spb/social-runtime-crawler

1import { describe, expect, it } from "vitest";2import { parseCount, parseDuration, tokenize, canonicalUrl } from "./util.ts";34describe("parseCount", () => {5  it("handles fr/en thousands, decimals, suffixes and unicode spaces", () => {6    expect(parseCount("3 624 visionnements")).toBe(3624);7    expect(parseCount("3 624 vues")).toBe(3624);8    expect(parseCount("12,345 views")).toBe(12345);9    expect(parseCount("1,2 M de vues")).toBe(1_200_000);10    expect(parseCount("48 k vues")).toBe(48_000);11    expect(parseCount("483")).toBe(483);12    expect(parseCount("17 k")).toBe(17_000);13    expect(parseCount("no number")).toBeUndefined();14  });15});16describe("parseDuration", () => {17  it("parses mm:ss and h:mm:ss", () => {18    expect(parseDuration("12:34")).toBe(754);19    expect(parseDuration("1:02:03")).toBe(3723);20  });21});22describe("tokenize", () => {23  it("keeps 2-letter domain tokens and drops stopwords", () => {24    const t = tokenize("Discover public Quebec creators discussing AI");25    expect(t.has("ai")).toBe(true);26    expect(t.has("quebec")).toBe(true);27    expect(t.has("public")).toBe(false);28  });29});30describe("canonicalUrl", () => {31  it("strips tracking params and www", () => {32    expect(canonicalUrl("https://www.youtube.com/watch?v=abc&si=xyz&t=10")).toBe("https://youtube.com/watch?v=abc");33  });34});35