fix: regression in header case-insensitivity

[erunion]

This fixes a regression that was introduced in the recent header case-insensitivity work (#178) that is causing edge cases where, on certain targets, it's possible for a header to be:

• Duplicated and both instances having differing data. This was mostly apparent on multipart/form-data requests where one header would just be multipart/form-data while the other would be that plus the boundary.
• Some targets would completely fail to interpret the header, resulting in either an undefined header being set, being set in ways that it shouldn't be, or the target completely failing to function because a header wasn't being removed.

To resolve this I've:

• Added a new headers helper with a couple methods to traverse the now case-insensitive object of headers so that these targets that need special case handling for something like content-type can safely search for that in the available headers.
• Updated every target that was looking for case-sensitive headers to now use this new helper to do the same work it needs to do.

For testing all of this, I've updated the multipart/form-data test fixture to have its content-type header as Content-Type to force this case-insensitive lookup for targets that need it.

😰

[DMarby]

Looks great, thank you for finding this regression and fixing it!

[jgiovaresco]

Nice catch 👏

I'm wondering if we should also use the helper for

• httpsnippet/src/targets/c/libcurl.js

  Line 31 in 675df1a

  .push('curl_easy_setopt(hnd, CURLOPT_COOKIE, "%s");', source.allHeaders.cookie)

• httpsnippet/src/targets/node/fetch.js

  Lines 79 to 82 in 675df1a

  	reqOpts.headers.cookie = cookies
  	} else {
  	reqOpts.headers = {}
  	reqOpts.headers.cookie = cookies

[jgiovaresco]

I think I miss something I don't understand why we would not use the new helper functions you have added?

we would have something like

const contentTypeHeader = helpers.hasHeader('content-type') ? helpers.getHeaderName('content-type'): 'content-type';
request.headersObj[contentTypeHeader] = 'multipart/form-data; boundary=' + boundary

[erunion]

yeah this was an oversight on my part. will update!

[erunion]

@jgiovaresco Updated with your feedback.

I'm wondering if we should also use the helper for

• httpsnippet/src/targets/c/libcurl.js

  Line 31 in 675df1a

  .push('curl_easy_setopt(hnd, CURLOPT_COOKIE, "%s");', source.allHeaders.cookie)

• httpsnippet/src/targets/node/fetch.js

  Lines 79 to 82 in 675df1a

  	reqOpts.headers.cookie = cookies
  	} else {
  	reqOpts.headers = {}
  	reqOpts.headers.cookie = cookies

I'd love to swap the reqOpts.headers object with an object that has some functionality like caseless so we could eliminate the need for allHeaders and this new helper utility added, but that's a larger refactor that I think is out of scope for resolving this regression.

[develohpanda]

caseless would be cool, but I agree out of scope for this regression.

Thank you for taking the time to investigate and fix this! 🎉 Sorry about the delay, I've been quite bogged down for a while!

[develohpanda]

Nice tests!