fix(v2): optimize markdown parser regex (fix #4726)

[nam-hle]

Motivation

Fix #4726 (origin: #4713)

Have you read the Contributing Guidelines on pull requests?

Yes.

Test Plan

1. Clone my prepared repo and install
2. Start. Everything should work well
3. Incremental uncomment import statements in docs/intro.md, the compilation time will increase exponentially from around 15 imported components.
4. Apply the changes in packages/docusaurus-utils/src/markdownParser.ts of this PR into the corresponding compiled file (at client-docusaurus/node_modules/@docusaurus/utils/lib/markdownParser.js) of my repo.
5. Do the step 2-3 again to see the compile time improvement.

Please feel free to suggest any concrete performat tests that need to do.

Related PRs

N/A.

[netlify]

[V2]

Built without sensitive environment variables with commit 5c36bf2

https://deploy-preview-4729--docusaurus-2.netlify.app

[github-actions]

⚡️ Lighthouse report for the changes in this PR:

Category	Score
🟠 Performance	66
🟢 Accessibility	96
🟢 Best practices	100
🟢 SEO	100
🟢 PWA	95

Lighthouse ran on https://deploy-preview-4729--docusaurus-2.netlify.app/

[slorber]

Thanks! I tried to hack around a bit locally and also found a similar potential solution.

That seems good enough to merge for now!

Seems to work, and could see your test being very slow without the regex change.

Not sure why but it seems to only affect the 2nd regex (alternateTitleMatch), not the first one.

Hey, if you are a regex pro and want to help, any improvement to support such syntax would be welcome, as the current regex does not cover edge cases:

          import Component1 from '@site/src/components/Component1';
          import Component2, {NamedImport, SecondImport} from '@site/src/components/Component2';
          import {Component3, NamedImport as NamedImportRenamed} from '@site/src/components/Component3';
          import * as Component4 from '@site/src/components/Component4';

Thanks ;)