­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ó ÇaWc@s dZddlZddlZddlmZmZmZejddkr\eZ e Z n e Z e Z e d„Zdeefd„ƒYZd efd „ƒYZejd ƒjZejd ƒjZejd ƒjZdefd„ƒYZdefd„ƒYZdS(s cssselect.xpath =============== Translation of parsed CSS selectors to XPath expressions. :copyright: (c) 2007-2012 Ian Bicking and contributors. See AUTHORS for more details. :license: BSD, see LICENSE for more details. iÿÿÿÿN(tparset parse_seriest SelectorErroriicCs+|jddƒjdƒ}t|||ƒS(Ntasciitreplace(tencodetdecodetgetattr(tobjtnametdefault((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt_unicode_safe_getattrstExpressionErrorcBseZdZRS(s3Unknown or unsupported selector (eg. pseudo-class).(t__name__t __module__t__doc__(((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyR #st XPathExprcBsSeZddded„Zd„Zd„Zd„Zd„Zd„Zd„Z RS( tt*cCs||_||_||_dS(N(tpathtelementt condition(tselfRRRt star_prefix((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt__init__+s  cCs=t|jƒt|jƒ}|jr9|d|j7}n|S(Ns[%s](t_unicodeRRR(RR((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt__str__0s cCsd|jj|fS(Ns%s[%s](t __class__R (R((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt__repr__6scCs/|jr"d|j|f|_n ||_|S(Ns %s and (%s)(R(RR((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt add_condition9s  cCs=|jdkrdS|jdtj|jƒƒd|_dS(NRs name() = %s(RRtGenericTranslatort xpath_literal(R((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt add_name_test@s cCs|jd7_dS(se Append '*/' to the path to keep the context constrained to a single parent. s*/N(R(R((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytadd_star_prefixHscCsTt|ƒ|}|jdkr/||j7}n||_|j|_|j|_|S(Ns*/(RRRR(RtcombinertotherR((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytjoinOs   ( R RtFalseRRRRR R!R$(((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyR)s     s('+)s^[a-zA-Z_][a-zA-Z0-9_.-]*$s^[^ \t\r\n\f]+$RcBsEeZdZidd6dd6dd6dd6Zid d 6d d 6d d 6dd6dd6dd6dd6dd6ZdZdZeZeZ eZ e Z dd„Z ded„Zd„Zed„ƒZd„Zd „Zd!„Zd"„Zd#„Zd$„Zd%„Zd&„Zd'„Zd(„Zd)„Zd*„Zd+„Zeed,„Z d-„Z!d.„Z"d/„Z#d0„Z$d1„Z%d2„Z&d3„Z'd4„Z(d5„Z)d6„Z*d7„Z+d8„Z,d9„Z-d:„Z.e.Z/e.Z0e.Z1e.Z2e.Z3e.Z4e.Z5e.Z6e.Z7d;„Z8d<„Z9d=„Z:d>„Z;d?„Z<d@„Z=dA„Z>dB„Z?RS(Cs¤ Translator for "generic" XML documents. Everything is case-sensitive, no assumption is made on the meaning of element names and attribute names. t descendantt tchildt>tdirect_adjacentt+tindirect_adjacentt~texiststequalst=tincludess~=t dashmatchs|=t prefixmatchs^=t suffixmatchs$=tsubstringmatchs*=t differents!=tidsxml:langsdescendant-or-self::cs&dj‡‡fd†t|ƒDƒƒS(sµTranslate a *group of selectors* to XPath. Pseudo-elements are not supported here since XPath only knows about "real" elements. :param css: A *group of selectors* as an Unicode string. :param prefix: This string is prepended to the XPath expression for each selector. The default makes selectors scoped to the context node’s subtree. :raises: :class:`SelectorSyntaxError` on invalid selectors, :class:`ExpressionError` on unknown/unsupported selectors, including pseudo-elements. :returns: The equivalent XPath 1.0 expression as an Unicode string. s | c3s'|]}ˆj|ˆdtƒVqdS(ttranslate_pseudo_elementsN(tselector_to_xpathtTrue(t.0tselector(tprefixR(s3/usr/lib/python2.7/site-packages/cssselect/xpath.pys ¾s(R$R(RtcssR=((R=Rs3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt css_to_xpath«scCst|ddƒ}|s.td|fƒ‚n|j|ƒ}t||jƒsUt‚|r||jr||j||jƒ}n|p…dt |ƒS(s4Translate a parsed selector to XPath. :param selector: A parsed :class:`Selector` object. :param prefix: This string is prepended to the resulting XPath expression. The default makes selectors scoped to the context node’s subtree. :param translate_pseudo_elements: Unless this is set to ``True`` (as :meth:`css_to_xpath` does), the :attr:`~Selector.pseudo_element` attribute of the selector is ignored. It is the caller's responsibility to reject selectors with pseudo-elements, or to account for them somehow. :raises: :class:`ExpressionError` on unknown/unsupported selectors. :returns: The equivalent XPath 1.0 expression as an Unicode string. t parsed_trees"Expected a parsed selector, got %rRN( RtNonet TypeErrortxpatht isinstancet xpathexpr_clstAssertionErrortpseudo_elementtxpath_pseudo_elementR(RR<R=R8ttreeRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyR9ÂscCstdƒ‚dS(sTranslate a pseudo-element. Defaults to not supporting pseudo-elements at all, but can be overridden by sub-classes. s"Pseudo-elements are not supported.N(R (RRCRG((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyRHáscCsŠt|ƒ}d|kr%d|}nad|kr>d|}nHddjgt|ƒD](}|rTd|krrdpud|^qTƒ}|S(Nt's'%s't"s"%s"s concat(%s)t,(RR$tsplit_at_single_quotes(tstpart((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyRês      <cCsTt|ƒj}t|d|jƒdƒ}|dkrJtd|ƒ‚n||ƒS(s%Translate any parsed selector object.sxpath_%ss%s is not supported.N(ttypeR RtlowerRAR (Rtparsed_selectort type_nametmethod((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyRCøs  cCsH|j|j}t|d|ƒ}||j|jƒ|j|jƒƒS(sTranslate a combined selector.sxpath_%s_combinator(tcombinator_mappingt combinatorRRCR<t subselector(RtcombinedRVRT((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_combinedselectorscCs\|j|jƒ}|j|jƒ}|jƒ|jrK|jd|jƒS|jdƒSdS(Nsnot(%s)t0(RCR<RWR RR(RtnegationRCt sub_xpath((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_negation s   cCs`d|jjddƒ}t||dƒ}|sGtd|jƒ‚n||j|jƒ|ƒS(s$Translate a functional pseudo-class.sxpath_%s_functiont-t_s!The pseudo-class :%s() is unknownN(R RR RAR RCR<(RtfunctionRT((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_functions cCs]d|jjddƒ}t||dƒ}|sGtd|jƒ‚n||j|jƒƒS(sTranslate a pseudo-class.sxpath_%s_pseudoR^R_sThe pseudo-class :%s is unknownN(tidentRR RAR RCR<(RtpseudoRT((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt xpath_pseudos cCsí|j|j}t|d|ƒ}|jr>|jjƒ}n |j}t|ƒ}|jr‡d|j|f}|ot|jƒ}n|ršd|}nd|j|ƒ}|j rÈ|j jƒ}n |j }||j |j ƒ||ƒS(s Translate an attribute selector.sxpath_attrib_%ss%s:%st@sattribute::*[name() = %s]( tattribute_operator_mappingtoperatorRtlower_case_attribute_namestattribRQt is_safe_namet namespaceRtlower_case_attribute_valuestvalueRCR<(RR<RgRTR tsafeRiRm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt xpath_attrib's        cCs(|j|jƒ}|j|d|jƒS(sTranslate a class selector.s@class(RCR<txpath_attrib_includest class_name(Rtclass_selectorRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt xpath_class=scCs(|j|jƒ}|j|d|jƒS(sTranslate an ID selector.s@id(RCR<txpath_attrib_equalsR7(Rt id_selectorRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt xpath_hashDscCsŸ|j}|sd}t}n$t|ƒ}|jrB|jƒ}n|jrvd|j|f}|opt|jƒ}n|jd|ƒ}|s›|jƒn|S(s'Translate a type or universal selector.Rs%s:%sR(RR:Rjtlower_case_element_namesRQRkRER (RR<RRnRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt xpath_elementIs      cCs|jd|ƒS(s;right is a child, grand-child or further descendant of lefts/descendant-or-self::*/(R$(Rtlefttright((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_descendant_combinator`scCs|jd|ƒS(s#right is an immediate child of leftt/(R$(RRyRz((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_child_combinatordscCs)|jd|ƒ}|jƒ|jdƒS(s)right is a sibling immediately after lefts/following-sibling::sposition() = 1(R$R R(RRyRzRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt xpath_direct_adjacent_combinatorhs cCs|jd|ƒS(s1right is a sibling after left, immediately or nots/following-sibling::(R$(RRyRz((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt"xpath_indirect_adjacent_combinatornsc Csbyt|jƒ\}}Wn$tk r?td|jƒ‚nX|rS|jƒn|jƒ|dkr|r|d|}n|jd|ƒS|r¤| }| }n|dkrÀt| ƒ}n d| }|dkríd||fg}ng}|dkr|jd|ƒn&|dkr9|r9|jd |ƒnd j |ƒ}|r^|j|ƒn|S( NsInvalid series: '%r'is last() - %ssposition() = %ss+%sis(position() %s) mod %s = 0sposition() >= %ssposition() < (last() %s)s and ( Rt argumentst ValueErrorR R R!RtstrtappendR$( RRCR`tlastR tatbtb_negtexpr((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_nth_child_functionus8          cCs|j||dtƒS(NR„(R‰R:(RRCR`((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_nth_last_child_functionžscCs4|jdkrtdƒ‚n|j||dtƒS(NRs"*:nth-of-type() is not implementedR (RR R‰R%(RRCR`((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_nth_of_type_function¡s  cCs:|jdkrtdƒ‚n|j||dtdtƒS(NRs"*:nth-of-type() is not implementedR„R (RR R‰R:R%(RRCR`((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_nth_last_of_type_function¨s  cCs^|jƒdgdgfkr4td|jƒ‚n|jdj}|jd|j|ƒƒS(NtSTRINGtIDENTs9Expected a single string or ident for :contains(), got %riscontains(., %s)(targument_typesR R€RmRR(RRCR`Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_contains_function¯scCs^|jƒdgdgfkr4td|jƒ‚n|jdj}|jd|j|ƒƒS(NRRŽs5Expected a single string or ident for :lang(), got %rislang(%s)(RR R€RmRR(RRCR`Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_lang_functionºscCs |jdƒS(Nsnot(parent::*)(R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_root_pseudoÆscCs!|jƒ|jƒ|jdƒS(Nsposition() = 1(R!R R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_first_child_pseudoÉs  cCs!|jƒ|jƒ|jdƒS(Nsposition() = last()(R!R R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_last_child_pseudoÎs  cCs5|jdkrtdƒ‚n|jƒ|jdƒS(NRs"*:first-of-type is not implementedsposition() = 1(RR R!R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_first_of_type_pseudoÓs   cCs5|jdkrtdƒ‚n|jƒ|jdƒS(NRs!*:last-of-type is not implementedsposition() = last()(RR R!R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_last_of_type_pseudoÚs   cCs!|jƒ|jƒ|jdƒS(Ns last() = 1(R R!R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_only_child_pseudoás  cCs+|jdkrtdƒ‚n|jdƒS(NRs!*:only-of-type is not implementeds last() = 1(RR R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_only_of_type_pseudoæs cCs |jdƒS(Nsnot(*) and not(string-length())(R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_empty_pseudoìscCs |jdƒS(s:Common implementation for pseudo-classes that never match.RZ(R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytpseudo_never_matchesïscCs| s t‚|j|ƒ|S(N(RFR(RRCR Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_attrib_existsÿs  cCs$|jd||j|ƒfƒ|S(Ns%s = %s(RR(RRCR Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyRts cCsP|r,|jd|||j|ƒfƒn |jd||j|ƒfƒ|S(Nsnot(%s) or %s != %ss%s != %s(RR(RRCR Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_attrib_differents   cCsKt|ƒr:|jd|||jd|dƒfƒn |jdƒ|S(Ns:%s and contains(concat(' ', normalize-space(%s), ' '), %s)R'RZ(tis_non_whitespaceRR(RRCR Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyRps  % c Cs:|jd|||j|ƒ||j|dƒfƒ|S(Ns'%s and (%s = %s or starts-with(%s, %s))R^(RR(RRCR Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_attrib_dashmatchs  cCs=|r,|jd|||j|ƒfƒn |jdƒ|S(Ns%s and starts-with(%s, %s)RZ(RR(RRCR Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_attrib_prefixmatch#s   cCsM|r<|jd|||t|ƒd|j|ƒfƒn |jdƒ|S(Ns/%s and substring(%s, string-length(%s)-%s) = %siRZ(RtlenR(RRCR Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_attrib_suffixmatch+s - cCs=|r,|jd|||j|ƒfƒn |jdƒ|S(Ns%s and contains(%s, %s)RZ(RR(RRCR Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pytxpath_attrib_substringmatch5s   (@R RRRURft id_attributetlang_attributeR%RwRhRlRRER?R9RHt staticmethodRRCRYR]RaRdRoRsRvRxR{R}R~RR:R‰RŠR‹RŒRR‘R’R“R”R•R–R—R˜R™Rštxpath_link_pseudotxpath_visited_pseudotxpath_hover_pseudotxpath_active_pseudotxpath_focus_pseudotxpath_target_pseudotxpath_enabled_pseudotxpath_disabled_pseudotxpath_checked_pseudoR›RtRœRpRžRŸR¡R¢(((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyRgs               (                 tHTMLTranslatorcBsMeZdZdZed„Zd„Zd„Zd„Zd„Z d„Z RS(sê Translator for (X)HTML documents. Has a more useful implementation of some pseudo-classes based on HTML-specific element names and attribute names, as described in the `HTML5 specification`_. It assumes no-quirks mode. The API is the same as :class:`GenericTranslator`. .. _HTML5 specification: http://www.w3.org/TR/html5/links.html#selectors :param xhtml: If false (the default), element names and attribute names are case-insensitive. tlangcCs(||_|s$t|_t|_ndS(N(txhtmlR:RwRh(RR±((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyRRs  cCs |jdƒS(NsŠ(@selected and name(.) = 'option') or (@checked and (name(.) = 'input' or name(.) = 'command')and (@type = 'checkbox' or @type = 'radio'))(R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyR®YscCsq|jƒdgdgfkr4td|jƒ‚n|jdj}|jd|j|j|jƒdƒfƒS(NRRŽs5Expected a single string or ident for :lang(), got %ris‡ancestor-or-self::*[@lang][1][starts-with(concat(translate(@%s, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz'), '-'), %s)]R^(RR R€RmRR¤RRQ(RRCR`Rm((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyR‘ascCs |jdƒS(NsA@href and (name(.) = 'a' or name(.) = 'link' or name(.) = 'area')(R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyR¦oscCs |jdƒS(Ns¥ ( @disabled and ( (name(.) = 'input' and @type != 'hidden') or name(.) = 'button' or name(.) = 'select' or name(.) = 'textarea' or name(.) = 'command' or name(.) = 'fieldset' or name(.) = 'optgroup' or name(.) = 'option' ) ) or ( ( (name(.) = 'input' and @type != 'hidden') or name(.) = 'button' or name(.) = 'select' or name(.) = 'textarea' ) and ancestor::fieldset[@disabled] ) (R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyR­vscCs |jdƒS(Ns' ( @href and ( name(.) = 'a' or name(.) = 'link' or name(.) = 'area' ) ) or ( ( name(.) = 'command' or name(.) = 'fieldset' or name(.) = 'optgroup' ) and not(@disabled) ) or ( ( (name(.) = 'input' and @type != 'hidden') or name(.) = 'button' or name(.) = 'select' or name(.) = 'textarea' or name(.) = 'keygen' ) and not (@disabled or ancestor::fieldset[@disabled]) ) or ( name(.) = 'option' and not( @disabled or ancestor::optgroup[@disabled] ) ) (R(RRC((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyR¬’s( R RRR¤R%RR®R‘R¦R­R¬(((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyR¯?s     (Rtsystretcssselect.parserRRRt version_infot basestringt _basestringtunicodeRR‚RAR t RuntimeErrorR tobjectRtcompiletsplitRMtmatchRjRRR¯(((s3/usr/lib/python2.7/site-packages/cssselect/xpath.pyt s"    1ÿÙ