def test_citation_to_citeproc_pubmed_book():
"""
Extracting CSL metadata from books in PubMed is not supported.
Logic not implemented to parse XML returned by
https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=29227604&rettype=full
"""
with pytest.raises(NotImplementedError):
citation_to_citeproc('pmid:29227604')
def test_citation_to_citeproc_doi_datacite():
citation = 'doi:10.7287/peerj.preprints.3100v1'
citeproc = citation_to_citeproc(citation)
assert citeproc['id'] == '11cb5HXoY'
assert citeproc['URL'] == 'https://doi.org/10.7287/peerj.preprints.3100v1'
assert citeproc['DOI'] == '10.7287/peerj.preprints.3100v1'
assert citeproc['type'] == 'report'
assert citeproc[
'title'] == 'Sci-Hub provides access to nearly all scholarly literature'
authors = citeproc['author']
assert authors[0]['family'] == 'Himmelstein'
assert authors[-1]['family'] == 'Greene'
def get_authors(cite_as):
"""Given a doi, get a list of Authors
"""
from manubot import cite
try:
citation = cite.citation_to_citeproc(parse_cite_as_url(cite_as))
authors = [parse_author(d) for d in citation['author']]
return authors
except Exception as e:
logging.warning("Unable to get the authors for: {}\n{}".format(cite_as, e))
return []
def test_citation_to_citeproc_arxiv():
citation = 'arxiv:cond-mat/0703470v2'
citeproc = citation_to_citeproc(citation)
assert citeproc['id'] == 'ES92tcdg'
assert citeproc['URL'] == 'https://arxiv.org/abs/cond-mat/0703470v2'
assert citeproc['number'] == 'cond-mat/0703470v2'
assert citeproc['version'] == '2'
assert citeproc['type'] == 'report'
assert citeproc['container-title'] == 'arXiv'
assert citeproc['title'] == 'Portraits of Complex Networks'
authors = citeproc['author']
assert authors[0]['literal'] == 'J. P. Bagrow'
assert citeproc['DOI'] == '10.1209/0295-5075/81/68004'
def test_citation_to_citeproc_pmc(identifier, citation_id):
citation = f'pmcid:{identifier}'
citeproc = citation_to_citeproc(citation)
assert citeproc['id'] == citation_id
assert citeproc[
'URL'] == 'https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3041534/'
assert citeproc[
'container-title'] == 'Summit on Translational Bioinformatics'
assert citeproc[
'title'] == 'Secondary Use of EHR: Data Quality Issues and Informatics Opportunities'
authors = citeproc['author']
assert authors[0]['family'] == 'Botsis'
assert citeproc['PMID'] == '21347133'
assert citeproc['PMCID'] == 'PMC3041534'
def generate_csl_items(args, citation_df):
"""
General CSL (citeproc) items for standard_citations in citation_df.
Writes references.json to disk and logs warnings for potential problems.
"""
# Read manual references (overrides) in JSON CSL
manual_refs = read_manual_references(args.manual_references_path)
requests_cache.install_cache(args.requests_cache_path,
include_get_headers=True)
cache = requests_cache.get_cache()
if args.clear_requests_cache:
logging.info('Clearing requests-cache')
requests_cache.clear()
logging.info(
f'requests-cache starting with {len(cache.responses)} cached responses'
)
csl_items = list()
failures = list()
for citation in citation_df.standard_citation.unique():
if citation in manual_refs:
csl_items.append(manual_refs[citation])
continue
try:
citeproc = citation_to_citeproc(citation)
csl_items.append(citeproc)
except Exception as error:
logging.exception(f'Citeproc retrieval failure for {citation}')
failures.append(citation)
logging.info(
f'requests-cache finished with {len(cache.responses)} cached responses'
)
requests_cache.uninstall_cache()
if failures:
message = 'Citeproc retrieval failed for:\n{}'.format(
'\n'.join(failures))
logging.error(message)
# Write JSON CSL bibliography for Pandoc.
with args.references_path.open('w') as write_file:
json.dump(csl_items, write_file, indent=2, ensure_ascii=False)
write_file.write('\n')
return csl_items
def cli_cite(args):
"""
Main function for the manubot cite command-line interface.
Does not allow user to directly specify Pandoc's --to argument, due to
inconsistent citaiton rendering by output format. See
https://github.com/jgm/pandoc/issues/4834
"""
# generate CSL JSON data
csl_list = list()
for citation in args.citations:
try:
if not is_valid_citation_string(f'@{citation}'):
continue
citation = standardize_citation(citation)
csl_item = citation_to_citeproc(citation, prune=args.prune_csl)
csl_list.append(csl_item)
except Exception as error:
logging.error(f'citation_to_citeproc for {citation} failed '
f'due to a {error.__class__.__name__}:\n{error}')
logging.info(error, exc_info=True)
# output CSL JSON data, if --render is False
if not args.render:
write_file = args.output.open('w') if args.output else sys.stdout
with write_file:
json.dump(csl_list, write_file, ensure_ascii=False, indent=2)
write_file.write('\n')
return
# use Pandoc to render citations
if not args.format and args.output:
vars(args)['format'] = extension_to_format.get(args.output.suffix)
if not args.format:
vars(args)['format'] = 'plain'
pandoc_metadata = {
'nocite': '@*',
'csl': args.csl,
'references': csl_list,
}
call_pandoc(
metadata=pandoc_metadata,
path=args.output,
format=args.format,
)
def cli_cite(args):
"""
Main function for the manubot cite command-line interface.
Does not allow user to directly specify Pandoc's --to argument, due to
inconsistent citaiton rendering by output format. See
https://github.com/jgm/pandoc/issues/4834
"""
from manubot.cite import (
citation_to_citeproc,
standardize_citation,
)
# generate CSL JSON data
csl_list = list()
for citation in args.citations:
citation = standardize_citation(citation)
csl_list.append(citation_to_citeproc(citation, prune=args.prune_csl))
# output CSL JSON data, if --render is False
if not args.render:
write_file = args.output.open('w') if args.output else sys.stdout
with write_file:
json.dump(csl_list, write_file, ensure_ascii=False, indent=2)
write_file.write('\n')
return
# use Pandoc to render citations
if not args.format and args.output:
vars(args)['format'] = extension_to_format.get(args.output.suffix)
if not args.format:
vars(args)['format'] = 'plain'
pandoc_metadata = {
'nocite': '@*',
'csl': args.csl,
'references': csl_list,
}
call_pandoc(
metadata=pandoc_metadata,
path=args.output,
format=args.format,
)
def test_citation_to_citeproc_pubmed_1():
"""
Generated from XML returned by
https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=21347133&rettype=full
"""
citation = 'pmid:21347133'
citeproc = citation_to_citeproc(citation)
assert citeproc['id'] == 'y9ONtSZ9'
assert citeproc['type'] == 'article-journal'
assert citeproc['URL'] == 'https://www.ncbi.nlm.nih.gov/pubmed/21347133'
assert citeproc[
'container-title'] == 'AMIA Joint Summits on Translational Science proceedings. AMIA Joint Summits on Translational Science'
assert citeproc[
'title'] == 'Secondary Use of EHR: Data Quality Issues and Informatics Opportunities.'
assert citeproc['issued']['date-parts'] == [[2010, 3, 1]]
authors = citeproc['author']
assert authors[0]['given'] == 'Taxiarchis'
assert authors[0]['family'] == 'Botsis'
assert citeproc['PMID'] == '21347133'
assert citeproc['PMCID'] == 'PMC3041534'
def test_citation_to_citeproc_pubmed_2():
"""
Generated from XML returned by
https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=27094199&rettype=full
"""
citation = 'pmid:27094199'
citeproc = citation_to_citeproc(citation)
print(citeproc)
assert citeproc['id'] == 'alaFV9OY'
assert citeproc['type'] == 'article-journal'
assert citeproc['URL'] == 'https://www.ncbi.nlm.nih.gov/pubmed/27094199'
assert citeproc[
'container-title'] == 'Circulation. Cardiovascular genetics'
assert citeproc['container-title-short'] == 'Circ Cardiovasc Genet'
assert citeproc['page'] == '179-84'
assert citeproc[
'title'] == 'Genetic Association-Guided Analysis of Gene Networks for the Study of Complex Traits.'
assert citeproc['issued']['date-parts'] == [[2016, 4]]
authors = citeproc['author']
assert authors[0]['given'] == 'Casey S'
assert authors[0]['family'] == 'Greene'
assert citeproc['PMID'] == '27094199'
assert citeproc['DOI'] == '10.1161/circgenetics.115.001181'