Python2/PyMuPDF: mupdf-source/thirdparty/gumbo-parser/python/gumbo/html5lib

annotate mupdf-source/thirdparty/gumbo-parser/python/gumbo/html5lib_adapter.py @ 9:b2d41d02518f

Use the LLVM compilers by symlinking them in the venv. This is because CC= and CXX= do not work for some subprojects.

author	Franz Glasner <fzglas.hg@dom66.de>
date	Wed, 17 Sep 2025 06:11:34 +0200
parents	b50eed0cc0ef
children

rev	line source
2 b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	1 # Copyright 2012 Google Inc. All Rights Reserved.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	2 #
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	3 # Licensed under the Apache License, Version 2.0 (the "License");
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	4 # you may not use this file except in compliance with the License.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	5 # You may obtain a copy of the License at
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	6 #
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	7 # http://www.apache.org/licenses/LICENSE-2.0
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	8 #
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	9 # Unless required by applicable law or agreed to in writing, software
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	10 # distributed under the License is distributed on an "AS IS" BASIS,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	11 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	12 # See the License for the specific language governing permissions and
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	13 # limitations under the License.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	14 #
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	15 """Adapter between Gumbo and html5lib.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	16
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	17 This exports one method, parse, with the same signature as html5lib.parse. It
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	18 takes the text to parse, and optionally an html5lib TreeBuilder to build the
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	19 tree, and gives back a DOM tree in that format. Example:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	20
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	21 doc = parse(text, treebuilder='lxml')
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	22 """
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	23
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	24 __author__ = 'jdtang@google.com (Jonathan Tang)'
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	25
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	26 import gumboc
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	27
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	28 # These should match html5lib.constants.namespaces, and be indexed by the enum
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	29 # values of gumboc.Namespace
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	30 _NAMESPACES = [
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	31 'http://www.w3.org/1999/xhtml',
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	32 'http://www.w3.org/2000/svg',
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	33 'http://www.w3.org/1998/Math/MathML',
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	34 ]
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	35
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	36
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	37 def _convert_doctype(treebuilder, source_node):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	38 if not source_node.has_doctype:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	39 # Mimic html5lib behavior: if no doctype token, no doctype node.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	40 return
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	41 treebuilder.insertDoctype({
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	42 'name': source_node.name.decode('utf-8'),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	43 'publicId': source_node.public_identifier.decode('utf-8'),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	44 'systemId': source_node.system_identifier.decode('utf-8'),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	45 })
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	46
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	47
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	48 def _convert_attributes(source_node):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	49 def maybe_namespace(attr):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	50 if attr.namespace != gumboc.AttributeNamespace.NONE:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	51 return (repr(attr.namespace).lower() if attr.name != 'xmlns' else None,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	52 attr.name.decode('utf-8'),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	53 attr.namespace.to_url())
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	54 else:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	55 return attr.name.decode('utf-8')
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	56 return dict((maybe_namespace(attr), attr.value.decode('utf-8'))
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	57 for attr in source_node.attributes)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	58
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	59
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	60 def _convert_element(source_node):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	61 if source_node.type not in ( gumboc.NodeType.ELEMENT, gumboc.NodeType.TEMPLATE):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	62 # If-statement instead of assert so it runs with -O
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	63 raise AssertionError(
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	64 '_convert_element only works with elements; found %r' %
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	65 source_node.type)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	66 return {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	67 'name': source_node.v.element.tag_name.decode('utf-8'),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	68 'namespace': _NAMESPACES[source_node.v.element.tag_namespace.value],
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	69 'data': _convert_attributes(source_node),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	70 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	71
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	72
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	73 def _insert_root(treebuilder, source_node, pop_element = True):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	74 treebuilder.insertRoot(_convert_element(source_node))
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	75 for child_node in source_node.children:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	76 _insert_node(treebuilder, child_node)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	77 if pop_element:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	78 treebuilder.openElements.pop()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	79
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	80 def _insert_node(treebuilder, source_node):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	81 assert source_node.type != gumboc.NodeType.DOCUMENT
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	82 if source_node.type == gumboc.NodeType.COMMENT:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	83 treebuilder.insertComment({'data': source_node.v.text.text.decode('utf-8')})
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	84 elif source_node.type in (
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	85 gumboc.NodeType.TEXT,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	86 gumboc.NodeType.WHITESPACE,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	87 gumboc.NodeType.CDATA):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	88 treebuilder.insertText(source_node.v.text.text.decode('utf-8'))
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	89 else:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	90 treebuilder.insertElementNormal(_convert_element(source_node))
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	91 for child_node in source_node.v.element.children:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	92 _insert_node(treebuilder, child_node)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	93 treebuilder.openElements.pop()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	94
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	95
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	96 class HTMLParser(object):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	97 def __init__(self, tree):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	98 self.tree = tree
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	99
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	100 def parse(self, text_or_file, **kwargs):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	101 try:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	102 text = text_or_file.read()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	103 except AttributeError:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	104 # Assume a string.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	105 text = text_or_file
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	106
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	107 with gumboc.parse(text, **kwargs) as output:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	108 _convert_doctype(self.tree, output.contents.document.contents)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	109 for node in output.contents.document.contents.children:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	110 if node.type == gumboc.NodeType.COMMENT:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	111 self.tree.insertComment({'data': node.v.text.text.decode('utf-8')},
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	112 self.tree.document)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	113 elif node.type in (gumboc.NodeType.ELEMENT, gumboc.NodeType.TEMPLATE):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	114 _insert_root(self.tree, output.contents.root.contents)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	115 else:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	116 assert 'Only comments and <html> nodes allowed at the root'
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	117 return self.tree.getDocument()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	118
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	119 def parseFragment(self, text_or_file, container, **kwargs):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	120 try:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	121 text = text_or_file.read()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	122 except AttributeError:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	123 # Assume a string.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	124 text = text_or_file
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	125 if ' ' in container:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	126 container_ns, container = container.split(' ')
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	127 else:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	128 container_ns = "html"
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	129
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	130 with gumboc.parse(
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	131 text,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	132 fragment_context=gumboc.Tag.from_str(container),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	133 fragment_namespace=getattr(gumboc.Namespace, container_ns.upper()),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	134 **kwargs) as output:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	135 for node in output.contents.document.contents.children:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	136 if node.type in (gumboc.NodeType.ELEMENT, gumboc.NodeType.TEMPLATE):
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	137 _insert_root(self.tree, output.contents.root.contents, False)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	138 else:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	139 assert 'Malformed fragment parse (??)'
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	140 return self.tree.getFragment()

Mercurial > hgrepos > Python2 > PyMuPDF

annotate mupdf-source/thirdparty/gumbo-parser/python/gumbo/html5lib_adapter.py @ 9:b2d41d02518f