@docen/import-docx 0.0.13 → 0.0.14

package/README.md

@@ -70,11 +70,6 @@ interface DocxImportOptions {
   /** Custom image converter (default: embed as base64) */
   convertImage?: (image: DocxImageInfo) => Promise<DocxImageResult>;
 
-  /** Whether to ignore empty paragraphs (default: false).
-   * Empty paragraphs are those without text content or images.
-   * Paragraphs containing only whitespace or images are not considered empty. */
-  ignoreEmptyParagraphs?: boolean;
-
   /**
    * Dynamic import function for @napi-rs/canvas
    * Required for image cropping in Node.js environment, ignored in browser
@@ -94,7 +89,15 @@ interface DocxImportOptions {
    *
    * @default false
    */
-  enableImageCrop?: boolean;
+  crop?: boolean;
+
+  /** Paragraph processing options */
+  paragraph?: {
+    /** Whether to ignore empty paragraphs (default: false).
+     * Empty paragraphs are those without text content or images.
+     * Paragraphs containing only whitespace or images are not considered empty. */
+    ignoreEmpty?: boolean;
+  };
 }
 ```
 
@@ -231,7 +234,7 @@ const buffer = readFileSync("document.docx");
 
 const content = await parseDOCX(buffer, {
   canvasImport: () => import("@napi-rs/canvas"),
-  enableImageCrop: true, // Enable cropping (default is false)
+  crop: true, // Enable cropping (default is false)
 });
 ```
 
@@ -243,7 +246,7 @@ If you want to explicitly ignore crop information in DOCX and use full images (t
 
 ```typescript
 const content = await parseDOCX(buffer, {
-  enableImageCrop: false,
+  crop: false,
 });
 ```
 
@@ -271,7 +274,7 @@ All colors are imported as hex values (e.g., "#FF0000", "#008000"). Color names
 - Only embedded images are supported (external image links are not fetched)
 - Image dimensions and title are extracted from DOCX metadata
 - **Image Cropping**: By default, images are imported in full size (crop information is ignored)
-  - To enable cropping, set `enableImageCrop: true` in options
+  - To enable cropping, set `crop: true` in options
   - In browser environments, cropping works natively with Canvas API
   - In Node.js, you must also provide `canvasImport` option with dynamic import of `@napi-rs/canvas`
   - If `@napi-rs/canvas` is not available in Node.js, images will be imported without cropping (graceful degradation)

package/dist/index.d.mts

@@ -2166,7 +2166,7 @@ can be lifted. Will not go across
 [isolating](https://prosemirror.net/docs/ref/#model.NodeSpec.isolating) parent nodes.
 */
 //#endregion
-//#region ../../node_modules/.pnpm/prosemirror-view@1.41.6/node_modules/prosemirror-view/dist/index.d.ts
+//#region ../../node_modules/.pnpm/prosemirror-view@1.41.7/node_modules/prosemirror-view/dist/index.d.ts
 type DOMNode = InstanceType<typeof window.Node>;
 type WidgetConstructor = ((view: EditorView, getPos: () => number | undefined) => DOMNode) | DOMNode;
 /**
@@ -3672,7 +3672,7 @@ point into textblock nodes. It can be empty (a regular cursor
 position).
 */
 //#endregion
-//#region ../../node_modules/.pnpm/@tiptap+core@3.20.0_@tiptap+pm@3.20.0/node_modules/@tiptap/core/dist/index.d.ts
+//#region ../../node_modules/.pnpm/@tiptap+core@3.20.4_@tiptap+pm@3.20.4/node_modules/@tiptap/core/dist/index.d.ts
 type StringKeyOf<T> = Extract<keyof T, string>;
 type CallbackType<T extends Record<string, any>, EventName extends StringKeyOf<T>> = T[EventName] extends any[] ? T[EventName] : [T[EventName]];
 type CallbackFunction<T extends Record<string, any>, EventName extends StringKeyOf<T>> = (...props: CallbackType<T, EventName>) => any;
@@ -4436,6 +4436,27 @@ interface ExtendableConfig<Options = any, Storage = any, Config extends Extensio
      * Defines if this markdown element should indent it's child elements
      */
     indentsContent?: boolean;
+    /**
+     * Lets a mark tell the Markdown serializer which inline HTML tags it can
+     * safely use when plain markdown delimiters would become ambiguous.
+     *
+     * This is mainly useful for overlapping marks. For example, bold followed
+     * by bold+italic followed by italic cannot always be written back with only
+     * `*` and `**` in a way that still parses correctly. In that case, the
+     * serializer can close the overlapping section with markdown and reopen the
+     * remaining tail with HTML instead.
+     *
+     * Example:
+     * - desired formatting: `**123` + `*456*` + `789 italic`
+     * - serialized result: `**123*456***<em>789</em>`
+     *
+     * If your extension defines custom mark names, set `htmlReopen` on that
+     * extension so the serializer can reuse its HTML form for overlap cases.
+     */
+    htmlReopen?: {
+      open: string;
+      close: string;
+    };
   };
   /**
    * This function extends the schema of the node.
@@ -5279,7 +5300,8 @@ type MarkdownToken = {
  */
 type MarkdownParseHelpers = {
   /** Parse an array of inline tokens into text nodes with marks */parseInline: (tokens: MarkdownToken[]) => JSONContent[]; /** Parse an array of block-level tokens */
-  parseChildren: (tokens: MarkdownToken[]) => JSONContent[]; /** Create a text node with optional marks */
+  parseChildren: (tokens: MarkdownToken[]) => JSONContent[]; /** Parse block-level tokens while preserving implicit empty paragraphs from blank lines */
+  parseBlockChildren?: (tokens: MarkdownToken[]) => JSONContent[]; /** Create a text node with optional marks */
   createTextNode: (text: string, marks?: Array<{
     type: string;
     attrs?: any;
@@ -5312,6 +5334,7 @@ type RenderContext = {
   level: number;
   meta?: Record<string, any>;
   parentType?: string | null;
+  previousNode?: JSONContent | null;
 };
 /** Extension contract for markdown parsing/serialization. */
 /**
@@ -5345,7 +5368,8 @@ type MarkdownRendererHelpers = {
    * @param separator An optional separator string (legacy) or RenderContext
    * @returns The rendered markdown string
    */
-  renderChildren: (nodes: JSONContent | JSONContent[], separator?: string) => string;
+  renderChildren: (nodes: JSONContent | JSONContent[], separator?: string) => string; /** Render a single child node with its sibling index preserved */
+  renderChild?: (node: JSONContent, index: number) => string;
   /**
    * Render a text token to a markdown string
    * @param prefix The prefix to add before the content
@@ -6638,9 +6662,11 @@ interface DocxImportOptions {
   image?: {
     handler?: DocxImageImportHandler;
     canvasImport?: () => Promise<typeof _napi_rs_canvas0>;
-    enableImageCrop?: boolean;
+    crop?: boolean;
+  };
+  paragraph?: {
+    ignoreEmpty?: boolean;
   };
-  ignoreEmptyParagraphs?: boolean;
 }
 //#endregion
 //#region ../../node_modules/.pnpm/@types+unist@3.0.3/node_modules/@types/unist/index.d.ts

package/dist/index.mjs

@@ -463,7 +463,7 @@ async function applyCropToImage(pic, imgInfo, params) {
 		if (!base64Data) return imgInfo;
 		const croppedBase64 = uint8ArrayToBase64(await cropImageIfNeeded(base64ToUint8Array(base64Data), crop, {
 			canvasImport: params.context.image?.canvasImport,
-			enabled: params.context.image?.enableImageCrop ?? false
+			enabled: params.context.image?.crop ?? false
 		}));
 		const originalWidth = imgInfo.width || 0;
 		const originalHeight = imgInfo.height || 0;
@@ -532,25 +532,6 @@ function findDrawingElement(run) {
 	return choice ? findChild(choice, "w:drawing") : null;
 }
 /**
-* Adjust image dimensions to fit within group bounds while preserving aspect ratio
-*/
-function fitToGroup(groupWidth, groupHeight, metaWidth, metaHeight) {
-	const metaRatio = metaWidth / metaHeight;
-	const groupRatio = groupWidth / groupHeight;
-	if (Math.abs(metaRatio - groupRatio) > .1) if (metaRatio > groupRatio) return {
-		width: groupWidth,
-		height: Math.round(groupWidth / metaRatio)
-	};
-	else return {
-		width: Math.round(groupHeight * metaRatio),
-		height: groupHeight
-	};
-	return {
-		width: groupWidth,
-		height: groupHeight
-	};
-}
-/**
 * Extract images from DOCX and convert to base64 data URLs or use custom handler
 * Returns Map of relationship ID to image info (src + dimensions)
 */
@@ -613,7 +594,7 @@ async function extractImageFromDrawing(drawing, params) {
 				try {
 					src = `${metadata},${uint8ArrayToBase64(await cropImageIfNeeded(bytes, crop, {
 						canvasImport: context.image?.canvasImport,
-						enabled: context.image?.enableImageCrop ?? false
+						enabled: context.image?.crop ?? false
 					}))}`;
 				} catch (error) {
 					console.warn("Image cropping failed, using original image:", error);
@@ -711,10 +692,57 @@ async function extractImagesFromDrawing(drawing, params) {
 	if (group) {
 		const groupSp = findChild(group, "wpg:grpSp");
 		const pictures = groupSp ? [...findDeepChildren(groupSp, "pic:pic"), ...findDeepChildren(groupSp, "pic")] : [...findDeepChildren(group, "pic:pic"), ...findDeepChildren(group, "pic")];
+		const wspShapes = groupSp ? findDeepChildren(groupSp, "wps:wsp") : findDeepChildren(group, "wps:wsp");
+		const childImages = [];
 		for (const pic of pictures) {
 			const picGraphic = findChild(pic, "a:graphic");
+			let relativeSize = null;
+			const spPr = findChild(pic, "pic:spPr");
+			if (spPr) {
+				const xfrm = findChild(spPr, "a:xfrm");
+				if (xfrm) {
+					const ext = findChild(xfrm, "a:ext");
+					if (ext && ext.attributes["cx"] && ext.attributes["cy"]) relativeSize = {
+						cx: parseInt(ext.attributes["cx"], 10),
+						cy: parseInt(ext.attributes["cy"], 10)
+					};
+				}
+			}
+			childImages.push({
+				pic,
+				picGraphic,
+				relativeSize,
+				isWsp: false
+			});
+		}
+		for (const wsp of wspShapes) {
+			const wspGraphic = findChild(wsp, "a:graphic");
+			let relativeSize = null;
+			const spPr = findChild(wsp, "wps:spPr");
+			if (spPr) {
+				const xfrm = findChild(spPr, "a:xfrm");
+				if (xfrm) {
+					const ext = findChild(xfrm, "a:ext");
+					if (ext && ext.attributes["cx"] && ext.attributes["cy"]) relativeSize = {
+						cx: parseInt(ext.attributes["cx"], 10),
+						cy: parseInt(ext.attributes["cy"], 10)
+					};
+				}
+			}
+			childImages.push({
+				pic: wsp,
+				picGraphic: wspGraphic,
+				relativeSize,
+				isWsp: true
+			});
+		}
+		let totalCx = 0;
+		for (const child of childImages) if (child.relativeSize) totalCx += child.relativeSize.cx;
+		const scaleFactor = totalCx > 0 && groupWidth ? groupWidth / totalCx : 1;
+		for (const child of childImages) {
+			const { pic, picGraphic, relativeSize, isWsp } = child;
 			if (!picGraphic) {
-				const blipFill = findChild(pic, "pic:blipFill") || findDeepChild(pic, "a:blipFill");
+				const blipFill = isWsp ? findChild(pic, "wps:blipFill") || findDeepChild(pic, "a:blipFill") : findChild(pic, "pic:blipFill") || findDeepChild(pic, "a:blipFill");
 				if (!blipFill) continue;
 				const blip = findChild(blipFill, "a:blip") || findDeepChild(blipFill, "a:blip");
 				if (!blip?.attributes["r:embed"]) continue;
@@ -722,13 +750,22 @@ async function extractImagesFromDrawing(drawing, params) {
 				const imgInfo = params.context.images.get(rId);
 				if (!imgInfo) continue;
 				const processedImgInfo = await applyCropToImage(pic, imgInfo, params);
+				let width = processedImgInfo.width;
+				let height = processedImgInfo.height;
+				if (groupWidth && groupHeight && relativeSize && totalCx > 0) {
+					width = Math.round(relativeSize.cx * scaleFactor);
+					height = Math.round(relativeSize.cy * scaleFactor);
+				} else if (groupWidth && groupHeight) {
+					width = groupWidth;
+					height = groupHeight;
+				}
 				result.push({
 					type: "image",
 					attrs: {
 						src: processedImgInfo.src,
 						alt: "",
-						width: processedImgInfo.width,
-						height: processedImgInfo.height
+						width,
+						height
 					}
 				});
 				continue;
@@ -748,7 +785,7 @@ async function extractImagesFromDrawing(drawing, params) {
 				if (base64Data) {
 					const croppedBase64 = uint8ArrayToBase64(await cropImageIfNeeded(base64ToUint8Array(base64Data), crop, {
 						canvasImport: params.context.image?.canvasImport,
-						enabled: params.context.image?.enableImageCrop ?? false
+						enabled: params.context.image?.crop ?? false
 					}));
 					image.attrs.src = `${metadata},${croppedBase64}`;
 					const rId = syntheticDrawing.children[0]?.type === "element" ? findDeepChild(syntheticDrawing.children[0], "a:blip")?.attributes["r:embed"] : void 0;
@@ -763,27 +800,25 @@ async function extractImagesFromDrawing(drawing, params) {
 							const visibleHeightPct = 1 - cropTopPct - cropBottomPct;
 							const croppedWidth = Math.round(imgInfo.width * visibleWidthPct);
 							const croppedHeight = Math.round(imgInfo.height * visibleHeightPct);
-							image.attrs.width = croppedWidth;
-							image.attrs.height = croppedHeight;
+							if (groupWidth && groupHeight && relativeSize && totalCx > 0) {
+								image.attrs.width = Math.round(relativeSize.cx * scaleFactor);
+								image.attrs.height = Math.round(relativeSize.cy * scaleFactor);
+							} else {
+								image.attrs.width = croppedWidth;
+								image.attrs.height = croppedHeight;
+							}
 						}
 					}
 				}
 			} catch (error) {
 				console.warn("Grouped image cropping failed, using original image:", error);
 			}
-			else {
-				const rId = syntheticDrawing.children[0]?.type === "element" ? findDeepChild(syntheticDrawing.children[0], "a:blip")?.attributes["r:embed"] : void 0;
-				if (groupWidth && groupHeight && rId) {
-					const imgInfo = params.context.images.get(rId);
-					if (imgInfo?.width && imgInfo?.height) {
-						const adjusted = fitToGroup(groupWidth, groupHeight, imgInfo.width, imgInfo.height);
-						image.attrs.width = adjusted.width;
-						image.attrs.height = adjusted.height;
-					} else {
-						image.attrs.width = groupWidth;
-						image.attrs.height = groupHeight;
-					}
-				}
+			else if (groupWidth && groupHeight && relativeSize && totalCx > 0) {
+				image.attrs.width = Math.round(relativeSize.cx * scaleFactor);
+				image.attrs.height = Math.round(relativeSize.cy * scaleFactor);
+			} else if (groupWidth && groupHeight) {
+				image.attrs.width = groupWidth;
+				image.attrs.height = groupHeight;
 			}
 			result.push(image);
 		}
@@ -1340,7 +1375,9 @@ function parseCellProperties(cellNode) {
 	if (!tcPr) return props;
 	const gridSpan = findChild(tcPr, "w:gridSpan");
 	if (gridSpan?.attributes["w:val"]) props.colspan = parseInt(gridSpan.attributes["w:val"]);
-	if (findChild(tcPr, "w:vMerge")?.attributes["w:val"] === "continue") props.rowspan = 0;
+	const vMerge = findChild(tcPr, "w:vMerge");
+	if (vMerge) if (vMerge.attributes["w:val"] === "continue") props.rowspan = 0;
+	else props.rowspan = 1;
 	const tcW = findChild(tcPr, "w:tcW");
 	if (tcW?.attributes["w:w"]) props.colwidth = [convertTwipToPixels(parseInt(tcW.attributes["w:w"]))];
 	const shd = findChild(tcPr, "w:shd");
@@ -1679,7 +1716,7 @@ async function convertElements(elements, params) {
 	for (let i = 0; i < elements.length; i++) {
 		if (processedIndices.has(i)) continue;
 		const element = elements[i];
-		if (params.context.ignoreEmptyParagraphs && element.name === "w:p" && isEmptyParagraph(element)) continue;
+		if (params.context.paragraph?.ignoreEmpty && element.name === "w:p" && isEmptyParagraph(element)) continue;
 		const node = await convertElement(element, elements, i, params, processedIndices);
 		if (Array.isArray(node)) result.push(...node);
 		else if (node) result.push(node);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@docen/import-docx",
-  "version": "0.0.13",
+  "version": "0.0.14",
   "description": "A powerful TipTap/ProseMirror extension that imports Microsoft Word DOCX files to editor content",
   "keywords": [
     "converter",
@@ -51,10 +51,10 @@
     "xast-util-from-xml": "4.0.0"
   },
   "devDependencies": {
-    "@tiptap/core": "3.20.0",
+    "@tiptap/core": "3.20.4",
     "@types/xast": "2.0.4",
-    "@docen/extensions": "0.0.13",
-    "@docen/utils": "0.0.13"
+    "@docen/extensions": "0.0.14",
+    "@docen/utils": "0.0.14"
   },
   "peerDependencies": {
     "@napi-rs/canvas": "^0.1.88"
@@ -65,7 +65,7 @@
     }
   },
   "optionalDependencies": {
-    "@napi-rs/canvas": "^0.1.96"
+    "@napi-rs/canvas": "^0.1.97"
   },
   "scripts": {
     "dev": "basis build --stub",