­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ó ÇaWc@sdZddlZddlZddlZejddkrLeZeZn e Ze Zd„Z de fd„ƒYZ de efd „ƒYZd efd „ƒYZd efd „ƒYZdefd„ƒYZdefd„ƒYZdefd„ƒYZdefd„ƒYZdefd„ƒYZdefd„ƒYZdefd„ƒYZdefd„ƒYZejdƒZejdƒZejd ƒZd!„Zd"„Z d#„Z!e"d$„Z#d%„Z$d&„Z%d'„Z&d(e'fd)„ƒYZ(d*e(fd+„ƒYZ)d,dBd-„ƒYZ*d.„Z+e+d/ƒZ,e+d0ƒZ-e+d1ƒZ.e+d2ƒZ/ie+d3ƒd46e+d5ƒd66Z0ejd7ƒj1Z2eje*j3ej4ƒj1Z5ejd8ƒj1Z6e7ed9ƒrÖej8d:d;ƒZ9n d<„Z9d=„Z:d>„Z;d?„Z<d@efdA„ƒYZ=dS(Cs cssselect.parser ================ Tokenizer, parser and parsed objects for CSS selectors. :copyright: (c) 2007-2012 Ian Bicking and contributors. See AUTHORS for more details. :license: BSD, see LICENSE for more details. iÿÿÿÿNiicCs|jdƒjƒjdƒS(s(Lower-case, but only in the ASCII range.tutf8(tencodetlowertdecode(tstring((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyt ascii_lowerst SelectorErrorcBseZdZRS(sáCommon parent for :class:`SelectorSyntaxError` and :class:`ExpressionError`. You can just use ``except SelectorError:`` when calling :meth:`~GenericTranslator.css_to_xpath` and handle both exceptions types. (t__name__t __module__t__doc__(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR!stSelectorSyntaxErrorcBseZdZRS(s3Parsing a selector that does not match the grammar.(RRR (((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR *stSelectorcBs,eZdZdd„Zd„Zd„ZRS(s" Represents a parsed selector. :meth:`~GenericTranslator.selector_to_xpath` accepts this object, but ignores :attr:`pseudo_element`. It is the user’s responsibility to account for pseudo-elements and reject selectors with unknown or unsupported pseudo-elements. cCsA||_|dk r4t|tƒ r4t|ƒ}n||_dS(N(t parsed_treetNonet isinstancetFunctionalPseudoElementRtpseudo_element(tselfttreeR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyt__init__:s   cCs]t|jtƒr$t|jƒ}n|jr=d|j}nd}d|jj|j|fS(Ns::%sts%s[%r%s](RRRtreprt __class__RR (RR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyt__repr__Us cCs;|jjƒ\}}}|jr.|d7}n|||fS(s’Return the specificity_ of this selector as a tuple of 3 integers. .. _specificity: http://www.w3.org/TR/selectors/#specificity i(R t specificityR(Rtatbtc((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR_s  N(RRR R RRR(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR 0s   tClasscBs)eZdZd„Zd„Zd„ZRS(s( Represents selector.class_name cCs||_||_dS(N(tselectort class_name(RRR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRos cCsd|jj|j|jfS(Ns %s[%r.%s](RRRR(R((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRsscCs/|jjƒ\}}}|d7}|||fS(Ni(RR(RRRR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRws (RRR RRR(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRks  RcBs2eZdZd„Zd„Zd„Zd„ZRS(sx Represents selector::name(arguments) .. attribute:: name The name (identifier) of the pseudo-element, as a string. .. attribute:: arguments The arguments of the pseudo-element, as a list of tokens. **Note:** tokens are not part of the public API, and may change between cssselect versions. Use at your own risks. cCst|ƒ|_||_dS(N(Rtnamet arguments(RRR ((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRŽscCs3d|jj|jg|jD]}|j^qfS(Ns %s[::%s(%r)](RRRR tvalue(Rttoken((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR’scCsg|jD]}|j^q S(N(R ttype(RR"((s4/usr/lib/python2.7/site-packages/cssselect/parser.pytargument_types—scCs/|jjƒ\}}}|d7}|||fS(Ni(RR(RRRR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRšs (RRR RRR$R(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR}s    tFunctioncBs2eZdZd„Zd„Zd„Zd„ZRS(s( Represents selector:name(expr) cCs%||_t|ƒ|_||_dS(N(RRRR (RRRR ((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR¤s cCs9d|jj|j|jg|jD]}|j^q"fS(Ns %s[%r:%s(%r)](RRRRR R!(RR"((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR©scCsg|jD]}|j^q S(N(R R#(RR"((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR$®scCs/|jjƒ\}}}|d7}|||fS(Ni(RR(RRRR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR±s (RRR RRR$R(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR% s    tPseudocBs)eZdZd„Zd„Zd„ZRS(s# Represents selector:ident cCs||_t|ƒ|_dS(N(RRtident(RRR'((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR»s cCsd|jj|j|jfS(Ns %s[%r:%s](RRRR'(R((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR¿scCs/|jjƒ\}}}|d7}|||fS(Ni(RR(RRRR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRÃs (RRR RRR(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR&·s  tNegationcBs)eZdZd„Zd„Zd„ZRS(s. Represents selector:not(subselector) cCs||_||_dS(N(Rt subselector(RRR)((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRÍs cCsd|jj|j|jfS(Ns%s[%r:not(%r)](RRRR)(R((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRÑscCsI|jjƒ\}}}|jjƒ\}}}||||||fS(N(RRR)(Rta1tb1tc1ta2tb2tc2((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRÕs(RRR RRR(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR(És  tAttribcBs)eZdZd„Zd„Zd„ZRS(s> Represents selector[namespace|attrib operator value] cCs1||_||_||_||_||_dS(N(Rt namespacetattribtoperatorR!(RRR1R2R3R!((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRßs     cCs~|jr"d|j|jf}n |j}|jdkrTd|jj|j|fSd|jj|j||j|jfSdS(Ns%s|%stexistss %s[%r[%s]]s%s[%r[%s %s %r]](R1R2R3RRRR!(RR2((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRæs  cCs/|jjƒ\}}}|d7}|||fS(Ni(RR(RRRR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRós (RRR RRR(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR0Ûs  tElementcBs/eZdZddd„Zd„Zd„ZRS(sU Represents namespace|element `None` is for the universal selector '*' cCs||_||_dS(N(R1telement(RR1R6((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRs cCsB|jp d}|jr.d|j|f}nd|jj|fS(Nt*s%s|%ss%s[%s](R6R1RR(RR6((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRs cCs|jr dSdSdS(Nii(iii(iii(R6(R((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR s N(RRR R RRR(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR5ùs tHashcBs)eZdZd„Zd„Zd„ZRS(s Represents selector#id cCs||_||_dS(N(Rtid(RRR9((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRs cCsd|jj|j|jfS(Ns %s[%r#%s](RRRR9(R((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRscCs/|jjƒ\}}}|d7}|||fS(Ni(RR(RRRR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRs (RRR RRR(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR8s  tCombinedSelectorcBs#eZd„Zd„Zd„ZRS(cCs1|dk st‚||_||_||_dS(N(R tAssertionErrorRt combinatorR)(RRR<R)((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR$s  cCsA|jdkrd}n |j}d|jj|j||jfS(Nt s s %s[%r %s %r](R<RRRR)(Rtcomb((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR*s   cCsI|jjƒ\}}}|jjƒ\}}}||||||fS(N(RRR)(RR*R+R,R-R.R/((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR2s(RRRRR(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR:#s  s%^[ \t\r\n\f]*([a-zA-Z]+)[ \t\r\n\f]*$s6^[ \t\r\n\f]*([a-zA-Z]*)#([a-zA-Z0-9_-]+)[ \t\r\n\f]*$s?^[ \t\r\n\f]*([a-zA-Z]*)\.([a-zA-Z][a-zA-Z0-9_-]*)[ \t\r\n\f]*$cCstj|ƒ}|r4ttd|jdƒƒƒgStj|ƒ}|dk r†tttd|jdƒpmdƒ|jdƒƒƒgStj|ƒ}|dk rØtt td|jdƒp¿dƒ|jdƒƒƒgSt t |ƒƒ}||_ t t|ƒƒS(sÀParse a CSS *group of selectors*. If you don't care about pseudo-elements or selector specificity, you can skip this and use :meth:`~GenericTranslator.css_to_xpath`. :param css: A *group of selectors* as an Unicode string. :raises: :class:`SelectorSyntaxError` on invalid selectors. :returns: A list of parsed :class:`Selector` objects, one for each selector in the comma-separated group. R6iiN(t_el_retmatchR R5tgroupt_id_reR R8t _class_reRt TokenStreamttokenizetsourcetlisttparse_selector_group(tcssR@tstream((s4/usr/lib/python2.7/site-packages/cssselect/parser.pytparseEs ! ! ccsP|jƒx?tt|ƒŒV|jƒdkrG|jƒ|jƒq Pq WdS(NtDELIMt,(RLRM(tskip_whitespaceR tparse_selectortpeektnext(RJ((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRHos   cCsÃt|ƒ\}}x¤|jƒ|jƒ}|d d fkrAPn|rZtd|ƒ‚n|jdddƒr‹|jƒj}|jƒnd}t|ƒ\}}t|||ƒ}qW||fS( NtEOFRLRMs4Got pseudo-element ::%s not at the end of a selectort+t>t~R=(RRN(RLRM( tparse_simple_selectorRNRPR R tis_delimRQR!R:(RJtresultRRPR<t next_selector((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyROys"    c Cs |jƒt|jƒ}|jƒ}|jdks@|dkr«|jdkra|jƒj}n|jƒd}|jƒdkrœ|jƒ|jƒ}qµ|}d}n d}}t ||ƒ}d}x˜|jƒ}|jdks|j ddd d ƒs|r|dkrPn|r/t d |ƒ‚n|jd krYt ||jƒjƒ}qÍ|d kr‡|jƒt ||jƒƒ}qÍ|d!kr¯|jƒt||ƒ}qÍ|d"krN|jƒ|jƒd#kr'|jƒ|jƒ}|jƒd$krÍ|jƒt|t|ƒƒ}qÍqÍn|jƒ}|jƒd%krWt|ƒ}qÍn|jƒd&kr~t||ƒ}qÍn|jƒ|jƒ|jƒdkr3|r¹t dƒ‚nt|dtƒ\} } |jƒ} | rÿt d| | jfƒ‚n| d'kr!t d| fƒ‚nt|| ƒ}qat||t|ƒƒ}qÍt d|fƒ‚qÍWt|jƒ|kr–t d|jƒfƒ‚n||fS((NtIDENTRLR7t|tSRRRMRSRTRUt)s4Got pseudo-element ::%s not at the end of a selectortHASHt.t[t:t(s first-lines first-lettertbeforetaftertnotsGot nested :not()tinside_negations+Got pseudo-element ::%s inside :not() at %ssExpected ')', got %ssExpected selector, got %s(RLR7(RLR[(R\RR(RLR](RLR_(RLR`(RLRa(RLRa(RLRb(s first-lines first-letterRcRd(RLRb(RLR](RNtlentusedRPR#RQR!R tnext_ident_or_starR5RWR R8Rt next_identt parse_attribRtparse_argumentsRt_unicodeR&RVtTruetposR(R%( RJRftselector_startRPR1R6RXRR'targumenttargument_pseudo_elementRQ((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRV‘s’       '                  cCs{g}xn|jƒ|jƒ}|jd ks@|d d gkrP|j|ƒq |d kr`|Std|fƒ‚q WdS( NRZtSTRINGtNUMBERRLRSt-R]sExpected an argument, got %s(RZRsRt(RLRS(RLRu(RLR](RNRQR#tappendR (RJR RQ((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRläs   cCs|jƒ|jƒ}|dkrP|jƒdkrPtd|jƒfƒ‚n|jƒdkr²|jƒ|jƒdkr—d}|jƒd}q¼|}|jƒ}d}n d}}|dkru|jƒ|jƒ}|dkrt|||ddƒS|dkrd}qu|jdd d d dd ƒr_|jƒdkr_|j d}|jƒqutd |fƒ‚n|jƒ|jƒ}|j dkr°td|fƒ‚n|jƒ|jƒ}|dkrètd|fƒ‚nt|||||j ƒS(NRLR[sExpected '|', got %st=s|=t]R4t^t$R7RUt!sOperator expected, got %sRZRss Expected string or ident, got %ssExpected ']', got %s(RLR[(RLR[(RLRw(RLRx(RLRw(RLRw(RZRs(RLRx( RNRiR RPR RQRjR0RWR!R#(RRJR2R1topRQR!((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRkósN                     cCsx,|D]$}|jdkrtdƒ‚qqWdjd„|Dƒƒjƒ}|dkr^dS|dkrndS|d kr~dSd |kršd t|ƒfS|jd dƒ\}}|sÁd}n7|d ksÙ|d krìt|d ƒ}n t|ƒ}|sd }n t|ƒ}||fS(sv Parses the arguments for :nth-child() and friends. :raises: A list of tokens :returns: :``(a, b)`` Rss$String tokens not allowed in series.Rcss|]}|jVqdS(N(R!(t.0R"((s4/usr/lib/python2.7/site-packages/cssselect/parser.pys +stoddiitevenitnRuRSt1(ii(ii(ii(R#t ValueErrortjointstriptinttsplit(ttokensR"tsRR((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyt parse_series s,         tTokencBsMeZd„Zd„Zd„ZeejdƒƒZeejdƒƒZ RS(cCs%tj|||fƒ}||_|S(N(ttuplet__new__Ro(tclsttype_R!Rotobj((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRŒFs cCsd|j|j|jfS(Ns<%s '%s' at %i>(R#R!Ro(R((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRKscGs|jdko|j|kS(NRL(R#R!(Rtvalues((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRWNsii( RRRŒRRWtpropertyR3t itemgetterR#R!(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRŠEs    tEOFTokencBseZd„Zd„ZRS(cCstj|dd|ƒS(NRR(RŠRŒR (RRo((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRŒVscCsd|j|jfS(Ns <%s at %i>(R#Ro(R((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRYs(RRRŒR(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR“Us t TokenMacroscBsHeZdZedZdeZdZdeefZdeefZRS(s&\\([0-9a-f]{1,6})(?:\r\n|[ \n\r\t\f])?s|\\[^\n\r\f0-9a-f]s\\(?:\n|\r\n|\r|\f)|s [^\0-\177]s[_a-z0-9-]|%s|%ss [_a-z]|%s|%s(RRtunicode_escapetescapet string_escapetnonasciitnmchartnmstart(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyR”`s   cCs tj|ttƒtjƒjS(N(tretcompiletvarsR”t IGNORECASER@(tpattern((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyt_compilehss [ \t\r\n\f]+s[+-]?(?:[0-9]*\.[0-9]+|[0-9]+)s#(?:%(nmchar)s)+s -?(?:%(nmstart)s)(?:%(nmchar)s)*s!([^\n\r\f\\']|%(string_escape)s)*t's!([^\n\r\f\\"]|%(string_escape)s)*t"s\\(.)s\\(?:\n|\r\n|\r|\f)t methodcallerRAicCs |jdƒS(Ni(RA(R@((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyt_replace_simple}scCs:t|jdƒdƒ}|tjkr0d}nt|ƒS(Niiiýÿ(R…RAtsyst maxunicodet_unichr(R@t codepoint((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyt_replace_unicode€s cCs"tt|ƒ}tt|ƒ}|S(N(t_sub_unicode_escapeR©t_sub_simple_escapeR¤(R!((s4/usr/lib/python2.7/site-packages/cssselect/parser.pytunescape_ident‡sccsÉd}t|ƒ}x“||kr§t|d|ƒ}|r\tdd|ƒV|jƒ}qnt|d|ƒ}|rµtttt|j ƒƒƒ}td||ƒV|jƒ}qnt |d|ƒ}|rtttt|j ƒdƒƒ}td||ƒV|jƒ}qn||}|t krêt ||d|dƒ}|sTt dƒ‚|jƒ}||krt d |ƒ‚n|||kr¢t d |ƒ‚ntttttd |j ƒƒƒƒ}td ||ƒV|d}qnt|d|ƒ}|r1|j ƒ}td ||ƒV|jƒ}qn|d}|||!dkr…|jd|ƒ}|dkru|}q|d7}qntd|||ƒV|d7}qW||ksºt ‚t|ƒVdS(NiRoR\R=RZiR^s)Should have found at least an empty matchsUnclosed string at %ssInvalid string at %sRRsRtis/*s*/iÿÿÿÿRL(Rgt_match_whitespaceRŠtendt _match_identR«R¤RªR©RAt _match_hasht_match_string_by_quoteR;R t_sub_newline_escapet _match_numbertfindR“(RˆRotlen_sR@R!tquotetend_postpos2((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyREsl               RDcBsAeZdd„Zd„Zd„Zd„Zd„Zd„ZRS(cCsmg|_t|ƒ|_||_d|_t|_y|jj|_ Wn t k rh|jj |_ nXdS(N( RhtiterR‡RFR tpeekedtFalset_peekingRQt next_tokentAttributeErrort__next__(RR‡RF((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRÏs     cCsP|jr,t|_|jj|jƒ|jS|jƒ}|jj|ƒ|SdS(N(R¼R»RhRvRºR½(RRQ((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRQÛs   cCs+|js$|jƒ|_t|_n|jS(N(R¼R½RºRn(R((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRPås  cCs8|jƒ}|jdkr1td|fƒ‚n|jS(NRZsExpected ident, got %s(RQR#R R!(RRQ((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRjës cCsI|jƒ}|jdkr"|jS|dkr2dStd|fƒ‚dS(NRZRLR7sExpected ident or '*', got %s(RLR7(RQR#R!R R (RRQ((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRiñs  cCs,|jƒ}|jdkr(|jƒndS(NR\(RPR#RQ(RRP((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRNûs N( RRR RRQRPRjRiRN(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyRDÎs    ((>R R¥R›R3t version_infotunicodeRmtunichrR§tstrtchrRt ExceptionRt SyntaxErrorR tobjectR RRR%R&R(R0R5R8R:RœR?RBRCRKRHROR»RVRlRkR‰R‹RŠR“R”R R­R³R°R¯R±tsubR«R•tIRªR²thasattrR£R¤R©R¬RERD(((s4/usr/lib/python2.7/site-packages/cssselect/parser.pyt sj      ;#  *  S  - %           A