def _convert_row_to_post(self, row): post = Post() claim_id = unicode(row['claim_id']) title = unicode(row['title'], errors='replace') post.content = title description = unicode(row['description'], errors='replace') post.description = description url = unicode(row['url']) post.url = url publication_date = row['publication_date'] post.date = date(publication_date) post_guid = compute_post_guid(self._social_network_url, claim_id, publication_date) post.guid = post_guid post.post_id = post_guid post.domain = self._domain post.author = self._author_name author_guid = compute_author_guid_by_author_name(self._author_name) post.author_guid = author_guid post.post_osn_guid = post_guid keywords = unicode(row['keywords']) post.tags = keywords post_type = unicode(row['post_type']) post.post_type = post_type return post
def _parse_post(self, json_content): post = Post() content = json_content['postText'][0] post.content = content str_post_timestamp = json_content["postTimestamp"] post.created_at = str(str_post_timestamp) # post_timestamp = dateutil.parser.parse(str_post_timestamp) # str_post_date = date_to_str(post_timestamp) post_timestamp, str_post_date = self._get_str_and_date_formats( str_post_timestamp) post.date = post_timestamp post_media = json_content["postMedia"] if len(post_media) > 0: post.media_path = post_media[0] else: post.media_path = None post_id = json_content["id"] post.post_id = post_id post.author = post_id post_guid = compute_post_guid(self._social_network_url, post_id, str_post_date) post.guid = post_guid post.domain = self._domain post.author_guid = post_guid post.post_osn_guid = post_guid return post
def _create_post_by_row(self, record_dict): post = Post() post_id = self._convert_to_unicode_value(record_dict["post_id"]) post.post_osn_id = post_id post.post_id = post_id author_name = self._convert_to_unicode_value(record_dict["tumblog_id"]) post.author = author_name post_short_url = self._convert_to_unicode_value( record_dict["post_short_url"]) self._set_post_url(post_short_url, author_name, post) post_creation_date = self._convert_to_unicode_value( record_dict["created_time_epoch"]) post.created_at = post_creation_date if post_creation_date is not None: post_formatted_creation_date, str_post_formatted_creation_date = convert_epoch_timestamp_to_datetime( post_creation_date) post.date = post_formatted_creation_date else: str_post_formatted_creation_date = self._set_start_date() post.guid = compute_post_guid(post.url, author_name, str_post_formatted_creation_date) post.post_osn_guid = post.guid post.title = self._convert_to_unicode_value(record_dict["post_title"]) post_content = record_dict["post_content"] if post_content != 'NULL': content = json.loads(post_content.decode("utf-8")) #content = eval(record_dict["post_content"]) final_content = "" if 'title' in content.keys(): title = content['title'] final_content += title if 'text' in content.keys(): text = content['text'] final_content += text post.content = self._convert_to_unicode_value(final_content) post.domain = self._domain post.author_guid = compute_author_guid_by_author_name(author_name) post.post_type = self._convert_to_unicode_value( record_dict["post_type"]) post.post_format = self._convert_to_unicode_value( record_dict["post_format"]) post.reblog_key = self._convert_to_unicode_value( record_dict["post_reblog_key"]) post.tags = self._convert_to_unicode_value(record_dict["post_tags"]) post.state = self._convert_to_unicode_value(record_dict["post_state"]) if post.post_osn_id not in self._post_dict: self._post_dict[post.post_osn_id] = post return post
def convert_claim_to_post(claim): from DB.schema_definition import Post post = Post() post.post_id = claim.claim_id post.content = claim.title post.description = claim.description post.url = claim.url post.date = claim.verdict_date post.domain = 'Claim' post.author = 'no author' post.author_guid = 'no author' post.guid = compute_post_guid(claim.url, post.author, date_to_str(post.date)) post.post_osn_guid = post.guid post.tags = claim.keywords post.post_type = claim.verdict return post
def _create_post(self, original_liar_dataset_id, speaker, targeted_label, statement): post = Post() post.post_id = str(original_liar_dataset_id) post_guid = compute_post_guid(self._social_network_url, original_liar_dataset_id, '2007-01-01 00:00:00') post.guid = post_guid post.domain = self._domain post.author = speaker author_guid = compute_author_guid_by_author_name(speaker) post.author_guid = author_guid post.post_osn_guid = post_guid post.date = date('2007-01-01 00:00:00') post.post_type = targeted_label post.content = statement return post