�
z fc @ s} d Z d d l Z e j d � j Z e j d � j Z e j d � Z e j d � Z e j d � Z [ d d
d � � YZ d S( s� Shared support for scanning document type declarations in HTML and XHTML.
This module is used as a foundation for the HTMLParser and sgmllib
modules (indirectly, for htmllib as well). It has no documented
public API and should not be used directly.
i����Ns [a-zA-Z][-_.a-zA-Z0-9]*\s*s (\'[^\']*\'|"[^"]*")\s*s --\s*>s ]\s*]\s*>s ]\s*>t
ParserBasec B s� e Z d Z d � Z d � Z d � Z d � Z d � Z d Z d � Z d d � Z
d d
� Z d � Z d � Z
d
� Z d � Z d � Z d � Z d � Z RS( si Parser base class which provides some common support methods used
by the SGML/HTML and XHTML parsers.c C s" | j t k r t d � � n d S( Ns( markupbase.ParserBase must be subclassed( t __class__R t RuntimeError( t self( ( s" /usr/lib64/python2.7/markupbase.pyt __init__ s c C s t d � � d S( Ns. subclasses of ParserBase must override error()( t NotImplementedError( R t message( ( s" /usr/lib64/python2.7/markupbase.pyt error! s c C s d | _ d | _ d S( Ni i ( t linenot offset( R ( ( s" /usr/lib64/python2.7/markupbase.pyt reset% s c C s | j | j f S( s& Return current line number and offset.( R R ( R ( ( s" /usr/lib64/python2.7/markupbase.pyt getpos) s c C s� | | k r | S| j } | j d | | � } | rm | j | | _ | j d | | � } | | d | _ n | j | | | _ | S( Ns
i ( t rawdatat countR t rindexR ( R t it jR t nlinest pos( ( s" /usr/lib64/python2.7/markupbase.pyt updatepos1 s t c
C se | j } | d } | | | !d k s2 t d � � | | | d !d k rQ | d S| | | d !d k rl d St | � } | | | d !d k r� | j | � S| | d
k r� | j | � S| j | | � \ } } | d k r� | S| d k r� d | _ n xe| | k r`| | } | d k r`| | d | !} | d k rK| j | � n
| j | � | d S| d
k r�t | | � } | s�d S| j
� } n� | d k r�| j | | � \ } } n� | | j k r�| d } nv | d
k r8| d k r| j | d | � } qM| d k r(| j d | � qM| j d � n | j d | | � | d k r� | Sq� Wd S( Ni s <!s$ unexpected call to parse_declarationi t >t -R i����s --t [i t doctypes "'t4 abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZt attlistt linktypet linkt elements& unsupported '[' char in %s declarations" unexpected '[' char in declarations! unexpected %r char in declaration( R R ( R R R R (
R t AssertionErrort lent
parse_commentt parse_marked_sectiont
_scan_namet _decl_othercharst handle_declt unknown_declt _declstringlit_matcht endt _parse_doctype_subsetR (
R R R R t nt decltypet ct datat mt name( ( s" /usr/lib64/python2.7/markupbase.pyt parse_declarationA sZ
i c C s | j } | | | d !d k s, t d � � | j | d | � \ } } | d k rX | S| d k r} t j | | d � } nA | d k r� t j | | d � } n | j d
| | d | !� | s� d S| r� | j d � } | j | | d | !� n | j d � S( Ni s <![s) unexpected call to parse_marked_section()i t tempt cdatat ignoret includet rcdatat ift elset endifs+ unknown status keyword %r in marked sectioni����( R0 R1 R2 R3 R4 ( R5 R6 R7 (
R R R" t _markedsectioncloset searcht _msmarkedsectioncloseR t startR% R' ( R R t reportR t sectNameR t match( ( s" /usr/lib64/python2.7/markupbase.pyR! � s # c C s� | j } | | | d !d k r0 | j d � n t j | | d � } | sP d S| r� | j d � } | j | | d | !� n | j d � S( Ni s <!--s" unexpected call to parse_comment()i����i ( R R t
_commentcloseR9 R; t handle_commentR' ( R R R< R R> R ( ( s" /usr/lib64/python2.7/markupbase.pyR � s c
C s� | j } t | � } | } x�| | k r�| | } | d k r�| | | d !} | d k ra d S| d k r� | j | | d � | j d | � n | d | k r� d S| d | k r� d S| | | d !d k r� | j | d d
�} | d
k r | Sq n | j | d | � \ } } | d k r+d S| d k r_| j | | d � | j d | � n t | d | � } | | | � } | d
k r�| Sq | d k r | d | k r�d S| j | d | � \ } } | d
k r�| S| | d k r�| d } q�q | d k r�| d } x* | | k rB| | j � rB| d } qW| | k r�| | d k rc| S| j | | � | j d � q�d Sq | j � r�| d } q | j | | � | j d | � q Wd S( Nt <i i����s <!i s* unexpected char in internal subset (in %r)i s <!--R< i R R t entityt notations) unknown declaration %r in internal subsett _parse_doctype_t %t ;t ]R s% unexpected char after internal subsets% unexpected char %r in internal subset( R R RB RC ( R R R R R R" t getattrt isspace(
R R t declstartposR R) R R+ t sR. t meth( ( s" /usr/lib64/python2.7/markupbase.pyR( � sn
c C sY | j | | � \ } } | d k r( d S| j } d | | k rU | j d | � d Sd S( Ni����R i ( R" R t find( R R RJ R. R R ( ( s" /usr/lib64/python2.7/markupbase.pyt _parse_doctype_element� s c C s | j } | j | | � \ } } | | | d !} | d k rB d S| d k rV | d Sx�| j | | � \ } } | d k r� | S| | | d !} | d k r� d S| d k rd | | k r� | j d | � d } n d Sx% | | | d !j � r| d } q� W| | s,d Sn | j | | � \ } } | | | d !} | sGd S| d k r�t | | � } | rw| j � } n d S| | | d !} | s�d Sn | d k r| | d k r�d S| j | d | � \ } } | d k r�| S| | | d !} | sd Sn | d k rY | d SqY Wd S(
Ni R i����R i t (t )s '"t #( R R" RM RI R& R' ( R R RJ R R. R R+ R- ( ( s" /usr/lib64/python2.7/markupbase.pyt _parse_doctype_attlist sZ
c C s� | j | | � \ } } | d k r( | S| j } x� | | | d !} | sO d S| d k rc | d S| d k r� t | | � } | s� d S| j � } q4 | j | | � \ } } | d k r4 | Sq4 Wd S( Ni i i����R s '"( R" R R& R' ( R R RJ R. R R R+ R- ( ( s" /usr/lib64/python2.7/markupbase.pyt _parse_doctype_notation> s$ c C s3 | j } | | | d !d k ri | d } xB | | | d !} | sH d S| j � ra | d } q- Pq- Wn | } | j | | � \ } } | d k r� | Sx� | j | | d !} | s� d S| d k r� t | | � } | r� | j � } q+d Sq� | d k r| d S| j | | � \ } } | d k r� | Sq� Wd S( Ni RE i����i s '"R ( R RI R" R& R' ( R R RJ R R R+ R. R- ( ( s" /usr/lib64/python2.7/markupbase.pyt _parse_doctype_entityU s8
c C s� | j } t | � } | | k r% d St | | � } | r� | j � } | j � } | t | � | k rl d S| j � | j � f S| j | | � | j d | | | d !� d S( Ni����s expected name token at %ri ( Ni����( Ni����(
R R t Nonet _declname_matcht groupt stript lowerR' R R ( R R RJ R R) R- RK R. ( ( s" /usr/lib64/python2.7/markupbase.pyR" y s c C s d S( N( ( R R, ( ( s" /usr/lib64/python2.7/markupbase.pyR% � s ( t __name__t
__module__t __doc__R R R
R R R# R/ R! R R( RN RR RS RT R" R% ( ( ( s" /usr/lib64/python2.7/markupbase.pyR s"
R C 9 $ ( (
R\ t ret compileR> RV R& R? R8 R: R ( ( ( s" /usr/lib64/python2.7/markupbase.pyt <module> s