mirror of
https://github.com/Hopiu/wagtail.git
synced 2026-05-03 04:54:45 +00:00
gracefully handle unrecognised elements at the top level
This commit is contained in:
Matt Westcott
Thibaud Colas
parent
1378b766ee
commit
808ad56e0f
2 changed files with 92 additions and 11 deletions
|
|
@ -10,7 +10,7 @@ class HandlerState(object):
|
|||
self.current_block = None
|
||||
self.current_inline_styles = []
|
||||
self.current_entity_ranges = []
|
||||
self.depth = 0
|
||||
self.list_depth = 0
|
||||
self.list_item_type = None
|
||||
self.pushed_states = []
|
||||
|
||||
|
|
@ -19,7 +19,7 @@ class HandlerState(object):
|
|||
'current_block': self.current_block,
|
||||
'current_inline_styles': self.current_inline_styles,
|
||||
'current_entity_ranges': self.current_entity_ranges,
|
||||
'depth': self.depth,
|
||||
'list_depth': self.list_depth,
|
||||
'list_item_type': self.list_item_type
|
||||
})
|
||||
|
||||
|
|
@ -28,7 +28,7 @@ class HandlerState(object):
|
|||
self.current_block = last_state['current_block']
|
||||
self.current_inline_styles = last_state['current_inline_styles']
|
||||
self.current_entity_ranges = last_state['current_entity_ranges']
|
||||
self.depth = last_state['depth']
|
||||
self.list_depth = last_state['list_depth']
|
||||
self.list_item_type = last_state['list_item_type']
|
||||
|
||||
|
||||
|
|
@ -45,7 +45,7 @@ class ListElementHandler(object):
|
|||
pass
|
||||
else:
|
||||
# start the next nesting level
|
||||
state.depth += 1
|
||||
state.list_depth += 1
|
||||
|
||||
state.list_item_type = self.list_item_type
|
||||
|
||||
|
|
@ -59,7 +59,7 @@ class BlockElementHandler(object):
|
|||
|
||||
def create_block(self, name, attrs, state, contentstate):
|
||||
assert state.current_block is None, "%s element found nested inside another block" % name
|
||||
return Block(self.block_type, depth=state.depth)
|
||||
return Block(self.block_type, depth=state.list_depth)
|
||||
|
||||
def handle_starttag(self, name, attrs, state, contentstate):
|
||||
block = self.create_block(name, dict(attrs), state, contentstate)
|
||||
|
|
@ -80,7 +80,7 @@ class ListItemElementHandler(BlockElementHandler):
|
|||
|
||||
def create_block(self, name, attrs, state, contentstate):
|
||||
assert state.list_item_type is not None, "%s element found outside of an enclosing list element" % name
|
||||
return Block(state.list_item_type, depth=state.depth)
|
||||
return Block(state.list_item_type, depth=state.list_depth)
|
||||
|
||||
|
||||
class InlineStyleElementHandler(object):
|
||||
|
|
@ -131,7 +131,7 @@ class AtomicBlockEntityElementHandler(object):
|
|||
entity = self.create_entity(name, dict(attrs), state, contentstate)
|
||||
key = contentstate.add_entity(entity)
|
||||
|
||||
block = Block('atomic', depth=state.depth)
|
||||
block = Block('atomic', depth=state.list_depth)
|
||||
contentstate.blocks.append(block)
|
||||
block.text = ' '
|
||||
entity_range = EntityRange(key)
|
||||
|
|
@ -208,6 +208,7 @@ class HtmlToContentStateHandler(HTMLParser):
|
|||
def reset(self):
|
||||
self.state = HandlerState()
|
||||
self.contentstate = ContentState()
|
||||
self.element_depth = 0 # number of unclosed start tags encountered, including the one currently being handled
|
||||
super().reset()
|
||||
|
||||
def add_block(self, block):
|
||||
|
|
@ -215,20 +216,35 @@ class HtmlToContentStateHandler(HTMLParser):
|
|||
self.current_block = block
|
||||
|
||||
def handle_starttag(self, name, attrs):
|
||||
self.element_depth += 1
|
||||
try:
|
||||
element_handler = self.element_handlers[name]
|
||||
except KeyError:
|
||||
return # ignore unrecognised elements
|
||||
if self.element_depth == 1:
|
||||
# treat unrecognised top-level elements as paragraphs
|
||||
element_handler = self.element_handlers['p']
|
||||
else:
|
||||
# ignore unrecognised elements below the top-level
|
||||
element_handler = None
|
||||
|
||||
element_handler.handle_starttag(name, attrs, self.state, self.contentstate)
|
||||
if element_handler:
|
||||
element_handler.handle_starttag(name, attrs, self.state, self.contentstate)
|
||||
|
||||
def handle_endtag(self, name):
|
||||
try:
|
||||
element_handler = self.element_handlers[name]
|
||||
except KeyError:
|
||||
return # ignore unrecognised elements
|
||||
if self.element_depth == 1:
|
||||
# treat unrecognised top-level elements as paragraphs
|
||||
element_handler = self.element_handlers['p']
|
||||
else:
|
||||
# ignore unrecognised elements below the top-level
|
||||
element_handler = None
|
||||
|
||||
element_handler.handle_endtag(name, self.state, self.contentstate)
|
||||
if element_handler:
|
||||
element_handler.handle_endtag(name, self.state, self.contentstate)
|
||||
|
||||
self.element_depth -= 1
|
||||
|
||||
def handle_data(self, content):
|
||||
if self.state.current_block is None:
|
||||
|
|
|
|||
65
wagtail/admin/tests/test_contentstate.py
Normal file
65
wagtail/admin/tests/test_contentstate.py
Normal file
|
|
@ -0,0 +1,65 @@
|
|||
import json
|
||||
|
||||
from django.test import TestCase
|
||||
|
||||
from wagtail.admin.rich_text.converters.contentstate import ContentstateConverter
|
||||
|
||||
|
||||
def content_state_equal(v1, v2):
|
||||
"Test whether two contentState structures are equal, ignoring 'key' properties"
|
||||
if type(v1) != type(v2):
|
||||
return False
|
||||
|
||||
if type(v1) == dict:
|
||||
if set(v1.keys()) != set(v2.keys()):
|
||||
return False
|
||||
return all(
|
||||
k == 'key' or content_state_equal(v, v2[k])
|
||||
for k, v in v1.items()
|
||||
)
|
||||
elif type(v1) == list:
|
||||
if len(v1) != len(v2):
|
||||
return False
|
||||
return all(
|
||||
content_state_equal(a, b) for a, b in zip(v1, v2)
|
||||
)
|
||||
else:
|
||||
return v1 == v2
|
||||
|
||||
|
||||
class TestHtmlToContentState(TestCase):
|
||||
def assertContentStateEqual(self, v1, v2):
|
||||
"Assert that two contentState structures are equal, ignoring 'key' properties"
|
||||
self.assertTrue(content_state_equal(v1, v2), "%r does not match %r" % (v1, v2))
|
||||
|
||||
def test_paragraphs(self):
|
||||
converter = ContentstateConverter(features=[])
|
||||
result = json.loads(converter.from_database_format(
|
||||
'''
|
||||
<p>Hello world!</p>
|
||||
<p>Goodbye world!</p>
|
||||
'''
|
||||
))
|
||||
self.assertContentStateEqual(result, {
|
||||
'entityMap': {},
|
||||
'blocks': [
|
||||
{'inlineStyleRanges': [], 'text': 'Hello world!', 'depth': 0, 'type': 'unstyled', 'key': '00000', 'entityRanges': []},
|
||||
{'inlineStyleRanges': [], 'text': 'Goodbye world!', 'depth': 0, 'type': 'unstyled', 'key': '00000', 'entityRanges': []},
|
||||
]
|
||||
})
|
||||
|
||||
def test_unknown_block_becomes_paragraph(self):
|
||||
converter = ContentstateConverter(features=[])
|
||||
result = json.loads(converter.from_database_format(
|
||||
'''
|
||||
<foo>Hello world!</foo>
|
||||
<p>Goodbye world!</p>
|
||||
'''
|
||||
))
|
||||
self.assertContentStateEqual(result, {
|
||||
'entityMap': {},
|
||||
'blocks': [
|
||||
{'inlineStyleRanges': [], 'text': 'Hello world!', 'depth': 0, 'type': 'unstyled', 'key': '00000', 'entityRanges': []},
|
||||
{'inlineStyleRanges': [], 'text': 'Goodbye world!', 'depth': 0, 'type': 'unstyled', 'key': '00000', 'entityRanges': []},
|
||||
]
|
||||
})
|
||||
Loading…
Reference in a new issue