initial ideas

lang2/lexer.ts

@@ -0,0 +1,54 @@
+type Token<T extends string> = {
+	type: T,
+	string: string,
+}
+
+type LexerProps<T extends string> = {
+	rules: {
+		[key in T]: RegExp;
+	},
+	// postprocess?: (token: Token<T>) =>
+}
+
+const entries = <K extends string, V>(obj: {[key in K]: V}) => {
+	return Object.entries(obj) as Array<[K, V]>;
+}
+
+const lexer = <T extends string>(props: LexerProps<T>) => {
+	const {rules} = props;
+	return (string: string): Array<Token<T>> => {
+		const tokens: Array<Token<T>> = [];
+		let str = string;
+		while (str.length) {
+			let matched = false;
+			for (const [type, matcher] of entries(rules)) {
+				const match = str.match(matcher);
+				if (match && match.index === 0) {
+					if (type !== "_skip") {
+						tokens.push({type, string: match[0]});
+					}
+					str = str.slice(match[0].length);
+					matched = true;
+					break;
+				}
+			}
+			if (!matched) {
+				console.log(tokens);
+				console.log(str);
+				throw 'Infinite Loop';
+			}
+		}
+		return tokens;
+	}
+}
+
+const my_lexer = lexer({
+	rules: {
+		symbol: /\w+/,
+		punctuation: /[\.,:;()[\]{}]/,
+		string_literal: /"([^"\\]|\\.)*"/,
+		_skip: /\s+/,
+	}
+});
+
+console.log(my_lexer(`abc def { ghi } "string literal!" "with \\"escaped\\" quote marks"`));

lang2/peg.js

@@ -0,0 +1,28 @@
+const Expression = rule(
+	[{head: Term}, {tail: star([_, slash("+", "-"), _, Term])}],
+	({head, tail}) => {
+		return tail.reduce(function(result, element) {
+			if (element[1] === "+") { return result + element[3]; }
+			if (element[1] === "-") { return result - element[3]; }
+		}, head);
+	}
+);
+
+const Term = rule(
+	[{head: Factor}, {tail: star([_, slash("*", "/"), _, Factor])}],
+	({head, tail}) => {
+		return tail.reduce(function(result, element) {
+			if (element[1] === "*") { return result * element[3]; }
+			if (element[1] === "/") { return result / element[3]; }
+		}, head);
+	}
+);
+
+const Factor = slash(
+	rule(["(", _, {expr: Expression}, _, ")"], ({expr}) => expr),
+	Integer
+);
+
+const Integer = rule([_, /[0-9]+/], () => { return parseInt(text(), 10); });
+
+const _ = /[ \t\n\r]*/;