InkNote/frontend/src/search.test.ts

import { describe, it, expect, beforeEach } from "vitest";

import * as search from "./search";
import { serializeNoteContent } from "./editor/content";

const body = (text: string) =>
  serializeNoteContent({
    type: "doc",
    content: [{ type: "paragraph", content: [{ type: "text", text }] }],
  });

const seed = () => {
  search.removeEntry("a");
  search.removeEntry("b");
  search.removeEntry("c");
  search.setEntry("a", "Haushalt", body("Der Kühlschrank braucht eine neue Dichtung."));
  search.setEntry("b", "Finanzen", body("Steuererklärung abgeben, Belege liegen im Ordner."));
  search.setEntry("c", "Dichtung", body("Nur der Titel enthält das Wort."));
};

beforeEach(seed);

describe("search", () => {
  it("returns null for an empty query, meaning 'no search'", () => {
    expect(search.search("")).toBeNull();
    expect(search.search("   ")).toBeNull();
  });

  it("finds a word that appears only in the body", () => {
    const hits = search.search("Belege");
    expect(hits?.map((h) => h.id)).toEqual(["b"]);
  });

  it("ranks a title match above a body match", () => {
    const hits = search.search("Dichtung");
    expect(hits?.map((h) => h.id)).toEqual(["c", "a"]);
  });

  it("requires every term to match", () => {
    expect(search.search("Belege Ordner")?.map((h) => h.id)).toEqual(["b"]);
    expect(search.search("Belege Kühlschrank")).toEqual([]);
  });

  it("ignores case", () => {
    expect(search.search("BELEGE")?.map((h) => h.id)).toEqual(["b"]);
  });

  it("ignores diacritics, so an ASCII keyboard still finds umlauts", () => {
    expect(search.search("kuhlschrank")?.map((h) => h.id)).toEqual(["a"]);
    expect(search.search("Steuererklarung")?.map((h) => h.id)).toEqual(["b"]);
  });

  it("returns a snippet around the match", () => {
    const hit = search.search("Dichtung")?.find((h) => h.id === "a");
    expect(hit?.snippet).toMatch(/Dichtung/);
  });

  it("respects the result limit", () => {
    for (let i = 0; i < 30; i++) search.setEntry(`x${i}`, `Note ${i}`, body("common word"));
    expect(search.search("common", { limit: 5 })?.length).toBe(5);
    for (let i = 0; i < 30; i++) search.removeEntry(`x${i}`);
  });

  it("stops matching a note once it is removed from the index", () => {
    expect(search.search("Belege")?.length).toBe(1);
    search.removeEntry("b");
    expect(search.search("Belege")).toEqual([]);
  });

  it("reflects an updated body immediately", () => {
    search.setEntry("a", "Haushalt", body("Völlig anderer Inhalt jetzt."));
    expect(search.search("Kühlschrank")).toEqual([]);
    expect(search.search("anderer")?.map((h) => h.id)).toEqual(["a"]);
  });
});

describe("attached document text", () => {
  /* A PDF note has two sources of text, written at different moments by
   * different code. Before these, whichever ran last was the only searchable
   * half — typing a note dropped the document, re-reading it dropped the note. */
  it("finds a note by a word from its attached document", () => {
    search.setEntry("p1", "Handbuch", body("Meine eigene Notiz."));
    search.setAttachedText("p1", "Handbuch", "Der Motor hat eine Zündkerze.");
    expect(search.search("Zündkerze")?.map((h) => h.id)).toEqual(["p1"]);
    search.removeEntry("p1");
  });

  it("and by a word the user typed beside it", () => {
    search.setEntry("p1", "Handbuch", body("Meine eigene Notiz."));
    search.setAttachedText("p1", "Handbuch", "Der Motor hat eine Zündkerze.");
    expect(search.search("eigene")?.map((h) => h.id)).toEqual(["p1"]);
    search.removeEntry("p1");
  });

  it("keeps the document text when the note is edited afterwards", () => {
    search.setAttachedText("p1", "Handbuch", "Der Motor hat eine Zündkerze.");
    search.setEntry("p1", "Handbuch", body("Ganz neuer Notiztext."));
    expect(search.search("Zündkerze")?.map((h) => h.id)).toEqual(["p1"]);
    expect(search.search("neuer")?.map((h) => h.id)).toEqual(["p1"]);
    search.removeEntry("p1");
  });

  it("keeps the note when the document is re-read afterwards", () => {
    search.setEntry("p1", "Handbuch", body("Meine eigene Notiz."));
    search.setAttachedText("p1", "Handbuch", "Der Motor hat eine Zündkerze.");
    search.setAttachedText("p1", "Handbuch", "Der Motor hat eine Zündkerze und einen Riemen.");
    expect(search.search("eigene")?.map((h) => h.id)).toEqual(["p1"]);
    expect(search.search("Riemen")?.map((h) => h.id)).toEqual(["p1"]);
    search.removeEntry("p1");
  });

  it("shows the sentence a document hit was found in, not an unrelated excerpt", () => {
    search.setEntry("p1", "Handbuch", body("Kurze Notiz."));
    search.setAttachedText("p1", "Handbuch", "Vorher. Der Motor hat eine Zündkerze. Nachher.");
    const hit = search.search("Zündkerze")?.[0];
    expect(hit?.snippet).toMatch(/Zündkerze/);
    search.removeEntry("p1");
  });

  it("clears the attached text when it is set to empty", () => {
    search.setEntry("p1", "Handbuch", body("Notiz."));
    search.setAttachedText("p1", "Handbuch", "Zündkerze");
    search.setEntry("p1", "Handbuch", body("Notiz."), "");
    expect(search.search("Zündkerze")).toEqual([]);
    search.removeEntry("p1");
  });
});

describe("contentToText", () => {
  it("reads a v2 document", () => {
    expect(search.contentToText(body("hello world"))).toBe("hello world");
  });

  it("reads a v1 body, HTML and task list alike", () => {
    const v1 = JSON.stringify({
      html: "<p>legacy <b>text</b></p>",
      tasks: [{ id: "t", text: "buy milk", done: false }],
    });
    const text = search.contentToText(v1);
    expect(text).toContain("legacy");
    expect(text).toContain("buy milk");
  });

  it("does not execute or leak markup from a v1 body", () => {
    const text = search.contentToText(JSON.stringify({ html: "<p>safe</p><script>bad()</script>" }));
    expect(text).toContain("safe");
    expect(text).not.toContain("<script>");
  });

  it("survives content that is not valid JSON", () => {
    expect(() => search.contentToText("<p>bare html</p>")).not.toThrow();
    expect(search.contentToText("<p>bare html</p>")).toContain("bare html");
  });

  it("returns an empty string for empty input", () => {
    expect(search.contentToText(null)).toBe("");
    expect(search.contentToText(undefined)).toBe("");
    expect(search.contentToText("")).toBe("");
  });
});

describe("notes pinned into a PDF", () => {
  beforeEach(() => {
    seed();
    search.removeEntry("pdf");
  });

  it("are searchable, even though they live outside the note body", () => {
    // They are stored in the page overlay, so nothing else puts them in the
    // index — and a note written onto page 14 that cannot be found again is
    // not much of a note.
    search.setEntry("pdf", "Handbuch", body("Meine Anmerkungen."));
    search.setPinnedText("pdf", "Handbuch", "S. 14: Achtung, Drehmoment beachten");
    expect(search.search("drehmoment")?.map((h) => h.id)).toEqual(["pdf"]);
  });

  it("show the pinned text as the excerpt, with its page", () => {
    search.setEntry("pdf", "Handbuch", body("Meine Anmerkungen."));
    search.setPinnedText("pdf", "Handbuch", "S. 14: Achtung, Drehmoment beachten");
    expect(search.search("drehmoment")?.[0]?.snippet).toContain("S. 14");
  });

  it("do not displace what the reader typed beside the document", () => {
    search.setEntry("pdf", "Handbuch", body("Meine Anmerkungen."));
    search.setPinnedText("pdf", "Handbuch", "Drehmoment");
    expect(search.search("anmerkungen")?.map((h) => h.id)).toEqual(["pdf"]);
    expect(search.search("drehmoment")?.map((h) => h.id)).toEqual(["pdf"]);
  });

  it("survive the document's own text being indexed afterwards", () => {
    search.setPinnedText("pdf", "Handbuch", "Drehmoment");
    search.setAttachedText("pdf", "Handbuch", "Der Hersteller empfiehlt regelmäßige Wartung.");
    expect(search.search("drehmoment")?.map((h) => h.id)).toEqual(["pdf"]);
    expect(search.search("wartung")?.map((h) => h.id)).toEqual(["pdf"]);
  });

  it("survive the note body being saved afterwards", () => {
    // Typing in the panel beside the PDF must not wipe what is pinned into it.
    search.setPinnedText("pdf", "Handbuch", "Drehmoment");
    search.setEntry("pdf", "Handbuch", body("Später weiterlesen."));
    expect(search.search("drehmoment")?.map((h) => h.id)).toEqual(["pdf"]);
  });

  it("are replaced wholesale, so a deleted pin stops being findable", () => {
    search.setPinnedText("pdf", "Handbuch", "Drehmoment");
    search.setPinnedText("pdf", "Handbuch", "");
    expect(search.search("drehmoment")).toEqual([]);
  });
});