From: Disassembler Date: Wed, 5 Feb 2020 11:07:27 +0000 (+0100) Subject: Add Czech stop words X-Git-Tag: 2.4~95^2 X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=refs%2Fpull%2F3253%2Fhead;p=rspamd.git Add Czech stop words --- diff --git a/contrib/languages-data/stop_words b/contrib/languages-data/stop_words index 3fe6d9338..b6566f122 100644 --- a/contrib/languages-data/stop_words +++ b/contrib/languages-data/stop_words @@ -19,4 +19,5 @@ "ne": ["\u0906\u092b\u0942","\u092f\u0940","\u0930\u0939\u0947\u0915\u093e","\u091a\u093e\u0939\u093f\u090f","\u0924\u093f\u092e\u0940","\u0905\u0928\u094d\u092f\u0924\u094d\u0930","\u0924\u0941\u0930\u0941\u0928\u094d\u0924\u0948","\u091b","\u0925\u093f\u092f\u094b","\u0924\u094d\u0938\u092a\u091b\u093f","\u0939\u0941\u0928\u094d","\u092d\u0928\u094d\u0928\u0947","\u0906\u0924\u094d\u092e","\u0905\u0930\u0941\u0932\u093e\u0908","\u0928\u094c","\u0925\u093f\u090f","\u092c\u0930\u0941","\u091c\u0939\u093e\u0901","\u092c\u093f\u0936\u0947\u0937","\u0932\u0917\u092d\u0917","\u0930","\u0930\u093e\u0916\u094d\u091b","\u0915\u093f","\u092f\u0938\u092c\u093e\u0939\u0947\u0915","\u092a\u0928\u093f","\u092a\u0939\u093f\u0932\u094d\u092f\u0948","\u0915\u0938\u0930\u0940","\u0924\u093f\u0930","\u0928\u091c\u093f\u0915\u0948","\u0928","\u092a\u091f\u0915","\u092c\u093f\u0930\u0941\u0926\u094d\u0927","\u0926\u0947\u0916\u093f\u0928\u094d\u091b","\u0924\u093e\u092a\u0928\u0940","\u0917\u0930\u094d\u0928\u0941\u092a\u0930\u094d\u091b","\u091c\u093e\u0928","\u0924\u094d\u092f\u094b","\u092a\u091b\u093f","\u0928\u093f\u092e\u094d\u0928","\u092e","\u0938\u093e\u0901\u091a\u094d\u091a\u0948","\u0938\u092c\u0948","\u092d\u0928\u094d\u091b\u0941","\u091c\u092c","\u0909\u0926\u093e\u0939\u0930\u0923","\u092a\u093e\u0901\u091a\u094c\u0902","\u0938\u093e\u0924","\u0938\u0902\u0917\u0948","\u092a\u0915\u094d\u0915\u093e","\u092c\u093e\u0939\u0947\u0915","\u0914\u0902","\u090f\u0915\u0926\u092e","\u0930\u093e\u092e\u094d\u0930\u094b","\u0915\u0938\u0948\u0932\u0947","\u090f\u0915","\u092a\u094d\u0930\u0924\u0947\u0915","\u091c\u0938\u0932\u093e\u0908","\u0917\u0930\u094d\u0926\u0948","\u092a\u0942\u0930\u094d\u0935","\u092c\u093e\u0930\u0947","\u0939\u0941\u0928\u094d\u091b","\u092f\u0938","\u091b\u0941","\u0938\u0927\u0948","\u0924\u093f\u0928\u093f\u0939\u0930\u0941\u0932\u093e\u0908","\u0926\u093f\u0928\u0941\u092d\u090f\u0915\u094b","\u090f\u0909\u091f\u0948","\u092d\u090f","\u092a\u0939\u093f\u0932\u0947","\u0909\u0939\u093e\u0932\u093e\u0908","\u0938\u0915\u094d\u091b","\u0925\u093f\u090f\u0928","\u0924\u0947\u0938\u094d\u0930\u094b","\u0915\u094b","\u092d\u093f\u0924\u094d\u0930","\u0930\u0939\u0947\u0915\u094b","\u092f\u0926\u093f","\u0938\u093e\u0925","\u0932\u0947","\u092e\u093e\u0925\u093f","\u091b\u0942","\u0939\u0941\u0928\u0947","\u0935\u093e\u0938\u094d\u0924\u0935\u092e\u093e","\u092d\u0928\u094d","\u0917\u0930\u0940","\u0909\u092a","\u0917\u0930\u094d\u091b","\u0924\u093f\u0928\u0940","\u0905\u0930\u094d\u0925\u093e\u0924\u094d","\u0924\u092a\u093e\u0908\u0915\u094b","\u0905\u091d\u0948","\u0928\u0948","\u0913\u0920","\u092d\u0928\u094d\u0928\u0941\u092d\u092f\u094b","\u0915\u093f\u0928\u092d\u0928\u0947","\u0906\u092f\u094b","\u0939\u0930\u0947\u0915","\u0906\u091c\u0915\u094b","\u091a\u093e\u0939\u0928\u094d\u091b\u0941","\u0938\u092e\u092f","\u0906\u0926\u093f","\u0905\u0930\u094d\u0925\u093e\u0924","\u0915\u092e \u0938\u0947 \u0915\u092e","\u0915\u094b\u0939\u0940","\u0928\u092f\u093e\u0901","\u0930\u093e\u0916\u0947","\u092c\u0940\u091a","\u0924\u0940","\u0926\u0947\u0916\u093f","\u0924\u0930","\u0905\u092c","\u092c\u0928\u0947","\u092d\u0928\u0947","\u092c\u093e\u0939\u093f\u0930","\u0924\u0924\u094d\u0915\u093e\u0932","\u0905\u0915\u094d\u0938\u0930","\u0935\u0930\u0940\u092a\u0930\u0940","\u091c\u0947","\u092a\u0915\u094d\u0915\u0948","\u092d\u093f\u0924\u094d\u0930\u0940","\u0926\u093f\u090f","\u0926\u0941\u0908","\u0924","\u092c\u0940\u091a\u092e\u093e","\u091c\u0938\u092e\u093e","\u0938\u093e\u092f\u0926","\u0938\u094d\u092a\u0937\u094d\u091f","\u0917\u0930\u094d\u0928\u0941","\u0926\u093f\u0928\u0941\u0939\u0941\u0928\u094d\u091b","\u0905\u0928\u094d\u0924\u0930\u094d\u0917\u0924","\u0924\u092a\u093e\u0908","\u092f\u0939\u093e\u0901","\u092a\u0939\u093f\u0932\u094b","\u0915\u0947","\u0905\u0930\u094d\u0915\u094b","\u092f\u0938\u0930\u0940","\u0917\u0930\u094d\u0928\u0947","\u0915\u0924\u0948","\u0917\u0930\u094c\u0902","\u0928\u093f","\u092e\u093e","\u092f\u0926\u094d\u092f\u092a\u093f","\u091c\u092c\u0915\u093f","\u0938\u0939\u0940","\u092a\u094d\u0932\u0938","\u0924\u094d\u0938\u0948\u0932\u0947","\u0917\u0930\u094d\u0928","\u0909\u0928\u0932\u0947","\u0917\u0948\u0930","\u092f\u0939\u093e\u0901\u0938\u092e\u094d\u092e","\u0928\u093f\u092e\u094d\u0924\u093f","\u0917\u090f","\u0932\u093e\u0908","\u091c\u0924\u093e\u0924\u0924\u0948","\u092f\u0938\u0915\u094b","\u091a\u093e\u0932\u0947","\u0906\u090f","\u092f\u0938\u094b","\u0909\u0928\u0915\u094b","\u091a\u093e\u0939\u0928\u0941\u0939\u0941\u0928\u094d\u091b","\u092a\u0930\u094d\u0925\u094d\u092f\u094b","\u091c\u0938\u094d\u0924\u094b\u0938\u0941\u0915\u0948","\u0924\u0940\u0928","\u092e\u0932\u093e\u0908","\u0924\u093f\u0928\u0940\u0939\u0930\u0941\u0915\u094b","\u092f\u0925\u094b\u091a\u093f\u0924","\u0930\u0942\u092a","\u0920\u0940\u0915","\u092b\u0947\u0930\u0940","\u0915\u0943\u092a\u092f\u093e","\u0917\u0930\u094d\u091b\u0941","\u0915\u0947\u0939\u0940","\u092d\u0928\u094d\u091b\u0928\u094d","\u0917\u0930\u093f","\u0938\u0902\u0917","\u092d\u0928","\u0917\u0930\u0947\u0930","\u0917\u092f\u094c","\u092f\u0938\u094d\u0924\u094b","\u0917\u0930\u0947\u0915\u094b","\u0906\u092b\u094d\u0928\u094b","\u091c\u0938\u092c\u093e\u091f","\u0938\u092e\u094d\u092d\u0935","\u092a\u094d\u0930\u0924\u093f","\u091c\u094b","\u0915\u094d\u0930\u092e\u0936\u0903","\u092e\u0941\u0916\u094d\u092f","\u092a\u0930\u094d\u091b","\u0932\u093e\u0917\u093f","\u091b\u0948\u0928","\u0906\u092b\u094d\u0928\u0948","\u092f\u093e","\u091c\u0941\u0928","\u0928\u093f\u092e\u094d\u0928\u093e\u0928\u0941\u0938\u093e\u0930","\u0930\u0939\u0940","\u092e\u093e\u0924\u094d\u0930","\u0905\u0928\u0941\u0938\u093e\u0930","\u091b\u0928\u094d","\u0926\u0947\u0916\u0947\u0930","\u091c\u0938\u094d\u0924\u094b","\u0915\u0941\u0930\u093e","\u0938\u093e\u0925\u0948","\u0926\u0947\u0916\u0947\u0915\u094b","\u0939\u0930\u0947","\u0915\u0941\u0928\u0948","\u0939\u094b","\u0906\u092b\u0942\u0932\u093e\u0908","\u092f\u0924\u093f","\u0905\u0930\u0941","\u0924\u0925\u093e","\u0936\u093e\u092f\u0926","\u0938\u092e\u094d\u092e","\u0939\u0941\u0928","\u0924\u094d\u092f\u0939\u093e\u0901","\u0938\u094b\u0939\u0940","\u0926\u094b\u0938\u094d\u0930\u094b","\u092f\u0938\u092a\u091b\u093f","\u0915\u0939\u093e\u0901\u092c\u093e\u091f","\u091c\u0938\u094d\u0924\u0948","\u091a\u093e\u0930","\u092e\u0947\u0930\u094b","\u092a\u093e\u0901\u091a","\u092f\u094b","\u092a\u0930\u094d\u092f\u093e\u092a\u094d\u0924","\u0924\u0926\u0928\u0941\u0938\u093e\u0930","\u091c\u0938\u0932\u0947","\u0926\u0947\u0916\u093f\u092f\u094b","\u0905\u0928\u094d\u092f","\u092d\u0928\u094d\u0926\u093e","\u0924\u0932","\u092d\u090f\u0915\u094b","\u0926\u0947\u0916\u0947","\u0927\u0947\u0930\u0948","\u0917\u0930\u0947\u0915\u093e","\u0905\u0917\u093e\u0921\u0940","\u091b\u094c\u0902","\u091c\u093e\u0939\u093f\u0930","\u0928\u0924\u094d\u0930","\u0928\u093f\u0930\u094d\u0926\u093f\u0937\u094d\u091f","\u0915\u093f\u0928","\u0938\u092c\u0948\u0932\u093e\u0908","\u0938\u094b","\u092d\u0930","\u0938\u093e\u0930\u093e","\u0915\u0939\u093f\u0932\u0947\u0915\u093e\u0939\u0940\u0902","\u0924\u0947\u0938\u094d\u0915\u093e\u0930\u0923","\u091c\u0938\u0915\u094b","\u092a\u091b\u093f\u0932\u094d\u0932\u094b","\u0905\u0932\u0917","\u0915\u0938\u0948","\u0924\u093f\u0928\u0940\u0939\u0930\u0942","\u0905\u0928\u094d\u092f\u0925\u093e","\u0938\u091f\u094d\u091f\u093e"], "nl": ["zich","zou","hebben","reeds","ge","veel","worden","daar","aan","want","zonder","het","omdat","niet","voor","iets","zijn","zij","wordt","nog","dit","ik","tegen","heb","ben","ze","ons","uw","maar","hij","iemand","naar","geweest","doen","mijn","een","werd","zo","mij","haar","door","meer","dus","niets","kon","zal","toen","uit","ook","wezen","wil","met","geen","bij","moet","onder","zelf","eens","toch","kunnen","altijd","deze","wat","heeft"], "ro": ["fii","aceea","t\u0103i","acestia","altfel","citi","atita","numai","cîtva","toata","pîn\u0103","c\u0103tre","acelea","atunci","acest","îmi","imi","dup\u0103","putini","parca","aia","unele","atit","decit","fi","altceva","ala","dou\u0103","tuturor","cei","s\u0103","prin","citeva","c\u0103ci","unuia","doilea","mîine","deja","inca","mele","unul","ea","pân\u0103","uneori","\u0103stea","alte","tale","unui","ceilalti","care","deci","sai","dat\u0103","prima","vostru","cind","abia","ace\u015fti","iar","oriunde","nostru","pai","fi\u0163i","ati","chiar","atata","dac\u0103","tocmai","despre","cat","ca","nici","nimeni","to\u0163i","pina","v\u0103","de\u015fi","drept","doi","niste","f\u0103r\u0103","no\u015ftri","nimic","acea","s\u0103u","atat","cînd","sint","cî\u0163i","isi","\u015fi","a\u015f","celor","mâine","nou","patru","cea","lor","vo\u015ftri","pe","foarte","anume","cine","dintr-","mereu","în","inapoi","undeva","toate","aceeasi","adica","sunt","carora","noua","uneia","deasupra","dintr","alta","câ\u0163i","cum","toat\u0103","c\u0103","î\u0163i","c\u0103rui","asupra","câtva","unei","va","dau","fata","acestei","zice","zi","nou\u0103","câte","acela","fara","vou\u0103","toti","printr-","totul","atitea","atitia","desi","cît","acei","fiu","ii","îi","orice","c\u0103ror","caruia","oricând","acel","ceva","cand","pot","patra","cite","mea","mai","insa","aceia","catre","ale","cel","poate","îl","pu\u0163in","asta","mult","\u0103sta","avem","vreo","doar","t\u0103u","unora","atare","pentru","daca","alea","a\u0163i","treilea","careia","avea","alti","\u0103lea","altul","spre","apoi","ace\u015ftia","multa","\u0163i","noastr\u0103","m\u0103","unor","oricât","ul","inainte","face","fim","intre","oricine","sau","citiva","aceste","voastr\u0103","unii","noastre","acestui","mult\u0103","ave\u0163i","dintre","fie","altcineva","sale","spate","ar","e\u015fti","sintem","\u0163ie","ori","ului","sunte\u0163i","oricare","sub","intr","pu\u0163in\u0103","asa","totusi","acestea","atatia","aceasta","voastre","atatea","fiecare","trei","mul\u0163i","lâng\u0103","incit","oricum","multi","multe","ni\u015fte","c\u0103rei","astfel","aici","cita","oricînd","ba","unu","cineva","d\u0103","treia","altii","unde","mei","\u0103\u015ftia","cam","acesta","iti","aib\u0103","primul","dupa","acele","aceast\u0103","sa-mi","fost","cele","ceea","prea","totu\u015fi","cit","dar","acelasi","pu\u0163ina","peste","cât","astea","lîng\u0103","s\u0103i","oricît","suntem","sa-ti","vreun","cumva","cîte","\u0103la","pic","tine","avut"], +"cs": ["ačkoli","ahoj","alespoň","anebo","aspoň","během","blízko","bohužel","brzo","brzy","bude","budeme","budeš","budete","budou","budu","byl","byla","byli","bylo","byly","bys","být","celá","celé","celý","či","čísel","čísla","číslo","dál","dále","další","dát","děkujeme","děkuji","dělá","dělat","dík","díky","dnes","dobrý","docela","dokud","druhá","druhé","druhý","druzí","hned","hodně","chce","chceme","chceš","chcete","chci","chtějí","chtít","já","jak","jaká","jaké","jaký","jde","jeho","její","jejich","jemu","jen","jenom","jenž","jestli","jestliže","ještě","jí","jich","jím","jiná","jinak","jiné","jiný","jít","již","jsem","jsi","jsme","jsou","jste","každá","každé","každý","kde","kdo","kdy","když","kolik","kromě","která","které","který","kteří","kvůli","má","mají","málo","mám","máme","máš","máte","mé","mě","měl","mém","mezi","mí","místo","mít","mně","mnou","moc","mohl","mohou","mohu","možná","možné","možný","můj","musí","může","můžeme","můžeš","můžete","nad","nade","nám","námi","naproti","nás","náš","naše","naši","našich","našim","našimi","nebo","nebyl","nebyla","nebyli","nebyly","něco","nějak","nějaká","nějaké","nějakou","nějaký","někam","někde","někdo","nemají","nemáme","nemáš","nemáte","neměl","němu","není","než","nich","nikam","nikde","nikdo","ním","nimi","nová","nové","nový","pak","paní","pod","podle","pokud","pořád","potom","pozdravem","proč","prosím","prostě","proti","protože","prvé","první","před","přejeme","přeji","přes","přese","při","příloha","přílohou","přílohy","příloze","přiložená","přiložené","přiloženém","přiložený","přiložených","přímo","sám","skoro","smí","smíme","smíš","smíte","snad","spolu","svá","svého","svému","svoje","svoji","svou","svůj","tady","také","takhle","taková","takové","takový","takto","taky","takže","tamhle","tamhleto","tamto","tě","tebou","teď","tedy","tento","této","tobě","tohle","trošku","třeba","tvá","tvé","tvoje","tvůj","určitě","už","vám","vámi","vás","váš","vaše","vaši","vaší","vážená","vážené","vážení","vážený","včera","vědět","však","vše","všeho","všechen","všechna","všechno","všechny","všichni","vždy","zač","zatím","zatímco","zda","zítra","žádám","žádná","žádné","žádný","že"], }