Version 2

[blakeembrey]

Opening this to centralize the V2 discussion.

Goal: Lenient parser, written in TypeScript, with no regexes when parsing to be used by multiparty.

We also need to investigate the existing open issues and resolve them. We may need to create two "versions" of parse/format, depending on how the investigation goes, to support the HTTP header vs the multi-part header.

• Better multipart support #3
• Support continuations #2
• allow parsing utf8 filenames #15
• Generate ext-parameter when utf8 and latin1 representations don't match #27
• Certain languages can't be matched correctly by the Regular Expression EXT_VALUE_REGEXP like en-US or zh_cn #47
• MIME encoded values #95
• Investigate RFC 8187 compliance (obsoletes RFC 5987) #116

[blakeembrey]

It should support node 18 so it can be adopted into Express.js v5, and we could use this as an opportunity for a dual ESM + CJS package @Phillip9587 is hoping to add.

[blakeembrey]

We probably need to build a local demo page to test download/upload of filenames and how they encode across browsers quickly.

[Phillip9587]

We probably need to build a local demo page to test download/upload of filenames and how they encode across browsers quickly.

And we can maybe use playwright to automatically test this in CI: https://playwright.dev/docs/api/class-download#download-suggested-filename

[blakeembrey]

Not sure we need it automated, so I wouldn't invest much in that, mostly just to confirm how different edge cases like https://github.com/pillarjs/multiparty/blob/master/test/fixture/http/special-chars-in-filename/issue-252-chrome.http

From reviewing multiparty and this library we have an issue where every browser has slightly different behavior. We should attempt to handle as much as possible, but I'm not sure how doable it is yet.