Your IP : 18.116.20.205


Current Path : /usr/lib/python2.7/site-packages/pip/_vendor/html5lib/
Upload File :
Current File : //usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyc

�
��abc!@`sddlmZmZmZddlmZmZddlmZm	Z	ddl
Z
ddlZddlm
Z
ddlmZmZmZmZddlmZdd	lmZdd
lmZyddlmZWnek
r�eZnXegeD]Zejd�^q��ZegeD]Zejd�^q"�ZegeD]Zejd�^qJ�Zeed
dg�BZ dZ!ej"r�e!ddkr�e!j#d�dks�t$�ej%e!d e&d�d�Z'nej%e!�Z'e(dddddddddddddd d!d"d#d$d%d&d'd(d)d*d+d,d-d.d/d0d1d2g �Z)ej%d3�Z*iZ+d4e,fd5��YZ-d6�Z.d7e,fd8��YZ/d9e/fd:��YZ0d;e1fd<��YZ2d=e,fd>��YZ3d?e,fd@��YZ4dA�Z5dS(Bi(tabsolute_importtdivisiontunicode_literals(t	text_typetbinary_type(thttp_clientturllibN(twebencodingsi(tEOFtspaceCharacterstasciiLetterstasciiUppercase(tReparseException(t_utils(tStringIO(tBytesIOuasciit>t<u�[---Ÿ﷐-﷯￾￿🿾🿿𯿾𯿿𿿾𿿿񏿾񏿿񟿾񟿿񯿾񯿿񿿾񿿿򏿾򏿿򟿾򟿿򯿾򯿿򿿾򿿿󏿾󏿿󟿾󟿿󯿾󯿿󿿾󿿿􏿾􏿿]i����u]u"\uD800-\uDFFF"i��i��i��i��i��i��i��i��i��i��i��i��i��i��i��i��i��	i��	i��
i��
i��i��i��i��i��
i��
i��i��i��i��i��i��u[	-
 -/:-@[-`{-~]tBufferedStreamcB`sMeZdZd�Zd�Zd�Zd�Zd�Zd�Zd�Z	RS(u�Buffering for streams that do not have buffering of their own

    The buffer is implemented as a list of chunks on the assumption that
    joining many strings will be slow since it is O(n**2)
    cC`s%||_g|_ddg|_dS(Ni����i(tstreamtbuffertposition(tselfR((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt__init__@s		cC`sJd}x,|j|jd D]}|t|�7}qW||jd7}|S(Nii(RRtlen(Rtpostchunk((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyttellEs
cC`sx||j�kst�|}d}x>t|j|�|krd|t|j|�8}|d7}q'W||g|_dS(Nii(t_bufferedBytestAssertionErrorRRR(RRtoffsetti((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytseekLscC`sp|js|j|�S|jdt|j�kr_|jdt|jd�kr_|j|�S|j|�SdS(Niii����(Rt_readStreamRRt_readFromBuffer(Rtbytes((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytreadUs	
 
cC`s&tg|jD]}t|�^q
�S(N(tsumRR(Rtitem((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR^scC`sL|jj|�}|jj|�|jdcd7<t|�|jd<|S(Nii(RR$RtappendRR(RR#tdata((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR!as
cC`s"|}g}|jd}|jd}x�|t|j�kr�|dkr�|dks\t�|j|}|t|�|kr�|}|||g|_n/t|�|}|t|�g|_|d7}|j||||!�||8}d}q)W|r|j|j|��ndj|�S(Niit(RRRRR'R!tjoin(RR#tremainingBytestrvtbufferIndextbufferOffsettbufferedDatatbytesToRead((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR"hs&

$



(
t__name__t
__module__t__doc__RRR R$RR!R"(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR9s								cK`s�t|tj�s<t|tjj�rEt|jtj�rEt}n9t|d�rot|j	d�t
�}nt|t
�}|r�g|D]}|jd�r�|^q�}|r�td|��nt
||�St||�SdS(Nureadiu	_encodingu3Cannot set an encoding with a unicode input, set %r(t
isinstanceRtHTTPResponseRtresponsetaddbasetfptFalsethasattrR$Rtendswitht	TypeErrortHTMLUnicodeInputStreamtHTMLBinaryInputStream(tsourcetkwargst	isUnicodetxt	encodings((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytHTMLInputStream�s	(
R=cB`s}eZdZdZd�Zd�Zd�Zd�Zd�Zd�Z	d
d�Zd	�Zd
�Z
ed�Zd�ZRS(u�Provides a unicode stream of characters to the HTMLTokenizer.

    This class takes care of character encoding and removing or replacing
    incorrect byte-sequences and also provides column and line tracking.

    i(cC`s�tjsd|_n-td�dkr6|j|_n|j|_dg|_td�df|_	|j
|�|_|j�dS(u�Initialises the HTMLInputStream.

        HTMLInputStream(source, [encoding]) -> Normalized stream from source
        for use by html5lib.

        source can be either a file-object, local filename or a string.

        The optional encoding parameter must be a string that indicates
        the encoding.  If specified, that encoding will be used,
        regardless of any BOM or later declaration (such as in a meta
        element)

        u􏿿iiuutf-8ucertainN(
R
tsupports_lone_surrogatestNonetreportCharacterErrorsRtcharacterErrorsUCS4tcharacterErrorsUCS2tnewLinestlookupEncodingtcharEncodingt
openStreamt
dataStreamtreset(RR?((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�s	cC`sCd|_d|_d|_g|_d|_d|_d|_dS(Nui(Rt	chunkSizetchunkOffsetterrorstprevNumLinestprevNumColsRFt_bufferedCharacter(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRO�s						cC`s(t|d�r|}nt|�}|S(uvProduces a file object from source.

        source can be either a file object, local filename or a string.

        uread(R:R(RR?R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRM�s	cC`st|j}|jdd|�}|j|}|jdd|�}|dkr\|j|}n||d}||fS(Nu
ii����i(RtcountRStrfindRT(RRRtnLinestpositionLinetlastLinePostpositionColumn((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt	_position�s	
cC`s&|j|j�\}}|d|fS(u:Returns (line, col) of the current position in the stream.i(R\RQ(Rtlinetcol((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�scC`sL|j|jkr%|j�s%tSn|j}|j|}|d|_|S(uo Read one character from the stream or queue if available. Return
            EOF when EOF is reached.
        i(RQRPt	readChunkRR(RRQtchar((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR`�s	

cC`sO|dkr|j}n|j|j�\|_|_d|_d|_d|_|jj	|�}|j
r�|j
|}d|_
n
|s�tSt|�dkr�t
|d�}|dks�d|ko�dknr�|d|_
|d }q�n|jr|j|�n|jdd	�}|jd
d	�}||_t|�|_tS(Nuiii����i
i�i��u
u
u
(RFt_defaultChunkSizeR\RPRSRTRRQRNR$RUR9RtordRGtreplacetTrue(RRPR(tlastv((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR_�s0				
(
		cC`s:x3tttj|���D]}|jjd�qWdS(Nuinvalid-codepoint(trangeRtinvalid_unicode_retfindallRRR'(RR(t_((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRH%s"cC`st}x�tj|�D]�}|r(qnt|j��}|j�}tj|||d!�r�tj|||d!�}|t	kr�|j
jd�nt}q|dkr�|dkr�|t
|�dkr�|j
jd�qt}|j
jd�qWdS(Niuinvalid-codepointi�i��i(R9RgtfinditerRbtgrouptstartR
tisSurrogatePairtsurrogatePairToCodepointtnon_bmp_invalid_codepointsRRR'RdR(RR(tskiptmatcht	codepointRtchar_val((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRI)s 	c
C`s}yt||f}Wn�tk
r�x&|D]}t|�dks+t�q+Wdjg|D]}dt|�^qZ�}|s�d|}ntjd|�}t||f<nXg}x�tri|j|j	|j
�}|dkr�|j
|jkr?Pq?nB|j
�}||jkr?|j|j	|j
|!�||_
Pn|j|j	|j
�|j�s�Pq�q�Wdj|�}	|	S(u� Returns a string of characters from the stream up to but not
        including any character in 'characters' or EOF. 'characters' must be
        a container that supports the 'in' method and iteration over its
        characters.
        i�uu\x%02xu^%su[%s]+N(tcharsUntilRegExtKeyErrorRbRR*tretcompileRdRqRRQRFRPtendR'R_(
Rt
characterstoppositetcharstctregexR,tmRxtr((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt
charsUntil@s2

,
%		cC`so|dk	rk|jdkr=||j|_|jd7_qk|jd8_|j|j|kskt�ndS(Nii(RFRQRRPR(RR`((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytungetosN(R1R2R3RaRRORMR\RR`RFR_RHRIR9R�R�(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR=�s	 					&		/R>cB`sbeZdZd	d	d	d	ded�Zd�Zd�Zed�Zd�Z	d�Z
d�ZRS(
u�Provides a unicode stream of characters to the HTMLTokenizer.

    This class takes care of character encoding and removing or replacing
    incorrect byte-sequences and also provides column and line tracking.

    uwindows-1252cC`s�|j|�|_tj||j�d|_d|_||_||_||_||_	||_
|j|�|_|jddk	s�t�|j�dS(u�Initialises the HTMLInputStream.

        HTMLInputStream(source, [encoding]) -> Normalized stream from source
        for use by html5lib.

        source can be either a file-object, local filename or a string.

        The optional encoding parameter must be a string that indicates
        the encoding.  If specified, that encoding will be used,
        regardless of any BOM or later declaration (such as in a meta
        element)

        iidiN(RMt	rawStreamR=RtnumBytesMetatnumBytesChardettoverride_encodingttransport_encodingtsame_origin_parent_encodingtlikely_encodingtdefault_encodingtdetermineEncodingRLRFRRO(RR?R�R�R�R�R�t
useChardet((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�s							cC`s3|jdjj|jd�|_tj|�dS(Niureplace(RLt
codec_infotstreamreaderR�RNR=RO(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRO�s"cC`sUt|d�r|}nt|�}y|j|j��Wnt|�}nX|S(uvProduces a file object from source.

        source can be either a file object, local filename or a string.

        uread(R:RR RR(RR?R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRM�s	cC`s!|j�df}|ddk	r&|St|j�df}|ddk	rO|St|j�df}|ddk	rx|S|j�df}|ddk	r�|St|j�df}|ddk	r�|djjd�r�|St|j	�df}|ddk	r|S|r�yddl
m}Wntk
r4q�Xg}|�}x[|j
s�|jj|j�}t|t�szt�|s�Pn|j|�|j|�qGW|j�t|jd�}|jjd�|dk	r�|dfSnt|j�df}|ddk	r|Std�dfS(Nucertainiu	tentativeuutf-16(tUniversalDetectoruencodinguwindows-1252(t	detectBOMRFRKR�R�tdetectEncodingMetaR�tnamet
startswithR�tchardet.universaldetectorR�tImportErrortdoneR�R$R�R4R#RR'tfeedtclosetresultR R�(RtchardetRLR�tbufferstdetectorRtencoding((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��sR'
	


cC`s�|jddkst�t|�}|dkr5dS|jdkretd�}|dk	s�t�nr||jdkr�|jddf|_nF|jjd�|df|_|j�td|jd|f��dS(	Niucertainuutf-16beuutf-16leuutf-8iuEncoding changed from %s to %s(uutf-16beuutf-16le(	RLRRKRFR�R�R ROR(RtnewEncoding((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytchangeEncodings
cC`s�idtj6dtj6dtj6dtj6dtj6}|jjd�}t|t	�s_t
�|j|d �}d}|s�|j|�}d}|s�|j|d �}d}q�n|r�|jj|�t
|�S|jjd	�d
Sd
S(u�Attempts to detect at BOM at the start of the stream. If
        an encoding can be determined from the BOM return the name of the
        encoding otherwise return Noneuutf-8uutf-16leuutf-16beuutf-32leuutf-32beiiiiN(tcodecstBOM_UTF8tBOM_UTF16_LEtBOM_UTF16_BEtBOM_UTF32_LEtBOM_UTF32_BER�R$R4R#RtgetR RKRF(RtbomDicttstringR�R ((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�s&

cC`s�|jj|j�}t|t�s*t�t|�}|jjd�|j�}|dk	r||j
dkr|td�}n|S(u9Report the encoding declared by the meta element
        iuutf-16beuutf-16leuutf-8N(uutf-16beuutf-16le(R�R$R�R4R#RtEncodingParserR tgetEncodingRFR�RK(RRtparserR�((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�9sN(R1R2R3RFRdRRORMR�R�R�R�(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR>�s(		>		"t
EncodingBytescB`s�eZdZd�Zd�Zd�Zd�Zd�Zd�Zd�Z	d�Z
ee
e	�Zd	�Z
ee
�Zed
�Zd�Zd�Zd
�ZRS(u�String-like object with an associated position and various extra methods
    If the position is ever greater than the string length then an exception is
    raisedcC`s+t|t�st�tj||j��S(N(R4R#Rt__new__tlower(Rtvalue((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�LscC`s
d|_dS(Ni����(R\(RR�((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRPscC`s|S(N((R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt__iter__TscC`sS|jd}|_|t|�kr/t�n|dkrDt�n|||d!S(Nii(R\Rt
StopIterationR<(Rtp((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt__next__Ws		cC`s
|j�S(N(R�(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytnext_scC`sY|j}|t|�kr$t�n|dkr9t�n|d|_}|||d!S(Nii(R\RR�R<(RR�((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytpreviouscs			cC`s+|jt|�krt�n||_dS(N(R\RR�(RR((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytsetPositionls	cC`s<|jt|�krt�n|jdkr4|jSdSdS(Ni(R\RR�RF(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytgetPositionqs
	cC`s||j|jd!S(Ni(R(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytgetCurrentByte{scC`sc|j}xJ|t|�krU|||d!}||krH||_|S|d7}qW||_dS(uSkip past a list of charactersiN(RRR\RF(RR{R�R|((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRp�s			cC`sc|j}xJ|t|�krU|||d!}||krH||_|S|d7}qW||_dS(Ni(RRR\RF(RR{R�R|((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt	skipUntil�s			cC`sQ|j}|||t|�!}|j|�}|rM|jt|�7_n|S(u�Look for a sequence of bytes at the start of a string. If the bytes
        are found return True and advance the position to the byte after the
        match. Otherwise return False and leave the position alone(RRR�(RR#R�R(R,((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt
matchBytes�s	cC`sh||jj|�}|dkr^|jdkr=d|_n|j|t|�d7_tSt�dS(u�Look for the next sequence of bytes matching a given sequence. If
        a match is found advance the position to the last byte of the matchi����iiN(RtfindR\RRdR�(RR#tnewPosition((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytjumpTo�s(R1R2R3R�RR�R�R�R�R�R�tpropertyRR�tcurrentBytetspaceCharactersBytesRpR�R�R�(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�Hs 												R�cB`s_eZdZd�Zd�Zd�Zd�Zd�Zd�Zd�Z	d�Z
d	�ZRS(
u?Mini parser for detecting character encoding from meta elementscC`st|�|_d|_dS(u3string - the data to work on for encoding detectionN(R�R(RFR�(RR(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�scC`s�d|jfd|jfd|jfd|jfd|jfd|jff}xv|jD]k}t}xR|D]J\}}|jj|�rky|�}PWq�tk
r�t	}Pq�XqkqkW|sXPqXqXW|j
S(Ns<!--s<metas</s<!s<?R(t
handleCommentt
handleMetathandlePossibleEndTagthandleOtherthandlePossibleStartTagR(RdR�R�R9R�(RtmethodDispatchRitkeepParsingtkeytmethod((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��s&	
cC`s|jjd�S(uSkip over commentss-->(R(R�(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��scC`sK|jjtkrtSt}d}x"trF|j�}|dkrGtS|ddkr�|ddk}|rC|dk	rC||_tSq%|ddkr�|d}t|�}|dk	rC||_tSq%|ddkr%t	t
|d��}|j�}|dk	rCt|�}|dk	r@|r4||_tS|}q@qCq%q%WdS(Nis
http-equiviscontent-typetcharsettcontent(R(R�R�RdR9RFtgetAttributeR�RKtContentAttrParserR�tparse(Rt	hasPragmatpendingEncodingtattrttentativeEncodingtcodect
contentParser((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��s:		
		cC`s
|jt�S(N(thandlePossibleTagR9(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��scC`st|j�|jt�S(N(R�R(R�Rd(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��s
cC`s�|j}|jtkr9|r5|j�|j�ntS|jt�}|dkra|j�n+|j�}x|dk	r�|j�}qpWtS(NR(
R(R�tasciiLettersBytesR�R�RdR�tspacesAngleBracketsR�RF(RtendTagR(R|R�((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��s	


cC`s|jjd�S(NR(R(R�(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�scC`s�|j}|jttdg�B�}|dksIt|�dksIt�|d	krYdSg}g}x�tr
|dkr�|r�Pnz|tkr�|j�}Pn^|d
kr�dj|�dfS|t	kr�|j
|j��n|dkr�dS|j
|�t|�}qhW|dkr7|j
�dj|�dfSt|�|j�}|dkr�|}x�tr�t|�}||kr�t|�dj|�dj|�fS|t	kr�|j
|j��qb|j
|�qbWn^|dkr�dj|�dfS|t	kr|j
|j��n|dkr-dS|j
|�x}tr�t|�}|tkrwdj|�dj|�fS|t	kr�|j
|j��q=|dkr�dS|j
|�q=WdS(u_Return a name,value pair for the next attribute in the stream,
        if one is found, or Nonet/iRt=R)t't"N(RN(R�R(R�R�(R(RpR�t	frozensetRFRRRdR*tasciiUppercaseBytesR'R�R�R�R�(RR(R|tattrNamet	attrValuet	quoteChar((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�sh	$	


	

	(R1R2R3RR�R�R�R�R�R�R�R�(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��s				$				R�cB`seZd�Zd�ZRS(cC`s"t|t�st�||_dS(N(R4R#RR((RR(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRfscC`s:y!|jjd�|jjd7_|jj�|jjdksHdS|jjd7_|jj�|jjdkr�|jj}|jjd7_|jj}|jj|�r�|j||jj!SdSnP|jj}y(|jjt�|j||jj!SWntk
r|j|SXWntk
r5dSXdS(NR�iR�R�R�(R�R�(	R(R�RRpR�RFR�R�R�(Rt	quoteMarktoldPosition((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�js.



(R1R2RR�(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�es	cC`swt|t�r:y|jd�}Wq:tk
r6dSXn|dk	roytj|�SWqstk
rkdSXndSdS(u{Return the python codec name corresponding to an encoding or None if the
    string doesn't correspond to a valid encoding.uasciiN(R4RtdecodetUnicodeDecodeErrorRFRtlookuptAttributeError(R�((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRK�s

(6t
__future__RRRtpip._vendor.sixRRtpip._vendor.six.movesRRR�Rvtpip._vendorRt	constantsRR	R
RRR)R
tioRRR�R�R&tencodeR�R�R�R�tinvalid_unicode_no_surrogateRERVRRwtevalRgtsetRotascii_punctuation_reRttobjectRRDR=R>R#R�R�R�RK(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt<module>sR"

(((	+
J	��h�'

?>