Compare commits

..

1 commit

Author SHA1 Message Date
5b45f164fb
macro wip 2024-02-27 03:37:09 +01:00
9 changed files with 70 additions and 223 deletions

View file

@ -4,3 +4,9 @@ edition = "2018"
name = "laddertypes" name = "laddertypes"
version = "0.1.0" version = "0.1.0"
#[lib]
#proc-macro = true
[dependencies]
laddertype-macro = { path = "./laddertype-macro" }

View file

@ -5,8 +5,6 @@ Rust Implementation of Ladder-Types (parsing, unification, rewriting, etc)
## Ladder Types ## Ladder Types
### Motivation
In order to implement complex datastructures and algorithms, usually In order to implement complex datastructures and algorithms, usually
many layers of abstraction are built ontop of each other. many layers of abstraction are built ontop of each other.
Consequently higher-level data types are encoded into lower-level data Consequently higher-level data types are encoded into lower-level data
@ -59,48 +57,6 @@ this:
1696093021:1696093039:1528324679:1539892301:1638141920:1688010253 1696093021:1696093039:1528324679:1539892301:1638141920:1688010253
``` ```
### Syntax
In their core form, type-terms can be one of the following:
- (**Atomic Type**) | `SomeTypeName`
- (**Literal Integer**) | `0` | `1` | `2` | ...
- (**Literal Character**) | `'a'` | `'b'` | `'c'` | ...
- (**Literal String**) | `"abc"`
- (**Parameter Application**) | `<T1 T2>` given `T1` and `T2` are type-terms
- (**Ladder**) | `T1 ~ T2` given `T1` and `T2` are type-terms
Ontop of that, the following syntax-sugar is defined:
#### Complex Types
- `[ T ]` <===> `<Seq T>`
- `{ a:A b:B }` <===> `<Struct <"a" A> <"b" B>>`
- `a:A | b:B` <===> `<Enum <"a" A> <"b" B>>`
#### Function Types
- `A -> B` <===> `<Fn A B>`
#### Reference Types
- `*A` <===> `<Ptr A>`
- `&A` <===> `<ConstRef A>`
- `&!A` <===> `<MutRef A>`
### Equivalences
#### Currying
`<<A B> C>` <===> `<A B C>`
#### Ladder-Normal-Form
exhaustively apply `<A B~C>` ===> `<A B>~<A C>`
e.g. `[<Digit 10>]~[Char]~[Ascii]` is in **LNF**
#### Parameter-Normal-Form
exhaustively apply `<A B>~<A C>` ===> `<A B~C>`
e.g. `[<Digit 10>~Char~Ascii]` is in **PNF**
## How to use this crate ## How to use this crate
```rust ```rust
@ -117,19 +73,6 @@ fn main() {
} }
``` ```
## Roadmap
- [x] (Un-)Parsing
- [x] (De-)Currying
- [x] Unification
- [x] Ladder-Normal-Form
- [x] Parameter-Normal-Form
- [ ] (De)-Sugaring
- [ ] Seq
- [ ] Enum
- [ ] Struct
- [ ] References
- [ ] Function
## License ## License
[GPLv3](COPYING) [GPLv3](COPYING)

View file

@ -7,7 +7,6 @@ pub mod parser;
pub mod unparser; pub mod unparser;
pub mod curry; pub mod curry;
pub mod lnf; pub mod lnf;
pub mod pnf;
pub mod subtype; pub mod subtype;
pub mod unification; pub mod unification;

View file

@ -20,10 +20,14 @@ pub enum ParseError {
//<<<<>>>><<>><><<>><<<*>>><<>><><<>><<<<>>>>\\ //<<<<>>>><<>><><<>><<<*>>><<>><><<>><<<<>>>>\\
impl TypeDict { impl TypeDict {
pub fn parse(&mut self, s: &str) -> Result<TypeTerm, ParseError> { pub fn parse(&mut self, s: &str) -> Result<TypeTerm, ParseError> {
let mut tokens = LadderTypeLexer::from(s.chars()).peekable(); let mut tokens = LadderTypeLexer::from(s.chars());
self.parse_tokens( tokens.peekable() )
}
pub fn parse_tokens<It>(&mut self, mut tokens: Peekable<It>) -> Result<TypeTerm, ParseError>
where It: Iterator<Item = Result<LadderTypeToken, LexError>> {
match self.parse_ladder(&mut tokens) { match self.parse_ladder(&mut tokens) {
Ok(t) => { Ok(t) => {
if let Some(_tok) = tokens.peek() { if let Some(_tok) = tokens.peek() {
@ -36,8 +40,8 @@ impl TypeDict {
} }
} }
fn parse_app<It>(&mut self, tokens: &mut Peekable<LadderTypeLexer<It>>) -> Result<TypeTerm, ParseError> fn parse_app<It>(&mut self, tokens: &mut Peekable<It>) -> Result<TypeTerm, ParseError>
where It: Iterator<Item = char> where It: Iterator<Item = Result<LadderTypeToken, LexError>>
{ {
let mut args = Vec::new(); let mut args = Vec::new();
while let Some(tok) = tokens.peek() { while let Some(tok) = tokens.peek() {
@ -57,8 +61,8 @@ impl TypeDict {
Err(ParseError::UnexpectedEnd) Err(ParseError::UnexpectedEnd)
} }
fn parse_rung<It>(&mut self, tokens: &mut Peekable<LadderTypeLexer<It>>) -> Result<TypeTerm, ParseError> fn parse_rung<It>(&mut self, tokens: &mut Peekable<It>) -> Result<TypeTerm, ParseError>
where It: Iterator<Item = char> where It: Iterator<Item = Result<LadderTypeToken, LexError>>
{ {
match tokens.next() { match tokens.next() {
Some(Ok(LadderTypeToken::Open)) => self.parse_app(tokens), Some(Ok(LadderTypeToken::Open)) => self.parse_app(tokens),
@ -79,8 +83,8 @@ impl TypeDict {
} }
} }
fn parse_ladder<It>(&mut self, tokens: &mut Peekable<LadderTypeLexer<It>>) -> Result<TypeTerm, ParseError> fn parse_ladder<It>(&mut self, tokens: &mut Peekable<It>) -> Result<TypeTerm, ParseError>
where It: Iterator<Item = char> where It: Iterator<Item = Result<LadderTypeToken, LexError>>
{ {
let mut rungs = Vec::new(); let mut rungs = Vec::new();

View file

@ -1,113 +0,0 @@
use crate::term::TypeTerm;
//<<<<>>>><<>><><<>><<<*>>><<>><><<>><<<<>>>>\\
impl TypeTerm {
/// transmute type into Parameter-Normal-Form (PNF)
///
/// Example:
/// ```ignore
/// <Seq <Digit 10>>~<Seq Char>
/// ⇒ <Seq <Digit 10>~Char>
/// ```
pub fn param_normalize(self) -> Self {
match self {
TypeTerm::Ladder(mut rungs) => {
if rungs.len() > 0 {
// normalize all rungs separately
for r in rungs.iter_mut() {
*r = r.clone().param_normalize();
}
// take top-rung
match rungs.remove(0) {
TypeTerm::App(params_top) => {
let mut params_ladders = Vec::new();
let mut tail : Vec<TypeTerm> = Vec::new();
// append all other rungs to ladders inside
// the application
for p in params_top {
params_ladders.push(vec![ p ]);
}
for r in rungs {
match r {
TypeTerm::App(mut params_rung) => {
if params_rung.len() > 0 {
let mut first_param = params_rung.remove(0);
if first_param == params_ladders[0][0] {
for (l, p) in params_ladders.iter_mut().skip(1).zip(params_rung) {
l.push(p.param_normalize());
}
} else {
params_rung.insert(0, first_param);
tail.push(TypeTerm::App(params_rung));
}
}
}
TypeTerm::Ladder(mut rs) => {
for r in rs {
tail.push(r.param_normalize());
}
}
atomic => {
tail.push(atomic);
}
}
}
let head = TypeTerm::App(
params_ladders.into_iter()
.map(
|mut l| {
l.dedup();
match l.len() {
0 => TypeTerm::unit(),
1 => l.remove(0),
_ => TypeTerm::Ladder(l).param_normalize()
}
}
)
.collect()
);
if tail.len() > 0 {
tail.insert(0, head);
TypeTerm::Ladder(tail)
} else {
head
}
}
TypeTerm::Ladder(mut r) => {
r.append(&mut rungs);
TypeTerm::Ladder(r)
}
atomic => {
rungs.insert(0, atomic);
TypeTerm::Ladder(rungs)
}
}
} else {
TypeTerm::unit()
}
}
TypeTerm::App(params) => {
TypeTerm::App(
params.into_iter()
.map(|p| p.param_normalize())
.collect())
}
atomic => atomic
}
}
}
//<<<<>>>><<>><><<>><<<*>>><<>><><<>><<<<>>>>\\

View file

@ -153,4 +153,3 @@ fn test_lexer_large() {
assert_eq!( lex.next(), None ); assert_eq!( lex.next(), None );
} }

View file

@ -3,7 +3,6 @@ pub mod lexer;
pub mod parser; pub mod parser;
pub mod curry; pub mod curry;
pub mod lnf; pub mod lnf;
pub mod pnf;
pub mod subtype; pub mod subtype;
pub mod substitution; pub mod substitution;
pub mod unification; pub mod unification;

View file

@ -143,7 +143,7 @@ fn test_parser_ladder_large() {
assert_eq!( assert_eq!(
TypeDict::new().parse( TypeDict::new().parse(
"<Seq Date "<Seq Date
~<TimeSince UnixEpoch> ~<TimeSince UnixEpoch>
~<Duration Seconds> ~<Duration Seconds>
~ ~
~<PosInt 10 BigEndian> ~<PosInt 10 BigEndian>
@ -204,3 +204,54 @@ fn test_parser_ladder_large() {
); );
} }
macro_rules! lt_tokenize {
($symbol:ident) => {
crate::lexer::LadderTypeToken::Symbol( "$symbol".into() )
}
(< $rest::tt) => {
crate::lexer::LadderTypeToken::Open,
lt_tokenize!($rest)
}
(> $rest::tt) => {
crate::lexer::LadderTypeToken::Close,
lt_tokenize!($rest)
}
(~ $rest::tt) => {
crate::lexer::LadderTypeToken::Ladder,
lt_tokenize!($rest)
}
}
macro_rules! lt_parse {
($dict:ident, $tokens:tt*) => {
$dict.parse_tokens(
vec![
lt_tokenize!($tokens)
].into_iter().peekable()
)
}
}
#[test]
fn test_proc_macro() {
use laddertype_macro::laddertype;
use crate::lexer::LadderTypeToken;
let mut dict = TypeDict::new();
let t1 = dict.parse_tokens(vec![
Ok(crate::lexer::LadderTypeToken::Open),
Ok(crate::lexer::LadderTypeToken::Symbol("Seq".into())),
Ok(crate::lexer::LadderTypeToken::Symbol("Char".into())),
Ok(crate::lexer::LadderTypeToken::Close)
].into_iter().peekable());
let t2 = dict.parse_tokens(vec![
lt_tokenize!{ <Seq Char> }
].into_iter().peekable());
//lt_parse!( dict, <Seq Char> );
assert_eq!(t1, t2);
}

View file

@ -1,41 +0,0 @@
use crate::dict::TypeDict;
#[test]
fn test_param_normalize() {
let mut dict = TypeDict::new();
assert_eq!(
dict.parse("A~B~C").expect("parse error"),
dict.parse("A~B~C").expect("parse error").param_normalize(),
);
assert_eq!(
dict.parse("<A B>~C").expect("parse error"),
dict.parse("<A B>~C").expect("parse error").param_normalize(),
);
assert_eq!(
dict.parse("<A B~C>").expect("parse error"),
dict.parse("<A B>~<A C>").expect("parse error").param_normalize(),
);
assert_eq!(
dict.parse("<A B~C D~E>").expect("parse error"),
dict.parse("<A B D>~<A C D>~<A C E>").expect("parse errror").param_normalize(),
);
assert_eq!(
dict.parse("<Seq <Digit 10>~Char>").expect("parse error"),
dict.parse("<Seq <Digit 10>>~<Seq Char>").expect("parse errror").param_normalize(),
);
assert_eq!(
dict.parse("<A <B C~D~E> F~G H H>").expect("parse error"),
dict.parse("<A <B C> F H H>
~<A <B D> F H H>
~<A <B E> F H H>
~<A <B E> G H H>").expect("parse errror")
.param_normalize(),
);
}