| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683 |
- # Generated by Snowball 2.2.0 - https://snowballstem.org/
- from .basestemmer import BaseStemmer
- from .among import Among
- class BasqueStemmer(BaseStemmer):
- '''
- This class implements the stemming algorithm defined by a snowball script.
- Generated by Snowball 2.2.0 - https://snowballstem.org/
- '''
- a_0 = [
- Among(u"idea", -1, 1),
- Among(u"bidea", 0, 1),
- Among(u"kidea", 0, 1),
- Among(u"pidea", 0, 1),
- Among(u"kundea", -1, 1),
- Among(u"galea", -1, 1),
- Among(u"tailea", -1, 1),
- Among(u"tzailea", -1, 1),
- Among(u"gunea", -1, 1),
- Among(u"kunea", -1, 1),
- Among(u"tzaga", -1, 1),
- Among(u"gaia", -1, 1),
- Among(u"aldia", -1, 1),
- Among(u"taldia", 12, 1),
- Among(u"karia", -1, 1),
- Among(u"garria", -1, 2),
- Among(u"karria", -1, 1),
- Among(u"ka", -1, 1),
- Among(u"tzaka", 17, 1),
- Among(u"la", -1, 1),
- Among(u"mena", -1, 1),
- Among(u"pena", -1, 1),
- Among(u"kina", -1, 1),
- Among(u"ezina", -1, 1),
- Among(u"tezina", 23, 1),
- Among(u"kuna", -1, 1),
- Among(u"tuna", -1, 1),
- Among(u"kizuna", -1, 1),
- Among(u"era", -1, 1),
- Among(u"bera", 28, 1),
- Among(u"arabera", 29, 4),
- Among(u"kera", 28, 1),
- Among(u"pera", 28, 1),
- Among(u"orra", -1, 1),
- Among(u"korra", 33, 1),
- Among(u"dura", -1, 1),
- Among(u"gura", -1, 1),
- Among(u"kura", -1, 1),
- Among(u"tura", -1, 1),
- Among(u"eta", -1, 1),
- Among(u"keta", 39, 1),
- Among(u"gailua", -1, 1),
- Among(u"eza", -1, 1),
- Among(u"erreza", 42, 1),
- Among(u"tza", -1, 2),
- Among(u"gaitza", 44, 1),
- Among(u"kaitza", 44, 1),
- Among(u"kuntza", 44, 1),
- Among(u"ide", -1, 1),
- Among(u"bide", 48, 1),
- Among(u"kide", 48, 1),
- Among(u"pide", 48, 1),
- Among(u"kunde", -1, 1),
- Among(u"tzake", -1, 1),
- Among(u"tzeke", -1, 1),
- Among(u"le", -1, 1),
- Among(u"gale", 55, 1),
- Among(u"taile", 55, 1),
- Among(u"tzaile", 55, 1),
- Among(u"gune", -1, 1),
- Among(u"kune", -1, 1),
- Among(u"tze", -1, 1),
- Among(u"atze", 61, 1),
- Among(u"gai", -1, 1),
- Among(u"aldi", -1, 1),
- Among(u"taldi", 64, 1),
- Among(u"ki", -1, 1),
- Among(u"ari", -1, 1),
- Among(u"kari", 67, 1),
- Among(u"lari", 67, 1),
- Among(u"tari", 67, 1),
- Among(u"etari", 70, 1),
- Among(u"garri", -1, 2),
- Among(u"karri", -1, 1),
- Among(u"arazi", -1, 1),
- Among(u"tarazi", 74, 1),
- Among(u"an", -1, 1),
- Among(u"ean", 76, 1),
- Among(u"rean", 77, 1),
- Among(u"kan", 76, 1),
- Among(u"etan", 76, 1),
- Among(u"atseden", -1, 3),
- Among(u"men", -1, 1),
- Among(u"pen", -1, 1),
- Among(u"kin", -1, 1),
- Among(u"rekin", 84, 1),
- Among(u"ezin", -1, 1),
- Among(u"tezin", 86, 1),
- Among(u"tun", -1, 1),
- Among(u"kizun", -1, 1),
- Among(u"go", -1, 1),
- Among(u"ago", 90, 1),
- Among(u"tio", -1, 1),
- Among(u"dako", -1, 1),
- Among(u"or", -1, 1),
- Among(u"kor", 94, 1),
- Among(u"tzat", -1, 1),
- Among(u"du", -1, 1),
- Among(u"gailu", -1, 1),
- Among(u"tu", -1, 1),
- Among(u"atu", 99, 1),
- Among(u"aldatu", 100, 1),
- Among(u"tatu", 100, 1),
- Among(u"baditu", 99, 5),
- Among(u"ez", -1, 1),
- Among(u"errez", 104, 1),
- Among(u"tzez", 104, 1),
- Among(u"gaitz", -1, 1),
- Among(u"kaitz", -1, 1)
- ]
- a_1 = [
- Among(u"ada", -1, 1),
- Among(u"kada", 0, 1),
- Among(u"anda", -1, 1),
- Among(u"denda", -1, 1),
- Among(u"gabea", -1, 1),
- Among(u"kabea", -1, 1),
- Among(u"aldea", -1, 1),
- Among(u"kaldea", 6, 1),
- Among(u"taldea", 6, 1),
- Among(u"ordea", -1, 1),
- Among(u"zalea", -1, 1),
- Among(u"tzalea", 10, 1),
- Among(u"gilea", -1, 1),
- Among(u"emea", -1, 1),
- Among(u"kumea", -1, 1),
- Among(u"nea", -1, 1),
- Among(u"enea", 15, 1),
- Among(u"zionea", 15, 1),
- Among(u"unea", 15, 1),
- Among(u"gunea", 18, 1),
- Among(u"pea", -1, 1),
- Among(u"aurrea", -1, 1),
- Among(u"tea", -1, 1),
- Among(u"kotea", 22, 1),
- Among(u"artea", 22, 1),
- Among(u"ostea", 22, 1),
- Among(u"etxea", -1, 1),
- Among(u"ga", -1, 1),
- Among(u"anga", 27, 1),
- Among(u"gaia", -1, 1),
- Among(u"aldia", -1, 1),
- Among(u"taldia", 30, 1),
- Among(u"handia", -1, 1),
- Among(u"mendia", -1, 1),
- Among(u"geia", -1, 1),
- Among(u"egia", -1, 1),
- Among(u"degia", 35, 1),
- Among(u"tegia", 35, 1),
- Among(u"nahia", -1, 1),
- Among(u"ohia", -1, 1),
- Among(u"kia", -1, 1),
- Among(u"tokia", 40, 1),
- Among(u"oia", -1, 1),
- Among(u"koia", 42, 1),
- Among(u"aria", -1, 1),
- Among(u"karia", 44, 1),
- Among(u"laria", 44, 1),
- Among(u"taria", 44, 1),
- Among(u"eria", -1, 1),
- Among(u"keria", 48, 1),
- Among(u"teria", 48, 1),
- Among(u"garria", -1, 2),
- Among(u"larria", -1, 1),
- Among(u"kirria", -1, 1),
- Among(u"duria", -1, 1),
- Among(u"asia", -1, 1),
- Among(u"tia", -1, 1),
- Among(u"ezia", -1, 1),
- Among(u"bizia", -1, 1),
- Among(u"ontzia", -1, 1),
- Among(u"ka", -1, 1),
- Among(u"joka", 60, 3),
- Among(u"aurka", 60, 10),
- Among(u"ska", 60, 1),
- Among(u"xka", 60, 1),
- Among(u"zka", 60, 1),
- Among(u"gibela", -1, 1),
- Among(u"gela", -1, 1),
- Among(u"kaila", -1, 1),
- Among(u"skila", -1, 1),
- Among(u"tila", -1, 1),
- Among(u"ola", -1, 1),
- Among(u"na", -1, 1),
- Among(u"kana", 72, 1),
- Among(u"ena", 72, 1),
- Among(u"garrena", 74, 1),
- Among(u"gerrena", 74, 1),
- Among(u"urrena", 74, 1),
- Among(u"zaina", 72, 1),
- Among(u"tzaina", 78, 1),
- Among(u"kina", 72, 1),
- Among(u"mina", 72, 1),
- Among(u"garna", 72, 1),
- Among(u"una", 72, 1),
- Among(u"duna", 83, 1),
- Among(u"asuna", 83, 1),
- Among(u"tasuna", 85, 1),
- Among(u"ondoa", -1, 1),
- Among(u"kondoa", 87, 1),
- Among(u"ngoa", -1, 1),
- Among(u"zioa", -1, 1),
- Among(u"koa", -1, 1),
- Among(u"takoa", 91, 1),
- Among(u"zkoa", 91, 1),
- Among(u"noa", -1, 1),
- Among(u"zinoa", 94, 1),
- Among(u"aroa", -1, 1),
- Among(u"taroa", 96, 1),
- Among(u"zaroa", 96, 1),
- Among(u"eroa", -1, 1),
- Among(u"oroa", -1, 1),
- Among(u"osoa", -1, 1),
- Among(u"toa", -1, 1),
- Among(u"ttoa", 102, 1),
- Among(u"ztoa", 102, 1),
- Among(u"txoa", -1, 1),
- Among(u"tzoa", -1, 1),
- Among(u"\u00F1oa", -1, 1),
- Among(u"ra", -1, 1),
- Among(u"ara", 108, 1),
- Among(u"dara", 109, 1),
- Among(u"liara", 109, 1),
- Among(u"tiara", 109, 1),
- Among(u"tara", 109, 1),
- Among(u"etara", 113, 1),
- Among(u"tzara", 109, 1),
- Among(u"bera", 108, 1),
- Among(u"kera", 108, 1),
- Among(u"pera", 108, 1),
- Among(u"ora", 108, 2),
- Among(u"tzarra", 108, 1),
- Among(u"korra", 108, 1),
- Among(u"tra", 108, 1),
- Among(u"sa", -1, 1),
- Among(u"osa", 123, 1),
- Among(u"ta", -1, 1),
- Among(u"eta", 125, 1),
- Among(u"keta", 126, 1),
- Among(u"sta", 125, 1),
- Among(u"dua", -1, 1),
- Among(u"mendua", 129, 1),
- Among(u"ordua", 129, 1),
- Among(u"lekua", -1, 1),
- Among(u"burua", -1, 1),
- Among(u"durua", -1, 1),
- Among(u"tsua", -1, 1),
- Among(u"tua", -1, 1),
- Among(u"mentua", 136, 1),
- Among(u"estua", 136, 1),
- Among(u"txua", -1, 1),
- Among(u"zua", -1, 1),
- Among(u"tzua", 140, 1),
- Among(u"za", -1, 1),
- Among(u"eza", 142, 1),
- Among(u"eroza", 142, 1),
- Among(u"tza", 142, 2),
- Among(u"koitza", 145, 1),
- Among(u"antza", 145, 1),
- Among(u"gintza", 145, 1),
- Among(u"kintza", 145, 1),
- Among(u"kuntza", 145, 1),
- Among(u"gabe", -1, 1),
- Among(u"kabe", -1, 1),
- Among(u"kide", -1, 1),
- Among(u"alde", -1, 1),
- Among(u"kalde", 154, 1),
- Among(u"talde", 154, 1),
- Among(u"orde", -1, 1),
- Among(u"ge", -1, 1),
- Among(u"zale", -1, 1),
- Among(u"tzale", 159, 1),
- Among(u"gile", -1, 1),
- Among(u"eme", -1, 1),
- Among(u"kume", -1, 1),
- Among(u"ne", -1, 1),
- Among(u"zione", 164, 1),
- Among(u"une", 164, 1),
- Among(u"gune", 166, 1),
- Among(u"pe", -1, 1),
- Among(u"aurre", -1, 1),
- Among(u"te", -1, 1),
- Among(u"kote", 170, 1),
- Among(u"arte", 170, 1),
- Among(u"oste", 170, 1),
- Among(u"etxe", -1, 1),
- Among(u"gai", -1, 1),
- Among(u"di", -1, 1),
- Among(u"aldi", 176, 1),
- Among(u"taldi", 177, 1),
- Among(u"geldi", 176, 8),
- Among(u"handi", 176, 1),
- Among(u"mendi", 176, 1),
- Among(u"gei", -1, 1),
- Among(u"egi", -1, 1),
- Among(u"degi", 183, 1),
- Among(u"tegi", 183, 1),
- Among(u"nahi", -1, 1),
- Among(u"ohi", -1, 1),
- Among(u"ki", -1, 1),
- Among(u"toki", 188, 1),
- Among(u"oi", -1, 1),
- Among(u"goi", 190, 1),
- Among(u"koi", 190, 1),
- Among(u"ari", -1, 1),
- Among(u"kari", 193, 1),
- Among(u"lari", 193, 1),
- Among(u"tari", 193, 1),
- Among(u"garri", -1, 2),
- Among(u"larri", -1, 1),
- Among(u"kirri", -1, 1),
- Among(u"duri", -1, 1),
- Among(u"asi", -1, 1),
- Among(u"ti", -1, 1),
- Among(u"ontzi", -1, 1),
- Among(u"\u00F1i", -1, 1),
- Among(u"ak", -1, 1),
- Among(u"ek", -1, 1),
- Among(u"tarik", -1, 1),
- Among(u"gibel", -1, 1),
- Among(u"ail", -1, 1),
- Among(u"kail", 209, 1),
- Among(u"kan", -1, 1),
- Among(u"tan", -1, 1),
- Among(u"etan", 212, 1),
- Among(u"en", -1, 4),
- Among(u"ren", 214, 2),
- Among(u"garren", 215, 1),
- Among(u"gerren", 215, 1),
- Among(u"urren", 215, 1),
- Among(u"ten", 214, 4),
- Among(u"tzen", 214, 4),
- Among(u"zain", -1, 1),
- Among(u"tzain", 221, 1),
- Among(u"kin", -1, 1),
- Among(u"min", -1, 1),
- Among(u"dun", -1, 1),
- Among(u"asun", -1, 1),
- Among(u"tasun", 226, 1),
- Among(u"aizun", -1, 1),
- Among(u"ondo", -1, 1),
- Among(u"kondo", 229, 1),
- Among(u"go", -1, 1),
- Among(u"ngo", 231, 1),
- Among(u"zio", -1, 1),
- Among(u"ko", -1, 1),
- Among(u"trako", 234, 5),
- Among(u"tako", 234, 1),
- Among(u"etako", 236, 1),
- Among(u"eko", 234, 1),
- Among(u"tariko", 234, 1),
- Among(u"sko", 234, 1),
- Among(u"tuko", 234, 1),
- Among(u"minutuko", 241, 6),
- Among(u"zko", 234, 1),
- Among(u"no", -1, 1),
- Among(u"zino", 244, 1),
- Among(u"ro", -1, 1),
- Among(u"aro", 246, 1),
- Among(u"igaro", 247, 9),
- Among(u"taro", 247, 1),
- Among(u"zaro", 247, 1),
- Among(u"ero", 246, 1),
- Among(u"giro", 246, 1),
- Among(u"oro", 246, 1),
- Among(u"oso", -1, 1),
- Among(u"to", -1, 1),
- Among(u"tto", 255, 1),
- Among(u"zto", 255, 1),
- Among(u"txo", -1, 1),
- Among(u"tzo", -1, 1),
- Among(u"gintzo", 259, 1),
- Among(u"\u00F1o", -1, 1),
- Among(u"zp", -1, 1),
- Among(u"ar", -1, 1),
- Among(u"dar", 263, 1),
- Among(u"behar", 263, 1),
- Among(u"zehar", 263, 7),
- Among(u"liar", 263, 1),
- Among(u"tiar", 263, 1),
- Among(u"tar", 263, 1),
- Among(u"tzar", 263, 1),
- Among(u"or", -1, 2),
- Among(u"kor", 271, 1),
- Among(u"os", -1, 1),
- Among(u"ket", -1, 1),
- Among(u"du", -1, 1),
- Among(u"mendu", 275, 1),
- Among(u"ordu", 275, 1),
- Among(u"leku", -1, 1),
- Among(u"buru", -1, 2),
- Among(u"duru", -1, 1),
- Among(u"tsu", -1, 1),
- Among(u"tu", -1, 1),
- Among(u"tatu", 282, 4),
- Among(u"mentu", 282, 1),
- Among(u"estu", 282, 1),
- Among(u"txu", -1, 1),
- Among(u"zu", -1, 1),
- Among(u"tzu", 287, 1),
- Among(u"gintzu", 288, 1),
- Among(u"z", -1, 1),
- Among(u"ez", 290, 1),
- Among(u"eroz", 290, 1),
- Among(u"tz", 290, 1),
- Among(u"koitz", 293, 1)
- ]
- a_2 = [
- Among(u"zlea", -1, 2),
- Among(u"keria", -1, 1),
- Among(u"la", -1, 1),
- Among(u"era", -1, 1),
- Among(u"dade", -1, 1),
- Among(u"tade", -1, 1),
- Among(u"date", -1, 1),
- Among(u"tate", -1, 1),
- Among(u"gi", -1, 1),
- Among(u"ki", -1, 1),
- Among(u"ik", -1, 1),
- Among(u"lanik", 10, 1),
- Among(u"rik", 10, 1),
- Among(u"larik", 12, 1),
- Among(u"ztik", 10, 1),
- Among(u"go", -1, 1),
- Among(u"ro", -1, 1),
- Among(u"ero", 16, 1),
- Among(u"to", -1, 1)
- ]
- g_v = [17, 65, 16]
- I_p2 = 0
- I_p1 = 0
- I_pV = 0
- def __r_mark_regions(self):
- self.I_pV = self.limit
- self.I_p1 = self.limit
- self.I_p2 = self.limit
- v_1 = self.cursor
- try:
- try:
- v_2 = self.cursor
- try:
- if not self.in_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab2()
- try:
- v_3 = self.cursor
- try:
- if not self.out_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab4()
- if not self.go_out_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab4()
- self.cursor += 1
- raise lab3()
- except lab4: pass
- self.cursor = v_3
- if not self.in_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab2()
- if not self.go_in_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab2()
- self.cursor += 1
- except lab3: pass
- raise lab1()
- except lab2: pass
- self.cursor = v_2
- if not self.out_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab0()
- try:
- v_4 = self.cursor
- try:
- if not self.out_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab6()
- if not self.go_out_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab6()
- self.cursor += 1
- raise lab5()
- except lab6: pass
- self.cursor = v_4
- if not self.in_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab0()
- if self.cursor >= self.limit:
- raise lab0()
- self.cursor += 1
- except lab5: pass
- except lab1: pass
- self.I_pV = self.cursor
- except lab0: pass
- self.cursor = v_1
- v_5 = self.cursor
- try:
- if not self.go_out_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab7()
- self.cursor += 1
- if not self.go_in_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab7()
- self.cursor += 1
- self.I_p1 = self.cursor
- if not self.go_out_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab7()
- self.cursor += 1
- if not self.go_in_grouping(BasqueStemmer.g_v, 97, 117):
- raise lab7()
- self.cursor += 1
- self.I_p2 = self.cursor
- except lab7: pass
- self.cursor = v_5
- return True
- def __r_RV(self):
- if not self.I_pV <= self.cursor:
- return False
- return True
- def __r_R2(self):
- if not self.I_p2 <= self.cursor:
- return False
- return True
- def __r_R1(self):
- if not self.I_p1 <= self.cursor:
- return False
- return True
- def __r_aditzak(self):
- self.ket = self.cursor
- among_var = self.find_among_b(BasqueStemmer.a_0)
- if among_var == 0:
- return False
- self.bra = self.cursor
- if among_var == 1:
- if not self.__r_RV():
- return False
- if not self.slice_del():
- return False
- elif among_var == 2:
- if not self.__r_R2():
- return False
- if not self.slice_del():
- return False
- elif among_var == 3:
- if not self.slice_from(u"atseden"):
- return False
- elif among_var == 4:
- if not self.slice_from(u"arabera"):
- return False
- else:
- if not self.slice_from(u"baditu"):
- return False
- return True
- def __r_izenak(self):
- self.ket = self.cursor
- among_var = self.find_among_b(BasqueStemmer.a_1)
- if among_var == 0:
- return False
- self.bra = self.cursor
- if among_var == 1:
- if not self.__r_RV():
- return False
- if not self.slice_del():
- return False
- elif among_var == 2:
- if not self.__r_R2():
- return False
- if not self.slice_del():
- return False
- elif among_var == 3:
- if not self.slice_from(u"jok"):
- return False
- elif among_var == 4:
- if not self.__r_R1():
- return False
- if not self.slice_del():
- return False
- elif among_var == 5:
- if not self.slice_from(u"tra"):
- return False
- elif among_var == 6:
- if not self.slice_from(u"minutu"):
- return False
- elif among_var == 7:
- if not self.slice_from(u"zehar"):
- return False
- elif among_var == 8:
- if not self.slice_from(u"geldi"):
- return False
- elif among_var == 9:
- if not self.slice_from(u"igaro"):
- return False
- else:
- if not self.slice_from(u"aurka"):
- return False
- return True
- def __r_adjetiboak(self):
- self.ket = self.cursor
- among_var = self.find_among_b(BasqueStemmer.a_2)
- if among_var == 0:
- return False
- self.bra = self.cursor
- if among_var == 1:
- if not self.__r_RV():
- return False
- if not self.slice_del():
- return False
- else:
- if not self.slice_from(u"z"):
- return False
- return True
- def _stem(self):
- self.__r_mark_regions()
- self.limit_backward = self.cursor
- self.cursor = self.limit
- while True:
- v_2 = self.limit - self.cursor
- try:
- if not self.__r_aditzak():
- raise lab0()
- continue
- except lab0: pass
- self.cursor = self.limit - v_2
- break
- while True:
- v_3 = self.limit - self.cursor
- try:
- if not self.__r_izenak():
- raise lab1()
- continue
- except lab1: pass
- self.cursor = self.limit - v_3
- break
- v_4 = self.limit - self.cursor
- self.__r_adjetiboak()
- self.cursor = self.limit - v_4
- self.cursor = self.limit_backward
- return True
- class lab0(BaseException): pass
- class lab1(BaseException): pass
- class lab2(BaseException): pass
- class lab3(BaseException): pass
- class lab4(BaseException): pass
- class lab5(BaseException): pass
- class lab6(BaseException): pass
- class lab7(BaseException): pass
|