add functions and variables to language

This commit is contained in:
2024-10-12 16:50:17 -04:00
parent 498d04c376
commit 2d269203e6
3 changed files with 328 additions and 50 deletions

View File

@@ -1,5 +1,8 @@
use crate::common::{Context, Error}; use crate::common::{Context, Error};
use std::collections::HashMap;
use std::borrow::Cow;
mod tokenizer; mod tokenizer;
mod parse; mod parse;
@@ -7,9 +10,17 @@ fn evaluate(expr: &str) -> Result<f64, Error> {
let tokens = tokenizer::Token::tokenize(expr)?; let tokens = tokenizer::Token::tokenize(expr)?;
let mut tokens = tokens.iter(); let mut tokens = tokens.iter();
let tree = parse::ParseTree::new(&mut tokens)?; let globals = HashMap::new();
let locals = HashMap::new();
let mut locals = Cow::Borrowed(&locals);
Ok(tree.evaluate()) let tree = parse::ParseTree::parse(&mut tokens, &globals, &mut locals)?;
let mut globals = HashMap::new();
let locals = HashMap::new();
let mut locals = Cow::Borrowed(&locals);
tree.evaluate(&mut globals, &mut locals)
} }
/// Evaluates an expression (uses Polish Notation) /// Evaluates an expression (uses Polish Notation)

View File

@@ -1,38 +1,192 @@
use std::error::Error; use std::error::Error;
use std::borrow::Cow;
use std::collections::HashMap;
use std::fmt::Display; use std::fmt::Display;
use super::tokenizer::{Token, Op}; use crate::common;
use super::tokenizer::{self, Token};
#[derive(Debug)] #[derive(Clone, Debug)]
pub enum ParseTree { pub enum ParseTree<'a> {
Leaf(Token), Add(Box<ParseTree<'a>>, Box<ParseTree<'a>>),
Branch(Op, Box<ParseTree>, Box<ParseTree>), Sub(Box<ParseTree<'a>>, Box<ParseTree<'a>>),
Mul(Box<ParseTree<'a>>, Box<ParseTree<'a>>),
Div(Box<ParseTree<'a>>, Box<ParseTree<'a>>),
Exp(Box<ParseTree<'a>>, Box<ParseTree<'a>>),
Equ(&'a str, Box<ParseTree<'a>>, Box<ParseTree<'a>>),
GlobalEqu(&'a str, Box<ParseTree<'a>>),
Compose(Box<ParseTree<'a>>, Box<ParseTree<'a>>),
Id(Box<ParseTree<'a>>),
FunctionDeclaration(&'a str, Vec<Object<'a>>, Box<ParseTree<'a>>, Box<ParseTree<'a>>),
FunctionApplication(&'a str, Vec<ParseTree<'a>>),
Variable(&'a str),
Scalar(f64),
}
#[derive(Clone, Debug)]
pub struct FunctionDeclaration<'a> {
name: &'a str,
args: Vec<Object<'a>>,
}
#[derive(Clone, Debug)]
pub struct Function<'a> {
decl: FunctionDeclaration<'a>,
body: Option<Box<ParseTree<'a>>>, // may be used in declarations where a value isn't provided
}
#[derive(Clone, Debug)]
pub struct Variable<'a> {
name: &'a str,
body: Option<Box<ParseTree<'a>>>, // may be used in declarations where a value isn't provided
}
impl<'a> Variable<'a> {
pub fn new(name: &'a str, body: Option<Box<ParseTree<'a>>>) -> Self {
Self { name, body }
}
}
#[derive(Clone, Debug)]
pub enum Object<'a> {
Variable(Variable<'a>),
Func(Function<'a>),
} }
#[derive(Debug, Clone)] #[derive(Debug, Clone)]
pub enum ParseError { pub enum ParseError {
UnexpectedEndInput, UnexpectedEndInput,
IdentifierUndefined(String),
InvalidIdentifier,
FunctionUndefined,
VariableUndefined,
} }
impl Display for ParseError { impl Display for ParseError {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self { match self {
ParseError::UnexpectedEndInput => write!(f, "Input ended unexpectedly"), ParseError::UnexpectedEndInput => write!(f, "Input ended unexpectedly"),
ParseError::IdentifierUndefined(name) => write!(f, "Undefined variable `{}`", name),
ParseError::InvalidIdentifier => write!(f, "Invalid identifier"),
ParseError::FunctionUndefined => write!(f, "Undefined function"),
ParseError::VariableUndefined => write!(f, "Undefined variable"),
} }
} }
} }
impl Error for ParseError {} impl Error for ParseError {}
impl ParseTree { impl<'a> ParseTree<'a> {
pub fn new<'a, I: Iterator<Item = &'a Token>>(tokens: &mut I) -> Result<Self, ParseError> { pub fn parse<I: Iterator<Item = &'a Token>>(
tokens: &mut I,
globals: &HashMap<String, FunctionDeclaration<'a>>,
locals: &mut Cow<HashMap<String, FunctionDeclaration<'a>>>) -> Result<Self, ParseError>
{
if let Some(token) = tokens.next() { if let Some(token) = tokens.next() {
match token { match token {
Token::Scalar(_) | Token::Identifier(_) => Ok(Self::Leaf(token.clone())), // Just return scalars
Token::Operator(op) => { Token::Scalar(x) => Ok(ParseTree::Scalar(*x)),
let left = ParseTree::new(tokens)?;
let right = ParseTree::new(tokens)?;
Ok(Self::Branch(op.clone(), Box::new(left), Box::new(right))) // Get any identifiers as objects from local first then global scope
Token::Identifier(ident) => {
// If it is found to be a function, get its argument count.
// During parsing, we only keep track of function definitions
// so that we know how many arguments it takes
if let Some(decl) = locals.clone().get(ident).or(globals.get(ident)) {
let args = decl.args.iter()
.map(|_| ParseTree::parse(tokens, globals, locals)).collect::<Result<Vec<_>, ParseError>>()?;
Ok(ParseTree::FunctionApplication(ident, args))
} else {
Ok(ParseTree::Variable(ident))
}
}
Token::Operator(op) => {
match op {
tokenizer::Op::Add => Ok(ParseTree::Add(
Box::new(ParseTree::parse(tokens, globals, locals)?),
Box::new(ParseTree::parse(tokens, globals, locals)?)
)),
tokenizer::Op::Sub => Ok(ParseTree::Sub(
Box::new(ParseTree::parse(tokens, globals, locals)?),
Box::new(ParseTree::parse(tokens, globals, locals)?)
)),
tokenizer::Op::Mul => Ok(ParseTree::Mul(
Box::new(ParseTree::parse(tokens, globals, locals)?),
Box::new(ParseTree::parse(tokens, globals, locals)?)
)),
tokenizer::Op::Div => Ok(ParseTree::Div(
Box::new(ParseTree::parse(tokens, globals, locals)?),
Box::new(ParseTree::parse(tokens, globals, locals)?)
)),
tokenizer::Op::Exp => Ok(ParseTree::Exp(
Box::new(ParseTree::parse(tokens, globals, locals)?),
Box::new(ParseTree::parse(tokens, globals, locals)?)
)),
tokenizer::Op::Equ | tokenizer::Op::LazyEqu => {
let token = tokens.next().ok_or(ParseError::UnexpectedEndInput)?;
if let Token::Identifier(ident) = token {
Ok(ParseTree::Equ(
ident,
Box::new(ParseTree::parse(tokens, globals, locals)?),
Box::new(ParseTree::parse(tokens, globals, locals)?)))
} else {
Err(ParseError::InvalidIdentifier)
}
}
tokenizer::Op::GlobalEqu | tokenizer::Op::LazyGlobalEqu => {
let token = tokens.next().ok_or(ParseError::UnexpectedEndInput)?;
if let Token::Identifier(ident) = token {
Ok(ParseTree::GlobalEqu(
ident,
Box::new(ParseTree::parse(tokens, globals, locals)?)
))
} else {
Err(ParseError::InvalidIdentifier)
}
}
tokenizer::Op::FunctionDeclare(arg_count) => {
let token = tokens.next().ok_or(ParseError::UnexpectedEndInput)?;
if let Token::Identifier(ident) = token {
let args: Vec<Object> = tokens.take(*arg_count)
.map(|token| match token {
Token::Identifier(s)
=> Ok(Object::Variable(Variable::new(s, None))),
_ => Err(ParseError::InvalidIdentifier),
}).collect::<Result<_, ParseError>>()?;
if args.len() < *arg_count {
return Err(ParseError::InvalidIdentifier);
}
let locals = locals.to_mut();
locals.insert(ident.clone(), FunctionDeclaration {
name: ident,
args: args.clone()
});
Ok(ParseTree::FunctionDeclaration(
ident,
args,
Box::new(ParseTree::parse(tokens, globals, &mut Cow::Borrowed(&*locals))?),
Box::new(ParseTree::parse(tokens, globals, &mut Cow::Borrowed(&*locals))?)))
} else {
Err(ParseError::InvalidIdentifier)
}
}
tokenizer::Op::Compose => {
Ok(ParseTree::Compose(
Box::new(ParseTree::parse(tokens, globals, locals)?),
Box::new(ParseTree::parse(tokens, globals, locals)?)
))
}
tokenizer::Op::Id =>
Ok(ParseTree::Id(Box::new(ParseTree::parse(tokens, globals, locals)?)))
}
} }
} }
} else { } else {
@@ -40,18 +194,121 @@ impl ParseTree {
} }
} }
pub fn evaluate(&self) -> f64 { pub fn evaluate(
self,
globals: &mut HashMap<String, Object<'a>>,
locals: &mut Cow<HashMap<String, Object<'a>>>) -> Result<f64, common::Error>
{
match self { match self {
ParseTree::Leaf(Token::Scalar(value)) => *value, ParseTree::Add(l, r) => Ok(l.evaluate(globals, locals)? + r.evaluate(globals, locals)?),
ParseTree::Leaf(Token::Identifier(_)) => unimplemented!(), ParseTree::Sub(l, r) => Ok(l.evaluate(globals, locals)? + r.evaluate(globals, locals)?),
ParseTree::Leaf(Token::Operator(_)) => panic!("This absolutely should not happen"), ParseTree::Mul(l, r) => Ok(l.evaluate(globals, locals)? * r.evaluate(globals, locals)?),
ParseTree::Branch(op, left, right) => match op { ParseTree::Div(l, r) => Ok(l.evaluate(globals, locals)? / r.evaluate(globals, locals)?),
Op::Add => left.evaluate() + right.evaluate(), ParseTree::Exp(l, r)
Op::Sub => left.evaluate() - right.evaluate(), => Ok(l.evaluate(globals, locals)?.powf(r.evaluate(globals, locals)?)),
Op::Mul => left.evaluate() * right.evaluate(), ParseTree::Equ(ident, value, body) => {
Op::Div => left.evaluate() / right.evaluate(), let value = value.evaluate(globals, locals)?;
Op::Exp => left.evaluate().powf(right.evaluate()),
let locals = locals.to_mut();
locals.insert(ident.to_string(),
Object::Variable(
Variable::new(ident, Some(Box::new(ParseTree::Scalar(value))))));
body.evaluate(globals, &mut Cow::Borrowed(&locals))
} }
ParseTree::GlobalEqu(ident, body) => {
globals.insert(ident.to_string(),
Object::Variable(Variable::new(ident, Some(body.clone()))));
Ok(0.0)
}
ParseTree::Compose(l, r) => {
let _ = l.evaluate(globals, locals);
r.evaluate(globals, locals)
}
ParseTree::Id(body) => body.evaluate(globals, locals),
ParseTree::FunctionDeclaration(name, args, body, cont) => {
let locals = locals.to_mut();
locals.insert(name.to_string(), Object::Func(Function {
decl: FunctionDeclaration {
name,
args: args.clone(),
},
body: Some(body.clone())
}));
cont.evaluate(globals, &mut Cow::Borrowed(&locals))
}
ParseTree::FunctionApplication(name, params) => {
let locals = locals.to_mut();
let obj = locals.get(name).or(globals.get(name)).cloned();
if let Some(Object::Func(func)) = obj {
for (param, arg) in params.iter().zip(func.decl.args.iter()) {
match arg {
Object::Variable(v)
=> locals.insert(
v.name.to_string(),
Object::Variable(
Variable::new(
&v.name,
Some(Box::new(
ParseTree::Scalar(
param.clone().evaluate(
globals, &mut Cow::Borrowed(&locals))?)))))),
Object::Func(func)
=> locals.insert(
func.decl.name.to_string(),
Object::Func(func.clone()))
};
}
func.body.ok_or(ParseError::FunctionUndefined.into())
.and_then(|body|
body.clone().evaluate(globals, &mut Cow::Borrowed(&locals)))
} else {
Err(ParseError::FunctionUndefined.into())
}
}
ParseTree::Variable(ident) => {
let locals = locals.to_mut();
let obj = locals.get(ident).or(globals.get(ident)).cloned();
if let Some(Object::Variable(obj)) = obj {
return obj.body.clone().ok_or(ParseError::VariableUndefined.into())
.and_then(|body| body.clone().evaluate(globals, &mut Cow::Borrowed(&locals)));
}
Err(ParseError::IdentifierUndefined(ident.to_string()).into())
}
ParseTree::Scalar(x) => Ok(x),
} }
} }
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_parse() {
let tokens = Token::tokenize("= x 15 : square x ** x 2 square x").expect("failed to tokenize");
let mut tokens = tokens.iter();
let globals = HashMap::new();
let locals = HashMap::new();
let mut locals = Cow::Borrowed(&locals);
let tree = ParseTree::parse(&mut tokens, &globals, &mut locals).expect("failed to parse");
eprintln!("{tree:?}");
let mut globals = HashMap::new();
let locals = HashMap::new();
let mut locals = Cow::Borrowed(&locals);
eprintln!("{}", tree.evaluate(&mut globals, &mut locals).expect("failed to evaluate"));
}
} }

View File

@@ -1,4 +1,5 @@
use std::error; use std::error;
use crate::common::Error; use crate::common::Error;
use std::fmt::{Display, Formatter}; use std::fmt::{Display, Formatter};
use std::str::FromStr; use std::str::FromStr;
@@ -10,6 +11,13 @@ pub enum Op {
Mul, Mul,
Div, Div,
Exp, Exp,
Equ,
LazyEqu,
GlobalEqu,
LazyGlobalEqu,
FunctionDeclare(usize),
Compose,
Id,
} }
#[derive(Debug, Clone)] #[derive(Debug, Clone)]
@@ -43,44 +51,46 @@ impl Token {
Token::Scalar(value) Token::Scalar(value)
} }
pub fn add() -> Token { pub fn operator(op: Op) -> Token {
Token::Operator(Op::Add) Token::Operator(op)
} }
pub fn sub() -> Token {
Token::Operator(Op::Sub)
}
pub fn mul() -> Token {
Token::Operator(Op::Mul)
}
pub fn div() -> Token {
Token::Operator(Op::Div)
}
pub fn exp() -> Token {
Token::Operator(Op::Exp)
}
pub fn tokenize(s: &str) -> Result<Vec<Self>, Error> { pub fn tokenize(s: &str) -> Result<Vec<Self>, Error> {
s.split_whitespace().map(Token::from_str).collect() s.split_whitespace().map(Token::from_str).collect()
} }
} }
fn get_dot_count<I: Iterator<Item = char>>(s: I) -> usize {
s.fold(0, |acc, c| acc + match c {
':' => 2,
'.' => 1,
_ => 0,
})
}
impl FromStr for Token { impl FromStr for Token {
type Err = Error; type Err = Error;
fn from_str(s: &str) -> Result<Self, Self::Err> { fn from_str(s: &str) -> Result<Self, Self::Err> {
match s { let s = if s.starts_with("\\") { s.chars().skip(1).collect() } else { s.to_string() };
match s.as_str() {
// First check if s is an operator // First check if s is an operator
"+" => Ok(Token::add()), "+" => Ok(Token::operator(Op::Add)),
"-" => Ok(Token::sub()), "-" => Ok(Token::operator(Op::Sub)),
"*" => Ok(Token::mul()), "*" => Ok(Token::operator(Op::Mul)),
"**" => Ok(Token::exp()), "**" => Ok(Token::operator(Op::Exp)),
"/" => Ok(Token::div()), "/" => Ok(Token::operator(Op::Div)),
"=" => Ok(Token::operator(Op::Equ)),
"." => Ok(Token::operator(Op::LazyEqu)),
"=>" => Ok(Token::operator(Op::GlobalEqu)),
".>" => Ok(Token::operator(Op::LazyGlobalEqu)),
"~" => Ok(Token::operator(Op::Compose)),
"," => Ok(Token::operator(Op::Id)),
_ => { _ => {
if s.starts_with(|c| char::is_digit(c, 10)) { // variable length operators
if s.starts_with(':') {
Ok(Token::operator(Op::FunctionDeclare(1 + get_dot_count(s[1..].chars()))))
} else if s.starts_with(|c| char::is_digit(c, 10)) {
Ok(Token::scalar(s.parse()?)) Ok(Token::scalar(s.parse()?))
} else if s.starts_with(char::is_alphabetic) } else if s.starts_with(char::is_alphabetic)
&& s.chars().skip(1).all(char::is_alphanumeric) { && s.chars().skip(1).all(char::is_alphanumeric) {