B
� f�" � @ s\ d Z ddlZddlZddlZdgZe�dd�ZG dd� d�ZG dd� d�Z G d 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 @ sj e Zd ZdZddd�Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� ZdS )r zs This class provides a set of methods to read, parse and answer
questions about a single robots.txt file.
� c C s, g | _ d | _d| _d| _| �|� d| _d S )NFr )�entries�
default_entry� disallow_all� allow_all�set_url� last_checked)�self�url� r
�7/opt/alt/python37/lib64/python3.7/urllib/robotparser.py�__init__ s
zRobotFileParser.__init__c C s | j S )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 r
r
r �mtime$ s zRobotFileParser.mtimec C s ddl }|� � | _dS )zYSets the time the robots.txt file was last fetched to the
current time.
r N)�timer
)r r r
r
r �modified- s zRobotFileParser.modifiedc C s& || _ tj�|�dd� \| _| _dS )z,Sets the URL referring to a robots.txt file.� � N)r �urllib�parse�urlparseZhost�path)r r r
r
r r 5 s zRobotFileParser.set_urlc
C s� yt j�| j�}W nR t jjk
rd } z0|jdkr:d| _n|jdkrT|jdk rTd| _W dd}~X Y n X |� � }| �
|� d�� � � dS )z4Reads the robots.txt URL and feeds it to the parser.)i� i� Ti� i� Nzutf-8)
r ZrequestZurlopenr �errorZ HTTPError�coder r �readr �decode�
splitlines)r �f�err�rawr
r
r r : s
zRobotFileParser.readc C s, d|j kr | jd kr(|| _n | j�|� d S )N�*)�
useragentsr r �append)r �entryr
r
r �
_add_entryG s
zRobotFileParser._add_entryc C s6 d}t � }| �� �x|D �]�}|sT|dkr8t � }d}n |dkrT| �|� t � }d}|�d�}|dkrr|d|� }|�� }|s�q|�dd�}t|�dkr|d �� �� |d< tj �
|d �� �|d< |d dk�r |dkr�| �|� t � }|j � |d � d}q|d dk�r4|dk�r|j
� t|d d �� d}q|d d
k�rh|dk�r|j
� t|d d �� d}q|d d k�r�|dk�r|d �� �� �r�t|d �|_d}q|d d
kr|dkr|d �d�}t|�dk�r|d �� �� �r|d �� �� �rtt|d �t|d ��|_d}qW |dk�r2| �|� 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-agentZdisallowFZallowTz crawl-delayz request-rate�/)�Entryr r% �find�strip�split�len�lowerr r �unquoter" r# � rulelines�RuleLine�isdigit�int�delayr �req_rate)r �lines�stater$ �line�iZnumbersr
r
r r P sd
zRobotFileParser.parsec C s� | j r
dS | jrdS | js dS tj�tj�|��}tj�dd|j|j |j
|j f�}tj� |�}|sfd}x"| j
D ]}|�|�rn|�|�S qnW | jr�| j�|�S dS )z=using the parsed robots.txt decide if useragent can fetch urlFTr r) )r r r
r r r r0 �
urlunparser ZparamsZqueryZfragment�quoter �
applies_to� allowancer )r � useragentr Z
parsed_urlr$ r
r
r � can_fetch� s$
zRobotFileParser.can_fetchc C s> | � � s d S x | jD ]}|�|�r|jS qW | jr:| jjS d S )N)r r r= r5 r )r r? r$ r
r
r � crawl_delay� s
zRobotFileParser.crawl_delayc C s> | � � s d S x | jD ]}|�|�r|jS qW | jr:| jjS d S )N)r r r= r6 r )r r? r$ r
r
r � request_rate� s
z RobotFileParser.request_ratec C s0 | j }| jd k r || jg }d�tt|��d S )N�
)r r �join�map�str)r r r
r
r �__str__� s
zRobotFileParser.__str__N)r )�__name__�
__module__� __qualname__�__doc__r r r r r r% r r@ rA rB rG r
r
r
r r s
C
c @ s( e Zd ZdZdd� Zdd� Zdd� ZdS ) r2 zoA rule line is a single "Allow:" (allowance==True) or "Disallow:"
(allowance==False) followed by a path.c C s< |dkr|sd}t j�t j�|��}t j�|�| _|| _d S )Nr T)r r r; r r< r r> )r r r> r
r
r r � s
zRuleLine.__init__c C s | j dkp|�| j �S )Nr! )r �
startswith)r �filenamer
r
r r= � s zRuleLine.applies_toc C s | j r
dndd | j S )NZAllowZDisallowz: )r> r )r r
r
r rG � s zRuleLine.__str__N)rH rI rJ rK r r= rG r
r
r
r r2 � s r2 c @ s0 e Zd ZdZdd� Zdd� Zdd� Zdd � Zd
S ) r* z?An entry has one or more user-agents and zero or more rulelinesc C s g | _ g | _d | _d | _d S )N)r" r1 r5 r6 )r r
r
r r � s zEntry.__init__c C s� g }x | j D ]}|�d|� �� q W | jd k r@|�d| j� �� | jd k rj| j}|�d|j� d|j� �� |�tt| j �� |�d� d�
|�S )Nz User-agent: z
Crawl-delay: zRequest-rate: r) r rC ) r" r# r5 r6 ZrequestsZseconds�extendrE rF r1 rD )r Zret�agentZrater
r
r rG � s
z
Entry.__str__c C sF |� d�d �� }x.| jD ]$}|dkr*dS |�� }||krdS qW dS )z2check if this entry applies to the specified agentr) r r! TF)r- r/ r" )r r? rO r
r
r r= � s zEntry.applies_toc C s$ x | j D ]}|�|�r|jS qW dS )zZPreconditions:
- our agent applies to this entry
- filename is URL decodedT)r1 r= r> )r rM r9 r
r
r r> � s
zEntry.allowanceN)rH rI rJ rK r rG r= r> r
r
r
r r* � s
r* ) rK � collectionsZ urllib.parser Zurllib.request�__all__�
namedtupler r r2 r* r
r
r
r �<module> s 6