Attributes are not parsed if they begin with a tab character

[eXplowar]

Ffl-file with tabs for attributes:

Result of parsing (Linguini incorrectly identifies this as Junk):

Ftl without tabs (no errors, attributes were defined correctly):

Result of parsing:

Not sure if this will be correct, but my suggestion is to adjust the condition on line 400: if (_reader.TryPeekChar(out var c) && c.IsAsciiAlphabetic() || c == '\u0009')

[Ygg01]

Result of parsing (Linguini incorrectly identifies this as Junk):

Small but important distinction. Linguini correctly identifies this as junk.

Look at https://github.com/projectfluent/fluent/issues/165

Tab was removed as valid indentation of attributes.

Further proof is the Fluent spec: https://github.com/projectfluent/fluent/blob/feffc720fa5d5621a703f42731d709e216fc6730/spec/fluent.ebnf#L32

Attribute           ::= line_end blank? "." Identifier blank_inline? 

/* Whitespace */
blank_inline        ::= "\u0020"+
line_end            ::= "\u000D\u000A"
                      | "\u000A"
                      | EOF
blank               ::= (blank_inline | line_end)+

So an attribute is a rule that says \n then OneOrMore('\n' or ' ') then . then identifier.

You would need to have U+0009 (tab) character to appear somewhere in blank definition. In fact, it doesn't.

By complying with this request we would be diverging from Fluent spec too much, for little gain.

[eXplowar]

Thanks for the answer, we'll know ;)

[Ygg01]

Thanks for the answer, we'll know ;)

What you can do is fix files to use spaces or fork Linguini.Syntax for your own usage. But then not sure how that would work with Nuget.