Gherkin table auto-format escapes single backslashes

[vitezslav-popovsky-solarwinds]

Used Visual Studio

Visual Studio 2022

Are the latest Visual Studio updates installed?

Yes

Content of reqnroll.json (if present)

No response

Issue Description

When using the auto-format feature on a Gherkin table in a .feature file, the Reqenroll plugin incorrectly converts single backslashes (\) into double backslashes (\\). This behavior is incorrect for Gherkin syntax This bug reduces the readability of the specification file by adding unnecessary escape characters.

Steps to Reproduce

1. Open a .feature file in a Visual Studio project where the Reqenroll plugin is active.
2. Create a Gherkin scenario that includes a data table.
3. In a cell of the table, enter a string that contains a single backslash, such as a Windows file path. For example:

| FilePath         |
| C:\Users\Test    |

4. Trigger the document auto-format command.

Link to a project repository that reproduces the issue

No response

[clrudolphi]

I'm inclined to say that this is not a bug.
From the Cucumber reference documentation, it is specifically mentioned that backslash characters in table cells can be/should be escaped.

Are you experiencing problems, other than human read-ability, because of this behavior?

[vitezslav-popovsky-solarwinds]

@clrudolphi only readability. Both

| FilePath      |
| C:\Users\Test |

and

| FilePath        |
| C:\\Users\\Test |

are translated to "C:\\Users\\Test" string

[Code-Grump]

Seems like our parser isn't handling the \\ escape sequence Cucumber specifies, but the auto-formatter expects it to be supported. Might need to address this in the parser? @gasparnagy

[gasparnagy]

@Code-Grump We would need to check how this test was passing then... https://github.com/cucumber/gherkin/blob/main/testdata/good/escaped_pipes.feature

[Code-Grump]

@gasparnagy Just checking my understanding is right: the parser reads the table cell in the example

\\\|a\\\\n

as

\|a\\n

which would be represented in a C-string as

\\|a\\\\n

@vitezslav-popovsky-solarwinds where are you seeing your value represented as C:\\Users\\Test?

[vitezslav-popovsky-solarwinds]

@Code-Grump by "C:\\Users\\Test" I meant C-string, i.e. characters C:\Users\Test, sorry for the confusion.

That means both variants of the following table cell

| FilePath      |
| C:\Users\Test |

and

| FilePath        |
| C:\\Users\\Test |

are read as

C:\Users\Test

which would be represented in a C-string as

"C:\\Users\\Test"

In other words, "\" and "" are interpreted the same way, and we use just "" (I initially perceived the auto-formatter's corrections as unnecessary because I had overlooked a relevant note in the documentation.)

[gasparnagy]

@Code-Grump by "C:\\Users\\Test" I meant C-string, i.e. characters C:\Users\Test, sorry for the confusion.

That means both variants of the following table cell

| FilePath |
| C:\Users\Test |
and

| FilePath |
| C:\Users\Test |
are read as

C:\Users\Test

which would be represented in a C-string as

"C:\Users\Test"
In other words, "" and "" are interpreted the same way, and we use just "" (I initially perceived the auto-formatter's corrections as unnecessary because I had overlooked a relevant note in the documentation.)

Ok. I understand now. Based on the Gherkin specification (to be double checked), the \ only needs to be escaped if there are | or n (and maybe some other) after it. If \ comes after, it might be escaped, but not necessary. So the formatter is a bit too greedy.