@teasel/parser 0.0.2 → 0.0.4

package/api.js

@@ -51,8 +51,9 @@ function stops(list) {
  * @property {() => ArrayLike<number>} shapes
  */
 
-// words the engine has to judge: keywords, the strict-mode reserved words and the contextual ones
-const KEYWORD = new Set('await break case catch class const continue debugger default delete do else enum export extends false finally for function if implements import in instanceof interface let new null package private protected public return static super switch this throw true try typeof var void while with yield'.split(' '));
+// words the engine has to judge: keywords, the strict-mode reserved words, the contextual ones,
+// and the two names strict mode refuses to bind
+const KEYWORD = new Set('arguments await break case catch class const continue debugger default delete do else enum eval export extends false finally for function if implements import in instanceof interface let new null package private protected public return static super switch this throw true try typeof var void while with yield'.split(' '));
 
 /**
  * The offset after an identifier the host's syntax follows directly, so the answer needs no
@@ -63,10 +64,12 @@ const KEYWORD = new Set('await break case catch class const continue debugger de
  * @returns {[number, number] | null} the identifier's end and where the parse ends
  */
 function bare(source, at, end, stopAt, typescript) {
+	if (at >= end) return null;
 	let i = at;
 	const first = source.codePointAt(i);
 	if (first === undefined || !isIdentifierStart(first) || first === 0x5c) return null;
 	i += first > 0xffff ? 2 : 1;
+	if (i > end) return null;
 	while (i < end) {
 		const code = /** @type {number} */ (source.codePointAt(i));
 		if (code === 0x5c) return null;
@@ -85,7 +88,7 @@ function bare(source, at, end, stopAt, typescript) {
 			const after = source.codePointAt(i + stop.length);
 			if (isIdentifierStart(/** @type {number} */ (stop.codePointAt(0))) && after !== undefined && isIdentifierChar(after)) continue;
 			// a stop TypeScript reads as its own is the engine's to judge
-			if (typescript && (stop === 'as' || stop === 'satisfies')) return null;
+			if (typescript && (stop === 'as' || stop === 'satisfies' || stop === ':')) return null;
 			return [name_end, name_end];
 		}
 	}
@@ -112,7 +115,8 @@ export function bind(engine) {
 		constructor(source, options) {
 			this.#held = engine.create(source, names(options), options?.host ?? '');
 			this.#source = source;
-			this.#options = options ?? {};
+			// what the engine was prepared with, however the caller's object changes after
+			this.#options = { ...options };
 			registry?.register(this, this.#held, this);
 		}
 
@@ -123,14 +127,14 @@ export function bind(engine) {
 		 */
 		parse(entry = 'program', offset = 0, { end, stopAt } = {}) {
 			if (this.#held === undefined) throw new TypeError('the source is freed');
-			const index = ENTRY[entry];
+			const index = Object.hasOwn(ENTRY, entry) ? ENTRY[entry] : undefined;
 			if (index === undefined) throw new TypeError(`${JSON.stringify(entry)} is not an entry`);
 			const stop = stops(stopAt);
 			if (this.#options.host !== undefined && index === ENTRY.program) return result(engine.parse(this.#held, index, 0, undefined, ''), this.#source);
 			if ((index === ENTRY.expression || index === ENTRY.pattern) && Number.isInteger(offset) && offset >= 0) {
 				const cut = end === undefined ? this.#source.length : end;
 				const found = Number.isInteger(cut) && cut <= this.#source.length && offset <= cut ? bare(this.#source, offset, cut, index === ENTRY.pattern ? [',', '(', ':', '='] : stopAt, !!this.#options.typescript) : null;
-				if (found !== null && (index === ENTRY.expression || !this.#source.startsWith(':', found[0]))) return this.#identifier(offset, found[0], index === ENTRY.pattern);
+				if (found !== null) return this.#identifier(offset, found[0], index === ENTRY.pattern);
 			}
 			return result(engine.parse(this.#held, index, offset, end, stop), this.#source);
 		}
@@ -164,7 +168,7 @@ export function bind(engine) {
 		#position(offset) {
 			if (this.#lines === undefined) {
 				this.#lines = [0];
-				for (let i = this.#source.indexOf('\n'); i !== -1; i = this.#source.indexOf('\n', i + 1)) this.#lines.push(i + 1);
+				for (const m of this.#source.matchAll(/\r\n?|[\n\u2028\u2029]/g)) this.#lines.push(m.index + m[0].length);
 			}
 			let lo = 0, hi = this.#lines.length - 1;
 			while (lo < hi) {

package/index.js

@@ -17,10 +17,12 @@ function bytes(text) {
 	return room.subarray(0, written);
 }
 
-export const Source = bind({
+export const engine = {
 	create: (source, names, host) => native.create(bytes(source), names, host),
 	parse: native.parse,
 	free: native.free,
 	constants: native.constants,
 	shapes: native.shapes,
-});
+};
+
+export const Source = bind(engine);

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@teasel/parser",
-	"version": "0.0.2",
+	"version": "0.0.4",
 	"description": "A JavaScript and TypeScript parser in Rust. It answers in ESTree.",
 	"license": "MIT",
 	"repository": {
@@ -53,10 +53,10 @@
 		"@changesets/cli": "^3.0.1"
 	},
 	"optionalDependencies": {
-		"@teasel/parser-linux-x64-gnu": "0.0.2",
-		"@teasel/parser-linux-arm64-gnu": "0.0.2",
-		"@teasel/parser-darwin-x64": "0.0.2",
-		"@teasel/parser-darwin-arm64": "0.0.2",
-		"@teasel/parser-win32-x64-msvc": "0.0.2"
+		"@teasel/parser-linux-x64-gnu": "0.0.4",
+		"@teasel/parser-linux-arm64-gnu": "0.0.4",
+		"@teasel/parser-darwin-x64": "0.0.4",
+		"@teasel/parser-darwin-arm64": "0.0.4",
+		"@teasel/parser-win32-x64-msvc": "0.0.4"
 	}
 }

package/wasm.js

@@ -8,17 +8,34 @@ const utf8 = new TextDecoder();
 
 // `teasel.wasm` next to this file, read where there is a file system and fetched elsewhere
 const url = new URL('./teasel.wasm', import.meta.url);
-const { instance } =
+const { module, instance } =
 	url.protocol === 'file:'
 		? await WebAssembly.instantiate(await (await import('node:fs/promises')).readFile(url), {})
 		: await WebAssembly.instantiateStreaming(fetch(url), {});
 /** @type {WebAssembly.Exports & Record<string, Function> & { memory: WebAssembly.Memory }} */
-const wasm = /** @type {any} */ (instance.exports);
+let wasm = /** @type {any} */ (instance.exports);
 /** @type {string[]} */
 let constants = [];
 /** @type {number[]} */
 let shapes = [];
 let shapes_known = 0;
+// a panic traps the instance for good: a fresh one takes over, and the sources held by the old one are gone
+let generation = 0;
+
+/** @template T @param {() => T} f @returns {T} */
+function guarded(f) {
+	try {
+		return f();
+	} catch (error) {
+		if (!(error instanceof WebAssembly.RuntimeError)) throw error;
+		wasm = /** @type {any} */ (new WebAssembly.Instance(module, {}).exports);
+		constants = [];
+		shapes = [];
+		shapes_known = 0;
+		generation++;
+		throw new Error('the engine panicked and started over; the sources it held are gone', { cause: error });
+	}
+}
 
 // the module takes the bytes over
 function bytes(text) {
@@ -29,9 +46,15 @@ function bytes(text) {
 }
 
 function create(source, names, host) {
-	const handle = wasm.source_new(...bytes(source), ...bytes(names), ...bytes(host));
+	const handle = guarded(() => wasm.source_new(...bytes(source), ...bytes(names), ...bytes(host)));
 	if (handle === 0) throw new Error(JSON.parse(text()).error.message);
-	return handle;
+	return { handle, generation };
+}
+
+/** @param {{ handle: number, generation: number }} held */
+function handle(held) {
+	if (held.generation !== generation) throw new Error('the source was held by an engine that panicked and started over');
+	return held.handle;
 }
 
 const text = () => utf8.decode(new Uint8Array(wasm.memory.buffer, wasm.text_ptr(), wasm.text_len()));
@@ -52,11 +75,15 @@ function answer(status) {
 	return words();
 }
 
-export const Source = bind({
+export const engine = {
 	create,
 	// the words outlive the source: they sit in the answer buffer until the next parse
-	parse: (held, entry, offset, end, stop) => answer(wasm.source_parse(held, entry, offset, end ?? 0, end === undefined ? 0 : 1, ...bytes(stop))),
-	free: (held) => wasm.source_free(held),
+	parse: (held, entry, offset, end, stop) => answer(guarded(() => wasm.source_parse(handle(held), entry, offset, end ?? 0, end === undefined ? 0 : 1, ...bytes(stop)))),
+	free: (held) => {
+		if (held.generation === generation) wasm.source_free(held.handle);
+	},
 	constants: () => constants,
 	shapes: () => shapes,
-});
+};
+
+export const Source = bind(engine);