fix(mdx-loader): resolve Markdown/MDX links with Remark instead of RegExp (#10168)

This commit is contained in:
Sébastien Lorber 2024-05-24 19:03:23 +02:00 committed by GitHub
parent aab332c2ae
commit e34614963e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
36 changed files with 902 additions and 1620 deletions

View file

@ -1,250 +0,0 @@
// Jest Snapshot v1, https://goo.gl/fbAQLP
exports[`replaceMarkdownLinks does basic replace 1`] = `
{
"brokenMarkdownLinks": [
{
"contentPaths": {
"contentPath": "docs",
"contentPathLocalized": "i18n/docs-localized",
},
"filePath": "docs/intro.md",
"link": "hmmm.md",
},
],
"newContent": "
[foo](/doc/foo)
[baz](/doc/baz)
[foo](/doc/foo)
[http](http://github.com/facebook/docusaurus/README.md)
[https](https://github.com/facebook/docusaurus/README.md)
[asset](./foo.js)
[asset as well](@site/docs/_partial.md)
[looks like http...](/doc/http)
[nonexistent](hmmm.md)
",
}
`;
exports[`replaceMarkdownLinks handles link titles 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "
[URL](/docs/file "title")
[URL](/docs/file 'title')
[URL](/docs/file (title))
",
}
`;
exports[`replaceMarkdownLinks handles stray spaces 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "
[URL]( /docs/file )
[ref]: /docs/file
",
}
`;
exports[`replaceMarkdownLinks handles unpaired fences 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "
\`\`\`foo
hello
\`\`\`foo
hello
\`\`\`
A [link](/docs/file)
",
}
`;
exports[`replaceMarkdownLinks ignores links in HTML comments 1`] = `
{
"brokenMarkdownLinks": [
{
"contentPaths": {
"contentPath": "docs",
"contentPathLocalized": "i18n/docs-localized",
},
"filePath": "docs/intro.md",
"link": "./foo.md",
},
{
"contentPaths": {
"contentPath": "docs",
"contentPathLocalized": "i18n/docs-localized",
},
"filePath": "docs/intro.md",
"link": "./foo.md",
},
],
"newContent": "
<!-- [foo](./foo.md) -->
<!--
[foo](./foo.md)
-->
",
}
`;
exports[`replaceMarkdownLinks ignores links in fenced blocks 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "
\`\`\`
[foo](foo.md)
\`\`\`
\`\`\`\`js
[foo](foo.md)
\`\`\`
[foo](foo.md)
\`\`\`
[foo](foo.md)
\`\`\`\`
\`\`\`\`js
[foo](foo.md)
\`\`\`
[foo](foo.md)
\`\`\`\`
~~~js
[foo](foo.md)
~~~
~~~js
[foo](foo.md)
\`\`\`
[foo](foo.md)
\`\`\`
[foo](foo.md)
~~~
",
}
`;
exports[`replaceMarkdownLinks ignores links in inline code 1`] = `
{
"brokenMarkdownLinks": [
{
"contentPaths": {
"contentPath": "docs",
"contentPathLocalized": "i18n/docs-localized",
},
"filePath": "docs/intro.md",
"link": "foo.md",
},
],
"newContent": "
\`[foo](foo.md)\`
",
}
`;
exports[`replaceMarkdownLinks preserves query/hash 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "
[URL](/docs/file?foo=bar#baz)
[URL](/docs/file#a)
[URL](/docs/file?c)
",
}
`;
exports[`replaceMarkdownLinks replaces Markdown links with spaces 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "
[doc a](/docs/doc%20a)
[doc a](</docs/doc%20a>)
[doc b](/docs/my%20docs/doc%20b)
[doc b](</docs/my%20docs/doc%20b>)
[doc]: </docs/my%20docs/doc%20b>
",
}
`;
exports[`replaceMarkdownLinks replaces links with same title as URL 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "
[foo.md](/docs/foo)
[./foo.md](</docs/foo>)
[./foo.md](/docs/foo)
[foo.md](/docs/foo)
[./foo.md](/docs/foo)
",
}
`;
exports[`replaceMarkdownLinks replaces multiple links on same line 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "
[a](/docs/a), [a](/docs/a), [b](/docs/b), [c](/docs/c)
",
}
`;
exports[`replaceMarkdownLinks replaces reference style Markdown links 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "
The following operations are defined for [URI]s:
* [info]: Returns metadata about the resource,
* [list]: Returns metadata about the resource's children (like getting the content of a local directory).
[URI]: /docs/api/classes/uri
[info]: /docs/api/classes/uri#info
[list]: /docs/api/classes/uri#list
",
}
`;
exports[`replaceMarkdownLinks replaces two links on the same line 1`] = `
{
"brokenMarkdownLinks": [],
"newContent": "[TypeScript](/programming-languages/typescript/) and [Go](/programming-languages/go/)",
}
`;
exports[`replaceMarkdownLinks resolves absolute and relative links differently 1`] = `
{
"brokenMarkdownLinks": [
{
"contentPaths": {
"contentPath": "docs",
"contentPathLocalized": "i18n/docs-localized",
},
"filePath": "docs/intro/intro.md",
"link": "./api/classes/divine_uri.URI.md",
},
{
"contentPaths": {
"contentPath": "docs",
"contentPathLocalized": "i18n/docs-localized",
},
"filePath": "docs/intro/intro.md",
"link": "/another.md",
},
],
"newContent": "
[Relative link](/docs/another)
[Relative link 2](/docs/api/classes/uri)
[Relative link that should be absolute](./api/classes/divine_uri.URI.md)
[Absolute link](/docs/api/classes/uri)
[Absolute link from site dir](/docs/api/classes/uri)
[Absolute link that should be relative](/another.md)
[Relative link that acts as absolute](/docs/api/classes/uri)
[Relative link that acts as relative](/docs/another)
",
}
`;

View file

@ -5,401 +5,70 @@
* LICENSE file in the root directory of this source tree.
*/
import {replaceMarkdownLinks} from '../markdownLinks';
import {resolveMarkdownLinkPathname} from '../markdownLinks';
describe('resolveMarkdownLinkPathname', () => {
type Context = Parameters<typeof resolveMarkdownLinkPathname>[1];
describe('replaceMarkdownLinks', () => {
it('does basic replace', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro.md': '/docs/intro',
'@site/docs/foo.md': '/doc/foo',
'@site/docs/bar/baz.md': '/doc/baz',
'@site/docs/http.foo.md': '/doc/http',
},
fileString: `
[foo](./foo.md)
[baz](./bar/baz.md)
[foo](foo.md)
[http](http://github.com/facebook/docusaurus/README.md)
[https](https://github.com/facebook/docusaurus/README.md)
[asset](./foo.js)
[asset as well](@site/docs/_partial.md)
[looks like http...](http.foo.md)
[nonexistent](hmmm.md)
`,
}),
).toMatchSnapshot();
});
const context: Context = {
siteDir: '.',
sourceFilePath: 'docs/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro.md': '/docs/intro',
'@site/docs/foo.md': '/doc/foo',
'@site/docs/bar/baz.md': '/doc/baz',
'@site/docs/http.foo.md': '/doc/http',
},
};
it('replaces two links on the same line', () => {
// cSpell:ignore Goooooooooo
// This is a very arcane bug: if we continue matching using the previous
// matching index (as is the behavior of RegExp#exec), it will go right over
// the next Markdown link and fail to match the "Go" link. This only happens
// when: (1) the replaced link is much shorter than the Markdown path, (2)
// the next link is very close to the current one (e.g. here if it's not
// "Go" but "Goooooooooo", or if every link has the /docs/ prefix, the bug
// will not trigger because it won't overshoot)
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro.md': '/',
'@site/docs/programming-languages/typescript/typescript.md':
'/programming-languages/typescript/',
'@site/docs/programming-languages/go/go.md':
'/programming-languages/go/',
},
fileString: `[TypeScript](programming-languages/typescript/typescript.md) and [Go](programming-languages/go/go.md)`,
}),
).toMatchSnapshot();
});
function test(linkPathname: string, expectedOutput: string) {
const output = resolveMarkdownLinkPathname(linkPathname, context);
expect(output).toEqual(expectedOutput);
}
it('replaces reference style Markdown links', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro/intro.md': '/docs/intro',
'@site/docs/api/classes/divine_uri.URI.md': '/docs/api/classes/uri',
},
fileString: `
The following operations are defined for [URI]s:
* [info]: Returns metadata about the resource,
* [list]: Returns metadata about the resource's children (like getting the content of a local directory).
[URI]: ../api/classes/divine_uri.URI.md
[info]: ../api/classes/divine_uri.URI.md#info
[list]: ../api/classes/divine_uri.URI.md#list
`,
}),
).toMatchSnapshot();
test('./foo.md', '/doc/foo');
test('foo.md', '/doc/foo');
test('./bar/baz.md', '/doc/baz');
test('http.foo.md', '/doc/http');
test('@site/docs/_partial.md', null);
test('foo.js', null);
test('nonexistent.md', null);
test('https://github.com/facebook/docusaurus/README.md', null);
});
it('resolves absolute and relative links differently', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
const context: Context = {
siteDir: '.',
sourceFilePath: 'docs/intro/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro/intro.md': '/docs/intro',
'@site/docs/intro/another.md': '/docs/another',
'@site/docs/api/classes/divine_uri.URI.md': '/docs/api/classes/uri',
},
sourceToPermalink: {
'@site/docs/intro/intro.md': '/docs/intro',
'@site/docs/intro/another.md': '/docs/another',
'@site/docs/api/classes/divine_uri.URI.md': '/docs/api/classes/uri',
},
};
fileString: `
[Relative link](./another.md)
[Relative link 2](../api/classes/divine_uri.URI.md)
[Relative link that should be absolute](./api/classes/divine_uri.URI.md)
[Absolute link](/api/classes/divine_uri.URI.md)
[Absolute link from site dir](/docs/api/classes/divine_uri.URI.md)
[Absolute link that should be relative](/another.md)
[Relative link that acts as absolute](api/classes/divine_uri.URI.md)
[Relative link that acts as relative](another.md)
`,
}),
).toMatchSnapshot();
});
function test(linkPathname: string, expectedOutput: string) {
const output = resolveMarkdownLinkPathname(linkPathname, context);
expect(output).toEqual(expectedOutput);
}
// TODO bad
it('ignores links in HTML comments', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro.md': '/docs/intro',
},
fileString: `
<!-- [foo](./foo.md) -->
<!--
[foo](./foo.md)
-->
`,
}),
).toMatchSnapshot();
});
it('ignores links in fenced blocks', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro.md': '/docs/intro',
},
fileString: `
\`\`\`
[foo](foo.md)
\`\`\`
\`\`\`\`js
[foo](foo.md)
\`\`\`
[foo](foo.md)
\`\`\`
[foo](foo.md)
\`\`\`\`
\`\`\`\`js
[foo](foo.md)
\`\`\`
[foo](foo.md)
\`\`\`\`
~~~js
[foo](foo.md)
~~~
~~~js
[foo](foo.md)
\`\`\`
[foo](foo.md)
\`\`\`
[foo](foo.md)
~~~
`,
}),
).toMatchSnapshot();
});
// FIXME
it('ignores links in inline code', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro.md': '/docs/intro',
},
fileString: `
\`[foo](foo.md)\`
`,
}),
).toMatchSnapshot();
});
it('replaces links with same title as URL', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro.md': '/docs/intro',
'@site/docs/foo.md': '/docs/foo',
},
fileString: `
[foo.md](foo.md)
[./foo.md](<./foo.md>)
[./foo.md](./foo.md)
[foo.md](./foo.md)
[./foo.md](foo.md)
`,
}),
).toMatchSnapshot();
});
it('replaces multiple links on same line', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/intro.md': '/docs/intro',
'@site/docs/a.md': '/docs/a',
'@site/docs/b.md': '/docs/b',
'@site/docs/c.md': '/docs/c',
},
fileString: `
[a](a.md), [a](a.md), [b](b.md), [c](c.md)
`,
}),
).toMatchSnapshot();
});
it('replaces Markdown links with spaces', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/intro.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/doc a.md': '/docs/doc%20a',
'@site/docs/my docs/doc b.md': '/docs/my%20docs/doc%20b',
},
fileString: `
[doc a](./doc%20a.md)
[doc a](<./doc a.md>)
[doc b](./my%20docs/doc%20b.md)
[doc b](<./my docs/doc b.md>)
[doc]: <./my docs/doc b.md>
`,
}),
).toMatchSnapshot();
});
it('does not replace non-Markdown links', () => {
const input = `
[asset](./file.md_asset/1.png)
[URL](<https://example.com/file_(1).md>)
[not a link]((foo)
[not a link](foo bar)
[not a link]: foo bar
[not a link]: (foo
[not a link]: bar)
`;
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/file.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/file.md': '/docs/file',
},
fileString: input,
}),
).toEqual({
newContent: input,
brokenMarkdownLinks: [],
});
});
it('handles stray spaces', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/file.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/file.md': '/docs/file',
},
fileString: `
[URL]( ./file.md )
[ref]: ./file.md
`,
}),
).toMatchSnapshot();
});
it('handles link titles', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/file.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/file.md': '/docs/file',
},
fileString: `
[URL](./file.md "title")
[URL](./file.md 'title')
[URL](./file.md (title))
`,
}),
).toMatchSnapshot();
});
it('preserves query/hash', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/file.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/file.md': '/docs/file',
},
fileString: `
[URL](./file.md?foo=bar#baz)
[URL](./file.md#a)
[URL](./file.md?c)
`,
}),
).toMatchSnapshot();
});
it('handles unpaired fences', () => {
expect(
replaceMarkdownLinks({
siteDir: '.',
filePath: 'docs/file.md',
contentPaths: {
contentPath: 'docs',
contentPathLocalized: 'i18n/docs-localized',
},
sourceToPermalink: {
'@site/docs/file.md': '/docs/file',
},
fileString: `
\`\`\`foo
hello
\`\`\`foo
hello
\`\`\`
A [link](./file.md)
`,
}),
).toMatchSnapshot();
test('./another.md', '/docs/another');
test('../api/classes/divine_uri.URI.md', '/docs/api/classes/uri');
test('./api/classes/divine_uri.URI.md', null);
test('/api/classes/divine_uri.URI.md', '/docs/api/classes/uri');
test('/docs/api/classes/divine_uri.URI.md', '/docs/api/classes/uri');
test('/another.md', null);
test('api/classes/divine_uri.URI.md', '/docs/api/classes/uri');
test('another.md', '/docs/another');
});
});

View file

@ -17,6 +17,9 @@ import {
hasSSHProtocol,
parseURLPath,
serializeURLPath,
parseURLOrPath,
toURLPath,
parseLocalURLPath,
} from '../urlUtils';
describe('normalizeUrl', () => {
@ -228,6 +231,166 @@ describe('isValidPathname', () => {
});
});
describe('toURLPath', () => {
it('url', () => {
const url = new URL('https://example.com/pathname?qs#hash');
expect(toURLPath(url)).toEqual({
pathname: '/pathname',
search: 'qs',
hash: 'hash',
});
});
it('pathname + qs', () => {
const url = parseURLOrPath('/pathname?qs');
expect(toURLPath(url)).toEqual({
pathname: '/pathname',
search: 'qs',
hash: undefined,
});
});
it('pathname + hash', () => {
const url = parseURLOrPath('/pathname#hash');
expect(toURLPath(url)).toEqual({
pathname: '/pathname',
search: undefined,
hash: 'hash',
});
});
it('pathname + qs + hash', () => {
const url = parseURLOrPath('/pathname?qs#hash');
expect(toURLPath(url)).toEqual({
pathname: '/pathname',
search: 'qs',
hash: 'hash',
});
});
it('pathname + empty qs + empty hash', () => {
const url = parseURLOrPath('/pathname?#');
expect(toURLPath(url)).toEqual({
pathname: '/pathname',
search: '',
hash: '',
});
});
});
describe('parseLocalURLPath', () => {
it('returns null for non-local URLs', () => {
expect(parseLocalURLPath('https://example')).toBeNull();
expect(parseLocalURLPath('https://example:80')).toBeNull();
expect(parseLocalURLPath('https://example.com/xyz')).toBeNull();
expect(parseLocalURLPath('https://example.com/xyz?qs#hash')).toBeNull();
expect(parseLocalURLPath('https://example.com:80/xyz?qs#hash')).toBeNull();
expect(parseLocalURLPath('https://u:p@example:80/xyz?qs#hash')).toBeNull();
});
it('parses pathname', () => {
expect(parseLocalURLPath('/pathname')).toEqual({
pathname: '/pathname',
search: undefined,
hash: undefined,
});
expect(parseLocalURLPath('pathname.md')).toEqual({
pathname: 'pathname.md',
search: undefined,
hash: undefined,
});
expect(parseLocalURLPath('./pathname')).toEqual({
pathname: './pathname',
search: undefined,
hash: undefined,
});
expect(parseLocalURLPath('../../pathname.mdx')).toEqual({
pathname: '../../pathname.mdx',
search: undefined,
hash: undefined,
});
});
it('parses qs', () => {
expect(parseLocalURLPath('?')).toEqual({
pathname: '',
search: '',
hash: undefined,
});
expect(parseLocalURLPath('?qs')).toEqual({
pathname: '',
search: 'qs',
hash: undefined,
});
expect(parseLocalURLPath('?age=42')).toEqual({
pathname: '',
search: 'age=42',
hash: undefined,
});
});
it('parses hash', () => {
expect(parseLocalURLPath('#')).toEqual({
pathname: '',
search: undefined,
hash: '',
});
expect(parseLocalURLPath('#hash')).toEqual({
pathname: '',
search: undefined,
hash: 'hash',
});
});
it('parses complex local paths', () => {
expect(
parseLocalURLPath('../../great/path name/doc.mdx?age=42#hash'),
).toEqual({
pathname: '../../great/path name/doc.mdx',
search: 'age=42',
hash: 'hash',
});
expect(parseLocalURLPath('my great path?=42#hash?qsInHash')).toEqual({
pathname: 'my great path',
search: '=42',
hash: 'hash?qsInHash',
});
expect(parseLocalURLPath('?qs1#hash1?qs2#hash2')).toEqual({
pathname: '',
search: 'qs1',
hash: 'hash1?qs2#hash2',
});
expect(parseLocalURLPath('../swizzling.mdx#wrapping')).toEqual({
pathname: '../swizzling.mdx',
search: undefined,
hash: 'wrapping',
});
});
it('parses is isomorphic with serialize', () => {
const testLocalPath = (url: string) => {
expect(serializeURLPath(parseLocalURLPath(url)!)).toBe(url);
};
[
'',
'doc',
'doc.mdx',
'./doc.mdx',
'.././doc.mdx',
'/some pathname/.././doc.mdx',
'?',
'?qs',
'#',
'#hash',
'?qs#hash',
'?qs#hash',
'doc.mdx?qs#hash',
'/some pathname/.././doc.mdx?qs#hash',
'/some pathname/.././doc.mdx?qs#hash?qs2#hash2',
].forEach(testLocalPath);
});
});
describe('parseURLPath', () => {
it('parse and resolve pathname', () => {
expect(parseURLPath('')).toEqual({