@docen/import-docx 0.0.8

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 Demo Macro
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,287 @@
+# @docen/import-docx
+
+![npm version](https://img.shields.io/npm/v/@docen/import-docx)
+![npm downloads](https://img.shields.io/npm/dw/@docen/import-docx)
+![npm license](https://img.shields.io/npm/l/@docen/import-docx)
+
+> Import Microsoft Word DOCX files to TipTap/ProseMirror content.
+
+## Features
+
+- 📝 **Rich Text Parsing** - Accurate parsing of headings, paragraphs, and blockquotes with formatting
+- 🖼️ **Image Extraction** - Automatic image extraction with base64 conversion and cropping support
+- 📊 **Table Support** - Complete table structure with colspan/rowspan detection algorithm
+- ✅ **Lists & Tasks** - Bullet lists, numbered lists with start number extraction, and task lists with checkbox detection
+- 🎨 **Text Formatting** - Bold, italic, underline, strikethrough, subscript, superscript, and highlights
+- 🎯 **Text Styles** - Comprehensive style support including colors, backgrounds, fonts, sizes, and line heights
+- 🔗 **Links** - Hyperlink extraction with href preservation
+- 💻 **Code Blocks** - Code block detection with language attribute extraction
+- 🌐 **Cross-Platform** - Works in both browser and Node.js environments
+- ✂️ **Image Cropping** - Automatic cropping of images based on DOCX crop metadata
+- 🧠 **Smart Parsing** - DOCX XML parsing with proper element grouping and structure reconstruction
+- ⚡ **Fast Processing** - Uses fflate for ultra-fast ZIP decompression
+
+## Installation
+
+```bash
+# Install with npm
+$ npm install @docen/import-docx
+
+# Install with yarn
+$ yarn add @docen/import-docx
+
+# Install with pnpm
+$ pnpm add @docen/import-docx
+```
+
+## Quick Start
+
+```typescript
+import { parseDOCX } from "@docen/import-docx";
+import { readFileSync } from "node:fs";
+
+// Read DOCX file
+const buffer = readFileSync("document.docx");
+
+// Parse DOCX to TipTap JSON
+const content = await parseDOCX(buffer);
+
+// Use in TipTap editor
+editor.commands.setContent(content);
+```
+
+## API Reference
+
+### `parseDOCX(input, options?)`
+
+Parses a DOCX file and converts it to TipTap/ProseMirror JSON content.
+
+**Parameters:**
+
+- `input: Buffer | ArrayBuffer | Uint8Array` - DOCX file data
+- `options?: DocxImportOptions` - Optional import configuration
+
+**Returns:** `Promise<JSONContent>` - TipTap/ProseMirror document content with images embedded
+
+**Options:**
+
+```typescript
+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
+   *
+   * @example
+   * import { parseDOCX } from '@docen/import-docx';
+   * const content = await parseDOCX(buffer, {
+   *   canvasImport: () => import('@napi-rs/canvas')
+   * });
+   */
+  canvasImport?: () => Promise<typeof import("@napi-rs/canvas")>;
+
+  /**
+   * Enable or disable image cropping during import
+   * When true (default), images with crop information in DOCX will be cropped
+   * When false, crop information is ignored and full image is used
+   *
+   * @default true
+   */
+  enableImageCrop?: boolean;
+}
+```
+
+**Default Image Converter:**
+
+The package exports `defaultImageConverter` which embeds images as base64 data URLs:
+
+```typescript
+import { defaultImageConverter } from "@docen/import-docx";
+
+// Use in custom converter
+await parseDOCX(buffer, {
+  convertImage: async (image) => {
+    if (shouldUploadToCDN) {
+      return uploadToCDN(image.data);
+    }
+    return defaultImageConverter(image);
+  },
+});
+```
+
+## Supported Content Types
+
+### Text Formatting
+
+- **Bold**, _Italic_, <u>Underline</u>, ~~Strikethrough~~
+- ^Superscript^ and ~Subscript~
+- Text highlights
+- Text colors and background colors
+- Font families and sizes
+- Line heights
+
+### Block Elements
+
+- **Headings** (H1-H6) with proper level detection
+- **Paragraphs** with text alignment (left, right, center, justify)
+- **Blockquotes** (Detected by indentation + left border formatting)
+- **Horizontal Rules** (Detected as page breaks in DOCX)
+- **Code Blocks** with language attribute support
+
+### Lists
+
+- **Bullet Lists** with proper nesting and structure
+- **Numbered Lists** with custom start number extraction
+- **Task Lists** with checked/unchecked state detection (☐/☑ symbols)
+
+### Tables
+
+- Complete table structure parsing
+- **Table Cells** with colspan detection using grid-based algorithm
+- **Table Cells** with rowspan detection using vMerge tracking
+- Cell alignment and formatting preservation
+- Merged cell handling (both horizontal and vertical)
+
+### Media & Embeds
+
+- **Images** with automatic base64 conversion
+- **Grouped Images** (DOCX image groups) support
+- **Links** (hyperlinks) with href extraction
+
+## Parsing Algorithm
+
+### Document Structure
+
+The parser follows a structured workflow:
+
+1. **Extract Relationships** - Parse `_rels/document.xml.rels` for hyperlinks and images
+2. **Parse Numbering** - Extract list definitions from `numbering.xml` (abstractNum → numFmt)
+3. **Process Document Body** - Iterate through document.xml elements:
+   - Detect content types (tables, lists, paragraphs, code blocks, etc.)
+   - Group consecutive elements into proper containers
+   - Convert XML nodes to TipTap JSON nodes
+
+### Table Processing
+
+Tables use specialized algorithms:
+
+- **Colspan Detection** - Grid-based algorithm tracks cell positions and detects horizontal merges
+- **Rowspan Detection** - Vertical merge (vMerge) tracking across rows with proper cell skipping
+- **Cell Content** - Recursive parsing of nested paragraphs and formatting
+- **Hyperlink Support** - Proper handling of links within table cells
+
+### List Processing
+
+Lists utilize the DOCX numbering system:
+
+- **Numbering ID Mapping** - Maps abstractNum to formatting (bullet vs decimal)
+- **Start Value Extraction** - Extracts and preserves start numbers for ordered lists
+- **Nesting Preservation** - Maintains proper list hierarchy
+- **Consecutive Grouping** - Groups consecutive list items into list containers
+
+## Examples
+
+### Basic Usage
+
+```typescript
+import { parseDOCX } from "@docen/import-docx";
+
+const buffer = readFileSync("example.docx");
+const { content } = await parseDOCX(buffer);
+
+console.log(JSON.stringify(content, null, 2));
+```
+
+### Use with TipTap Editor
+
+```typescript
+import { Editor } from "@tiptap/core";
+import { parseDOCX } from "@docen/import-docx";
+
+const editor = new Editor({
+  extensions: [...],
+  content: "",
+});
+
+// Import DOCX file
+async function importDocx(file: File) {
+  const buffer = await file.arrayBuffer();
+  const content = await parseDOCX(buffer);
+  editor.commands.setContent(content);
+}
+```
+
+### Node.js Environment with Image Cropping
+
+In Node.js environment, you need to provide `@napi-rs/canvas` for image cropping:
+
+```typescript
+import { parseDOCX } from "@docen/import-docx";
+import { readFileSync } from "node:fs";
+
+// Install @napi-rs/canvas first: pnpm add @napi-rs/canvas
+const buffer = readFileSync("document.docx");
+
+const content = await parseDOCX(buffer, {
+  canvasImport: () => import("@napi-rs/canvas"),
+  enableImageCrop: true, // default is true
+});
+```
+
+### Disable Image Cropping
+
+If you want to ignore crop information in DOCX and use full images:
+
+```typescript
+const content = await parseDOCX(buffer, {
+  enableImageCrop: false,
+});
+```
+
+## Known Limitations
+
+### Blockquote Detection
+
+DOCX does not have a semantic blockquote structure. Blockquotes are detected by:
+
+- Left indentation ≥ 720 twips (0.5 inch)
+- Presence of left border (single line)
+
+This detection method may produce false positives for documents with custom indentation similar to blockquotes.
+
+### Code Marks
+
+The `code` mark is NOT automatically detected from monospace fonts (Consolas, Courier New, etc.). This is intentional to avoid false positives. Code marks should be explicitly added in the source document or through editor UI.
+
+### Color Format
+
+All colors are imported as hex values (e.g., "#FF0000", "#008000"). Color names from the original document are not preserved.
+
+### Image Limitations
+
+- Only embedded images are supported (external image links are not fetched)
+- Image dimensions and title are extracted from DOCX metadata
+- **Image Cropping in Node.js**: Requires `@napi-rs/canvas` as an optional dependency
+  - In browser environments, cropping works natively with Canvas API
+  - In Node.js, you must provide `canvasImport` option with dynamic import of `@napi-rs/canvas`
+  - If `@napi-rs/canvas` is not available, images will be imported without cropping (graceful degradation)
+- Some DOCX image features (like advanced positioning or text wrapping) have limited support
+
+### Table Cell Types
+
+DOCX format does not distinguish between header and body cells at a semantic level. All cells are imported as `tableCell` type for consistency. This is a DOCX format limitation.
+
+## Contributing
+
+Contributions are welcome! Please read our [Contributor Covenant](https://www.contributor-covenant.org/version/2/1/code_of_conduct/) and submit pull requests to the [main repository](https://github.com/DemoMacro/docen).
+
+## License
+
+- [MIT](LICENSE) &copy; [Demo Macro](https://imst.xyz/)

package/dist/chunks/index.cjs

@@ -0,0 +1 @@
+"use strict";const index=require("../index.cjs");require("xast-util-from-xml"),require("fflate"),require("undio"),require("image-meta"),exports.convertParagraph=index.convertParagraph,exports.convertTable=index.convertTable,exports.convertTaskItem=index.convertTaskItem,exports.extractAlignment=index.extractAlignment,exports.extractMarks=index.extractMarks,exports.extractRuns=index.extractRuns,exports.getCodeBlockLanguage=index.getCodeBlockLanguage,exports.getListInfo=index.getListInfo,exports.getTaskItemChecked=index.getTaskItemChecked,exports.isCodeBlock=index.isCodeBlock,exports.isHorizontalRule=index.isHorizontalRule,exports.isListItem=index.isListItem,exports.isTable=index.isTable,exports.isTaskItem=index.isTaskItem;

package/dist/chunks/index.mjs

@@ -0,0 +1 @@
+export{convertParagraph,convertTable,convertTaskItem,extractAlignment,extractMarks,extractRuns,getCodeBlockLanguage,getListInfo,getTaskItemChecked,isCodeBlock,isHorizontalRule,isListItem,isTable,isTaskItem}from"../index.mjs";import"xast-util-from-xml";import"fflate";import"undio";import"image-meta";

package/dist/index.cjs

@@ -0,0 +1 @@
+"use strict";const xastUtilFromXml=require("xast-util-from-xml"),fflate=require("fflate"),undio=require("undio"),imageMeta=require("image-meta");function findChild(n,t){for(const e of n.children)if(e.type==="element"&&e.name===t)return e}function findDeepChild(n,t){for(const e of n.children){if(e.type==="element"&&e.name===t)return e;if(e.type==="element"){const r=findDeepChild(e,t);if(r)return r}}}function findDeepChildren(n,t){const e=[];for(const r of n.children)r.type==="element"&&r.name===t&&e.push(r),r.type==="element"&&e.push(...findDeepChildren(r,t));return e}const s="ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";function uint8ArrayToBase64(n){const t=n.length,e=Math.ceil(t/3)*4,r=Array.from({length:e});let i=0;for(let c=0;c<t;c+=3){const l=n[c],f=c+1<t?n[c+1]:0,b=c+2<t?n[c+2]:0,I=l>>2,k=(l&3)<<4|f>>4,R=(f&15)<<2|b>>6,P=b&63;r[i++]=s[I],r[i++]=s[k],r[i++]=c+1<t?s[R]:"=",r[i++]=c+2<t?s[P]:"="}return r.join("")}function base64ToUint8Array(n){const t=atob(n),e=new Uint8Array(t.length);for(let r=0;r<t.length;r++)e[r]=t.charCodeAt(r);return e}const isNode=globalThis.process?.release?.name==="node",isBrowser=typeof window<"u";async function h$2(n){const t=await n;return t.default||t}let o,u$1=class{#t=!1;constructor({enableHWA:t=!1}={}){this.#t=t}create(t,e){const r=this._createCanvas(t,e);return{canvas:r,context:r.getContext("2d",{willReadFrequently:!this.#t})}}reset({canvas:t},e,r){if(!t)throw new Error("Canvas is not specified");t.width=e,t.height=r}destroy(t){if(!t.canvas)throw new Error("Canvas is not specified");t.canvas.width=0,t.canvas.height=0,t.canvas=void 0,t.context=void 0}_createCanvas(t,e){throw new Error("Not implemented")}};class DOMCanvasFactory extends u$1{_document;constructor({ownerDocument:t=globalThis.document,enableHWA:e=!1}={}){super({enableHWA:e}),this._document=t}_createCanvas(t,e){const r=this._document.createElement("canvas");return r.width=t,r.height=e,r}}class NodeCanvasFactory extends u$1{constructor({enableHWA:t=!1}={}){super({enableHWA:t})}_createCanvas(t,e){if(!o)throw new Error("@napi-rs/canvas module is not resolved");return o.createCanvas(t,e)}}async function resolveCanvasModule(n){o??=await h$2(n())}async function createCanvasFactory(n){if(isBrowser)return DOMCanvasFactory;if(isNode){if(!n)throw new Error("In Node.js environment, @napi-rs/canvas is required for image cropping. Please provide canvasImport parameter or install it: pnpm add @napi-rs/canvas");return await resolveCanvasModule(n),NodeCanvasFactory}throw new Error("Unsupported environment for canvas operations")}async function cropImageIfNeeded(n,t,e={}){if(!t||!t.left&&!t.top&&!t.right&&!t.bottom||e.enabled===!1)return n;try{const r=await createCanvasFactory(e.canvasImport),i=await w$1(n,r),c=(t.left||0)/1e5*i.width,l=(t.top||0)/1e5*i.height,f=(t.right||0)/1e5*i.width,b=(t.bottom||0)/1e5*i.height,I=Math.round(i.width-c-f),k=Math.round(i.height-l-b);if(I<=0||k<=0)return console.warn("Invalid crop dimensions, returning original image"),n;const R=new r().create(I,k);if(!R.context)throw new Error("Failed to get 2D context from canvas");R.context.drawImage(i,c,l,I,k,0,0,I,k);const P=R.canvas.toDataURL(),F=await(await fetch(P)).arrayBuffer();return new Uint8Array(F)}catch(r){return console.warn("Image cropping failed, returning original image:",r),n}}async function w$1(n,t){if(isBrowser){const e=new Blob([n.buffer]),r=URL.createObjectURL(e);try{const i=new Image;return new Promise((c,l)=>{i.onload=()=>{URL.revokeObjectURL(r),c(i)},i.onerror=()=>{URL.revokeObjectURL(r),l(new Error("Failed to load image"))},i.src=r})}catch(i){throw URL.revokeObjectURL(r),i}}else{if(!o)throw new Error("@napi-rs/canvas module is not resolved");return await o.loadImage(Buffer.from(n))}}const j="http://schemas.openxmlformats.org/officeDocument/2006/relationships/image";function C$1(n){const t=parseInt(n,10);if(!isNaN(t))return Math.round(t/9525)}function B(n){const t=n.attributes.l,e=n.attributes.t,r=n.attributes.r,i=n.attributes.b;if(!(!t&&!e&&!r&&!i))return{left:t?parseInt(t,10):void 0,top:e?parseInt(e,10):void 0,right:r?parseInt(r,10):void 0,bottom:i?parseInt(i,10):void 0}}function N(n){const t=findChild(n,"wp:align"),e=findChild(n,"wp:posOffset"),r=t?.children[0]?.type==="text"?t.children[0].value:void 0,i=e?.children[0]?.type==="text"?parseInt(e.children[0].value,10):void 0;if(!(!r&&i===void 0))return{...r&&{align:r},...i!==void 0&&{offset:i}}}function findDrawingElement(n){let t=findChild(n,"w:drawing");if(t)return t;const e=findChild(n,"mc:AlternateContent"),r=e&&findChild(e,"mc:Choice");return r&&findChild(r,"w:drawing")}function O$1(n,t,e,r){const i=e/r,c=n/t;return Math.abs(i-c)>.1?i>c?{width:n,height:Math.round(n/i)}:{width:Math.round(t*i),height:t}:{width:n,height:t}}function extractImages(n){const t=new Map,e=n["word/_rels/document.xml.rels"];if(!e)return t;const r=xastUtilFromXml.fromXml(new TextDecoder().decode(e)),i=findChild(r,"Relationships");if(!i)return t;const c=findDeepChildren(i,"Relationship");for(const l of c)if(l.attributes.Type===j&&l.attributes.Id&&l.attributes.Target){const f="word/"+l.attributes.Target,b=n[f];if(!b)continue;let I,k,R="png";try{const $=imageMeta.imageMeta(b);I=$.width,k=$.height,$.type&&(R=$.type)}catch{}const P=uint8ArrayToBase64(b),F=`data:image/${R};base64,${P}`;t.set(l.attributes.Id,{src:F,width:I,height:k})}return t}async function extractImageFromDrawing(n,t){const{images:e,options:r}=t,i=findDeepChild(n,"a:blip");if(!i?.attributes["r:embed"])return null;const c=i.attributes["r:embed"],l=e.get(c);if(!l)return null;let f=l.src;const b=findDeepChild(n,"a:srcRect");if(b){const D=B(b);if(D&&f.startsWith("data:")){const[U,_]=f.split(",");if(_){const V=base64ToUint8Array(_);try{const q=await cropImageIfNeeded(V,D,{canvasImport:r?.canvasImport,enabled:r?.enableImageCrop!==!1}),G=uint8ArrayToBase64(q);f=`${U},${G}`}catch(q){console.warn("Image cropping failed, using original image:",q)}}}}const I=findDeepChild(n,"wp:extent");let k,R;if(I){const D=I.attributes.cx,U=I.attributes.cy;typeof D=="string"&&(k=C$1(D)),typeof U=="string"&&(R=C$1(U))}const P=findDeepChild(n,"a:xfrm");let F;if(P?.attributes.rot){const D=parseInt(P.attributes.rot,10);isNaN(D)||(F=D/6e4)}const $=findDeepChild(n,"wp:docPr")?.attributes.title,W=findDeepChild(n,"wp:positionH"),z=findDeepChild(n,"wp:positionV");let E;if(W||z){const D=W?N(W):void 0,U=z?N(z):void 0;E={horizontalPosition:{relative:W?.attributes.relativeFrom||"page",...D?.align&&{align:D.align},...D?.offset!==void 0&&{offset:D.offset}},verticalPosition:{relative:z?.attributes.relativeFrom||"page",...U?.align&&{align:U.align},...U?.offset!==void 0&&{offset:U.offset}}}}const H=findDeepChild(n,"pic:spPr");let X;if(H){const D=findDeepChild(H,"a:ln"),U=D&&findDeepChild(D,"a:solidFill"),_=U&&findDeepChild(U,"a:srgbClr");_?.attributes.val&&(X={type:"solidFill",solidFillType:"rgb",value:_.attributes.val})}return{type:"image",attrs:{src:f,alt:"",...k!==void 0&&{width:k},...R!==void 0&&{height:R},...F!==void 0&&{rotation:F},...$&&{title:$},...E&&{floating:E},...X&&{outline:X}}}}function S$2(n,t,e){if(t&&e&&n.width&&n.height){const r=O$1(t,e,n.width,n.height);return{type:"image",attrs:{src:n.src,alt:"",width:r.width,height:r.height}}}return{type:"image",attrs:{src:n.src,alt:"",...t!==void 0&&{width:t},...e!==void 0&&{height:e}}}}async function extractImagesFromDrawing(n,t){const e=[],r=findChild(n,"wp:inline")||findChild(n,"wp:anchor");if(!r)return e;const i=findChild(r,"wp:extent");let c,l;if(i){const k=i.attributes.cx,R=i.attributes.cy;typeof k=="string"&&(c=C$1(k)),typeof R=="string"&&(l=C$1(R))}const f=findChild(r,"a:graphic");if(!f)return e;const b=findChild(f,"a:graphicData");if(!b)return e;const I=findChild(b,"wpg:wgp");if(I){const k=findChild(I,"wpg:grpSp"),R=k?[...findDeepChildren(k,"pic:pic"),...findDeepChildren(k,"pic")]:[...findDeepChildren(I,"pic:pic"),...findDeepChildren(I,"pic")];for(const P of R){const F=findChild(P,"a:graphic");if(!F){const E=findChild(P,"pic:blipFill")||findDeepChild(P,"a:blipFill");if(!E)continue;const H=findChild(E,"a:blip")||findDeepChild(E,"a:blip");if(!H?.attributes["r:embed"])continue;const X=H.attributes["r:embed"],D=t.images.get(X);if(!D)continue;e.push(S$2(D,c,l));continue}const $={children:[F]},W=await extractImageFromDrawing($,t);if(!W)continue;const z=$.children[0]?.type==="element"?findDeepChild($.children[0],"a:blip")?.attributes["r:embed"]:void 0;if(c&&l&&z){const E=t.images.get(z);if(E?.width&&E?.height){const H=O$1(c,l,E.width,E.height);W.attrs.width=H.width,W.attrs.height=H.height}else W.attrs.width=c,W.attrs.height=l}e.push(W)}}else{const k=await extractImageFromDrawing(n,t);k&&e.push(k)}return e}const p="http://schemas.openxmlformats.org/officeDocument/2006/relationships/hyperlink";function extractHyperlinks(n){const t=new Map,e=n["word/_rels/document.xml.rels"];if(!e)return t;const r=xastUtilFromXml.fromXml(new TextDecoder().decode(e)),i=findChild(r,"Relationships");if(!i)return t;const c=findDeepChildren(i,"Relationship");for(const l of c)l.attributes.Type===p&&l.attributes.Id&&l.attributes.Target&&t.set(l.attributes.Id,l.attributes.Target);return t}function parseNumberingXml(n){const t=new Map,e=new Map,r=n["word/numbering.xml"];if(!r)return t;const i=xastUtilFromXml.fromXml(new TextDecoder().decode(r)),c=new Map,l=findChild(i,"w:numbering");if(!l)return t;const f=findDeepChildren(l,"w:abstractNum");for(const I of f){const k=I.attributes["w:abstractNumId"],R=findChild(I,"w:lvl");if(!R)continue;const P=findChild(R,"w:numFmt");P?.attributes["w:val"]&&c.set(k,P.attributes["w:val"]);const F=findChild(R,"w:start");F?.attributes["w:val"]&&e.set(k,parseInt(F.attributes["w:val"],10))}const b=findDeepChildren(l,"w:num");for(const I of b){const k=I.attributes["w:numId"],R=findChild(I,"w:abstractNumId");if(!R?.attributes["w:val"])continue;const P=R.attributes["w:val"],F=c.get(P);if(!F)continue;const $=e.get(P);F==="bullet"?t.set(k,{type:"bullet"}):t.set(k,{type:"ordered",...$!==void 0&&{start:$}})}return t}function parseStylesXml(n){const t=new Map,e=n["word/styles.xml"];if(!e)return t;const r=xastUtilFromXml.fromXml(new TextDecoder().decode(e)),i=findChild(r,"w:styles");if(!i)return t;const c=findDeepChildren(i,"w:style").filter(l=>l.attributes["w:type"]==="paragraph");for(const l of c){const f=l.attributes["w:styleId"];if(!f)continue;const b={styleId:f},I=findChild(l,"w:name");I?.attributes["w:val"]&&(b.name=I.attributes["w:val"]);const k=findChild(l,"w:pPr");if(k){const P=findChild(k,"w:outlineLvl");P?.attributes["w:val"]!==void 0&&(b.outlineLvl=parseInt(P.attributes["w:val"],10))}const R=findChild(l,"w:rPr");if(R){const P={},F=findChild(R,"w:color");if(F?.attributes["w:val"]&&F.attributes["w:val"]!=="auto"){const z=F.attributes["w:val"];P.color=z.startsWith("#")?z:`#${z}`}findChild(R,"w:b")&&(P.bold=!0),findChild(R,"w:i")&&(P.italic=!0),findChild(R,"w:u")&&(P.underline=!0),findChild(R,"w:strike")&&(P.strike=!0);const $=findChild(R,"w:sz");if($?.attributes["w:val"]){const z=$.attributes["w:val"],E=parseInt(z,10);isNaN(E)||(P.fontSize=E)}const W=findChild(R,"w:rFonts");W?.attributes["w:ascii"]&&(P.fontFamily=W.attributes["w:ascii"]),Object.keys(P).length>0&&(b.charFormat=P)}t.set(f,b)}return t}function d(n,t){const e=findChild(n,"w:t");if(!e)return null;const r=e.children.find(c=>c.type==="text");if(!r?.value)return null;const i=extractMarks(n,t);return{type:"text",text:r.value,...i.length&&{marks:i}}}async function extractRuns(n,t){const e=[];for(const r of n.children)if(r.type==="element"){if(r.name==="w:hyperlink"){const i=r,c=i.attributes["r:id"],l=t.hyperlinks.get(c);if(!l)continue;for(const f of i.children){if(f.type!=="element"||f.name!=="w:r")continue;const b=f,I=findDrawingElement(b);if(I){const R=await extractImageFromDrawing(I,t);if(R){e.push(R);continue}const P=await extractImagesFromDrawing(I,t);if(P.length){e.push(...P);continue}}const k=d(b,t.styleInfo);k&&(k.marks=k.marks||[],k.marks.push({type:"link",attrs:{href:l}}),e.push(k))}}else if(r.name==="w:r"){const i=r,c=findDrawingElement(i);if(c){const f=await extractImagesFromDrawing(c,t);if(f.length){e.push(...f);continue}}if(findChild(i,"w:br")){const f=extractMarks(i,t.styleInfo);e.push({type:"hardBreak",...f.length&&{marks:f}})}const l=d(i,t.styleInfo);l&&e.push(l)}}return e}function extractMarks(n,t){const e=[],r=findChild(n,"w:rPr");let i={};if(t?.charFormat&&(i={...t.charFormat}),r){const c=findChild(r,"w:b");c&&(c.attributes["w:val"]==="false"?i.bold=!1:i.bold=!0);const l=findChild(r,"w:i");l&&(l.attributes["w:val"]==="false"?i.italic=!1:i.italic=!0),findChild(r,"w:u")&&(i.underline=!0),findChild(r,"w:strike")&&(i.strike=!0);const f=findChild(r,"w:color");if(f?.attributes["w:val"]&&f.attributes["w:val"]!=="auto"){const P=f.attributes["w:val"];i.color=P.startsWith("#")?P:`#${P}`}const b=findChild(r,"w:sz");if(b?.attributes["w:val"]){const P=b.attributes["w:val"],F=parseInt(P,10);isNaN(F)||(i.fontSize=F)}const I=findChild(r,"w:rFonts");I?.attributes["w:ascii"]&&(i.fontFamily=I.attributes["w:ascii"]);const k=findChild(r,"w:shd");if(k?.attributes["w:fill"]&&k.attributes["w:fill"]!=="auto"){const P=k.attributes["w:fill"];i.backgroundColor=P.startsWith("#")?P:`#${P}`}findChild(r,"w:highlight")&&e.push({type:"highlight"});const R=findChild(r,"w:vertAlign");if(R){const P=R.attributes["w:val"];P==="subscript"?e.push({type:"subscript"}):P==="superscript"&&e.push({type:"superscript"})}}if(i.bold&&e.push({type:"bold"}),i.italic&&e.push({type:"italic"}),i.underline&&e.push({type:"underline"}),i.strike&&e.push({type:"strike"}),i.color||i.backgroundColor||i.fontSize||i.fontFamily){const c={color:i.color||"",backgroundColor:i.backgroundColor||"",fontSize:"",fontFamily:"",lineHeight:""};if(i.fontSize){const l=Math.round(i.fontSize/1.5*10)/10;c.fontSize=`${l}px`}i.fontFamily&&(c.fontFamily=i.fontFamily),e.push({type:"textStyle",attrs:c})}return e}function extractAlignment(n){const t=findChild(n,"w:pPr");if(!t)return;const e=findChild(t,"w:jc");if(!e?.attributes["w:val"])return;const r=e.attributes["w:val"],i={left:"left",right:"right",center:"center",both:"justify"}[r];return i?{textAlign:i}:void 0}function y$2(n){return`${Math.round(n/15)}px`}function m$1(n){const t=findChild(n,"w:pPr");if(!t)return null;const e={},r=findChild(t,"w:ind");if(r){const c=I=>{const k=r.attributes[I];if(typeof k!="string")return null;const R=parseInt(k,10);return isNaN(R)?null:y$2(R)},l=c("w:left");l&&(e.indentLeft=l);const f=c("w:right");f&&(e.indentRight=f);const b=c("w:firstLine");if(b)e.indentFirstLine=b;else{const I=c("w:hanging");I&&(e.indentFirstLine=`-${I}`)}}const i=findChild(t,"w:spacing");if(i){const c=b=>{const I=i.attributes[b];if(typeof I!="string")return null;const k=parseInt(I,10);return isNaN(k)?null:y$2(k)},l=c("w:before");l&&(e.spacingBefore=l);const f=c("w:after");f&&(e.spacingAfter=f)}return Object.keys(e).length?e:null}async function convertParagraph(n,t){const e=findChild(n,"w:pPr"),r=(e&&findChild(e,"w:pStyle"))?.attributes["w:val"];if(r&&t.styleMap){const f=t.styleMap.get(r);if(f?.outlineLvl!==void 0&&f.outlineLvl>=0&&f.outlineLvl<=5){const I=f.outlineLvl+1;return h$1(n,t,f,I)}const b=r.match(/^Heading(\d+)$/);if(b){const I=parseInt(b[1],10);return h$1(n,t,f,I)}}const i=r&&t.styleMap?t.styleMap.get(r):void 0,c=await extractRuns(n,{...t,styleInfo:i}),l={...extractAlignment(n),...m$1(n)};if(w(n)){const f=c.filter(b=>b.type!=="hardBreak");return[{type:"paragraph",...Object.keys(l).length&&{attrs:l},content:f.length?f:void 0},{type:"horizontalRule"}]}if(c.length===1&&c[0].type==="hardBreak"){const f=findChild(n,"w:r");if((f&&findChild(f,"w:br"))?.attributes["w:type"]==="page")return{type:"horizontalRule"}}return c.length===1&&c[0].type==="image"?c[0]:{type:"paragraph",...Object.keys(l).length&&{attrs:l},content:c}}function w(n){const t=[],e=r=>{if(r.name==="w:r")t.push(r);else for(const i of r.children)i.type==="element"&&e(i)};return e(n),t.some(r=>findChild(r,"w:br")?.attributes["w:type"]==="page")}async function h$1(n,t,e,r){return{type:"heading",attrs:{level:r,...m$1(n)},content:await extractRuns(n,{...t,styleInfo:e})}}function parseBorder(n){if(!n)return null;const t=n.attributes["w:val"],e=n.attributes["w:sz"],r=n.attributes["w:color"],i={single:"solid",dashed:"dashed",dotted:"dotted",double:"double",none:"none",nil:"none"},c={};if(r&&r!=="auto"&&(c.color=`#${r}`),e){const l=parseInt(e);isNaN(l)||(c.width=Math.round(l/6))}return t&&i[t]&&(c.style=i[t]),Object.keys(c).length>0?c:null}function parseTableProperties(n){const t={marginTop:void 0,marginBottom:void 0,marginLeft:void 0,marginRight:void 0},e=findChild(n,"w:tblPr");if(!e)return null;const r=findChild(e,"w:tblCellMar");if(!r)return null;const i=findChild(r,"w:top");if(i?.attributes["w:w"]){const b=parseInt(i.attributes["w:w"]);isNaN(b)||(t.marginTop=b)}const c=findChild(r,"w:bottom");if(c?.attributes["w:w"]){const b=parseInt(c.attributes["w:w"]);isNaN(b)||(t.marginBottom=b)}const l=findChild(r,"w:left");if(l?.attributes["w:w"]){const b=parseInt(l.attributes["w:w"]);isNaN(b)||(t.marginLeft=b)}const f=findChild(r,"w:right");if(f?.attributes["w:w"]){const b=parseInt(f.attributes["w:w"]);isNaN(b)||(t.marginRight=b)}return t.marginTop===void 0&&t.marginBottom===void 0&&t.marginLeft===void 0&&t.marginRight===void 0?null:t}function parseRowProperties(n){const t={rowHeight:null},e=findChild(n,"w:trPr");if(!e)return t;const r=findChild(e,"w:trHeight");if(r?.attributes["w:val"]){const i=parseInt(r.attributes["w:val"]),c=Math.round(i/15);t.rowHeight=`${c}px`}return t}function parseCellProperties(n){const t={colSpan:1,rowSpan:1,colWidth:null},e=findChild(n,"w:tcPr");if(!e)return t;const r=findChild(e,"w:gridSpan");r?.attributes["w:val"]&&(t.colSpan=parseInt(r.attributes["w:val"])),findChild(e,"w:vMerge")?.attributes["w:val"]==="continue"&&(t.rowSpan=0);const i=findChild(e,"w:tcW");if(i?.attributes["w:w"]){const b=parseInt(i.attributes["w:w"]);t.colWidth=Math.round(b/15)}const c=findChild(e,"w:shd");c?.attributes["w:fill"]&&(t.backgroundColor=`#${c.attributes["w:fill"]}`);const l=findChild(e,"w:vAlign");l?.attributes["w:val"]&&(t.verticalAlign=l.attributes["w:val"]);const f=findChild(e,"w:tcBorders");if(f){const b=parseBorder(findChild(f,"w:top"));b&&(t.borderTop=b);const I=parseBorder(findChild(f,"w:bottom"));I&&(t.borderBottom=I);const k=parseBorder(findChild(f,"w:left"));k&&(t.borderLeft=k);const R=parseBorder(findChild(f,"w:right"));R&&(t.borderRight=R)}return t}function isTable(n){return n.name==="w:tbl"}async function convertTable(n,t){const e=[];for(const l of n.children)l.type==="element"&&l.name==="w:tr"&&e.push(l);const r=new Map,i=await Promise.all(e.map((l,f)=>u(l,{...t,activeRowspans:r,rows:e,rowIndex:f}))),c=parseTableProperties(n);return{type:"table",...c&&{attrs:c},content:i}}async function u(n,t){const e=[];let r=0;const i=parseRowProperties(n);for(const c of n.children){if(c.type!=="element"||c.name!=="w:tc")continue;const l=t.activeRowspans.get(r);if(l&&l>0){t.activeRowspans.set(r,l-1),r++;continue}let f=parseCellProperties(c);if(f?.rowSpan===1){const I=y$1({...t,colIndex:r});I>1&&(f={...f,rowSpan:I})}if(f?.rowSpan&&f.rowSpan>1&&t.activeRowspans.set(r,f.rowSpan-1),f?.rowSpan===0){r++;continue}const b=await g(c,t);e.push({type:"tableCell",...f&&{attrs:f},content:b}),r+=f?.colSpan||1}return{type:"tableRow",...i&&{attrs:i},content:e}}function y$1(n){let t=1,e=n.colIndex;for(let r=n.rowIndex+1;r<n.rows.length;r++){const i=n.rows[r];let c=!1;for(const l of i.children){if(l.type!=="element"||l.name!=="w:tc")continue;const f=parseCellProperties(l),b=f?.colSpan||1;if(e>=0&&e<b){if(f?.rowSpan===0)t++,c=!0;else return t;break}e-=b}if(!c)break}return t}async function g(n,t){const e=[];for(const r of n.children)if(r.type==="element"&&r.name==="w:p"){const i=await convertParagraph(r,t);Array.isArray(i)?e.push(...i):e.push(i)}return e.length?e:[{type:"paragraph",content:[]}]}function isListItem(n){const t=findChild(n,"w:pPr");return!!t&&findChild(t,"w:numPr")!==void 0}function getListInfo(n){const t=findChild(n,"w:pPr"),e=t&&findChild(t,"w:numPr");if(!e)return null;const r=findChild(e,"w:ilvl"),i=findChild(e,"w:numId");return!r||!i?null:{numId:i.attributes["w:val"],level:parseInt(r.attributes["w:val"]||"0",10)}}function isCodeBlock(n){const t=findChild(n,"w:pPr"),e=(t&&findChild(t,"w:pStyle"))?.attributes["w:val"];return e==="CodeBlock"||e?.startsWith("Code")||!1}function getCodeBlockLanguage(n){const t=findChild(n,"w:pPr"),e=(t&&findChild(t,"w:pStyle"))?.attributes["w:val"];return e?.startsWith("CodeBlock")&&e.replace("CodeBlock","").toLowerCase()||void 0}const x="\u2610",a="\u2611";function m(n){const t=findChild(n,"w:r");if(!t)return null;const e=findChild(t,"w:t");if(!e)return null;const r=e.children.find(i=>i.type==="text");return r?.value&&r||null}function isTaskItem(n){const t=m(n);if(!t)return!1;const e=t.value;return e.startsWith(x)||e.startsWith(a)}function getTaskItemChecked(n){return m(n)?.value.startsWith(a)||!1}function convertTaskItem(n){return{type:"taskItem",attrs:{checked:getTaskItemChecked(n)},content:[h(n)]}}function h(n){const t=[];let e=!1;for(const i of n.children){if(i.type!=="element"||i.name!=="w:r")continue;if(!e){const f=findChild(i,"w:t")?.children.find(b=>b.type==="text");if(f?.value){const b=f.value;if(b.startsWith(x)||b.startsWith(a)){e=!0;const I=b.substring(2).trimStart();I&&t.push({type:"text",text:I});continue}}}const c=y(i),l=findChild(i,"w:t")?.children.find(f=>f.type==="text");if(l?.value){const f={type:"text",text:l.value};c.length&&(f.marks=c),t.push(f)}}const r=extractAlignment(n);return{type:"paragraph",...r&&{attrs:r},content:t.length?t:void 0}}function y(n){const t=[],e=findChild(n,"w:rPr");return e&&(findChild(e,"w:b")&&t.push({type:"bold"}),findChild(e,"w:i")&&t.push({type:"italic"}),findChild(e,"w:u")&&t.push({type:"underline"}),findChild(e,"w:strike")&&t.push({type:"strike"})),t}function isHorizontalRule(n){const t=findChild(n,"w:r");if(!t)return!1;let e=!1,r=!1;for(const i of t.children)i.type==="element"&&(i.name==="w:br"&&i.attributes["w:type"]==="page"?e=!0:i.name==="w:t"?i.children.find(c=>c.type==="text")?.value?.trim().length&&(r=!0):i.name!=="w:rPr"&&(r=!0));return e&&!r}const O=n=>{const t=[],e=findDeepChildren(n,"w:r");for(const r of e){const i=findChild(r,"w:t");if(!i)continue;const c=i.children.find(l=>l.type==="text");c&&"value"in c&&c.value&&t.push({type:"text",text:c.value})}return t},S$1=async(n,t,e)=>{const r=await convertTable(n[t],{hyperlinks:e.hyperlinks,images:e.images,options:e.options,styleMap:e.styleMap});let i=1;return t+1<n.length&&n[t+1].name==="w:p"&&C(n[t+1])&&i++,{nodes:[r],consumed:i}},J=async(n,t)=>{const e=[];let r=t;for(;r<n.length;){const i=n[r];if(i.name!=="w:p"||!isCodeBlock(i))break;const c=getCodeBlockLanguage(i),l={type:"codeBlock",...c&&{attrs:{language:c}},content:O(i)};e.push(l),r++}return{nodes:e,consumed:r-t}},M=async(n,t,e)=>{const{listTypeMap:r}=e,i=[];let c=t;for(;c<n.length;){const l=n[c];if(l.name!=="w:p"||!isListItem(l))break;const f=getListInfo(l);if(!f)break;const b=r.get(f.numId),I=b?.type||"bullet",k=[];for(;c<n.length;){const P=n[c];if(P.name!=="w:p"||!isListItem(P))break;const F=getListInfo(P);if(!F||F.numId!==f.numId)break;const $=await convertParagraph(P,e),W=Array.isArray($)?$[0]:$;k.push({type:"listItem",content:[W]}),c++}const R={type:I==="bullet"?"bulletList":"orderedList",content:k};I==="ordered"&&(R.attrs={type:null,...b?.start!==void 0&&{start:b.start}}),i.push(R)}return{nodes:i,consumed:c-t}},T=async(n,t)=>{const e=[];let r=t;for(;r<n.length;){const i=n[r];if(i.name!=="w:p"||!isTaskItem(i))break;const{convertTaskItem:c}=await import("./chunks/index.cjs"),l=c(i);e.push(l),r++}return{nodes:[{type:"taskList",content:e}],consumed:r-t}},v=async()=>({nodes:[{type:"horizontalRule"}],consumed:1}),L=async(n,t,e)=>{const r=await convertParagraph(n[t],e);return Array.isArray(r)?{nodes:r,consumed:1}:{nodes:[r],consumed:1}},C=n=>{const t=findDeepChildren(n,"w:r");for(const e of t){const r=findChild(e,"w:t");if(r){const c=r.children.find(l=>l.type==="text");if(c&&"value"in c&&c.value&&c.value.trim().length>0)return!1}if(findChild(e,"w:drawing")||findChild(e,"mc:AlternateContent")||findChild(e,"w:pict"))return!1;const i=findChild(e,"w:br");if(i&&i.attributes["w:type"]==="page")return!1}return!0},A=n=>n.name==="w:tbl"?S$1:n.name==="w:p"?isCodeBlock(n)?J:isTaskItem(n)?T:isListItem(n)?M:isHorizontalRule(n)?v:L:null,processElements=async(n,t)=>{const e=[];let r=0;for(;r<n.length;){const i=n[r],c=A(i);if(!c){r++;continue}if(i.name==="w:p"&&t.ignoreEmptyParagraphs&&C(i)){r++;continue}const{nodes:l,consumed:f}=await c(n,r,t);e.push(...l),r+=f}return e};async function parseDOCX(n,t={}){const{ignoreEmptyParagraphs:e=!1}=t,r=await undio.toUint8Array(n),i=fflate.unzipSync(r),c=extractHyperlinks(i),l=extractImages(i),f=i["word/document.xml"];if(!f)throw new Error("Invalid DOCX file: missing word/document.xml");const b=xastUtilFromXml.fromXml(new TextDecoder().decode(f)),I=parseNumberingXml(i),k=parseStylesXml(i);return await S(b,l,c,I,k,e,t)}async function S(n,t,e,r,i,c,l){if(n.type!=="root")return{type:"doc",content:[]};const f=findChild(n,"w:document");if(!f)return{type:"doc",content:[]};const b=findChild(f,"w:body");if(!b)return{type:"doc",content:[]};const I={hyperlinks:e,images:t,listTypeMap:r,styleMap:i,ignoreEmptyParagraphs:c,options:l};return{type:"doc",content:await processElements(b.children.filter(k=>k.type==="element"),I)}}exports.convertParagraph=convertParagraph,exports.convertTable=convertTable,exports.convertTaskItem=convertTaskItem,exports.extractAlignment=extractAlignment,exports.extractMarks=extractMarks,exports.extractRuns=extractRuns,exports.getCodeBlockLanguage=getCodeBlockLanguage,exports.getListInfo=getListInfo,exports.getTaskItemChecked=getTaskItemChecked,exports.isCodeBlock=isCodeBlock,exports.isHorizontalRule=isHorizontalRule,exports.isListItem=isListItem,exports.isTable=isTable,exports.isTaskItem=isTaskItem,exports.parseDOCX=parseDOCX;