3
B;WU7 @ 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 ddkZ xe ej D ]p\Z Z!e re e!dkse re e!dkrqe!dkre e!dkrej"e!Z!ne#e!Z!e!eks4e j$ re ee!< qW dd Z%ede% d ddZ&G dd de'Z(G dd de)Z*dS ) )absolute_importdivisionunicode_literals) text_typeN)register_errorxmlcharrefreplace_errors ) voidElementsbooleanAttributesspaceCharacters)rcdataElementsentities xmlEntities) treewalkers_utils)escape z"'=<>`[]u_
/` ᠏ ]u &c
C s" t | ttfrg }g }d}xt| j| j| j D ]n\}}|rFd}q4|| j }tj| j|t | j|d g rtj
| j||d }d}nt |}|j | q4W x^|D ]V}t
j|} | r|j d |j | | jds|j d q|j dt|dd qW dj|| jfS t| S d S )NFr Tr ;z&#x%s;r )
isinstanceUnicodeEncodeErrorUnicodeTranslateError enumerateobjectstartendr ZisSurrogatePairminsurrogatePairToCodepointordappend_encode_entity_mapgetendswithhexjoinr )
excresZ
codepointsskipicindexZ codepointZcpe r/ /usr/lib/python3.6/serializer.pyhtmlentityreplace_errors* s0
"
r1 htmlentityreplaceetreec K s$ t j|}tf |}|j|| |S )N)r Z
getTreeWalkerHTMLSerializerrender)inputZtreeencodingZserializer_optsZwalkersr/ r/ r0 serializeJ s
r9 c @ s~ e Zd ZdZdZdZdZdZdZdZ dZ
dZ dZ dZ
dZdZdZd!Zdd Zdd Zdd Zd"ddZd#d d Zd$d d ZdS )%r4 legacy"TFquote_attr_values
quote_charuse_best_quote_charomit_optional_tagsminimize_boolean_attributesuse_trailing_solidus space_before_trailing_solidusescape_lt_in_attrs
escape_rcdataresolve_entitiesalphabetical_attributesinject_meta_charsetstrip_whitespacesanitizec 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) frozensetoptionslen TypeErrornextiterr> setattrr$ getattrerrorsstrict)selfkwargsZunexpected_argsattrr/ r/ r0 __init__p s . zHTMLSerializer.__init__c C s* t |tst| jr"|j| jdS |S d S )Nr2 )r r AssertionErrorr7 encode)rT stringr/ r/ r0 rY s zHTMLSerializer.encodec C s* t |tst| jr"|j| jdS |S d S )NrS )r r rX r7 rY )rT rZ r/ r/ r0 encodeStrict s zHTMLSerializer.encodeStrictNc c s || _ d}g | _|r0| jr0ddlm} |||}| jrJddlm} ||}| jrdddlm} ||}| j r~ddl
m} ||}| j rddl m} ||}xR|D ]H}|d }|dkr`d|d }|d r|d |d 7 }n|d
r|d 7 }|d
rJ|d
j
d d
kr0|d
j
dd
kr*| jd d}nd }|d||d
|f 7 }|d7 }| j|V q|d5kr|dksz|r|r|d j
dd
kr| jd | j|d V n| jt|d V q|d6kr|d } | jd| V | tkr| j rd}n|r| jd x|d j D ]\\}
} } | }
| }| jd V | j|
V | j s|
tj| t kr"|
tjd t kr"| jd V | jd kst|d
krd}n@| jd krtj |d k }n$| jd!krt j |d k }nt d"|j d#d$}| j r|j d%d&}|r| j!}| j"rTd|kr
CharactersSpaceCharactersdatazCommentz--zComment contains --z ZEntityr zEntity %s not recognizedz&%s;)ra rb )rd re ))r7 rR rG Zfilters.inject_meta_charsetr\ rF Z filters.alphabeticalattributesrH Zfilters.whitespacerI Zfilters.sanitizerr? Zfilters.optionaltagsfindserializeErrorr[ rY r r rD itemsr@ r
r$ tupler< rL _quoteAttributeSpecsearch_quoteAttributeLegacy
ValueErrorreplacerC r= r> r rA rB r
rE r )rT
treewalkerr7 Zin_cdatar\ tokenr] Zdoctyper= r^ _Z attr_nameZ
attr_valuekvZ
quote_attrrc keyr/ r/ r0 r9 s
zHTMLSerializer.serializec C s2 |rdj t| j||S dj t| j|S d S )N r )r' listr9 )rT rv r7 r/ r/ r0 r5 ? s zHTMLSerializer.renderXXX ERROR MESSAGE NEEDEDc C s | j j| | jrtd S )N)rR r" rS SerializeError)rT rc r/ r/ r0 rn 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 rY r[ r9 r5 rn 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 Zsixr recodecsr r Z constantsr r
r r r
r r r r Zxml.sax.saxutilsr r' Z_quoteAttributeSpecCharscompilerq rs r# rL Z_is_ucs4r} ro ry rz r r! islowerr1 r9 r r4 Exceptionr r/ r/ r/ r0 s:
|