��Rea~� @s�ddlmZmZmZddlmZddlmZmZddl Z ddl Z ddlmZm Z ddlmZddlmZmZmZmZdd lmZdd lmZedd�eD��Zed d�eD��Zedd�eD��Zeeddg�BZdZejr�edEdkrWejd�dks]t �e j!eddF�e"d�d�Z#ne j!e�Z#ddddddddddddd d!d"d#d$d%d&d'd(d)d*d+d,d-d.d/d0d1d2d3h Z$e j!d4�Z%iZ&Gd5d6�d6e'�Z(d7d8�Z)Gd9d:�d:e'�Z*Gd;d<�d<e*�Z+Gd=d>�d>e,�Z-Gd?d@�d@e'�Z.GdAdB�dBe'�Z/dCdD�Z0dS)G�)�absolute_import�division�unicode_literals)� text_type)�http_client�urllibN)�BytesIO�StringIO)�webencodings�)�EOF�spaceCharacters�asciiLetters�asciiUppercase)�_ReparseException)�_utilscCsg|]}|jd��qS)�ascii)�encode)�.0�item�r��/builddir/build/BUILDROOT/alt-python35-pip-20.2.4-5.el8.x86_64/opt/alt/python35/lib/python3.5/site-packages/pip/_vendor/html5lib/_inputstream.py� s rcCsg|]}|jd��qS)r)r)rrrrrrs cCsg|]}|jd��qS)r)r)rrrrrrs �>�t|j|�|krd|t|j|�8}|d7}q'W||g|_dS)Nrr)�_bufferedBytes�AssertionErrorr#rr )r!r$�offset�irrr�seekFszBufferedStream.seekcCsp|js|j|�S|jdt|j�kr_|jdt|jd�kr_|j|�S|j|�SdS)Nrrr)r�_readStreamr r#�_readFromBuffer)r!�bytesrrr�readOs zBufferedStream.readcCstdd�|jD��S)NcSsg|]}t|��qSr)r#)rrrrrrYs z1BufferedStream._bufferedBytes..)�sumr)r!rrrr'XszBufferedStream._bufferedBytescCsJ|jj|�}|jj|�|jdd7.z3Cannot set an encoding with a unicode input, set %r) � isinstancer�HTTPResponser�response�addbase�fp�hasattrr/r� TypeError�HTMLUnicodeInputStream�HTMLBinaryInputStream)�source�kwargs� isUnicode� encodingsrrr�HTMLInputStream}s rOc@s�eZdZdZdZdd�Zdd�Zdd�Zd d �Zdd�Z d d�Z ddd�Zdd�Zdd�Z ddd�Zdd�ZdS)rIz�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(cCs�tjsd|_n-td�dkr6|j|_n|j|_dg|_td�df|_|j |�|_ |j�dS)a�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) Nu􏿿rrzutf-8�certain)r�supports_lone_surrogates�reportCharacterErrorsr#�characterErrorsUCS4�characterErrorsUCS2�newLines�lookupEncoding�charEncoding� openStream� dataStream�reset)r!rKrrrr"�s zHTMLUnicodeInputStream.__init__cCsCd|_d|_d|_g|_d|_d|_d|_dS)N�r)r%� chunkSize�chunkOffset�errors�prevNumLines�prevNumCols�_bufferedCharacter)r!rrrrZ�s zHTMLUnicodeInputStream.resetcCs(t|d�r|}nt|�}|S)zvProduces a file object from source. source can be either a file object, local filename or a string. r/)rGr )r!rKrrrrrX�s z!HTMLUnicodeInputStream.openStreamcCst|j}|jdd|�}|j|}|jdd|�}|dkr\|j|}n||d}||fS)N� rrr)r%�countr_�rfindr`)r!r)r%�nLines�positionLine�lastLinePos�positionColumnrrr� _position�s z HTMLUnicodeInputStream._positioncCs&|j|j�\}}|d|fS)z:Returns (line, col) of the current position in the stream.r)rir])r!�line�colrrrr �szHTMLUnicodeInputStream.positioncCsI|j|jkr"|j�s"tS|j}|j|}|d|_|S)zo Read one character from the stream or queue if available. Return EOF when EOF is reached. r)r]r\� readChunkrr%)r!r]�charrrrrm�s zHTMLUnicodeInputStream.charNcCsI|dkr|j}|j|j�\|_|_d|_d|_d|_|jj|�}|j r�|j |}d|_ n |s�dSt |�dkr�t|d�}|dks�d|ko�dknr�|d |_ |dd�}|jr |j|�|j dd �}|j d d �}||_t |�|_dS)Nr[rFr� i�i��z rb� Trrr)�_defaultChunkSizerir\r_r`r%r]rYr/rar#�ordrR�replace)r!r\r2�lastvrrrrl�s0 ( z HTMLUnicodeInputStream.readChunkcCs:x3tttj|��D]}|jjd�qWdS)Nzinvalid-codepoint)�ranger#�invalid_unicode_re�findallr^r1)r!r2�_rrrrSs"z*HTMLUnicodeInputStream.characterErrorsUCS4cCsd}x�tj|�D]�}|r%qt|j��}|j�}tj|||d��r�tj|||d��}|tkr�|j j d�d}q|dkr�|dkr�|t|�dkr�|j j d�qd}|j j d�qWdS)NF�zinvalid-codepointTi�i��r)ru�finditerrq�group�startr�isSurrogatePair�surrogatePairToCodepoint�non_bmp_invalid_codepointsr^r1r#)r!r2�skip�match� codepointr$�char_valrrrrT#s z*HTMLUnicodeInputStream.characterErrorsUCS2Fc Cseyt||f}Wn�tk r�x&|D]}t|�dks+t�q+Wdjdd�|D��}|syd|}tjd|�}t||f.z^%sz[%s]+N)�charsUntilRegEx�KeyErrorrqr(r4�re�compiler�r%r]r\�endr1rl) r!� characters�opposite�charsr��regexr6�mr��rrrr� charsUntil:s2 & z!HTMLUnicodeInputStream.charsUntilcCsl|tk rh|jdkr=||j|_|jd7_n+|jd8_|j|j|ksht�dS)Nrr)rr]r%r\r()r!rmrrr�ungetiszHTMLUnicodeInputStream.unget)r;r<r=r>rpr"rZrXrir rmrlrSrTr�r�rrrrrI�s &/rIc@seZdZdZdddddddd�Zdd�Zd d �Zddd�Zd d�Zdd�Z dd�Z dS)rJz�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. Nzwindows-1252TcCs�|j|�|_tj||j�d|_d|_||_||_||_||_ ||_ |j|�|_|jddk s�t �|j�dS)a�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) i�drN)rX� rawStreamrIr"�numBytesMeta�numBytesChardet�override_encoding�transport_encoding�same_origin_parent_encoding�likely_encoding�default_encoding�determineEncodingrWr(rZ)r!rKr�r�r�r�r�� useChardetrrrr"�s zHTMLBinaryInputStream.__init__cCs3|jdjj|jd�|_tj|�dS)Nrrr)rW� codec_info�streamreaderr�rYrIrZ)r!rrrrZ�s"zHTMLBinaryInputStream.resetcCs`t|d�r|}nt|�}y|j|j��Wntk r[t|�}YnX|S)zvProduces a file object from source. source can be either a file object, local filename or a string. r/)rGrr+r&� Exceptionr)r!rKrrrrrX�s z HTMLBinaryInputStream.openStreamcCs|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 r5Yn�Xg}|�}xX|js�|j j|j�}t|t�s{t�|s�P|j|�|j|�qHW|j�t|jd�}|j jd�|dk r�|dfSt|j�df}|ddk r|Std�dfS)NrPr� tentativezutf-16)�UniversalDetector�encodingzwindows-1252)� detectBOMrVr�r��detectEncodingMetar��name� startswithr��%pip._vendor.chardet.universaldetectorr��ImportError�doner�r/r�rBr.r(r1�feed�close�resultr+r�)r!�chardetrWr��buffers�detectorrr�rrrr��sR' z'HTMLBinaryInputStream.determineEncodingcCs�|jddkst�t|�}|dkr5dS|jdkretd�}|dk s�t�nr||jdkr�|jddf|_nF|jjd�|df|_|j�td|jd|f��dS) NrrP�utf-16be�utf-16lezutf-8rzEncoding changed from %s to %s)r�r�)rWr(rVr�r�r+rZr)r!�newEncodingrrr�changeEncodings z$HTMLBinaryInputStream.changeEncodingc Cs�tjdtjdtjdtjdtjdi}|jjd�}t|t �sZt �|j|dd��}d}|s�|j|�}d}|s�|j|dd ��}d }|r�|jj|�t |�S|jjd �dSdS)z�Attempts to detect at BOM at the start of the stream. If an encoding can be determined from the BOM return the name of the encoding otherwise return Nonezutf-8zutf-16lezutf-16bezutf-32lezutf-32be�N�rxr)�codecs�BOM_UTF8�BOM_UTF16_LE�BOM_UTF16_BE�BOM_UTF32_LE�BOM_UTF32_BEr�r/rBr.r(�getr+rV)r!�bomDict�stringr�r+rrrr�s$ zHTMLBinaryInputStream.detectBOMcCs}|jj|j�}t|t�s*t�t|�}|jjd�|j�}|dk ry|j dkryt d�}|S)z9Report the encoding declared by the meta element rN�utf-16be�utf-16lezutf-8)r�r�)r�r/r�rBr.r(�EncodingParserr+�getEncodingr�rV)r!r�parserr�rrrr�3sz(HTMLBinaryInputStream.detectEncodingMeta)r;r<r=r>r"rZrXr�r�r�r�rrrrrJzs(>"rJc@s�eZdZdZdd�Zdd�Zdd�Zdd �Zd d�Zdd �Z dd�Z dd�Zeee �Z dd�Zee�Zedd�Zdd�Zdd�Zdd�ZdS)� EncodingBytesz�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 raisedcCs+t|t�st�tj||j��S)N)rBr.r(�__new__�lower)r!�valuerrrr�FszEncodingBytes.__new__cCs d|_dS)Nrr)ri)r!r�rrrr"JszEncodingBytes.__init__cCs|S)Nr)r!rrr�__iter__NszEncodingBytes.__iter__cCsS|jd}|_|t|�kr/t�n|dkrAt�|||d�S)Nrr)rir#� StopIterationrH)r!�prrr�__next__Qs zEncodingBytes.__next__cCs |j�S)N)r�)r!rrr�nextYszEncodingBytes.nextcCsY|j}|t|�kr$t�n|dkr6t�|d|_}|||d�S)Nrr)rir#r�rH)r!r�rrr�previous]s zEncodingBytes.previouscCs(|jt|�krt�||_dS)N)rir#r�)r!r rrr�setPositionfszEncodingBytes.setPositioncCs9|jt|�krt�|jdkr1|jSdSdS)Nr)rir#r�)r!rrr�getPositionks zEncodingBytes.getPositioncCs||j|jd�S)Nr)r )r!rrr�getCurrentByteuszEncodingBytes.getCurrentBytecCsf|j}xM|t|�krX|||d�}||krK||_|S|d7}qW||_dS)zSkip past a list of charactersrN)r r#ri)r!r�r�r�rrrrzs zEncodingBytes.skipcCsf|j}xM|t|�krX|||d�}||krK||_|S|d7}qW||_dS)Nr)r r#ri)r!r�r�r�rrr� skipUntil�s zEncodingBytes.skipUntilcCs4|j||j�}|r0|jt|�7_|S)z�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)r�r r#)r!r.r6rrr� matchBytes�szEncodingBytes.matchBytescCsIy*|j||j�t|�d|_Wntk rDt�YnXdS)z�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 matchrT)�indexr r#ri� ValueErrorr�)r!r.rrr�jumpTo�s * zEncodingBytes.jumpToN)r;r<r=r>r�r"r�r�r�r�r�r��propertyr r��currentByte�spaceCharactersBytesrr�r�r�rrrrr�Bs r�c@s�eZdZdZdd�Zdd�Zdd�Zdd �Zd d�Zdd �Z dd�Z dd�Zdd�ZdS)r�z?Mini parser for detecting character encoding from meta elementscCst|�|_d|_dS)z3string - the data to work on for encoding detectionN)r�r2r�)r!r2rrrr"�szEncodingParser.__init__cCsd|jkrdSd|jfd|jfd|jfd|jfd|jfd|jff}x�|jD]�}d}y|jjd�Wntk r�PYnXxP|D]H\}}|jj|�r�y|�}PWq�tk r�d}PYq�Xq�W|skPqkW|j S) Ns)r2r�)r!rrrr��szEncodingParser.handleCommentcCs<|jjtkrdSd}d}x|j�}|dkrAdS|ddkr�|ddk}|r4|dk r4||_dSq%|ddkr�|d}t|�}|dk r4||_dSq%|ddkr%tt|d��}|j�}|dk r%t|�}|dk r%|r.||_dS|}q%WdS) NTFrs http-equivrscontent-typescharsetscontent) r2r�r��getAttributer�rV�ContentAttrParserr��parse)r!� hasPragma�pendingEncoding�attr�tentativeEncoding�codec� contentParserrrrr��s: zEncodingParser.handleMetacCs |jd�S)NF)�handlePossibleTag)r!rrrr��sz%EncodingParser.handlePossibleStartTagcCst|j�|jd�S)NT)r�r2r�)r!rrrr��s z#EncodingParser.handlePossibleEndTagcCs�|j}|jtkr6|r2|j�|j�dS|jt�}|dkr^|j�n+|j�}x|dk r�|j�}qmWdS)NTr)r2r��asciiLettersBytesr�r�r��spacesAngleBracketsr�)r!�endTagr2r�r�rrrr��s z EncodingParser.handlePossibleTagcCs|jjd�S)Nr)r2r�)r!rrrr�szEncodingParser.handleOthercCs�|j}|jttdg�B�}|dksIt|�dksIt�|d krYdSg}g}x�|dkr~|r~Pnz|tkr�|j�}Pn^|d kr�dj|�dfS|tkr�|j|j ��n|dkr�dS|j|�t |�}qhW|dkr1|j�dj|�dfSt |�|j�}|dkr�|}x�t |�}||kr�t |�dj|�dj|�fS|tkr�|j|j ��q\|j|�q\Wn^|dkr�dj|�dfS|tkr|j|j ��n|dkr!dS|j|�xwt |�}|tkredj|�dj|�fS|tkr�|j|j ��q1|dkr�dS|j|�q1WdS)z_Return a name,value pair for the next attribute in the stream, if one is found, or None�/Nrr�=r3�'�")rN)r�r)r�r�) r2rr�� frozensetr#r(r4�asciiUppercaseBytesr1r�r�r�r�)r!r2r��attrName� attrValue� quoteCharrrrr�sh $ zEncodingParser.getAttributeN) r;r<r=r>r"r�r�r�r�r�r�r�r�rrrrr��s$r�c@s(eZdZdd�Zdd�ZdS)r�cCs"t|t�st�||_dS)N)rBr.r(r2)r!r2rrrr"aszContentAttrParser.__init__cCsNy1|jjd�|jjd7_|jj�|jjdksHdS|jjd7_|jj�|jjdkr�|jj}|jjd7_|jj}|jj|�r�|j||jj�SdSn]|jj}y+|jjt�|j||jj�SWn#tk r/|j|d�SYnXWntk rIdSYnXdS)Nscharsetrr�r�r�)r�r�)r2r�r rr�r�r�r�)r!� quoteMark�oldPositionrrrr�es. zContentAttrParser.parseN)r;r<r=r"r�rrrrr�`sr�cCs|t|t�r;y|jd�}Wntk r:dSYnX|dk rtytj|�SWqxtk rpdSYqxXndSdS)z{Return the python codec name corresponding to an encoding or None if the string doesn't correspond to a valid encoding.rN)rBr.�decode�UnicodeDecodeErrorr �lookup�AttributeError)r�rrrrV�s rVrr)1� __future__rrrZpip._vendor.sixr�pip._vendor.six.movesrrr�r��iorr �pip._vendorr � constantsrr rrrr[rr�r�r�r�r��invalid_unicode_no_surrogaterQrcr(r��evalrur~�ascii_punctuation_rer��objectrrOrIrJr.r�r�r�rVrrrr�sJ" + J��b�'