U
iÛf9 ã @ sV d Z ddlZe d¡jZe d¡jZe d¡Ze d¡Ze d¡Z[G dd „ d ƒZ dS )
zÎShared support for scanning document type declarations in HTML and XHTML.
This module is used as a foundation for the html.parser module. It has no
documented public API and should not be used directly.
é Nz[a-zA-Z][-_.a-zA-Z0-9]*\s*z(\'[^\']*\'|"[^"]*")\s*z--\s*>z ]\s*]\s*>z]\s*>c @ s e Zd ZdZdd„ Zdd„ Zdd„ Zdd „ Zd
d „ Zd Z d
d„ Z
d#dd„Z d$dd„Z dd„ Z
dd„ Zdd„ Zdd„ Zd d „ Zd d „ Zd d!„ Zd"S )%Ú
ParserBaseziParser base class which provides some common support methods used
by the SGML/HTML and XHTML parsers.c C s | j tkrtdƒ‚d S )Nz)_markupbase.ParserBase must be subclassed)Ú __class__r Ú RuntimeError©Úself© r ú0/opt/alt/python38/lib64/python3.8/_markupbase.pyÚ__init__ s
ÿzParserBase.__init__c C s t dƒ‚d S )Nz.subclasses of ParserBase must override error())ÚNotImplementedError)r Úmessager r r Úerror s ÿzParserBase.errorc C s d| _ d| _d S )Né r ©ÚlinenoÚoffsetr r r r Úreset$ s zParserBase.resetc C s | j | jfS )z&Return current line number and offset.r r r r r Úgetpos( s zParserBase.getposc C sb ||kr |S | j }| d||¡}|rN| j| | _| d||¡}||d | _n| j| | | _|S )NÚ
r
)ÚrawdataÚcountr Úrindexr )r ÚiÚjr ZnlinesÚposr r r Ú updatepos0 s zParserBase.updateposÚ c
C sÞ | j }|d }|||… dks&tdƒ‚|||d … dkrB|d S |||d … dkrZdS t|ƒ}|||d … dkr€| |¡S || d kr–| |¡S | ||¡\}}|d
k r²|S |d krÀd | _||k rÚ|| }|dkr||d |… }|d kr| |¡ n
| |¡ |d S |d
kr)ú-r éÿÿÿÿz--ú[r Zdoctyper z"'Z4abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ> ÚattlistÚelementÚlinkÚlinktypez&unsupported '[' char in %s declarationz"unexpected '[' char in declarationz!unexpected %r char in declaration)
r ÚAssertionErrorÚlenÚ
parse_commentÚparse_marked_sectionÚ
_scan_nameÚ_decl_othercharsZ handle_declÚ unknown_declÚ_declstringlit_matchÚendÚ_parse_doctype_subsetr )
r r r r ÚnZdecltypeÚcÚdataÚmÚnamer r r Úparse_declaration@ s\
ÿz ParserBase.parse_declarationr
c C sÆ | j }|||d … dks"tdƒ‚| |d |¡\}}|dk rB|S |dkr\t ||d ¡}n4|dkrvt ||d ¡}n| d||d |… ¡ |s˜dS |r¼| d¡}| ||d |… ¡ | d¡S ) Né z ÚincludeÚcdataÚtempÚrcdataÚignore> ÚelseÚendifÚifz+unknown status keyword %r in marked sectionr )
r r& r* Ú_markedsectioncloseÚsearchÚ_msmarkedsectioncloser Ústartr, r. )r r Úreportr ZsectNamer Úmatchr r r r) ’ s
z ParserBase.parse_marked_sectionc C sj | j }|||d … dkr$| d¡ t ||d ¡}|s r" r# ÚentityÚnotationz)unknown declaration %r in internal subsetZ_parse_doctype_ú%ú;ú]r z%unexpected char after internal subsetz%unexpected char %r in internal subset)r r' r r r( r* ÚgetattrÚisspace)
r r Ú declstartposr r0 r r1 Úsr4 Zmethr r r r/ ¶ sp
ÿ
z ParserBase._parse_doctype_subsetc C sF | ||¡\}}|dkr dS | j}d||d … krB| d|¡d S dS )Nr r r
)r* r Úfind)r r rP r4 r r r r r Ú_parse_doctype_elementù s z!ParserBase._parse_doctype_elementc C s¶ | j }| ||¡\}}|||d … }|dkr2dS |dkrB|d S | ||¡\}}|dk r^|S |||d … }|dkrzdS |dkrØd||d … kr¤| d|¡d }ndS |||d … ¡ rÆ|d }q¨||d … sèdS n| ||¡\}}|||d … }|sdS |dkrDt||ƒ}|r&| ¡ }ndS |||d … }|sDdS |d kr ||d … d krddS | |d |¡\}}|dk r†|S |||d … }|s dS |dkrB|d S qBd S )
Nr
r r r r ú(ú)ú'"ú#)r r* rR rO r- r. )r r rP r r4 r r1 r3 r r r Ú_parse_doctype_attlist sX
z!ParserBase._parse_doctype_attlistc C s | ||¡\}}|dk r |S | j}|||d … }|s:dS |dkrJ|d S |dkrnt||ƒ}|sddS | ¡ }q"| ||¡\}}|dk r"|S q"d S )Nr r
r r rV )r* r r- r. )r r rP r4 r r r1 r3 r r r Ú_parse_doctype_notation= s"
z"ParserBase._parse_doctype_notationc C sä | j }|||d … dkrR|d }|||d … }|s:dS | ¡ rP|d }q"qVq"n|}| ||¡\}}|dk rr|S | j ||d … }|sŒdS |dkr²t||ƒ}|r¬| ¡ }qÞdS qr|dkrÂ|d S | ||¡\}}|dk rr|S qrd S )Nr
rK r r rV r )r rO r* r- r. )r r rP r r r1 r4 r3 r r r Ú_parse_doctype_entityT s4
z ParserBase._parse_doctype_entityc C s† | j }t|ƒ}||krdS t||ƒ}|r\| ¡ }| ¡ }|t|ƒ |krLdS | ¡ | ¡ fS | ||¡ | d|||d … ¡ d S )N)Nr zexpected name token at %ré ) r r' Ú_declname_matchÚgroupÚstripÚlowerr. r r )r r rP r r0 r3 rQ r4 r r r r* x s
ÿzParserBase._scan_namec C s d S )Nr )r r2 r r r r, Š s zParserBase.unknown_declN)r
)r
)Ú__name__Ú
__module__Ú __qualname__Ú__doc__r r r r r r+ r5 r) r( r/ rS rX rY rZ r* r, r r r r r s"
R
C 9$r )
rc ÚreÚcompilerD r\ r- rG r? rA r r r r r Ú s