Port TransferCoding/Transfer-Encoding to cats-parse #4023
Conversation
aeons
force-pushed
the
transfer-encoding-cats-parse
branch
from
ecd40f7
to
256b7c6
Compare
aeons
force-pushed
the
transfer-encoding-cats-parse
branch
from
256b7c6
to
02df3dc
Compare
|
I think you took non-converted files, so I believe nobody has covered them yet. Please notify in the thread #3984 that few more parsers are covered. |
| @@ -77,6 +78,10 @@ object ParseResult { | |||
| case NonFatal(e) => Left(ParseFailure(sanitized, e.getMessage)) | |||
| } | |||
|
|
|||
| private[http4s] def fromParser[A](parser: Parser[A], errorMessage: => String)( | |||
| s: String): ParseResult[A] = | |||
| parser.parseAll(s).leftMap(e => ParseFailure(errorMessage, e.toString)) | |||
There was a problem hiding this comment.
It's nice to centralize this, because we're going to need to do better than that e.toString.
There was a problem hiding this comment.
Yep, this was what I saw in the other places this pattern was used. Will investigate.
There was a problem hiding this comment.
So, what we get out of Parser.Error is an offset where the parsing failed, and a list of expected inputs that it didn't find. I can format that in a decently nice way, but it's not going be as good as the ErrorFormatter from parboiled.
An alternative is to do something like https://github.com/http4s/http4s/blob/main/core/src/main/scala/org/http4s/HttpVersion.scala#L58 and repeat the input instead of giving parser-specific errors.
There was a problem hiding this comment.
The new LocationMap in cats-parse helps turn input into a row-column index and is designed to help with formatting. Oscar mentioned some code in bosatsu that formats errors, but I haven't gone hunting for it.
I suspect there's an opportunity to do a decent default rendering in cats-parse on that LocationMap.
core/src/main/scala/org/http4s/internal/parsing/Rfc2616BasicRules.scala
Outdated
Show resolved
Hide resolved
See #3984
I don't have the complete overview of what is being done on which branch, so I hope noone has done this already.
I added a little helper, that I could envision being used in many of these header parsers, I can remove that again if we want to avoid that kind of thing.