def createLinkDatum(parent_url, url): """Create Link CrawlDbDatum mock data.""" key = ndb.Key(CrawlDbDatum, parent_url) CrawlDbDatum.get_or_insert(url, parent=key, url=url, last_status=pipelines.UNFETCHED)
def createMockFetchedDatum(url, html_text, status): """Create FetchedDatum mock data.""" key = ndb.Key(CrawlDbDatum, url) crawl = CrawlDbDatum.get_or_insert(url, parent=key, url=url, last_status=status) if status != pipelines.UNFETCHED: fetched_datum = FetchedDbDatum(parent=crawl.key, url=url, fetched_url = url, fetched_content = html_text, content_type="text/html") fetched_datum.put()
def createMockFetchedDatum(url, html_text, status): """Create FetchedDatum mock data.""" key = ndb.Key(CrawlDbDatum, url) crawl = CrawlDbDatum.get_or_insert(url, parent=key, url=url, last_status=status) if status != pipelines.UNFETCHED: fetched_datum = FetchedDbDatum(parent=crawl.key, url=url, fetched_url=url, fetched_content=html_text, content_type="text/html") fetched_datum.put()
def createMockCrawlDbDatum(url): """Create CrawlDbDatum mock data.""" CrawlDbDatum.get_or_insert(url, url=url, last_status=pipelines.UNFETCHED)