�
z fc @ s} d Z d d l Z d d l Z d g Z d d d � � YZ d d d � � YZ d d
d � � YZ d e j f d
� � YZ d S( s& 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
i����Nt RobotFileParserc B sb e Z d Z d d � Z d � Z d � Z d � Z d � Z d � Z d � Z d � Z
d
� Z RS( ss This class provides a set of methods to read, parse and answer
questions about a single robots.txt file.
t c C s> g | _ d | _ t | _ t | _ | j | � d | _ d S( Ni ( t entriest Nonet
default_entryt Falset disallow_allt allow_allt set_urlt last_checked( t selft url( ( s# /usr/lib64/python2.7/robotparser.pyt __init__ s
c C s | j S( s� 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
( ( s# /usr/lib64/python2.7/robotparser.pyt mtime! s c C s d d l } | j � | _ d S( sY Sets the time the robots.txt file was last fetched to the
current time.
i����N( t timeR ( R
R ( ( s# /usr/lib64/python2.7/robotparser.pyt modified* s c C s/ | | _ t j | � d d !\ | _ | _ d S( s, Sets the URL referring to a robots.txt file.i i N( R t urlparset hostt path( R
R ( ( s# /usr/lib64/python2.7/robotparser.pyR 2 s c C s� t � } | j | j � } g | D] } | j � ^ q" } | j � | j | _ | j d k rk t | _ nO | j d k r� | j d k r� t | _ n% | j d k r� | r� | j | � n d S( s4 Reads the robots.txt URL and feeds it to the parser.i� i� i� i� i� N( i� i� (
t URLopenert openR t stript closet errcodet TrueR R t parse( R
t openert ft linet lines( ( s# /usr/lib64/python2.7/robotparser.pyt read7 s
c C sA d | j k r- | j d k r= | | _ q= n | j j | � d S( Nt *( t
useragentsR R R t append( R
t entry( ( s# /usr/lib64/python2.7/robotparser.pyt
_add_entryE s c C s& d } d } t � } | j � x�| D]�} | d 7} | s� | d k rZ t � } d } q� | d k r� | j | � t � } d } q� n | j d � } | d k r� | | } n | j � } | s� q& n | j d d � } t | � d k r& | d j � j � | d <t j | d j � � | d <| d d k ru| d k rX| j | � t � } n | j
j | d � d } q| d d k r�| d k r�| j j t
| d t � � d } q�q| d d k r| d k r�| j j t
| d t � � d } q�qq& q& W| d k r"| j | � n d S(
s� 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.i i i t #t :s
user-agentt disallowt allowN( t EntryR R# t findR t splitt lent lowert urllibt unquoteR R! t rulelinest RuleLineR R ( R
R t statet
linenumberR" R t i( ( s# /usr/lib64/python2.7/robotparser.pyR N sP
c C s� | j r
t S| j r t S| j s' t St j t j | � � } t j d d | j | j
| j | j f � } t j
| � } | s� d } n x- | j D]" } | j | � r� | j | � Sq� W| j r� | j j | � St S( s= using the parsed robots.txt decide if useragent can fetch urlR t /( R R R R R R R- R. t
urlunparseR t paramst queryt fragmentt quoteR t
applies_tot allowanceR ( R
t useragentR t
parsed_urlR" ( ( s# /usr/lib64/python2.7/robotparser.pyt can_fetch� s$ c C sE | j } | j d k r+ | | j g } n d j t t | � � d S( Ns
( R R R t joint mapt str( R
R ( ( s# /usr/lib64/python2.7/robotparser.pyt __str__� s ( t __name__t
__module__t __doc__R R
R R R R# R R> RB ( ( ( s# /usr/lib64/python2.7/robotparser.pyR s 4 R0 c B s) e Z d Z d � Z d � Z d � Z RS( so A rule line is a single "Allow:" (allowance==True) or "Disallow:"
(allowance==False) followed by a path.c C sS | d k r | r t } n t j t j | � � } t j | � | _ | | _ d S( NR ( R R R5 R- R9 R R; ( R
R R; ( ( s# /usr/lib64/python2.7/robotparser.pyR � s
c C s | j d k p | j | j � S( NR ( R t
startswith( R
t filename( ( s# /usr/lib64/python2.7/robotparser.pyR: � s c C s | j r d p d d | j S( Nt Allowt Disallows : ( R; R ( R
( ( s# /usr/lib64/python2.7/robotparser.pyRB � s ( RC RD RE R R: RB ( ( ( s# /usr/lib64/python2.7/robotparser.pyR0 � s R( c B s2 e Z d Z d � Z d � Z d � Z d � Z RS( s? An entry has one or more user-agents and zero or more rulelinesc C s g | _ g | _ d S( N( R R/ ( R
( ( s# /usr/lib64/python2.7/robotparser.pyR � s c C sj g } x' | j D] } | j d | d g � q Wx* | j D] } | j t | � d g � q: Wd j | � S( Ns User-agent: s
R ( R t extendR/ RA R? ( R
t rett agentR ( ( s# /usr/lib64/python2.7/robotparser.pyRB � s c C s] | j d � d j � } x= | j D]2 } | d k r9 t S| j � } | | k r# t Sq# Wt S( s2 check if this entry applies to the specified agentR4 i R ( R* R, R R R ( R
R< RL ( ( s# /usr/lib64/python2.7/robotparser.pyR: � s c C s. x' | j D] } | j | � r
| j Sq
Wt S( sZ Preconditions:
- our agent applies to this entry
- filename is URL decoded( R/ R: R; R ( R
RG R ( ( s# /usr/lib64/python2.7/robotparser.pyR; � s ( RC RD RE R RB R: R; ( ( ( s# /usr/lib64/python2.7/robotparser.pyR( � s
R c B s# e Z d � Z d � Z d � Z RS( c G s t j j | | � d | _ d S( Ni� ( R- t FancyURLopenerR R ( R
t args( ( s# /usr/lib64/python2.7/robotparser.pyR � s c C s d S( N( NN( R ( R
R t realm( ( s# /usr/lib64/python2.7/robotparser.pyt prompt_user_passwd� s c C s( | | _ t j j | | | | | | � S( N( R R- RM t http_error_default( R
R t fpR t errmsgt headers( ( s# /usr/lib64/python2.7/robotparser.pyRQ � s ( RC RD R RP RQ ( ( ( s# /usr/lib64/python2.7/robotparser.pyR � s ( ( ( ( RE R R- t __all__R R0 R( RM R ( ( ( s# /usr/lib64/python2.7/robotparser.pyt <module> s �$