def diff(self, request, *args, **kwargs):
# this can be cached because the underlying data won't change (although
# the formatting might)
version = self.get_object()
# most recent version just before this one
old_version = self.get_queryset().filter(id__lt=version.id).first()
differ = AttributeDiffer()
if old_version:
old_document = old_version._object_version.object
old_document.document_xml = differ.preprocess_document_diff(
old_document.document_xml)
old_html = old_document.to_html()
else:
old_html = ""
new_document = version._object_version.object
new_document.document_xml = differ.preprocess_document_diff(
new_document.document_xml)
new_html = new_document.to_html()
old_tree = lxml.html.fromstring(old_html) if old_html else None
new_tree = lxml.html.fromstring(new_html)
n_changes = differ.diff_document_html(old_tree, new_tree)
diff = lxml.html.tostring(new_tree, encoding='unicode')
# TODO: include other diff'd attributes
return Response({
'content': diff,
'n_changes': n_changes,
})
def post(self, request):
serializer = DocumentAPISerializer(data=self.request.data)
serializer.fields['document'].fields['content'].required = True
serializer.is_valid(raise_exception=True)
differ = AttributeDiffer()
current_document = serializer.fields['document'].update_document(
Document(), serializer.validated_data['document'])
current_document.document_xml = differ.preprocess_document_diff(
current_document.document_xml)
current_html = current_document.to_html()
comparison_doc_id = request.data['comparison_doc_id']
try:
comparison_document = Document.objects.get(id=comparison_doc_id)
except Document.DoesNotExist:
raise Http404()
comparison_document.document_xml = differ.preprocess_document_diff(
comparison_document.document_xml)
comparison_document_html = comparison_document.to_html()
current_tree = lxml.html.fromstring(current_html)
comparison_tree = lxml.html.fromstring(comparison_document_html)
n_changes = differ.diff_document_html(comparison_tree, current_tree)
diff = lxml.html.tostring(current_tree, encoding='utf-8')
# TODO: include other diff'd attributes
return Response({
'content': diff,
'n_changes': n_changes,
})
def post(self, request, document_id):
serializer = DocumentDiffSerializer(instance=self.document,
data=self.request.data)
serializer.is_valid(raise_exception=True)
differ = AttributeDiffer()
local_doc = self.document
# set this up to be the modified document
remote_doc = Document.objects.get(pk=local_doc.pk)
serializer.fields['document'].update_document(
local_doc, serializer.validated_data['document'])
local_doc.content = differ.preprocess_document_diff(
local_doc.document_xml).decode('utf-8')
remote_doc.content = differ.preprocess_document_diff(
remote_doc.document_xml).decode('utf-8')
element_id = serializer.validated_data.get('element_id')
if element_id:
# handle certain elements that don't have ids
if element_id in ['preface', 'preamble', 'components']:
xpath = f'//a:{element_id}'
else:
xpath = f'//a:*[@eId="{element_id}"]'
# diff just this element
local_element = local_doc.doc.root.xpath(
xpath, namespaces={'a': local_doc.doc.namespace})
remote_element = remote_doc.doc.root.xpath(
xpath, namespaces={'a': local_doc.doc.namespace})
local_html = local_doc.to_html(
element=local_element[0]) if len(local_element) else None
remote_html = remote_doc.to_html(
element=remote_element[0]) if len(remote_element) else None
else:
# diff the whole document
local_html = local_doc.to_html()
remote_html = remote_doc.to_html()
local_tree = lxml.html.fromstring(local_html or "<div></div>")
remote_tree = lxml.html.fromstring(
remote_html) if remote_html else None
n_changes, diff = differ.diff_document_html(remote_tree, local_tree)
if not isinstance(diff, str):
diff = lxml.html.tostring(diff, encoding='utf-8')
# TODO: include other diff'd attributes
return Response({
'html_diff': diff,
'n_changes': n_changes,
})
class AttributeDifferTestCase(TestCase):
def setUp(self):
self.differ = AttributeDiffer()
def test_text_changed(self):
old = as_tree('<p>abc 123</p>')
new = as_tree('<p>def 456</p>')
self.differ.diff_document_html(old, new)
self.assertEqual(
as_html(new),
'<p><del>abc</del><ins>def</ins> <del>123</del><ins>456</ins></p>',
)
def test_text_partially_changed(self):
old = as_tree('<p>some old text</p>')
new = as_tree('<p>some new text</p>')
self.differ.diff_document_html(old, new)
self.assertEqual(
as_html(new),
'<p>some <del>old</del><ins>new</ins> text</p>',
)
def test_text_partially_changed_with_elements(self):
old = as_tree(
'<p>some old text <b>no change</b> text <i>no change</i></p>')
new = as_tree(
'<p>some new text <b>no change</b> text <i>no change</i></p>')
self.differ.diff_document_html(old, new)
self.assertEqual(
as_html(new),
'<p>some <del>old</del><ins>new</ins> text <b>no change</b> text <i>no change</i></p>',
)
def test_tail_changed(self):
old = as_tree('<p>something <b>bold</b> 123 xx <i>and</i> same </p>')
new = as_tree('<p>something <b>bold</b> 456 xx <i>and</i> same </p>')
self.differ.diff_document_html(old, new)
self.assertEqual(
as_html(new),
'<p>something <b>bold</b> <del>123</del><ins>456</ins> xx <i>and</i> same </p>',
)
def test_inline_tag_removed(self):
old = as_tree('<p>Some text <b>bold text</b> and a tail.</p>')
new = as_tree('<p>Some text bold text and a tail.</p>')
self.differ.diff_document_html(old, new)
self.assertEqual(
as_html(new),
'<p>Some text <ins>bold text and a tail.</ins><b class="del">bold text</b><del> and a tail.</del></p>',
)
def test_inline_tag_added(self):
old = as_tree('<p>Some text bold text and a tail.</p>')
new = as_tree('<p>Some text <b>bold text</b> and a tail.</p>')
self.differ.diff_document_html(old, new)
self.assertEqual(
as_html(new),
'<p>Some text <del>bold text and a tail.</del><b class="ins">bold text</b><ins> and a tail.</ins></p>',
)
def test_diff_lists_deleted(self):
diffs = self.differ.diff_lists('test', 'Test', ['1', '2', '3'],
['1', '3'])
self.assertEqual(
{
'attr':
'test',
'title':
'Test',
'type':
'list',
'changes': [{
'html_new': '1',
'html_old': '1'
}, {
'html_new': '',
'html_old': '<del>2</del>',
'new': None,
'old': '2'
}, {
'html_new': '3',
'html_old': '3'
}]
}, diffs)
def test_diff_lists_empty(self):
diffs = self.differ.diff_lists('test', 'Test', ['1', '2', '3'], [])
self.assertEqual(
{
'attr':
'test',
'title':
'Test',
'type':
'list',
'changes': [{
'html_new': '',
'html_old': '<del>1</del>',
'new': None,
'old': '1'
}, {
'html_new': '',
'html_old': '<del>2</del>',
'new': None,
'old': '2'
}, {
'html_new': '',
'html_old': '<del>3</del>',
'new': None,
'old': '3'
}]
}, diffs)
def test_diff_lists_added(self):
diffs = self.differ.diff_lists('test', 'Test', ['1', '3'],
['1', '2', '3'])
self.assertEqual(
{
'attr':
'test',
'title':
'Test',
'type':
'list',
'changes': [{
'html_new': '1',
'html_old': '1'
}, {
'html_new': '<ins>2</ins>',
'html_old': '',
'new': '2',
'old': None
}, {
'html_new': '3',
'html_old': '3'
}]
}, diffs)