3
Ug= @ 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 isSurrogatePairminsurrogatePairToCodepointordappend_encode_entity_mapgetendswithhexjoinr )
excres
codepointsskipicindex codepointcpe r3 C/tmp/pip-unpacked-wheel-0ht26j5g/pip/_vendor/html5lib/serializer.pyhtmlentityreplace_errors* s0
"
r5 htmlentityreplaceetreec K s$ t j|}tf |}|j|| |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
getTreeWalkerHTMLSerializerrender)inputtreeencodingserializer_optswalkersr3 r3 r4 serializeK s
rA c @ s~ e Zd ZdZdZdZdZdZdZdZ dZ
dZ dZ dZ
dZdZdZd!Zdd Zdd Zdd Zd"ddZd#d d Zd$d d ZdS )%r9 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 )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 s* t |tst| jr"|j| jdS |S d S )Nr6 )r r AssertionErrorr= encode)r\ stringr3 r3 r4 rb s zHTMLSerializer.encodec C s* t |tst| jr"|j| jdS |S d S )Nr[ )r r ra r= rb )r\ rc r3 r3 r4 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$}| j r|j d%d&}|r| j!}| j"rTd|kr
CharactersSpaceCharactersdatazCommentz--zComment contains --z Entityr zEntity %s not recognizedz&%s;)rm rn )rp rq ))r= rZ rO filters.inject_meta_charsetre rN filters.alphabeticalattributesrP filters.whitespacerQ filters.sanitizerrG filters.optionaltagsfindserializeErrorrd rb r r rL itemsrH r
r% tuplerD rT _quoteAttributeSpecsearch_quoteAttributeLegacy
ValueErrorreplacerK rE rF r rI rJ r
rM r )r\
treewalkerr= in_cdatare tokenrf doctyperE rh _ attr_name
attr_valuekv
quote_attrro keyr3 r3 r4 rA s
zHTMLSerializer.serializec C s2 |rdj t| j||S dj t| j|S dS )an Serializes the stream from the treewalker into a string
:arg treewalker: the treewalker to serialize
:arg encoding: the string encoding to use
:returns: the serialized tree
Example:
>>> from html5lib import parse, getTreeWalker
>>> from html5lib.serializer import HTMLSerializer
>>> token_stream = parse('Hi!')
>>> walker = getTreeWalker('etree')
>>> serializer = HTMLSerializer(omit_optional_tags=False)
>>> serializer.render(walker(token_stream))
'Hi!'
r N)r( listrA )r\ r r= r3 r3 r4 r: w s zHTMLSerializer.renderXXX ERROR MESSAGE NEEDEDc C s | j j| | jrtd S )N)rZ r# r[ SerializeError)r\ ro r3 r3 r4 r s z HTMLSerializer.serializeError)rD rE rF rG rH rI rJ rK rL rM rN rO rP rQ )N)N)r )__name__
__module__ __qualname__rD rE rF rG rH rI rJ rK rL rM rN rO rP rQ rS r` rb rd rA r: r r3 r3 r3 r4 r9 h s6 Y
r9 c @ s e Zd ZdZdS )r zError in serialized treeN)r r r __doc__r3 r3 r3 r4 r s r )r7 N)+
__future__r r r pip._vendor.sixr recodecsr r constantsr r
r r r
r r r r xml.sax.saxutilsr r( _quoteAttributeSpecCharscompiler r r$ rT _is_ucs4r r r r r! r" islowerr5 rA r r9 Exceptionr r3 r3 r3 r4 s<
1