def extract_post(self, data, post_type):
post = Post()
if data['publish_date'] is None:
publish_date_date = calendar.timegm(time.gmtime()) * 1000
else:
publish_date_date = data['publish_date']['$date']
date_str = datetime.datetime.fromtimestamp(
publish_date_date / 1000).strftime('%Y-%m-%d %H:%M:%S')
post.post_id = compute_post_guid(data['url'], data['source'], date_str)
post.guid = post.post_id
post.author_guid = compute_author_guid_by_author_name(data['source'])
post.author = str(data['source'])
post.date = convert_str_to_unicode_datetime(date_str)
post.title = str(data['title'])
post.url = str(data['url'])
post.source_url = str(data['source'])
post.content = str(data['text'])
post.tags = ','.join(data['keywords'])
post.domain = self._domain
post.post_type = post_type
if 'description' not in data['meta_data']:
post.description = ""
else:
post.description = str(data['meta_data']['description'])
return post
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 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
