Add Decodal formatter

2026-07-09 17:39:32 +09:00
parent 16006d000a
commit 22f6bc1ab9
16 changed files with 606 additions and 24 deletions
@@ -0,0 +1,442 @@
+use alloc::string::{String, ToString};
+use core::fmt::Write;
+
+use crate::{
+    Ast, Diagnostic, Expr, ExprId, Field, Literal, Param, SourceForm, Span,
+    ast::{BinaryOp, CompareOp, MatchArm, UnaryOp},
+    diagnostic::Result,
+    parse_source_with_source_id,
+    span::SourceId,
+};
+
+const INDENT: usize = 4;
+
+pub fn format_source(source: &str) -> Result<String> {
+    format_source_with_source_id(SourceId(0), source)
+}
+
+pub fn format_source_with_source_id(source_id: SourceId, source: &str) -> Result<String> {
+    reject_comments(source_id, source)?;
+    let parsed = parse_source_with_source_id(source_id, source)?;
+    Ok(Formatter::new(&parsed.ast).format_root(parsed.root, parsed.source_form))
+}
+
+struct Formatter<'a> {
+    ast: &'a Ast,
+}
+
+impl<'a> Formatter<'a> {
+    fn new(ast: &'a Ast) -> Self {
+        Self { ast }
+    }
+
+    fn format_root(&self, root: ExprId, source_form: SourceForm) -> String {
+        let mut out = String::new();
+        match source_form {
+            SourceForm::Fields => {
+                if let Expr::Object(fields) = &self.ast.get(root).expr {
+                    self.write_fields(&mut out, fields, 0);
+                } else {
+                    self.write_expr(&mut out, root, 0, 0);
+                }
+            }
+            SourceForm::Expr => self.write_expr(&mut out, root, 0, 0),
+        }
+        if !out.ends_with('\n') {
+            out.push('\n');
+        }
+        out
+    }
+
+    fn write_expr(&self, out: &mut String, id: ExprId, indent: usize, parent_prec: u8) {
+        let expr = &self.ast.get(id).expr;
+        let prec = expr_prec(expr);
+        let paren = prec < parent_prec;
+        if paren {
+            out.push('(');
+        }
+        match expr {
+            Expr::Literal(literal) => self.write_literal(out, literal),
+            Expr::Ident(name) => out.push_str(name),
+            Expr::Object(fields) => self.write_object(out, fields, indent),
+            Expr::Array(items) => self.write_array(out, items, indent),
+            Expr::Let { bindings, body } => self.write_let(out, bindings, *body, indent),
+            Expr::Import(path) => {
+                out.push_str("import ");
+                write_quoted(out, path);
+            }
+            Expr::Path { base, field } => {
+                self.write_expr(out, *base, indent, prec);
+                out.push('.');
+                out.push_str(field);
+            }
+            Expr::Call { callee, args } => {
+                self.write_expr(out, *callee, indent, prec);
+                out.push('(');
+                for (index, arg) in args.iter().enumerate() {
+                    if index > 0 {
+                        out.push_str(", ");
+                    }
+                    self.write_expr(out, *arg, indent, 0);
+                }
+                out.push(')');
+            }
+            Expr::Function { params, body } => self.write_function(out, params, *body, indent),
+            Expr::Match { scrutinee, arms } => self.write_match(out, *scrutinee, arms, indent),
+            Expr::Unary { op, expr } => {
+                out.push_str(match op {
+                    UnaryOp::Neg => "-",
+                    UnaryOp::Not => "!",
+                });
+                self.write_expr(out, *expr, indent, prec);
+            }
+            Expr::Binary { op, lhs, rhs } => {
+                self.write_expr(out, *lhs, indent, prec);
+                out.push(' ');
+                out.push_str(binary_op_text(*op));
+                out.push(' ');
+                self.write_expr(out, *rhs, indent, prec + 1);
+            }
+            Expr::Default { base, fallback } => {
+                self.write_expr(out, *base, indent, prec);
+                out.push_str(" default ");
+                self.write_expr(out, *fallback, indent, prec + 1);
+            }
+            Expr::CompareConstraint { op, value } => {
+                out.push_str(compare_op_text(*op));
+                out.push(' ');
+                self.write_expr(out, *value, indent, prec);
+            }
+            Expr::RegexConstraint(pattern) => {
+                out.push('/');
+                out.push_str(pattern);
+                out.push('/');
+            }
+            Expr::Wildcard => out.push('_'),
+        }
+        if paren {
+            out.push(')');
+        }
+    }
+
+    fn write_literal(&self, out: &mut String, literal: &Literal) {
+        match literal {
+            Literal::String(value) => write_quoted(out, value),
+            Literal::Int(value) => {
+                let _ = write!(out, "{value}");
+            }
+            Literal::Float(value) => {
+                let text = value.to_string();
+                out.push_str(&text);
+                if !text.contains('.') && !text.contains('e') && !text.contains('E') {
+                    out.push_str(".0");
+                }
+            }
+            Literal::Bool(value) => out.push_str(if *value { "true" } else { "false" }),
+        }
+    }
+
+    fn write_object(&self, out: &mut String, fields: &[Field], indent: usize) {
+        if fields.is_empty() {
+            out.push_str("{}");
+            return;
+        }
+        out.push('{');
+        out.push('\n');
+        self.write_fields(out, fields, indent + INDENT);
+        write_indent(out, indent);
+        out.push('}');
+    }
+
+    fn write_fields(&self, out: &mut String, fields: &[Field], indent: usize) {
+        for field in fields {
+            write_indent(out, indent);
+            self.write_field(out, field, indent);
+            out.push_str(";\n");
+        }
+    }
+
+    fn write_field(&self, out: &mut String, field: &Field, indent: usize) {
+        for (index, part) in field.path.iter().enumerate() {
+            if index > 0 {
+                out.push('.');
+            }
+            out.push_str(part);
+        }
+        out.push_str(" = ");
+        self.write_expr(out, field.value, indent, 0);
+    }
+
+    fn write_array(&self, out: &mut String, items: &[ExprId], indent: usize) {
+        if items.is_empty() {
+            out.push_str("[]");
+            return;
+        }
+        if items.iter().all(|item| self.is_inline(*item)) {
+            out.push('[');
+            for (index, item) in items.iter().enumerate() {
+                if index > 0 {
+                    out.push_str(", ");
+                }
+                self.write_expr(out, *item, indent, 0);
+            }
+            out.push(']');
+            return;
+        }
+        out.push('[');
+        out.push('\n');
+        for item in items {
+            write_indent(out, indent + INDENT);
+            self.write_expr(out, *item, indent + INDENT, 0);
+            out.push_str(",\n");
+        }
+        write_indent(out, indent);
+        out.push(']');
+    }
+
+    fn write_let(&self, out: &mut String, bindings: &[Field], body: ExprId, indent: usize) {
+        out.push_str("let");
+        out.push('\n');
+        self.write_fields(out, bindings, indent + INDENT);
+        write_indent(out, indent);
+        out.push_str("in");
+        out.push('\n');
+        write_indent(out, indent + INDENT);
+        self.write_expr(out, body, indent + INDENT, 0);
+    }
+
+    fn write_function(&self, out: &mut String, params: &[Param], body: ExprId, indent: usize) {
+        out.push('(');
+        for (index, param) in params.iter().enumerate() {
+            if index > 0 {
+                out.push_str(", ");
+            }
+            self.write_param(out, param, indent);
+        }
+        out.push(')');
+        out.push_str(" =>");
+        if self.is_inline(body) {
+            out.push(' ');
+            self.write_expr(out, body, indent, 0);
+        } else {
+            out.push('\n');
+            write_indent(out, indent + INDENT);
+            self.write_expr(out, body, indent + INDENT, 0);
+        }
+    }
+
+    fn write_param(&self, out: &mut String, param: &Param, indent: usize) {
+        out.push_str(&param.name);
+        if let Some(constraint) = param.constraint {
+            out.push_str(": ");
+            self.write_expr(out, constraint, indent, 0);
+        }
+    }
+
+    fn write_match(&self, out: &mut String, scrutinee: ExprId, arms: &[MatchArm], indent: usize) {
+        out.push_str("match ");
+        self.write_expr(out, scrutinee, indent, 0);
+        out.push_str(" {");
+        if !arms.is_empty() {
+            out.push('\n');
+            for arm in arms {
+                write_indent(out, indent + INDENT);
+                self.write_expr(out, arm.pattern, indent + INDENT, 0);
+                out.push_str(": ");
+                self.write_expr(out, arm.body, indent + INDENT, 0);
+                out.push_str(";\n");
+            }
+            write_indent(out, indent);
+        }
+        out.push('}');
+    }
+
+    fn is_inline(&self, id: ExprId) -> bool {
+        match &self.ast.get(id).expr {
+            Expr::Literal(_)
+            | Expr::Ident(_)
+            | Expr::Import(_)
+            | Expr::RegexConstraint(_)
+            | Expr::Wildcard => true,
+            Expr::Path { base, .. } => self.is_inline(*base),
+            Expr::Call { callee, args } => {
+                self.is_inline(*callee) && args.iter().all(|arg| self.is_inline(*arg))
+            }
+            Expr::Unary { expr, .. } => self.is_inline(*expr),
+            Expr::Binary { lhs, rhs, .. } => self.is_inline(*lhs) && self.is_inline(*rhs),
+            Expr::Default { base, fallback } => self.is_inline(*base) && self.is_inline(*fallback),
+            Expr::CompareConstraint { value, .. } => self.is_inline(*value),
+            Expr::Array(items) => items.iter().all(|item| self.is_inline(*item)),
+            Expr::Object(fields) => fields.is_empty(),
+            Expr::Let { .. } | Expr::Function { .. } | Expr::Match { .. } => false,
+        }
+    }
+}
+
+fn expr_prec(expr: &Expr) -> u8 {
+    match expr {
+        Expr::Default { .. } => 1,
+        Expr::Binary { op, .. } => binary_prec(*op),
+        Expr::Unary { .. } | Expr::CompareConstraint { .. } => 17,
+        Expr::Path { .. } | Expr::Call { .. } => 19,
+        _ => 20,
+    }
+}
+
+fn binary_prec(op: BinaryOp) -> u8 {
+    match op {
+        BinaryOp::Patch => 3,
+        BinaryOp::And => 5,
+        BinaryOp::LogicalOr => 7,
+        BinaryOp::LogicalAnd => 9,
+        BinaryOp::Equal
+        | BinaryOp::NotEqual
+        | BinaryOp::Greater
+        | BinaryOp::GreaterEqual
+        | BinaryOp::Less
+        | BinaryOp::LessEqual => 11,
+        BinaryOp::Concat => 12,
+        BinaryOp::Add | BinaryOp::Sub => 13,
+        BinaryOp::Mul | BinaryOp::Div => 15,
+    }
+}
+
+fn binary_op_text(op: BinaryOp) -> &'static str {
+    match op {
+        BinaryOp::Add => "+",
+        BinaryOp::Sub => "-",
+        BinaryOp::Mul => "*",
+        BinaryOp::Div => "/",
+        BinaryOp::Concat => "++",
+        BinaryOp::Equal => "==",
+        BinaryOp::NotEqual => "!=",
+        BinaryOp::Greater => ">",
+        BinaryOp::GreaterEqual => ">=",
+        BinaryOp::Less => "<",
+        BinaryOp::LessEqual => "<=",
+        BinaryOp::LogicalAnd => "&&",
+        BinaryOp::LogicalOr => "||",
+        BinaryOp::And => "&",
+        BinaryOp::Patch => "//",
+    }
+}
+
+fn compare_op_text(op: CompareOp) -> &'static str {
+    match op {
+        CompareOp::Gt => ">",
+        CompareOp::Gte => ">=",
+        CompareOp::Lt => "<",
+        CompareOp::Lte => "<=",
+        CompareOp::Eq => "==",
+    }
+}
+
+fn write_indent(out: &mut String, indent: usize) {
+    for _ in 0..indent {
+        out.push(' ');
+    }
+}
+
+fn write_quoted(out: &mut String, value: &str) {
+    out.push('"');
+    for ch in value.chars() {
+        match ch {
+            '"' => out.push_str("\\\""),
+            '\\' => out.push_str("\\\\"),
+            '\n' => out.push_str("\\n"),
+            '\r' => out.push_str("\\r"),
+            '\t' => out.push_str("\\t"),
+            ch if ch.is_control() => {
+                let _ = write!(out, "\\u{:04x}", ch as u32);
+            }
+            ch => out.push(ch),
+        }
+    }
+    out.push('"');
+}
+
+fn reject_comments(source_id: SourceId, source: &str) -> Result<()> {
+    if let Some(offset) = comment_offset(source) {
+        return Err(Diagnostic::syntax(
+            Span::new(source_id, offset, offset + 1),
+            "formatter does not yet preserve comments",
+        ));
+    }
+    Ok(())
+}
+
+fn comment_offset(source: &str) -> Option<usize> {
+    let bytes = source.as_bytes();
+    let mut index = 0;
+    while index < bytes.len() {
+        match bytes[index] {
+            b'"' => {
+                index += 1;
+                while index < bytes.len() {
+                    match bytes[index] {
+                        b'\\' => index += 2,
+                        b'"' => {
+                            index += 1;
+                            break;
+                        }
+                        _ => index += 1,
+                    }
+                }
+            }
+            b'#' => return Some(index),
+            _ => index += 1,
+        }
+    }
+    None
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::parse_source;
+
+    #[test]
+    fn formats_top_level_fields() {
+        let formatted = format_source("port=Int&>=1 default 8080;name=String;").unwrap();
+        assert_eq!(
+            formatted,
+            "port = Int & >= 1 default 8080;\nname = String;\n"
+        );
+        parse_source(&formatted).unwrap();
+    }
+
+    #[test]
+    fn formats_nested_objects() {
+        let formatted =
+            format_source("Server={host=String default \"localhost\";port=Int&>0;};").unwrap();
+        assert_eq!(
+            formatted,
+            "Server = {\n    host = String default \"localhost\";\n    port = Int & > 0;\n};\n"
+        );
+        parse_source(&formatted).unwrap();
+    }
+
+    #[test]
+    fn formats_let_function_match_and_array() {
+        let source = "let f=(x:Int)=>match x{_: [1,{a=2;}];};in f(1)";
+        let formatted = format_source(source).unwrap();
+        assert_eq!(
+            formatted,
+            "let\n    f = (x: Int) =>\n        match x {\n            _: [\n                1,\n                {\n                    a = 2;\n                },\n            ];\n        };\nin\n    f(1)\n"
+        );
+        parse_source(&formatted).unwrap();
+    }
+
+    #[test]
+    fn preserves_needed_parentheses() {
+        let formatted = format_source("value=(1+2)*3;").unwrap();
+        assert_eq!(formatted, "value = (1 + 2) * 3;\n");
+        parse_source(&formatted).unwrap();
+    }
+
+    #[test]
+    fn rejects_comments_instead_of_dropping_them() {
+        let error = format_source("# keep me\nvalue = 1;").unwrap_err();
+        assert_eq!(error.message, "formatter does not yet preserve comments");
+    }
+}
@@ -7,6 +7,7 @@ pub mod constraints;
 pub mod diagnostic;
 pub mod embedding;
 pub mod eval;
+pub mod formatter;
 mod lexer;
 pub mod module;
 pub mod parser;
@@ -21,6 +22,7 @@ pub use decodal_derive::Decodal;
 pub use diagnostic::{Diagnostic, DiagnosticKind, Result};
 pub use embedding::{HostField, HostValue};
 pub use eval::{Engine, format_diagnostic_with};
+pub use formatter::{format_source, format_source_with_source_id};
 pub use module::{EmptyLoader, LoadedSource, Module, SourceLoader};
 pub use parser::{ParseOutput, Parser, SourceForm, parse_source, parse_source_with_source_id};
 pub use runtime::{Constraint, Data, ExprRef, LiteralValue, ModuleId, PrimitiveType, RuntimeValue};