3
�Pf � ) @ s� d dl mZmZmZ d dlmZmZ d dlmZm Z d dl
Z
d dl Z d dl m
Z
ddlmZmZmZmZ ddlmZ dd lmZ d d
lmZ yd d lmZ W n ek
r� eZY nX ed d
� eD ��Zedd
� eD ��Z edd
� eD ��Z eeddg�B Z dZ ej �r(e j!e ddF� e"d� d �Z#n
e j!e �Z#e$dddddddd d d d d d!d"d#d$d%d&d'd(d)d*d+d,d-d.d/d0d1d2d3d4g �Z%e j!d5�Z&i Z'G d6d7� d7e(�Z)d8d9� Z*G d:d;� d;e(�Z+G d<d=� d=e+�Z,G d>d?� d?e-�Z.G d@dA� dAe(�Z/G dBdC� dCe(�Z0dDdE� Z1dS )G� )�absolute_import�division�unicode_literals)� text_type� binary_type)� http_client�urllibN)� webencodings� )�EOF�spaceCharacters� asciiLetters�asciiUppercase)�ReparseException)�_utils)�StringIO)�BytesIOc C s g | ]}|j d ��qS )�ascii)�encode)�.0�item� r �"/usr/lib/python3.6/_inputstream.py�
<listcomp> s r c C s g | ]}|j d ��qS )r )r )r r r r r r s c C s g | ]}|j d ��qS )r )r )r r r r r r s � >� <u� [- - --]z"\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�� z[ -
-/:-@[-`{-~]c @ sH e Zd ZdZdd� Zdd� Zdd� Zdd � Zd
d � Zd d
� Z dd� Z
dS )�BufferedStreamz�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)
c C s || _ g | _ddg| _d S )Nr
r ���)�stream�buffer�position)�selfr r r r �__init__@ s zBufferedStream.__init__c C s@ d}x(| j d | jd � D ]}|t|�7 }qW || jd 7 }|S )Nr r
)r r! �len)r" �pos�chunkr r r �tellE s
zBufferedStream.tellc C sH |}d}x0t | j| �|k r8|t | j| �8 }|d7 }q
W ||g| _d S )Nr r
)r$ r r! )r" r% �offset�ir r r �seekL s zBufferedStream.seekc C sT | j s| j|�S | jd t| j �krF| jd t| j d �krF| j|�S | j|�S d S )Nr r
r )r � _readStreamr! r$ �_readFromBuffer)r" �bytesr r r �readU s
zBufferedStream.readc C s t dd� | jD ��S )Nc S s g | ] }t |��qS r )r$ )r r r r r r _ s z1BufferedStream._bufferedBytes.<locals>.<listcomp>)�sumr )r" r r r �_bufferedBytes^ s z BufferedStream._bufferedBytesc C s< | j j|�}| jj|� | jd d7 < t|�| jd< |S )Nr r
)r r. r �appendr! r$ )r" r- �datar r r r+ a s
zBufferedStream._readStreamc C s� |}g }| j d }| j d }x�|t| j�k r�|dkr�| j| }|t|�| krb|}||| g| _ n"t|�| }|t|�g| _ |d7 }|j|||| � � ||8 }d}q W |r�|j| j|�� dj|�S )Nr r
� )r! r$ r r1 r+ �join)r" r- ZremainingBytes�rvZ bufferIndexZ bufferOffsetZ bufferedDataZ bytesToReadr r r r, h s$
z BufferedStream._readFromBufferN) �__name__�
__module__� __qualname__�__doc__r# r' r* r. r0 r+ r, r r r r r 9 s r c K s� t | tj�s(t | tjj�r.t | jtj�r.d}n&t| d�rJt | jd�t �}n
t | t �}|r�dd� |D �}|rvt
d| ��t | f|�S t | f|�S d S )NFr. r c S s g | ]}|j d �r|�qS )Z _encoding)�endswith)r �xr r r r � s z#HTMLInputStream.<locals>.<listcomp>z3Cannot set an encoding with a unicode input, set %r)
�
isinstancer Z HTTPResponser ZresponseZaddbase�fp�hasattrr. r � TypeError�HTMLUnicodeInputStream�HTMLBinaryInputStream)�source�kwargsZ isUnicodeZ encodingsr r r �HTMLInputStream� s
rD c @ sp e Zd ZdZdZdd� Zdd� Zdd� Zd d
� Zd d � Z d
d� Z
ddd�Z dd� Z dd� Z
d dd�Zdd� ZdS ) r@ z�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 ( c C sZ t jsd| _n td�dkr$| j| _n| j| _dg| _td�df| _| j |�| _
| j � dS )a� Initialises the HTMLInputStream.
HTMLInputStream(source, [encoding]) -> Normalized stream from source
for use by html5lib.
source can be either a file-object, local filename or a string.
The optional encoding parameter must be a string that indicates
the encoding. If specified, that encoding will be used,
regardless of any BOM or later declaration (such as in a meta
element)
Nu r
r zutf-8�certain) r �supports_lone_surrogates�reportCharacterErrorsr$ �characterErrorsUCS4�characterErrorsUCS2ZnewLines�lookupEncoding� charEncoding�
openStream�
dataStream�reset)r" rB r r r r# � s
z HTMLUnicodeInputStream.__init__c C s. d| _ d| _d| _g | _d| _d| _d | _d S )N� r )r& � chunkSize� chunkOffset�errors� prevNumLines� prevNumCols�_bufferedCharacter)r" r r r rN � s z HTMLUnicodeInputStream.resetc C s t |d�r|}nt|�}|S )zvProduces a file object from source.
source can be either a file object, local filename or a string.
r. )r> r )r" rB r r r r rL � s
z!HTMLUnicodeInputStream.openStreamc C sT | j }|jdd|�}| j| }|jdd|�}|dkr@| j| }n ||d }||fS )N�
r r
r )r& �countrS �rfindrT )r" r( r& ZnLinesZ positionLineZ lastLinePosZpositionColumnr r r � _position� s
z HTMLUnicodeInputStream._positionc C s | j | j�\}}|d |fS )z:Returns (line, col) of the current position in the stream.r
)rY rQ )r" �line�colr r r r! � s z HTMLUnicodeInputStream.positionc C s6 | j | jkr| j� stS | j }| j| }|d | _ |S )zo Read one character from the stream or queue if available. Return
EOF when EOF is reached.
r
)rQ rP � readChunkr r& )r" rQ �charr r r r] � s
zHTMLUnicodeInputStream.charNc C s� |d kr| j }| j| j�\| _| _d| _d| _d| _| jj|�}| j rX| j | }d | _ n|s`dS t
|�dkr�t |d �}|dks�d| ko�dkn r�|d
| _ |d d� }| j r�| j |� |j
dd �}|j
d
d �}|| _t
|�| _d S )NrO r Fr
�
i � i�� z
rV �
Tr r r )�_defaultChunkSizerY rP rS rT r& rQ rM r. rU r$ �ordrG �replace)r" rP r2 Zlastvr r r r\ � s0
z HTMLUnicodeInputStream.readChunkc C s, x&t ttj|���D ]}| jjd� qW d S )Nzinvalid-codepoint)�ranger$ �invalid_unicode_re�findallrR r1 )r" r2 �_r r r rH % s z*HTMLUnicodeInputStream.characterErrorsUCS4c C s� d}x�t j|�D ]�}|rqt|j� �}|j� }tj|||d � �rttj|||d � �}|tkrn| j j
d� d}q|dkr�|dkr�|t |�d kr�| j j
d� qd}| j j
d� qW d S )NF� zinvalid-codepointTi � i�� r
) rd �finditerra �group�startr ZisSurrogatePairZsurrogatePairToCodepoint�non_bmp_invalid_codepointsrR r1 r$ )r" r2 �skip�matchZ codepointr% Zchar_valr r r rI ) s z*HTMLUnicodeInputStream.characterErrorsUCS2Fc C s� yt ||f }W nN tk
r^ djdd� |D ��}|s@d| }tjd| � }t ||f< Y nX g }x||j| j| j�}|dkr�| j| jkr�P n0|j � }|| jkr�|j
| j| j|� � || _P |j
| j| jd� � | j � sfP qfW dj|�}|S )z� 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.
rO c S s g | ]}d t |� �qS )z\x%02x)ra )r �cr r r r N s z5HTMLUnicodeInputStream.charsUntil.<locals>.<listcomp>z^%sz[%s]+N) �charsUntilRegEx�KeyErrorr4 �re�compilerm r& rQ rP �endr1 r\ ) r" Z
charactersZopposite�charsZregexr5 �mrs �rr r r �
charsUntil@ s.
z!HTMLUnicodeInputStream.charsUntilc C s@ |d k r<| j dkr.|| j | _| jd7 _n| j d8 _ d S )Nr r
)rQ r& rP )r" r] r r r �ungeto s
z HTMLUnicodeInputStream.unget)N)F)r6 r7 r8 r9 r` r# rN rL rY r! r] r\ rH rI rw rx r r r r r@ � s
&
/r@ c @ sL e Zd ZdZddd�Zdd� Zd d
� Zdd d �Zd
d� Zdd� Z dd� Z
dS )rA z�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.
N� windows-1252Tc C s\ | j |�| _tj| | j� d| _d| _|| _|| _|| _|| _ || _
| 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)
i �d N)rL � rawStreamr@ r# � numBytesMeta�numBytesChardet�override_encoding�transport_encoding�same_origin_parent_encoding�likely_encoding�default_encoding�determineEncodingrK rN )r" rB r~ r r� r� r� Z
useChardetr r r r# � s z HTMLBinaryInputStream.__init__c C s& | j d jj| jd�| _tj| � d S )Nr rb )rK Z
codec_info� streamreaderr{ rM r@ rN )r" r r r rN � s zHTMLBinaryInputStream.resetc C sD t |d�r|}nt|�}y|j|j� � W n t|�}Y nX |S )zvProduces a file object from source.
source can be either a file object, local filename or a string.
r. )r> r r* r' r )r" rB r r r r rL � s
z HTMLBinaryInputStream.openStreamc C s� | j � df}|d d k r |S t| j�df}|d d k r:|S t| j�df}|d d k rX|S | j� df}|d d k rt|S t| j�df}|d d k r�|d jjd� r�|S t| j�df}|d d k r�|S |�rdyddl m
} W n t k
r� Y nxX g }|� }x6|j �s.| j
j| j�}|�sP |j|� |j|� q�W |j� t|jd �}| j
jd� |d k �rd|dfS t| j�df}|d d k �r�|S td�dfS )NrE r Z tentativezutf-16)�UniversalDetector�encodingz windows-1252)� detectBOMrJ r~ r �detectEncodingMetar� �name�
startswithr� Zchardet.universaldetectorr� � ImportError�doner{ r. r} r1 Zfeed�close�resultr* r� )r" ZchardetrK r� ZbuffersZdetectorr r� r r r r� � sP
z'HTMLBinaryInputStream.determineEncodingc C s� t |�}|d krd S |jdkr(t d�}nT|| jd krH| jd df| _n4| jjd� |df| _| j� td| jd |f ��d S )N�utf-16be�utf-16lezutf-8r rE z Encoding changed from %s to %s)r� r� )rJ r� rK r{ r* rN r )r" Z newEncodingr r r �changeEncoding s
z$HTMLBinaryInputStream.changeEncodingc
C s� t jdt jdt jdt jdt jdi}| jjd�}|j|dd� �}d}|sp|j|�}d}|sp|j|dd � �}d }|r�| jj |� t
|�S | jj d
� dS dS ) z�Attempts to detect at BOM at the start of the stream. If
an encoding can be determined from the BOM return the name of the
encoding otherwise return Nonezutf-8zutf-16lezutf-16bezutf-32lezutf-32be� N� rg r ) �codecs�BOM_UTF8� BOM_UTF16_LE� BOM_UTF16_BE� BOM_UTF32_LE� BOM_UTF32_BEr{ r. �getr* rJ )r" ZbomDict�stringr� r* r r r r� s"
z HTMLBinaryInputStream.detectBOMc C sH | j j| j�}t|�}| j jd� |j� }|dk rD|jdkrDtd�}|S )z9Report the encoding declared by the meta element
r N�utf-16be�utf-16lezutf-8)r� r� )r{ r. r| �EncodingParserr* � getEncodingr� rJ )r" r �parserr� r r r r� 9 s z(HTMLBinaryInputStream.detectEncodingMeta)NNNNry T)T) r6 r7 r8 r9 r# rN rL r� r� r� r� r r r r rA � s
(
>"rA c @ s� e Zd ZdZdd� Zdd� Zdd� Zdd � Zd
d � Zd d
� Z dd� Z
dd� Z e e e
�Z
dd� Ze e�Zefdd�Zdd� Zdd� Zdd� Zd S ) �
EncodingBytesz�String-like object with an associated position and various extra methods
If the position is ever greater than the string length then an exception is
raisedc C s t j| |j� �S )N)r- �__new__�lower)r" �valuer r r r� L s zEncodingBytes.__new__c C s
d| _ d S )Nr
r )rY )r" r� r r r r# P s zEncodingBytes.__init__c C s | S )Nr )r" r r r �__iter__T s zEncodingBytes.__iter__c C s> | j d }| _ |t| �kr"t�n |dk r.t�| ||d � S )Nr
r )rY r$ �
StopIterationr? )r" �pr r r �__next__W s zEncodingBytes.__next__c C s | j � S )N)r� )r" r r r �next_ s zEncodingBytes.nextc C sB | j }|t| �krt�n |dk r$t�|d | _ }| ||d � S )Nr r
)rY r$ r� r? )r" r� r r r �previousc s zEncodingBytes.previousc C s | j t| �krt�|| _ d S )N)rY r$ r� )r" r! r r r � setPositionl s zEncodingBytes.setPositionc C s* | j t| �krt�| j dkr"| j S d S d S )Nr )rY r$ r� )r" r r r � getPositionq s
zEncodingBytes.getPositionc C s | | j | j d � S )Nr
)r! )r" r r r �getCurrentByte{ s z EncodingBytes.getCurrentBytec C sL | j }x:|t| �k r@| ||d � }||kr6|| _|S |d7 }qW || _dS )z Skip past a list of charactersr
N)r! r$ rY )r" rt r� rn r r r rl � s zEncodingBytes.skipc C sL | j }x:|t| �k r@| ||d � }||kr6|| _|S |d7 }qW || _d S )Nr
)r! r$ rY )r" rt r� rn r r r � skipUntil� s zEncodingBytes.skipUntilc C s> | j }| ||t|� � }|j|�}|r:| j t|�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- r� r2 r5 r r r �
matchBytes� s
zEncodingBytes.matchBytesc C sR | | j d� j|�}|dkrJ| jdkr,d| _| j|t|� d 7 _dS t�dS )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 matchNr
r Tr r )r! �findrY r$ r� )r" r- Z newPositionr r r �jumpTo� s
zEncodingBytes.jumpToN)r6 r7 r8 r9 r� r# r� r� r� r� r� r� �propertyr! r� � currentByte�spaceCharactersBytesrl r� r� r� r r r r r� H s
r� c @ sX e Zd ZdZdd� Zdd� Zdd� Zdd � Zd
d � Zd d
� Z dd� Z
dd� Z dd� Z dS )r� z?Mini parser for detecting character encoding from meta elementsc C s t |�| _d| _dS )z3string - the data to work on for encoding detectionN)r� r2 r� )r" r2 r r r r# � s
zEncodingParser.__init__c
C s� d| j fd| jfd| jfd| jfd| jfd| jff}x^| jD ]T}d}xD|D ]<\}}| jj|�rJy |� }P W qJ tk
r� d}P Y qJX qJW |s<P q<W | jS ) Ns <!--s <metas </s <!s <?r TF) �
handleComment�
handleMeta�handlePossibleEndTag� handleOther�handlePossibleStartTagr2 r� r� r� )r" ZmethodDispatchrf Z keepParsing�key�methodr r r r� � s& zEncodingParser.getEncodingc C s | j jd�S )zSkip over commentss -->)r2 r� )r" r r r r� � s z EncodingParser.handleCommentc C s� | j jtkrdS d}d }x�| j� }|d kr.dS |d dkr^|d dk}|r�|d k r�|| _dS q|d dkr�|d }t|�}|d k r�|| _dS q|d dkrtt|d ��}|j� }|d k rt|�}|d k r|r�|| _dS |}qW d S ) NTFr s
http-equivr
s content-types charsets content) r2 r� r� � getAttributer� rJ �ContentAttrParserr� �parse)r" Z hasPragmaZpendingEncoding�attrZtentativeEncoding�codecZ
contentParserr r r r� � s: zEncodingParser.handleMetac C s
| j d�S )NF)�handlePossibleTag)r" r r r r� � s z%EncodingParser.handlePossibleStartTagc C s t | j� | jd�S )NT)r� r2 r� )r" r r r r� � s
z#EncodingParser.handlePossibleEndTagc C sf | j }|jtkr(|r$|j� | j� dS |jt�}|dkrD|j� n | j� }x|d k r`| j� }qNW dS )NTr )r2 r� �asciiLettersBytesr� r� r� �spacesAngleBracketsr� )r" ZendTagr2 rn r� r r r r� � s
z EncodingParser.handlePossibleTagc C s | j jd�S )Nr )r2 r� )r" r r r r� s zEncodingParser.handleOtherc C s� | j }|jttdg�B �}|dkr&dS g }g }xt|dkr@|r@P nX|tkrT|j� }P nD|d krjdj|�dfS |tkr�|j|j� � n|dkr�dS |j|� t|�}q0W |dkr�|j � dj|�dfS t|� |j� }|d
k�r:|}x�t|�}||k�rt|� dj|�dj|�fS |tk�r*|j|j� � q�|j|� q�W nJ|dk�rRdj|�dfS |tk�rl|j|j� � n|dk�rzdS |j|� x^t|�}|t
k�r�dj|�dj|�fS |tk�r�|j|j� � n|dk�r�dS |j|� �q�W dS ) z_Return a name,value pair for the next attribute in the stream,
if one is found, or None� /r N� =r3 � '� ")r N)r� r )r� r� ) r2 rl r� � frozensetr4 �asciiUppercaseBytesr1 r� r� r� r� )r" r2 rn ZattrNameZ attrValueZ quoteCharr r r r� sf
zEncodingParser.getAttributeN)
r6 r7 r8 r9 r# r� r� r� r� r� r� r� r� r r r r r� � s $r� c @ s e Zd Zdd� Zdd� ZdS )r� c C s
|| _ d S )N)r2 )r" r2 r r r r# f s zContentAttrParser.__init__c C s y�| j jd� | j jd7 _| j j� | j jdks8d S | j jd7 _| j j� | j jdkr�| j j}| j jd7 _| j j}| j j|�r�| j || j j� S d S nF| j j}y | j jt� | j || j j� S tk
r� | j |d � S X W n tk
�r d S X d S )Ns charsetr
r� r� r� )r� r� )r2 r� r! rl r� r� r� r� )r" Z quoteMarkZ oldPositionr r r r� j s.
zContentAttrParser.parseN)r6 r7 r8 r# r� r r r r r� e s r� c C s` t | t�r.y| jd�} W n tk
r, dS X | dk rXy
tj| �S tk
rT dS X ndS dS )z{Return the python codec name corresponding to an encoding or None if the
string doesn't correspond to a valid encoding.r N)r< r �decode�UnicodeDecodeErrorr �lookup�AttributeError)r� r r r rJ � s
rJ r )2Z
__future__r r r Zpip._vendor.sixr r Zpip._vendor.six.movesr r r� rq Z pip._vendorr Z constantsr r r
r r rO r �ior r r� r� r� r� r� r� Z invalid_unicode_no_surrogaterF rr �evalrd �setrk Zascii_punctuation_rero �objectr rD r@ rA r- r� r� r� rJ r r r r �<module> sV
J g Ih 6'