Skip to content

DeepTutor: Lifelong Personalized Tutoring. https://deeptutor.info/.

16.8 KB
Raw Blame
import test from "node:test";
import assert from "node:assert/strict";
import {
  appendWithEmphasisRepair,
  decodeEscapedUnicodeForDisplay,
  escapeUnknownHtmlTagsForDisplay,
  hasVisibleMarkdownContent,
  markdownUrlTransform,
  normalizeMarkdownForDisplay,
  repairChineseEmphasis,
  repairMalformedStrongEmphasis,
  safeDecodeURIComponent,
} from "../lib/markdown-display";

test("repairChineseEmphasis repairs punctuation boundaries only for Chinese", () => {
  assert.equal(repairChineseEmphasis("中文*,重点*", "zh-CN"), "中文 *,重点*");
  assert.equal(
    repairChineseEmphasis("中文**,重点**", "zh"),
    "中文 **,重点**",
  );
  assert.equal(repairChineseEmphasis("中文*,重点*", "en"), "中文*,重点*");
  assert.equal(repairChineseEmphasis("中文*,重点*"), "中文*,重点*");
});

test("repairChineseEmphasis balances the opposite delimiter when exactly one boundary is repaired", () => {
  assert.equal(
    repairChineseEmphasis("中文*,重点*内容", "zh"),
    "中文 *,重点* 内容",
  );
});

test("repairChineseEmphasis is fed the complete raw stream, not prior display output", () => {
  const chunks = ["中文*,重", "点*", "内容"];
  const rawContent = chunks.join("");

  assert.equal(repairChineseEmphasis(rawContent, "zh"), "中文 *,重点* 内容");
  assert.equal(
    repairChineseEmphasis(
      repairChineseEmphasis(chunks.slice(0, 2).join(""), "zh") + chunks[2],
      "zh",
    ),
    "中文 *,重点*内容",
  );
});
test("repairChineseEmphasis leaves inner whitespace and literal stars unchanged", () => {
  const inputs = [
    "* 重点 *",
    "** 重点 **",
    "*重点 内容*",
    "前言 * 第一项 * 与 * 第二项 * 后记",
    "Use * as multiplication *",
    String.raw`\* 重点 \*`,
  ];
  for (const input of inputs) {
    assert.equal(repairChineseEmphasis(input, "zh"), input);
  }
});

test("repairChineseEmphasis leaves code, math, escaped, and triple-star spans untouched", () => {
  const input = [
    "`中文*,重点*`",
    "```md",
    "中文**,重点**",
    "```",
    "$中文*,重点*$",
    "\\中文\\*,重点\\*",
    "中文***,重点***内容",
  ].join("\n");
  assert.equal(repairChineseEmphasis(input, "zh"), input);
});

test("repairMalformedStrongEmphasis moves label whitespace outside the closing marker", () => {
  assert.equal(
    repairMalformedStrongEmphasis("**發布日期: **2026 年 7 月 30 日"),
    "**發布日期:** 2026 年 7 月 30 日",
  );
});

test("repairMalformedStrongEmphasis preserves valid and incomplete Markdown", () => {
  const inputs = [
    "**發布日期:** 2026 年 7 月 30 日",
    "Use **bold text** normally.",
    "**發布日期: 2026 年 7 月 30 日",
    "**發布日期: **",
  ];

  for (const input of inputs) {
    assert.equal(repairMalformedStrongEmphasis(input), input);
  }
});

test("repairMalformedStrongEmphasis leaves code and math spans untouched", () => {
  const input = [
    "`**label: **value`",
    "",
    "```md",
    "**label: **value",
    "```",
    "",
    "$\\text{**label: **value}$",
    "",
    "\\[",
    "**label: **value",
    "\\]",
  ].join("\n");

  assert.equal(repairMalformedStrongEmphasis(input), input);
});

test("repairMalformedStrongEmphasis leaves lines whose ** markers don't pair off", () => {
  // Each of these renders correctly today; pairing the first two ``**`` would
  // break emphasis the renderer already gets right.
  const inputs = [
    // Prose that mentions ** literally, followed by a real bold span.
    "In Markdown, use ** to make text **bold**.",
    // A malformed label immediately followed by a legitimate bold span:
    // repairing the label would leave a stray ** behind.
    "**Note: **Important**",
    // Nested strong emphasis, which CommonMark renders as all-bold.
    "**重點 **必讀** 內容**",
  ];

  for (const input of inputs) {
    assert.equal(repairMalformedStrongEmphasis(input), input);
  }
});

test("repairMalformedStrongEmphasis leaves indented code blocks verbatim", () => {
  const input = "Example:\n\n    **label: **value";

  assert.equal(repairMalformedStrongEmphasis(input), input);
});

test("repairMalformedStrongEmphasis repairs multiple occurrences idempotently", () => {
  const input = "**Date: **2026 and **Source: **Official";
  const expected = "**Date:** 2026 and **Source:** Official";
  const repaired = repairMalformedStrongEmphasis(input);

  assert.equal(repaired, expected);
  assert.equal(repairMalformedStrongEmphasis(repaired), expected);
});

test("normalizeMarkdownForDisplay removes empty details blocks", () => {
  const input = "Before\n\n<details><summary></summary></details>\n\nAfter";
  assert.equal(normalizeMarkdownForDisplay(input), "Before\n\nAfter");
});

test("normalizeMarkdownForDisplay repairs missing ATX heading separators", () => {
  const input = ["##Title", "###第一部分", "# Already spaced", "#######"].join(
    "\n",
  );
  const expected = [
    "## Title",
    "### 第一部分",
    "# Already spaced",
    "#######",
  ].join("\n");

  assert.equal(normalizeMarkdownForDisplay(input), expected);
  assert.equal(normalizeMarkdownForDisplay(expected), expected);
});

test("normalizeMarkdownForDisplay leaves heading-like fenced code verbatim", () => {
  const input = [
    "```c",
    "##define FEATURE",
    "```",
    "",
    "~~~text",
    "###literal",
    "~~~",
    "",
    "```python",
    "##unfinished",
  ].join("\n");

  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay leaves raw HTML code blocks verbatim", () => {
  const input = [
    "<pre",
    '  class="example">',
    "###literal",
    "</pre>",
    "",
    "<code>",
    "##define FEATURE",
    "</code>",
    "",
    "###Heading",
  ].join("\n");
  const expected = input.replace("###Heading", "### Heading");

  assert.equal(normalizeMarkdownForDisplay(input), expected);
});

test("normalizeMarkdownForDisplay does not mistake code-like prose for HTML blocks", () => {
  const inputs = [
    "`<pre>`\n###Heading",
    String.raw`\<pre>` + "\n###Heading",
    "<!-- example: <pre> -->\n###Heading",
    "<[email protected]>\n###Heading",
  ];

  for (const input of inputs) {
    assert.equal(
      normalizeMarkdownForDisplay(input),
      input.replace("###Heading", "### Heading"),
    );
  }
});

test("normalizeMarkdownForDisplay decodes dense non-ASCII JSON escapes", () => {
  const input = "\\u300c\\u6570\\u5236\\u8f6c\\u6362\\u300d";
  assert.equal(normalizeMarkdownForDisplay(input), "「数制转换」");
  assert.equal(decodeEscapedUnicodeForDisplay(input), "「数制转换」");
});

test("decoded unicode cannot reintroduce invisible or bidi controls", () => {
  assert.equal(normalizeMarkdownForDisplay("\\u200b\\u4e2d\\u6587"), "中文");
  assert.equal(normalizeMarkdownForDisplay("\\u202e\\u0061\\u0062"), "ab");
});

test("normalizeMarkdownForDisplay keeps isolated and ASCII unicode escape examples", () => {
  const inputs = [
    "A JSON string can encode A as \\u0041.",
    "Three ASCII escapes: \\u0041\\u0042\\u0043.",
  ];

  for (const input of inputs) {
    assert.equal(normalizeMarkdownForDisplay(input), input);
  }
});

test("normalizeMarkdownForDisplay keeps unicode escapes inside code verbatim", () => {
  const input = [
    "Escaped text: `\\u300c\\u6570\\u5236\\u8f6c\\u6362\\u300d`",
    "",
    "```json",
    '"label": "\\u300c\\u6570\\u5236\\u8f6c\\u6362\\u300d"',
    "```",
  ].join("\n");

  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay keeps unicode escapes in indented code verbatim", () => {
  const input =
    'Example:\n\n    "label": "\\u300c\\u6570\\u5236\\u8f6c\\u6362\\u300d"';

  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay removes raw html control placeholders", () => {
  const input =
    'Before\n\n<progress></progress>\n<input type="text" />\n<textarea> </textarea>\n\nAfter';
  assert.equal(normalizeMarkdownForDisplay(input), "Before\n\nAfter");
});

test("normalizeMarkdownForDisplay removes empty markdown tables", () => {
  const input = "Before\n\n| |\n|---|\n\nAfter";
  assert.equal(normalizeMarkdownForDisplay(input), "Before\n\nAfter");
});

test("normalizeMarkdownForDisplay removes empty html tables", () => {
  const input = "Before\n\n<table><tr><td>&nbsp;</td></tr></table>\n\nAfter";
  assert.equal(normalizeMarkdownForDisplay(input), "Before\n\nAfter");
});

test("normalizeMarkdownForDisplay keeps meaningful tables", () => {
  const input = "Before\n\n| Topic |\n|---|\n| Math |\n\nAfter";
  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay linkifies bare citations in prose", () => {
  assert.equal(
    normalizeMarkdownForDisplay("Reference [1]."),
    'Reference [1](#references "citation").',
  );
});

test("normalizeMarkdownForDisplay links research citations to exact references", () => {
  assert.equal(
    normalizeMarkdownForDisplay(
      "Agentic loops [CIT-1-01] and plans [PLAN-01].",
    ),
    'Agentic loops [1](#ref-cit-1-01 "citation") and plans [2](#ref-plan-01 "citation").',
  );
});

test("normalizeMarkdownForDisplay numbers research citations from reference list order", () => {
  const refs =
    '<details id="references" open><summary>参考资料</summary><ol>' +
    '<li id="ref-cit-1-01" data-citation-id="CIT-1-01">' +
    "<strong>[1]</strong> <code>CIT-1-01</code> A</li>" +
    '<li id="ref-cit-2-01" data-citation-id="CIT-2-01">' +
    "<strong>[2]</strong> <code>CIT-2-01</code> B</li>" +
    "</ol></details>";
  const input = `First [CIT-2-01], then [CIT-1-01].\n\n${refs}`;
  assert.equal(
    normalizeMarkdownForDisplay(input),
    `First [2](#ref-cit-2-01 "citation"), then [1](#ref-cit-1-01 "citation").\n\n${refs}`,
  );
});

test("normalizeMarkdownForDisplay keeps array indexes inside fenced code", () => {
  const input = "```js\nconst item = values[0];\n```";
  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay keeps array indexes inside inline code", () => {
  const input = "Use `values[0]` for the first item.";
  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay keeps bracketed vectors inside display math", () => {
  const input = ["The row for `sat` is:", "", "\\[", "[1, 1, 2]", "\\]"].join(
    "\n",
  );
  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay keeps bracketed vectors inside weighted math", () => {
  const input = ["\\[", "0.212[1, 1] + 0.212[2, 0] + 0.576[0, 3]", "\\]"].join(
    "\n",
  );
  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay keeps number arrays in prose untouched", () => {
  const input = "线性卷积结果 [1, 5, 9, 5, 3, 2, 7] 与 [8, 5, 3, 6]。";
  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay keeps number arrays inside inline math", () => {
  const input = "序列 $x = [1, 5, 9, 5, 3, 2, 7]$ 与 $h = [8, 5, 3, 6]$。";
  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay still linkifies small distinct numeric citation groups", () => {
  assert.equal(
    normalizeMarkdownForDisplay("See [1, 3] for details."),
    'See [1, 3](#references "citation") for details.',
  );
});

test("normalizeMarkdownForDisplay keeps backticked number arrays as code", () => {
  const input = "Result `[1, 5, 9, 5, 3, 2, 7]` here.";
  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("normalizeMarkdownForDisplay unwraps explicit citation code spans outside code", () => {
  assert.equal(
    normalizeMarkdownForDisplay("See `[web-1]` for details."),
    'See [web-1](#references "citation") for details.',
  );
});

test("normalizeMarkdownForDisplay unwraps research citation code spans", () => {
  assert.equal(
    normalizeMarkdownForDisplay("See `[CIT-1-01]` for details."),
    'See [1](#ref-cit-1-01 "citation") for details.',
  );
});

test("normalizeMarkdownForDisplay does not linkify research reference list ids", () => {
  const input =
    '<details id="references" open><summary>参考资料</summary><ol>' +
    '<li id="ref-cit-1-01" data-citation-id="CIT-1-01">' +
    "<strong>[1]</strong> <code>CIT-1-01</code> Web Search: q</li>" +
    "</ol></details>";
  assert.equal(normalizeMarkdownForDisplay(input), input);
});

test("escapeUnknownHtmlTagsForDisplay escapes LLM pseudo tags", () => {
  const input = "Before\n<think>internal scratchpad</think>\nAfter";
  assert.equal(
    escapeUnknownHtmlTagsForDisplay(input),
    "Before\n`<think>`internal scratchpad`</think>`\nAfter",
  );
});

test("escapeUnknownHtmlTagsForDisplay preserves line count for previews", () => {
  const input = "A\n\n<thinking>hidden</thinking>\nB";
  const output = escapeUnknownHtmlTagsForDisplay(input);
  assert.equal(output.split("\n").length, input.split("\n").length);
});

test("escapeUnknownHtmlTagsForDisplay keeps allowed html tags", () => {
  const input = "<details><summary>More</summary>Body</details>";
  assert.equal(escapeUnknownHtmlTagsForDisplay(input), input);
});

test("escapeUnknownHtmlTagsForDisplay escapes active html containers", () => {
  const input = '<iframe src="https://example.com"></iframe>';
  assert.equal(
    escapeUnknownHtmlTagsForDisplay(input),
    '`<iframe src="https://example.com">``</iframe>`',
  );
});

test("escapeUnknownHtmlTagsForDisplay strips unsafe html attributes", () => {
  const input =
    '<a href="javascript:alert(1)" onclick="alert(2)" style="color:red">link</a>';
  assert.equal(escapeUnknownHtmlTagsForDisplay(input), "<a>link</a>");
});

test("markdownUrlTransform keeps raster data images on img src", () => {
  const png = "data:image/png;base64,iVBORw0KGgo=";
  assert.equal(markdownUrlTransform(png, "src", { tagName: "img" }), png);
});

test("markdownUrlTransform rejects active data URLs", () => {
  assert.equal(
    markdownUrlTransform("data:text/html;base64,PHNjcmlwdD4=", "src", {
      tagName: "img",
    }),
    "",
  );
  assert.equal(
    markdownUrlTransform(
      "data:image/svg+xml;base64,PHN2ZyBvbmxvYWQ9YWxlcnQoMSk+",
      "src",
      { tagName: "img" },
    ),
    "",
  );
});

test("markdownUrlTransform only allows data images on img src", () => {
  assert.equal(
    markdownUrlTransform("data:image/png;base64,iVBORw0KGgo=", "href", {
      tagName: "a",
    }),
    "",
  );
});

test("safeDecodeURIComponent decodes valid hash components", () => {
  assert.equal(safeDecodeURIComponent("section%201"), "section 1");
});

test("safeDecodeURIComponent keeps malformed hash components intact", () => {
  assert.doesNotThrow(() => safeDecodeURIComponent("%E0%A4%A"));
  assert.equal(safeDecodeURIComponent("%E0%A4%A"), "%E0%A4%A");
});

test("hasVisibleMarkdownContent rejects empty raw-html placeholders", () => {
  assert.equal(
    hasVisibleMarkdownContent("<details><summary></summary></details>"),
    false,
  );
});

test("hasVisibleMarkdownContent rejects raw html control placeholders", () => {
  assert.equal(
    hasVisibleMarkdownContent('<progress></progress>\n<input type="text" />'),
    false,
  );
});

test("hasVisibleMarkdownContent rejects empty markdown tables", () => {
  assert.equal(hasVisibleMarkdownContent("| |\n|---|"), false);
});

test("hasVisibleMarkdownContent keeps meaningful markdown", () => {
  assert.equal(
    hasVisibleMarkdownContent("这是一个正常回复。\n\n- 第一条"),
    true,
  );
});

/** Replay a reply arriving in chunks the way the chat reducer does. */
function streamThrough(chunks: string[], language: string): string {
  let raw = "";
  let display = "";
  for (const chunk of chunks) {
    raw += chunk;
    display = appendWithEmphasisRepair(display, chunk, raw, language);
  }
  // What STREAM_END does: settle the trailing line, which had no newline.
  return repairChineseEmphasis(raw, language);
}

test("appendWithEmphasisRepair lands on the same text as repairing every chunk", () => {
  const chunks = [
    "第一行",
    "**,重点**",
    "\n",
    "第二行也有",
    "**,重点**",
    "\n\n",
    "```\n",
    "code**,not touched**\n",
    "```\n",
    "结尾",
    "**,重点**",
  ];
  const whole = chunks.join("");

  assert.equal(streamThrough(chunks, "zh-CN"), repairChineseEmphasis(whole, "zh-CN"));
});

test("appendWithEmphasisRepair repairs a line as soon as its newline arrives", () => {
  const raw = "中文**,重点**\n";
  // Mid-line the partial tail shows through unrepaired...
  assert.equal(
    appendWithEmphasisRepair("中文", "**,重点**", "中文**,重点**", "zh"),
    "中文**,重点**",
  );
  // ...and the newline that completes the line settles it.
  assert.equal(appendWithEmphasisRepair("中文**,重点**", "\n", raw, "zh"), "中文 **,重点**\n");
});

test("appendWithEmphasisRepair leaves non-Chinese replies untouched", () => {
  assert.equal(
    appendWithEmphasisRepair("ignored", "**bold**", "text**bold**", "en"),
    "text**bold**",
  );
});