unicode: Recognise multiple index literals in input

1 files changed, 30 insertions, 15 deletions

diff --git a/modules/unicode/unicode.js b/modules/unicode/unicode.js
index ee83179..ab8ad53 100644
--- a/modules/unicode/unicode.js
+++ b/modules/unicode/unicode.js
@@ -82,22 +82,37 @@ function searchDescription(text) {
 	return result;
 }
 
-function recogniseIndex(text) {
-	let m;
-
-	m = text.match(/^([0-9]+)$/);  // 1234
-	if (m) return lookupCode(parseInt(m[1], 10));
-
-	m = text.match(/^(?:0[Xx]|[Uu]\+)([0-9a-fA-F]+)$/);  // 0x34ab / U+34ab
-	if (m) return lookupCode(parseInt(m[1], 16));
-
-	m = text.match(/^&#([0-9]+);$/);  // Ӓ
-	if (m) return lookupCode(parseInt(m[1], 10));
+function recogniseIndices(text) {
+	const results = [];
+
+	// A: 1234 (decimal)
+	// B: 0x34ab / U+34ab (hexadecimal)
+	// C: Ӓ (xml-style decimal)
+	// D: 㒫 (xml-style hexadecimal)
+	// within word boundaries.
+	//                AAAAAA BBBBBBBBBBBBBBBBBBBBBBBBBBBBBB CCCCCCCCCCC DDDDDDDDDDDDDDDDDDDDD
+	const regex = /\b([0-9]+|(?:0[Xx]|[Uu]\+)([0-9a-fA-F]+)|&#([0-9]+);|&#[Xx]([0-9a-fA-F]+);)\b/g;
+	const matches = text.match(regex);
+	if (matches == null) return [];
+
+	for (const m of matches) {
+		let code = -1;
+
+		if ("Xx+".indexOf(m[1]) != -1) {  // hexadecimal
+			code = parseInt(m.slice(2), 16);
+		} else if (m[0] == "&" && "Xx".indexOf(m[2]) != -1) {  // xml-style hexadecimal
+			code = parseInt(m.slice(3, -1), 16);
+		} else if (m[0] == "&") {  // xml-style decimal
+			code = parseInt(m.slice(2, -1), 10);
+		} else {  // decimal
+			code = parseInt(m, 10);
+		}
 
-	m = text.match(/^&#[Xx]([0-9a-fA-F]+);$/);  // 㒫
-	if (m) return lookupCode(parseInt(m[1], 16));
+		const row = lookupCode(code);
+		if (row != null) results.push(row);
+	}
 
-	return null;
+	return results;
 }
 
 module.exports = function (app, io, moddir) {
@@ -123,7 +138,7 @@ module.exports = function (app, io, moddir) {
 		}
 
 		res.json({
-			index: recogniseIndex(req.params.query) || undefined,
+			indices: recogniseIndices(req.params.query),
 			codepoints: codepoints,
 			notfound: notfound,
 			search: searchDescription(req.params.query),