"htJzUddlmZddlZddlZddlmZddlmZddl m Z ddl m Z m Z mZmZmZmZddlmZmZdd lmZmZdd lmZmZmZmZdd lmZdd lm Z dd l!m"Z"dZ#dZ$dZ%dZ&ia'e e(e"fe)d<de e(e"ffdZ*de(de"ddfdZ+de(ddfdZ,edd d7de-de(de-fdZ.de/fd Z0 d8de-d!e/d"eede e(e"ffd#Z1Gd$d%eZ2Gd&d'e2Z3Gd(d)e2Z4dddde3fd*ee(efde-d+e-d,e-d-e-d"eedeee e(e/efeddefffd.Z5 d9d*ee(efde-d+e-d,e-d-e-d"eedee(fd/Z6 d:d*ee(efde-d0e-d+e-d,e-d-e-d"eedeee(e ffd1Z7 d9d*ee(efde-d+e-d,e-d-e-d"eedeed2ee(e(e(fffd3Z8 d;d4ee(efd,e-d-e-d"eede-f d5Z9deZdZe ddededeeee ffdZ y)rr8rVr)c|dkDr |rytt}|j|vr||j|S t|jr |j}nt |j}t |dd}t }|j}d}|j} |D]F} d| vrd}|r| sn:d | vr| jd} | d d vr.|| j| Ht|j||j |j|S#t$rM|j| |dz }|j||cY jS#t $rYSwxYwt $r.} |r%Yd} ~ jy#t $rYywxYw| d} ~ wwxYw#t $rY|SwxYw# jw#t $rYwwxYwxYw)z[Parse. :param fail_silently: :param retry_count: :return: rNrutf8)encodingFz===BEGIN PRIVATE DOMAINS===Tz// xn--r)/ )privater;rY)rr% local_pathrr codecs_openraddinclude_privatesplitstripr(closeIOErrorr&r Exception) clsr8rV _tld_namesrb local_filetrietrie_addprivate_sectionrelineerrs r.rz(BaseMozillaTLDSourceParser.get_tld_namess ?   NNj (3>>*6 2 S^^$ ^^ (8 $ZvFJ6DxxH#O!11O" E0D8&*O"?$::<?D7k)DJJL>OD E 's~~t <        ")   } = 1 K$$+%    "      "          "  sCD#F1#3F.GE)) E54E58F.F)GF F$#F$'F))F..G1 F>=F>G#GG# G G#G  G#N)Fr) __name__ __module__ __qualname__ classmethodboolrSr rstrrrr,r-r.rrsA=>N N7:N $sDy/ "NNr-rc<eZdZUdZdZeed<dZeed<dZeed<y) rMozilla TLD source.mozillauidz4https://publicsuffix.org/list/public_suffix_list.datr>zres/effective_tld_names.dat.txtrbN) rsrtru__doc__r|rx__annotations__r>rbr,r-r.rrs#CLJL7J7r-rcJeZdZUdZdZeed<dZeed<dZeed<dZ e ed <y ) r rzmozilla_public_onlyr|z?https://publicsuffix.org/list/public_suffix_list.dat?publiconlyr>z+res/effective_tld_names_public_only.dat.txtrbFreN) rsrtrur}r|rxr~r>rbrerwr,r-r.r r s5$C$IDJC!OT!r-r url fix_protocol search_publicsearch_privatec|s |s tdt||}t|ts%|r|j dsd|} t |}n|}|j} | s|rdd|fSt|| j} | jdr| jd} | jd} |j} || j} d} d}d}t| }t!|d z d d D]z}| |}| j"ng|| j$k(rnV| j"j'|}|| j"j'd }|n| d z } |} | j(sw| }| }||(|j(r|s |j*r|s|j*r|rdd|fSt-| ||k(rd }nt/d ||z }| ||fS#t $r}|r dd|fcYd}~S|d}~wwxYw) zProcess URL. :param parser_class: :param url: :param fail_silently: :param fix_protocol: :param search_public: :param search_private: :return: zJEither `search_public` or `search_private` (or both) shall be set to True.)r8rW)z//zhttp://https://rN)r.rr*) domain_name)rr isinstancer startswithr ValueErrorhostnamerlowerendswithrstriprfrbrootlenrangechildren exceptionr=leafrarmax)rr8rrrrWrl parsed_urler domain_partsr/nodecurrent_length tld_lengthmatchlen_domain_partsipartchild non_zero_is r.r#r#s:$ ^%   #,J c; ' /L MSE"C !#J %%K  z) )$ $##%KC !((- $$S)L'22 * + 0 0DNJ E<( #a'R 0A ==  4>> !  !!$' =MM%%c*E =  ! 99'JE7>  emm5== z) )# < <:% ,z9: Z //_ T3&  s$ G G&G!G&G!!G&c d|vr td|s|rtnt}t||||||\}}} |y|J|dkr | jSdj ||dz dS)a?Extract the first level domain. Extract the top level domain based on the mozilla's effective TLD names dat file. Returns a string. May throw ``TldBadUrl`` or ``TldDomainNotFound`` exceptions if there's bad URL provided or no TLD match found respectively. :param url: URL to get top level domain from. :param fail_silently: If set to True, no exceptions are raised and None is returned on failure. :param fix_protocol: If set to True, missing or wrong protocol is ignored (https is appended instead). :param search_public: If set to True, search in public domains. :param search_private: If set to True, search in private domains. :param parser_class: :type url: str | SplitResult :type fail_silently: bool :type fix_protocol: bool :type search_public: bool :type search_private: bool :return: String with top level domain (if ``as_object`` argument is set to False) or a ``tld.utils.Result`` object (if ``as_object`` argument is set to True); returns None on failure. :rtype: str as_objectzH`as_object` argument is deprecated for `get_fld`. Use `get_tld` instead.rr8rrrrWNrrr)rrr r#rjoin) rr8rrrrWkwargsrrrs r.rrsDf%    #1  ,7 #!#%! ,(L*j  !! !A~""" 88La!12 33r-rc|s|rtnt}t||||||\}}} |y|J|s%|dkr | jSdj ||dS|dkr$d} d} | jJd| j} nBdj |d|dz } dj ||dz |} dj ||d} t | | | | S) aExtract the top level domain. Extract the top level domain based on the mozilla's effective TLD names dat file. Returns a string. May throw ``TldBadUrl`` or ``TldDomainNotFound`` exceptions if there's bad URL provided or no TLD match found respectively. :param url: URL to get top level domain from. :param fail_silently: If set to True, no exceptions are raised and None is returned on failure. :param as_object: If set to True, ``tld.utils.Result`` object is returned, ``domain``, ``suffix`` and ``tld`` properties. :param fix_protocol: If set to True, missing or wrong protocol is ignored (https is appended instead). :param search_public: If set to True, search in public domains. :param search_private: If set to True, search in private domains. :param parser_class: :type url: str | SplitResult :type fail_silently: bool :type as_object: bool :type fix_protocol: bool :type search_public: bool :type search_private: bool :return: String with top level domain (if ``as_object`` argument is set to False) or a ``tld.utils.Result`` object (if ``as_object`` argument is set to True); returns None on failure. :rtype: str rNrrzNo hostname in URLr) subdomaindomaintldr)rr r#rrr) rr8rrrrrWrrrrr_tlds r.rrsJ  #1  ,7 #!#%! ,(L*j  !! !  >&& &xx Z[122A~ "".D0DD.""HH\*:JN;< ,zA~ CDxx Z[12 F r-NNNc |s|rtnt} t||d||||}|y|j|j|j fS#t tttf$rYywxYw)zParse TLD into parts. :param url: :param fail_silently: :param fix_protocol: :param search_public: :param search_private: :param parser_class: :return: Tuple (tld, domain, subdomain) :rtype: tuple T)r8rrrrrWr) rr rrrrrrrr)rr8rrrrWobjs r.r!r!&s&  #1    '%')%  ;#ww CMM11 (*A: N   sA "A A&%A&valuecN|s|rtnt}t|dd|||}||k(S)agCheck if given URL is tld. :param value: URL to get top level domain from. :param search_public: If set to True, search in public domains. :param search_private: If set to True, search in private domains. :param parser_class: :type value: str :type search_public: bool :type search_private: bool :return: :rtype: bool Tr)rr r)rrrrWrs r.rrUsC$  #1   #%!  D D=r-c$|r t|yiay)zReset the ``tld_names`` to empty value. If ``tld_names_local_path`` is given, removes specified entry from ``tld_names`` instead. :param tld_names_local_path: :type tld_names_local_path: str :return: N)r"r%r4s r.r$r$ys 45 r-)FN)FrN)FFTTN)FFFTTN)TTN)N); __future__rrMrQcodecsrrc functoolsros.pathrtypingrrr r r r urllib.parser rbaserr exceptionsrrrrhelpersrresultrrnr __author__ __copyright__ __license____all__r%rxr~rr(r"rwr&rSr'rrrr r#rrr!rr$r,r-r.rs' &;;./ !  , <  ( 4T ?c4i / /)- /  /.#.$. 3d#37-0 $6c: .2*+ #t)_ >P!4Pf878"'A"$ .D s0 sK s0s0s0 s0  s0 *+ s0 5cC, -uT45L/M MNs0p .2 B4 sK B4B4B4 B4  B4 *+ B4c]B4N .2Q sK QQQ Q  Q  Q*+QeCK !Ql .2 , sK ,,, ,  , *+ , 5! "E#sC-$8 89,b.2 ! k! "!!!*+ !  !H#r-