Ñò rùMc@sSdZyddklZWn#ej oddklZnXddklZlZddkl Z ddk l Z ddk Z d„Z d„Z d „Zd „Zhhhe d „Zd he d „Zhdd„Zd„Ze d„Ze d„Zd„Ze d„Ze d„Zd„Zd„Ze d„Zhe d„Ze d„Zehe d„Zde d„Z!e d„Z"d„Z#d„Z$d „Z%d!„Z&d"„Z'd#„Z(d$„Z)d%„Z*d&„Z+d'„Z,d(„Z-e.d)jo[ddk/Z/e/i0d*ed+e/i1e/i2Bƒ\Z3Z4e4oe3 o d,e4GHne ƒndS(-sc Bachelor AI project 2009. Andreas van Cranenburgh 0440949. Two-word stage dialogue simulator using a corpus of exemplars. Interactive usage: $ python model.py Other uses: - Re-enact dialogue in 01.cha (should be a text file with one utterance per line, with each line preceeded by *MOT or *CHI, as in the CHILDES corpus): $ python # >>> import model # >>> model.evaluate(open('01.cha').read()) ... - Talk to self for 25 utterances: # >>> model.selftalk(25) ... - Demonstrate some sample utterances that work (but not from corpus): # >>> model.test() ... iÿÿÿÿ(tchoice(t uppercaset lowercase(t defaultdict(tpprintNcCs~dGdGHHdGHtƒiƒGHHtƒ}|idƒxCto;dGtƒ}|djoPn|i|ƒ}dG|GHq7WdS(s% Interactive textual user interface. sChild dialogue simulator.s+Enter parent utterance (or `quit' to quit).sUtterances in corpus: sParent: tquitsChild: N(t getexemplarstkeystdialoguetsendtNonetTruet raw_input(tdtutttreply((s /home/andreas/ai/baproj/model.pytmain3s     cGs"ditd„|DƒƒƒGHdS(s Print debug output (also see nodbg). Takes a variable number of arguments of any type. >>> dbg('choices', range(3)) choices [0, 1, 2] t css%x|]}t|ƒiƒVqWdS(N(tstrt expandtabs(t.0ta((s /home/andreas/ai/baproj/model.pys Os N(tjointlist(tm((s /home/andreas/ai/baproj/model.pytdbgHscGsdS(N((R((s /home/andreas/ai/baproj/model.pytnodbgPscCscddkl}l}y|t|ddƒƒSWn+|ƒ}||t|ddƒƒ|SXdS(Niÿÿÿÿ(tdumptloads.pickletrbtwb(tcPickleRRtopen(titemtfunRRttemp((s /home/andreas/ai/baproj/model.pytloadorgenerateRs c #sòˆptdtƒ‰nˆptd‡fd†ƒ‰n|ptd‡‡fd†ƒ}ndGˆGHdgdd}}}xhto`|V}t||ˆˆƒ}d|jo-t|ƒd jot| | ||ˆƒn|} |d | ƒtob|ˆjoU| oN| t|| ˆƒjo'|d ƒ| ˆ|dst constructionscs tˆˆƒS((tinferconstructions((R&R%(s /home/andreas/ai/baproj/model.pyR)gss!Lexicon (distilled from corpus): tt reinforcementis interpretation:sadded utterance to exemplarss/generation check failed, not added to exemplarss reaction:N(R,R,( R$RR t interprettlent reinforceR(t lexiconcheckR+tresponsetexpress2tappendt findtopic( R%R*R&tdebugt discoursettopict reactionuttRt newmeaningtmeaningtreaction((R&R%s /home/andreas/ai/baproj/model.pyR[s>      ic Cs+|p tƒ}nt|iƒƒ}d\}}t|gƒ}g}t|ƒ}|idƒx²t|ƒD]¤} |id||fƒ||dƒ||}}|i|ƒ} || joJ|djo=|dƒtt t|iƒƒ|ƒƒ}|i |ƒqm| }qmWdGdi |ƒGHdS( sº Talk to oneself, picking a random utterance from the exemplars when repetition is detected. Input: number of utterances to generate, exemplars Output: transcript of conversation s*MOTs*CHIs%s: %siÿÿÿÿspicking random utterances N(s*MOTs*CHI( RRRtsetRR R trangeR4RtaddR( tuteR6Rtspeakingt listeningtsaidtcR txtr((s /home/andreas/ai/baproj/model.pytselftalk…s(       " ic Cs|p tƒ}nt|ƒ}|idƒx»t|ƒD]­}g}xž|idƒD]}y|iddƒ\}}Wntj o qVnXd|joE|GHd|i|ƒ} | GHx%|| fD]} |i| ƒqÈWqVqVWq:WHdid„|DƒƒS(s When presented with a corpus fragment, feed parent utterances to model and generate model responses in place of child utterances. n = number of iterations to perform Input: string with transcript of corpus Output: same transcript with replies from model instead of child s t:itMOTs*CHI: cssx|]}t|ƒVqWdS(N(R(RR((s /home/andreas/ai/baproj/model.pys Âs N( RRR R R>tsplitt ValueErrorR4R( tconvR%tnR Rt modelconvtbtwhoRtansRF((s /home/andreas/ai/baproj/model.pytevaluate¥s,       cCs«did„tdƒDƒƒ}|dd|jodGndG|d|||ddÊs iiÿÿÿÿitupdatedtaddedRsreinforcement:twithN(RR>(R;R<R9R7R%trandomid((s /home/andreas/ai/baproj/model.pyR0Äs csˆd„‰|ddiƒ|ddiƒ‰‰tˆdˆƒ}|o8ˆ‡‡‡fd†|Dƒƒd}|d|ƒ|SdS(s² Look for a recurring element in the last two utterances, if found, this becomes the new topic. Input: discourse (list of utterances and meanings) Output: clause, or None cSsEt|ƒ}yd„|DƒiƒSWntj ot|ƒSXdS(Ncss6x/|](\}}|djo||fVqqWdS(iN((RRRP((s /home/andreas/ai/baproj/model.pys ãs (Rtnextt StopIterationtmin(tseq((s /home/andreas/ai/baproj/model.pyt minprefertwoàs  iÿÿÿÿiiþÿÿÿc3sCx<|]5}ˆˆi|ƒ|fˆi|ƒ|ffƒVqWdS(N(tindex(RR(tm2R\tm1(s /home/andreas/ai/baproj/model.pys îs s topic:N(RKt intersect(R7R6tcommonR8((R_R^R\s /home/andreas/ai/baproj/model.pyR5Ùs ) cCsgd„}|iddƒ}|iƒ}g}|iƒD]%\}} ||||ƒ|| fq8~} t| ƒd} | o>g} | D]&}|d| jo| |dqˆqˆ~ } ndSg} x³| D]«\}}g}|D]} | |jo || qãqã~}t||||tƒ\} }|oH| i| tt|||ƒ|ƒdt t |ƒƒ||ffƒqÌqÌW| oÝt | ƒd}|d ƒ|d |ƒt ||dt | ƒƒ}|o |} n |d} g}|D](}||diƒjo ||qëqë~}|d td i|ƒƒƒ|ot|| |||ƒdS| SdSd S(sö Return semantic representation for linguistic utterance. This function is in charge of backtracking over initial exemplars and picking the best result, the rest is done by interpretwith() Input: utterance Output: meaning representation cSstt||iƒƒƒS(N(R/R`RK(RRP((s /home/andreas/ai/baproj/model.pytmatchûss[= s[=iiR,gð?iÿÿÿÿsinitial exemplar:s s words left:RN(treplaceRKtitemstmaxt interpretwithRR4t edit_distR1R/ttokensRZt conciliatet varclausestreprR(t utteranceR8R%R&R6Rbtwordst_[1]RRPtinitialRNt_[2]titexuttt exmeaningt_[3]twR;REt_[4]((s /home/andreas/ai/baproj/model.pyR.ôsB  ?> 2     <csrd„‰tˆƒ‰di‡‡fd†|iƒDƒƒ‰‡‡fd†‰di‡fd†|iƒDƒƒS(NcSs.|id|dd!|dd!ƒiƒiƒS(Ns[=%s]iiÿÿÿÿ(Rctstriptlower(R((s /home/andreas/ai/baproj/model.pytelimBsRc3s:x3|],}ˆ|ƒˆjoˆˆ|ƒVqqWdS(N((RRP(Rytlex(s /home/andreas/ai/baproj/model.pys Fs cs(|ˆjp|ˆjp|ddjS(NiÿÿÿÿRI((R(tuttlextrevlex(s /home/andreas/ai/baproj/model.pytcheckKsc3s*x#|]}ˆ|ƒo |VqqWdS(N((RR(R}(s /home/andreas/ai/baproj/model.pys Ms (trevdictRRK(RR;Rz((R|RyR{RzR}s /home/andreas/ai/baproj/model.pyR1>s   (cCset|ƒ}|od|dddjoO|ddi|iƒƒiƒjo%|dt|dƒƒ|d}qwn|o'g}|D]}d|jo ||q‰q‰~oôg}|D]}d|jo ||q¼q¼~ddd!iƒiƒ} | |jodt|| |d|ƒ} | o| }|d|ƒqkdS|dcd|| 7<|d |ƒng} |D]"} | ddjo | | qvqv~ }nx§t |ƒD]™} di| ƒ}x|D]y}||joft | ƒi |iƒƒoJ|} g}| D]"} | |jo||| qq~}t||||d g|ƒ} | oãg}|D]}|| jo ||qkqk~}g}|D]}|| jo ||qq~}|d tdi|ƒƒd t|ƒƒ|d t||ƒƒ|dt| ƒƒt || |||ƒ\}}d||fSqÎqÎWq²W|odSd|fS(s‘ Interpretation helper function called by interpret(), work out meaning of remaining words by stitching together the best matching fragments. it[Rtskippingiiiÿÿÿÿsdemonstrative dereferenced:R,sappended demonstrative:t tins ands matchesN(iR,(iR,( R/RRRKRkRwRxRiR t substr_iterR=tissubsetRf(RmtpartialmeaningR%R&R6tlRnRuRptwordtpmRtRtsubt substringRPRvtsubstt_[5]REtmatchest_[6]Rtp((s /home/andreas/ai/baproj/model.pyRfPsH &:I : )6 22(c sðd„‰d|jod|idƒdSd|jotdƒ}|Sd |joƒd|jod Sd |iddƒd}‡fd †}g}|iƒD]}||||ƒ|fq£~}t|ƒd djoóg} |D]&}|d djo| |dqêqê~ } |dt| ƒƒ| ot| ƒ} nd St| |t|ƒƒ}|oht| |ƒ} | o)t| ƒdjo|d|ƒn| St| ƒdjo|d|ƒn|Sqì|Snd|jo|Sd S(s Transform a meaning into a response using adjacency pairs of speech acts. Input: meaning representation Output: meaning representation cSs<tg}t|ƒD]}t|ƒo ||qq~ƒS(N(R/Rhtvar(RRnR((s /home/andreas/ai/baproj/model.pytvarsŽst imperativesacknowledgement RIit ynquestiont agreementtdenialt whquestionR,s assertion:cs'tt|ƒt|ƒƒdˆ|ƒS(Ngà?(RgRh(RR(R‘(s /home/andreas/ai/baproj/model.pytscorežsispossible reactions:schoice:t assertion(R”R•( RKRtvaluesRZtsortedRiRjtunifiesR/( R;R%R6topRR—RnRR RptchoicesRGRN((R‘s /home/andreas/ai/baproj/model.pyR2ˆs@      6:  c CsÛ||iƒjott|ƒ|ƒSd„}g}|iƒD]"\}}||||ƒ|fqB~}t|ƒdt|ƒjoItg}|D],}|dt|ƒdjo ||q˜q˜~ƒdSdSdS(s; Express `meaning' by returning the most similar exemplar. cSst|iƒ|iƒƒS(N(RgRK(RR((s /home/andreas/ai/baproj/model.pytdistÇsiit0N(R™RR~RdRZR/( R;R%R&RžRnRRPR Rp((s /home/andreas/ai/baproj/model.pytexpressÀs <Ic sE‡fd†‰ˆ|jo"di‡fd†|ˆDƒƒSˆˆiƒjottˆƒˆƒSt‡fd†ˆiƒDƒƒoCtg}ˆiƒD]$\}}ˆ|jo ||qq~ƒSt‡fd†ˆiƒDƒƒoCtg}ˆiƒD]$\}}|ˆjo ||qq~ƒStˆ|ˆ|ƒS(sD Express `meaning' by returning a matching exemplar or construction cs|diˆiƒƒjS(NR(RR(R(R*(s /home/andreas/ai/baproj/model.pyR}ÒsRc3s*x#|]}ˆ|ƒo |VqqWdS(N((RR(R}(s /home/andreas/ai/baproj/model.pys Õs c3sx|]}ˆ|jVqWdS(N((Rtclauses(R;(s /home/andreas/ai/baproj/model.pys Ús c3sx|]}|ˆjVqWdS(N((RR¡(R;(s /home/andreas/ai/baproj/model.pys Ýs (RR™RR~tanyRdR3(R;R%R*R&RnRR¡Rp((R;R*R}s /home/andreas/ai/baproj/model.pyt expressmultiÏs "# ## #cs«‡fd†}ˆ|iƒjo.tt|ƒˆƒ}|d|ƒ||ƒSˆ|iƒjo.tt|ƒˆƒ}|d|ƒ||ƒSt‡fd†|iƒDƒƒo\tg}|iƒD]$\}}ˆ|jo ||qÈqÈ~ƒ}|d|ƒ||ƒStˆƒ} dˆjoˆiƒ‰nˆ| jo | ˆSdStˆdƒ ptˆdƒdtjoˆd} ndS|d| ƒx.ˆdD]"} | | jot| | ƒSqªW| | jo| | iƒStˆdƒ} | oF| dtjo5t ˆƒd jot ˆd ƒ} qIˆd} nt ˆdƒ} g}ˆiƒD]$\} }| |jo || qjqj~}|o[tg}ˆiƒD]$\} }| |jo || q²q²~ƒ}|d | d |ƒnd}tddgƒo|Sg}ˆiƒD]1\} }| |jo|| jo || q(q(~}|o1t|ƒ} | d |}|d | d |ƒn|d| ƒ|S(sÕ Transform a meaning into a two word utterance using exemplars, constructions or the lexicon. Filter result according to lexical knowledge. Input: meaning representation Output: one or two word utterance cs;di‡fd†|iƒDƒƒ}|djodS|S(NRc3s0x)|]"}|iƒˆjo |VqqWdS(N(Rx(RR(R&(s /home/andreas/ai/baproj/model.pys ìs R,(RRKR (RRG(R&(s /home/andreas/ai/baproj/model.pytreduceës% sreduced:c3sx|]}ˆ|jVqWdS(N((RR¡(R;(s /home/andreas/ai/baproj/model.pys ýs RIR,istrying to express:iiscomment:R‚Rstopic:stopic not expressible?( R™RR~R¢RdRKtargRtpopR/tpred(R;R%R*R&R6R¤RRnR¡R|tclauseRR8tcommentRpRPt candidatesRtRv((R&R;s /home/andreas/ai/baproj/model.pyR3ãsd    # $      -    >DK  c sz|dƒttƒ}t|ƒ}xê|iƒD]Ü\}}xÍt|iƒƒD]¹}t|ƒdjoPndid„|Dƒƒ‰dˆjoqNnt‡fd†|DƒƒpQtg} |D]} ˆ| jo | | qÄqÄ~ ƒ} | o| |ˆHs N(tallRx(RRP((s /home/andreas/ai/baproj/model.pys Hs c3sx|]}ˆ|jVqWdS(N((RRP(Rm(s /home/andreas/ai/baproj/model.pys Ls isconstruction (score %d)s (score %d)snot found in candidates %ssmeaning of construction(RtintR/RdRƒRKRR¢RšR~RR¦t findmeaningRRkRR+(R%R&R*R6tscoresRNRR;RRnRER—tformRptmst candidate((Rms /home/andreas/ai/baproj/model.pyR+9sH     8     > c s„g}|iƒD]$\}}ˆ|jo ||qq~‰t|ƒ‰tˆƒdjog}ˆD]}||iƒqh~} tt| d| dƒ} | o6g} ‡‡fd†‰di‡fd†| DƒƒSt‡fd†ˆDƒƒdStt ƒ} xJˆD]B}x9t |iƒƒD]%}di|ƒ} | | cd7‚s it(s(X)(R¢(R(R|R¯(s /home/andreas/ai/baproj/model.pytabstracts .Rc3sx|]}ˆ|ƒVqWdS(N((RR(R³(s /home/andreas/ai/baproj/model.pys ‰s c3s(x!|]}ˆi|ƒ|fVqWdS(N(tcount(RR(R°(s /home/andreas/ai/baproj/model.pys s N( RdR~R/RKR¤R`RReRR¬RƒR¦( R¯R%R&R6RnRR;RpRRRNR®((R³R|R°R¯s /home/andreas/ai/baproj/model.pyR­rs(> ' cs"td„ˆiƒDƒƒ}t|ƒ}xóˆiƒD]å\}}d|joÌt||idƒd|idƒ!ƒ}||idƒ iƒd}|||idƒdiƒd7}|iƒ|iƒ} ||idƒ ||idƒdˆ|<| |jo||| ¢s Rit]iÿÿÿÿiRcs"diˆiƒƒit|ƒƒS(NR(RR™R´R§(RP(R%(s /home/andreas/ai/baproj/model.pytfreq»sc3s%x|]}ˆ|ƒ|fVqWdS(N((RRP(R¶(s /home/andreas/ai/baproj/model.pys ½s R,slexicon: not found:s (tdictRdR/R¬R]RKRxR=RRRR¤R`RZR(R¦R'(R%tverboseR&R6texlcaseRNRRR¨R‡RnRt utterancesRpRPtmeaningsRªtnotfound((R%R¶s /home/andreas/ai/baproj/model.pyR'žsT   )%, 2 DJ  #4.1c Csn|}|d jo|iƒ}n|djo|S|}tSt}x|iƒD]}t|ƒd|jo"t|ƒo tt|ƒƒpt|ƒ|j oät|ƒt|ƒit|ƒƒd}tg}|D]} |t|ƒ| jqÚ~joƒ|dt jo9|d|dt|ƒƒ|i |t|ƒƒ}}qƒ|dt|ƒd|ƒ|i |t|ƒƒ}}q‡nt}nt|ƒt jo|Sng} |iƒD]*} t| ƒt jo| t| ƒq½q½~ } | ocd|joVt|ƒ|joC|i | d t|ƒƒ}|d t|ƒƒt ||||ƒSqRW|o|SdS( se Test whether meaning can be made comptabible with partialmeaning, by looking for a family resemblance with partialmeanings, if found, perform a substitution if necessary. Returns empty string on failure. >>> conciliate('question: animal(bunny) do(X)', 'assertion: point(dog) animal(dog)') substituted (bunny) for (dog) 'assertion: point(bunny) animal(bunny)' >>> conciliate('assertion: do(hop) animal(bunny)', 'assertion: animal(bunny) do(X)') instantiated (X) with (hop) 'assertion: animal(bunny) do(hop)' >>> conciliate('assertion: do(hop) animal(bunny)', 'assertion: point(bunny)') '' R,R²it instantiatedRVt substitutedtforRIis$instantiated variable predicate withN( R RKR R(R§R¥RRhR]RRcRi( R;R…R‹R6RˆtsuccesstparttoldargRnRRpR‘((s /home/andreas/ai/baproj/model.pyRi×sB   4)4 $ D' cCsß|}|djo|S|}tSx¶t|iƒ|iƒƒD]™\}}||joq>q>tt|ƒƒobt|ƒt|ƒit|ƒƒd}|d|dt|ƒƒ|it|ƒ|ƒ}}q>dSq>W|S(sù succeed if everything in "partialmeaning" is compatible with "meaning", substituting along the way. Similar to conciliate() but operates on the whole meaning instead of looking at individual clauses, and does not perform substitution, merely instantiation. >>> unifies('assertion: do(hop) animal(bunny)', 'assertion: do(X) animal(bunny)') substituted (hop) for (X) 'assertion: do(hop) animal(bunny)' >>> unifies('assertion: do(hop) animal(bunny)', 'assertion: animal(bunny) do(X)') '' R,iR¾R¿( R tzipRKRR¥RhR]R§Rc(R;R…R6Rˆtpart1tpart2tnewarg((s /home/andreas/ai/baproj/model.pyR›s    ) cCsyddkl}Wntj on X|ƒSh!dd6dd6dd6dd 6dd 6dd 6d d 6dd6dd6dd6dd6dd6dd6dd6dd6dd6d d!6d"d#6d$d%6d&d'6d&d(6d)d*6d+d,6d-d.6d/d06d1d26d3d46d5d66d7d86d9d:6d;d<6d=d>6d?d@6S(Asl Obtain corpus, either by importing from module `exemplars,' or by falling back to a small sample corpus. iÿÿÿÿ(RR,tuhtehtacknowledgementtaht confirmationtoktyestyeahR•tnoR-sthat's right !s assertion: point(ball) toy(ball)tballsimperative: throw[0](X) toy(X)sthrow it to mes(imperative: kick(football) toy(football)syou kick the football !swhquestion: point(X) Y(X)swhat is this ?s"whquestion: point(cow) animal(cow)sis that a cow ?swhquestion: do(X) animal(bunny)swhat does a bunny do ?swhquestion: do(X) animal(cow)swhat does a cow say ?s!whquestion: do(X) animal(duck[3])swhat does a duckie say ?swhquestion: animal(X) do(woof)swhat animal does woof woof ?s whquestion: do(X) animal(cat[2])swhat's a kitty say ?s$assertion: point(cat) animal(cat[2])sthat's a kittys'assertion: point(donkey) animal(donkey)tdonkeysthat's a donkeys assertion: do(meouw) animal(cat)tmeouws!assertion: do(quack) animal(duck)tquacksassertion: do(moo) animal(cow)tmoos assertion: animal(bunny) do(hop)tbunnysassertion: animal(dog) do(woof)tdogs assertion: do(hop) animal(bunny)thopsassertion: do(woof) animal(dog)s woof woofs#ynquestion: want(juice) food(juice)swant some juice ?sassertion: food(chocolate)t chocolateswhquestion: inside(X) toy(box)swhat's inside ?s assertion: inside(book) toy(box)tbooksassertion: point(book)sbook !(R%Rt ImportError(R((s /home/andreas/ai/baproj/model.pyR<sL c Cswddddddddd d d g }tƒ}t|ƒ}x4|D],}|GHt|d ||tƒ}d G|GHqCWdS(s# Interpret some sample utterances. swhat does a bunny do ?s kitty do ?sthrow the ballscatch the ballswhere lives birdie ?s that's a ballscan you throw it ?swhat does a duckie say ?s!what does this animal [=cow] do ?s'do you want to play with that [=ball] ?swhat does a ball say ?R,sinterpretation:N(RR'R.R(ttestR%R&RRP((s /home/andreas/ai/baproj/model.pyRÛos$   c sƒ‡‡fd†}d„}g}tt|ƒdƒD]‰|dgt|ƒdq6~}x$t|ƒD]\}}||d>> edit_dist('foo', 'bar') 2.4890507991136213csˆˆjodSdSdS(Nii(((RRP(s /home/andreas/ai/baproj/model.pyt substcost‘s cSstid|S(Ngš™™™™™É¿(tmathRA(RN((s /home/andreas/ai/baproj/model.pytdecay—siiiÿÿÿÿ(R>R/t enumerateRZ( tsourcettargetRÜRÞRntdistanceRqtrowtjtinsctsubstctdelc((RRPs /home/andreas/ai/baproj/model.pyRg‰s& B   -ccs[xTtt|ƒddƒD]:}x1t|t|ƒƒD]}||||d!Vq5WqWdS(s‹ Return all substrings of a sequence, in descending order of length. >>> list(substr_iter('abc')) ['abc', 'ab', 'bc', 'a', 'b', 'c']iÿÿÿÿiN(R>R/(R[RRP((s /home/andreas/ai/baproj/model.pyRƒ¯s cs t‡fd†ˆiƒDƒƒS(s¿ Reverse dictionary, ie. swap values and keys; since a value may occur with multiple keys, return a list of all keys associated with a value. >>> revdict({0: 1, 1: 2}) {1: [0], 2: [1]}c3sWxP|]I}|g}ˆiƒD]$\}}||jo ||q q ~fVqWdS(N(Rd(RRRnRPRE(R (s /home/andreas/ai/baproj/model.pys Às (R·R™(R ((R s /home/andreas/ai/baproj/model.pyR~¸scCs@g}|diƒD]$}tt|ƒƒo ||qq~S(s1 >>> varclauses(['foo(bar) zed(X)']) ['zed(X)']i(RKRR¥(RRnR((s /home/andreas/ai/baproj/model.pyRjÂscCs'd|jo||idƒ S|SdS(s >>> pred('foo(bar)') 'foo'R²N(R](ttoken((s /home/andreas/ai/baproj/model.pyR§Çs cCs'd|jo||idƒSdSdS(s >>> arg('foo(bar)') '(bar)'R²R,N(R](Rè((s /home/andreas/ai/baproj/model.pyR¥Îs cCs&|o|dtjp|dtjS(se Test whether an argument string is variable: >>> var('(bar)') False >>> var('(X)') True ii(R(R¥((s /home/andreas/ai/baproj/model.pyRÕscCs|iddƒiƒS(s© Turn meaning into a list of operator, predicates and arguments. >>> tokens('whquestion: do(X) animal(bunny)') ['whquestion:', 'do', '(X)', 'animal', '(bunny)'] R²s ((RcRK(R((s /home/andreas/ai/baproj/model.pyRhßscCst|ƒit|ƒƒS(s, >>> intersect([1,2,3], [2,3]) set([2, 3])(R=t intersection(RRP((s /home/andreas/ai/baproj/model.pyR`çst__main__R¸t optionflagss%d doctests succeeded!(5t__doc__t free_willRRÚtrandomtstringRRt collectionsRRRÝRRRR$RRHRSR0R5R.R1RfR2R R£R3R+R­R(R'R RiR›RRÛRgRƒR~RjR§R¥RRhR`t__name__tdoctestttestmodtNORMALIZE_WHITESPACEtELLIPSIStfailt attempted(((s /home/andreas/ai/baproj/model.pyts^     *    J  8 8   V9 ,9F  3  &