@origints/mammoth 0.1.0

package/dist/convert.d.ts

@@ -0,0 +1,52 @@
+import { TransformAst, TransformImpl } from '@origints/core';
+import { DocxToHtmlOptions, DocxToTextOptions } from './options';
+/**
+ * Creates a TransformAst for converting DOCX to HTML.
+ *
+ * @example
+ * ```ts
+ * const plan = Planner.in(loadFile('document.docx'))
+ *   .mapIn(docxToHtml())
+ *   .emit((out, $) => out.add('html', $.get('html').asString()))
+ *   .compile()
+ * ```
+ *
+ * @example With custom style mapping
+ * ```ts
+ * const plan = Planner.in(loadFile('document.docx'))
+ *   .mapIn(docxToHtml({
+ *     styleMap: [
+ *       "p[style-name='Title'] => h1.document-title",
+ *       "p[style-name='Subtitle'] => h2.document-subtitle",
+ *     ],
+ *     idPrefix: 'doc-',
+ *   }))
+ *   .emit((out, $) => out.add('content', $.get('html').asString()))
+ *   .compile()
+ * ```
+ */
+export declare function docxToHtml(options?: DocxToHtmlOptions): TransformAst;
+/**
+ * Creates a TransformAst for extracting raw text from DOCX.
+ *
+ * @example
+ * ```ts
+ * const plan = Planner.in(loadFile('document.docx'))
+ *   .mapIn(docxToText())
+ *   .emit((out, $) => out.add('text', $.get('text').asString()))
+ *   .compile()
+ * ```
+ */
+export declare function docxToText(options?: DocxToTextOptions): TransformAst;
+/**
+ * Transform implementation for docxToHtml.
+ *
+ * Accepts Buffer or ReadableStream<Uint8Array> input.
+ */
+export declare const docxToHtmlImpl: TransformImpl;
+/**
+ * Transform implementation for docxToText.
+ *
+ * Accepts Buffer or ReadableStream<Uint8Array> input.
+ */
+export declare const docxToTextImpl: TransformImpl;

package/dist/index.cjs

@@ -0,0 +1,2 @@
+"use strict";Object.defineProperty(exports,Symbol.toStringTag,{value:"Module"});const m=require("mammoth");async function s(e){const r=e.getReader(),t=[];try{for(;;){const{done:a,value:n}=await r.read();if(a)break;t.push(n)}return Buffer.concat(t)}finally{r.releaseLock()}}function u(e){return{kind:"transform",namespace:"@origints/mammoth",name:"docxToHtml",args:g(e)}}function o(e){return{kind:"transform",namespace:"@origints/mammoth",name:"docxToText",args:e}}const c={namespace:"@origints/mammoth",name:"docxToHtml",async execute(e,r){const t=y(r),a=await d(e),n=p(t),i=await m.convertToHtml({buffer:a},n);return{html:i.value,messages:i.messages.map(f)}}},l={namespace:"@origints/mammoth",name:"docxToText",async execute(e){const r=await d(e),t=await m.extractRawText({buffer:r});return{text:t.value,messages:t.messages.map(f)}}};function g(e){if(!e)return;const r={};return e.styleMap&&(r.styleMap=e.styleMap),e.includeEmbeddedStyleMap!==void 0&&(r.includeEmbeddedStyleMap=e.includeEmbeddedStyleMap),e.includeDefaultStyleMap!==void 0&&(r.includeDefaultStyleMap=e.includeDefaultStyleMap),e.idPrefix&&(r.idPrefix=e.idPrefix),(e.imageHandling==="inline"||e.imageHandling==="omit")&&(r.imageHandling=e.imageHandling),e.preserveEmptyParagraphs!==void 0&&(r.preserveEmptyParagraphs=e.preserveEmptyParagraphs),Object.keys(r).length>0?r:void 0}function y(e){return e||{}}async function d(e){if(Buffer.isBuffer(e))return e;if(e instanceof ReadableStream)return s(e);throw new Error(`docxToHtml expects Buffer or ReadableStream input, got ${typeof e}`)}function p(e){const r={};return e.styleMap&&(r.styleMap=e.styleMap),e.includeEmbeddedStyleMap!==void 0&&(r.includeEmbeddedStyleMap=e.includeEmbeddedStyleMap),e.includeDefaultStyleMap!==void 0&&(r.includeDefaultStyleMap=e.includeDefaultStyleMap),e.idPrefix&&(r.idPrefix=e.idPrefix),e.preserveEmptyParagraphs===!1?r.ignoreEmptyParagraphs=!0:e.preserveEmptyParagraphs===!0&&(r.ignoreEmptyParagraphs=!1),e.imageHandling&&(r.convertImage=x(e.imageHandling)),e.transformDocument&&(r.transformDocument=e.transformDocument),r}function x(e){return e==="omit"?m.images.imgElement(()=>Promise.resolve({src:""})):e==="inline"?m.images.imgElement(r=>r.readAsBase64String().then(t=>({src:`data:${r.contentType};base64,${t}`}))):m.images.imgElement(async r=>{const t={contentType:r.contentType,read:(n=>n==="base64"?r.readAsBase64String():r.readAsArrayBuffer())},a=await e(t);return a===null?{src:""}:a})}function f(e){return{type:e.type==="error"?"error":"warning",message:e.message}}function T(e){e.register(c),e.register(l)}exports.docxToHtml=u;exports.docxToHtmlImpl=c;exports.docxToText=o;exports.docxToTextImpl=l;exports.registerMammothTransforms=T;exports.streamToBuffer=s;
+//# sourceMappingURL=index.cjs.map

package/dist/index.cjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.cjs","sources":["../src/util.ts","../src/convert.ts","../src/index.ts"],"sourcesContent":["/**\n * Utility functions for mammoth package.\n *\n * @module mammoth/util\n */\n\n/**\n * Convert a ReadableStream<Uint8Array> to a Buffer.\n */\nexport async function streamToBuffer(\n  stream: ReadableStream<Uint8Array>\n): Promise<Buffer> {\n  const reader = stream.getReader()\n  const chunks: Uint8Array[] = []\n\n  try {\n    while (true) {\n      const { done, value } = await reader.read()\n      if (done) break\n      chunks.push(value)\n    }\n    return Buffer.concat(chunks)\n  } finally {\n    reader.releaseLock()\n  }\n}\n","/**\n * DOCX to HTML conversion transform for Origins.\n *\n * @module mammoth/convert\n */\n\nimport mammoth from 'mammoth'\nimport type { TransformAst, TransformImpl } from '@origints/core'\nimport type {\n  DocxToHtmlOptions,\n  DocxToTextOptions,\n  MammothImageHandler,\n} from './options'\nimport type {\n  DocxConversionResult,\n  DocxTextResult,\n  MammothMessage,\n} from './result'\nimport { streamToBuffer } from './util'\n\n/**\n * Mammoth options type extracted from the library.\n */\ninterface MammothOptions {\n  styleMap?: string | string[]\n  includeEmbeddedStyleMap?: boolean\n  includeDefaultStyleMap?: boolean\n  convertImage?: MammothImageConverter\n  ignoreEmptyParagraphs?: boolean\n  idPrefix?: string\n  transformDocument?: (element: unknown) => unknown\n}\n\n/**\n * Mammoth image converter (opaque branded type).\n */\ninterface MammothImageConverter {\n  __mammothBrand: 'ImageConverter'\n}\n\n/**\n * Mammoth image interface for custom converters.\n */\ninterface MammothImage {\n  contentType: string\n  readAsBase64String: () => Promise<string>\n  readAsBuffer: () => Promise<Buffer>\n  readAsArrayBuffer: () => Promise<ArrayBuffer>\n}\n\n/**\n * Creates a TransformAst for converting DOCX to HTML.\n *\n * @example\n * ```ts\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToHtml())\n *   .emit((out, $) => out.add('html', $.get('html').asString()))\n *   .compile()\n * ```\n *\n * @example With custom style mapping\n * ```ts\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToHtml({\n *     styleMap: [\n *       \"p[style-name='Title'] => h1.document-title\",\n *       \"p[style-name='Subtitle'] => h2.document-subtitle\",\n *     ],\n *     idPrefix: 'doc-',\n *   }))\n *   .emit((out, $) => out.add('content', $.get('html').asString()))\n *   .compile()\n * ```\n */\nexport function docxToHtml(options?: DocxToHtmlOptions): TransformAst {\n  return {\n    kind: 'transform',\n    namespace: '@origints/mammoth',\n    name: 'docxToHtml',\n    args: serializeOptions(options),\n  }\n}\n\n/**\n * Creates a TransformAst for extracting raw text from DOCX.\n *\n * @example\n * ```ts\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToText())\n *   .emit((out, $) => out.add('text', $.get('text').asString()))\n *   .compile()\n * ```\n */\nexport function docxToText(options?: DocxToTextOptions): TransformAst {\n  return {\n    kind: 'transform',\n    namespace: '@origints/mammoth',\n    name: 'docxToText',\n    args: options,\n  }\n}\n\n/**\n * Transform implementation for docxToHtml.\n *\n * Accepts Buffer or ReadableStream<Uint8Array> input.\n */\nexport const docxToHtmlImpl: TransformImpl = {\n  namespace: '@origints/mammoth',\n  name: 'docxToHtml',\n\n  async execute(\n    input: unknown,\n    args?: unknown\n  ): Promise<DocxConversionResult> {\n    const options = deserializeOptions(\n      args as SerializedDocxOptions | undefined\n    )\n    const buffer = await toBuffer(input)\n    const mammothOptions = toMammothOptions(options)\n\n    const result = await mammoth.convertToHtml({ buffer }, mammothOptions)\n\n    return {\n      html: result.value,\n      messages: result.messages.map(toMammothMessage),\n    }\n  },\n}\n\n/**\n * Transform implementation for docxToText.\n *\n * Accepts Buffer or ReadableStream<Uint8Array> input.\n */\nexport const docxToTextImpl: TransformImpl = {\n  namespace: '@origints/mammoth',\n  name: 'docxToText',\n\n  async execute(input: unknown): Promise<DocxTextResult> {\n    const buffer = await toBuffer(input)\n\n    // Note: extractRawText doesn't accept options in mammoth's API\n    const result = await mammoth.extractRawText({ buffer })\n\n    return {\n      text: result.value,\n      messages: result.messages.map(toMammothMessage),\n    }\n  },\n}\n\n// ---------------------------------------------------------------------------\n// Internal helpers\n// ---------------------------------------------------------------------------\n\n/**\n * Serialized options that can be stored in TransformAst.args.\n * Function handlers are converted to string identifiers.\n */\ninterface SerializedDocxOptions {\n  styleMap?: string[]\n  includeEmbeddedStyleMap?: boolean\n  includeDefaultStyleMap?: boolean\n  idPrefix?: string\n  imageHandling?: 'inline' | 'omit'\n  preserveEmptyParagraphs?: boolean\n  // Note: transformDocument and custom imageHandling functions cannot be serialized\n}\n\n/**\n * Serialize options for storage in TransformAst.\n * Custom functions cannot be serialized and are dropped.\n */\nfunction serializeOptions(\n  options?: DocxToHtmlOptions\n): SerializedDocxOptions | undefined {\n  if (!options) return undefined\n\n  const serialized: SerializedDocxOptions = {}\n\n  if (options.styleMap) {\n    serialized.styleMap = options.styleMap\n  }\n  if (options.includeEmbeddedStyleMap !== undefined) {\n    serialized.includeEmbeddedStyleMap = options.includeEmbeddedStyleMap\n  }\n  if (options.includeDefaultStyleMap !== undefined) {\n    serialized.includeDefaultStyleMap = options.includeDefaultStyleMap\n  }\n  if (options.idPrefix) {\n    serialized.idPrefix = options.idPrefix\n  }\n  if (\n    options.imageHandling === 'inline' ||\n    options.imageHandling === 'omit'\n  ) {\n    serialized.imageHandling = options.imageHandling\n  }\n  if (options.preserveEmptyParagraphs !== undefined) {\n    serialized.preserveEmptyParagraphs = options.preserveEmptyParagraphs\n  }\n\n  return Object.keys(serialized).length > 0 ? serialized : undefined\n}\n\n/**\n * Deserialize options from TransformAst.args.\n */\nfunction deserializeOptions(\n  serialized?: SerializedDocxOptions\n): DocxToHtmlOptions {\n  if (!serialized) return {}\n  return serialized\n}\n\n/**\n * Convert input to Buffer.\n */\nasync function toBuffer(input: unknown): Promise<Buffer> {\n  if (Buffer.isBuffer(input)) {\n    return input\n  }\n  if (input instanceof ReadableStream) {\n    return streamToBuffer(input as ReadableStream<Uint8Array>)\n  }\n  throw new Error(\n    `docxToHtml expects Buffer or ReadableStream input, got ${typeof input}`\n  )\n}\n\n/**\n * Convert our options to mammoth options.\n */\nfunction toMammothOptions(options: DocxToHtmlOptions): MammothOptions {\n  const mammothOpts: MammothOptions = {}\n\n  if (options.styleMap) {\n    mammothOpts.styleMap = options.styleMap\n  }\n\n  if (options.includeEmbeddedStyleMap !== undefined) {\n    mammothOpts.includeEmbeddedStyleMap = options.includeEmbeddedStyleMap\n  }\n\n  if (options.includeDefaultStyleMap !== undefined) {\n    mammothOpts.includeDefaultStyleMap = options.includeDefaultStyleMap\n  }\n\n  if (options.idPrefix) {\n    mammothOpts.idPrefix = options.idPrefix\n  }\n\n  if (options.preserveEmptyParagraphs === false) {\n    mammothOpts.ignoreEmptyParagraphs = true\n  } else if (options.preserveEmptyParagraphs === true) {\n    mammothOpts.ignoreEmptyParagraphs = false\n  }\n\n  if (options.imageHandling) {\n    mammothOpts.convertImage = createImageConverter(options.imageHandling)\n  }\n\n  if (options.transformDocument) {\n    mammothOpts.transformDocument = options.transformDocument\n  }\n\n  return mammothOpts\n}\n\n/**\n * Create a mammoth image converter from our options.\n */\nfunction createImageConverter(\n  handling: 'inline' | 'omit' | MammothImageHandler\n): MammothImageConverter {\n  if (handling === 'omit') {\n    return mammoth.images.imgElement(() => Promise.resolve({ src: '' }))\n  }\n\n  if (handling === 'inline') {\n    // Use default mammoth behavior (base64 inline)\n    return mammoth.images.imgElement((image: MammothImage) =>\n      image.readAsBase64String().then(data => ({\n        src: `data:${image.contentType};base64,${data}`,\n      }))\n    )\n  }\n\n  // Custom handler - adapt our interface to mammoth's\n  return mammoth.images.imgElement(async (image: MammothImage) => {\n    // Adapt mammoth's Image to our MammothImageElement interface\n    const adaptedImage = {\n      contentType: image.contentType,\n      read: ((encoding: 'base64' | 'buffer') => {\n        if (encoding === 'base64') {\n          return image.readAsBase64String()\n        }\n        return image.readAsArrayBuffer()\n      }) as {\n        (encoding: 'base64'): Promise<string>\n        (encoding: 'buffer'): Promise<ArrayBuffer>\n      },\n    }\n\n    const result = await handling(adaptedImage)\n    if (result === null) {\n      return { src: '' }\n    }\n    return result\n  })\n}\n\n/**\n * Convert mammoth message to our message type.\n */\nfunction toMammothMessage(msg: { type: string; message: string }): MammothMessage {\n  return {\n    type: msg.type === 'error' ? 'error' : 'warning',\n    message: msg.message,\n  }\n}\n","/**\n * @origints/mammoth - DOCX to HTML conversion for Origins using mammoth.js\n *\n * This package provides transforms for converting Word documents (.docx) to HTML.\n * It wraps the mammoth.js library and exposes all its conversion options.\n *\n * @packageDocumentation\n *\n * @example Basic usage\n * ```ts\n * import { Planner, loadFile } from '@origints/core'\n * import { docxToHtml, registerMammothTransforms } from '@origints/mammoth'\n *\n * // Register transforms\n * registerMammothTransforms(globalRegistry)\n *\n * // Create a plan\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToHtml())\n *   .emit((out, $) => out.add('html', $.get('html').asString()))\n *   .compile()\n * ```\n *\n * @example With custom style mapping\n * ```ts\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToHtml({\n *     styleMap: [\n *       \"p[style-name='Title'] => h1.document-title\",\n *       \"p[style-name='Heading 1'] => h1\",\n *       \"p[style-name='Heading 2'] => h2\",\n *     ],\n *     idPrefix: 'doc-',\n *     imageHandling: 'omit',\n *   }))\n *   .emit((out, $) => out.add('content', $.get('html').asString()))\n *   .compile()\n * ```\n */\n\n// Re-export option types\nexport type {\n  DocxToHtmlOptions,\n  DocxToTextOptions,\n  MammothImageElement,\n  MammothImageHandler,\n  MammothImageResult,\n} from './options'\n\n// Re-export result types\nexport type {\n  DocxConversionResult,\n  DocxTextResult,\n  MammothMessage,\n  MammothMessageType,\n} from './result'\n\n// Re-export transform creators and implementations\nexport {\n  docxToHtml,\n  docxToText,\n  docxToHtmlImpl,\n  docxToTextImpl,\n} from './convert'\n\n// Re-export utilities\nexport { streamToBuffer } from './util'\n\n// ---------------------------------------------------------------------------\n// Auto-registration of transforms\n// ---------------------------------------------------------------------------\n\nimport { docxToHtmlImpl, docxToTextImpl } from './convert'\n\n/**\n * Register the mammoth transforms with a registry.\n * Call this to enable docxToHtml() and docxToText() in your plans.\n *\n * @example\n * ```ts\n * import { globalRegistry } from '@origints/core'\n * import { registerMammothTransforms } from '@origints/mammoth'\n *\n * registerMammothTransforms(globalRegistry)\n * ```\n */\nexport function registerMammothTransforms(registry: {\n  register(impl: {\n    namespace: string\n    name: string\n    execute: (...args: unknown[]) => unknown\n  }): void\n}): void {\n  registry.register(docxToHtmlImpl)\n  registry.register(docxToTextImpl)\n}\n"],"names":["streamToBuffer","stream","reader","chunks","done","value","docxToHtml","options","serializeOptions","docxToText","docxToHtmlImpl","input","args","deserializeOptions","buffer","toBuffer","mammothOptions","toMammothOptions","result","mammoth","toMammothMessage","docxToTextImpl","serialized","mammothOpts","createImageConverter","handling","image","data","adaptedImage","encoding","msg","registerMammothTransforms","registry"],"mappings":"2GASA,eAAsBA,EACpBC,EACiB,CACjB,MAAMC,EAASD,EAAO,UAAA,EAChBE,EAAuB,CAAA,EAE7B,GAAI,CACF,OAAa,CACX,KAAM,CAAE,KAAAC,EAAM,MAAAC,CAAA,EAAU,MAAMH,EAAO,KAAA,EACrC,GAAIE,EAAM,MACVD,EAAO,KAAKE,CAAK,CACnB,CACA,OAAO,OAAO,OAAOF,CAAM,CAC7B,QAAA,CACED,EAAO,YAAA,CACT,CACF,CCkDO,SAASI,EAAWC,EAA2C,CACpE,MAAO,CACL,KAAM,YACN,UAAW,oBACX,KAAM,aACN,KAAMC,EAAiBD,CAAO,CAAA,CAElC,CAaO,SAASE,EAAWF,EAA2C,CACpE,MAAO,CACL,KAAM,YACN,UAAW,oBACX,KAAM,aACN,KAAMA,CAAA,CAEV,CAOO,MAAMG,EAAgC,CAC3C,UAAW,oBACX,KAAM,aAEN,MAAM,QACJC,EACAC,EAC+B,CAC/B,MAAML,EAAUM,EACdD,CAAA,EAEIE,EAAS,MAAMC,EAASJ,CAAK,EAC7BK,EAAiBC,EAAiBV,CAAO,EAEzCW,EAAS,MAAMC,EAAQ,cAAc,CAAE,OAAAL,CAAA,EAAUE,CAAc,EAErE,MAAO,CACL,KAAME,EAAO,MACb,SAAUA,EAAO,SAAS,IAAIE,CAAgB,CAAA,CAElD,CACF,EAOaC,EAAgC,CAC3C,UAAW,oBACX,KAAM,aAEN,MAAM,QAAQV,EAAyC,CACrD,MAAMG,EAAS,MAAMC,EAASJ,CAAK,EAG7BO,EAAS,MAAMC,EAAQ,eAAe,CAAE,OAAAL,EAAQ,EAEtD,MAAO,CACL,KAAMI,EAAO,MACb,SAAUA,EAAO,SAAS,IAAIE,CAAgB,CAAA,CAElD,CACF,EAwBA,SAASZ,EACPD,EACmC,CACnC,GAAI,CAACA,EAAS,OAEd,MAAMe,EAAoC,CAAA,EAE1C,OAAIf,EAAQ,WACVe,EAAW,SAAWf,EAAQ,UAE5BA,EAAQ,0BAA4B,SACtCe,EAAW,wBAA0Bf,EAAQ,yBAE3CA,EAAQ,yBAA2B,SACrCe,EAAW,uBAAyBf,EAAQ,wBAE1CA,EAAQ,WACVe,EAAW,SAAWf,EAAQ,WAG9BA,EAAQ,gBAAkB,UAC1BA,EAAQ,gBAAkB,UAE1Be,EAAW,cAAgBf,EAAQ,eAEjCA,EAAQ,0BAA4B,SACtCe,EAAW,wBAA0Bf,EAAQ,yBAGxC,OAAO,KAAKe,CAAU,EAAE,OAAS,EAAIA,EAAa,MAC3D,CAKA,SAAST,EACPS,EACmB,CACnB,OAAKA,GAAmB,CAAA,CAE1B,CAKA,eAAeP,EAASJ,EAAiC,CACvD,GAAI,OAAO,SAASA,CAAK,EACvB,OAAOA,EAET,GAAIA,aAAiB,eACnB,OAAOX,EAAeW,CAAmC,EAE3D,MAAM,IAAI,MACR,0DAA0D,OAAOA,CAAK,EAAA,CAE1E,CAKA,SAASM,EAAiBV,EAA4C,CACpE,MAAMgB,EAA8B,CAAA,EAEpC,OAAIhB,EAAQ,WACVgB,EAAY,SAAWhB,EAAQ,UAG7BA,EAAQ,0BAA4B,SACtCgB,EAAY,wBAA0BhB,EAAQ,yBAG5CA,EAAQ,yBAA2B,SACrCgB,EAAY,uBAAyBhB,EAAQ,wBAG3CA,EAAQ,WACVgB,EAAY,SAAWhB,EAAQ,UAG7BA,EAAQ,0BAA4B,GACtCgB,EAAY,sBAAwB,GAC3BhB,EAAQ,0BAA4B,KAC7CgB,EAAY,sBAAwB,IAGlChB,EAAQ,gBACVgB,EAAY,aAAeC,EAAqBjB,EAAQ,aAAa,GAGnEA,EAAQ,oBACVgB,EAAY,kBAAoBhB,EAAQ,mBAGnCgB,CACT,CAKA,SAASC,EACPC,EACuB,CACvB,OAAIA,IAAa,OACRN,EAAQ,OAAO,WAAW,IAAM,QAAQ,QAAQ,CAAE,IAAK,EAAA,CAAI,CAAC,EAGjEM,IAAa,SAERN,EAAQ,OAAO,WAAYO,GAChCA,EAAM,mBAAA,EAAqB,KAAKC,IAAS,CACvC,IAAK,QAAQD,EAAM,WAAW,WAAWC,CAAI,EAAA,EAC7C,CAAA,EAKCR,EAAQ,OAAO,WAAW,MAAOO,GAAwB,CAE9D,MAAME,EAAe,CACnB,YAAaF,EAAM,YACnB,MAAQG,GACFA,IAAa,SACRH,EAAM,mBAAA,EAERA,EAAM,kBAAA,EACf,EAMIR,EAAS,MAAMO,EAASG,CAAY,EAC1C,OAAIV,IAAW,KACN,CAAE,IAAK,EAAA,EAETA,CACT,CAAC,CACH,CAKA,SAASE,EAAiBU,EAAwD,CAChF,MAAO,CACL,KAAMA,EAAI,OAAS,QAAU,QAAU,UACvC,QAASA,EAAI,OAAA,CAEjB,CC7OO,SAASC,EAA0BC,EAMjC,CACPA,EAAS,SAAStB,CAAc,EAChCsB,EAAS,SAASX,CAAc,CAClC"}

package/dist/index.d.ts

@@ -0,0 +1,62 @@
+/**
+ * @origints/mammoth - DOCX to HTML conversion for Origins using mammoth.js
+ *
+ * This package provides transforms for converting Word documents (.docx) to HTML.
+ * It wraps the mammoth.js library and exposes all its conversion options.
+ *
+ * @packageDocumentation
+ *
+ * @example Basic usage
+ * ```ts
+ * import { Planner, loadFile } from '@origints/core'
+ * import { docxToHtml, registerMammothTransforms } from '@origints/mammoth'
+ *
+ * // Register transforms
+ * registerMammothTransforms(globalRegistry)
+ *
+ * // Create a plan
+ * const plan = Planner.in(loadFile('document.docx'))
+ *   .mapIn(docxToHtml())
+ *   .emit((out, $) => out.add('html', $.get('html').asString()))
+ *   .compile()
+ * ```
+ *
+ * @example With custom style mapping
+ * ```ts
+ * const plan = Planner.in(loadFile('document.docx'))
+ *   .mapIn(docxToHtml({
+ *     styleMap: [
+ *       "p[style-name='Title'] => h1.document-title",
+ *       "p[style-name='Heading 1'] => h1",
+ *       "p[style-name='Heading 2'] => h2",
+ *     ],
+ *     idPrefix: 'doc-',
+ *     imageHandling: 'omit',
+ *   }))
+ *   .emit((out, $) => out.add('content', $.get('html').asString()))
+ *   .compile()
+ * ```
+ */
+export type { DocxToHtmlOptions, DocxToTextOptions, MammothImageElement, MammothImageHandler, MammothImageResult, } from './options';
+export type { DocxConversionResult, DocxTextResult, MammothMessage, MammothMessageType, } from './result';
+export { docxToHtml, docxToText, docxToHtmlImpl, docxToTextImpl, } from './convert';
+export { streamToBuffer } from './util';
+/**
+ * Register the mammoth transforms with a registry.
+ * Call this to enable docxToHtml() and docxToText() in your plans.
+ *
+ * @example
+ * ```ts
+ * import { globalRegistry } from '@origints/core'
+ * import { registerMammothTransforms } from '@origints/mammoth'
+ *
+ * registerMammothTransforms(globalRegistry)
+ * ```
+ */
+export declare function registerMammothTransforms(registry: {
+    register(impl: {
+        namespace: string;
+        name: string;
+        execute: (...args: unknown[]) => unknown;
+    }): void;
+}): void;

package/dist/index.es.js

@@ -0,0 +1,105 @@
+import m from "mammoth";
+async function f(e) {
+  const r = e.getReader(), a = [];
+  try {
+    for (; ; ) {
+      const { done: t, value: n } = await r.read();
+      if (t) break;
+      a.push(n);
+    }
+    return Buffer.concat(a);
+  } finally {
+    r.releaseLock();
+  }
+}
+function x(e) {
+  return {
+    kind: "transform",
+    namespace: "@origints/mammoth",
+    name: "docxToHtml",
+    args: u(e)
+  };
+}
+function h(e) {
+  return {
+    kind: "transform",
+    namespace: "@origints/mammoth",
+    name: "docxToText",
+    args: e
+  };
+}
+const l = {
+  namespace: "@origints/mammoth",
+  name: "docxToHtml",
+  async execute(e, r) {
+    const a = o(
+      r
+    ), t = await s(e), n = g(a), i = await m.convertToHtml({ buffer: t }, n);
+    return {
+      html: i.value,
+      messages: i.messages.map(c)
+    };
+  }
+}, d = {
+  namespace: "@origints/mammoth",
+  name: "docxToText",
+  async execute(e) {
+    const r = await s(e), a = await m.extractRawText({ buffer: r });
+    return {
+      text: a.value,
+      messages: a.messages.map(c)
+    };
+  }
+};
+function u(e) {
+  if (!e) return;
+  const r = {};
+  return e.styleMap && (r.styleMap = e.styleMap), e.includeEmbeddedStyleMap !== void 0 && (r.includeEmbeddedStyleMap = e.includeEmbeddedStyleMap), e.includeDefaultStyleMap !== void 0 && (r.includeDefaultStyleMap = e.includeDefaultStyleMap), e.idPrefix && (r.idPrefix = e.idPrefix), (e.imageHandling === "inline" || e.imageHandling === "omit") && (r.imageHandling = e.imageHandling), e.preserveEmptyParagraphs !== void 0 && (r.preserveEmptyParagraphs = e.preserveEmptyParagraphs), Object.keys(r).length > 0 ? r : void 0;
+}
+function o(e) {
+  return e || {};
+}
+async function s(e) {
+  if (Buffer.isBuffer(e))
+    return e;
+  if (e instanceof ReadableStream)
+    return f(e);
+  throw new Error(
+    `docxToHtml expects Buffer or ReadableStream input, got ${typeof e}`
+  );
+}
+function g(e) {
+  const r = {};
+  return e.styleMap && (r.styleMap = e.styleMap), e.includeEmbeddedStyleMap !== void 0 && (r.includeEmbeddedStyleMap = e.includeEmbeddedStyleMap), e.includeDefaultStyleMap !== void 0 && (r.includeDefaultStyleMap = e.includeDefaultStyleMap), e.idPrefix && (r.idPrefix = e.idPrefix), e.preserveEmptyParagraphs === !1 ? r.ignoreEmptyParagraphs = !0 : e.preserveEmptyParagraphs === !0 && (r.ignoreEmptyParagraphs = !1), e.imageHandling && (r.convertImage = y(e.imageHandling)), e.transformDocument && (r.transformDocument = e.transformDocument), r;
+}
+function y(e) {
+  return e === "omit" ? m.images.imgElement(() => Promise.resolve({ src: "" })) : e === "inline" ? m.images.imgElement(
+    (r) => r.readAsBase64String().then((a) => ({
+      src: `data:${r.contentType};base64,${a}`
+    }))
+  ) : m.images.imgElement(async (r) => {
+    const a = {
+      contentType: r.contentType,
+      read: ((n) => n === "base64" ? r.readAsBase64String() : r.readAsArrayBuffer())
+    }, t = await e(a);
+    return t === null ? { src: "" } : t;
+  });
+}
+function c(e) {
+  return {
+    type: e.type === "error" ? "error" : "warning",
+    message: e.message
+  };
+}
+function M(e) {
+  e.register(l), e.register(d);
+}
+export {
+  x as docxToHtml,
+  l as docxToHtmlImpl,
+  h as docxToText,
+  d as docxToTextImpl,
+  M as registerMammothTransforms,
+  f as streamToBuffer
+};
+//# sourceMappingURL=index.es.js.map

package/dist/index.es.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.es.js","sources":["../src/util.ts","../src/convert.ts","../src/index.ts"],"sourcesContent":["/**\n * Utility functions for mammoth package.\n *\n * @module mammoth/util\n */\n\n/**\n * Convert a ReadableStream<Uint8Array> to a Buffer.\n */\nexport async function streamToBuffer(\n  stream: ReadableStream<Uint8Array>\n): Promise<Buffer> {\n  const reader = stream.getReader()\n  const chunks: Uint8Array[] = []\n\n  try {\n    while (true) {\n      const { done, value } = await reader.read()\n      if (done) break\n      chunks.push(value)\n    }\n    return Buffer.concat(chunks)\n  } finally {\n    reader.releaseLock()\n  }\n}\n","/**\n * DOCX to HTML conversion transform for Origins.\n *\n * @module mammoth/convert\n */\n\nimport mammoth from 'mammoth'\nimport type { TransformAst, TransformImpl } from '@origints/core'\nimport type {\n  DocxToHtmlOptions,\n  DocxToTextOptions,\n  MammothImageHandler,\n} from './options'\nimport type {\n  DocxConversionResult,\n  DocxTextResult,\n  MammothMessage,\n} from './result'\nimport { streamToBuffer } from './util'\n\n/**\n * Mammoth options type extracted from the library.\n */\ninterface MammothOptions {\n  styleMap?: string | string[]\n  includeEmbeddedStyleMap?: boolean\n  includeDefaultStyleMap?: boolean\n  convertImage?: MammothImageConverter\n  ignoreEmptyParagraphs?: boolean\n  idPrefix?: string\n  transformDocument?: (element: unknown) => unknown\n}\n\n/**\n * Mammoth image converter (opaque branded type).\n */\ninterface MammothImageConverter {\n  __mammothBrand: 'ImageConverter'\n}\n\n/**\n * Mammoth image interface for custom converters.\n */\ninterface MammothImage {\n  contentType: string\n  readAsBase64String: () => Promise<string>\n  readAsBuffer: () => Promise<Buffer>\n  readAsArrayBuffer: () => Promise<ArrayBuffer>\n}\n\n/**\n * Creates a TransformAst for converting DOCX to HTML.\n *\n * @example\n * ```ts\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToHtml())\n *   .emit((out, $) => out.add('html', $.get('html').asString()))\n *   .compile()\n * ```\n *\n * @example With custom style mapping\n * ```ts\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToHtml({\n *     styleMap: [\n *       \"p[style-name='Title'] => h1.document-title\",\n *       \"p[style-name='Subtitle'] => h2.document-subtitle\",\n *     ],\n *     idPrefix: 'doc-',\n *   }))\n *   .emit((out, $) => out.add('content', $.get('html').asString()))\n *   .compile()\n * ```\n */\nexport function docxToHtml(options?: DocxToHtmlOptions): TransformAst {\n  return {\n    kind: 'transform',\n    namespace: '@origints/mammoth',\n    name: 'docxToHtml',\n    args: serializeOptions(options),\n  }\n}\n\n/**\n * Creates a TransformAst for extracting raw text from DOCX.\n *\n * @example\n * ```ts\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToText())\n *   .emit((out, $) => out.add('text', $.get('text').asString()))\n *   .compile()\n * ```\n */\nexport function docxToText(options?: DocxToTextOptions): TransformAst {\n  return {\n    kind: 'transform',\n    namespace: '@origints/mammoth',\n    name: 'docxToText',\n    args: options,\n  }\n}\n\n/**\n * Transform implementation for docxToHtml.\n *\n * Accepts Buffer or ReadableStream<Uint8Array> input.\n */\nexport const docxToHtmlImpl: TransformImpl = {\n  namespace: '@origints/mammoth',\n  name: 'docxToHtml',\n\n  async execute(\n    input: unknown,\n    args?: unknown\n  ): Promise<DocxConversionResult> {\n    const options = deserializeOptions(\n      args as SerializedDocxOptions | undefined\n    )\n    const buffer = await toBuffer(input)\n    const mammothOptions = toMammothOptions(options)\n\n    const result = await mammoth.convertToHtml({ buffer }, mammothOptions)\n\n    return {\n      html: result.value,\n      messages: result.messages.map(toMammothMessage),\n    }\n  },\n}\n\n/**\n * Transform implementation for docxToText.\n *\n * Accepts Buffer or ReadableStream<Uint8Array> input.\n */\nexport const docxToTextImpl: TransformImpl = {\n  namespace: '@origints/mammoth',\n  name: 'docxToText',\n\n  async execute(input: unknown): Promise<DocxTextResult> {\n    const buffer = await toBuffer(input)\n\n    // Note: extractRawText doesn't accept options in mammoth's API\n    const result = await mammoth.extractRawText({ buffer })\n\n    return {\n      text: result.value,\n      messages: result.messages.map(toMammothMessage),\n    }\n  },\n}\n\n// ---------------------------------------------------------------------------\n// Internal helpers\n// ---------------------------------------------------------------------------\n\n/**\n * Serialized options that can be stored in TransformAst.args.\n * Function handlers are converted to string identifiers.\n */\ninterface SerializedDocxOptions {\n  styleMap?: string[]\n  includeEmbeddedStyleMap?: boolean\n  includeDefaultStyleMap?: boolean\n  idPrefix?: string\n  imageHandling?: 'inline' | 'omit'\n  preserveEmptyParagraphs?: boolean\n  // Note: transformDocument and custom imageHandling functions cannot be serialized\n}\n\n/**\n * Serialize options for storage in TransformAst.\n * Custom functions cannot be serialized and are dropped.\n */\nfunction serializeOptions(\n  options?: DocxToHtmlOptions\n): SerializedDocxOptions | undefined {\n  if (!options) return undefined\n\n  const serialized: SerializedDocxOptions = {}\n\n  if (options.styleMap) {\n    serialized.styleMap = options.styleMap\n  }\n  if (options.includeEmbeddedStyleMap !== undefined) {\n    serialized.includeEmbeddedStyleMap = options.includeEmbeddedStyleMap\n  }\n  if (options.includeDefaultStyleMap !== undefined) {\n    serialized.includeDefaultStyleMap = options.includeDefaultStyleMap\n  }\n  if (options.idPrefix) {\n    serialized.idPrefix = options.idPrefix\n  }\n  if (\n    options.imageHandling === 'inline' ||\n    options.imageHandling === 'omit'\n  ) {\n    serialized.imageHandling = options.imageHandling\n  }\n  if (options.preserveEmptyParagraphs !== undefined) {\n    serialized.preserveEmptyParagraphs = options.preserveEmptyParagraphs\n  }\n\n  return Object.keys(serialized).length > 0 ? serialized : undefined\n}\n\n/**\n * Deserialize options from TransformAst.args.\n */\nfunction deserializeOptions(\n  serialized?: SerializedDocxOptions\n): DocxToHtmlOptions {\n  if (!serialized) return {}\n  return serialized\n}\n\n/**\n * Convert input to Buffer.\n */\nasync function toBuffer(input: unknown): Promise<Buffer> {\n  if (Buffer.isBuffer(input)) {\n    return input\n  }\n  if (input instanceof ReadableStream) {\n    return streamToBuffer(input as ReadableStream<Uint8Array>)\n  }\n  throw new Error(\n    `docxToHtml expects Buffer or ReadableStream input, got ${typeof input}`\n  )\n}\n\n/**\n * Convert our options to mammoth options.\n */\nfunction toMammothOptions(options: DocxToHtmlOptions): MammothOptions {\n  const mammothOpts: MammothOptions = {}\n\n  if (options.styleMap) {\n    mammothOpts.styleMap = options.styleMap\n  }\n\n  if (options.includeEmbeddedStyleMap !== undefined) {\n    mammothOpts.includeEmbeddedStyleMap = options.includeEmbeddedStyleMap\n  }\n\n  if (options.includeDefaultStyleMap !== undefined) {\n    mammothOpts.includeDefaultStyleMap = options.includeDefaultStyleMap\n  }\n\n  if (options.idPrefix) {\n    mammothOpts.idPrefix = options.idPrefix\n  }\n\n  if (options.preserveEmptyParagraphs === false) {\n    mammothOpts.ignoreEmptyParagraphs = true\n  } else if (options.preserveEmptyParagraphs === true) {\n    mammothOpts.ignoreEmptyParagraphs = false\n  }\n\n  if (options.imageHandling) {\n    mammothOpts.convertImage = createImageConverter(options.imageHandling)\n  }\n\n  if (options.transformDocument) {\n    mammothOpts.transformDocument = options.transformDocument\n  }\n\n  return mammothOpts\n}\n\n/**\n * Create a mammoth image converter from our options.\n */\nfunction createImageConverter(\n  handling: 'inline' | 'omit' | MammothImageHandler\n): MammothImageConverter {\n  if (handling === 'omit') {\n    return mammoth.images.imgElement(() => Promise.resolve({ src: '' }))\n  }\n\n  if (handling === 'inline') {\n    // Use default mammoth behavior (base64 inline)\n    return mammoth.images.imgElement((image: MammothImage) =>\n      image.readAsBase64String().then(data => ({\n        src: `data:${image.contentType};base64,${data}`,\n      }))\n    )\n  }\n\n  // Custom handler - adapt our interface to mammoth's\n  return mammoth.images.imgElement(async (image: MammothImage) => {\n    // Adapt mammoth's Image to our MammothImageElement interface\n    const adaptedImage = {\n      contentType: image.contentType,\n      read: ((encoding: 'base64' | 'buffer') => {\n        if (encoding === 'base64') {\n          return image.readAsBase64String()\n        }\n        return image.readAsArrayBuffer()\n      }) as {\n        (encoding: 'base64'): Promise<string>\n        (encoding: 'buffer'): Promise<ArrayBuffer>\n      },\n    }\n\n    const result = await handling(adaptedImage)\n    if (result === null) {\n      return { src: '' }\n    }\n    return result\n  })\n}\n\n/**\n * Convert mammoth message to our message type.\n */\nfunction toMammothMessage(msg: { type: string; message: string }): MammothMessage {\n  return {\n    type: msg.type === 'error' ? 'error' : 'warning',\n    message: msg.message,\n  }\n}\n","/**\n * @origints/mammoth - DOCX to HTML conversion for Origins using mammoth.js\n *\n * This package provides transforms for converting Word documents (.docx) to HTML.\n * It wraps the mammoth.js library and exposes all its conversion options.\n *\n * @packageDocumentation\n *\n * @example Basic usage\n * ```ts\n * import { Planner, loadFile } from '@origints/core'\n * import { docxToHtml, registerMammothTransforms } from '@origints/mammoth'\n *\n * // Register transforms\n * registerMammothTransforms(globalRegistry)\n *\n * // Create a plan\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToHtml())\n *   .emit((out, $) => out.add('html', $.get('html').asString()))\n *   .compile()\n * ```\n *\n * @example With custom style mapping\n * ```ts\n * const plan = Planner.in(loadFile('document.docx'))\n *   .mapIn(docxToHtml({\n *     styleMap: [\n *       \"p[style-name='Title'] => h1.document-title\",\n *       \"p[style-name='Heading 1'] => h1\",\n *       \"p[style-name='Heading 2'] => h2\",\n *     ],\n *     idPrefix: 'doc-',\n *     imageHandling: 'omit',\n *   }))\n *   .emit((out, $) => out.add('content', $.get('html').asString()))\n *   .compile()\n * ```\n */\n\n// Re-export option types\nexport type {\n  DocxToHtmlOptions,\n  DocxToTextOptions,\n  MammothImageElement,\n  MammothImageHandler,\n  MammothImageResult,\n} from './options'\n\n// Re-export result types\nexport type {\n  DocxConversionResult,\n  DocxTextResult,\n  MammothMessage,\n  MammothMessageType,\n} from './result'\n\n// Re-export transform creators and implementations\nexport {\n  docxToHtml,\n  docxToText,\n  docxToHtmlImpl,\n  docxToTextImpl,\n} from './convert'\n\n// Re-export utilities\nexport { streamToBuffer } from './util'\n\n// ---------------------------------------------------------------------------\n// Auto-registration of transforms\n// ---------------------------------------------------------------------------\n\nimport { docxToHtmlImpl, docxToTextImpl } from './convert'\n\n/**\n * Register the mammoth transforms with a registry.\n * Call this to enable docxToHtml() and docxToText() in your plans.\n *\n * @example\n * ```ts\n * import { globalRegistry } from '@origints/core'\n * import { registerMammothTransforms } from '@origints/mammoth'\n *\n * registerMammothTransforms(globalRegistry)\n * ```\n */\nexport function registerMammothTransforms(registry: {\n  register(impl: {\n    namespace: string\n    name: string\n    execute: (...args: unknown[]) => unknown\n  }): void\n}): void {\n  registry.register(docxToHtmlImpl)\n  registry.register(docxToTextImpl)\n}\n"],"names":["streamToBuffer","stream","reader","chunks","done","value","docxToHtml","options","serializeOptions","docxToText","docxToHtmlImpl","input","args","deserializeOptions","buffer","toBuffer","mammothOptions","toMammothOptions","result","mammoth","toMammothMessage","docxToTextImpl","serialized","mammothOpts","createImageConverter","handling","image","data","adaptedImage","encoding","msg","registerMammothTransforms","registry"],"mappings":";AASA,eAAsBA,EACpBC,GACiB;AACjB,QAAMC,IAASD,EAAO,UAAA,GAChBE,IAAuB,CAAA;AAE7B,MAAI;AACF,eAAa;AACX,YAAM,EAAE,MAAAC,GAAM,OAAAC,EAAA,IAAU,MAAMH,EAAO,KAAA;AACrC,UAAIE,EAAM;AACV,MAAAD,EAAO,KAAKE,CAAK;AAAA,IACnB;AACA,WAAO,OAAO,OAAOF,CAAM;AAAA,EAC7B,UAAA;AACE,IAAAD,EAAO,YAAA;AAAA,EACT;AACF;ACkDO,SAASI,EAAWC,GAA2C;AACpE,SAAO;AAAA,IACL,MAAM;AAAA,IACN,WAAW;AAAA,IACX,MAAM;AAAA,IACN,MAAMC,EAAiBD,CAAO;AAAA,EAAA;AAElC;AAaO,SAASE,EAAWF,GAA2C;AACpE,SAAO;AAAA,IACL,MAAM;AAAA,IACN,WAAW;AAAA,IACX,MAAM;AAAA,IACN,MAAMA;AAAA,EAAA;AAEV;AAOO,MAAMG,IAAgC;AAAA,EAC3C,WAAW;AAAA,EACX,MAAM;AAAA,EAEN,MAAM,QACJC,GACAC,GAC+B;AAC/B,UAAML,IAAUM;AAAA,MACdD;AAAA,IAAA,GAEIE,IAAS,MAAMC,EAASJ,CAAK,GAC7BK,IAAiBC,EAAiBV,CAAO,GAEzCW,IAAS,MAAMC,EAAQ,cAAc,EAAE,QAAAL,EAAA,GAAUE,CAAc;AAErE,WAAO;AAAA,MACL,MAAME,EAAO;AAAA,MACb,UAAUA,EAAO,SAAS,IAAIE,CAAgB;AAAA,IAAA;AAAA,EAElD;AACF,GAOaC,IAAgC;AAAA,EAC3C,WAAW;AAAA,EACX,MAAM;AAAA,EAEN,MAAM,QAAQV,GAAyC;AACrD,UAAMG,IAAS,MAAMC,EAASJ,CAAK,GAG7BO,IAAS,MAAMC,EAAQ,eAAe,EAAE,QAAAL,GAAQ;AAEtD,WAAO;AAAA,MACL,MAAMI,EAAO;AAAA,MACb,UAAUA,EAAO,SAAS,IAAIE,CAAgB;AAAA,IAAA;AAAA,EAElD;AACF;AAwBA,SAASZ,EACPD,GACmC;AACnC,MAAI,CAACA,EAAS;AAEd,QAAMe,IAAoC,CAAA;AAE1C,SAAIf,EAAQ,aACVe,EAAW,WAAWf,EAAQ,WAE5BA,EAAQ,4BAA4B,WACtCe,EAAW,0BAA0Bf,EAAQ,0BAE3CA,EAAQ,2BAA2B,WACrCe,EAAW,yBAAyBf,EAAQ,yBAE1CA,EAAQ,aACVe,EAAW,WAAWf,EAAQ,YAG9BA,EAAQ,kBAAkB,YAC1BA,EAAQ,kBAAkB,YAE1Be,EAAW,gBAAgBf,EAAQ,gBAEjCA,EAAQ,4BAA4B,WACtCe,EAAW,0BAA0Bf,EAAQ,0BAGxC,OAAO,KAAKe,CAAU,EAAE,SAAS,IAAIA,IAAa;AAC3D;AAKA,SAAST,EACPS,GACmB;AACnB,SAAKA,KAAmB,CAAA;AAE1B;AAKA,eAAeP,EAASJ,GAAiC;AACvD,MAAI,OAAO,SAASA,CAAK;AACvB,WAAOA;AAET,MAAIA,aAAiB;AACnB,WAAOX,EAAeW,CAAmC;AAE3D,QAAM,IAAI;AAAA,IACR,0DAA0D,OAAOA,CAAK;AAAA,EAAA;AAE1E;AAKA,SAASM,EAAiBV,GAA4C;AACpE,QAAMgB,IAA8B,CAAA;AAEpC,SAAIhB,EAAQ,aACVgB,EAAY,WAAWhB,EAAQ,WAG7BA,EAAQ,4BAA4B,WACtCgB,EAAY,0BAA0BhB,EAAQ,0BAG5CA,EAAQ,2BAA2B,WACrCgB,EAAY,yBAAyBhB,EAAQ,yBAG3CA,EAAQ,aACVgB,EAAY,WAAWhB,EAAQ,WAG7BA,EAAQ,4BAA4B,KACtCgB,EAAY,wBAAwB,KAC3BhB,EAAQ,4BAA4B,OAC7CgB,EAAY,wBAAwB,KAGlChB,EAAQ,kBACVgB,EAAY,eAAeC,EAAqBjB,EAAQ,aAAa,IAGnEA,EAAQ,sBACVgB,EAAY,oBAAoBhB,EAAQ,oBAGnCgB;AACT;AAKA,SAASC,EACPC,GACuB;AACvB,SAAIA,MAAa,SACRN,EAAQ,OAAO,WAAW,MAAM,QAAQ,QAAQ,EAAE,KAAK,GAAA,CAAI,CAAC,IAGjEM,MAAa,WAERN,EAAQ,OAAO;AAAA,IAAW,CAACO,MAChCA,EAAM,mBAAA,EAAqB,KAAK,CAAAC,OAAS;AAAA,MACvC,KAAK,QAAQD,EAAM,WAAW,WAAWC,CAAI;AAAA,IAAA,EAC7C;AAAA,EAAA,IAKCR,EAAQ,OAAO,WAAW,OAAOO,MAAwB;AAE9D,UAAME,IAAe;AAAA,MACnB,aAAaF,EAAM;AAAA,MACnB,OAAO,CAACG,MACFA,MAAa,WACRH,EAAM,mBAAA,IAERA,EAAM,kBAAA;AAAA,IACf,GAMIR,IAAS,MAAMO,EAASG,CAAY;AAC1C,WAAIV,MAAW,OACN,EAAE,KAAK,GAAA,IAETA;AAAA,EACT,CAAC;AACH;AAKA,SAASE,EAAiBU,GAAwD;AAChF,SAAO;AAAA,IACL,MAAMA,EAAI,SAAS,UAAU,UAAU;AAAA,IACvC,SAASA,EAAI;AAAA,EAAA;AAEjB;AC7OO,SAASC,EAA0BC,GAMjC;AACP,EAAAA,EAAS,SAAStB,CAAc,GAChCsB,EAAS,SAASX,CAAc;AAClC;"}

package/dist/options.d.ts

@@ -0,0 +1,136 @@
+/**
+ * Options for DOCX to HTML conversion using mammoth.
+ *
+ * @module mammoth/options
+ */
+/**
+ * Image element provided during image conversion.
+ * Contains metadata and methods to read image content.
+ */
+export interface MammothImageElement {
+    /** MIME content type of the image (e.g., 'image/png', 'image/jpeg') */
+    contentType: string;
+    /**
+     * Read the image content.
+     *
+     * @param encoding - 'base64' for base64 string, 'buffer' for ArrayBuffer
+     * @returns Promise resolving to the image data
+     */
+    read(encoding: 'base64'): Promise<string>;
+    read(encoding: 'buffer'): Promise<ArrayBuffer>;
+}
+/**
+ * Result of custom image conversion.
+ * Return null to omit the image from output.
+ */
+export interface MammothImageResult {
+    /** The src attribute for the img element */
+    src: string;
+}
+/**
+ * Custom image handler function.
+ * Called for each image in the document during conversion.
+ *
+ * @example
+ * ```ts
+ * const handler: MammothImageHandler = async (image) => {
+ *   const base64 = await image.read('base64')
+ *   return { src: `data:${image.contentType};base64,${base64}` }
+ * }
+ * ```
+ */
+export type MammothImageHandler = (image: MammothImageElement) => Promise<MammothImageResult | null>;
+/**
+ * Options for DOCX to HTML conversion.
+ *
+ * These options map to mammoth.js convertToHtml options.
+ *
+ * @example
+ * ```ts
+ * const options: DocxToHtmlOptions = {
+ *   styleMap: [
+ *     "p[style-name='Title'] => h1.doc-title",
+ *     "p[style-name='Heading 1'] => h1",
+ *   ],
+ *   idPrefix: 'doc-',
+ *   imageHandling: 'inline',
+ * }
+ * ```
+ */
+export interface DocxToHtmlOptions {
+    /**
+     * Custom style mappings from Word styles to HTML elements.
+     *
+     * Each string follows mammoth style map syntax:
+     * - `"p[style-name='Heading 1'] => h1"` - Map paragraph style to h1
+     * - `"r[style-name='Emphasis'] => em"` - Map run style to em
+     * - `"p[style-name='Code'] => pre > code"` - Nested elements
+     *
+     * @see https://github.com/mwilliamson/mammoth.js#style-map
+     */
+    styleMap?: string[];
+    /**
+     * Include the document's embedded style map.
+     *
+     * When true (default), any style map embedded in the .docx file
+     * is combined with the provided styleMap.
+     *
+     * @default true
+     */
+    includeEmbeddedStyleMap?: boolean;
+    /**
+     * Include mammoth's default style mappings.
+     *
+     * When true (default), mammoth's default style map is used as a base.
+     * Set to false to only use explicit styleMap mappings.
+     *
+     * @default true
+     */
+    includeDefaultStyleMap?: boolean;
+    /**
+     * Prefix for generated element IDs.
+     *
+     * Used for bookmarks, footnotes, and other elements that need unique IDs.
+     * Useful when embedding multiple documents in a single page.
+     *
+     * @example 'doc1-' would generate IDs like 'doc1-footnote-1'
+     */
+    idPrefix?: string;
+    /**
+     * How to handle images in the document.
+     *
+     * - `'inline'` (default): Embed images as base64 data URIs
+     * - `'omit'`: Remove all images from output
+     * - Custom handler: Function to process each image
+     *
+     * @default 'inline'
+     */
+    imageHandling?: 'inline' | 'omit' | MammothImageHandler;
+    /**
+     * Preserve empty paragraphs in output.
+     *
+     * When false (default), empty paragraphs are omitted from HTML.
+     * Set to true to include them as `<p></p>` elements.
+     *
+     * @default false
+     */
+    preserveEmptyParagraphs?: boolean;
+    /**
+     * Transform the internal document representation before conversion.
+     *
+     * Advanced option for modifying the document structure.
+     * The function receives the mammoth document object and should return
+     * a transformed version.
+     */
+    transformDocument?: (document: unknown) => unknown;
+}
+/**
+ * Options for extracting raw text from DOCX.
+ */
+export interface DocxToTextOptions {
+    /**
+     * Prefix for generated element IDs.
+     * @see DocxToHtmlOptions.idPrefix
+     */
+    idPrefix?: string;
+}

package/dist/result.d.ts

@@ -0,0 +1,52 @@
+/**
+ * Result types for DOCX conversion.
+ *
+ * @module mammoth/result
+ */
+/**
+ * Message severity level.
+ */
+export type MammothMessageType = 'warning' | 'error';
+/**
+ * Message generated during DOCX conversion.
+ *
+ * Mammoth produces messages for issues like:
+ * - Unrecognized styles
+ * - Unsupported features
+ * - Missing images
+ */
+export interface MammothMessage {
+    /** Severity of the message */
+    type: MammothMessageType;
+    /** Human-readable description of the issue */
+    message: string;
+}
+/**
+ * Result of DOCX to HTML conversion.
+ *
+ * Contains the converted HTML and any messages generated during conversion.
+ *
+ * @example
+ * ```ts
+ * const result = await docxToHtmlImpl.execute(buffer)
+ * console.log(result.html)
+ * if (result.messages.length > 0) {
+ *   console.warn('Conversion warnings:', result.messages)
+ * }
+ * ```
+ */
+export interface DocxConversionResult {
+    /** The converted HTML string */
+    html: string;
+    /** Messages (warnings/errors) generated during conversion */
+    messages: MammothMessage[];
+}
+/**
+ * Result of DOCX to text extraction.
+ */
+export interface DocxTextResult {
+    /** The extracted plain text */
+    text: string;
+    /** Messages (warnings/errors) generated during extraction */
+    messages: MammothMessage[];
+}

package/dist/util.d.ts

@@ -0,0 +1,9 @@
+/**
+ * Utility functions for mammoth package.
+ *
+ * @module mammoth/util
+ */
+/**
+ * Convert a ReadableStream<Uint8Array> to a Buffer.
+ */
+export declare function streamToBuffer(stream: ReadableStream<Uint8Array>): Promise<Buffer>;

package/package.json

@@ -0,0 +1,47 @@
+{
+  "name": "@origints/mammoth",
+  "version": "0.1.0",
+  "description": "DOCX to HTML conversion for Origins using mammoth.js",
+  "type": "module",
+  "license": "MIT",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.es.js",
+      "require": "./dist/index.cjs"
+    }
+  },
+  "main": "./dist/index.cjs",
+  "module": "./dist/index.es.js",
+  "types": "./dist/index.d.ts",
+  "files": [
+    "dist"
+  ],
+  "publishConfig": {
+    "access": "public"
+  },
+  "dependencies": {
+    "mammoth": "^1.11.0"
+  },
+  "peerDependencies": {
+    "@origints/core": "^0.1.0"
+  },
+  "devDependencies": {
+    "@types/node": "25.0.6",
+    "@vitest/coverage-v8": "^4.0.16",
+    "eslint": "9.39.2",
+    "jszip": "^3.10.1",
+    "typescript": "5.9.3",
+    "vite": "7.3.1",
+    "vite-plugin-dts": "4.5.4",
+    "vitest": "4.0.16",
+    "@origints/core": "0.1.0"
+  },
+  "scripts": {
+    "build": "vite build",
+    "test": "vitest run",
+    "test:coverage": "vitest run --coverage",
+    "lint": "eslint \"{src,tests}/**/*.{ts,tsx}\" --max-warnings 0",
+    "typecheck": "tsc -p tsconfig.json --noEmit"
+  }
+}