def make_biorxiv_pages(pub):
if pub.doi.startswith('10.1101/') and pub.genre == 'posted-content':
url = 'https://doi.org/{}'.format(pub.doi)
pmh_page = page.PageTitleMatch()
pmh_page.url = url
pmh_page.doi = pub.doi
pmh_page.title = pub.title
pmh_page.normalized_title = pub.normalized_title
pmh_page.authors = _pmh_authors(pub)
pmh_page.scrape_version = 'submittedVersion'
pmh_page.scrape_metadata_url = url
xref_institution = pub.crossref_api_raw_new.get('institution',
{}).get('name', None)
if xref_institution == 'medRxiv':
pmh_page.pmh_id = _medrxiv_pmh_id(pub.doi)
pmh_page.endpoint_id = medrxiv_endpoint_id
else:
pmh_page.pmh_id = _biorxiv_pmh_id(pub.doi)
pmh_page.endpoint_id = biorxiv_endpoint_id
if _existing_page(page.PageTitleMatch, pmh_page.url, pmh_page.pmh_id):
return []
else:
return [pmh_page]
else:
return []
def make_research_square_pages(pub):
if pub.doi.startswith('10.21203/rs.') and pub.genre == 'posted-content':
url = u'https://doi.org/{}'.format(pub.doi)
pmh_page = page.PageTitleMatch()
pmh_page.pmh_id = _research_square_pmh_id(pub.doi)
pmh_page.url = url
pmh_page.doi = pub.doi
pmh_page.title = pub.title
pmh_page.normalized_title = pub.normalized_title
pmh_page.authors = _pmh_authors(pub)
pmh_page.endpoint_id = research_square_endpoint_id
pmh_page.scrape_version = 'submittedVersion'
pmh_page.scrape_metadata_url = url
if _existing_page(page.PageTitleMatch, pmh_page.url, pmh_page.pmh_id):
return []
else:
return [pmh_page]
else:
return []
def make_authorea_pages(pub):
if pub.publisher and 'authorea' in pub.publisher.lower(
) and pub.genre == 'posted-content':
url = 'https://doi.org/{}'.format(pub.doi)
pmh_page = page.PageTitleMatch()
pmh_page.pmh_id = _authorea_pmh_id(pub.doi)
pmh_page.url = url
pmh_page.doi = pub.doi
pmh_page.title = pub.title
pmh_page.normalized_title = pub.normalized_title
pmh_page.authors = _pmh_authors(pub)
pmh_page.endpoint_id = authorea_endpoint_id
pmh_page.scrape_version = 'submittedVersion'
pmh_page.scrape_metadata_url = url
if _existing_page(page.PageTitleMatch, pmh_page.url, pmh_page.pmh_id):
return []
else:
return [pmh_page]
else:
return []
def make_eartharxiv_pages(pub):
if pub.doi.startswith(
'10.31223/'
) and pub.publisher and 'california digital library' in pub.publisher.lower(
) and pub.genre == 'posted-content':
url = 'https://doi.org/{}'.format(pub.doi)
pmh_page = page.PageTitleMatch()
pmh_page.pmh_id = _eartharxiv_pmh_id(pub.doi)
pmh_page.url = url
pmh_page.doi = pub.doi
pmh_page.title = pub.title
pmh_page.normalized_title = pub.normalized_title
pmh_page.authors = _pmh_authors(pub)
pmh_page.endpoint_id = eartharxiv_endpoint_id
pmh_page.scrape_version = 'submittedVersion'
pmh_page.scrape_metadata_url = url
if _existing_page(page.PageTitleMatch, pmh_page.url, pmh_page.pmh_id):
return []
else:
return [pmh_page]
else:
return []