fixing parser marker bombs for some js libs (#1817)

* fixing parser marker bombs

crates/rslint_parser/src/syntax/typescript.rs

@@ -41,6 +41,13 @@ macro_rules! no_recover {
 			return None;
 		}
 	};
+	($p:expr, $m:expr, $t:expr, $res:expr) => {
+		if $res.is_none() && $p.state.no_recovery {
+			$m.abandon($p);
+			$p.rewind($t);
+			return None;
+		}
+	};
 }
 
 pub(crate) fn abstract_readonly_modifiers(
@@ -245,16 +252,26 @@ pub(crate) fn ts_decl(p: &mut Parser) -> Option<CompletedMarker> {
 }
 
 pub fn ts_type_alias_decl(p: &mut Parser) -> Option<CompletedMarker> {
+	let t = p.checkpoint();
+
 	let m = p.start();
 	let start = p.cur_tok().range.start;
 	p.bump_any();
-	no_recover!(p, identifier_name(p));
+	no_recover!(p, m, t, identifier_name(p));
 	if p.at(T![<]) {
 		no_recover!(p, ts_type_params(p));
 	}
+
 	let end = p.cur_tok().range.end;
-	p.expect_no_recover(T![=])?;
-	no_recover!(p, ts_type(p));
+
+	if p.expect_no_recover(T![=]).is_none() {
+		m.abandon(p);
+		p.rewind(t);
+		return None;
+	}
+
+	no_recover!(p, m, t, ts_type(p));
+
 	semi(p, start..end);
 	Some(m.complete(p, TS_TYPE_ALIAS_DECL))
 }
@@ -860,6 +877,8 @@ pub fn ts_type_operator_or_higher(p: &mut Parser) -> Option<CompletedMarker> {
 }
 
 pub fn ts_array_type_or_higher(p: &mut Parser) -> Option<CompletedMarker> {
+	let t = p.checkpoint();
+
 	let mut ty = ts_non_array_type(p);
 
 	while !p.has_linebreak_before_n(0) && p.at(T!['[']) {
@@ -868,8 +887,12 @@ pub fn ts_array_type_or_higher(p: &mut Parser) -> Option<CompletedMarker> {
 		if p.eat(T![']']) {
 			ty = Some(m.complete(p, TS_ARRAY));
 		} else {
-			no_recover!(p, ts_type(p));
-			p.expect_no_recover(T![']'])?;
+			no_recover!(p, m, t, ts_type(p));
+			if p.expect_no_recover(T![']']).is_none() {
+				m.abandon(p);
+				p.rewind(t);
+				return None;
+			}
 			ty = Some(m.complete(p, TS_INDEXED_ARRAY));
 		}
 	}
@@ -985,14 +1008,17 @@ pub fn ts_non_array_type(p: &mut Parser) -> Option<CompletedMarker> {
 			Some(m.complete(p, TS_TEMPLATE))
 		}
 		T![-] => {
+			let t = p.checkpoint();
 			let m = p.start();
 			p.bump_any();
 			if p.at(JS_NUMBER_LITERAL) {
 				let _m = p.start();
 				p.bump_any();
 				_m.complete(p, JS_NUMBER_LITERAL_EXPRESSION);
-			} else {
-				p.expect_no_recover(JS_NUMBER_LITERAL)?;
+			} else if p.expect_no_recover(JS_NUMBER_LITERAL).is_none() {
+				m.abandon(p);
+				p.rewind(t);
+				return None;
 			}
 			Some(m.complete(p, TS_LITERAL))
 		}
@@ -1025,10 +1051,15 @@ pub fn ts_non_array_type(p: &mut Parser) -> Option<CompletedMarker> {
 		}
 		T!['['] => ts_tuple(p),
 		T!['('] => {
+			let t = p.checkpoint();
 			let m = p.start();
 			p.bump_any();
-			no_recover!(p, ts_type(p));
-			p.expect_no_recover(T![')'])?;
+			no_recover!(p, m, t, ts_type(p));
+			if p.expect_no_recover(T![')']).is_none() {
+				m.abandon(p);
+				p.rewind(t);
+				return None;
+			};
 			Some(m.complete(p, TS_PAREN))
 		}
 		_ => {
@@ -1345,6 +1376,8 @@ pub fn ts_type_ref(
 	p: &mut Parser,
 	recovery_set: impl Into<Option<TokenSet>> + Clone,
 ) -> Option<CompletedMarker> {
+	let t = p.checkpoint();
+
 	let m = p.start();
 	if let Some(err_m) = maybe_eat_incorrect_modifier(p) {
 		let err = p
@@ -1354,9 +1387,14 @@ pub fn ts_type_ref(
 		p.error(err);
 	}
 
-	ts_entity_name(p, recovery_set, true)?;
+	if ts_entity_name(p, recovery_set, true).is_none() {
+		m.abandon(p);
+		p.rewind(t);
+		return None;
+	}
+
 	if !p.has_linebreak_before_n(0) && p.at(T![<]) {
-		no_recover!(p, ts_type_args(p));
+		no_recover!(p, m, t, ts_type_args(p));
 	}
 
 	Some(m.complete(p, TS_TYPE_REF))

xtask/Cargo.toml

@@ -29,4 +29,6 @@ yastl = "0.1"
 num_cpus = "1.13"
 ungrammar = "1.14.9"
 ureq = { version = "2.3.1" } 
-url = "2.2.2"
+url = "2.2.2"
+timing = "0.2.3"
+ansi_rgb = "0.2.0"

xtask/src/libs/libs.txt

@@ -1,2 +1,10 @@
 https://cdn.jsdelivr.net/npm/jquery@3.2.1/dist/jquery.min.js
-https://cdn.jsdelivr.net/npm/vue@3.2.22/dist/vue.global.prod.js
+https://cdn.jsdelivr.net/npm/vue@3.2.22/dist/vue.global.prod.js
+https://cdn.jsdelivr.net/npm/react@17.0.2/cjs/react.production.min.js
+https://cdn.jsdelivr.net/npm/react-dom@17.0.2/cjs/react-dom.production.min.js
+https://cdn.jsdelivr.net/npm/svelte@3.44.2/compiler.js
+https://cdn.jsdelivr.net/npm/dojo@1.16.4/dojo.js
+https://cdn.jsdelivr.net/npm/d3@7.1.1/dist/d3.min.js
+https://cdn.jsdelivr.net/npm/pixi.js@6.2.0/dist/browser/pixi.min.js
+https://cdn.jsdelivr.net/npm/three@0.134.0/build/three.min.js
+https://cdn.jsdelivr.net/npm/mathjax@3.2.0/es5/tex-chtml-full.js

xtask/src/libs/mod.rs

@@ -1,3 +1,4 @@
+use ansi_rgb::{red, Foreground};
 use std::{path::PathBuf, str::FromStr};
 
 fn err_to_string<E: std::fmt::Debug>(e: E) -> String {
@@ -18,7 +19,7 @@ pub fn get_code(lib: &str) -> Result<String, String> {
 
 	match std::fs::read_to_string(&file) {
 		Ok(code) => {
-			println!("[{}] - using [{}]", filename, file.display());
+			println!("[{}] - using [{}]", filename.fg(red()), file.display());
 			Ok(code)
 		}
 		Err(_) => {
@@ -44,13 +45,49 @@ pub fn get_code(lib: &str) -> Result<String, String> {
 	}
 }
 
-pub fn run() {
+pub fn run(filter: String) {
+	let regex = regex::Regex::new(filter.as_str()).unwrap();
 	let libs = include_str!("libs.txt").lines();
 	for lib in libs {
+		if !regex.is_match(lib) {
+			continue;
+		}
+
 		let code = get_code(lib);
 		match code {
 			Ok(code) => {
-				let _ = std::panic::catch_unwind(|| rslint_parser::parse_module(code.as_str(), 0));
+				let t = timing::start();
+				let _ = std::panic::catch_unwind(|| {
+					let text = code;
+
+					let tokenizer_timing = timing::start();
+					let (tokens, mut errors) = rslint_parser::tokenize(text.as_str(), 0);
+					let tok_source = rslint_parser::TokenSource::new(text.as_str(), &tokens);
+					println!("\ttokenizer took {:?}", tokenizer_timing.stop());
+
+					let parser_timing = timing::start();
+					let (events, errors, tokens) = {
+						let mut parser = rslint_parser::Parser::new(
+							tok_source,
+							0,
+							rslint_parser::Syntax::default().module(),
+						);
+						rslint_parser::syntax::program::parse(&mut parser);
+						let (events, p_errs) = parser.finish();
+						errors.extend(p_errs);
+						(events, errors, tokens)
+					};
+					println!("\tparser took {:?}", parser_timing.stop());
+
+					let treesink_timing = timing::start();
+					let mut tree_sink =
+						rslint_parser::LosslessTreeSink::new(text.as_str(), &tokens);
+					rslint_parser::process(&mut tree_sink, events, errors);
+					let (_green, _parse_errors) = tree_sink.finish();
+					println!("\ttree sink took {:?}", treesink_timing.stop());
+				});
+				let dur = t.stop();
+				println!("total: {:?}", dur);
 			}
 			Err(e) => println!("{:?}", e),
 		}

xtask/src/main.rs

@@ -51,7 +51,11 @@ fn main() -> Result<()> {
 			Ok(())
 		}
 		"coverage-libs" => {
-			xtask::libs::run();
+			let filter: String = args
+				.opt_value_from_str("--filter")
+				.unwrap()
+				.unwrap_or_else(|| ".*".to_string());
+			xtask::libs::run(filter);
 			Ok(())
 		}
 		_ => {