HEX

File: //lib64/python3.9/urllib/__pycache__/robotparser.cpython-39.pyc
a

�DOg�$�@s\dZddlZddlZddlZdgZe�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@sreZdZdZddd�Zdd�Zdd�Zd	d
�Zdd�Zd
d�Z	dd�Z
dd�Zdd�Zdd�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/python3.9/urllib/robotparser.py�__init__s
zRobotFileParser.__init__cCs|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.mtimecCsddl}|��|_dS)zYSets the time the robots.txt file was last fetched to the
        current time.

        rN)�timer)r
rrrr�modified.szRobotFileParser.modifiedcCs&||_tj�|�dd�\|_|_dS)z,Sets the URL referring to a robots.txt file.��N)r�urllib�parse�urlparse�host�pathrrrrr
6szRobotFileParser.set_urlc
Cs�ztj�|j�}WnTtjjyf}z8|jdvr8d|_n|jdkrR|jdkrRd|_WYd}~n&d}~00|�	�}|�
|�d����dS)z4Reads the robots.txt URL and feeds it to the parser.)i�i�Ti�i�Nzutf-8)
rZrequest�urlopenr�error�	HTTPError�coderr	�readr�decode�
splitlines)r
�f�err�rawrrrr!;s
zRobotFileParser.readcCs,d|jvr|jdur(||_n|j�|�dS�N�*)�
useragentsrr�append)r
�entryrrr�
_add_entryHs

zRobotFileParser._add_entrycCsPd}t�}|��|D�]}|sP|dkr4t�}d}n|dkrP|�|�t�}d}|�d�}|dkrn|d|�}|��}|s|q|�dd�}t|�dkr|d����|d<tj	�
|d���|d<|ddkr�|dkr�|�|�t�}|j�|d�d}q|ddk�r.|dk�r6|j
�t|dd	��d}q|dd
k�rb|dk�r6|j
�t|dd��d}q|ddk�r�|dk�r6|d�����r�t|d�|_d}q|dd
k�r|dk�r6|d�d�}t|�dk�r|d�����r|d�����rtt|d�t|d��|_d}q|ddkr|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.
        rr��#N�:z
user-agentZdisallowFZallowTzcrawl-delayzrequest-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
dS|jrdS|jsdStj�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 urlFTrr0)rr	rrrrr7�
urlunparser�params�query�fragment�quoter�
applies_to�	allowancer)r
�	useragentrZ
parsed_urlr+rrr�	can_fetch�s&�

zRobotFileParser.can_fetchcCs>|��sdS|jD]}|�|�r|jSq|jr:|jjSdS�N)rrrGr<r�r
rIr+rrr�crawl_delay�s

zRobotFileParser.crawl_delaycCs>|��sdS|jD]}|�|�r|jSq|jr:|jjSdSrK)rrrGr=rrLrrr�request_rate�s

zRobotFileParser.request_ratecCs|js
dS|jSrK)rrrrr�	site_maps�szRobotFileParser.site_mapscCs,|j}|jdur||jg}d�tt|��S)Nz

)rr�join�map�str)r
rrrr�__str__�s
zRobotFileParser.__str__N)r)�__name__�
__module__�__qualname__�__doc__rrrr
r!r,rrJrMrNrOrSrrrrrs
		
	I

c@s(eZdZdZdd�Zdd�Zdd�ZdS)	r9zoA rule line is a single "Allow:" (allowance==True) or "Disallow:"
       (allowance==False) followed by a path.cCs<|dkr|sd}tj�tj�|��}tj�|�|_||_dS)NrT)rrrBrrFrrH)r
rrHrrrr�s
zRuleLine.__init__cCs|jdkp|�|j�Sr')r�
startswith)r
�filenamerrrrG�szRuleLine.applies_tocCs|jr
dndd|jS)NZAllowZDisallowz: )rHrrrrrrS�szRuleLine.__str__N)rTrUrVrWrrGrSrrrrr9�sr9c@s0eZdZdZdd�Zdd�Zdd�Zdd	�Zd
S)r1z?An entry has one or more user-agents and zero or more rulelinescCsg|_g|_d|_d|_dSrK)r)r8r<r=rrrrr�szEntry.__init__cCs�g}|jD]}|�d|���q
|jdur<|�d|j���|jdurf|j}|�d|j�d|j���|�tt|j	��d�
|�S)NzUser-agent: z
Crawl-delay: zRequest-rate: r0�
)r)r*r<r=ZrequestsZseconds�extendrQrRr8rP)r
�ret�agentZraterrrrS�s


z
Entry.__str__cCsF|�d�d��}|jD](}|dkr*dS|��}||vrdSqdS)z2check if this entry applies to the specified agentr0rr(TF)r4r6r))r
rIr]rrrrG�s
zEntry.applies_tocCs$|jD]}|�|�r|jSqdS)zZPreconditions:
        - our agent applies to this entry
        - filename is URL decodedT)r8rGrH)r
rYr@rrrrH
s

zEntry.allowanceN)rTrUrVrWrrSrGrHrrrrr1�s

r1)rW�collections�urllib.parserZurllib.request�__all__�
namedtuplerrr9r1rrrr�<module>sB