'
)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
d„Zddd„Zddd„Zd 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 sz t |ƒt | jƒ }t|ƒdkr2tdtt|ƒƒ ƒ‚d|kr@d| _x(| jD ] }t| || |t | |ƒ¡ƒ 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) Ú frozensetÚoptionsÚlenÚ TypeErrorÚnextÚiterrF Úsetattrr% ÚgetattrÚerrorsÚstrict)ÚselfÚkwargsÚunexpected_argsÚattrr3 r3 r4 Ú__init__‡ s O zHTMLSerializer.__init__c C s* t |tƒst‚| jr"| | jd¡S |S d S )Nr6 )r r ÚAssertionErrorr= Úencode)r\ Ústringr3 r3 r4 rb à s zHTMLSerializer.encodec C s* t |tƒst‚| jr"| | jd¡S |S d S )Nr[ )r r ra r= rb )r\ rc r3 r3 r4 Ú encodeStrictç s zHTMLSerializer.encodeStrictNc c sî || _ d}g | _|r0| jr0ddlm} |||ƒ}| jrJddlm} ||ƒ}| jrdddlm} ||ƒ}| j r~ddl
m} ||ƒ}| j r˜ddl m} ||ƒ}xN|D ]D}|d }|dkr`d|d }|d rÞ|d |d 7 }n|d
rî|d 7 }|d
rJ|d
d ¡d
kr0|d
d¡d
kr*| d¡ d}nd }|d||d
|f 7 }|d7 }| |¡V q |dkrÆ|dksz|r°|rž|d
d¡d
krž| d¡ | |d ¡V n| t|d ƒ¡V q |dkrî|d } | d| ¡V | tkr | js d}n|r| d¡ x˜|d ¡ D ]†\\}
} } | }
| }| d¡V | |
¡V | jr~|
t | tƒ ¡kr |
t d tƒ ¡kr | d ¡V | jd ks¤t|ƒd
krªd}n@| jd krÆt |¡d k }n$| jd krât |¡d k }nt d!ƒ‚| d"d#¡}| j r
| d$d%¡}|rš| j!}| j"rPd|kr8d |kr8d }nd |krPd|krPd}|dkrh| dd&¡}n | d d'¡}| |¡V | |¡V | |¡V n | |¡V q W | t#krà| j$rà| j%rÔ| d(¡V n | d)¡V | d¡V q |d*kr2|d } | tkrd}n|r | d¡ | d+| ¡V q |d,krt|d }|
d-¡d
kr^| d.¡ | d/|d ¡V q |d0krØ|d } | d1 }|t&kr¦| d2| ¡ | j'rÂ|t(krÂt&| }nd3| }| |¡V q | |d ¡ q W d S )4NFr )ÚFilterÚtypeÚDoctypez )Ú
CharactersÚSpaceCharactersrn ÚdatazÚCommentz--zComment contains --z ÚEntityr zEntity %s not recognizedz&%s;))r= rZ rO Úfilters.inject_meta_charsetre rN Ú filters.alphabeticalattributesrP Úfilters.whitespacerQ Úfilters.sanitizerrG Úfilters.optionaltagsÚfindÚserializeErrorrd rb r r rL ÚitemsrH r
r% ÚtuplerD rT Ú_quoteAttributeSpecÚsearchÚ_quoteAttributeLegacyÚ
ValueErrorÚreplacerK rE rF r rI rJ r
rM r )r\ Ú
treewalkerr= Úin_cdatare Útokenrf ÚdoctyperE rh Ú_Ú attr_nameÚ
attr_valueÚkÚvÚ
quote_attrro Úkeyr3 r3 r4 rA î sØ
zHTMLSerializer.serializec C s2 |rd t| ||¡ƒ¡S d t| |¡ƒ¡S dS )an Serializes the stream from the treewalker into a string
:arg treewalker: the treewalker to serialize
:arg encoding: the string encoding to use
:returns: the serialized tree
Example:
>>> from html5lib import parse, getTreeWalker
>>> from html5lib.serializer import HTMLSerializer
>>> token_stream = parse('Hi!')
>>> walker = getTreeWalker('etree')
>>> serializer = HTMLSerializer(omit_optional_tags=False)
>>> serializer.render(walker(token_stream))
'Hi!'
ó r N)r( ÚlistrA )r\ r‰ r= r3 r3 r4 r: w s zHTMLSerializer.renderúXXX ERROR MESSAGE NEEDEDc C s | j |¡ | jrt‚d S )N)rZ r# r[ ÚSerializeError)r\ ro r3 r3 r4 r s z HTMLSerializer.serializeError)N)N)r– )Ú__name__Ú
__module__Ú __qualname__rD rE rF rG rH rI rJ rK rL rM rN rO rP rQ rS r` rb rd rA r: r r3 r3 r3 r4 r9 h s, Y
r9 c @ s e Zd ZdZdS )r— zError in serialized treeN)r˜ r™ rš Ú__doc__r3 r3 r3 r4 r— — s r— )r7 N)+Ú
__future__r r r Úpip._vendor.sixr ÚreÚcodecsr r Ú constantsr r
r r r
r r r r Úxml.sax.saxutilsr r( Ú_quoteAttributeSpecCharsÚcompiler„ r† r$ rT Ú_is_ucs4r• r‚ r r‘ r! r" Úislowerr5 rA r r9 Ú Exceptionr— r3 r3 r3 r4 Ú s<
1