From 29d26068ad98e865272198090375fca57e5cf156 Mon Sep 17 00:00:00 2001
From: Tom Smeding <tom@tomsmeding.com>
Date: Mon, 12 Jun 2023 23:07:05 +0200
Subject: unicode: Recognise multiple index literals in input

---
 modules/unicode/index.html | 32 ++++++++++++++++++++++----------
 modules/unicode/unicode.js | 45 ++++++++++++++++++++++++++++++---------------
 2 files changed, 52 insertions(+), 25 deletions(-)

diff --git a/modules/unicode/index.html b/modules/unicode/index.html
index 924f86e..8738b6f 100644
--- a/modules/unicode/index.html
+++ b/modules/unicode/index.html
@@ -49,7 +49,7 @@ function do_lookup(override) {
 	xhr.onreadystatechange = function() {
 		if (xhr.readyState == 4) {
 			if (xhr.status == 200) {
-				handleResponse(JSON.parse(xhr.responseText), input);
+				handleResponse(JSON.parse(xhr.responseText));
 			} else {
 				alert("Request failed: " + xhr.responseText);
 			}
@@ -62,7 +62,7 @@ function do_lookup(override) {
 	xhr.send();
 }
 
-function handleResponse(json, input) {
+function handleResponse(json) {
 	var elem;
 
 	/* elem = document.getElementById("json");
@@ -70,12 +70,20 @@ function handleResponse(json, input) {
 	elem.appendChild(document.createTextNode(JSON.stringify(json))); */
 
 	var index_container = document.getElementById("index_container");
-	if ("index" in json) {
+	if (json.indices.length > 0) {
 		index_container.classList.remove("invisible");
-		document.getElementById("index_input").innerHTML = input;
-		elem = document.getElementById("index");
-		elem.innerHTML = "";
-		elem.appendChild(makeCodepointDiv(json["index"]));
+		setTableRows("indices", json.indices);
+		if (json.indices.length > 1) {
+			document.getElementById("indices_concatenated_container").classList.remove("invisible");
+			elem = document.getElementById("indices_concatenated");
+			elem.innerHTML = "";
+			var str = "";
+			for (var i = 0; i < json.indices.length; i++)
+				str += String.fromCodePoint(parseInt(json.indices[i][0], 16));
+			elem.appendChild(document.createTextNode(str));
+		} else {
+			document.getElementById("indices_concatenated_container").classList.add("invisible");
+		}
 	} else if (!index_container.classList.contains("invisible")) {
 		index_container.classList.add("invisible");
 	}
@@ -141,15 +149,19 @@ window.addEventListener("load", function() {
 	<input type="button" onclick="do_lookup()" value="Lookup">
 	<br>
 
-	<!-- <pre id="json"></pre><br> -->
+	<!-- <code id="json"></code><br> -->
 
 	<div id="notfound_container" class="invisible">
 		These were not found, perhaps the server needs to be update its unicode table? <code id="notfound"></code>
 	</div>
 
 	<div id="index_container" class="invisible">
-		<h3>Codepoint <span id="index_input"></span></h3>
-		<div id="index" class="table"></div>
+		<h3>Numeric codepoints <span id="indices_num"></span></h3>
+		<div id="indices" class="table"></div>
+		<br>
+		<div id="indices_concatenated_container">
+			Concatenated: &#x27e8; <span id="indices_concatenated"></span> &#x27e9;
+		</div>
 	</div>
 
 	<h3>Codepoints <span id="codepoints_num"></span></h3>
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]+);$/);  // &#1234;
-	if (m) return lookupCode(parseInt(m[1], 10));
+function recogniseIndices(text) {
+	const results = [];
+
+	// A: 1234 (decimal)
+	// B: 0x34ab / U+34ab (hexadecimal)
+	// C: &#1234; (xml-style decimal)
+	// D: &#x34ab; (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]+);$/);  // &#x34ab;
-	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),
-- 
cgit v1.2.3-54-g00ecf