def test_date_from_dateparts(self):
self.assertEqual(date_from_dateparts([]), datetime.date(1970, 1, 1))
self.assertEqual(date_from_dateparts([2015]), datetime.date(2015, 1, 1))
self.assertEqual(date_from_dateparts([2015,2]), datetime.date(2015, 2, 1))
self.assertEqual(date_from_dateparts([2015,2,16]), datetime.date(2015, 2, 16))
self.assertEqual(date_from_dateparts([2015,2,16]), datetime.date(2015, 2, 16))
with self.assertRaises(ValueError):
date_from_dateparts([2015,2,35])
def parse_crossref_date(date):
"""
Parse the date representation from CrossRef to a python object
"""
ret = None
if 'date-parts' in date:
try:
for date in date['date-parts']:
ret = date_from_dateparts(date)
if ret is not None:
return ret
except ValueError:
pass
if 'raw' in date:
ret = tolerant_datestamp_to_datetime(date['raw']).date()
return ret
def parse_crossref_date(date):
"""
Parse the date representation from CrossRef to a python object
"""
ret = None
if 'date-parts' in date:
try:
for date in date['date-parts']:
ret = date_from_dateparts(date)
if ret is not None:
return ret
except ValueError:
pass
if 'raw' in date:
ret = tolerant_datestamp_to_datetime(date['raw']).date()
return ret
def parse_crossref_date(date):
"""
Parse the date representation from CrossRef to a python object
"""
if not date:
return None
ret = None
if 'date-parts' in date:
try:
for date in date['date-parts']:
ret = date_from_dateparts(date)
if ret == datetime.date(year=1970, month=1, day=1):
ret = None
if ret is not None:
return ret
except ValueError:
pass
if 'raw' in date:
ret = tolerant_datestamp_to_datetime(date['raw']).date()
if valid_publication_date(ret):
return ret
def parse_crossref_date(date):
"""
Parse the date representation from CrossRef to a python object
"""
if not date:
return None
ret = None
if 'date-parts' in date:
try:
for date in date['date-parts']:
ret = date_from_dateparts(date)
if ret == datetime.date(year=1970, month=1, day=1):
ret = None
if ret is not None:
return ret
except ValueError:
pass
if 'raw' in date:
ret = tolerant_datestamp_to_datetime(date['raw']).date()
if valid_publication_date(ret):
return ret
def _create_publication(paper, metadata):
if not metadata:
return
if not metadata.get('container-title'):
return
doi = to_doi(metadata.get('DOI', None))
title = metadata['container-title']
if isinstance(title, list):
title = title[0]
title = title[:512]
issn = metadata.get('ISSN', None)
if issn and isinstance(issn, list):
issn = issn[0] # TODO pass all the ISSN to the RoMEO interface
volume = metadata.get('volume', None)
pages = metadata.get('page', None)
issue = metadata.get('issue', None)
date_dict = metadata.get('issued', dict())
pubdate = None
if 'date-parts' in date_dict:
dateparts = date_dict.get('date-parts')[0]
pubdate = date_from_dateparts(dateparts)
# for instance it outputs dates like 2014-2-3
publisher_name = metadata.get('publisher', None)
if publisher_name:
publisher_name = publisher_name[:512]
pubtype = metadata.get('type', 'unknown')
pubtype = CROSSREF_PUBTYPE_ALIASES.get(pubtype, pubtype)
splash_url = doi_to_url(doi)
# PDF availability
pdf_url = None
licenses = set([(license or {}).get('URL')
for license in metadata.get('license', [])])
doi_prefix = doi.split('/')[0]
if doi_prefix in free_doi_prefixes or any(map(is_oa_license, licenses)):
pdf_url = splash_url
# Lookup journal
search_terms = {'jtitle': title}
if issn:
search_terms['issn'] = issn
journal = fetch_journal(search_terms)
publisher = None
if journal:
publisher = journal.publisher
AliasPublisher.increment(publisher_name, journal.publisher)
else:
publisher = fetch_publisher(publisher_name)
barepub = BareOaiRecord(
paper=paper,
journal_title=title,
issue=issue,
volume=volume,
pubdate=pubdate,
pages=pages,
doi=doi,
pubtype=pubtype,
publisher_name=publisher_name,
journal=journal,
publisher=publisher,
pdf_url=pdf_url,
splash_url=splash_url,
source=OaiSource.objects.get(identifier='crossref'),
identifier=doi_to_crossref_identifier(doi))
rec = paper.add_oairecord(barepub)
paper.update_availability()
return paper, rec
def _create_publication(paper, metadata):
if not metadata:
return
if not metadata.get('container-title'):
return
doi = to_doi(metadata.get('DOI', None))
title = metadata['container-title']
if isinstance(title, list):
title = title[0]
title = title[:512]
issn = metadata.get('ISSN', None)
if issn and isinstance(issn, list):
issn = issn[0] # TODO pass all the ISSN to the RoMEO interface
volume = metadata.get('volume', None)
pages = metadata.get('page', None)
issue = metadata.get('issue', None)
date_dict = metadata.get('issued', dict())
pubdate = None
if 'date-parts' in date_dict:
dateparts = date_dict.get('date-parts')[0]
pubdate = date_from_dateparts(dateparts)
# for instance it outputs dates like 2014-2-3
publisher_name = metadata.get('publisher', None)
if publisher_name:
publisher_name = publisher_name[:512]
pubtype = metadata.get('type', 'unknown')
pubtype = CROSSREF_PUBTYPE_ALIASES.get(pubtype, pubtype)
splash_url = doi_to_url(doi)
# PDF availability
pdf_url = None
licenses = set([(license or {}).get('URL')
for license in metadata.get('license', [])])
doi_prefix = doi.split('/')[0]
if doi_prefix in free_doi_prefixes or any(map(is_oa_license, licenses)):
pdf_url = splash_url
# Lookup journal
journal = Journal.find(issn=issn, title=title)
publisher = None
if journal:
publisher = journal.publisher
AliasPublisher.increment(publisher_name, journal.publisher)
else:
publisher = Publisher.find(publisher_name)
barepub = BareOaiRecord(
paper=paper,
journal_title=title,
issue=issue,
volume=volume,
pubdate=pubdate,
pages=pages,
doi=doi,
pubtype=pubtype,
publisher_name=publisher_name,
journal=journal,
publisher=publisher,
pdf_url=pdf_url,
splash_url=splash_url,
source=OaiSource.objects.get(identifier='crossref'),
identifier=doi_to_crossref_identifier(doi))
rec = paper.add_oairecord(barepub)
paper.update_availability()
return paper, rec
def _create_publication(paper, metadata):
if not metadata:
return
if not 'container-title' in metadata or not metadata['container-title']:
return
doi = to_doi(metadata.get('DOI', None))
title = metadata['container-title']
if type(title) == type([]):
title = title[0]
title = title[:512]
issn = metadata.get('ISSN', None)
if issn and type(issn) == type([]):
issn = issn[0] # TODO pass all the ISSN to the RoMEO interface
volume = metadata.get('volume', None)
pages = metadata.get('page', None)
issue = metadata.get('issue', None)
date_dict = metadata.get('issued', dict())
pubdate = None
if 'date-parts' in date_dict:
dateparts = date_dict.get('date-parts')[0]
pubdate = date_from_dateparts(dateparts)
# for instance it outputs dates like 2014-2-3
publisher_name = metadata.get('publisher', None)
if publisher_name:
publisher_name = publisher_name[:512]
pubtype = metadata.get('type', 'unknown')
pubtype = CROSSREF_PUBTYPE_ALIASES.get(pubtype, pubtype)
# PDF availability
pdf_url = None
licenses = set([(license or {}).get('URL')
for license in metadata.get('license', [])])
if any(map(is_oa_license, licenses)):
pdf_url = 'http://dx.doi.org/' + doi
# Lookup journal
search_terms = {'jtitle': title}
if issn:
search_terms['issn'] = issn
journal = fetch_journal(search_terms)
publisher = None
if journal:
publisher = journal.publisher
AliasPublisher.increment(publisher_name, journal.publisher)
else:
publisher = fetch_publisher(publisher_name)
barepub = BarePublication(title=title,
issue=issue,
volume=volume,
pubdate=pubdate,
paper=paper,
pages=pages,
doi=doi,
pubtype=pubtype,
publisher_name=publisher_name,
journal=journal,
publisher=publisher,
pdf_url=pdf_url)
pub = paper.add_publication(barepub)
cur_pubdate = paper.pubdate
if type(cur_pubdate) != type(pubdate):
cur_pubdate = cur_pubdate.date()
if pubdate and pubdate > cur_pubdate:
paper.pubdate = pubdate
paper.update_availability()
return paper, pub
def _create_publication(paper, metadata):
if not metadata:
return
if not 'container-title' in metadata or not metadata['container-title']:
return
doi = to_doi(metadata.get('DOI',None))
title = metadata['container-title']
if type(title) == type([]):
title = title[0]
title = title[:512]
issn = metadata.get('ISSN',None)
if issn and type(issn) == type([]):
issn = issn[0] # TODO pass all the ISSN to the RoMEO interface
volume = metadata.get('volume',None)
pages = metadata.get('page',None)
issue = metadata.get('issue',None)
date_dict = metadata.get('issued',dict())
pubdate = None
if 'date-parts' in date_dict:
dateparts = date_dict.get('date-parts')[0]
pubdate = date_from_dateparts(dateparts)
# for instance it outputs dates like 2014-2-3
publisher_name = metadata.get('publisher', None)
if publisher_name:
publisher_name = publisher_name[:512]
pubtype = metadata.get('type','unknown')
pubtype = CROSSREF_PUBTYPE_ALIASES.get(pubtype, pubtype)
# PDF availability
pdf_url = None
licenses = set([(license or {}).get('URL') for license in metadata.get('license', [])])
if any(map(is_oa_license, licenses)):
pdf_url = 'http://dx.doi.org/'+doi
# Lookup journal
search_terms = {'jtitle':title}
if issn:
search_terms['issn'] = issn
journal = fetch_journal(search_terms)
publisher = None
if journal:
publisher = journal.publisher
AliasPublisher.increment(publisher_name, journal.publisher)
else:
publisher = fetch_publisher(publisher_name)
barepub = BarePublication(title=title, issue=issue, volume=volume,
pubdate=pubdate, paper=paper, pages=pages,
doi=doi, pubtype=pubtype, publisher_name=publisher_name,
journal=journal, publisher=publisher, pdf_url=pdf_url)
pub = paper.add_publication(barepub)
cur_pubdate = paper.pubdate
if type(cur_pubdate) != type(pubdate):
cur_pubdate = cur_pubdate.date()
if pubdate and pubdate > cur_pubdate:
paper.pubdate = pubdate
paper.update_availability()
return paper, pub
