Viewing File: /opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyo


t_c @`sddlmZmZmZddlmZddlmZmZddl	Z	ddl
Z
ddlmZm
Z
ddlmZddlmZmZmZmZdd	lmZdd
lmZegeD]Zejd^qZegeD]Zejd^qZegeD]Zejd^qZeedd
gBZdZej re
j!ed e"ddZ#ne
j!eZ#ddddddddddddddd d!d"d#d$d%d&d'd(d)d*d+d,d-d.d/d0d1h Z$e
j!d2Z%iZ&d3e'fd4YZ(d5Z)d6e'fd7YZ*d8e*fd9YZ+d:e,fd;YZ-d<e'fd=YZ.d>e'fd?YZ/d@Z0dS(Ai(tabsolute_importtdivisiontunicode_literals(t	text_type(thttp_clientturllibN(tBytesIOtStringIO(twebencodingsi(tEOFtspaceCharacterstasciiLetterstasciiUppercase(t_ReparseException(t_utilsuasciit>t<u[---Ÿ﷐-﷯￾￿🿾🿿𯿾𯿿𿿾𿿿񏿾񏿿񟿾񟿿񯿾񯿿񿿾񿿿򏿾򏿿򟿾򟿿򯿾򯿿򿿾򿿿󏿾󏿿󟿾󟿿󯿾󯿿󿿾󿿿􏿾􏿿]iu"\uD800-\uDFFF"u]iiiiiiiiiiiiiiiii	i	i
i
iiiii
i
iiiiiiu[	-
 -/:-@\[-`{-~]tBufferedStreamcB`sMeZdZdZdZdZdZdZdZdZ	RS(uBuffering 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(Nii(tstreamtbuffertposition(tselfR((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyt__init__:s		cC`sJd}x,|j|jd D]}|t|7}qW||jd7}|S(Nii(RRtlen(Rtpostchunk((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyttell?s
cC`s`|}d}x>t|j||krL|t|j|8}|d7}qW||g|_dS(Nii(RRR(RRtoffsetti((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytseekFscC`sp|js|j|S|jdt|jkr_|jdt|jdkr_|j|S|j|SdS(Niii(Rt_readStreamRRt_readFromBuffer(Rtbytes((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytreadOs	
 
cC`s&tg|jD]}t|^q
S(N(tsumRR(Rtitem((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyt_bufferedBytesXscC`sL|jj|}|jj||jdcd7<t||jd<|S(Nii(RR!RtappendRR(RR tdata((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyR[s
cC`s|}g}|jd}|jd}x|t|jkr|dkr|j|}|t||kr|}|||g|_n/t||}|t|g|_|d7}|j||||!||8}d}q)W|r|j|j|ndj|S(Niit(RRRR%Rtjoin(RR tremainingBytestrvtbufferIndextbufferOffsettbufferedDatatbytesToRead((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRbs$

$



(
t__name__t
__module__t__doc__RRRR!R$RR(((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyR3s								cK`st|tjs<t|tjjrEt|jtjrEt}n9t|drot|j	dt
}nt|t
}|rg|D]}|jdr|^q}|rtd|nt
||St||SdS(Nureadiu	_encodingu3Cannot set an encoding with a unicode input, set %r(t
isinstanceRtHTTPResponseRtresponsetaddbasetfptFalsethasattrR!Rtendswitht	TypeErrortHTMLUnicodeInputStreamtHTMLBinaryInputStream(tsourcetkwargst	isUnicodetxt	encodings((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytHTMLInputStream}s	(
R;cB`s}eZdZdZdZdZdZdZdZdZ	d
dZd	Zd
Z
edZdZRS(uProvides 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`stjsd|_n-tddkr6|j|_n|j|_dg|_tddf|_	|j
||_|jdS(uInitialises 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(
Rtsupports_lone_surrogatestNonetreportCharacterErrorsRtcharacterErrorsUCS4tcharacterErrorsUCS2tnewLinestlookupEncodingtcharEncodingt
openStreamt
dataStreamtreset(RR=((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRs	cC`sCd|_d|_d|_g|_d|_d|_d|_dS(Nui(Rt	chunkSizetchunkOffsetterrorstprevNumLinestprevNumColsRDt_bufferedCharacter(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRMs						cC`s(t|dr|}nt|}|S(uvProduces a file object from source.

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

        uread(R8R(RR=R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRKs	cC`st|j}|jdd|}|j|}|jdd|}|dkr\|j|}n||d}||fS(Nu
iii(RtcountRQtrfindRR(RRRtnLinestpositionLinetlastLinePostpositionColumn((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyt	_positions	
cC`s&|j|j\}}|d|fS(u:Returns (line, col) of the current position in the stream.i(RZRO(Rtlinetcol((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRscC`sL|j|jkr%|js%tSn|j}|j|}|d|_|S(uo Read one character from the stream or queue if available. Return
            EOF when EOF is reached.
        i(RORNt	readChunkR	R(RROtchar((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyR^s	

cC`sO|dkr|j}n|j|j\|_|_d|_d|_d|_|jj	|}|j
r|j
|}d|_
n
|stSt|dkrt
|d}|dksd|kodknr|d|_
|d }qn|jr|j|n|jdd	}|jd
d	}||_t||_tS(Nuiiii
iiu
u
u
(RDt_defaultChunkSizeRZRNRQRRRRORLR!RSR7RtordREtreplacetTrue(RRNR&tlastv((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyR]s0				
(
		cC`s:x3tttj|D]}|jjdqWdS(Nuinvalid-codepoint(trangeRtinvalid_unicode_retfindallRPR%(RR&t_((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRFs"cC`st}xtj|D]}|r(qnt|j}|j}tj|||d!rtj|||d!}|t	kr|j
jdnt}q|dkr|dkr|t
|dkr|j
jdqt}|j
jdqWdS(Niuinvalid-codepointiii(R7RetfinditerR`tgrouptstartRtisSurrogatePairtsurrogatePairToCodepointtnon_bmp_invalid_codepointsRPR%RbR(RR&tskiptmatcht	codepointRtchar_val((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRG#s 	c
C`sTyt||f}Wnqtk
rdjg|D]}dt|^q1}|scd|}ntjd|}t||f<nXg}xtr@|j|j|j	}|dkr|j	|jkrPqnB|j}||jkr|j
|j|j	|!||_	Pn|j
|j|j	|jsPqqWdj|}	|	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(R`tretcompileRbRoRRORDRNtendR%R](
Rt
characterstoppositetcharstctregexR*tmRvtr((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyt
charsUntil:s.
,
%		cC`sS|tk	rO|jdkr=||j|_|jd7_qO|jd8_ndS(Nii(R	RORRN(RR^((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytungetisN(R/R0R1R_RRMRKRZRR^RDR]RFRGR7R~R(((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyR;s	 					&		/R<cB`sbeZdZd	d	d	d	dedZdZdZedZdZ	dZ
dZRS(
uProvides 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||jd|_d|_||_||_||_||_	||_
|j||_|j
dS(uInitialises 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(RKt	rawStreamR;RtnumBytesMetatnumBytesChardettoverride_encodingttransport_encodingtsame_origin_parent_encodingtlikely_encodingtdefault_encodingtdetermineEncodingRJRM(RR=RRRRRt
useChardet((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRs							cC`s3|jdjj|jd|_tj|dS(Niureplace(RJt
codec_infotstreamreaderRRLR;RM(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRMs"cC`s_t|dr|}nt|}y|j|jWntk
rZt|}nX|S(uvProduces a file object from source.

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

        uread(R8RRRt	ExceptionR(RR=R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRKs	
cC`s|jdf}|ddk	r&|St|jdf}|ddk	rO|St|jdf}|ddk	rx|S|jdf}|ddk	r|St|jdf}|ddk	r|djjdr|St|j	df}|ddk	r|S|ryddl
m}Wntk
r4qXg}|}xF|j
s|jj|j}|soPn|j||j|qGW|jt|jd}|jjd|dk	r|dfSnt|jdf}|ddk	r|StddfS(Nucertainiu	tentativeuutf-16(tUniversalDetectoruencodinguwindows-1252(t	detectBOMRDRIRRtdetectEncodingMetaRtnamet
startswithRt%pip._vendor.chardet.universaldetectorRtImportErrortdoneRR!RR%tfeedtclosetresultRR(RtchardetRJRtbufferstdetectorRtencoding((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRsP'
	


cC`st|}|dkrdS|jdkr:td}nr||jdkrf|jddf|_nF|jjd|df|_|jtd|jd|fdS(Nuutf-16beuutf-16leuutf-8iucertainuEncoding changed from %s to %s(uutf-16beuutf-16le(RIRDRRJRRRMR
(RtnewEncoding((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytchangeEncodings
cC`sidtj6dtj6dtj6dtj6dtj6}|jjd}|j|d }d}|s|j|}d}|s|j|d }d}qn|r|jj	|t
|S|jj	d	d
Sd
S(uAttempts 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_BERR!tgetRRIRD(RtbomDicttstringRR((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRs$

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(	RR!RtEncodingParserRtgetEncodingRDRRI(RRtparserR((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyR3sN(R/R0R1RDRbRRMRKRRRR(((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyR<zs(		>		"t
EncodingBytescB`seZdZdZdZdZdZdZdZdZ	dZ
ee
e	Zd	Z
ee
Zed
ZdZdZd
ZRS(uString-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||jS(N(R t__new__tlower(Rtvalue((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRFscC`s
d|_dS(Ni(RZ(RR((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRJscC`s|S(N((R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyt__iter__NscC`sS|jd}|_|t|kr/tn|dkrDtn|||d!S(Nii(RZRt
StopIterationR:(Rtp((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyt__next__Qs		cC`s
|jS(N(R(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytnextYscC`sY|j}|t|kr$tn|dkr9tn|d|_}|||d!S(Nii(RZRRR:(RR((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytprevious]s			cC`s+|jt|krtn||_dS(N(RZRR(RR((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytsetPositionfs	cC`s<|jt|krtn|jdkr4|jSdSdS(Ni(RZRRRD(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytgetPositionks
	cC`s||j|jd!S(Ni(R(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytgetCurrentByteuscC`sc|j}xJ|t|krU|||d!}||krH||_|S|d7}qW||_dS(uSkip past a list of charactersiN(RRRZRD(RRyRRz((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRnzs			cC`sc|j}xJ|t|krU|||d!}||krH||_|S|d7}qW||_dS(Ni(RRRZRD(RRyRRz((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyt	skipUntils			cC`s7|j||j}|r3|jt|7_n|S(uLook 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*((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyt
matchBytesscC`sHy*|j||jt|d|_Wntk
rCtnXtS(uLook 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(tindexRRRZt
ValueErrorRRb(RR ((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pytjumpTos
*

(R/R0R1RRRRRRRRtpropertyRRtcurrentBytetspaceCharactersBytesRnRRR(((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRBs 													RcB`s_eZdZdZdZdZdZdZdZdZ	dZ
d	ZRS(
u?Mini parser for detecting character encoding from meta elementscC`st||_d|_dS(u3string - the data to work on for encoding detectionN(RR&RDR(RR&((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRscC`s
d|jkrdSd|jfd|jfd|jfd|jfd|jfd|jff}x|jD]}t}y|jjdWnt	k
rPnXxR|D]J\}}|jj
|ry|}PWqt	k
rt}PqXqqW|skPqkqkW|jS(Ns<metas<!--s</s<!s<?R(
R&RDt
handleCommentt
handleMetathandlePossibleEndTagthandleOtherthandlePossibleStartTagRbRRRR7R(RtmethodDispatchRgtkeepParsingtkeytmethod((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRs2
	
cC`s|jjdS(uSkip over commentss-->(R&R(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRscC`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&RRRbR7RDtgetAttributeRRItContentAttrParserRtparse(Rt	hasPragmatpendingEncodingtattrttentativeEncodingtcodect
contentParser((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRs:		
		cC`s
|jtS(N(thandlePossibleTagR7(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRscC`st|j|jtS(N(RR&RRb(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRs
cC`s|j}|jtkr9|r5|j|jntS|jt}|dkra|jn+|j}x|dk	r|j}qpWtS(NR(
R&RtasciiLettersBytesRRRbRtspacesAngleBracketsRRD(RtendTagR&RzR((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRs	


cC`s|jjdS(NR(R&R(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRscC`s|j}|jttdgB}|dkr5dSg}g}xtr|dkr`|r`Pnz|tkr||j}Pn^|d	krdj|dfS|tkr|j|j	n|dkrdS|j|t
|}qDW|dkr|jdj|dfSt
||j}|d
kr|}xtrt
|}||krt
|dj|dj|fS|tkr|j|j	q>|j|q>Wn^|dkrdj|dfS|tkr|j|j	n|dkr	dS|j|x}trt
|}|tkrSdj|dj|fS|tkru|j|j	q|dkrdS|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(RR(RR(
R&RnRt	frozensetRDRbR(tasciiUppercaseBytesR%RRRR(RR&RztattrNamet	attrValuet	quoteChar((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRsf		


	

	(R/R0R1RRRRRRRRR(((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRs				$				RcB`seZdZdZRS(cC`s
||_dS(N(R&(RR&((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRascC`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(NRiRRR(RR(	R&RRRnRRDRRR(Rt	quoteMarktoldPosition((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRes.



(R/R0RR(((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyR`s	cC`swt|tr: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(R2R tdecodetUnicodeDecodeErrorRDRtlookuptAttributeError(R((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyRIs

(1t
__future__RRRtpip._vendor.sixRtpip._vendor.six.movesRRRRttioRRtpip._vendorRt	constantsR	R
RRR
R'RRR#tencodeRRRRtinvalid_unicode_no_surrogateRCRutevalReRmtascii_punctuation_reRrtobjectRRBR;R<R RRRRI(((sR/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.pyt<module>sH"(((	
J	b'
Back to Directory File Manager