def import_article(identifier):
"""Import a new article from arXiv or Crossref based on the identifier.
This function attempts to parse and normalize the identifier as a valid
arXiv id or DOI. If the identifier is valid and there is no record in
Inspire matching the ID, it queries the arXiv/CrossRef APIs and parses
the record to make it inspire compliant.
Args:
identifier(str): the ID of the record to import
Returns:
dict: the serialized article
Raises:
ExistingArticleError: if the record is already in Inspire.
ImportArticleError: if no article is found.
ImportConnectionError: if the importing request fails.
ImportParsingError: if an error occurs while parsing the result.
UnknownIdentifierError: if the identifier is neither "arxiv" or "doi".
"""
if is_arxiv(identifier):
pid_type = "arxiv"
pid_value = normalize_arxiv(identifier)
elif is_doi(identifier):
pid_type = "doi"
pid_value = normalize_doi(identifier)
else:
raise UnknownImportIdentifierError(identifier)
pid = PersistentIdentifier.query.filter_by(
pid_type=pid_type, pid_value=pid_value).one_or_none()
if pid:
raise ExistingArticleError(
f"Article {identifier} already in Inspire. UUID: {pid.object_uuid}"
)
importers = {"arxiv": import_arxiv, "doi": import_doi}
importer = importers.get(pid_type, UnknownImportIdentifierError)
article = importer(pid_value)
if not article:
raise ImportArticleError(f"No article found for {identifier}")
if pid_type == "arxiv":
article = merge_article_with_crossref_data(article)
return article
def import_article(identifier):
"""Import a new article from arXiv or Crossref based on the identifier.
This function attempts to parse and normalize the identifier as a valid
arXiv id or DOI. If the identifier is valid and there is no record in
Inspire matching the ID, it queries the arXiv/CrossRef APIs and parses
the record to make it inspire compliant.
Args:
identifier(str): the ID of the record to import
Returns:
dict: the serialized article
Raises:
ExistingArticleError: if the record is already in Inspire.
ImportArticleNotFoundError: if no article is found.
ImportParsingError: if an error occurs while parsing the result.
UnknownIdentifierError: if the identifier is neither "arxiv" or "doi".
"""
if is_arxiv(identifier):
pid_type = "arxiv"
pid_value = normalize_arxiv(identifier)
elif is_doi(identifier):
pid_type = "doi"
pid_value = normalize_doi(identifier).lower()
else:
raise UnknownImportIdentifierError(identifier)
recid = get_pid_for_pid(pid_type, pid_value, provider="recid")
if recid:
raise ExistingArticleError(
f"The article {identifier} already exists in Inspire", recid)
importers = {"arxiv": import_arxiv, "doi": import_doi}
importer = importers.get(pid_type, UnknownImportIdentifierError)
article = importer(pid_value)
if not article:
raise ImportArticleNotFoundError(f"No article found for {identifier}")
if pid_type == "arxiv":
article = merge_article_with_crossref_data(article)
return article
def test_normalize_arxiv_handles_uppercase_with_cls():
expected = 'math/0312059'
result = utils.normalize_arxiv('MatH.AC/0312059')
assert expected == result
def test_normalize_arxiv_handles_uppercase():
expected = 'math/0312059'
result = utils.normalize_arxiv('MATH/0312059')
assert expected == result
def test_normalize_arxiv_handles_new_identifiers_with_cls_prefix_and_ver():
expected = '1501.00001'
result = utils.normalize_arxiv('arXiv:hep-th.GT/1501.00001v1')
assert expected == result
def test_normalize_arxiv_handles_new_identifiers_with_class_and_wo_version():
expected = '1501.00001'
result = utils.normalize_arxiv('arXiv:hep-th/1501.00001')
assert expected == result
def test_normalize_arxiv_handles_solv_int():
expected = 'solv-int/9611008'
result = utils.normalize_arxiv('solv-int/9611008')
assert expected == result
def test_normalize_arxiv_handles_old_identifiers_with_prefix_and_version():
expected = 'math/0309136'
result = utils.normalize_arxiv('arXiv:math.GT/0309136v2')
assert expected == result
def test_normalize_arxiv_handles_new_identifiers_wo_prefix_and_with_version():
expected = '1501.00001'
result = utils.normalize_arxiv('1501.00001v1')
assert expected == result
def test_normalize_arxiv_handles_category_in_brackets():
expected = '1406.1599'
result = utils.normalize_arxiv('arXiv:1406.1599v2[physics.ins-det]')
assert expected == result
