# Generated from catalan.sbl by Snowball 3.1.1 - https://snowballstem.org/ from .basestemmer import BaseStemmer from .among import Among class CatalanStemmer(BaseStemmer): ''' This class implements the stemming algorithm defined by a snowball script. Generated from catalan.sbl by Snowball 3.1.1 - https://snowballstem.org/ ''' g_v = {"a", "e", "i", "o", "u", "à", "á", "è", "é", "í", "ï", "ò", "ó", "ú", "ü"} I_p2 = 0 I_p1 = 0 def __r_mark_regions(self): self.I_p1 = self.limit self.I_p2 = self.limit v_1 = self.cursor try: if not self.go_out_grouping(CatalanStemmer.g_v): raise lab0() self.cursor += 1 if not self.go_in_grouping(CatalanStemmer.g_v): raise lab0() self.cursor += 1 self.I_p1 = self.cursor if not self.go_out_grouping(CatalanStemmer.g_v): raise lab0() self.cursor += 1 if not self.go_in_grouping(CatalanStemmer.g_v): raise lab0() self.cursor += 1 self.I_p2 = self.cursor except lab0: pass self.cursor = v_1 return True def __r_cleaning(self): while True: v_1 = self.cursor try: self.bra = self.cursor among_var = self.find_among(CatalanStemmer.a_0) self.ket = self.cursor if among_var == 1: self.slice_from("a") elif among_var == 2: self.slice_from("e") elif among_var == 3: self.slice_from("i") elif among_var == 4: self.slice_from("o") elif among_var == 5: self.slice_from("u") elif among_var == 6: self.slice_from(".") else: if self.cursor >= self.limit: raise lab0() self.cursor += 1 continue except lab0: pass self.cursor = v_1 break return True def __r_R1(self): return self.I_p1 <= self.cursor def __r_R2(self): return self.I_p2 <= self.cursor def __r_attached_pronoun(self): self.ket = self.cursor if self.find_among_b(CatalanStemmer.a_1) == 0: return False self.bra = self.cursor if not self.__r_R1(): return False self.slice_del() return True def __r_standard_suffix(self): self.ket = self.cursor among_var = self.find_among_b(CatalanStemmer.a_2) if among_var == 0: return False self.bra = self.cursor if among_var == 1: if not self.__r_R1(): return False self.slice_del() elif among_var == 2: if not self.__r_R2(): return False self.slice_del() elif among_var == 3: if not self.__r_R2(): return False self.slice_from("log") elif among_var == 4: if not self.__r_R2(): return False self.slice_from("ic") else: if not self.__r_R1(): return False self.slice_from("c") return True def __r_verb_suffix(self): self.ket = self.cursor among_var = self.find_among_b(CatalanStemmer.a_3) if among_var == 0: return False self.bra = self.cursor if among_var == 1: if not self.__r_R1(): return False self.slice_del() else: if not self.__r_R2(): return False self.slice_del() return True def __r_residual_suffix(self): self.ket = self.cursor among_var = self.find_among_b(CatalanStemmer.a_4) if among_var == 0: return False self.bra = self.cursor if among_var == 1: if not self.__r_R1(): return False self.slice_del() else: if not self.__r_R1(): return False self.slice_from("ic") return True def _stem(self): self.__r_mark_regions() self.limit_backward = self.cursor self.cursor = self.limit v_1 = self.limit - self.cursor self.__r_attached_pronoun() self.cursor = self.limit - v_1 v_2 = self.limit - self.cursor try: while True: v_3 = self.limit - self.cursor try: if not self.__r_standard_suffix(): raise lab1() break except lab1: pass self.cursor = self.limit - v_3 if not self.__r_verb_suffix(): raise lab0() break except lab0: pass self.cursor = self.limit - v_2 v_4 = self.limit - self.cursor self.__r_residual_suffix() self.cursor = self.limit - v_4 self.cursor = self.limit_backward v_5 = self.cursor self.__r_cleaning() self.cursor = v_5 return True a_0 = [ Among("", -1, 7), Among("·", 0, 6), Among("à", 0, 1), Among("á", 0, 1), Among("è", 0, 2), Among("é", 0, 2), Among("ì", 0, 3), Among("í", 0, 3), Among("ï", 0, 3), Among("ò", 0, 4), Among("ó", 0, 4), Among("ú", 0, 5), Among("ü", 0, 5) ] a_1 = [ Among("la", -1, 1), Among("-la", 0, 1), Among("sela", 0, 1), Among("le", -1, 1), Among("me", -1, 1), Among("-me", 4, 1), Among("se", -1, 1), Among("-te", -1, 1), Among("hi", -1, 1), Among("'hi", 8, 1), Among("li", -1, 1), Among("-li", 10, 1), Among("'l", -1, 1), Among("'m", -1, 1), Among("-m", -1, 1), Among("'n", -1, 1), Among("-n", -1, 1), Among("ho", -1, 1), Among("'ho", 17, 1), Among("lo", -1, 1), Among("selo", 19, 1), Among("'s", -1, 1), Among("las", -1, 1), Among("selas", 22, 1), Among("les", -1, 1), Among("-les", 24, 1), Among("'ls", -1, 1), Among("-ls", -1, 1), Among("'ns", -1, 1), Among("-ns", -1, 1), Among("ens", -1, 1), Among("los", -1, 1), Among("selos", 31, 1), Among("nos", -1, 1), Among("-nos", 33, 1), Among("vos", -1, 1), Among("us", -1, 1), Among("-us", 36, 1), Among("'t", -1, 1) ] a_2 = [ Among("ica", -1, 4), Among("lógica", 0, 3), Among("enca", -1, 1), Among("ada", -1, 2), Among("ancia", -1, 1), Among("encia", -1, 1), Among("ència", -1, 1), Among("ícia", -1, 1), Among("logia", -1, 3), Among("inia", -1, 1), Among("íinia", 9, 1), Among("eria", -1, 1), Among("ària", -1, 1), Among("atòria", -1, 1), Among("alla", -1, 1), Among("ella", -1, 1), Among("ívola", -1, 1), Among("ima", -1, 1), Among("íssima", 17, 1), Among("quíssima", 18, 5), Among("ana", -1, 1), Among("ina", -1, 1), Among("era", -1, 1), Among("sfera", 22, 1), Among("ora", -1, 1), Among("dora", 24, 1), Among("adora", 25, 1), Among("adura", -1, 1), Among("esa", -1, 1), Among("osa", -1, 1), Among("assa", -1, 1), Among("essa", -1, 1), Among("issa", -1, 1), Among("eta", -1, 1), Among("ita", -1, 1), Among("ota", -1, 1), Among("ista", -1, 1), Among("ialista", 36, 1), Among("ionista", 36, 1), Among("iva", -1, 1), Among("ativa", 39, 1), Among("nça", -1, 1), Among("logía", -1, 3), Among("ic", -1, 4), Among("ístic", 43, 1), Among("enc", -1, 1), Among("esc", -1, 1), Among("ud", -1, 1), Among("atge", -1, 1), Among("ble", -1, 1), Among("able", 49, 1), Among("ible", 49, 1), Among("isme", -1, 1), Among("ialisme", 52, 1), Among("ionisme", 52, 1), Among("ivisme", 52, 1), Among("aire", -1, 1), Among("icte", -1, 1), Among("iste", -1, 1), Among("ici", -1, 1), Among("íci", -1, 1), Among("logi", -1, 3), Among("ari", -1, 1), Among("tori", -1, 1), Among("al", -1, 1), Among("il", -1, 1), Among("all", -1, 1), Among("ell", -1, 1), Among("ívol", -1, 1), Among("isam", -1, 1), Among("issem", -1, 1), Among("ìssem", -1, 1), Among("íssem", -1, 1), Among("íssim", -1, 1), Among("quíssim", 73, 5), Among("amen", -1, 1), Among("ìssin", -1, 1), Among("ar", -1, 1), Among("ificar", 77, 1), Among("egar", 77, 1), Among("ejar", 77, 1), Among("itar", 77, 1), Among("itzar", 77, 1), Among("fer", -1, 1), Among("or", -1, 1), Among("dor", 84, 1), Among("dur", -1, 1), Among("doras", -1, 1), Among("ics", -1, 4), Among("lógics", 88, 3), Among("uds", -1, 1), Among("nces", -1, 1), Among("ades", -1, 2), Among("ancies", -1, 1), Among("encies", -1, 1), Among("ències", -1, 1), Among("ícies", -1, 1), Among("logies", -1, 3), Among("inies", -1, 1), Among("ínies", -1, 1), Among("eries", -1, 1), Among("àries", -1, 1), Among("atòries", -1, 1), Among("bles", -1, 1), Among("ables", 103, 1), Among("ibles", 103, 1), Among("imes", -1, 1), Among("íssimes", 106, 1), Among("quíssimes", 107, 5), Among("formes", -1, 1), Among("ismes", -1, 1), Among("ialismes", 110, 1), Among("ines", -1, 1), Among("eres", -1, 1), Among("ores", -1, 1), Among("dores", 114, 1), Among("idores", 115, 1), Among("dures", -1, 1), Among("eses", -1, 1), Among("oses", -1, 1), Among("asses", -1, 1), Among("ictes", -1, 1), Among("ites", -1, 1), Among("otes", -1, 1), Among("istes", -1, 1), Among("ialistes", 124, 1), Among("ionistes", 124, 1), Among("iques", -1, 4), Among("lógiques", 127, 3), Among("ives", -1, 1), Among("atives", 129, 1), Among("logíes", -1, 3), Among("allengües", -1, 1), Among("icis", -1, 1), Among("ícis", -1, 1), Among("logis", -1, 3), Among("aris", -1, 1), Among("toris", -1, 1), Among("ls", -1, 1), Among("als", 138, 1), Among("ells", 138, 1), Among("ims", -1, 1), Among("íssims", 141, 1), Among("quíssims", 142, 5), Among("ions", -1, 1), Among("cions", 144, 1), Among("acions", 145, 2), Among("esos", -1, 1), Among("osos", -1, 1), Among("assos", -1, 1), Among("issos", -1, 1), Among("ers", -1, 1), Among("ors", -1, 1), Among("dors", 152, 1), Among("adors", 153, 1), Among("idors", 153, 1), Among("ats", -1, 1), Among("itats", 156, 1), Among("bilitats", 157, 1), Among("ivitats", 157, 1), Among("ativitats", 159, 1), Among("ïtats", 156, 1), Among("ets", -1, 1), Among("ants", -1, 1), Among("ents", -1, 1), Among("ments", 164, 1), Among("aments", 165, 1), Among("ots", -1, 1), Among("uts", -1, 1), Among("ius", -1, 1), Among("trius", 169, 1), Among("atius", 169, 1), Among("ès", -1, 1), Among("és", -1, 1), Among("ís", -1, 1), Among("dís", 174, 1), Among("ós", -1, 1), Among("itat", -1, 1), Among("bilitat", 177, 1), Among("ivitat", 177, 1), Among("ativitat", 179, 1), Among("ïtat", -1, 1), Among("et", -1, 1), Among("ant", -1, 1), Among("ent", -1, 1), Among("ient", 184, 1), Among("ment", 184, 1), Among("ament", 186, 1), Among("isament", 187, 1), Among("ot", -1, 1), Among("isseu", -1, 1), Among("ìsseu", -1, 1), Among("ísseu", -1, 1), Among("triu", -1, 1), Among("íssiu", -1, 1), Among("atiu", -1, 1), Among("ó", -1, 1), Among("ió", 196, 1), Among("ció", 197, 1), Among("ació", 198, 1) ] a_3 = [ Among("aba", -1, 1), Among("esca", -1, 1), Among("isca", -1, 1), Among("ïsca", -1, 1), Among("ada", -1, 1), Among("ida", -1, 1), Among("uda", -1, 1), Among("ïda", -1, 1), Among("ia", -1, 1), Among("aria", 8, 1), Among("iria", 8, 1), Among("ara", -1, 1), Among("iera", -1, 1), Among("ira", -1, 1), Among("adora", -1, 1), Among("ïra", -1, 1), Among("ava", -1, 1), Among("ixa", -1, 1), Among("itza", -1, 1), Among("ía", -1, 1), Among("aría", 19, 1), Among("ería", 19, 1), Among("iría", 19, 1), Among("ïa", -1, 1), Among("isc", -1, 1), Among("ïsc", -1, 1), Among("ad", -1, 1), Among("ed", -1, 1), Among("id", -1, 1), Among("ie", -1, 1), Among("re", -1, 1), Among("dre", 30, 1), Among("ase", -1, 1), Among("iese", -1, 1), Among("aste", -1, 1), Among("iste", -1, 1), Among("ii", -1, 1), Among("ini", -1, 1), Among("esqui", -1, 1), Among("eixi", -1, 1), Among("itzi", -1, 1), Among("am", -1, 1), Among("em", -1, 1), Among("arem", 42, 1), Among("irem", 42, 1), Among("àrem", 42, 1), Among("írem", 42, 1), Among("àssem", 42, 1), Among("éssem", 42, 1), Among("iguem", 42, 1), Among("ïguem", 42, 1), Among("avem", 42, 1), Among("àvem", 42, 1), Among("ávem", 42, 1), Among("irìem", 42, 1), Among("íem", 42, 1), Among("aríem", 55, 1), Among("iríem", 55, 1), Among("assim", -1, 1), Among("essim", -1, 1), Among("issim", -1, 1), Among("àssim", -1, 1), Among("èssim", -1, 1), Among("éssim", -1, 1), Among("íssim", -1, 1), Among("ïm", -1, 1), Among("an", -1, 1), Among("aban", 66, 1), Among("arian", 66, 1), Among("aran", 66, 1), Among("ieran", 66, 1), Among("iran", 66, 1), Among("ían", 66, 1), Among("arían", 72, 1), Among("erían", 72, 1), Among("irían", 72, 1), Among("en", -1, 1), Among("ien", 76, 1), Among("arien", 77, 1), Among("irien", 77, 1), Among("aren", 76, 1), Among("eren", 76, 1), Among("iren", 76, 1), Among("àren", 76, 1), Among("ïren", 76, 1), Among("asen", 76, 1), Among("iesen", 76, 1), Among("assen", 76, 1), Among("essen", 76, 1), Among("issen", 76, 1), Among("éssen", 76, 1), Among("ïssen", 76, 1), Among("esquen", 76, 1), Among("isquen", 76, 1), Among("ïsquen", 76, 1), Among("aven", 76, 1), Among("ixen", 76, 1), Among("eixen", 96, 1), Among("ïxen", 76, 1), Among("ïen", 76, 1), Among("in", -1, 1), Among("inin", 100, 1), Among("sin", 100, 1), Among("isin", 102, 1), Among("assin", 102, 1), Among("essin", 102, 1), Among("issin", 102, 1), Among("ïssin", 102, 1), Among("esquin", 100, 1), Among("eixin", 100, 1), Among("aron", -1, 1), Among("ieron", -1, 1), Among("arán", -1, 1), Among("erán", -1, 1), Among("irán", -1, 1), Among("iïn", -1, 1), Among("ado", -1, 1), Among("ido", -1, 1), Among("ando", -1, 2), Among("iendo", -1, 1), Among("io", -1, 1), Among("ixo", -1, 1), Among("eixo", 121, 1), Among("ïxo", -1, 1), Among("itzo", -1, 1), Among("ar", -1, 1), Among("tzar", 125, 1), Among("er", -1, 1), Among("eixer", 127, 1), Among("ir", -1, 1), Among("ador", -1, 1), Among("as", -1, 1), Among("abas", 131, 1), Among("adas", 131, 1), Among("idas", 131, 1), Among("aras", 131, 1), Among("ieras", 131, 1), Among("ías", 131, 1), Among("arías", 137, 1), Among("erías", 137, 1), Among("irías", 137, 1), Among("ids", -1, 1), Among("es", -1, 1), Among("ades", 142, 1), Among("ides", 142, 1), Among("udes", 142, 1), Among("ïdes", 142, 1), Among("atges", 142, 1), Among("ies", 142, 1), Among("aries", 148, 1), Among("iries", 148, 1), Among("ares", 142, 1), Among("ires", 142, 1), Among("adores", 142, 1), Among("ïres", 142, 1), Among("ases", 142, 1), Among("ieses", 142, 1), Among("asses", 142, 1), Among("esses", 142, 1), Among("isses", 142, 1), Among("ïsses", 142, 1), Among("ques", 142, 1), Among("esques", 161, 1), Among("ïsques", 161, 1), Among("aves", 142, 1), Among("ixes", 142, 1), Among("eixes", 165, 1), Among("ïxes", 142, 1), Among("ïes", 142, 1), Among("abais", -1, 1), Among("arais", -1, 1), Among("ierais", -1, 1), Among("íais", -1, 1), Among("aríais", 172, 1), Among("eríais", 172, 1), Among("iríais", 172, 1), Among("aseis", -1, 1), Among("ieseis", -1, 1), Among("asteis", -1, 1), Among("isteis", -1, 1), Among("inis", -1, 1), Among("sis", -1, 1), Among("isis", 181, 1), Among("assis", 181, 1), Among("essis", 181, 1), Among("issis", 181, 1), Among("ïssis", 181, 1), Among("esquis", -1, 1), Among("eixis", -1, 1), Among("itzis", -1, 1), Among("áis", -1, 1), Among("aréis", -1, 1), Among("eréis", -1, 1), Among("iréis", -1, 1), Among("ams", -1, 1), Among("ados", -1, 1), Among("idos", -1, 1), Among("amos", -1, 1), Among("ábamos", 197, 1), Among("áramos", 197, 1), Among("iéramos", 197, 1), Among("íamos", 197, 1), Among("aríamos", 201, 1), Among("eríamos", 201, 1), Among("iríamos", 201, 1), Among("aremos", -1, 1), Among("eremos", -1, 1), Among("iremos", -1, 1), Among("ásemos", -1, 1), Among("iésemos", -1, 1), Among("imos", -1, 1), Among("adors", -1, 1), Among("ass", -1, 1), Among("erass", 212, 1), Among("ess", -1, 1), Among("ats", -1, 1), Among("its", -1, 1), Among("ents", -1, 1), Among("às", -1, 1), Among("aràs", 218, 1), Among("iràs", 218, 1), Among("arás", -1, 1), Among("erás", -1, 1), Among("irás", -1, 1), Among("és", -1, 1), Among("arés", 224, 1), Among("ís", -1, 1), Among("iïs", -1, 1), Among("at", -1, 1), Among("it", -1, 1), Among("ant", -1, 1), Among("ent", -1, 1), Among("int", -1, 1), Among("ut", -1, 1), Among("ït", -1, 1), Among("au", -1, 1), Among("erau", 235, 1), Among("ieu", -1, 1), Among("ineu", -1, 1), Among("areu", -1, 1), Among("ireu", -1, 1), Among("àreu", -1, 1), Among("íreu", -1, 1), Among("asseu", -1, 1), Among("esseu", -1, 1), Among("eresseu", 244, 1), Among("àsseu", -1, 1), Among("ésseu", -1, 1), Among("igueu", -1, 1), Among("ïgueu", -1, 1), Among("àveu", -1, 1), Among("áveu", -1, 1), Among("itzeu", -1, 1), Among("ìeu", -1, 1), Among("irìeu", 253, 1), Among("íeu", -1, 1), Among("aríeu", 255, 1), Among("iríeu", 255, 1), Among("assiu", -1, 1), Among("issiu", -1, 1), Among("àssiu", -1, 1), Among("èssiu", -1, 1), Among("éssiu", -1, 1), Among("íssiu", -1, 1), Among("ïu", -1, 1), Among("ix", -1, 1), Among("eix", 265, 1), Among("ïx", -1, 1), Among("itz", -1, 1), Among("ià", -1, 1), Among("arà", -1, 1), Among("irà", -1, 1), Among("itzà", -1, 1), Among("ará", -1, 1), Among("erá", -1, 1), Among("irá", -1, 1), Among("irè", -1, 1), Among("aré", -1, 1), Among("eré", -1, 1), Among("iré", -1, 1), Among("í", -1, 1), Among("iï", -1, 1), Among("ió", -1, 1) ] a_4 = [ Among("a", -1, 1), Among("e", -1, 1), Among("i", -1, 1), Among("ïn", -1, 1), Among("o", -1, 1), Among("ir", -1, 1), Among("s", -1, 1), Among("is", 6, 1), Among("os", 6, 1), Among("ïs", 6, 1), Among("it", -1, 1), Among("eu", -1, 1), Among("iu", -1, 1), Among("iqu", -1, 2), Among("itz", -1, 1), Among("à", -1, 1), Among("á", -1, 1), Among("é", -1, 1), Among("ì", -1, 1), Among("í", -1, 1), Among("ï", -1, 1), Among("ó", -1, 1) ] class lab0(BaseException): pass class lab1(BaseException): pass