refactor: split cargo workspace

crates/toml-parser/src/lib.rs

@@ -0,0 +1,11 @@
+mod spanned_ext;
+#[cfg(test)]
+mod tests;
+mod toml_lexer;
+pub mod toml_parser;
+pub mod toml_span;
+pub mod toml_value;
+mod value_ext;
+pub mod value_parser;
+
+pub use spanned_ext::SpannedErrorExt;

crates/toml-parser/src/spanned_ext.rs

@@ -0,0 +1,64 @@
+use crate::{
+    toml_span::Spanned,
+    toml_value::Value,
+    value_parser::{ParseResult, Parser},
+};
+
+impl Spanned<&Value> {
+    pub fn parse<P: Parser>(&self, parser: &mut P) -> ParseResult<P> {
+        self.value.parse(self.span, parser)
+    }
+
+    pub fn parse_map<P: Parser, E>(
+        &self,
+        parser: &mut P,
+    ) -> Result<<P as Parser>::Value, Spanned<E>>
+    where
+        <P as Parser>::Error: Into<E>,
+    {
+        self.parse(parser).map_spanned_err(|e| e.into())
+    }
+}
+
+impl Spanned<Value> {
+    pub fn parse<P: Parser>(&self, parser: &mut P) -> ParseResult<P> {
+        self.as_ref().parse(parser)
+    }
+
+    pub fn parse_map<P: Parser, E>(
+        &self,
+        parser: &mut P,
+    ) -> Result<<P as Parser>::Value, Spanned<E>>
+    where
+        <P as Parser>::Error: Into<E>,
+    {
+        self.as_ref().parse_map(parser)
+    }
+}
+
+pub trait SpannedErrorExt {
+    type T;
+    type E;
+
+    fn map_spanned_err<U, F>(self, f: F) -> Result<Self::T, Spanned<U>>
+    where
+        F: FnOnce(Self::E) -> U;
+}
+
+impl<T, E> SpannedErrorExt for Result<T, Spanned<E>> {
+    type T = T;
+    type E = E;
+
+    fn map_spanned_err<U, F>(self, f: F) -> Result<Self::T, Spanned<U>>
+    where
+        F: FnOnce(Self::E) -> U,
+    {
+        match self {
+            Ok(v) => Ok(v),
+            Err(e) => Err(Spanned {
+                span: e.span,
+                value: f(e.value),
+            }),
+        }
+    }
+}

crates/toml-parser/src/tests.rs

@@ -0,0 +1,136 @@
+use {
+    crate::{
+        toml_parser::{ErrorHandler, ParserError, parse},
+        toml_span::{Span, Spanned, SpannedExt},
+        toml_value::Value,
+    },
+    bstr::{BStr, ByteSlice},
+    std::{
+        os::unix::ffi::OsStrExt,
+        panic::{AssertUnwindSafe, catch_unwind},
+        str::FromStr,
+    },
+    walkdir::WalkDir,
+};
+
+#[test]
+fn test() {
+    let mut have_failures = false;
+    let mut num = 0;
+    let tests = std::path::Path::new(concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../toml-config/toml-test/tests/valid"
+    ));
+    if !tests.exists() {
+        eprintln!("skipping TOML conformance tests because fixtures are not present");
+        return;
+    }
+    for path in WalkDir::new(tests) {
+        let path = path.unwrap();
+        if let Some(prefix) = path.path().as_os_str().as_bytes().strip_suffix(b".toml") {
+            num += 1;
+            let res = catch_unwind(AssertUnwindSafe(|| {
+                have_failures |= run_test(prefix.as_bstr());
+            }));
+            if res.is_err() {
+                eprintln!("panic while running {}", prefix.as_bstr());
+            }
+        }
+    }
+    if have_failures {
+        panic!("There were test failures");
+    }
+    eprintln!("ran {num} tests");
+}
+
+fn run_test(prefix: &BStr) -> bool {
+    let toml = std::fs::read(&format!("{}.toml", prefix)).unwrap();
+    let json = std::fs::read_to_string(&format!("{}.json", prefix)).unwrap();
+
+    let json: serde_json::Value = serde_json::from_str(&json).unwrap();
+    let json_as_toml = json_to_value(json);
+    let toml = match parse(toml.as_bytes(), &NoErrorHandler(prefix)) {
+        Ok(t) => t,
+        Err(e) => {
+            eprintln!("toml could not be parsed in test {}", prefix);
+            NoErrorHandler(prefix).handle(e);
+            return true;
+        }
+    };
+
+    if toml != json_as_toml {
+        eprintln!("toml and json differ in test {}", prefix);
+        eprintln!("toml: {:#?}", toml);
+        eprintln!("json: {:#?}", json_as_toml);
+        true
+    } else {
+        false
+    }
+}
+
+fn json_to_value(json: serde_json::Value) -> Spanned<Value> {
+    let span = Span { lo: 0, hi: 0 };
+    let val = match json {
+        serde_json::Value::String(_)
+        | serde_json::Value::Number(_)
+        | serde_json::Value::Null
+        | serde_json::Value::Bool(_) => panic!("Unexpected type"),
+        serde_json::Value::Array(v) => Value::Array(v.into_iter().map(json_to_value).collect()),
+        serde_json::Value::Object(v) => {
+            if v.len() == 2 && v.contains_key("type") && v.contains_key("value") {
+                let ty = v.get("type").unwrap().as_str().unwrap();
+                let val = v.get("value").unwrap().as_str().unwrap();
+                match ty {
+                    "string" => Value::String(val.to_owned()),
+                    "integer" => Value::Integer(i64::from_str(val).unwrap()),
+                    "float" => Value::Float(f64::from_str(val).unwrap()),
+                    "bool" => Value::Boolean(bool::from_str(val).unwrap()),
+                    _ => panic!("unexpected type {}", ty),
+                }
+            } else {
+                Value::Table(
+                    v.into_iter()
+                        .map(|(k, v)| (k.spanned(span), json_to_value(v)))
+                        .collect(),
+                )
+            }
+        }
+    };
+    val.spanned(span)
+}
+
+struct NoErrorHandler<'a>(&'a BStr);
+
+impl<'a> ErrorHandler for NoErrorHandler<'a> {
+    fn handle(&self, err: Spanned<ParserError>) {
+        eprintln!(
+            "{}: An error occurred during validation: {}",
+            self.0,
+            FormatError(err.span, Some(err.value)),
+        );
+    }
+
+    fn redefinition(&self, err: Spanned<ParserError>, prev: Span) {
+        eprintln!(
+            "{}: Redefinition: {}",
+            self.0,
+            FormatError(err.span, Some(err.value)),
+        );
+        eprintln!(
+            "{}: Previous: {}",
+            self.0,
+            FormatError(prev, None),
+        );
+    }
+}
+
+struct FormatError(Span, Option<ParserError>);
+
+impl std::fmt::Display for FormatError {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        if let Some(cause) = &self.1 {
+            write!(f, "{cause}: ")?;
+        }
+        write!(f, "span {}..{}", self.0.lo, self.0.hi)
+    }
+}

crates/toml-parser/src/toml_lexer.rs

@@ -0,0 +1,191 @@
+use crate::toml_span::{Span, Spanned, SpannedExt};
+
+#[derive(Copy, Clone, Debug, Eq, PartialEq)]
+pub enum Token<'a> {
+    Dot,
+    Equals,
+    Comma,
+    LeftBracket,
+    RightBracket,
+    LeftBrace,
+    RightBrace,
+    LiteralString(&'a [u8]),
+    CookedString(&'a [u8]),
+    Literal(&'a [u8]),
+}
+
+impl Token<'_> {
+    pub fn name(self, value_context: bool) -> &'static str {
+        match self {
+            Token::Dot => "`.`",
+            Token::Equals => "`=`",
+            Token::Comma => "`,`",
+            Token::LeftBracket => "`[`",
+            Token::RightBracket => "`]`",
+            Token::LeftBrace => "`{`",
+            Token::RightBrace => "`}`",
+            Token::LiteralString(_) | Token::CookedString(_) => "a string",
+            Token::Literal(_) if value_context => "a literal",
+            Token::Literal(_) => "a key",
+        }
+    }
+}
+
+pub struct Lexer<'a> {
+    input: &'a [u8],
+    pos: usize,
+    peek: Option<Spanned<Token<'a>>>,
+    peek_value_context: bool,
+}
+
+impl<'a> Lexer<'a> {
+    pub fn new(input: &'a [u8]) -> Self {
+        Self {
+            input,
+            pos: 0,
+            peek: None,
+            peek_value_context: false,
+        }
+    }
+
+    pub fn pos(&mut self) -> usize {
+        self.skip_ws();
+        self.pos
+    }
+
+    fn skip_ws(&mut self) {
+        while let Some(char) = self.input.get(self.pos).copied() {
+            match char {
+                b' ' | b'\t' | b'\n' => self.pos += 1,
+                b'#' => {
+                    self.pos += 1;
+                    while let Some(char) = self.input.get(self.pos).copied() {
+                        self.pos += 1;
+                        if char == b'\n' {
+                            break;
+                        }
+                    }
+                }
+                _ => break,
+            }
+        }
+    }
+
+    pub fn peek(&mut self, value_context: bool) -> Option<Spanned<Token<'a>>> {
+        let next = self.next(value_context);
+        self.peek = next;
+        self.peek_value_context = value_context;
+        next
+    }
+
+    pub fn next(&mut self, value_context: bool) -> Option<Spanned<Token<'a>>> {
+        if let Some(peek) = self.peek.take() {
+            if self.peek_value_context == value_context {
+                return Some(peek);
+            }
+            self.pos = peek.span.lo;
+        }
+
+        use Token::*;
+
+        macro_rules! get {
+            ($off:expr) => {
+                self.input.get(self.pos + $off).copied()
+            };
+        }
+
+        self.skip_ws();
+
+        let c = get!(0)?;
+        let pos = self.pos;
+
+        macro_rules! span {
+            () => {
+                Span {
+                    lo: pos,
+                    hi: self.pos,
+                }
+            };
+        }
+
+        'simple: {
+            let t = match c {
+                b'.' => Dot,
+                b',' => Comma,
+                b'=' => Equals,
+                b'[' => LeftBracket,
+                b']' => RightBracket,
+                b'{' => LeftBrace,
+                b'}' => RightBrace,
+                _ => break 'simple,
+            };
+            self.pos += 1;
+            return Some(t.spanned(span!()));
+        }
+
+        macro_rules! try_string {
+            ($delim:expr, $escaping:expr, $ident:ident) => {
+                if c == $delim {
+                    'ml_string: {
+                        let delim = ($delim, Some($delim), Some($delim));
+                        if (c, get!(1), get!(2)) != delim {
+                            break 'ml_string;
+                        }
+                        self.pos += 3;
+                        if get!(0) == Some(b'\n') {
+                            self.pos += 1;
+                        }
+                        let start = self.pos;
+                        let end = loop {
+                            let c = match get!(0) {
+                                Some(c) => c,
+                                _ => break self.pos,
+                            };
+                            self.pos += 1;
+                            if $escaping && c == b'\\' {
+                                self.pos += 1;
+                            } else if c == $delim {
+                                if (c, get!(0), get!(1)) == delim && get!(2) != Some($delim) {
+                                    self.pos += 2;
+                                    break self.pos - 3;
+                                }
+                            }
+                        };
+                        return Some($ident(&self.input[start..end]).spanned(span!()));
+                    }
+                    self.pos += 1;
+                    let start = self.pos;
+                    let end = loop {
+                        let c = match get!(0) {
+                            Some(c) => c,
+                            _ => break self.pos,
+                        };
+                        self.pos += 1;
+                        if $escaping && c == b'\\' {
+                            self.pos += 1;
+                        } else if c == $delim {
+                            break self.pos - 1;
+                        }
+                    };
+                    return Some($ident(&self.input[start..end]).spanned(span!()));
+                }
+            };
+        }
+
+        try_string!(b'\'', false, LiteralString);
+        try_string!(b'"', true, CookedString);
+
+        let start = self.pos;
+        while let Some(c) = get!(0) {
+            match c {
+                b' ' | b'\t' | b'\n' | b'#' | b',' | b'=' | b'{' | b'}' | b'[' | b']' => break,
+                b'.' if !value_context => break,
+                _ => {}
+            }
+            self.pos += 1;
+        }
+        let end = self.pos;
+
+        Some(Literal(&self.input[start..end]).spanned(span!()))
+    }
+}

crates/toml-parser/src/toml_parser.rs

@@ -0,0 +1,530 @@
+use {
+    crate::{
+        toml_lexer::{Lexer, Token},
+        toml_span::{Span, Spanned, SpannedExt},
+        toml_value::Value,
+    },
+    bstr::ByteSlice,
+    indexmap::{
+        IndexMap,
+        map::{RawEntryApiV1, raw_entry_v1::RawEntryMut},
+    },
+    std::{collections::VecDeque, mem, str::FromStr},
+    thiserror::Error,
+};
+
+pub trait ErrorHandler {
+    fn handle(&self, err: Spanned<ParserError>);
+    fn redefinition(&self, err: Spanned<ParserError>, prev: Span);
+}
+
+#[derive(Debug, Error)]
+pub enum ParserError {
+    #[error("Unexpected end of file")]
+    UnexpectedEof,
+    #[error("Expected a key")]
+    MissingKey,
+    #[error("Expected {0} but found {1}")]
+    Expected(&'static str, &'static str),
+    #[error("Duplicate key overwrites the previous definition")]
+    Redefined,
+    #[error("Literal is not valid UTF-8")]
+    NonUtf8Literal,
+    #[error("Could not parse the literal")]
+    UnknownLiteral,
+    #[error("Ignoring key due to following error")]
+    IgnoringKey,
+    #[error("Unnecessary comma")]
+    UnnecessaryComma,
+}
+
+pub fn parse(
+    input: &[u8],
+    error_handler: &dyn ErrorHandler,
+) -> Result<Spanned<Value>, Spanned<ParserError>> {
+    let parser = Parser {
+        lexer: Lexer::new(input),
+        error_handler,
+        last_span: None,
+    };
+    parser.parse()
+}
+
+struct Parser<'a, 'b> {
+    lexer: Lexer<'a>,
+    error_handler: &'b dyn ErrorHandler,
+    last_span: Option<Span>,
+}
+
+type Key = VecDeque<Spanned<String>>;
+
+impl<'a> Parser<'a, '_> {
+    fn parse(mut self) -> Result<Spanned<Value>, Spanned<ParserError>> {
+        self.parse_document()
+    }
+
+    fn unexpected_eof(&self) -> Spanned<ParserError> {
+        let span = self.last_span.unwrap_or(Span { lo: 0, hi: 0 });
+        ParserError::UnexpectedEof.spanned(span)
+    }
+
+    fn next(&mut self, value_context: bool) -> Result<Spanned<Token<'a>>, Spanned<ParserError>> {
+        match self.lexer.next(value_context) {
+            Some(t) => {
+                self.last_span = Some(t.span);
+                Ok(t)
+            }
+            _ => Err(self.unexpected_eof()),
+        }
+    }
+
+    fn peek(&mut self, value_context: bool) -> Result<Spanned<Token<'a>>, Spanned<ParserError>> {
+        match self.lexer.peek(value_context) {
+            Some(t) => Ok(t),
+            _ => Err(self.unexpected_eof()),
+        }
+    }
+
+    fn parse_value(&mut self) -> Result<Spanned<Value>, Spanned<ParserError>> {
+        let token = self.peek(true)?;
+        match token.value {
+            Token::LiteralString(s) => self.parse_literal_string(s),
+            Token::CookedString(s) => self.parse_cooked_string(s),
+            Token::LeftBracket => self.parse_array(),
+            Token::Literal(l) => self.parse_literal_value(l),
+            Token::LeftBrace => self.parse_inline_table(),
+            Token::Dot | Token::Equals | Token::Comma | Token::RightBrace | Token::RightBracket => {
+                Err(ParserError::Expected("a value", token.value.name(true)).spanned(token.span))
+            }
+        }
+    }
+
+    fn parse_literal_value(
+        &mut self,
+        literal: &[u8],
+    ) -> Result<Spanned<Value>, Spanned<ParserError>> {
+        let span = self.next(true)?.span;
+        let Ok(s) = std::str::from_utf8(literal) else {
+            return Err(ParserError::NonUtf8Literal.spanned(span));
+        };
+        if s == "true" {
+            return Ok(Value::Boolean(true).spanned(span));
+        }
+        if s == "false" {
+            return Ok(Value::Boolean(false).spanned(span));
+        }
+        let s = s.replace('_', "");
+        if let Ok(n) = i64::from_str(&s) {
+            return Ok(Value::Integer(n).spanned(span));
+        }
+        'radix: {
+            let b = s.as_bytes();
+            if b.len() >= 2 && b[0] == b'0' {
+                let radix = match b[1] {
+                    b'x' => 16,
+                    b'o' => 8,
+                    b'b' => 2,
+                    _ => break 'radix,
+                };
+                if let Ok(n) = i64::from_str_radix(&s[2..], radix) {
+                    return Ok(Value::Integer(n).spanned(span));
+                }
+            }
+        }
+        if let Ok(n) = f64::from_str(&s) {
+            return Ok(Value::Float(n).spanned(span));
+        }
+        Err(ParserError::UnknownLiteral.spanned(span))
+    }
+
+    fn parse_literal_string(&mut self, s: &[u8]) -> Result<Spanned<Value>, Spanned<ParserError>> {
+        let span = self.next(true)?.span;
+        let s = s.as_bstr().to_string();
+        Ok(Value::String(s).spanned(span))
+    }
+
+    fn parse_cooked_string(&mut self, s: &[u8]) -> Result<Spanned<Value>, Spanned<ParserError>> {
+        let span = self.next(true)?.span;
+        let s = self.cook_string(s);
+        Ok(Value::String(s).spanned(span))
+    }
+
+    fn cook_string(&self, s: &[u8]) -> String {
+        use std::io::Write;
+
+        if !s.contains(&b'\\') {
+            return s.as_bstr().to_string();
+        }
+        let mut res = vec![];
+        let mut pos = 0;
+        while pos < s.len() {
+            let c = s[pos];
+            pos += 1;
+            match c {
+                b'\\' => {
+                    let c = s[pos];
+                    pos += 1;
+                    match c {
+                        b'\\' => res.push(b'\\'),
+                        b'"' => res.push(b'"'),
+                        b'b' => res.push(0x8),
+                        b'f' => res.push(0xc),
+                        b'n' => res.push(b'\n'),
+                        b'r' => res.push(b'\r'),
+                        b't' => res.push(b'\t'),
+                        b'e' => res.push(0x1b),
+                        b'x' | b'u' | b'U' => 'unicode: {
+                            let len = match c {
+                                b'x' => 2,
+                                b'u' => 4,
+                                _ => 8,
+                            };
+                            if s.len() - pos >= len
+                                && let Ok(s) = std::str::from_utf8(&s[pos..pos + len])
+                                && let Ok(n) = u32::from_str_radix(s, 16)
+                                && let Some(c) = char::from_u32(n)
+                            {
+                                pos += len;
+                                let _ = write!(res, "{c}");
+                                break 'unicode;
+                            }
+                            res.extend_from_slice(&s[pos - 2..]);
+                        }
+                        b' ' | b'\t' | b'\n' => {
+                            let mut t = pos;
+                            let mut saw_nl = c == b'\n';
+                            while t < s.len() && matches!(s[t], b' ' | b'\t' | b'\n') {
+                                saw_nl |= s[t] == b'\n';
+                                t += 1;
+                            }
+                            if saw_nl {
+                                pos = t;
+                            } else {
+                                res.extend_from_slice(&[b'\\', c]);
+                            }
+                        }
+                        _ => {
+                            res.extend_from_slice(&[b'\\', c]);
+                        }
+                    }
+                }
+                _ => res.push(c),
+            }
+        }
+        res.as_bstr().to_string()
+    }
+
+    fn parse_array(&mut self) -> Result<Spanned<Value>, Spanned<ParserError>> {
+        let lo = self.next(true)?.span.lo;
+        let mut entries = vec![];
+        let mut consumed_comma = false;
+        loop {
+            if let Some(v) = self.lexer.peek(true) {
+                if v.value == Token::RightBracket {
+                    let _ = self.next(true);
+                    let hi = v.span.hi;
+                    let span = Span { lo, hi };
+                    return Ok(Value::Array(entries).spanned(span));
+                }
+                if entries.len() > 0 && !mem::take(&mut consumed_comma) {
+                    self.error_handler.handle(
+                        ParserError::Expected("`,` or `]`", v.value.name(true)).spanned(v.span),
+                    );
+                }
+            }
+            match self.parse_value() {
+                Ok(v) => {
+                    entries.push(v);
+                    consumed_comma = self.skip_comma(true);
+                }
+                Err(e) => {
+                    self.skip_tree(Token::LeftBracket, Token::RightBracket);
+                    return Err(e);
+                }
+            }
+        }
+    }
+
+    fn parse_inline_table(&mut self) -> Result<Spanned<Value>, Spanned<ParserError>> {
+        let lo = self.next(true)?.span.lo;
+        let mut map = IndexMap::new();
+        let mut consumed_comma = false;
+        loop {
+            let token = match self.peek(false) {
+                Ok(t) => t,
+                Err(e) => {
+                    self.error_handler.handle(e);
+                    break;
+                }
+            };
+            if token.value == Token::RightBrace {
+                let _ = self.next(false);
+                break;
+            }
+            if !map.is_empty() && !mem::take(&mut consumed_comma) {
+                self.error_handler.handle(
+                    ParserError::Expected("`,` or `}`", token.value.name(false))
+                        .spanned(token.span),
+                );
+            }
+            let res = match self.parse_key_value_with_recovery() {
+                Ok(res) => res,
+                Err(e) => {
+                    self.skip_tree(Token::LeftBrace, Token::RightBrace);
+                    return Err(e);
+                }
+            };
+            if let Some((mut key, value)) = res {
+                self.insert(&mut map, &mut key, value, false, false);
+            };
+            consumed_comma = self.skip_comma(false);
+        }
+        let hi = self.last_span().hi;
+        let span = Span { lo, hi };
+        Ok(Value::Table(map).spanned(span))
+    }
+
+    fn skip_comma(&mut self, value_context: bool) -> bool {
+        if let Some(token) = self.lexer.peek(value_context) {
+            if token.value != Token::Comma {
+                return false;
+            }
+            let _ = self.next(value_context);
+        }
+        while let Some(token) = self.lexer.peek(value_context) {
+            if token.value != Token::Comma {
+                break;
+            }
+            let _ = self.next(value_context);
+            self.error_handler
+                .handle(ParserError::UnnecessaryComma.spanned(token.span));
+        }
+        true
+    }
+
+    fn parse_document(&mut self) -> Result<Spanned<Value>, Spanned<ParserError>> {
+        let mut map = IndexMap::new();
+        self.parse_table_body(&mut map)?;
+        while self.lexer.peek(false).is_some() {
+            let (mut key, append) = self.parse_table_header()?;
+            let mut inner_map = IndexMap::new();
+            self.parse_table_body(&mut inner_map)?;
+            let value = Value::Table(inner_map).spanned(key.span);
+            self.insert(&mut map, &mut key.value, value, true, append);
+        }
+        let hi = self.last_span().hi;
+        let span = Span { lo: 0, hi };
+        Ok(Value::Table(map).spanned(span))
+    }
+
+    fn parse_table_header(&mut self) -> Result<(Spanned<Key>, bool), Spanned<ParserError>> {
+        let lo = self.next(false)?.span.lo;
+        let mut append = false;
+        if let Some(token) = self.lexer.peek(false)
+            && token.value == Token::LeftBracket
+        {
+            let _ = self.next(false);
+            append = true;
+        }
+        let key = self.parse_key()?;
+        let mut hi = self.parse_exact(Token::RightBracket, false)?.hi;
+        if append {
+            hi = self.parse_exact(Token::RightBracket, false)?.hi;
+        }
+        let span = Span { lo, hi };
+        Ok((key.spanned(span), append))
+    }
+
+    fn parse_table_body(
+        &mut self,
+        dst: &mut IndexMap<Spanned<String>, Spanned<Value>>,
+    ) -> Result<(), Spanned<ParserError>> {
+        while let Some(e) = self.lexer.peek(false) {
+            if e.value == Token::LeftBracket {
+                return Ok(());
+            }
+            let Some((mut key, value)) = self.parse_key_value_with_recovery()? else {
+                continue;
+            };
+            self.insert(dst, &mut key, value, false, false);
+        }
+        Ok(())
+    }
+
+    fn insert(
+        &self,
+        dst: &mut IndexMap<Spanned<String>, Spanned<Value>>,
+        keys: &mut Key,
+        value: Spanned<Value>,
+        modify_array_element: bool,
+        append_last: bool,
+    ) {
+        let key = keys.pop_front().unwrap();
+        if keys.is_empty() {
+            if let RawEntryMut::Occupied(mut old) =
+                dst.raw_entry_mut_v1().from_key(key.value.as_str())
+            {
+                if append_last && let Value::Array(array) = &mut old.get_mut().value {
+                    array.push(value);
+                    return;
+                }
+                if let Value::Table(old) = &mut old.get_mut().value
+                    && let Value::Table(new) = value.value
+                {
+                    for (k, v) in new {
+                        let mut keys = Key::new();
+                        keys.push_back(k);
+                        self.insert(old, &mut keys, v, false, false);
+                    }
+                    return;
+                }
+                self.error_handler
+                    .redefinition(ParserError::Redefined.spanned(key.span), old.key().span);
+                old.shift_remove();
+            }
+            let span = value.span;
+            let value = match append_last {
+                true => Value::Array(vec![value]).spanned(span),
+                false => value,
+            };
+            dst.insert(key, value);
+        } else {
+            if let RawEntryMut::Occupied(mut o) = dst.raw_entry_mut_v1().from_key(&key) {
+                match &mut o.get_mut().value {
+                    Value::Table(dst) => {
+                        self.insert(dst, keys, value, modify_array_element, append_last);
+                        return;
+                    }
+                    Value::Array(array) if modify_array_element => {
+                        if let Some(Value::Table(dst)) =
+                            array.last_mut().as_mut().map(|v| &mut v.value)
+                        {
+                            self.insert(dst, keys, value, modify_array_element, append_last);
+                            return;
+                        }
+                    }
+                    _ => {}
+                }
+                self.error_handler
+                    .redefinition(ParserError::Redefined.spanned(key.span), o.key().span);
+                o.shift_remove();
+            }
+            let mut map = IndexMap::new();
+            let span = value.span;
+            self.insert(&mut map, keys, value, modify_array_element, append_last);
+            dst.insert(key, Value::Table(map).spanned(span));
+        }
+    }
+
+    fn parse_key_value_with_recovery(
+        &mut self,
+    ) -> Result<Option<(Key, Spanned<Value>)>, Spanned<ParserError>> {
+        let pos = self.lexer.pos();
+        match self.parse_key_value() {
+            Ok(kv) => Ok(Some(kv)),
+            Err((e, key)) => {
+                if let Some(key) = key {
+                    let span = key.back().unwrap().span;
+                    self.error_handler
+                        .handle(ParserError::IgnoringKey.spanned(span));
+                }
+                if self.lexer.pos() == pos {
+                    Err(e)
+                } else {
+                    self.error_handler.handle(e);
+                    Ok(None)
+                }
+            }
+        }
+    }
+
+    #[expect(clippy::type_complexity)]
+    fn parse_key_value(
+        &mut self,
+    ) -> Result<(Key, Spanned<Value>), (Spanned<ParserError>, Option<Key>)> {
+        let key = self.parse_key();
+        let eq = self.parse_exact(Token::Equals, true);
+        let value = self.parse_value();
+        let key = match key {
+            Ok(k) => k,
+            Err(e) => return Err((e, None)),
+        };
+        if let Err(e) = eq {
+            return Err((e, Some(key)));
+        }
+        let value = match value {
+            Ok(v) => v,
+            Err(e) => return Err((e, Some(key))),
+        };
+        Ok((key, value))
+    }
+
+    fn parse_key(&mut self) -> Result<Key, Spanned<ParserError>> {
+        let mut parts = Key::new();
+        loop {
+            if parts.len() > 0 {
+                if self.parse_exact(Token::Dot, false).is_err() {
+                    break;
+                }
+            }
+            let Some(token) = self.lexer.peek(false) else {
+                break;
+            };
+            let s = match token.value {
+                Token::LiteralString(s) => s.as_bstr().to_string(),
+                Token::CookedString(s) => self.cook_string(s),
+                Token::Literal(l) => l.as_bstr().to_string(),
+                _ => break,
+            };
+            parts.push_back(s.spanned(token.span));
+            let _ = self.next(false);
+        }
+        if parts.is_empty() {
+            Err(ParserError::MissingKey.spanned(self.next_span()))
+        } else {
+            Ok(parts)
+        }
+    }
+
+    fn parse_exact(
+        &mut self,
+        token: Token<'a>,
+        value_context: bool,
+    ) -> Result<Span, Spanned<ParserError>> {
+        let actual = match self.peek(value_context) {
+            Ok(t) if t.value == token => {
+                let _ = self.next(value_context);
+                return Ok(t.span);
+            }
+            Ok(t) => t.value.name(value_context),
+            Err(_) => "end of file",
+        };
+        let span = self.next_span();
+        Err(ParserError::Expected(token.name(value_context), actual).spanned(span))
+    }
+
+    fn last_span(&self) -> Span {
+        self.last_span.unwrap_or(Span { lo: 0, hi: 0 })
+    }
+
+    fn next_span(&mut self) -> Span {
+        self.lexer.peek(false).map(|v| v.span).unwrap_or_else(|| {
+            let hi = self.last_span().hi;
+            Span { lo: hi, hi }
+        })
+    }
+
+    fn skip_tree(&mut self, start: Token, end: Token) {
+        let mut depth = 1;
+        while let Ok(next) = self.next(false) {
+            if next.value == start {
+                depth += 1;
+            } else if next.value == end {
+                depth -= 1;
+                if depth == 0 {
+                    return;
+                }
+            }
+        }
+    }
+}

crates/toml-parser/src/toml_span.rs

@@ -0,0 +1,108 @@
+use std::{
+    borrow::Borrow,
+    fmt::{Debug, Formatter},
+    hash::{Hash, Hasher},
+};
+
+#[derive(Copy, Clone, Eq, PartialEq)]
+pub struct Span {
+    pub lo: usize,
+    pub hi: usize,
+}
+
+impl Debug for Span {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}..{}", self.lo, self.hi)
+    }
+}
+
+#[derive(Copy, Clone, Eq)]
+pub struct Spanned<T> {
+    pub span: Span,
+    pub value: T,
+}
+
+impl<T> Spanned<T> {
+    pub fn as_ref(&self) -> Spanned<&T> {
+        Spanned {
+            span: self.span,
+            value: &self.value,
+        }
+    }
+
+    pub fn map<U, F: FnOnce(T) -> U>(self, f: F) -> Spanned<U> {
+        Spanned {
+            span: self.span,
+            value: f(self.value),
+        }
+    }
+
+    pub fn into<U>(self) -> Spanned<U>
+    where
+        T: Into<U>,
+    {
+        Spanned {
+            span: self.span,
+            value: self.value.into(),
+        }
+    }
+}
+
+impl<T: Debug> Debug for Spanned<T> {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        self.value.fmt(f)?;
+        write!(f, " @ {:?}", self.span)
+    }
+}
+
+impl<T: PartialEq> PartialEq for Spanned<T> {
+    fn eq(&self, other: &Self) -> bool {
+        self.value.eq(&other.value)
+    }
+}
+
+impl<T: Hash> Hash for Spanned<T> {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.value.hash(state)
+    }
+}
+
+impl Borrow<str> for Spanned<String> {
+    fn borrow(&self) -> &str {
+        &self.value
+    }
+}
+
+pub trait SpannedExt: Sized {
+    fn spanned(self, span: Span) -> Spanned<Self>;
+}
+
+impl<T> SpannedExt for T {
+    fn spanned(self, span: Span) -> Spanned<Self> {
+        Spanned { span, value: self }
+    }
+}
+
+pub trait DespanExt: Sized {
+    type T;
+
+    fn despan(self) -> Option<Self::T>;
+    fn despan_into<U>(self) -> Option<U>
+    where
+        Self::T: Into<U>;
+}
+
+impl<T> DespanExt for Option<Spanned<T>> {
+    type T = T;
+
+    fn despan(self) -> Option<Self::T> {
+        self.map(|v| v.value)
+    }
+
+    fn despan_into<U>(self) -> Option<U>
+    where
+        Self::T: Into<U>,
+    {
+        self.map(|v| v.value.into())
+    }
+}

crates/toml-parser/src/toml_value.rs

@@ -0,0 +1,63 @@
+use {
+    crate::toml_span::Spanned,
+    indexmap::IndexMap,
+    std::{
+        cmp::Ordering,
+        fmt::{Debug, Formatter},
+    },
+};
+
+pub enum Value {
+    String(String),
+    Integer(i64),
+    Float(f64),
+    Boolean(bool),
+    Array(Vec<Spanned<Value>>),
+    Table(IndexMap<Spanned<String>, Spanned<Value>>),
+}
+
+impl Value {
+    pub fn name(&self) -> &'static str {
+        match self {
+            Value::String(_) => "a string",
+            Value::Integer(_) => "an integer",
+            Value::Float(_) => "a float",
+            Value::Boolean(_) => "a boolean",
+            Value::Array(_) => "an array",
+            Value::Table(_) => "a table",
+        }
+    }
+}
+
+impl Debug for Value {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Value::String(v) => v.fmt(f),
+            Value::Integer(v) => v.fmt(f),
+            Value::Float(v) => v.fmt(f),
+            Value::Boolean(v) => v.fmt(f),
+            Value::Array(v) => v.fmt(f),
+            Value::Table(v) => v.fmt(f),
+        }
+    }
+}
+
+impl PartialEq for Value {
+    fn eq(&self, other: &Self) -> bool {
+        match (self, other) {
+            (Value::String(v1), Value::String(v2)) => v1 == v2,
+            (Value::Integer(v1), Value::Integer(v2)) => v1 == v2,
+            (Value::Float(v1), Value::Float(v2)) => {
+                if v1.is_nan() && v2.is_nan() {
+                    true
+                } else {
+                    v1.total_cmp(v2) == Ordering::Equal
+                }
+            }
+            (Value::Boolean(v1), Value::Boolean(v2)) => v1 == v2,
+            (Value::Array(v1), Value::Array(v2)) => v1 == v2,
+            (Value::Table(v1), Value::Table(v2)) => v1 == v2,
+            _ => false,
+        }
+    }
+}

crates/toml-parser/src/value_ext.rs

@@ -0,0 +1,18 @@
+use crate::{
+    toml_span::Span,
+    toml_value::Value,
+    value_parser::{ParseResult, Parser},
+};
+
+impl Value {
+    pub fn parse<P: Parser>(&self, span: Span, parser: &mut P) -> ParseResult<P> {
+        match self {
+            Value::String(a) => parser.parse_string(span, a),
+            Value::Integer(a) => parser.parse_integer(span, *a),
+            Value::Float(a) => parser.parse_float(span, *a),
+            Value::Boolean(a) => parser.parse_bool(span, *a),
+            Value::Array(a) => parser.parse_array(span, a),
+            Value::Table(a) => parser.parse_table(span, a),
+        }
+    }
+}

crates/toml-parser/src/value_parser.rs

@@ -0,0 +1,118 @@
+use {
+    crate::{
+        toml_span::{Span, Spanned, SpannedExt},
+        toml_value::Value,
+    },
+    indexmap::IndexMap,
+    std::{
+        error::Error,
+        fmt::{self, Display, Formatter},
+    },
+};
+
+#[derive(Copy, Clone, Debug)]
+pub enum DataType {
+    String,
+    Integer,
+    Float,
+    Boolean,
+    Array,
+    Table,
+}
+
+impl Display for DataType {
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        let s = match self {
+            DataType::String => "a string",
+            DataType::Integer => "an integer",
+            DataType::Float => "a float",
+            DataType::Boolean => "a bool",
+            DataType::Array => "an array",
+            DataType::Table => "a table",
+        };
+        f.write_str(s)
+    }
+}
+
+pub struct DataTypes(&'static [DataType]);
+
+impl Display for DataTypes {
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        let d = self.0;
+        match d.len() {
+            0 => Ok(()),
+            1 => d[0].fmt(f),
+            2 => write!(f, "{} or {}", d[0], d[1]),
+            _ => {
+                let mut first = true;
+                #[expect(clippy::needless_range_loop)]
+                for i in 0..d.len() - 1 {
+                    if !first {
+                        f.write_str(", ")?;
+                    }
+                    first = false;
+                    d[i].fmt(f)?;
+                }
+                write!(f, ", or {}", d[d.len() - 1])
+            }
+        }
+    }
+}
+
+#[derive(Debug)]
+pub struct UnexpectedDataType(&'static [DataType], DataType);
+
+impl Display for UnexpectedDataType {
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        write!(f, "Expected {} but found {}", DataTypes(self.0), self.1)
+    }
+}
+
+impl Error for UnexpectedDataType {}
+
+pub type ParseResult<P> = Result<<P as Parser>::Value, Spanned<<P as Parser>::Error>>;
+
+pub trait Parser {
+    type Value;
+    type Error: From<UnexpectedDataType>;
+
+    const EXPECTED: &'static [DataType];
+
+    fn parse_string(&mut self, span: Span, string: &str) -> ParseResult<Self> {
+        let _ = string;
+        expected(self, span, DataType::String)
+    }
+
+    fn parse_integer(&mut self, span: Span, integer: i64) -> ParseResult<Self> {
+        let _ = integer;
+        expected(self, span, DataType::Integer)
+    }
+
+    fn parse_float(&mut self, span: Span, float: f64) -> ParseResult<Self> {
+        let _ = float;
+        expected(self, span, DataType::Float)
+    }
+
+    fn parse_bool(&mut self, span: Span, bool: bool) -> ParseResult<Self> {
+        let _ = bool;
+        expected(self, span, DataType::Boolean)
+    }
+
+    fn parse_array(&mut self, span: Span, array: &[Spanned<Value>]) -> ParseResult<Self> {
+        let _ = array;
+        expected(self, span, DataType::Array)
+    }
+
+    fn parse_table(
+        &mut self,
+        span: Span,
+        table: &IndexMap<Spanned<String>, Spanned<Value>>,
+    ) -> ParseResult<Self> {
+        let _ = table;
+        expected(self, span, DataType::Table)
+    }
+}
+
+fn expected<P: Parser + ?Sized>(_p: &P, span: Span, actual: DataType) -> ParseResult<P> {
+    Err(P::Error::from(UnexpectedDataType(P::EXPECTED, actual)).spanned(span))
+}