def import_docx_file(self, user, work, date, language, docx_file, filesize):
document = Document()
document.work = work
document.expression_date = date
document.language = language
document.created_by_user = user
document.save()
importer = plugins.for_document('importer', document)
# hard-coded for Namibian docxes
importer.section_number_position = 'after-title'
upload = UploadedFile(file=docx_file,
content_type='application/vnd.openxmlformats-officedocument.wordprocessingml.document',
size=filesize)
try:
importer.create_from_upload(upload, document, None)
except ValueError as e:
print("Error during import: %s" % e.message)
raise ValidationError(e.message or "error during import")
docx_file.seek(0)
filename = os.path.split(docx_file.name)[1]
att = Attachment()
att.filename = filename
att.mime_type = 'application/vnd.openxmlformats-officedocument.wordprocessingml.document'
att.document = document
att.size = filesize
att.file.save(att.filename, docx_file)
document.updated_by_user = user
document.save_with_revision(user)
self.create_review_task(document, user, filename)
def test_get_subcomponent(self):
d = Document(language=self.eng)
d.work = self.work
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-2">
<num>2.</num>
<heading>The Chapter</heading>
<content>
<p>hi</p>
</content>
</chapter>
</body>
""")
assert_is_not_none(d.doc.components()['main'])
elem = d.get_subcomponent('main', 'chapter/2')
assert_equal(elem.get('id'), "chapter-2")
elem = d.get_subcomponent('main', 'section/1')
assert_equal(elem.get('id'), "section-1")
assert_is_none(d.get_subcomponent('main', 'chapter/99'))
assert_is_none(d.get_subcomponent('main', 'section/99'))
def form_valid(self, form):
data = form.cleaned_data
upload = data['file']
opts = data.get('options', {})
document = Document()
document.work = self.work
document.expression_date = data['expression_date']
document.language = data['language']
document.created_by_user = self.request.user
document.save()
importer = plugins.for_document('importer', document)
importer.section_number_position = opts.get('section_number_position',
'guess')
importer.cropbox = opts.get('cropbox', None)
try:
importer.create_from_upload(upload, document, self.request)
except ValueError as e:
log.error("Error during import: %s" % e.message, exc_info=e)
raise ValidationError(e.message or "error during import")
document.updated_by_user = self.request.user
document.save_with_revision(self.request.user)
# add source file as an attachment
AttachmentSerializer(context={
'document': document
}).create({'file': upload})
return JsonResponse(
{'location': reverse('document', kwargs={'doc_id': document.id})})
def form_valid(self, form):
data = form.cleaned_data
upload = data['file']
opts = data.get('options', {})
document = Document()
document.work = self.work
document.expression_date = data['expression_date']
document.language = data['language']
document.created_by_user = self.request.user
document.save()
importer = plugins.for_document('importer', document)
importer.section_number_position = opts.get('section_number_position',
'guess')
importer.cropbox = opts.get('cropbox', None)
importer.page_nums = form.cleaned_data['page_nums']
try:
importer.create_from_upload(upload, document, self.request)
except ValueError as e:
log.error("Error during import: %s" % str(e), exc_info=e)
return JsonResponse({'file': str(e) or "error during import"},
status=400)
document.updated_by_user = self.request.user
document.save_with_revision(self.request.user)
return JsonResponse(
{'location': reverse('document', kwargs={'doc_id': document.id})})
def test_preamble_and_friends_in_table_of_contents(self):
d = Document()
d.work = self.work
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>
""")
d.language = Language.objects.get(language__pk='en')
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_subpart_without_number(self):
d = Document()
d.work = self.work
d.content = document_fixture(xml="""
<body>
<subpart eId="subpart_1">
<heading>My subpart</heading>
</subpart>
</body>
""")
d.language = Language.objects.get(language__pk='en')
toc = d.table_of_contents()
toc = [t.as_dict() for t in toc]
self.maxDiff = None
self.assertEqual([
{
'type': 'subpart',
'component': 'main',
'subcomponent': 'subpart',
'title': 'My subpart',
'heading': 'My subpart',
'id': 'subpart_1'
},
], toc)
def test_set_content(self):
d = Document()
d.work = self.work
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_table_of_contents_afr(self):
d = Document()
d.work = self.work
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 = Language.objects.get(language__pk='af')
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'},
]
},
]},
])
def test_component_table_of_contents(self):
d = Document()
d.work = self.work
d.content = """<akomaNtoso xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://www.akomantoso.org/2.0" xsi:schemaLocation="http://www.akomantoso.org/2.0 akomantoso20.xsd">
<act contains="singleVersion">
<meta>
<identification source="#openbylaws">
<FRBRWork>
<FRBRthis value="/za/by-law/cape-town/2002/community-fire-safety/main"/>
<FRBRuri value="/za/by-law/cape-town/2002/community-fire-safety"/>
<FRBRalias value="Community Fire Safety By-law"/>
<FRBRdate date="2002-02-28" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRcountry value="za"/>
</FRBRWork>
<FRBRExpression>
<FRBRthis value="/za/by-law/cape-town/2002/community-fire-safety/main/eng@"/>
<FRBRuri value="/za/by-law/cape-town/2002/community-fire-safety/eng@"/>
<FRBRdate date="2002-02-28" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRlanguage language="eng"/>
</FRBRExpression>
<FRBRManifestation>
<FRBRthis value="/za/by-law/cape-town/2002/community-fire-safety/main/eng@"/>
<FRBRuri value="/za/by-law/cape-town/2002/community-fire-safety/eng@"/>
<FRBRdate date="2014-01-14" name="Generation"/>
<FRBRauthor href="#openbylaws" as="#author"/>
</FRBRManifestation>
</identification>
</meta>
<body></body>
</act>
<components>
<component id="component-1">
<doc name="schedule1">
<meta>
<identification source="#slaw">
<FRBRWork>
<FRBRthis value="/za/act/1980/01/schedule1"/>
<FRBRuri value="/za/act/1980/01"/>
<FRBRalias value="A Title"/>
<FRBRdate date="1980-01-01" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRcountry value="za"/>
</FRBRWork>
<FRBRExpression>
<FRBRthis value="/za/act/1980/01/eng@/schedule1"/>
<FRBRuri value="/za/act/1980/01/eng@"/>
<FRBRdate date="1980-01-01" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRlanguage language="eng"/>
</FRBRExpression>
<FRBRManifestation>
<FRBRthis value="/za/act/1980/01/eng@/schedule1"/>
<FRBRuri value="/za/act/1980/01/eng@"/>
<FRBRdate date="2015-04-22" name="Generation"/>
<FRBRauthor href="#slaw" as="#author"/>
</FRBRManifestation>
</identification>
</meta>
<mainBody>
<section id="schedule-1.section-0">
<content>
<p>1. Foo</p>
<p>2. Bar</p>
</content>
</section>
</mainBody>
</doc>
</component>
<component id="component-2">
<doc name="schedule2">
<meta>
<identification source="#slaw">
<FRBRWork>
<FRBRthis value="/za/act/1980/01/schedule2"/>
<FRBRuri value="/za/act/1980/01"/>
<FRBRdate date="1980-01-01" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRcountry value="za"/>
</FRBRWork>
<FRBRExpression>
<FRBRthis value="/za/act/1980/01/eng@/schedule2"/>
<FRBRuri value="/za/act/1980/01/eng@"/>
<FRBRdate date="1980-01-01" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRlanguage language="eng"/>
</FRBRExpression>
<FRBRManifestation>
<FRBRthis value="/za/act/1980/01/eng@/schedule2"/>
<FRBRuri value="/za/act/1980/01/eng@"/>
<FRBRdate date="2015-04-22" name="Generation"/>
<FRBRauthor href="#slaw" as="#author"/>
</FRBRManifestation>
</identification>
</meta>
<mainBody>
<section id="schedule-2.section-0">
<content>
<p>Baz</p>
<p>Boom</p>
</content>
</section>
</mainBody>
</doc>
</component>
</components>
</akomaNtoso>
"""
d.language = Language.objects.get(language__pk='en')
toc = d.table_of_contents()
toc = [t.as_dict() for t in toc]
self.maxDiff = None
self.assertEqual([
{
'component':
'schedule1',
'type':
'doc',
'subcomponent':
None,
'heading':
'A Title',
'title':
'A Title',
'children': [{
'component': 'schedule1',
'type': 'section',
'id': 'schedule-1.section-0',
'subcomponent': 'section',
'title': 'Section'
}]
},
{
'component':
'schedule2',
'type':
'doc',
'subcomponent':
None,
'heading':
'Schedule 2',
'title':
'Schedule 2',
'children': [{
'component': 'schedule2',
'type': 'section',
'id': 'schedule-2.section-0',
'subcomponent': 'section',
'title': 'Section'
}]
},
], toc)
def test_component_table_of_contents(self):
d = Document()
d.work = self.work
d.content = """<akomaNtoso xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
<act contains="singleVersion">
<meta>
<identification source="#openbylaws">
<FRBRWork>
<FRBRthis value="/za/by-law/cape-town/2002/community-fire-safety/!main"/>
<FRBRuri value="/za/by-law/cape-town/2002/community-fire-safety"/>
<FRBRalias value="Community Fire Safety By-law"/>
<FRBRdate date="2002-02-28" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRcountry value="za"/>
</FRBRWork>
<FRBRExpression>
<FRBRthis value="/za/by-law/cape-town/2002/community-fire-safety/main/eng@"/>
<FRBRuri value="/za/by-law/cape-town/2002/community-fire-safety/eng@"/>
<FRBRdate date="2002-02-28" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRlanguage language="eng"/>
</FRBRExpression>
<FRBRManifestation>
<FRBRthis value="/za/by-law/cape-town/2002/community-fire-safety/main/eng@"/>
<FRBRuri value="/za/by-law/cape-town/2002/community-fire-safety/eng@"/>
<FRBRdate date="2014-01-14" name="Generation"/>
<FRBRauthor href="#openbylaws" as="#author"/>
</FRBRManifestation>
</identification>
</meta>
<body></body>
<attachments>
<attachment eId="att_1">
<heading>A Title</heading>
<doc name="schedule">
<meta>
<identification source="#slaw">
<FRBRWork>
<FRBRthis value="/za/act/1980/01/!schedule1"/>
<FRBRuri value="/za/act/1980/01"/>
<FRBRalias value="A Title"/>
<FRBRdate date="1980-01-01" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRcountry value="za"/>
</FRBRWork>
<FRBRExpression>
<FRBRthis value="/za/act/1980/01/eng@/schedule1"/>
<FRBRuri value="/za/act/1980/01/eng@"/>
<FRBRdate date="1980-01-01" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRlanguage language="eng"/>
</FRBRExpression>
<FRBRManifestation>
<FRBRthis value="/za/act/1980/01/eng@/schedule1"/>
<FRBRuri value="/za/act/1980/01/eng@"/>
<FRBRdate date="2015-04-22" name="Generation"/>
<FRBRauthor href="#slaw" as="#author"/>
</FRBRManifestation>
</identification>
</meta>
<mainBody>
<section eId="sec_1">
<content>
<p>1. Foo</p>
<p>2. Bar</p>
</content>
</section>
</mainBody>
</doc>
</attachment>
<attachment eId="att_2">
<heading>Schedule 2</heading>
<doc name="schedule">
<meta>
<identification source="#slaw">
<FRBRWork>
<FRBRthis value="/za/act/1980/01/!schedule2"/>
<FRBRuri value="/za/act/1980/01"/>
<FRBRdate date="1980-01-01" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRcountry value="za"/>
</FRBRWork>
<FRBRExpression>
<FRBRthis value="/za/act/1980/01/eng@/schedule2"/>
<FRBRuri value="/za/act/1980/01/eng@"/>
<FRBRdate date="1980-01-01" name="Generation"/>
<FRBRauthor href="#council" as="#author"/>
<FRBRlanguage language="eng"/>
</FRBRExpression>
<FRBRManifestation>
<FRBRthis value="/za/act/1980/01/eng@/schedule2"/>
<FRBRuri value="/za/act/1980/01/eng@"/>
<FRBRdate date="2015-04-22" name="Generation"/>
<FRBRauthor href="#slaw" as="#author"/>
</FRBRManifestation>
</identification>
</meta>
<mainBody>
<section eId="sec_1">
<content>
<p>Baz</p>
<p>Boom</p>
</content>
</section>
</mainBody>
</doc>
</attachment>
</attachments>
</act>
</akomaNtoso>
"""
d.language = Language.objects.get(language__pk='en')
toc = d.table_of_contents()
toc = [t.as_dict() for t in toc]
self.maxDiff = None
self.assertEqual([{
'component':
'schedule1',
'type':
'attachment',
'subcomponent':
None,
'id':
'att_1',
'heading':
'A Title',
'title':
'A Title',
'children': [{
'component': 'schedule1',
'type': 'section',
'id': 'sec_1',
'subcomponent': 'section',
'title': 'Section',
}]
}, {
'component':
'schedule2',
'type':
'attachment',
'subcomponent':
None,
'id':
'att_2',
'heading':
'Schedule 2',
'title':
'Schedule 2',
'children': [{
'component': 'schedule2',
'type': 'section',
'id': 'sec_1',
'subcomponent': 'section',
'title': 'Section'
}]
}], toc)