docx

Eino - DOCX Parser

The Docx parser is Eino's document parsing component that implements the 'Parser' interface for parsing Microsoft Word (docx) files. This package is designed to parse Microsoft Word (.docx) files and extract their content into a structured, plain-text format.

📜 Overview

The docx package offers a configurable parser to read .docx documents from an io.Reader. It can selectively extract various parts of the document, including the main body text, headers, footers and tables. The output can be returned as a single consolidated document or split into separate sections.

This parser is built upon the docx2md library to handle the underlying XML structure of .docx files.

✨ Features

• Main Content Extraction: Parses all paragraphs and text runs from the main document body.
• Configurable Extraction: Easily enable or disable the inclusion of:
  • Headers
  • Footers
  • Tables
• Flexible Output:
  • Combine all extracted content into a single document.
  • Split content into separate sections (e.g., main, headers, footers).
• Lightweight wrapper around the docx2md library.

⚙️ Configuration

The behavior of the DocxParser is controlled by a Config struct. If no configuration is provided, a default one is used.

Here are the available configuration options:

🚀 Usage Example

Below is a basic example of how to use the DocxParser to read a .docx file and print its contents.

First, ensure you have the necessary packages:

go get github.com/cloudwego/eino
go get github.com/eino-contrib/docx2md

Example Code

package main

import (
	"context"
	"fmt"
	"log"
	"os"

	"github.com/cloudwego/eino-ext/components/document/parser/docx"
)

func main() {
    // 1. Open the DOCX file.
    file, err := os.Open("your_document.docx")
    if err != nil {
        log.Fatalf("Failed to open file: %v", err)
    }
    defer file.Close()

    ctx := context.Background()

    // 2. Configure the parser to include everything.
    config := &docx.Config{
        ToSections:      true, // Split content into sections
        IncludeHeaders:  true,
        IncludeFooters:  true,
        IncludeTables:   true,
    }

    // 3. Create a new parser instance.
    docxParser, err := docx.NewDocxParser(ctx, config)
    if err != nil {
        log.Fatalf("Failed to create parser: %v", err)
    }

    // 4. Parse the document.
    docs, err := docxParser.Parse(ctx, file)
    if err != nil {
        log.Fatalf("Failed to parse document: %v", err)
    }

    // 5. Print the extracted content.
    fmt.Printf("Successfully parsed %d section(s).\n\n", len(docs))
    for _, doc := range docs {
        fmt.Printf("--- Section ID: %s ---\n", doc.ID)
        fmt.Println(doc.Content)
        fmt.Println("--- End of Section ---")
        fmt.Println()
    }
}

Configuration Options

config := &docx.Config{
    ToSections:      false, // Whether to split content by sections
    IncludeHeaders:  true,  // Include headers in output
    IncludeFooters:  true,  // Include footers in output
    IncludeTables:   true,  // Include table content
}

parser, err := docx.NewDocxParser(context.Background(), config)

Output Format

When ToSections is false (default), the parser returns a single document with all content concatenated.

When ToSections is true, the parser returns multiple documents split by section type:

• "main" - Main document content
• "headers" - Header content (if enabled)
• "footers" - Footer content (if enabled)
• "tables" - Table content (if enabled)

Each section is preceded by a header line (e.g., "=== MAIN CONTENT ===") to identify the section type.

Limitations

• Currently only extracts plain text content
• Formatting, images, and other rich content are not preserved
• Complex table structures may not be perfectly represented
• Note on Comments: The IncludeComments option is now deprecated. The underlying DOCX parsing library was switched from an AGPL-licensed dependency to docx2md (MIT-licensed) to resolve a license conflict with the project's Apache 2.0 license. The new library does not currently support comment extraction.

Examples

See the following examples for more usage:

• Test Data Example