File: //lib64/python3.9/urllib/__pycache__/robotparser.cpython-39.opt-2.pyc
a
�DOg�$ � @ 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 �*/usr/lib64/python3.9/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�urlparse�host�pathr r r r r
6 s zRobotFileParser.set_urlc
C s� zt j�| j�}W nT t jjyf } z8|jdv r8d| _n|jdkrR|jdk rRd| _W Y d }~n&d }~0 0 |� � }| �
|� d�� � � d S )N)i� i� Ti� i� zutf-8)
r Zrequest�urlopenr �error� HTTPError�coder r �readr �decode�
splitlines)r
�f�err�rawr r r r" ; s
zRobotFileParser.readc C s, d|j v r | jd u r(|| _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 r1 )r r r r r r r8 �
urlunparser �params�query�fragment�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 rH r= r �r
rJ 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 rH r> r rL 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 ur || 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 rK rM rN rO rS r r r r r s
I
c @ s$ e Zd Zdd� Zdd� Zdd� ZdS )r: c C s< |dkr|sd}t j�t j�|��}t j�|�| _|| _d S )Nr T)r r rC r rG r rI )r
r rI r r r r � s
zRuleLine.__init__c C s | j dkp|�| j �S r( )r �
startswith)r
�filenamer r r rH � s zRuleLine.applies_toc C s | j r
dndd | j S )NZAllowZDisallowz: )rI r r r r r rS � s zRuleLine.__str__N)rT rU rV r rH rS r r r r r: � s r: c @ s, e Zd Zdd� Zdd� Zdd� Zdd� Zd S )
r2 c C s g | _ g | _d | _d | _d S r )r* r9 r= r> r r r r r � s zEntry.__init__c C s� g }| j D ]}|�d|� �� q
| jd ur<|�d| j� �� | jd urf| j}|�d|j� d|j� �� |�tt| j �� d�
|�S )Nz User-agent: z
Crawl-delay: zRequest-rate: r1 �
) r* r+ r= r> ZrequestsZseconds�extendrQ rR r9 rP )r
�ret�agentZrater r r rS � s
z
Entry.__str__c C sF |� d�d �� }| jD ](}|dkr* dS |�� }||v r dS qdS )Nr1 r r) TF)r5 r7 r* )r
rJ r\ r r r rH � s
zEntry.applies_toc C s$ | j D ]}|�|�r|j S qdS )NT)r9 rH rI )r
rX rA r r r rI
s
zEntry.allowanceN)rT rU rV r rS rH rI r r r r r2 � s
r2 )
� collections� urllib.parser Zurllib.request�__all__�
namedtupler r r: r2 r r r r �<module>
s B