@ai-matrx/print 0.8.0 → 0.8.2

package/CHANGELOG.md

@@ -1,5 +1,24 @@
 # Changelog
 
+## 0.8.2
+
+- **PDF font fallback covers Japanese / Chinese / Korean.** Characters the
+  body, emoji and symbol faces cannot draw now fall back to Noto Sans JP
+  (SIL OFL, fontsource) — loading only the unicode-range subsets the document
+  uses (`fonts.cjk` loader; default jsDelivr). Found on a real conversation
+  whose citations carried Japanese source titles. A glyph is announced only
+  when no face in the chain can draw it.
+- Consumer action: none — refresh to latest.
+
+## 0.8.1
+
+- **Fixed (Word):** exporting DOCX laid out a whole PDF to number the table of
+  contents even when the document has no table of contents; now only with a
+  TOC, and a failed measuring pass never fails the export — the TOC lists every
+  heading and Word fills in the page numbers on open (`toc-pages-deferred`
+  notice).
+- Consumer action: none — refresh to latest.
+
 ## 0.8.0
 
 - **Citations through a real CSL processor.** citeproc-js (the engine Zotero

package/dist/document.cjs

@@ -35,12 +35,14 @@ __export(document_exports, {
   CitationEngine: () => CitationEngine,
   DEFAULT_FOOTER: () => DEFAULT_FOOTER,
   DOCUMENT_FORMATS: () => DOCUMENT_FORMATS,
+  NOTO_CJK_BASE: () => NOTO_CJK_BASE,
   NOTO_EMOJI_URL: () => NOTO_EMOJI_URL,
   NOTO_SYMBOLS_URL: () => NOTO_SYMBOLS_URL,
   PAGE_SIZES: () => PAGE_SIZES,
   buildPdfDefinition: () => buildPdfDefinition,
   countUnplacedWords: () => countUnplacedWords,
   cslStyleId: () => cslStyleId,
+  defaultCjkFontLoader: () => defaultCjkFontLoader,
   defaultCslFetcher: () => defaultCslFetcher,
   defaultEmojiFontLoader: () => defaultEmojiFontLoader,
   defaultImageLoader: () => defaultImageLoader,
@@ -63,6 +65,7 @@ __export(document_exports, {
   parseLength: () => parseLength,
   parseLocator: () => parseLocator,
   parseMathML: () => parseMathML,
+  parseUnicodeRangeCss: () => parseUnicodeRangeCss,
   plainText: () => plainText,
   prepareDocument: () => prepareDocument,
   renderBlocksHtml: () => renderBlocksHtml,
@@ -4533,12 +4536,13 @@ async function loadTypography(tree, loaders = {}) {
   const base = await loadFaces(vfs, s.fontSize);
   let needEmoji = false;
   let needSymbols = 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 needSymbols = true;
-    if (needEmoji && needSymbols) break;
   }
   const fallbacks = [];
   if (needEmoji) {
@@ -4549,6 +4553,26 @@ 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 (cjk.size) {
+    let subsets = [];
+    try {
+      subsets = await (loaders.cjk ?? defaultCjkFontLoader)([...cjk]);
+    } catch {
+      subsets = [];
+    }
+    for (const sub of subsets) {
+      const font = `NotoCJK_${sub.name.replace(/[^\w]/g, "_")}`;
+      const b = await registerFace(font, `${sub.name}.woff`, async () => sub.bytes, tree, "Japanese / Chinese");
+      if (b) fallbacks.push({ font, bytes: b });
+    }
+    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).",
+        remedy: "Check the internet connection (the font comes from cdn.jsdelivr.net) and export again."
+      });
+    }
+  }
   const faces = fallbacks.length ? await loadFaces(vfs, s.fontSize, fallbacks) : base;
   if (!s.hyphenate) return { faces };
   const split = await loadHyphenator(s.lang);
@@ -4571,6 +4595,37 @@ 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";
+function parseUnicodeRangeCss(css) {
+  const out = [];
+  for (const block3 of css.split("@font-face").slice(1)) {
+    const file = /url\(\.\/files\/([\w-]+)\.woff\)/.exec(block3)?.[1];
+    const range = /unicode-range:\s*([^;]+);/.exec(block3)?.[1];
+    if (!file || !range) continue;
+    const ranges = range.split(",").map((r) => {
+      const [a, b] = r.trim().replace(/^U\+/i, "").split("-");
+      const lo = parseInt(a, 16);
+      return [lo, b ? parseInt(b, 16) : lo];
+    });
+    out.push({ file, ranges });
+  }
+  return out;
+}
+var defaultCjkFontLoader = async (cps) => {
+  if (typeof fetch !== "function" || cps.length === 0) return [];
+  const res = await fetch(`${NOTO_CJK_BASE}/400.css`);
+  if (!res.ok) return [];
+  const faces = parseUnicodeRangeCss(await res.text()).filter((f) => cps.some((cp) => f.ranges.some(([a, b]) => cp >= a && cp <= b)));
+  const out = [];
+  await Promise.all(
+    faces.map(async (f) => {
+      const r = await fetch(`${NOTO_CJK_BASE}/files/${f.file}.woff`);
+      if (r.ok) out.push({ name: f.file, bytes: new Uint8Array(await r.arrayBuffer()) });
+    })
+  );
+  return out;
+};
+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;
 function documentText(tree) {
   const s = tree.settings;
@@ -5548,7 +5603,21 @@ async function exportDocument(source, format, options = {}) {
       bytes = r.bytes;
       pageCount = r.pageCount;
     } else if (format === "docx") {
-      const pages = tree.toc.length ? await headingPages(tree, images, options.fonts) : void 0;
+      const hasToc = tree.sections.some((sec) => sec.blocks.some((blk) => blk.type === "toc"));
+      let pages;
+      if (hasToc) {
+        try {
+          pages = await headingPages(tree, images, options.fonts);
+        } catch (err) {
+          const n = {
+            code: "toc-pages-deferred",
+            message: `The Word file's table of contents lists every heading; its page numbers fill in when the file is opened (they could not be measured here: ${err.message}).`,
+            remedy: "Open the file in Word (or update fields) to show the page numbers."
+          };
+          tree.notices.push(n);
+          options.onNotice?.(n);
+        }
+      }
       bytes = await documentToDocx(tree, { images, headingPages: pages });
     } else bytes = await documentToEpub(tree, { images });
   }