Some escapes are unexpectedly modified in character literal patterns

[vain0x]

Issue created from fantomas-online

Code

match char 0 with
| '\''
| '\"'
| '\x00'
| '\u0000'
| _ -> ()

Result

match char 0 with
| '''
| '"'
| '�'
| '�'
| _ -> ()

Problem description

In character literal patterns, hex escape sequences, unicode escape sequences and single/double quote escape sequences are modified (unescaped). Expected behavior is that these literals are kept unchanged.

Extra information

• The formatted result breaks by code.
• The formatted result gives compiler warnings.
• I or my company would be willing to help fix this.

Options

Fantomas Master at 01/17/2021 13:01:42 - b36e5a5

Default Fantomas configuration

_{Did you know that you can ignore files when formatting from fantomas-tool or the FAKE targets by using a .fantomasignore file?}

[nojaf]

Hello @vain0x, thank you for reporting this issue and showing interest in the project.

The nature of this problem is that the AST does not represent anymore what the user originally typed.

ImplFile
  (ParsedImplFileInput
     ("tmp.fsx", true, QualifiedNameOfFile Tmp$fsx, [], [],
      [SynModuleOrNamespace
         ([Tmp], false, AnonModule,
          [DoExpr
             (NoDebugPointAtDoBinding,
              Match
                (DebugPointAtBinding tmp.fsx (1,0--1,17) IsSynthetic=false,
                 App
                   (NonAtomic, false, Ident char,
                    Const (Int32 0, tmp.fsx (1,11--1,12) IsSynthetic=false),
                    tmp.fsx (1,6--1,12) IsSynthetic=false),
                 [Clause
                    (Or
                       (Or
                          (Or
                             (Or
                                (Const
                                   (Char '\'',
                                    tmp.fsx (2,2--2,6) IsSynthetic=false),
                                 Const
                                   (Char '"',
                                    tmp.fsx (3,2--3,6) IsSynthetic=false),
                                 tmp.fsx (2,2--3,6) IsSynthetic=false),
                              Const
                                (Char '\000',
                                 tmp.fsx (4,2--4,8) IsSynthetic=false),
                              tmp.fsx (2,2--4,8) IsSynthetic=false),
                           Const
                             (Char '\000', tmp.fsx (5,2--5,10) IsSynthetic=false),
                           tmp.fsx (2,2--5,10) IsSynthetic=false),
                        Wild tmp.fsx (6,2--6,3) IsSynthetic=false,
                        tmp.fsx (2,2--6,3) IsSynthetic=false), None,
                     Const (Unit, tmp.fsx (6,7--6,9) IsSynthetic=false),
                     tmp.fsx (2,2--6,3) IsSynthetic=false, Yes)],
                 tmp.fsx (1,0--6,9) IsSynthetic=false),
              tmp.fsx (1,0--6,9) IsSynthetic=false)], PreXmlDocEmpty, [], None,
          tmp.fsx (1,0--7,0) IsSynthetic=false)], (true, true)))

Notice the last two Char '\000',. In the future, I'd like to solve these things in the F# compiler.

For now, we can work around this with trivia. (capturing the original char contents via the F# tokens).

We have a helper to detect the char content of the trivia:

fantomas/src/Fantomas/CodePrinter.fs

Line 4981 in b36e5a5

TriviaHelpers.``get CharContent`` r ctx.TriviaMainNodes

This helper needs a fix, please move ``get CharContent`` to the end of the file and change it to

    let ``get CharContent`` range (nodes: Map<FsAstType, TriviaNode list>) =
        getNodesForTypes
            [ SynExpr_Const
              SynPat_Const  ]
            nodes
        |> List.tryFind (fun tv -> RangeHelpers.rangeEq tv.Range range)
        |> Option.bind
            (fun tv ->
                match tv.ContentItself with
                | Some (CharContent c) -> Some c
                | _ -> None)

This will solve the problem I believe.
Many thanks for your contribution.

[vain0x]

Thanks for help!

[nojaf]

Thanks for your first contribution to this project!