From ff05c13844791ed9537d64d3c8c5496886eb04bf Mon Sep 17 00:00:00 2001
From: Louis Capitanchik <contact@louiscap.co>
Date: Sun, 28 May 2023 05:02:09 +0100
Subject: [PATCH] Replace peg parse_ fns with lalrpop parse_ fns
---
.gitignore | 2 +-
forge-script-lang/src/build.rs | 3 +-
forge-script-lang/src/error/mod.rs | 296 ++++++++++++++----
forge-script-lang/src/lexer/keywords.rs | 2 +-
forge-script-lang/src/lexer/mod.rs | 18 ++
forge-script-lang/src/parser/ast.rs | 64 ++--
...e_script.lalrpop => forge_grammar.lalrpop} | 123 +++++++-
forge-script-lang/src/parser/forge_script.rs | 71 +++++
forge-script-lang/src/parser/lr_grammar.rs | 45 ---
forge-script-lang/src/parser/mod.rs | 85 ++---
forge-script-lang/src/utilities.rs | 20 ++
11 files changed, 548 insertions(+), 181 deletions(-)
rename forge-script-lang/src/parser/{forge_script.lalrpop => forge_grammar.lalrpop} (51%)
create mode 100644 forge-script-lang/src/parser/forge_script.rs
delete mode 100644 forge-script-lang/src/parser/lr_grammar.rs
diff --git a/.gitignore b/.gitignore
index ff2439c..e28572c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -130,4 +130,4 @@ target/
# Generated Parser
-forge-script-lang/src/parser/forge_script.rs
\ No newline at end of file
+forge-script-lang/src/parser/forge_grammar.rs
\ No newline at end of file
diff --git a/forge-script-lang/src/build.rs b/forge-script-lang/src/build.rs
index 3094868..194123d 100644
--- a/forge-script-lang/src/build.rs
+++ b/forge-script-lang/src/build.rs
@@ -1,5 +1,6 @@
pub fn main() {
lalrpop::Configuration::new()
- .process_file("src/parser/forge_script.lalrpop")
+ .generate_in_source_tree()
+ .process_file("src/parser/forge_grammar.lalrpop")
.expect("Failed to parse language grammar");
}
diff --git a/forge-script-lang/src/error/mod.rs b/forge-script-lang/src/error/mod.rs
index b3986af..f741161 100644
--- a/forge-script-lang/src/error/mod.rs
+++ b/forge-script-lang/src/error/mod.rs
@@ -1,8 +1,12 @@
-use crate::lexer::Span;
+use crate::lexer::{ScriptTokenType, Span};
use crate::parse::ScriptToken;
-use peg::error::ExpectedSet;
+use crate::utilities::offset_to_line_column;
+use lalrpop_util::ParseError as BaseLalrError;
use std::error::Error;
use std::fmt::{Display, Formatter};
+use std::process::{ExitCode, Termination};
+
+pub type LalrError<'a> = BaseLalrError<usize, ScriptTokenType<'a>, TokenError<'a>>;
#[derive(Debug)]
pub enum TokenErrorKind<'a> {
@@ -61,13 +65,53 @@ pub enum ForgeErrorKind<'a> {
found: ScriptToken<'a>,
expected: peg::error::ExpectedSet,
},
+ InvalidToken {
+ location: TokenIndex,
+ },
+ UnexpectedEof {
+ expected: Vec<String>,
+ },
+ UnrecognizedToken {
+ token: ScriptTokenType<'a>,
+ span: ErrorSpan,
+ expected: Vec<String>,
+ },
+ ExpectedEof {
+ token: ScriptTokenType<'a>,
+ span: ErrorSpan,
+ },
+ Custom(String),
}
+pub type TokenIndex = usize;
+pub type ErrorSpan = (TokenIndex, TokenIndex);
+
#[derive(Debug, PartialEq)]
pub struct ForgeError<'a> {
pub kind: ForgeErrorKind<'a>,
}
+impl<'a> Termination for ForgeError<'a> {
+ fn report(self) -> ExitCode {
+ match self.kind {
+ ForgeErrorKind::IncompleteInput => ExitCode::from(101),
+ ForgeErrorKind::LexerError(_) => ExitCode::from(102),
+ ForgeErrorKind::UnexpectedToken { .. } => ExitCode::from(103),
+ ForgeErrorKind::InvalidToken { .. } => ExitCode::from(104),
+ ForgeErrorKind::UnexpectedEof { .. } => ExitCode::from(105),
+ ForgeErrorKind::UnrecognizedToken { .. } => ExitCode::from(106),
+ ForgeErrorKind::ExpectedEof { .. } => ExitCode::from(107),
+ ForgeErrorKind::Custom(_) => ExitCode::from(1),
+ }
+ }
+}
+
+impl<'a> From<ForgeErrorKind<'a>> for ForgeError<'a> {
+ fn from(value: ForgeErrorKind<'a>) -> Self {
+ Self { kind: value }
+ }
+}
+
impl<'a> From<ParseError<'a>> for ForgeError<'a> {
fn from(value: ParseError<'a>) -> Self {
match value.kind {
@@ -91,6 +135,31 @@ impl<'a> From<TokenError<'a>> for ForgeError<'a> {
}
}
+impl<'a> From<LalrError<'a>> for ForgeError<'a> {
+ fn from(value: LalrError<'a>) -> Self {
+ match value {
+ LalrError::InvalidToken { location } => {
+ ForgeErrorKind::InvalidToken { location }.into()
+ }
+ LalrError::UnrecognizedEof { expected, .. } => {
+ ForgeErrorKind::UnexpectedEof { expected }.into()
+ }
+ LalrError::UnrecognizedToken { token, expected } => ForgeErrorKind::UnrecognizedToken {
+ expected,
+ token: token.1,
+ span: (token.0, token.2),
+ }
+ .into(),
+ LalrError::ExtraToken { token } => ForgeErrorKind::ExpectedEof {
+ token: token.1,
+ span: (token.0, token.2),
+ }
+ .into(),
+ LalrError::User { error } => ForgeErrorKind::Custom(format!("{}", error)).into(),
+ }
+ }
+}
+
pub type ForgeResult<'a, T> = Result<T, ForgeError<'a>>;
pub fn print_unexpected_token<'a>(
@@ -134,79 +203,178 @@ pub fn print_unexpected_token<'a>(
eprintln!("|\n| Failed To Parse: expected {}", expected);
}
-pub struct ErrorFormat<'a>(pub &'a str, pub &'a ScriptToken<'a>, pub &'a ExpectedSet);
+pub fn print_forge_error<'a>(source: &'a str, fe: &'a ForgeError) {
+ eprintln!("{}", format_forge_error(source, fe));
+}
+
+pub fn format_forge_error<'a>(source: &'a str, fe: &'a ForgeError) -> String {
+ match &fe.kind {
+ ForgeErrorKind::IncompleteInput => String::from("| Unexpected end of file"),
+ ForgeErrorKind::LexerError(err) => format!("| {}", err),
+ ForgeErrorKind::UnexpectedToken { found, expected } => {
+ format!(
+ "{}",
+ SourcePrinter(
+ source,
+ format!("{}", found.token_type),
+ HighlightConfig {
+ line: found.position.location_line() as usize,
+ column: found.position.get_column(),
+ highlight_len: found.token_type.len(),
+ },
+ Some(format!(
+ "| Found {}, expected one of {}",
+ found.token_type,
+ expected
+ .tokens()
+ .collect::<Vec<&str>>()
+ .as_slice()
+ .join(", ")
+ ))
+ ),
+ )
+ }
+ ForgeErrorKind::InvalidToken { location } => {
+ let (line, column) = offset_to_line_column(source, *location);
+ format!(
+ "{}",
+ SourcePrinter(
+ source,
+ String::from("invalid token"),
+ HighlightConfig {
+ line,
+ column,
+ highlight_len: 1,
+ },
+ None,
+ )
+ )
+ }
+ ForgeErrorKind::UnexpectedEof { expected } => {
+ format!(
+ "| Unexpected end of file, expected one of {}",
+ expected.as_slice().join(", ")
+ )
+ }
+ ForgeErrorKind::UnrecognizedToken {
+ token,
+ span,
+ expected,
+ } => {
+ let (line, column) = offset_to_line_column(source, span.0);
+ format!(
+ "{}",
+ SourcePrinter(
+ source,
+ format!("{}", token),
+ HighlightConfig {
+ line,
+ column,
+ highlight_len: token.len(),
+ },
+ Some(format!(
+ "| Found {}, expected one of {}",
+ token,
+ expected.as_slice().join(", ")
+ ))
+ ),
+ )
+ }
+ ForgeErrorKind::ExpectedEof { token, span } => {
+ let (line, column) = offset_to_line_column(source, span.0);
+ format!(
+ "{}",
+ SourcePrinter(
+ source,
+ format!("{}", token),
+ HighlightConfig {
+ line,
+ column,
+ highlight_len: token.len(),
+ },
+ Some(format!("| Expected EOF, found {}", token))
+ ),
+ )
+ }
+ ForgeErrorKind::Custom(msg) => format!("| {}", msg),
+ }
+}
+
+#[derive(Clone, Copy)]
+pub struct HighlightConfig {
+ line: usize,
+ column: usize,
+ highlight_len: usize,
+}
-impl<'a> Display for ErrorFormat<'a> {
+struct SourcePrinter<'a>(&'a str, String, HighlightConfig, Option<String>);
+impl<'a> Display for SourcePrinter<'a> {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
- let ErrorFormat(source, token, expected) = self;
-
- let line = token.position.location_line() as usize;
- let column = token.position.get_column();
-
- let previous_line = if line > 1 {
- source.lines().nth(line - 2)
- } else {
- None
- };
- let source_line = source.lines().nth(line - 1).expect("Missing line");
- let next_line = source.lines().nth(line);
-
- let largest_line_num = line.max(line.saturating_sub(1)).max(line.saturating_add(1));
- let number_length = format!("{}", largest_line_num).len();
-
- writeln!(f, "| Script error on line {} at \"{}\"\n|", line, token)?;
- if let Some(prev) = previous_line {
- writeln!(
- f,
- "| [{:>width$}] {}",
- line - 1,
- prev,
- width = number_length
- )?;
- }
+ highlight_source_snippet(self.0, &self.1, self.2, self.3.clone(), f)
+ }
+}
+
+pub fn highlight_source_snippet(
+ source: &str,
+ name: impl Display,
+ highlight: HighlightConfig,
+ additional: Option<String>,
+ f: &mut Formatter<'_>,
+) -> std::fmt::Result {
+ let HighlightConfig {
+ line,
+ column,
+ highlight_len,
+ } = highlight;
+
+ let previous_line = if line > 1 {
+ source.lines().nth(line - 2)
+ } else {
+ None
+ };
+ let source_line = source.lines().nth(line - 1).expect("Missing line");
+ let next_line = source.lines().nth(line);
+
+ let largest_line_num = line.max(line.saturating_sub(1)).max(line.saturating_add(1));
+ let number_length = format!("{}", largest_line_num).len();
+
+ writeln!(f, "| Script error on line {} at \"{}\"\n|", line, name)?;
+ if let Some(prev) = previous_line {
writeln!(
f,
"| [{:>width$}] {}",
- line,
- source_line,
+ line - 1,
+ prev,
width = number_length
)?;
+ }
+ writeln!(
+ f,
+ "| [{:>width$}] {}",
+ line,
+ source_line,
+ width = number_length
+ )?;
+ writeln!(
+ f,
+ "| {} {}{}",
+ vec![" "; number_length + 2].join(""),
+ vec![" "; column - 1].join(""),
+ vec!["^"; highlight_len].join(""),
+ )?;
+ if let Some(next) = next_line {
writeln!(
f,
- "| {} {}{}",
- vec![" "; number_length + 2].join(""),
- vec![" "; column - 1].join(""),
- vec!["^"; token.token_type.len()].join(""),
+ "| [{:>width$}] {}",
+ line + 1,
+ next,
+ width = number_length
)?;
- if let Some(next) = next_line {
- writeln!(
- f,
- "| [{:>width$}] {}",
- line + 1,
- next,
- width = number_length
- )?;
- }
- writeln!(f, "|\n| Failed To Parse: expected {}", expected)
- }
-}
-
-pub fn print_forge_error<'a>(source: &'a str, fe: &'a ForgeError) {
- match &fe.kind {
- ForgeErrorKind::IncompleteInput => eprintln!("| Unexpected end of file"),
- ForgeErrorKind::LexerError(err) => eprintln!("| {}", err),
- ForgeErrorKind::UnexpectedToken { found, expected } => {
- print_unexpected_token(source, found, expected)
- }
}
-}
-pub fn format_forge_error<'a>(source: &'a str, fe: &'a ForgeError) -> String {
- match &fe.kind {
- ForgeErrorKind::IncompleteInput => String::from("| Unexpected end of file"),
- ForgeErrorKind::LexerError(err) => format!("| {}", err),
- ForgeErrorKind::UnexpectedToken { found, expected } => {
- format!("{}", ErrorFormat(source, found, expected))
- }
+ if let Some(extra) = additional {
+ writeln!(f, "|\n| {}", extra)
+ } else {
+ writeln!(f, "|")
}
}
diff --git a/forge-script-lang/src/lexer/keywords.rs b/forge-script-lang/src/lexer/keywords.rs
index 892876e..62a92c9 100644
--- a/forge-script-lang/src/lexer/keywords.rs
+++ b/forge-script-lang/src/lexer/keywords.rs
@@ -66,7 +66,7 @@ pub fn token_if(input: Span) -> IResult<Span, ScriptToken> {
pub fn token_null(input: Span) -> IResult<Span, ScriptToken> {
let (input, pos) = position(input)?;
- let (input, _) = tag_ws("null", input)?;
+ let (input, _) = tag("null")(input)?;
Ok((
input,
ScriptToken {
diff --git a/forge-script-lang/src/lexer/mod.rs b/forge-script-lang/src/lexer/mod.rs
index 685a1bc..f0b0937 100644
--- a/forge-script-lang/src/lexer/mod.rs
+++ b/forge-script-lang/src/lexer/mod.rs
@@ -108,6 +108,24 @@ mod _lex {
Ok(tokens)
}
+
+ #[cfg(test)]
+ mod lex_test {
+ use super::*;
+ use test_case::test_case;
+
+ #[test_case("123" => matches Ok(ScriptTokenType::Integer(123)))]
+ #[test_case("0.123" => matches Ok(ScriptTokenType::Float(_)))]
+ #[test_case("null" => matches Ok(ScriptTokenType::Null))]
+ #[test_case("foobar" => matches Ok(ScriptTokenType::Identifier(_)))]
+ #[test_case("true" => matches Ok(ScriptTokenType::Boolean(true)))]
+ #[test_case("false" => matches Ok(ScriptTokenType::Boolean(false)))]
+ fn correct_lexing(inp: &str) -> Result<ScriptTokenType, ()> {
+ any_token(Span::new(inp))
+ .map_err(|_| ())
+ .map(|tok| tok.1.token_type)
+ }
+ }
}
use crate::parser::TokenSlice;
diff --git a/forge-script-lang/src/parser/ast.rs b/forge-script-lang/src/parser/ast.rs
index feab320..24690c2 100644
--- a/forge-script-lang/src/parser/ast.rs
+++ b/forge-script-lang/src/parser/ast.rs
@@ -6,13 +6,13 @@ use std::ops::Deref;
pub trait AstNode {}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct Program(pub ExpressionList);
impl AstNode for Program {}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct ExpressionList {
pub expressions: Vec<Expression>,
@@ -51,7 +51,7 @@ impl Deref for ExpressionList {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(
feature = "serde",
derive(serde::Serialize, serde::Deserialize),
@@ -144,7 +144,7 @@ impl Display for BinaryOp {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(
feature = "serde",
derive(serde::Serialize, serde::Deserialize),
@@ -158,14 +158,14 @@ pub enum VoidExpression {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct GroupedExpression {
pub inner: Box<ValueExpression>,
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(
feature = "serde",
derive(serde::Serialize, serde::Deserialize),
@@ -194,12 +194,12 @@ pub enum ValueExpression {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct TypeofValue(pub Box<ValueExpression>);
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct ConditionalLoop {
pub block: GuardedBlock,
@@ -222,7 +222,7 @@ impl ConditionalLoop {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct Conditional {
pub blocks: Vec<GuardedBlock>,
@@ -230,7 +230,7 @@ pub struct Conditional {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct GuardedBlock {
pub guard: Box<ValueExpression>,
@@ -238,7 +238,7 @@ pub struct GuardedBlock {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct Import {
pub source: String,
@@ -246,14 +246,14 @@ pub struct Import {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct Export {
pub items: IdentifierList,
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct Print {
pub expr: Box<ValueExpression>,
@@ -270,7 +270,7 @@ pub type IdentifierList = Vec<IdentifierNode>;
pub type ParameterList = Vec<ValueExpression>;
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct Identifier(pub String);
impl Display for Identifier {
@@ -282,7 +282,7 @@ impl Display for Identifier {
/// Alias an identifier, to create a new way of referring to it
/// IdentifierAlias(original, alias) => identifier "as" alias
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct IdentifierAlias(pub String, pub String);
impl Display for IdentifierAlias {
@@ -292,7 +292,7 @@ impl Display for IdentifierAlias {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(
feature = "serde",
derive(serde::Serialize, serde::Deserialize),
@@ -328,7 +328,7 @@ impl IdentifierNode {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(
feature = "serde",
derive(serde::Serialize, serde::Deserialize),
@@ -353,7 +353,7 @@ impl Display for LiteralNode {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct Assignment {
pub ident: Identifier,
@@ -361,7 +361,7 @@ pub struct Assignment {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(
feature = "serde",
derive(serde::Serialize, serde::Deserialize),
@@ -373,7 +373,7 @@ pub enum DeclareIdent {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct DeclareFunction {
pub ident: Identifier,
@@ -382,14 +382,25 @@ pub struct DeclareFunction {
}
#[derive(Clone)]
-#[cfg_attr(feature = "debug-ast", derive(Debug))]
+#[cfg_attr(any(feature = "debug-ast", test), derive(Debug))]
#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
pub struct FunctionCall {
pub name: Identifier,
pub params: ParameterList,
}
-macro_rules! impl_into_ast {
+macro_rules! impl_into_void_expr {
+ ($($type: ty => $variant: expr),+) => {
+ $(
+ impl From<$type> for VoidExpression {
+ fn from(other: $type) -> Self {
+ $variant(other)
+ }
+ }
+ )+
+ };
+}
+macro_rules! impl_into_value_expr {
($($type: ty => $variant: expr),+) => {
$(
impl From<$type> for ValueExpression {
@@ -401,7 +412,7 @@ macro_rules! impl_into_ast {
};
}
-impl_into_ast!(
+impl_into_value_expr!(
GroupedExpression => ValueExpression::Grouped,
ExpressionList => ValueExpression::Block,
LiteralNode => ValueExpression::Literal,
@@ -413,3 +424,10 @@ impl_into_ast!(
DeclareFunction => ValueExpression::DeclareFunction,
TypeofValue => ValueExpression::Typeof
);
+
+impl_into_void_expr!(
+ Print => VoidExpression::Print,
+ ConditionalLoop => VoidExpression::ConditionLoop,
+ Import => VoidExpression::Import,
+ Export => VoidExpression::Export
+);
diff --git a/forge-script-lang/src/parser/forge_script.lalrpop b/forge-script-lang/src/parser/forge_grammar.lalrpop
similarity index 51%
rename from forge-script-lang/src/parser/forge_script.lalrpop
rename to forge-script-lang/src/parser/forge_grammar.lalrpop
index 06cb127..1066ca2 100644
--- a/forge-script-lang/src/parser/forge_script.lalrpop
+++ b/forge-script-lang/src/parser/forge_grammar.lalrpop
@@ -4,17 +4,78 @@ use crate::runtime::numbers::Number;
grammar<'a>;
+pub Program: Program = {
+ ExpressionList => Program(<>)
+};
+
+Block: Option<ExpressionList> = {
+ "{" <ExpressionList?> "}" => <>,
+};
+
+ExpressionList: ExpressionList = {
+ <mut v:(<Expression> ";")+> <e:Expression?> => {
+ match e {
+ Some(val) => {
+ v.push(val);
+ ExpressionList {
+ expressions: v,
+ is_void: false,
+ }
+ },
+ None => {
+ ExpressionList {
+ expressions: v,
+ is_void: true,
+ }
+ },
+ }
+ }
+};
+
pub Expression: Expression = {
ValueExpression => Expression::Value(<>),
-}
+ VoidExpression => Expression::Void(<>),
+};
+
+VoidExpression: VoidExpression = {
+ "print" <expr:ValueExpression> => Print { expr: Box::new(expr) }.into(),
+ "import" "{" <items:IdentifierList> "}" "from" <source:StringValue> => Import { source, items }.into(),
+ "export" "{" <items:IdentifierList> "}" => Export { items }.into(),
+};
ValueExpression: ValueExpression = {
#[precedence(level="1")]
Literal => ValueExpression::Literal(<>),
#[precedence(level="1")]
+ "(" <expr:ValueExpression> ")" => GroupedExpression { inner: Box::new(expr) }.into(),
+
+ #[precedence(level="2")] #[assoc(side="right")]
"typeof" <expr:ValueExpression> => TypeofValue(Box::new(expr)).into(),
+ #[precedence(level="2")] #[assoc(side="right")]
+ "-" <expr:ValueExpression> => ValueExpression::Unary { operand: Box::new(expr), operator: UnaryOp::Negate },
+ #[precedence(level="2")] #[assoc(side="right")]
+ "!" <expr:ValueExpression> => ValueExpression::Unary { operand: Box::new(expr), operator: UnaryOp::Not },
+ #[precedence(level = "3")]
+ IfStatement => <>.into(),
+
+ #[precedence(level="4")] #[assoc(side="left")]
+ <lhs:ValueExpression> "*" <rhs:ValueExpression> => {
+ ValueExpression::Binary {
+ lhs: Box::new(lhs),
+ rhs: Box::new(rhs),
+ operator: BinaryOp::Multiply,
+ }
+ },
#[precedence(level="4")] #[assoc(side="left")]
+ <lhs:ValueExpression> "/" <rhs:ValueExpression> => {
+ ValueExpression::Binary {
+ lhs: Box::new(lhs),
+ rhs: Box::new(rhs),
+ operator: BinaryOp::Divide,
+ }
+ },
+ #[precedence(level="6")] #[assoc(side="left")]
<lhs:ValueExpression> "+" <rhs:ValueExpression> => {
ValueExpression::Binary {
lhs: Box::new(lhs),
@@ -22,7 +83,7 @@ ValueExpression: ValueExpression = {
operator: BinaryOp::Add,
}
},
- #[precedence(level="4")] #[assoc(side="left")]
+ #[precedence(level="6")] #[assoc(side="left")]
<lhs:ValueExpression> "-" <rhs:ValueExpression> => {
ValueExpression::Binary {
lhs: Box::new(lhs),
@@ -30,7 +91,49 @@ ValueExpression: ValueExpression = {
operator: BinaryOp::Subtract,
}
},
-}
+
+ #[precedence(level="1")]
+ Identifier => <>.into()
+};
+
+IfStatement: Conditional = {
+ BareIfStatement => Conditional { blocks: vec![<>], fallback: None },
+ <fi: BareIfStatement> "else" <bl:Block> => Conditional { blocks: vec![fi], fallback: bl },
+ <fi:BareIfStatement> "else" <ls:IfStatement> => {
+ let mut ls = ls;
+ let mut new = vec![fi];
+ new.append(&mut ls.blocks);
+ Conditional {
+ blocks: new,
+ fallback: ls.fallback,
+ }
+ }
+};
+
+BareIfStatement: GuardedBlock = {
+ "if" <guard:ValueExpression> <bl:Block> => {
+ GuardedBlock {
+ block: bl.unwrap_or_default(),
+ guard: Box::new(guard),
+ }
+ }
+};
+
+ParameterList = ListOf<ValueExpression>;
+IdentifierList = ListOf<IdentifierNode>;
+
+IdentifierNode: IdentifierNode = {
+ IdentifierAlias => IdentifierNode::Alias(<>),
+ Identifier => IdentifierNode::Direct(<>),
+};
+
+IdentifierAlias: IdentifierAlias = {
+ <base:Identifier> "as" <alias:Identifier> => IdentifierAlias(base.0, alias.0),
+};
+
+Identifier: Identifier = {
+ "identifier" => Identifier(String::from(<>))
+};
Literal: LiteralNode = {
"boolean" => LiteralNode::Boolean(<>),
@@ -39,12 +142,22 @@ Literal: LiteralNode = {
"integer" => LiteralNode::Number(Number::Integer(<>)),
"string" => LiteralNode::String(String::from(<>)),
"owned_string" => LiteralNode::String(<>),
-}
+};
StringValue: String = {
"string" => String::from(<>),
"owned_string" => <>,
-}
+};
+
+ListOf<T>: Vec<T> = {
+ <mut v:(<T> ",")+> <e:T?> => match e {
+ None => v,
+ Some(e) => {
+ v.push(e);
+ v
+ }
+ }
+};
extern {
type Location = usize;
diff --git a/forge-script-lang/src/parser/forge_script.rs b/forge-script-lang/src/parser/forge_script.rs
new file mode 100644
index 0000000..dc9abcc
--- /dev/null
+++ b/forge-script-lang/src/parser/forge_script.rs
@@ -0,0 +1,71 @@
+// use lalrpop_util::lalrpop_mod;
+// lalrpop_mod!(pub forge_script);
+
+use crate::error::ForgeResult;
+use crate::lexer::{script_to_tokens, ScriptTokenType};
+use crate::parser::ast::{Expression, Program};
+use crate::TokenError;
+use peg::Parse;
+
+pub type InputSpan<'a, Loc, Tok> = Result<(Loc, Tok, Loc), TokenError<'a>>;
+type ExprSpan<'a> = InputSpan<'a, usize, ScriptTokenType<'a>>;
+
+macro_rules! export_grammar_fn {
+ ($name:ident = $output:ty => $part: tt) => {
+ pub fn $name(source: &str) -> ForgeResult<$output> {
+ let tokens = script_to_tokens(source)?
+ .iter()
+ .map(|tok| {
+ Ok((
+ tok.position.start(),
+ tok.token_type.clone(),
+ tok.position.start() + tok.position.len(),
+ ))
+ })
+ .collect::<Vec<ExprSpan>>();
+
+ let value =
+ super::forge_grammar::$part::new().parse::<ExprSpan, Vec<ExprSpan>>(tokens)?;
+
+ Ok(value)
+ }
+ };
+}
+
+export_grammar_fn!(parse_program = Program => ProgramParser);
+export_grammar_fn!(parse_expression = Expression => ExpressionParser);
+
+// pub fn parse_expression(source: &str) -> ForgeResult<Expression> {
+// let tokens = script_to_tokens(source)?
+// .iter()
+// .map(|tok| {
+// Ok((
+// tok.position.start(),
+// tok.token_type.clone(),
+// tok.position.start() + tok.position.len(),
+// ))
+// })
+// .collect::<Vec<ExprSpan>>();
+//
+// let value =
+// super::forge_grammar::ExpressionParser::new().parse::<ExprSpan, Vec<ExprSpan>>(tokens)?;
+//
+// Ok(value)
+// }
+
+#[cfg(test)]
+mod grammar_test {
+ use super::parse_expression;
+ use crate::error::ForgeResult;
+ use crate::parse::ast::Expression;
+ use test_case::test_case;
+
+ #[test_case("123" => matches Ok(_) ; "Parse literal number")]
+ #[test_case(r#""Basic String""# => matches Ok(_) ; "Parse literal string")]
+ #[test_case("false" => matches Ok(_) ; "Parse literal false")]
+ #[test_case("true" => matches Ok(_) ; "Parse literal true")]
+ #[test_case("null" => matches Ok(_) ; "Parse literal null")]
+ fn expression_parsing<'a>(prog: &'a str) -> ForgeResult<'a, Expression> {
+ parse_expression(prog)
+ }
+}
diff --git a/forge-script-lang/src/parser/lr_grammar.rs b/forge-script-lang/src/parser/lr_grammar.rs
deleted file mode 100644
index c7d0e39..0000000
--- a/forge-script-lang/src/parser/lr_grammar.rs
+++ /dev/null
@@ -1,45 +0,0 @@
-// use lalrpop_util::lalrpop_mod;
-// lalrpop_mod!(pub forge_script);
-
-use crate::lexer::{ScriptToken, ScriptTokenType};
-use crate::TokenError;
-use peg::Parse;
-use std::slice::Iter;
-
-pub struct TokenInput<'a>(std::slice::Iter<'a, ScriptToken<'a>>);
-
-impl<'a> From<Iter<'a, ScriptToken<'a>>> for TokenInput<'a> {
- fn from(value: Iter<'a, ScriptToken<'a>>) -> Self {
- TokenInput(value)
- }
-}
-
-pub type InputSpan<'a, Loc, Tok> = Result<(Loc, Tok, Loc), TokenError<'a>>;
-
-impl<'a> Iterator for TokenInput<'a> {
- type Item = InputSpan<'a, usize, ScriptTokenType<'a>>;
-
- fn next(&mut self) -> Option<Self::Item> {
- self.0.next().map(|tok| {
- Ok((
- tok.position.start(),
- tok.token_type.clone(),
- tok.position.start() + tok.position.len(),
- ))
- })
- }
-}
-
-#[cfg(test)]
-mod grammar_test {
- use crate::lexer::{script_to_tokens, ScriptTokenType};
- use crate::parser::lr_grammar::{InputSpan, TokenInput};
-
- #[test]
- fn basic_parsing() {
- let code = script_to_tokens("4 + 4").expect("Failed to parse");
- crate::parser::forge_script::ExpressionParser::new()
- .parse::<InputSpan<usize, ScriptTokenType>, TokenInput>(code.iter().into())
- .expect("Failed to boop");
- }
-}
diff --git a/forge-script-lang/src/parser/mod.rs b/forge-script-lang/src/parser/mod.rs
index 2745c9f..03f85bf 100644
--- a/forge-script-lang/src/parser/mod.rs
+++ b/forge-script-lang/src/parser/mod.rs
@@ -1,11 +1,11 @@
pub mod ast;
mod atoms;
+mod forge_script;
mod grammar;
-mod lr_grammar;
#[cfg(test)]
mod test_suite;
-mod forge_script;
+pub(crate) mod forge_grammar;
use crate::error::{ForgeError, ForgeErrorKind, ForgeResult};
use crate::print_forge_error;
@@ -15,44 +15,47 @@ pub fn slice<'a>(toks: &'a [crate::lexer::ScriptToken]) -> TokenSlice<'a> {
TokenSlice(toks)
}
-pub fn parse_expression(expr: &str) -> ForgeResult<ast::Expression> {
- let tokens = crate::lexer::script_to_tokens(expr)?;
- let result = match grammar::expression(&TokenSlice(tokens.as_slice())) {
- Ok(expr) => Ok(expr),
- Err(parse_error) => {
- let bad_token = &tokens[parse_error.location];
- Err(ForgeError {
- kind: ForgeErrorKind::UnexpectedToken {
- found: bad_token.clone(),
- expected: parse_error.expected,
- },
- })
- }
- };
+pub use forge_script::parse_expression;
+pub use forge_script::parse_program;
- result.map_err(|e| {
- print_forge_error(expr, &e);
- e
- })
-}
+// pub fn parse_expression(expr: &str) -> ForgeResult<ast::Expression> {
+// let tokens = crate::lexer::script_to_tokens(expr)?;
+// let result = match grammar::expression(&TokenSlice(tokens.as_slice())) {
+// Ok(expr) => Ok(expr),
+// Err(parse_error) => {
+// let bad_token = &tokens[parse_error.location];
+// Err(ForgeError {
+// kind: ForgeErrorKind::UnexpectedToken {
+// found: bad_token.clone(),
+// expected: parse_error.expected,
+// },
+// })
+// }
+// };
+//
+// result.map_err(|e| {
+// print_forge_error(expr, &e);
+// e
+// })
+// }
-pub fn parse_program(prog: &str) -> ForgeResult<ast::Program> {
- let tokens = crate::lexer::script_to_tokens(prog)?;
- let result = match grammar::program(&TokenSlice(tokens.as_slice())) {
- Ok(prog) => Ok(prog.clone()),
- Err(parse_error) => {
- let bad_token = &tokens[parse_error.location];
- Err(ForgeError {
- kind: ForgeErrorKind::UnexpectedToken {
- found: bad_token.clone(),
- expected: parse_error.expected,
- },
- })
- }
- };
-
- result.map_err(|e| {
- print_forge_error(prog, &e);
- e
- })
-}
+// pub fn parse_program(prog: &str) -> ForgeResult<ast::Program> {
+// let tokens = crate::lexer::script_to_tokens(prog)?;
+// let result = match grammar::program(&TokenSlice(tokens.as_slice())) {
+// Ok(prog) => Ok(prog.clone()),
+// Err(parse_error) => {
+// let bad_token = &tokens[parse_error.location];
+// Err(ForgeError {
+// kind: ForgeErrorKind::UnexpectedToken {
+// found: bad_token.clone(),
+// expected: parse_error.expected,
+// },
+// })
+// }
+// };
+//
+// result.map_err(|e| {
+// print_forge_error(prog, &e);
+// e
+// })
+// }
diff --git a/forge-script-lang/src/utilities.rs b/forge-script-lang/src/utilities.rs
index 6c5c1c2..0233d1a 100644
--- a/forge-script-lang/src/utilities.rs
+++ b/forge-script-lang/src/utilities.rs
@@ -42,3 +42,23 @@ macro_rules! deref_as {
pub trait Clown<Target> {
fn clown(&self) -> Target;
}
+
+type LineNum = usize;
+type ColumnNum = usize;
+
+pub fn offset_to_line_column(source: &str, offset: usize) -> (LineNum, ColumnNum) {
+ let mut remaining = offset;
+ let mut current_line = 0;
+ for line in source.lines() {
+ current_line += 1;
+ if remaining < line.len() {
+ return (current_line, remaining + 1);
+ }
+ remaining -= line.len();
+ }
+
+ (
+ current_line,
+ source.lines().last().map(|l| l.len()).unwrap_or(0),
+ )
+}
--
GitLab