ó õ^‚Rc@s}dZddlZddlZdgZdd d„ƒYZdd d„ƒYZdd d„ƒYZd ejfd „ƒYZdS(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://info.webcrawler.com/mak/projects/robots/norobots-rfc.html iÿÿÿÿNtRobotFileParsercBsbeZdZdd„Zd„Zd„Zd„Zd„Zd„Zd„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. tcCs>g|_d|_t|_t|_|j|ƒd|_dS(Ni(tentriestNonet default_entrytFalset disallow_allt allow_alltset_urlt last_checked(tselfturl((s!/usr/lib/python2.7/robotparser.pyt__init__s      cCs|jS(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/lib/python2.7/robotparser.pytmtime scCsddl}|jƒ|_dS(sYSets the time the robots.txt file was last fetched to the current time. iÿÿÿÿN(ttimeR (R R((s!/usr/lib/python2.7/robotparser.pytmodified)s cCs/||_tj|ƒdd!\|_|_dS(s,Sets the URL referring to a robots.txt file.iiN(R turlparsethosttpath(R R ((s!/usr/lib/python2.7/robotparser.pyR1s cCs¯tƒ}|j|jƒ}g|D]}|jƒ^q"}|jƒ|j|_|jdkrkt|_n@|jdkr†t|_n%|jdkr«|r«|j |ƒndS(s4Reads the robots.txt URL and feeds it to the parser.i‘i“iiÈN(i‘i“( t URLopenertopenR tstriptcloseterrcodetTrueRRtparse(R topenertftlinetlines((s!/usr/lib/python2.7/robotparser.pytread6s     cCsAd|jkr-|jdkr=||_q=n|jj|ƒdS(Nt*(t useragentsRRRtappend(R tentry((s!/usr/lib/python2.7/robotparser.pyt _add_entryDscCsd}d}tƒ}xä|D]Ü}|d7}|s~|dkrPtƒ}d}q~|dkr~|j|ƒtƒ}d}q~n|jdƒ}|dkr¦|| }n|jƒ}|s¾qn|jddƒ}t|ƒdkr|djƒjƒ|dRA(((s!/usr/lib/python2.7/robotparser.pyRs     3 R0cBs)eZdZd„Zd„Zd„ZRS(soA rule line is a single "Allow:" (allowance==True) or "Disallow:" (allowance==False) followed by a path.cCs;|dkr| rt}ntj|ƒ|_||_dS(NR(RR-R9RR;(R RR;((s!/usr/lib/python2.7/robotparser.pyR Ÿs cCs|jdkp|j|jƒS(NR(Rt startswith(R tfilename((s!/usr/lib/python2.7/robotparser.pyR:¦scCs|jrdpdd|jS(NtAllowtDisallows: (R;R(R ((s!/usr/lib/python2.7/robotparser.pyRA©s(RBRCRDR R:RA(((s!/usr/lib/python2.7/robotparser.pyR0œs  R(cBs2eZdZd„Zd„Zd„Zd„ZRS(s?An entry has one or more user-agents and zero or more rulelinescCsg|_g|_dS(N(R R/(R ((s!/usr/lib/python2.7/robotparser.pyR ¯s cCsjg}x'|jD]}|jd|dgƒqWx*|jD]}|jt|ƒdgƒq:Wdj|ƒS(Ns User-agent: s R(R textendR/R@R?(R trettagentR((s!/usr/lib/python2.7/robotparser.pyRA³s cCs]|jdƒdjƒ}x=|jD]2}|dkr9tS|jƒ}||kr#tSq#WtS(s2check if this entry applies to the specified agentR4iR(R*R,R RR(R R<RK((s!/usr/lib/python2.7/robotparser.pyR:»s   cCs.x'|jD]}|j|ƒr |jSq WtS(sZPreconditions: - our agent applies to this entry - filename is URL decoded(R/R:R;R(R RFR((s!/usr/lib/python2.7/robotparser.pyR;Ès (RBRCRDR RAR:R;(((s!/usr/lib/python2.7/robotparser.pyR(­s    RcBs#eZd„Zd„Zd„ZRS(cGs tjj||Œd|_dS(NiÈ(R-tFancyURLopenerR R(R targs((s!/usr/lib/python2.7/robotparser.pyR ÒscCsdS(N(NN(R(R Rtrealm((s!/usr/lib/python2.7/robotparser.pytprompt_user_passwdÖscCs(||_tjj||||||ƒS(N(RR-RLthttp_error_default(R R tfpRterrmsgtheaders((s!/usr/lib/python2.7/robotparser.pyRPÛs (RBRCR RORP(((s!/usr/lib/python2.7/robotparser.pyRÑs  (((( RDRR-t__all__RR0R(RLR(((s!/usr/lib/python2.7/robotparser.pyt s   Š$