File: //opt/alt/python37/lib/python3.7/site-packages/bs4/__pycache__/diagnose.cpython-37.pyc
B
�.�`� � @ s d Z dZddlZddlmZ ddlmZ ddlZddlmZm Z ddl
m Z ddl Z ddl
Z
ddlZddlZddlZddlZddlZddlZdd � Zd#d d �ZG d
d� de�Zdd� ZdZdZd$dd�Zd%dd�Zd&dd�Zd'd d �Z d(d d!�Z e d"k�reej � � � dS ))z=Diagnostic functions, mainly for use when doing tech support.�MIT� N)�StringIO)�
HTMLParser)�
BeautifulSoup� __version__)�builder_registryc C s< t dt � t dtj � dddg}x>|D ]6}x0tjD ]}||jkr6P q6W |�|� t d| � q*W d|kr�|�ddg� y*dd l m
} t d
d � t t
|j�� � W n* tk
r� } z t d � W d
d
}~X Y nX d|k�r"ydd
l}t d|j � W n, tk
�r } z t d� W d
d
}~X Y nX t| d��r8| �� } nhtj�| ��rpt d| � t| ��}|�� } W d
Q R X n0| �d��s�| �d��r�t d| � t d� d
S t � x�|D ]�}t d| � d} yt| |�}
d} W n8 tk
�r
} zt d| � t�� W d
d
}~X Y nX | �r*t d| � t |
� � � t d� �q�W d
S ) z/Diagnostic suite for isolating common problems.z'Diagnostic running on Beautiful Soup %szPython version %sz html.parser�html5lib�lxmlz;I noticed that %s is not installed. Installing it may help.�xmlr )�etreezFound lxml version %s�.z.lxml is not installed or couldn't be imported.NzFound html5lib version %sz2html5lib is not installed or couldn't be imported.�readz7"%s" looks like a filename. Reading data from the file.zhttp:zhttps:z<"%s" looks like a URL. Beautiful Soup is not an HTTP client.zpYou need to use some other library to get the document behind the URL, and feed that document to Beautiful Soup.z#Trying to parse your markup with %sFTz %s could not parse the markup.z#Here's what %s did with the markup:zP--------------------------------------------------------------------------------) �printr �sys�versionr Zbuilders�features�remove�appendr r �join�map�strZ LXML_VERSION� ImportErrorr �hasattrr
�os�path�exists�open�
startswithr � Exception� traceback� print_excZprettify) �dataZ
basic_parsers�nameZbuilderr �er �fp�parser�success�soup� r( �=/opt/alt/python37/lib/python3.7/site-packages/bs4/diagnose.py�diagnose sd
r* Tc K sN ddl m} x<|jt| �fd|i|��D ] \}}td||j|jf � q(W dS )z�Print out the lxml events that occur during parsing.
This lets you see how lxml parses a document when no Beautiful
Soup code is running.
r )r �htmlz %s, %4s, %sN)r r Z iterparser r �tag�text)r! r+ �kwargsr �event�elementr( r( r) �
lxml_traceT s $r1 c @ s` e Zd ZdZdd� Zdd� Zdd� Zdd � Zd
d � Zd d
� Z dd� Z
dd� Z dd� Z dd� Z
dS )�AnnouncingParserz?Announces HTMLParser parse events, without doing anything else.c C s t |� d S )N)r )�self�sr( r( r) �_pa s zAnnouncingParser._pc C s | � d| � d S )Nz%s START)r5 )r3 r" �attrsr( r( r) �handle_starttagd s z AnnouncingParser.handle_starttagc C s | � d| � d S )Nz%s END)r5 )r3 r" r( r( r) �
handle_endtagg s z AnnouncingParser.handle_endtagc C s | � d| � d S )Nz%s DATA)r5 )r3 r! r( r( r) � handle_dataj s z AnnouncingParser.handle_datac C s | � d| � d S )Nz
%s CHARREF)r5 )r3 r" r( r( r) �handle_charrefm s z AnnouncingParser.handle_charrefc C s | � d| � d S )Nz %s ENTITYREF)r5 )r3 r" r( r( r) �handle_entityrefp s z!AnnouncingParser.handle_entityrefc C s | � d| � d S )Nz
%s COMMENT)r5 )r3 r! r( r( r) �handle_comments s z AnnouncingParser.handle_commentc C s | � d| � d S )Nz%s DECL)r5 )r3 r! r( r( r) � handle_declv s z AnnouncingParser.handle_declc C s | � d| � d S )Nz%s UNKNOWN-DECL)r5 )r3 r! r( r( r) � unknown_decly s z AnnouncingParser.unknown_declc C s | � d| � d S )Nz%s PI)r5 )r3 r! r( r( r) � handle_pi| s zAnnouncingParser.handle_piN)�__name__�
__module__� __qualname__�__doc__r5 r7 r8 r9 r: r; r< r= r> r? r( r( r( r) r2 ^ s r2 c C s t � }|�| � dS )z�Print out the HTMLParser events that occur during parsing.
This lets you see how HTMLParser parses a document when no
Beautiful Soup code is running.
N)r2 �feed)r! r% r( r( r) �htmlparser_trace s rE ZaeiouZbcdfghjklmnpqrstvwxyz� c C s> d}x4t | �D ](}|d dkr$t}nt}|t�|�7 }qW |S )z#Generate a random word-like string.� � r )�range� _consonants�_vowels�random�choice)�lengthr4 �i�tr( r( r) �rword� s rQ � c C s d� dd� t| �D ��S )z'Generate a random sentence-like string.� c s s | ]}t t�d d��V qdS )rR � N)rQ rL �randint)�.0rO r( r( r) � <genexpr>� s z rsentence.<locals>.<genexpr>)r rI )rN r( r( r) � rsentence� s rX �� c C s� dddddddg}g }x~t | �D ]r}t�dd �}|dkrRt�|�}|�d
| � q |d krr|�tt�d d ��� q |d
kr t�|�}|�d| � q W dd�|� d S )z+Randomly generate an invalid HTML document.�pZdiv�spanrO �b�script�tabler � z<%s>� rR rH z</%s>z<html>�
z</html>)rI rL rU rM r rX r )� num_elementsZ tag_names�elementsrO rM Ztag_namer( r( r) �rdoc� s
rd 順 c
C s( t dt � t| �}t dt|� � x�dddgddgD ]z}d}y"t�� }t||�}t�� }d}W n6 tk
r� } zt d | � t�� W d
d
}~X Y nX |r6t d ||| f � q6W d d
l m
} t�� }|� |� t�� }t d|| � d d
l } | �
� }t�� }|�|� t�� }t d|| � d
S )z.Very basic head-to-head performance benchmark.z1Comparative parser benchmark on Beautiful Soup %sz3Generated a large invalid HTML document (%d bytes).r r+ r z html.parserFTz %s could not parse the markup.Nz"BS4+%s parsed the markup in %.2fs.r )r z$Raw lxml parsed the markup in %.2fs.z(Raw html5lib parsed the markup in %.2fs.)r r rd �len�timer r r r r r ZHTMLr r �parse)
rb r! r% r& �ar' r\ r# r r r( r( r) �benchmark_parsers� s4
rj r c C sX t �� }|j}t| �}tt||d�}t�d|||� t� |�}|�
d� |� dd� d S )N)�bs4r! r% z bs4.BeautifulSoup(data, parser)Z
cumulativez
_html5lib|bs4�2 ) �tempfile�NamedTemporaryFiler" rd �dictrk �cProfileZrunctx�pstatsZStatsZ
sort_statsZ print_stats)rb r% Z
filehandle�filenamer! �varsZstatsr( r( r) �profile� s
rt �__main__)T)rF )rR )rY )re )re r )!rC � __license__rp �ior Z html.parserr rk r r Z bs4.builderr r rq rL rm rg r r r* r1 r2 rE rK rJ rQ rX rd rj rt r@ �stdinr
r( r( r( r) �<module> s8 =
!