@bendyline/squisq-formats 2.2.0 → 2.3.1

package/NOTICE.md

@@ -6,14 +6,13 @@ Third-party components remain under their respective license terms.
 
 ## Runtime dependencies
 
-| Package           | Version | License                 | Repository                          |
-| ----------------- | ------- | ----------------------- | ----------------------------------- |
-| @bendyline/squisq | 2.1.0   | MIT                     | https://github.com/bendyline/squisq |
-| @pdf-lib/upng     | 1.0.1   | MIT                     | https://github.com/Hopding/upng     |
-| @xmldom/xmldom    | 0.9.10  | MIT                     | https://github.com/xmldom/xmldom    |
-| jszip             | 3.10.1  | MIT OR GPL-3.0-or-later | https://github.com/Stuk/jszip       |
-| pdf-lib           | 1.17.1  | MIT                     | https://pdf-lib.js.org              |
-| pdfjs-dist        | 4.10.38 | Apache-2.0              | https://mozilla.github.io/pdf.js    |
+| Package        | Version | License                 | Repository                       |
+| -------------- | ------- | ----------------------- | -------------------------------- |
+| @pdf-lib/upng  | 1.0.1   | MIT                     | https://github.com/Hopding/upng  |
+| @xmldom/xmldom | 0.9.10  | MIT                     | https://github.com/xmldom/xmldom |
+| jszip          | 3.10.1  | MIT OR GPL-3.0-or-later | https://github.com/Stuk/jszip    |
+| pdf-lib        | 1.17.1  | MIT                     | https://pdf-lib.js.org           |
+| pdfjs-dist     | 4.10.38 | Apache-2.0              | https://mozilla.github.io/pdf.js |
 
 Squisq uses jszip under its MIT license option. Copyright and complete license
 texts for these dependencies are included in their respective npm

package/README.md

@@ -299,7 +299,7 @@ Also exported: `createRegistry` / `defaultRegistry` / `defaultFormats`, `BUILTIN
 
 The package root re-exports the common converters; `./container`, the plain-HTML/bundle functions, `docxToContainer`, `pdfToContainer`, `PdfPageSize`, and the image utilities are subpath-only.
 
-See the full [API Reference](../../docs/API.md#bendylinesquisq-formats) for every signature and options interface.
+See the full [API Reference](https://github.com/bendyline/squisq/blob/main/docs/API.md#bendylinesquisq-formats) for every signature and options interface.
 
 ## Related Packages
 

package/THIRD_PARTY_LICENSES.txt

@@ -0,0 +1,8 @@
+THIRD-PARTY LICENSES FOR @bendyline/squisq-formats
+
+Generated from the actual esbuild input graph. Package-local license, copying,
+and notice files are reproduced verbatim. When an npm tarball omits its
+repository license, the pinned upstream copy is vendored and identified below.
+
+COMPONENTS
+

package/dist/{chunk-PJXJI2LY.js → chunk-GYIVES2E.js}

@@ -1046,7 +1046,20 @@ function hexToRgb(hex) {
 // src/pdf/import.ts
 import { markdownToDoc } from "@bendyline/squisq/doc";
 import { stringifyMarkdown } from "@bendyline/squisq/markdown";
+
+// src/pdf/upng.ts
 import UPNG from "@pdf-lib/upng";
+function isUpngEncoder(value) {
+  return typeof value === "object" && value !== null && "encode" in value && typeof value.encode === "function";
+}
+function resolveUpngEncoder(value) {
+  if (isUpngEncoder(value)) return value;
+  if (typeof value !== "object" || value === null || !("default" in value)) return null;
+  return isUpngEncoder(value.default) ? value.default : null;
+}
+var upngEncoder = resolveUpngEncoder(UPNG);
+
+// src/pdf/import.ts
 async function pdfToMarkdownDoc(data, options = {}) {
   options.signal?.throwIfAborted();
   const bytes = data instanceof Blob ? new Uint8Array(await data.arrayBuffer()) : data instanceof ArrayBuffer ? new Uint8Array(data) : data;
@@ -1206,7 +1219,8 @@ function imageDataToPng(img) {
     }
     const owned = new Uint8Array(rgba.byteLength);
     owned.set(rgba);
-    return UPNG.encode([owned.buffer], img.width, img.height, 0);
+    if (!upngEncoder) return null;
+    return upngEncoder.encode([owned.buffer], img.width, img.height, 0);
   } catch {
     return null;
   }
@@ -1255,6 +1269,22 @@ function insertImageBlocks(blocks, blockPages, images) {
   }
   return result;
 }
+function hasInterItemSpace(previous, current) {
+  if (!previous) return false;
+  if (/\s$/.test(previous.str) || /^\s/.test(current.str)) return false;
+  const gap = current.x - (previous.x + previous.width);
+  const referenceHeight = Math.min(previous.height, current.height);
+  return gap > Math.max(0.5, referenceHeight * 0.08);
+}
+function joinTextItems(items) {
+  let text = "";
+  for (let index = 0; index < items.length; index++) {
+    const item = items[index];
+    if (hasInterItemSpace(items[index - 1], item)) text += " ";
+    text += item.str;
+  }
+  return text;
+}
 function configurePdfWorker(workerSrc) {
   _workerSrc = workerSrc;
 }
@@ -1337,7 +1367,7 @@ async function extractTextLines(pdf, options) {
       const fontNames = lineItems.map((i) => i.fontName);
       const fontName = modeStr(fontNames) || "";
       const minX = Math.min(...lineItems.map((i) => i.x));
-      const text = lineItems.map((i) => i.str).join(" ");
+      const text = joinTextItems(lineItems);
       allLines.push({
         items: lineItems,
         y: key,
@@ -1538,9 +1568,13 @@ function isItalicFont(fontName) {
 function buildInlineNodes(line, options) {
   const nodes = [];
   const detectLinksOpt = options.detectLinks !== false;
-  for (const item of line.items) {
+  for (let itemIndex = 0; itemIndex < line.items.length; itemIndex++) {
+    const item = line.items[itemIndex];
     const text = item.str;
     if (!text || text.trim().length === 0) continue;
+    if (hasInterItemSpace(line.items[itemIndex - 1], item)) {
+      nodes.push({ type: "text", value: " " });
+    }
     const bold = isBoldFont(item.fontName);
     const italic = isItalicFont(item.fontName);
     const mono = isMonospaceItem(item);

package/dist/{chunk-JE6LSIHE.js → chunk-T7M52LHF.js}

@@ -167,10 +167,11 @@ ${mode === "static" ? "#squisq-root{display:block}" : ""}
 </head>
 <body>
 <div id="squisq-root"></div>
+<script type="application/json" id="squisq-doc" data-squisq-doc="1">${docJson}</script>
 <script>${escapeForScript(playerScript)}</script>
 <script>
 (function(){
-  var doc = JSON.parse(${JSON.stringify(docJson)});
+  var doc = JSON.parse(document.getElementById("squisq-doc").textContent);
   var images = JSON.parse(${JSON.stringify(imageMapJson)});
   // Re-expand path/basename aliases that share a payload (see
   // buildInlineImageMaps) so \`images\` has every key the resolver may ask for.
@@ -221,10 +222,11 @@ ${mode === "static" ? "#squisq-root{display:block}" : ""}
 </head>
 <body>
 <div id="squisq-root"></div>
+<script type="application/json" id="squisq-doc" data-squisq-doc="1">${docJson}</script>
 <script src="${escapeHtml(playerScriptPath)}"></script>
 <script>
 (function(){
-  var doc = JSON.parse(${JSON.stringify(docJson)});
+  var doc = JSON.parse(document.getElementById("squisq-doc").textContent);
   var images = JSON.parse(${JSON.stringify(imageMapJson)});
   var audio = ${audioMapJson};
   SquisqPlayer.mount(document.getElementById("squisq-root"), doc, {
@@ -1223,6 +1225,7 @@ import {
   sanitizeHtmlNodes as sanitizeHtmlNodes2,
   stringifyMarkdown
 } from "@bendyline/squisq/markdown";
+import { docToMarkdown } from "@bendyline/squisq/doc";
 var HEADINGS = {
   h1: 1,
   h2: 2,
@@ -1409,6 +1412,33 @@ function toHtmlString(data) {
   const bytes = data instanceof Uint8Array ? data : new Uint8Array(data);
   return new TextDecoder("utf-8").decode(bytes);
 }
+function findEmbeddedSquisqDoc(nodes) {
+  for (const node of nodes) {
+    if (!isElement(node)) continue;
+    if (node.tagName.toLowerCase() === "script" && "data-squisq-doc" in node.attributes) {
+      return node;
+    }
+    const nested = findEmbeddedSquisqDoc(node.children);
+    if (nested) return nested;
+  }
+  return void 0;
+}
+function parseEmbeddedSquisqDoc(script) {
+  const version = script.attributes["data-squisq-doc"];
+  if (version !== "1") {
+    throw new Error(`Unsupported embedded Squisq Doc version "${version || "(missing)"}"`);
+  }
+  let value;
+  try {
+    value = JSON.parse(textContent(script));
+  } catch {
+    throw new SyntaxError("Invalid embedded Squisq Doc JSON");
+  }
+  if (value === null || typeof value !== "object" || typeof value.articleId !== "string" || !Number.isFinite(value.duration) || !Array.isArray(value.blocks) || value.audio === null || typeof value.audio !== "object") {
+    throw new TypeError("Embedded Squisq Doc has an invalid shape");
+  }
+  return value;
+}
 function htmlToMarkdownDocSync(html, options = {}) {
   options.signal?.throwIfAborted();
   const maxInputChars = options.maxInputChars ?? 16 * 1024 * 1024;
@@ -1419,6 +1449,8 @@ function htmlToMarkdownDocSync(html, options = {}) {
     throw new RangeError(`HTML exceeds the ${maxInputChars}-character safety limit`);
   }
   let nodes = parseHtmlToNodes(html);
+  const embedded = findEmbeddedSquisqDoc(nodes);
+  if (embedded) return docToMarkdown(parseEmbeddedSquisqDoc(embedded));
   if (options.sanitize !== false) nodes = sanitizeHtmlNodes2(nodes);
   return { type: "document", children: blocksFromNodes(nodes) };
 }

package/dist/html/index.js

@@ -9,7 +9,7 @@ import {
   markdownDocToPlainHtml,
   markdownDocsToHtmlBundle,
   markdownDocsToPlainHtmlBundle
-} from "../chunk-JE6LSIHE.js";
+} from "../chunk-T7M52LHF.js";
 import {
   arrayBufferToBase64DataUrl,
   extractFilename,

package/dist/index.js

@@ -11,7 +11,7 @@ import {
   defaultRegistry,
   prepareConversion,
   resolveConversionLimits
-} from "./chunk-3NKXBZSR.js";
+} from "./chunk-LBHZSY7N.js";
 import {
   inferThemeFromFile
 } from "./chunk-6S6GU3ZG.js";
@@ -62,7 +62,7 @@ import {
   markdownDocToPdf,
   pdfToDoc,
   pdfToMarkdownDoc
-} from "./chunk-PJXJI2LY.js";
+} from "./chunk-GYIVES2E.js";
 import "./chunk-IIQYS2YH.js";
 import {
   collectImagePaths,
@@ -71,7 +71,7 @@ import {
   htmlToMarkdown,
   htmlToMarkdownDoc,
   htmlToMarkdownDocSync
-} from "./chunk-JE6LSIHE.js";
+} from "./chunk-T7M52LHF.js";
 import "./chunk-6RQOV3B3.js";
 import "./chunk-AONELFLA.js";
 export {

package/dist/pdf/index.js

@@ -5,7 +5,7 @@ import {
   pdfToContainer,
   pdfToDoc,
   pdfToMarkdownDoc
-} from "../chunk-PJXJI2LY.js";
+} from "../chunk-GYIVES2E.js";
 import "../chunk-IIQYS2YH.js";
 export {
   configurePdfWorker,

package/dist/registry/index.js

@@ -7,7 +7,7 @@ import {
   defaultRegistry,
   prepareConversion,
   resolveConversionLimits
-} from "../chunk-3NKXBZSR.js";
+} from "../chunk-LBHZSY7N.js";
 import {
   ConversionError
 } from "../chunk-KXOZMWBS.js";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bendyline/squisq-formats",
-  "version": "2.2.0",
+  "version": "2.3.1",
   "description": "Document format converters — DOCX, PDF, OOXML import/export",
   "license": "MIT",
   "author": "Bendyline",
@@ -24,6 +24,9 @@
     "access": "public"
   },
   "type": "module",
+  "engines": {
+    "node": ">=22.14.0"
+  },
   "main": "./dist/index.js",
   "module": "./dist/index.js",
   "types": "./dist/index.d.ts",
@@ -31,7 +34,8 @@
     "dist",
     "!dist/**/*.map",
     "LICENSE",
-    "NOTICE.md"
+    "NOTICE.md",
+    "THIRD_PARTY_LICENSES.txt"
   ],
   "exports": {
     ".": {
@@ -96,14 +100,14 @@
     }
   },
   "scripts": {
-    "build": "tsup",
+    "build": "tsup && node ../../scripts/generate-bundle-licenses.mjs . && node ../../scripts/remove-empty-chunks.mjs dist",
     "dev": "concurrently -n js,dts -c blue,gray -r \"tsup --watch --no-dts --no-clean\" \"tsup --watch --dts-only --no-clean\"",
     "typecheck": "tsc --noEmit"
   },
   "dependencies": {
     "@pdf-lib/upng": "1.0.1",
     "@xmldom/xmldom": "0.9.10",
-    "@bendyline/squisq": "2.2.0",
+    "@bendyline/squisq": "2.3.1",
     "jszip": "3.10.1",
     "pdf-lib": "1.17.1",
     "pdfjs-dist": "4.10.38"