Parsing delimited text where delimiter appears in the text?

[zmitchell]

I’m trying to parse a custom file format with an example shown below, where the <shell code> is one or more line of possibly valid code for a certain shell.

@shells(bash,fish)
@test does_a_thing {
  <shell code>
}

@test does_different_thing {
  <shell code>
}

The problem I run into with my own grammar is that } is a valid character in Bash code, but is also the delimiter for the test body, so the closing delimiter appears as the first character in a valid bit of shell code. How would you approach parsing this?

[zmitchell]

Here’s what I had put together before getting stuck.

supported_shell = { "bash" | "zsh" | "tcsh" | "fish" }
comma = _{ "," }
shells = { supported_shell ~ (comma ~ supported_shell)* }
shells_decorator = { "@shells(" ~  shells ~ ")" }

test_name = { ASCII_ALPHA ~ ( ASCII_ALPHANUMERIC | "_" )* }
test_header = { "@test " ~ test_name ~ " " }
punctuation = { '!'..'@' | '['..'`' | '{'..'~' }
shell_character = { ASCII_ALPHANUMERIC | " " | "\t" | punctuation }
shell_line = { (shell_character)* ~ NEWLINE }
test_body = { shell_line+ }
test = { test_header ~ "{" ~ NEWLINE ~ test_body ~ "}" }

test_with_decorators = { (shells_decorator ~ NEWLINE)? ~ test }

test_file = { NEWLINE* ~ (test_with_decorators ~ NEWLINE)* }