1#![feature(box_syntax)]2#![feature(test)]34extern crate test;56use peg::parser;7use std::{path::PathBuf, rc::Rc};8mod expr;9mod string_processing;10pub use expr::*;11pub use peg;12use string_processing::deent;1314pub struct ParserSettings {15 pub loc_data: bool,16 pub file_name: PathBuf,17}1819parser! {20 grammar jsonnet_parser() for str {21 use peg::ParseLiteral;2223 24 rule comment()25 = "//" (!['\n'][_])* "\n"26 / "/*" ("\\*/" / "\\\\" / (!("*/")[_]))* "*/"27 / "#" (!['\n'][_])* "\n"2829 rule single_whitespace() = quiet!{([' ' | '\r' | '\n' | '\t'] / comment())} / expected!("<whitespace>")30 rule _() = single_whitespace()*3132 33 rule comma() = quiet!{_ "," _} / expected!("<comma>")34 rule alpha() -> char = c:$(['_' | 'a'..='z' | 'A'..='Z']) {c.chars().next().unwrap()}35 rule digit() -> char = d:$(['0'..='9']) {d.chars().next().unwrap()}36 rule end_of_ident() = !['0'..='9' | '_' | 'a'..='z' | 'A'..='Z']37 38 rule uint() -> u64 = a:$(digit()+) { a.parse().unwrap() }39 40 rule number() -> f64 = quiet!{a:$(uint() ("." uint())? (['e'|'E'] (s:['+'|'-'])? uint())?) { a.parse().unwrap() }} / expected!("<number>")4142 43 rule reserved() = ("assert" / "else" / "error" / "false" / "for" / "function" / "if" / "import" / "importstr" / "in" / "local" / "null" / "tailstrict" / "then" / "self" / "super" / "true") end_of_ident()44 rule id() -> String = quiet!{ !reserved() s:$(alpha() (alpha() / digit())*) {s.to_owned()}} / expected!("<identifier>")4546 rule keyword(id: &'static str)47 = ##parse_string_literal(id) end_of_ident()48 49 rule l(s: &ParserSettings, x: rule<Expr>) -> LocExpr50 = start:position!() v:x() end:position!() {loc_expr!(v, s.loc_data, (s.file_name.clone(), start, end))}5152 pub rule param(s: &ParserSettings) -> expr::Param = name:id() expr:(_ "=" _ expr:expr(s){expr})? { expr::Param(name, expr) }53 pub rule params(s: &ParserSettings) -> expr::ParamsDesc54 = params:(param(s) ** comma()) {55 let mut defaults_started = false;56 for param in ¶ms {57 defaults_started = defaults_started || param.1.is_some();58 assert_eq!(defaults_started, param.1.is_some(), "defauld parameters should be used after all positionals");59 }60 expr::ParamsDesc(params)61 }62 / { expr::ParamsDesc(Vec::new()) }6364 pub rule arg(s: &ParserSettings) -> expr::Arg65 = name:id() _ "=" _ expr:expr(s) {expr::Arg(Some(name), expr)}66 / expr:expr(s) {expr::Arg(None, expr)}67 pub rule args(s: &ParserSettings) -> expr::ArgsDesc68 = args:arg(s) ** comma() comma()? {69 let mut named_started = false;70 for arg in &args {71 named_started = named_started || arg.0.is_some();72 assert_eq!(named_started, arg.0.is_some(), "named args should be used after all positionals");73 }74 expr::ArgsDesc(args)75 }76 / { expr::ArgsDesc(Vec::new()) }7778 pub rule bind(s: &ParserSettings) -> expr::BindSpec79 = name:id() _ "=" _ expr:expr(s) {expr::BindSpec{name, params: None, value: expr}}80 / name:id() _ "(" _ params:params(s) _ ")" _ "=" _ expr:expr(s) {expr::BindSpec{name, params: Some(params), value: expr}}81 pub rule assertion(s: &ParserSettings) -> expr::AssertStmt82 = keyword("assert") _ cond:expr(s) msg:(_ ":" _ e:expr(s) {e})? { expr::AssertStmt(cond, msg) }8384 pub rule whole_line() -> String85 = str:$((!['\n'][_])* "\n") {str.to_owned()}86 pub rule string() -> String87 = "\"" str:$(("\\\"" / "\\\\" / (!['"'][_]))*) "\"" {unescape::unescape(str).unwrap()}88 / "'" str:$(("\\'" / "\\\\" / (!['\''][_]))*) "'" {unescape::unescape(str).unwrap()}89 / "@'" str:$(("''" / (!['\''][_]))*) "'" {str.replace("''", "'")}90 / "@\"" str:$(("\"\"" / (!['"'][_]))*) "\"" {str.replace("\"\"", "\"")}91 92 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<1, 1> whole_line())+) " "*<0, 0> "|||" {deent(str)}93 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<2, 2> whole_line())+) " "*<1, 1> "|||" {deent(str)}94 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<3, 3> whole_line())+) " "*<2, 2> "|||" {deent(str)}95 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<4, 4> whole_line())+) " "*<3, 3> "|||" {deent(str)}96 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<5, 5> whole_line())+) " "*<4, 4> "|||" {deent(str)}97 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<6, 6> whole_line())+) " "*<5, 5> "|||" {deent(str)}98 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<7, 7> whole_line())+) " "*<6, 6> "|||" {deent(str)}99 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<8, 8> whole_line())+) " "*<7, 7> "|||" {deent(str)}100 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<9, 9> whole_line())+) " "*<8, 8> "|||" {deent(str)}101 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<10, 10> whole_line())+) " "*<9, 9> "|||" {deent(str)}102 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<11, 11> whole_line())+) " "*<10, 10> "|||" {deent(str)}103 / "|||" (!['\n']single_whitespace())* "\n" str:$((" "*<12, 12> whole_line())+) " "*<11, 10> "|||" {deent(str)}104105 pub rule field_name(s: &ParserSettings) -> expr::FieldName106 = name:id() {expr::FieldName::Fixed(name)}107 / name:string() {expr::FieldName::Fixed(name)}108 / "[" _ expr:expr(s) _ "]" {expr::FieldName::Dyn(expr)}109 pub rule visibility() -> expr::Visibility110 = ":::" {expr::Visibility::Unhide}111 / "::" {expr::Visibility::Hidden}112 / ":" {expr::Visibility::Normal}113 pub rule field(s: &ParserSettings) -> expr::FieldMember114 = name:field_name(s) _ plus:"+"? _ visibility:visibility() _ value:expr(s) {expr::FieldMember{115 name,116 plus: plus.is_some(),117 params: None,118 visibility,119 value,120 }}121 / name:field_name(s) _ "(" _ params:params(s) _ ")" _ visibility:visibility() _ value:expr(s) {expr::FieldMember{122 name,123 plus: false,124 params: Some(params),125 visibility,126 value,127 }}128 pub rule obj_local(s: &ParserSettings) -> BindSpec129 = keyword("local") _ bind:bind(s) {bind}130 pub rule member(s: &ParserSettings) -> expr::Member131 = bind:obj_local(s) {expr::Member::BindStmt(bind)}132 / assertion:assertion(s) {expr::Member::AssertStmt(assertion)}133 / field:field(s) {expr::Member::Field(field)}134 pub rule objinside(s: &ParserSettings) -> expr::ObjBody135 = 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})? {136 expr::ObjBody::ObjComp {137 pre_locals,138 key,139 value,140 post_locals,141 compspecs: [vec![CompSpec::ForSpec(forspec)], others.unwrap_or_default()].concat(),142 }143 }144 / members:(member(s) ** comma()) comma()? {expr::ObjBody::MemberList(members)}145 pub rule ifspec(s: &ParserSettings) -> IfSpecData146 = keyword("if") _ expr:expr(s) {IfSpecData(expr)}147 pub rule forspec(s: &ParserSettings) -> ForSpecData148 = keyword("for") _ id:id() _ keyword("in") _ cond:expr(s) {ForSpecData(id, cond)}149 pub rule compspec(s: &ParserSettings) -> Vec<expr::CompSpec>150 = s:(i:ifspec(s) { expr::CompSpec::IfSpec(i) } / f:forspec(s) {expr::CompSpec::ForSpec(f)} ) ** _ {s}151 pub rule local_expr(s: &ParserSettings) -> LocExpr152 = l(s,<keyword("local") _ binds:bind(s) ** comma() _ ";" _ expr:expr(s) { Expr::LocalExpr(binds, expr) }>)153 pub rule string_expr(s: &ParserSettings) -> LocExpr154 = l(s, <s:string() {Expr::Str(s)}>)155 pub rule obj_expr(s: &ParserSettings) -> LocExpr156 = l(s,<"{" _ body:objinside(s) _ "}" {Expr::Obj(body)}>)157 pub rule array_expr(s: &ParserSettings) -> LocExpr158 = l(s,<"[" _ elems:(expr(s) ** comma()) _ comma()? "]" {Expr::Arr(elems)}>)159 pub rule array_comp_expr(s: &ParserSettings) -> LocExpr160 = 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())}>)161 pub rule number_expr(s: &ParserSettings) -> LocExpr162 = l(s,<n:number() { expr::Expr::Num(n) }>)163 pub rule var_expr(s: &ParserSettings) -> LocExpr164 = l(s,<n:id() { expr::Expr::Var(n) }>)165 pub rule if_then_else_expr(s: &ParserSettings) -> LocExpr166 = l(s,<cond:ifspec(s) _ keyword("then") _ cond_then:expr(s) cond_else:(_ keyword("else") _ e:expr(s) {e})? {Expr::IfElse{167 cond,168 cond_then,169 cond_else,170 }}>)171172 pub rule literal(s: &ParserSettings) -> LocExpr173 = l(s,<v:(174 keyword("null") {LiteralType::Null}175 / keyword("true") {LiteralType::True}176 / keyword("false") {LiteralType::False}177 / keyword("self") {LiteralType::This}178 / keyword("$") {LiteralType::Dollar}179 / keyword("super") {LiteralType::Super}180 ) {Expr::Literal(v)}>)181182 pub rule expr_basic(s: &ParserSettings) -> LocExpr183 = literal(s)184185 / string_expr(s) / number_expr(s)186 / array_expr(s)187 / obj_expr(s)188 / array_expr(s)189 / array_comp_expr(s)190191 / l(s,<keyword("importstr") _ path:string() {Expr::ImportStr(PathBuf::from(path))}>)192 / l(s,<keyword("import") _ path:string() {Expr::Import(PathBuf::from(path))}>)193194 / var_expr(s)195 / local_expr(s)196 / if_then_else_expr(s)197198 / l(s,<keyword("function") _ "(" _ params:params(s) _ ")" _ expr:expr(s) {Expr::Function(params, expr)}>)199 / l(s,<assertion:assertion(s) _ ";" _ expr:expr(s) { Expr::AssertExpr(assertion, expr) }>)200201 / l(s,<keyword("error") _ expr:expr(s) { Expr::Error(expr) }>)202203 pub rule slice_desc(s: &ParserSettings) -> SliceDesc204 = start:expr(s)? _ ":" _ pair:(end:expr(s)? _ step:(":" _ e:expr(s) {e})? {(end, step)})? {205 if let Some((end, step)) = pair {206 SliceDesc { start, end, step }207 }else{208 SliceDesc { start, end: None, step: None }209 }210 }211212 rule expr(s: &ParserSettings) -> LocExpr213 = start:position!() a:precedence! {214 a:(@) _ "||" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Or, b))}215 --216 a:(@) _ "&&" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::And, b))}217 --218 a:(@) _ "|" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::BitOr, b))}219 --220 a:@ _ "^" _ b:(@) {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::BitXor, b))}221 --222 a:(@) _ "&" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::BitAnd, b))}223 --224 a:(@) _ "==" _ b:@ {loc_expr_todo!(Expr::Apply(225 el!(Expr::Index(226 el!(Expr::Var("std".to_owned())),227 el!(Expr::Str("equals".to_owned()))228 )),229 ArgsDesc(vec![Arg(None, a), Arg(None, b)]),230 true231 ))}232 a:(@) _ "!=" _ b:@ {loc_expr_todo!(Expr::UnaryOp(UnaryOpType::Not, el!(Expr::Apply(233 el!(Expr::Index(234 el!(Expr::Var("std".to_owned())),235 el!(Expr::Str("equals".to_owned()))236 )),237 ArgsDesc(vec![Arg(None, a), Arg(None, b)]),238 true239 ))))}240 --241 a:(@) _ "<" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Lt, b))}242 a:(@) _ ">" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Gt, b))}243 a:(@) _ "<=" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Lte, b))}244 a:(@) _ ">=" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Gte, b))}245 a:(@) _ keyword("in") _ b:@ {loc_expr_todo!(Expr::Apply(246 el!(Expr::Index(247 el!(Expr::Var("std".to_owned())),248 el!(Expr::Str("objectHasEx".to_owned()))249 )), ArgsDesc(vec![Arg(None, b), Arg(None, a), Arg(None, el!(Expr::Literal(LiteralType::True)))]),250 true251 ))}252 --253 a:(@) _ "<<" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Lhs, b))}254 a:(@) _ ">>" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Rhs, b))}255 --256 a:(@) _ "+" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Add, b))}257 a:(@) _ "-" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Sub, b))}258 --259 a:(@) _ "*" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Mul, b))}260 a:(@) _ "/" _ b:@ {loc_expr_todo!(Expr::BinaryOp(a, BinaryOpType::Div, b))}261 a:(@) _ "%" _ b:@ {loc_expr_todo!(Expr::Apply(262 el!(Expr::Index(263 el!(Expr::Var("std".to_owned())),264 el!(Expr::Str("mod".to_owned()))265 )), ArgsDesc(vec![Arg(None, a), Arg(None, b)]),266 true267 ))}268 --269 "-" _ b:@ {loc_expr_todo!(Expr::UnaryOp(UnaryOpType::Minus, b))}270 "!" _ b:@ {loc_expr_todo!(Expr::UnaryOp(UnaryOpType::Not, b))}271 "~" _ b:@ { loc_expr_todo!(Expr::UnaryOp(UnaryOpType::BitNot, b)) }272 --273 a:(@) _ "[" _ s:slice_desc(s) _ "]" {loc_expr_todo!(Expr::Slice(a, s))}274 a:(@) _ "." _ s:id() {loc_expr_todo!(Expr::Index(a, el!(Expr::Str(s))))}275 a:(@) _ "[" _ s:expr(s) _ "]" {loc_expr_todo!(Expr::Index(a, s))}276 a:(@) _ "(" _ args:args(s) _ ")" ts:(_ keyword("tailstrict"))? {loc_expr_todo!(Expr::Apply(a, args, ts.is_some()))}277 a:(@) _ "{" _ body:objinside(s) _ "}" {loc_expr_todo!(Expr::ObjExtend(a, body))}278 --279 e:expr_basic(s) {e}280 "(" _ e:expr(s) _ ")" {loc_expr_todo!(Expr::Parened(e))}281 } end:position!() {282 let LocExpr(e, _) = a;283 LocExpr(e, if s.loc_data {284 Some(Rc::new(ExprLocation(s.file_name.to_owned(), start, end)))285 } else {286 None287 })288 }289 / e:expr_basic(s) {e}290291 pub rule jsonnet(s: &ParserSettings) -> LocExpr = _ e:expr(s) _ {e}292 }293}294295pub type ParseError = peg::error::ParseError<peg::str::LineCol>;296pub fn parse(str: &str, settings: &ParserSettings) -> Result<LocExpr, ParseError> {297 jsonnet_parser::jsonnet(str, settings)298}299300#[macro_export]301macro_rules! el {302 ($expr:expr) => {303 LocExpr(std::rc::Rc::new($expr), None)304 };305}306307#[cfg(test)]308pub mod tests {309 use super::{expr::*, parse};310 use crate::ParserSettings;311 use std::path::PathBuf;312313 macro_rules! parse {314 ($s:expr) => {315 parse(316 $s,317 &ParserSettings {318 loc_data: false,319 file_name: PathBuf::from("/test.jsonnet"),320 },321 )322 .unwrap()323 };324 }325326 mod expressions {327 use super::*;328329 pub fn basic_math() -> LocExpr {330 el!(Expr::BinaryOp(331 el!(Expr::Num(2.0)),332 BinaryOpType::Add,333 el!(Expr::BinaryOp(334 el!(Expr::Num(2.0)),335 BinaryOpType::Mul,336 el!(Expr::Num(2.0)),337 )),338 ))339 }340 }341342 #[test]343 fn multiline_string() {344 assert_eq!(345 parse!("|||\n Hello world!\n a\n|||"),346 el!(Expr::Str(" Hello world!\na\n".to_owned())),347 )348 }349350 #[test]351 fn string_escaping() {352 assert_eq!(353 parse!(r#""Hello, \"world\"!""#),354 el!(Expr::Str(r#"Hello, "world"!"#.to_owned())),355 );356 assert_eq!(357 parse!(r#"'Hello \'world\'!'"#),358 el!(Expr::Str("Hello 'world'!".to_owned())),359 );360 assert_eq!(parse!(r#"'\\\\'"#), el!(Expr::Str("\\\\".to_owned())),);361 }362363 #[test]364 fn string_unescaping() {365 assert_eq!(366 parse!(r#""Hello\nWorld""#),367 el!(Expr::Str("Hello\nWorld".to_owned())),368 );369 }370371 #[test]372 fn string_verbantim() {373 assert_eq!(374 parse!(r#"@"Hello\n""World""""#),375 el!(Expr::Str("Hello\\n\"World\"".to_owned())),376 );377 }378379 #[test]380 fn imports() {381 assert_eq!(382 parse!("import \"hello\""),383 el!(Expr::Import(PathBuf::from("hello"))),384 );385 assert_eq!(386 parse!("importstr \"garnish.txt\""),387 el!(Expr::ImportStr(PathBuf::from("garnish.txt")))388 );389 }390391 #[test]392 fn empty_object() {393 assert_eq!(parse!("{}"), el!(Expr::Obj(ObjBody::MemberList(vec![]))));394 }395396 #[test]397 fn basic_math() {398 assert_eq!(399 parse!("2+2*2"),400 el!(Expr::BinaryOp(401 el!(Expr::Num(2.0)),402 BinaryOpType::Add,403 el!(Expr::BinaryOp(404 el!(Expr::Num(2.0)),405 BinaryOpType::Mul,406 el!(Expr::Num(2.0))407 ))408 ))409 );410 }411412 #[test]413 fn basic_math_with_indents() {414 assert_eq!(parse!("2 + 2 * 2 "), expressions::basic_math());415 }416417 #[test]418 fn basic_math_parened() {419 assert_eq!(420 parse!("2+(2+2*2)"),421 el!(Expr::BinaryOp(422 el!(Expr::Num(2.0)),423 BinaryOpType::Add,424 el!(Expr::Parened(expressions::basic_math())),425 ))426 );427 }428429 430 #[test]431 fn comments() {432 assert_eq!(433 parse!("2//comment\n+//comment\n3/*test*/*/*test*/4"),434 el!(Expr::BinaryOp(435 el!(Expr::Num(2.0)),436 BinaryOpType::Add,437 el!(Expr::BinaryOp(438 el!(Expr::Num(3.0)),439 BinaryOpType::Mul,440 el!(Expr::Num(4.0))441 ))442 ))443 );444 }445446 447 #[test]448 fn comment_escaping() {449 assert_eq!(450 parse!("2/*\\*/+*/ - 22"),451 el!(Expr::BinaryOp(452 el!(Expr::Num(2.0)),453 BinaryOpType::Sub,454 el!(Expr::Num(22.0))455 ))456 );457 }458459 #[test]460 fn suffix() {461 462 463 464 465 }466467 #[test]468 fn array_comp() {469 use Expr::*;470 assert_eq!(471 parse!("[std.deepJoin(x) for x in arr]"),472 el!(ArrComp(473 el!(Apply(474 el!(Index(475 el!(Var("std".to_owned())),476 el!(Str("deepJoin".to_owned()))477 )),478 ArgsDesc(vec![Arg(None, el!(Var("x".to_owned())))]),479 false,480 )),481 vec![CompSpec::ForSpec(ForSpecData(482 "x".to_owned(),483 el!(Var("arr".to_owned()))484 ))]485 )),486 )487 }488489 #[test]490 fn reserved() {491 use Expr::*;492 assert_eq!(parse!("null"), el!(Literal(LiteralType::Null)));493 assert_eq!(parse!("nulla"), el!(Var("nulla".to_owned())));494 }495496 #[test]497 fn multiple_args_buf() {498 parse!("a(b, null_fields)");499 }500501 #[test]502 fn infix_precedence() {503 use Expr::*;504 assert_eq!(505 parse!("!a && !b"),506 el!(BinaryOp(507 el!(UnaryOp(UnaryOpType::Not, el!(Var("a".to_owned())))),508 BinaryOpType::And,509 el!(UnaryOp(UnaryOpType::Not, el!(Var("b".to_owned()))))510 ))511 );512 }513514 #[test]515 fn infix_precedence_division() {516 use Expr::*;517 assert_eq!(518 parse!("!a / !b"),519 el!(BinaryOp(520 el!(UnaryOp(UnaryOpType::Not, el!(Var("a".to_owned())))),521 BinaryOpType::Div,522 el!(UnaryOp(UnaryOpType::Not, el!(Var("b".to_owned()))))523 ))524 );525 }526527 #[test]528 fn double_negation() {529 use Expr::*;530 assert_eq!(531 parse!("!!a"),532 el!(UnaryOp(533 UnaryOpType::Not,534 el!(UnaryOp(UnaryOpType::Not, el!(Var("a".to_owned()))))535 ))536 )537 }538539 #[test]540 fn array_test_error() {541 parse!("[a for a in b if c for e in f]");542 543 }544545 #[test]546 fn can_parse_stdlib() {547 parse!(jsonnet_stdlib::STDLIB_STR);548 }549550 use test::Bencher;551552 553 #[bench]554 fn bench_parse_peg(b: &mut Bencher) {555 b.iter(|| parse!(jsonnet_stdlib::STDLIB_STR))556 }557558 559 #[bench]560 fn bench_parse_serde_bincode(b: &mut Bencher) {561 let serialized = bincode::serialize(&parse!(jsonnet_stdlib::STDLIB_STR)).unwrap();562 b.iter(|| bincode::deserialize::<LocExpr>(&serialized))563 }564}