test(integrations/client/packages): batch 2-4 unit coverage + zip-slip guard extraction
Batch 2-4 of the test-strategy rollout. Test-only except one minimal, behaviour-preserving extraction in file.utils.ts. All suites green: server 82 suites/836+1todo, editor-ext 86, mcp 270, client (new files) 86. integrations (server): - file.utils.ts: extract pure `isEntryPathSafe(entryName, targetDir)` from extractZipInternal so the zip-slip/path-traversal guard is unit-testable; call site rerouted, behaviour identical (only a warn-message string merged). - file.utils.zip-safety.spec.ts: traversal/strip/__MACOSX/prefix-confusion cases (mutation-resistant: fails if containment loses the path.sep). - import-formatter / import.utils / table-utils / export utils / import.service extractTitleAndRemoveHeading: pure import/export transforms, Notion/XWiki formatting, table colspan widths (idempotent), slug/link rewriting. client: - safeRedirectPath: open-redirect guard, every reject branch independently. - buildChatMarkdown (fence anti-breakout), label-colors, normalize-label, share tree build, page URL builders, notification time-grouping (fake clock). packages: - editor-ext: deriveFootnoteId golden table, parseHtmlEmbedHeight crafted values, orphan footnote extraction. - mcp: deriveFootnoteId parity (drift guard vs editor-ext), applyTextEdits idempotency + cross-block replaceAll, diffDocs/summarizeChange on reorder. Reviewed (APPROVE): extraction behaviour-preserving, assertions mutation-resistant. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
claude_code
@@ -30,6 +30,52 @@ export function getFileTaskFolderPath(
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Pure path-safety decision for a single ZIP entry (zip-slip / path-traversal guard).
|
||||
*
|
||||
* Reproduces exactly the inline check previously embedded in `extractZipInternal`:
|
||||
* 1. Strip any leading slashes from the entry name.
|
||||
* 2. Reject names that fail `yauzl.validateFileName` (e.g. backslashes,
|
||||
* relative `..` segments, drive letters).
|
||||
* 3. Reject `__MACOSX/` metadata entries.
|
||||
* 4. Resolve the entry against the target directory and require it to stay
|
||||
* strictly inside `targetDir` using a `targetResolved + path.sep` prefix check
|
||||
* (the trailing separator prevents sibling-directory prefix confusion, e.g.
|
||||
* `/tmp/x` must not match `/tmp/x-evil`).
|
||||
*
|
||||
* @param entryName The decoded (UTF-8) entry file name from the archive.
|
||||
* @param targetDir Directory the archive is being extracted into.
|
||||
* @returns `{ safe }` and, when safe, the resolved absolute path of the entry.
|
||||
*/
|
||||
export function isEntryPathSafe(
|
||||
entryName: string,
|
||||
targetDir: string,
|
||||
): { safe: boolean; resolved?: string } {
|
||||
// Strip leading slashes so absolute-looking entries cannot escape the target.
|
||||
const safe = entryName.replace(/^\/+/, '');
|
||||
|
||||
const validationError = yauzl.validateFileName(safe);
|
||||
if (validationError) {
|
||||
return { safe: false };
|
||||
}
|
||||
|
||||
// Skip macOS resource-fork metadata entries.
|
||||
if (safe.startsWith('__MACOSX/')) {
|
||||
return { safe: false };
|
||||
}
|
||||
|
||||
const fullPath = path.join(targetDir, safe);
|
||||
const resolved = path.resolve(fullPath);
|
||||
const targetResolved = path.resolve(targetDir);
|
||||
|
||||
// Containment check: resolved path must live strictly inside the target dir.
|
||||
if (!resolved.startsWith(targetResolved + path.sep)) {
|
||||
return { safe: false };
|
||||
}
|
||||
|
||||
return { safe: true, resolved };
|
||||
}
|
||||
|
||||
/**
|
||||
* Extracts a ZIP archive.
|
||||
*/
|
||||
@@ -103,29 +149,15 @@ function extractZipInternal(
|
||||
const name = entry.fileName.toString('utf8');
|
||||
const safe = name.replace(/^\/+/, '');
|
||||
|
||||
const validationError = yauzl.validateFileName(safe);
|
||||
if (validationError) {
|
||||
console.warn(`Skipping invalid entry (${validationError})`);
|
||||
zipfile.readEntry();
|
||||
return;
|
||||
}
|
||||
|
||||
if (safe.startsWith('__MACOSX/')) {
|
||||
// Zip-slip / path-traversal guard (see isEntryPathSafe).
|
||||
if (!isEntryPathSafe(name, target).safe) {
|
||||
console.warn(`Skipping unsafe entry: ${safe}`);
|
||||
zipfile.readEntry();
|
||||
return;
|
||||
}
|
||||
|
||||
const fullPath = path.join(target, safe);
|
||||
|
||||
const resolved = path.resolve(fullPath);
|
||||
const targetResolved = path.resolve(target);
|
||||
|
||||
if (!resolved.startsWith(targetResolved + path.sep)) {
|
||||
console.warn(`Skipping entry (path outside target): ${safe}`);
|
||||
zipfile.readEntry();
|
||||
return;
|
||||
}
|
||||
|
||||
// Handle directories
|
||||
if (/\/$/.test(name)) {
|
||||
try {
|
||||
|
||||
@@ -0,0 +1,105 @@
|
||||
import * as path from 'path';
|
||||
import { isEntryPathSafe } from './file.utils';
|
||||
|
||||
/**
|
||||
* Unit tests for isEntryPathSafe: the pure zip-slip / path-traversal guard
|
||||
* extracted from extractZipInternal. The contract reproduced from the
|
||||
* production inline check is, in order:
|
||||
* 1. strip leading slashes from the entry name;
|
||||
* 2. reject names that fail yauzl.validateFileName (relative `..` segments,
|
||||
* backslashes, drive letters, etc.);
|
||||
* 3. reject `__MACOSX/` metadata entries;
|
||||
* 4. resolve the (stripped) entry under the target dir and require it to stay
|
||||
* strictly inside the target via a `targetResolved + path.sep` prefix check.
|
||||
*
|
||||
* The separator in step 4 is the load-bearing detail: it prevents sibling-dir
|
||||
* prefix confusion (e.g. target `/tmp/x` vs `/tmp/x-evil`). The tests below are
|
||||
* written so that weakening that check to a bare `startsWith(targetResolved)`
|
||||
* makes at least one test fail.
|
||||
*/
|
||||
describe('isEntryPathSafe', () => {
|
||||
// Use an absolute target; on the test platform path.sep is '/'.
|
||||
const target = path.resolve('/tmp/x');
|
||||
|
||||
it('accepts a normal nested entry and resolves it inside the target', () => {
|
||||
const result = isEntryPathSafe('a/b/c.png', target);
|
||||
expect(result.safe).toBe(true);
|
||||
expect(result.resolved).toBe(path.join(target, 'a/b/c.png'));
|
||||
// Resolved path must live strictly under the target directory.
|
||||
expect(result.resolved!.startsWith(target + path.sep)).toBe(true);
|
||||
});
|
||||
|
||||
it('strips a single leading slash and then treats the entry as safe', () => {
|
||||
const result = isEntryPathSafe('/a/b/c.png', target);
|
||||
expect(result.safe).toBe(true);
|
||||
expect(result.resolved).toBe(path.join(target, 'a/b/c.png'));
|
||||
});
|
||||
|
||||
it('strips multiple leading slashes and then treats the entry as safe', () => {
|
||||
const result = isEntryPathSafe('///a/b.png', target);
|
||||
expect(result.safe).toBe(true);
|
||||
expect(result.resolved).toBe(path.join(target, 'a/b.png'));
|
||||
});
|
||||
|
||||
it('skips (marks unsafe) __MACOSX metadata entries', () => {
|
||||
const result = isEntryPathSafe('__MACOSX/foo', target);
|
||||
expect(result.safe).toBe(false);
|
||||
expect(result.resolved).toBeUndefined();
|
||||
});
|
||||
|
||||
it('rejects a relative ../../ traversal entry', () => {
|
||||
// yauzl.validateFileName flags this as an "invalid relative path", so it is
|
||||
// rejected before the containment check ever runs. Either way: unsafe.
|
||||
const result = isEntryPathSafe('../../etc/passwd', target);
|
||||
expect(result.safe).toBe(false);
|
||||
expect(result.resolved).toBeUndefined();
|
||||
});
|
||||
|
||||
it('rejects an entry whose resolved path would land in a sibling directory (prefix confusion)', () => {
|
||||
// The classic off-by-one: target `/tmp/x` must NOT contain `/tmp/x-evil`.
|
||||
// Such an escape can only be expressed with a `..` segment, which the guard
|
||||
// rejects. This asserts the guard holds for the sibling-escape attempt.
|
||||
const result = isEntryPathSafe('../x-evil/p', target);
|
||||
expect(result.safe).toBe(false);
|
||||
expect(result.resolved).toBeUndefined();
|
||||
});
|
||||
|
||||
it('rejects an entry that resolves to exactly the target dir (no trailing separator)', () => {
|
||||
// `.` resolves to the target itself. The strict `targetResolved + path.sep`
|
||||
// prefix check rejects it; a weakened `startsWith(targetResolved)` (without
|
||||
// the separator) would WRONGLY accept it. This test is the mutation killer
|
||||
// for the separator: if the separator is dropped, this assertion fails.
|
||||
const result = isEntryPathSafe('.', target);
|
||||
expect(result.safe).toBe(false);
|
||||
expect(result.resolved).toBeUndefined();
|
||||
});
|
||||
|
||||
it('keeps the target/sibling boundary: a bare-prefix sibling is not inside the target', () => {
|
||||
// Direct statement of the invariant the separator protects. The resolved
|
||||
// sibling path shares the target's basename as a prefix but is a different
|
||||
// directory; only the `+ path.sep` form correctly classifies it as outside.
|
||||
const target2 = path.resolve('/tmp/x');
|
||||
const siblingResolved = path.resolve(path.join(target2, '..', 'x-evil', 'p'));
|
||||
expect(siblingResolved.startsWith(target2)).toBe(true); // weak (buggy) check matches
|
||||
expect(siblingResolved.startsWith(target2 + path.sep)).toBe(false); // strict check rejects
|
||||
});
|
||||
|
||||
it('rejects an entry containing a backslash via yauzl.validateFileName', () => {
|
||||
// Backslashes are flagged by yauzl.validateFileName as invalid characters,
|
||||
// so such entries are unsafe regardless of where they would resolve.
|
||||
const result = isEntryPathSafe('a\\b.png', target);
|
||||
expect(result.safe).toBe(false);
|
||||
expect(result.resolved).toBeUndefined();
|
||||
});
|
||||
|
||||
it('accepts a stripped absolute path that lands inside the target', () => {
|
||||
// Documented ACTUAL behaviour: an entry like `/etc/passwd` has its leading
|
||||
// slash stripped to `etc/passwd`, which resolves to <target>/etc/passwd —
|
||||
// strictly inside the target, hence safe. (This is the point of the strip:
|
||||
// an absolute-looking entry is re-anchored under the target rather than
|
||||
// escaping to the filesystem root.)
|
||||
const result = isEntryPathSafe('/etc/passwd', target);
|
||||
expect(result.safe).toBe(true);
|
||||
expect(result.resolved).toBe(path.join(target, 'etc/passwd'));
|
||||
});
|
||||
});
|
||||
@@ -0,0 +1,403 @@
|
||||
// @sindresorhus/slugify ships as ESM and is not in jest's transform allowlist,
|
||||
// so it cannot be imported under ts-jest here. Mock it with a deterministic
|
||||
// lowercase/dash slugifier that matches the real output for the simple ASCII
|
||||
// titles used in these tests (e.g. "Real Title" -> "real-title"). This keeps
|
||||
// the test focused on the formatter's own slug-composition logic.
|
||||
jest.mock('@sindresorhus/slugify', () => ({
|
||||
__esModule: true,
|
||||
default: (input: string) =>
|
||||
String(input)
|
||||
.trim()
|
||||
.toLowerCase()
|
||||
.replace(/[^a-z0-9]+/g, '-')
|
||||
.replace(/^-+|-+$/g, ''),
|
||||
}));
|
||||
|
||||
import { load, CheerioAPI, Cheerio } from 'cheerio';
|
||||
import {
|
||||
rewriteInternalLinksToMentionHtml,
|
||||
notionFormatter,
|
||||
xwikiFormatter,
|
||||
defaultHtmlFormatter,
|
||||
unwrapFromParagraph,
|
||||
} from './import-formatter';
|
||||
|
||||
/**
|
||||
* Unit tests for import-formatter.ts. These are pure DOM transforms driven by
|
||||
* cheerio. Each test loads a snippet, runs the target function against the
|
||||
* cheerio root, and asserts the mutated markup / return value. Assertions are
|
||||
* written to fail if the corresponding branch were silently removed.
|
||||
*/
|
||||
|
||||
type PageMeta = { id: string; title: string; slugId: string };
|
||||
|
||||
function makeRoot(html: string): { $: CheerioAPI; $root: Cheerio<any> } {
|
||||
const $ = load(html);
|
||||
return { $, $root: $.root() };
|
||||
}
|
||||
|
||||
describe('rewriteInternalLinksToMentionHtml', () => {
|
||||
const creatorId = 'creator-1';
|
||||
const sourcePageId = 'source-page-1';
|
||||
const workspaceId = 'workspace-1';
|
||||
|
||||
it('replaces an internal link whose text equals the page title with a mention span', async () => {
|
||||
const meta: PageMeta = {
|
||||
id: 'target-id-1',
|
||||
title: 'Design Doc',
|
||||
slugId: 'slugABC',
|
||||
};
|
||||
// currentFilePath dir is "docs"; href "./target.md" resolves to "docs/target.md"
|
||||
const map = new Map<string, PageMeta>([['docs/target.md', meta]]);
|
||||
const { $, $root } = makeRoot(
|
||||
'<a href="./target.md">Design Doc</a>',
|
||||
);
|
||||
|
||||
const backlinks = await rewriteInternalLinksToMentionHtml(
|
||||
$,
|
||||
$root,
|
||||
'docs/index.md',
|
||||
map,
|
||||
creatorId,
|
||||
sourcePageId,
|
||||
workspaceId,
|
||||
);
|
||||
|
||||
const $mention = $root.find('span[data-type="mention"]');
|
||||
expect($mention.length).toBe(1);
|
||||
expect($mention.attr('data-entity-type')).toBe('page');
|
||||
expect($mention.attr('data-entity-id')).toBe('target-id-1');
|
||||
expect($mention.attr('data-label')).toBe('Design Doc');
|
||||
expect($mention.attr('data-slug-id')).toBe('slugABC');
|
||||
expect($mention.attr('data-creator-id')).toBe(creatorId);
|
||||
expect($mention.attr('data-id')).toBeTruthy();
|
||||
expect($mention.text()).toBe('Design Doc');
|
||||
// original anchor must be gone
|
||||
expect($root.find('a').length).toBe(0);
|
||||
|
||||
expect(backlinks).toEqual([
|
||||
{ sourcePageId, targetPageId: 'target-id-1', workspaceId },
|
||||
]);
|
||||
});
|
||||
|
||||
it('rewrites href to /s/{space}/p/{slug} when text differs from the title', async () => {
|
||||
const meta: PageMeta = {
|
||||
id: 'target-id-2',
|
||||
title: 'Real Title',
|
||||
slugId: 'slug999',
|
||||
};
|
||||
const map = new Map<string, PageMeta>([['docs/target.md', meta]]);
|
||||
const { $, $root } = makeRoot(
|
||||
'<a href="./target.md">click here</a>',
|
||||
);
|
||||
|
||||
const backlinks = await rewriteInternalLinksToMentionHtml(
|
||||
$,
|
||||
$root,
|
||||
'docs/index.md',
|
||||
map,
|
||||
creatorId,
|
||||
sourcePageId,
|
||||
workspaceId,
|
||||
'myspace',
|
||||
);
|
||||
|
||||
// still an anchor, no mention span
|
||||
expect($root.find('span[data-type="mention"]').length).toBe(0);
|
||||
const $a = $root.find('a');
|
||||
expect($a.length).toBe(1);
|
||||
// slugify('Real Title') => 'real-title'
|
||||
expect($a.attr('href')).toBe('/s/myspace/p/real-title-slug999');
|
||||
expect($a.attr('data-internal')).toBe('true');
|
||||
expect($a.text()).toBe('click here');
|
||||
|
||||
expect(backlinks).toEqual([
|
||||
{ sourcePageId, targetPageId: 'target-id-2', workspaceId },
|
||||
]);
|
||||
});
|
||||
|
||||
it('uses /p/{slug} when no spaceSlug is provided', async () => {
|
||||
const meta: PageMeta = {
|
||||
id: 'target-id-3',
|
||||
title: 'Other Page',
|
||||
slugId: 'slug777',
|
||||
};
|
||||
const map = new Map<string, PageMeta>([['docs/target.md', meta]]);
|
||||
const { $, $root } = makeRoot('<a href="./target.md">label</a>');
|
||||
|
||||
await rewriteInternalLinksToMentionHtml(
|
||||
$,
|
||||
$root,
|
||||
'docs/index.md',
|
||||
map,
|
||||
creatorId,
|
||||
sourcePageId,
|
||||
workspaceId,
|
||||
);
|
||||
|
||||
expect($root.find('a').attr('href')).toBe('/p/other-page-slug777');
|
||||
});
|
||||
|
||||
it('leaves external http and /api/ hrefs untouched and records no backlink', async () => {
|
||||
const map = new Map<string, PageMeta>();
|
||||
const { $, $root } = makeRoot(
|
||||
'<a href="https://example.com/page">ext</a><a href="/api/files/x">api</a>',
|
||||
);
|
||||
|
||||
const backlinks = await rewriteInternalLinksToMentionHtml(
|
||||
$,
|
||||
$root,
|
||||
'docs/index.md',
|
||||
map,
|
||||
creatorId,
|
||||
sourcePageId,
|
||||
workspaceId,
|
||||
);
|
||||
|
||||
const hrefs = $root
|
||||
.find('a')
|
||||
.map((_, el) => $(el).attr('href'))
|
||||
.get();
|
||||
expect(hrefs).toEqual(['https://example.com/page', '/api/files/x']);
|
||||
expect($root.find('a').first().attr('data-internal')).toBeUndefined();
|
||||
expect(backlinks).toEqual([]);
|
||||
});
|
||||
|
||||
it('falls back without throwing on a malformed decodeURIComponent href', async () => {
|
||||
const meta: PageMeta = {
|
||||
id: 'target-id-4',
|
||||
title: 'Broken',
|
||||
slugId: 'slug000',
|
||||
};
|
||||
// The raw (un-decodable) href is what gets joined: "docs/%E0%A4%A.md".
|
||||
const map = new Map<string, PageMeta>([['docs/%E0%A4%A.md', meta]]);
|
||||
const { $, $root } = makeRoot('<a href="%E0%A4%A.md">Broken</a>');
|
||||
|
||||
let backlinks: any;
|
||||
await expect(
|
||||
(async () => {
|
||||
backlinks = await rewriteInternalLinksToMentionHtml(
|
||||
$,
|
||||
$root,
|
||||
'docs/index.md',
|
||||
map,
|
||||
creatorId,
|
||||
sourcePageId,
|
||||
workspaceId,
|
||||
);
|
||||
})(),
|
||||
).resolves.not.toThrow();
|
||||
|
||||
// Because the raw path matched the map, it still produced a mention.
|
||||
expect($root.find('span[data-type="mention"]').length).toBe(1);
|
||||
expect(backlinks).toEqual([
|
||||
{ sourcePageId, targetPageId: 'target-id-4', workspaceId },
|
||||
]);
|
||||
});
|
||||
|
||||
it('accumulates one backlink per resolved link', async () => {
|
||||
const a: PageMeta = { id: 'id-a', title: 'A', slugId: 's-a' };
|
||||
const b: PageMeta = { id: 'id-b', title: 'B', slugId: 's-b' };
|
||||
const map = new Map<string, PageMeta>([
|
||||
['docs/a.md', a],
|
||||
['docs/b.md', b],
|
||||
]);
|
||||
const { $, $root } = makeRoot(
|
||||
'<a href="./a.md">A</a><a href="./b.md">B</a>',
|
||||
);
|
||||
|
||||
const backlinks = await rewriteInternalLinksToMentionHtml(
|
||||
$,
|
||||
$root,
|
||||
'docs/index.md',
|
||||
map,
|
||||
creatorId,
|
||||
sourcePageId,
|
||||
workspaceId,
|
||||
);
|
||||
|
||||
expect(backlinks).toEqual([
|
||||
{ sourcePageId, targetPageId: 'id-a', workspaceId },
|
||||
{ sourcePageId, targetPageId: 'id-b', workspaceId },
|
||||
]);
|
||||
});
|
||||
});
|
||||
|
||||
describe('notionFormatter', () => {
|
||||
it('converts a multi-column column-list to data-type="columns" with the right layout', () => {
|
||||
const html =
|
||||
'<div class="column-list">' +
|
||||
'<div class="column"><p>one</p></div>' +
|
||||
'<div class="column"><p>two</p></div>' +
|
||||
'<div class="column"><p>three</p></div>' +
|
||||
'</div>';
|
||||
const { $, $root } = makeRoot(html);
|
||||
|
||||
notionFormatter($, $root);
|
||||
|
||||
const $cols = $root.find('div[data-type="columns"]');
|
||||
expect($cols.length).toBe(1);
|
||||
// 3 columns => COLUMN_LAYOUTS[3] === 'three_equal'
|
||||
expect($cols.attr('data-layout')).toBe('three_equal');
|
||||
expect($root.find('div[data-type="column"]').length).toBe(3);
|
||||
// original column-list wrapper is gone
|
||||
expect($root.find('div.column-list').length).toBe(0);
|
||||
});
|
||||
|
||||
it('uses two_equal layout for exactly two columns', () => {
|
||||
const html =
|
||||
'<div class="column-list">' +
|
||||
'<div class="column"><p>one</p></div>' +
|
||||
'<div class="column"><p>two</p></div>' +
|
||||
'</div>';
|
||||
const { $, $root } = makeRoot(html);
|
||||
|
||||
notionFormatter($, $root);
|
||||
|
||||
expect($root.find('div[data-type="columns"]').attr('data-layout')).toBe(
|
||||
'two_equal',
|
||||
);
|
||||
});
|
||||
|
||||
it('converts figure.equation into a mathBlock with the tex text', () => {
|
||||
const html =
|
||||
'<figure class="equation">' +
|
||||
'<annotation encoding="application/x-tex">E = mc^2</annotation>' +
|
||||
'</figure>';
|
||||
const { $, $root } = makeRoot(html);
|
||||
|
||||
notionFormatter($, $root);
|
||||
|
||||
const $math = $root.find('div[data-type="mathBlock"]');
|
||||
expect($math.length).toBe(1);
|
||||
expect($math.attr('data-katex')).toBe('true');
|
||||
expect($math.text()).toBe('E = mc^2');
|
||||
expect($root.find('figure.equation').length).toBe(0);
|
||||
});
|
||||
|
||||
it('converts ul.to-do-list items to a taskList with data-checked reflecting checkbox-on', () => {
|
||||
const html =
|
||||
'<ul class="to-do-list">' +
|
||||
'<li><div class="checkbox checkbox-on"></div>' +
|
||||
'<span class="to-do-children-checked">done item</span></li>' +
|
||||
'<li><div class="checkbox checkbox-off"></div>' +
|
||||
'<span class="to-do-children-unchecked">open item</span></li>' +
|
||||
'</ul>';
|
||||
const { $, $root } = makeRoot(html);
|
||||
|
||||
notionFormatter($, $root);
|
||||
|
||||
const $list = $root.find('ul[data-type="taskList"]');
|
||||
expect($list.length).toBe(1);
|
||||
const $items = $list.find('li[data-type="taskItem"]');
|
||||
expect($items.length).toBe(2);
|
||||
expect($items.eq(0).attr('data-checked')).toBe('true');
|
||||
expect($items.eq(1).attr('data-checked')).toBe('false');
|
||||
// checked item has a checked input; unchecked does not
|
||||
expect($items.eq(0).find('input[checked]').length).toBe(1);
|
||||
expect($items.eq(1).find('input[checked]').length).toBe(0);
|
||||
// text is carried over
|
||||
expect($items.eq(0).find('p').text()).toBe('done item');
|
||||
expect($items.eq(1).find('p').text()).toBe('open item');
|
||||
});
|
||||
});
|
||||
|
||||
describe('xwikiFormatter', () => {
|
||||
it('replaces the root with the contents of #xwikicontent when present', () => {
|
||||
const html =
|
||||
'<div id="header">junk</div>' +
|
||||
'<div id="xwikicontent"><p>real body</p><h2>heading</h2></div>';
|
||||
const { $, $root } = makeRoot(html);
|
||||
|
||||
xwikiFormatter($, $root);
|
||||
|
||||
expect($root.find('#header').length).toBe(0);
|
||||
expect($root.find('#xwikicontent').length).toBe(0);
|
||||
expect($root.find('p').text()).toBe('real body');
|
||||
expect($root.find('h2').text()).toBe('heading');
|
||||
});
|
||||
|
||||
it('leaves HTML without #xwikicontent unchanged', () => {
|
||||
const html = '<div id="header">junk</div><p>body</p>';
|
||||
const { $, $root } = makeRoot(html);
|
||||
const before = $root.html();
|
||||
|
||||
xwikiFormatter($, $root);
|
||||
|
||||
expect($root.html()).toBe(before);
|
||||
});
|
||||
});
|
||||
|
||||
describe('defaultHtmlFormatter', () => {
|
||||
it('replaces a recognized provider anchor with a data-type="embed" div', () => {
|
||||
const url = 'https://www.youtube.com/watch?v=dQw4w9WgXcQ';
|
||||
const { $, $root } = makeRoot(`<a href="${url}">video</a>`);
|
||||
|
||||
defaultHtmlFormatter($, $root);
|
||||
|
||||
const $embed = $root.find('div[data-type="embed"]');
|
||||
expect($embed.length).toBe(1);
|
||||
expect($embed.attr('data-provider')).toBe('youtube');
|
||||
expect($embed.attr('data-src')).toBe(url);
|
||||
// the anchor is gone
|
||||
expect($root.find('a').length).toBe(0);
|
||||
});
|
||||
|
||||
it('leaves an anchor as a link when provider resolves to iframe', () => {
|
||||
// A plain non-provider URL falls through to the default iframe provider,
|
||||
// which the formatter explicitly skips.
|
||||
const url = 'https://example.com/some/page';
|
||||
const { $, $root } = makeRoot(`<a href="${url}">site</a>`);
|
||||
|
||||
defaultHtmlFormatter($, $root);
|
||||
|
||||
expect($root.find('div[data-type="embed"]').length).toBe(0);
|
||||
const $a = $root.find('a');
|
||||
expect($a.length).toBe(1);
|
||||
expect($a.attr('href')).toBe(url);
|
||||
});
|
||||
});
|
||||
|
||||
describe('unwrapFromParagraph', () => {
|
||||
it('replaces the wrapper entirely when the node is the only child of a <p>', () => {
|
||||
const { $, $root } = makeRoot('<p><img src="x.png"></p>');
|
||||
const $node = $root.find('img');
|
||||
|
||||
unwrapFromParagraph($, $node);
|
||||
|
||||
// the <p> wrapper is gone, the img is hoisted to the root
|
||||
expect($root.find('p').length).toBe(0);
|
||||
expect($root.find('img').length).toBe(1);
|
||||
});
|
||||
|
||||
it('moves the node before the wrapper when there are sibling contents', () => {
|
||||
const { $, $root } = makeRoot('<p>text before <img src="x.png"></p>');
|
||||
const $node = $root.find('img');
|
||||
|
||||
unwrapFromParagraph($, $node);
|
||||
|
||||
// img moved out; the paragraph still holds the sibling text
|
||||
const html = $root.html() || '';
|
||||
// img must appear before the paragraph in document order
|
||||
const imgIndex = html.indexOf('<img');
|
||||
const pIndex = html.indexOf('<p');
|
||||
expect(imgIndex).toBeGreaterThanOrEqual(0);
|
||||
expect(pIndex).toBeGreaterThanOrEqual(0);
|
||||
expect(imgIndex).toBeLessThan(pIndex);
|
||||
expect($root.find('p').text()).toContain('text before');
|
||||
});
|
||||
|
||||
it('returns (does not infinite-loop) on adversarial nesting', () => {
|
||||
// Node wrapped in nested <a> and <p> wrappers.
|
||||
const { $, $root } = makeRoot(
|
||||
'<p><a href="#"><img src="x.png"></a></p>',
|
||||
);
|
||||
const $node = $root.find('img');
|
||||
|
||||
// If unwrapFromParagraph looped forever this call would hang the test.
|
||||
expect(() => unwrapFromParagraph($, $node)).not.toThrow();
|
||||
// It fully unwrapped: no surrounding p/a left around the img.
|
||||
expect($node.closest('p, a').length).toBe(0);
|
||||
expect($root.find('img').length).toBe(1);
|
||||
});
|
||||
});
|
||||
@@ -0,0 +1,137 @@
|
||||
import {
|
||||
stripNotionID,
|
||||
extractNotionPartialId,
|
||||
resolveRelativeAttachmentPath,
|
||||
} from './import.utils';
|
||||
|
||||
/**
|
||||
* Unit tests for the pure helpers in import.utils.ts:
|
||||
* - stripNotionID / extractNotionPartialId: filename suffix parsing.
|
||||
* - resolveRelativeAttachmentPath: maps an HTML-relative attachment href onto
|
||||
* a key that exists in the extracted-archive candidate map.
|
||||
*/
|
||||
|
||||
describe('stripNotionID', () => {
|
||||
it('strips a 32-hex suffix preceded by a space separator', () => {
|
||||
// 32 hex chars with a leading space.
|
||||
const id = 'a1b2c3d4e5f60718293a4b5c6d7e8f90';
|
||||
expect(stripNotionID(`My Page ${id}`)).toBe('My Page');
|
||||
});
|
||||
|
||||
it('strips a 32-hex suffix preceded by a dash separator', () => {
|
||||
const id = 'a1b2c3d4e5f60718293a4b5c6d7e8f90';
|
||||
expect(stripNotionID(`My-Page-${id}`)).toBe('My-Page');
|
||||
});
|
||||
|
||||
it('strips a 32-hex suffix with no separator', () => {
|
||||
const id = 'a1b2c3d4e5f60718293a4b5c6d7e8f90';
|
||||
expect(stripNotionID(`MyPage${id}`)).toBe('MyPage');
|
||||
});
|
||||
|
||||
it('strips a partial UUID suffix "{4}-{4}"', () => {
|
||||
expect(stripNotionID('Cool 324d-35ab')).toBe('Cool');
|
||||
});
|
||||
|
||||
it('leaves a name without an ID unchanged', () => {
|
||||
expect(stripNotionID('Just A Title')).toBe('Just A Title');
|
||||
});
|
||||
});
|
||||
|
||||
describe('extractNotionPartialId', () => {
|
||||
it('returns prefix/suffix (lowercased) for a partial UUID folder name', () => {
|
||||
expect(extractNotionPartialId('Cool 324D-35AB')).toEqual({
|
||||
prefix: '324d',
|
||||
suffix: '35ab',
|
||||
});
|
||||
});
|
||||
|
||||
it('returns null when there is no partial UUID suffix', () => {
|
||||
expect(extractNotionPartialId('No Id Here')).toBeNull();
|
||||
});
|
||||
|
||||
it('returns null when the suffix lacks the leading space', () => {
|
||||
// The regex requires a leading space before "{4}-{4}".
|
||||
expect(extractNotionPartialId('Name324d-35ab')).toBeNull();
|
||||
});
|
||||
});
|
||||
|
||||
describe('resolveRelativeAttachmentPath', () => {
|
||||
it('returns the direct candidate when it exists', () => {
|
||||
const candidates = new Map<string, string>([
|
||||
['attachments/file.png', '/abs/attachments/file.png'],
|
||||
]);
|
||||
expect(
|
||||
resolveRelativeAttachmentPath(
|
||||
'./attachments/file.png',
|
||||
'pages',
|
||||
candidates,
|
||||
),
|
||||
).toBe('attachments/file.png');
|
||||
});
|
||||
|
||||
it('strips the Confluence "download/attachments/" prefix to match the archive layout', () => {
|
||||
const candidates = new Map<string, string>([
|
||||
['attachments/123/diagram.png', '/abs/attachments/123/diagram.png'],
|
||||
]);
|
||||
expect(
|
||||
resolveRelativeAttachmentPath(
|
||||
'download/attachments/123/diagram.png',
|
||||
'pages',
|
||||
candidates,
|
||||
),
|
||||
).toBe('attachments/123/diagram.png');
|
||||
});
|
||||
|
||||
it('decodes a percent-encoded name before matching', () => {
|
||||
const candidates = new Map<string, string>([
|
||||
['attachments/my file.png', '/abs/attachments/my file.png'],
|
||||
]);
|
||||
expect(
|
||||
resolveRelativeAttachmentPath(
|
||||
'attachments/my%20file.png',
|
||||
'pages',
|
||||
candidates,
|
||||
),
|
||||
).toBe('attachments/my file.png');
|
||||
});
|
||||
|
||||
it('falls back to the raw (still-encoded) value on a malformed escape without throwing', () => {
|
||||
// "%E0%A4" is an incomplete UTF-8 sequence; decodeURIComponent throws and
|
||||
// the helper keeps the raw string, which then matches the candidate key.
|
||||
const candidates = new Map<string, string>([
|
||||
['attachments/%E0%A4.png', '/abs/attachments/%E0%A4.png'],
|
||||
]);
|
||||
let result: string | null = null;
|
||||
expect(() => {
|
||||
result = resolveRelativeAttachmentPath(
|
||||
'attachments/%E0%A4.png',
|
||||
'pages',
|
||||
candidates,
|
||||
);
|
||||
}).not.toThrow();
|
||||
expect(result).toBe('attachments/%E0%A4.png');
|
||||
});
|
||||
|
||||
it('returns null when nothing matches', () => {
|
||||
const candidates = new Map<string, string>([
|
||||
['attachments/other.png', '/abs/attachments/other.png'],
|
||||
]);
|
||||
expect(
|
||||
resolveRelativeAttachmentPath(
|
||||
'./attachments/missing.png',
|
||||
'pages',
|
||||
candidates,
|
||||
),
|
||||
).toBeNull();
|
||||
});
|
||||
|
||||
it('matches via the pageDir-joined fallback path', () => {
|
||||
// raw resolves under pageDir when neither the direct nor confluence key hit.
|
||||
const candidates = new Map<string, string>([
|
||||
['pages/sub/img.png', '/abs/pages/sub/img.png'],
|
||||
]);
|
||||
expect(
|
||||
resolveRelativeAttachmentPath('sub/img.png', 'pages', candidates),
|
||||
).toBe('pages/sub/img.png');
|
||||
});
|
||||
});
|
||||
@@ -0,0 +1,105 @@
|
||||
import { load, CheerioAPI, Cheerio } from 'cheerio';
|
||||
import { normalizeTableColumnWidths } from './table-utils';
|
||||
|
||||
/**
|
||||
* Unit tests for normalizeTableColumnWidths: it writes a `colwidth` attribute
|
||||
* onto the first-row cells of every <table>, deriving widths from a <colgroup>
|
||||
* or the first row, accounting for colspan, and falling back to a default
|
||||
* per-column width (150px) when no pixel widths are present. Re-running the
|
||||
* transform on its own output must be a no-op (idempotent).
|
||||
*/
|
||||
|
||||
const DEFAULT = 150;
|
||||
|
||||
function run(html: string): { $: CheerioAPI; $root: Cheerio<any> } {
|
||||
const $ = load(html);
|
||||
const $root = $.root();
|
||||
normalizeTableColumnWidths($, $root);
|
||||
return { $, $root };
|
||||
}
|
||||
|
||||
function firstRowColwidths($root: Cheerio<any>): (string | undefined)[] {
|
||||
return $root
|
||||
.find('table')
|
||||
.first()
|
||||
.find('> tbody > tr, > thead > tr, > tr')
|
||||
.first()
|
||||
.children('td, th')
|
||||
.map((_, el) => (el as any).attribs?.colwidth)
|
||||
.get();
|
||||
}
|
||||
|
||||
describe('normalizeTableColumnWidths', () => {
|
||||
it('applies colgroup <col width> to the first-row cells', () => {
|
||||
const html =
|
||||
'<table>' +
|
||||
'<colgroup><col width="120"><col width="80"></colgroup>' +
|
||||
'<tbody><tr><td>a</td><td>b</td></tr></tbody>' +
|
||||
'</table>';
|
||||
const { $root } = run(html);
|
||||
|
||||
expect(firstRowColwidths($root)).toEqual(['120', '80']);
|
||||
});
|
||||
|
||||
it('falls back to first-row cell widths when there is no colgroup', () => {
|
||||
const html =
|
||||
'<table><tbody>' +
|
||||
'<tr><td style="width: 200px">a</td><td width="90">b</td></tr>' +
|
||||
'</tbody></table>';
|
||||
const { $root } = run(html);
|
||||
|
||||
expect(firstRowColwidths($root)).toEqual(['200', '90']);
|
||||
});
|
||||
|
||||
it('splits a colspan width across the spanned columns', () => {
|
||||
// colspan=2 with width 100 => each derived column ~50, the spanning cell
|
||||
// then gets the joined slice "50,50".
|
||||
const html =
|
||||
'<table><tbody>' +
|
||||
'<tr><td colspan="2" width="100">merged</td></tr>' +
|
||||
'</tbody></table>';
|
||||
const { $root } = run(html);
|
||||
|
||||
expect(firstRowColwidths($root)).toEqual(['50,50']);
|
||||
});
|
||||
|
||||
it('ignores em/% widths (treated as no width) and applies the default', () => {
|
||||
const html =
|
||||
'<table><tbody>' +
|
||||
'<tr><td style="width: 10em">a</td><td style="width: 50%">b</td></tr>' +
|
||||
'</tbody></table>';
|
||||
const { $root } = run(html);
|
||||
|
||||
expect(firstRowColwidths($root)).toEqual([String(DEFAULT), String(DEFAULT)]);
|
||||
});
|
||||
|
||||
it('applies the default per-column width to a markdown-style table with no widths', () => {
|
||||
const html =
|
||||
'<table><tbody>' +
|
||||
'<tr><td>a</td><td>b</td><td>c</td></tr>' +
|
||||
'<tr><td>1</td><td>2</td><td>3</td></tr>' +
|
||||
'</tbody></table>';
|
||||
const { $root } = run(html);
|
||||
|
||||
expect(firstRowColwidths($root)).toEqual([
|
||||
String(DEFAULT),
|
||||
String(DEFAULT),
|
||||
String(DEFAULT),
|
||||
]);
|
||||
});
|
||||
|
||||
it('is idempotent: re-running on its own output changes nothing', () => {
|
||||
const html =
|
||||
'<table>' +
|
||||
'<colgroup><col width="120"><col width="80"></colgroup>' +
|
||||
'<tbody><tr><td>a</td><td>b</td></tr></tbody>' +
|
||||
'</table>';
|
||||
const { $, $root } = run(html);
|
||||
const afterFirst = $root.html();
|
||||
|
||||
// second pass
|
||||
normalizeTableColumnWidths($, $root);
|
||||
expect($root.html()).toBe(afterFirst);
|
||||
expect(firstRowColwidths($root)).toEqual(['120', '80']);
|
||||
});
|
||||
});
|
||||
Reference in New Issue
Block a user