U
.e= @ sz d dl mZmZmZ d dlmZ d dlZd dlmZm Z ddl
m Z m Z m
Z
ddl
mZmZmZ ddlmZmZ d d lmZ d
e
d Zed e d
Zed e d Zi Ze ddkZ e e D ]n\Z Z!e re e!dkse se e!dkrqe!dkre e!dkre"e!Z!ne#e!Z!e!eks0e $ re ee!< qdd 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}t| j| j| j D ]n\}}|rDd}q2|| j }t| j|t | j|d g rt
| j||d }d}nt |}| | q2|D ]V}t
|} | r| d | | | ds| d q| dt|dd qd|| jfS t| S d S )NFr Tr ;z&#x%s;r )
isinstanceUnicodeEncodeErrorUnicodeTranslateError enumerateobjectstartendr ZisSurrogatePairminsurrogatePairToCodepointordappend_encode_entity_mapgetendswithhexjoinr )
excresZ
codepointsskipicindexZ codepointZcpe r/ C/usr/lib/python3.8/site-packages/pip/_vendor/html5lib/serializer.pyhtmlentityreplace_errors* s0
"
r1 htmlentityreplaceetreec K s$ t |}tf |}||| |S )a Serializes the input token stream using the specified treewalker
:arg input: the token stream to serialize
:arg tree: the treewalker to use
:arg encoding: the encoding to use
:arg serializer_opts: any options to pass to the
:py:class:`html5lib.serializer.HTMLSerializer` that gets created
:returns: the tree serialized as a string
Example:
>>> from html5lib.html5parser import parse
>>> from html5lib.serializer import serialize
>>> token_stream = parse('
'
)r Z
getTreeWalkerHTMLSerializerrender)inputZtreeencodingZserializer_optsZwalkersr/ r/ r0 serializeK s
r9 c @ s~ e Zd ZdZdZdZdZdZdZdZ dZ
dZ dZ dZ
dZdZdZdZdd Zdd Zd
d Zdd
dZdddZdddZd S )r4 legacy"TF)quote_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 sv t |t | j }t|dkr2tdtt| d|kr@d| _| jD ] }t| |||t | | qFg | _
d| _ dS )aB
Initialize HTMLSerializer
:arg inject_meta_charset: Whether or not to inject the meta charset.
Defaults to ``True``.
:arg quote_attr_values: Whether to quote attribute values that don't
require quoting per legacy browser behavior (``"legacy"``), when
required by the standard (``"spec"``), or always (``"always"``).
Defaults to ``"legacy"``.
:arg quote_char: Use given quote character for attribute quoting.
Defaults to ``"`` which will use double quotes unless attribute
value contains a double quote, in which case single quotes are
used.
:arg escape_lt_in_attrs: Whether or not to escape ``<`` in attribute
values.
Defaults to ``False``.
:arg escape_rcdata: Whether to escape characters that need to be
escaped within normal elements within rcdata elements such as
style.
Defaults to ``False``.
:arg resolve_entities: Whether to resolve named character entities that
appear in the source tree. The XML predefined entities < >
& " ' are unaffected by this setting.
Defaults to ``True``.
:arg strip_whitespace: Whether to remove semantically meaningless
whitespace. (This compresses all whitespace to a single space
except within ``pre``.)
Defaults to ``False``.
:arg minimize_boolean_attributes: Shortens boolean attributes to give
just the attribute value, for example::
becomes::
Defaults to ``True``.
:arg use_trailing_solidus: Includes a close-tag slash at the end of the
start tag of void elements (empty elements whose end tag is
forbidden). E.g. ````.
Defaults to ``False``.
:arg space_before_trailing_solidus: Places a space immediately before
the closing slash in a tag using a trailing solidus. E.g.
````. Requires ``use_trailing_solidus=True``.
Defaults to ``True``.
:arg sanitize: Strip all unsafe or unknown constructs from output.
See :py:class:`html5lib.filters.sanitizer.Filter`.
Defaults to ``False``.
:arg omit_optional_tags: Omit start/end tags that are optional.
Defaults to ``True``.
:arg alphabetical_attributes: Reorder attributes to be in alphabetical order.
Defaults to ``False``.
r z2__init__() got an unexpected keyword argument '%s'r= FN) frozensetoptionslen TypeErrornextiterr> setattrr$ getattrerrorsstrict)selfkwargsZunexpected_argsattrr/ r/ r0 __init__ s O
zHTMLSerializer.__init__c C s* t |tst| jr"|| jdS |S d S )Nr2 r r AssertionErrorr7 encoderT stringr/ r/ r0 rZ s zHTMLSerializer.encodec C s* t |tst| jr"|| jdS |S d S )NrS rX r[ r/ r/ r0 encodeStrict s zHTMLSerializer.encodeStrictNc c s || _ d}g | _|r0| jr0ddlm} |||}| jrJddlm} ||}| jrdddlm} ||}| j r~ddl
m} ||}| j rddl m} ||}|D ]>}|d }|dkr\d|d }|d r|d |d 7 }n|d
r|d 7 }|d
rF|d
d d
kr,|d
dd
kr&| d d}nd }|d||d
|f 7 }|d7 }| |V q|dkr|dksv|r|r|d
dd
kr| d | |d V n| t|d V q|dkr|d } | d| V | tkr| jsd}n|r | d |d D ]\\}
} } | }
| }| dV | |
V | jrv|
t| t kr|
td t kr| d V | jd kst|d
krd}n@| jd krt |d k }n$| jd krt |d k }nt d!| d"d#}| j r| d$d%}|r| j!}| j"rHd|kr0d |kr0d }nd |krHd|krHd}|dkr`| dd&}n | d d'}| |V | |V | |V n | |V q| t#kr| j$r| j%r| d(V n | d)V | dV q|d*kr(|d } | tkrd}n|r| d | d+| V q|d,krj|d }|
d-d
krT| d. | d/|d V q|d0kr|d } | d1 }|t&kr| d2| | j'r|t(krt&| }nd3| }| |V q| |d qd S )4NFr )FiltertypeZDoctypez )Z
CharactersSpaceCharactersrc datazCommentz--zComment contains --z ZEntityr zEntity %s not recognizedz&%s;))r7 rR rG Zfilters.inject_meta_charsetr^ rF Z filters.alphabeticalattributesrH Zfilters.whitespacerI Zfilters.sanitizerr? Zfilters.optionaltagsfindserializeErrorr] rZ r r rD itemsr@ r
r$ 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_attrrd keyr/ r/ r0 r9 s
zHTMLSerializer.serializec C s2 |rd t| ||S d t| |S dS )an Serializes the stream from the treewalker into a string
:arg treewalker: the treewalker to serialize
:arg encoding: the string encoding to use
:returns: the serialized tree
Example:
>>> from html5lib import parse, getTreeWalker
>>> from html5lib.serializer import HTMLSerializer
>>> token_stream = parse('Hi!')
>>> walker = getTreeWalker('etree')
>>> serializer = HTMLSerializer(omit_optional_tags=False)
>>> serializer.render(walker(token_stream))
'Hi!'
r N)r' listr9 )rT ru r7 r/ r/ r0 r5 w s zHTMLSerializer.renderXXX ERROR MESSAGE NEEDEDc C s | j | | jrtd S )N)rR r" rS SerializeError)rT rd r/ r/ r0 rm s z HTMLSerializer.serializeError)N)N)r} )__name__
__module__ __qualname__r< r= r> r? r@ rA rB rC rD rE rF rG rH rI rK rW rZ r] r9 r5 rm r/ r/ r/ r0 r4 h s, Y
r4 c @ s e Zd ZdZdS )r~ zError in serialized treeN)r r r __doc__r/ r/ r/ r0 r~ s r~ )r3 N)+Z
__future__r r r Zpip._vendor.sixr recodecsr r Z constantsr r
r r r
r r r r Zxml.sax.saxutilsr r' Z_quoteAttributeSpecCharscompilerp rr r# rL Z_is_ucs4r| rn rx ry r r! islowerr1 r9 r r4 Exceptionr~ r/ r/ r/ r0 sD
1