4Ic @sRdZddkZddkZddklZddklZddklZddk l Z ddk l Z ddk TddkTd efd YZd efd YZd efdYZhdd<dd<dd<dd<dd<dd<dd<dd<dd <d!d"<d#d$<d%d&<d'd(<d)d*<d+d,<d-d.<d/d0<d1d2<d3d4<d5d6<d7d8<d9d:<d;d<<d=d><d?d@<dAdB<dCdD<dEdF<dGdH<dIdJ<dKdL<dMdN<dOdP<dOdQ<dRdS<dTdU<dVdW<dXdY<dZd[<d\d]<d^d_<d`da<dbdc<ddde<dfdg<dhdi<djdk<dldm<dndo<dpdq<drds<dtdu<dvdw<dxdy<dzd{<d|d}<d~d<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd<dd1scssx|]}|d VqWdS(iN((R R ((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pys 2ss5Items in "psd" and "pos" subdirectories do not match.s%s.psds%s.pos( t CorpusReadert__init__tYCOEParseCorpusReadertroottjoint _psd_readertYCOETaggedCorpusReadert _pos_readertsettfileidst ValueErrortsortedt _documents(tselfRRt documentst_[1]tdoct_[2]R((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR (s$$&%)cCs|tjo |iSnt|to |g}nx2|D]*}||ijotd|q<q<Wttd|DS(s Return a list of document identifiers for all documents in this corpus, or for the documents with the given file(s) if specified. sFile id %s not foundcssx|]}|d VqWdS(iN((R R ((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pys Is(tNoneRt isinstancet basestringt_fileidstKeyErrorRR(RRR ((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR;s   c Cs|tjo |iSnt|to |g}nttg}|D]}|d|qF~g}|D]}|d|qh~S(s Return a list of file identifiers for the files that make up this corpus, or that store the given document(s) if specified. s%s.poss%s.psd(RR!RR RR(RRRRR((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyRKs    (cCs|tjo |i}ntt|to |g}nxS|D]K}||ijo5|ddjotdqtd|q>q>Wg}|D]}|d||fq~S(s Helper that selects the appropriate fileids for a given set of documents from a given subcorpus (pos or psd). is.poss.psdsvExpected a document identifier, not a file identifier. (Use corpus.documents() to get a list of document identifiers.s Document identifier %s not founds%s.%s(s.poss.psd(RRRR R(RRt subcorpustdocumentRtd((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyt _getfileidsWs    cCs|ii|i|dS(NR (RtwordsR&(RR((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR'nscCs|ii|i|dS(NR (RtsentsR&(RR((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR(pscCs|ii|i|dS(NR (RtparasR&(RR((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR)rscCs|ii|i|dS(NR (Rt tagged_wordsR&(RR((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR*tscCs|ii|i|dS(NR (Rt tagged_sentsR&(RR((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR+vscCs|ii|i|dS(NR (Rt tagged_parasR&(RR((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR,xscCs|ii|i|dS(NR(Rt parsed_sentsR&(RR((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR-zssEUse .raw() or .words() or .tagged_words() or .parsed_sents() instead.tparsedcCs|djo|i|Sn|djo|i|Sn|djo|i|Sn|djo|i|Sn|djotdntd|dS(NR.trawt tokenizedttaggedtchunkedsno longer supporteds bad format %r(R-R/R'R*R(Rtitemstformat((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pytread~s sUse .parsed_sents() instead.cCs |i|S(N(R-(RR3((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR.ssUse .words() instead.cCs |i|S(N(R'(RR3((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR0ssUse .tagged_words() instead.cCs |i|S(N(R*(RR3((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR1ssOperation no longer supported.cCstddS(Ns$format "chunked" no longer supported(R(RR3((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR2s(t__name__t __module__t__doc__RR RRR&R'R(R)R*R+R,R-RR5R.R0R1R2(((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR"s,               RcBseZdZdZRS(srSpecialized version of the standard bracket parse corpus reader that strips out (CODE ...) and (ID ...) nodes.cCs@tidd|}tid|odSnti||S(Ns(?u)\((CODE|ID)[^\)]*\)ts\s*\(\s*\)\s*$(tretsubtmatchRRt_parse(Rtt((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR=s(R6R7R8R=(((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyRsRcBseZedZRS(c CsAd}t|dt}ti|||ddd|d|dS(Ns,(?u)\(?<=/\.)\s+|\s*\S*_CODE\s*|\s*\S*_ID\s*tgapstsept_tsent_tokenizerR(RtTrueRR (RRR3Rtgaps_reRB((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyR s (R6R7RR (((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyRssAdrian and Ritheuss coadrian.o34slfric, Supplemental Homiliess coaelhom.o3slfric's Lives of Saintss coaelive.o3sAlcuin De virtutibus et vitiistcoalcuinsAlexander's Letter to Aristotles coalex.o23sApollonius of Tyres coapollo.o3t Augustinetcoaugusts$Bede's History of the English Churchs cobede.o2sBenedictine Rules cobenrul.o3sBlickling Homiliess coblick.o23s#Boethius' Consolation of Philosophys coboeth.o2sByrhtferth's Manuals cobyrhtf.o3sCanons of Edgar (D)t cocanedgDsCanons of Edgar (X)t cocanedgXslfric's Catholic Homilies Is cocathom1.o3slfric's Catholic Homilies IIs cocathom2.o3s Saint Chads cochad.o24sChrodegang of Metz, RuletcochdrulsSaint Christophert cochristophsAnglo-Saxon Chronicle As cochronA.o23sAnglo-Saxon Chronicle CtcochronCsAnglo-Saxon Chronicle DtcochronDsAnglo-Saxon Chronicle Es cochronE.o34sCura Pastoraliss cocura.o2sCura Pastoralis (Cotton)tcocuraCs Dicts of Catos codicts.o34sDocuments 1 (O1)s codocu1.o1sDocuments 2 (O1/O2)s codocu2.o12sDocuments 2 (O2)s codocu2.o2sDocuments 3 (O2/O3)s codocu3.o23sDocuments 3 (O3)s codocu3.o3sDocuments 4 (O2/O4)s codocu4.o24s Honorius of Autun, Elucidarium 1tcoeluc1tcoeluc2slfric's Epilogue to Genesiss coepigen.o3sSaint Euphrosynetcoeuphrs Saint Eustace and his companionstcoeusts Exodus (P)t coexodusPs Genesis (C)t cogenesiCsGregory's Dialogues (C)s cogregdC.o24sGregory's Dialogues (H)s cogregdH.o23sPseudo-Apuleius, Herbariumtcoherbars"Wulfstan's Institute of Polity (D)s coinspolD.o34s"Wulfstan's Institute of Polity (X)t coinspolXs Saint JamestcojamestLacnungas colacnu.o23t Leechdomss colaece.o2s Laws, Cnut Is colaw1cn.o3s Laws, Cnut IIs colaw2cn.o3sLaws, thelred Vs colaw5atr.o3sLaws, thelred VIs colaw6atr.o3s Laws, Alfreds colawaf.o2sAlfred's Introduction to Lawss colawafint.o2s Laws, Gerefas colawger.o34s Laws, Ines colawine.ox2sNorthumbra Preosta Laguscolawnorthu.o3sLaws, William I, Lads colawwllad.o4tLeofrics coleofri.o4slfric's Letter to Sigefyrths colsigef.o3s lfric's Letter to Sigeweard (B)t colsigewBs lfric's Letter to Sigeweard (Z)s colsigewZ.o34slfric's Letter to Wulfgeattcolwgeatslfric's Letter to Wulfsige (T)t colwsigeTs lfric's Letter to Wulfsige (Xa)scolwsigeXa.o34slfric's Letter to Wulfstan Is colwstan1.o3slfric's Letter to Wulfstan IIs colwstan2.o3sSaint Margaret (C)s comargaC.o34sSaint Margaret (T)tcomargaTsMartyrology, Itcomart1sMartyrology, IItcomart2sMartyrology, IIIs comart3.o23sMarvels of the Easts comarvel.o23s Mary of Egypttcomarys Saint NeottconeotsGospel of Nicodemus (A)tconicodAsGospel of Nicodemus (C)tconicodCsGospel of Nicodemus (D)tconicodDsGospel of Nicodemus (E)tconicodEtOrosiuss coorosiu.o2t Heptateuchs cootest.o3s'lfric's Preface to Catholic Homilies Iscoprefcath1.o3s(lfric's Preface to Catholic Homilies IIscoprefcath2.o3sPreface to the Cura Pastoraliss coprefcura.o2slfric's Preface to Genesiss coprefgen.o3s#lfric's Preface to Lives of Saintsscopreflives.o3s"Preface to Augustine's Soliloquiest coprefsolilos*Pseudo-Apuleius, Medicina de quadrupedibuss coquadru.o23sHistory of the Holy Rood-TreetcoroodsSeven Sleeperst cosevenslsSt. Augustine's SoliloquiestcosolilosSolomon and Saturn Is cosolsat1.o4sSolomon and Saturn IIt cosolsat2slfric's De Temporibus Annis cotempo.o3sVercelli HomiliestcoverhomsVercelli Homilies (E)t coverhomEsVercelli Homilies (L)t coverhomLsSaint Vincent (Bodley 343)tcovinceBsVindicta SalvatoristcovinsalsWest-Saxon Gospelss cowsgosp.o3sWulfstan's Homiliess cowulf.o34(R8tosR:t nltk.tokenizeRt nltk.corpus.reader.bracket_parseRtnltk.corpus.reader.taggedRtstringRtnltk.internalsRtutiltapiR RRRR(((s-/p/zhu/06/nlp/nltk/nltk/corpus/reader/ycoe.pyss    s