edn -> dsl

This commit is contained in:
🪞👃🪞 2025-03-14 23:44:13 +02:00
parent d30eda33d1
commit 877b344765
35 changed files with 197 additions and 225 deletions

169
dsl/src/token.rs Normal file
View file

@ -0,0 +1,169 @@
//! [Token]s are parsed substrings with an associated [Value].
//!
//! * [ ] FIXME: Value may be [Err] which may shadow [Result::Err]
//! * [Value::Exp] wraps an expression depth and a [SourceIter]
//! with the remaining part of the expression.
//! * expression depth other that 0 mean unclosed parenthesis.
//! * closing and unopened parenthesis panics during reading.
//! * [ ] TODO: signed depth might be interesting
//! * [Value::Sym] and [Value::Key] are stringish literals
//! with slightly different parsing rules.
//! * [Value::Num] is an unsigned integer literal.
//!```
//! use tengri_dsl::{*, Value::*};
//! let source = include_str!("../test.edn");
//! let mut view = TokenIter::new(source);
//! assert_eq!(view.peek(), Some(Token {
//! source,
//! start: 0,
//! length: source.len(),
//! value: Exp(0, TokenIter::new(&source[1..]))
//! }));
//!```
use crate::*;
use self::Value::*;
#[derive(Debug, Copy, Clone, Default, PartialEq)] pub struct Token<'a> {
pub source: &'a str,
pub start: usize,
pub length: usize,
pub value: Value<'a>,
}
#[derive(Debug, Copy, Clone, Default, PartialEq)] pub enum Value<'a> {
#[default] Nil,
Err(ParseError),
Num(usize),
Sym(&'a str),
Key(&'a str),
Exp(usize, TokenIter<'a>),
}
impl<'a> Token<'a> {
pub const fn new (source: &'a str, start: usize, length: usize, value: Value<'a>) -> Self {
Self { source, start, length, value }
}
pub const fn end (&self) -> usize {
self.start.saturating_add(self.length)
}
pub const fn slice (&'a self) -> &'a str {
self.slice_source(self.source)
//str_range(self.source, self.start, self.end())
}
pub const fn slice_source <'b> (&'a self, source: &'b str) -> &'b str {
str_range(source, self.start, self.end())
}
pub const fn slice_source_exp <'b> (&'a self, source: &'b str) -> &'b str {
str_range(source, self.start.saturating_add(1), self.end())
}
pub const fn value (&self) -> Value {
self.value
}
pub const fn error (self, error: ParseError) -> Self {
Self { value: Value::Err(error), ..self }
}
pub const fn grow (self) -> Self {
Self { length: self.length.saturating_add(1), ..self }
}
pub const fn grow_num (self, m: usize, c: char) -> Self {
match to_digit(c) {
Ok(n) => Self { value: Num(10*m+n), ..self.grow() },
Result::Err(e) => Self { value: Err(e), ..self.grow() },
}
}
pub const fn grow_key (self) -> Self {
let mut token = self.grow();
token.value = Key(token.slice_source(self.source));
token
}
pub const fn grow_sym (self) -> Self {
let mut token = self.grow();
token.value = Sym(token.slice_source(self.source));
token
}
pub const fn grow_exp (self) -> Self {
let mut token = self.grow();
if let Exp(depth, _) = token.value {
token.value = Exp(depth, TokenIter::new(token.slice_source_exp(self.source)));
} else {
unreachable!()
}
token
}
pub const fn grow_in (self) -> Self {
let mut token = self.grow_exp();
if let Value::Exp(depth, source) = token.value {
token.value = Value::Exp(depth.saturating_add(1), source)
} else {
unreachable!()
}
token
}
pub const fn grow_out (self) -> Self {
let mut token = self.grow_exp();
if let Value::Exp(depth, source) = token.value {
if depth > 0 {
token.value = Value::Exp(depth - 1, source)
} else {
return self.error(Unexpected(')'))
}
} else {
unreachable!()
}
token
}
}
#[cfg(test)] mod test_token_prop {
use proptest::prelude::*;
proptest! {
#[test] fn test_token_prop (
source in "\\PC*",
start in usize::MIN..usize::MAX,
length in usize::MIN..usize::MAX,
) {
let token = crate::Token {
source: &source,
start,
length,
value: crate::Value::Nil
};
let _ = token.slice();
}
}
}
#[cfg(test)] #[test] fn test_token () -> Result<(), Box<dyn std::error::Error>> {
let source = ":f00";
let mut token = Token { source, start: 0, length: 1, value: Sym(":") };
token = token.grow_sym();
assert_eq!(token, Token { source, start: 0, length: 2, value: Sym(":f") });
token = token.grow_sym();
assert_eq!(token, Token { source, start: 0, length: 3, value: Sym(":f0") });
token = token.grow_sym();
assert_eq!(token, Token { source, start: 0, length: 4, value: Sym(":f00") });
let src = "";
assert_eq!(None, SourceIter(src).next());
let src = " \n \r \t ";
assert_eq!(None, SourceIter(src).next());
let src = "7";
assert_eq!(Num(7), SourceIter(src).next().unwrap().0.value);
let src = " 100 ";
assert_eq!(Num(100), SourceIter(src).next().unwrap().0.value);
let src = " 9a ";
assert_eq!(Err(Unexpected('a')), SourceIter(src).next().unwrap().0.value);
let src = " :123foo ";
assert_eq!(Sym(":123foo"), SourceIter(src).next().unwrap().0.value);
let src = " \r\r\r\n\n\n@bar456\t\t\t\t\t\t";
assert_eq!(Sym("@bar456"), SourceIter(src).next().unwrap().0.value);
let src = "foo123";
assert_eq!(Key("foo123"), SourceIter(src).next().unwrap().0.value);
let src = "foo/bar";
assert_eq!(Key("foo/bar"), SourceIter(src).next().unwrap().0.value);
Ok(())
}