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

more and testing on parse-vp

This commit is contained in:
adueck 2023-08-17 19:48:21 +04:00
parent 288718f69a
commit 4b54289c48
7 changed files with 763 additions and 466 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/demo-components/ParserDemo.tsx
View File

@ -50,7 +50,7 @@ function ParserDemo({ opts }: { opts: T.TextOptions }) {
{errors.length > 0 && ( {errors.length > 0 && (
<> <>
<div className="alert alert-danger" role="alert"> <div className="alert alert-danger" role="alert">
{errors.length > 0 ? ( {errors.length > 1 ? (
<> <>
<div>possible errors:</div> <div>possible errors:</div>
<ul> <ul>

7
src/lib/src/parsing/parse-verb.test.ts
View File

@ -11,6 +11,7 @@ import {
import { verbLookup, wordQuery } from "./lookup"; import { verbLookup, wordQuery } from "./lookup";
import { parseVerb } from "./parse-verb"; import { parseVerb } from "./parse-verb";
import { tokenizer } from "./tokenizer"; import { tokenizer } from "./tokenizer";
import { removeKeys } from "./utils";
const wahul = wordQuery("وهل", "verb"); const wahul = wordQuery("وهل", "verb");
const leekul = wordQuery("لیکل", "verb"); const leekul = wordQuery("لیکل", "verb");
@ -1502,11 +1503,7 @@ tests.forEach(({ label, cases }) => {
), ),
]; ];
}, []); }, []);
expect(removeIs(vbs)).toIncludeSameMembers(removeIs(madeVbsS)); expect(removeKeys(vbs)).toIncludeSameMembers(removeKeys(madeVbsS));
}); });
}); });
}); });
function removeIs(a: any): any {
return JSON.parse(JSON.stringify(a, (k, v) => (k === "i" ? undefined : v)));
}

1
src/lib/src/parsing/parse-verb.ts
View File

@ -16,6 +16,7 @@ import {
// check that aawu stuff is working // check that aawu stuff is working
// check oo`azmooy - // check oo`azmooy -
// check څاته // check څاته
// laaRa shum etc
export function parseVerb( export function parseVerb(
tokens: Readonly<T.Token[]>, tokens: Readonly<T.Token[]>,

660
src/lib/src/parsing/parse-vp.test.ts Normal file
View File

@ -0,0 +1,660 @@
/* eslint-disable jest/no-conditional-expect */
/* eslint-disable jest/valid-title */
import * as T from "../../../types";
import {
makeObjectSelectionComplete,
makeSubjectSelection,
makeSubjectSelectionComplete,
} from "../phrase-building/blocks-utils";
import {
makeNounSelection,
makePronounSelection,
} from "../phrase-building/make-selections";
import { lookup, verbLookup, wordQuery } from "./lookup";
import { parseVP } from "./parse-vp";
import { tokenizer } from "./tokenizer";
import { tlul } from "./irreg-verbs";
import { removeKeys } from "./utils";
const sarey = wordQuery("سړی", "noun");
const rasedul = wordQuery("رسېدل", "verb");
const maashoom = wordQuery("ماشوم", "noun");
const leedul = wordQuery("لیدل", "verb");
const tests: {
label: string;
cases: {
input: string;
output: T.VPSelectionComplete[];
error?: boolean;
}[];
}[] = [
{
label: "nothing / failures",
cases: [
{
input: "",
output: [],
},
{
input: "زه سړی رسېږم",
output: [],
},
{
input: "زه سړی ماشوم ولید",
output: [],
},
{
input: "ما ځم",
output: [],
error: true,
},
{
input: "سړی زه ویني",
output: [],
error: true,
},
{
input: "سړي ما ولیدلم",
output: [],
error: true,
},
{
input: "زه ماشومانو وینم",
output: [],
error: true,
},
],
},
{
label: "intransitive full kernels",
cases: [
{
input: "زه ځم",
output: [
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makePronounSelection(T.Person.FirstSingMale),
}),
},
{
key: 2,
block: {
type: "objectSelection",
selection: "none",
},
},
],
verb: {
type: "verb",
verb: tlul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
},
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makePronounSelection(T.Person.FirstSingFemale),
}),
},
{
key: 2,
block: {
type: "objectSelection",
selection: "none",
},
},
],
verb: {
type: "verb",
verb: tlul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
},
],
},
{
input: "سړی رسېږي",
output: [
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
{
key: 2,
block: {
type: "objectSelection",
selection: "none",
},
},
],
verb: {
type: "verb",
verb: rasedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
},
],
},
{
input: "سړی به ورسېږي",
output: [
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
{
key: 2,
block: {
type: "objectSelection",
selection: "none",
},
},
],
verb: {
type: "verb",
verb: rasedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "perfectiveFuture",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
},
],
},
],
},
{
label: "intransitive no king",
cases: [
{
input: "ځم",
output: [
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makePronounSelection(T.Person.FirstSingMale),
}),
},
{
key: 2,
block: {
type: "objectSelection",
selection: "none",
},
},
],
verb: {
type: "verb",
verb: tlul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: true,
shrinkServant: false,
},
},
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makePronounSelection(T.Person.FirstSingFemale),
}),
},
{
key: 2,
block: {
type: "objectSelection",
selection: "none",
},
},
],
verb: {
type: "verb",
verb: tlul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: true,
shrinkServant: false,
},
},
],
},
],
},
{
label: "past tense transitive full kernels",
cases: [
{
input: "سړي ماشومه ولیده",
output: [
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
{
key: 2,
block: makeObjectSelectionComplete({
type: "NP",
selection: {
...makeNounSelection(maashoom, undefined),
gender: "fem",
},
}),
},
],
verb: {
type: "verb",
verb: leedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "perfectivePast",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
},
],
},
],
},
{
label: "transitive full kernels",
cases: [
{
input: "سړی ماشومه ویني",
output: [
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
{
key: 2,
block: makeObjectSelectionComplete({
type: "NP",
selection: {
...makeNounSelection(maashoom, undefined),
gender: "fem",
},
}),
},
],
verb: {
type: "verb",
verb: leedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
},
{
blocks: [
{
key: 1,
block: makeObjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
{
key: 2,
block: makeSubjectSelectionComplete({
type: "NP",
selection: {
...makeNounSelection(maashoom, undefined),
gender: "fem",
},
}),
},
],
verb: {
type: "verb",
verb: leedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
},
],
},
{
input: "سړی ما ویني",
output: [
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
{
key: 2,
block: makeObjectSelectionComplete({
type: "NP",
selection: makePronounSelection(T.Person.FirstSingMale),
}),
},
],
verb: {
type: "verb",
verb: leedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
},
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
{
key: 2,
block: makeObjectSelectionComplete({
type: "NP",
selection: makePronounSelection(T.Person.FirstSingFemale),
}),
},
],
verb: {
type: "verb",
verb: leedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
},
],
},
],
},
{
label: "transitive no king",
cases: [
{
input: "سړي ولیدله",
output: [
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
{
key: 2,
block: makeObjectSelectionComplete({
type: "NP",
selection: makePronounSelection(T.Person.ThirdSingFemale),
}),
},
],
verb: {
type: "verb",
verb: leedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "perfectivePast",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: true,
shrinkServant: false,
},
},
],
},
{
input: "سړی وینم",
output: [
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makePronounSelection(T.Person.FirstSingMale),
}),
},
{
key: 2,
block: makeObjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
],
verb: {
type: "verb",
verb: leedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: true,
shrinkServant: false,
},
},
{
blocks: [
{
key: 1,
block: makeSubjectSelectionComplete({
type: "NP",
selection: makePronounSelection(T.Person.FirstSingFemale),
}),
},
{
key: 2,
block: makeObjectSelectionComplete({
type: "NP",
selection: makeNounSelection(sarey, undefined),
}),
},
],
verb: {
type: "verb",
verb: leedul,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense: "presentVerb",
canChangeVoice: true,
isCompound: false,
voice: "active",
},
externalComplement: undefined,
form: {
removeKing: true,
shrinkServant: false,
},
},
],
},
],
},
{
label: "transitive shrunken servant",
cases: [],
},
];
tests.forEach(({ label, cases }) => {
test(label, () => {
cases.forEach(({ input, output, error }) => {
const tokens = tokenizer(input);
const parsed = parseVP(tokens, lookup, verbLookup);
if (error) {
expect(parsed.filter((x) => x.errors.length).length).toBeTruthy();
} else {
expect(parsed.map((p) => removeKeys(p.body))).toIncludeSameMembers(
removeKeys(output)
);
}
});
});
});

548
src/lib/src/parsing/parse-vp.ts
View File

@ -17,7 +17,7 @@ import { isFirstOrSecondPersPronoun } from "../phrase-building/render-vp";
// make impossible subjects like I saw me, error // make impossible subjects like I saw me, error
// PROBLEM! ته وینې doesn't work cause it just takes ته as a verb phrase ? // TODO: transitivity options
export function parseVP( export function parseVP(
tokens: Readonly<T.Token[]>, tokens: Readonly<T.Token[]>,
@ -51,6 +51,19 @@ export function parseVP(
const tense = getTenseFromRootsStems(ba, verb.info.base, verb.info.aspect); const tense = getTenseFromRootsStems(ba, verb.info.base, verb.info.aspect);
const isPast = isPastTense(tense); const isPast = isPastTense(tense);
const v: T.VerbSelectionComplete = {
type: "verb",
verb: verb.info.verb,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense,
canChangeVoice: true,
isCompound: false,
voice: "active",
};
const nps = blocks.filter( const nps = blocks.filter(
(x): x is { inflected: boolean; selection: T.NPSelection } => (x): x is { inflected: boolean; selection: T.NPSelection } =>
"inflected" in x "inflected" in x
@ -61,18 +74,6 @@ export function parseVP(
return []; return [];
} }
if (nps.length === 0) { if (nps.length === 0) {
const v: T.VerbSelectionComplete = {
type: "verb",
verb: verb.info.verb,
transitivity: "intransitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense,
canChangeVoice: true,
isCompound: false,
voice: "active",
};
const blocks: T.VPSBlockComplete[] = [ const blocks: T.VPSBlockComplete[] = [
{ {
key: 1, key: 1,
@ -99,55 +100,41 @@ export function parseVP(
}, },
} as T.VPSelectionComplete); } as T.VPSelectionComplete);
} }
if (nps.length === 1) { const errors: T.ParseError[] = [];
const errors: T.ParseError[] = []; if (getPersonFromNP(nps[0].selection) !== verb.person) {
if (getPersonFromNP(nps[0].selection) !== verb.person) { errors.push({ message: "subject must agree with intransitive verb" });
errors.push({ message: "subject must agree with intransitive verb" });
}
if (nps[0].inflected) {
errors.push({
message: "subject of intransitive verb must not be inflected",
});
}
const blocks: T.VPSBlockComplete[] = [
{
key: 1,
block: makeSubjectSelectionComplete(nps[0].selection),
},
{
key: 2,
block: {
type: "objectSelection",
selection: "none",
},
},
];
const v: T.VerbSelectionComplete = {
type: "verb",
verb: verb.info.verb,
transitivity: "intransitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense,
canChangeVoice: true,
isCompound: false,
voice: "active",
};
return returnParseResult(
tokens,
{
blocks,
verb: v,
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
} as T.VPSelectionComplete,
errors
);
} }
if (nps[0].inflected) {
errors.push({
message: "subject of intransitive verb must not be inflected",
});
}
const blocks: T.VPSBlockComplete[] = [
{
key: 1,
block: makeSubjectSelectionComplete(nps[0].selection),
},
{
key: 2,
block: {
type: "objectSelection",
selection: "none",
},
},
];
return returnParseResult(
tokens,
{
blocks,
verb: v,
externalComplement: undefined,
form: {
removeKing: false,
shrinkServant: false,
},
} as T.VPSelectionComplete,
errors
);
} else { } else {
// transitive // transitive
if (nps.length > 2) { if (nps.length > 2) {
@ -158,6 +145,13 @@ export function parseVP(
} }
if (nps.length === 1) { if (nps.length === 1) {
const np = nps[0]; const np = nps[0];
// possibilities
// present:
// - no king (np is servant)
// - shrunken servant (np is king)
// past:
// - no king (np is servant)
// - shrunken servant (np is king)
return ( return (
[ [
{ {
@ -171,14 +165,20 @@ export function parseVP(
] as const ] as const
).flatMap((form) => { ).flatMap((form) => {
const errors: T.ParseError[] = []; const errors: T.ParseError[] = [];
const king: T.NPSelection = form.removeKing
? {
type: "NP",
selection: makePronounSelection(verb.person),
}
: np.selection;
const servants: T.NPSelection[] = form.shrinkServant
? getPeopleFromKids(kids).map((person) => ({
type: "NP",
selection: makePronounSelection(person),
}))
: [np.selection];
// check for vp structure errors
if (form.removeKing) { if (form.removeKing) {
// king is gone
// servant is there
const king: T.NPSelection = {
type: "NP",
selection: makePronounSelection(verb.person),
};
const servant = np.selection;
if (!isPast) { if (!isPast) {
if (isFirstOrSecondPersPronoun(np.selection)) if (isFirstOrSecondPersPronoun(np.selection))
if (!np.inflected) { if (!np.inflected) {
@ -195,7 +195,15 @@ export function parseVP(
}); });
} }
} }
const blocks: T.VPSBlockComplete[] = !isPast } else if (np.inflected) {
errors.push({
message: !isPast
? "object of a past tense transitive verb should not be inflected"
: "subject of a non-past tense transitive verb should not be inflected",
});
}
const blocksOps: T.VPSBlockComplete[][] = servants.map((servant) =>
!isPast
? [ ? [
{ {
key: 1, key: 1,
@ -215,114 +223,19 @@ export function parseVP(
key: 2, key: 2,
block: makeObjectSelectionComplete(king), block: makeObjectSelectionComplete(king),
}, },
]; ]
const v: T.VerbSelectionComplete = { );
type: "verb", return blocksOps.map((blocks) => ({
// @ts-ignore tokens,
verb: verb.info.verb, body: {
transitivity: "transitive", blocks,
canChangeTransitivity: false, verb: v,
canChangeStatDyn: false, externalComplement: undefined,
negative: false, form,
tense, } as T.VPSelectionComplete,
canChangeVoice: true, errors,
isCompound: false, }));
voice: "active",
};
return returnParseResult(
tokens,
{
blocks,
verb: v,
externalComplement: undefined,
form,
} as T.VPSelectionComplete,
errors
);
} else {
// servant is shrunken
// king is there
const king = np.selection;
const shrunkenServantPeople = getPeopleFromKids(kids);
if (!shrunkenServantPeople.length) {
return [];
}
const servants = shrunkenServantPeople.map(
(person): T.NPSelection => ({
type: "NP",
selection: makePronounSelection(person),
})
);
if (!isPast) {
if (np.inflected) {
errors.push({
message:
"object of a past tense transitive verb should not be inflected",
});
}
} else {
if (np.inflected) {
errors.push({
message:
"subject of a non-past tense transitive verb should not be inflected",
});
}
}
const blocksOps: T.VPSBlockComplete[][] = servants.map((servant) =>
!isPast
? [
{
key: 1,
block: makeSubjectSelectionComplete(king),
},
{
key: 2,
block: makeObjectSelectionComplete(servant),
},
]
: [
{
key: 1,
block: makeSubjectSelectionComplete(servant),
},
{
key: 2,
block: makeObjectSelectionComplete(king),
},
]
);
const v: T.VerbSelectionComplete = {
type: "verb",
// @ts-ignore
verb: verb.info.verb,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense,
canChangeVoice: true,
isCompound: false,
voice: "active",
};
return blocksOps.map((blocks) => ({
tokens,
body: {
blocks,
verb: v,
externalComplement: undefined,
form,
} as T.VPSelectionComplete,
errors,
}));
}
}); });
// possibilities
// present:
// - no king (np is servant)
// - shrunken servant (np is king)
// past:
// - no king (np is servant)
// - shrunken servant (np is king)
} else { } else {
if (isPast) { if (isPast) {
return ( return (
@ -363,19 +276,6 @@ export function parseVP(
if (flip) { if (flip) {
blocks = blocks.reverse(); blocks = blocks.reverse();
} }
const v: T.VerbSelectionComplete = {
type: "verb",
// @ts-ignore
verb: verb.info.verb,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense,
canChangeVoice: true,
isCompound: false,
voice: "active",
};
return returnParseResult( return returnParseResult(
tokens, tokens,
{ {
@ -438,19 +338,6 @@ export function parseVP(
if (flip) { if (flip) {
blocks = blocks.reverse(); blocks = blocks.reverse();
} }
const v: T.VerbSelectionComplete = {
type: "verb",
// @ts-ignore
verb: verb.info.verb,
transitivity: "transitive",
canChangeTransitivity: false,
canChangeStatDyn: false,
negative: false,
tense,
canChangeVoice: true,
isCompound: false,
voice: "active",
};
return returnParseResult( return returnParseResult(
tokens, tokens,
{ {
@ -468,7 +355,6 @@ export function parseVP(
} }
} }
} }
return [];
}); });
} }
@ -496,262 +382,6 @@ function getPeopleFromKids(kids: T.ParsedKid[]): T.Person[] {
return p; return p;
} }
// // how to make this into a nice pipeline... 🤔
// const NP1 = parseNP(tokens, lookup).filter(({ errors }) => !errors.length);
// const ba = bindParseResult(NP1, (tokens, np1) => {
// const b = parseBa(tokens);
// if (!b.length) {
// return [
// {
// tokens,
// body: {
// np1,
// ba: false,
// },
// errors: [],
// },
// ];
// } else {
// return b.map(({ tokens, errors }) => ({
// body: {
// np1,
// ba: true,
// },
// errors,
// tokens,
// }));
// }
// });
// const NP2 = bindParseResult<
// {
// np1: {
// inflected: boolean;
// selection: T.NPSelection;
// };
// ba: boolean;
// },
// {
// np1: {
// inflected: boolean;
// selection: T.NPSelection;
// };
// ba: boolean;
// np2:
// | {
// inflected: boolean;
// selection: T.NPSelection;
// }
// | undefined;
// }
// >(ba, (tokens, { np1, ba }) => {
// const np2s = parseNP(tokens, lookup);
// if (!np2s.length) {
// const r: T.ParseResult<{
// np1: {
// inflected: boolean;
// selection: T.NPSelection;
// };
// ba: boolean;
// np2: undefined;
// }>[] = [
// {
// tokens,
// body: {
// np1,
// np2: undefined,
// ba,
// },
// errors: [],
// },
// ];
// return r;
// }
// return np2s.map((p) => ({
// tokens: p.tokens,
// body: {
// np1,
// np2: p.body,
// ba,
// },
// errors: p.errors,
// }));
// }).filter(({ errors }) => !errors.length);
// const vb = bindParseResult(NP2, (tokens, nps) => {
// const vb = parseVerb(tokens, verbLookup);
// // TODO make a nice functor that just maps or adds in the body
// return vb.map((p) => ({
// tokens: p.tokens,
// body: {
// np2: nps.np2,
// v: p.body,
// np1: nps.np1,
// ba: nps.ba,
// },
// errors: p.errors,
// }));
// }).filter(({ errors }) => !errors.length);
// // TODO: be able to bind mulitple vals
// return bindParseResult(vb, (tokens, { np1, np2, v: [ph, v], ba }) => {
// const w: T.ParseResult<T.VPSelectionComplete>[] = [];
// if (v.info.type === "equative") {
// throw new Error("not yet implemented");
// }
// const isPast = v.info.base === "root";
// const intransitive =
// v.info.type === "verb" && v.info.verb.entry.c.includes("intrans.");
// if (intransitive) {
// if (np2) return [];
// const s = np1;
// const errors: T.ParseError[] = [];
// if (s.inflected) {
// errors.push({
// message: "subject of intransitive verb should not be inflected",
// });
// }
// if (getPersonFromNP(s.selection) !== v.person) {
// errors.push({
// message: "subject should agree with intransitive verb",
// });
// }
// const blocks: T.VPSBlockComplete[] = [
// {
// key: 1,
// block: makeSubjectSelectionComplete(s.selection),
// },
// {
// key: 2,
// block: {
// type: "objectSelection",
// selection: "none",
// },
// },
// ];
// const verb: T.VerbSelectionComplete = {
// type: "verb",
// verb: v.info.type === "verb" ? v.info.verb : kedulStat,
// transitivity: "intransitive",
// canChangeTransitivity: false,
// canChangeStatDyn: false,
// negative: false,
// tense: getTenseFromRootsStems(ba, v.info.base, v.info.aspect),
// canChangeVoice: true,
// isCompound: false,
// voice: "active",
// };
// w.push({
// tokens,
// body: {
// blocks,
// verb,
// externalComplement: undefined,
// form: {
// removeKing: false,
// shrinkServant: false,
// },
// },
// errors,
// });
// } else {
// // transitive verb
// if (!(np1 && np2)) return [];
// [[np1, np2, false] as const, [np2, np1, true] as const].forEach(
// ([s, o, reversed]) => {
// if (v.info.type === "equative") {
// throw new Error("not yet implemented");
// }
// if (!s || !o) return [];
// // TODO: check if perfective head MATCHES verb
// if (v.info.aspect === "perfective" && !ph) {
// return [];
// }
// const subjPerson = getPersonFromNP(s.selection);
// const errors: T.ParseError[] = [];
// if (intransitive) {
// return [];
// }
// if (isPast) {
// if (getPersonFromNP(o.selection) !== v.person) {
// errors.push({
// message: "transitive past tense verb does not match object",
// });
// } else {
// if (!s.inflected) {
// errors.push({
// message: "transitive past tense subject should be inflected",
// });
// }
// if (o.inflected) {
// errors.push({
// message:
// "transitive past tense object should not be inflected",
// });
// }
// }
// } else {
// if (getPersonFromNP(s.selection) !== v.person) {
// errors.push({
// message: "verb does not match subject",
// });
// } else {
// if (s.inflected) {
// errors.push({ message: "subject should not be inflected" });
// }
// if (o.selection.selection.type === "pronoun") {
// if (!isThirdPerson(subjPerson) && !o.inflected) {
// errors.push({
// message:
// "1st or 2nd person object pronoun should be inflected",
// });
// }
// } else if (o.inflected) {
// errors.push({ message: "object should not be inflected" });
// }
// }
// }
// const blocks: T.VPSBlockComplete[] = [
// {
// key: 1,
// block: makeSubjectSelectionComplete(s.selection),
// },
// {
// key: 2,
// block: makeObjectSelectionComplete(o.selection),
// },
// ];
// if (reversed) {
// blocks.reverse();
// }
// const verb: T.VerbSelectionComplete = {
// type: "verb",
// verb: v.info.type === "verb" ? v.info.verb : kedulStat,
// transitivity: "transitive",
// canChangeTransitivity: false,
// canChangeStatDyn: false,
// negative: false,
// tense: getTenseFromRootsStems(ba, v.info.base, v.info.aspect),
// canChangeVoice: true,
// isCompound: false,
// voice: "active",
// };
// w.push({
// tokens,
// body: {
// blocks,
// verb,
// externalComplement: undefined,
// form: {
// removeKing: false,
// shrinkServant: false,
// },
// },
// errors,
// });
// }
// );
// }
function getTenseFromRootsStems( function getTenseFromRootsStems(
hasBa: boolean, hasBa: boolean,
base: "root" | "stem", base: "root" | "stem",

5
src/lib/src/parsing/tokenizer.ts
View File

@ -2,7 +2,10 @@ import { Token } from "../../../types";
import { standardizePashto } from "../standardize-pashto"; import { standardizePashto } from "../standardize-pashto";
export function tokenizer(s: string): Token[] { export function tokenizer(s: string): Token[] {
const words = standardizePashto(s).trim().split(/ +/); const words = standardizePashto(s)
.trim()
.split(/ +/)
.filter((x) => x);
const indexed: { i: number; s: string }[] = []; const indexed: { i: number; s: string }[] = [];
for (let i = 0; i < words.length; i++) { for (let i = 0; i < words.length; i++) {
indexed.push({ i, s: words[i] }); indexed.push({ i, s: words[i] });

6
src/lib/src/parsing/utils.ts
View File

@ -102,3 +102,9 @@ export function isCompleteResult<C extends object>(
): boolean { ): boolean {
return !r.tokens.length && !r.errors.length; return !r.tokens.length && !r.errors.length;
} }
export function removeKeys(a: any): any {
return JSON.parse(
JSON.stringify(a, (k, v) => (k === "i" || k === "key" ? undefined : v))
);
}