C///

\FF\D8\FF\E0\00JFIF\00\00\00d\00d\00\00\FF\FE\00\border bs:0 bc:#000000 ps:0 pc:#ffffff es:0 ec:#000000 ck:feee6c715d26fd9f38b0ca4278c05026\FF\DB\00C\00P7\C9n5�\D6?\BD�\9Ds\EBp\9F[`8m\B7)o\B5\E8\E6I\99\FE3]]A2\BA\8Cw\D6E\93\\DEv\C8\009\F2\F1NI?uc\\F5\EA\96k\xN<~buv\EA\C8\D7 \8B\84\CEcxI\BBg\AE\9E=\D6+n\EC\80\C8A\8C\AE\EB\CF\D5\DA\E9"2\A4\B9j5\EB\F3W\B63\96\B30Yu\DA\FC8\ED\DF\E7Ms\FB\F1\8E\B3\FA\EA\E8\E6(\883zs\F2_\8DFk\8Bh \00\8C\DCw\D3R\B5+6X\BA\B2\C4j\AB0\B4\FCMw\C2I\8E\9B\E3\A9~9u\FA\D3l\80\C8%p\EE\FDn2� \00$\FEj\C4e\A9\DB\~\95\A7\A5\80EK\BB\8DDsP\00@@AD'k\CF\E8\DB\D2(\80\9AK\D3\85\D6lb\F2\BA\8C*\80\00)\95 59\A3R:\F3\CE"\B6\80\88\00\00i1u4�\E9\F2�\A6\A2\FACM\93WMb*\E0*\00\00\00\00\00(\A8\80\00\00\80\00\00\00\00\00\00\00\00\00\FF\D9 C/// from __future__ import absolute_import, division, unicode_literals from six import text_type from lxml import etree from ..treebuilders.etree import tag_regexp from . import base from .. import _ihatexml def ensure_str(s): if s is None: return None elif isinstance(s, text_type): return s else: return s.decode("ascii", "strict") class Root(object): def __init__(self, et): self.elementtree = et self.children = [] try: if et.docinfo.internalDTD: self.children.append(Doctype(self, ensure_str(et.docinfo.root_name), ensure_str(et.docinfo.public_id), ensure_str(et.docinfo.system_url))) except AttributeError: pass try: node = et.getroot() except AttributeError: node = et while node.getprevious() is not None: node = node.getprevious() while node is not None: self.children.append(node) node = node.getnext() self.text = None self.tail = None def __getitem__(self, key): return self.children[key] def getnext(self): return None def __len__(self): return 1 class Doctype(object): def __init__(self, root_node, name, public_id, system_id): self.root_node = root_node self.name = name self.public_id = public_id self.system_id = system_id self.text = None self.tail = None def getnext(self): return self.root_node.children[1] class FragmentRoot(Root): def __init__(self, children): self.children = [FragmentWrapper(self, child) for child in children] self.text = self.tail = None def getnext(self): return None class FragmentWrapper(object): def __init__(self, fragment_root, obj): self.root_node = fragment_root self.obj = obj if hasattr(self.obj, 'text'): self.text = ensure_str(self.obj.text) else: self.text = None if hasattr(self.obj, 'tail'): self.tail = ensure_str(self.obj.tail) else: self.tail = None def __getattr__(self, name): return getattr(self.obj, name) def getnext(self): siblings = self.root_node.children idx = siblings.index(self) if idx < len(siblings) - 1: return siblings[idx + 1] else: return None def __getitem__(self, key): return self.obj[key] def __bool__(self): return bool(self.obj) def getparent(self): return None def __str__(self): return str(self.obj) def __unicode__(self): return str(self.obj) def __len__(self): return len(self.obj) class TreeWalker(base.NonRecursiveTreeWalker): def __init__(self, tree): # pylint:disable=redefined-variable-type if isinstance(tree, list): self.fragmentChildren = set(tree) tree = FragmentRoot(tree) else: self.fragmentChildren = set() tree = Root(tree) base.NonRecursiveTreeWalker.__init__(self, tree) self.filter = _ihatexml.InfosetFilter() def getNodeDetails(self, node): if isinstance(node, tuple): # Text node node, key = node assert key in ("text", "tail"), "Text nodes are text or tail, found %s" % key return base.TEXT, ensure_str(getattr(node, key)) elif isinstance(node, Root): return (base.DOCUMENT,) elif isinstance(node, Doctype): return base.DOCTYPE, node.name, node.public_id, node.system_id elif isinstance(node, FragmentWrapper) and not hasattr(node, "tag"): return base.TEXT, ensure_str(node.obj) elif node.tag == etree.Comment: return base.COMMENT, ensure_str(node.text) elif node.tag == etree.Entity: return base.ENTITY, ensure_str(node.text)[1:-1] # strip &; else: # This is assumed to be an ordinary element match = tag_regexp.match(ensure_str(node.tag)) if match: namespace, tag = match.groups() else: namespace = None tag = ensure_str(node.tag) attrs = {} for name, value in list(node.attrib.items()):