o
7]h= @ 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 Zed e d
Zed e d Zi Ze ddkZ e e D ]5\Z Z!e rqe e!dksye sze e!dkrzqee!dkre e!dkre"e!Z!ne#e!Z!e!evse $ re ee!< qedd 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 ]7\}}|r!d}q|| j }t| j|t | j|d g rFt
| j||d }d}nt |}| | q|D ]+}t
|} | rp| d | | | dso| d qR| dt|dd qRd|| jfS t| S )NFr Tr ;z&#x%s;r )
isinstanceUnicodeEncodeErrorUnicodeTranslateError enumerateobjectstartendr isSurrogatePairminsurrogatePairToCodepointordappend_encode_entity_mapgetendswithhexjoinr )
excres
codepointsskipicindex codepointcpe r3 A/usr/lib/python3/dist-packages/pip/_vendor/html5lib/serializer.pyhtmlentityreplace_errors* s2
"
r5 htmlentityreplaceetreec K s( t |}tdi |}||| |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('
'
Nr3 )r
getTreeWalkerHTMLSerializerrender)inputtreeencodingserializer_optswalkersr3 r3 r4 serializeK s
rA c @ s~ e Zd ZdZdZdZdZdZdZdZ dZ
dZ dZ dZ
dZdZdZdZdd Zdd Zd
d Zdd
dZdddZdddZd S )r9 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|dkrtdtt| d|v r d| _| jD ]}t| |||t | | 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'rE FN) frozensetoptionslen TypeErrornextiterrF setattrr% getattrerrorsstrict)selfkwargsunexpected_argsattrr3 r3 r4 __init__ s O
zHTMLSerializer.__init__c C & t |tsJ | jr|| jdS |S )Nr6 r r r= encoder\ stringr3 r3 r4 rc zHTMLSerializer.encodec C ra )Nr[ rb rd r3 r3 r4 encodeStrict rf zHTMLSerializer.encodeStrictNc c s || _ d}g | _|r| jrddlm} |||}| jr&ddlm} ||}| jr3ddlm} ||}| j r@ddl
m} ||}| j rMddl m} ||}|D ]}|d }|dkrd|d }|d rm|d |d 7 }n|d
ru|d 7 }|d
r|d
d d
kr|d
dd
kr| d d}nd }|d||d
|f 7 }|d7 }| |V qO|dv r|dks|r|r|d
dd
kr| d | |d V qO| t|d V qO|dv r|d } | d| V | tv r| jsd}n|r| d |d D ]\\}
} } | }
| }| dV | |
V | jr/|
t| t vr|
td t vr| d V | jd ksBt|d
krEd}n | jd krSt |d u}n| jd krat |d u}nt d!| d"d#}| j ru| d$d%}|r| j!}| j"rd|v rd |vrd }n d |v rd|vrd}|dkr| dd&}n| d d'}| |V | |V | |V q| |V q| t#v r| j$r| j%r| d(V n| d)V | dV qO|d*kr |d } | tv rd}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 )FiltertypeDoctypez )
CharactersSpaceCharactersrq datazCommentz--zComment contains --z Entityr zEntity %s not recognizedz&%s;))r= rZ rO filters.inject_meta_charsetrh rN filters.alphabeticalattributesrP filters.whitespacerQ filters.sanitizerrG filters.optionaltagsfindserializeErrorrg rc r r rL itemsrH r
r% tuplerD rT _quoteAttributeSpecsearch_quoteAttributeLegacy
ValueErrorreplacerK rE rF r rI rJ r
rM r )r\
treewalkerr= in_cdatarh tokenri doctyperE rk _ attr_name
attr_valuekv
quote_attrrr keyr3 r3 r4 rA s
zHTMLSerializer.serializec C s. |r
d t| ||S d t| |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!'
r )r( listrA )r\ r r= r3 r3 r4 r: w s zHTMLSerializer.renderXXX ERROR MESSAGE NEEDEDc C s | j | | jr td S N)rZ r# r[ SerializeError)r\ rr r3 r3 r4 r s z HTMLSerializer.serializeErrorr )r )__name__
__module__ __qualname__rD rE rF rG rH rI rJ rK rL rM rN rO rP rQ rS r` rc rg 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 __doc__r3 r3 r3 r4 r s r )r7 N)+
__future__r r r pip._vendor.sixr recodecsr r constantsr r
r r r
r r r r xml.sax.saxutilsr r( _quoteAttributeSpecCharscompiler r r$ rT _is_ucs4r r r r r! r" islowerr5 rA r r9 Exceptionr r3 r3 r3 r4 sF
1