U
|
H=®d= ã @ 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 Ze d e d ¡Ze d e d ¡Zi Ze ddkZ e e ¡ D ]n\Z Z!e râe e!dksÊe sôe e!dkrôqÊe!dkrÊe e!dkre "e!¡Z!ne#e!Z!e!eks0e $¡ rÊe ee!< qÊdd Z%ede% d ddZ&G dd de'Z(G dd de)Z*dS ) é )Úabsolute_importÚdivisionÚunicode_literals)Ú text_typeN)Úregister_errorÚxmlcharrefreplace_errorsé )Ú voidElementsÚbooleanAttributesÚspaceCharacters)ÚrcdataElementsÚentitiesÚ xmlEntities)Ú treewalkersÚ_utils)ÚescapeÚ z"'=<>`ú[ú]u_
|
/`Â áá á ââââââ
ââââââ¨â©â¯âã]u ô¿¿é ú&c
|
C s t | ttfrg }g }d}t| j| j| j
D ]n\}}|rDd}q2|| j }t | j|t | j|d g
¡rt
|
| j||d
¡}d}nt |}| |¡ q2|D ]V}t |¡} | râ| d¡ | | ¡ | d¡sü| d¡ q¦| dt|dd
¡ q¦d |¡| jfS t| S d S )NFr Tr ú;z&#x%s;r )Ú
|
isinstanceÚUnicodeEncodeErrorÚUnicodeTranslateErrorÚ enumerateÚobjectÚstartÚendr ÚisSurrogatePairÚminÚsurrogatePairToCodepointÚordÚappendÚ_encode_entity_mapÚgetÚendswithÚhexÚjoinr )
|
ÚexcÚresÚ
|
codepointsÚskipÚiÚcÚindexÚ codepointÚcpÚe© r3 úVD:\z\workplace\VsCode\pyvenv\venv\Lib\site-packages\pip/_vendor/html5lib/serializer.pyÚhtmlentityreplace_errors* s0
|
"
|
|
|
|
r5 ÚhtmlentityreplaceÚetreec K s$ 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('<html><body><p>Hi!</p></body></html>')
|
>>> serialize(token_stream, omit_optional_tags=False)
|
'<html><head></head><body><p>Hi!</p></body></html>'
|
|
)r Ú getTreeWalkerÚHTMLSerializerÚrender)ÚinputÚtreeÚencodingÚserializer_optsÚwalkerÚsr3 r3 r4 Ú serializeK s
|
|
rA c @ s~ e Zd ZdZdZdZdZdZdZdZ dZ
|
dZ dZ dZ dZdZdZdZdd Zdd Zd
|
d Zdd dZdddZdddZd S )r9 Úlegacyú"TF)Úquote_attr_valuesÚ
|
quote_charÚuse_best_quote_charÚomit_optional_tagsÚminimize_boolean_attributesÚuse_trailing_solidusÚ space_before_trailing_solidusÚescape_lt_in_attrsÚ escape_rcdataÚresolve_entitiesÚalphabetical_attributesÚinject_meta_charsetÚstrip_whitespaceÚsanitizec
|
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::
|
|
<input disabled="disabled">
|
|
becomes::
|
|
<input disabled>
|
|
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. ``<hr/>``.
|
|
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.
|
``<hr />``. 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'rE FN) Ú frozensetÚoptionsÚlenÚ TypeErrorÚnextÚiterrF Úsetattrr% ÚgetattrÚerrorsÚstrict)ÚselfÚkwargsÚunexpected_argsÚattrr3 r3 r4 Ú__init__ s O
|
zHTMLSerializer.__init__c C s* t |tst| jr"| | jd¡S |S d S )Nr6 ©r r ÚAssertionErrorr= Úencode©r\ Ústringr3 r3 r4 rc à s zHTMLSerializer.encodec C s* t |tst| jr"| | jd¡S |S d S )Nr[ ra rd r3 r3 r4 Ú encodeStrictç s zHTMLSerializer.encodeStrictNc c sâ || _ d}g | _|r0| jr0ddlm} |||}| jrJddlm} ||}| jrdddlm} ||}| j r~ddl
|
m} ||}| j rddl m} ||}|D ]>}|d }|dkr\d|d }|d rÚ|d |d 7 }n|d
|
rê|d 7 }|d
|
rF|d
|
d ¡d kr,|d
|
d¡d kr&| d¡ d}nd }|d||d
|
|f 7 }|d7 }| |¡V q|dkrÂ|dksv|r¬|r|d d¡d kr| d¡ | |d ¡V n| t|d ¡V q|dkrä|d } | d| ¡V | tkrü| jsüd}n|r | d¡ |d ¡ D ]\\}
|
} } | } | }| d¡V | | ¡V | jrv| t | t ¡kr| t d t ¡kr| d ¡V | jd kst|d kr¢d}n@| jd kr¾t |¡d k }n$| jd krÚt |¡d k }nt d!| d"d#¡}| j r| d$d%¡}|r| j!}| j"rHd|kr0d |kr0d }nd |krHd|krHd}|dkr`| dd&¡}n | d d'¡}| |¡V | |¡V | |¡V n | |¡V q| t#krÖ| j$rÖ| j%rÊ| d(¡V n | d)¡V | d¡V q|d*kr(|d } | tkrd}n|r| d¡ | d+| ¡V q|d,krj|d }| d-¡d krT| d.¡ | d/|d ¡V q|d0krÎ|d } | d1 }|t&kr| d2| ¡ | j'r¸|t(kr¸t&| }nd3| }| |¡V q| |d ¡ qd S )4NFr )ÚFilterÚtypeÚDoctypez <!DOCTYPE %sÚnameÚpublicIdz PUBLIC "%s"ÚsystemIdz SYSTEMrC r ú'zBSystem identifier contains both single and double quote charactersz %s%s%sú>)Ú
|
CharactersÚSpaceCharactersrp Údataz</zUnexpected </ in CDATA)ÚStartTagÚEmptyTagz<%sTz+Unexpected child element of a CDATA elementú r ú=ÚalwaysÚspecrB z?quote_attr_values must be one of: 'always', 'spec', or 'legacy'r z&ú<z<z'z"z /ú/ÚEndTagz</%s>ÚCommentz--zComment contains --z <!--%s-->ÚEntityr zEntity %s not recognizedz&%s;))r= rZ rO Úfilters.inject_meta_charsetrg rN Ú filters.alphabeticalattributesrP Úfilters.whitespacerQ Úfilters.sanitizerrG Úfilters.optionaltagsÚfindÚserializeErrorrf rc r r rL ÚitemsrH r
|
r% ÚtuplerD rT Ú_quoteAttributeSpecÚsearchÚ_quoteAttributeLegacyÚ
|
ValueErrorÚreplacerK rE rF r rI rJ r rM r )r\ Ú
|
treewalkerr= Úin_cdatarg Útokenrh ÚdoctyperE rj Ú_Ú attr_nameÚ
|
attr_valueÚkÚvÚ
|
quote_attrrq Úkeyr3 r3 r4 rA î sÜ
|
|
|
|
|
|
|
|
|
|
ÿþ
|
|
|
|
|
|
|
|
|
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('<html><body>Hi!</body></html>')
|
>>> walker = getTreeWalker('etree')
|
>>> serializer = HTMLSerializer(omit_optional_tags=False)
|
>>> serializer.render(walker(token_stream))
|
'<html><head></head><body>Hi!</body></html>'
|
|
ó r N)r( ÚlistrA )r\ r r= r3 r3 r4 r: w s zHTMLSerializer.renderúXXX ERROR MESSAGE NEEDEDc C s | j |¡ | jrtd S )N)rZ r# r[ ÚSerializeError)r\ rq r3 r3 r4 r s z HTMLSerializer.serializeError)N)N)r )Ú__name__Ú
|
__module__Ú __qualname__rD rE rF rG rH rI rJ rK rL rM rN rO rP rQ rS r` rc rf rA r: r r3 r3 r3 r4 r9 h s, Y
|
|
|
r9 c @ s e Zd ZdZdS )r zError in serialized treeN)r r r Ú__doc__r3 r3 r3 r4 r s r )r7 N)+Ú
|
__future__r r r Zpip._vendor.sixr ÚreÚcodecsr r Ú constantsr r
|
r r r r r r r Úxml.sax.saxutilsr r( Ú_quoteAttributeSpecCharsÚcompiler r r$ rT Ú_is_ucs4r r r r r! r" Úislowerr5 rA r r9 Ú Exceptionr r3 r3 r3 r4 Ú<module> sD
|
ÿ
|
ÿ
|
ÿ
|
|
|
1
|
