Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-07-09 03:04:10 -04:00
Code Issues Packages Projects Releases Wiki Activity

HTML 5 parser: Preserve non-core namespace declarations on the <html> tag

Browse Source
This commit is contained in:
Kovid Goyal 2014-02-01 09:46:31 +05:30
parent cc731ecd46
commit d9806a5e6e
2 changed files with 36 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
src/calibre/ebooks/oeb/polish/parsing.py
View File

@ -275,7 +275,7 @@ def makeelement_ns(ctx, namespace, prefix, name, attrib, nsmap):
for k, v in elem.items(): # Only elem.items() preserves attrib order for k, v in elem.items(): # Only elem.items() preserves attrib order
nelem.set(k, v) nelem.set(k, v)
for (prefix, name), v in namespaced_attribs.iteritems(): for (prefix, name), v in namespaced_attribs.iteritems():
ns = nsmap.get('prefix', None) ns = nsmap.get(prefix, None)
if ns is not None: if ns is not None:
try: try:
nelem.set('{%s}%s' % (ns, name), v) nelem.set('{%s}%s' % (ns, name), v)
@ -331,6 +331,13 @@ class TreeBuilder(BaseTreeBuilder):
self.openElements.append(element) self.openElements.append(element)
self.document.appendChild(element) self.document.appendChild(element)
def promote_elem(self, elem, tag_name):
' Add the paraphernalia to elem that the html5lib infrastructure needs '
self.proxy_cache.append(elem)
elem.name = tag_name
elem.namespace = elem.nsmap[elem.prefix]
elem.nameTuple = (elem.nsmap[elem.prefix], elem.name)
def createElement(self, token, nsmap=None): def createElement(self, token, nsmap=None):
"""Create an element but don't insert it anywhere""" """Create an element but don't insert it anywhere"""
nsmap = nsmap or {} nsmap = nsmap or {}
@ -346,10 +353,7 @@ class TreeBuilder(BaseTreeBuilder):
# Keep a reference to elem so that lxml does not delete and re-create # Keep a reference to elem so that lxml does not delete and re-create
# it, losing the name related attributes # it, losing the name related attributes
self.proxy_cache.append(elem) self.promote_elem(elem, token_name)
elem.name = token_name
elem.namespace = elem.nsmap[elem.prefix]
elem.nameTuple = (elem.nsmap[elem.prefix], elem.name)
position = token.get('position', None) position = token.get('position', None)
if position is not None: if position is not None:
# Unfortunately, libxml2 can only store line numbers upto 65535 # Unfortunately, libxml2 can only store line numbers upto 65535
@ -388,6 +392,18 @@ class TreeBuilder(BaseTreeBuilder):
self.openElements.append(element) self.openElements.append(element)
return element return element
def clone_node(self, elem, nsmap_update):
assert len(elem) == 0
nsmap = elem.nsmap.copy()
nsmap.update(nsmap_update)
nelem = self.lxml_context.makeelement(elem.tag, nsmap=nsmap)
self.promote_elem(nelem, elem.tag.rpartition('}')[2])
nelem.sourceline = elem.sourceline
for k, v in elem.items():
nelem.set(k, v)
nelem.text, nelem.tail = elem.text, elem.tail
return nelem
def apply_html_attributes(self, attrs): def apply_html_attributes(self, attrs):
if not attrs: if not attrs:
return return
@ -403,7 +419,18 @@ class TreeBuilder(BaseTreeBuilder):
continue continue
if k == 'xml:lang' and 'lang' not in html.attrib: if k == 'xml:lang' and 'lang' not in html.attrib:
k = 'lang' k = 'lang'
html.set(to_xml_name(k), v) html.set(k, v)
continue
if k.startswith('xmlns:') and v not in known_namespaces and v != namespaces['html'] and len(html) == 0:
# We have a namespace declaration, the only way to add
# it to the existing html node is to replace it.
prefix = k[len('xmlns:'):]
if not prefix:
continue
self.openElements[0] = html = self.clone_node(html, {prefix:v})
self.document.appendChild(html)
else:
html.set(to_xml_name(k), v)
def apply_body_attributes(self, attrs): def apply_body_attributes(self, attrs):
if not attrs: if not attrs:

3
src/calibre/ebooks/oeb/polish/tests/parsing.py
View File

@ -180,6 +180,9 @@ class ParsingTests(BaseTest):
root = parse('<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en-US" xmlns:xml="http://www.w3.org/XML/1998/namespace"><body/></html>') root = parse('<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en-US" xmlns:xml="http://www.w3.org/XML/1998/namespace"><body/></html>')
self.assertNotIn('xmlnsU0003Axml', root.attrib, 'xml namespace declaration not removed') self.assertNotIn('xmlnsU0003Axml', root.attrib, 'xml namespace declaration not removed')
root = parse('<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en-US" xmlns:extra="extra"><body/></html>')
self.assertIn('extra', root.nsmap, 'Extra namespace declaration on <html> tag not preserved')
def timing(): def timing():
import time, sys import time, sys
from calibre.ebooks.chardet import xml_to_unicode from calibre.ebooks.chardet import xml_to_unicode