HOME

sh-3ll 1.0
DIR:/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/
Current File : //usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyo
�
��abc!@`s�ddlmZmZmZddlmZmZddlmZm	Z	ddl
Z
ddlZddlm
Z
ddlmZmZmZmZddlmZdd	lmZdd
lmZyddlmZWnek
r�eZnXegeD]Zejd�^q��ZegeD]Zejd�^q"�ZegeD]Zejd�^qJ�Zeed
dg�BZ dZ!ej"r�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:��YZ.d;e/fd<��YZ0d=e*fd>��YZ1d?e*fd@��YZ2dA�Z3dS(Bi(tabsolute_importtdivisiontunicode_literals(t	text_typetbinary_type(thttp_clientturllibN(twebencodingsi(tEOFtspaceCharacterstasciiLetterstasciiUppercase(tReparseException(t_utils(tStringIO(tBytesIOuasciit>t<u�[---﷐-﷯🿾🿿𯿾𯿿𿿾𿿿񏿾񏿿񟿾񟿿񯿾񯿿񿿾񿿿򏿾򏿿򟿾򟿿򯿾򯿿򿿾򿿿󏿾󏿿󟿾󟿿󯿾󯿿󿿾󿿿􏿾􏿿]i����u"\uD800-\uDFFF"u]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`s`|}d}x>t|j|�|krL|t|j|�8}|d7}qW||g|_dS(Nii(RRR(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.pyt_bufferedBytes^scC`sL|jj|�}|jj|�|jdcd7<t|�|jd<|S(Nii(RR"RtappendRR(RR!tdata((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRas
cC`s|}g}|jd}|jd}x�|t|j�kr�|dkr�|j|}|t|�|kr�|}|||g|_n/t|�|}|t|�g|_|d7}|j||||!�||8}d}q)W|r|j|j|��ndj|�S(Niit(RRRR&Rtjoin(RR!tremainingBytestrvtbufferIndextbufferOffsettbufferedDatatbytesToRead((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR hs$

$



(
t__name__t
__module__t__doc__RRRR"R%RR (((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�td|��nt
||�St||�SdS(Nureadiu	_encodingu3Cannot set an encoding with a unicode input, set %r(t
isinstanceRtHTTPResponseRtresponsetaddbasetfptFalsethasattrR"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	�Zd
�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_surrogatestNonetreportCharacterErrorsRtcharacterErrorsUCS4tcharacterErrorsUCS2tnewLinestlookupEncodingtcharEncodingt
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	chunkSizetchunkOffsetterrorstprevNumLinestprevNumColsREt_bufferedCharacter(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRN�s						cC`s(t|d�r|}nt|�}|S(uvProduces a file object from source.

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

        uread(R9R(RR>R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRL�s	cC`st|j}|jdd|�}|j|}|jdd|�}|dkr\|j|}n||d}||fS(Nu
ii����i(RtcountRRtrfindRS(RRRtnLinestpositionLinetlastLinePostpositionColumn((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[RP(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(RPROt	readChunkRR(RRPtchar((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
(REt_defaultChunkSizeR[RORRRSRRPRMR"RTR8RtordRFtreplacetTrue(RROR'tlastv((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR^�s0				
(
		cC`s:x3tttj|���D]}|jjd�qWdS(Nuinvalid-codepoint(trangeRtinvalid_unicode_retfindallRQR&(RR't_((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRG%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(R8RftfinditerRatgrouptstartR
tisSurrogatePairtsurrogatePairToCodepointtnon_bmp_invalid_codepointsRQR&RcR(RR'tskiptmatcht	codepointRtchar_val((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRH)s 	c
C`sTyt||f}Wnqtk
r�djg|D]}dt|�^q1�}|scd|}ntjd|�}t||f<nXg}x�tr@|j|j|j	�}|dkr�|j	|jkrPqnB|j�}||jkr|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.
        uu\x%02xu^%su[%s]+N(tcharsUntilRegExtKeyErrorR)RatretcompileRcRpRRPREROtendR&R^(
Rt
characterstoppositetcharstctregexR+tmRwtr((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt
charsUntil@s.
,
%		cC`sS|dk	rO|jdkr=||j|_|jd7_qO|jd8_ndS(Nii(RERPRRO(RR_((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytungetosN(R0R1R2R`RRNRLR[RR_RER^RGRHR8RR�(((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.

    uwindows-1252cC`s�|j|�|_tj||j�d|_d|_||_||_||_||_	||_
|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)

        iidN(RLt	rawStreamR<RtnumBytesMetatnumBytesChardettoverride_encodingttransport_encodingtsame_origin_parent_encodingtlikely_encodingtdefault_encodingtdetermineEncodingRKRN(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(RKt
codec_infotstreamreaderR�RMR<RN(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRN�s"cC`sUt|d�r|}nt|�}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(R9RRRR(RR>R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRL�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}Wntk
r4q�Xg}|�}xF|j
s�|jj|j�}|soPn|j|�|j|�qGW|j�t|jd�}|jjd�|dk	r�|dfSnt|j�df}|ddk	r�|Std�dfS(Nucertainiu	tentativeuutf-16(tUniversalDetectoruencodinguwindows-1252(t	detectBOMRERJR�R�tdetectEncodingMetaR�tnamet
startswithR�tchardet.universaldetectorR�tImportErrortdoneR�R"R�R&tfeedtclosetresultRR�(RtchardetRKR�tbufferstdetectorRtencoding((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��sP'
	


cC`s�t|�}|dkrdS|jdkr:td�}nr||jdkrf|jddf|_nF|jjd�|df|_|j�td|jd|f��dS(Nuutf-16beuutf-16leuutf-8iucertainuEncoding changed from %s to %s(uutf-16beuutf-16le(RJRER�RKR�RRNR(RtnewEncoding((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytchangeEncodings
cC`s�idtj6dtj6dtj6dtj6dtj6}|jjd�}|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_UTF8tBOM_UTF16_LEtBOM_UTF16_BEtBOM_UTF32_LEtBOM_UTF32_BER�R"tgetRRJRE(RtbomDicttstringR�R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�s$

cC`sk|jj|j�}t|�}|jjd�|j�}|dk	rg|jdkrgtd�}n|S(u9Report the encoding declared by the meta element
        iuutf-16beuutf-16leuutf-8N(uutf-16beuutf-16le(	R�R"R�tEncodingParserRtgetEncodingRER�RJ(RRtparserR�((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�9sN(R0R1R2RERcRRNRLR�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	�Zd	�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`stj||j��S(N(R!t__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�RE(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}qW||_dS(uSkip past a list of charactersiN(RRR[RE(RRzR�R{((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRo�s			cC`sc|j}xJ|t|�krU|||d!}||krH||_|S|d7}qW||_dS(Ni(RRR[RE(RRzR�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[RRcR�(RR!tnewPosition((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pytjumpTo�s(R0R1R2R�RR�R�R�R�R�R�tpropertyRR�tcurrentBytetspaceCharactersBytesRoR�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'RER�(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'RcR�R�R8R�(RtmethodDispatchRhtkeepParsingtkeytmethod((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-equiviscontent-typetcharsettcontent(R'R�R�RcR8REtgetAttributeR�RJtContentAttrParserR�tparse(Rt	hasPragmatpendingEncodingtattrttentativeEncodingtcodect
contentParser((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��s:		
		cC`s
|jt�S(N(thandlePossibleTagR8(R((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR��scC`st|j�|jt�S(N(R�R'R�Rc(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�RcR�tspacesAngleBracketsR�RE(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�}|dkr5dSg}g}x�tr�|dkr`|r`Pnz|tkr||j�}Pn^|d	kr�dj|�dfS|tkr�|j|j	��n|dkr�dS|j|�t
|�}qDW|dkr|j�dj|�dfSt
|�|j�}|d
kr�|}x�tr�t
|�}||kr�t
|�dj|�dj|�fS|tkr�|j|j	��q>|j|�q>Wn^|dkr�dj|�dfS|tkr�|j|j	��n|dkr	dS|j|�x}tr�t
|�}|tkrSdj|�dj|�fS|tkru|j|j	��q|dkr�dS|j|�qWdS(u_Return a name,value pair for the next attribute in the stream,
        if one is found, or Nonet/Rt=R(t't"N(RN(R�R(R�R�(
R'RoR�t	frozensetRERcR)tasciiUppercaseBytesR&R�R�R�R�(RR'R{tattrNamet	attrValuet	quoteChar((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�sf		


	

	(R0R1R2RR�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
||_dS(N(R'(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�RRoR�RER�R�R�(Rt	quoteMarktoldPosition((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyR�js.



(R0R1RR�(((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(R3RtdecodetUnicodeDecodeErrorRERtlookuptAttributeError(R�((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyRJ�s

(4t
__future__RRRtpip._vendor.sixRRtpip._vendor.six.movesRRR�Rutpip._vendorRt	constantsRR	R
RRR(R
tioRRR�R�R$tencodeR�R�R�R�tinvalid_unicode_no_surrogateRDRvtevalRftsetRntascii_punctuation_reRstobjectRRCR<R=R!R�R�R�RJ(((sE/usr/lib/python2.7/site-packages/pip/_vendor/html5lib/_inputstream.pyt<module>sP"
(((	
J	��h�'