grammar.pest

//! # Pest parsing grammar
//!
//! Pest is really easy to use, and it forms simple parsing trees
//!
//! I may eventually switch to Nom if I have the time to convert everything.


// Core items: whitespace and comments

WHITESPACE = _{ " " | "\n" | "\r" | "\t" | "\\\n" }

COMMENT = _{
    "/*" ~ (!"*/" ~ ANY)* ~ "*/"
  | ("//" | "--" | "#") ~ (!"\n" ~ ANY)* ~ "\n"
}

white = _{WHITESPACE | COMMENT}

string = ${
    "\"" ~ (
        "\\" ~ ANY
      | !"\"" ~ ANY
    )* ~ "\""
  | "`" ~ (!"`" ~ ANY)* ~ "`"
}

date = ${
    ('0'..'9'){4}
  ~ "/"
  ~ ('0'..'9'){2}
  ~ "/"
  ~ ('0'..'9'){2}
}

msecs = ${"." ~ ('0'..'9'){3}}

time = ${
    ('0'..'9'){2}
  ~ ":"
  ~ ('0'..'9'){2}
  ~ ":"
  ~ ('0'..'9'){2}
  ~ msecs?
}

utc = ${"-UTC"}

datetime = ${
    date
  ~ (WHITESPACE | COMMENT)+
  ~ time
  ~ utc?
}

days = ${('0'..'9')+}

duration = @{
    (days ~ "d:")?
  ~ time
}

number_raw = ${"-"? ~ ('0'..'9')+}
number_suffix = ${"L" | "BD"}
number = ${
    number_raw
  ~ number_suffix?
}

decimal_raw = ${"-"? ~ ('0'..'9')+ ~ "." ~ ('0'..'9')+}
decimal_suffix = ${"f"}
decimal = ${
    decimal_raw
  ~ decimal_suffix?
}

bool_true = ${"true" | "on"}
bool_false = ${"false" | "off"}
boolean = ${bool_true | bool_false}

null = ${"null"}

base64_char = {'a'..'z' | 'A'..'Z' | '0'..'9' | "+" | "/" | "="}
base64 = ${"[" ~ (base64_char | (WHITESPACE | COMMENT)+)* ~ "]"}

value = ${
    string  | base64
  | datetime| date      | duration
  | decimal | number
  | boolean | null
}

ident = ${
	  ('a'..'z' | 'A'..'Z' | "_")
  ~ ('a'..'z' | 'A'..'Z' | '0'..'9' | "." | "$" | "-" | "_")*
}

attribute = ${ident ~ "=" ~ value}

namespace = ${ident ~ ":"}
tag = ${
    (value | namespace? ~ (!"\n" ~ white)* ~ ident)
  ~ ((!"\n" ~ white)+ ~ value)*
  ~ ((!"\n" ~ white)+ ~ attribute)*
  ~ ((!"\n" ~ white)+ ~ "{" ~ white* ~ tags ~ white* ~ "}" ~ (!"\n" ~ white)*)?
}

tags_sep = _{";" | "\n"}
tags = { tag? ~ (tags_sep ~ tag?)* }
tagtree = {SOI ~ tags ~ EOI}