# Generated from estonian.sbl by Snowball 3.1.1 - https://snowballstem.org/ from .basestemmer import BaseStemmer from .among import Among class EstonianStemmer(BaseStemmer): ''' This class implements the stemming algorithm defined by a snowball script. Generated from estonian.sbl by Snowball 3.1.1 - https://snowballstem.org/ ''' g_V1 = {"a", "e", "i", "o", "u", "ä", "õ", "ö", "ü"} g_RV = {"'", "a", "e", "i", "o", "u"} g_KI = {"b", "d", "f", "g", "h", "k", "p", "s", "t", "z", "š", "ž"} g_GI = {"a", "c", "e", "i", "j", "l", "m", "n", "o", "q", "r", "u", "v", "w", "x", "ä", "õ", "ö", "ü"} I_p1 = 0 def __r_mark_regions(self): self.I_p1 = self.limit while True: v_1 = self.cursor try: if self.cursor + 2 > self.limit: raise lab0() self.cursor += 2 while True: try: if self.cursor == self.limit or self.current[self.cursor] != "'": raise lab1() self.cursor += 1 break except lab1: pass if self.cursor >= self.limit: raise lab0() self.cursor += 1 break except lab0: pass self.cursor = v_1 if not self.go_out_grouping(EstonianStemmer.g_V1): return False self.cursor += 1 if not self.go_in_grouping(EstonianStemmer.g_V1): return False self.cursor += 1 break self.I_p1 = self.cursor return True def __r_emphasis(self): if self.cursor < self.I_p1: return False v_2 = self.limit_backward self.limit_backward = self.I_p1 self.ket = self.cursor among_var = self.find_among_b(EstonianStemmer.a_0) if among_var == 0: self.limit_backward = v_2 return False self.bra = self.cursor self.limit_backward = v_2 v_3 = self.limit - self.cursor if self.cursor - 4 < self.limit_backward: return False self.cursor -= 4 self.cursor = self.limit - v_3 if among_var == 1: v_4 = self.limit - self.cursor if not self.in_grouping_b(EstonianStemmer.g_GI): return False self.cursor = self.limit - v_4 v_5 = self.limit - self.cursor try: if not self.__r_LONGV(): raise lab0() return False except lab0: pass self.cursor = self.limit - v_5 self.slice_del() else: if not self.in_grouping_b(EstonianStemmer.g_KI): return False self.slice_del() return True def __r_verb(self): if self.cursor < self.I_p1: return False v_2 = self.limit_backward self.limit_backward = self.I_p1 self.ket = self.cursor among_var = self.find_among_b(EstonianStemmer.a_1) if among_var == 0: self.limit_backward = v_2 return False self.bra = self.cursor self.limit_backward = v_2 if among_var == 1: self.slice_del() elif among_var == 2: self.slice_from("a") else: if not self.in_grouping_b(EstonianStemmer.g_V1): return False self.slice_del() return True def __r_LONGV(self): return self.find_among_b(EstonianStemmer.a_2) != 0 def __r_i_plural(self): if self.cursor < self.I_p1: return False v_2 = self.limit_backward self.limit_backward = self.I_p1 self.ket = self.cursor if self.cursor <= self.limit_backward or self.current[self.cursor - 1] != "i": self.limit_backward = v_2 return False self.cursor -= 1 self.bra = self.cursor self.limit_backward = v_2 if not self.in_grouping_b(EstonianStemmer.g_RV): return False self.slice_del() return True def __r_special_noun_endings(self): if self.cursor < self.I_p1: return False v_2 = self.limit_backward self.limit_backward = self.I_p1 self.ket = self.cursor among_var = self.find_among_b(EstonianStemmer.a_3) if among_var == 0: self.limit_backward = v_2 return False self.bra = self.cursor self.limit_backward = v_2 self.slice_from(EstonianStemmer.as_3[among_var - 1]) return True def __r_case_ending(self): if self.cursor < self.I_p1: return False v_2 = self.limit_backward self.limit_backward = self.I_p1 self.ket = self.cursor among_var = self.find_among_b(EstonianStemmer.a_4) if among_var == 0: self.limit_backward = v_2 return False self.bra = self.cursor self.limit_backward = v_2 if among_var == 1: while True: try: if not self.in_grouping_b(EstonianStemmer.g_RV): raise lab0() break except lab0: pass if not self.__r_LONGV(): return False break else: v_3 = self.limit - self.cursor if self.cursor - 4 < self.limit_backward: return False self.cursor -= 4 self.cursor = self.limit - v_3 self.slice_del() return True def __r_plural_three_first_cases(self): if self.cursor < self.I_p1: return False v_2 = self.limit_backward self.limit_backward = self.I_p1 self.ket = self.cursor among_var = self.find_among_b(EstonianStemmer.a_6) if among_var == 0: self.limit_backward = v_2 return False self.bra = self.cursor self.limit_backward = v_2 if among_var == 1: self.slice_from("iku") elif among_var == 2: v_3 = self.limit - self.cursor try: if not self.__r_LONGV(): raise lab0() return False except lab0: pass self.cursor = self.limit - v_3 self.slice_del() elif among_var == 3: while True: v_4 = self.limit - self.cursor try: v_5 = self.limit - self.cursor if self.cursor - 4 < self.limit_backward: raise lab0() self.cursor -= 4 self.cursor = self.limit - v_5 among_var = self.find_among_b(EstonianStemmer.a_5) if among_var > 0: self.slice_from(EstonianStemmer.as_5[among_var - 1]) break except lab0: pass self.cursor = self.limit - v_4 self.slice_from("t") break else: while True: try: if not self.in_grouping_b(EstonianStemmer.g_RV): raise lab0() break except lab0: pass if not self.__r_LONGV(): return False break self.slice_del() return True def __r_nu(self): if self.cursor < self.I_p1: return False v_2 = self.limit_backward self.limit_backward = self.I_p1 self.ket = self.cursor if self.find_among_b(EstonianStemmer.a_7) == 0: self.limit_backward = v_2 return False self.bra = self.cursor self.limit_backward = v_2 self.slice_del() return True def __r_undouble_kpt(self): if not self.in_grouping_b(EstonianStemmer.g_V1): return False if self.I_p1 > self.cursor: return False self.ket = self.cursor among_var = self.find_among_b(EstonianStemmer.a_8) if among_var == 0: return False self.bra = self.cursor self.slice_from(EstonianStemmer.as_8[among_var - 1]) return True def __r_degrees(self): if self.cursor < self.I_p1: return False v_2 = self.limit_backward self.limit_backward = self.I_p1 self.ket = self.cursor among_var = self.find_among_b(EstonianStemmer.a_9) if among_var == 0: self.limit_backward = v_2 return False self.bra = self.cursor self.limit_backward = v_2 if among_var == 1: if not self.in_grouping_b(EstonianStemmer.g_RV): return False self.slice_del() else: self.slice_del() return True def __r_substantive(self): v_1 = self.limit - self.cursor self.__r_special_noun_endings() self.cursor = self.limit - v_1 v_2 = self.limit - self.cursor self.__r_case_ending() self.cursor = self.limit - v_2 v_3 = self.limit - self.cursor self.__r_plural_three_first_cases() self.cursor = self.limit - v_3 v_4 = self.limit - self.cursor self.__r_degrees() self.cursor = self.limit - v_4 v_5 = self.limit - self.cursor self.__r_i_plural() self.cursor = self.limit - v_5 v_6 = self.limit - self.cursor self.__r_nu() self.cursor = self.limit - v_6 return True def __r_verb_exceptions(self): self.bra = self.cursor among_var = self.find_among(EstonianStemmer.a_10) if among_var == 0: return False self.ket = self.cursor if self.cursor < self.limit: return False self.slice_from(EstonianStemmer.as_10[among_var - 1]) return True def _stem(self): v_1 = self.cursor try: if not self.__r_verb_exceptions(): raise lab0() return False except lab0: pass self.cursor = v_1 v_2 = self.cursor self.__r_mark_regions() self.cursor = v_2 self.limit_backward = self.cursor self.cursor = self.limit v_3 = self.limit - self.cursor self.__r_emphasis() self.cursor = self.limit - v_3 v_4 = self.limit - self.cursor try: while True: v_5 = self.limit - self.cursor try: if not self.__r_verb(): raise lab1() break except lab1: pass self.cursor = self.limit - v_5 self.__r_substantive() break except lab0: pass self.cursor = self.limit - v_4 v_6 = self.limit - self.cursor self.__r_undouble_kpt() self.cursor = self.limit - v_6 self.ket = self.cursor if self.cursor <= self.limit_backward or self.current[self.cursor - 1] != "'": return False self.cursor -= 1 self.bra = self.cursor self.slice_del() self.cursor = self.limit_backward return True a_0 = [ Among("gi", -1, 1), Among("ki", -1, 2) ] a_1 = [ Among("da", -1, 3), Among("mata", -1, 1), Among("b", -1, 3), Among("ksid", -1, 1), Among("nuksid", 3, 1), Among("me", -1, 3), Among("sime", 5, 1), Among("ksime", 6, 1), Among("nuksime", 7, 1), Among("akse", -1, 2), Among("dakse", 9, 1), Among("takse", 9, 1), Among("site", -1, 1), Among("ksite", 12, 1), Among("nuksite", 13, 1), Among("n", -1, 3), Among("sin", 15, 1), Among("ksin", 16, 1), Among("nuksin", 17, 1), Among("daks", -1, 1), Among("taks", -1, 1) ] a_2 = [ Among("aa", -1, -1), Among("ee", -1, -1), Among("ii", -1, -1), Among("oo", -1, -1), Among("uu", -1, -1), Among("ää", -1, -1), Among("õõ", -1, -1), Among("öö", -1, -1), Among("üü", -1, -1) ] a_3 = [ Among("lane", -1, 1), Among("line", -1, 3), Among("mine", -1, 2), Among("lasse", -1, 1), Among("lisse", -1, 3), Among("misse", -1, 2), Among("lasi", -1, 1), Among("lisi", -1, 3), Among("misi", -1, 2), Among("last", -1, 1), Among("list", -1, 3), Among("mist", -1, 2) ] as_3 = ("lase", "mise", "lise") a_4 = [ Among("ga", -1, 1), Among("ta", -1, 1), Among("le", -1, 1), Among("sse", -1, 1), Among("l", -1, 1), Among("s", -1, 1), Among("ks", 5, 1), Among("t", -1, 2), Among("lt", 7, 1), Among("st", 7, 1) ] a_5 = [ Among("", -1, 2), Among("las", 0, 1), Among("lis", 0, 1), Among("mis", 0, 1), Among("t", 0, -1) ] as_5 = ("e", "") a_6 = [ Among("d", -1, 4), Among("sid", 0, 2), Among("de", -1, 4), Among("ikkude", 2, 1), Among("ike", -1, 1), Among("ikke", -1, 1), Among("te", -1, 3) ] a_7 = [ Among("va", -1, -1), Among("du", -1, -1), Among("nu", -1, -1), Among("tu", -1, -1) ] a_8 = [ Among("kk", -1, 1), Among("pp", -1, 2), Among("tt", -1, 3) ] as_8 = ("k", "p", "t") a_9 = [ Among("ma", -1, 2), Among("mai", -1, 1), Among("m", -1, 1) ] a_10 = [ Among("joob", -1, 1), Among("jood", -1, 1), Among("joodakse", 1, 1), Among("jooma", -1, 1), Among("joomata", 3, 1), Among("joome", -1, 1), Among("joon", -1, 1), Among("joote", -1, 1), Among("joovad", -1, 1), Among("juua", -1, 1), Among("juuakse", 9, 1), Among("jäi", -1, 12), Among("jäid", 11, 12), Among("jäime", 11, 12), Among("jäin", 11, 12), Among("jäite", 11, 12), Among("jääb", -1, 12), Among("jääd", -1, 12), Among("jääda", 17, 12), Among("jäädakse", 18, 12), Among("jäädi", 17, 12), Among("jääks", -1, 12), Among("jääksid", 21, 12), Among("jääksime", 21, 12), Among("jääksin", 21, 12), Among("jääksite", 21, 12), Among("jääma", -1, 12), Among("jäämata", 26, 12), Among("jääme", -1, 12), Among("jään", -1, 12), Among("jääte", -1, 12), Among("jäävad", -1, 12), Among("jõi", -1, 1), Among("jõid", 32, 1), Among("jõime", 32, 1), Among("jõin", 32, 1), Among("jõite", 32, 1), Among("keeb", -1, 4), Among("keed", -1, 4), Among("keedakse", 38, 4), Among("keeks", -1, 4), Among("keeksid", 40, 4), Among("keeksime", 40, 4), Among("keeksin", 40, 4), Among("keeksite", 40, 4), Among("keema", -1, 4), Among("keemata", 45, 4), Among("keeme", -1, 4), Among("keen", -1, 4), Among("kees", -1, 4), Among("keeta", -1, 4), Among("keete", -1, 4), Among("keevad", -1, 4), Among("käia", -1, 8), Among("käiakse", 53, 8), Among("käib", -1, 8), Among("käid", -1, 8), Among("käidi", 56, 8), Among("käiks", -1, 8), Among("käiksid", 58, 8), Among("käiksime", 58, 8), Among("käiksin", 58, 8), Among("käiksite", 58, 8), Among("käima", -1, 8), Among("käimata", 63, 8), Among("käime", -1, 8), Among("käin", -1, 8), Among("käis", -1, 8), Among("käite", -1, 8), Among("käivad", -1, 8), Among("laob", -1, 16), Among("laod", -1, 16), Among("laoks", -1, 16), Among("laoksid", 72, 16), Among("laoksime", 72, 16), Among("laoksin", 72, 16), Among("laoksite", 72, 16), Among("laome", -1, 16), Among("laon", -1, 16), Among("laote", -1, 16), Among("laovad", -1, 16), Among("loeb", -1, 14), Among("loed", -1, 14), Among("loeks", -1, 14), Among("loeksid", 83, 14), Among("loeksime", 83, 14), Among("loeksin", 83, 14), Among("loeksite", 83, 14), Among("loeme", -1, 14), Among("loen", -1, 14), Among("loete", -1, 14), Among("loevad", -1, 14), Among("loob", -1, 7), Among("lood", -1, 7), Among("loodi", 93, 7), Among("looks", -1, 7), Among("looksid", 95, 7), Among("looksime", 95, 7), Among("looksin", 95, 7), Among("looksite", 95, 7), Among("looma", -1, 7), Among("loomata", 100, 7), Among("loome", -1, 7), Among("loon", -1, 7), Among("loote", -1, 7), Among("loovad", -1, 7), Among("luua", -1, 7), Among("luuakse", 106, 7), Among("lõi", -1, 6), Among("lõid", 108, 6), Among("lõime", 108, 6), Among("lõin", 108, 6), Among("lõite", 108, 6), Among("lööb", -1, 5), Among("lööd", -1, 5), Among("löödakse", 114, 5), Among("löödi", 114, 5), Among("lööks", -1, 5), Among("lööksid", 117, 5), Among("lööksime", 117, 5), Among("lööksin", 117, 5), Among("lööksite", 117, 5), Among("lööma", -1, 5), Among("löömata", 122, 5), Among("lööme", -1, 5), Among("löön", -1, 5), Among("lööte", -1, 5), Among("löövad", -1, 5), Among("lüüa", -1, 5), Among("lüüakse", 128, 5), Among("müüa", -1, 13), Among("müüakse", 130, 13), Among("müüb", -1, 13), Among("müüd", -1, 13), Among("müüdi", 133, 13), Among("müüks", -1, 13), Among("müüksid", 135, 13), Among("müüksime", 135, 13), Among("müüksin", 135, 13), Among("müüksite", 135, 13), Among("müüma", -1, 13), Among("müümata", 140, 13), Among("müüme", -1, 13), Among("müün", -1, 13), Among("müüs", -1, 13), Among("müüte", -1, 13), Among("müüvad", -1, 13), Among("näeb", -1, 18), Among("näed", -1, 18), Among("näeks", -1, 18), Among("näeksid", 149, 18), Among("näeksime", 149, 18), Among("näeksin", 149, 18), Among("näeksite", 149, 18), Among("näeme", -1, 18), Among("näen", -1, 18), Among("näete", -1, 18), Among("näevad", -1, 18), Among("nägema", -1, 18), Among("nägemata", 158, 18), Among("näha", -1, 18), Among("nähakse", 160, 18), Among("nähti", -1, 18), Among("põeb", -1, 15), Among("põed", -1, 15), Among("põeks", -1, 15), Among("põeksid", 165, 15), Among("põeksime", 165, 15), Among("põeksin", 165, 15), Among("põeksite", 165, 15), Among("põeme", -1, 15), Among("põen", -1, 15), Among("põete", -1, 15), Among("põevad", -1, 15), Among("saab", -1, 2), Among("saad", -1, 2), Among("saada", 175, 2), Among("saadakse", 176, 2), Among("saadi", 175, 2), Among("saaks", -1, 2), Among("saaksid", 179, 2), Among("saaksime", 179, 2), Among("saaksin", 179, 2), Among("saaksite", 179, 2), Among("saama", -1, 2), Among("saamata", 184, 2), Among("saame", -1, 2), Among("saan", -1, 2), Among("saate", -1, 2), Among("saavad", -1, 2), Among("sai", -1, 2), Among("said", 190, 2), Among("saime", 190, 2), Among("sain", 190, 2), Among("saite", 190, 2), Among("sõi", -1, 9), Among("sõid", 195, 9), Among("sõime", 195, 9), Among("sõin", 195, 9), Among("sõite", 195, 9), Among("sööb", -1, 9), Among("sööd", -1, 9), Among("söödakse", 201, 9), Among("söödi", 201, 9), Among("sööks", -1, 9), Among("sööksid", 204, 9), Among("sööksime", 204, 9), Among("sööksin", 204, 9), Among("sööksite", 204, 9), Among("sööma", -1, 9), Among("söömata", 209, 9), Among("sööme", -1, 9), Among("söön", -1, 9), Among("sööte", -1, 9), Among("söövad", -1, 9), Among("süüa", -1, 9), Among("süüakse", 215, 9), Among("teeb", -1, 17), Among("teed", -1, 17), Among("teeks", -1, 17), Among("teeksid", 219, 17), Among("teeksime", 219, 17), Among("teeksin", 219, 17), Among("teeksite", 219, 17), Among("teeme", -1, 17), Among("teen", -1, 17), Among("teete", -1, 17), Among("teevad", -1, 17), Among("tegema", -1, 17), Among("tegemata", 228, 17), Among("teha", -1, 17), Among("tehakse", 230, 17), Among("tehti", -1, 17), Among("toob", -1, 10), Among("tood", -1, 10), Among("toodi", 234, 10), Among("tooks", -1, 10), Among("tooksid", 236, 10), Among("tooksime", 236, 10), Among("tooksin", 236, 10), Among("tooksite", 236, 10), Among("tooma", -1, 10), Among("toomata", 241, 10), Among("toome", -1, 10), Among("toon", -1, 10), Among("toote", -1, 10), Among("toovad", -1, 10), Among("tuua", -1, 10), Among("tuuakse", 247, 10), Among("tõi", -1, 10), Among("tõid", 249, 10), Among("tõime", 249, 10), Among("tõin", 249, 10), Among("tõite", 249, 10), Among("viia", -1, 3), Among("viiakse", 254, 3), Among("viib", -1, 3), Among("viid", -1, 3), Among("viidi", 257, 3), Among("viiks", -1, 3), Among("viiksid", 259, 3), Among("viiksime", 259, 3), Among("viiksin", 259, 3), Among("viiksite", 259, 3), Among("viima", -1, 3), Among("viimata", 264, 3), Among("viime", -1, 3), Among("viin", -1, 3), Among("viisime", -1, 3), Among("viisin", -1, 3), Among("viisite", -1, 3), Among("viite", -1, 3), Among("viivad", -1, 3), Among("võib", -1, 11), Among("võid", -1, 11), Among("võida", 274, 11), Among("võidakse", 275, 11), Among("võidi", 274, 11), Among("võiks", -1, 11), Among("võiksid", 278, 11), Among("võiksime", 278, 11), Among("võiksin", 278, 11), Among("võiksite", 278, 11), Among("võima", -1, 11), Among("võimata", 283, 11), Among("võime", -1, 11), Among("võin", -1, 11), Among("võis", -1, 11), Among("võite", -1, 11), Among("võivad", -1, 11) ] as_10 = ("joo", "saa", "viima", "keesi", "löö", "lõi", "loo", "käisi", "söö", "too", "võisi", "jääma", "müüsi", "luge", "põde", "ladu", "tegi", "nägi") class lab0(BaseException): pass class lab1(BaseException): pass