@ai-matrx/print 0.8.2 → 0.8.4

package/CHANGELOG.md

@@ -1,5 +1,35 @@
 # Changelog
 
+## 0.8.4
+
+- **Right-to-left scripts in PDF (Arabic, Hebrew).** Lines are put in visual
+  order by the Unicode Bidirectional Algorithm (bidi-js; words reversed at RTL
+  levels, numbers kept left-to-right, brackets mirrored), RTL paragraphs,
+  headings, list items and cells are right-aligned, and letters are shaped by
+  the font (Arabic joining forms). New fallback faces Noto Sans Arabic and
+  Noto Sans Hebrew (`fonts.arabic` / `fonts.hebrew`). HTML/EPUB blocks carry
+  `dir="auto"`; Word marks RTL paragraphs `bidi` and RTL runs `rtl`.
+- **Korean** in PDF through Noto Sans KR (unicode-range subsets on demand,
+  same `fonts.cjk` loader; Hangul → KR, other CJK → JP).
+- **Table cells are hyphenated** to their own column width in PDF; a word
+  that fits a line whole is never hyphenated.
+- **APA 7 narrative citations use "and"** (`Kim and Alvarez (2021)`),
+  parenthetical keep "&"; `@key [p. 4]` gives a narrative citation with a
+  locator (pandoc syntax).
+- Consumer action: none — refresh to latest.
+
+## 0.8.3
+
+- **Fixed:** an `<artifact>` envelope (how chat answers store tables and code)
+  printed as raw pipes in documents. The document pipeline now unwraps it with
+  the same rules the print skins use (`unwrapArtifactsToMarkdown`, exported
+  from `/markdown`): a table stays a table, code stays code.
+- **Fixed (EPUB):** a `$$…$$` formula inside a sentence rendered as a `<div>`
+  inside `<p>` (epubcheck RSC-005); it is now a block-styled span.
+- **Word:** East-Asian text carries a CJK font hint, so Word/LibreOffice pick
+  an installed Japanese/Chinese face.
+- Consumer action: none — refresh to latest.
+
 ## 0.8.2
 
 - **PDF font fallback covers Japanese / Chinese / Korean.** Characters the

package/dist/document.cjs

@@ -35,8 +35,11 @@ __export(document_exports, {
   CitationEngine: () => CitationEngine,
   DEFAULT_FOOTER: () => DEFAULT_FOOTER,
   DOCUMENT_FORMATS: () => DOCUMENT_FORMATS,
+  NOTO_ARABIC_URL: () => NOTO_ARABIC_URL,
   NOTO_CJK_BASE: () => NOTO_CJK_BASE,
   NOTO_EMOJI_URL: () => NOTO_EMOJI_URL,
+  NOTO_HEBREW_URL: () => NOTO_HEBREW_URL,
+  NOTO_KR_BASE: () => NOTO_KR_BASE,
   NOTO_SYMBOLS_URL: () => NOTO_SYMBOLS_URL,
   PAGE_SIZES: () => PAGE_SIZES,
   buildPdfDefinition: () => buildPdfDefinition,
@@ -78,6 +81,71 @@ __export(document_exports, {
 });
 module.exports = __toCommonJS(document_exports);
 
+// src/document/bidi.ts
+var import_bidi_js = __toESM(require("bidi-js"), 1);
+var BIDI = /* @__PURE__ */ Symbol.for("ai-matrx.print.document.bidi");
+function bidi() {
+  const g = globalThis;
+  if (!g[BIDI]) g[BIDI] = (0, import_bidi_js.default)();
+  return g[BIDI];
+}
+var RTL_RE = /[֐-ࣿיִ-﷿ﹰ-]/;
+function isRtlParagraph(text) {
+  if (!RTL_RE.test(text)) return false;
+  return (bidi().getEmbeddingLevels(text).paragraphs[0]?.level ?? 0) % 2 === 1;
+}
+function visualOrder(input, rtl) {
+  const b = bidi();
+  const runs3 = input.map((r) => typeof r === "string" ? { text: r } : { ...r, text: String(r.text ?? "") });
+  const lines = [[]];
+  runs3.forEach((r, i) => {
+    for (const m of r.text.matchAll(/\n|[^\S\n]+|[\u0590-\u08FF\uFB1D-\uFDFF\uFE70-\uFEFF]+|[^\s\u0590-\u08FF\uFB1D-\uFDFF\uFE70-\uFEFF]+/g)) {
+      if (m[0] === "\n") lines.push([]);
+      else lines[lines.length - 1].push({ run: i, text: m[0] });
+    }
+  });
+  const out = [];
+  lines.forEach((line, li) => {
+    if (li > 0) out.push({ ...runs3[0], text: "\n" });
+    const text = line.map((t) => t.text).join("");
+    if (!text) return;
+    const { levels } = b.getEmbeddingLevels(text, rtl ? "rtl" : "ltr");
+    let at = 0;
+    const tokens = line.map((t) => {
+      const level = levels[at] ?? 0;
+      let s = t.text;
+      if (level % 2 === 1) {
+        s = [...s].map((ch) => b.getMirroredCharacter(ch) ?? ch).join("");
+        if (!RTL_RE.test(s) && !/^\d/.test(s)) s = [...s].reverse().join("");
+      }
+      at += t.text.length;
+      return { ...t, text: s, level };
+    });
+    const max = Math.max(...tokens.map((t) => t.level));
+    const minOdd = Math.min(...tokens.map((t) => t.level % 2 === 1 ? t.level : Infinity));
+    for (let lvl = max; lvl >= minOdd && Number.isFinite(minOdd); lvl--) {
+      for (let i = 0; i < tokens.length; ) {
+        if (tokens[i].level < lvl) {
+          i++;
+          continue;
+        }
+        let j = i;
+        while (j < tokens.length && tokens[j].level >= lvl) j++;
+        tokens.splice(i, j - i, ...tokens.slice(i, j).reverse());
+        i = j;
+      }
+    }
+    for (const t of tokens) {
+      const last = out[out.length - 1];
+      const src = runs3[t.run];
+      if (last && last.text !== "\n" && last.__run === t.run) last.text += t.text;
+      else out.push({ ...src, text: t.text, __run: t.run });
+    }
+  });
+  for (const r of out) delete r.__run;
+  return out;
+}
+
 // src/document/math.ts
 var import_source = require("@ai-matrx/content-ir/source");
 var MATH_OPEN = "\uE100";
@@ -2125,6 +2193,8 @@ var CitationEngine = class {
   /** A note style: every citation becomes a numbered note. */
   isNote;
   numeric;
+  /** APA 7 §8.11: narrative citations join authors with "and", not "&". */
+  apaNarrativeAnd;
   constructor(styleXml, localeXml, items, lang) {
     const byId = new Map(items.map((i) => [i.id, toCslJson(i)]));
     const sys = {
@@ -2135,6 +2205,7 @@ var CitationEngine = class {
     this.engine = new Engine(sys, styleXml, cslLocaleTag(lang), true);
     this.engine.setOutputFormat("html");
     this.isNote = this.engine.opt.xclass === "note";
+    this.apaNarrativeAnd = /<title>[^<]*(?:\bAPA\b|American Psychological Association)/i.test(styleXml) || /<link[^>]*href="[^"]*\/styles\/apa(?:-[\w-]*)?"[^>]*rel="(?:self|template|independent-parent)"/i.test(styleXml);
     this.numeric = /citation-number/.test(styleXml.slice(styleXml.indexOf("<citation"), styleXml.indexOf("</citation>")));
   }
   /** Cite a group; `noteIndex` is the note number for note styles (0 = in text). */
@@ -2171,7 +2242,8 @@ var CitationEngine = class {
     if (opts.narrative && refs.length === 1 && !this.isNote && !this.numeric) {
       const author = make(items);
       const rest = make(refs.map((r, i) => ({ ...items[i], "author-only": void 0, "suppress-author": true, id: r.key })));
-      const join = () => `${author.label}${rest.label ? ` ${/^\(/.test(rest.label) ? rest.label : `(${rest.label})`}` : ""}`;
+      const who = () => this.apaNarrativeAnd ? author.label.replace(/,? & /g, (m) => m.startsWith(",") ? ", and " : " and ") : author.label;
+      const join = () => `${who()}${rest.label ? ` ${/^\(/.test(rest.label) ? rest.label : `(${rest.label})`}` : ""}`;
       const listeners = [];
       author.listeners.push(() => listeners.forEach((l) => l(join())));
       rest.listeners.push(() => listeners.forEach((l) => l(join())));
@@ -2650,7 +2722,7 @@ function runs(nodes, c, st = {}) {
   for (const n of nodes) {
     switch (n.type) {
       case "text":
-        out.push(tr(n.value));
+        out.push(RTL_RE.test(n.value) ? tr(n.value, { rightToLeft: true }) : tr(n.value));
         break;
       case "strong":
         out.push(...runs(n.children, c, { ...st, bold: true }));
@@ -2802,6 +2874,10 @@ function numberingRef(c, ordered, start) {
   }
   return ref;
 }
+function bidiPara(nodes) {
+  const text = nodes.map((n) => "value" in n ? n.value : "").join("");
+  return isRtlParagraph(text) ? { bidirectional: true } : {};
+}
 function blocks(list, c, o = {}) {
   const out = [];
   for (const b of list) out.push(...block(b, c, o));
@@ -2826,6 +2902,7 @@ function block(b, c, o) {
       const kids = [...b.number ? [new d.TextRun({ text: `${b.number}  `, color: HEX.muted })] : [], ...runs(b.children, c)];
       return [
         new d.Paragraph({
+          ...bidiPara(b.children),
           heading,
           keepNext: true,
           keepLines: true,
@@ -2835,7 +2912,7 @@ function block(b, c, o) {
       ];
     }
     case "paragraph":
-      return [para(c, o, { children: runs(b.children, c), widowControl: true })];
+      return [para(c, o, { children: runs(b.children, c), widowControl: true, ...bidiPara(b.children) })];
     case "list": {
       const level = o.listLevel ?? 0;
       const ref = numberingRef(c, b.ordered, b.start);
@@ -2852,7 +2929,14 @@ function block(b, c, o) {
             })
           );
         } else {
-          out.push(new d.Paragraph({ numbering: { reference: ref, level, instance }, children: firstRuns, widowControl: true }));
+          out.push(
+            new d.Paragraph({
+              numbering: { reference: ref, level, instance },
+              children: firstRuns,
+              widowControl: true,
+              ...first?.type === "paragraph" ? bidiPara(first.children) : {}
+            })
+          );
         }
         const tail = first && first.type !== "paragraph" ? [first, ...rest] : rest;
         for (const child of tail) {
@@ -2894,6 +2978,7 @@ function block(b, c, o) {
             borders: { top: none, left: none, right: none, bottom: head ? { ...border, size: 10, color: HEX.heading } : border },
             children: [
               new d.Paragraph({
+                ...bidiPara(cells[i] ?? []),
                 alignment: alignOf(i),
                 keepNext: keep && !last,
                 children: runs(cells[i] ?? [], c, head ? { bold: true } : {})
@@ -3107,7 +3192,11 @@ async function documentToDocx(tree, options = {}) {
     numbering: { config: [...c.numbering.entries()].map(([reference, levels2]) => ({ reference, levels: levels2 })) },
     styles: {
       default: {
-        document: { run: { font: "Cambria", size, color: HEX.text }, paragraph: { spacing: { after: 140, line: 300 } } }
+        // East-Asian text gets a CJK face (Word and LibreOffice substitute an installed one).
+        document: {
+          run: { font: { ascii: "Cambria", hAnsi: "Cambria", cs: "Arial", eastAsia: "Yu Gothic" }, size, sizeComplexScript: size, color: HEX.text },
+          paragraph: { spacing: { after: 140, line: 300 } }
+        }
       },
       paragraphStyles: [
         heading("Heading1", "Heading 1", s.fontSize * 1.8, 0),
@@ -3250,7 +3339,7 @@ function renderInlineHtml(nodes, x) {
       case "cite":
         return n.resolved ? `<a class="doc-cite" href="${(0, import_html_escape.escapeHtml)(href("ref-" + (n.keys[0] ?? ""), x))}">${(0, import_html_escape.escapeHtml)(n.label)}</a>` : `<span class="doc-cite doc-cite-missing">${(0, import_html_escape.escapeHtml)(n.label)}</span>`;
       case "math":
-        return mathHtml(n.tex, n.display, x);
+        return mathHtml(n.tex, n.display, x, true);
       case "noteRef":
         return `<sup class="doc-note-ref"><a href="${(0, import_html_escape.escapeHtml)(href(`note-${n.n}`, x))}" id="noteref-${n.n}">${n.n}</a></sup>`;
       default:
@@ -3258,10 +3347,11 @@ function renderInlineHtml(nodes, x) {
     }
   }).join("");
 }
-function mathHtml(tex, display, x) {
+function mathHtml(tex, display, x, inText = false) {
   const m = x.tree.math.get(mathKey(tex, display));
   if (m?.mathml && !m.error) {
-    return display ? `<div class="doc-math-display">${m.mathml}</div>` : `<span class="doc-math">${m.mathml}</span>`;
+    if (!display) return `<span class="doc-math">${m.mathml}</span>`;
+    return inText ? `<span class="doc-math-display">${m.mathml}</span>` : `<div class="doc-math-display">${m.mathml}</div>`;
   }
   return `<code class="doc-math-tex">${(0, import_html_escape.escapeHtml)(display ? `$$${tex}$$` : `$${tex}$`)}</code>`;
 }
@@ -3292,10 +3382,10 @@ function renderBlockHtml(b, x) {
   switch (b.type) {
     case "heading": {
       const num = b.number ? `<span class="doc-heading-number">${b.number}</span> ` : "";
-      return `<h${b.level} id="${(0, import_html_escape.escapeHtml)(b.id)}">${num}${renderInlineHtml(b.children, x)}</h${b.level}>`;
+      return `<h${b.level} id="${(0, import_html_escape.escapeHtml)(b.id)}" dir="auto">${num}${renderInlineHtml(b.children, x)}</h${b.level}>`;
     }
     case "paragraph":
-      return `<p>${renderInlineHtml(b.children, x)}</p>`;
+      return `<p dir="auto">${renderInlineHtml(b.children, x)}</p>`;
     case "list": {
       const tag = b.ordered ? "ol" : "ul";
       const start = b.ordered && b.start !== 1 ? ` start="${b.start}"` : "";
@@ -3303,12 +3393,12 @@ function renderBlockHtml(b, x) {
       const items = b.items.map((i) => {
         const box = i.checked === null ? "" : `<span class="doc-task-box${i.checked ? " is-checked" : ""}" role="img" aria-label="${i.checked ? "done" : "not done"}"></span>`;
         const inner = i.children.length === 1 && i.children[0]?.type === "paragraph" ? renderInlineHtml(i.children[0].children, x) : renderBlocksHtml(i.children, x);
-        return `<li${i.checked !== null ? ' class="doc-task"' : ""}>${box}${inner}</li>`;
+        return `<li dir="auto"${i.checked !== null ? ' class="doc-task"' : ""}>${box}${inner}</li>`;
       }).join("");
       return `<${tag}${start}${task ? ' class="doc-task-list"' : ""}>${items}</${tag}>`;
     }
     case "blockquote":
-      return `<blockquote>${renderBlocksHtml(b.children, x)}</blockquote>`;
+      return `<blockquote dir="auto">${renderBlocksHtml(b.children, x)}</blockquote>`;
     case "code":
       return `<pre class="doc-code"${b.lang ? ` data-language="${(0, import_html_escape.escapeHtml)(b.lang)}"` : ""}><code>${(0, import_html_escape.escapeHtml)(
         b.value
@@ -3318,8 +3408,8 @@ function renderBlockHtml(b, x) {
         const a = b.align[i];
         return a ? ` style="text-align:${a}"` : "";
       };
-      const head = `<thead><tr>${b.header.map((c, i) => `<th${al(i)}>${renderInlineHtml(c, x)}</th>`).join("")}</tr></thead>`;
-      const body = `<tbody>${b.rows.map((r) => `<tr>${r.map((c, i) => `<td${al(i)}>${renderInlineHtml(c, x)}</td>`).join("")}</tr>`).join("")}</tbody>`;
+      const head = `<thead><tr>${b.header.map((c, i) => `<th dir="auto"${al(i)}>${renderInlineHtml(c, x)}</th>`).join("")}</tr></thead>`;
+      const body = `<tbody>${b.rows.map((r) => `<tr>${r.map((c, i) => `<td dir="auto"${al(i)}>${renderInlineHtml(c, x)}</td>`).join("")}</tr>`).join("")}</tbody>`;
       const id = b.caption?.id ? ` id="tbl:${(0, import_html_escape.escapeHtml)(b.caption.id)}"` : "";
       return `<div class="doc-table-wrap"${id}><table>${b.caption ? captionHtml(b.caption, "caption", x) : ""}${head}${body}</table></div>`;
     }
@@ -3423,7 +3513,7 @@ table caption{caption-side:top;text-align:left;margin-bottom:4pt;}
 .doc-figure figcaption{margin-top:4pt;}
 .doc-image-missing{border:1px dashed var(--matrx-print-rule,#d1d5db);padding:16pt;color:var(--matrx-print-muted,#6b7280);}
 .doc-inline-img{max-height:1.2em;vertical-align:middle;}
-.doc-math-display{margin:0.6em 0 0.9em;text-align:center;break-inside:avoid;overflow-x:auto;}
+.doc-math-display{display:block;margin:0.6em 0 0.9em;text-align:center;break-inside:avoid;overflow-x:auto;}
 .doc-math-display math{font-size:1.15em;}
 .doc-math-tex{color:var(--matrx-print-danger,#b91c1c);}
 .doc-task-list{list-style:none;padding-left:0.2em;}
@@ -3502,7 +3592,7 @@ blockquote{margin:0 0 0.8em;padding-left:1em;border-left:3px solid #d1d5db;color
 .doc-cover-logo{max-height:5em;}
 .doc-cover-title{font-size:2em;}
 .matrx-page-break{page-break-after:always;break-after:page;}
-.doc-math-display{margin:0.6em 0;text-align:center;}
+.doc-math-display{display:block;margin:0.6em 0;text-align:center;}
 `.trim();
 function extOf(mime) {
   return mime === "image/png" ? "png" : mime === "image/gif" ? "gif" : "jpg";
@@ -3789,7 +3879,7 @@ function breakRuns(input, width, t) {
     for (let guard = 0; guard < 50 && atoms.length; guard++) {
       const gw = atoms.reduce((n, a) => n + w(a), 0);
       const sw = pendingSpace && line > 0 ? w(pendingSpace) : 0;
-      if (line + sw + gw <= W || line === 0 && atoms.length !== 1) {
+      if (line + sw + gw <= W || line === 0 && (atoms.length !== 1 || gw <= width)) {
         if (sw && pendingSpace) out.push(pendingSpace);
         out.push(...atoms);
         line += sw + gw;
@@ -3913,11 +4003,14 @@ function withFallbackFonts(nodes, c) {
       const fb = fallbackFor(cp);
       if (fb) {
         flush();
-        out.push({ ...n, text: ch, font: fb, bold: false, italics: false });
+        const prev = out[out.length - 1];
+        if (prev && typeof prev === "object" && prev.font === fb && prev.__fb) prev.text = `${prev.text}${ch}`;
+        else out.push({ ...n, text: ch, font: fb, bold: false, italics: false, __fb: true });
       } else dropped = true;
     }
     flush();
   }
+  for (const o of out) if (typeof o === "object") delete o.__fb;
   if (dropped && !c.glyphs.announced) {
     c.glyphs.announced = true;
     c.tree.notices.push({
@@ -3944,8 +4037,13 @@ function drawable(text, c) {
 }
 function bodyRuns(nodes, c, width = c.width) {
   const r = runs2(nodes, c);
-  return c.typo ? breakRuns(r, width, c.typo) : r;
+  const text = plain(nodes);
+  if (!RTL_RE.test(text)) return c.typo ? breakRuns(r, width, c.typo) : r;
+  const typo = c.typo ?? (c.faces ? { split: (w) => [w], measure: c.faces.measure } : void 0);
+  const lines = typo ? breakRuns(r, width, typo) : r;
+  return visualOrder(lines, isRtlParagraph(text));
 }
+var alignFor = (nodes) => isRtlParagraph(plain(nodes)) ? { alignment: "right" } : {};
 function runs2(nodes, c, st = {}) {
   const out = [];
   const base = (raw) => {
@@ -4137,6 +4235,7 @@ function estimateLines(text, c, size = c.s.fontSize) {
 function columnWidths(b, n, size, c) {
   const measure = (text, bold) => c.faces ? c.faces.measure(text, { bold, fontSize: size }) : text.length * size * 0.5;
   const pad = 10;
+  const slack = 1;
   const min = new Array(n).fill(0);
   const max = new Array(n).fill(0);
   const rows = [[b.header, true], ...b.rows.map((r) => [r, false])];
@@ -4144,8 +4243,8 @@ function columnWidths(b, n, size, c) {
     for (let i = 0; i < n; i++) {
       const t = plain(r[i] ?? []);
       const longest = t.split(/\s+/).reduce((m, w) => Math.max(m, measure(w, head)), 0);
-      min[i] = Math.max(min[i], Math.min(longest, c.width * 0.35) + pad);
-      max[i] = Math.max(max[i], measure(t, head) + pad);
+      min[i] = Math.max(min[i], Math.min(longest * slack, c.width * 0.35) + pad);
+      max[i] = Math.max(max[i], measure(t, head) * slack + pad);
     }
   }
   const avail = c.width;
@@ -4209,8 +4308,11 @@ function block2(b, c) {
   switch (b.type) {
     case "heading": {
       const rel = Math.min(6, b.level - c.minLevel + 1);
+      const rtlHeading = RTL_RE.test(b.plain);
+      const headRuns = rtlHeading ? visualOrder(runs2(b.children, c), isRtlParagraph(b.plain)) : runs2(b.children, c);
       const node = {
-        text: [...b.number ? [{ text: `${b.number}  `, color: COLOR.muted }] : [], ...runs2(b.children, c)],
+        ...rtlHeading ? alignFor(b.children) : {},
+        text: [...b.number ? [{ text: `${b.number}  `, color: COLOR.muted }] : [], ...headRuns],
         style: `h${Math.min(6, b.level)}`,
         headlineLevel: b.level,
         id: b.id
@@ -4224,12 +4326,12 @@ function block2(b, c) {
     }
     case "paragraph": {
       const text = bodyRuns(b.children, c);
-      const p = { text, style: "p" };
+      const p = { text, style: "p", ...alignFor(b.children) };
       return estimateLines(plain(b.children), c) <= 5 ? { stack: [p], unbreakable: true } : p;
     }
     case "list": {
       const items = b.items.map((it) => {
-        const content = it.children.length === 1 && it.children[0]?.type === "paragraph" ? [{ text: bodyRuns(it.children[0].children, c, c.width - 20 - (it.checked !== null ? 8 : 0)), margin: [0, 0, 0, 2] }] : blocks2(it.children, { ...c, width: c.width - 18 });
+        const content = it.children.length === 1 && it.children[0]?.type === "paragraph" ? [{ text: bodyRuns(it.children[0].children, c, c.width - 20 - (it.checked !== null ? 8 : 0)), margin: [0, 0, 0, 2], ...alignFor(it.children[0].children) }] : blocks2(it.children, { ...c, width: c.width - 18 });
         if (it.checked !== null) return { ...taskItem(it.checked, content, c), listType: "none" };
         return content.length === 1 ? content[0] : { stack: content };
       });
@@ -4282,15 +4384,24 @@ function block2(b, c) {
     case "table": {
       const n = Math.max(b.header.length, ...b.rows.map((r) => r.length), 1);
       const size = c.s.fontSize * 0.92;
+      const widths = columnWidths(b, n, size, c);
+      const cellRuns = (inl, i, head) => {
+        const r = runs2(inl, c).map((x) => typeof x === "object" ? { ...x, fontSize: x.fontSize ?? size, ...head ? { bold: true } : {} } : x);
+        const w = (widths[i] ?? c.width) - 10;
+        if (RTL_RE.test(plain(inl))) {
+          const typo = c.typo ?? (c.faces ? { split: (x) => [x], measure: c.faces.measure } : void 0);
+          return visualOrder(typo ? breakRuns(r, w, typo) : r, isRtlParagraph(plain(inl)));
+        }
+        return c.typo && !head ? breakRuns(r, w, c.typo) : r;
+      };
       const cell = (inl, i, head) => ({
-        text: runs2(inl ?? [], c),
+        text: cellRuns(inl ?? [], i, head),
         bold: head,
-        alignment: b.align[i] ?? "left",
+        alignment: b.align[i] ?? (isRtlParagraph(plain(inl ?? [])) ? "right" : "left"),
         fontSize: size,
         color: head ? COLOR.heading : COLOR.text
       });
       const row = (r, head) => Array.from({ length: n }, (_, i) => cell(r[i], i, head));
-      const widths = columnWidths(b, n, size, c);
       const lineH = size * 1.3;
       const pageH = c.pageHeight;
       const rowHeight = (r) => Math.max(...r.map((inl, i) => estimateLines(plain(inl), { ...c, width: (widths[i] ?? c.width) - 10 }, size) * lineH)) + 8;
@@ -4536,12 +4647,16 @@ async function loadTypography(tree, loaders = {}) {
   const base = await loadFaces(vfs, s.fontSize);
   let needEmoji = false;
   let needSymbols = false;
+  let needArabic = false;
+  let needHebrew = false;
   const cjk = /* @__PURE__ */ new Set();
   for (const ch of documentText(tree)) {
     const cp = ch.codePointAt(0);
     if (cp < 32 || cp === 8205 || cp === 65039 || base.body(cp)) continue;
     if (EMOJI_CP.test(ch)) needEmoji = true;
-    else if (isCjk(cp)) cjk.add(cp);
+    else if (isCjk(cp) || isHangul(cp)) cjk.add(cp);
+    else if (isArabic(cp)) needArabic = true;
+    else if (isHebrew(cp)) needHebrew = true;
     else needSymbols = true;
   }
   const fallbacks = [];
@@ -4553,6 +4668,14 @@ async function loadTypography(tree, loaders = {}) {
     const b = await registerFace("NotoSymbols", "NotoSansMath.woff", loaders.symbols ?? defaultSymbolFontLoader, tree, "symbols");
     if (b) fallbacks.push({ font: "NotoSymbols", bytes: b });
   }
+  if (needArabic) {
+    const b = await registerFace("NotoArabic", "NotoSansArabic.woff", loaders.arabic ?? urlLoader(NOTO_ARABIC_URL), tree, "Arabic");
+    if (b) fallbacks.push({ font: "NotoArabic", bytes: b });
+  }
+  if (needHebrew) {
+    const b = await registerFace("NotoHebrew", "NotoSansHebrew.woff", loaders.hebrew ?? urlLoader(NOTO_HEBREW_URL), tree, "Hebrew");
+    if (b) fallbacks.push({ font: "NotoHebrew", bytes: b });
+  }
   if (cjk.size) {
     let subsets = [];
     try {
@@ -4568,7 +4691,7 @@ async function loadTypography(tree, loaders = {}) {
     if (!subsets.length) {
       tree.notices.push({
         code: "glyph-substituted",
-        message: "The Japanese / Chinese font could not be loaded, so those characters were left out of the PDF (Word, EPUB and HTML keep them).",
+        message: "The Chinese / Japanese / Korean font could not be loaded, so those characters were left out of the PDF (Word, EPUB and HTML keep them).",
         remedy: "Check the internet connection (the font comes from cdn.jsdelivr.net) and export again."
       });
     }
@@ -4596,6 +4719,12 @@ var urlLoader = (url) => async () => {
 var defaultEmojiFontLoader = urlLoader(NOTO_EMOJI_URL);
 var defaultSymbolFontLoader = urlLoader(NOTO_SYMBOLS_URL);
 var NOTO_CJK_BASE = "https://cdn.jsdelivr.net/npm/@fontsource/noto-sans-jp@5.3.0";
+var NOTO_KR_BASE = "https://cdn.jsdelivr.net/npm/@fontsource/noto-sans-kr@5.3.0";
+var NOTO_ARABIC_URL = "https://cdn.jsdelivr.net/npm/@fontsource/noto-sans-arabic@5.3.0/files/noto-sans-arabic-arabic-400-normal.woff";
+var NOTO_HEBREW_URL = "https://cdn.jsdelivr.net/npm/@fontsource/noto-sans-hebrew@5.3.0/files/noto-sans-hebrew-hebrew-400-normal.woff";
+var isHangul = (cp) => cp >= 44032 && cp <= 55215 || cp >= 4352 && cp <= 4607 || cp >= 12592 && cp <= 12687;
+var isArabic = (cp) => cp >= 1536 && cp <= 1791 || cp >= 1872 && cp <= 1919 || cp >= 2208 && cp <= 2303 || cp >= 64336 && cp <= 65023 || cp >= 65136 && cp <= 65279;
+var isHebrew = (cp) => cp >= 1424 && cp <= 1535 || cp >= 64285 && cp <= 64335;
 function parseUnicodeRangeCss(css) {
   const out = [];
   for (const block3 of css.split("@font-face").slice(1)) {
@@ -4611,19 +4740,25 @@ function parseUnicodeRangeCss(css) {
   }
   return out;
 }
-var defaultCjkFontLoader = async (cps) => {
+async function unicodeRangeFaces(base, cps) {
   if (typeof fetch !== "function" || cps.length === 0) return [];
-  const res = await fetch(`${NOTO_CJK_BASE}/400.css`);
+  const res = await fetch(`${base}/400.css`);
   if (!res.ok) return [];
   const faces = parseUnicodeRangeCss(await res.text()).filter((f) => cps.some((cp) => f.ranges.some(([a, b]) => cp >= a && cp <= b)));
   const out = [];
   await Promise.all(
     faces.map(async (f) => {
-      const r = await fetch(`${NOTO_CJK_BASE}/files/${f.file}.woff`);
+      const r = await fetch(`${base}/files/${f.file}.woff`);
       if (r.ok) out.push({ name: f.file, bytes: new Uint8Array(await r.arrayBuffer()) });
     })
   );
   return out;
+}
+var defaultCjkFontLoader = async (cps) => {
+  const kr = cps.filter(isHangul);
+  const jp = cps.filter((cp) => !isHangul(cp));
+  const [a, b] = await Promise.all([unicodeRangeFaces(NOTO_KR_BASE, kr), unicodeRangeFaces(NOTO_CJK_BASE, jp)]);
+  return [...a, ...b];
 };
 var isCjk = (cp) => cp >= 11904 && cp <= 40959 || cp >= 44032 && cp <= 55215 || cp >= 63744 && cp <= 64255 || cp >= 65072 && cp <= 65103 || cp >= 65280 && cp <= 65519 || cp >= 131072 && cp <= 201551;
 var EMOJI_CP = /\p{Extended_Pictographic}|\p{Regional_Indicator}|\u20E3/u;
@@ -4792,6 +4927,70 @@ function removeThinkingContent(content) {
   return result.trim();
 }
 var PAGE_BREAK_HTML = `<div class="${PAGE_BREAK_CLASS}" role="separator" aria-label="${PAGE_BREAK_LABEL}" data-label="${PAGE_BREAK_LABEL}"></div>`;
+var CODE_ARTIFACT_LANGUAGE = {
+  html: "html",
+  react: "tsx",
+  svg: "svg",
+  mermaid: "mermaid",
+  diff: "diff",
+  iframe: "html",
+  code: ""
+};
+function parseTagAttributes(tag) {
+  const attrs = {};
+  for (const m of tag.matchAll(/([a-zA-Z_][\w-]*)\s*=\s*(?:"([^"]*)"|'([^']*)')/g)) {
+    attrs[m[1].toLowerCase()] = m[2] ?? m[3] ?? "";
+  }
+  return attrs;
+}
+function unwrapArtifactsToMarkdown(markdown) {
+  if (!/<artifact\b/i.test(markdown)) return markdown;
+  return markdown.split(/(^[ \t]*(?:`{3,}|~{3,})[\s\S]*?^[ \t]*(?:`{3,}|~{3,})[ \t]*$)/m).map(
+    (part, i) => i % 2 === 1 ? part : part.replace(/<artifact\b((?:[^>"']|"[^"]*"|'[^']*')*?)>([\s\S]*?)<\/artifact>/gi, (_m, tag, body) => {
+      const attrs = parseTagAttributes(tag);
+      const type = (attrs.type ?? "").toLowerCase();
+      const trimmed = body.replace(/^\n+|\n+$/g, "");
+      if (!trimmed.trim()) return "";
+      if (/^\s*(`{3,}|~{3,})/.test(trimmed)) return `
+
+${trimmed}
+
+`;
+      if (type in CODE_ARTIFACT_LANGUAGE) {
+        const lang = CODE_ARTIFACT_LANGUAGE[type] || attrs.language || attrs.lang || "";
+        return `
+
+\`\`\`\`${lang}
+${trimmed}
+\`\`\`\`
+
+`;
+      }
+      if (looksLikeJson(trimmed)) return `
+
+\`\`\`\`json
+${JSON.stringify(JSON.parse(trimmed), null, 2)}
+\`\`\`\`
+
+`;
+      return `
+
+${trimmed}
+
+`;
+    })
+  ).join("");
+}
+function looksLikeJson(body) {
+  const t = body.trim();
+  if (!/^[[{]/.test(t)) return false;
+  try {
+    JSON.parse(t);
+    return true;
+  } catch {
+    return false;
+  }
+}
 var LINK_URL = String.raw`((?:[^()\s]|\([^()\s]*\))+)(?:\s+"[^"]*")?`;
 var IMAGE_RE = new RegExp(String.raw`!\[([^\]]*)\]\(` + LINK_URL + String.raw`\)`, "g");
 var LINK_RE = new RegExp(String.raw`\[([^\]]+)\]\(` + LINK_URL + String.raw`\)`, "g");
@@ -4878,7 +5077,7 @@ function uniqueId(ctx, base) {
   return id;
 }
 var CITE_GROUP_RE = /\[(-?@[^[\]]+)\]/g;
-var BARE_RE = /(^|[^\w@.\\])@((?:fig|tbl|sec|eq):[\w-]+|[A-Za-z0-9_](?:[\w:.#$%&+?<>~/-]*\w)?)/g;
+var BARE_RE = /(^|[^\w@.\\])@((?:fig|tbl|sec|eq):[\w-]+|[A-Za-z0-9_](?:[\w:.#$%&+?<>~/-]*\w)?)(?: \[([^\[\]@]+)\])?/g;
 function scanText(value, ctx) {
   const out = [];
   const pushText = (s) => {
@@ -4899,8 +5098,8 @@ function scanText(value, ctx) {
         last2 = start + key.length + 1;
       } else if (ctx.refs.has(key)) {
         pushText(s.slice(last2, start));
-        out.push(cite([{ key, suppressAuthor: false }], ctx, true));
-        last2 = start + key.length + 1;
+        out.push(cite([{ key, locator: m[3]?.trim(), suppressAuthor: false }], ctx, true));
+        last2 = start + key.length + 1 + (m[3] !== void 0 ? m[3].length + 3 : 0);
       }
     }
     pushText(s.slice(last2));
@@ -5246,7 +5445,7 @@ function parseDocument(source, options = {}) {
   const settings = resolveSettings(fm.data, sink, options);
   for (const n of sink) notify(n);
   const body = removeThinkingContent(fm.body);
-  const lifted = extractMath(body);
+  const lifted = extractMath(unwrapArtifactsToMarkdown(body));
   const root = (0, import_mdast_util_from_markdown.fromMarkdown)(prepass(lifted.text), {
     extensions: [(0, import_micromark_extension_gfm.gfm)()],
     mdastExtensions: [(0, import_mdast_util_gfm.gfmFromMarkdown)()]