def test_citations_check_collisions(caplog):
input_ids = [
"citekey-1",
"citekey-1",
"citekey-2",
"Citekey-2",
]
citations = Citations(input_ids)
citations.check_collisions()
assert not caplog.records
def test_citations_filter_pandoc_xnos():
input_ids = [
"fig:pandoc-fignos-key", # should filter
"eq:pandoc-eqnos-key", # should filter
"tbl:pandoc-tablenos-key", # should filter
"not-pandoc-xnos-key", # should keep
]
citations = Citations(input_ids)
citations.filter_pandoc_xnos()
assert len(citations.citekeys) == 1
assert citations.citekeys[0].input_id == "not-pandoc-xnos-key"
def test_citations_filter_unhandled():
input_ids = [
"citekey-with-no-prefix",
"bad-prefix:citekey",
":empty-prefix",
"doi:handled-prefix",
]
citations = Citations(input_ids)
citations.filter_unhandled()
assert len(citations.citekeys) == 1
assert citations.citekeys[0].input_id == "doi:handled-prefix"
def test_citations_inspect():
input_ids = [
"citekey-1", # passes inspection
"arXiv:1806.05726v1", # passes inspection
"arXiv:bad-id",
"DOI:bad-id",
"pmid:bad-id",
"DOID:not-disease-ontology-id",
]
citations = Citations(input_ids)
report = citations.inspect(log_level="INFO")
print(report)
assert len(report.splitlines()) == 4
assert "pmid:bad-id -- PubMed Identifiers should be 1-8 digits" in report
def test_citations_check_multiple_input_ids(caplog):
input_ids = [
"doi:10/b6vnmd",
"DOI:10/B6VNMD",
"doi:10.1016/s0933-3657(96)00367-3",
"ugly-doi-alias",
"other-citekey",
]
citekey_aliases = {"ugly-doi-alias": "DOI:10.1016/s0933-3657(96)00367-3"}
citations = Citations(input_ids, citekey_aliases)
citations.check_multiple_input_ids()
expected = "Multiple citekey input_ids refer to the same standard_id doi:10.1016/s0933-3657(96)00367-3:"
"['doi:10/b6vnmd', 'DOI:10/B6VNMD', 'doi:10.1016/s0933-3657(96)00367-3', 'ugly-doi-alias']"
assert expected in caplog.text
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 citation rendering by output format. See
https://github.com/jgm/pandoc/issues/4834
"""
_parse_cli_cite_args(args)
citations = Citations(
input_ids=args.citekeys, prune_csl_items=args.prune_csl, sort_csl_items=False
)
citations.load_manual_references(paths=args.bibliography)
citations.inspect(log_level="WARNING")
csl_items = citations.get_csl_items()
# output CSL JSON data, if --render is False
if not args.render:
write_file = args.output.open("wb") if args.output else sys.stdout.buffer
with write_file:
write_file.write(citations.csl_json.encode())
return
# use Pandoc to render references
pandoc_metadata = {"nocite": "@*", "csl": args.csl, "references": csl_items}
call_pandoc(metadata=pandoc_metadata, path=args.output, format=args.format)
def test_citations_citekeys_tsv():
input_ids = [
"citekey-1",
"arXiv:1806.05726v1",
"DOI:10.7717/peerj.338",
"pmid:29618526",
]
citations = Citations(input_ids)
citekeys_tsv = citations.citekeys_tsv
assert isinstance(citekeys_tsv, str)
assert "arxiv:1806.05726v1" in citekeys_tsv.splitlines()[2].split("\t")
def test_citations_csl_serialization(csl_format):
ccr_dir = pathlib.Path(__file__).parent.joinpath("cite-command-rendered")
citations = Citations(
["arxiv:1806.05726v1", "doi:10.7717/peerj.338", "pubmed:29618526"])
citations.load_manual_references(
paths=[ccr_dir.joinpath("input-bibliography.json")])
citations.get_csl_items()
path_out = ccr_dir.joinpath(f"output-bibliography.{csl_format}")
# uncomment the following line to regenerate test output
# citations.write_csl_items(path_out)
csl_out = getattr(citations, f"csl_{csl_format}")
assert csl_out == path_out.read_text()
def process_citations(doc: pf.Doc) -> None:
"""
Apply citation-by-identifier to a Python object representation of
Pandoc's Abstract Syntax Tree.
"""
# process metadata.manubot-bibliography-cache
bib_cache = doc.get_metadata(key="manubot-bibliography-cache")
if not (bib_cache is None or isinstance(bib_cache, str)):
logging.warning(
f"Expected metadata.manubot-bibliography-cache to be a string or null (None), "
f"but received a {bib_cache.__class__.__name__}. Setting to None.")
bib_cache = None
doc.manubot["bibliography_cache"] = bib_cache
# process metadata.citekey-aliases
citekey_aliases = doc.get_metadata("citekey-aliases", default={})
if not isinstance(citekey_aliases, dict):
logging.warning(
f"Expected metadata.citekey-aliases to be a dict, "
f"but received a {citekey_aliases.__class__.__name__}. Disregarding."
)
citekey_aliases = dict()
doc.manubot["citekey_aliases"] = citekey_aliases
doc.walk(_get_reference_link_citekey_aliases)
doc.walk(_get_citekeys_action)
manuscript_citekeys = doc.manubot["manuscript_citekeys"]
citations = Citations(input_ids=manuscript_citekeys,
aliases=citekey_aliases)
citations.csl_item_failure_log_level = "ERROR"
requests_cache_path = doc.get_metadata("manubot-requests-cache-path")
if requests_cache_path:
from manubot.process.requests_cache import RequestsCache
req_cache = RequestsCache(requests_cache_path)
req_cache.mkdir()
req_cache.install()
if doc.get_metadata("manubot-clear-requests-cache", default=False):
req_cache.clear()
citations.filter_pandoc_xnos()
citations.load_manual_references(**_get_load_manual_references_kwargs(doc))
citations.inspect(log_level="WARNING")
citations.get_csl_items()
doc.manubot["citekey_shortener"] = citations.input_to_csl_id
doc.walk(_citation_to_id_action)
if requests_cache_path:
req_cache.close()
citations.write_citekeys_tsv(
path=doc.get_metadata("manubot-output-citekeys"))
citations.write_csl_items(
path=doc.get_metadata("manubot-output-bibliography"))
citations.write_csl_items(path=doc.manubot["bibliography_cache"])
# Update pandoc metadata with fields that this filter
# has either consumed, created, or modified.
doc.metadata["bibliography"] = []
doc.metadata["references"] = citations.csl_items
doc.metadata["citekey_aliases"] = citekey_aliases
def process_citations(doc):
"""
Apply citation-by-identifier to a Python object representation of
Pandoc's Abstract Syntax Tree.
The following Pandoc metadata fields are considered:
- bibliography (use to define reference metadata manually)
- citekey-aliases (use to define tags for cite-by-id citations)
- manubot-requests-cache-path
- manubot-clear-requests-cache
- manubot-output-citekeys: path to write TSV table of citekeys
- manubot-output-bibliography: path to write generated CSL JSON bibliography
"""
citekey_aliases = doc.get_metadata("citekey-aliases", default={})
if not isinstance(citekey_aliases, dict):
logging.warning(
f"Expected metadata.citekey-aliases to be a dict, "
f"but received a {citekey_aliases.__class__.__name__}. Disregarding."
)
citekey_aliases = dict()
global_variables["citekey_aliases"] = citekey_aliases
doc.walk(_get_reference_link_citekey_aliases)
doc.walk(_get_citekeys_action)
manuscript_citekeys = global_variables["manuscript_citekeys"]
citations = Citations(input_ids=manuscript_citekeys,
aliases=citekey_aliases)
citations.csl_item_failure_log_level = "ERROR"
requests_cache_path = doc.get_metadata("manubot-requests-cache-path")
if requests_cache_path:
from manubot.process.requests_cache import RequestsCache
req_cache = RequestsCache(requests_cache_path)
req_cache.mkdir()
req_cache.install()
if doc.get_metadata("manubot-clear-requests-cache", default=False):
req_cache.clear()
citations.filter_pandoc_xnos()
citations.load_manual_references(**_get_load_manual_references_kwargs(doc))
citations.inspect(log_level="WARNING")
citations.get_csl_items()
global_variables["citekey_shortener"] = citations.input_to_csl_id
doc.walk(_citation_to_id_action)
if requests_cache_path:
req_cache.close()
citations.write_citekeys_tsv(
path=doc.get_metadata("manubot-output-citekeys"))
citations.write_csl_json(
path=doc.get_metadata("manubot-output-bibliography"))
# Update pandoc metadata with fields that this filter
# has either consumed, created, or modified.
doc.metadata["bibliography"] = []
doc.metadata["references"] = citations.csl_items
doc.metadata["citekey_aliases"] = citekey_aliases