a  ze$@sXddlZddlZddlZdgZeddZGdddZGdddZGdd d Z dS) NRobotFileParser RequestRatezrequests secondsc@sneZdZdddZddZddZdd Zd d Zd d ZddZ ddZ ddZ ddZ ddZ ddZdS)rcCs2g|_g|_d|_d|_d|_||d|_dS)NFr)entriessitemaps default_entry disallow_all allow_allset_url last_checkedselfurlrF/opt/bitninja-python-dojo/embedded/lib/python3.9/urllib/robotparser.py__init__s zRobotFileParser.__init__cCs|jSN)r r rrrmtime%szRobotFileParser.mtimecCsddl}||_dS)Nr)timer )r rrrrmodified.szRobotFileParser.modifiedcCs&||_tj|dd\|_|_dS)N)rurllibparseurlparsehostpathr rrrr 6szRobotFileParser.set_urlc Csztj|j}WnTtjjyf}z8|jdvr8d|_n|jdkrR|jdkrRd|_WYd}~n&d}~00| }| | d dS)N)iiTiizutf-8) rZrequesturlopenrerror HTTPErrorcoderr readrdecode splitlines)r ferrrawrrrr";s zRobotFileParser.readcCs,d|jvr|jdur(||_n |j|dSN*) useragentsrrappend)r entryrrr _add_entryHs  zRobotFileParser._add_entrycCsPd}t}||D]}|sP|dkr4t}d}n|dkrP||t}d}|d}|dkrn|d|}|}|s|q|dd}t|dkr|d|d<tj |d|d<|ddkr|dkr||t}|j |dd}q|ddkr.|dkr6|j t|ddd}q|dd krb|dkr6|j t|dd d}q|dd kr|dkr6|drt|d|_d}q|dd kr|dkr6|dd }t|dkr|dr|drtt|dt|d|_d}q|ddkr|j |dq|dkrL||dS)Nrr#:z user-agentZdisallowFZallowTz crawl-delayz request-rate/Zsitemap)Entryrr-findstripsplitlenlowerrrunquoter*r+ rulelinesRuleLineisdigitintdelayrreq_rater)r linesstater,lineiZnumbersrrrrQsj                zRobotFileParser.parsecCs|jr dS|jrdS|jsdStjtj|}tjdd|j|j |j |j f}tj |}|sfd}|j D]}||rl||Sql|jr|j|SdS)NFTrr1)rr r rrrr8 urlunparserparamsqueryfragmentquoter applies_to allowancer)r useragentrZ parsed_urlr,rrr can_fetchs&     zRobotFileParser.can_fetchcCs>|s dS|jD]}||r|jSq|jr:|jjSdSr)rrrHr=rr rJr,rrr crawl_delays   zRobotFileParser.crawl_delaycCs>|s dS|jD]}||r|jSq|jr:|jjSdSr)rrrHr>rrLrrr request_rates   zRobotFileParser.request_ratecCs|js dS|jSr)rrrrr site_mapsszRobotFileParser.site_mapscCs,|j}|jdur||jg}dtt|S)Nz )rrjoinmapstr)r rrrr__str__s  zRobotFileParser.__str__N)r)__name__ __module__ __qualname__rrrr r"r-rrKrMrNrOrSrrrrrs    I  c@s$eZdZddZddZddZdS)r:cCs<|dkr|sd}tjtj|}tj||_||_dS)NrT)rrrCrrGrrI)r rrIrrrrs  zRuleLine.__init__cCs|jdkp||jSr()r startswith)r filenamerrrrHszRuleLine.applies_tocCs|jr dndd|jS)NZAllowZDisallowz: )rIrrrrrrSszRuleLine.__str__N)rTrUrVrrHrSrrrrr:sr:c@s,eZdZddZddZddZddZd S) r2cCsg|_g|_d|_d|_dSr)r*r9r=r>rrrrrszEntry.__init__cCsg}|jD]}|d|q |jdur<|d|j|jdurf|j}|d|jd|j|tt|j d |S)Nz User-agent: z Crawl-delay: zRequest-rate: r1 ) r*r+r=r>ZrequestsZsecondsextendrQrRr9rP)r retagentZraterrrrSs   z Entry.__str__cCsF|dd}|jD](}|dkr*dS|}||vrdSqdS)Nr1rr)TF)r5r7r*)r rJr\rrrrHs zEntry.applies_tocCs$|jD]}||r|jSqdS)NT)r9rHrI)r rXrArrrrI s   zEntry.allowanceN)rTrUrVrrSrHrIrrrrr2s  r2) collections urllib.parserZurllib.request__all__ namedtuplerrr:r2rrrr s B