@uniweb/semantic-parser 1.3.0 → 1.4.0

package/README.md

@@ -162,7 +162,7 @@ When a group begins with headings, the parser reads one **headline stack** with
 2. **One step smaller** — the next line of the headline: the subtitle first, then further subtitle lines (`subtitle` is a string for one line, an array for several — a three-line header like name / role / affiliation fits in one headline).
 3. **Two or more steps smaller, or a step back up** — the stack ends; the rest of the run starts a new group (an item), read the same way.
 
-Before the title, `#>` label lines (headings carrying `role: "pretitle"`) and smaller headings ascending to a more important one all join the `pretitle`. A label line anywhere names the block it sits in; a label with no following heading labels an untitled block.
+Before the title, `#>` label lines (headings carrying `role: "pretitle"`) and every smaller heading stacked above the most important one — in whatever order — join the `pretitle`. A label line anywhere names the block it sits in; a label with no following heading labels an untitled block.
 
 ### Main content vs items
 
@@ -180,7 +180,7 @@ This is the mechanism that creates repeating content groups (cards, features, FA
 Two spellings fill the `pretitle` slot:
 
 - **A `#>` label line** — a heading with `role: "pretitle"`. It names the block that starts next, at any depth, with no level arithmetic; the hash count is carried for round-tripping and means nothing.
-- **Position** — headings followed by a more important heading at the start of a group (H3 before H1, H6 before H5, …). Stacked smaller headings all join: `#### / ### / # Title` yields a two-line pretitle.
+- **Position** — smaller headings stacked above the most important heading of a run (H3 before H1, H6 before H5, …), in any order. `#### / ### / # Title` and `## / ### / # Title` both yield a two-line pretitle.
 
 ### Banner image
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@uniweb/semantic-parser",
-  "version": "1.3.0",
+  "version": "1.4.0",
   "description": "Semantic parser for ProseMirror/TipTap content structures",
   "type": "module",
   "main": "./src/index.js",
@@ -30,9 +30,7 @@
   "directories": {
     "test": "tests"
   },
-  "dependencies": {
-    "yaml": "^2.8.2"
-  },
+  "dependencies": {},
   "scripts": {
     "test": "vitest run",
     "test-report": "vitest run --reporter=json --outputFile=test-results.json",

package/src/processors/groups.js

@@ -232,15 +232,20 @@ function readStack(sequence, startIdx, options = {}) {
         i++;
     }
 
-    // Ascending positional pretitle: smaller headings stacked directly above
-    // a more important one (e.g. #### / ### / # Title) all join the pretitle.
-    while (
-        i + 1 < sequence.length &&
-        sequence[i].type === "heading" &&
-        sequence[i + 1].type === "heading" &&
-        !isLabel(sequence[i + 1]) &&
-        sequence[i + 1].level < sequence[i].level
-    ) {
+    // Positional pretitle: the title is the FIRST occurrence of the run's most
+    // important level; every smaller heading stacked above it joins the
+    // pretitle, in whatever order they were written (#### / ### / # and
+    // ## / ### / # both give a two-line pretitle). An ascending-only walk
+    // read a descending stack as a bodiless headline followed by a bigger
+    // heading, which demoted the real title into the items.
+    let best = i;
+    for (let k = i; k < sequence.length; k++) {
+        const el = sequence[k];
+        if (el.type !== "heading") break;
+        if (isLabel(el)) continue;
+        if (el.level < sequence[best].level || isLabel(sequence[best])) best = k;
+    }
+    while (i < best) {
         pre.push(sequence[i]);
         i++;
     }

package/src/processors/sequence.js

@@ -1,16 +1,3 @@
-import { parse as parseYaml } from "yaml";
-
-/**
- * Get code block data - prefers pre-parsed attrs.data, falls back to parsing text
- *
- * Content can come from two sources:
- * 1. Pre-parsed at build time: attrs.data contains parsed JS object
- * 2. Legacy/runtime: text needs to be parsed based on language
- *
- * @param {string} text - Raw code block text
- * @param {Object} attrs - Code block attributes (language, tag, data)
- * @returns {*} Parsed data or raw text
- */
 /**
  * HTML-attribute encoding for values placed inside double-quoted
  * attributes. Backslashes and braces (common in LaTeX) pass through
@@ -28,42 +15,6 @@ function escapeAttr(s) {
         .replace(/>/g, ">");
 }
 
-function getCodeBlockData(text, attrs) {
-    const { language, tag, data } = attrs || {};
-
-    // Only process tagged blocks
-    if (!tag) {
-        return text;
-    }
-
-    // Prefer pre-parsed data from build time (attrs.data)
-    if (data !== undefined) {
-        return data;
-    }
-
-    // Fallback: parse text at runtime (for backwards compatibility)
-    const lang = (language || "").toLowerCase();
-
-    if (lang === "json") {
-        try {
-            return JSON.parse(text);
-        } catch {
-            return text;
-        }
-    }
-
-    if (lang === "yaml" || lang === "yml") {
-        try {
-            return parseYaml(text);
-        } catch {
-            return text;
-        }
-    }
-
-    // Unknown language - return raw text
-    return text;
-}
-
 /**
  * Check if an inline node is an icon.
  * TipTap editor uses type "UniwebIcon"; markdown pipeline uses type "image" with role "icon".
@@ -172,9 +123,10 @@ function createSequenceElement(node, options = {}) {
             // to a text string and lose the author's prose.
             //
             // No markdown is parsed here and none may ever be: this package
-            // depends on `yaml` alone, and reaching for a markdown parser would
-            // put content-reader inside the runtime. The reader already built
-            // this node; the job here is only to WALK it.
+            // parses NOTHING and has no parser dependency at all, and reaching
+            // for a markdown parser would put content-reader inside the
+            // runtime. The reader already built this node; the job here is only
+            // to WALK it.
             //
             // `options` rides along so nested content is sequenced under the
             // same options as its parent. (`inset_block` and `blockquote` used
@@ -198,12 +150,23 @@ function createSequenceElement(node, options = {}) {
 
         case "codeBlock": {
             const codeText = getTextContent(content, options);
-            // Tagged code blocks are semantically data blocks, not code
+            // Tagged code blocks are semantically data blocks, not code — the
+            // tag is what routes them to `data[tag]`, so the routing holds even
+            // though the payload here is raw text.
+            //
+            // ⛔ NOTHING IS PARSED HERE, and nothing may be. A tagged block the
+            // reader could parse arrives as a `dataBlock` carrying `attrs.data`
+            // (the case above); content-reader emits THIS shape only when
+            // `parseCodeBlockData` already returned null — i.e. the text is not
+            // valid JSON or YAML. Re-running a parser on it in the browser
+            // cannot succeed, and it used to cost every site the whole `yaml`
+            // package: 98 KB minified, ~29 KB gzip, pinned into the bundle by a
+            // static import that no bundler could shake. Removed 2026-09-01.
             if (attrs?.tag) {
                 return {
                     type: "dataBlock",
                     tag: attrs.tag,
-                    data: getCodeBlockData(codeText, attrs),
+                    data: codeText,
                 };
             }
             return {