o
V=ñ^ƒ= ã @ sv 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 dƒdkZ e e ¡ ƒD ]5\Z Z!e rqe e!ƒdksye sze e!ƒdkrzqee!dkrše e!ƒdkrŠe "e!¡Z!ne#e!ƒZ!e!evs–e $¡ rše ee!< qedd„ Z%ede%ƒ d dd„Z&G dd„ de'ƒZ(G dd„ de)ƒZ*dS ) é )Úabsolute_importÚdivisionÚunicode_literals)Ú text_typeN)Úregister_errorÚxmlcharrefreplace_errorsé )Ú voidElementsÚbooleanAttributesÚspaceCharacters)ÚrcdataElementsÚentitiesÚ xmlEntities)Ú treewalkersÚ_utils)ÚescapeÚ z"'=<>`ú[ú]u_
/`Â áš€á Žá  â€â€‚        

 âŸã€€]u ô¿¿é ú&c
C s t | ttfƒr†g }g }d}t| j| j| j… ƒD ]7\}}|r!d}q|| j }t | j|t | j|d gƒ… ¡rFt
| j||d … ¡}d}nt |ƒ}| |¡ q|D ]+}t
|¡} | rp| d¡ | | ¡ | d¡so| d¡ qR| dt|ƒdd … ¡ qRd |¡| jfS t| ƒS )NFr Tr ú;z&#x%s;r )Ú
isinstanceÚUnicodeEncodeErrorÚUnicodeTranslateErrorÚ enumerateÚobjectÚstartÚendr ZisSurrogatePairÚminÚsurrogatePairToCodepointÚordÚappendÚ_encode_entity_mapÚgetÚendswithÚhexÚjoinr )
ÚexcÚresZ
codepointsÚskipÚiÚcÚindexZ codepointÚcpÚe© r0 ú5/usr/lib/python3/dist-packages/html5lib/serializer.pyÚhtmlentityreplace_errors* s2
"
€ r2 ÚhtmlentityreplaceÚetreec K s( t |¡}tdi |¤Ž}| || ƒ|¡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('Hi!
')
>>> serialize(token_stream, omit_optional_tags=False)
'Hi!
'
Nr0 )r Z
getTreeWalkerÚHTMLSerializerÚrender)ÚinputZtreeÚencodingZserializer_optsZwalkerÚsr0 r0 r1 Ú serializeK s
r: c @ s~ e Zd ZdZdZdZdZdZdZdZ dZ
dZ dZ dZ
dZdZdZdZdd„ Zdd „ Zd
d „ Zdd
d„Zddd„Zddd„Zd S )r5 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 |ƒt | jƒ }t|ƒdkrtdtt|ƒƒ ƒ‚d|v r d| _| jD ]}t| || |t | |ƒ¡ƒ q#g | _
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) Ú frozensetÚoptionsÚlenÚ TypeErrorÚnextÚiterr? Úsetattrr$ ÚgetattrÚerrorsÚstrict)ÚselfÚkwargsZunexpected_argsÚattrr0 r0 r1 Ú__init__‡ s O
zHTMLSerializer.__init__c C ó& t |tƒsJ ‚| jr| | jd¡S |S )Nr3 ©r r r8 Úencode©rU Ústringr0 r0 r1 r[ à ó zHTMLSerializer.encodec C rY )NrT rZ r\ r0 r0 r1 Ú encodeStrictç r^ zHTMLSerializer.encodeStrictNc c sÌ || _ d}g | _|r| jrddlm} |||ƒ}| jr&ddlm} ||ƒ}| jr3ddlm} ||ƒ}| j r@ddl
m} ||ƒ}| j rMddl m} ||ƒ}|D ]}|d }|dkr«d|d }|d rm|d |d 7 }n|d
ru|d 7 }|d
r |d
d ¡d
kr“|d
d¡d
kr| d¡ d}nd }|d||d
|f 7 }|d7 }| |¡V qO|dv rÙ|dksµ|rÎ|rÅ|d
d¡d
krÅ| d¡ | |d ¡V qO| t|d ƒ¡V qO|dv rç|d } | d| ¡V | tv rô| jsôd}n|rû| d¡ |d ¡ D ]Ä\\}
} } | }
| }| d¡V | |
¡V | jr/|
t | tƒ ¡vrÄ|
t d tƒ ¡vrÄ| d ¡V | jd ksBt|ƒd
krEd}n | jd krSt |¡d u}n| jd krat |¡d u}nt d!ƒ‚| d"d#¡}| j ru| d$d%¡}|r¾| j!}| j"r˜d|v rŒd |vrŒd }n d |v r˜d|vr˜d}|dkr¤| dd&¡}n| d d'¡}| |¡V | |¡V | |¡V q| |¡V q| t#v rà| j$rà| j%rÚ| d(¡V n| d)¡V | d¡V qO|d*kr |d } | tv rød}n|r | d¡ | d+| ¡V qO|d,kr*|d }|
d-¡d
kr | d.¡ | d/|d ¡V qO|d0kr\|d } | d1 }|t&vrC| d2| ¡ | j'rQ|t(vrQt&| }nd3| }| |¡V qO| |d ¡ qOd S )4NFr )ÚFilterÚtypeZDoctypez )Z
CharactersÚSpaceCharactersre ÚdatazÚCommentz--zComment contains --z ZEntityr zEntity %s not recognizedz&%s;))r8 rS rH Zfilters.inject_meta_charsetr` rG Z filters.alphabeticalattributesrI Zfilters.whitespacerJ Zfilters.sanitizerr@ Zfilters.optionaltagsÚfindÚserializeErrorr_ r[ r r rE ÚitemsrA r
r$ Útupler= rM Ú_quoteAttributeSpecÚsearchÚ_quoteAttributeLegacyÚ
ValueErrorÚreplacerD r> r? r rB rC r
rF r )rU Ú
treewalkerr8 Zin_cdatar` Útokenra Zdoctyper> rb Ú_Z attr_nameZ
attr_valueÚkÚvZ
quote_attrrf Úkeyr0 r0 r1 r: î sâ €
ÿþ
€
”zHTMLSerializer.serializec C s. |r
d t| ||¡ƒ¡S d t| |¡ƒ¡S )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!'
ó r )r' Úlistr: )rU rw r8 r0 r0 r1 r6 w s zHTMLSerializer.renderúXXX ERROR MESSAGE NEEDEDc C s | j |¡ | jr t‚d S ©N)rS r" rT ÚSerializeError)rU rf r0 r0 r1 ro s ÿz HTMLSerializer.serializeErrorr€ )r )Ú__name__Ú
__module__Ú __qualname__r= r> r? r@ rA rB rC rD rE rF rG rH rI rJ rL rX r[ r_ r: r6 ro r0 r0 r0 r1 r5 h s. Y
r5 c @ s e Zd ZdZdS )r zError in serialized treeN)r‚ rƒ r„ Ú__doc__r0 r0 r0 r1 r — s r )r4 N)+Z
__future__r r r Zsixr ÚreÚcodecsr r Z constantsr r
r r r
r r r r Zxml.sax.saxutilsr r' Z_quoteAttributeSpecCharsÚcompilerr rt r# rM Z_is_ucs4r~ rp rz r{ r r! Úislowerr2 r: r r5 Ú Exceptionr r0 r0 r0 r1 Ú sH
ÿ
ÿ
ÿ €
1