def get_content_for_date(self,
date: datetime.date) -> Iterable[ContentInfo]:
date_string = date.strftime("%Y-%m-%d")
params = {
"expand": ["content.version.number", "content.metadata.mediatype"],
"includeArchivedSpaces":
True,
"cql":
f"lastModified={date_string} or created={date_string} order by lastModified,created asc"
}
for r in self._client.paginated_get("search", params):
content = r.get("content")
result_global_container = r.get("resultGlobalContainer")
if not content or not result_global_container:
continue
space = SpaceInfo(
result_global_container["displayUrl"].split("/")[-1],
result_global_container["title"])
content_id = content["id"]
latest_version = content["version"]["number"]
title = ConfluenceRepository._extract_title(r)
mime_type = None
if content["type"] == "attachment":
mime_type = ConfluenceRepository._extract_mime_type(r, title)
if not mime_type.startswith(
"text/"
) and mime_type not in self.supported_attachment_types:
if not mime_type.startswith("image/"):
logging.warning(
f"Content type {mime_type} not supported. Skipping attachment {title}"
)
continue
yield ContentInfo(content_id, content["type"], latest_version,
title, space, mime_type)
def makeContentInfo(self,
checkin_json,
content,
url=None,
text=None,
photoId=None,
reply=False,
post=False):
assert (reply ^ post), "Must pass exactly one of reply or post"
assert (text or photoId)
# Avoid posting duplicate content.
request = ContentInfo.all()
request = request.filter('checkin_id = ', checkin_json['id'])
existing_contents = request.fetch(10)
for existing_content in existing_contents:
# Check that they're the same type of content
if existing_content.reply_id and not reply:
continue
if existing_content.post_id and not post:
continue
# Check if the content payload is the same
if existing_content.content == content:
logging.info('Avoided posting duplicate content %s' % content)
return existing_content
content_id = utils.generateId()
checkin_id = checkin_json['id']
content_info = ContentInfo()
content_info.content_id = content_id
content_info.fs_id = checkin_json['user']['id']
content_info.checkin_id = checkin_id
content_info.venue_id = checkin_json['venue']['id']
content_info.content = content
if not url:
url = self.generateContentUrl(content_id)
access_token = self.fetchAccessToken(content_info.fs_id)
client = utils.makeFoursquareClient(access_token)
params = {'contentId': content_id, 'url': url}
if text:
params['text'] = text
if photoId:
params['photoId'] = photoId
logging.info('creating content with params=%s' % params)
if post:
if CONFIG['local_dev']:
content_info.post_id = utils.generateId()
else:
response_json = client.checkins.addpost(checkin_id, params)
content_info.post_id = response_json['post']['id']
elif reply:
if CONFIG['local_dev']:
content_info.reply_id = utils.generateId()
else:
response_json = client.checkins.reply(checkin_id, params)
reply_id = None
if 'replies' in response_json:
reply_id = response_json['replies']['id']
elif 'reply' in response_json:
# Right now we return "replies" but we should probably return "reply"
# adding this so I don't have to do it later in the event we rename
reply_id = response_json['reply']['id']
else:
logging.error(
"Could not find reply id in /checkins/reply response: %s"
% response_json)
content_info.reply_id = reply_id
content_info.put()
return content_info
