rea/src/domain/javascript/htmlArtifactReferences.test.ts

import { expect, it } from "vitest";

import {
  HTML_INLINE_SCRIPT_LOCATION_LIMIT,
  htmlArtifactReferences,
} from "./htmlArtifactReferences.js";

it.each([
  ['<script data-src="decoy.js"></script>', []],
  ['<script x:src="decoy.js"></script>', []],
  [`<script title='src="decoy.js"'></script>`, []],
  ["<script src=actual.js></script>", ["actual.js"]],
  ['<script title=">" src="actual.js"></script>', ["actual.js"]],
  ['<SCRIPT SRC="actual.js"></SCRIPT>', ["actual.js"]],
  ["<script src='actual.js'></script>", ["actual.js"]],
  ['<script data-src="decoy.js" src="actual.js"></script>', ["actual.js"]],
  ['<script src="actual.js" src="decoy.js"></script>', ["actual.js"]],
  ['<script src="" src="decoy.js"></script>', []],
  ['<script src="actual&#46;js"></script>', ["actual.js"]],
  ['<!-- <script src="decoy.js"></script> -->', []],
  ['<textarea><script src="decoy.js"></script></textarea>', []],
  ['<style><script src="decoy.js"></script></style>', []],
  [`<script>const text = '<script src="decoy.js">';</script>`, []],
  [
    '<template><base href="decoy/"><script src="decoy.js"></script></template>',
    [],
  ],
  ['<svg><script src="decoy.js"></script></svg>', []],
  [
    '<svg><foreignObject><script src="actual.js"></script></foreignObject></svg>',
    ["actual.js"],
  ],
  // The existing extraction contract does not classify script MIME types.
  ['<script type="application/json" src="actual.js"></script>', ["actual.js"]],
])("extracts document references from %s", (text, expected) => {
  expect(
    htmlArtifactReferences(text).scripts.map(({ scriptPath }) => scriptPath),
  ).toEqual(expected);
});

it("retains exact opening-tag coordinates rather than the script body", () => {
  const opening = '<script\n title=">" src=actual.js>';
  const prefix = "<!doctype html>\n\n";
  const text = `${prefix}${opening}ignored body</script>`;
  expect(htmlArtifactReferences(text)).toEqual({
    baseHref: null,
    inlineScripts: [],
    inlineScriptCount: 0,
    scripts: [
      {
        scriptPath: "actual.js",
        location: {
          start: { line: 3, column: 0 },
          end: { line: 4, column: ' title=">" src=actual.js>'.length },
        },
      },
    ],
  });
});

it.each([
  ['<base data-href="decoy/"><base href=actual/>', "actual/"],
  ['<base href="actual&#47;" href="decoy/"><base href="later/">', "actual/"],
  ['<base href=""><base href="decoy/">', ""],
  ['<template><base href="decoy/"></template><base href="actual/">', "actual/"],
  ['<!-- <base href="decoy/"> --><base title=">" href="actual/">', "actual/"],
])("selects the first real base href from %s", (text, expected) => {
  expect(htmlArtifactReferences(text).baseHref).toBe(expected);
});

it("traverses deep valid document content without recursive stack growth", () => {
  const depth = 12_000;
  const text = `${"<div>".repeat(depth)}<script src=actual.js></script>${"</div>".repeat(depth)}`;
  expect(
    htmlArtifactReferences(text).scripts.map(({ scriptPath }) => scriptPath),
  ).toEqual(["actual.js"]);
});

it.each([
  ["<script>run();</script>", ["classic"]],
  ['<script type="">run();</script>', ["classic"]],
  ['<script type=" Text/JavaScript ">run();</script>', ["classic"]],
  ['<script type="application/x-javascript">run();</script>', ["classic"]],
  ['<script language="javascript">run();</script>', ["classic"]],
  ['<script language="">run();</script>', ["classic"]],
  ['<script type="MODULE">import "./x.js";</script>', ["module"]],
  // Only the JavaScript MIME check strips whitespace; `module` must match.
  ['<script type=" module ">import "./x.js";</script>', []],
  ['<script language=" javascript ">run();</script>', []],
  ["<script nomodule>run();</script>", ["classic"]],
  // A src attribute, even an empty one, means the browser ignores the body.
  ['<script src="actual.js">run();</script>', []],
  ['<script src="">run();</script>', []],
  ["<script>  \n\t </script>", []],
  ['<script type="application/json">{"a":1}</script>', []],
  ['<script type="application/ld+json">{}</script>', []],
  ['<script type="importmap">{"imports":{}}</script>', []],
  ['<script type="speculationrules">{}</script>', []],
  ['<script type="text/template"><p>x</p></script>', []],
  ['<script language="vbscript">run()</script>', []],
  ["<!-- <script>run();</script> -->", []],
  ["<template><script>run();</script></template>", []],
  ["<svg><script>run();</script></svg>", []],
])("classifies inline JavaScript in %s", (text, expected) => {
  expect(
    htmlArtifactReferences(text).inlineScripts.map(
      ({ scriptType }) => scriptType,
    ),
  ).toEqual(expected);
});

it("retains exact inline script body coordinates, including unclosed scripts", () => {
  const prefix = "<!doctype html>\n<script>";
  const body = "\nfetch('/api');\n";
  const closed = `${prefix}${body}</script><p>after</p>`;
  expect(htmlArtifactReferences(closed).inlineScripts).toEqual([
    {
      scriptType: "classic",
      location: { start: { line: 2, column: 8 }, end: { line: 4, column: 0 } },
    },
  ]);
  const unclosed = `${prefix}${body}`;
  expect(htmlArtifactReferences(unclosed).inlineScripts).toEqual([
    {
      scriptType: "classic",
      location: { start: { line: 2, column: 8 }, end: { line: 4, column: 0 } },
    },
  ]);
});

it("counts every inline body but retains bounded locations", () => {
  const count = 150_000;
  const text = "<script>x</script>".repeat(count);
  const { inlineScripts, inlineScriptCount } = htmlArtifactReferences(text);
  expect(inlineScriptCount).toBe(count);
  expect(inlineScripts).toHaveLength(HTML_INLINE_SCRIPT_LOCATION_LIMIT);
  const column = 18 * (HTML_INLINE_SCRIPT_LOCATION_LIMIT - 1) + 8;
  expect(inlineScripts.at(-1)).toEqual({
    scriptType: "classic",
    location: {
      start: { line: 1, column },
      end: { line: 1, column: column + 1 },
    },
  });
});