@haklex/rich-litexml 0.15.4 → 0.15.6

package/dist/browser.mjs

@@ -1,4 +1,4 @@
-import { a as createDefaultRegistry, c as LitexmlRegistry, d as registerParseHTML, i as deserializeNodesFromXml, l as registerCustomReaders, n as serializeToXml, o as registerCustomWriters, r as deserializeFromXml, s as registerBuiltinWriters, t as serializeNodesToXml, u as registerBuiltinReaders } from "./src-DmI3DOeG.js";
+import { a as createDefaultRegistry, c as LitexmlRegistry, d as registerParseHTML, i as deserializeNodesFromXml, l as registerCustomReaders, n as serializeToXml, o as registerCustomWriters, r as deserializeFromXml, s as registerBuiltinWriters, t as serializeNodesToXml, u as registerBuiltinReaders } from "./src-z3l9E6En.js";
 //#region src/index-browser.ts
 registerParseHTML((html) => new DOMParser().parseFromString(html, "text/html"));
 //#endregion

package/dist/deserializer.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"deserializer.d.ts","sourceRoot":"","sources":["../src/deserializer.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,qBAAqB,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AAG5E,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAQlD,wBAAgB,kBAAkB,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,EAAE,eAAe,GAAG,qBAAqB,CAkBhG;AAED,wBAAgB,uBAAuB,CACrC,GAAG,EAAE,MAAM,EACX,QAAQ,EAAE,eAAe,GACxB,qBAAqB,EAAE,CAKzB"}
+{"version":3,"file":"deserializer.d.ts","sourceRoot":"","sources":["../src/deserializer.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,qBAAqB,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AAG5E,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,YAAY,CAAC;AAwElD,wBAAgB,kBAAkB,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,EAAE,eAAe,GAAG,qBAAqB,CAkBhG;AAED,wBAAgB,uBAAuB,CACrC,GAAG,EAAE,MAAM,EACX,QAAQ,EAAE,eAAe,GACxB,qBAAqB,EAAE,CAKzB"}

package/dist/node.mjs

@@ -1,4 +1,4 @@
-import { a as createDefaultRegistry, c as LitexmlRegistry, d as registerParseHTML, i as deserializeNodesFromXml, l as registerCustomReaders, n as serializeToXml, o as registerCustomWriters, r as deserializeFromXml, s as registerBuiltinWriters, t as serializeNodesToXml, u as registerBuiltinReaders } from "./src-DmI3DOeG.js";
+import { a as createDefaultRegistry, c as LitexmlRegistry, d as registerParseHTML, i as deserializeNodesFromXml, l as registerCustomReaders, n as serializeToXml, o as registerCustomWriters, r as deserializeFromXml, s as registerBuiltinWriters, t as serializeNodesToXml, u as registerBuiltinReaders } from "./src-z3l9E6En.js";
 import { parseHTML } from "linkedom";
 //#region src/index-node.ts
 registerParseHTML((html) => parseHTML(html).document);

package/dist/{src-DmI3DOeG.js → src-z3l9E6En.js}

@@ -1025,8 +1025,66 @@ function getFormatBit(tagName) {
 }
 //#endregion
 //#region src/deserializer.ts
+/**
+* HTML5 void elements — tags the HTML parser already treats as self-closing
+* regardless of `/>` syntax. Custom or non-void tags written as `<tag … />`
+* are otherwise interpreted as opening tags by the HTML parser, which then
+* swallows all following siblings as children until a matching close tag (or
+* end of document) is found. We expand those into explicit `<tag …></tag>`
+* before handing the string to linkedom / DOMParser.
+*/
+var HTML_VOID_ELEMENTS = new Set([
+	"area",
+	"base",
+	"br",
+	"col",
+	"embed",
+	"hr",
+	"img",
+	"input",
+	"link",
+	"meta",
+	"param",
+	"source",
+	"track",
+	"wbr"
+]);
+var SELF_CLOSING_RE = /<([\w-]+)((?:\s+[\w-]+(?:\s*=\s*(?:"[^"]*"|'[^']*'))?)*)\s*\/>/g;
+var CDATA_OPEN = "<![CDATA[";
+var CDATA_CLOSE = "]]>";
+function expandSelfClosing(chunk) {
+	return chunk.replaceAll(SELF_CLOSING_RE, (match, tag, attrs) => {
+		if (HTML_VOID_ELEMENTS.has(tag.toLowerCase())) return match;
+		return `<${tag}${attrs}></${tag}>`;
+	});
+}
+/**
+* Expand `<custom-tag … />` to `<custom-tag …></custom-tag>` outside CDATA.
+* Leaves HTML void elements and already-paired tags untouched. CDATA content
+* is sliced out verbatim with `indexOf` to avoid regex pitfalls around `]]>`.
+*/
+function normalizeSelfClosingTags(xml) {
+	const out = [];
+	let cursor = 0;
+	while (cursor < xml.length) {
+		const start = xml.indexOf(CDATA_OPEN, cursor);
+		if (start < 0) {
+			out.push(expandSelfClosing(xml.slice(cursor)));
+			break;
+		}
+		out.push(expandSelfClosing(xml.slice(cursor, start)));
+		const end = xml.indexOf(CDATA_CLOSE, start + 9);
+		if (end < 0) {
+			out.push(xml.slice(start));
+			break;
+		}
+		out.push(xml.slice(start, end + 3));
+		cursor = end + 3;
+	}
+	return out.join("");
+}
 function parseXml(xml) {
-	return parseHTML(`<!DOCTYPE html><html><body>${xml}</body></html>`);
+	return parseHTML(`<!DOCTYPE html><html><body>${normalizeSelfClosingTags(xml)}</body></html>`);
 }
 function deserializeFromXml(xml, registry) {
 	const doc = parseXml(xml);
@@ -1075,16 +1133,24 @@ function createReaderContext(registry) {
 		parseChildren(element) {
 			const blockLevel = isBlockContainer(element);
 			const nodes = [];
-			for (const child of element.childNodes) if (child.nodeType === 3) {
-				const text = child.textContent ?? "";
-				if (blockLevel && text.trim() === "") continue;
-				if (text === "") continue;
-				nodes.push(makeTextNode(text, 0));
-			} else if (child.nodeType === 1) {
+			let pendingText = "";
+			const flushText = () => {
+				if (pendingText === "") return;
+				if (blockLevel && pendingText.trim() === "") {
+					pendingText = "";
+					return;
+				}
+				nodes.push(makeTextNode(pendingText, 0));
+				pendingText = "";
+			};
+			for (const child of element.childNodes) if (child.nodeType === 3) pendingText += child.textContent ?? "";
+			else if (child.nodeType === 1) {
+				flushText();
 				const parsed = parseElement(child, registry, ctx, 0);
 				if (parsed) if (Array.isArray(parsed)) nodes.push(...parsed);
 				else nodes.push(parsed);
 			}
+			flushText();
 			return nodes;
 		},
 		parseNestedState(xml) {
@@ -1110,15 +1176,20 @@ function parseElement(element, registry, ctx, inheritedFormat) {
 }
 function parseInlineChildren(element, registry, ctx, format) {
 	const nodes = [];
-	for (const child of element.childNodes) if (child.nodeType === 3) {
-		const text = child.textContent ?? "";
-		if (text === "") continue;
-		nodes.push(makeTextNode(text, format));
-	} else if (child.nodeType === 1) {
+	let pendingText = "";
+	const flushText = () => {
+		if (pendingText === "") return;
+		nodes.push(makeTextNode(pendingText, format));
+		pendingText = "";
+	};
+	for (const child of element.childNodes) if (child.nodeType === 3) pendingText += child.textContent ?? "";
+	else if (child.nodeType === 1) {
+		flushText();
 		const parsed = parseElement(child, registry, ctx, format);
 		if (parsed) if (Array.isArray(parsed)) nodes.push(...parsed);
 		else nodes.push(parsed);
 	}
+	flushText();
 	return nodes;
 }
 function parseFallbackNode(element) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@haklex/rich-litexml",
-  "version": "0.15.4",
+  "version": "0.15.6",
   "description": "Bidirectional Lexical SerializedNode <-> XML conversion with plugin registry",
   "repository": {
     "type": "git",