Your IP : 18.188.218.219
U
i�f�$ � @ sX d dl Z d dlZd dlZdgZe �dd�ZG dd� d�ZG dd� d�ZG dd � d �Z dS )
� N�RobotFileParser� RequestRatezrequests secondsc @ sn e Zd Zddd�Zdd� Zdd� Zdd � Zd
d � Zd d
� Zdd� Z dd� Z
dd� Z dd� Z dd� Z
dd� ZdS ) r � c C s2 g | _ g | _d | _d| _d| _| �|� d| _d S )NFr )�entries�sitemaps�
default_entry� disallow_all� allow_all�set_url� last_checked��self�url� r �7/opt/alt/python38/lib64/python3.8/urllib/robotparser.py�__init__ s
zRobotFileParser.__init__c C s | j S �N)r �r
r r r �mtime% s zRobotFileParser.mtimec C s dd l }|� � | _d S )Nr )�timer )r
r r r r �modified. s zRobotFileParser.modifiedc C s& || _ tj�|�dd� \| _| _d S )N� � )r �urllib�parse�urlparseZhost�pathr r r r r
6 s zRobotFileParser.set_urlc
C s� zt j�| j�}W nR t jjk
rd } z0|jdkr:d| _n|jdkrT|jdk rTd| _W 5 d }~X Y n X |� � }| �
|� d�� � � d S )N)i� i� Ti� i� zutf-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_entryH s
zRobotFileParser._add_entryc C sP d}t � }| �� |D �] }|sP|dkr4t � }d}n |dkrP| �|� t � }d}|�d�}|dkrn|d |� }|�� }|s|q|�dd�}t|�dkr|d �� �� |d< tj �
|d �� �|d< |d dkr�|dkr�| �|� t � }|j � |d � d}q|d dk�r.|dk�r6|j
� t|d d�� d}q|d d k�rb|dk�r6|j
� t|d d
�� d}q|d d k�r�|dk�r6|d �� �� �r�t|d �|_d}q|d d k�r|dk�r6|d �d
�}t|�dk�r|d �� �� �r|d �� �� �rtt|d �t|d ��|_d}q|d dkr| j� |d � q|dk�rL| �|� d S )Nr r � �#�:z
user-agentZdisallowFZallowTz crawl-delayz request-rate�/Zsitemap)�Entryr r* �find�strip�split�len�lowerr r �unquoter' r( � rulelines�RuleLine�isdigit�int�delayr �req_rater )r
�lines�stater) �line�iZnumbersr r r r Q sj
�
zRobotFileParser.parsec C s� | j r
dS | jrdS | js dS tj�tj�|��}tj�dd|j|j |j
|j f�}tj� |�}|sfd}| j
D ] }|�|�rl|�|� S ql| jr�| j�|�S dS )NFTr r. )r r r r r r r5 �
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 | jD ]}|�|�r|j S q| jr:| jjS d S r )r r rB r: r �r
rD r) r r r � crawl_delay� s
zRobotFileParser.crawl_delayc C s> | � � s d S | jD ]}|�|�r|j S q| jr:| jjS d S r )r r rB r; r rF r r r � request_rate� s
z RobotFileParser.request_ratec C s | j s
d S | j S r )r r r r r � site_maps� s zRobotFileParser.site_mapsc C s, | j }| jd k r || jg }d�tt|��S )Nz
)r r �join�map�str)r
r r r r �__str__� s
zRobotFileParser.__str__N)r )�__name__�
__module__� __qualname__r r r r
r r* r rE rG rH rI rM r r r r r s
I
c @ s$ e Zd Zdd� Zdd� Zdd� ZdS )r7 c C s< |dkr|sd}t j�t j�|��}t j�|�| _|| _d S )Nr T)r r r@ r rA r rC )r
r rC r r r r � s
zRuleLine.__init__c C s | j dkp|�| j �S r% )r �
startswith)r
�filenamer r r rB � s zRuleLine.applies_toc C s | j r
dndd | j S )NZAllowZDisallowz: )rC r r r r r rM � s zRuleLine.__str__N)rN rO rP r rB rM r r r r r7 � s r7 c @ s, e Zd Zdd� Zdd� Zdd� Zdd� Zd S )
r/ c C s g | _ g | _d | _d | _d S r )r' r6 r: r; r r r r r � s zEntry.__init__c C s� g }| j D ]}|�d|� �� q
| jd k r<|�d| j� �� | jd k rf| j}|�d|j� d|j� �� |�tt| j �� d�
|�S )Nz User-agent: z
Crawl-delay: zRequest-rate: r. �
) r' r( r: r; ZrequestsZseconds�extendrK rL r6 rJ )r
Zret�agentZrater r r rM � s
z
Entry.__str__c C sF |� d�d �� }| jD ](}|dkr* dS |�� }||kr dS qdS )Nr. r r& TF)r2 r4 r' )r
rD rU r r r rB � s
zEntry.applies_toc C s$ | j D ]}|�|�r|j S qdS )NT)r6 rB rC )r
rR r> r r r rC
s
zEntry.allowanceN)rN rO rP r rM rB rC r r r r r/ � s
r/ )
� collectionsZ urllib.parser Zurllib.request�__all__�
namedtupler r r7 r/ r r r r �<module>
s B