diff --git a/src/icelandic/DictIce.gf b/src/icelandic/DictIce.gf new file mode 100644 index 000000000..8a8dc10f8 --- /dev/null +++ b/src/icelandic/DictIce.gf @@ -0,0 +1,337 @@ +concrete DictIce of DictIceAbs = CatIce ** + open + ParadigmsIce, (I=WiktIrregIce) + in { + +-- monolingual dictionary bootstrapped from LexIce +-- AR 2019-08-07 + +lin anda_V = mkV "anda" "anda" "andaði" "andaður"; -- some bug somewhere... +lin aska_N = mkN "aska" feminine ; +lin auga_N = mkN "auga" neuter ; +lin auðveldur_A = mkA "auðveldur" ; +lin bak_N = mkN "bak" neuter ; +lin banki_N = mkN "banki" masculine ; +lin barn_N = mkN "barn" neuter ; +lin bein_N = mkN "bein" neuter ; +lin beinn_A = mkA "beinn" ; +lin beittur_A = mkA "beittur" ; +lin belja_N = mkN "belja" feminine ; +lin berja_V = (mkV "berja") ; +lin beygja_V = mkV "beygja" "beygi" ; +lin binda_V = I.binda_V ; +lin biðja_V = I.biðja_V ; +lin blautur_A = mkA "blautur" ; +lin blár_A = mkA "blár" "blá" ; +lin blása_V = I.blása_V ; +lin blóm_N = mkN "blóm" neuter ; +lin blóð_N = mkN "blóð" neuter ; +lin borg_N = mkN "borg" "borgir" feminine ; +lin borð_N = mkN "borð" neuter ; +lin borða_V = (mkV "borða" "borða" "borðaði" "borðaður") ; +lin brauð_N = mkN "brauð" neuter ; +lin breiður_A = mkA "breiður" ; +lin brenna_V = I.brenna_V ; +lin brjóst_N = mkN "brjóst" neuter ; +lin brjóta_V = I.brjóta_V ; +lin bróðir_N = mkN "bróðir" "bróður" "bræður" masculine ; +lin brúnn_A = mkA "brúnn" ; +lin bátur_N = mkN "bátur" masculine ; +lin bæta_V = mkV "bæta" "bæti" "bætti" "bættur" ; +lin bíll_N = mkN "bíll" masculine ; +lin bíta_V = I.bíta_V ; +lin bók_N = mkN "bók" "bókar" "bækur" feminine ; +lin bólgna_V = mkV "bólgna" "bólgna" "bólgnaði" "bólgnaður" ; +lin börkur_N = mkN "börkur" "barkar" "berkir" masculine ; +lin búð_N = mkN "búð" "búðir" feminine ; +lin dagblað_N = mkN "dagblað" neuter ; +lin dagur_N = mkN "dagur" masculine ; +lin detta_V = I.detta_V ; +lin deyja_V = I.deyja_V ; +lin drekka_V = I.drekka_V ; +lin drepa_V = I.drepa_V ; +lin drottning_N = mkN "drottning" feminine ; +lin dýr_N = mkN "dýr" neuter ; +lin eftir_V = (I.bíða_V) ; +lin egg_N = mkN "egg" neuter ; +lin eiginkona_N = mkCompoundN "eigin" (kona_N) ; +lin eiginmaður_N = mkCompoundN "eigin" (maður_N) ; +lin eldavél_N = mkN "eldavél" "eldavélar" feminine ; +lin eldur_N = mkN "eldur" "eldar" masculine ; +lin elska_V = (mkV "elska" "elska" "elskaði" "elskaður") ; +lin epli_N = mkN "epli" neuter ; +lin eyra_N = mkN "eyra" neuter ; +lin fallegur_A = mkA "fallegur" ; +lin fara_V = I.fara_V ; +lin faðir_N = mkN "faðir" "föður" "feður" masculine ; +lin ferða_V = mkV "ferða" "ferða" "ferðaði" "ferðaður" ; -- this verb only exists in the middle voice +lin finna_V = I.finna_V ; +lin fiskur_N = mkN "fiskur" masculine ; +lin fita_N = mkN "fita" feminine ; +lin fjall_N = mkN "fjall" neuter ; +lin fjöður_N = mkN "fjöður" "fjaðrir" feminine ; +lin fljóta_V = I.fljóta_V ; +lin fljúga_V = I.fljúga_V ; +lin flugvél_N = mkN "flugvél" "flugvélar" feminine ; +lin flæða_V = mkV "flæða" ; +lin friður_N = mkN "friður" masculine ; +lin frjósa_V = I.frjósa_V ; +lin fræ_N = mkN "fræ" neuter ; +lin frændi_N = mkN "frændi" masculine ; -- frænka is the feminine equivalent +lin fugl_N = mkN "fugl" masculine ; +lin fullur_A = mkA "fullur" ; +lin fótleggur_N = mkCompoundN "fót" (mkN "leggur" "leggjar" "leggir" masculine) ; +lin fótur_N = mkN "fótur" "fætur" masculine ; +lin gamall_A = mkA "gamall" ; +lin ganga_V = I.ganga_V ; +lin garður_N = mkN "garður" masculine ; +lin gefa_V = I.gefa_V ; +lin gera_V = (mkV "gera") ; +lin giftur_A = mkA "giftur" ; +lin gleyma_V = (mkV "gleyma") ; +lin gluggi_N = mkN "gluggi" masculine ; +lin grafa_V = I.grafa_V ; +lin gras_N = mkN "gras" neuter ; +lin grænn_A = mkA "grænn" ; +lin gull_N = mkN "gull" neuter ; +lin gulur_A = mkA "gulur" ; +lin gólf_N = mkN "gólf" neuter ; +lin góður_A = mkA "góður" "góð" "betri" ; +lin görn_N = mkN "görn" "garnir" feminine ; +lin gúmmí_N = mkN "gúmmí" neuter ; +lin halda_V = I.halda_V ; +lin hanski_N = mkN "hanski" masculine ; +lin hata_V = (mkV "hata" "hata" "hataði" "hataður") ; +lin hattur_N = mkN "hattur" masculine ; +lin haus_N = mkN "haus" masculine ; +lin heimskur_A = mkA "heimskur" ; +lin heitur_A = mkA "heitur" ; +lin hestur_N = mkN "hestur" masculine ; +lin heyra_V = (mkV "heyra") ; +lin himinn_N = mkN "himinn" masculine ; +lin hjarta_N = mkN "hjarta" neuter ; +lin hjól_N = mkN "hjól" neuter ; +lin hlaupa_V = I.hlaupa_V ; +lin hlusta_V = (mkV "hlusta" "hlusta" "hlustaði" "hlustaður") ; +lin hlæja_V = I.hlæja_V ; +lin hné_N = mkN "hné" neuter ; +lin hoppa_V = mkV "hoppa" "hoppa" ; +lin horfa_V = (mkV "horfa") ; +lin horn_N = mkN "horn" neuter ; +lin hreinn_A = mkA "hreinn" ; +lin hrækja_V = mkV "hrækja" ; +lin hugsa_V = mkV "hugsa" "hugsa" "hugsaði" "hugsaður" ; +lin hundur_N = mkN "hundur" masculine ; +lin hurð_N = mkN "hurð" "hurðir" feminine ; +lin hvítur_A = mkA "hvítur" ; +lin háls_N = mkN "háls" masculine ; +lin hár_N = mkN "hár" neuter ; +lin háskóli_N = mkN "háskóli" masculine ; +lin hæð_N = mkN "hæð" "hæðir" feminine ; +lin höfn_N = mkN "höfn" "hafnir" feminine ; +lin hönd_N = mkN "hönd" "hönd" "hendi" "handar" "hendur" "hendur" "höndum" "handa" feminine ; +lin húfa_N = mkN "húfa" feminine ; +lin hús_N = mkN "hús" neuter ; +lin húð_N = mkN "húð" "húðir" feminine ; +lin iðnaður_N = mkN "iðnaður" masculine ; +lin járn_N = mkN "járn" neuter ; +lin jörð_N = mkN "jörð" "jarðir" feminine ; +lin kaldur_A = mkA "kaldur" ; +lin kasta_V = (mkV "kasta" "kasta" "kastaði" "kastaður") ; +lin kaupa_V = (mkV "kaupa" "kaupi" "keypti" "keyptur" "keypt") ; +lin kenna_V = (mkV "kenna") ; +lin kennari_N = mkN "kennari" masculine ; +lin kind_N = mkN "kind" "kindar" "kindur" feminine ; +lin kirkja_N = mkN "kirkja" feminine ; +lin kjöt_N = mkN "kjöt" neuter ; +lin klettur_N = mkN "klettur" masculine ; +lin klippa_V = (mkV "klippa") ; +lin klár_A = mkA "klár" "klár" ; +lin klóra_V = (mkV "klóra" "klóra" "klóraði" "klóraður") ; +lin koma_V = I.koma_V ; +lin kona_N = mkN "kona" "konu" "konur" "kvenna" feminine ; +lin konungur_N = mkN "konungur" masculine ; +lin kreista_V = (mkV "kreista") ; +lin kringlóttur_A = mkA "kringlóttur" ; +lin kápa_N = mkN "kápa" feminine ; +lin köttur_N = mkN "köttur" "kattar" "kettir" masculine ; +lin lag_N = mkN "lag" neuter ; +lin lampi_N = mkN "lampi" masculine ; +lin land_N = mkN "land" neuter ; +lin langt_Adv = mkAdv "langt" ; +lin langur_A = mkA "langur" ; +lin lauf_N = mkN "lauf" neuter ; +lin leika_V = I.leika_V ; +lin leita_V = (mkV "leita" "leita" "leitaði" "leitaður") ; +lin lesa_V = I.lesa_V ; +lin lest_N = mkN "lest" "lestir" feminine ; +lin leðinilegur_A = mkA "leðinilegur" ; +lin leður_N = mkN "leður" neuter ; +lin lifa_V = mkV "lifa" ; +lin lifur_N = mkN "lifur" "lifrar" feminine ; +lin list_N = mkN "list" "listir" feminine ; +lin ljótur_A = mkA "ljótur" ; +lin ljúga_V = I.ljúga_V ; +lin loft_N = mkN "loft" neuter ; +lin loka_V = (mkV "loka" "loka" "lokaði" "lokaður") ; +lin láta_V = I.láta_V ; +lin læknir_N = mkN "læknir" "lækni" "lækni" "læknis" "læknar" "lækna" "læknum" "lækna" masculine ; +lin læra_V = (mkV "læra") ; +lin líklegur_A = mkA "líklegur" ; +lin lítill_A = mkA "lítill" "lítil" "minni" ; +lin lögregluþjónn_N = mkCompoundN "lögreglu" (mkN "þjónn" masculine) ; +lin lús_N = mkN "lús" "lýs" feminine ; +lin magi_N = mkN "magi" masculine ; +lin maður_N = mkN "maður" "manns" "menn" masculine ; +lin mikilvægur_A = mkA "mikilvægur" ; +lin mjólk_N = mkN "mjólk" "mjólk" "mjólk" "mjólkur" "mjólk" "mjólk" "mjólk" "mjólkur" feminine ; +lin munnur_N = mkN "munnur" masculine ; +lin myndavél_N = mkN "myndavél" "myndavélar" feminine ; +lin mála_V = (mkV "mála" "mála" "málaði" "málaður") ; +lin málfræði_N = mkN "málfræði" "málfræði" "málfræði" "málfræði" "málfræði" "málfræði" "málfræði" "málfræði" feminine ; +lin móðir_N = mkN "móðir" "mæður" feminine ; +lin nafn_N = mkN "nafn" neuter ; +lin nef_N = mkN "nef" neuter ; +lin nemandi_N = mkN "nemandi" masculine ; +lin nudda_V = (mkV "nudda" "nudda" "nuddaði" "nuddaður") ; +lin nálægur_A = mkA "nálægur" ; +lin nótt_N = mkN "nótt" "nætur" "nætur" "nótta" feminine ; +lin nögl_N = mkN "nögl" "naglar" "neglur" feminine ; +lin númer_N = mkN "númer" neuter ; +lin núna_Adv = mkAdv "núna" ; +lin nýr_A = mkA "nýr" "ný" ; +lin olía_N = mkN "olía" feminine ; +lin opna_V = (mkV "opna" "opna" "opnaði" "opnaður") ; +lin ormur_N = mkN "ormur" masculine ; +lin ostur_N = mkN "ostur" masculine ; +lin other_A = mkA "other" ; +lin pappír_N = mkN "pappír" masculine ; +lin penni_N = mkN "penni" masculine ; +lin persóna_N = mkN "persóna" feminine ; +lin plat_N = mkN "plat" neuter ; +lin pláneta_N = mkN "pláneta" feminine ; +lin prestur_N = mkN "prestur" masculine ; +lin prik_N = mkN "prik" neuter ; +lin rauður_A = mkA "rauður" ; +lin regla_N = mkN "regla" feminine ; +lin reipi_N = mkN "reipi" neuter ; +lin reykur_N = mkN "reykur" "reykjar" "reykir" masculine; +lin rigning_N = mkN "rigning" feminine ; +lin rotinn_A = mkA "rotinn" ; +lin ryk_N = mkN "ryk" neuter ; +lin réttur_A = mkA "réttur" ; +lin rófa_N = mkN "rófa" feminine ; +lin rót_N = mkN "rót" "rótar" "rætur" feminine ; +lin salt_N = mkN "salt" neuter ; +lin sandur_N = mkN "sandur" masculine ; +lin sauma_V = mkV "sauma" "sauma" "saumaði" "saumaður" ; +lin segja_V = mkV "segja" "segi" "sagði" "sagður"; +lin selja_V = (mkV "selja") ; +lin senda_V = (mkV "senda") ; +lin silfur_N = mkN "silfur" neuter ; +lin sitja_V = I.sitja_V ; +lin sjá_V = I.sjá_V ; +lin sjónvarp_N = mkN "sjónvarp" neuter ; +lin sjór_N = mkN "sjór" "sjó" "sjó" "sjós" "sjóir" "sjói" "sjóum" "sjóa" masculine ; +lin sjúga_V = I.sjúga_V ; +lin skemmtilegur_A = mkA "skemmtilegur" ; +lin skilja_V = (mkV "skilja") ; +lin skip_N = mkN "skip" neuter ; +lin skipta_V = (mkV "skipta") ; +lin skrifa_V = (mkV "skrifa" "skrifa" "skrifaði" "skrifaður") ; +lin skyrta_N = mkN "skyrta" feminine ; +lin skítugur_A = mkA "skítugur" ; +lin skógur_N = mkN "skógur" masculine ; +lin skóli_N = mkN "skóli" masculine ; +lin skór_N = mkN "skór" "skór" masculine ; +lin ský_N = mkN "ský" neuter ; +lin sléttur_A = mkA "sléttur" ; +lin smjör_N = mkN "smjör" neuter ; +lin snjór_N = mkN "snjór" masculine ; +lin snákur_N = mkN "snákur" masculine ; +lin sofa_V = I.sofa_V ; +lin sokkur_N = mkN "sokkur" masculine ; +lin spurning_N = mkN "spurning" feminine ; +lin spyrja_V = (mkV "spyrja" "spyr" "spurði" "spurður") ; +lin standa_V = I.standa_V ; +lin stansa_V = mkV "stansa" "stansa" "stansaði" "stansaður" ; +lin steinn_N = mkN "steinn" masculine ; +lin stelpa_N = mkNPlGen "stelpa" feminine ; +lin stinga_V = I.stinga_V ; +lin stjarna_N = mkN "stjarna" feminine ; +lin stjóri_N = mkN "stjóri" masculine ; +lin strjúka_V = I.strjúka_V ; +lin strákur_N = mkN "strákur" masculine ; +lin stríð_N = mkN "stríð" neuter ; +lin stuttur_A = mkA "stuttur" "stutt" "syttri" ; +lin stál_N = mkN "stál" neuter ; +lin stóll_N = mkN "stóll" masculine ; +lin stór_A = mkA "stór" "stór" "stærri" ; +lin svara_V = (mkV "svara" "svara" "svaraði" "svaraður") ; +lin svartur_A = mkA "svartur" ; +lin synda_V = mkV "synda" ; +lin syngja_V = I.syngja_V ; +lin systir_N = mkN "systir" "systur" feminine ; +lin sígvél_N = mkN "sígvél" neuter ; +lin sól_N = mkN "sól" "sólir" feminine ; +lin tala_V = (mkV "tala" "tala" "talaði" "talaður") ; +lin telja_V = (mkV "telja") ; +lin teppi_N = mkN "teppi" neuter ; +lin tilbúinn_A = mkA "tilbúinn" ; +lin toga_V = (mkV "toga" "toga" "togaði" "togaður") ; +lin tré_N = mkN "tré" neuter ; +lin trúarbragð_N = mkN "trúarbragð" neuter ; +lin tunga_N = mkNPlGen "tunga" feminine ; +lin tungl_N = mkN "tungl" neuter ; +lin tungumál_N = mkN "tungumál" neuter ; +lin tómur_A = mkA "tómur" ; +lin tónlist_N = mkN "tónlist" "tónlist" "tónlist" "tónlistar" "tónlist" "tónlist" "tónlist" "tónlistar" feminine ; +lin tölva_N = mkN "tölva" feminine ; +lin tönn_N = mkN "tönn" "tannar" "tennur" feminine ; +lin týna_V = (mkV "týna") ; +lin ungur_A = mkA "ungur" "ung" "yngri" ; +lin vatn_N = mkN "vatn" neuter ; +lin vegalend_N = mkN "vegalend" "vegalengdir" feminine ; +lin vegur_N = mkN "vegur" "vegar" "vegir" masculine ; +lin veitingastaður_N = mkCompoundN "veitinga" (mkN "staður" "staðar" "staðir" masculine) ; +lin veiða_V = (mkV "veiða") ; +lin velta_V = mkV "velta" ; +lin verksmiðja_N = mkN "verksmiðja" feminine ; +lin verða_V = I.verða_V ; +lin vindur_N = mkN "vindur" masculine ; +lin vinna_V = I.vinna_V ; +lin vinur_N = mkN "vinur" "vinar" "vinir" masculine ; +lin vita_V = I.vita_V ; +lin viður_N = mkN "viður" "viðar" "viðir" masculine ; +lin vona_V = mkV "vona" "vona" "vonaði" "vonaður" ; +lin vondur_A = mkA "vondur" "vond" "verri" ; +lin vængur_N = mkN "vængur" masculine ; +lin vín_N = mkN "vín" neuter ; +lin víður_A = mkA "víður" ; +lin á_N = mkN "á" "ár" feminine ; +lin ár_N = mkN "ár" neuter ; +lin ást_N = mkN "ást" "ástir" feminine ; +lin ástæða_N = mkNPlGen "ástæða" feminine ; +lin ávöxtur_N = mkN "ávöxtur" "ávaxtar" "ávextir" masculine ; +lin æla_V = mkV "æla" ; +lin í_Adv = mkAdv "í dag" ; +lin íbúð_N = mkN "íbúð" "íbúðir" feminine ; +lin ísskápur_N = mkCompoundN "ís" (mkN "skápur" "skáps" "skápar" masculine) ; +lin ís_N = mkN "ís" masculine ; +lin óvinur_N = mkCompoundN "ó" (vinur_N) ; +lin óviss_A = mkA "óviss" "óviss" "óvissari" ; +lin öl_N = mkN "öl" neuter ; +lin útvarp_N = mkN "útvarp" neuter ; +lin ýta_V = (mkV "ýta") ; +lin þak_N = mkN "þak" neuter ; +lin þefa_V = mkV "þefa" "þefa" "þefaði" "þefaður" ; +lin þegar_Adv = mkAdv "þegar" ; +lin þoka_N = mkN "þoka" feminine ; +lin þorp_N = mkN "þorp" neuter ; +lin þrífa_V = I.þrífa_V ; +lin þröngur_A = mkA "þröngur" ; +lin þungur_A = mkA "þungur" "þung" "þyngri" ; +lin þunnur_A = mkA "þunnur" "þunn" "þynnri" ; +lin þurr_A = mkA "þurr" "þurr" ; +lin þykkur_A = mkA "þykkur" ; +} \ No newline at end of file diff --git a/src/icelandic/DictIceAbs.gf b/src/icelandic/DictIceAbs.gf new file mode 100644 index 000000000..4295f937c --- /dev/null +++ b/src/icelandic/DictIceAbs.gf @@ -0,0 +1,334 @@ +abstract DictIceAbs = Cat ** { + +-- monolingual dictionary bootstrapped from LexIce +-- AR 2019-08-07 + +fun anda_V : V ; +fun aska_N : N ; +fun auga_N : N ; +fun auðveldur_A : A ; +fun bak_N : N ; +fun banki_N : N ; +fun barn_N : N ; +fun bein_N : N ; +fun beinn_A : A ; +fun beittur_A : A ; +fun belja_N : N ; +fun berja_V : V ; +fun beygja_V : V ; +fun binda_V : V ; +fun biðja_V : V ; +fun blautur_A : A ; +fun blár_A : A ; +fun blása_V : V ; +fun blóm_N : N ; +fun blóð_N : N ; +fun borg_N : N ; +fun borð_N : N ; +fun borða_V : V ; +fun brauð_N : N ; +fun breiður_A : A ; +fun brenna_V : V ; +fun brjóst_N : N ; +fun brjóta_V : V ; +fun bróðir_N : N ; +fun brúnn_A : A ; +fun bátur_N : N ; +fun bæta_V : V ; +fun bíll_N : N ; +fun bíta_V : V ; +fun bók_N : N ; +fun bólgna_V : V ; +fun börkur_N : N ; +fun búð_N : N ; +fun dagblað_N : N ; +fun dagur_N : N ; +fun detta_V : V ; +fun deyja_V : V ; +fun drekka_V : V ; +fun drepa_V : V ; +fun drottning_N : N ; +fun dýr_N : N ; +fun eftir_V : V ; +fun egg_N : N ; +fun eiginkona_N : N ; +fun eiginmaður_N : N ; +fun eldavél_N : N ; +fun eldur_N : N ; +fun elska_V : V ; +fun epli_N : N ; +fun eyra_N : N ; +fun fallegur_A : A ; +fun fara_V : V ; +fun faðir_N : N ; +fun ferða_V : V ; +fun finna_V : V ; +fun fiskur_N : N ; +fun fita_N : N ; +fun fjall_N : N ; +fun fjöður_N : N ; +fun fljóta_V : V ; +fun fljúga_V : V ; +fun flugvél_N : N ; +fun flæða_V : V ; +fun friður_N : N ; +fun frjósa_V : V ; +fun fræ_N : N ; +fun frændi_N : N ; +fun fugl_N : N ; +fun fullur_A : A ; +fun fótleggur_N : N ; +fun fótur_N : N ; +fun gamall_A : A ; +fun ganga_V : V ; +fun garður_N : N ; +fun gefa_V : V ; +fun gera_V : V ; +fun giftur_A : A ; +fun gleyma_V : V ; +fun gluggi_N : N ; +fun grafa_V : V ; +fun gras_N : N ; +fun grænn_A : A ; +fun gull_N : N ; +fun gulur_A : A ; +fun gólf_N : N ; +fun góður_A : A ; +fun görn_N : N ; +fun gúmmí_N : N ; +fun halda_V : V ; +fun hanski_N : N ; +fun hata_V : V ; +fun hattur_N : N ; +fun haus_N : N ; +fun heimskur_A : A ; +fun heitur_A : A ; +fun hestur_N : N ; +fun heyra_V : V ; +fun himinn_N : N ; +fun hjarta_N : N ; +fun hjól_N : N ; +fun hlaupa_V : V ; +fun hlusta_V : V ; +fun hlæja_V : V ; +fun hné_N : N ; +fun hoppa_V : V ; +fun horfa_V : V ; +fun horn_N : N ; +fun hreinn_A : A ; +fun hrækja_V : V ; +fun hugsa_V : V ; +fun hundur_N : N ; +fun hurð_N : N ; +fun hvítur_A : A ; +fun háls_N : N ; +fun hár_N : N ; +fun háskóli_N : N ; +fun hæð_N : N ; +fun höfn_N : N ; +fun hönd_N : N ; +fun húfa_N : N ; +fun hús_N : N ; +fun húð_N : N ; +fun iðnaður_N : N ; +fun járn_N : N ; +fun jörð_N : N ; +fun kaldur_A : A ; +fun kasta_V : V ; +fun kaupa_V : V ; +fun kenna_V : V ; +fun kennari_N : N ; +fun kind_N : N ; +fun kirkja_N : N ; +fun kjöt_N : N ; +fun klettur_N : N ; +fun klippa_V : V ; +fun klár_A : A ; +fun klóra_V : V ; +fun koma_V : V ; +fun kona_N : N ; +fun konungur_N : N ; +fun kreista_V : V ; +fun kringlóttur_A : A ; +fun kápa_N : N ; +fun köttur_N : N ; +fun lag_N : N ; +fun lampi_N : N ; +fun land_N : N ; +fun langt_Adv : Adv ; +fun langur_A : A ; +fun lauf_N : N ; +fun leika_V : V ; +fun leita_V : V ; +fun lesa_V : V ; +fun lest_N : N ; +fun leðinilegur_A : A ; +fun leður_N : N ; +fun lifa_V : V ; +fun lifur_N : N ; +fun list_N : N ; +fun ljótur_A : A ; +fun ljúga_V : V ; +fun loft_N : N ; +fun loka_V : V ; +fun láta_V : V ; +fun læknir_N : N ; +fun læra_V : V ; +fun líklegur_A : A ; +fun lítill_A : A ; +fun lögregluþjónn_N : N ; +fun lús_N : N ; +fun magi_N : N ; +fun maður_N : N ; +fun mikilvægur_A : A ; +fun mjólk_N : N ; +fun munnur_N : N ; +fun myndavél_N : N ; +fun mála_V : V ; +fun málfræði_N : N ; +fun móðir_N : N ; +fun nafn_N : N ; +fun nef_N : N ; +fun nemandi_N : N ; +fun nudda_V : V ; +fun nálægur_A : A ; +fun nótt_N : N ; +fun nögl_N : N ; +fun númer_N : N ; +fun núna_Adv : Adv ; +fun nýr_A : A ; +fun olía_N : N ; +fun opna_V : V ; +fun ormur_N : N ; +fun ostur_N : N ; +fun other_A : A ; +fun pappír_N : N ; +fun penni_N : N ; +fun persóna_N : N ; +fun plat_N : N ; +fun pláneta_N : N ; +fun prestur_N : N ; +fun prik_N : N ; +fun rauður_A : A ; +fun regla_N : N ; +fun reipi_N : N ; +fun reykur_N : N ; +fun rigning_N : N ; +fun rotinn_A : A ; +fun ryk_N : N ; +fun réttur_A : A ; +fun rófa_N : N ; +fun rót_N : N ; +fun salt_N : N ; +fun sandur_N : N ; +fun sauma_V : V ; +fun segja_V : V ; +fun selja_V : V ; +fun senda_V : V ; +fun silfur_N : N ; +fun sitja_V : V ; +fun sjá_V : V ; +fun sjónvarp_N : N ; +fun sjór_N : N ; +fun sjúga_V : V ; +fun skemmtilegur_A : A ; +fun skilja_V : V ; +fun skip_N : N ; +fun skipta_V : V ; +fun skrifa_V : V ; +fun skyrta_N : N ; +fun skítugur_A : A ; +fun skógur_N : N ; +fun skóli_N : N ; +fun skór_N : N ; +fun ský_N : N ; +fun sléttur_A : A ; +fun smjör_N : N ; +fun snjór_N : N ; +fun snákur_N : N ; +fun sofa_V : V ; +fun sokkur_N : N ; +fun spurning_N : N ; +fun spyrja_V : V ; +fun standa_V : V ; +fun stansa_V : V ; +fun steinn_N : N ; +fun stelpa_N : N ; +fun stinga_V : V ; +fun stjarna_N : N ; +fun stjóri_N : N ; +fun strjúka_V : V ; +fun strákur_N : N ; +fun stríð_N : N ; +fun stuttur_A : A ; +fun stál_N : N ; +fun stóll_N : N ; +fun stór_A : A ; +fun svara_V : V ; +fun svartur_A : A ; +fun synda_V : V ; +fun syngja_V : V ; +fun systir_N : N ; +fun sígvél_N : N ; +fun sól_N : N ; +fun tala_V : V ; +fun telja_V : V ; +fun teppi_N : N ; +fun tilbúinn_A : A ; +fun toga_V : V ; +fun tré_N : N ; +fun trúarbragð_N : N ; +fun tunga_N : N ; +fun tungl_N : N ; +fun tungumál_N : N ; +fun tómur_A : A ; +fun tónlist_N : N ; +fun tölva_N : N ; +fun tönn_N : N ; +fun týna_V : V ; +fun ungur_A : A ; +fun vatn_N : N ; +fun vegalend_N : N ; +fun vegur_N : N ; +fun veitingastaður_N : N ; +fun veiða_V : V ; +fun velta_V : V ; +fun verksmiðja_N : N ; +fun verða_V : V ; +fun vindur_N : N ; +fun vinna_V : V ; +fun vinur_N : N ; +fun vita_V : V ; +fun viður_N : N ; +fun vona_V : V ; +fun vondur_A : A ; +fun vængur_N : N ; +fun vín_N : N ; +fun víður_A : A ; +fun á_N : N ; +fun ár_N : N ; +fun ást_N : N ; +fun ástæða_N : N ; +fun ávöxtur_N : N ; +fun æla_V : V ; +fun í_Adv : Adv ; +fun íbúð_N : N ; +fun ísskápur_N : N ; +fun ís_N : N ; +fun óvinur_N : N ; +fun óviss_A : A ; +fun öl_N : N ; +fun útvarp_N : N ; +fun ýta_V : V ; +fun þak_N : N ; +fun þefa_V : V ; +fun þegar_Adv : Adv ; +fun þoka_N : N ; +fun þorp_N : N ; +fun þrífa_V : V ; +fun þröngur_A : A ; +fun þungur_A : A ; +fun þunnur_A : A ; +fun þurr_A : A ; +fun þykkur_A : A ; +} \ No newline at end of file diff --git a/src/icelandic/wiktionary/CheckWithWiktionary.hs b/src/icelandic/wiktionary/CheckWithWiktionary.hs index 57687dc1f..81cf80d48 100644 --- a/src/icelandic/wiktionary/CheckWithWiktionary.hs +++ b/src/icelandic/wiktionary/CheckWithWiktionary.hs @@ -1,6 +1,7 @@ import qualified Data.Map -import qualified Data.Char -import qualified Data.List +import qualified Data.Text.IO +import Data.Char +import Data.List -- AR 2019-08-06 -- checking IrregIce wrt Wikipedia: @@ -135,3 +136,62 @@ getGroups ls = let (v,vs) = splitAt 105 ls in if null v then [] else v:getGroups actInf v@(i:_) = last (words i) +jumpToIcelandic ls = dropWhile (\l -> not (isPrefixOf "

" l && isPrefixOf "Icelandic" (untag l))) ls + +------------------------------- +-- just retrieving ------------ + +-- to be run in verbs/ +getAllWiktVerbs = do + vs <- readFile "wikt-verbs.txt" >>= return . lines + writeFile "v.tmp" "" + mapM_ (\v -> getWiktVerb v >>= appendFile "v.tmp" . unlines . emitGF) vs + +-- to be run in adjectives/ +getAllWiktAdjectives = do + vs <- readFile "wikt-verbs.txt" >>= return . lines + mapM_ (\v -> getWiktAdjective v >>= putStrLn . unlines . emitGF) vs + + +-- return ([relevant Wikt lines], (fun,cat,lin), message) +getWiktWord :: Int -> (String -> [String] -> ([String],((String,String,String),Message))) -> FilePath -> IO ([String],((String,String,String),Message)) +getWiktWord number check file = do + s <- readFile file >>= return . map untag . take number . getTD . jumpToIcelandic . lines + return $ check file s + +getWiktNoun = getWiktWord 17 checkNoun +getWiktAdjective = getWiktWord 120 noCheck +getWiktVerb = getWiktWord 75 checkVerb + +noCheck :: String -> [String] -> ([String],((String,String,String),Message)) +noCheck s ss = (ss, (noGF, MMissing s)) +noGF = ("--","--","--") + +checkNoun noun forms = noCheck noun forms ---- + +checkVerb verb forms = + if length forms < 75 + then (forms, (noGF, MBad (verb ++ ": only " ++ show (length forms) ++ " lines"))) + else case unexpectedWikLines forms of + [] -> (forms, ((verb ++ "_V", "V", app "mkV" [verb, forms!!5, forms!!18, forms!!74, forms!!1]), MGood verb)) ---- + us -> (forms, (noGF, MBad (verb ++ ": unexpected lines " ++ show (length us)))) + +data Message = + MGood String + | MBad String + | MMissing String + deriving (Show,Eq) + +app f xs = unwords $ f : map (quote . wform . words) xs +quote s = "\"" ++ s ++ "\"" + +emitGF (ss,((fun,cat,lin),msg)) = case msg of + MGood _ -> [unwords ["fun",fun,":",cat,";"],unwords ["lin",fun,"=",lin,";"]] + _ -> ["-- " ++ show msg] + + +-- mkN : (x1,_,_,_,_,_,_,x8 : Str) -> Gender -> N = mk8N ; nForms8 a b c d e f g h ; sgNom,sgAcc,sgDat,sgGen,plNom,plAcc,plDat,plGen +-- mkA : (_,_,_ : Str) -> A = mk3A ; mk3A : (_,_,_ : Str) -> A = \mas,fem,com +-- mkV : (_,_,_,_,_ : Str) -> V = \telja,tel,taldi,talinn,talið -> -- inf,presIndSg1,pastIndSg1, weak past part, sup + +