Project Files

docs

initial-docs

ANNOTATE-IMAGE-DETECTION.md

MEMORIZE-DOC-IMAGES.md

SETUP.md

USER-DOCS.md

CHANGELOG.md

DEPLOYMENT.md

USER_GUIDE.md

python

docling_parser.py

extract_image_page.py

pymupdf_parser.py

requirements.txt

src

documents

parsers

doclingRunner.ts

docxParser.ts

pdfParser.ts

textParser.ts

fileWatcher.ts

loader.ts

fpzip

decompress.ts

fpzip_bridge.cpp

fpzip_loader.ts

fpzip_wasm.js

fpzip_wasm.wasm

package.json

helpers

documentImages.ts

drawBboxesOnImage.ts

embedLocalImages.ts

frontmatter.ts

globalConfigReader.ts

pngMetadata.ts

readPngMetadata.ts

sequenceExtractor.ts

toolProgress.ts

videoAssembler.ts

visionCapabilityPrimer.ts

visionPromotionLog.ts

rag

bm25.ts

chunker.ts

embeddings.ts

retriever.ts

retrieverSingleton.ts

vectorStore.ts

services

chatExporter.ts

lmStudioVisionAnalyzer.ts

toolResultHarvester.ts

userDocsGuidePrimer.ts

sources

adapters

githubMarkdownSourceAdapter.ts

huggingFaceMarkdownSourceAdapter.ts

lmStudioConversationSourceAdapter.ts

staticHtmlSourceAdapter.ts

http.ts

lmStudioConversationMarkdown.ts

normalizer.ts

registry.ts

remoteImageResolver.ts

types.ts

tools

analyse_image.ts

annotate_image.ts

detect_object.ts

export_doc.ts

extract_image.ts

fetch_image.ts

find_doc.ts

forget_doc.ts

memorize_doc.ts

read_config.ts

read_doc.ts

rewrite_doc.ts

show_image.ts

skip_doc.ts

types

external-shims.d.ts

utils

language.ts

pythonRunner.ts

ragLogger.ts

ragVenvSetup.ts

config.ts

core-bundle.mjs

index.ts

orchestrator.ts

paths.ts

promptPreprocessor.ts

thinkingToolCallParser.ts

toolsProvider.ts

types.ts

.gitignore

.lmsignore

.npmignore

.swcrc

build.mjs

LICENSE

manifest.json

mcp-shims.d.ts

package-lock.json

package.json

README.md

rollup.config.mjs

tsconfig.json

src / documents / parsers / docxParser.ts

/**
 * DOCX Parser with Docling Integration
 * Uses IBM's docling for professional document parsing (layout, tables, embedded images).
 * No fallback chain: unlike PDF, there is no PyMuPDF/pdf-parse equivalent for DOCX,
 * so Docling is the only strategy.
 */

import type { ParsedDocument, ParserOptions } from '../../types';
import { parseWithDoclingScript } from './doclingRunner.js';

export class DocxParser {
  static async parse(filePath: string, options?: ParserOptions): Promise<ParsedDocument> {
    return parseWithDoclingScript(filePath, 'docx', 'DOCX Parser', options);
  }
}

user-docs

user-docs

user-docs

user-docs