@localess/richtext 4.0.1-dev.20260916085732 → 4.0.1

package/SKILL.md

@@ -33,7 +33,8 @@ const html = renderRichTextToHtml(data.body);
 - Output is byte-identical to TipTap's `generateHTML` for the Studio's
   extension set, except link `href`s pass a protocol allowlist
   (`http:`/`https:`/`mailto:`/`tel:`/scheme-less); `javascript:`/`data:`
-  hrefs become `""`.
+  hrefs become `""`, including obfuscated forms (`java\tscript:`, leading
+  control characters) — the href is normalized like a browser URL parser first.
 - Never throws; malformed input renders `''`.
 
 ## Supported node set
@@ -152,7 +153,7 @@ ordered lists (nested, `start`), fenced and indented code blocks, `**bold**`, `*
 Anything else goes through `unsupported` and is listed in the result — return the report to the
 caller, do not assume a clean parse. One warning per element type per parse, silent in production.
 
-Link hrefs pass the renderer's `sanitizeUrl` allowlist: `javascript:` and `data:` become `""`.
+Link hrefs pass the renderer's `sanitizeUrl` allowlist: `javascript:` and `data:` become `""`, including entity-encoded tab/newline/control-character obfuscations.
 
 Never throws except under `unsupported: 'throw'`; `null`, `undefined`, and `''` give an empty doc.
 

package/dist/escape.d.ts

@@ -10,5 +10,9 @@ export declare function escapeAttr(value: string): string;
  * Allowlist URL sanitizer for link hrefs: `http:`, `https:`, `mailto:`, `tel:`
  * and scheme-less (relative/protocol-relative/fragment/query) URLs pass;
  * everything else (e.g. `javascript:`, `data:`) becomes `''`.
+ *
+ * The input is first normalized the way the WHATWG URL parser does it (leading/trailing
+ * C0 control or space stripped, tab/newline removed everywhere), so obfuscated schemes
+ * such as `java\tscript:` are caught; the normalized URL is what gets returned.
  */
 export declare function sanitizeUrl(url: string): string;

package/dist/html-parser/index.js

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_parse_common = require("../parse-common-DEYhfKnK.js");
+const require_parse_common = require("../parse-common-ColxMxGw.js");
 //#region src/html-parser/to-model.ts
 /** Tag → mark, the inverse of `MARK_RENDER_MAP`. `b`/`i`/`del`/`s` are accepted as aliases. */
 var MARK_TAGS = {

package/dist/html-parser/index.mjs

@@ -1,4 +1,4 @@
-import { n as UnsupportedTracker, o as sanitizeUrl, r as emptyDocument, t as RichTextParseError } from "../parse-common-C-dAtFji.mjs";
+import { n as UnsupportedTracker, o as sanitizeUrl, r as emptyDocument, t as RichTextParseError } from "../parse-common-CmQhI2Da.mjs";
 //#region src/html-parser/to-model.ts
 /** Tag → mark, the inverse of `MARK_RENDER_MAP`. `b`/`i`/`del`/`s` are accepted as aliases. */
 var MARK_TAGS = {

package/dist/index.js

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_parse_common = require("./parse-common-DEYhfKnK.js");
+const require_parse_common = require("./parse-common-ColxMxGw.js");
 //#region src/attrs.ts
 /**
 * Normalizes a node/mark's stored attrs into the attributes to emit, in the

package/dist/index.mjs

@@ -1,4 +1,4 @@
-import { a as escapeHtml, i as escapeAttr, n as UnsupportedTracker, o as sanitizeUrl, r as emptyDocument, t as RichTextParseError } from "./parse-common-C-dAtFji.mjs";
+import { a as escapeHtml, i as escapeAttr, n as UnsupportedTracker, o as sanitizeUrl, r as emptyDocument, t as RichTextParseError } from "./parse-common-CmQhI2Da.mjs";
 //#region src/attrs.ts
 /**
 * Normalizes a node/mark's stored attrs into the attributes to emit, in the

package/dist/markdown-parser/index.js

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_parse_common = require("../parse-common-DEYhfKnK.js");
+const require_parse_common = require("../parse-common-ColxMxGw.js");
 //#region src/markdown-parser/inline.ts
 /**
 * Parses Markdown inline content into text nodes with marks.

package/dist/markdown-parser/index.mjs

@@ -1,4 +1,4 @@
-import { n as UnsupportedTracker, o as sanitizeUrl, r as emptyDocument, t as RichTextParseError } from "../parse-common-C-dAtFji.mjs";
+import { n as UnsupportedTracker, o as sanitizeUrl, r as emptyDocument, t as RichTextParseError } from "../parse-common-CmQhI2Da.mjs";
 //#region src/markdown-parser/inline.ts
 /**
 * Parses Markdown inline content into text nodes with marks.

package/dist/{parse-common-C-dAtFji.mjs → parse-common-CmQhI2Da.mjs}

@@ -22,13 +22,19 @@ function escapeAttr(value) {
 }
 var SAFE_SCHEME = /^(?:https?:|mailto:|tel:)/i;
 var HAS_SCHEME = /^[a-z][a-z0-9+.-]*:/i;
+var C0_CONTROL_OR_SPACE_EDGES = /^[\u0000- ]+|[\u0000- ]+$/g;
+var TAB_OR_NEWLINE = /[\t\n\r]/g;
 /**
 * Allowlist URL sanitizer for link hrefs: `http:`, `https:`, `mailto:`, `tel:`
 * and scheme-less (relative/protocol-relative/fragment/query) URLs pass;
 * everything else (e.g. `javascript:`, `data:`) becomes `''`.
+*
+* The input is first normalized the way the WHATWG URL parser does it (leading/trailing
+* C0 control or space stripped, tab/newline removed everywhere), so obfuscated schemes
+* such as `java\tscript:` are caught; the normalized URL is what gets returned.
 */
 function sanitizeUrl(url) {
-	const trimmed = url.trim();
+	const trimmed = url.replace(C0_CONTROL_OR_SPACE_EDGES, "").replace(TAB_OR_NEWLINE, "").trim();
 	if (trimmed === "") return "";
 	if (SAFE_SCHEME.test(trimmed)) return trimmed;
 	if (!HAS_SCHEME.test(trimmed)) return trimmed;

package/dist/{parse-common-DEYhfKnK.js → parse-common-ColxMxGw.js}

@@ -22,13 +22,19 @@ function escapeAttr(value) {
 }
 var SAFE_SCHEME = /^(?:https?:|mailto:|tel:)/i;
 var HAS_SCHEME = /^[a-z][a-z0-9+.-]*:/i;
+var C0_CONTROL_OR_SPACE_EDGES = /^[\u0000- ]+|[\u0000- ]+$/g;
+var TAB_OR_NEWLINE = /[\t\n\r]/g;
 /**
 * Allowlist URL sanitizer for link hrefs: `http:`, `https:`, `mailto:`, `tel:`
 * and scheme-less (relative/protocol-relative/fragment/query) URLs pass;
 * everything else (e.g. `javascript:`, `data:`) becomes `''`.
+*
+* The input is first normalized the way the WHATWG URL parser does it (leading/trailing
+* C0 control or space stripped, tab/newline removed everywhere), so obfuscated schemes
+* such as `java\tscript:` are caught; the normalized URL is what gets returned.
 */
 function sanitizeUrl(url) {
-	const trimmed = url.trim();
+	const trimmed = url.replace(C0_CONTROL_OR_SPACE_EDGES, "").replace(TAB_OR_NEWLINE, "").trim();
 	if (trimmed === "") return "";
 	if (SAFE_SCHEME.test(trimmed)) return trimmed;
 	if (!HAS_SCHEME.test(trimmed)) return trimmed;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@localess/richtext",
-  "version": "4.0.1-dev.20260916085732",
+  "version": "4.0.1",
   "description": "Framework-neutral rich text model and renderer for Localess's TipTap JSON content.",
   "keywords": [
     "localess",
@@ -58,7 +58,7 @@
   },
   "license": "MIT",
   "dependencies": {
-    "@localess/model": "4.0.1-dev.20260916085732"
+    "@localess/model": "4.0.1"
   },
   "devDependencies": {
     "@tiptap/extension-bold": "^3.22.5",