U \Shy@s dZddlmZddlZddlZddlZddlmZmZm Z zddl Z dZ Wne k rddZ YnXzddl ZdZWne k rdZYnXddlmZmZddlmZdd lmZGd d d ZGd d d eZGdddeZGdddeZGdddeZdS)z:Japanese search language: includes routine to split words.) annotationsN)AnyDictListTF)ExtensionError SphinxError)SearchLanguage) import_objectc@s,eZdZdddddZdddd d Zd S) BaseSplitterdictNoneoptionsreturncCs ||_dSN)rselfrrC/root/rtd-docs/venv/lib/python3.8/site-packages/sphinx/search/ja.py__init__$szBaseSplitter.__init__str list[str]inputrcCstdS)zN :param str input: :return: :rtype: list[str] N)NotImplementedErrorrrrrrsplit'szBaseSplitter.splitN)__name__ __module__ __qualname__rrrrrrr #sr csbeZdZdddfdd Zdddd d Zdddd d Zdddd dZddddZZS) MecabSplitterr r r csDt|d|_d|_ts(||n |||dd|_dS)NZdic_enczutf-8) superrctypes_libmecab ctypes_mecab native_module init_ctypes init_nativeget dict_encoder __class__rrr1s   zMecabSplitter.__init__rrrcCs4tr|j|}n|j|j||j}|dS)N ) r$nativeparser"mecab_sparse_tostrr#encoder(rrrresultrrrr;s zMecabSplitter.splitcCs.d}|d}|r|d|7}t||_dS)Nz-Owakatir  -d %s)r'MeCabZTaggerr,)rrparamr rrrr&Cs   zMecabSplitter.init_nativec Csddl}|d}|dkr>tjdr,d}nd}|j|}n2tj ||kr\|j|}nd}tj |rp|}|dkrt dd}|d}|r|d |7}t pt }|||_|jf|jj_|j|jj_|j|jf|jj_|j|jj_|j|||_|jdkrtd dS) Nrlibwinz libmecab.dllZmecabz&MeCab dynamic library is not availablezmecab -Owakatir r2zmecab initialization failed) ctypes.utilr'sysplatform startswithutil find_libraryospathbasenameexists RuntimeErrorgetfilesystemencodinggetdefaultencodingCDLLr"c_char_pZ mecab_new2ZargtypesZc_void_prestyper.r/r#r) rrctypesr5libnameZlibpathr4r Zfs_encrrrr%Js6         zMecabSplitter.init_ctypesrcCs|jr|j|jdSr)r"Z mecab_destroyr#rrrr__del__nszMecabSplitter.__del__) rrrrrr&r%rK __classcell__rrr)rr 0s  $r csBeZdZdddfdd ZddddZd d d d d ZZS)JanomeSplitterr r r cs2t||d|_|dd|_|dS)NZuser_dicZ user_dic_encutf8)r!rr' user_dict user_dict_encinit_tokenizerrr)rrrts  zJanomeSplitter.__init__rIcCs&ts tdtjj|j|jd|_dS)NzJanome is not available)ZudicZudic_enc) janome_modulerAjanome tokenizer TokenizerrOrPrJrrrrQzszJanomeSplitter.init_tokenizerrrrcCs&ddd|j|D}|dS)Nr+css|] }|jVqdSr)Zsurface).0tokenrrr sz'JanomeSplitter.split..)joinrTtokenizerr0rrrrszJanomeSplitter.split)rrrrrQrrLrrr)rrMssrMc@sveZdZddddddddd DZd Zd d d ddZddddddddddddddddd d!Zd"d#d$d%d&d'd(d)d*d+d,d- Zd.d/d0d1d2Z d3d4d5Z d6d7d8d9d:d;dd?d@dAdBdCdDZ dEdFdGdHdIdJdKdLdMdNdO Z dPdQdRdSdTdUdVdWdXdYdZd[d\d]d^d_d`Z daddbdcdddedfdgd,dhdidj Zdkdldmdndkdldodpdqdrdsdtdudvdwdxdydzd{d|d}d~dddddddddddddddddddddddddddddddddddddddddddddddmdndodGZdddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd d d d d dd ddddddddddddddddddd d!d"dd#xZd$d%d&d'd'd(d)d*d+d,d-d.d/d0d1d2d2d3d4dd5d6d7d7d8d9dd:d;d<d=d>d?d@dAdBdCdCdDdEdFdGdHdIdIdJdKdKdLdMdNdNdOdPdQdRdSdSdTdUdVdWdXdYdZd[d\d]d^d_d`dadbdcdddededfdgdhdddidjdjdkdldmdndod\dpdqdrdsdsdtdudvdwdxdydzd{d|d}d~ddddddd{drZddddVddddddddd ZdddddddZddddddddddddddddddddddddZddddddddddddddddddddddÐdLdĜZdŐdƐdd3dȐdɐdʐdːd̐d͐dΐdϐdАdќ ZdddӐdԐd՜Zd֐dאdؐdِdڐdېdܐdݐdސddߐddddddddddZd$ddddddddddddd#dddZdddZdddddddddddPddddddZd d d d d ddddddd Zddddddddddddd Zdd d!d"d#Zd$d%d&d'd(d)d*Zd+d,d-Z d.d/d0d1d2d3d4d5Z!d6d7d8d9d:d;Z"d<d=d>d?d@d;Z#dAdBiZ$dCdDdEZ%dFdGiZ&dHdIdJdKdLdMdNdOdPdQ Z'dRdSdTdUZ(dVdWdXdYdZd[d\d]d^d_ Z)d`d`dadbdcdddedfdgdhdidjdkdldmdnddodpdqdrdsdtdudvddwdxdydzd{d|d}d~dddadtd&Z*dddddddfddd1ddddddddVddddddvdddddddddwddudddddddddddddddddiddddddddddddddddddddddddÐdĐdŐdƐdǐdȐdɐdʐdːd̐d͐dΐdϐdАdѐdҐdӐdԐdՐdɐd֐dאdؐdِdڐdېdddddddܜkZ+dݐdސdߐdݐddddddddddddTddddddddddddddddddddddddddddddd dld d d d dddddddddddddddddddddd d!d"d#d$d%d&d'd(d)d*d+d,d-d.d/d0d1d2d3ddd4dd5d6d7d8d9d:d;d<dd=d>d?d@dAdBddCdDdEdFdGdHdIdJdKdLdMdNdOdPdQdRdSdTdUdVdWdXdYdZd[d\d]d^d_d`dadbdcdddedfdgdhdiddjdkdאdldmdndodpdqdrdsdtdud dvd֐dwddxdydzd{d|d}d~dddddddddddduddddddHd dddddddddސddddddddddddZ,dddddddddddddddddddddddddddddddddddddddddddddddÐdĐdŐdɐdƐdǐdȐdɐdʐdːd̐d͐dΐdϐdАdѐdҐdӐdԐdՐd֐dQdאdؐdddڐdېdܐdݐddސdߐdddddddddddddddddddddddddddddddddddddϐddddddddddddd d d d d ddddddddd[ddddddddddd̐dӐddsdddd d!d"d#d$d%d&d'dd(d)d*d+d,d-d.d/dאd0d1dPd dd2did3d4d5d6d7d8dTd9d:d;d<d=d>d?d@ddddPdAdBdCddddddאdؐdِdڐdېdܐdݐddސdDZ-dEdRdFdGdHdEdRdIdJdKdLddMdNdOdPdQdRdSdTdUd1dVddWdXdYdZdcdod[d\d]d^d_d`dadbdsdcdddedSdfdgdhd>diddjdkdldmdndodpdqdrdsdtdudXdvdwdxdyddzd{d|d}d~ddddddddddddddאdBddd)ddddFdGdIdhd>didcZ.ddddddddGdhddddddddddېd=ddddddtddddddސddYdddzdxdddddddddddddddd5Z/dddddZ0ddddddZ1dddddZ2dS(DefaultSplittercCsi|]\}}t||qSr)recompile)rVpatternvaluerrr szDefaultSplitter.MHIKAN)u/[一二三四五六七八九十百千万億兆]u[一-龠々〆ヵヶ]u [ぁ-ん]u[ァ-ヴーｱ-ﾝﾞｰ]u[a-zA-Zａ-ｚＡ-Ｚ]u [0-9０-９]ii ii)HHIIZKHOHi=i iiiQiii/i`iiitiiii i)ZAAZAIANrhZHMHNHOZIAIHriZIKIOZKIKKZMHMKOOiiri/iiii i7iii ) rhZHIZHKrlrmrnrprqZMMZOArji'i0ii`)BBZOBrrZUB<i)BOrri~iiziviiiiiii;iii )BHHBHMBIIZBIMBMHZBNHBOHBOOZOHIOIHZOKAOKHZOKKOOOviyiiiiHi`iiKi) rvBHIrwBIHBKKZBKOOHHOHMr|ZUHIiih iiii iii~iiiii>ii-)rvrrxZBKIryZBMMZBNNrzrrOIIr}ZOKIZOKOZOMHr~iiiDiiiii i3) rvrrxZBIKrZBMIr{ZOAHrZOHKZONNiii|iiiiiii/ii iXii#iiiiQ iPiiiibiYi ii%iigiu iLiii/iQi iii iiipi ii ii( iieiiA ioiDiii i iiiii=iii,)Gu,とu,同uB1あuB1同u、とu、同u」とuあっいういっいるuうしuうんからがらuこうことuこんuさらさんしたしてuすでuそこuそれたちuためったuついていてきできですではといとみuどこないuなっなどなんにしにはuにもに対uの一uの中uませまでuままuまるuやむuよっれたuれでuをしuを見u亡くu京都u取りu大き大阪u平方u引き日本u本当u毎日u目指u Ｂ１あu Ｂ１同u｣とiiciiii iiiiiUiiii8iiip3iwiiiiii`iijiiiiUiioiii8i iiij ii igiibi.iii0iiHiii6 i ii i iiiii-imigii1iiiii#i!iii'iPiiiiTiiQiiiiiiiZiiiiiisiiYiii#iiiBii.idi, iiiwiiii"iiT)xz..Z11u――u−−ruうかuかしuかもruかれuがいrきたuくなrこのuさせuされrしいuしかrrしなuそのたいuたたuただruたとuたはrってuっとrruてくuてはてもuでいruでしruでもruとこuとととのrともruながruなのruにおrになuによruに従uに関uのかuのでuのにuののuはいuはがuはずuばれましrまれuめてuもいuものuらからしuらにuりしuりまrれてればuろうわれuを通んだuんなu一人u一方u一日u一部u上がu会社u出てu分のu同党u同日ru委員u少なu年度u年間u府県u手権新聞u日新ru日米u曜日u朝鮮u本人u東京u然と社会u立てu第にu米国u１１iniii_iii ii[iii]iii) iiixiii iiiimi^iii;i6iii~i ii0 iiiyib iIii"ii,iaiiiiti,i`iWi.ii6iiiii)i8iiyiWiiiriT ii9iiYiii iiiiVi:i_iRiCii^iii\iiiiiAii)ruあたuありuあるuい.uい。uいいuいえuいくuいたrruいわuうちuうとuえとuか.uか。uかけuかっuかにruかりuが,uが、uがきuがけuがっruがりruけどrruころuさいuさをuし,uし、rrrruしまuす.uす。uするuず,uず、uずにuそうuた.uた。ruたのuたりuたるuだ.uだ。uだっrrruておrruでにruとうuとしrruどうruなくruに,uに、rrruの,uの、uの子uは,uは、uべきruますuまっrrruられuるるuれ,uれ、rrruれるrruんでuカ月u会議u入りu大会u始め市ru日,u日、ruｶ月iEiiDiziiiri ii) AAAHHHHHMHIIHOHZHOMIHIZIOHIOIZIOMMMHZOOIi(ii~iSii)HHOrZHMMrKKHriiZiii2iiqiiii@ii?iizi7{i)igii)rrHHIrZHIKrIHHrIHOIIHZIIMrZKHHKKArZKOKMHHZMHMZMHOrZNNHZNNOZOHOi5i@imiii$iiiiiAi6ii: i ikii)rrZHHKrZHHNrZHIHrrrrrZIIIZIIOZIOOZKAKrZKKKrZMHIrZMMMZMOMii<i|i;iitiii) BHHHZBHHIBHIHBIHHBIIIZBNHHZBOHHBOOOOAKKOHHHOHIHOIHIOIIHiii))rrZBKAKriiiiiii0iliiiioi iKniS)rBHHMrBHIIBIIHrZBMHIZBMHMZBOMHrOHHIZOHIIZOHMHOIHHrOKAKZOKHHZOKKAZOOHHZOOIIii:ii{i5i iiiiiYi$)rrrrrZOAAArrrZOHHOrrrrZOIIIrii) についu 東京都iii~i!iii)ii[iZii_ii_ip)u ある程u いったu ころがu しょうu その後u だってu ていた としてu ともにu ものでu 一気にu 初めてu 同時にu 大きなu 対してu 社会党i:i"i&iiii)iiXi) u いただ していrru にとっu に当たuので,u ので、u のものu れからu 十二月i!ii iiiEiiIi^i) uいう.u いう。u からなruたが,u たが、u ているu というu ましたu ませんu ようとu よるとi]ii)rerdraOi3i#iiii)rerbrcrarfrii )rerciiiii i$i)rerbrcrdrarfri9i*iii)rbrcrdrariiWi}ri*Ei)Brriiiiiii) BHBIBKBNrurjOIOKrrqi)rrri!*iyiiX ii9iit) ZBArrrZBMrrurONi1iSiiyii7iiiiGii.i*iiBiwiyiniigi4i1iieisiiihi~i2)&,、「あうがきこでとどにのはもやよらりれをん・主京区午国大委r日理生県都｢･ii|i{iI ii iidiuiniii]i i-iKii+ii'iigiiiigiiJi;i,i/ii2i8i?i iii8iiiiji$i/iiiZiiiiiii'iii+iiiniiJiiii]igiiLi4iiiiisii*iiiFiiii ii)krr〇r」rいrおかrくrさuざしすせそただつてrrrなrr r ひべまr r r rるrrrアカuキッu三u不世中r事人会保u入初副北rru天u太子学実小r年u強u手u揺政文新ru明最u朝u本東u果u次民気ru発目相r立第米自行u西見調議込開間r ｣ｯｱｶuｷiiiEiiin i ixii& iiiui iii iiiuiHieiiJii\iii ii}ii iiiiiitiiii ii6iii iIiii'i'iji iUi7iii.ii!iiiiWi iiaii+iiiyii i&i*iiii ii iiiiUiii&iiiiiiiBiiiimiiiiYii6iiii} iip iiEiiiiqiiii{i#i9i i/ii3iiki iIiWiihi ilii$i i5 ibiLiiiiiiiiiiii@iiWi iOihi{iii9iiiiii#ii+i iiWii4iihiiiii)r1u−ru々r%r&〓rr'rえr(r)rr*けげrごr+r,r-r.r/r0ちっr2r3rrrr4rr r r5ふへほr7みめr r r rrr8rわrrr9uグuスr;トuムルuロンr一u下r<u両r=r予u二r?u今u他以u低u何作rAu元党u全u公u六共円u再u分rBu別前rC力動務化rDru千r協u口u右u各合u同和員ru型場u外u妻rFu安rGu家rHu少u州ru平rI広u府u度u建u当u得思性u戸u指rJ教u数rKrLru旧u昨時u曜rM月u期u李u村rNu核u森業氏rOu決u法u海u無物u特ru用町的u直省ru知u私u税rSrTrU系約u級統総線者能rVu英rXrYu財u費車軍r[u通選郎u郡u部u金長r\r]u関u雨電u非u駅１r^r!r_r`uｸﾞuｽﾄuﾑﾙuﾛﾝiZi iiigiiiiii.ie iivi~ii4iiiii0 ii6i%iiii iiiiiii9ici=iii!iiiiAi iiiiiJiIii ii iiiii&i i^3iaiiimiiiiiiiiiiBii i iiii*iDiiCiWi'iieiiiiiiriiiiPibi~i{i>iiiivii$iii ivi4ii)iiXiiii i%iMiiiFiii idi!i iii"i iici iiiy ii{izioiiii&ii ii7iigiii$ iiii[ieiii@iiH)r.u―r。r%rr&rcrr'rrdr(r)rrぎr*rerfrrgr+r,じr-uずr.r/r0r1rhrir2r3rrr4ruぬuねr r uばr5uびrjrkr6rlr7rmuむrnr ゃr ょr rrr8ruろrorrr:uコuセr;rpuメuラuリrqrrruーrsr=rtr>u井r?rur@u体rvu側u先rwrxu内ryrBrzrCr{r|r}r~rru参rrru器u回ru園u地ru塁u士u多rrErF定u寺rHu屋u山u島u川rrIu庁後ru感所u支u改rJrrKu方rrrMru木u来u校rrrOrPu沢u済ru率u球ru産田rrrQrr空rSrTrUrru経rrru署rru般rWu規u警rZu谷u賞rru輪r[u近u道rrru野u銀rr]u院u際ru領題館u首u高r r^r!r_uｰrauｺuｾruﾒuﾗuﾘrriiii6ikiwiKi iii[iXii8i)iii^4iiiii8iiiiiai i0iiiiGiiiiri i{iPi8iiiiiQiihiiRiiiLiiiiii]iiiiGiiiii@ iO)crrrbE2]rrrrr'rrdr(r)rrrr*rfr+r,r-r1rhrir2r3rrrr4rr r r6rmrnrrrr8rrorruイrqrrr=rr@rwr}ru告rrrFru嵐ru席rIrru挙rLrru査u格u機rrrrrRrru研社rrru表u語rZrrrr]rrruＥ２r uｲrri(iiiii8iiiTi=i eiki_ii2i_iiipiiwiIiiGiiiiii:)5rrrbE1rrrrr)rr*rrr-r0rir3rrr4rr r r rr8rrqrrrsr=u件r@rzrrrrFrrrrrRru福rru連rruＥ１rrr)charrcCs*|jD]\}}||r |Sq dS)Nr) patterns_itemsmatch)rrr^r_rrrctype_s  zDefaultSplitter.ctype_zdict[str, int]int)r keyrcCs||kr||SdS)Nrr)rr rrrrts_szDefaultSplitter.ts_rrcCs:|sgSg}dddg}dddg}|D]}|||||q$|d|d|d|d|d|d|d}d }d }d } td t|dD]~} |j} || d} || d } || d }|| }|| d }|| d }|| d}|| d }|| d }|| }|| d }|| d }| ||j|7} | ||j|7} | ||j| 7} | ||j ||7} | ||j || 7} | ||j | 7} | ||j | 7} | ||j |7} | ||j|7} | ||j|7} | ||j|7} | ||j| |7} | ||j||7} | ||j||7} | ||j| | |7} | ||j| ||7} | ||j|||7} | ||j|||7} | ||j|7} | ||j|7} | ||j|7} | ||j|7} | ||j|7} | ||j|7} | ||j||7} | ||j||7} | ||j ||7} | ||j!|||7} | ||j"|||7} | ||j#|||7} | ||j$|||7} | ||j%||7} | ||j&||7} | ||j%| |7} | ||j'|||7} | ||j(|||7} | ||j)| ||7} | ||j*| ||7} | ||j+||||7} | ||j,||||7} | ||j-| |||7} | ||j.| |||7} d}| d kr||/d}d}|}| }|} ||| 7}q||/|S)NZB3ZB2ZB1rrrZE3Urr)0appendrrangelenBIAS__rUP1__UP2__UP3__BP1__BP2__UW1__UW2__UW3__UW4__UW5__UW6__BW1__BW2__BW3__TW1__TW2__TW3__TW4__UC1__UC2__UC3__UC4__UC5__UC6__BC1__BC2__BC3__TC1__TC2__TC3__TC4__UQ1__UQ2__BQ1__BQ2__BQ3__BQ4__TQ1__TQ2__TQ3__TQ4__strip)rrr1segctypetwordp1Zp2Zp3iZscoreZw1Zw2Zw3Zw4Zw5Zw6c1c2c3Zc4Zc5Zc6prrrrs                    zDefaultSplitter.splitN)3rrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrZUQ3__rrrrrrrrrrrrrr[s             $% r[c@sXeZdZdZdZdZdddddZd d d d d Zd ddddZd d dddZ dS)SearchJapanesezg Japanese search implementation: uses no stemmer, but word splitting is quite complicated. jaZJapaneser r r c CsT|dd}zt|||_Wn0tk rN}ztd||W5d}~XYnXdS)Ntypez sphinx.search.ja.DefaultSplitterz$Splitter module %r can't be imported)r'r splitterr)rr dotted_pathexcrrrinits zSearchJapanese.initrrrcCs |j|Sr)rrrrrrr szSearchJapanese.splitbool) stemmed_wordrcCs t|dkS)Nr)r)rrrrr word_filterszSearchJapanese.word_filter)rrcCs|Srr)rrrrrstemszSearchJapanese.stemN) rrr__doc__langZ language_namerrr r rrrrrsr)r  __future__rr=r\r8typingrrrr3r$ ImportErrorZjanome.tokenizerrSrRZ sphinx.errorsrrZ sphinx.searchrZ sphinx.utilr r r rMr[rrrrrs2      Cz