File manager

U

e5d�$�@s\dZd
dl
Z
d
dlZd
dlZdg
Ze
�dd�ZGdd�d�ZGdd�d�Z	Gd	d
�d
�Z
dS)a%
 robotparser.py

    Copyright (C) 2000  Bastian Kleineidam

    You can choose between two licenses when using this package:
    1) GNU GPLv2
    2) PSF license for Python 2.2

    The robots.txt Exclusion Protocol is implemented as specified in
    http://www.robotstxt.org/norobots-rfc.txt
�N�RobotFileParser�RequestRatezrequests secondsc@sreZ
dZd
Zddd�
Zdd�Zdd�Zd	d
�Zdd�Zd
d�Z	dd�Z
dd�Zdd�Zdd�Z
dd�Zdd�ZdS)rzs This class provides a set of methods to read, parse and answer
    questions about a single robots.txt file.

    �cCs2g|_g|_
d|_d
|_d
|_|�|
�

d|_dS)NFr
)�entries�sitemaps�
default_entry�disallow_all�	allow_all�set_url�last_checked��self�url�r�*/usr/lib64/python3.8/urllib/robotparser.py�__init__s







zRobotFileParser.__init__c


Cs|jS)
z�Returns the time the robots.txt file was last fetched.

        This is useful for long-running web spiders that need to
        check for new robots.txt files periodically.

        )
r�
r
rrr�mtime%szRobotFileParser.mtimec
Csd
dl}
|
��|_
dS)zYSets the time the robots.txt file was last fetched to the
        current time.

        r
N)�timer)r
rrrr�modified.s
zRobotFileParser.modifiedcCs&|
|_t
j�|
�
d
d�\|_|_dS)z,Sets the URL referring to a robots.txt file.�
�N)r�urllib�parse�urlparseZhost�pathrrrrr
6s
zRobotFileParser.set_urlc

Cs�ztj
�|j�
}
WnRtjjk
rd
}
z0|jd
kr:d|_n|jdkrT|jdkrTd|_W5d}~XYnX|
�	�}|�
|�d�
���

dS)z4Reads the robots.txt URL and feeds it to the parser.)i�
i�
Ti�
i�
Nzutf-8)
rZrequestZurlopenr�errorZ	HTTPError�coderr	�readr�decode�
splitlines)r
�
f�err�rawrrrr;s







zRobotFileParser.readcCs,d
|
jkr|j
dkr(|
|_
n|j�|
�

dS�N�
*)�
useragentsrr�append)r
�entryrrr�
_add_entryHs


zRobotFileParser._add_entrycCsPd
}t�}|�
�
|
D�]}|sP|dkr4t�}d
}n|dkrP|�|�

t�}d
}|�d�
}|d
krn|d|�}|��}|s|q|�dd�}t|�
dkr|d
����|d
<tj	�
|d���
|d<|d
dkr�|dkr�|�|�

t�}|j�|d�

d}q|d
dk�
r.|d
k�r6|j
�t|dd	��

d}q|d
d
k�
rb|d
k�r6|j
�t|dd��

d}q|d
dk�
r�|d
k�r6|d�����
r�t|d�
|_d}q|d
d
k�r|d
k�r6|d�d�
}t|�
dk�r|d
�����r|d�����rtt|d
�
t|d�
�|_d}q|d
dkr|j�|d�

q|dk�rL|�|�

dS)z�Parse the input lines from a robots.txt file.

        We allow that a user-agent: line is not preceded by
        one or more blank lines.
        r
r��
#N�
:z
user-agentZdisallowFZallowTzcrawl-delayzrequest-rate�
/Zsitemap)�Entryrr)�find�strip�split�len�lowerrr�unquoter&r'�	rulelines�RuleLine�isdigit�int�delayr�req_rater)r
�lines�stater(�line�
iZnumbersrrrrQsj
















































 
�




zRobotFileParser.parsecCs�|jr
d
S|j
rdS|jsd
Stj�tj�|�
�
}tj�dd|j|j	|j
|jf�
}tj�|�
}|sfd}|j
D]}|�|
�
rl|�|�

Sql|jr�|j�|�
SdS)z=using the parsed robots.txt decide if useragent can fetch urlFTrr-)rr	rrrrr4�
urlunparserZparamsZqueryZfragment�quoter�
applies_to�	allowancer)r
�	useragentrZ
parsed_urlr(rrr�	can_fetch�s*





�







zRobotFileParser.can_fetchcCs>|��sdS|j
D]}|�|
�
r|j
Sq|jr:|jjSdS�
N)rrrAr9r�r
rCr(rrr�crawl_delay�s









zRobotFileParser.crawl_delaycCs>|��sdS|j
D]}|�|
�
r|j
Sq|jr:|jjSdSrE)rrrAr:rrFrrr�request_rate�s









zRobotFileParser.request_ratec


Cs|js
dS|jSrE)
rrrrr�	site_maps�s


zRobotFileParser.site_mapsc
Cs,|j}
|j
dk	r|
|j
g
}
d
�tt|
��
S)Nz

)rr�join�map�str)r
rrrr�__str__�s




zRobotFileParser.__str__N)
r)�__name__�
__module__�__qualname__�__doc__rrrr
rr)rrDrGrHrIrMrrrrrs

		
	I

c@s(eZ
dZd
Zdd�Zdd�Zdd�ZdS)	r6zoA rule line is a single "Allow:" (allowance==True) or "Disallow:"
       (allowance==False) followed by a path.cCs<|
d
kr|sd}tj
�tj
�|
�
�
}
tj
�|
�
|_||_dS)NrT)rrr?rr@rrB)r
rrBrrrr�s




zRuleLine.__init__cCs|jd
kp|
�
|j�
Sr$)r�
startswith)r
�filenamerrrrA�s
zRuleLine.applies_toc

Cs|jr
d
ndd|j
S)NZAllowZDisallowz: )rBrrrrrrM�s
zRuleLine.__str__N)rNrOrPrQrrArMrrrrr6�s
r6c@s0eZ
dZd
Zdd�Zdd�Zdd�Zdd	�Zd
S)r.z?An entry has one or more user-agents and zero or more rulelinesc

Csg|_g|_
d|_d|_dSrE)r&r5r9r:rrrrr�s



zEntry.__init__c
Cs�g}
|jD]}|
�
d
|���

q
|jdk	r<|
�
d|j���

|jdk	rf|j}|
�
d|j�d|j���

|
�tt|j	��

d�
|
�
S)NzUser-agent: z
Crawl-delay: zRequest-rate: r-�

)r&r'r9r:ZrequestsZseconds�extendrKrLr5rJ)r
Zret�agentZraterrrrM�s












z
Entry.__str__cCsF|
�d
�
d�
�}
|jD](}|dkr*
dS|�
�}||
kr
dSqdS)z2check if this entry applies to the specified agentr-r
r%TF)r1r3r&)r
rCrVrrrrA�s






zEntry.applies_tocCs$|jD]}|�
|
�
r|j
Sqd
S)zZPreconditions:
        - our agent applies to this entry
        - filename is URL decodedT)r5rArB)r
rSr=rrrrB

s




zEntry.allowanceN)rNrOrPrQrrMrArBrrrrr.�s



r.)rQ�collectionsZurllib.parserZurllib.request�__all__�
namedtuplerrr6r.rrrr�<module>
s

B