def citation_to_citeproc(citation, prune=True):
"""
Return a dictionary with citation metadata
"""
citation == standardize_citation(citation, warn_if_changed=True)
source, identifier = citation.split(':', 1)
if source in citeproc_retrievers:
citeproc_retriever = import_function(citeproc_retrievers[source])
csl_item = citeproc_retriever(identifier)
else:
msg = f'Unsupported citation source {source} in {citation}'
raise ValueError(msg)
from manubot import __version__ as manubot_version
from manubot.cite.citeproc import (
citeproc_passthrough,
append_to_csl_item_note,
)
note_text = f'This CSL JSON Item was automatically generated by Manubot v{manubot_version} using citation-by-identifier.'
note_dict = {
'standard_id': citation,
}
append_to_csl_item_note(csl_item, note_text, note_dict)
short_id = get_citation_short_id(citation)
csl_item = citeproc_passthrough(csl_item, set_id=short_id, prune=prune)
return csl_item
def citekey_to_csl_item(citekey, prune=True):
"""
Generate a CSL Item (Python dictionary) for the input citekey.
"""
citekey == standardize_citekey(citekey, warn_if_changed=True)
source, identifier = citekey.split(':', 1)
if source in citeproc_retrievers:
citeproc_retriever = import_function(citeproc_retrievers[source])
csl_item = citeproc_retriever(identifier)
else:
msg = f'Unsupported citation source {source!r} in {citekey!r}'
raise ValueError(msg)
from manubot import __version__ as manubot_version
from manubot.cite.citeproc import (
csl_item_passthrough,
append_to_csl_item_note,
)
note_text = f'This CSL JSON Item was automatically generated by Manubot v{manubot_version} using citation-by-identifier.'
note_dict = {
'standard_id': citekey,
}
append_to_csl_item_note(csl_item, note_text, note_dict)
short_citekey = shorten_citekey(citekey)
csl_item = csl_item_passthrough(csl_item,
set_id=short_citekey,
prune=prune)
return csl_item
def test_append_to_csl_item_note(input_note, text, dictionary, expected_note):
csl_item = {
'id': 'test_csl_item',
'type': 'entry',
'note': input_note,
}
append_to_csl_item_note(csl_item, text, dictionary)
output_note = csl_item['note']
assert output_note == expected_note
def csl_item_set_standard_id(csl_item):
"""
Extract the standard_id (standard citation key) for a csl_item and modify the csl_item in-place to set its "id" field.
The standard_id is extracted from a "standard_citation" field, the "note" field, or the "id" field.
If extracting the citation from the "id" field, uses the infer_citekey_prefix function to set the prefix.
For example, if the extracted standard_id does not begin with a supported prefix (e.g. "doi:", "pmid:"
or "raw:"), the citation is assumed to be raw and given a "raw:" prefix. The extracted citation
(referred to as "original_standard_id") is checked for validity and standardized, after which it is
the final "standard_id".
Regarding csl_item modification, the csl_item "id" field is set to the standard_citation and the note field
is created or updated with key-value pairs for standard_id, original_standard_id, and original_id.
Note that the Manubot software generally refers to the "id" of a CSL Item as a citekey.
However, in this context, we use "id" rather than "citekey" for consistency with CSL's "id" field.
"""
if not isinstance(csl_item, dict):
raise ValueError(
"csl_item must be a CSL Data Item represented as a Python dictionary")
from manubot.cite.citeproc import (
append_to_csl_item_note,
parse_csl_item_note,
)
note_dict = parse_csl_item_note(csl_item.get('note', ''))
original_id = None
original_standard_id = None
if 'id' in csl_item:
original_id = csl_item['id']
original_standard_id = infer_citekey_prefix(original_id)
if 'standard_id' in note_dict:
original_standard_id = note_dict['standard_id']
if 'standard_citation' in csl_item:
original_standard_id = csl_item.pop('standard_citation')
if original_standard_id is None:
raise ValueError(
'csl_item_set_standard_id could not detect a field with a citation / standard_citation. '
'Consider setting the CSL Item "id" field.')
assert is_valid_citekey(original_standard_id, allow_raw=True)
standard_id = standardize_citekey(
original_standard_id, warn_if_changed=False)
add_to_note = {}
if original_id and original_id != standard_id:
if original_id != note_dict.get('original_id'):
add_to_note['original_id'] = original_id
if original_standard_id and original_standard_id != standard_id:
if original_standard_id != note_dict.get('original_standard_id'):
add_to_note['original_standard_id'] = original_standard_id
if standard_id != note_dict.get('standard_id'):
add_to_note['standard_id'] = standard_id
append_to_csl_item_note(csl_item, dictionary=add_to_note)
csl_item['id'] = standard_id
return csl_item
def load_manual_references(paths=[], extra_csl_items=[]):
"""
Read manual references (overrides) from files specified by a list of paths.
Returns a standard_citation to CSL Item dictionary. extra_csl_items specifies
JSON CSL stored as a Python object, to be used in addition to the CSL JSON
stored as text in the file specified by path. Set paths=[] to only use extra_csl_items.
"""
csl_items = []
for path in paths:
path = pathlib.Path(path)
if not path.is_file():
logging.warning(
f'process.load_bibliographies is skipping a non-existent path: {path}'
)
continue
for csl_item in load_bibliography(path):
append_to_csl_item_note(
csl_item,
text=
f'This CSL JSON Item was loaded by Manubot v{manubot_version} from a manual reference file.',
dictionary={'manual_reference_filename': path.name},
)
csl_items.append(csl_item)
csl_items.extend(extra_csl_items)
manual_refs = dict()
for csl_item in csl_items:
try:
csl_item_set_standard_id(csl_item)
except Exception:
csl_item_str = json.dumps(csl_item, indent=2)
logging.info(
f'Skipping csl_item where setting standard_id failed:\n{csl_item_str}',
exc_info=True)
continue
standard_id = csl_item['id']
csl_item = citeproc_passthrough(
csl_item, set_id=get_citation_short_id(standard_id))
manual_refs[standard_id] = csl_item
return manual_refs