def test_extracts_authors_from_multiple_citations():
# given
citations = [
Citation(authors=[Person(raw='Ozkut Korkmaz')]),
Citation(authors=[Person(raw='Smith, Jordan')]),
]
# when
known_authors = extract_known_authors(citations)
# then
assert known_authors == {'Ozkut Korkmaz', 'Smith, Jordan'}
def test_parse_citation_fields():
field_parsed_citation = parse_fields_in_citation(CITATION)
assert field_parsed_citation == Citation(
volume=1,
number=1,
type=CitationType.COLLECTION,
title='Lexikon der islamischen Welt',
editors=[
Person(first='Klaus', last='Kreiser', raw='Klaus Kreiser'),
Person(first='Werner', last='Diem', raw='Werner Diem'),
Person(first='Hans',
middle='Georg',
last='Majer',
raw='Hans Georg Majer')
],
keywords=[{
'raw': 'A'
}, {
'raw': 'B'
}],
reviews=[],
comments=[],
date_published={'year': 1974},
location='Stuttgart',
series='Urban-Taschenbücher, 200/1-3',
number_of_volumes='3',
authors=[Person(first='L.', last='Bazin', raw='Bazin, L.')],
raw_text='1. Lexikon der islamischen Welt. '
'Klaus Kreiser, Werner Diem, Hans Georg Majer ed. '
'3 Bde., Stuttgart, 1974 (Urban-Taschenbücher, 200/1-3).',
remaining_text=
'{{{ title }}}. {{{ editors }}} {{{ number_of_volumes }}} '
'{{{ location }}} {{{ date_published }}}{{{ series }}}.')
def parse_fields_in_citation(intermediate: IntermediateCitation) -> Citation:
reviews, comments = parse_amendments_or_comments([intermediate.comment])
more_reviews, amendments = parse_amendments_or_comments(intermediate.amendments)
reviews.extend(more_reviews)
reviews.extend(parse_reviews(intermediate.reviews))
return Citation(
volume=intermediate.volume,
number=int(intermediate.number),
type=intermediate.type,
title=intermediate.title,
location=intermediate.location,
series=intermediate.series,
keywords=[{'raw': keyword} for keyword in intermediate.keywords],
number_of_volumes=intermediate.number_of_volumes,
number_of_pages=intermediate.number_of_pages,
authors=parse_authors(intermediate.authors),
editors=parse_editors_or_translators(intermediate.editors),
translators=parse_editors_or_translators(intermediate.translators),
comments=comments,
reviews=reviews,
published_in=parse_reference(intermediate.published_in),
amendments=amendments,
date_published=parse_date_published(intermediate.date_published),
raw_text=intermediate.raw_text,
material=parse_material(intermediate.material),
date=parse_date(intermediate.date),
ta_references=parse_ta_references(intermediate.ta_references),
remaining_text=intermediate.remaining_text
)
def test_normalize_keywords_with_super_keyword():
citation = Citation(keywords=[{'raw': 'AB. IRGENDWAS'}])
keyword_mapping = {
'A': {
'de': 'Allgemeines',
'en': 'General'
},
'AB': {
'de': 'Spezielles',
'en': 'Specific'
}
}
citation = normalize_keywords_for_citation(citation, keyword_mapping)
assert citation.keywords == [{
'code': 'AB',
'nameDE': 'Spezielles',
'nameEN': 'Specific',
'raw': 'AB. IRGENDWAS',
'super': {
'code': 'A',
'nameDE': 'Allgemeines',
'nameEN': 'General',
'raw': None,
'super': None
}
}]
def test_extracts_nothing_if_no_authors_present():
# given
citations = [Citation(authors=[])]
# when
known_authors = extract_known_authors(citations)
# then
assert len(known_authors) == 0
def test_extracts_single_author_from_one_citation():
# given
citations = [Citation(authors=[Person(raw='Smith, Jordan')])]
# when
known_authors = extract_known_authors(citations)
# then
assert known_authors == {'Smith, Jordan'}
def test_normalize_keywords():
citation = Citation(keywords=[{'raw': 'A. ASDFERAWER'}])
keyword_mapping = {'A': {'de': 'Allgemeines', 'en': 'General'}}
citation = normalize_keywords_for_citation(citation, keyword_mapping)
assert citation.keywords == [{
'code': 'A',
'nameDE': 'Allgemeines',
'nameEN': 'General',
'raw': 'A. ASDFERAWER',
'super': None
}]
def test_assign_duplicate():
# given
citations = [
Citation(volume=1, number=1),
Citation(volume=1, number=2),
Citation(volume=1, number=1),
Citation(volume=1, number=1),
]
# when
citations_with_ids = list(assign_citation_ids(citations))
# then
assert citations_with_ids == [
Citation(volume=1, number=1, id='1-1'),
Citation(volume=1, number=2, id='1-2'),
Citation(volume=1, number=1, id='1-1-1'),
Citation(volume=1, number=1, id='1-1-2'),
]
def test_insert_known_authors():
raw_text = ''.join((
'12. Handžić, Adem ',
'Problematika sakupljanja i izdavanja turskih '
'istorij-skih izvora u radu Orijentalnog Instituta. ',
'In: POF 20-21.1970/71 (1974).213-221. ',
'[Die Problematik der Erfassung und Herausgabe der türkischen historischen Quellen ',
'im Rahmen der Arbeiten des Orientalischen Instituts in Sarajevo, Jugoslavien.]',
))
raw_citation = IntermediateCitation(volume=1, raw_text=raw_text)
parsed_citation = parse_fields_in_citation(parse_citation(raw_citation))
queue = Queue()
insert_known_authors([parsed_citation], ['Handžić, Adem'], queue)
assert queue.get() == Citation(
volume=1,
number=12,
type=CitationType.ARTICLE,
authors=[Person(last='Handžić', first='Adem', raw='Handžić, Adem')],
title='Problematika sakupljanja i izdavanja turskih '
'istorij-skih izvora u radu Orijentalnog Instituta',
comments=[
'Die Problematik der Erfassung und Herausgabe der türkischen historischen Quellen '
'im Rahmen der Arbeiten des Orientalischen Instituts in Sarajevo, Jugoslavien'
],
reviews=[],
raw_text=raw_text,
published_in={
'journal': 'POF',
'volumeStart': 20,
'volumeEnd': 21,
'yearStart': 1970,
'yearEnd': 1971,
'yearParentheses': 1974,
'pageStart': 213,
'pageEnd': 221,
'type': 'journal',
'raw': 'POF 20-21.1970/71 (1974).213-221'
},
ta_references=[],
remaining_text='{{{ authors }}} {{{ title }}} {{{ in }}}'
)
assert queue.empty()
def _citation_as_dict(cls, citation: Citation):
citation_dict = asdict(replace(
citation, type=citation.type.value if citation.type else None),
dict_factory=cls._to_dict)
citation_dict['fullyParsed'] = citation.fully_parsed()
return citation_dict