Scrape VejînLex to find the parts of speech of a Kurdish word and return the results as an array

README.md

Usage

import { getPos } from "./getPos.ts";

const pos = await getPos("سڵاو");

console.log(pos); // Expected result: [ "noun" ]

getPos.ts

import { DOMParser } from "https://deno.land/x/deno_dom@v0.1.30-alpha/deno-dom-wasm.ts";

export async function getPos(word: string) {
  return [
    ...new Set(
      [
        ...(new DOMParser()
          .parseFromString(
            await (
              await fetch(
                `https://lex.vejin.net/ck/search?${new URLSearchParams({
                  t: word,
                  f: "0",
                  d: "3",
                })}`,
              )
            ).text(),
            "text/html",
          )
          ?.querySelectorAll(".wah > span.tag.pos") || []),
      ]
        // deno-lint-ignore no-explicit-any
        .map((v: any) => v.innerText)
        .map((v) => v.replace("and", ""))
        .map((v) => v.split(","))
        .flat()
        .map((v) => v.split(/\s/))
        .flat()
        .map((v) => v.trim())
        .filter((v) => v),
    ),
  ];
}