rantlr-core 0.2.0

//! In-process interpreter: run a validated [`Grammar`] against test input.
//! Powers the Tauri playground without requiring generated host-language code.

use std::collections::BTreeMap;

use regex::Regex;
use serde::Serialize;

use crate::ast::{Expr, Grammar, GrammarItem, RuleDecl, TokenBody};
use crate::error::RantlrError;
use crate::span::Span;
use crate::token::BuiltinType;
use crate::{analyze, Diagnostic};

#[derive(Debug, Clone, Serialize)]
pub struct ParseTree {
    pub kind: String,
    pub text: String,
    pub children: Vec<ParseTree>,
}

#[derive(Debug, Clone, Serialize)]
pub struct AnalyzeResult {
    pub grammar_name: String,
    pub tree: ParseTree,
}

#[derive(Debug, Clone, Serialize)]
#[serde(rename_all = "camelCase")]
pub struct DiagnosticJson {
    pub severity: String,
    pub kind: String,
    pub line: usize,
    pub column: usize,
    pub end_line: usize,
    pub end_column: usize,
    pub message: String,
    pub help: Option<String>,
    /// Stable code for code actions, e.g. `left-recursion`.
    pub code: Option<String>,
    pub fix_hint: Option<crate::FixHint>,
}

/// Parse + lint a `.gr` source; return structured diagnostics (empty = ok).
pub fn diagnose(source: &str) -> Vec<DiagnosticJson> {
    match crate::parser::parse(source) {
        Err(err) => vec![diagnostic_from_error(source, &err)],
        Ok(grammar) => {
            let mut out = Vec::new();
            for diag in crate::linter::lint_all(source, &grammar) {
                out.push(diagnostic_from_diag(source, &diag));
            }
            out
        }
    }
}

/// Validate grammar, then parse `test_input` with an in-process RD interpreter.
pub fn analyze_and_parse(source: &str, test_input: &str) -> Result<AnalyzeResult, RantlrError> {
    let grammar = analyze(source)?;
    let tree = parse_input(&grammar, test_input)?;
    Ok(AnalyzeResult {
        grammar_name: grammar.name,
        tree,
    })
}

pub fn parse_input(grammar: &Grammar, input: &str) -> Result<ParseTree, RantlrError> {
    let engine = Engine::from_grammar(grammar).map_err(|msg| {
        RantlrError::parse(input, msg, Span::from_offsets(0, 0))
    })?;
    engine.parse(input)
}

fn diagnostic_from_error(_source: &str, err: &RantlrError) -> DiagnosticJson {
    let end_col = err.column.saturating_add(1);
    DiagnosticJson {
        severity: "Error".into(),
        kind: format!("{:?}", err.kind),
        line: err.line,
        column: err.column,
        end_line: err.line,
        end_column: end_col,
        message: err.message.clone(),
        help: err.help.clone(),
        code: err.code.clone(),
        fix_hint: err.fix_hint.clone(),
    }
}

fn diagnostic_from_diag(source: &str, diag: &Diagnostic) -> DiagnosticJson {
    let (line, column) = diag.span.line_col(source);
    let end = diag.span.end.as_usize().min(source.len());
    let (end_line, end_column) = Span::from_offsets(end, end).line_col(source);
    DiagnosticJson {
        severity: format!("{:?}", diag.severity),
        kind: format!("{:?}", diag.kind),
        line,
        column,
        end_line,
        end_column: end_column.max(column + 1),
        message: diag.message.clone(),
        help: diag.help.clone(),
        code: diag.code.clone(),
        fix_hint: diag.fix.clone(),
    }
}

struct TokDef {
    name: String,
    re: Regex,
    skip: bool,
}

struct Engine<'g> {
    rules: BTreeMap<String, &'g RuleDecl>,
    start: String,
    tokens: Vec<TokDef>,
}

#[derive(Clone)]
struct LexTok {
    name: String,
    text: String,
}

struct Parser<'a> {
    tokens: &'a [LexTok],
    pos: usize,
    rules: &'a BTreeMap<String, &'a RuleDecl>,
}

impl<'g> Engine<'g> {
    fn from_grammar(grammar: &'g Grammar) -> Result<Self, String> {
        let mut rules = BTreeMap::new();
        let mut start = None;
        let mut declared = Vec::new();

        for item in &grammar.items {
            match item {
                GrammarItem::Rule(r) => {
                    if start.is_none() {
                        start = Some(r.name.clone());
                    }
                    rules.insert(r.name.clone(), r);
                }
                GrammarItem::Token(t) => {
                    declared.push((t.name.clone(), pattern_for_body(&t.body)?, t.skip));
                }
                _ => {}
            }
        }

        let start = start.ok_or_else(|| "grammar has no rules".to_string())?;

        let mut inferred: BTreeMap<String, String> = BTreeMap::new();
        for rule in rules.values() {
            collect_literals(&rule.body, &mut inferred);
        }

        let mut tokens = Vec::new();
        for (name, pat, skip) in declared {
            tokens.push(TokDef {
                name,
                re: Regex::new(&format!("^{pat}")).map_err(|e| e.to_string())?,
                skip,
            });
        }
        for (lit, name) in inferred {
            if tokens.iter().any(|t| t.name == name) {
                continue;
            }
            let pat = regex_escape(&lit);
            tokens.push(TokDef {
                name,
                re: Regex::new(&format!("^{pat}")).map_err(|e| e.to_string())?,
                skip: false,
            });
        }

        tokens.sort_by(|a, b| {
            // Prefer specific literals over broad Identifier / Number patterns.
            token_specificity(b).cmp(&token_specificity(a)).then_with(|| {
                b.re.as_str().len().cmp(&a.re.as_str().len())
            })
        });

        Ok(Self {
            rules,
            start,
            tokens,
        })
    }

    fn parse(&self, input: &str) -> Result<ParseTree, RantlrError> {
        let lexed = self.tokenize(input)?;
        let mut parser = Parser {
            tokens: &lexed,
            pos: 0,
            rules: &self.rules,
        };
        let tree = parser.parse_rule(&self.start)?;
        if !parser.at_end() {
            let t = parser.peek();
            return Err(RantlrError::parse(
                input,
                format!("unexpected trailing input (`{}`)", t.name),
                Span::from_offsets(0, 0),
            ));
        }
        Ok(tree)
    }

    fn tokenize(&self, input: &str) -> Result<Vec<LexTok>, RantlrError> {
        let mut out = Vec::new();
        let mut i = 0;
        let bytes = input.as_bytes();
        while i < input.len() {
            let rest = &input[i..];
            let mut best: Option<(usize, i32, &TokDef)> = None;
            for tok in &self.tokens {
                if let Some(m) = tok.re.find(rest) {
                    if m.start() == 0 {
                        let len = m.end();
                        let spec = token_specificity(tok);
                        let better = match best {
                            None => true,
                            Some((bl, bs, _)) => len > bl || (len == bl && spec > bs),
                        };
                        if better {
                            best = Some((len, spec, tok));
                        }
                    }
                }
            }
            let Some((len, _, tok)) = best else {
                let (line, col) = Span::from_offsets(i, i).line_col(input);
                return Err(RantlrError {
                    kind: crate::error::ErrorKind::Parse,
                    message: format!(
                        "unexpected character `{}`",
                        input[i..].chars().next().unwrap_or('?')
                    ),
                    line,
                    column: col,
                    snippet: crate::error::context_snippet(input, Span::from_offsets(i, i + 1)),
                    help: None,
                    code: None,
                    fix_hint: None,
                });
            };
            if !tok.skip {
                out.push(LexTok {
                    name: tok.name.clone(),
                    text: input[i..i + len].to_string(),
                });
            }
            i += len;
            let _ = bytes; // silence
        }
        out.push(LexTok {
            name: "EOF".into(),
            text: String::new(),
        });
        Ok(out)
    }
}

impl<'a> Parser<'a> {
    fn at_end(&self) -> bool {
        self.peek().name == "EOF"
    }

    fn peek(&self) -> &LexTok {
        &self.tokens[self.pos.min(self.tokens.len() - 1)]
    }

    fn mark(&self) -> usize {
        self.pos
    }

    fn reset(&mut self, m: usize) {
        self.pos = m;
    }

    fn expect_token(&mut self, name: &str) -> Result<ParseTree, RantlrError> {
        let tok = self.peek().clone();
        if tok.name == name {
            if tok.name != "EOF" {
                self.pos += 1;
            }
            return Ok(ParseTree {
                kind: name.into(),
                text: tok.text,
                children: vec![],
            });
        }
        Err(self.err(format!("expected `{name}`, found `{}`", tok.name)))
    }

    fn parse_rule(&mut self, name: &str) -> Result<ParseTree, RantlrError> {
        let rule = self
            .rules
            .get(name)
            .copied()
            .ok_or_else(|| self.err(format!("unknown rule `{name}`")))?;
        let children = self.parse_expr(&rule.body)?;
        let text = children.iter().map(|c| c.text.as_str()).collect::<String>();
        Ok(ParseTree {
            kind: name.into(),
            text,
            children,
        })
    }

    fn parse_expr(&mut self, expr: &Expr) -> Result<Vec<ParseTree>, RantlrError> {
        match expr {
            Expr::Seq { items } => {
                let mut out = Vec::new();
                for item in items {
                    out.extend(self.parse_expr(item)?);
                }
                Ok(out)
            }
            Expr::Group { body } => self.parse_expr(body),
            Expr::Ref { name, .. } => {
                if self.rules.contains_key(name) {
                    Ok(vec![self.parse_rule(name)?])
                } else {
                    Ok(vec![self.expect_token(name)?])
                }
            }
            Expr::Literal { value, .. } => {
                let tok_name = literal_token_name(value);
                Ok(vec![self.expect_token(&tok_name)?])
            }
            Expr::Optional { body } => {
                let m = self.mark();
                match self.parse_expr(body) {
                    Ok(nodes) => Ok(nodes),
                    Err(_) => {
                        self.reset(m);
                        Ok(vec![])
                    }
                }
            }
            Expr::Repeat { min, max, body } => {
                let mut out = Vec::new();
                let mut n = 0u64;
                loop {
                    if let Some(max) = *max {
                        if n >= max {
                            break;
                        }
                    }
                    let m = self.mark();
                    match self.parse_expr(body) {
                        Ok(nodes) => {
                            out.extend(nodes);
                            n += 1;
                        }
                        Err(_) => {
                            self.reset(m);
                            break;
                        }
                    }
                }
                if let Some(min) = *min {
                    if n < min {
                        return Err(self.err(format!("expected at least {min} repetitions")));
                    }
                }
                Ok(out)
            }
            Expr::Match { arms } | Expr::Alt { alts: arms } => {
                let mut last_err = None;
                for arm in arms {
                    let m = self.mark();
                    match self.parse_expr(arm) {
                        Ok(nodes) => return Ok(nodes),
                        Err(e) => {
                            self.reset(m);
                            last_err = Some(e);
                        }
                    }
                }
                Err(last_err.unwrap_or_else(|| self.err("no alternative matched")))
            }
        }
    }

    fn err(&self, message: impl Into<String>) -> RantlrError {
        RantlrError {
            kind: crate::error::ErrorKind::Parse,
            message: message.into(),
            line: 1,
            column: 1,
            snippet: String::new(),
            help: Some(format!("at token `{}`", self.peek().name)),
            code: None,
            fix_hint: None,
        }
    }
}

fn pattern_for_body(body: &TokenBody) -> Result<String, String> {
    Ok(match body {
        TokenBody::Literal(s) => regex_escape(s),
        TokenBody::Builtin(b) => builtin_regex(*b).to_string(),
    })
}

/// Higher = win ties against broader patterns (Identifier swallowing keywords).
fn token_specificity(tok: &TokDef) -> i32 {
    if tok.name.starts_with("Lit_") {
        return 100;
    }
    let pat = tok.re.as_str();
    // Anchored exact-ish literals from inferred keywords are short; Identifier is broad.
    if pat.contains("[A-Za-z_") || pat.contains("[0-9]") {
        return 10;
    }
    50
}

fn builtin_regex(b: BuiltinType) -> &'static str {
    match b {
        BuiltinType::Number => r"[0-9]+(?:\.[0-9]+)?",
        BuiltinType::QuotedString => r#"(?:"(?:\\.|[^"\\])*"|'(?:\\.|[^'\\])*')"#,
        BuiltinType::Email => r"[A-Za-z0-9._%+\-]+@[A-Za-z0-9.\-]+\.[A-Za-z]{2,}",
        BuiltinType::Url => r"https?://[^\s]+",
        BuiltinType::DateTime => r"\d{4}-\d{2}-\d{2}(?:[T ]\d{2}:\d{2}:\d{2})?",
        BuiltinType::Identifier => r"[A-Za-z_][A-Za-z0-9_]*",
    }
}

fn regex_escape(s: &str) -> String {
    let mut out = String::new();
    for c in s.chars() {
        if matches!(
            c,
            '\\' | '.' | '+' | '*' | '?' | '(' | ')' | '[' | ']' | '{' | '}' | '^' | '$' | '|'
        ) {
            out.push('\\');
        }
        out.push(c);
    }
    out
}

fn collect_literals(expr: &Expr, out: &mut BTreeMap<String, String>) {
    match expr {
        Expr::Alt { alts } | Expr::Seq { items: alts } | Expr::Match { arms: alts } => {
            for e in alts {
                collect_literals(e, out);
            }
        }
        Expr::Repeat { body, .. } | Expr::Optional { body } | Expr::Group { body } => {
            collect_literals(body, out);
        }
        Expr::Literal { value, .. } => {
            out.entry(value.clone())
                .or_insert_with(|| literal_token_name(value));
        }
        Expr::Ref { .. } => {}
    }
}

fn literal_token_name(lit: &str) -> String {
    let mapped: String = lit
        .chars()
        .map(|c| match c {
            '+' => "Plus".into(),
            '-' => "Minus".into(),
            '*' => "Star".into(),
            '/' => "Slash".into(),
            '(' => "LParen".into(),
            ')' => "RParen".into(),
            c if c.is_ascii_alphanumeric() => c.to_string(),
            _ => format!("U{:04X}", c as u32),
        })
        .collect();
    format!("Lit_{mapped}")
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn interprets_calculator() {
        let src = include_str!("../testdata/calculator.gr");
        let result = analyze_and_parse(src, "1+2*3").unwrap();
        assert_eq!(result.grammar_name, "Calculator");
        assert_eq!(result.tree.kind, "prog");
        assert_eq!(result.tree.text, "1+2*3");
    }

    #[test]
    fn diagnose_left_recursion() {
        let src = include_str!("../testdata/left_recursive.gr");
        let diags = diagnose(src);
        assert_eq!(diags.len(), 1);
        assert_eq!(diags[0].code.as_deref(), Some("left-recursion"));
    }

    #[test]
    fn interprets_api_gateway() {
        let src = include_str!("../testdata/api_gateway.gr");
        let input = r#"gateway api_v1 { route /users/:id { methods GET POST limit 10000 per ip balance RoundRobin auth jwt oauth nested posts { constraint "^[a-z0-9-]{3,40}$" methods GET limit 5000 per ip balance LeastConn } } }"#;
        let result = analyze_and_parse(src, input).expect("api gateway");
        assert_eq!(result.grammar_name, "ApiGateway");
        assert_eq!(result.tree.kind, "gateway");
    }
}