@uniweb/semantic-parser 1.1.12 → 1.1.14

package/README.md

@@ -229,6 +229,29 @@ Bracketed spans (`[text]{.class}`) are converted to `<span>` elements with their
 
 Spans can have classes, IDs, and custom attributes. They combine with other marks—a span with bold becomes `<strong><span class="...">text</span></strong>`.
 
+### Inline Insets and Cross-references
+
+The parser handles inline `inset_placeholder` nodes — generated by
+`@uniweb/build`'s content collector when it pulls
+`![alt](@Component){k=v}`, `[text](@Component){k=v}`, `[@key]{k=v}`,
+or `[#id]{k=v}` references out of paragraph content. Two artifacts
+ship to consumers:
+
+1. **`children` carries an `inset` entry per placeholder** — each
+   carrying the `refId` and `embedKind` (`'text'` for textual sugars
+   like `[@key]` / `[#id]` / `[text](@Component)`; `'visual'` for
+   `![alt](@Component)`). Renderers that walk `paragraph.children`
+   pick these up alongside icons, math, and links.
+
+2. **`text` carries a `<uniweb-inset data-ref-id="…"></uniweb-inset>`
+   marker** at the placeholder position — useful for renderers that
+   read `paragraph.text` as a flat HTML string (kit's `<Prose>` is the
+   canonical example; it splits on the marker and intersperses
+   React-rendered insets via the framework's child-block renderer).
+
+The two views stay in lockstep — both record every inline inset the
+parser sees, in source order.
+
 ## Use Cases
 
 - **Component-based websites**: Extract structured data for React/Vue components

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@uniweb/semantic-parser",
-  "version": "1.1.12",
+  "version": "1.1.14",
   "description": "Semantic parser for ProseMirror/TipTap content structures",
   "type": "module",
   "main": "./src/index.js",

package/src/processors/sequence.js

@@ -572,6 +572,8 @@ function parseImgBlock(itemAttrs) {
         theme,
         role,
         credit = "",
+        id, // {#fig-id} cross-reference label — preserved so Press
+            // adapters can emit \label{id} (LaTeX) / <id> (Typst).
     } = itemAttrs;
 
     let { contentType, viewType, contentId, identifier } = imgInfo || {};
@@ -605,6 +607,7 @@ function parseImgBlock(itemAttrs) {
         theme,
         role,
         credit,
+        id: id || undefined,
     };
 }
 
@@ -639,16 +642,24 @@ function parseVideoBlock(itemAttrs) {
 function stripTags(htmlString) {
     if (!htmlString || typeof htmlString !== "string") return "";
 
-    // Remove HTML tags using regular expression
+    // Remove HTML tags using regular expression.
     const plainString = htmlString.replace(/<[^>]*>/g, "");
 
-    // Decode HTML entities
-    const decodedString = new DOMParser().parseFromString(
-        plainString,
-        "text/html"
-    ).body.textContent;
-
-    return decodedString;
+    // Decode the HTML entity subset citestyle / content-reader / kit emit.
+    // The previous DOMParser-based path covered every entity in the spec,
+    // but DOMParser is browser-only and crashes Node SSR (which the
+    // unipress compile pipeline runs in). The regex below covers the
+    // entities that actually appear in this codebase's emitted HTML.
+    return plainString
+        .replace(/&amp;/g, "&")
+        .replace(/&lt;/g, "<")
+        .replace(/&gt;/g, ">")
+        .replace(/&quot;/g, '"')
+        .replace(/&#39;/g, "'")
+        .replace(/&apos;/g, "'")
+        .replace(/&nbsp;/g, " ")
+        .replace(/&#(\d+);/g, (_, n) => String.fromCharCode(Number(n)))
+        .replace(/&#x([0-9a-fA-F]+);/g, (_, n) => String.fromCharCode(parseInt(n, 16)));
 }
 
 function isLink(item) {