src/paca.mjs (tokenizeRegexStep): Support tokenizing range exps {m,n}

1 files changed, 98 insertions, 2 deletions

diff --git a/src/paca.mjs b/src/paca.mjs
index e68cd77..53a9c11 100644
--- a/src/paca.mjs
+++ b/src/paca.mjs
@@ -1,5 +1,5 @@
 import {
-	butlast, explode, last, mapValues, max, reduce, reduced,
+	butlast, explode, isNumeric, last, mapValues, max, reduce, reduced,
 } from "sjs";
 
 
@@ -9,6 +9,7 @@ export class SyntaxError extends Error {}
 const ConcatStep = {
 	ACCEPTING: "accepting",
 	ESCAPING:  "escaping",
+	RANGE:     "range",
 };
 
 const nonConcatOperators = new Set(["*", "+", "?", "|", ")"]);
@@ -23,6 +24,11 @@ const shouldConcat = (char, next) =>
 const isOperator = char =>
 	nonConcatOperators.has(char) || char == "(";
 
+const numFromDigits = digits =>
+	digits.length === 0
+		? -1
+		: Number(digits.join(""));
+
 const tokenizeRegexStep = chars => ({ out, state, context }, char, index) => {
 	const next = chars[index + 1];
 	const maybeConcat = shouldConcat(char, next)
@@ -37,6 +43,84 @@ const tokenizeRegexStep = chars => ({ out, state, context }, char, index) => {
 		};
 	}
 
+	if (state === ConcatStep.RANGE) {
+		if (char === "}") {
+			if (context.where !== "to") {
+				return reduced({
+					out,
+					state,
+					context,
+					error: new SyntaxError(
+						"missing comma in range operator",
+					),
+				});
+			}
+
+			const from = numFromDigits(context.from);
+			const to   = numFromDigits(context.to);
+			if (from > to && to != -1) {
+				return reduced({
+					out,
+					state,
+					context,
+					error: new Error(`bad range values: {${from},${to}}`),
+				});
+			}
+			return {
+				out: out.concat({
+					operator: "range",
+					from,
+					to,
+				}),
+				state: ConcatStep.ACCEPTING,
+				context: null,
+			};
+		}
+
+		if (char === ",") {
+			if (context.where === "to") {
+				return reduced({
+					out,
+					state,
+					context,
+					error: new SyntaxError(
+						"extraneuos comma in range expression",
+					),
+				});
+			} else {
+				return {
+					out,
+					state,
+					context: {
+						...context,
+						where: "to",
+					},
+				};
+			}
+		}
+
+		if (!isNumeric(char)) {
+			return reduced({
+				out,
+				state,
+				context,
+				error: new SyntaxError(
+					"bad char in range expression: " +
+						char,
+				),
+			});
+		}
+
+		return {
+			out,
+			state,
+			context: {
+				...context,
+				[context.where]: context[context.where].concat(char),
+			},
+		};
+	}
+
 	if (char === "\\") {
 		return {
 			out,
@@ -45,6 +129,18 @@ const tokenizeRegexStep = chars => ({ out, state, context }, char, index) => {
 		};
 	}
 
+	if (char === "{") {
+		return {
+			out,
+			state: ConcatStep.RANGE,
+			context: {
+				from:  [],
+				to:    [],
+				where: "from",
+			},
+		};
+	}
+
 	const op = isOperator(char) ? { operator: char } : char;
 	return {
 		out: out.concat(op, maybeConcat),
@@ -54,7 +150,7 @@ const tokenizeRegexStep = chars => ({ out, state, context }, char, index) => {
 };
 
 const tokenizeRegexFn = chars =>
-	chars.reduce(tokenizeRegexStep(chars), {
+	reduce(chars, tokenizeRegexStep(chars), {
 		out:     [],
 		state:   ConcatStep.ACCEPTING,
 		context: null,