3
�Pfa7 � @ s� d dl mZmZmZ d dlmZ d dlZd dlmZm Z ddl
m Z m Z m
Z
ddl
mZmZmZ ddlmZmZ d d lmZ d
je
�d Zejd e d
�Zejd e d �Zi Ze d�dkZ x�e ej � �D ]p\Z Z!e r�e e!�dks�e r�e e!�dkr�q�e!dkr�e e!�dk�rej"e!�Z!ne#e!�Z!e!ek�s4e j$� r�e ee!< q�W dd� Z%ede%� d dd�Z&G dd� de'�Z(G dd� de)�Z*dS ) � )�absolute_import�division�unicode_literals)� text_typeN)�register_error�xmlcharrefreplace_errors� )� voidElements�booleanAttributes�spaceCharacters)�rcdataElements�entities� xmlEntities)� treewalkers�_utils)�escape� z"'=<>`�[�]u_
/` ᠏ ]u � �&c
C s" t | ttf��rg }g }d}x�t| j| j| j� �D ]n\}}|rFd}q4|| j }tj| j|t | j|d g�� �r�tj
| j||d � �}d}nt |�}|j |� q4W x^|D ]V}t
j|�} | r�|j d� |j | � | jd�s�|j d� q�|j dt|�dd � � q�W dj|�| jfS t| �S d S )NFr Tr �;z&#x%s;r )�
isinstance�UnicodeEncodeError�UnicodeTranslateError� enumerate�object�start�endr ZisSurrogatePair�min�surrogatePairToCodepoint�ord�append�_encode_entity_map�get�endswith�hex�joinr )
�exc�resZ
codepoints�skip�i�c�indexZ codepointZcp�e� r/ � /usr/lib/python3.6/serializer.py�htmlentityreplace_errors* s0
"
r1 �htmlentityreplace�etreec K s$ t j|�}tf |�}|j|| �|�S )N)r Z
getTreeWalker�HTMLSerializer�render)�inputZtree�encodingZserializer_optsZwalker�sr/ r/ r0 � serializeJ s
r9 c @ s~ e Zd ZdZdZdZdZdZdZdZ dZ
dZ dZ dZ
dZdZdZd!Zdd� Zdd� Zdd� Zd"dd�Zd#d d �Zd$d d �ZdS )%r4 �legacy�"TF�quote_attr_values�
quote_char�use_best_quote_char�omit_optional_tags�minimize_boolean_attributes�use_trailing_solidus� space_before_trailing_solidus�escape_lt_in_attrs�
escape_rcdata�resolve_entities�alphabetical_attributes�inject_meta_charset�strip_whitespace�sanitizec K sz t |�t | j� }t|�dkr2tdtt|�� ��d|kr@d| _x(| jD ] }t| ||j|t | |��� qHW g | _
d| _ dS )a6 Initialize HTMLSerializer.
Keyword options (default given first unless specified) include:
inject_meta_charset=True|False
Whether it insert a meta element to define the character set of the
document.
quote_attr_values="legacy"|"spec"|"always"
Whether to quote attribute values that don't require quoting
per legacy browser behaviour, when required by the standard, or always.
quote_char=u'"'|u"'"
Use given quote character for attribute quoting. Default is to
use double quote unless attribute value contains a double quote,
in which case single quotes are used instead.
escape_lt_in_attrs=False|True
Whether to escape < in attribute values.
escape_rcdata=False|True
Whether to escape characters that need to be escaped within normal
elements within rcdata elements such as style.
resolve_entities=True|False
Whether to resolve named character entities that appear in the
source tree. The XML predefined entities < > & " '
are unaffected by this setting.
strip_whitespace=False|True
Whether to remove semantically meaningless whitespace. (This
compresses all whitespace to a single space except within pre.)
minimize_boolean_attributes=True|False
Shortens boolean attributes to give just the attribute value,
for example becomes .
use_trailing_solidus=False|True
Includes a close-tag slash at the end of the start tag of void
elements (empty elements whose end tag is forbidden). E.g. .
space_before_trailing_solidus=True|False
Places a space immediately before the closing slash in a tag
using a trailing solidus. E.g. . Requires use_trailing_solidus.
sanitize=False|True
Strip all unsafe or unknown constructs from output.
See `html5lib user documentation`_
omit_optional_tags=True|False
Omit start/end tags that are optional.
alphabetical_attributes=False|True
Reorder attributes to be in alphabetical order.
.. _html5lib user documentation: http://code.google.com/p/html5lib/wiki/UserDocumentation
r z2__init__() got an unexpected keyword argument '%s'r= FN) � frozenset�options�len� TypeError�next�iterr> �setattrr$ �getattr�errors�strict)�self�kwargsZunexpected_args�attrr/ r/ r0 �__init__p s . zHTMLSerializer.__init__c C s | j r|j| j d�S |S d S )Nr2 )r7 �encode)rT �stringr/ r/ r0 rX � s zHTMLSerializer.encodec C s | j r|j| j d�S |S d S )NrS )r7 rX )rT rY r/ r/ r0 � encodeStrict� s zHTMLSerializer.encodeStrictNc c s� || _ d}g | _|r0| jr0ddlm} |||�}| jrJddlm} ||�}| jrdddlm} ||�}| j r~ddl
m} ||�}| j r�ddl m} ||�}�xR|D �]H}|d }|dk�r`d|d }|d r�|d |d 7 }n|d
r�|d 7 }|d
�rJ|d
j
d �d
k�r0|d
j
d�d
k�r*| jd� d}nd }|d||d
|f 7 }|d7 }| j|�V q�|d5k�r�|dk�sz|�r�|�r�|d j
d�d
k�r�| jd� | j|d �V n| jt|d ��V q�|d6k�r�|d } | jd| �V | tk�r| j �rd}n|�r| jd� �x�|d j� D �]�\\}
} } | }
| }| jd �V | j|
�V | j �s�|
tj| t� �k�r"|
tjd t� �k�r"| jd �V | jd k�s�t|�d
k�r�d}n@| jd k�r�tj |�d k }n$| jd!k�r�t j |�d k }nt d"��|j d#d$�}| j �r|j d%d&�}|�r�| j!}| j"�rTd|k�r�
Characters�SpaceCharacters�dataz�Commentz--zComment contains --z ZEntityr zEntity %s not recognizedz&%s;)r` ra )rc rd ))r7 rR rG Zfilters.inject_meta_charsetr[ rF Z filters.alphabeticalattributesrH Zfilters.whitespacerI Zfilters.sanitizerr? Zfilters.optionaltags�find�serializeErrorrZ rX r r rD �itemsr@ r
r$ �tupler< rL �_quoteAttributeSpec�search�_quoteAttributeLegacy�
ValueError�replacerC r= r> r rA rB r
rE r )rT �
treewalkerr7 Zin_cdatar[ �tokenr\ Zdoctyper= r] �_Z attr_nameZ
attr_value�k�vZ
quote_attrrb �keyr/ r/ r0 r9 � s�
zHTMLSerializer.serializec C s2 |rdj t| j||���S dj t| j|���S d S )N� r )r' �listr9 )rT ru r7 r/ r/ r0 r5 ? s zHTMLSerializer.render�XXX ERROR MESSAGE NEEDEDc C s | j j|� | jrt�d S )N)rR r" rS �SerializeError)rT rb r/ r/ r0 rm E s z HTMLSerializer.serializeError)r< r= r> r? r@ rA rB rC rD rE rF rG rH rI )N)N)r} )�__name__�
__module__� __qualname__r< r= r> r? r@ rA rB rC rD rE rF rG rH rI rK rW rX rZ r9 r5 rm r/ r/ r/ r0 r4 Q s6 8
r4 c @ s e Zd ZdZdS )r~ zError in serialized treeN)r r� r� �__doc__r/ r/ r/ r0 r~ L s r~ )r3 N)+Z
__future__r r r Zpip._vendor.sixr �re�codecsr r Z constantsr r
r r r
r r r r Zxml.sax.saxutilsr r' Z_quoteAttributeSpecChars�compilerp rr r# rL Z_is_ucs4r| rn rx ry r r! �islowerr1 r9 r r4 � Exceptionr~ r/ r/ r/ r0 � s:
|