Wrong request detection with leading whitespace & non http method

[rngtng]

Hey hey

I'm quite surprised to find out that leading whitespace in front of the http method is supported. Can you elaborate why? IMHO it leads to bad side effects and allow writing (IMHO) rather unreadable code. Like this is a valid syntax:

            GET https://httpbin.org/anything

One bad side effect is that then body is dropped in the following:

POST https://webhook.site/....
Content-Type: text/plain
     Post the following content

And even more surprising, it fails with non http method too. This is rendered as three requests:

POST https://webhook.site/....
PostCode is 1234
Gettogether

My expectation is that lines with leading whitespace are not detected as new requests. Same for lines which have start with HTTP method. wyt?

[AnWeber]

I actually don't use whitespaces before the HTTP method either. I would like to check how Intellij behaves here and act accordingly. I think I would adjust the regex here to only accept uppercase letters.

[rngtng]

Yes. Sounds like a good fix

[AnWeber]

So Intellij allows whitespaces, but the request method must be specified with uppercase. I adjust my logic accordingly to avoid false positives. Additionally I add a new switch # @force-separator to force the explicit use of ### to terminate the current region.

[rngtng]

@AnWeber ok, I see. Thanks for Introducing # @force-separator although it feels a little bit like a hack and not very intuitive. How come PostCode is matched at all? I guess the line in question is this one:
https://github.com/AnWeber/httpyac/blob/main/src/plugins/core/parse/requestHttpRegionParser.ts#L8

I added it here https://regex101.com/r/XwdLRc/1

I'd assume a single space after the method needs to be enforced (to seperate method and url) so what about

... RAPHQL)\s+(?<url>.+?)$
or
... RAPHQL) \s*(?<url>.+?)$

IMHO this would sharpen the parsing and fix my cases?

[rngtng]

on top, a url can't have whitespaces? so you probably could sharpen the matching here too..

I quick found this reference for allowed URL characters: https://www.freecodecamp.org/news/url-encoded-characters-reference/#:~:text=There%20are%20only%20certain%20characters,that%20can%20have%20special%20meanings.

[AnWeber]

@rngtng Unfortunately more difficult. The url can also contain variables that maybe allow whitespace. Now I could check if there is a {{ before the whitespace. But I lose the independence of the parser. And additionally I would limit the extensibility, because there might be use cases for whitespaces without {{ (not encoded query params?).

[AnWeber]

I forgot to make the whitespace after the method mandatory. I will add +. It was only missing in http parser :-(

[rngtng]

Ah right. Forgot about the variables. But imho enforcing a single whitespace would solve it already. With the leading one i can live with.

so imho just add the + and remove the force separator to keep it simple?

[AnWeber]

The force separator would just be a workaround in case I run into a problem again. But it would also be solvable by creative variables. I remove it again. The + is already in there. I just forgot to reference the ticket.

[rngtng]

Perfect. Thanks a lot

[AnWeber]

change is released with 6.4.3

[rngtng]

thanks a lot!