def test_preamble_and_friends_in_table_of_contents(self):
d = Document()
d.content = document_fixture(xml="""
<coverpage>
<content><p>hi</p></content>
</coverpage>
<preface>
<content><p>hi</p></content>
</preface>
<preamble>
<content><p>hi</p></content>
</preamble>
<body>
<content><p>hi</p></content>
</body>
<conclusions>
<content><p>hi></p></content>
</conclusions>
""")
toc = d.table_of_contents()
toc = [t.as_dict() for t in toc]
self.maxDiff = None
self.assertEqual(toc, [
{'type': 'coverpage', 'component': 'main', 'subcomponent': 'coverpage', 'title': 'Coverpage'},
{'type': 'preface', 'component': 'main', 'subcomponent': 'preface', 'title': 'Preface'},
{'type': 'preamble', 'component': 'main', 'subcomponent': 'preamble', 'title': 'Preamble'},
{'type': 'conclusions', 'component': 'main', 'subcomponent': 'conclusions', 'title': 'Conclusions'},
])
def test_find_simple(self):
document = Document(work=self.work,
document_xml=document_fixture(xml="""
<section eId="sec_1">
<num>1.</num>
<heading>Tester</heading>
<paragraph eId="sec_1.paragraph-0">
<content>
<p>Something to do with GN no 102 of 2012.</p>
<p>And another thing about SI 4 of 1998.</p>
</content>
</paragraph>
</section>"""),
language=self.eng)
expected = Document(work=self.work,
document_xml=document_fixture(xml="""
<section eId="sec_1">
<num>1.</num>
<heading>Tester</heading>
<paragraph eId="sec_1.paragraph-0">
<content>
<p>Something to do with <ref href="/akn/za/act/gn/2012/102">GN no 102 of 2012</ref>.</p>
<p>And another thing about <ref href="/akn/za/act/si/1998/4">SI 4 of 1998</ref>.</p>
</content>
</paragraph>
</section>"""),
language=self.eng)
self.finder.find_references_in_document(document)
root = etree.fromstring(expected.content)
expected.content = etree.tostring(root,
encoding='utf-8').decode('utf-8')
self.assertEqual(expected.content, document.content)
def test_italics_markup(self):
document = Document(work=self.work,
document_xml=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Application of Act</heading>
<content>
<p>In the Gazette it says that habeus corpus is XYZ. As per the evidence of person X, Y.</p>
</content>
</section>
"""))
expected = Document(work=self.work,
document_xml=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Application of Act</heading>
<content>
<p>In the <i>Gazette</i> it says that <i>habeus corpus</i> is XYZ. As <i>per</i> the evidence of person X, Y.</p>
</content>
</section>
"""))
self.italics_terms_finder.mark_up_italics_in_document(
document, self.italics_terms)
root = etree.fromstring(expected.content)
expected.content = etree.tostring(root,
encoding='utf-8').decode('utf-8')
self.assertEqual(expected.content, document.content)
def test_find_simple(self):
document = Document(document_xml=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with Act no 22 of 2012.</p>
<p>And another thing about Act 4 of 1998.</p>
</content>
</paragraph>
</section>"""),
language=self.eng)
expected = Document(document_xml=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with Act <ref href="/za/act/2012/22">no 22 of 2012</ref>.</p>
<p>And another thing about Act <ref href="/za/act/1998/4">4 of 1998</ref>.</p>
</content>
</paragraph>
</section>"""),
language=self.eng)
self.finder.find_references_in_document(document)
root = etree.fromstring(expected.content)
expected.content = etree.tostring(root,
encoding='utf-8').decode('utf-8')
self.assertEqual(expected.content, document.content)
def test_set_content(self):
d = Document()
d.content = document_fixture(u'γνωρίζω')
assert_equal(d.frbr_uri, '/za/act/1900/1')
assert_equal(d.country, 'za')
assert_equal(d.doc.publication_date, date(2005, 7, 24))
def test_section_of_this(self):
document = Document(
work=self.work,
document_xml=document_fixture(
xml="""
<section eId="sec_7">
<num>7.</num>
<heading>Active ref heading</heading>
<content>
<p>As given in section 26 of this Act, blah.</p>
<p>As given in section 26 (1) of this Act, blah.</p>
<p>As given in section 26 (1) of this Proclamation, blah.</p>
<p>As given in section 26(1)(b)(iii)(dd)(A) of this Act, blah.</p>
<p>In section 26 of Act 5 of 2012 it says one thing and in section 26 of this Act it says another.</p>
<p>As <i>given</i> in (we're now in a tail) section 26 of this Act, blah.</p>
</content>
</section>
<section eId="sec_26">
<num>26.</num>
<heading>Important heading</heading>
<content>
<p>An important provision.</p>
</content>
</section>
"""
),
language=self.eng)
expected = Document(
work=self.work,
document_xml=document_fixture(
xml="""
<section eId="sec_7">
<num>7.</num>
<heading>Active ref heading</heading>
<content>
<p>As given in <ref href="#sec_26">section 26</ref> of this Act, blah.</p>
<p>As given in <ref href="#sec_26">section 26</ref> (1) of this Act, blah.</p>
<p>As given in <ref href="#sec_26">section 26</ref> (1) of this Proclamation, blah.</p>
<p>As given in <ref href="#sec_26">section 26</ref>(1)(b)(iii)(dd)(A) of this Act, blah.</p>
<p>In section 26 of Act 5 of 2012 it says one thing and in <ref href="#sec_26">section 26</ref> of this Act it says another.</p>
<p>As <i>given</i> in (we're now in a tail) <ref href="#sec_26">section 26</ref> of this Act, blah.</p>
</content>
</section>
<section eId="sec_26">
<num>26.</num>
<heading>Important heading</heading>
<content>
<p>An important provision.</p>
</content>
</section>
"""
),
language=self.eng)
self.section_refs_finder.find_references_in_document(document)
root = etree.fromstring(expected.content)
expected.content = etree.tostring(root, encoding='utf-8').decode('utf-8')
self.assertEqual(expected.content, document.content)
def test_empty_expression_date(self):
d = Document()
d.content = document_fixture('test')
d.expression_date = ''
assert_equal(d.expression_date, '')
d.expression_date = None
assert_equal(d.expression_date, None)
def test_inherit_from_work(self):
w = Work.objects.create(frbr_uri='/za/act/2009/test',
title='Test document')
d = Document(work=w, expression_date='2011-02-01')
d.save()
d = Document.objects.get(pk=d.id)
assert_equal(w.frbr_uri, d.frbr_uri)
assert_equal(w.title, d.title)
def test_inherit_from_work(self):
user = User.objects.get(pk=1)
w = Work.objects.create(frbr_uri='/za/act/2009/test', title='Test document', country=Country.for_code('za'), created_by_user=user)
d = Document(work=w, expression_date='2011-02-01', language=self.eng, created_by_user=user)
d.save()
d = Document.objects.get(pk=d.id)
assert_equal(w.frbr_uri, d.frbr_uri)
assert_equal(w.title, d.title)
def test_section_valid_and_invalid(self):
document = Document(
work=self.work,
document_xml=document_fixture(
xml="""
<section eId="sec_7">
<num>7.</num>
<heading>Active ref heading</heading>
<content>
<p>As given in sections 26 and 35, one of which isn't in this document, blah.</p>
<p>As given in sections 35 and 26, one of which isn't in this document, blah.</p>
<p>In section 200 it says one thing and in section 26 it says another.</p>
<p>As <i>given</i> in (we're now in a tail) section 200, blah, but section 26 says something else.</p>
<p>As <i>given</i> in (we're now in a tail) section 26 of Act 5 of 2012, blah, but section 26 of this Act says something else.</p>
</content>
</section>
<section eId="sec_26">
<num>26.</num>
<heading>The section we want</heading>
<content>
<p>The provision you're looking for.</p>
</content>
</section>
"""
),
language=self.eng)
expected = Document(
work=self.work,
document_xml=document_fixture(
xml="""
<section eId="sec_7">
<num>7.</num>
<heading>Active ref heading</heading>
<content>
<p>As given in sections <ref href="#sec_26">26</ref> and 35, one of which isn't in this document, blah.</p>
<p>As given in sections 35 and <ref href="#sec_26">26</ref>, one of which isn't in this document, blah.</p>
<p>In section 200 it says one thing and in <ref href="#sec_26">section 26</ref> it says another.</p>
<p>As <i>given</i> in (we're now in a tail) section 200, blah, but <ref href="#sec_26">section 26</ref> says something else.</p>
<p>As <i>given</i> in (we're now in a tail) section 26 of Act 5 of 2012, blah, but <ref href="#sec_26">section 26</ref> of this Act says something else.</p>
</content>
</section>
<section eId="sec_26">
<num>26.</num>
<heading>The section we want</heading>
<content>
<p>The provision you're looking for.</p>
</content>
</section>
"""
),
language=self.eng)
self.section_refs_finder.find_references_in_document(document)
root = etree.fromstring(expected.content)
expected_content = etree.tostring(root, encoding='utf-8').decode('utf-8')
self.assertEqual(expected_content, document.content)
def test_find_simple(self):
za = Country.objects.get(pk=1)
user1 = User.objects.get(pk=1)
settings.INDIGO['WORK_PROPERTIES'] = {
'za': {
'cap': 'Chapter (cap)',
}
}
work = Work(
frbr_uri='/akn/za/act/2002/5',
title='Act 5 of 2002',
country=za,
created_by_user=user1,
)
work.properties['cap'] = '12'
work.updated_by_user = user1
work.save()
document = Document(
document_xml=document_fixture(
xml="""
<section eId="sec_1">
<num>1.</num>
<heading>Tester</heading>
<paragraph eId="sec_1.paragraph-0">
<content>
<p>Something to do with Cap. 12.</p>
</content>
</paragraph>
</section>"""
),
language=self.eng,
work=work)
expected = Document(
document_xml=document_fixture(
xml="""
<section eId="sec_1">
<num>1.</num>
<heading>Tester</heading>
<paragraph eId="sec_1.paragraph-0">
<content>
<p>Something to do with <ref href="/akn/za/act/2002/5">Cap. 12</ref>.</p>
</content>
</paragraph>
</section>"""
),
language=self.eng,
work=work)
self.finder.find_references_in_document(document)
root = etree.fromstring(expected.content)
expected.content = etree.tostring(root, encoding='utf-8').decode('utf-8')
self.assertEqual(expected.content, document.content)
def test_find_multiple_in_tail(self):
doc = Document(work=self.work,
content=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something Act 4 of 2000 and "<term>City</term>" means the (Act No. 117 of 1998) and also Act 5 of 2020;</p>
</content>
</paragraph>
</section>"""))
self.finder.find_references_in_document(doc)
self.assertMultiLineEqual(
'''<body xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something <ref href="/akn/za/act/2000/4">Act 4 of 2000</ref> and "<term>City</term>" means the (<ref href="/akn/za/act/1998/117">Act No. 117 of 1998</ref>) and also <ref href="/akn/za/act/2020/5">Act 5 of 2020</ref>;</p>
</content>
</paragraph>
</section>
</body>
''',
etree.tostring(doc.doc.body, encoding='utf-8',
pretty_print=True).decode('utf-8'))
def test_constitution(self):
doc = Document(work=self.work,
content=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>die Grondwet</p>
<p>die Grondwet, 1996</p>
<p>die Grondwet van Suid Afrika</p>
<p>die Grondwet van die Republiek van Suid-Afrika</p>
<p>die Grondwet van die Republiek van Suid-Afrika, 1996</p>
<p>die Grondwet van die Republiek van Suid-Afrika 1996</p>
<p>die Grondwet van die Republiek van Suid-Afrika Wet, 1996</p>
<p>die Grondwet van die Republiek van Suid-Afrika, 1996 ( Wet 108 van 1996 )</p>
<p>die Grondwet van die Republiek van Suid-Afrika, 1996 ( Wet No 108 van 1996 )</p>
<p>die Grondwet van die Republiek van Suid-Afrika, 1996 ( Wet No. 108 van 1996 )</p>
<p>die Grondwet van die Republiek van Suid-Afrika Wet, 1996 ( Wet No. 108 van 1996 )</p>
<p>die Grondwet van die Republiek van Suid-Afrika Wet 108 van 1996</p>
<p>die Grondwet van die Republiek van Suid-Afrika (Wet 108 van 1996)</p>
<p>the below shouldn't match</p>
<p>enige grondwet</p>
</content>
</paragraph>
</section>"""))
self.finder.find_references_in_document(doc)
self.maxDiff = None
self.assertMultiLineEqual(
'''<body xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet</ref></p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet, 1996</ref></p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van Suid Afrika</ref></p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika</ref></p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika, 1996</ref></p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika 1996</ref></p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika Wet, 1996</ref></p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika, 1996</ref> ( <ref href="/akn/za/act/1996/constitution">Wet 108 van 1996</ref> )</p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika, 1996</ref> ( <ref href="/akn/za/act/1996/constitution">Wet No 108 van 1996</ref> )</p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika, 1996</ref> ( <ref href="/akn/za/act/1996/constitution">Wet No. 108 van 1996</ref> )</p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika Wet, 1996</ref> ( <ref href="/akn/za/act/1996/constitution">Wet No. 108 van 1996</ref> )</p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika</ref> <ref href="/akn/za/act/1996/constitution">Wet 108 van 1996</ref></p>
<p>die <ref href="/akn/za/act/1996/constitution">Grondwet van die Republiek van Suid-Afrika</ref> (<ref href="/akn/za/act/1996/constitution">Wet 108 van 1996</ref>)</p>
<p>the below shouldn't match</p>
<p>enige grondwet</p>
</content>
</paragraph>
</section>
</body>
''',
etree.tostring(doc.doc.body, encoding='utf-8',
pretty_print=True).decode('utf-8'))
def test_find_without_act_in_parens(self):
doc = Document(work=self.work,
content=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with Income Tax Act, 1962 (No 58 of 1962).</p>
</content>
</paragraph>
</section>"""))
self.finder.find_references_in_document(doc)
self.maxDiff = None
self.assertMultiLineEqual(
'''<body xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with Income Tax Act, 1962 (<ref href="/akn/za/act/1962/58">No 58 of 1962</ref>).</p>
</content>
</paragraph>
</section>
</body>
''',
etree.tostring(doc.doc.body, encoding='utf-8',
pretty_print=True).decode('utf-8'))
def test_toc_with_schedule_no_heading(self):
doc = Document(work=self.work,
document_xml=component_fixture(text="hi"),
language=self.eng)
# strip the heading element, the builder will use the FRBRalias instead
for node in doc.doc.root.xpath('//a:attachment/a:heading',
namespaces={'a': doc.doc.namespace}):
node.getparent().remove(node)
toc = self.builder.table_of_contents_for_document(doc)
self.assertEqual([{
'type':
'attachment',
'component':
'schedule1',
'subcomponent':
None,
'title':
'Schedule alias',
'heading':
'Schedule alias',
'id':
'att_1',
'children': [{
'component': 'schedule1',
'title': 'Section',
'type': 'section',
'id': 'sec_1',
'subcomponent': 'section',
}],
}], [t.as_dict() for t in toc])
self.assertEqual("att_1/sec_1", toc[0].children[0].qualified_id)
def test_find_with_empty_string(self):
doc = Document(work=self.work,
content=document_fixture(xml="""
<section eId="sec_1">
<num>1.</num>
<heading/>
<hcontainer eId="sec_1__hcontainer_1">
<content>
<p>In this By-law, unless the context indicates otherwise-</p>
<p>"" means the National Road Traffic Act, 1996 (<ref href="/akn/za/act/1996/93">Act No. 93 of 1996</ref>);</p>
</content>
</hcontainer>
</section>"""))
self.maxDiff = None
self.finder.find_terms_in_document(doc)
self.assertMultiLineEqual(
'''<body xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
<section eId="sec_1">
<num>1.</num>
<heading/>
<hcontainer eId="sec_1__hcontainer_1">
<content>
<p>In this By-law, unless the context indicates otherwise-</p>
<p>"" means the National Road Traffic Act, 1996 (<ref href="/akn/za/act/1996/93">Act No. 93 of 1996</ref>);</p>
</content>
</hcontainer>
</section>
</body>
''',
etree.tostring(doc.doc.body, pretty_print=True).decode('utf-8'))
def test_fancy_quotes(self):
doc = Document(work=self.work,
content=document_fixture(xml="""
<section eId="sec_1">
<num>1.</num>
<heading>Definitions</heading>
<hcontainer eId="sec_1__hcontainer_1">
<content>
<p>“Act“ means the National Road Traffic Act, 1996 (<ref href="/akn/za/act/1996/93">Act No. 93 of 1996</ref>);</p>
</content>
</hcontainer>
</section>"""))
self.maxDiff = None
self.finder.find_terms_in_document(doc)
self.assertMultiLineEqual(
'''<body xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
<section eId="sec_1">
<num>1.</num>
<heading>Definitions</heading>
<hcontainer eId="sec_1__hcontainer_1">
<content>
<p refersTo="#term-Act">“<def refersTo="#term-Act">Act</def>“ means the National Road Traffic Act, 1996 (<ref href="/akn/za/act/1996/93">Act No. 93 of 1996</ref>);</p>
</content>
</hcontainer>
</section>
</body>
''',
etree.tostring(doc.doc.body, pretty_print=True,
encoding='UTF-8').decode('utf-8'))
def import_from_file(self, fname, frbr_uri):
cmd = ['bundle', 'exec', 'slaw', 'parse']
if self.fragment:
cmd.extend(['--fragment', self.fragment])
if self.fragment_id_prefix:
cmd.extend(['--id-prefix', self.fragment_id_prefix])
if self.section_number_position:
cmd.extend(
['--section-number-position', self.section_number_position])
cmd.extend(['--grammar', self.slaw_grammar])
cmd.append(fname)
code, stdout, stderr = self.shell(cmd)
if code > 0:
raise ValueError(stderr)
if not stdout:
raise ValueError("We couldn't get any useful text out of the file")
if self.fragment:
doc = Fragment(stdout.decode('utf-8'))
else:
doc = Document.randomized(frbr_uri)
doc.content = stdout.decode('utf-8')
doc.frbr_uri = frbr_uri # reset it
doc.title = None
doc.copy_attributes()
self.log.info("Successfully imported from %s" % fname)
return doc
def test_toc_with_schedule(self):
doc = Document(work=self.work,
document_xml=component_fixture(text="hi"),
language=self.eng)
toc = self.builder.table_of_contents_for_document(doc)
self.assertEqual([{
'type':
'attachment',
'component':
'schedule1',
'subcomponent':
None,
'title':
'Schedule 1',
'heading':
'Schedule 1',
'id':
'att_1',
'children': [{
'component': 'schedule1',
'title': 'Section',
'type': 'section',
'id': 'sec_1',
'subcomponent': 'section',
}],
}], [t.as_dict() for t in toc])
self.assertEqual("att_1/sec_1", toc[0].children[0].qualified_id)
def test_ignore_existing(self):
doc = Document(work=self.work,
content=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with <ref href="/akn/za/act/2012/22">Act no 22 of 2012</ref>.</p>
<p>And another thing about <ref href="/akn/za/act/1998/4"><b>Act 4 of 1998</b></ref>.</p>
</content>
</paragraph>
</section>"""))
self.finder.find_references_in_document(doc)
self.assertMultiLineEqual(
'''<body xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with <ref href="/akn/za/act/2012/22">Act no 22 of 2012</ref>.</p>
<p>And another thing about <ref href="/akn/za/act/1998/4"><b>Act 4 of 1998</b></ref>.</p>
</content>
</paragraph>
</section>
</body>
''',
etree.tostring(doc.doc.body, encoding='utf-8',
pretty_print=True).decode('utf-8'))
def import_from_upload(self, upload, frbr_uri, request):
""" Create a new Document by importing it from a
:class:`django.core.files.uploadedfile.UploadedFile` instance.
"""
self.reformat = True
if upload.content_type in ['text/xml', 'application/xml']:
# just assume it's valid AKN xml
doc = Document.randomized(frbr_uri)
doc.content = upload.read().decode('utf-8')
return doc
if upload.content_type == 'application/vnd.openxmlformats-officedocument.wordprocessingml.document':
# pre-process docx to HTML and then import html
html = self.docx_to_html(upload)
doc = self.import_from_text(html, frbr_uri, '.html')
elif upload.content_type == 'application/pdf':
doc = self.import_from_pdf(upload, frbr_uri)
else:
# slaw will do its best
with self.tempfile_for_upload(upload) as f:
doc = self.import_from_file(f.name, frbr_uri)
self.analyse_after_import(doc)
return doc
def test_ignore_existing(self):
doc = Document(content=document_fixture(xml=u"""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with <ref href="/za/act/2012/22">Act no 22 of 2012</ref>.</p>
<p>And another thing about <ref href="/za/act/1998/4"><b>Act 4 of 1998</b></ref>.</p>
</content>
</paragraph>
</section>"""))
self.finder.find_references_in_document(doc)
self.assertMultiLineEqual(
'''<body xmlns="http://www.akomantoso.org/2.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with <ref href="/za/act/2012/22">Act no 22 of 2012</ref>.</p>
<p>And another thing about <ref href="/za/act/1998/4"><b>Act 4 of 1998</b></ref>.</p>
</content>
</paragraph>
</section>
</body>
''', etree.tostring(doc.doc.body, pretty_print=True))
def test_find_multiple_in_tail(self):
doc = Document(content=document_fixture(xml=u"""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something Act 4 of 2000 and "<term>City</term>" means the (Act No. 117 of 1998) and also Act 5 of 2020;</p>
</content>
</paragraph>
</section>"""))
self.finder.find_references_in_document(doc)
self.assertMultiLineEqual(
'''<body xmlns="http://www.akomantoso.org/2.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something <ref href="/za/act/2000/4">Act 4 of 2000</ref> and "<term>City</term>" means the (<ref href="/za/act/1998/117">Act No. 117 of 1998</ref>) and also <ref href="/za/act/2020/5">Act 5 of 2020</ref>;</p>
</content>
</paragraph>
</section>
</body>
''', etree.tostring(doc.doc.body, pretty_print=True))
def test_find_without_act_in_parens(self):
doc = Document(content=document_fixture(xml=u"""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with Income Tax Act, 1962 (No 58 of 1962).</p>
</content>
</paragraph>
</section>"""))
self.finder.find_references_in_document(doc)
self.maxDiff = None
self.assertMultiLineEqual(
'''<body xmlns="http://www.akomantoso.org/2.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>Something to do with Income Tax Act, 1962 (<ref href="/za/act/1962/58">No 58 of 1962</ref>).</p>
</content>
</paragraph>
</section>
</body>
''', etree.tostring(doc.doc.body, pretty_print=True))
def test_unicode(self):
doc = Document(content=document_fixture(xml=u"""
<section id="section-1">
<num>1.</num>
<heading>Definitions</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>"Actë" means the National Road Traffic Act, 1996 (<ref href="/za/act/1996/93">Act No. 93 of 1996</ref>);</p>
</content>
</paragraph>
</section>
"""))
self.maxDiff = None
self.finder.find_terms_in_document(doc)
self.assertMultiLineEqual(
'''<body xmlns="http://www.akomantoso.org/2.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<section id="section-1">
<num>1.</num>
<heading>Definitions</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p refersTo="#term-Actë">"<def refersTo="#term-Actë">Actë</def>" means the National Road Traffic Act, 1996 (<ref href="/za/act/1996/93">Act No. 93 of 1996</ref>);</p>
</content>
</paragraph>
</section>
</body>
''', etree.tostring(doc.doc.body, pretty_print=True, encoding='UTF-8'))
def test_constitution(self):
doc = Document(work=self.work, content=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>the Constitution</p>
<p>the Constitution, 1996</p>
<p>the Constitution of South Africa</p>
<p>the Constitution of the Republic of South Africa</p>
<p>the Constitution of the Republic of South Africa, 1996</p>
<p>the Constitution of the Republic of South Africa 1996</p>
<p>the Constitution of the Republic of South Africa Act, 1996</p>
<p>the Constitution of the Republic of South Africa, 1996 ( Act 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa, 1996 ( Act No 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa, 1996 ( Act No. 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa Act, 1996 ( Act No. 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa Act 108 of 1996</p>
<p>the Constitution of the Republic of South Africa (Act 108 of 1996)</p>
<p>the below shouldn't match</p>
<p>Constitutionally unsound</p>
<p>is unconstitutional</p>
<p>their constitution is poor</p>
</content>
</paragraph>
</section>"""))
self.finder.find_references_in_document(doc)
self.maxDiff = None
self.assertMultiLineEqual('''<body xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution</ref></p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution, 1996</ref></p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of South Africa</ref></p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa</ref></p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa, 1996</ref></p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa 1996</ref></p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa Act, 1996</ref></p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa, 1996</ref> ( <ref href="/akn/za/act/1996/constitution">Act 108 of 1996</ref> )</p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa, 1996</ref> ( <ref href="/akn/za/act/1996/constitution">Act No 108 of 1996</ref> )</p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa, 1996</ref> ( <ref href="/akn/za/act/1996/constitution">Act No. 108 of 1996</ref> )</p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa Act, 1996</ref> ( <ref href="/akn/za/act/1996/constitution">Act No. 108 of 1996</ref> )</p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa</ref> <ref href="/akn/za/act/1996/constitution">Act 108 of 1996</ref></p>
<p>the <ref href="/akn/za/act/1996/constitution">Constitution of the Republic of South Africa</ref> (<ref href="/akn/za/act/1996/constitution">Act 108 of 1996</ref>)</p>
<p>the below shouldn't match</p>
<p>Constitutionally unsound</p>
<p>is unconstitutional</p>
<p>their constitution is poor</p>
</content>
</paragraph>
</section>
</body>
''', etree.tostring(doc.doc.body, encoding='utf-8', pretty_print=True).decode('utf-8'))
def test_find_with_empty_string(self):
doc = Document(content=document_fixture(xml=u"""
<section id="section-1">
<num>1.</num>
<heading/>
<paragraph id="section-1.paragraph-0">
<content>
<p>In this By-law, unless the context indicates otherwise-</p>
<p>"" means the National Road Traffic Act, 1996 (<ref href="/za/act/1996/93">Act No. 93 of 1996</ref>);</p>
</content>
</paragraph>
</section>
"""))
self.maxDiff = None
self.finder.find_terms_in_document(doc)
self.assertMultiLineEqual(
'''<body xmlns="http://www.akomantoso.org/2.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<section id="section-1">
<num>1.</num>
<heading/>
<paragraph id="section-1.paragraph-0">
<content>
<p>In this By-law, unless the context indicates otherwise-</p>
<p>"" means the National Road Traffic Act, 1996 (<ref href="/za/act/1996/93">Act No. 93 of 1996</ref>);</p>
</content>
</paragraph>
</section>
</body>
''', etree.tostring(doc.doc.body, pretty_print=True))
def test_dont_link_constitution_in_constitution(self):
constitution = Work(frbr_uri='/akn/za/act/1996/constitution')
doc = Document(work=constitution, content=document_fixture(xml="""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>the Constitution</p>
<p>the Constitution, 1996</p>
<p>the Constitution of South Africa</p>
<p>the Constitution of the Republic of South Africa</p>
<p>the Constitution of the Republic of South Africa, 1996</p>
<p>the Constitution of the Republic of South Africa 1996</p>
<p>the Constitution of the Republic of South Africa Act, 1996</p>
<p>the Constitution of the Republic of South Africa, 1996 ( Act 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa, 1996 ( Act No 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa, 1996 ( Act No. 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa Act, 1996 ( Act No. 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa Act 108 of 1996</p>
<p>the Constitution of the Republic of South Africa (Act 108 of 1996)</p>
</content>
</paragraph>
</section>"""))
unchanged = doc.document_xml
doc.doc.frbr_uri = FrbrUri.parse(constitution.frbr_uri)
self.finder.find_references_in_document(doc)
self.assertMultiLineEqual(unchanged, doc.document_xml)
def test_constitution(self):
doc = Document(content=document_fixture(xml=u"""
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>the Constitution</p>
<p>the Constitution, 1996</p>
<p>the Constitution of South Africa</p>
<p>the Constitution of the Republic of South Africa</p>
<p>the Constitution of the Republic of South Africa, 1996</p>
<p>the Constitution of the Republic of South Africa 1996</p>
<p>the Constitution of the Republic of South Africa Act, 1996</p>
<p>the Constitution of the Republic of South Africa, 1996 ( Act 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa, 1996 ( Act No 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa, 1996 ( Act No. 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa Act, 1996 ( Act No. 108 of 1996 )</p>
<p>the Constitution of the Republic of South Africa Act 108 of 1996</p>
<p>the Constitution of the Republic of South Africa (Act 108 of 1996)</p>
<p>the below shouldn't match</p>
<p>Constitutionally unsound</p>
<p>is unconstitutional</p>
<p>their constitution is poor</p>
</content>
</paragraph>
</section>
"""))
self.finder.find_references_in_document(doc)
self.maxDiff = None
self.assertMultiLineEqual(
'''<body xmlns="http://www.akomantoso.org/2.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<section id="section-1">
<num>1.</num>
<heading>Tester</heading>
<paragraph id="section-1.paragraph-0">
<content>
<p>the <ref href="/za/act/1996/constitution">Constitution</ref></p>
<p>the <ref href="/za/act/1996/constitution">Constitution, 1996</ref></p>
<p>the <ref href="/za/act/1996/constitution">Constitution of South Africa</ref></p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa</ref></p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa, 1996</ref></p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa 1996</ref></p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa Act, 1996</ref></p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa, 1996</ref> ( <ref href="/za/act/1996/constitution">Act 108 of 1996</ref> )</p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa, 1996</ref> ( <ref href="/za/act/1996/constitution">Act No 108 of 1996</ref> )</p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa, 1996</ref> ( <ref href="/za/act/1996/constitution">Act No. 108 of 1996</ref> )</p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa Act, 1996</ref> ( <ref href="/za/act/1996/constitution">Act No. 108 of 1996</ref> )</p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa</ref> <ref href="/za/act/1996/constitution">Act 108 of 1996</ref></p>
<p>the <ref href="/za/act/1996/constitution">Constitution of the Republic of South Africa</ref> (<ref href="/za/act/1996/constitution">Act 108 of 1996</ref>)</p>
<p>the below shouldn't match</p>
<p>Constitutionally unsound</p>
<p>is unconstitutional</p>
<p>their constitution is poor</p>
</content>
</paragraph>
</section>
</body>
''', etree.tostring(doc.doc.body, pretty_print=True))
def test_table_of_contents_afr(self):
d = Document()
d.content = document_fixture(xml="""
<body xmlns="http://www.akomantoso.org/2.0">
<section id="section-1">
<num>1.</num>
<heading>Foo</heading>
<content>
<p>hello</p>
</content>
</section>
<chapter id="chapter-1">
<num>1.</num>
<heading>The Chapter</heading>
<part id="part-A">
<num>A</num>
<heading>The Part</heading>
<section id="section-2">
<num>2.</num>
<heading>Other</heading>
<content>
<p>hi</p>
</content>
</section>
</part>
</chapter>
</body>
""")
d.language = 'afr'
toc = d.table_of_contents()
toc = [t.as_dict() for t in toc]
self.maxDiff = None
self.assertEqual(toc, [
{'id': 'section-1', 'num': '1.', 'type': 'section', 'heading': 'Foo',
'component': 'main', 'subcomponent': 'section/1', 'title': '1. Foo'},
{'id': 'chapter-1', 'num': '1.', 'type': 'chapter', 'heading': 'The Chapter',
'component': 'main', 'subcomponent': 'chapter/1', 'title': 'Hoofstuk 1. - The Chapter', 'children': [
{'id': 'part-A', 'num': 'A', 'type': 'part', 'heading': 'The Part',
'component': 'main', 'subcomponent': 'chapter/1/part/A', 'title': 'Deel A - The Part', 'children': [
{'id': 'section-2', 'num': '2.', 'type': 'section', 'heading': 'Other',
'component': 'main', 'subcomponent': 'section/2', 'title': '2. Other'},
]
},
]},
])
