3 d @sdZddlmZddlmZddlmZddlZddlZddlZddlZddl Z ddl Z ddl m Z ddl mZddlmZmZmZmZmZmZdd lmZmZmZdd lmZdd lmZdd lmZdd lmZddlm Z ddlm!Z!ddlm"Z"dddddddddg Z#ddl$m%Z%diddZ&ddZ'ddZ(ddZ)dZ*d Z+d!Z,d"Z-e+e,e-fZ.e*e+e,e-fZ/dZ0d Z1d!Z2d#Z3d$Z4d"Z5d%Z6d&Z7e7d'Z8e7d(Z9e7d)Z:d*Z;d+Ze7d.Z?d/Z@eZPe9ZQe:ZRd?ZSe0e;fZTe0e>fZUd@ZVedAZWedBZXedCZYedDZZedEZ[dZ\dFZ]e]dGZ^e]dHZ_d+Z`dIZadJZbdKdLZce"rdMdLZcejddNZeejddOZfd?ZgejddPZhejddQZiejddRZjejddSZkejddSZlejddTZmejddUZnejddVZoejddWZpejddXZqGdYdZdZZrGd[ddesZtGd\d]d]euZvejdd^ZwGd_d`d`euZxdadZyGdbdde%ZzGdcddezZ{dddeZ|dfdgZ}e~dhkre}dS)ja notation3.py - Standalone Notation3 Parser Derived from CWM, the Closed World Machine Authors of the original suite: * Dan Connolly <@@> * Tim Berners-Lee <@@> * Yosi Scharf <@@> * Joseph M. Reagle Jr. * Rich Salz http://www.w3.org/2000/10/swap/notation3.py Copyright 2000-2007, World Wide Web Consortium. Copyright 2001, MIT. Copyright 2001, Zolera Systems Inc. License: W3C Software License http://www.w3.org/Consortium/Legal/copyright-software Modified by Sean B. Palmer Copyright 2007, Sean B. Palmer. Modified to work with rdflib by Gunnar Aastrand Grimnes Copyright 2010, Gunnar A. Grimnes )absolute_import)division)print_functionN)Decimal)uuid4)URIRefBNodeLiteralVariable_XSD_PFX _unique_id) QuotedGraphConjunctiveGraphGraph)b) binary_type) long_type) string_types) text_type)unichr) narrow_build BadSyntaxN3Parser TurtleParser splitFragPjoinbase runNamespace uniqueURIhexify)ParsercCs6|jd}|dkr*|d|||dfS|dfSdS)zsplit a URI reference before the fragment Punctuation is kept. e.g. >>> splitFragP("abc#def") ('abc', '#def') >>> splitFragP("abcdef") ('abcdef', '') #rN)rfind)Zurirefpunctir&B/tmp/pip-build-7vycvbft/rdflib/rdflib/plugins/parsers/notation3.pyrAs c Cs|jd}|jd}|dkr0|dks,||kr0|S|jd}|dksNtd|t|\}}|sf||S||d|ddkrtd||f||d|dd kr|jd|d}n|d}|dkrt|}|d}|d dd kr|d |d|S|d ddkr|d ||S|jd}x|d dd krH|dd }|d krXd }nX|d ddkst|dkr|dd }|jd||} | dkr|d | d}| }nPq*W|d |d||S)ajoin an absolute URI and URI reference (non-ascii characters are supported/doctested; haven't checked the details of the IRI spec though) ``here`` is assumed to be absolute. ``there`` is URI reference. >>> join('http://example/x/y/z', '../abc') 'http://example/x/abc' Raise ValueError if there uses relative path syntax but here has no hierarchical path. >>> join('mid:foo@example', '../foo') # doctest: +NORMALIZE_WHITESPACE Traceback (most recent call last): raise ValueError(here) ValueError: Base has no slash after colon - with relative '../foo'. >>> join('http://example/x/y/z', '') 'http://example/x/y/z' >>> join('mid:foo@example', '#foo') 'mid:foo@example#foo' We grok IRIs >>> len(u'Andr\xe9') 5 >>> join('http://example.org/', u'#Andr\xe9') u'http://example.org/#Andr\xe9' /:rzBase uri '%s' is not absolutez8Base <%s> has no slash after colon - with relative '%s'.z//Nz./.r"z../z..)findAssertionErrorr ValueErrorlenr#) hereZthereZslashlZcolonlZbcolonlpathfragbpathZslashrr%r&r&r'rXsL&            cCsdttjdS)aThe base URI for this process - the Web equiv of cwd Relative or abolute unix-standard filenames parsed relative to this yeild the URI of the file. If we had a reliable way of getting a computer name, we should put it in the hostname just to prevent ambiguity zfile://r() _fixslashosgetcwdr&r&r&r'rs cCs4|jdd}|ddkr0|ddkr0|dd}|S)z8 Fix windowslike filename to unixlike - (#ifdef WINDOWS)\r(rr*r)r+N)replace)sr&r&r'r6s  r6r*r+r,z#http://www.w3.org/2000/10/swap/log#isforSomeforAllz/http://www.w3.org/1999/02/22-rdf-syntax-ns#typez+http://www.w3.org/1999/02/22-rdf-syntax-ns#zhttp://www.w3.org/2002/07/owl#ZsameAsZparsesToz$http://www.w3.org/TR/REC-rdf-syntax/z&http://www.w3.org/2000/10/swap/log.n3#firstrestlinilListEmptycCstdkrtttdatS)z:Return a URI suitable as a namespace for run-local objectsNr!)runNamespaceValuerrr r&r&r&r'rscCstd7atdttS)z A unique URIr*Zu_)nexturstrr&r&r&r'rsF2cOsdS)Nr&)argskargsr&r&r' BecauseOfDatasrNcOsdS)Nr&)rLrMr&r&r'becauseSubexpressionsrOr!z*http://www.w3.org/2000/10/swap/log#impliesbooleandecimaldoublefloatintegerz !"#$&'()*,+/;<=>?@[\]^`{|}~r-r)ZABCDEFabcdef0123456789z(_~.-!$&'()*+,;=/?#@%)c Cs8ytt|jddStd|jdYnXdS)Nr*zInvalid unicode code point: )rintgroup Exception)mr&r&r' unicodeExpand;srZc CsPytt|jddStk rJtjd|jdtj|jddSXdS)Nr*rUzEncountered a unicode char > 0xFFFF in a narrow python build. Trying to degrade gracefully, but this can cause problems later when working with the string: %srZunicode_escape)rrVrWr0warningswarncodecsdecode)rYr&r&r'rZBsz\\u([0-9a-fA-F]{4})z\\U([0-9a-fA-F]{8})z[ \t]*(#[^\n]*)?\r?\nz[ \t]*(#[^\n]*)?$z[ \t]*z [-+]?[0-9]+z[-+]?[0-9]*\.[0-9]+z9[-+]?(?:[0-9]+\.[0-9]*|\.[0-9]+|[0-9]+)(?:e|E)[-+]?[0-9]+z[0-9]+z [\\\r\n\"\']z[a-zA-Z0-9]+(-[a-zA-Z0-9]+)*c@sdeZdZdXddZddZdd Zd d Zd d ZddZddZ dYddZ ddZ ddZ ddZ ddZddZddZd d!Zd"d#Zd$d%Zd&d'Zd(d)Zd*d+Zd,d-ZdZd.d/Zd0d1Zd2d3Zd[d4d5Zd6d7Zd8d9Zd:d;Zdd?Z d@dAZ!dBdCZ"dDdEZ#dFdGZ$dHdIZ%dJdKZ&dLdMZ'dNdOZ(dPdQZ)dRdSZ*dTdUZ+dVdWZ,dS)\ SinkParserNr"FcCsfi|_|dkr0d|ks"td||d|jd<||_|rD|j|||_d|_d|_||_dddd d d d d g|_d|_ i|_ i|_ i|_ ||_ ||_|rdnd|_d|_trt|j||j d|_|r||_n|r||_nd|_|j sd|jkst|js|jr|jd|_nt|_|dkrN|jrB|j|d|_n |j|_n||_|j|_d|_dS)zf note: namespace names should *not* end in # ; the # will get added during qname processing r"r)zDocument URI not absolute: <%s>r!rathisbindhasr?oftruefalse"'N)Zbecausez#_gz #_formula)rgrh)rg) _bindingsr/_storeZ setGenPrefix_thisDoclines startOfLine _genPrefixkeywords keywordsSet_anonymousNodes _variables_parentVariablesZ_reasonturtlestring_delimiters_reason2trackingrN newSymbol_baseURIr newFormula_formula_context_parentContext)selfstoreZ openFormulaZthisDocbaseURIZ genPrefixwhyrtr&r&r'__init__gsR     zSinkParser.__init__cCsd|j|j||jdfS)a(String generated from position in file This is for repeatability when refering people to bnodes in a document. This has diagnostic uses less formally, as it should point one to which bnode the arbitrary identifier actually is. It gives the line and character number of the '[' charcacter or path character which introduced the blank node. The first blank node is boringly _L1C1. It used to be used only for tracking, but for tests in general it makes the canonical ordering of bnodes repeatable.z %s_L%iC%ir*)rnrlrm)r~r%r&r&r'r2s zSinkParser.herecCs|jS)N)r{)r~r&r&r'formulaszSinkParser.formulacCs|j|jS)N)loadBufread)r~streamr&r&r' loadStreamszSinkParser.loadStreamcCs|j|j||jS)z1Parses a buffer and returns its top level formula)startDocfeedendDoc)r~bufr&r&r'rs zSinkParser.loadBufcCst|tsB|jd}t|dkrF|dtjjdkrF|dd}n|}d}xF|dkr|j||}|dkrldS|j||}|dkrL|j||dqLWdS)aFeed an octet stream tothe parser if BadSyntax is raised, the string passed in the exception object is the remainder after any statements have been parsed. So if there is more data to feed to the parser, it should be straightforward to recover.zutf-8rr*Nzexpected directive or statement) isinstancerr^r1r]BOM_UTF8 skipSpacedirectiveOrStatementr)r~octetsr;r%jr&r&r'rs      zSinkParser.feedcCsz|j||}|dkr|S|jr6|j||}|dkr6|S|j||}|dkrV|j||S|j||}|dkrv|j||S|S)Nr)rrtsparqlDirective directivecheckDot statement)r~argstrhr%rr&r&r'rs      zSinkParser.directiveOrStatementcCs|dtkst|||ddkr.|d}n||jkr is ok, rdf:type shortcut a must be followed by ws) rr*@r)Nr) _notNameCharsr/ror1_notKeywordsChars)r~tokrr%colonr&r&r'rs   zSinkParser.tokcCs\|dtkst|||t|j|jkrT||t|tkrT|t|}|SdSdS)zCheck for SPARQL keyword. Space must have been stripped on entry and we must not be at end of file. Case insensitive and not preceeded by @ rr*Nr)rr/r1lower_notQNameChars)r~rrr%r&r&r' sparqlToks   zSinkParser.sparqlTokcCs6|j||}|dkr|Sg}|jd||}|dkr@|j||d|jd||}|dkr|jrj|j||d|j||||j}|dkr|j||d|j|dd|S|jd||}|dkr:|jr|j||d|j||||j}|dkr|j||d x8|D]0}||jks||j kr|j j ||j|<qW|S|jd ||}|dkr|jrh|j||d |j||||j}|dkr|j||d x|D]}|j j |qW|S|jd ||dd}|dkrg}|j |||}|dkr|j||d|j|||}|dkr|j||d|j|d}|jrBt|j|}n d|krb|j||d|dd|kspt||j|dd<|j|ddt||S|jd||}|dkr2g}|j|||}|dkr|j||d|j|d}|jrt|j|}n|j||d|dd|ks(t||_|SdS)Nrrbz%keyword bind is obsolete: use @prefixroz%Found 'keywords' when in Turtle mode.z/'@keywords' needs comma separated list of wordsrAz#Found 'forAll' when in Turtle mode.zBad variable list after @forAllr@z$Found 'forSome' when in Turtle mode.z Bad variable list after @forSomeprefixT)rzexpected qname after @prefixz'expected after @prefix _qname_r*r)zWith no base URI, cannot use zrelative URI in @prefix <>rzexpected after @base z&With no previous base URI, cannot use zrelative URI in @base With no previous base URI, cannot use relative URI in @base after @prefix _qname_r*r)zWith no base URI, cannot use zrelative URI in @prefix after @base z&With no previous base URI, cannot use zrelative URI in @base With no previous base URI, cannot use relative URI in @base - prop -> <- prop -< _operator_rrcz"Found 'has' keyword in Turtle modezexpected property after 'has'z->r?z!Found 'is' keyword in Turtle modezexpected after 'is'z/End of file found, expected property after 'is'rdzexpected 'of' after 'is' z<-r`r+z<=zFound '<=' in Turtle mode. Zimpliesr*=zFound '=' in Turtle moderz:=zFound ':=' in Turtle modeZbecomesz>-z>- ... -> syntax is obsolete.r) rrrtrpropappendRDF_typerjrxLogic_NS DAML_sameAs)r~rr%rrrr&r&r'verbs      ,zSinkParser.verbcCs|j|||S)N)r)r~rr%rr&r&r'rAszSinkParser.propcCs|j|||S)N)r3)r~rr%rr&r&r'rDszSinkParser.itemcCs|jj|j||jdS)N)r)rj newBlankNoder|rv)r~rr&r&r' blankNodeGszSinkParser.blankNodec Cs|j|||}|dkr|Sx|||ddkr|||d}|j}|j|j|d}|j||d|}|dkr|j||d|j}|dkr|j|j|||fn|j|j|||f|j|qW|S)z#Parse the path production. rr*z!^)rz"EOF found in middle of path syntax^) nodeOrLiteralpoprr2noderrr|r) r~rr%rrchsubjobjpredr&r&r'r3Js"zSinkParser.pathcCs<|jj|d}|dk r|S|jj|j|jd}||j|<|S)z?Remember or generate a term for one of these _: anonymous nodesN)r)rqgetrjrr|rv)r~lnZtermr&r&r' anonymousNodeas  zSinkParser.anonymousNodecCs|}|j||}|dkr|S|}|||d}|dkr|j|}|j||d}|dkrj|j||d|||ddkr@|jr|j||d|d}g} |j||| }|dkr2| d}t| dkrx | D]} |j|jt|| fqW|j||}|dkr|j||d|||ddkr@|d}n|j||d |d krV|j |d }|j |||}|dkr||j||d |j||}|dkr|j||d |||ddkr|j||d|j ||dS|j r |dkr ||d|d} | dkr|d7}|d}g} d} x|j||}|dkrF|j||d|||ddkrf|d}P| s|||ddkr|d7}n|j||dnd} g}|j |||}|dkr|j||d| j |j j|dq"W|j |j j| |j|S|d}|j}|j|_|j}|j}|j|_i|_|jj|_|j}t|_|d kr\|j j}||_xn|j||}|dkr|j||d|||ddkr|d}P|j||}|dkrd|j||dqdW||_|j|_||_|j|_||_||_|j |j|S|dkr|j j}||d|d} | dkrL|j j}|d7}|d}g} x|j||}|dkr~|j||d|||ddkr|d}Pg}|j |||}|dkr|j||d | j |j j|dqZW|j || |j|S|jd!||}|dkr |j||d&|jd$||}|dkrF|j d|S|jd%||}|dkrl|j d|S|d kr|j|||}|dkr|Sd'S)(zParse the production. Space is now skipped once at the beginning instead of in multipe calls to self.skipSpace(). rr*[z EOF after '['rz(Found '[=' or '[ =' when in turtle mode.z'EOF when objectList expected after [ = ;zobjectList expected after [= N)rzproperty_list expectedz,EOF when ']' expected after [ ]z ']' expected{r+$Tzneeded '$}', found end.z$},z expected: ','Fzexpected item in set or '$}'zneeded '}', found end.}zexpected statement or '}'(zneeded ')', found end.)zexpected item in list or ')'raz'Keyword 'this' was ancient N3. Now use z@forSome and @forAll keywords.rerfzEKeyword 'this' was ancient N3. Now use @forSome and @forAll keywords.r)rr2rrt objectListr1rr|rrrrrrjinternnewSetr}rqrsrrcopyrvrOrzrclosenewListrr)r~rr%rZsubjectAlreadyrrrZbnodeIDobjsrZch2rFZ first_runrZoldParentContextZparentAnonymousNodesZgrandParentVariablesZreason2Z thing_typer&r&r'rjs2                                            zSinkParser.nodec Csxx<|j||}|dkr(|j||d||dkr6P|d}qW|||ddkr|jrj|j||d|d}g}|j||||}|dkr|j||d|}q|}g}|j|||}|dkr|Sg}|j|||}|dkr|j||d xL|D]D}|d\} } | d kr"|j|j| ||fq|j|j| ||fqW|j||}|dkr^|j||d |||ddkrx|S|d}qWd S) zUParse property list Leaves the terminating punctuation in the buffer rz-EOF found when expected verb in property listrr*r+z:-zFound in ':-' in Turtle modez!bad {} or () or [] node after :- zobjectList expectedz->zEOF found in list of objectsN)rrrtrrrrr|) r~rr%rrrvrrZdirasymr&r&r'r'sT        zSinkParser.property_listcCs|j||}|dkr"|j||d||dkr2|S||||}|dkrJdSxj|j||}|dkrd|S|||d}|dkr|dkrd S|S|||d|}|dkrL|j||dqLWdS) zereturn value: -1 bad syntax; >1 new position in argstr res has things found appended rz"EOF found expecting comma sep listr-r*rzbad list contentNrr)rr)r~rrrwhatr%rr&r&r'rYs,    zSinkParser.commaSeparatedListcCs||j|||}|dkrdSx\|j||}|dkr>|j||d|||ddkrV|S|j||d|}|dkr|SqWdS)Nrr*zEOF found after objectrr)rrr)r~rr%rrr&r&r'rus zSinkParser.objectListcCsv|j||}|dkr|S|||ddkr4|dS|||ddkrL|S|||ddkrd|S|j||ddS)Nrr*r-rrz.expected '.' or '}' or ']' at end of statement)rr)r~rr%rr&r&r'rs zSinkParser.checkDotc Csg}|j|||}|dkr|d\}}|dkrFds:td|jt}nty|j|}Wndtk r|dkr|j|j||S|j r|dkrt |jpdd}n|j ||d|YnX|j j ||} | |j kr|j|j | n |j| |S|j||}|dkrdS||d krNg} |j||| }|dkrJ|j| d|SdS||d kr^|d}|} x|t|krL||d kr@|| |} tjt| } tjt| } |jrt |j| } nd | kstd ||d|dkr| dddk r| d} |j j | } | |j kr.|j|j | n |j| |dS|d}qjW|j ||dnv|jrg} |j||| }|dkrdS| d|jkr|j ||d| d|j|j j |jd| d|SdSdS)zGenerate uri from n3 representation. Note that the RDF convention of directly concatenating NS and local name is now used though I prefer inserting a '#' to make the namesapces look more like what XML folks expect. rNz not used?_r"r!zPrefix "%s:" not boundr*? variable(:abc) rr*rz 0123456789-z"Varible name can't start with '%s'Nr!)rrr)rrr1rr}rjrxryrrr|rrvrrs)r~rr%rrZvarURIr&r&r'rs2      zSinkParser.variablecCst|j||}|dkrdS||dks0||tkr4dS|}x$|t|kr\||tkr\|d}q:W|j||||S)z abc -> :abc rr*z 0123456789-rr)rrr1r)r~rr%rrr&r&r'rs  zSinkParser.bareWordc CsL|j||}|dkrdS||}|dkr,dS|tkr|}|d}x4|t|krt||}|tkrp||}|d}qBPqBW||ddkr|dd}|sdS|d8}nd}|t|ko||dkr|}|dkrt}nt}|d}d }d}x|t|kr||}| r |d kr d }|d7}q|s0||kr|rZ|tkrt|j|j||d |nF|d kr||dtks||dtkrt|j|j||d|||}|d}d }qPqW|rt|j|j||d||ddkr|dd}|sdS|d8}|j ||f|S|rD|j rD||j krD|j d|f|SdSdS)z xyz:def -> ('xyz', 'def') If not in keywords and keywordsSet: def -> ('', 'def') :def -> ('', 'def') rr*z 0123456789-+.r-Nr"r)rFr9Tzillegal escape %r+zillegal hex escape zqname cannot end with \rrrrrrr) rrr1r escapeCharsrrklinehexCharsrrpro) r~rr%rcrrZ allowedCharsZ lastslashr&r&r'r,s|        $   zSinkParser.qnamecCs|j|||}|dkr|S|j||}|dkr2dS|}|||jkr|||d||dkrn||d}n||}|t|}|j|||\}}|j|jj||SdSdS)Nrr*r,rr)rrrur1strconstrrj newLiteral)r~rr%rrdelimr;r&r&r'rs   zSinkParser.objectc Cs|j|||}|j}|dkr |S|j||}|dkr8d S|}||}|dkrtj||}|r~|j}|jt||||Stj||}|r|j}|jt ||||St j||}|r|j}|jt ||||S|||j kr|||d||dkr||d}n||}|t |}d} |j|||\}} d} |||ddkrtj||d}|dkrt|j|||d |j}||d|} |}|||dd krg} |j||d| }| d} |j|jj| | | |Sd SdS) Nrr*z -+0987654321.r,rz#Bad language code syntax on string zliteral, after @r+z^^rz3Bad language code syntax on string literal, after @r)rrlrexponent_syntaxrrrrSdecimal_syntaxrinteger_syntaxrrur1rlangcoderrkrrjr) r~rr%rr startlinerrYrdtr;langZres2r&r&r'rsd       zSinkParser.nodeOrLiteralcCst|tr|dS|S)Nr*)rtuple)r~rr&r&r'rs zSinkParser.uriOfcCs<|d}|d|d|d|df\}}}}|} d} |j} x| t|kr(|| |kr ||krr| d}|| fS||kr || | d|kr| d}| |} || fS|| | d|kr| d}| |} || fS|| | d|kr| d}|| fS| d} | |} q>tj|| } | sDtd|| d | || | d f| j}y| || |} Wnptk rd} x&|| |D]}| d t|} qWtj dj }t |j | || d d | |fYnX||}||kr|} q>q>|dkr||kr| |} |d} q>q>|dkr^||kr:t |j | ||d|jd|_| |} |d} | |_ q>|dkr>|d} || | d}|st |j | ||ddj|}|dkrd|}| |} | d} q>|dkr|j|| d| \} }| |} q>|dkr|j|| d| \} }| |} q>|j ||dq>W|j ||ddS)zRparse an N3 string constant delimited by delim. return index, val rr+r,r"r*z&Quote expected in string at ^ in %s^%sz %02xz"Unicode error appending charactersz %s to string, because %srgrhz znewline found in string literalr9zunterminated string literal (2)z abfrtvn\"'z  \"'uUz bad escapezunterminated string literalN)rgrh)rlr1 interestingsearchr/start UnicodeErrorordsysexc_info__str__rrkrmr.uEscapeUEscape)r~rr%rZdelim1Zdelim2Zdelim3Zdelim4Zdelim5rustrrrYerrrZstreasonrrZuchr&r&r'rs$   $               zSinkParser.strconstcCs~t|||kr"t|j|||dy(|||jtd|||||fSt|j|||d||||YnXdS)Nzunterminated string literal(3)r9zbad string literal hex escape: )r1rrkrrZ)r~rr%rregnrr&r&r'_unicodeEscapeDs ( zSinkParser._unicodeEscapecCs|j|||tddS)Nrr)rr)r~rr%rr&r&r'rPszSinkParser.uEscapecCs|j|||tddS)Nr)rr)r~rr%rr&r&r'rSszSinkParser.UEscapecCst|j|j|||dS)N)rrkrl)r~rr%msgr&r&r'rVszSinkParser.BadSyntax)Nr"Nr"NF)F)N)N)-__name__ __module__ __qualname__rr2rrrrrrrrrrbrrrrrrrrrrr3rrrrrrrrrrrrrrrrrrrr&r&r&r'r_fsX @ `9Z  >2 W#Y?a r_c@s(eZdZddZddZeddZdS)rcCs(|jd|_||_||_||_||_dS)Nzutf-8)encode_str_i_whyrl_uri)r~rrlrr%rr&r&r'r^s zBadSyntax.__init__c Csx|j}|j}d}|dkr&d}|d}nd}t||dkr@d}nd}d|jd|j|j|||||||d|fS)Nr<z...r"z8at line %i of <%s>: Bad syntax (%s) at ^ in: "%s%s^%s%s"r*)rrr1rlrr)r~rr%rprepostr&r&r'rfs zBadSyntax.__str__cCst|S)N)rJ)r~r&r&r'messagexszBadSyntax.messageN)rrrrrpropertyr#r&r&r&r'r]sc@sLeZdZdZddZddZddZdd d Zdd d Zd dZ ddZ dS)FormularcCsJtj|_d|_tjd7_tj|_i|_i|_t|j |j d|_ dS)Nrr*)r identifier) rhexuuidcounterr%number existentialsZ universalsr rid quotedgraph)r~parentr&r&r'rs zFormula.__init__cCs d|jS)Nz _:Formula%s)r*)r~r&r&r'rszFormula.__str__cCstd|jS)Nz _:Formula%s)rr*)r~r&r&r'r,sz Formula.idNcCsJ|dkr,|jd7_td|j|jf}nt|jdjjdd}|S)Nr*zf%sb%sr!rr)r)rr(splitrr:)r~rrbnr&r&r'rs zFormula.newBlankNodecCst|jdjS)Nr!)r r/r)r~rrr&r&r'rszFormula.newUniversalcCs|j|j|<dS)N)rr+)r~rr&r&r'rszFormula.declareExistentialcCs|jS)N)r-)r~r&r&r'rsz Formula.close)NN)N) rrrr*rrr,rrrrr&r&r&r'r%s   r%z ([\x80-\xff])c@seZdZddZddZddZddZd d d Zd d ZddZ ddZ ddZ d!ddZ ddZ ddZddZddZddZd S)"RDFSinkcCsd|_d|_||_dS)Nr) rootFormular)graph)r~r3r&r&r'rszRDFSink.__init__cCs|jjjstt|j}|S)N)r3r formula_awarer/r%)r~fr&r&r'rzs zRDFSink.newFormulacCst|jj|S)N)rr3r)r~r&r&r&r'newGraphszRDFSink.newGraphcGs t|dS)Nr)r)r~rLr&r&r'rxszRDFSink.newSymbolNcCsnt|tr|j|St|ts&|dkrH|jd7_tdt|j}n"tt|djdjj dd}|S)Nr*rrr!rr) rr%rrr)rrJr/rr:)r~argrrr0r&r&r'rs  "zRDFSink.newBlankNodecCs |rt||dSt||dSdS)N)datatype)r)r )r~r;rrr&r&r'rs zRDFSink.newLiteralc Cs|jd}|s|S|jd}|jd}|j|}}xF|ddD]6}|j||||f|j|} |j|||| f| }qBW|j||||df|j||||f|S)Nz.http://www.w3.org/1999/02/22-rdf-syntax-ns#nilz0http://www.w3.org/1999/02/22-rdf-syntax-ns#firstz/http://www.w3.org/1999/02/22-rdf-syntax-ns#restr*rr)rxrr) r~rr5rErBrCafr`neZanr&r&r'rs" zRDFSink.newListcGst|S)N)set)r~rLr&r&r'rszRDFSink.newSetcGsdjdd|DS)Nr)css|]}t|VqdS)N)repr).0rr&r&r' sz.RDFSink.setDefaultNamespace..)r)r~rLr&r&r'rszRDFSink.setDefaultNamespacecCs|\}}}}t|drtd|j||}|j||}|j||}||jkr`|jj|||fn.t|tr~|jj|||fn|j|||fdS)NrzFormula used as predicate) hasattrrX normaliser2r3addrr%r-)r~rrr5pr;or&r&r'rs       zRDFSink.makeStatementcCst|trtt|dSt|tr}t|dks&t|dkr4dt|}n dt|}||}q Wt|S)zUse URL encoding to return an ASCII string corresponding to the given UTF8 string >>> hexify("http://example/a b") %(b)s'http://example/a%%20b' r"~!z%%%02Xz%c)r r)rr;rr&r&r'r3s    c@s"eZdZdZddZd ddZdS) rzK An RDFLib parser for Turtle See http://www.w3.org/TR/turtle/ cCsdS)Nr&)r~r&r&r'rNszTurtleParser.__init__utf-8Tc Csz|dkrtd|t|}|j|jp2|jp2d}t|||d}|j|jx"|jj D]\}} |j || q^WdS)Nutf-8*N3/Turtle files are always utf-8 encoded, I was passed: %sr")rrt)NrN)rOrP) rXr1Z absolutize getPublicId getSystemIdr_r getByteStreamriitemsrb) r~sourcer3encodingrtsinkrrBr namespacer&r&r'parseQs zTurtleParser.parseN)rMT)rrr__doc__rrYr&r&r&r'rFsc@s"eZdZdZddZdddZdS) rz` An RDFLib parser for Notation3 See http://www.w3.org/DesignIssues/Notation3.html cCsdS)Nr&)r~r&r&r'rmszN3Parser.__init__utf-8cCsJ|jjs t|jjstt|jd}||_|j|_tj||||dddS)N)rF)rt) rZ context_awarer/r4rZdefault_contextZnamespace_managerrrY)r~rUr3rVZ conj_graphr&r&r'rYps    zN3Parser.parseN)r[)rrrrZrrYr&r&r&r'rdscCsddl}|jdS)Nr)doctesttestmod)r\r&r&r'_test~sr^cCst}t|}dtjjtjtjd}t||d}|j d|j d<|j t tjdd}|j }|j|j||jx|jdD] }t|qWdS)Nzfile://r*)rr!r"rb)NNN)rr1r7r3rr8r argvr_ryriropenrrrrZquadsprint)grWZbase_urirBr5Zrdbytesrr&r&r'mains  rd__main__)r)rZ __future__rrrtypesr r7rer]r[rQrr(rZ rdflib.termrrr r r r Z rdflib.graphr rrsixrrZ rdflib.compatrrrrr__all__Z rdflib.parserr rrrr6ZCONTEXTZPREDZSUBJOBJZPARTSZALL4SYMBOLZFORMULALITERALZ LITERAL_DTZ LITERAL_LANGZ ANONYMOUSZ XMLLITERALrZNODE_MERGE_URIZ forSomeSymZ forAllSymZ RDF_type_URIZ RDF_NS_URIZOWL_NSZDAML_sameAs_URIZ parsesTo_URIZRDF_specZList_NSZ _Old_Logic_NSZN3_firstZN3_restZN3_liZN3_nilZN3_ListZN3_EmptyrHrrIrrwZ chatty_flagrNrOZN3_forSome_URIZ N3_forAll_URIrrrZLOG_implies_URIrFrHrIZFLOAT_DATATYPErGZoption_noregenrrrZ_rdfnsrrrZcompilerrZN3CommentCharacterrrrsigned_integerrrrZ digitstringrrr_ SyntaxErrorrrr%Zr_hibyter1rrrr^rdrr&r&r&r's                `                  "'