Skip to content

Commit

Permalink
ddt
Browse files Browse the repository at this point in the history
  • Loading branch information
@Trondtr
Trondtr committed Jun 18, 2024
1 parent d1dbe93 commit 7c4670e
Showing 1 changed file with 334 additions and 0 deletions.
334 changes: 334 additions & 0 deletions scripts/dict_smefin.dtd
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,334 @@
<!ELEMENT r (lics?, e+, xhtml:script?) >
<!ATTLIST r xmlns:xhtml CDATA #FIXED "http://www.w3.org/1999/xhtml"
xml:lang ( sme ) #IMPLIED
>
<!ATTLIST r
id ( smefin ) #IMPLIED >


<!-- the header -->
<!ELEMENT lics (lic, ref*, sourcenote*) >
<!ATTLIST lics xml:space (default|preserve) 'preserve'
xmlns:xhtml CDATA #FIXED "http://www.w3.org/1999/xhtml"
>
<!ELEMENT lic (#PCDATA|a)*>
<!ATTLIST lic xml:lang ( en | nno ) #IMPLIED >
<!ELEMENT ref (#PCDATA|a|i)*>
<!ELEMENT sourcenote (#PCDATA|a)*>
<!ELEMENT a (#PCDATA)>
<!ELEMENT i (#PCDATA)>

<!ELEMENT e (lg, source?, semantics?, stem*, sources?, mg+) >
<!ATTLIST e
usage ( mt | vd | dict | other | nds | fad | KB ) #IMPLIED
src ( nj | sk | SvSt | gt_fad | gt | fad | other | gg | SK |
PS_SA | KAL | nou94 | PS | husbykuhmunen | lmm | sammallahti |
BergslandMagga | Lagercrantz | nsr ) #IMPLIED
exclude ( smanob | nobsma ) #IMPLIED
reverse ( yes | no ) #IMPLIED
illpl ( yes | no ) #IMPLIED
id CDATA #IMPLIED
xpos CDATA #IMPLIED
mc CDATA #IMPLIED
>
<!ATTLIST e r ( LR | RL ) #IMPLIED > <!-- found in all_smnsme.xml -->
<!ATTLIST e note CDATA #IMPLIED >
<!ATTLIST e oa_unif ( done | s | N | E ) #IMPLIED >

<!ELEMENT lg (l+, l_ref*, lemma_ref*, lsub*, lc*, stem*, analysis*, mini_paradigm*, algu?) >
<!ELEMENT lemma_ref (#PCDATA) >
<!ATTLIST lemma_ref lemmaID CDATA #IMPLIED >
<!ATTLIST lg freq CDATA #IMPLIED >
<!ELEMENT analysis (#PCDATA|wordform)* >
<!ATTLIST analysis ms CDATA #IMPLIED >
<!ELEMENT wordform (#PCDATA) >
<!ELEMENT mini_paradigm (analysis+) >
<!ELEMENT l (#PCDATA) >
<!ATTLIST l
g_stem ( Sg | Pl ) #IMPLIED
sg_gen CDATA #IMPLIED
sg_ill CDATA #IMPLIED
pl_gen CDATA #IMPLIED
pl_ill CDATA #IMPLIED
pl3 CDATA #IMPLIED
sg1 CDATA #IMPLIED
sg3 CDATA #IMPLIED
gen CDATA #IMPLIED
ill CDATA #IMPLIED
ader CDATA #IMPLIED
v_type ( rv ) #IMPLIED
pos ( A | N | V | Adv | Po | Pr | Pron | Rel | Pers | Det | I | Interj | Pcle |
Num | CC | CS | Prop | Proppl | G3 | Actor | Npl | Adp |
Prsprc | Phrase | Fp-Cmp | Lp-Cmp | Cmp | mwe | X | Xxx | Qu | Abbr | Clt |
Phrase_N | Phrase_A | Phrase_V | Phrase_Adv ) #REQUIRED
attr CDATA #IMPLIED
decl ( 1 | 2 | 3 | 4 | 5 | x | i ) #IMPLIED
nr ( Sg | Du | Pl ) #IMPLIED
gen_only CDATA #IMPLIED
context CDATA #IMPLIED
type ( Rel | Pers | Indef | Refl | Dem | interr |
pre | post | ord | TV | IV | actor | G3 | G7 | NomAg | Ord
| Prop | Recipr | Coll | Interr | Logo ) #IMPLIED
spec ( ABBR | ACR | Attr | Cmpar.Comp | Cmp_SgNom.Cmpnd | Cmp_Sh.Cmpnd | Coll | Dem | G3 | Ind | Indef | Indef.Attr | Indef.Pl.Nom | Indef.Sg.Loc.Attr | Interr | IV | IV.VAbess | Logo.3p.Du | Logo.3p.Pl | Logo.3p.Sg | NomAg | Ord | Pers.1p.Du | Pers.1p.Pl | Pers.1p.Sg | Pers.2p.Du | Pers.2p.Pl | Pers.2p.pl | Pers.2p.Sg | Pers.3p.Du | Pers.3p.Pl | Pers.3p.Sg | Pl | PrfPrc | PrsPrc | Qst | Refl | Rel | Sem_Act | Sem_Ani | Sem_Ani_Group.Sg | Sem_Body.Pl | Sem_Build | Sem_Clth.Pl | Sem_Dummytag | Sem_Dummytag.Attr | Sem_Dummytag.Pl | Sem_Dummytag.Sg | Sem_Dummytag.Sg.Nom | Sem_Event.Pl | Sem_Feat-psych.Sg | Sem_Hum.Attr | Sem_Hum_Plc | Sem_Money.Sg | Sem_Org | Sem_Org.Sg | Sem_Plc | Sem_Plc.Pl | Sem_Process.Sg.Nom | Sem_Prod-cogn.Sg | Sem_Substnc.Pl | Sem_Substnc.Sg | Sem_Time | Sem_Time.Pl | Sem_Tool-catch.Sg | Sg | Sg.Ela | Sg.Gen | Sg.Ill | Sg.Ine | Sg.Nom | Superl | Superl.Attr | Superl.Sg.Nom | TV | TV.Der_PassL.V | TV.Der_PassL.V.IV | .V.IV ) #IMPLIED
class ( I | II | III | IV | V | VI | xxx ) #IMPLIED
stem ( even | odd | 2syll | 3syll | xsyll | weak ) #IMPLIED
pg ( yes | no ) #IMPLIED
p3p CDATA #IMPLIED
minip ( notSg1 | onlyPl ) #IMPLIED
illpl ( ok | no ) #IMPLIED
status ( sjekk | ok | _unknown_N ) #IMPLIED
grad ( pos | comp | sup ) #IMPLIED
vmax ( 2 | 3 | 4) #IMPLIED
base CDATA #IMPLIED
dialect CDATA #IMPLIED
soggi CDATA #IMPLIED
rime CDATA #IMPLIED
margo CDATA #IMPLIED
vow CDATA #IMPLIED
tt_auto CDATA #IMPLIED
diph ( yes ) #IMPLIED
umlaut ( full | A | C | B | D | E | F ) #IMPLIED
num ( pl ) #IMPLIED
val ( IV | TV ) #IMPLIED
r1_par CDATA #IMPLIED
r2_par CDATA #IMPLIED
par CDATA #IMPLIED
dera CDATA #IMPLIED

>
<!ATTLIST l mwe CDATA #IMPLIED >
<!ATTLIST l comma CDATA #IMPLIED >
<!ATTLIST l syn CDATA #IMPLIED >
<!ATTLIST l sem_type ( Obj | Plc | Fem | Org | Mal | Txt ) #IMPLIED >
<!ATTLIST l paradigme ( cases ) #IMPLIED >
<!ATTLIST l comment CDATA #IMPLIED >
<!ATTLIST l case CDATA #IMPLIED >
<!ATTLIST l mod CDATA #IMPLIED >
<!ATTLIST l hid CDATA #IMPLIED >
<!ATTLIST l src CDATA #IMPLIED >
<!ATTLIST l value CDATA #IMPLIED >
<!ATTLIST l t_type ( Pers | expl ) #IMPLIED >
<!ATTLIST l re CDATA #IMPLIED >
<!ATTLIST l tt CDATA #IMPLIED >
<!ATTLIST l syn_or CDATA #IMPLIED >

<!-- decl: i indeclineable -->

<!-- lsub = frequent subform of lemma -->
<!ELEMENT lsub (#PCDATA) >
<!ATTLIST lsub
attr ( ok | no ) #IMPLIED
decl ( 1 | 2 | 3 | 4 | 5 | x | i ) #IMPLIED
nr ( sg | pl ) #IMPLIED
context CDATA #IMPLIED
type ( rel | pers | indef | refl | pre | post | ord ) #IMPLIED
class ( I | II | III | IV | V | VI ) #IMPLIED
stem ( even | odd | 2syll | 3syll | xsyll | weak ) #IMPLIED
p3p CDATA #IMPLIED
minip ( notSg1 | onlyPl ) #IMPLIED
illpl ( ok | no ) #IMPLIED
status ( sjekk | ok ) #IMPLIED
grad ( pos | comp | sup ) #IMPLIED
base CDATA #IMPLIED
dialect CDATA #IMPLIED
>
<!ATTLIST lsub src CDATA #IMPLIED >
<!ATTLIST lsub x ( fad ) #IMPLIED >

<!-- lc = lemma comment, misused as stem field -->
<!ELEMENT lc (#PCDATA) >
<!ELEMENT source (book) >
<!ATTLIST book
name CDATA #IMPLIED >
<!ELEMENT semantics (sem | sem* ) >
<!ATTLIST sem
name CDATA #IMPLIED >

<!-- algu = lemma IDs in the algu database -->
<!ELEMENT algu EMPTY >
<!ATTLIST algu
lekseemi_id CDATA #REQUIRED >
<!ATTLIST algu
sanue_id CDATA #REQUIRED >


<!-- the real stem field -->
<!ELEMENT stem (#PCDATA) >
<!ATTLIST stem
class ( bisyllabic | trisyllabic | contracted ) #IMPLIED
>

<!ELEMENT sources (book*, frequency?, geography*) >
<!ELEMENT book EMPTY >
<!-- name ( a1 | dej | s1 | s2 | s3 | s4 | åa1 | åa2 | åa3 | åa4 | åa5 | åa6 | xxx ) #IMPLIED -->

<!ELEMENT frequency EMPTY >
<!ATTLIST frequency
class ( common | rare ) #IMPLIED
>
<!ELEMENT geography EMPTY >
<!ATTLIST geography
class ( other | south | mid | north ) #IMPLIED
>
<!-- name CDATA #REQUIRED -->

<!ELEMENT sem EMPTY >
<!ATTLIST sem
class CDATA #IMPLIED
>

<!ELEMENT mg (semantics?, l_ref?, tg+, xg* ) >
<!ATTLIST mg
xml:lang ( sme | nob | fin | swe | rus | smn ) #IMPLIED
rest CDATA #IMPLIED
>
<!ATTLIST mg id CDATA #IMPLIED >
<!ATTLIST mg src CDATA #IMPLIED >
<!ATTLIST mg x ( fad ) #IMPLIED >

<!ELEMENT tg (semantics?, re?, ( ((t, stem*, l_ref?, morph_expl?) | tf | te ))+, xg*, syng?) >
<!ATTLIST tg
xml:lang ( fin ) #REQUIRED
check CDATA #IMPLIED
>
<!ELEMENT re (#PCDATA) > <!-- Domain restriction: bot, phys, ... -->
<!ATTLIST re
xml:lang ( fin | sme | lat) #IMPLIED
>
<!ATTLIST re x ( fad ) #IMPLIED >
<!ATTLIST re comment CDATA #IMPLIED >

<!ELEMENT t (#PCDATA) >
<!ATTLIST t freq CDATA #IMPLIED >
<!ATTLIST t oa ( first | yes | no ) #IMPLIED >
<!ATTLIST t dict ( yes | no ) #IMPLIED >
<!ATTLIST t t_type ( Pers | expl ) #IMPLIED >
<!ATTLIST t nr ( Sg | Pl ) #IMPLIED >
<!ATTLIST t pos ( A | S | N | V | Adp | Clt |
Adv | P | Pr | Po | Det |
Pron | Prop | I | Interj | CC | CS | Im | Npl | Num |
Pcle | Prsprc | Phrase | Fp-Cmp | Lp-Cmp | Cmp | X | Xxx |
phrase_N | phrase_A | phrase_V | phrase_Adv | g3 | NomAg | mwe | fp-cmp
| Abbr | Foc_ba
| Adv-N | N-Pron | Adv-CC | Adv-Pron | Adv-CS | CC-CS | A-N | Adv-Po | CMP | A-Adv | Adv-Po-Pron | A-Adv-Pron | A-Adv-N | Adv-N-Po | N-Po | Po-Pron | A-Pron | Foc_gis | Foc_han ) #IMPLIED >
<!ATTLIST t type ( expl
| grammarexpl |Prop | Indef | G3 | NomAg | Coll | Recipr | Pers
| grammar | Refl | Dem | Poss | Logo | Rel | Interr | TV | IV ) #IMPLIED >
<!ATTLIST t xml:lang ( nob | sme | sma | swe | eng | deu ) #IMPLIED >
<!ATTLIST t alt_str CDATA #IMPLIED >
<!ATTLIST t attr CDATA #IMPLIED >
<!ATTLIST t case CDATA #IMPLIED >
<!ATTLIST t comment CDATA #IMPLIED >
<!ATTLIST t context CDATA #IMPLIED >
<!ATTLIST t diph CDATA #IMPLIED >
<!ATTLIST t gen_only CDATA #IMPLIED > <!-- Fetched from l -->
<!ATTLIST t hid CDATA #IMPLIED >
<!ATTLIST t href CDATA #IMPLIED >
<!ATTLIST t illpl CDATA #IMPLIED >
<!ATTLIST t l_par CDATA #IMPLIED >
<!ATTLIST t margo CDATA #IMPLIED >
<!ATTLIST t minip CDATA #IMPLIED >
<!ATTLIST t mod CDATA #IMPLIED >
<!ATTLIST t mwe CDATA #IMPLIED >
<!ATTLIST t num CDATA #IMPLIED >
<!ATTLIST t r_par CDATA #IMPLIED >
<!ATTLIST t rest CDATA #IMPLIED >
<!ATTLIST t sem-cl CDATA #IMPLIED >
<!ATTLIST t sem_type ( Obj | Plc | Org | Mal ) #IMPLIED >
<!ATTLIST t soggi CDATA #IMPLIED >
<!ATTLIST t spec ( .A.Attr | ABBR | ACR | Attr | Cmpar | Cmpar.Comp | .Cmpnd | Cmp_SgNom.Cmpnd | Cmp_Sh.Cmpnd | Coll | Dem | Dem.Sg.Ela | Der_lasj.A | Der_NomAg.Sem_Hum.Pl | Der_t.A | G3 | Indef | Indef.Attr | Indef.Attr.Foc_Neg-ge | Indef.Pl.Acc | Indef.Pl.Nom | Indef.Sg.Acc.Foc_Neg-ge | Indef.Sg.Nom | Indef.Sg.Nom.Foc_Neg-ge | Inf | Interr | Interr.Pl.Acc | Interr.Sg.Acc. | Interr.Sg.Acc | Interr.Sg.Nom | IV | IV.Der_NomAct.N | IV.Inf | IV..V | IV.VAbess | Logo.3p.Pl | Logo.3p.Sg | NomAg | Ord | Pers.1p.Du | Pers.1p.Pl | Pers.1p.Sg | Pers.2p.Du | Pers.2p.Pl | Pers.2p.Sg | Pers.3p.Du | Pers.3p.Pl | Pers.3p.Sg | Pl | Pl.Gen | Pl.Nom | PrfPrc | PrsPrc | Qst | Recipr | Refl | Rel | Rel.Pl | Rel.Pl.Acc | Rel.Sg | Rel.Sg.Acc. | Rel.Sg.Acc | Sem_Act | Sem_Amount.Sg.Ill | Sem_Ani | Sem_Build | Sem_Clth.Pl | Sem_Dummytag | Sem_Dummytag.Attr | Sem_Dummytag.Cmpar.Comp | Sem_Dummytag.Pl | Sem_Event.Pl | Sem_Feat.Sg.Loc | Sem_Hum.Attr | Sem_Hum.Cmpar.Comp | Sem_Hum.Sg.Nom | Sem_Org | Sem_Plc | Sem_Plc.Foc_Neg-ge | Sem_Process.Sg.Nom | Sem_Time | Sem_Time.Pl | Sem_Time.Sg.Ill | Sg | Sg.Ela | Sg.Gen | Sg.Ill | Sg.Ine | Sg.Nom | Superl | Superl.Attr | Superl.Sg.Nom | TV | TV.Der_Caus.V.TV.Der_PassL.V.IV | TV.Der_PassL.V | TV.Der_PassL.V.IV | .V.IV ) #IMPLIED > <!-- Is this perhaps the same as type in l? -->
<!ATTLIST t src CDATA #IMPLIED >
<!ATTLIST t stat ( pref | notpref | perhaps ) #IMPLIED >
<!ATTLIST t stem CDATA #IMPLIED >
<!ATTLIST t syn CDATA #IMPLIED >
<!ATTLIST t syn_dash CDATA #IMPLIED >
<!ATTLIST t t_tld CDATA #IMPLIED >
<!ATTLIST t tcomm ( yes | no ) #IMPLIED >
<!ATTLIST t umlaut ( full | A | C | B | D | E | F ) #IMPLIED >
<!ATTLIST t value CDATA #IMPLIED >
<!ATTLIST t var ( sw | no | fi ) #IMPLIED >
<!ATTLIST t vow CDATA #IMPLIED >
<!ATTLIST t wf CDATA #IMPLIED >

<!-- perhaps not first for attlist oa, if reverse-leksa shall be taken from nobsma -->

<!ELEMENT l_ref (#PCDATA) >

<!-- translation_comment = for Oahpa use -->

<!ELEMENT tf (#PCDATA) > <!-- Translation phrase -->
<!ATTLIST tf oa ( first | yes | no ) #IMPLIED >
<!ATTLIST tf dict (yes | no ) #IMPLIED >
<!ATTLIST tf pos ( phrase | phrase_n | phrase_a | phrase_v | phrase_adv ) #IMPLIED >
<!ATTLIST tf gen ( m | f | n ) #IMPLIED >
<!ATTLIST tf lang ( swe | nob ) #IMPLIED >
<!ATTLIST tf xml:lang ( nob | sme | sma | swe | eng | deu ) #IMPLIED >
<!ATTLIST tf tcomm ( yes | no ) #IMPLIED >
<!ATTLIST tf stat ( pref ) #IMPLIED >


<!ELEMENT te (#PCDATA) > <!-- Translation explanation -->
<!ATTLIST te lang ( swe | nob ) #IMPLIED >
<!ATTLIST te xml:lang ( nob | sme | sma | swe | eng | deu ) #IMPLIED >
<!ATTLIST te dict (yes | no ) #IMPLIED >
<!ATTLIST te oa ( first | yes | no ) #IMPLIED >
<!ATTLIST te stat ( pref ) #IMPLIED >
<!ATTLIST te pos ( phrase | phrase_n | phrase_a | phrase_v | phrase_adv ) #IMPLIED >

<!ELEMENT morph_expl (#PCDATA) >


<!ELEMENT xg ((x, xt+)+)>
<!ATTLIST xg re CDATA #IMPLIED >
<!ELEMENT x (#PCDATA)>
<!ATTLIST x src CDATA #IMPLIED >
<!ELEMENT xt (#PCDATA)>
<!ATTLIST xt src CDATA #IMPLIED >

<!ELEMENT syng (syn+)>
<!ELEMENT syn (#PCDATA)>

<!ELEMENT xhtml:script EMPTY >
<!ATTLIST xhtml:script type CDATA #FIXED "text/javascript" >
<!ATTLIST xhtml:script src CDATA #IMPLIED >
<!ATTLIST xhtml:script xmlns:xhtml CDATA #FIXED "http://www.w3.org/1999/xhtml" >



<!--
This document is not quite finished yet...

Here are the pos tags:
a = Adjective
m = Masculine (nob)
f = Feminine (nob)
n = Neuter (nob), noun (sme)
S = noun, still not given gender (nob)
v = verb
adv = adverb
num = numeral
pcle = particle
p = preposition (nob)
pr = preposition (sme)
po = postposition (sme)
pron = pronoun
i = interjection
cc = conjunction
cs = subjunction
im = infinitive mark å (nob)
s = sentence, or explanation
x = still no pos

Documenting
(the tags are short due to a manual cleanup)
r root
e entry
l lemma
lc lemma comment
m meaning
re restriction (semantic restriction of the following t)
t translation
x example
xt example translation
.g group
-->

0 comments on commit 7c4670e

Please sign in to comment.