pashto-inflector/src/lib/pashto-inflector.ts

/**
 * Copyright (c) 2021 lingdocs.com
 *
 * This source code is licensed under the MIT license found in the
 * LICENSE file in the root directory of this source tree.
 *
 */

import { pashtoConsonants } from "./pashto-consonants";
import {
  concatInflections,
  splitDoubleWord,
  ensureUnisexInflections,
  makePsString,
  removeFVarients,
  concatPsString,
  endsInConsonant,
  endsInAaOrOo,
  addOEnding,
  endsInShwa,
  splitPsByVarients,
  removeEndTick,
  endsWith,
} from "./p-text-helpers";
import {
  accentFSylsOnNFromEnd,
  hasAccents,
  removeAccents,
  splitUpSyllables,
} from "./accent-helpers";
import * as T from "../types";
import { splitFIntoPhonemes } from "./phonetics-to-diacritics";

const endingInSingleARegex = /[^a]'?’?[aá]'?’?$/;
const endingInHeyOrAynRegex = /[^ا][هع]$/;
// const endingInAlefRegex = /اع?$/;

export function inflectWord(word: T.DictionaryEntry): T.InflectorOutput {
  // If it's a noun/adj, inflect accordingly
  // TODO: What about n. f. / adj. that end in ي ??
  const w = removeFVarients(word);
  if (w.c?.includes("doub.")) {
    const words = splitDoubleWord(w);
    const inflected = words.map((x) => ensureUnisexInflections(inflectWord(x), x));
    return {
      inflections: concatInflections(
        inflected[0].inflections,
        inflected[1].inflections,
      ) as T.UnisexInflections,
    };
  }
  if (w.c && w.c.includes("pl.")) {
    return handlePluralNoun(w);
  }
  if (w.c && (w.c.includes("adj.") || w.c.includes("unisex") || w.c.includes("num"))) {
    return handleUnisexWord(w);
  }
  if (w.c && (w.c.includes("n. m."))) {
    return handleMascNoun(w);
  }
  if (w.c && (w.c.includes("n. f."))) {
    return handleFemNoun(w);
  }
  // It's not a noun/adj
  return false;
}

// LEVEL 2 FUNCTIONS
function handleUnisexWord(word: T.DictionaryEntryNoFVars): T.InflectorOutput {
  // Get last letter of Pashto and last two letters of phonetics
  // TODO: !!! Handle weird endings / symbols ' etc.
  const pEnd = word.p.slice(-1);
  const plurals = makePlural(word);
  if (word.noInf) {
    return !plurals ? false : { ...plurals };
  }
  if (word.infap && word.infaf && word.infbp && word.infbf) {
    return {
      inflections: inflectIrregularUnisex(word.p, word.f, [
        {p: word.infap, f: word.infaf},
        {p: word.infbp, f: word.infbf},
      ]),
      ...plurals,
    };
  }
  if (pEnd === "ی" && word.f.slice(-2) === "ey") {
    return { inflections: inflectRegularYeyUnisex(word.p, word.f), ...plurals };
  }
  if (pEnd === "ه" && word.g.slice(-1) === "u") {
    return { inflections: inflectRegularShwaEndingUnisex(word.p, word.f), ...plurals };
  }
  if (pEnd === "ی" && word.f.slice(-2) === "éy") {
    return { inflections: inflectEmphasizedYeyUnisex(word.p, word.f), ...plurals };
  }
  if (
    pashtoConsonants.includes(pEnd) ||
    word.p.slice(-2) === "وی" ||
    word.p.slice(-2) === "ای" ||
    word.f.slice(-1) === "w" ||
    (word.p.slice(-1) === "ه" && word.f.slice(-1) === "h")
  ) {
    return { inflections: inflectConsonantEndingUnisex(word.p, word.f), ...plurals };
  }
  if (plurals) return plurals;
  return false;
}

function handlePluralNoun(w: T.DictionaryEntryNoFVars): T.InflectorOutput {
  if (!w.c || !w.c.includes("n.")) return false;
  const plurals = makePlural(w);
  if (w.noInf) {
    return !plurals ? false : { ...plurals };
  }
  if (!plurals) return false;
  return { ...plurals };
}

function handleMascNoun(w: T.DictionaryEntryNoFVars): T.InflectorOutput {
  // Get last letter of Pashto and last two letters of phonetics
  // TODO: !!! Handle weird endings / symbols ' etc.
  const plurals = makePlural(w);
  if (w.noInf) {
    return !plurals ? false : { ...plurals };
  }
  const pEnd = w.p.slice(-1);
  const fEnd = w.f.slice(-2);
  if (w.infap && w.infaf && w.infbp && w.infbf) {
    return {
      inflections: inflectIrregularMasc(w.p, w.f, [
        {p: w.infap, f: w.infaf},
        {p: w.infbp, f: w.infbf},
      ]),
      ...plurals,
    };
  }
  const isTobEnding = (w.p.slice(-3) === "توب" && ["tób", "tob"].includes(w.f.slice(-3)) && w.p.length > 3);
  if (isTobEnding) {
    return { inflections: inflectTobMasc(w.p, w.f), ...plurals };
  }
  if (pEnd === "ی" && fEnd === "ey") {
    return { inflections: inflectRegularYeyMasc(w.p, w.f), ...plurals };
  }
  if (pEnd === "ی" && fEnd === "éy") {
    return { inflections: inflectRegularEmphasizedYeyMasc(w.p, w.f), ...plurals };
  }
  return plurals ? { ...plurals } : false
}

function handleFemNoun(word: T.DictionaryEntryNoFVars): T.InflectorOutput {
  // Get first of comma seperated phonetics entries
  /* istanbul ignore next */ // will always have word.c at this point
  const c = word.c || "";
  const animate = c.includes("anim.");
  const pEnd = word.p.slice(-1);

  const plurals = makePlural(word);
  if (word.noInf) {
    return !plurals ? false : { ...plurals };
  }

  if (endingInHeyOrAynRegex.test(word.p) && endingInSingleARegex.test(word.f)) {
    return { inflections: inflectRegularAFem(word.p, word.f), ...plurals };
  }
  if (word.p.slice(-1) === "ح" && endingInSingleARegex.test(word.f)) {
    return { inflections: inflectRegularAWithHimPEnding(word.p, word.f), ...plurals };
  }
  // TODO: better reusable function to check if something ends with a consonant
  if ((pashtoConsonants.includes(pEnd) || word.f.slice(-1) === "w") && !animate) {
    return { inflections: inflectRegularInanMissingAFem(word.p, word.f), ...plurals };
  }
  if (pEnd === "ي" && (!animate)) {
    return { inflections: inflectRegularInanEeFem(word.p, word.f), ...plurals };
  }
  if (pEnd === "ۍ") {
    return { inflections: inflectRegularUyFem(word.p, word.f), ...plurals };
  }
  // if (endingInAlefRegex.test(word.p)) {
  //   return { inflections: inflectRegularAaFem(word.p, f) };
  // }
  return plurals ? { ...plurals } : false;
}

// LEVEL 3 FUNCTIONS
function inflectIrregularUnisex(p: string, f: string, inflections: Array<{p: string, f: string}>): T.Inflections {
  const inf1 = removeAccents(inflections[1]);
  const inf0syls = splitFIntoPhonemes(inflections[0].f);
  const inf0f = accentFSylsOnNFromEnd(inf0syls, 0);
  return {
    masc: [
      [{p, f}],
      [{p: inflections[0].p, f: inf0f, }],
      [{p: `${inf1.p}و`, f: `${inf1.f}ó`}],
    ],
    fem: [
      [{p: `${inf1.p}ه`, f: `${inf1.f}á`}],
      [{p: `${inf1.p}ې`, f: `${inf1.f}é`}],
      [{p: `${inf1.p}و`, f: `${inf1.f}ó`}],
    ],
  };
}

export function inflectRegularYeyUnisex(p: string, f: string): T.UnisexInflections {
  const baseP = p.slice(0, -1);
  const baseF = f.slice(0, -2);
  return {
    masc: [
      [{p, f}],
      [{p: `${baseP}ي`, f: `${baseF}ee`}],
      [
        {p: `${baseP}یو`, f: `${baseF}iyo`},
        {p: `${baseP}و`, f: `${baseF}o`},
      ],
    ],
    fem: [
      [{p: `${baseP}ې`, f: `${baseF}e`}],
      [{p: `${baseP}ې`, f: `${baseF}e`}],
      [{p: `${baseP}و`, f: `${baseF}o`}],
    ],
  };
}

export function inflectRegularShwaEndingUnisex(pr: string, fr: string): T.UnisexInflections {
  const { p, f } = removeAccents(makePsString(pr, fr));
  const accented = fr.slice(-1) === "ú";
  const baseP = p.slice(0, -1);
  const baseF = f.slice(0, -1);
  return {
    masc: [
      [{p: `${baseP}ه`, f: `${baseF}${accented ? "ú" : "u"}`}],
      [{p: `${baseP}ه`, f: `${baseF}${accented ? "ú" : "u"}`}],
      [{p: `${baseP}و`, f: `${baseF}${accented ? "ó" : "o"}`}],
    ],
    fem: [
      [{p: `${baseP}ه`, f: `${baseF}${accented ? "á" : "a"}`}],
      [{p: `${baseP}ې`, f: `${baseF}${accented ? "é" : "e"}`}],
      [{p: `${baseP}و`, f: `${baseF}${accented ? "ó" : "o"}`}],
    ],
  };
}

function inflectEmphasizedYeyUnisex(p: string, f: string): T.UnisexInflections {
  const baseP = p.slice(0, -1);
  const baseF = f.slice(0, -2);
  return {
    masc: [
      [{p, f}],
      [{p: `${baseP}ي`, f: `${baseF}ée`}],
      [
        {p: `${baseP}یو`, f: `${baseF}iyo`},
        {p: `${baseP}و`, f: `${baseF}ó`},
      ],
    ],
    fem: [
      [{p: `${baseP}ۍ`, f: `${baseF}úy`}],
      [{p: `${baseP}ۍ`, f: `${baseF}úy`}],
      [
        { p: `${baseP}یو`, f: `${baseF}úyo` },
        { p: `${baseP}و`, f: `${baseF}ó`, },
      ],
    ],
  };
}

function inflectConsonantEndingUnisex(p: string, f: string): T.UnisexInflections {
  const fSyls = splitUpSyllables(f);
  const iBase = fSyls.length === 1
    ? makePsString(p, accentFSylsOnNFromEnd(fSyls, 0))
    : makePsString(p, f);
  return {
    masc: [
      [{p, f}],
      [{p, f}],
      [{p: `${iBase.p}و`, f: `${iBase.f}o`}],
    ],
    fem: [
      [{p: `${iBase.p}ه`, f: `${iBase.f}a`}],
      [{p: `${iBase.p}ې`, f: `${iBase.f}e`}],
      [{p: `${iBase.p}و`, f: `${iBase.f}o`}],
    ],
  };
}

function inflectRegularYeyMasc(p: string, f: string): T.Inflections {
  const baseP = p.slice(0, -1);
  const baseF = f.slice(0, -2);
  return {
    masc: [
      [{p, f}],
      [{p: `${baseP}ي`, f: `${baseF}ee`}],
      [
        {p: `${baseP}یو`, f: `${baseF}iyo`},
        {p: `${baseP}و`, f: `${baseF}o`},
      ],
    ],
  };
}

function inflectTobMasc(p: string, f: string): T.Inflections {
  const baseP = p.slice(0, -3);
  const baseF = f.slice(0, -3);
  return {
    masc: [
      [{p, f}],
      [{p: `${baseP}تابه`, f: `${baseF}taabu`}],
      [{p: `${baseP}تبو`, f: `${baseF}tabo`}],
    ],
  };
}

function inflectRegularEmphasizedYeyMasc(p: string, f: string): T.Inflections {
  const baseP = p.slice(0, -1);
  const baseF = f.slice(0, -2);
  return {
    masc: [
      [{p, f}],
      [{p: `${baseP}ي`, f: `${baseF}ée`}],
      [
        {p: `${baseP}یو`, f: `${baseF}iyo`},
        {p: `${baseP}و`, f: `${baseF}o`},
      ],
    ],
  };
}

function inflectIrregularMasc(p: string, f: string, inflections: Array<{p: string, f: string}>): T.Inflections {
  const inf0f = splitUpSyllables(inflections[0].f).length > 1
    ? accentFSylsOnNFromEnd(inflections[0].f, 0)
    : inflections[0].f
  return {
    masc: [
      [{p, f}],
      [{p: inflections[0].p, f: inf0f}],
      [{p: `${inflections[1].p}و`, f: `${removeAccents(inflections[1].f)}ó`}],
    ],
  };
}

function inflectRegularAFem(p: string, f: string): T.Inflections {
  const withoutTrailingComma = ["'", "’"].includes(f.slice(-1)) ? f.slice(0, -1) : f;
  const accentLast = hasAccents(withoutTrailingComma.slice(-1));
  const baseF = withoutTrailingComma.slice(0, -1);
  const baseP = p.slice(-1) === "ع" ? p : p.slice(0, -1);
  return {
    fem: [
      [{p, f}],
      [{p: `${baseP}ې`, f: `${baseF}${accentLast ? "é" : "e"}`}],
      [{p: `${baseP}و`, f: `${baseF}${accentLast ? "ó" : "o"}`}],
    ],
  };
}

function inflectRegularAWithHimPEnding(p: string, f: string): T.Inflections {
  const baseF = f.slice(0, -1);
  return {
    fem: [
      [{p, f}],
      [{p: `${p}ې`, f: `${baseF}e`}],
      [{p: `${p}و`, f: `${baseF}o`}],
    ],
  };
}

function inflectRegularInanMissingAFem(p: string, f: string): T.Inflections {
  const fBase = splitUpSyllables(f).length === 1
    ? accentFSylsOnNFromEnd(f, 0)
    : f;
    return {
    fem: [
      [{p, f}],
      [{p: `${p}ې`, f: `${fBase}e`}],
      [{p: `${p}و`, f: `${fBase}o`}],
    ],
  };
}

function inflectRegularInanEeFem(p: string, f: string): T.Inflections {
  const baseP = p.slice(0, -1);
  const baseF = f.slice(0, -2);
  return {
    fem: [
      [{p, f}],
      [{p: `${baseP}ۍ`, f: `${baseF}úy`}],
      [{p: `${baseP}یو`, f: `${baseF}úyo`}],
    ],
  };
}

function inflectRegularUyFem(p: string, f: string): T.Inflections {
  const baseP = p.slice(0, -1);
  const baseF = removeAccents(f.slice(0, -2));
  return {
    fem: [
      [{p, f: `${baseF}úy`}],
      [{p, f: `${baseF}úy`}],
      [
        {p: `${baseP}یو`, f: `${baseF}úyo`},
        {p: `${baseP}و`, f: `${baseF}o`},
      ],
    ],
  };
}

function makePashtoPlural(word: T.DictionaryEntryNoFVars): T.PluralInflections | undefined {
  if (!(word.ppp && word.ppf)) return undefined;
  const base = splitPsByVarients(
    makePsString(word.ppp, word.ppf)
  );
  function getBaseAndO(): T.PluralInflectionSet {
    return [
      base,
      base.flatMap(addOEnding) as T.ArrayOneOrMore<T.PsString>,
    ];
  }
  if (word.c?.includes("n. m.")) {
    return { masc: getBaseAndO() };
  }
  if (word.c?.includes("n. f.")) {
    return { fem: getBaseAndO() };
  }
  // TODO: handle masculine and unisex
  return undefined;
}

function makeArabicPlural(word: T.DictionaryEntryNoFVars): T.PluralInflections | undefined {
  if (!(word.apf && word.app)) return undefined;
  const w = makePsString(word.app, word.apf);
  const plural = splitPsByVarients(w);
  const end = removeAccents(removeEndTick(word.apf).slice(-1));
  // again typescript being dumb and not letting me use a typed key here
  const value = [
    plural,
    plural.flatMap(addOEnding) as T.ArrayOneOrMore<T.PsString>,
  ] as T.PluralInflectionSet;
  // feminine words that have arabic plurals stay feminine with the plural - ie مرجع - مراجع
  // but masculine words that appear feminine in the plural aren't femening with the Arabic plural - ie. نبي - انبیا
  if (["i", "e", "a"].includes(end) && word.c?.includes("n. f.")) {
    return { fem: value };
  }
  return { masc: value };
}

function makePlural(w: T.DictionaryEntryNoFVars): { plural: T.PluralInflections } | { arabicPlural: T.PluralInflections } | undefined {
  function addSecondInf(plur: T.ArrayOneOrMore<T.PsString> | T.PsString): T.PluralInflectionSet {
    if (!Array.isArray(plur)) {
      return addSecondInf([plur]);
    }
    return [
      plur,
      plur.flatMap(addOEnding) as T.ArrayOneOrMore<T.PsString>,
    ];
  }
  if (w.c && w.c.includes("pl.")) {
    const plural = addSecondInf(makePsString(w.p, w.f));
    // Typescript being dumb and not letting me do a typed variable for the key
    // could try refactoring with an updated TypeScript dependency
    if (w.c.includes("n. m.")) return { plural: { masc: plural }};
    if (w.c.includes("n. f.")) return { plural: { fem: plural }};
  }
  const arabicPlural = makeArabicPlural(w);
  const pashtoPlural = makePashtoPlural(w);
  function addMascPluralSuffix(animate?: boolean, shortSquish?: boolean): T.PluralInflectionSet {
    if (shortSquish && (w.infap === undefined || w.infaf === undefined)) {
      throw new Error(`no irregular inflection info for ${w.p} - ${w.ts}`);
    }
    const b = removeAccents(shortSquish
      ? makePsString((w.infap as string).slice(0, -1), (w.infaf as string).slice(0, -1))
      : w
    );
    const base = endsInShwa(b)
      ? makePsString(b.p.slice(0, -1), b.f.slice(0, -1))
      : b;
    return addSecondInf(
      concatPsString(base, (animate && !shortSquish) ? { p: "ان", f: "áan" } : { p: "ونه", f: "óona" }),
    );
  } 
  function addAnimUnisexPluralSuffix(): T.UnisexSet<T.PluralInflectionSet> {
    const base = removeAccents(w);
    return {
      masc: addMascPluralSuffix(true),
      fem: addSecondInf(concatPsString(base, { p: "انې", f: "áane" })),
    };
  }
  function addEePluralSuffix(gender: T.Gender): T.PluralInflectionSet {
    const b = removeAccents(w);
    const base = {
      p: b.p.slice(0, -1),
      f: b.f.slice(0, -2),
    };
    const firstInf: T.ArrayOneOrMore<T.PsString> = [
      concatPsString(base, { p: "یان", f: "iyáan" }, gender === "fem" ? { p: "ې", f: "e" } : ""),
      ...gender === "fem" 
        ? [concatPsString(base, { p: "یګانې", f: "eegáane" })]
        : [],
    ];
    return [
      firstInf,
      firstInf.flatMap(addOEnding),
      // firstInf.map(addOEnding),
    ] as T.PluralInflectionSet;
  }
  function addAnimN3UnisexPluralSuffix(): T.UnisexSet<T.PluralInflectionSet> {
    const b = removeAccents(w);
    const base = {
      p: b.p.slice(0, -1),
      f: b.f.slice(0, -2),
    };
    return {
      masc: [
        [concatPsString(base, { p: "یان", f: "iyáan" })],
        [concatPsString(base, { p: "یانو", f: "iyáano" })],
        // TODO: or use addSecondInf method above?
      ],
      fem: [
        [concatPsString(base, { p: "یانې", f: "iyáane" })],
        [concatPsString(base, { p: "یانو", f: "iyáano" })],
      ],
    };
  }
  function addFemLongVowelSuffix(): T.PluralInflectionSet {
    const base = removeEndTick(makePsString(w.p, w.f));
    const baseWOutAccents = removeAccents(base);
    const space = (w.p.slice(-1) === "ع" || w.p.slice(-1) === "ه") ? { p: " ", f: "" } : "";
    return addSecondInf([
      concatPsString(base, space, { p: "وې", f: "we" }),
      concatPsString(baseWOutAccents, space, { p: "ګانې", f: "gáane" })
    ]);
  }
  // TODO: This should be possible for words like پلویان but not for words like ترورزامن 🤔
  // function addFemToPashtoPlural(i: T.PluralInflections): T.UnisexSet<T.PluralInflectionSet> {
  //   if ("fem" in i && "masc" in i) return i;
  //   if (!("masc" in i)) throw new Error("bad pashto plural doesn't even have masculine");
  //   if (endsInConsonant(i.masc[0][0])) {
  //     return {
  //       ...i,
  //       fem: [
  //         i.masc[0].map((x) => concatPsString(x, { p: "ې", f: "e" })) as T.ArrayOneOrMore<T.PsString>,
  //         i.masc[0].map((x) => concatPsString(x, { p: "و", f: "o" })) as T.ArrayOneOrMore<T.PsString>,
  //       ],
  //     };
  //   }
  //   return {
  //     ...i,
  //     fem: i.masc,
  //   };
  // }

  const shortSquish = !!w.infap && !w.infap.includes("ا");
  const anim = w.c?.includes("anim.");
  const type = (w.c?.includes("unisex"))
    ? "unisex noun"
    : (w.c?.includes("n. m."))
    ? "masc noun"
    : (w.c?.includes("n. f."))
    ? "fem noun"
    : "other";
  if (pashtoPlural) return {
    plural: pashtoPlural,
    arabicPlural,
  }; 
  if (type === "unisex noun") {
    // doesn't need to be labelled anim - because it's only with animate nouns that you get the unisex - I THINK
    if (endsInConsonant(w) && (!w.infap)) {
      return { arabicPlural, plural: addAnimUnisexPluralSuffix() };
    }
    if (shortSquish && !anim) {
      return { arabicPlural, plural: { masc: addMascPluralSuffix(anim, shortSquish) }};
    }
    if (endsWith([{ p: "ی", f: "éy" }, { p: "ي" }], w, true)) {
      return { arabicPlural, plural: addAnimN3UnisexPluralSuffix() };
    }
    // usually shortSquish nouns would never have arabicPlurals -- so we don't have to worry about catching
    // arabic plurals for the animat ones, right?
  }
  if (
      type === "masc noun" &&
      (shortSquish || ((endsInConsonant(w) || endsInShwa(w)) && (!w.infap))) &&
      (w.p.slice(-3) !== "توب")
    ) {
    return {
      arabicPlural,
      plural: {
        masc: addMascPluralSuffix(anim, shortSquish),
      },
    };
  }
  if (
    type === "masc noun" &&
    endsWith({ p: "ی", f: "éy" }, w, true) &&
    anim
  ) {
    const { masc } = addAnimN3UnisexPluralSuffix();
    return {
      arabicPlural,
      plural: {
        masc,
      },
    };
  }
  if (type === "masc noun" && endsWith({ p: "ي" }, w)) {
    const masc = addEePluralSuffix("masc");
    return {
      arabicPlural,
      plural: { masc },
    };
  }
  // TODO: What about endings in long ee / animate at inanimate
  if (type === "fem noun" && endsInAaOrOo(w) && (!w.infap)) {
    return {
      arabicPlural,
      plural: {
        fem: addFemLongVowelSuffix(),
      },
    };
  }
  if (type === "fem noun" && endsWith({ p: "ي" }, w)) {
    return {
      arabicPlural,
      plural: {
        fem: addEePluralSuffix("fem"),
      },
    };
  }
  if (arabicPlural) {
    return { arabicPlural, plural: pashtoPlural };
  }
  return undefined;
}