def test_extract_paragraphs(WMLParser):
WMLParser.return_value = iter([
Paragraph(text='some text', volume=1),
Paragraph(text='some more', volume=1)
])
paragraphs = list(extract_paragraphs('TA99_01.xml'))
assert paragraphs == [
Paragraph(text='some text', volume=99),
Paragraph(text='some more', volume=99)
]
def test_assembles_one_citation_with_keyword():
paragraphs = [
Paragraph(type=ParagraphType.KEYWORD, volume=5, text='A'),
Paragraph(type=ParagraphType.CITATION, volume=5, text='1. Some citation')
]
citations = list(assemble_citations(paragraphs))
assert citations == [
IntermediateCitation(
volume=5,
raw_text='1. Some citation',
keywords=['A']
)
]
def test_assembles_one_citation_with_amendment():
paragraphs = [
Paragraph(type=ParagraphType.CITATION, volume=5, text='1. Some citation'),
Paragraph(type=ParagraphType.AMENDMENT, volume=5, text='Some amendment')
]
citations = list(assemble_citations(paragraphs))
assert citations == [
IntermediateCitation(
volume=5,
keywords=[],
raw_text='1. Some citation',
amendments=['Some amendment']
)
]
def test_detect_paragraph_types():
sample_paragraphs_with_expected_types = [
('Something', None),
('ZEITSCHRIFTEN UND', ParagraphType.JOURNAL_SECTION_BEGIN),
('Something', None),
('A. Allgemeines', ParagraphType.KEYWORD),
('1. First citation', ParagraphType.CITATION),
('• Some bullet point', ParagraphType.AMENDMENT),
('3. Second citation', ParagraphType.CITATION),
('Ac. bibliotheken', ParagraphType.KEYWORD),
('4. Third citation', ParagraphType.CITATION),
('Autoren, Herausgeber, Übersetzer, Rezensenten',
ParagraphType.AUTHOR_INDEX_BEGIN),
('Something', None),
]
paragraphs = [
Paragraph(text=p, volume='130')
for p, _ in sample_paragraphs_with_expected_types
]
paragraphs = list(detect_paragraph_types(paragraphs, KEYWORD_MAPPING))
detected_types = [p.type for p in paragraphs]
assert detected_types == [
paragraph_type
for _, paragraph_type in sample_paragraphs_with_expected_types
]
def __iter__(self):
self._parse_xml()
paragraph_nodes = self._xpath('//w:p')
for paragraph_index, paragraph_node in enumerate(paragraph_nodes):
yield Paragraph(
originalIndex=paragraph_index,
text=self._get_paragraph_text(paragraph_node),
)
def test_assembles_three_citations():
paragraphs = [
Paragraph(type=ParagraphType.CITATION, volume=5, text='1. Citation 1'),
Paragraph(type=ParagraphType.CITATION, volume=5, text='2. Citation 2'),
Paragraph(type=ParagraphType.CITATION, volume=5, text='3. Citation 3')
]
citations = list(assemble_citations(paragraphs))
assert citations == [
IntermediateCitation(
volume=5,
raw_text='1. Citation 1'
),
IntermediateCitation(
volume=5,
raw_text='2. Citation 2',
),
IntermediateCitation(
volume=5,
raw_text='3. Citation 3',
)
]
def test_assembles_one_citation():
paragraphs = [
Paragraph(type=ParagraphType.CITATION, volume=5, text='1. Some citation')
]
citations = list(assemble_citations(paragraphs))
assert citations == [
IntermediateCitation(
volume=5,
keywords=[],
raw_text='1. Some citation'
)
]
def test_assembles_two_citations_with_keywords():
paragraphs = [
Paragraph(type=ParagraphType.KEYWORD, volume=5, text='A'),
Paragraph(type=ParagraphType.CITATION, volume=5, text='1. Citation 1'),
Paragraph(type=ParagraphType.KEYWORD, volume=5, text='B'),
Paragraph(type=ParagraphType.AMENDMENT, volume=5, text='Some amendment'),
Paragraph(type=ParagraphType.CITATION, volume=5, text='2. Citation 2')
]
citations = list(assemble_citations(paragraphs))
assert citations == [
IntermediateCitation(
volume=5,
raw_text='1. Citation 1',
keywords=['A'],
amendments=['Some amendment'],
),
IntermediateCitation(
volume=5,
raw_text='2. Citation 2',
keywords=['B'],
amendments=[]
)
]