verb,
'published': prop.get('published', ''),
'updated': prop.get('updated', ''),
'startTime': prop.get('start'),
'endTime': prop.get('end'),
'displayName': get_text(prop.get('name')),
'summary': get_text(prop.get('summary')),
'content': get_html(prop.get('content')),
'url': urls[0] if urls else None,
'urls': [{'value': u} for u in urls] if urls and len(urls) > 1 else None,
# image is special cased below, to handle alt
'stream': [{'url': url} for url in get_string_urls(props.get('video', []))],
'location': json_to_object(prop.get('location')),
'replies': {'items': [json_to_object(c) for c in props.get('comment', [])]},
'tags': [{'objectType': 'hashtag', 'displayName': cat}
if isinstance(cat, basestring)
else json_to_object(cat)
for cat in props.get('category', [])],
'attachments': attachments,
}
# images, including alt text
photo_urls = set()
obj['image'] = []
for photo in props.get('photo', []) + props.get('featured', []):
url = photo
alt = None
if isinstance(photo, dict):
photo = photo.get('properties') or photo
url = get_first(photo, 'value') or get_first(photo, 'url')
alt = get_first(photo, 'alt')
if url and url not in photo_urls and is_absolute(url):
photo_urls.add(url)
obj['image'].append({'url': url, 'displayName': alt})
# mf2util uses the indieweb/mf2 location algorithm to collect location properties.
interpreted = mf2util.interpret({'items': [mf2]}, None)
if interpreted:
loc = interpreted.get('location')
if loc:
obj['location']['objectType'] = 'place'
lat, lng = loc.get('latitude'), loc.get('longitude')
if lat and lng:
try:
obj['location'].update({
'latitude': float(lat),
'longitude': float(lng),
})
except ValueError:
logging.warn(
'Could not convert latitude/longitude (%s, %s) to decimal', lat, lng)
if as_type == 'activity':
objects = []
for target in itertools.chain.from_iterable(
props.get(field, []) for field in (
'follow-of', 'like', 'like-of', 'repost', 'repost-of', 'in-reply-to',
'invitee')):
t = json_to_object(target) if isinstance(target, dict) else {'url': target}
# eliminate duplicates from redundant backcompat properties
if t not in objects:
objects.append(t)
obj.update({
'object': objects[0] if len(objects) == 1 else objects,
'actor': author,
})
if as_verb == 'tag':
obj['target'] = {'url': prop['tag-of']}
if obj.get('object'):
raise NotImplementedError(
'Combined in-reply-to and tag-of is not yet supported.')
obj['object'] = obj.pop('tags')
else:
obj.update({
'inReplyTo': [{'url': url} for url in in_reply_tos],
'author': author,
})
return source.Source.postprocess_object(obj)
def object_to_json(obj, trim_nulls=True, entry_class='h-entry',
default_object_type=None, synthesize_content=True):
"""Converts an ActivityStreams object to microformats2 JSON.
Args:
obj: dict, a decoded JSON ActivityStreams object
trim_nulls: boolean, whether to remove elements with null or empty values
entry_class: string or sequence, the mf2 class(es) that entries should be
given (e.g. 'h-cite' when parsing a reference to a foreign entry).
defaults to 'h-entry'
default_object_type: string, the ActivityStreams objectType to use if one
is not present. defaults to None
synthesize_content: whether to generate synthetic content if the object
doesn't have its own, e.g. 'likes this.' or 'shared this.'
Returns:
dict, decoded microformats2 JSON
"""
if not obj or not isinstance(obj, dict):
return {}
obj_type = source.object_type(obj) or default_object_type
# if the activity type is a post, then it's really just a conduit
# for the object. for other verbs, the activity itself is the
# interesting thing
if obj_type == 'post':
primary = obj.get('object', {})
obj_type = source.object_type(primary) or default_object_type
else:
primary = obj
# TODO: extract snippet
name = primary.get('displayName', primary.get('title'))
summary = primary.get('summary')
author = obj.get('author', obj.get('actor', {}))
in_reply_tos = obj.get('inReplyTo', obj.get('context', {}).get('inReplyTo', []))
is_rsvp = obj_type in ('rsvp-yes', 'rsvp-no', 'rsvp-maybe')
if (is_rsvp or obj_type == 'react') and obj.get('object'):
objs = obj['object']
in_reply_tos.extend(objs if isinstance(objs, list) else [objs])
# maps objectType to list of objects
attachments = defaultdict(list)
for prop in 'attachments', 'tags':
for elem in get_list(primary, prop):
attachments[elem.get('objectType')].append(elem)
# construct mf2!
ret = {
'type': (AS_TO_MF2_TYPE.get(obj_type) or
[entry_class] if isinstance(entry_class, basestring)
else list(entry_class)),
'properties': {
'uid': [obj.get('id') or ''],
'numeric-id': [obj.get('numeric_id') or ''],
'name': [name],
'nickname': [obj.get('username') or ''],
'summary': [summary],
'url': (list(object_urls(obj) or object_urls(primary)) +
obj.get('upstreamDuplicates', [])),
# photo is special cased below, to handle alt
'video': dedupe_urls(get_urls(attachments, 'video', 'stream') +
get_urls(primary, 'stream')),
'audio': get_urls(attachments, 'audio', 'stream'),
'published': [obj.get('published', primary.get('published', ''))],
'updated': [obj.get('updated', primary.get('updated', ''))],
'content': [{
'value': xml.sax.saxutils.unescape(primary.get('content', '')),
'html': render_content(primary, include_location=False,
synthesize_content=synthesize_content),
}],
'in-reply-to': util.trim_nulls([o.get('url') for o in in_reply_tos]),
'author': [object_to_json(
author, trim_nulls=False, default_object_type='person')],
'location': [object_to_json(
primary.get('location', {}), trim_nulls=False,
default_object_type='place')],
'comment': [object_to_json(c, trim_nulls=False, entry_class='h-cite')
for c in obj.get('replies', {}).get('items', [])],
'start': [primary.get('startTime')],
'end': [primary.get('endTime')],
},
'children': (
# silly hack: i haven't found anywhere in AS1 or AS2 to indicate that
# something is being "quoted," like in a quote tweet, so i cheat and use
# extra knowledge here that quoted tweets are converted to note
# attachments, but URLs in the tweet text are converted to article tags.
[object_to_json(a, trim_nulls=False, entry_class=['u-quotation-of', 'h-cite'])
for a in attachments['note'] if 'startIndex' not in a] +
[object_to_json(a, trim_nulls=False, entry_class=['h-cite'])
for a in attachments['article'] if 'startIndex' not in a])
}
# photos, including alt text
photo_urls = set()
ret['properties']['photo'] = []
for image in get_list(attachments, 'image') + [primary]:
for url in get_urls(image, 'image'):
if url and url not in photo_urls:
photo_urls.add(url)
name = get_first(image, 'image', {}).get('displayName')
ret['properties']['photo'].append({'value': url, 'alt': name} if name else url)
# hashtags and person tags
if obj_type == 'tag':
ret['properties']['tag-of'] = util.get_urls(obj, 'target')
tags = obj.get('tags', []) or get_first(obj, 'object', {}).get('tags', [])
if not tags and obj_type == 'tag':
tags = util.get_list(obj, 'object')
ret['properties']['category'] = []
for tag in tags:
if tag.get('objectType') == 'person':
ret['properties']['category'].append(
object_to_json(tag, entry_class='u-category h-card'))
elif tag.get('objectType') == 'hashtag' or obj_type == 'tag':
name = tag.get('displayName')
if name:
ret['properties']['category'].append(name)
# rsvp
if is_rsvp:
ret['properties']['rsvp'] = [obj_type[len('rsvp-'):]]
elif obj_type == 'invite':
invitee = object_to_json(obj.get('object'), trim_nulls=False,
default_object_type='person')
ret['properties']['invitee'] = [invitee]
# like and repost mentions
for type, prop in (
('favorite', 'like'),
('follow', 'follow'),
('like', 'like'),
('share', 'repost'),
):
if obj_type == type:
# The ActivityStreams spec says the object property should always be a
# single object, but it's useful to let it be a list, e.g. when a like has
# multiple targets, e.g. a like of a post with original post URLs in it,
# which brid.gy does.
objs = get_list(obj, 'object')
ret['properties'][prop + '-of'] = [
# flatten contexts that are just a url
o['url'] if 'url' in o and set(o.keys()) <= set(['url', 'objectType'])
else object_to_json(o, trim_nulls=False, entry_class='h-cite')
for o in objs]
else:
# received likes and reposts
ret['properties'][prop] = [
object_to_json(t, trim_nulls=False, entry_class='h-cite')
for t in tags if source.object_type(t) == type]
# latitude & longitude
lat = long = None
position = ISO_6709_RE.match(primary.get('position') or '')
if position:
lat, long = position.groups()
if not lat:
lat = primary.get('latitude')
if not long:
long = primary.get('longitude')
if lat:
ret['properties']['latitude'] = [str(lat)]
if long:
ret['properties']['longitude'] = [str(long)]
if trim_nulls:
ret = util.trim_nulls(ret)
return ret
def image_url(obj):
return util.get_first(obj, 'image', {}).get('url')
def json_to_object(mf2, actor=None, fetch_mf2=False):
"""Converts a single microformats2 JSON item to an ActivityStreams object.
Supports h-entry, h-event, h-card, and other single item times. Does *not* yet
support h-feed.
Args:
mf2: dict, decoded JSON microformats2 object
actor: optional author AS actor object. usually comes from a rel="author"
link. if mf2 has its own author, that will override this.
fetch_mf2: boolean, whether to fetch additional pages via HTTP if necessary,
e.g. to determine authorship: https://indieweb.org/authorship
Returns:
dict, ActivityStreams object
"""
if not mf2 or not isinstance(mf2, dict):
return {}
mf2 = copy.copy(mf2)
props = mf2.setdefault('properties', {})
prop = first_props(props)
rsvp = prop.get('rsvp')
# convert author
mf2_author = prop.get('author')
if mf2_author and isinstance(mf2_author, dict):
author = json_to_object(mf2_author)
else:
# the author h-card may be on another page. run full authorship algorithm:
# https://indieweb.org/authorship
author = mf2util.find_author({'items': [mf2]}, hentry=mf2,
fetch_mf2_func=util.fetch_mf2 if fetch_mf2 else None)
if author:
author = {
'objectType': 'person',
'url': author.get('url'),
'displayName': author.get('name'),
'image': [{'url': author.get('photo')}],
}
if not author:
author = actor
mf2_types = mf2.get('type') or []
if 'h-geo' in mf2_types or 'p-location' in mf2_types:
mf2_type = 'location'
elif 'tag-of' in props:
# TODO: remove once this is in mf2util
# https://github.com/kylewm/mf2util/issues/18
mf2_type = 'tag'
elif 'follow-of' in props: # ditto
mf2_type = 'follow'
else:
# mf2 'photo' type is a note or article *with* a photo, but AS 'photo' type
# *is* a photo. so, special case photo type to fall through to underlying
# mf2 type without photo.
# https://github.com/snarfed/bridgy/issues/702
without_photo = copy.deepcopy(mf2)
without_photo.get('properties', {}).pop('photo', None)
mf2_type = mf2util.post_type_discovery(without_photo)
as_type, as_verb = MF2_TO_AS_TYPE_VERB.get(mf2_type, (None, None))
if rsvp:
as_verb = 'rsvp-%s' % rsvp
# special case GitHub issues that are in-reply-to the repo or its issues URL
in_reply_tos = get_string_urls(props.get('in-reply-to', []))
for url in in_reply_tos:
if re.match(r'^https?://github.com/[^/]+/[^/]+(/issues)?/?$', url):
as_type = 'issue'
def is_absolute(url):
"""Filter out relative and invalid URLs (mf2py gives absolute urls)."""
return urllib.parse.urlparse(url).netloc
urls = props.get('url') and get_string_urls(props.get('url'))
# quotations: https://indieweb.org/quotation#How_to_markup
attachments = [
json_to_object(quote)
for quote in mf2.get('children', []) + props.get('quotation-of', [])
if isinstance(quote, dict) and 'h-cite' in set(quote.get('type', []))]
# audio and video
#
# the duration mf2 property is still emerging. examples in the wild use both
# integer seconds and ISO 8601 durations.
# https://indieweb.org/duration
# https://en.wikipedia.org/wiki/ISO_8601#Durations
duration = prop.get('duration') or prop.get('length')
if duration:
if util.is_int(duration):
duration = int(duration)
else:
parsed = util.parse_iso8601_duration(duration)
if parsed:
duration = int(parsed.total_seconds())
else:
logging.debug('Unknown format for length or duration %r', duration)
duration = None
stream = None
bytes = size_to_bytes(prop.get('size'))
for type in 'audio', 'video':
atts = [{
'objectType': type,
'stream': {
'url': url,
# integer seconds: http://activitystrea.ms/specs/json/1.0/#media-link
'duration': duration,
# file size in bytes. nonstandard, not in AS1 or AS2
'size': bytes,
},
} for url in get_string_urls(props.get(type, []))]
attachments.extend(atts)
if atts:
stream = atts[0]['stream']
obj = {
'id': prop.get('uid'),
'objectType': as_type,
'verb': as_verb,
'published': prop.get('published', ''),
'updated': prop.get('updated', ''),
'startTime': prop.get('start'),
'endTime': prop.get('end'),
'displayName': get_text(prop.get('name')),
'summary': get_text(prop.get('summary')),
'content': get_html(prop.get('content')),
'url': urls[0] if urls else None,
'urls': [{'value': u} for u in urls] if urls and len(urls) > 1 else None,
# image is special cased below, to handle alt
'stream': [stream],
'location': json_to_object(prop.get('location')),
'replies': {'items': [json_to_object(c) for c in props.get('comment', [])]},
'tags': [{'objectType': 'hashtag', 'displayName': cat}
if isinstance(cat, str)
else json_to_object(cat)
for cat in props.get('category', [])],
'attachments': attachments,
}
# images, including alt text
photo_urls = set()
obj['image'] = []
for photo in props.get('photo', []) + props.get('featured', []):
url = photo
alt = None
if isinstance(photo, dict):
photo = photo.get('properties') or photo
url = get_first(photo, 'value') or get_first(photo, 'url')
alt = get_first(photo, 'alt')
if url and url not in photo_urls and is_absolute(url):
photo_urls.add(url)
obj['image'].append({'url': url, 'displayName': alt})
# mf2util uses the indieweb/mf2 location algorithm to collect location properties.
interpreted = mf2util.interpret({'items': [mf2]}, None)
if interpreted:
loc = interpreted.get('location')
if loc:
obj['location']['objectType'] = 'place'
lat, lng = loc.get('latitude'), loc.get('longitude')
if lat and lng:
try:
obj['location'].update({
'latitude': float(lat),
'longitude': float(lng),
})
except ValueError:
logging.debug(
'Could not convert latitude/longitude (%s, %s) to decimal', lat, lng)
if as_type == 'activity':
objects = []
for target in itertools.chain.from_iterable(
props.get(field, []) for field in (
'follow-of', 'like', 'like-of', 'repost', 'repost-of', 'in-reply-to',
'invitee')):
t = json_to_object(target) if isinstance(target, dict) else {'url': target}
# eliminate duplicates from redundant backcompat properties
if t not in objects:
objects.append(t)
obj.update({
'object': objects[0] if len(objects) == 1 else objects,
'actor': author,
})
if as_verb == 'tag':
obj['target'] = {'url': prop['tag-of']}
if obj.get('object'):
raise NotImplementedError(
'Combined in-reply-to and tag-of is not yet supported.')
obj['object'] = obj.pop('tags')
else:
obj.update({
'inReplyTo': [{'url': url} for url in in_reply_tos],
'author': author,
})
return source.Source.postprocess_object(obj)
def object_to_json(obj,
trim_nulls=True,
entry_class='h-entry',
default_object_type=None,
synthesize_content=True):
"""Converts an ActivityStreams object to microformats2 JSON.
Args:
obj: dict, a decoded JSON ActivityStreams object
trim_nulls: boolean, whether to remove elements with null or empty values
entry_class: string, the mf2 class that entries should be given (e.g.
'h-cite' when parsing a reference to a foreign entry). defaults to
'h-entry'
default_object_type: string, the ActivityStreams objectType to use if one
is not present. defaults to None
synthesize_content: whether to generate synthetic content if the object
doesn't have its own, e.g. 'likes this.' or 'shared this.'
Returns: dict, decoded microformats2 JSON
"""
if not obj:
return {}
obj_type = source.object_type(obj) or default_object_type
# if the activity type is a post, then it's really just a conduit
# for the object. for other verbs, the activity itself is the
# interesting thing
if obj_type == 'post':
primary = obj.get('object', {})
obj_type = source.object_type(primary) or default_object_type
else:
primary = obj
# TODO: extract snippet
name = primary.get('displayName', primary.get('title'))
summary = primary.get('summary')
author = obj.get('author', obj.get('actor', {}))
in_reply_tos = obj.get('inReplyTo',
obj.get('context', {}).get('inReplyTo', []))
is_rsvp = obj_type in ('rsvp-yes', 'rsvp-no', 'rsvp-maybe')
if (is_rsvp or obj_type == 'react') and obj.get('object'):
objs = obj['object']
in_reply_tos.extend(objs if isinstance(objs, list) else [objs])
# TODO: more tags. most will be p-category?
ret = {
'type':
(['h-card'] if obj_type == 'person' else
['h-card', 'p-location'] if obj_type == 'place' else [entry_class]),
'properties': {
'uid': [obj.get('id', '')],
'name': [name],
'summary': [summary],
'url': (list(object_urls(obj) or object_urls(primary)) +
obj.get('upstreamDuplicates', [])),
'photo': [
image.get('url')
for image in (util.get_list(obj, 'image')
or util.get_list(primary, 'image'))
],
'video': [obj.get('stream', primary.get('stream', {})).get('url')],
'published': [obj.get('published', primary.get('published', ''))],
'updated': [obj.get('updated', primary.get('updated', ''))],
'content': [{
'value':
xml.sax.saxutils.unescape(primary.get('content', '')),
'html':
render_content(primary,
include_location=False,
synthesize_content=synthesize_content),
}],
'in-reply-to':
util.trim_nulls([o.get('url') for o in in_reply_tos]),
'author': [
object_to_json(author,
trim_nulls=False,
default_object_type='person')
],
'location': [
object_to_json(primary.get('location', {}),
trim_nulls=False,
default_object_type='place')
],
'latitude':
primary.get('latitude'),
'longitude':
primary.get('longitude'),
'comment': [
object_to_json(c, trim_nulls=False, entry_class='h-cite')
for c in obj.get('replies', {}).get('items', [])
],
},
'children': [
object_to_json(c, trim_nulls=False, entry_class='h-cite')
for c in primary.get('attachments', [])
if c.get('objectType') in ('note', 'article')
],
}
# hashtags and person tags
tags = obj.get('tags', []) or util.get_first(obj, 'object', {}).get(
'tags', [])
ret['properties']['category'] = []
for tag in tags:
if tag.get('objectType') == 'person':
cls = 'u-category h-card'
elif tag.get('objectType') == 'hashtag':
cls = 'u-category'
else:
continue
ret['properties']['category'].append(
object_to_json(tag, entry_class=cls))
# rsvp
if is_rsvp:
ret['properties']['rsvp'] = [obj_type[len('rsvp-'):]]
elif obj_type == 'invite':
invitee = object_to_json(obj.get('object'),
trim_nulls=False,
default_object_type='person')
ret['properties']['invitee'] = [invitee]
# like and repost mentions
for type, prop in ('like', 'like'), ('share', 'repost'):
if obj_type == type:
# The ActivityStreams spec says the object property should always be a
# single object, but it's useful to let it be a list, e.g. when a like has
# multiple targets, e.g. a like of a post with original post URLs in it,
# which brid.gy does.
objs = util.get_list(obj, 'object')
ret['properties'][prop + '-of'] = [
# flatten contexts that are just a url
o['url']
if 'url' in o and set(o.keys()) <= set(['url', 'objectType'])
else object_to_json(o, trim_nulls=False, entry_class='h-cite')
for o in objs
]
else:
# received likes and reposts
ret['properties'][prop] = [
object_to_json(t, trim_nulls=False, entry_class='h-cite')
for t in tags if source.object_type(t) == type
]
if trim_nulls:
ret = util.trim_nulls(ret)
return ret
def _prepare_activity(a, reader=True):
"""Preprocesses an activity to prepare it to be rendered as Atom.
Modifies a in place.
Args:
a: ActivityStreams 1 activity dict
reader: boolean, whether the output will be rendered in a feed reader.
Currently just includes location if True, not otherwise.
"""
act_type = source.object_type(a)
if not act_type or act_type == 'post':
primary = a.get('object', {})
else:
primary = a
obj = a.setdefault('object', {})
# Render content as HTML; escape &s
obj['rendered_content'] = _encode_ampersands(
microformats2.render_content(primary,
include_location=reader,
render_attachments=True))
# Make sure every activity has the title field, since Atom <entry> requires
# the title element.
if not a.get('title'):
a['title'] = util.ellipsize(
_encode_ampersands(
a.get('displayName') or a.get('content') or obj.get('title')
or obj.get('displayName') or obj.get('content') or 'Untitled'))
# strip HTML tags. the Atom spec says title is plain text:
# http://atomenabled.org/developers/syndication/#requiredEntryElements
a['title'] = xml.sax.saxutils.escape(
BeautifulSoup(a['title']).get_text(''))
children = []
image_urls_seen = set()
image_atts = []
# normalize attachments, render attached notes/articles
attachments = a.get('attachments') or obj.get('attachments') or []
for att in attachments:
att['stream'] = util.get_first(att, 'stream')
type = att.get('objectType')
if type == 'image':
image_atts.append(util.get_first(att, 'image'))
continue
image_urls_seen |= set(util.get_urls(att, 'image'))
if type in ('note', 'article'):
html = microformats2.render_content(att,
include_location=reader,
render_attachments=True)
author = att.get('author')
if author:
name = microformats2.maybe_linked_name(
microformats2.object_to_json(author).get('properties', []))
html = '%s: %s' % (name.strip(), html)
children.append(html)
# render image(s) that we haven't already seen
for image in image_atts + util.get_list(obj, 'image'):
if not image:
continue
url = image.get('url')
parsed = urllib.parse.urlparse(url)
rest = urllib.parse.urlunparse(('', '') + parsed[2:])
img_src_re = re.compile(
r"""src *= *['"] *((https?:)?//%s)?%s *['"]""" %
(re.escape(parsed.netloc), re.escape(rest)))
if (url and url not in image_urls_seen
and not img_src_re.search(obj['rendered_content'])):
children.append(microformats2.img(url))
image_urls_seen.add(url)
obj['rendered_children'] = [
_encode_ampersands(child) for child in children
]
# make sure published and updated are strict RFC 3339 timestamps
for prop in 'published', 'updated':
val = obj.get(prop)
if val:
obj[prop] = util.maybe_iso8601_to_rfc3339(val)
# Atom timestamps are even stricter than RFC 3339: they can't be naive ie
# time zone unaware. They must have either an offset or the Z suffix.
# https://www.feedvalidator.org/docs/error/InvalidRFC3339Date.html
if not util.TIMEZONE_OFFSET_RE.search(obj[prop]):
obj[prop] += 'Z'
def activities_to_jsonfeed(activities, actor=None, title=None, feed_url=None,
home_page_url=None):
"""Converts ActivityStreams activities to a JSON feed.
Args:
activities: sequence of ActivityStreams activity dicts
actor: ActivityStreams actor dict, the author of the feed
title: string, the feed title
home_page_url: string, the home page URL
feed_url: the URL of the JSON Feed, if any. Included in the feed_url field.
Returns:
dict, JSON Feed data, ready to be JSON-encoded
"""
try:
iter(activities)
except TypeError:
raise TypeError('activities must be iterable')
if isinstance(activities, (dict, basestring)):
raise TypeError('activities may not be a dict or string')
def image_url(obj):
return util.get_first(obj, 'image', {}).get('url')
def actor_name(obj):
return obj.get('displayName') or obj.get('username')
if not actor:
actor = {}
items = []
for activity in activities:
obj = activity.get('object') or activity
if obj.get('objectType') == 'person':
continue
author = obj.get('author', {})
content = obj.get('content')
obj_title = obj.get('title') or obj.get('displayName')
item = {
'id': obj.get('id') or obj.get('url'),
'url': obj.get('url'),
'image': image_url(obj),
'title': obj_title if mf2util.is_name_a_title(obj_title, content) else None,
'summary': obj.get('summary'),
'content_html': content,
'date_published': obj.get('published'),
'date_modified': obj.get('updated'),
'author': {
'name': actor_name(author),
'url': author.get('url'),
'avatar': image_url(author),
},
'attachments': [],
}
for att in obj.get('attachments', []):
url = (util.get_first(att, 'stream') or util.get_first(att, 'image') or att
).get('url')
mime = mimetypes.guess_type(url)[0] if url else None
if (att.get('objectType') in ATTACHMENT_TYPES or
mime and mime.split('/')[0] in ATTACHMENT_TYPES):
item['attachments'].append({
'url': url or '',
'mime_type': mime,
'title': att.get('title'),
})
if not item['content_html']:
item['content_text'] = ''
items.append(item)
return util.trim_nulls({
'version': 'https://jsonfeed.org/version/1',
'title': title or actor_name(actor) or 'JSON Feed',
'feed_url': feed_url,
'home_page_url': home_page_url or actor.get('url'),
'author': {
'name': actor_name(actor),
'url': actor.get('url'),
'avatar': image_url(actor),
},
'items': items,
}, ignore='content_text')
def from_activities(activities,
actor=None,
title=None,
feed_url=None,
home_page_url=None,
hfeed=None):
"""Converts ActivityStreams activities to an RSS 2.0 feed.
Args:
activities: sequence of ActivityStreams activity dicts
actor: ActivityStreams actor dict, the author of the feed
title: string, the feed title
feed_url: string, the URL for this RSS feed
home_page_url: string, the home page URL
hfeed: dict, parsed mf2 h-feed, if available
Returns:
unicode string with RSS 2.0 XML
"""
try:
iter(activities)
except TypeError:
raise TypeError('activities must be iterable')
if isinstance(activities, (dict, str)):
raise TypeError('activities may not be a dict or string')
fg = FeedGenerator()
fg.id(feed_url)
assert feed_url
fg.link(href=feed_url, rel='self')
if home_page_url:
fg.link(href=home_page_url, rel='alternate')
# TODO: parse language from lang attribute:
# https://github.com/microformats/mf2py/issues/150
fg.language('en')
fg.generator('granary', uri='https://granary.io/')
hfeed = hfeed or {}
actor = actor or {}
image = (util.get_url(hfeed.get('properties', {}), 'photo')
or util.get_url(actor, 'image'))
if image:
fg.image(image)
props = hfeed.get('properties') or {}
content = microformats2.get_text(util.get_first(props, 'content', ''))
summary = util.get_first(props, 'summary', '')
desc = content or summary or '-'
fg.description(desc) # required
fg.title(title or util.ellipsize(desc)) # required
latest = None
feed_has_enclosure = False
for activity in activities:
obj = activity.get('object') or activity
if obj.get('objectType') == 'person':
continue
item = fg.add_entry()
url = obj.get('url')
id = obj.get('id') or url
item.id(id)
item.link(href=url)
item.guid(url, permalink=True)
# title (required)
title = (obj.get('title') or obj.get('displayName')
or util.ellipsize(obj.get('content', '-')))
# strip HTML tags
title = util.parse_html(title).get_text('').strip()
item.title(title)
content = microformats2.render_content(obj,
include_location=True,
render_attachments=True,
render_image=True)
if not content:
content = obj.get('summary')
if content:
item.content(content, type='CDATA')
categories = [
{
'term': t['displayName']
} for t in obj.get('tags', [])
if t.get('displayName') and t.get('verb') not in ('like', 'react',
'share')
and t.get('objectType') not in ('article', 'person', 'mention')
]
item.category(categories)
author = obj.get('author', {})
author = {
'name': author.get('displayName') or author.get('username'),
'uri': author.get('url'),
'email': author.get('email') or '-',
}
item.author(author)
published = obj.get('published') or obj.get('updated')
if published and isinstance(published, str):
try:
dt = mf2util.parse_datetime(published)
if not isinstance(dt, datetime):
dt = datetime.combine(dt, time.min)
if not dt.tzinfo:
dt = dt.replace(tzinfo=util.UTC)
item.published(dt)
if not latest or dt > latest:
latest = dt
except ValueError: # bad datetime string
pass
item_has_enclosure = False
for att in obj.get('attachments', []):
stream = util.get_first(att, 'stream') or att
if not stream:
continue
url = stream.get('url') or ''
mime = mimetypes.guess_type(url)[0] or ''
if (att.get('objectType') in ENCLOSURE_TYPES
or mime and mime.split('/')[0] in ENCLOSURE_TYPES):
if item_has_enclosure:
logging.info(
'Warning: item %s already has an RSS enclosure, skipping additional enclosure %s',
id, url)
continue
item_has_enclosure = feed_has_enclosure = True
item.enclosure(url=url,
type=mime,
length=str(stream.get('size', '')))
item.load_extension('podcast')
duration = stream.get('duration')
if duration:
item.podcast.itunes_duration(duration)
if feed_has_enclosure:
fg.load_extension('podcast')
fg.podcast.itunes_author(
actor.get('displayName') or actor.get('username'))
if summary:
fg.podcast.itunes_summary(summary)
fg.podcast.itunes_explicit('no')
fg.podcast.itunes_block(False)
name = author.get('name')
if name:
fg.podcast.itunes_author(name)
if image:
fg.podcast.itunes_image(image)
fg.podcast.itunes_category(categories)
if latest:
fg.lastBuildDate(latest)
return fg.rss_str(pretty=True).decode('utf-8')
def _create(self, obj, preview=None, include_link=source.OMIT_LINK,
ignore_formatting=False):
"""Creates or previews creating a tweet, reply tweet, retweet, or favorite.
https://dev.twitter.com/docs/api/1.1/post/statuses/update
https://dev.twitter.com/docs/api/1.1/post/statuses/retweet/:id
https://dev.twitter.com/docs/api/1.1/post/favorites/create
Args:
obj: ActivityStreams object
preview: boolean
include_link: string
ignore_formatting: boolean
Returns:
a CreationResult
If preview is True, the content will be a unicode string HTML
snippet. If False, it will be a dict with 'id' and 'url' keys
for the newly created Twitter object.
"""
assert preview in (False, True)
type = obj.get('objectType')
verb = obj.get('verb')
base_obj = self.base_object(obj)
base_id = base_obj.get('id')
base_url = base_obj.get('url')
is_reply = type == 'comment' or 'inReplyTo' in obj
image_urls = [image.get('url') for image in util.get_list(obj, 'image')]
video_url = util.get_first(obj, 'stream', {}).get('url')
has_media = (image_urls or video_url) and (type in ('note', 'article') or is_reply)
lat = obj.get('location', {}).get('latitude')
lng = obj.get('location', {}).get('longitude')
# prefer displayName over content for articles
type = obj.get('objectType')
base_url = self.base_object(obj).get('url')
prefer_content = type == 'note' or (base_url and (type == 'comment'
or obj.get('inReplyTo')))
content = self._content_for_create(obj, ignore_formatting=ignore_formatting,
prefer_name=not prefer_content,
strip_first_video_tag=bool(video_url))
if not content:
if type == 'activity':
content = verb
elif has_media:
content = ''
else:
return source.creation_result(
abort=False, # keep looking for things to publish,
error_plain='No content text found.',
error_html='No content text found.')
if is_reply and base_url:
# Twitter *used* to require replies to include an @-mention of the
# original tweet's author
# https://dev.twitter.com/docs/api/1.1/post/statuses/update#api-param-in_reply_to_status_id
# ...but now we use the auto_populate_reply_metadata query param instead:
# https://dev.twitter.com/overview/api/upcoming-changes-to-tweets
# the embed URL in the preview can't start with mobile. or www., so just
# hard-code it to twitter.com. index #1 is netloc.
parsed = urlparse.urlparse(base_url)
parts = parsed.path.split('/')
if len(parts) < 2 or not parts[1]:
raise ValueError('Could not determine author of in-reply-to URL %s' % base_url)
reply_to_prefix = '@%s ' % parts[1].lower()
if content.lower().startswith(reply_to_prefix):
content = content[len(reply_to_prefix):]
parsed = list(parsed)
parsed[1] = self.DOMAIN
base_url = urlparse.urlunparse(parsed)
# need a base_url with the tweet id for the embed HTML below. do this
# *after* checking the real base_url for in-reply-to author username.
if base_id and not base_url:
base_url = 'https://twitter.com/-/statuses/' + base_id
if is_reply and not base_url:
return source.creation_result(
abort=True,
error_plain='Could not find a tweet to reply to.',
error_html='Could not find a tweet to <a href="http://indiewebcamp.com/reply">reply to</a>. '
'Check that your post has an <a href="http://indiewebcamp.com/comment">in-reply-to</a> '
'link a Twitter URL or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Twitter.')
# truncate and ellipsize content if it's over the character
# count. URLs will be t.co-wrapped, so include that when counting.
content = self._truncate(
content, obj.get('url'), include_link, type)
# linkify defaults to Twitter's link shortening behavior
preview_content = util.linkify(content, pretty=True, skip_bare_cc_tlds=True)
if type == 'activity' and verb == 'like':
if not base_url:
return source.creation_result(
abort=True,
error_plain='Could not find a tweet to like.',
error_html='Could not find a tweet to <a href="http://indiewebcamp.com/favorite">favorite</a>. '
'Check that your post has a like-of link to a Twitter URL or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Twitter.')
if preview:
return source.creation_result(
description='<span class="verb">favorite</span> <a href="%s">'
'this tweet</a>:\n%s' % (base_url, self.embed_post(base_obj)))
else:
data = urllib.urlencode({'id': base_id})
self.urlopen(API_POST_FAVORITE, data=data)
resp = {'type': 'like'}
elif type == 'activity' and verb == 'share':
if not base_url:
return source.creation_result(
abort=True,
error_plain='Could not find a tweet to retweet.',
error_html='Could not find a tweet to <a href="http://indiewebcamp.com/repost">retweet</a>. '
'Check that your post has a repost-of link to a Twitter URL or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Twitter.')
if preview:
return source.creation_result(
description='<span class="verb">retweet</span> <a href="%s">'
'this tweet</a>:\n%s' % (base_url, self.embed_post(base_obj)))
else:
data = urllib.urlencode({'id': base_id})
resp = self.urlopen(API_POST_RETWEET % base_id, data=data)
resp['type'] = 'repost'
elif type in ('note', 'article') or is_reply: # a tweet
content = unicode(content).encode('utf-8')
data = {'status': content}
if is_reply:
description = \
'<span class="verb">@-reply</span> to <a href="%s">this tweet</a>:\n%s' % (
base_url, self.embed_post(base_obj))
data.update({
'in_reply_to_status_id': base_id,
'auto_populate_reply_metadata': 'true',
})
else:
description = '<span class="verb">tweet</span>:'
if video_url:
preview_content += ('<br /><br /><video controls src="%s"><a href="%s">'
'this video</a></video>' % (video_url, video_url))
if not preview:
ret = self.upload_video(video_url)
if isinstance(ret, source.CreationResult):
return ret
data['media_ids'] = ret
elif image_urls:
num_urls = len(image_urls)
if num_urls > MAX_MEDIA:
image_urls = image_urls[:MAX_MEDIA]
logging.warning('Found %d photos! Only using the first %d: %r',
num_urls, MAX_MEDIA, image_urls)
preview_content += '<br /><br />' + ' '.join(
'<img src="%s" />' % url for url in image_urls)
if not preview:
ret = self.upload_images(image_urls)
if isinstance(ret, source.CreationResult):
return ret
data['media_ids'] = ','.join(ret)
if lat and lng:
preview_content += (
'<div>at <a href="https://maps.google.com/maps?q=%s,%s">'
'%s, %s</a></div>' % (lat, lng, lat, lng))
data['lat'] = lat
data['long'] = lng
if preview:
return source.creation_result(content=preview_content, description=description)
else:
resp = self.urlopen(API_POST_TWEET, data=urllib.urlencode(data))
resp['type'] = 'comment' if is_reply else 'post'
elif (verb and verb.startswith('rsvp-')) or verb == 'invite':
return source.creation_result(
abort=True,
error_plain='Cannot publish RSVPs to Twitter.',
error_html='This looks like an <a href="http://indiewebcamp.com/rsvp">RSVP</a>. '
'Publishing events or RSVPs to Twitter is not supported.')
else:
return source.creation_result(
abort=False,
error_plain='Cannot publish type=%s, verb=%s to Twitter' % (type, verb),
error_html='Cannot publish type=%s, verb=%s to Twitter' % (type, verb))
id_str = resp.get('id_str')
if id_str:
resp.update({'id': id_str, 'url': self.tweet_url(resp)})
elif 'url' not in resp:
resp['url'] = base_url
return source.creation_result(resp)
def _create(self,
obj,
preview,
include_link=source.OMIT_LINK,
ignore_formatting=False):
"""Creates or previews creating for the previous two methods.
https://www.flickr.com/services/api/upload.api.html
https://www.flickr.com/services/api/flickr.photos.comments.addComment.html
https://www.flickr.com/services/api/flickr.favorites.add.html
https://www.flickr.com/services/api/flickr.photos.people.add.html
Args:
obj: ActivityStreams object
preview: boolean
include_link: string
ignore_formatting: boolean
Return:
a CreationResult
"""
# photo, comment, or like
type = source.object_type(obj)
logging.debug('publishing object type %s to Flickr', type)
link_text = '(Originally published at: %s)' % obj.get('url')
image_url = util.get_first(obj, 'image', {}).get('url')
video_url = util.get_first(obj, 'stream', {}).get('url')
content = self._content_for_create(
obj,
ignore_formatting=ignore_formatting,
strip_first_video_tag=bool(video_url))
if (video_url or image_url) and type in ('note', 'article'):
name = obj.get('displayName')
people = self._get_person_tags(obj)
hashtags = [
t.get('displayName') for t in obj.get('tags', [])
if t.get('objectType') == 'hashtag' and t.get('displayName')
]
lat = obj.get('location', {}).get('latitude')
lng = obj.get('location', {}).get('longitude')
# if name does not represent an explicit title, then we'll just
# use it as the title and wipe out the content
if name and content and not mf2util.is_name_a_title(name, content):
name = content
content = None
# add original post link
if include_link == source.INCLUDE_LINK:
content = ((content + '\n\n') if content else '') + link_text
if preview:
preview_content = ''
if name:
preview_content += '<h4>%s</h4>' % name
if content:
preview_content += '<div>%s</div>' % content
if hashtags:
preview_content += '<div> %s</div>' % ' '.join(
'#' + t for t in hashtags)
if people:
preview_content += '<div> with %s</div>' % ', '.join(
('<a href="%s">%s</a>' %
(p.get('url'), p.get('displayName')
or 'User %s' % p.get('id')) for p in people))
if lat and lng:
preview_content += '<div> at <a href="https://maps.google.com/maps?q=%s,%s">%s, %s</a></div>' % (
lat, lng, lat, lng)
if video_url:
preview_content += (
'<video controls src="%s"><a href="%s">this video'
'</a></video>' % (video_url, video_url))
else:
preview_content += '<img src="%s" />' % image_url
return source.creation_result(content=preview_content,
description='post')
params = []
if name:
params.append(('title', name))
if content:
params.append(('description', content.encode('utf-8')))
if hashtags:
params.append(('tags', ','.join(
('"%s"' % t if ' ' in t else t).encode('utf-8')
for t in hashtags)))
file = util.urlopen(video_url or image_url)
try:
resp = self.upload(params, file)
except requests.exceptions.ConnectionError as e:
if e.args[0].message.startswith(
'Request exceeds 10 MiB limit'):
msg = 'Sorry, photos and videos must be under 10MB.'
return source.creation_result(error_plain=msg,
error_html=msg)
else:
raise
photo_id = resp.get('id')
resp.update({
'type':
'post',
'url':
self.photo_url(self.path_alias() or self.user_id(), photo_id),
})
if video_url:
resp['granary_message'] = \
"Note that videos take time to process before they're visible."
# add person tags
for person_id in sorted(p.get('id') for p in people):
self.call_api_method('flickr.photos.people.add', {
'photo_id': photo_id,
'user_id': person_id,
})
# add location
if lat and lng:
self.call_api_method('flickr.photos.geo.setLocation', {
'photo_id': photo_id,
'lat': lat,
'lon': lng,
})
return source.creation_result(resp)
base_obj = self.base_object(obj)
base_id = base_obj.get('id')
base_url = base_obj.get('url')
# maybe a comment on a flickr photo?
if type == 'comment' or obj.get('inReplyTo'):
if not base_id:
return source.creation_result(
abort=True,
error_plain='Could not find a photo to comment on.',
error_html=
'Could not find a photo to <a href="http://indiewebcamp.com/reply">comment on</a>. '
'Check that your post has an <a href="http://indiewebcamp.com/comment">in-reply-to</a> '
'link to a Flickr photo or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Flickr.'
)
if include_link == source.INCLUDE_LINK:
content += '\n\n' + link_text
if preview:
return source.creation_result(
content=content,
description='comment on <a href="%s">this photo</a>.' %
base_url)
resp = self.call_api_method(
'flickr.photos.comments.addComment', {
'photo_id': base_id,
'comment_text': content.encode('utf-8'),
})
resp = resp.get('comment', {})
resp.update({
'type': 'comment',
'url': resp.get('permalink'),
})
return source.creation_result(resp)
if type == 'like':
if not base_id:
return source.creation_result(
abort=True,
error_plain='Could not find a photo to favorite.',
error_html=
'Could not find a photo to <a href="http://indiewebcamp.com/like">favorite</a>. '
'Check that your post has an <a href="http://indiewebcamp.com/like">like-of</a> '
'link to a Flickr photo or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Flickr.'
)
if preview:
return source.creation_result(
description='favorite <a href="%s">this photo</a>.' %
base_url)
# this method doesn't return any data
self.call_api_method('flickr.favorites.add', {
'photo_id': base_id,
})
# TODO should we canonicalize the base_url (e.g. removing trailing path
# info like "/in/contacts/")
return source.creation_result({
'type':
'like',
'url':
'%s#favorited-by-%s' % (base_url, self.user_id()),
})
return source.creation_result(
abort=False,
error_plain='Cannot publish type=%s to Flickr.' % type,
error_html='Cannot publish type=%s to Flickr.' % type)
def _prepare_activity(a, reader=True):
"""Preprocesses an activity to prepare it to be rendered as Atom.
Modifies a in place.
Args:
a: ActivityStreams 1 activity dict
reader: boolean, whether the output will be rendered in a feed reader.
Currently just includes location if True, not otherwise.
"""
act_type = source.object_type(a)
obj = util.get_first(a, 'object', default={})
primary = obj if (not act_type or act_type == 'post') else a
# Render content as HTML; escape &s
obj['rendered_content'] = _encode_ampersands(microformats2.render_content(
primary, include_location=reader, render_attachments=True))
# Make sure every activity has the title field, since Atom <entry> requires
# the title element.
if not a.get('title'):
a['title'] = util.ellipsize(_encode_ampersands(
a.get('displayName') or a.get('content') or obj.get('title') or
obj.get('displayName') or obj.get('content') or 'Untitled'))
# strip HTML tags. the Atom spec says title is plain text:
# http://atomenabled.org/developers/syndication/#requiredEntryElements
a['title'] = xml.sax.saxutils.escape(BeautifulSoup(a['title']).get_text(''))
children = []
image_urls_seen = set()
image_atts = []
# normalize actor images
for elem in a, obj:
actor = elem.get('actor')
if actor:
actor['image'] = util.get_first(actor, 'image')
# normalize attachments, render attached notes/articles
attachments = a.get('attachments') or obj.get('attachments') or []
for att in attachments:
att['stream'] = util.get_first(att, 'stream')
type = att.get('objectType')
if type == 'image':
att['image'] = util.get_first(att, 'image')
image_atts.append(att['image'])
continue
image_urls_seen |= set(util.get_urls(att, 'image'))
if type in ('note', 'article'):
html = microformats2.render_content(att, include_location=reader,
render_attachments=True)
author = att.get('author')
if author:
name = microformats2.maybe_linked_name(
microformats2.object_to_json(author).get('properties') or {})
html = '%s: %s' % (name.strip(), html)
children.append(html)
# render image(s) that we haven't already seen
for image in image_atts + util.get_list(obj, 'image'):
if not image:
continue
url = image.get('url')
parsed = urllib.parse.urlparse(url)
rest = urllib.parse.urlunparse(('', '') + parsed[2:])
img_src_re = re.compile(r"""src *= *['"] *((https?:)?//%s)?%s *['"]""" %
(re.escape(parsed.netloc), re.escape(rest)))
if (url and url not in image_urls_seen and
not img_src_re.search(obj['rendered_content'])):
children.append(microformats2.img(url))
image_urls_seen.add(url)
obj['rendered_children'] = [_encode_ampersands(child) for child in children]
# make sure published and updated are strict RFC 3339 timestamps
for prop in 'published', 'updated':
val = obj.get(prop)
if val:
obj[prop] = util.maybe_iso8601_to_rfc3339(val)
# Atom timestamps are even stricter than RFC 3339: they can't be naive ie
# time zone unaware. They must have either an offset or the Z suffix.
# https://www.feedvalidator.org/docs/error/InvalidRFC3339Date.html
if not util.TIMEZONE_OFFSET_RE.search(obj[prop]):
obj[prop] += 'Z'
def try_activitypub(self):
source = util.get_required_param(self, 'source')
# fetch source page, convert to ActivityStreams
source_resp = common.requests_get(source)
source_url = source_resp.url or source
source_mf2 = mf2py.parse(source_resp.text, url=source_url)
# logging.debug('Parsed mf2 for %s: %s', source_resp.url, json.dumps(source_mf2, indent=2))
entry = mf2util.find_first_entry(source_mf2, ['h-entry'])
logging.info('First entry: %s', json.dumps(entry, indent=2))
# make sure it has url, since we use that for AS2 id, which is required
# for ActivityPub.
props = entry.setdefault('properties', {})
if not props.get('url'):
props['url'] = [source_url]
source_obj = microformats2.json_to_object(entry, fetch_mf2=True)
logging.info('Converted to AS: %s', json.dumps(source_obj, indent=2))
# fetch target page as AS object. target is first in-reply-to, like-of,
# or repost-of, *not* target query param.)
target = util.get_url(util.get_first(source_obj, 'inReplyTo') or
util.get_first(source_obj, 'object'))
if not target:
common.error(self, 'No u-in-reply-to, u-like-of, or u-repost-of '
'found in %s' % source_url)
try:
target_resp = common.get_as2(target)
except (requests.HTTPError, exc.HTTPBadGateway) as e:
if (e.response.status_code // 100 == 2 and
common.content_type(e.response).startswith('text/html')):
self.resp = Response.get_or_create(
source=source_url, target=e.response.url or target,
direction='out', source_mf2=json.dumps(source_mf2))
return self.send_salmon(source_obj, target_resp=e.response)
raise
target_url = target_resp.url or target
self.resp = Response.get_or_create(
source=source_url, target=target_url, direction='out',
protocol='activitypub', source_mf2=json.dumps(source_mf2))
# find actor's inbox
target_obj = target_resp.json()
inbox_url = target_obj.get('inbox')
if not inbox_url:
# TODO: test actor/attributedTo and not, with/without inbox
actor = target_obj.get('actor') or target_obj.get('attributedTo')
if isinstance(actor, dict):
inbox_url = actor.get('inbox')
actor = actor.get('url')
if not inbox_url and not actor:
common.error(self, 'Target object has no actor or attributedTo URL')
if not inbox_url:
# fetch actor as AS object
actor = common.get_as2(actor).json()
inbox_url = actor.get('inbox')
if not inbox_url:
# TODO: probably need a way to save errors like this so that we can
# return them if ostatus fails too.
# common.error(self, 'Target actor has no inbox')
return self.send_salmon(source_obj, target_resp=target_resp)
# convert to AS2
source_domain = urlparse.urlparse(source_url).netloc
key = MagicKey.get_or_create(source_domain)
source_activity = common.postprocess_as2(
as2.from_as1(source_obj), target=target_obj, key=key)
if self.resp.status == 'complete':
source_activity['type'] = 'Update'
# prepare HTTP Signature (required by Mastodon)
# https://w3c.github.io/activitypub/#authorization-lds
# https://tools.ietf.org/html/draft-cavage-http-signatures-07
# https://github.com/tootsuite/mastodon/issues/4906#issuecomment-328844846
acct = 'acct:%s@%s' % (source_domain, source_domain)
auth = HTTPSignatureAuth(secret=key.private_pem(), key_id=acct,
algorithm='rsa-sha256')
# deliver source object to target actor's inbox.
headers = {
'Content-Type': common.CONTENT_TYPE_AS2,
# required for HTTP Signature
# https://tools.ietf.org/html/draft-cavage-http-signatures-07#section-2.1.3
'Date': datetime.datetime.utcnow().strftime('%a, %d %b %Y %H:%M:%S GMT'),
}
inbox_url = urlparse.urljoin(target_url, inbox_url)
resp = common.requests_post(inbox_url, json=source_activity, auth=auth,
headers=headers)
self.response.status_int = resp.status_code
if resp.status_code == 202:
self.response.write('202 response! If this is Mastodon 1.x, their '
'signature verification probably failed. :(\n')
self.response.write(resp.text)
def render_content(obj, include_location=True, synthesize_content=True):
"""Renders the content of an ActivityStreams object.
Includes tags, mentions, and non-note/article attachments. (Note/article
attachments are converted to mf2 children in object_to_json and then rendered
in json_to_html.)
Args:
obj: decoded JSON ActivityStreams object
include_location: whether to render location, if provided
synthesize_content: whether to generate synthetic content if the object
doesn't have its own, e.g. 'likes this.' or 'shared this.'
Returns:
string, rendered HTML
"""
content = obj.get('content', '')
# extract tags. preserve order but de-dupe, ie don't include a tag more than
# once.
seen_ids = set()
mentions = []
tags = {} # maps string objectType to list of tag objects
for t in obj.get('tags', []):
id = t.get('id')
if id and id in seen_ids:
continue
seen_ids.add(id)
if 'startIndex' in t and 'length' in t:
mentions.append(t)
else:
tags.setdefault(source.object_type(t), []).append(t)
# linkify embedded mention tags inside content.
if mentions:
mentions.sort(key=lambda t: t['startIndex'])
last_end = 0
orig = content
content = ''
for tag in mentions:
start = tag['startIndex']
end = start + tag['length']
content += orig[last_end:start]
content += '<a href="%s">%s</a>' % (tag['url'], orig[start:end])
last_end = end
content += orig[last_end:]
# convert newlines to <br>s
# do this *after* linkifying tags so we don't have to shuffle indices over
content = content.replace('\n', '<br />\n')
# linkify embedded links. ignore the "mention" tags that we added ourselves.
# TODO: fix the bug in test_linkify_broken() in webutil/util_test.py, then
# uncomment this.
# if content:
# content = util.linkify(content)
# attachments, e.g. links (aka articles)
# TODO: use oEmbed? http://oembed.com/ , http://code.google.com/p/python-oembed/
attachments = [a for a in obj.get('attachments', [])
if a.get('objectType') not in ('note', 'article')]
for tag in attachments + tags.pop('article', []):
name = tag.get('displayName', '')
open_a_tag = False
if tag.get('objectType') == 'video':
video = util.get_first(tag, 'stream') or util.get_first(obj, 'stream')
poster = util.get_first(tag, 'image', {})
if video and video.get('url'):
content += '\n<p>%s' % vid(video['url'], poster.get('url'), 'thumbnail')
else:
content += '\n<p>'
url = tag.get('url') or obj.get('url')
if url:
content += '\n<a class="link" href="%s">' % url
open_a_tag = True
image = util.get_first(tag, 'image') or util.get_first(obj, 'image')
if image and image.get('url'):
content += '\n' + img(image['url'], 'thumbnail', name)
if name:
content += '\n<span class="name">%s</span>' % name
if open_a_tag:
content += '\n</a>'
summary = tag.get('summary')
if summary and summary != name:
content += '\n<span class="summary">%s</span>' % summary
content += '\n</p>'
# generate share/like contexts if the activity does not have content
# of its own
for as_type, verb in [('share', 'Shared'), ('like', 'Likes')]:
obj_type = source.object_type(obj)
if (not synthesize_content or obj_type != as_type or 'object' not in obj or
'content' in obj):
continue
targets = util.get_list(obj, 'object')
if not targets:
continue
for target in targets:
# sometimes likes don't have enough content to render anything
# interesting
if 'url' in target and set(target) <= set(['url', 'objectType']):
content += '<a href="%s">%s this.</a>' % (
target.get('url'), verb.lower())
else:
author = target.get('author', target.get('actor', {}))
# special case for twitter RT's
if obj_type == 'share' and 'url' in obj and re.search(
'^https?://(?:www\.|mobile\.)?twitter\.com/', obj.get('url')):
content += 'RT <a href="%s">@%s</a> ' % (
target.get('url', '#'), author.get('username'))
else:
# image looks bad in the simplified rendering
author = {k: v for k, v in author.iteritems() if k != 'image'}
content += '%s <a href="%s">%s</a> by %s' % (
verb, target.get('url', '#'),
target.get('displayName', target.get('title', 'a post')),
hcard_to_html(object_to_json(author, default_object_type='person')),
)
content += render_content(target, include_location=include_location,
synthesize_content=synthesize_content)
# only include the first context in the content (if there are
# others, they'll be included as separate properties)
break
break
# location
loc = obj.get('location')
if include_location and loc:
content += '\n' + hcard_to_html(
object_to_json(loc, default_object_type='place'),
parent_props=['p-location'])
# these are rendered manually in json_to_html()
for type in 'like', 'share', 'react', 'person':
tags.pop(type, None)
# render the rest
content += tags_to_html(tags.pop('hashtag', []), 'p-category')
content += tags_to_html(tags.pop('mention', []), 'u-mention')
content += tags_to_html(sum(tags.values(), []), 'tag')
return content
def object_to_json(obj,
trim_nulls=True,
entry_class='h-entry',
default_object_type=None,
synthesize_content=True):
"""Converts an ActivityStreams object to microformats2 JSON.
Args:
obj: dict, a decoded JSON ActivityStreams object
trim_nulls: boolean, whether to remove elements with null or empty values
entry_class: string or sequence, the mf2 class(es) that entries should be
given (e.g. 'h-cite' when parsing a reference to a foreign entry).
defaults to 'h-entry'
default_object_type: string, the ActivityStreams objectType to use if one
is not present. defaults to None
synthesize_content: whether to generate synthetic content if the object
doesn't have its own, e.g. 'likes this.' or 'shared this.'
Returns:
dict, decoded microformats2 JSON
"""
if not obj or not isinstance(obj, dict):
return {}
obj_type = source.object_type(obj) or default_object_type
# if the activity type is a post, then it's really just a conduit
# for the object. for other verbs, the activity itself is the
# interesting thing
if obj_type == 'post':
primary = obj.get('object', {})
obj_type = source.object_type(primary) or default_object_type
else:
primary = obj
# TODO: extract snippet
name = primary.get('displayName', primary.get('title'))
summary = primary.get('summary')
author = obj.get('author', obj.get('actor', {}))
in_reply_tos = obj.get('inReplyTo',
obj.get('context', {}).get('inReplyTo', []))
is_rsvp = obj_type in ('rsvp-yes', 'rsvp-no', 'rsvp-maybe')
if (is_rsvp or obj_type == 'react') and obj.get('object'):
objs = obj['object']
in_reply_tos.extend(objs if isinstance(objs, list) else [objs])
# maps objectType to list of objects
attachments = defaultdict(list)
for prop in 'attachments', 'tags':
for elem in get_list(primary, prop):
attachments[elem.get('objectType')].append(elem)
# construct mf2!
ret = {
'type': (AS_TO_MF2_TYPE.get(obj_type) or [entry_class] if isinstance(
entry_class, basestring) else list(entry_class)),
'properties': {
'uid': [obj.get('id') or ''],
'numeric-id': [obj.get('numeric_id') or ''],
'name': [name],
'nickname': [obj.get('username') or ''],
'summary': [summary],
'url': (list(object_urls(obj) or object_urls(primary)) +
obj.get('upstreamDuplicates', [])),
'photo':
dedupe_urls(
get_urls(attachments, 'image', 'image') +
get_urls(primary, 'image')),
'video':
dedupe_urls(
get_urls(attachments, 'video', 'stream') +
get_urls(primary, 'stream')),
'audio':
get_urls(attachments, 'audio', 'stream'),
'published': [obj.get('published', primary.get('published', ''))],
'updated': [obj.get('updated', primary.get('updated', ''))],
'content': [{
'value':
xml.sax.saxutils.unescape(primary.get('content', '')),
'html':
render_content(primary,
include_location=False,
synthesize_content=synthesize_content),
}],
'in-reply-to':
util.trim_nulls([o.get('url') for o in in_reply_tos]),
'author': [
object_to_json(author,
trim_nulls=False,
default_object_type='person')
],
'location': [
object_to_json(primary.get('location', {}),
trim_nulls=False,
default_object_type='place')
],
'comment': [
object_to_json(c, trim_nulls=False, entry_class='h-cite')
for c in obj.get('replies', {}).get('items', [])
],
'start': [primary.get('startTime')],
'end': [primary.get('endTime')],
},
'children': [
object_to_json(a,
trim_nulls=False,
entry_class=['u-quotation-of', 'h-cite'])
for a in attachments['note'] + attachments['article']
]
}
# hashtags and person tags
tags = obj.get('tags', []) or get_first(obj, 'object', {}).get('tags', [])
ret['properties']['category'] = []
for tag in tags:
if tag.get('objectType') == 'person':
ret['properties']['category'].append(
object_to_json(tag, entry_class='u-category h-card'))
elif tag.get('objectType') == 'hashtag':
name = tag.get('displayName')
if name:
ret['properties']['category'].append(name)
# rsvp
if is_rsvp:
ret['properties']['rsvp'] = [obj_type[len('rsvp-'):]]
elif obj_type == 'invite':
invitee = object_to_json(obj.get('object'),
trim_nulls=False,
default_object_type='person')
ret['properties']['invitee'] = [invitee]
# like and repost mentions
for type, prop in ('favorite', 'like'), ('like', 'like'), ('share',
'repost'):
if obj_type == type:
# The ActivityStreams spec says the object property should always be a
# single object, but it's useful to let it be a list, e.g. when a like has
# multiple targets, e.g. a like of a post with original post URLs in it,
# which brid.gy does.
objs = get_list(obj, 'object')
ret['properties'][prop + '-of'] = [
# flatten contexts that are just a url
o['url']
if 'url' in o and set(o.keys()) <= set(['url', 'objectType'])
else object_to_json(o, trim_nulls=False, entry_class='h-cite')
for o in objs
]
else:
# received likes and reposts
ret['properties'][prop] = [
object_to_json(t, trim_nulls=False, entry_class='h-cite')
for t in tags if source.object_type(t) == type
]
# latitude & longitude
lat = long = None
position = ISO_6709_RE.match(primary.get('position') or '')
if position:
lat, long = position.groups()
if not lat:
lat = primary.get('latitude')
if not long:
long = primary.get('longitude')
if lat:
ret['properties']['latitude'] = [str(lat)]
if long:
ret['properties']['longitude'] = [str(long)]
if trim_nulls:
ret = util.trim_nulls(ret)
return ret
def render_content(obj, include_location=True, synthesize_content=True):
"""Renders the content of an ActivityStreams object.
Includes tags, mentions, and non-note/article attachments. (Note/article
attachments are converted to mf2 children in object_to_json and then rendered
in json_to_html.)
Args:
obj: decoded JSON ActivityStreams object
include_location: whether to render location, if provided
synthesize_content: whether to generate synthetic content if the object
doesn't have its own, e.g. 'likes this.' or 'shared this.'
Returns: string, rendered HTML
"""
content = obj.get('content', '')
# extract tags. preserve order but de-dupe, ie don't include a tag more than
# once.
seen_ids = set()
mentions = []
tags = {} # maps string objectType to list of tag objects
for t in obj.get('tags', []):
id = t.get('id')
if id and id in seen_ids:
continue
seen_ids.add(id)
if 'startIndex' in t and 'length' in t:
mentions.append(t)
else:
tags.setdefault(source.object_type(t), []).append(t)
# linkify embedded mention tags inside content.
if mentions:
mentions.sort(key=lambda t: t['startIndex'])
last_end = 0
orig = content
content = ''
for tag in mentions:
start = tag['startIndex']
end = start + tag['length']
content += orig[last_end:start]
content += '<a href="%s">%s</a>' % (tag['url'], orig[start:end])
last_end = end
content += orig[last_end:]
# convert newlines to <br>s
# do this *after* linkifying tags so we don't have to shuffle indices over
content = content.replace('\n', '<br />\n')
# linkify embedded links. ignore the "mention" tags that we added ourselves.
# TODO: fix the bug in test_linkify_broken() in webutil/util_test.py, then
# uncomment this.
# if content:
# content = util.linkify(content)
# attachments, e.g. links (aka articles)
# TODO: use oEmbed? http://oembed.com/ , http://code.google.com/p/python-oembed/
attachments = [
a for a in obj.get('attachments', [])
if a.get('objectType') not in ('note', 'article')
]
for tag in attachments + tags.pop('article', []):
name = tag.get('displayName', '')
open_a_tag = False
if tag.get('objectType') == 'video':
video = util.get_first(tag, 'stream') or util.get_first(
obj, 'stream')
poster = util.get_first(tag, 'image', {})
if video and video.get('url'):
content += '\n<p>%s' % vid(video['url'], poster.get('url'),
'thumbnail')
else:
content += '\n<p>'
url = tag.get('url') or obj.get('url')
if url:
content += '\n<a class="link" href="%s">' % url
open_a_tag = True
image = util.get_first(tag, 'image') or util.get_first(
obj, 'image')
if image and image.get('url'):
content += '\n' + img(image['url'], 'thumbnail', name)
if name:
content += '\n<span class="name">%s</span>' % name
if open_a_tag:
content += '\n</a>'
summary = tag.get('summary')
if summary and summary != name:
content += '\n<span class="summary">%s</span>' % summary
content += '\n</p>'
# generate share/like contexts if the activity does not have content
# of its own
for as_type, verb in [('share', 'Shared'), ('like', 'Likes')]:
obj_type = source.object_type(obj)
if (not synthesize_content or obj_type != as_type
or 'object' not in obj or 'content' in obj):
continue
targets = util.get_list(obj, 'object')
if not targets:
continue
for target in targets:
# sometimes likes don't have enough content to render anything
# interesting
if 'url' in target and set(target) <= set(['url', 'objectType']):
content += '<a href="%s">%s this.</a>' % (target.get('url'),
verb.lower())
else:
author = target.get('author', target.get('actor', {}))
# special case for twitter RT's
if obj_type == 'share' and 'url' in obj and re.search(
'^https?://(?:www\.|mobile\.)?twitter\.com/',
obj.get('url')):
content += 'RT <a href="%s">@%s</a> ' % (target.get(
'url', '#'), author.get('username'))
else:
# image looks bad in the simplified rendering
author = {
k: v
for k, v in author.iteritems() if k != 'image'
}
content += '%s <a href="%s">%s</a> by %s' % (
verb,
target.get('url', '#'),
target.get('displayName', target.get(
'title', 'a post')),
hcard_to_html(
object_to_json(author,
default_object_type='person')),
)
content += render_content(
target,
include_location=include_location,
synthesize_content=synthesize_content)
# only include the first context in the content (if there are
# others, they'll be included as separate properties)
break
break
# location
loc = obj.get('location')
if include_location and loc:
content += '\n' + hcard_to_html(object_to_json(
loc, default_object_type='place'),
parent_props=['p-location'])
# these are rendered manually in json_to_html()
for type in 'like', 'share', 'react', 'person':
tags.pop(type, None)
# render the rest
content += tags_to_html(tags.pop('hashtag', []), 'p-category')
content += tags_to_html(tags.pop('mention', []), 'u-mention')
content += tags_to_html(sum(tags.values(), []), 'tag')
return content
def object_to_json(obj, trim_nulls=True, entry_class='h-entry',
default_object_type=None, synthesize_content=True):
"""Converts an ActivityStreams object to microformats2 JSON.
Args:
obj: dict, a decoded JSON ActivityStreams object
trim_nulls: boolean, whether to remove elements with null or empty values
entry_class: string, the mf2 class that entries should be given (e.g.
'h-cite' when parsing a reference to a foreign entry). defaults to
'h-entry'
default_object_type: string, the ActivityStreams objectType to use if one
is not present. defaults to None
synthesize_content: whether to generate synthetic content if the object
doesn't have its own, e.g. 'likes this.' or 'shared this.'
Returns:
dict, decoded microformats2 JSON
"""
if not obj or not isinstance(obj, dict):
return {}
obj_type = source.object_type(obj) or default_object_type
# if the activity type is a post, then it's really just a conduit
# for the object. for other verbs, the activity itself is the
# interesting thing
if obj_type == 'post':
primary = obj.get('object', {})
obj_type = source.object_type(primary) or default_object_type
else:
primary = obj
# TODO: extract snippet
name = primary.get('displayName', primary.get('title'))
summary = primary.get('summary')
author = obj.get('author', obj.get('actor', {}))
in_reply_tos = obj.get(
'inReplyTo', obj.get('context', {}).get('inReplyTo', []))
is_rsvp = obj_type in ('rsvp-yes', 'rsvp-no', 'rsvp-maybe')
if (is_rsvp or obj_type == 'react') and obj.get('object'):
objs = obj['object']
in_reply_tos.extend(objs if isinstance(objs, list) else [objs])
# TODO: more tags. most will be p-category?
ret = {
'type': (['h-card'] if obj_type == 'person'
else ['h-card', 'p-location'] if obj_type == 'place'
else [entry_class]),
'properties': {
'uid': [obj.get('id', '')],
'name': [name],
'summary': [summary],
'url': (list(object_urls(obj) or object_urls(primary)) +
obj.get('upstreamDuplicates', [])),
'photo': [image.get('url') for image in
(util.get_list(obj, 'image') or util.get_list(primary, 'image'))],
'video': [obj.get('stream', primary.get('stream', {})).get('url')],
'published': [obj.get('published', primary.get('published', ''))],
'updated': [obj.get('updated', primary.get('updated', ''))],
'content': [{
'value': xml.sax.saxutils.unescape(primary.get('content', '')),
'html': render_content(primary, include_location=False,
synthesize_content=synthesize_content),
}],
'in-reply-to': util.trim_nulls([o.get('url') for o in in_reply_tos]),
'author': [object_to_json(
author, trim_nulls=False, default_object_type='person')],
'location': [object_to_json(
primary.get('location', {}), trim_nulls=False,
default_object_type='place')],
'latitude': primary.get('latitude'),
'longitude': primary.get('longitude'),
'comment': [object_to_json(c, trim_nulls=False, entry_class='h-cite')
for c in obj.get('replies', {}).get('items', [])],
},
'children': [object_to_json(c, trim_nulls=False, entry_class='h-cite')
for c in primary.get('attachments', [])
if c.get('objectType') in ('note', 'article')],
}
# hashtags and person tags
tags = obj.get('tags', []) or util.get_first(obj, 'object', {}).get('tags', [])
ret['properties']['category'] = []
for tag in tags:
if tag.get('objectType') == 'person':
cls = 'u-category h-card'
elif tag.get('objectType') == 'hashtag':
cls = 'u-category'
else:
continue
ret['properties']['category'].append(object_to_json(tag, entry_class=cls))
# rsvp
if is_rsvp:
ret['properties']['rsvp'] = [obj_type[len('rsvp-'):]]
elif obj_type == 'invite':
invitee = object_to_json(obj.get('object'), trim_nulls=False,
default_object_type='person')
ret['properties']['invitee'] = [invitee]
# like and repost mentions
for type, prop in ('like', 'like'), ('share', 'repost'):
if obj_type == type:
# The ActivityStreams spec says the object property should always be a
# single object, but it's useful to let it be a list, e.g. when a like has
# multiple targets, e.g. a like of a post with original post URLs in it,
# which brid.gy does.
objs = util.get_list(obj, 'object')
ret['properties'][prop + '-of'] = [
# flatten contexts that are just a url
o['url'] if 'url' in o and set(o.keys()) <= set(['url', 'objectType'])
else object_to_json(o, trim_nulls=False, entry_class='h-cite')
for o in objs]
else:
# received likes and reposts
ret['properties'][prop] = [
object_to_json(t, trim_nulls=False, entry_class='h-cite')
for t in tags if source.object_type(t) == type]
if trim_nulls:
ret = util.trim_nulls(ret)
return ret
def jsonfeed_to_activities(jsonfeed):
"""Converts a JSON feed to ActivityStreams activities and actor.
Args:
jsonfeed: dict, JSON Feed data
Returns:
(activities, actor) tuple, where activities and actor are both
ActivityStreams object dicts
Raises:
ValueError, if jsonfeed isn't a valid JSON Feed dict
"""
if not hasattr(jsonfeed, 'get'):
raise ValueError('Expected dict (or compatible), got %s' %
jsonfeed.__class__.__name__)
author = jsonfeed.get('author', {})
actor = {
'objectType': 'person',
'url': author.get('url'),
'image': [{
'url': author.get('avatar')
}],
'displayName': author.get('name'),
}
def attachment(jf):
if not hasattr(jf, 'get'):
raise ValueError('Expected attachment to be dict; got %s' % jf)
url = jf.get('url')
type = jf.get('mime_type', '').split('/')[0]
as1 = {
'objectType': type,
'title': jf.get('title'),
}
if type in ('audio', 'video'):
as1['stream'] = {'url': url}
else:
as1['url'] = url
return as1
activities = [{
'object': {
'objectType':
'article' if item.get('title') else 'note',
'title':
item.get('title'),
'summary':
item.get('summary'),
'content':
util.get_first(item, 'content_html')
or util.get_first(item, 'content_text'),
'id':
str(item.get('id') or ''),
'published':
item.get('date_published'),
'updated':
item.get('date_modified'),
'url':
item.get('url'),
'image': [{
'url': item.get('image')
}],
'author': {
'displayName': item.get('author', {}).get('name'),
'image': [{
'url': item.get('author', {}).get('avatar')
}]
},
'attachments':
[attachment(a) for a in item.get('attachments', [])],
}
} for item in jsonfeed.get('items', [])]
return (util.trim_nulls(activities), util.trim_nulls(actor))
def from_activities(activities, actor=None, title=None, feed_url=None,
home_page_url=None, hfeed=None):
"""Converts ActivityStreams activities to an RSS 2.0 feed.
Args:
activities: sequence of ActivityStreams activity dicts
actor: ActivityStreams actor dict, the author of the feed
title: string, the feed title
feed_url: string, the URL for this RSS feed
home_page_url: string, the home page URL
hfeed: dict, parsed mf2 h-feed, if available
Returns:
unicode string with RSS 2.0 XML
"""
try:
iter(activities)
except TypeError:
raise TypeError('activities must be iterable')
if isinstance(activities, (dict, basestring)):
raise TypeError('activities may not be a dict or string')
fg = FeedGenerator()
fg.id(feed_url)
assert feed_url
fg.link(href=feed_url, rel='self')
if home_page_url:
fg.link(href=home_page_url, rel='alternate')
# TODO: parse language from lang attribute:
# https://github.com/microformats/mf2py/issues/150
fg.language('en')
fg.generator('granary', uri='https://granary.io/')
hfeed = hfeed or {}
actor = actor or {}
image = util.get_url(hfeed, 'image') or util.get_url(actor, 'image')
if image:
fg.image(image)
props = hfeed.get('properties') or {}
content = microformats2.get_text(util.get_first(props, 'content', ''))
summary = util.get_first(props, 'summary', '')
desc = content or summary or '-'
fg.description(desc) # required
fg.title(title or util.ellipsize(desc)) # required
latest = None
enclosures = False
for activity in activities:
obj = activity.get('object') or activity
if obj.get('objectType') == 'person':
continue
item = fg.add_entry()
url = obj.get('url')
item.id(obj.get('id') or url)
item.link(href=url)
item.guid(url, permalink=True)
item.title(obj.get('title') or obj.get('displayName') or '-') # required
content = microformats2.render_content(
obj, include_location=True, render_attachments=False) or obj.get('summary')
if content:
item.content(content, type='CDATA')
item.category(
[{'term': t['displayName']} for t in obj.get('tags', [])
if t.get('displayName') and t.get('verb') not in ('like', 'react', 'share')])
author = obj.get('author', {})
item.author({
'name': author.get('displayName') or author.get('username'),
'uri': author.get('url'),
})
published = obj.get('published') or obj.get('updated')
if published:
try:
dt = mf2util.parse_datetime(published)
if not isinstance(dt, datetime):
dt = datetime.combine(dt, time.min)
if not dt.tzinfo:
dt = dt.replace(tzinfo=util.UTC)
item.published(dt)
if not latest or dt > latest:
latest = dt
except ValueError: # bad datetime string
pass
for att in obj.get('attachments', []):
stream = util.get_first(att, 'stream') or att
if not stream:
continue
url = stream.get('url') or ''
mime = mimetypes.guess_type(url)[0] or ''
if (att.get('objectType') in ENCLOSURE_TYPES or
mime and mime.split('/')[0] in ENCLOSURE_TYPES):
enclosures = True
item.enclosure(url=url, type=mime, length='REMOVEME') # TODO: length (bytes)
item.load_extension('podcast')
duration = stream.get('duration')
if duration:
item.podcast.itunes_duration(duration)
if enclosures:
fg.load_extension('podcast')
fg.podcast.itunes_author(actor.get('displayName') or actor.get('username'))
if summary:
fg.podcast.itunes_summary(summary)
fg.podcast.itunes_explicit('no')
fg.podcast.itunes_block(False)
if latest:
fg.lastBuildDate(latest)
return fg.rss_str(pretty=True).decode('utf-8').replace(' length="REMOVEME"', '')
def activities_to_jsonfeed(activities,
actor=None,
title=None,
feed_url=None,
home_page_url=None):
"""Converts ActivityStreams activities to a JSON feed.
Args:
activities: sequence of ActivityStreams activity dicts
actor: ActivityStreams actor dict, the author of the feed
title: string, the feed title
home_page_url: string, the home page URL
feed_url: the URL of the JSON Feed, if any. Included in the feed_url field.
Returns:
dict, JSON Feed data, ready to be JSON-encoded
"""
try:
iter(activities)
except TypeError:
raise TypeError('activities must be iterable')
if isinstance(activities, (dict, basestring)):
raise TypeError('activities may not be a dict or string')
def image_url(obj):
return util.get_first(obj, 'image', {}).get('url')
def actor_name(obj):
return obj.get('displayName') or obj.get('username')
if not actor:
actor = {}
items = []
for activity in activities:
obj = activity.get('object') or activity
if obj.get('objectType') == 'person':
continue
author = obj.get('author', {})
content = obj.get('content')
# The JSON Feed spec (https://jsonfeed.org/version/1#items) says that the
# URL from the "image" property may also appear in "content_html", in which
# case it should be interpreted as the "main, featured image" of the
# post. It does not specify the behavior or semantics in the case that the
# image does *not* appear in "content_html", but currently at least one
# feed reader (Feedbin) will not display the image as part of the post
# content unless it is explicitly included in "content_html".
if content and image_url(obj):
content += HTML_IMAGE_TEMPLATE.format(image_url(obj))
obj_title = obj.get('title') or obj.get('displayName')
item = {
'id': obj.get('id') or obj.get('url'),
'url': obj.get('url'),
'image': image_url(obj),
'title':
obj_title if mf2util.is_name_a_title(obj_title, content) else None,
'summary': obj.get('summary'),
'content_html': content,
'date_published': obj.get('published'),
'date_modified': obj.get('updated'),
'author': {
'name': actor_name(author),
'url': author.get('url'),
'avatar': image_url(author),
},
'attachments': [],
}
for att in obj.get('attachments', []):
url = (util.get_first(att, 'stream')
or util.get_first(att, 'image') or att).get('url')
mime = mimetypes.guess_type(url)[0] if url else None
if (att.get('objectType') in ATTACHMENT_TYPES
or mime and mime.split('/')[0] in ATTACHMENT_TYPES):
item['attachments'].append({
'url': url or '',
'mime_type': mime,
'title': att.get('title'),
})
if not item['content_html']:
item['content_text'] = ''
items.append(item)
return util.trim_nulls(
{
'version': 'https://jsonfeed.org/version/1',
'title': title or actor_name(actor) or 'JSON Feed',
'feed_url': feed_url,
'home_page_url': home_page_url or actor.get('url'),
'author': {
'name': actor_name(actor),
'url': actor.get('url'),
'avatar': image_url(actor),
},
'items': items,
},
ignore='content_text')
def activities_to_jsonfeed(activities, actor=None, title=None, feed_url=None,
home_page_url=None):
"""Converts ActivityStreams activities to a JSON feed.
Args:
activities: sequence of ActivityStreams activity dicts
actor: ActivityStreams actor dict, the author of the feed
title: string, the feed title
home_page_url: string, the home page URL
feed_url: the URL of the JSON Feed, if any. Included in the feed_url field.
Returns:
dict, JSON Feed data, ready to be JSON-encoded
"""
try:
iter(activities)
except TypeError:
raise TypeError('activities must be iterable')
if isinstance(activities, (dict, basestring)):
raise TypeError('activities may not be a dict or string')
def image_url(obj):
return util.get_first(obj, 'image', {}).get('url')
def actor_name(obj):
return obj.get('displayName') or obj.get('username')
if not actor:
actor = {}
items = []
for activity in activities:
obj = activity.get('object') or activity
if obj.get('objectType') == 'person':
continue
author = obj.get('author', {})
content = microformats2.render_content(
obj, include_location=True, render_attachments=True)
obj_title = obj.get('title') or obj.get('displayName')
item = {
'id': obj.get('id') or obj.get('url'),
'url': obj.get('url'),
'image': image_url(obj),
'title': obj_title if mf2util.is_name_a_title(obj_title, content) else None,
'summary': obj.get('summary'),
'content_html': content,
'date_published': obj.get('published'),
'date_modified': obj.get('updated'),
'author': {
'name': actor_name(author),
'url': author.get('url'),
'avatar': image_url(author),
},
'attachments': [],
}
for att in obj.get('attachments', []):
url = (util.get_first(att, 'stream') or util.get_first(att, 'image') or att
).get('url')
mime = mimetypes.guess_type(url)[0] if url else None
if (att.get('objectType') in ATTACHMENT_TYPES or
mime and mime.split('/')[0] in ATTACHMENT_TYPES):
item['attachments'].append({
'url': url or '',
'mime_type': mime,
'title': att.get('title'),
})
if not item['content_html']:
item['content_text'] = ''
items.append(item)
return util.trim_nulls({
'version': 'https://jsonfeed.org/version/1',
'title': title or actor_name(actor) or 'JSON Feed',
'feed_url': feed_url,
'home_page_url': home_page_url or actor.get('url'),
'author': {
'name': actor_name(actor),
'url': actor.get('url'),
'avatar': image_url(actor),
},
'items': items,
}, ignore='content_text')
def object_to_json(obj, trim_nulls=True, entry_class='h-entry',
default_object_type=None, synthesize_content=True):
"""Converts an ActivityStreams object to microformats2 JSON.
Args:
obj: dict, a decoded JSON ActivityStreams object
trim_nulls: boolean, whether to remove elements with null or empty values
entry_class: string or sequence, the mf2 class(es) that entries should be
given (e.g. 'h-cite' when parsing a reference to a foreign entry).
defaults to 'h-entry'
default_object_type: string, the ActivityStreams objectType to use if one
is not present. defaults to None
synthesize_content: whether to generate synthetic content if the object
doesn't have its own, e.g. 'likes this.' or 'shared this.'
Returns:
dict, decoded microformats2 JSON
"""
if not obj or not isinstance(obj, dict):
return {}
obj_type = source.object_type(obj) or default_object_type
# if the activity type is a post, then it's really just a conduit
# for the object. for other verbs, the activity itself is the
# interesting thing
if obj_type == 'post':
primary = obj.get('object', {})
obj_type = source.object_type(primary) or default_object_type
else:
primary = obj
# TODO: extract snippet
name = primary.get('displayName', primary.get('title'))
summary = primary.get('summary')
author = obj.get('author', obj.get('actor', {}))
in_reply_tos = obj.get('inReplyTo') or []
if not in_reply_tos:
context = obj.get('context')
if context and isinstance(context, dict):
in_reply_tos = context.get('inReplyTo') or []
is_rsvp = obj_type in ('rsvp-yes', 'rsvp-no', 'rsvp-maybe')
if (is_rsvp or obj_type == 'react') and obj.get('object'):
objs = obj['object']
in_reply_tos.extend(objs if isinstance(objs, list) else [objs])
# maps objectType to list of objects
attachments = defaultdict(list)
for prop in 'attachments', 'tags':
for elem in get_list(primary, prop):
attachments[elem.get('objectType')].append(elem)
# prefer duration and size from object's stream, then first video, then first
# audio
stream = {}
for candidate in [obj] + attachments['video'] + attachments['audio']:
for stream in get_list(candidate, 'stream'):
if stream:
break
duration = stream.get('duration')
if duration is not None:
if util.is_int(duration):
duration = str(duration)
else:
logging('Ignoring duration %r; expected int, got %s', duration.__class__)
duration = None
sizes = []
size = stream.get('size')
if size:
sizes = [str(size)]
# construct mf2!
ret = {
'type': (AS_TO_MF2_TYPE.get(obj_type) or
[entry_class] if isinstance(entry_class, str)
else list(entry_class)),
'properties': {
'uid': [obj.get('id') or ''],
'numeric-id': [obj.get('numeric_id') or ''],
'name': [name],
'nickname': [obj.get('username') or ''],
'summary': [summary],
'url': (list(object_urls(obj) or object_urls(primary)) +
obj.get('upstreamDuplicates', [])),
# photo is special cased below, to handle alt
'video': dedupe_urls(get_urls(attachments, 'video', 'stream') +
get_urls(primary, 'stream')),
'audio': get_urls(attachments, 'audio', 'stream'),
'duration': [duration],
'size': sizes,
'published': [obj.get('published', primary.get('published', ''))],
'updated': [obj.get('updated', primary.get('updated', ''))],
'in-reply-to': util.trim_nulls([o.get('url') for o in in_reply_tos]),
'author': [object_to_json(
author, trim_nulls=False, default_object_type='person')],
'location': [object_to_json(
primary.get('location', {}), trim_nulls=False,
default_object_type='place')],
'comment': [object_to_json(c, trim_nulls=False, entry_class='h-cite')
for c in obj.get('replies', {}).get('items', [])],
'start': [primary.get('startTime')],
'end': [primary.get('endTime')],
},
'children': (
# silly hack: i haven't found anywhere in AS1 or AS2 to indicate that
# something is being "quoted," like in a quote tweet, so i cheat and use
# extra knowledge here that quoted tweets are converted to note
# attachments, but URLs in the tweet text are converted to article tags.
[object_to_json(a, trim_nulls=False, entry_class=['u-quotation-of', 'h-cite'])
for a in attachments['note'] if 'startIndex' not in a] +
[object_to_json(a, trim_nulls=False, entry_class=['h-cite'])
for a in attachments['article'] if 'startIndex' not in a])
}
# content. emulate e- vs p- microformats2 parsing: e- if there are HTML tags,
# otherwise p-.
# https://indiewebcamp.com/note#Indieweb_whitespace_thinking
text = xml.sax.saxutils.unescape(primary.get('content', ''))
html = render_content(primary, include_location=False,
synthesize_content=synthesize_content)
if '<' in html:
ret['properties']['content'] = [{'value': text, 'html': html}]
else:
ret['properties']['content'] = [text]
# photos, including alt text
photo_urls = set()
ret['properties']['photo'] = []
for image in get_list(attachments, 'image') + [primary]:
for url in get_urls(image, 'image'):
if url and url not in photo_urls:
photo_urls.add(url)
name = get_first(image, 'image', {}).get('displayName')
ret['properties']['photo'].append({'value': url, 'alt': name} if name else url)
# hashtags and person tags
if obj_type == 'tag':
ret['properties']['tag-of'] = util.get_urls(obj, 'target')
tags = obj.get('tags', []) or get_first(obj, 'object', {}).get('tags', [])
if not tags and obj_type == 'tag':
tags = util.get_list(obj, 'object')
ret['properties']['category'] = []
for tag in tags:
if tag.get('objectType') == 'person':
ret['properties']['category'].append(
object_to_json(tag, entry_class='u-category h-card'))
elif tag.get('objectType') == 'hashtag' or obj_type == 'tag':
name = tag.get('displayName')
if name:
ret['properties']['category'].append(name)
# rsvp
if is_rsvp:
ret['properties']['rsvp'] = [obj_type[len('rsvp-'):]]
elif obj_type == 'invite':
invitee = object_to_json(obj.get('object'), trim_nulls=False,
default_object_type='person')
ret['properties']['invitee'] = [invitee]
# like and repost mentions
for type, prop in (
('favorite', 'like'),
('follow', 'follow'),
('like', 'like'),
('share', 'repost'),
):
if obj_type == type:
# The ActivityStreams spec says the object property should always be a
# single object, but it's useful to let it be a list, e.g. when a like has
# multiple targets, e.g. a like of a post with original post URLs in it,
# which brid.gy does.
objs = get_list(obj, 'object')
ret['properties'][prop + '-of'] = [
# flatten contexts that are just a url
o['url'] if 'url' in o and set(o.keys()) <= set(['url', 'objectType'])
else object_to_json(o, trim_nulls=False, entry_class='h-cite')
for o in objs]
else:
# received likes and reposts
ret['properties'][prop] = [
object_to_json(t, trim_nulls=False, entry_class='h-cite')
for t in tags if source.object_type(t) == type]
# latitude & longitude
lat = long = None
position = ISO_6709_RE.match(primary.get('position') or '')
if position:
lat, long = position.groups()
if not lat:
lat = primary.get('latitude')
if not long:
long = primary.get('longitude')
if lat:
ret['properties']['latitude'] = [str(lat)]
if long:
ret['properties']['longitude'] = [str(long)]
if trim_nulls:
ret = util.trim_nulls(ret)
return ret
def _activitypub_targets(self):
"""
Returns: list of (Response, string inbox URL)
"""
# if there's in-reply-to, like-of, or repost-of, they're the targets.
# otherwise, it's all followers' inboxes.
targets = self._targets()
if not targets:
# interpret this as a Create or Update, deliver it to followers
inboxes = []
for follower in Follower.query().filter(
Follower.key > Key('Follower', self.source_domain + ' '),
Follower.key < Key(
'Follower', self.source_domain + chr(ord(' ') + 1))):
if follower.status != 'inactive' and follower.last_follow:
actor = json_loads(follower.last_follow).get('actor')
if actor and isinstance(actor, dict):
inboxes.append(
actor.get('endpoints', {}).get('sharedInbox')
or actor.get('publicInbox') or actor.get('inbox'))
return [(Response.get_or_create(source=self.source_url,
target=inbox,
direction='out',
protocol='activitypub',
source_mf2=json_dumps(
self.source_mf2)), inbox)
for inbox in inboxes if inbox]
resps_and_inbox_urls = []
for target in targets:
# fetch target page as AS2 object
try:
self.target_resp = common.get_as2(target)
except (requests.HTTPError, exc.HTTPBadGateway) as e:
self.target_resp = getattr(e, 'response', None)
if self.target_resp and self.target_resp.status_code // 100 == 2:
content_type = common.content_type(self.target_resp) or ''
if content_type.startswith('text/html'):
# TODO: pass e.response to try_salmon()'s target_resp
continue # give up
raise
target_url = self.target_resp.url or target
resp = Response.get_or_create(source=self.source_url,
target=target_url,
direction='out',
protocol='activitypub',
source_mf2=json_dumps(
self.source_mf2))
# find target's inbox
target_obj = self.target_resp.json()
resp.target_as2 = json_dumps(target_obj)
inbox_url = target_obj.get('inbox')
if not inbox_url:
# TODO: test actor/attributedTo and not, with/without inbox
actor = (util.get_first(target_obj, 'actor')
or util.get_first(target_obj, 'attributedTo'))
if isinstance(actor, dict):
inbox_url = actor.get('inbox')
actor = actor.get('url') or actor.get('id')
if not inbox_url and not actor:
self.error(
'Target object has no actor or attributedTo with URL or id.'
)
elif not isinstance(actor, str):
self.error(
'Target actor or attributedTo has unexpected url or id object: %r'
% actor)
if not inbox_url:
# fetch actor as AS object
actor = common.get_as2(actor).json()
inbox_url = actor.get('inbox')
if not inbox_url:
# TODO: probably need a way to save errors like this so that we can
# return them if ostatus fails too.
# self.error('Target actor has no inbox')
continue
inbox_url = urllib.parse.urljoin(target_url, inbox_url)
resps_and_inbox_urls.append((resp, inbox_url))
return resps_and_inbox_urls
def dispatch_request(self, site):
logger.info(f'Params: {list(request.values.items())}')
# strip fragments from source and target url
self.source_url = urllib.parse.urldefrag(request.form['source'])[0]
self.target_url = urllib.parse.urldefrag(request.form['target'])[0]
# follow target url through any redirects, strip utm_* query params
resp = util.follow_redirects(self.target_url)
redirected_target_urls = [r.url for r in resp.history]
self.target_url = util.clean_url(resp.url)
# parse and validate target URL
domain = util.domain_from_link(self.target_url)
if not domain:
self.error(f'Could not parse target URL {self.target_url}')
# look up source by domain
source_cls = models.sources[site]
domain = domain.lower()
self.source = (source_cls.query()
.filter(source_cls.domains == domain)
.filter(source_cls.features == 'webmention')
.filter(source_cls.status == 'enabled')
.get())
if not self.source:
# check for a rel-canonical link. Blogger uses these when it serves a post
# from multiple domains, e.g country TLDs like epeus.blogspot.co.uk vs
# epeus.blogspot.com.
# https://github.com/snarfed/bridgy/issues/805
mf2 = self.fetch_mf2(self.target_url, require_mf2=False)
if not mf2:
# fetch_mf2() already wrote the error response
return
domains = util.dedupe_urls(
util.domain_from_link(url)
for url in mf2[1]['rels'].get('canonical', []))
if domains:
self.source = (source_cls.query()
.filter(source_cls.domains.IN(domains))
.filter(source_cls.features == 'webmention')
.filter(source_cls.status == 'enabled')
.get())
if not self.source:
self.error(
f'Could not find {source_cls.GR_CLASS.NAME} account for {domain}. Is it registered with Bridgy?')
# check that the target URL path is supported
target_path = urllib.parse.urlparse(self.target_url).path
if target_path in ('', '/'):
msg = 'Home page webmentions are not currently supported.'
logger.info(msg)
return {'error': msg}, 202
for pattern in self.source.PATH_BLOCKLIST:
if pattern.match(target_path):
msg = f'{self.source.GR_CLASS.NAME} webmentions are not supported for URL path: {target_path}'
logger.info(msg)
return {'error': msg}, 202
# create BlogWebmention entity
id = f'{self.source_url} {self.target_url}'
self.entity = BlogWebmention.get_or_insert(
id, source=self.source.key, redirected_target_urls=redirected_target_urls)
if self.entity.status == 'complete':
# TODO: response message saying update isn't supported
return self.entity.published
logger.debug(f'BlogWebmention entity: {self.entity.key.urlsafe().decode()}')
# fetch source page
fetched = self.fetch_mf2(self.source_url)
if not fetched:
return
resp, mf2 = fetched
item = self.find_mention_item(mf2.get('items', []))
if not item:
self.error(f'Could not find target URL {self.target_url} in source page {resp.url}', data=mf2, log_exception=False)
# default author to target domain
author_name = domain
author_url = f'http://{domain}/'
# extract author name and URL from h-card, if any
props = item['properties']
author = get_first(props, 'author')
if author:
if isinstance(author, str):
author_name = author
else:
author_props = author.get('properties', {})
author_name = get_first(author_props, 'name')
author_url = get_first(author_props, 'url')
# if present, u-url overrides source url
u_url = get_first(props, 'url')
if u_url:
self.entity.u_url = u_url
# generate content
content = props['content'][0] # find_mention_item() guaranteed this is here
text = (content.get('html') or content.get('value')).strip()
source_url = self.entity.source_url()
text += f' <br /> <a href="{source_url}">via {util.domain_from_link(source_url)}</a>'
# write comment
try:
self.entity.published = self.source.create_comment(
self.target_url, author_name, author_url, text)
except Exception as e:
code, body = util.interpret_http_exception(e)
msg = f'Error: {code}: {e}; {body}'
if code == '401':
logger.warning(f'Disabling source due to: {e}', exc_info=True)
self.source.status = 'disabled'
self.source.put()
self.error(msg, status=code, report=self.source.is_beta_user())
elif code == '404':
# post is gone
self.error(msg, status=code, report=False)
elif util.is_connection_failure(e) or (code and int(code) // 100 == 5):
self.error(msg, status=502, report=False)
elif code or body:
self.error(msg, status=code, report=True)
else:
raise
# write results to datastore
self.entity.status = 'complete'
self.entity.put()
return self.entity.published
def image_url(obj):
return util.get_first(obj, 'image', {}).get('url')
def _create(self, obj, preview=None, include_link=False, ignore_formatting=False):
"""Creates or previews creating a tweet, reply tweet, retweet, or favorite.
https://dev.twitter.com/docs/api/1.1/post/statuses/update
https://dev.twitter.com/docs/api/1.1/post/statuses/retweet/:id
https://dev.twitter.com/docs/api/1.1/post/favorites/create
Args:
obj: ActivityStreams object
preview: boolean
include_link: boolean
Returns:
a CreationResult
If preview is True, the content will be a unicode string HTML
snippet. If False, it will be a dict with 'id' and 'url' keys
for the newly created Twitter object.
"""
assert preview in (False, True)
type = obj.get('objectType')
verb = obj.get('verb')
base_obj = self.base_object(obj)
base_id = base_obj.get('id')
base_url = base_obj.get('url')
is_reply = type == 'comment' or 'inReplyTo' in obj
image_urls = [image.get('url') for image in util.get_list(obj, 'image')]
video_url = util.get_first(obj, 'stream', {}).get('url')
has_media = (image_urls or video_url) and (type in ('note', 'article') or is_reply)
lat = obj.get('location', {}).get('latitude')
lng = obj.get('location', {}).get('longitude')
# prefer displayName over content for articles
type = obj.get('objectType')
base_url = self.base_object(obj).get('url')
prefer_content = type == 'note' or (base_url and (type == 'comment'
or obj.get('inReplyTo')))
content = self._content_for_create(obj, ignore_formatting=ignore_formatting,
prefer_name=not prefer_content,
strip_first_video_tag=bool(video_url))
if not content:
if type == 'activity':
content = verb
elif has_media:
content = ''
else:
return source.creation_result(
abort=False, # keep looking for things to publish,
error_plain='No content text found.',
error_html='No content text found.')
if is_reply and base_url:
# extract username from in-reply-to URL so we can @-mention it, if it's
# not already @-mentioned, since Twitter requires that to make our new
# tweet a reply.
# https://dev.twitter.com/docs/api/1.1/post/statuses/update#api-param-in_reply_to_status_id
# TODO: this doesn't handle an in-reply-to username that's a prefix of
# another username already mentioned, e.g. in reply to @foo when content
# includes @foobar.
parsed = urlparse.urlparse(base_url)
parts = parsed.path.split('/')
if len(parts) < 2 or not parts[1]:
raise ValueError('Could not determine author of in-reply-to URL %s' % base_url)
mention = '@' + parts[1]
if mention.lower() not in content.lower():
content = mention + ' ' + content
# the embed URL in the preview can't start with mobile. or www., so just
# hard-code it to twitter.com. index #1 is netloc.
parsed = list(parsed)
parsed[1] = self.DOMAIN
base_url = urlparse.urlunparse(parsed)
# need a base_url with the tweet id for the embed HTML below. do this
# *after* checking the real base_url for in-reply-to author username.
if base_id and not base_url:
base_url = 'https://twitter.com/-/statuses/' + base_id
if is_reply and not base_url:
return source.creation_result(
abort=True,
error_plain='Could not find a tweet to reply to.',
error_html='Could not find a tweet to <a href="http://indiewebcamp.com/reply">reply to</a>. '
'Check that your post has an <a href="http://indiewebcamp.com/comment">in-reply-to</a> '
'link a Twitter URL or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Twitter.')
# truncate and ellipsize content if it's over the character
# count. URLs will be t.co-wrapped, so include that when counting.
include_url = obj.get('url') if include_link else None
content = self._truncate(content, include_url, has_media)
# linkify defaults to Twitter's link shortening behavior
preview_content = util.linkify(content, pretty=True, skip_bare_cc_tlds=True)
if type == 'activity' and verb == 'like':
if not base_url:
return source.creation_result(
abort=True,
error_plain='Could not find a tweet to like.',
error_html='Could not find a tweet to <a href="http://indiewebcamp.com/favorite">favorite</a>. '
'Check that your post has a like-of link to a Twitter URL or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Twitter.')
if preview:
return source.creation_result(
description='<span class="verb">favorite</span> <a href="%s">'
'this tweet</a>:\n%s' % (base_url, self.embed_post(base_obj)))
else:
data = urllib.urlencode({'id': base_id})
self.urlopen(API_POST_FAVORITE, data=data)
resp = {'type': 'like'}
elif type == 'activity' and verb == 'share':
if not base_url:
return source.creation_result(
abort=True,
error_plain='Could not find a tweet to retweet.',
error_html='Could not find a tweet to <a href="http://indiewebcamp.com/repost">retweet</a>. '
'Check that your post has a repost-of link to a Twitter URL or to an original post that publishes a '
'<a href="http://indiewebcamp.com/rel-syndication">rel-syndication</a> link to Twitter.')
if preview:
return source.creation_result(
description='<span class="verb">retweet</span> <a href="%s">'
'this tweet</a>:\n%s' % (base_url, self.embed_post(base_obj)))
else:
data = urllib.urlencode({'id': base_id})
resp = self.urlopen(API_POST_RETWEET % base_id, data=data)
resp['type'] = 'repost'
elif type in ('note', 'article') or is_reply: # a tweet
content = unicode(content).encode('utf-8')
data = {'status': content}
if is_reply:
description = \
'<span class="verb">@-reply</span> to <a href="%s">this tweet</a>:\n%s' % (
base_url, self.embed_post(base_obj))
data['in_reply_to_status_id'] = base_id
else:
description = '<span class="verb">tweet</span>:'
if video_url:
preview_content += ('<br /><br /><video controls src="%s"><a href="%s">'
'this video</a></video>' % (video_url, video_url))
if not preview:
ret = self.upload_video(video_url)
if isinstance(ret, source.CreationResult):
return ret
data['media_ids'] = ret
elif image_urls:
num_urls = len(image_urls)
if num_urls > MAX_MEDIA:
image_urls = image_urls[:MAX_MEDIA]
logging.warning('Found %d photos! Only using the first %d: %r',
num_urls, MAX_MEDIA, image_urls)
preview_content += '<br /><br />' + ' '.join(
'<img src="%s" />' % url for url in image_urls)
if not preview:
data['media_ids'] = ','.join(self.upload_images(image_urls))
if lat and lng:
preview_content += (
'<div>at <a href="https://maps.google.com/maps?q=%s,%s">'
'%s, %s</a></div>' % (lat, lng, lat, lng))
data['lat'] = lat
data['long'] = lng
if preview:
return source.creation_result(content=preview_content, description=description)
else:
resp = self.urlopen(API_POST_TWEET, data=urllib.urlencode(data))
resp['type'] = 'comment' if is_reply else 'post'
elif (verb and verb.startswith('rsvp-')) or verb == 'invite':
return source.creation_result(
abort=True,
error_plain='Cannot publish RSVPs to Twitter.',
error_html='This looks like an <a href="http://indiewebcamp.com/rsvp">RSVP</a>. '
'Publishing events or RSVPs to Twitter is not supported.')
else:
return source.creation_result(
abort=False,
error_plain='Cannot publish type=%s, verb=%s to Twitter' % (type, verb),
error_html='Cannot publish type=%s, verb=%s to Twitter' % (type, verb))
id_str = resp.get('id_str')
if id_str:
resp.update({'id': id_str, 'url': self.tweet_url(resp)})
elif 'url' not in resp:
resp['url'] = base_url
return source.creation_result(resp)