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