Alexis Palmer commited on
Commit ·
ba43e42
1
Parent(s): 97bf531
Mapudungun hangman test
Browse files- map_avenue/map-test.txt +268 -0
- util.py +69 -0
map_avenue/map-test.txt
ADDED
|
@@ -0,0 +1,268 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
fillantü witrakeymi fillantü
|
| 2 |
+
epu, epu antü küla antü ka ta yagel kay ka falikalu tüfanay ngewelay kuyfi ngelafuy yagel anay
|
| 3 |
+
welu eymi rumel puliwen witrakeymi
|
| 4 |
+
puliwen witraken <*SPA>la <*SPA>hora <*SPA>que <*SPA>quiere <*SPA>levantar witrakey fey iñche tüfa petu mongelen nay ünikerkey ta wenu mapu ta ünikerkey ta mongelealu che nay iñche epu rupa meli rupachi ünin
|
| 5 |
+
¡Aa!
|
| 6 |
+
ünin mongen, tüfa petu mongelen nay meli rupachi ünin taiñ peuma
|
| 7 |
+
¡Aa!
|
| 8 |
+
taiñ peuma tati ünikey che nay ngünechen femkunuel kay nay üniy üni ta che mongelealu ta ünikerkellinay
|
| 9 |
+
¡Aa!
|
| 10 |
+
mongelealu ünikey <!1üniki> anay refalte mongelekelay che tüfa iñche may petu mongelellen may nay müte fütalan anay newe pelowelan ke llemay
|
| 11 |
+
chumngelu chi pelowelaymi
|
| 12 |
+
feyta ñi kintun nem ka na nay llampaykoketuy chi nay femkey chi nay
|
| 13 |
+
amukeymi pitalmu kay
|
| 14 |
+
amukelan müte konkelan rume pitalmuanay kiñeke rupamu apuranen kutran fey ngemeken pitalmu
|
| 15 |
+
chempikeymu wingka
|
| 16 |
+
<*SPA>serio <*SPA>que <*SPA>tiene <*SPA>tanto <*SPA>años <*SPA>éste Camargo <*SPA>hombre pingeken ka
|
| 17 |
+
<Laugh> chumalanta küpaymi pingekeymi
|
| 18 |
+
feypingeken ka fey <*SPA>tanto, <*SPA>tanto <*SPA>este <*SPA>no <*SPA>quiere <*SPA>entrar niel <*SPA>el <*SPA>kuchillo <*SPA>éste wewon pikeneu ta pu wingka ayekalkeneu ka
|
| 19 |
+
eymi müley tami feypiafel iñche ta petu weche wentrulen piafimi
|
| 20 |
+
feypikellefiñ may <%> <*SPA>quiero <*SPA>casar <*SPA>hombre pikefiñ engün ayekaukingün
|
| 21 |
+
¡Aa! kurengean pikefimi
|
| 22 |
+
may iñche ta petu mongelen llemay
|
| 23 |
+
elumulayafun küdau may pikelafimi
|
| 24 |
+
feypikelafiñ puwelan küdaumu anay pelotuwelan
|
| 25 |
+
welu re feypikantuafiel
|
| 26 |
+
may pelowelan tüfa <Noise>
|
| 27 |
+
eymi tami chau chunten tripantu mongey
|
| 28 |
+
uta <*SPA>más doy <*SPA>cien
|
| 29 |
+
doy pataka
|
| 30 |
+
may
|
| 31 |
+
tami ñuke
|
| 32 |
+
tañi ñuke ta inangechi laynga ka epe, ka epe pataka niefuynga
|
| 33 |
+
epe pataka niefuy
|
| 34 |
+
may
|
| 35 |
+
<Cough> ¡Aa! cheu tripapay tami ñuke faupüle müten tuwi
|
| 36 |
+
<*SPA>no <*SPA>afuera, <*SPA>afuera
|
| 37 |
+
chuchipüle
|
| 38 |
+
<*SPA>selva <*SPA>oscura <!1selvacura>
|
| 39 |
+
üyüu tuwüy
|
| 40 |
+
may <*SPA>perquenco pingey tati mapu ta feypüle wingka mapefiñ pilay ta mapu nay
|
| 41 |
+
¡Aa!
|
| 42 |
+
ka eymingelu tukulelfi üy kakelu femngechi kampiayka <!1cambiayca>
|
| 43 |
+
¡Aa!
|
| 44 |
+
niekekelafuy ta wingka ta tüfa nay eymiam chuchi mapuche ta ka tüfa nay femyawiymi ka nay weniy
|
| 45 |
+
iñche ta Alfarinu pingey tañi mapu
|
| 46 |
+
Alfarinu
|
| 47 |
+
may
|
| 48 |
+
Alfarinu
|
| 49 |
+
may
|
| 50 |
+
iñche kayta kimlafiñ kayta Alfarino, Alfarinu pikay ta che ka lle
|
| 51 |
+
Temuko kay
|
| 52 |
+
may, may Temuko ta feyta miauken
|
| 53 |
+
kuyfi deuma amuwelaymi ka?
|
| 54 |
+
tüfa ngemewekelan anay plataam ka fütra <*SPA>escaso nay
|
| 55 |
+
pensionauküleymi?
|
| 56 |
+
pensionaukülefun <*SPA>pero pichiken nüken
|
| 57 |
+
pichiken llemay
|
| 58 |
+
pichiken
|
| 59 |
+
¡Aa! welu <uh> fülküleymi wariamu tati
|
| 60 |
+
fülkülefun wariamu pekanka wül-lay wingka rume kam paysanumu anay pekanka wül-lay kuyfitu doy pekankakefeyu fütrake aweluyem
|
| 61 |
+
¡Aa!
|
| 62 |
+
<*SPA>y pu wingka dpy pekanka tripayal tati pofüre <*SPA>hombre pifuy tüfa pu wingka femwelay anay tenga pülata kontau pipiey
|
| 63 |
+
kuyfi doy elungekefuymün
|
| 64 |
+
doy pichi elungekeyiñ chemu rume tati nielan pülata pingipiy regalangekay iyagel fey tüfa femwelay anay re pülatamu müten amuley
|
| 65 |
+
nienole pülata fey amulay?
|
| 66 |
+
ya amulay, amulay
|
| 67 |
+
¡Aa!
|
| 68 |
+
femkey che tüfa anay
|
| 69 |
+
tami <uh> chemchi ta kümey ta kutran furinmu
|
| 70 |
+
kuyfi
|
| 71 |
+
may kutran furinmu chem kümey, kutraniem furi
|
| 72 |
+
<%> tati chem fütake awelungati l'awen' ta kimüy ka anay iñchiñ ta kimlayiñ ta anay <%> tüfa kay ta tañi domo nay palusantu piy
|
| 73 |
+
palusantu yengu ta nalka
|
| 74 |
+
rume kümey pingey <!1pingi> fey feyngey
|
| 75 |
+
feyngey, feyngey
|
| 76 |
+
eymi putulafimi fey
|
| 77 |
+
<*SPA>no putukelafiñ tati
|
| 78 |
+
¡Aa!<%>
|
| 79 |
+
tati <*SPA>limpiaplata ka müley
|
| 80 |
+
<uh> pewen anta l'awen' ngekey kay chachay
|
| 81 |
+
fey tati trantupatuy allfeñngelu ta +/wor/+ workükakeychi nay fey notukey ta che tati allfeñngelulle fey
|
| 82 |
+
pewenmu
|
| 83 |
+
may pewenmu fayümngey
|
| 84 |
+
fey kümey
|
| 85 |
+
fey kümey
|
| 86 |
+
¿alüley pewen tapüle?
|
| 87 |
+
pütrakamaley chi nay repül nentukatuy mawidamu re pewen nentulley chi nay
|
| 88 |
+
feymuanta pewenche pingekey?
|
| 89 |
+
feymulle ta pewenche pingey tati nay
|
| 90 |
+
¡Aa! fentrekeley pewen
|
| 91 |
+
pütrakamagel tüfa kay ta wingka ta apümüy anay apümüy
|
| 92 |
+
chumngelu chi
|
| 93 |
+
<hm> seriakefiyngün kay nay
|
| 94 |
+
traflafingün
|
| 95 |
+
traflafingün fey ta tau newentulelu müten ta paysanu kay feyta küpaymangelay tañi feranamullenay
|
| 96 |
+
newentulelu
|
| 97 |
+
may feypingey wentrüu mekey kay nay
|
| 98 |
+
¡Aa!
|
| 99 |
+
femüymamukelchi pipingey feymu ta kapalekellinanay ngülliumu pütrümna küñeke tripantumu müna fünkey ngülliu anay, kiñeke tripantu müten kiñeke tripantu weda pukemiyem tati wül-lay ta ngünechen müna nongikelley tati ngilliu anay
|
| 100 |
+
ngilliu anta ingekey ka
|
| 101 |
+
chemuam müna kümey mudayünmu ngülliu anay
|
| 102 |
+
¡Aa!
|
| 103 |
+
pütrümna kümey ta ngülliu ta müdayünmu lle ka tati füntu pingey tati y deumangey korünmu ellangey tati angküchi medküngeyanay rumeñma kümey
|
| 104 |
+
ka kümey
|
| 105 |
+
kümey
|
| 106 |
+
¡Aa!
|
| 107 |
+
fey felekelleychi
|
| 108 |
+
ilo korümu ta püstuy
|
| 109 |
+
alü niekeymün anta kapüra faupüle
|
| 110 |
+
nielayiñ kapüra
|
| 111 |
+
kuyfi ta mülekefuy ka tüfa ta ngewelay kiñekenekakefuyiñ kuyfi tüfa ta niewelayiñ anay
|
| 112 |
+
niewelay
|
| 113 |
+
eskasiay anta chem rume ta wülngerkelaynay keya mongeltual tañi <*SPA>familia ta che
|
| 114 |
+
may, may
|
| 115 |
+
eskasiafun kompüle ta faltakey ka che nay
|
| 116 |
+
fey femngechi kuyfi niekefuymi
|
| 117 |
+
kuyfi niekellefuyiñ pero tüfa ta femwelayiñ anay tüfa rumeñma eskaseakelu itrokom nay
|
| 118 |
+
¡Aa!
|
| 119 |
+
rumeñma eskaseakeyiñ <Noise> pewelayin ta tüfa anay pekan pengewelay tati
|
| 120 |
+
pekan pengewelay
|
| 121 |
+
<*SPA>no
|
| 122 |
+
¡Aa! eymi pilukelaymi?
|
| 123 |
+
<*SPA>no pilukelan chemkülekan müten
|
| 124 |
+
kümelkaley tami pilun
|
| 125 |
+
may
|
| 126 |
+
allküleymi küme
|
| 127 |
+
allkülen
|
| 128 |
+
¡Aa!
|
| 129 |
+
<*SPA>eso <*SPA>si <*SPA>que pelowelanke anay, pelowenke
|
| 130 |
+
welu chafokeymi kay
|
| 131 |
+
¡Aa!
|
| 132 |
+
¿chafokeymi?
|
| 133 |
+
<*SPA>no
|
| 134 |
+
chafokelaymi
|
| 135 |
+
<*SPA>no
|
| 136 |
+
ya, ya, ya, ya feyti kiñe kutran müley kimlafimi wauyuyem che?
|
| 137 |
+
<*SPA>no kimlafiñ anay
|
| 138 |
+
¡Aa!
|
| 139 |
+
kimlafiñ
|
| 140 |
+
wauyukelaymi?
|
| 141 |
+
<*SPA>no iñche tüfa petu mongelen nay
|
| 142 |
+
welu eymi kutranchengelaymi <*SPA>pues <!1pu'>
|
| 143 |
+
kutranchengekelay kiñekemu rumeñma kutranken rüfanay chuman kay
|
| 144 |
+
chem doy kutrankey?
|
| 145 |
+
tañi piukeyem, tañi pu ngeyem kutrankey kiñekemu tulliuken anay, femkey che anay
|
| 146 |
+
piuke kutraniyem chem l'awen' putukeymi
|
| 147 |
+
feyti foye workükefiñ
|
| 148 |
+
foye
|
| 149 |
+
may
|
| 150 |
+
¡Aa!
|
| 151 |
+
feytamu ngellelay foye nay
|
| 152 |
+
kañpüle müley <*SPA>pues <!1pu'>
|
| 153 |
+
kañpüle <*SPA>afuera müten müley ka
|
| 154 |
+
feyta müley ta re ngülliu faupüle
|
| 155 |
+
femay re ngilliu ta mülelli nay
|
| 156 |
+
welu ngelay foye
|
| 157 |
+
ngelay, ngelay
|
| 158 |
+
¡Aa!
|
| 159 |
+
ngellelay foye tamu
|
| 160 |
+
<uh> pafmu chem kümey pafmuyem?
|
| 161 |
+
+/pafmuye/+ +/pa/+ pafmuyem
|
| 162 |
+
may
|
| 163 |
+
fey tati chem ka tukungekey ama
|
| 164 |
+
feyta kümey
|
| 165 |
+
pafmulu
|
| 166 |
+
pafmunmu
|
| 167 |
+
palki
|
| 168 |
+
may
|
| 169 |
+
müley palki faupüle kay
|
| 170 |
+
ngelaychi nay
|
| 171 |
+
ngelay palki
|
| 172 |
+
may
|
| 173 |
+
welu mawidamu fentrelepelafuy l'awen' peñi
|
| 174 |
+
tüfamu
|
| 175 |
+
may
|
| 176 |
+
mülellefuy may
|
| 177 |
+
<uhm> tüfamu keta müley l'awen' <%>
|
| 178 |
+
welu kuyfi kimfuymi l'awen'
|
| 179 |
+
kimkallefun mayta tati kutran *pütrawe pingey feyta tungefuy kay nay feyta tüfa ta pelotuwelan ka nay
|
| 180 |
+
welu üy kimneymafimi l'awen'
|
| 181 |
+
llipañmawe pingey kangelu tati chempingekey chi pe fey engün ta mülefuyanay
|
| 182 |
+
ya <Noise> welu <uh> alün l'awen' putupelaymi kuyfi mapu l'awen' eymi
|
| 183 |
+
worküngey fayümngey tati l'awen' tati, tati *matragelu ta fey tutekey tati anay tutekey
|
| 184 |
+
tutekey
|
| 185 |
+
may
|
| 186 |
+
¡Aa! feyta <uh> nielu mongen tutey ka
|
| 187 |
+
tutellimay, tutey iñche ta petu mongelellen may pütra kuyfiche
|
| 188 |
+
welu kümey tami alütun mongen tati peñi
|
| 189 |
+
kürasia may nay kürasia
|
| 190 |
+
kümey tami alünmongen tun
|
| 191 |
+
fey kürasia pipiepan tati
|
| 192 |
+
müley tami
|
| 193 |
+
taiñ choyeteu ka ta ngünechen kayta witrañpüramnieneu feyta mongelen ka nay <*SPA>mucha gürasia tami pipiepafiel
|
| 194 |
+
ngillatuleymukeymi welu
|
| 195 |
+
¡Aa!
|
| 196 |
+
ngillatulekeymi
|
| 197 |
+
ngüllatuken nay, workam chem l'awen' nay ngillatu pichiken püntefnentukaken kay ka
|
| 198 |
+
¡Aa!
|
| 199 |
+
femkey che anay
|
| 200 |
+
welu ngillatuley che mongelekey pingey
|
| 201 |
+
femkey
|
| 202 |
+
eymi rumel ngillatukeymi
|
| 203 |
+
ngillatumeken worküayem fey ngillatuken tüfa kay ta felen müte chumlekalafun anay <*SPA>eso <*SPA>si <*SPA>que pelowellelan
|
| 204 |
+
peumakeymi kay
|
| 205 |
+
peumaken ünin melike rupachi ünin.
|
| 206 |
+
¡Aa! feyta akukelay pütrümna *wetanki iñamnga akuay pipeyu
|
| 207 |
+
¡Aa!
|
| 208 |
+
tüfa aku akungepey tañi witan chemuam femekey tüfa fey tüfi llemay pipingepatun ta, ta
|
| 209 |
+
witaymi
|
| 210 |
+
may
|
| 211 |
+
¿küme witaymiam?
|
| 212 |
+
müna küme witan tayi pipiepefiñ ta tüfa
|
| 213 |
+
fey mayta iñche tañi akual
|
| 214 |
+
femüy fey tami akukellal may tati femekepen
|
| 215 |
+
pemean ñi peñi doy alütunielu mongen pipenga?
|
| 216 |
+
tüfey may
|
| 217 |
+
doy alünielu mongen fey pemeafiñ pipen fey
|
| 218 |
+
fey fütra kuyfi che iñche tüfa kay
|
| 219 |
+
feymu mayta witapeymi?
|
| 220 |
+
feymu
|
| 221 |
+
welu küme witapeymi?
|
| 222 |
+
küme witapen <*SPA>porque müna kimyey tañi witan ta che anay
|
| 223 |
+
¡Aa!
|
| 224 |
+
tüfa man lipang tayiye chemmu amuy fente mekepen pipen
|
| 225 |
+
inichi akuay pipeymi
|
| 226 |
+
may ta tüfa llemay ngütramkamekepatuymi tüfa tati nay
|
| 227 |
+
iñche duamtukeyu pu fey
|
| 228 |
+
may, may
|
| 229 |
+
doy alü nielu mongen feyta ngütramkameafiñ piken
|
| 230 |
+
may kümeyka fey kürasia may nay kürasia
|
| 231 |
+
kimkelafun müten cheu tami mülen
|
| 232 |
+
may nay
|
| 233 |
+
wingkulmu mülerkeymi <!1mülerkimi>
|
| 234 |
+
wingkulüukülen, wingkülüukülen
|
| 235 |
+
winkülmu müleymi may
|
| 236 |
+
may fey
|
| 237 |
+
küme dungu tati tami mülen wingkulmeu
|
| 238 |
+
fey chumngechianta kimüymi ta tüfa femngechi püfkükonpaymi tapüle peñi
|
| 239 |
+
ngütramkapefiñ tami fotüm
|
| 240 |
+
may, may
|
| 241 |
+
Domingo
|
| 242 |
+
Domingo Cayul
|
| 243 |
+
Domingu Camargo
|
| 244 |
+
feylle fey
|
| 245 |
+
Cayul
|
| 246 |
+
Cayul feyti ñi chau pütrüney mongen pipey
|
| 247 |
+
may, may
|
| 248 |
+
<uh> petu allküley piy ngütramkameafimi kichu alü kimney dungu pipey
|
| 249 |
+
feyta iney <!1iñi> Domingo feyta ñi fotüm nay
|
| 250 |
+
fey pelelmeam ñi chau pipeneu
|
| 251 |
+
may, may
|
| 252 |
+
pemeafimi piy petu nieyta mongeley tañi fücha chau piy
|
| 253 |
+
petu mongelellen may tüfa
|
| 254 |
+
fentepuay ngütram peñi feyta pepapeyu feymu fey peumangelmi doy alüniaymi mongen <*SPA>pues
|
| 255 |
+
kürasia may nay kürasia
|
| 256 |
+
iñche ta rupaletuan tati amuleanta kañpüle
|
| 257 |
+
may, may kañpüle ka wariamu amukayaymi ka ka paysanumu ka, ka
|
| 258 |
+
may
|
| 259 |
+
kam pewenko chi tüfachi pewenko ka pewenko, <*SPA>selva <*SPA>oscura <!1selvecura> pingey tati rupauyimi <*SPA>selva <*SPA>oscura <!1selvecura> eymi ta tüfaka
|
| 260 |
+
feypüle amuan feula
|
| 261 |
+
may, may <*SPA>selva <*SPA>oscura <!1selvecura>
|
| 262 |
+
may
|
| 263 |
+
wiñoltuayminge mapu ka
|
| 264 |
+
may
|
| 265 |
+
may, may
|
| 266 |
+
feyta kümeleaymi pu peñi pepapeyu tati
|
| 267 |
+
kürasia may nay kürasia, fey tañi ngütramkapafiel ka nay
|
| 268 |
+
feley.
|
util.py
ADDED
|
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import re
|
| 3 |
+
import unicodedata
|
| 4 |
+
|
| 5 |
+
def strip_accents(text: str) -> str:
|
| 6 |
+
"""Removes accents from text."""
|
| 7 |
+
return ''.join(c for c in unicodedata.normalize('NFD', text)
|
| 8 |
+
if unicodedata.category(c) != 'Mn')
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
def load_raw_text(corpus_directory: str, file_names=None) -> str:
|
| 12 |
+
"""Loads all the text files in a directory into one large string"""
|
| 13 |
+
corpus = ""
|
| 14 |
+
|
| 15 |
+
for file_name in os.listdir(corpus_directory):
|
| 16 |
+
# Read the file as a string
|
| 17 |
+
file_path = os.path.join(corpus_directory, file_name)
|
| 18 |
+
if os.path.isdir(file_path):
|
| 19 |
+
continue
|
| 20 |
+
|
| 21 |
+
# Make sure we only read text files
|
| 22 |
+
if ".txt" not in file_name:
|
| 23 |
+
continue
|
| 24 |
+
|
| 25 |
+
with open(file_path, 'r') as file:
|
| 26 |
+
file_contents = file.read()
|
| 27 |
+
corpus += (file_contents + "\n")
|
| 28 |
+
return corpus
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
word_regex = r"[\w|\']+"
|
| 32 |
+
def tokenize(text):
|
| 33 |
+
return re.findall(word_regex, text)
|
| 34 |
+
|
| 35 |
+
|
| 36 |
+
def preprocess(text):
|
| 37 |
+
"""Tokenizes and processes text which is already separated by spaces into words. Designed for English punctuation."""
|
| 38 |
+
text = strip_accents(text)
|
| 39 |
+
text = text.lower()
|
| 40 |
+
|
| 41 |
+
tokens = text.split(" ")
|
| 42 |
+
|
| 43 |
+
tokens_filtered = []
|
| 44 |
+
for token in tokens:
|
| 45 |
+
# Skip any tokens with special characters
|
| 46 |
+
if re.match(r"[\w|\']+|[\.|\,|\?|\!]", token):
|
| 47 |
+
tokens_filtered.append(token)
|
| 48 |
+
return tokens_filtered
|
| 49 |
+
|
| 50 |
+
|
| 51 |
+
def pad(text: list, num_padding: int):
|
| 52 |
+
"""Pads the given text, as a list of strings, with <s> characters between sentences."""
|
| 53 |
+
padded_text = []
|
| 54 |
+
|
| 55 |
+
# Add initial padding to the first sentence
|
| 56 |
+
for _ in range(num_padding):
|
| 57 |
+
padded_text.append("<s>")
|
| 58 |
+
|
| 59 |
+
for word in text:
|
| 60 |
+
padded_text.append(word)
|
| 61 |
+
|
| 62 |
+
# Every time we see an end punctuation mark, add <s> tokens before it
|
| 63 |
+
# REPLACE IF YOUR LANGUAGE USES DIFFERENT END PUNCTUATION
|
| 64 |
+
if word in [".", "?", "!"]:
|
| 65 |
+
for _ in range(num_padding):
|
| 66 |
+
padded_text.append("<s>")
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
return padded_text
|