f#]c@sZdZddlZddlZddlZddlmZddlmZej e Z e Z y ejejdeZ WnnXddlZdefdYZejjjdefd YZdd lmZdd lmZd efd YZiedd6edd6edd6Zx$dD]Zeeeee} of the XML Schema definition. In particular, we need to be able to identify character properties and block escapes, as defined in F.1.1, by name. - Block data: U{http://www.unicode.org/Public/3.1-Update/Blocks-4.txt} - Property list data: U{http://www.unicode.org/Public/3.1-Update/PropList-3.1.0.txt} - Full dataset: U{http://www.unicode.org/Public/3.1-Update/UnicodeData-3.1.0.txt} The Unicode database active at the time XML Schema 1.0 was defined is archived at U{http://www.unicode.org/Public/3.1-Update/UnicodeCharacterDatabase-3.1.0.html}, and refers to U{Unicode Standard Annex #27: Unicode 3.1 }. iN(tsix(txranges[\U0001d7ce-\U0001d7ff]tCodePointSetErrorcBseZdZRS(s8Raised when some abuse of a L{CodePointSet} is detected.(t__name__t __module__t__doc__(((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyR4st CodePointSetcBseZdZdZdZes'eZndZdZdZ dZ dZ dZ dZ d Zd Zd Zd ed ededededfZdZedZdZdZdZRS(sRepresent a set of Unicode code points. Each code point is an integral value between 0 and 0x10FFFF. This class is used to represent a set of code points in a manner suitable for use as regular expression character sets.iicCs|jS(sUFor testing purrposes only, access to the codepoints internal representation.(t_CodePointSet__codepoints(tself((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyt _codepointsSscCs t|jS(N(thashR(R((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyt__hash__XscCs|j|jkS(s-Equality is delegated to the codepoints list.(R(Rtother((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyt__eq__[scCs|j|jkS(N(R(RR ((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyt__lt___scGsg|_dt|krlt|dtrI|jj|djdSt|dtrl|d}qlnx|D]}|j|qsWdS(Nii(Rtlent isinstanceRtextendtlisttadd(Rtargsta((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyt__init__bs  cCst|tr(|\}}|d7}n_t|tjrqdt|krXtnt|}|d}nt|}|d}||krtdn||j kr|S||j kr|j d}nt j |j |}t j |j |}|d@d>|d@B}|s'd|}nd|krC|j ||5npd|krp|j |d|5||j |            cCs|j|tS(sAdd the given value to the code point set. @param value: An integral value denoting a code point, or a tuple C{(s,e)} denoting the start and end (inclusive) code points in a range. @return: C{self}(t_CodePointSet__mutatetTrue(RR!((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyRscCsJt|tr%|j|jn!x|D]}|j|tq,W|S(sAdd multiple values to a code point set. @param values: Either a L{CodePointSet} instance, or an iterable whose members are valid parameters to L{add}. @return: C{self}(RRRtasTuplesR)R*(Rtvaluestv((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyRs  cCsGt|tr7x!|jD]}|j|qW|S|j|tS(sRemove the given value from the code point set. @param value: An integral value denoting a code point, or a tuple C{(s,e)} denoting the start and end (inclusive) code points in a range, or a L{CodePointSet}. @return: C{self}(RRR+tsubtractR)tFalse(RR!R-((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyR.s it^s\t[t]t-cCsVtj|}d|kr-tjd}n||jkrRtjd|}n|S(Nitx00i\(Rtunichrtut-_CodePointSet__XMLtoPythonREEscapedCodepoints(Rt code_pointtrv((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyt__unichrs  cCsg}|r%|jtjdnxg|jD]Y\}}||krc|j|j|q2|j|j|d|j|gq2W|r|jtjdntjdj|S(sReturn the code point set as Unicode regular expression character group consisting of a sequence of characters or character ranges. This returns a regular expression fragment using Python's regular expression syntax. Note that different regular expression syntaxes are not compatible, often in subtle ways. @param with_brackets: If C{True} (default), square brackets are added to enclose the returned character group.R1R3R2t(tappendRR6R+t_CodePointSet__unichrRtjoin(Rt with_bracketstrvaR#R$((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyt asPatterns  ,cCsg}d}x]tt|jD]F}|dk r[|j||j|dfd}q"|j|}q"W|dk r||jkr|j||jfn|S(sReturn the codepoints as tuples denoting the ranges that are in the set. Each tuple C{(s, e)} indicates that the code points from C{s} (inclusive) to C{e}) (inclusive) are in the set.iN(tNoneRRRR<R(RR9tstartR&((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyR+s  cCsxt|}dt|jkrQd|jdkrQ|jj|jdn#|jjd|jj|j|S(s;Return an instance that represents the inverse of this set.ii(ttypeRRRR<(RR9((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pytnegate s (cCsKdt|jks3d|jd|jdkr7dStj|jdS(stIf this set represents a single character, return it as its unicode string value. Otherwise return C{None}.iiiN(RRRBRR5(R((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pytasSingleCharacters3N(RRRRtMaxShortCodePointtSupportsWideUnicodeRBRR R R RRR)RRR.RR7R=R*RAR+RERF(((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyR8s4      0          (t PropertyMap(tBlockMaptXML1p0e2cBsOeZdZeddddCdDZerJejdejejfnedEdFdGdHdIdJdKdLdMdNdOdPdQdRdSd'dTd*dUdVdWd1d2d3d4dXdYdZd[d\d]d^d_d`dadbdcdMdddedfdgdhdidjdkdld`dmdndedodpdqdrdsdpdtdudvdwdxdydzd{d|d}d~dddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd dddddddddddddddd d!d"ddd'd(d)d*ddd/dd2d3d4ddddddddCdDdEddddLdddddddd[dd^dddddZ eddkdZ ee j e Z edddddddzdd}dddddddddddddddddddddddddddddddddddddddddddddddddddd d d d d ddddddddddddddddddddddddddd dd d_Z edd d!d"d#d$d%d&d'd(d)d*d+d,d-Zed-d.d/d0d1d2d3d4d.d/d0 Zee Zejed;ejed<ee Zejed;ee Zej eejed=ejed>ejed;ejed<ej e ej eee Zej eejed=ejed>ejed;ej e ej ed?ejejfZejd@efZdAejfZejd@efZd?ejejfZejd@efZdBeefZejd@efZRS(1s-Regular expression support for XML Schema Data Types. This class holds character classes and regular expressions used to constrain the lexical space of XML Schema datatypes derived from U{string}. They are from U{XML 1.0 (Second Edition)} and U{Namespaces in XML }. Unlike the regular expressions used for pattern constraints in XML Schema, which are derived from the Unicode 3.1 specification, these are derived from the Unicode 2.0 specification. The XML Schema definition refers explicitly to the second edition of XML, so we have to use these code point sets and patterns. Be aware that U{subsequent updates to the XML specification } have changed the corresponding patterns for other uses of XML. One significant change is that the original specification, used here, does not allow wide unicode characters.i i i i iiiiiAiZiaiziiiiiiii1i4i>iAiHiJi~iiiiiiiiiPiiiiiiiiiiiiiiiiiiiii iiOiQi\i^iiiiiiiiiiiiii1iViYiaiiiiii!i:iAiJiqiiiiiiiiiii i9 i= iX ia i i i i i i i i i i i i i i i i i i i i i i i( i* i0 i2 i3 i5 i6 i8 i9 iY i\ i^ ir it i i i i i i i i i i i i i i i i i i i i i( i* i0 i2 i3 i6 i9 i= i\ i] i_ ia i i i i i i i i i i i i i i i i i i i i i i i i i( i* i3 i5 i9 i` ia i i i i i i i i i i i i i i i i i i i( i* i9 i` ia ii.i0i2i3i@iEiiiiiiiiiiiiiiiiiiiiiiiiii@iGiIiiiiiiiiiiii i i iii<i>i@iLiNiPiTiUiYi_iaicieigiiiminirisiuiiiiiiiiiiiiiiiiiiiiii iEiHiMiPiWiYi[i]i_i}iiiiiiiiiiiiiiiiiiii&!i*!i+!i.!i!i!iA0i0i0i0i1i,1iiiNii0i!0i)0iiEi`iaiiiiiiiiiiiiiKiRipiiiiiiiiiii i i< i> iL iM iQ iT ib ic i i i i i i i i i i i i i i i i< i> i? i@ iB iG iH iK iM ip iq i i i i i i i i i i i i< i> iC iG iH iK iM iV iW i i i i i i i i i i i i> iD iF iH iJ iM iU iV i i i i i i i i i i i i i> iC iF iH iJ iM iW i1i4i:iGiNiiiiiiiiii5i7i9i>i?iqiiiiiiiiiiii i i i*0i/0i0i0i0i9i`iiiiif io i i if io i i if io i i if io i i if io iPiYiii i)iiiii@iFii0i10i50i0i0i0i0t_t:t.R3s%s%s*s^%s$s%s+s(%s:)?%s(i i(ii(iAiZ(iaiz(ii(ii(ii(ii1(i4i>(iAiH(iJi~(ii(ii(ii(ii(iPi(ii(ii(ii(ii(ii(ii(ii (iiO(iQi\(i^i(ii(ii(ii(ii(ii(ii(i1iV(iai(ii(ii(i!i:(iAiJ(iqi(ii(ii(ii(ii(i i9 (iX ia (i i (i i (i i (i i (i i (i i (i i (i i (i i (i i (i i( (i* i0 (i2 i3 (i5 i6 (i8 i9 (iY i\ (ir it (i i (i i (i i (i i (i i (i i (i i (i i (i i( (i* i0 (i2 i3 (i6 i9 (i\ i] (i_ ia (i i (i i (i i (i i (i i (i i (i i (i i (i i (i i (i i (i i( (i* i3 (i5 i9 (i` ia (i i (i i (i i (i i (i i (i i (i i (i i (i i( (i* i9 (i` ia (ii.(i2i3(i@iE(ii(ii(ii(ii(ii(ii(ii(ii(ii(i@iG(iIii(ii(ii(ii(ii(i i (ii(iTiU(i_ia(imin(iris(ii(ii(ii(ii(ii(ii(ii(i iE(iHiM(iPiW(i_i}(ii(ii(ii(ii(ii(ii(ii(ii(ii(i*!i+!(i!i!(iA0i0(i0i0(i1i,1(ii(iNi(i!0i)0(iiE(i`ia(ii(ii(ii(ii(ii(iKiR(ii(ii(ii(ii(ii(i i (i> iL (iQ iT (ib ic (i i (i i (i i (i i (i i (i@ iB (iG iH (iK iM (ip iq (i i (i i (i i (i i (i i (i> iC (iG iH (iK iM (iV iW (i i (i i (i i (i i (i i (i> iD (iF iH (iJ iM (iU iV (i i (i i (i i (i i (i i (i i (i> iC (iF iH (iJ iM (i4i:(iGiN(ii(ii(ii(ii(iqi(ii(ii(ii(ii(i i (i*0i/0(i0i9(i`ii(ii(if io (i i (if io (i i (if io (i i (if io (i i (if io (iPiY(ii(i i)(i10i50(i0i0(i0i0( RRRRtCharRHRRGRtBaseChart IdeographicRtLettert CombiningChartDigittExtendert NameStartCharRtNCNameStartChartNameChart NCNameCharRAtName_pattretcompiletName_ret NmToken_patt NmToken_ret NCName_patt NCName_ret QName_pattQName_re(((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyRKs                 i tni tri tts\|.-^?*+{}()[]sp{%s}sP{%s}sp{Is%s}sP{Is%s}s s i s R#tSRLRMtitItctCtNdtdtDtPtZtWtw(-RR[tloggingtpyxb.utils.utilitytpyxbt pyxb.utilsRtpyxb.utils.six.movesRt getLoggerRt_logR/RHR\R6R*Rt LookupErrorRtutilstutilitytBackfillComparisonstobjectRtpyxb.utils.unicode_dataRIRJRKt SingleCharEscRjRtcatEsctcomplEsct iteritemstkR-REt IsBlockEsct WildcardEsct MultiCharEscRRRRXR(((sG/data/av2000/b2b/venv/lib/python2.7/site-packages/pyxb/utils/unicode.pyt!s\       $$!+1.