def _convert_tweet_to_post(self, tweet, post_type): post = Post() post.post_osn_id = unicode(tweet.id) post_creation_date = tweet.date created_at = unicode(date_to_str(post_creation_date)) post.created_at = created_at post.date = post_creation_date post.favorite_count = tweet.favorites post.retweet_count = tweet.retweets post.content = unicode(tweet.text) author_name = unicode(tweet.username) post.author = author_name # post.author_guid = compute_author_guid_by_author_name(author_name) post_url = tweet.permalink post.url = unicode(post_url) post_guid = compute_post_guid(post_url, author_name, created_at) post.guid = post_guid post.post_id = post_guid post.domain = self._domain post.post_type = post_type return post
def create_dummy_post(self): post = Post() post.post_id = unicode(self.post_id) post.author = u"author" post.guid = unicode(generate_random_guid()) post.title = u"title" post.url = u"http://google.com" post.date = str_to_date("2016-08-24 10:00:15") post.content = u"text" post.is_detailed = True post.is_LB = False post.is_valid = True post.domain = u"Google" post.author_guid = unicode(self.author_guid) post.post_osn_id = 123455678 post.retweet_count = 11 post.favorite_count = 10 post.created_at = u"2016-08-24 10:00:15" return post
def photo_xml_to_post(self, child): p = Post() p.title = str(child.find('title').text) p.url = str(child.find('urls').find('url').text) try: p.tags = ','.join(tag.text for tag in child.find('tags').findall('tag')) except: pass p.created_at = str(child.find('dates').get('posted')) p.date = datetime.datetime.fromtimestamp(int(p.created_at)) p.author = str(child.find('owner').get('nsid')) p.domain = 'flickr' p.author_guid = compute_author_guid_by_author_name(p.author) p.retweet_count = int(child.find('comments').text) p.post_id = compute_post_guid(p.url, p.author, date_to_str(p.date)) p.post_osn_id = str(child.get('id')) if child.find('labels') is not None: p.post_type = ','.join( tag.text for tag in child.find('labels').findall('label')) return p
def _convert_tweet_dict_to_post(self, tweet_dict): post = Post() post_osn_id = tweet_dict['id_str'] post.post_osn_id = post_osn_id author_osn_id = tweet_dict['author_osn_id'] author = self._author_osn_id_author_dict[author_osn_id] author_screen_name = author.author_screen_name post.author = author_screen_name post.author_guid = compute_author_guid_by_author_name( author_screen_name) created_at = tweet_dict['created_at'] post.created_at = created_at creation_date_str = extract_tweet_publiction_date(created_at) creation_date = str_to_date(creation_date_str) post.date = creation_date post.favorite_count = tweet_dict['favorite_count'] post.retweet_count = tweet_dict['retweet_count'] post.reply_count = tweet_dict['reply_count'] post.content = str(tweet_dict['full_text']) post.domain = self._domain post.language = str(tweet_dict['lang']) post_url = "https://twitter.com/{0}/status/{1}".format( author_screen_name, post_osn_id) post.url = post_url post_guid = compute_post_guid(post_url, author_screen_name, creation_date_str) post.guid = post_guid post.post_id = post_guid return post
def _generate_post(self, instagram_post): post = Post() post.author_guid = str(instagram_post['owner']['id']) post.date = datetime.datetime.fromtimestamp(instagram_post['taken_at_timestamp']) post.post_osn_id = instagram_post['id'] try: post.content = instagram_post['edge_media_to_caption']['edges'][0]['node']['text'] except: pass post.retweet_count = instagram_post['edge_media_to_comment']['count'] post.favorite_count = instagram_post['edge_media_preview_like']['count'] post.url = 'https://www.instagram.com/p/{}/'.format(instagram_post['shortcode']) image_names = [] # for url in instagram_post['urls']: # image_name_contaner = url.split('/')[-1] # end = image_name_contaner.index('?') # image_names.append(image_name_contaner[:end]) image_name = self._get_image_name_from_url(instagram_post['display_url']) post.media_path = str(instagram_post['display_url']) post.post_format = '{}'.format(image_name) post.domain = 'Instagram' post.post_type = 'post' post.post_id = str(post.post_osn_id) return post