3 �Pfa7�@s�ddlmZmZmZddlmZddlZddlmZm Z ddl mZmZm Z ddl mZmZmZddlmZmZdd lmZd je �dZejded �Zejded�ZiZed�dkZx�eej��D]p\Z Z!er�ee!�dks�er�ee!�dkr�q�e!dkr�ee!�dk�rej"e!�Z!ne#e!�Z!e!ek�s4e j$�r�e ee!<q�Wdd�Z%ede%�ddd�Z&Gdd�de'�Z(Gdd�de)�Z*dS)�)�absolute_import�division�unicode_literals)� text_typeN)�register_error�xmlcharrefreplace_errors�)�voidElements�booleanAttributes�spaceCharacters)�rcdataElements�entities�xmlEntities)�treewalkers�_utils)�escape�z"'=<>`�[�]u_ /`  ᠎᠏     　]u􏿿��&c Cs"t|ttf��rg}g}d}x�t|j|j|j��D]n\}}|rFd}q4||j}tj|j|t |j|dg��r�tj |j||d��}d}nt|�}|j|�q4Wx^|D]V}t j|�} | r�|jd�|j| �| jd�s�|jd�q�|jdt|�dd��q�Wdj|�|jfSt|�SdS)NFrTr�;z&#x%s;r)� isinstance�UnicodeEncodeError�UnicodeTranslateError� enumerate�object�start�endrZisSurrogatePair�min�surrogatePairToCodepoint�ord�append�_encode_entity_map�get�endswith�hex�joinr) �exc�resZ codepoints�skip�i�c�indexZ codepointZcp�e�r/� /usr/lib/python3.6/serializer.py�htmlentityreplace_errors*s0 " r1�htmlentityreplace�etreecKs$tj|�}tf|�}|j||�|�S)N)rZ getTreeWalker�HTMLSerializer�render)�inputZtree�encodingZserializer_optsZwalker�sr/r/r0� serializeJs r9c@s~eZdZdZdZdZdZdZdZdZ dZ dZdZdZ dZdZdZd!Zdd�Zdd�Zdd�Zd"dd�Zd#dd�Zd$dd �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 Kszt|�t|j�}t|�dkr2tdtt|��d|kr@d|_x(|jD]}t|||j|t ||��qHWg|_ 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 rz2__init__() got an unexpected keyword argument '%s'r=FN)� frozenset�options�len� TypeError�next�iterr>�setattrr$�getattr�errors�strict)�self�kwargsZunexpected_args�attrr/r/r0�__init__ps.zHTMLSerializer.__init__cCs*t|t�st�|jr"|j|jd�S|SdS)Nr2)rr�AssertionErrorr7�encode)rT�stringr/r/r0rY�szHTMLSerializer.encodecCs*t|t�st�|jr"|j|jd�S|SdS)NrS)rrrXr7rY)rTrZr/r/r0�encodeStrict�szHTMLSerializer.encodeStrictNccs�||_d}g|_|r0|jr0ddlm}|||�}|jrJddlm}||�}|jrdddlm}||�}|j r~ddl m}||�}|jr�ddlm}||�}�xR|D�]H}|d}|dk�r`d|d}|dr�|d |d7}n|d r�|d7}|d �rJ|d j d�d k�r0|d j d�d k�r*|jd�d}nd}|d||d |f7}|d7}|j|�Vq�|d5k�r�|dk�sz|�r�|�r�|dj d�d k�r�|jd�|j|d�Vn|jt|d��Vq�|d6k�r�|d} |jd| �V| tk�r|j�rd}n|�r|jd��x�|dj�D�]�\\} }}|} |}|jd�V|j| �V|j�s�| tj| t��k�r"| tjdt��k�r"|jd�V|jdk�s�t|�d k�r�d}n@|jd k�r�tj|�dk }n$|jd!k�r�tj|�dk }ntd"��|jd#d$�}|j �r|jd%d&�}|�r�|j!}|j"�rTd|k�r� Characters�SpaceCharacters�dataz�Commentz--zComment contains --z ZEntityrzEntity %s not recognizedz&%s;)rarb)rdre))r7rRrGZfilters.inject_meta_charsetr\rFZfilters.alphabeticalattributesrHZfilters.whitespacerIZfilters.sanitizerr?Zfilters.optionaltags�find�serializeErrorr[rYrrrD�itemsr@r r$�tupler<rL�_quoteAttributeSpec�search�_quoteAttributeLegacy� ValueError�replacerCr=r>r rArBr rEr)rT� treewalkerr7Zin_cdatar\�tokenr]Zdoctyper=r^�_Z attr_nameZ attr_value�k�vZ quote_attrrc�keyr/r/r0r9�s� zHTMLSerializer.serializecCs2|rdjt|j||��Sdjt|j|��SdS)N�r)r'�listr9)rTrvr7r/r/r0r5?szHTMLSerializer.render�XXX ERROR MESSAGE NEEDEDcCs|jj|�|jrt�dS)N)rRr"rS�SerializeError)rTrcr/r/r0rnEszHTMLSerializer.serializeError)r<r=r>r?r@rArBrCrDrErFrGrHrI)N)N)r~)�__name__� __module__�__qualname__r<r=r>r?r@rArBrCrDrErFrGrHrIrKrWrYr[r9r5rnr/r/r/r0r4Qs68 r4c@seZdZdZdS)rzError in serialized treeN)r�r�r��__doc__r/r/r/r0rLsr)r3N)+Z __future__rrrZpip._vendor.sixr�re�codecsrrZ constantsr r rrr rrrrZxml.sax.saxutilsrr'Z_quoteAttributeSpecChars�compilerqrsr#rLZ_is_ucs4r}roryrzr r!�islowerr1r9rr4� Exceptionrr/r/r/r0�s: |