lingdocs
/
pashto-inflector
mirror of https://github.com/lingdocs/pashto-inflector.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

more work and tests

This commit is contained in:
adueck 2023-04-12 18:36:27 -04:00
parent 092c64f967
commit bbeb219b01
5 changed files with 1025 additions and 122 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
.vscode/sn.code-snippets vendored Normal file
View File

@ -0,0 +1,52 @@
{
"PS Sententce": {
"prefix": "pssentence",
"body": [
"{",
"\tp: \"$1\",",
"\tf: \"$2\",",
"\te: \"$3\",",
"}"
],
},
"PS String": {
"prefix": "psstring",
"body": "{ p: \"$1\", f: \"$2\" }",
},
"InlinePs": {
"prefix": "inlineps",
"body": "<InlinePs opts={opts} ps={{ p: \"$1\", f: \"$2\", e: \"$3\" }} />",
},
"EditableVPEx": {
"prefix": "editablevpex",
"body": [
"<EditableVPEx opts={opts}>{",
"\t$1",
"}</EditableVPEx>"
]
},
"EditableEPEx": {
"prefix": "editableepex",
"body": [
"<EditableEPEx opts={opts} hideOmitSubject>{",
"\t$1",
"}</EditableEPEx>"
]
},
"Examples": {
"prefix": "examples",
"body": [
"<Examples opts={opts}>{[",
"\t{",
"\t\tp: \"$1\",",
"\t\tf: \"$2\",",
"\t\te: \"$3\",",
"\t},",
"]}</Examples>",
],
},
"Link": {
"prefix": "link",
"body": "<Link to=\"$1\">$2</Link>",
}
}

8
src/lib/src/accent-helpers.ts
View File

@ -27,7 +27,7 @@ export function accentOnFront(s: T.SingleOrLengthOpts<T.PsString>): T.SingleOrLe
} }
return { return {
...s, ...s,
f: accentSyllable(removeAccents(s.f)), f: accentLetter(removeAccents(s.f)),
}; };
} }
@ -78,7 +78,7 @@ export function accentFSylsOnNFromEnd(syls: string[] | string, n: number): strin
} }
return [ return [
...syls.slice(0, syls.length-(n+1)), // before accent ...syls.slice(0, syls.length-(n+1)), // before accent
accentSyllable(syls[syls.length-(n+1)]), // syllable to be accented accentLetter(syls[syls.length-(n+1)]), // syllable to be accented
...(n !== 0) ? syls.slice(syls.length-n) : [], // after syllable to be accented ...(n !== 0) ? syls.slice(syls.length-n) : [], // after syllable to be accented
].join(""); ].join("");
} }
@ -104,7 +104,7 @@ const accentReplacer = [
{ vowel: "U", accented: "Ú" }, { vowel: "U", accented: "Ú" },
]; ];
export function accentSyllable(s: string): string { export function accentLetter(s: string): string {
return s.replace(/a|e|i|o|u|U/, (match) => { return s.replace(/a|e|i|o|u|U/, (match) => {
const r = accentReplacer.find((x) => x.vowel === match); const r = accentReplacer.find((x) => x.vowel === match);
/* istanbul ignore next */ /* istanbul ignore next */
@ -115,7 +115,7 @@ export function accentSyllable(s: string): string {
export function accentPsSyllable(ps: T.PsString): T.PsString { export function accentPsSyllable(ps: T.PsString): T.PsString {
return { return {
p: ps.p, p: ps.p,
f: accentSyllable(ps.f), f: accentLetter(ps.f),
}; };
} }

843
src/lib/src/new-verb-engine/roots-and-stems.test.ts
View File

@ -1,7 +1,6 @@
import * as T from "../../../types"; import * as T from "../../../types";
import { getPastParticiple } from "./roots-and-stems"; import { getPastParticiple, getRootStem } from "./roots-and-stems";
import { vEntry } from "./rs-helpers"; import { vEntry } from "./rs-helpers";
import { getAllRs } from "./rs-helpers";
const wahul = vEntry({"ts":1527815399,"i":15049,"p":"وهل","f":"wahul","g":"wahul","e":"to hit","r":4,"c":"v. trans.","tppp":"واهه","tppf":"waahu","ec":"hit,hits,hitting,hit,hit"}); const wahul = vEntry({"ts":1527815399,"i":15049,"p":"وهل","f":"wahul","g":"wahul","e":"to hit","r":4,"c":"v. trans.","tppp":"واهه","tppf":"waahu","ec":"hit,hits,hitting,hit,hit"});
const achawul = vEntry({"ts":1527811872,"i":224,"p":"اچول","f":"achawul","g":"achawul","e":"to put, pour, drop, throw, put on","r":4,"c":"v. trans.","ec":"put,puts,putting,put,put"}); const achawul = vEntry({"ts":1527811872,"i":224,"p":"اچول","f":"achawul","g":"achawul","e":"to put, pour, drop, throw, put on","r":4,"c":"v. trans.","ec":"put,puts,putting,put,put"});
@ -26,6 +25,719 @@ const kenaastul = vEntry({"ts":1527812759,"i":11124,"p":"کېناستل","f":"ke
const ghadzedul = vEntry({"ts":1527812615,"i":9500,"p":"غځېدل","f":"ghadzedul","g":"ghadzedul","e":"stretch out, lie, be extended, expand","r":3,"c":"v. intrans.","ec":"stretch","ep":"out"}); const ghadzedul = vEntry({"ts":1527812615,"i":9500,"p":"غځېدل","f":"ghadzedul","g":"ghadzedul","e":"stretch out, lie, be extended, expand","r":3,"c":"v. intrans.","ec":"stretch","ep":"out"});
const prexodul = vEntry({"ts":1527815190,"i":2495,"p":"پرېښودل","f":"prexodúl","g":"prexodul","e":"to leave, abandon, forsake, let go, allow","r":4,"c":"v. trans.","psp":"پرېږد","psf":"preGd","noOo":true,"separationAtP":3,"separationAtF":3,"ec":"abandon"}); const prexodul = vEntry({"ts":1527815190,"i":2495,"p":"پرېښودل","f":"prexodúl","g":"prexodul","e":"to leave, abandon, forsake, let go, allow","r":4,"c":"v. trans.","psp":"پرېږد","psf":"preGd","noOo":true,"separationAtP":3,"separationAtF":3,"ec":"abandon"});
const raawustul = vEntry({"ts":1527819827,"i":6955,"p":"راوستل","f":"raawustúl","g":"raawustul","e":"to bring, deliver (animate objects), obtain, extract","r":3,"c":"v. trans.","psp":"راول","psf":"raawul","noOo":true,"separationAtP":2,"separationAtF":3,"ec":"bring,brings,bringing,brought,brought"}); const raawustul = vEntry({"ts":1527819827,"i":6955,"p":"راوستل","f":"raawustúl","g":"raawustul","e":"to bring, deliver (animate objects), obtain, extract","r":3,"c":"v. trans.","psp":"راول","psf":"raawul","noOo":true,"separationAtP":2,"separationAtF":3,"ec":"bring,brings,bringing,brought,brought"});
const bandawul = vEntry(
{"ts":1527821309,"i":1792,"p":"بندول","f":"bandawul","g":"bandawul","e":"to close, block, stop, barricade, cut off, restrain, hold back","r":3,"c":"v. stat. comp. trans.","l":1577301753727,"ec":"close"},
{"ts":1577301753727,"i":1780,"p":"بند","f":"band","g":"band","e":"closed, blocked, stopped","c":"adj."},
);
const bandedul = vEntry(
{"ts":1588781671306,"i":1796,"p":"بندېدل","f":"bandedúl","g":"bandedul","e":"to be closed, blocked, stopped","r":4,"c":"v. stat. comp. intrans.","l":1577301753727,"ec":"be","ep":"closed"},
{"ts":1577301753727,"i":1780,"p":"بند","f":"band","g":"band","e":"closed, blocked, stopped","c":"adj."},
);
const stureyKawul = vEntry(
{"ts":1591033078746,"i":7877,"p":"ستړی کول","f":"stuRey kawul","g":"stuReykawul","e":"to make tired, wear out","r":4,"c":"v. stat. comp. trans.","l":1527815306,"ec":"make","ep":"tired"},
{"ts":1527815306,"i":7876,"p":"ستړی","f":"stúRey","g":"stuRey","e":"tired","r":4,"c":"adj. / adv."},
);
const stureyKedul = vEntry(
{"ts":1591033069786,"i":7878,"p":"ستړی کېدل","f":"stuRey kedul","g":"stuReykedul","e":"to get tired, fatigued","r":4,"c":"v. stat. comp. intrans.","l":1527815306,"ec":"get","ep":"tired"},
{"ts":1527815306,"i":7876,"p":"ستړی","f":"stúRey","g":"stuRey","e":"tired","r":4,"c":"adj. / adv."},
);
const ooPH: T.PH= { type: "PH", ps: { p: "و", f: "óo" }};
describe("imperfective stems", () => {
const tests: {
title: string,
tests: {
verb: T.VerbEntry,
genderNumber?: T.GenderNumber,
result: T.RootsStemsOutput,
}[],
}[] = [
{
title: "is the shortened infinitive for regular verbs",
tests: [
{
verb: ganul,
result: [[], [{ type: "VB", ps: [{ p: "ګڼ", f: "gaN" }]}]],
},
{
verb: wahul,
result: [[], [{ type: "VB", ps: [{ p: "وه", f: "wah" }]}]],
},
],
},
{
title: "is the -eG for for regular intransitive verbs",
tests: [
{
verb: ghadzedul,
result: [[], [{ type: "VB", ps: [{ p: "غځېږ", f: "ghadzéG" }]}]],
},
{
verb: rasedul,
result: [[], [{
type: "VB",
ps: {
long: [{ p: "رسېږ", f: "raséG" }],
short: [{ p: "رس", f: "ras" }],
},
}]],
},
],
},
{
title: "is the special imperfective stem for irregular verbs with a special imperfective stem",
tests: [
{
verb: khorul,
result: [[], [{ type: "VB", ps: [{ p: "خور", f: "khor" }]}]],
},
{
verb: kexodul,
result: [[], [{ type: "VB", ps: [{ p: "ږد", f: "Gd" }] }]],
},
],
},
{
title: "is the fused word for stative compounds that swallow the k",
tests: [
{
verb: bandawul,
genderNumber: { gender: "masc", number: "singular" },
result: [[], [{ type: "VB", ps: [{ p: "بندو", f: "bandawX" }]}]],
},
{
verb: bandedul,
genderNumber: { gender: "masc", number: "singular" },
result: [[], [{ type: "VB", ps: [{ p: "بندېږ", f: "bandéG" }]}]],
},
],
},
{
title: "is welded together with the complement on seperated stative compounds",
tests: [
{
verb: stureyKawul,
genderNumber: { gender: "fem", number: "singular" },
result: [[], [{
type: "welded",
left: {
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "ستړې", f: "stuRe" },
number: "singular",
gender: "fem",
},
},
right: {
type: "VB",
ps: [{ p: "کو", f: "kawX" }],
},
}]],
},
{
verb: stureyKedul,
genderNumber: { gender: "fem", number: "singular" },
result: [[], [{
type: "welded",
left: {
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "ستړې", f: "stuRe" },
number: "singular",
gender: "fem",
},
},
right: {
type: "VB",
ps: [{ p: "کېږ", f: "kéG" }],
},
}]],
},
]
}
];
tests.forEach(x => {
test(x.title, () => {
x.tests.forEach(y => {
expect(getRootStem({
verb: y.verb,
aspect: "imperfective",
type: "basic",
rs: "stem",
voice: "active",
genderNumber: y.genderNumber || {
gender: "masc",
number: "singular",
},
})).toEqual(y.result);
});
})
});
});
describe("imperfective roots", () => {
const tests: {
title: string,
tests: {
verb: T.VerbEntry,
genderNumber?: T.GenderNumber,
result: T.RootsStemsOutput,
}[],
}[] = [
{
title: "is the infinitive with and without ل for regular verbs, with a trailing accent on the short form",
tests: [
{
verb: wahul,
result: [[], [{
type: "VB",
ps: {
long: [{ p: "وهل", f: "wahúl" }],
short: [{ p: "وه", f: "wahX" }],
},
}]
],
},
],
},
{
title: "is the fused word for stative compounds that swallow the k",
tests: [
{
verb: bandawul,
genderNumber: { gender: "masc", number: "singular" },
result: [
[],
[
{
type: "VB",
ps: {
long: [{ p: "بندول", f: "bandawúl" }],
short: [{ p: "بندو", f: "bandawX" }],
}
},
],
],
},
{
verb: bandedul,
genderNumber: { gender: "masc", number: "singular" },
result: [
[],
[
{
type: "VB",
ps: {
long: [{ p: "بندېدل", f: "bandedúl" }],
short: [{ p: "بندېد", f: "bandedX" }]
},
},
],
],
},
],
},
{
title: "is welded together with the complement on seperated stative compounds",
tests: [
{
verb: stureyKawul,
genderNumber: { gender: "fem", number: "singular" },
result: [[], [{
type: "welded",
left: {
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "ستړې", f: "stuRe" },
number: "singular",
gender: "fem",
},
},
right: {
type: "VB",
ps: {
long: [{ p: "کول", f: "kawúl" }],
short: [{ p: "کو", f: "kawX" }],
},
},
}]],
},
{
verb: stureyKedul,
genderNumber: { gender: "fem", number: "singular" },
result: [[], [{
type: "welded",
left: {
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "ستړې", f: "stuRe" },
number: "singular",
gender: "fem",
},
},
right: {
type: "VB",
ps: {
long: [{ p: "کېدل", f: "kedúl" }],
short: [{ p: "کېد", f: "kedX" }],
},
},
}]],
},
]
}
];
tests.forEach(x => {
test(x.title, () => {
x.tests.forEach(y => {
expect(getRootStem({
verb: y.verb,
aspect: "imperfective",
type: "basic",
rs: "root",
voice: "active",
genderNumber: y.genderNumber || {
gender: "masc",
number: "singular",
},
})).toEqual(y.result);
});
})
});
});
describe("perfective stems", () => {
const tests: {
title: string,
tests: {
verb: T.VerbEntry,
genderNumber?: T.GenderNumber,
result: T.RootsStemsOutput,
}[],
}[] = [
{
title: "is the imprefective stem with an oo- prefix for regular verbs",
tests: [
{
verb: ganul,
result: [
[ooPH],
[{ type: "VB", ps: [{ p: "ګڼ", f: "gaN" }]}],
],
},
{
verb: wahul,
result: [[ooPH], [{ type: "VB", ps: [{ p: "وه", f: "wah" }]}]],
},
],
},
{
title: "the imperfective stem the perfective stem is regularly built off could be irregular",
tests: [
{
verb: khorul,
result: [[ooPH], [{ type: "VB", ps: [{ p: "خور", f: "khor" }]}]],
},
],
},
{
title: "is the irregular perfective stem if the entry has a ssp/ssf",
tests: [
{
verb: wurul,
result: [
[{ type: "PH", ps: { p: "یو", f: "yó" }}],
[{ type: "VB", ps: [{ p: "س", f: "s" }]}],
],
}
]
},
{
title: "includes extra irregular short roots for kawul verbs",
tests: [
{
verb: kawulDyn,
result: [
[ooPH],
[
{
type: "VB",
ps: {
long: [{ p: "کړ", f: "kR" }],
short: [{ p: "ک", f: "k" }],
},
},
]
],
},
],
},
{
title: "doesn't have a perfective head for kawul stative",
tests: [
{
verb: kawulStat,
result: [
[],
[
{
type: "VB",
ps: {
long: [{ p: "کړ", f: "kR" }],
short: [{ p: "ک", f: "k" }],
},
},
]
],
},
],
},
{
title: "irregular, inflecting stem for tlul",
tests: [
{
verb: tlul,
genderNumber: { gender: "fem", number: "singular" },
result: [
[{ type: "PH", ps: { p: "لاړه ", f: "láaRa " }}],
[
{ type: "VB", ps: [{ p: "ش", f: "sh" }]},
],
],
},
{
verb: tlul,
genderNumber: { gender: "fem", number: "plural" },
result: [
[{ type: "PH", ps: { p: "لاړې ", f: "láaRe " }}],
[
{ type: "VB", ps: [{ p: "ش", f: "sh" }]},
],
],
},
{
verb: tlul,
genderNumber: { gender: "masc", number: "plural" },
result: [
[{ type: "PH", ps: { p: "لاړ ", f: "láaR " }}],
[
{ type: "VB", ps: [{ p: "ش", f: "sh" }]},
],
],
},
],
},
{
title: "broken apart with complement seperately in the perfective",
tests: [
{
verb: bandawul,
genderNumber: { gender: "masc", number: "singular" },
result: [
[
{
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "بند", f: "bánd" },
gender: "masc",
number: "singular",
},
},
],
[
{
type: "VB",
ps: {
long: [{ p: "کړ", f: "kR" }],
short: [{ p: "ک", f: "k" }],
},
},
],
],
},
{
verb: bandedul,
genderNumber: { gender: "fem", number: "plural" },
result: [
[
{
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "بندې", f: "bánde" },
gender: "fem",
number: "plural",
},
},
],
[
{
type: "VB",
ps: [{ p: "ش", f: "sh" }],
},
],
],
},
{
verb: stureyKedul,
genderNumber: { gender: "masc", number: "plural" },
result: [
[
{
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "ستړي", f: "stúRee" },
gender: "masc",
number: "plural",
},
},
],
[
{
type: "VB",
ps: [{ p: "ش", f: "sh" }],
},
],
],
},
],
},
];
tests.forEach(x => {
test(x.title, () => {
x.tests.forEach(y => {
expect(getRootStem({
verb: y.verb,
aspect: "perfective",
type: "basic",
rs: "stem",
voice: "active",
genderNumber: y.genderNumber || {
gender: "masc",
number: "singular",
},
})).toEqual(y.result);
});
})
});
});
describe("perfective roots", () => {
const tests: {
title: string,
tests: {
verb: T.VerbEntry,
genderNumber?: T.GenderNumber,
result: T.RootsStemsOutput,
}[],
}[] = [
{
title: "is the imprefective root with an oo- prefix for regular verbs",
tests: [
{
verb: ganul,
result: [
[ooPH],
[
{
type: "VB",
ps: {
long: [{ p: "ګڼل", f: "gaNul" }],
short: [{ p: "ګڼ", f: "gaN" }],
},
},
],
],
},
{
verb: wahul,
result: [
[ooPH],
[
{
type: "VB",
ps: {
long: [{ p: "وهل", f: "wahul" }],
short: [{ p: "وه", f: "wah" }],
},
},
],
],
},
],
},
{
title: "is the irregular perfective root if the entry has a ssp/ssf",
tests: [
{
verb: wurul,
result: [
[{ type: "PH", ps: { p: "یو", f: "yó" }}],
[
{
type: "VB",
ps: {
long: [{ p: "ړل", f: "Rul" }],
short: [{ p: "ړ", f: "R" }],
}
},
],
],
},
],
},
{
title: "includes extra irregular short roots for kawul verbs",
tests: [
{
verb: kawulDyn,
result: [
[ooPH],
[
{
type: "VB",
ps: {
long: [{ p: "کړل", f: "kRul" }],
short: [{ p: "کړ", f: "kR" }],
mini: [{ p: "ک", f: "k" }],
},
},
],
],
},
],
},
{
title: "doesn't have a perfective head for kawul stative",
tests: [
{
verb: kawulStat,
result: [
[],
[
{
type: "VB",
ps: {
long: [{ p: "کړل", f: "kRul" }],
short: [{ p: "کړ", f: "kR" }],
mini: [{ p: "ک", f: "k" }],
},
},
]
],
},
],
},
{
title: "broken apart with complement seperately in the perfective",
tests: [
{
verb: bandawul,
genderNumber: { gender: "masc", number: "singular" },
result: [
[
{
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "بند", f: "bánd" },
gender: "masc",
number: "singular",
},
},
],
[
{
type: "VB",
ps: {
long: [{ p: "کړل", f: "kRul" }],
short: [{ p: "کړ", f: "kR" }],
mini: [{ p: "ک", f: "k" }],
},
},
],
],
},
{
verb: bandedul,
genderNumber: { gender: "fem", number: "plural" },
result: [
[
{
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "بندې", f: "bánde" },
gender: "fem",
number: "plural",
},
},
],
[
{
type: "VB",
ps: {
long: [{ p: "شول", f: "shwul" }],
short: [{ p: "شو", f: "shw" }],
}
},
],
],
},
{
verb: stureyKedul,
genderNumber: { gender: "masc", number: "plural" },
result: [
[
{
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "ستړي", f: "stúRee" },
gender: "masc",
number: "plural",
},
},
],
[
{
type: "VB",
ps: {
long: [{ p: "شول", f: "shwul" }],
short: [{ p: "شو", f: "shw" }],
}
},
],
],
},
],
},
];
tests.forEach(x => {
test(x.title, () => {
x.tests.forEach(y => {
expect(getRootStem({
verb: y.verb,
aspect: "perfective",
type: "basic",
rs: "root",
voice: "active",
genderNumber: y.genderNumber || {
gender: "masc",
number: "singular",
},
})).toEqual(y.result);
});
})
});
});
describe("past participles", () => { describe("past participles", () => {
test("for most verbs are just the imperfective root (imperative) plus ی - ey", () => { test("for most verbs are just the imperfective root (imperative) plus ی - ey", () => {
@ -97,7 +809,19 @@ describe("past participles", () => {
number: "plural", number: "plural",
}); });
}); });
test("kawul/kedul verbs have an irregular pprt fields that give us the irregular past participle", () => { test("special short form with تلل - tlul", () => {
expect(getPastParticiple(tlul, "active", { gender: "masc", number: "plural" }))
.toEqual({
type: "VB",
ps: {
long: [{ p: "تللي", f: "tlúlee" }],
short: [{ p: "تلي", f: "túlee" }],
},
gender: "masc",
number: "plural",
});
});
test("kawul/kedul/raatlul verbs have an irregular pprt fields that give us the irregular past participle", () => {
expect(getPastParticiple(kawulDyn, "active", { gender: "masc", number: "singular" })) expect(getPastParticiple(kawulDyn, "active", { gender: "masc", number: "singular" }))
.toEqual({ .toEqual({
type: "VB", type: "VB",
@ -127,7 +851,118 @@ describe("past participles", () => {
number: "plural", number: "plural",
}); });
}); });
test("stative compounds weld the complement to the kawul/kedul participle", () => {
expect(getPastParticiple(bandawul, "active", { gender: "fem", number: "singular" }))
.toEqual({
type: "welded",
left: {
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "بنده", f: "banda" },
gender: "fem",
number: "singular",
},
},
right: {
type: "VB",
ps: [{ p: "کړې", f: "kúRe" }],
gender: "fem",
number: "singular",
},
});
expect(getPastParticiple(bandedul, "active", { gender: "fem", number: "plural" }))
.toEqual({
type: "welded",
left: {
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "بندې", f: "bande" },
gender: "fem",
number: "plural",
},
},
right: {
type: "VB",
ps: [{ p: "شوې", f: "shúwe" }],
gender: "fem",
number: "plural",
},
});
});
test("for passive with simple verbs, long perfective root welded to kedul participle", () => {
expect(getPastParticiple(ganul, "passive", { gender: "fem", number: "singular" }))
.toEqual({
type: "welded",
left: {
type: "VB",
ps: [{ p: "ګڼل", f: "gaNul" }],
},
right: {
type: "VB",
ps: [{ p: "شوې", f: "shúwe" }],
gender: "fem",
number: "singular",
},
});
});
test("special passive forms for kawul verbs - kRul perfective root + shúwey", () => {
expect(getPastParticiple(kawulStat, "passive", { gender: "masc", number: "singular"}))
.toEqual({
type: "welded",
left: {
type: "VB",
ps: [{ p: "کړل", f: "kRul" }],
},
right: {
type: "VB",
ps: [{ p: "شوی", f: "shúwey" }],
gender: "masc",
number: "singular",
},
});
expect(getPastParticiple(kawulDyn, "passive", { gender: "masc", number: "singular"}))
.toEqual({
type: "welded",
left: {
type: "VB",
ps: [{ p: "کړل", f: "kRul" }],
},
right: {
type: "VB",
ps: [{ p: "شوی", f: "shúwey" }],
gender: "masc",
number: "singular",
},
});
expect(getPastParticiple(bandawul, "passive", { gender: "fem", number: "plural" }))
.toEqual({
type: "welded",
left: {
type: "welded",
left: {
type: "NComp",
comp: {
type: "AdjComp",
ps: { p: "بندې", f: "bande" },
gender: "fem",
number: "plural",
},
},
right: {
type: "VB",
ps: [{ p: "کړل", f: "kRul" }],
},
},
right: {
type: "VB",
ps: [{ "p": "شوې", "f": "shúwe" }],
gender: "fem",
number: "plural"
},
});
})
}); });
// const ooPh: T.PH = { // const ooPh: T.PH = {

208
src/lib/src/new-verb-engine/roots-and-stems.ts
View File

@ -11,13 +11,15 @@ import {
} from "../p-text-helpers"; } from "../p-text-helpers";
import * as T from "../../../types"; import * as T from "../../../types";
import { makePsString, removeFVarientsFromVerb } from "../accent-and-ps-utils"; import { makePsString, removeFVarientsFromVerb } from "../accent-and-ps-utils";
import { accentOnNFromEnd, accentSyllable, removeAccents } from "../accent-helpers"; import { accentOnNFromEnd, countSyllables, removeAccents } from "../accent-helpers";
import { isKawulVerb, isTlulVerb } from "../type-predicates"; import { isKawulVerb, isTlulVerb } from "../type-predicates";
import { vEntry, addAbilityEnding, weld, removeL, addTrailingAccent, tlulPerfectiveStem, getLongVB, possiblePPartLengths, isStatComp, statCompImperfectiveSpace, makeComplement } from "./rs-helpers"; import { vEntry, addAbilityEnding, weld, removeL, addTrailingAccent, tlulPerfectiveStem, getLongVB, possiblePPartLengths, isStatComp, statCompImperfectiveSpace, makeComplement, vTransitivity } from "./rs-helpers";
import { inflectPattern3 } from "./new-inflectors"; import { inflectPattern3 } from "./new-inflectors";
const kedulStat = vEntry({"ts":1581086654898,"i":11100,"p":"کېدل","f":"kedul","g":"kedul","e":"to become _____","r":2,"c":"v. intrans.","ssp":"ش","ssf":"sh","prp":"شول","prf":"shwul","pprtp":"شوی","pprtf":"shúwey","noOo":true,"ec":"become"}); const statVerb = {
const kawulStat = vEntry({"ts":1579015359582,"i":11030,"p":"کول","f":"kawul","g":"kawul","e":"to make ____ ____ (as in \"He's making me angry.\")","r":4,"c":"v. trans.","ssp":"کړ","ssf":"kR","prp":"کړل","prf":"kRul","pprtp":"کړی","pprtf":"kúRey","noOo":true,"ec":"make,makes,making,made,made"}); intransitive: vEntry({"ts":1581086654898,"i":11100,"p":"کېدل","f":"kedul","g":"kedul","e":"to become _____","r":2,"c":"v. intrans.","ssp":"ش","ssf":"sh","prp":"شول","prf":"shwul","pprtp":"شوی","pprtf":"shúwey","noOo":true,"ec":"become"}),
transitive: vEntry({"ts":1579015359582,"i":11030,"p":"کول","f":"kawul","g":"kawul","e":"to make ____ ____ (as in \"He's making me angry.\")","r":4,"c":"v. trans.","ssp":"کړ","ssf":"kR","prp":"کړل","prf":"kRul","pprtp":"کړی","pprtf":"kúRey","noOo":true,"ec":"make,makes,making,made,made"}),
};
const shwulVB: T.VBBasic = { const shwulVB: T.VBBasic = {
type: "VB", type: "VB",
@ -54,9 +56,7 @@ export function getRootStem({ verb, rs, aspect, type, genderNumber, voice }: {
? aspect === "imperfective" ? aspect === "imperfective"
? getImperfectiveStem(v, genderNumber) ? getImperfectiveStem(v, genderNumber)
: getPerfectiveStem(v, genderNumber) : getPerfectiveStem(v, genderNumber)
: aspect === "imperfective" : getRoot(v, genderNumber, aspect);
? getImperfectiveRoot(v, genderNumber)
: getPerfectiveRoot(v, genderNumber);
} }
function getAbilityRs( function getAbilityRs(
@ -66,13 +66,10 @@ function getAbilityRs(
voice: T.Voice, voice: T.Voice,
genderNum: T.GenderNumber, genderNum: T.GenderNumber,
): [[] | [T.VHead], [T.VB, T.VBA]] { ): [[] | [T.VHead], [T.VB, T.VBA]] {
const losesAspect = isTlulVerb(verb) || (isStatComp(verb) && verb.entry.c?.includes("intrans.")); const losesAspect = isTlulVerb(verb) || (isStatComp(verb) && vTransitivity(verb) === "intransitive");
const [vhead, [basicroot]] = voice === "passive" const [vhead, [basicroot]] = voice === "passive"
// passive ability loses aspect
? getPassiveRs(verb, "imperfective", "root", genderNum) ? getPassiveRs(verb, "imperfective", "root", genderNum)
: aspect === "imperfective" || losesAspect : getRoot(verb, genderNum, losesAspect ? "imperfective" : aspect);
? getImperfectiveRoot(verb, genderNum)
: getPerfectiveRoot(verb, genderNum);
return [ return [
vhead, vhead,
[ [
@ -88,6 +85,16 @@ export function getPastParticiple(verb: T.VerbEntry, voice: T.Voice, { gender, n
if (voice === "passive") { if (voice === "passive") {
return getPassivePp(v, { gender, number }); return getPassivePp(v, { gender, number });
} }
if (isStatComp(v) && v.complement) {
return weld(
makeComplement(v.complement, { gender, number }),
getPastParticiple(
statVerb[vTransitivity(verb)],
voice,
{ gender, number },
) as T.VBGenNum,
);
}
if (verb.entry.pprtp && verb.entry.pprtf) { if (verb.entry.pprtp && verb.entry.pprtf) {
const base = makePsString(verb.entry.pprtp, verb.entry.pprtf); const base = makePsString(verb.entry.pprtp, verb.entry.pprtf);
return { return {
@ -97,28 +104,19 @@ export function getPastParticiple(verb: T.VerbEntry, voice: T.Voice, { gender, n
number, number,
}; };
} }
const basicRoot = getImperfectiveRoot(removeFVarientsFromVerb(verb), { gender, number })[1][0]; const basicRoot = getRoot(removeFVarientsFromVerb(verb), { gender, number }, "imperfective")[1][0];
const longRoot = getLongVB(basicRoot); const longRoot = getLongVB(basicRoot);
const rootWLengths = possiblePPartLengths(longRoot); const rootWLengths = possiblePPartLengths(longRoot);
/* istanbul ignore next */
if ("right" in rootWLengths) { if ("right" in rootWLengths) {
return { throw new Error("should not have welded here");
...rootWLengths, }
right: { return {
...rootWLengths.right, ...rootWLengths,
ps: addTail(rootWLengths.right.ps), ps: addTail(rootWLengths.ps),
gender, gender,
number, number,
}, };
};
} else {
return {
...rootWLengths,
ps: addTail(rootWLengths.ps),
gender,
number,
};
}
function addTail(ps: T.SingleOrLengthOpts<T.PsString[]>): T.SingleOrLengthOpts<T.PsString[]> { function addTail(ps: T.SingleOrLengthOpts<T.PsString[]>): T.SingleOrLengthOpts<T.PsString[]> {
if ("long" in ps) { if ("long" in ps) {
@ -133,97 +131,87 @@ export function getPastParticiple(verb: T.VerbEntry, voice: T.Voice, { gender, n
} }
function getPassivePp(verb: T.VerbEntryNoFVars, genderNumber: T.GenderNumber): T.WeldedGN { function getPassivePp(verb: T.VerbEntryNoFVars, genderNumber: T.GenderNumber): T.WeldedGN {
const basicRoot = getImperfectiveRoot(verb, genderNumber)[1][0]; if (isStatComp(verb) && verb.complement) {
return weld(
makeComplement(verb.complement, genderNumber),
getPassivePp(statVerb.transitive, genderNumber),
);
}
const basicRoot = getRoot(verb, genderNumber, isKawulVerb(verb) ? "perfective" : "imperfective")[1][0];
const longRoot = getLongVB(basicRoot); const longRoot = getLongVB(basicRoot);
const kedulVb: T.VBGenNum = getPastParticiple(kedulStat, "active", genderNumber) as T.VBGenNum; const kedulVb: T.VBGenNum = getPastParticiple(statVerb.intransitive, "active", genderNumber) as T.VBGenNum;
return weld(longRoot, kedulVb); return weld(longRoot, kedulVb);
} }
// TODO: could combine these two functions...
function getPassiveRs(verb: T.VerbEntryNoFVars, aspect: T.Aspect, rs: "root" | "stem", genderNumber: T.GenderNumber): [[] | [T.VHead], [T.VBA]] { function getPassiveRs(verb: T.VerbEntryNoFVars, aspect: T.Aspect, rs: "root" | "stem", genderNumber: T.GenderNumber): [[] | [T.VHead], [T.VBA]] {
const [vHead, [basicRoot]] = (aspect === "imperfective" const [vHead, [basicRoot]] = getRoot(verb, genderNumber, aspect);
? getImperfectiveRoot
: getPerfectiveRoot
)(verb, genderNumber);
const longRoot = getLongVB(basicRoot); const longRoot = getLongVB(basicRoot);
const kedulVba = getRootStem({ verb: kedulStat, aspect, rs, type: "basic", voice: "active", genderNumber: { gender: "masc", number: "singular" }})[1][0] as T.VBBasic; const kedulVba = getRootStem({ verb: statVerb.intransitive, aspect, rs, type: "basic", voice: "active", genderNumber: { gender: "masc", number: "singular" }})[1][0] as T.VBBasic;
return [ return [
vHead, vHead,
[weld(longRoot, kedulVba)], [weld(longRoot, kedulVba)],
]; ];
} }
function getImperfectiveRoot(verb: T.VerbEntryNoFVars, genderNum: T.GenderNumber): [[], [T.VBA]] { function getRoot(verb: T.VerbEntryNoFVars, genderNum: T.GenderNumber, aspect: T.Aspect): [[T.VHead] | [], [T.VBA]] {
if (verb.complement && isStatComp(verb) && statCompImperfectiveSpace(verb)) { if (verb.complement && isStatComp(verb) && (aspect === "perfective" || statCompImperfectiveSpace(verb))) {
const auxStem = getImperfectiveRoot(verb.entry.c?.includes("intrans.") const auxStem = getRoot(
? kedulStat statVerb[vTransitivity(verb)],
: kawulStat genderNum,
, genderNum)[1][0]; aspect,
return [ )[1][0] as T.VBBasic;
[], const comp = makeComplement(verb.complement, genderNum);
[ return aspect === "perfective"
weld( ? [[comp], [auxStem]]
makeComplement(verb.complement, genderNum), : [
auxStem as T.VBBasic, [],
), [
], weld(
]; makeComplement(verb.complement, genderNum),
auxStem as T.VBBasic,
),
],
];
} }
const infinitive = accentOnNFromEnd(makePsString(verb.entry.p, verb.entry.f), 0); const base = aspect === "imperfective"
return [ ? accentOnNFromEnd(makePsString(verb.entry.p, verb.entry.f), 0)
[], : removeAccents(
[ (verb.entry.prp && verb.entry.prf)
{ ? makePsString(verb.entry.prp, verb.entry.prf)
type: "VB", : makePsString(verb.entry.p, verb.entry.f)
ps: { );
long: [infinitive], const [perfectiveHead, rest] = aspect === "perfective"
short: [addTrailingAccent(removeL(infinitive))], ? getPerfectiveHead(base, verb)
}, : [undefined, base];
},
],
];
}
function getPerfectiveRoot(verb: T.VerbEntryNoFVars, genderNum: T.GenderNumber): [[T.VHead] | [], [T.VBA]] {
if (verb.complement && isStatComp(verb)) {
const auxStem = getPerfectiveRoot(verb.entry.c?.includes("intrans.")
? kedulStat
: kawulStat,
genderNum)[1][0];
return [
[makeComplement(verb.complement, genderNum)],
[auxStem as T.VBBasic],
];
}
const base = removeAccents(
(verb.entry.prp && verb.entry.prf)
? makePsString(verb.entry.prp, verb.entry.prf)
: makePsString(verb.entry.p, verb.entry.f)
);
const [perfectiveHead, rest] = getPerfectiveHead(base, verb);
return [ return [
perfectiveHead ? [perfectiveHead] : [], perfectiveHead ? [perfectiveHead] : [],
[ [
{ {
type: "VB", type: "VB",
ps: { ps: aspect === "imperfective"
long: [rest], ? {
short: [removeL(rest)], long: [rest],
...isKawulVerb(verb) ? { short: [addTrailingAccent(removeL(rest))],
mini: [{ p: "ک", f: "k" }], }
} : {}, : {
}, long: [rest],
short: [removeL(rest)],
...(aspect === "perfective" && isKawulVerb(verb)) ? {
mini: [{ p: "ک", f: "k" }],
} : {},
},
}, },
], ],
]; ];
} }
function getImperfectiveStem(verb: T.VerbEntryNoFVars, genderNum: T.GenderNumber): [[], [T.VB]] { function getImperfectiveStem(verb: T.VerbEntryNoFVars, genderNum: T.GenderNumber): [[], [T.VB]] {
if (verb.complement && isStatComp(verb) && statCompImperfectiveSpace(verb)) { const statComp = isStatComp(verb);
const auxStem = getImperfectiveStem(verb.entry.c?.includes("intrans.") if (verb.complement && statComp && statCompImperfectiveSpace(verb)) {
? kedulStat const auxStem = getImperfectiveStem(
: kawulStat, statVerb[vTransitivity(verb)],
genderNum)[1][0]; genderNum,
)[1][0];
return [ return [
[], [],
[ [
@ -245,12 +233,17 @@ function getImperfectiveStem(verb: T.VerbEntryNoFVars, genderNum: T.GenderNumber
], ],
]; ];
} }
const rawBase = removeL(makePsString(verb.entry.p, verb.entry.f)); const extraRawBase = removeL(makePsString(verb.entry.p, verb.entry.f));
if (verb.entry.c?.includes("intrans.") && rawBase.p.endsWith("ېد")) { // welded stative -awul verbs should always have a trailing accent
const shortBase = trimOffPs(rawBase, 1, 2); const rawBase = isKawulVerb(verb) || statComp || (countSyllables(extraRawBase) > 1 && extraRawBase.f.endsWith("aw"))
? addTrailingAccent(extraRawBase)
: extraRawBase;
if (vTransitivity(verb) === "intransitive" && extraRawBase.p.endsWith("ېد")) {
// TODO: Factor out into a function
const shortBase = trimOffPs(extraRawBase, 2, 2);
const long: T.PsString[] = [concatPsString( const long: T.PsString[] = [concatPsString(
shortBase, shortBase,
{ p: "ږ", f: "éG" }, { p: "ېږ", f: "éG" },
)]; )];
const short: T.PsString[] | undefined = (verb.entry.shortIntrans) const short: T.PsString[] | undefined = (verb.entry.shortIntrans)
? [shortBase] ? [shortBase]
@ -277,10 +270,7 @@ function getImperfectiveStem(verb: T.VerbEntryNoFVars, genderNum: T.GenderNumber
function getPerfectiveStem(verb: T.VerbEntryNoFVars, genderNum: { gender: T.Gender, number: T.NounNumber }): [[T.VHead] | [], [T.VB]] { function getPerfectiveStem(verb: T.VerbEntryNoFVars, genderNum: { gender: T.Gender, number: T.NounNumber }): [[T.VHead] | [], [T.VB]] {
if (verb.complement && isStatComp(verb)) { if (verb.complement && isStatComp(verb)) {
const auxStem = getPerfectiveStem(verb.entry.c?.includes("intrans.") const auxStem = getPerfectiveStem(statVerb[vTransitivity(verb)], genderNum)[1][0];
? kedulStat
: kawulStat,
genderNum)[1][0];
return [ return [
[makeComplement(verb.complement, genderNum)], [makeComplement(verb.complement, genderNum)],
[auxStem as T.VBBasic], [auxStem as T.VBBasic],
@ -319,10 +309,10 @@ function getPerfectiveHead(base: T.PsString, { entry }: T.VerbEntryNoFVars): [T.
if (entry.separationAtP && entry.separationAtF) { if (entry.separationAtP && entry.separationAtF) {
const ph: T.PH = { const ph: T.PH = {
type: "PH", type: "PH",
ps: { ps: accentOnNFromEnd({
p: base.p.slice(0, entry.separationAtP), p: base.p.slice(0, entry.separationAtP),
f: accentSyllable(base.f.slice(0, entry.separationAtF)), f: base.f.slice(0, entry.separationAtF),
}, }, 0),
}; };
const rest = { const rest = {
p: base.p.slice(entry.separationAtP), p: base.p.slice(entry.separationAtP),

36
src/lib/src/new-verb-engine/rs-helpers.ts
View File

@ -51,7 +51,7 @@ export function vEntry(e: any, c?: any): T.VerbEntryNoFVars {
return { return {
entry: removeFVarients(e), entry: removeFVarients(e),
...c ? { ...c ? {
c, complement: c,
} : {}, } : {},
} as T.VerbEntryNoFVars; } as T.VerbEntryNoFVars;
} }
@ -108,9 +108,17 @@ export function verbEndingConcat(ps: T.PsString[], end: T.PsString[]): T.PsStrin
)); ));
} }
export function weld(left: T.Welded["left"], right: T.VBGenNum): T.WeldedGN; // TODO: THIS IS UGGGGLY NEED TO THINK THROUGH THE TYPING ON THE WELDING
export function weld(left: T.Welded["left"], right: T.VBBasic): T.Welded; export function weld(left: T.Welded["left"], right: T.VBGenNum | T.WeldedGN): T.WeldedGN;
export function weld(left: T.Welded["left"], right: T.VBBasic | T.VBGenNum): T.Welded | T.WeldedGN { export function weld(left: T.Welded["left"], right: T.VBBasic | T.NComp | T.Welded): T.Welded;
export function weld(left: T.Welded["left"], right: T.VBBasic | T.VBGenNum | T.Welded | T.NComp | T.WeldedGN): T.Welded | T.WeldedGN {
if (right.type === "welded") {
return weld(weld(left, right.left), right.right);
}
/* istanbul ignore next */
if (right.type === "NComp") {
throw new Error("can't weld a complement on the right side");
}
return { return {
type: "welded", type: "welded",
left: removeAccentsFromLeft(left), left: removeAccentsFromLeft(left),
@ -154,13 +162,22 @@ export function addTrailingAccent(ps: T.PsString): T.PsString {
export function removeL(ps: T.PsString): T.PsString { export function removeL(ps: T.PsString): T.PsString {
return trimOffPs(ps, 1, 2); return trimOffPs(ps, 1, 2);
} }
/**
* returns a simple polar transitivity of the verb, only intransitive or transitive
*
* @param v
* @returns
*/
export function vTransitivity(v: T.VerbEntry): "intransitive" | "transitive" {
return v.entry.c?.includes("intrans.") ? "intransitive" : "transitive";
}
export function tlulPerfectiveStem(person: { gender: T.Gender, number: T.NounNumber }): [[T.PH], [T.VB]] { export function tlulPerfectiveStem(person: { gender: T.Gender, number: T.NounNumber }): [[T.PH], [T.VB]] {
return [ return [
[ [
{ {
type: "PH", type: "PH",
ps: inflectPattern1({ p: "لاړ", f: "laaR" }, person).map(x => concatPsString(x, " "))[0], ps: inflectPattern1({ p: "لاړ", f: "láaR" }, person).map(x => concatPsString(x, " "))[0],
}, },
], ],
[ [
@ -214,6 +231,15 @@ export function possiblePPartLengths(vba: T.VBNoLenghts<T.VBA>): T.VBA {
}; };
} }
const infinitive = vba.ps[0]; const infinitive = vba.ps[0];
if (infinitive.f === "tlúl") {
return {
type: "VB",
ps: {
long: [infinitive],
short: [{ p: "تل", f: "túl" }],
},
};
}
const [trimP, trimF] = (infinitive.p.slice(-4) === "ښودل" && infinitive.p.length > 4 && infinitive.p !== "کېښودل" && infinitive.p !== "کښېښودل") const [trimP, trimF] = (infinitive.p.slice(-4) === "ښودل" && infinitive.p.length > 4 && infinitive.p !== "کېښودل" && infinitive.p !== "کښېښودل")
// special thing for اېښودل - پرېښودل // special thing for اېښودل - پرېښودل
? [3, 4] ? [3, 4]