3Ic @sddkZddklZddkZddklZlZddkTdZdZ dZ dZ d Z d Z d Zhd d <dd<dd<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:YZd?Zed@jo endAgZdS(BiN(tjoin(tcorpustin_idle(t*s[^/ ]+s\bs <>s <>s <>s <>s)English: Brown Corpus (Humor, simplified)cCstiidtS(t simplify_tags(Rtcess_catt tagged_sentstTrue(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pytss%Catalan: CESS-CAT Corpus (simplified)cCs tiiS((RtbrownR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR!ssEnglish: Brown CorpuscCstiidtS(R(RR RR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR#ss"English: Brown Corpus (simplified)cCs"tiiddddgdtS(t categoriestnewst editorialtreviewsR(RR RR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR%ss)English: Brown Corpus (Press, simplified)cCstiidddtS(R treligionR(RR RR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR'ss,English: Brown Corpus (Religion, simplified)cCstiidddtS(R tlearnedR(RR RR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR)ss+English: Brown Corpus (Learned, simplified)cCstiidddtS(R tscience_fictionR(RR RR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR+ss3English: Brown Corpus (Science Fiction, simplified)cCstiidddtS(R tromanceR(RR RR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR-ss+English: Brown Corpus (Romance, simplified)cCstiidddtS(R thumorR(RR RR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR/scCs tiiS((Rtnps_chatt tagged_posts(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR1ssEnglish: NPS Chat CorpuscCstiidtS(R(RRRR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR3ss%English: NPS Chat Corpus (simplified)cCs tiiS((RttreebankR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR5ss#English: Wall Street Journal CorpuscCstiidtS(R(RRRR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR7ss0English: Wall Street Journal Corpus (simplified)cCs tiiS((Rtsinica_treebankR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR9ssChinese: Sinica CorpuscCstiidtS(R(RRRR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR;ss#Chinese: Sinica Corpus (simplified)cCs tiiS((RtalpinoR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR=ssDutch: Alpino CorpuscCstiidtS(R(RRRR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR?ss!Dutch: Alpino Corpus (simplified)cCstiiddS(tfiless hindi.pos(RtindianR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRAssHindi: Indian Languages CorpuscCstiidddtS(Rs hindi.posR(RRRR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRCss+Hindi: Indian Languages Corpus (simplified)cCs tiiS((RtflorestaR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyREss&Portuguese: Floresta Corpus (Portugal)cCstiidtS(R(RRRR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRGss2Portuguese: Floresta Corpus (Portugal, simplified)cCs tiiS((Rt mac_morphoR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRIss&Portuguese: MAC-MORPHO Corpus (Brazil)cCstiidtS(R(RRRR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRKss2Portuguese: MAC-MORPHO Corpus (Brazil, simplified)cCstiidtS(R(Rtcess_espRR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRMss%Spanish: CESS-ESP Corpus (simplified)tConcordanceSearchViewcBsgeZdZdZdZdZdZdZdZdZ dZ d Z d Z d Z d Zd ZdZdZdZdZdZdZdZdZdZdZdZdZdZdZdZdZdZ dZ!d Z"d!Z#d"Z$d#Z%d$Z&d%Z'd&Z(d'Z)d(Z*RS()s#FFFs#F00t HL_WRD_TAGs#C0C0C0t HL_LBL_TAGg333333?cCsst|_|ii|t|_|i|i|i|i|i|i|i |ii dS(N( tConcordanceSearchModeltmodelt add_listenertTkttopt _init_topt _init_menubart _init_widgetst_bind_event_handlerst load_corpustDEFAULT_CORPUS(tself((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyt__init__^s    cCsA|id|id|id|i|idddS(Ns 950x680+50+50sNLTK Concordance Searchs ii(tgeometryttitletbindtdestroytminsize(R+R$((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR%hs  c Cst|td|idddddd|_|i|i|i|i|i|i|i|i|i|i|ii dddt dS( Nt backgroundtpadxitpadytbordertfilltbothtexpand( tFrametdictt_BACKGROUND_COLOURt main_framet_init_corpus_selectt_init_query_boxt_init_results_boxt _init_pagingt _init_statustpackR(R+tparent((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR'ns0c Cs`t|i|_t|i|_t|i|_t|i}t|dddd}|iddddd|id d |idd ddd |t|dd}t|dd}|i dd d|iddddd|i |i ddd|iddddd|i |i ddd|iddddd|i |i d|iddddd |t|dd}t|dd}|i ddd|iddddd|i |i ddd|iddddd|i |i ddd|iddddd|i |i d|iddddd |t|dd}|i ddd|iddddd|i |i ddd|iddddd|i |i dd d|idddd!d|i |i d|idd"ddd ||idd#ddd ||idd$ddd ||iid |dS(%Nttearoffit borderwidthtlabeltExitt underlineitcommandt acceleratorsCtrl-qtFiletmenut20tvariabletvalueit50i2t100ids Result Counts 60 charactersi<s 80 charactersiPs100 characterstBefores 70 charactersiFs 90 charactersiZs110 charactersintAftertContexttEdit(tIntVarR$t _result_sizet _cntx_bf_lent _cntx_af_lentMenut add_commandR0t add_cascadetadd_radiobuttontset_result_sizetinvoketset_cntx_bf_lentset_cntx_af_lentconfig(R+tmenubartfilemenuteditmenut rescntmenutcntxmenut cntxbfmenut cntxafmenu((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR&wsP   cKs|ii|i_dS(N(RWtgetR!t result_count(R+tkwargs((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR^scKs|ii|_dS(N(RYRjt _char_after(R+Rl((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRascKs|ii|_dS(N(RXRjt _char_before(R+Rl((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR`scCs t|d|i}t||_|ii|iit|dtddd|iddddd d i d d |ii i i |ii}t ||i|iid |i|ii}d |dR$Rv( R9R;tEntryt query_boxRBtButtontsearcht search_buttonR/tsearch_enter_keypress_handler(R+RCRtanother((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR>s%*% cGs|idS(N(R(R+tevent((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRscCst|}t|}t|}t|dd}t|dddd}t|dtidddd d d ddd |id |idddddddd |_|iidddddt|ii |i d|i |ii |i d|i |idddddd|id|ii|idddddtdd |id|iit|d!d"d#|iidddd|idd$dddtdd%|idd&dddd'|idd$dddtdS((NREitorientthoriztfonttfamilytcouriertsizet16tstatetdisabledtyscrollcommandtxscrollcommandtwraptnoneRt40theightRMtexportselectionRqRrR6R7R8t foregroundtyRuteRIRttwRps R2R$Rvtbottomts(R9t ScrollbartTextttkFonttFontRyt results_boxRBRt tag_configt_HIGHLIGHT_WORD_TAGt_HIGHLIGHT_WORD_COLOURt_HIGHLIGHT_LABEL_TAGt_HIGHLIGHT_LABEL_COLOURRbtyviewtxviewRzR;(R+RCRti1ti2t vscrollbart hscrollbar((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR?s*      '"+"cCst|d|i}t|ddd|iddddd dd d |_}|id d ddt|ddd|iddddd dd d |_}|id ddd|id dddd|_dS(NR2RptPreviousRIRt10REiRsRRRqRrRuRtNexttrightR$R6Ri(R9R;RtprevioustprevRBtnextt current_page(R+RCRRR((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR@s::cCs/|i|i|ii|iddS(Ni(tclear_results_boxtfreeze_editableR!RR(R+((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs  cCs/|i|i|ii|iddS(Ni(RRR!RR(R+((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs  c Gscd}d}y6ddkl}|d|d|d|iiWnt|i||nXdS(NsNLTK Concordance Search Demo s#About: NLTK Concordance Search Demoi(tMessagetmessageR.RC(t tkMessageBoxRR<tshowtShowTextR$(R+RtABOUTtTITLER((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pytabouts&cCs\|iit|i|iit|i|iit|i|iit|i dS(N( R$R/tCORPUS_LOADED_EVENTthandle_corpus_loadedtSEARCH_TERMINATED_EVENTthandle_search_terminatedtSEARCH_ERROR_EVENTthandle_search_errortERROR_LOADING_CORPUS_EVENTthandle_error_loading_corpus(R+((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR(scCs<d|ii|id<|i|i|idS(NsError in loading Rp(RxRjRtunfreeze_editablet clear_allR(R+R((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR s  cCs?|iid|id<|i|i|iidS(Ns is loadedRp(RxRjRRRRt focus_set(R+R((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs  cCs|ii}|i|d|idRR?R@RRRR(RRRRRR)RRRRR0RRRRRR R (((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRPsR  1                       R cBseZdZdZdZdZdZdZdZdZ dZ d Z d Z d Z d Zd eifdYZdeifdYZRS(cCsNg|_t|_t|_d|_|i|id|_ d|_ dS(Ni( t listenerst_CORPORAR|t_DEFAULTR*RRRRRktlast_sent_searched(R+((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR,s       cCs:g}|i|ii|i|i|i|S(N(textendR|R}R~R*tsort(R+tcopy((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs  cCs2||_g|_|i||}|idS(N(RRt LoadCorpuststart(R+tnamet runner_thread((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR)s  cCs2||_||_|i|||iidS(N(RRt SearchCorpusRkR(R+Rtpage((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs  cCsG||_t|i|jo|i|i|n|itdS(N(RRRRRtnotify_listenersR(R+R((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs cCs||_|itdS(N(RRR(R+R((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs cCs|ii|dS(N(RR(R+tlistener((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR"scCs%x|iD]}|i|q WdS(N(RR (R+RR((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs cCsd|_g|_d|_dS(Ni(RRRt last_page(R+((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs  cCs d|_dS(N(RR(R+((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRscCs|ii|d|dS(Ni(RR(R+Rt resultset((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyt set_resultsscCs|i|idS(Ni(RR(R+((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRscCsQ|igjp|idgjotSn|idjotSn||ijS(Ni(RtFalseRRR(R+R((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs $RcBseZdZdZRS(cCs'|||_|_tii|dS(N(R!Rt threadingtThreadR,(R+RR!((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR,sc Csya|ii|i}g}|D]}|td|Dq$~|i_|iitWn*tj o}|GH|iitnXdS(Ncss'x |]\}}|d|VqWdS(RN((t.0Rtt((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pys s( R!R|RRRRRt ExceptionR(R+ttst_[1]RR((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pytruns7(RRR,R*(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs RcBs#eZdZdZdZRS(cCs1||||_|_|_tii|dS(N(R!tcountRR#R$R,(R+R!RR+((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR,sc Cs|i}gdd}}}x|ii|iiD]}yti||}Wn7tij o(|ii|iit dSnX|o[|i ||i |i f|d7}||i jo|ii|d7_Pqn|d7}q7W|i t|jo?|ii|d7_|i|i_|ii|i|n|ii|i|d |iitdS(Niii(tprocessed_queryR!RRtreRterrorRRRRRtendR+RRRR!R(R+tqtsent_postit sent_countRtm((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR*s.   "  cCsg}x|iiiD]}tidd|}tid|o!|ittd|tqd|jo|it|tq|it|dttqWdi |S(Ns\.s[^/ ]s[A-Z]+$RR( R!RRR-tsubtmatchRtBOUNDARYt WORD_OR_TAGR(R+tnewtterm((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR,s! !(RRR,R*R,(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyRs  (RRR,RR)RRRR"RRRR!RRR#R$RR(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pyR s            cCst}|idS(N(RR (R((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pytapps t__main__R;(R-tstringRR#tnltkRRtnltk.draw.utilR8R7RRRRRRtobjectRR R;Rt__all__(((s./p/zhu/06/nlp/nltk/nltk/app/concordance_app.pys sV                          C|