git.delta.rocks / jrsonnet / refs/commits / 8eff8514f067

1#![feature(box_syntax)]2#![feature(test)]34extern crate test;56use peg::parser;7use std::{path::PathBuf, rc::Rc};8mod expr;9pub use expr::*;10pub use peg;1112pub struct ParserSettings {13	pub loc_data: bool,14	pub file_name: PathBuf,15}1617parser! {18	grammar jsonnet_parser() for str {19		use peg::ParseLiteral;2021		/// Standard C-like comments22		rule comment()23			= "//" (!['\n'][_])* "\n"24			/ "/*" ((!("*/")[_][_])/("\\" "*/"))* "*/"25			/ "#" (!['\n'][_])* "\n"2627		rule _() = ([' ' | '\n' | '\t'] / comment())*2829		/// For comma-delimited elements30		rule comma() = quiet!{_ "," _} / expected!("<comma>")31		rule alpha() -> char = c:$(['_' | 'a'..='z' | 'A'..='Z']) {c.chars().next().unwrap()}32		rule digit() -> char = d:$(['0'..='9']) {d.chars().next().unwrap()}33		rule end_of_ident() = !['0'..='9' | '_' | 'a'..='z' | 'A'..='Z']34		/// Sequence of digits35		rule uint() -> u64 = a:$(digit()+) { a.parse().unwrap() }36		/// Number in scientific notation format37		rule number() -> f64 = quiet!{a:$(uint() ("." uint())? (['e'|'E'] (s:['+'|'-'])? uint())?) { a.parse().unwrap() }} / expected!("<number>")3839		/// Reserved word followed by any non-alphanumberic40		rule reserved() = ("assert" / "else" / "error" / "false" / "for" / "function" / "if" / "import" / "importstr" / "in" / "local" / "null" / "tailstrict" / "then" / "self" / "super" / "true") end_of_ident()41		rule id() -> String = quiet!{ !reserved() s:$(alpha() (alpha() / digit())*) {s.to_owned()}} / expected!("<identifier>")4243		rule keyword(id: &'static str)44			= ##parse_string_literal(id) end_of_ident()45		// Adds location data information to existing expression46		rule l(s: &ParserSettings, x: rule<Expr>) -> LocExpr47			= start:position!() v:x() end:position!() {loc_expr!(v, s.loc_data, (s.file_name.clone(), start, end))}4849		pub rule param(s: &ParserSettings) -> expr::Param = name:id() expr:(_ "=" _ expr:expr(s){expr})? { expr::Param(name, expr) }50		pub rule params(s: &ParserSettings) -> expr::ParamsDesc51			= params:(param(s) ** comma()) {52				let mut defaults_started = false;53				for param in &params {54					defaults_started = defaults_started || param.1.is_some();55					assert_eq!(defaults_started, param.1.is_some(), "defauld parameters should be used after all positionals");56				}57				expr::ParamsDesc(params)58			}59			/ { expr::ParamsDesc(Vec::new()) }6061		pub rule arg(s: &ParserSettings) -> expr::Arg62			= name:id() _ "=" _ expr:expr(s) {expr::Arg(Some(name), expr)}63			/ expr:expr(s) {expr::Arg(None, expr)}64		pub rule args(s: &ParserSettings) -> expr::ArgsDesc65			= args:arg(s) ** comma() comma()? {66				let mut named_started = false;67				for arg in &args {68					named_started = named_started || arg.0.is_some();69					assert_eq!(named_started, arg.0.is_some(), "named args should be used after all positionals");70				}71				expr::ArgsDesc(args)72			}73			/ { expr::ArgsDesc(Vec::new()) }7475		pub rule bind(s: &ParserSettings) -> expr::BindSpec76			= name:id() _ "=" _ expr:expr(s) {expr::BindSpec{name, params: None, value: expr}}77			/ name:id() _ "(" _ params:params(s) _ ")" _ "=" _ expr:expr(s) {expr::BindSpec{name, params: Some(params), value: expr}}78		pub rule assertion(s: &ParserSettings) -> expr::AssertStmt79			= keyword("assert") _ cond:expr(s) msg:(_ ":" _ e:expr(s) {e})? { expr::AssertStmt(cond, msg) }80		pub rule string() -> String81			= v:("\"" str:$(("\\\"" / !['"'][_])*) "\"" {str.to_owned()}82			/ "'" str:$((!['\''][_])*) "'" {str.to_owned()}) {v.replace("\\n", "\n")}83		pub rule field_name(s: &ParserSettings) -> expr::FieldName84			= name:id() {expr::FieldName::Fixed(name)}85			/ name:string() {expr::FieldName::Fixed(name)}86			/ "[" _ expr:expr(s) _ "]" {expr::FieldName::Dyn(expr)}87		pub rule visibility() -> expr::Visibility88			= ":::" {expr::Visibility::Unhide}89			/ "::" {expr::Visibility::Hidden}90			/ ":" {expr::Visibility::Normal}91		pub rule field(s: &ParserSettings) -> expr::FieldMember92			= name:field_name(s) _ plus:"+"? _ visibility:visibility() _ value:expr(s) {expr::FieldMember{93				name,94				plus: plus.is_some(),95				params: None,96				visibility,97				value,98			}}99			/ name:field_name(s) _ "(" _ params:params(s) _ ")" _ visibility:visibility() _ value:expr(s) {expr::FieldMember{100				name,101				plus: false,102				params: Some(params),103				visibility,104				value,105			}}106		pub rule obj_local(s: &ParserSettings) -> BindSpec107			= keyword("local") _ bind:bind(s) {bind}108		pub rule member(s: &ParserSettings) -> expr::Member109			= bind:obj_local(s) {expr::Member::BindStmt(bind)}110			/ assertion:assertion(s) {expr::Member::AssertStmt(assertion)}111			/ field:field(s) {expr::Member::Field(field)}112		pub rule objinside(s: &ParserSettings) -> expr::ObjBody113			= pre_locals:(b: obj_local(s) comma() {b})* "[" _ key:expr(s) _ "]" _ ":" _ value:expr(s) post_locals:(comma() b:obj_local(s) {b})* _ forspec:forspec(s) others:(_ rest:compspec(s) {rest})? {114				expr::ObjBody::ObjComp {115					pre_locals,116					key,117					value,118					post_locals,119					rest: [vec![CompSpec::ForSpec(forspec)], others.unwrap_or_default()].concat(),120				}121			}122			/ members:(member(s) ** comma()) comma()? {expr::ObjBody::MemberList(members)}123		pub rule ifspec(s: &ParserSettings) -> IfSpecData124			= keyword("if") _ expr:expr(s) {IfSpecData(expr)}125		pub rule forspec(s: &ParserSettings) -> ForSpecData126			= keyword("for") _ id:id() _ keyword("in") _ cond:expr(s) {ForSpecData(id, cond)}127		pub rule compspec(s: &ParserSettings) -> Vec<expr::CompSpec>128			= s:(i:ifspec(s) { expr::CompSpec::IfSpec(i) } / f:forspec(s) {expr::CompSpec::ForSpec(f)} ) ** _ {s}129		pub rule local_expr(s: &ParserSettings) -> LocExpr130			= l(s,<keyword("local") _ binds:bind(s) ** comma() _ ";" _ expr:expr(s) { Expr::LocalExpr(binds, expr) }>)131		pub rule string_expr(s: &ParserSettings) -> LocExpr132			= l(s, <s:string() {Expr::Str(s)}>)133		pub rule obj_expr(s: &ParserSettings) -> LocExpr134			= l(s,<"{" _ body:objinside(s) _ "}" {Expr::Obj(body)}>)135		pub rule array_expr(s: &ParserSettings) -> LocExpr136			= l(s,<"[" _ elems:(expr(s) ** comma()) _ comma()? "]" {Expr::Arr(elems)}>)137		pub rule array_comp_expr(s: &ParserSettings) -> LocExpr138			= l(s,<"[" _ expr:expr(s) _ comma()? _ forspec:forspec(s) _ others:(others: compspec(s) _ {others})? "]" {Expr::ArrComp(expr, [vec![CompSpec::ForSpec(forspec)], others.unwrap_or_default()].concat())}>)139		pub rule number_expr(s: &ParserSettings) -> LocExpr140			= l(s,<n:number() { expr::Expr::Num(n) }>)141		pub rule var_expr(s: &ParserSettings) -> LocExpr142			= l(s,<n:id() { expr::Expr::Var(n) }>)143		pub rule if_then_else_expr(s: &ParserSettings) -> LocExpr144			= l(s,<cond:ifspec(s) _ keyword("then") _ cond_then:expr(s) cond_else:(_ keyword("else") _ e:expr(s) {e})? {Expr::IfElse{145				cond,146				cond_then,147				cond_else,148			}}>)149150		pub rule literal(s: &ParserSettings) -> LocExpr151			= l(s,<v:(152				keyword("null") {LiteralType::Null}153				/ keyword("true") {LiteralType::True}154				/ keyword("false") {LiteralType::False}155				/ keyword("self") {LiteralType::This}156				/ keyword("$") {LiteralType::Dollar}157				/ keyword("super") {LiteralType::Super}158			) {Expr::Literal(v)}>)159160		pub rule expr_basic(s: &ParserSettings) -> LocExpr161			= literal(s)162163			/ string_expr(s) / number_expr(s)164			/ array_expr(s)165			/ obj_expr(s)166			/ array_expr(s)167			/ array_comp_expr(s)168169			/ var_expr(s)170			/ local_expr(s)171			/ if_then_else_expr(s)172173			/ l(s,<keyword("function") _ "(" _ params:params(s) _ ")" _ expr:expr(s) {Expr::Function(params, expr)}>)174			/ l(s,<assertion:assertion(s) _ ";" _ expr:expr(s) { Expr::AssertExpr(assertion, expr) }>)175176			/ l(s,<keyword("error") _ expr:expr(s) { Expr::Error(expr) }>)177178		pub rule slice_desc(s: &ParserSettings) -> SliceDesc179			= start:expr(s)? _ ":" _ pair:(end:expr(s)? _ step:(":" _ e:expr(s) {e})? {(end, step)})? {180				if let Some((end, step)) = pair {181					SliceDesc { start, end, step }182				}else{183					SliceDesc { start, end: None, step: None }184				}185			}186187		rule expr(s: &ParserSettings) -> LocExpr188			= start:position!() a:precedence! {189				a:(@) _ "||" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Or, b))}190				--191				a:(@) _ "&&" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::And, b))}192				--193				a:(@) _ "|" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::BitOr, b))}194				--195				a:@ _ "^" _ b:(@) {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::BitXor, b))}196				--197				a:(@) _ "&" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::BitAnd, b))}198				--199				a:(@) _ "==" _ b:@ {loc_expr_todo!(Expr::Apply(200					el!(Expr::Index(201						el!(Expr::Var("std".to_owned())),202						el!(Expr::Str("equals".to_owned()))203					)), ArgsDesc(vec![Arg(None, a), Arg(None, b)])204				))}205				a:(@) _ "!=" _ b:@ {loc_expr_todo!(Expr::UnaryOp(UnaryOpType::Not, el!(Expr::Apply(206					el!(Expr::Index(207						el!(Expr::Var("std".to_owned())),208						el!(Expr::Str("equals".to_owned()))209					)), ArgsDesc(vec![Arg(None, a), Arg(None, b)])210				))))}211				--212				a:(@) _ "<" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Lt, b))}213				a:(@) _ ">" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Gt, b))}214				a:(@) _ "<=" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Lte, b))}215				a:(@) _ ">=" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Gte, b))}216				--217				a:(@) _ "<<" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Lhs, b))}218				a:(@) _ ">>" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Rhs, b))}219				--220				a:(@) _ "+" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Add, b))}221				a:(@) _ "-" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Sub, b))}222				--223				a:(@) _ "*" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Mul, b))}224				a:(@) _ "/" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Div, b))}225				a:(@) _ "%" _ b:@ {loc_expr_todo!(Expr::Apply(226					el!(Expr::Index(227						el!(Expr::Var("std".to_owned())),228						el!(Expr::Str("mod".to_owned()))229					)), ArgsDesc(vec![Arg(None, a), Arg(None, b)])230				))}231				--232						"-" _ b:@ {loc_expr_todo!(Expr::UnaryOp(UnaryOpType::Minus, b))}233						"!" _ b:@ {loc_expr_todo!(Expr::UnaryOp(UnaryOpType::Not, b))}234						"~" _ b:@ { loc_expr_todo!(Expr::UnaryOp(UnaryOpType::BitNot, b)) }235				--236				a:(@) _ "[" _ s:slice_desc(s) _ "]" {loc_expr_todo!(Expr::Slice(a, s))}237				a:(@) _ "." _ s:id() {loc_expr_todo!(Expr::Index(a, el!(Expr::Str(s))))}238				a:(@) _ "[" _ s:expr(s) _ "]" {loc_expr_todo!(Expr::Index(a, s))}239				a:(@) _ "(" _ args:args(s) _ ")" (_ keyword("tailstrict"))? {loc_expr_todo!(Expr::Apply(a, args))}240				a:(@) _ "{" _ body:objinside(s) _ "}" {loc_expr_todo!(Expr::ObjExtend(a, body))}241				--242				e:expr_basic(s) {e}243				"(" _ e:expr(s) _ ")" {loc_expr_todo!(Expr::Parened(e))}244			} end:position!() {245				let LocExpr(e, _) = a;246				LocExpr(e, if s.loc_data {247					Some(Rc::new(ExprLocation(s.file_name.to_owned(), start, end)))248				} else {249					None250				})251			}252			/ e:expr_basic(s) {e}253254		pub rule jsonnet(s: &ParserSettings) -> LocExpr = _ e:expr(s) _ {e}255	}256}257258pub type ParseError = peg::error::ParseError<peg::str::LineCol>;259pub fn parse(str: &str, settings: &ParserSettings) -> Result<LocExpr, ParseError> {260	jsonnet_parser::jsonnet(str, settings)261}262263#[macro_export]264macro_rules! el {265	($expr:expr) => {266		LocExpr(std::rc::Rc::new($expr), None)267	};268}269270#[cfg(test)]271pub mod tests {272	use super::{expr::*, parse};273	use crate::ParserSettings;274	use std::path::PathBuf;275276	macro_rules! parse {277		($s:expr) => {278			parse(279				$s,280				&ParserSettings {281					loc_data: false,282					file_name: PathBuf::from("/test.jsonnet"),283					},284				)285			.unwrap()286		};287	}288289	mod expressions {290		use super::*;291292		pub fn basic_math() -> LocExpr {293			el!(Expr::BinaryOp(294				el!(Expr::Num(2.0)),295				BinaryOpType::Add,296				el!(Expr::BinaryOp(297					el!(Expr::Num(2.0)),298					BinaryOpType::Mul,299					el!(Expr::Num(2.0)),300				)),301			))302		}303	}304305	#[test]306	fn empty_object() {307		assert_eq!(parse!("{}"), el!(Expr::Obj(ObjBody::MemberList(vec![]))));308	}309310	#[test]311	fn basic_math() {312		assert_eq!(313			parse!("2+2*2"),314			el!(Expr::BinaryOp(315				el!(Expr::Num(2.0)),316				BinaryOpType::Add,317				el!(Expr::BinaryOp(318					el!(Expr::Num(2.0)),319					BinaryOpType::Mul,320					el!(Expr::Num(2.0))321				))322			))323		);324	}325326	#[test]327	fn basic_math_with_indents() {328		assert_eq!(parse!("2	+ 	  2	  *	2   	"), expressions::basic_math());329	}330331	#[test]332	fn basic_math_parened() {333		assert_eq!(334			parse!("2+(2+2*2)"),335			el!(Expr::BinaryOp(336				el!(Expr::Num(2.0)),337				BinaryOpType::Add,338				el!(Expr::Parened(expressions::basic_math())),339			))340		);341	}342343	/// Comments should not affect parsing344	#[test]345	fn comments() {346		assert_eq!(347			parse!("2//comment\n+//comment\n3/*test*/*/*test*/4"),348			el!(Expr::BinaryOp(349				el!(Expr::Num(2.0)),350				BinaryOpType::Add,351				el!(Expr::BinaryOp(352					el!(Expr::Num(3.0)),353					BinaryOpType::Mul,354					el!(Expr::Num(4.0))355				))356			))357		);358	}359360	/// Comments should be able to be escaped361	#[test]362	fn comment_escaping() {363		assert_eq!(364			parse!("2/*\\*/+*/ - 22"),365			el!(Expr::BinaryOp(366				el!(Expr::Num(2.0)),367				BinaryOpType::Sub,368				el!(Expr::Num(22.0))369			))370		);371	}372373	#[test]374	fn suffix() {375		// assert_eq!(parse!("std.test"), el!(Expr::Num(2.2)));376		// assert_eq!(parse!("std(2)"), el!(Expr::Num(2.2)));377		// assert_eq!(parse!("std.test(2)"), el!(Expr::Num(2.2)));378		// assert_eq!(parse!("a[b]"), el!(Expr::Num(2.2)))379	}380381	#[test]382	fn array_comp() {383		use Expr::*;384		assert_eq!(385			parse!("[std.deepJoin(x) for x in arr]"),386			el!(ArrComp(387				el!(Apply(388					el!(Index(389						el!(Var("std".to_owned())),390						el!(Str("deepJoin".to_owned()))391					)),392					ArgsDesc(vec![Arg(None, el!(Var("x".to_owned())))])393				)),394				vec![CompSpec::ForSpec(ForSpecData(395					"x".to_owned(),396					el!(Var("arr".to_owned()))397				))]398			)),399		)400	}401402	#[test]403	fn reserved() {404		use Expr::*;405		assert_eq!(parse!("null"), el!(Literal(LiteralType::Null)));406		assert_eq!(parse!("nulla"), el!(Var("nulla".to_owned())));407	}408409	#[test]410	fn multiple_args_buf() {411		parse!("a(b, null_fields)");412	}413414	#[test]415	fn infix_precedence() {416		use Expr::*;417		assert_eq!(418			parse!("!a && !b"),419			el!(BinaryOp(420				el!(UnaryOp(UnaryOpType::Not, el!(Var("a".to_owned())))),421				BinaryOpType::And,422				el!(UnaryOp(UnaryOpType::Not, el!(Var("b".to_owned()))))423			))424		);425	}426427	#[test]428	fn infix_precedence_division() {429		use Expr::*;430		assert_eq!(431			parse!("!a / !b"),432			el!(BinaryOp(433				el!(UnaryOp(UnaryOpType::Not, el!(Var("a".to_owned())))),434				BinaryOpType::Div,435				el!(UnaryOp(UnaryOpType::Not, el!(Var("b".to_owned()))))436			))437		);438	}439440	#[test]441	fn double_negation() {442		use Expr::*;443		assert_eq!(444			parse!("!!a"),445			el!(UnaryOp(446				UnaryOpType::Not,447				el!(UnaryOp(UnaryOpType::Not, el!(Var("a".to_owned()))))448			))449		)450	}451452	#[test]453	fn array_test_error() {454		parse!("[a for a in b if c for e in f]");455		//                    ^^^^ failed code456	}457458	#[test]459	fn can_parse_stdlib() {460		parse!(jsonnet_stdlib::STDLIB_STR);461	}462463	use test::Bencher;464465	// From source code466	#[bench]467	fn bench_parse_peg(b: &mut Bencher) {468		b.iter(|| parse!(jsonnet_stdlib::STDLIB_STR))469	}470471	// From serialized blob472	#[bench]473	fn bench_parse_serde_bincode(b: &mut Bencher) {474		let serialized = bincode::serialize(&parse!(jsonnet_stdlib::STDLIB_STR)).unwrap();475		b.iter(|| bincode::deserialize::<LocExpr>(&serialized))476	}477}