B
3UDB @ sV d Z ddlZedjZedjZedZedZedZ[G dd d Z dS )
zShared 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.
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#ddZ d$ddZ dd Z
dd Zdd Zdd Zd d Zd d Zd d! Zd"S )%
ParserBaseziParser base class which provides some common support methods used
by the SGML/HTML and XHTML parsers.c C s | j tkrtdd S )Nz)_markupbase.ParserBase must be subclassed) __class__r RuntimeError)self r 0/opt/alt/python37/lib64/python3.7/_markupbase.py__init__ s
zParserBase.__init__c C s t dd S )Nz.subclasses of ParserBase must override error())NotImplementedError)r messager r r error s zParserBase.errorc C s d| _ d| _d S )N r )linenooffset)r r r r reset$ s zParserBase.resetc C s | j | jfS )z&Return current line number and offset.)r
r )r r r r getpos( s zParserBase.getposc C sb ||kr |S | j }|d||}|rN| j| | _|d||}||d | _n| j| | | _|S )N
r )rawdatacountr
rindexr )r ijr Znlinesposr r r updatepos0 s zParserBase.updatepos c
C s | j }|d }||| dks&td|||d dkrB|d S |||d dkrZdS t|}|||d dkr| |S || d kr| |S | ||\}}|d
k r|S |d krd | _x ||k r|| }|dkr||d | }|d kr| | n
| | |d S |d
kr@t ||}|s6dS |
}n|dkr\| ||\} }nt|| jkrr|d }n^|d kr|d kr| |d |}n$|dkr| d| n
| d n| d|| |d
k r|S qW dS )N z)-r z--[r Zdoctyper z"'Z4abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ> linkattlistlinktypeelementz&unsupported '[' char in %s declarationz"unexpected '[' char in declarationz!unexpected %r char in declaration)
r AssertionErrorlen
parse_commentparse_marked_section
_scan_name_decl_othercharsZ handle_decl unknown_decl_declstringlit_matchend_parse_doctype_subsetr )
r r r r nZdecltypecdatamnamer r r parse_declaration@ sZ
z ParserBase.parse_declarationr c C s | j }|||d dks"td| |d |\}}|dk rB|S |dkr\t||d }n4|dkrvt||d }n| d||d | |sdS |r|d}| ||d | | dS ) N z ignorecdatatemprcdatainclude> elseifendifz+unknown status keyword %r in marked sectionr )
r r# r' _markedsectionclosesearch_msmarkedsectioncloser startr) r+ )r r reportr ZsectNamer matchr r r r& s
z ParserBase.parse_marked_sectionc C sj | j }|||d dkr$| d t||d }|s entitynotationr r" z)unknown declaration %r in internal subsetZ_parse_doctype_%;]r z%unexpected char after internal subsetz%unexpected char %r in internal subset)r r$ r r r% r' getattrisspace)
r r declstartposr r- r r. sr1 Zmethr r r r, sn
z ParserBase._parse_doctype_subsetc C sF | ||\}}|dkr dS | j}d||d krB|d|d S dS )Nr r r )r' r find)r r rL r1 r r r r r _parse_doctype_element s z!ParserBase._parse_doctype_elementc C s | j }| ||\}}|||d }|dkr2dS |dkrB|d S xv| ||\}}|dk rb|S |||d }|dkr~dS |dkrd||d kr|d|d }ndS x |||d r|d }qW ||d sdS n| ||\}}|||d }|s
dS |dkrLt||}|r.| }ndS |||d }|sLdS |d kr||d d krldS | |d |\}}|dk r|S |||d }|sdS |dkrF|d S qFW d S )
Nr r r r r ()z'"#)r r' rN rK r* r+ )r r rL r r1 r r. r0 r r r _parse_doctype_attlist sZ
z!ParserBase._parse_doctype_attlistc C s | ||\}}|dk r |S | j}xl|||d }|s r r r r r s