Uniword supports MHTML (MIME HTML) format for Word 2003+ compatibility. MHTML files encode an entire Word document — including text, formatting, and embedded resources — into a single MIME multipart file.

1. MIME Structure

An MHTML file is a MIME multipart document with the following structure:

  • MIME headers — Content-Type and boundary definitions

  • HTML part — The document content rendered as HTML with Word-specific CSS

  • Embedded resources — Images and other binary data encoded as Base64 within MIME parts

The format uses multipart/related MIME type with each part referenced by Content-ID (CID) URIs.

2. CSS Generation

Uniword generates Word-compatible CSS from document styles:

  • Paragraph styles are translated to CSS block-level properties

  • Character styles become inline CSS or <span> styles

  • Table styles map to CSS table formatting

  • Theme colors and fonts are resolved to CSS values

3. DOCX to MHTML Conversion 

# Load a DOCX file and convert to MHTML
doc = Uniword.load('document.docx')
doc.save_as('document.mhtml')

# Or use the format handler directly
doc.save('document.mhtml', format: :mhtml)

The conversion process:

  1. Parses the DOCX package into the Document Layer

  2. Generates HTML from document elements

  3. Produces CSS from styles and themes

  4. Encodes embedded images as Base64 MIME parts

  5. Wraps everything in a MIME multipart structure

4. MHTML to DOCX Conversion 

# Load an MHTML file and convert to DOCX
doc = Uniword.load('document.mhtml')
doc.save('document.docx')

The reverse conversion:

  1. Parses the MIME multipart structure

  2. Extracts HTML content and CSS

  3. Converts HTML elements back to OOXML document parts

  4. Decodes Base64 resources into image parts

  5. Packages as a valid DOCX ZIP file

5. Format Auto-Detection

Uniword automatically detects the file format based on file extension and content:

  • .docx files are processed by the DOCX handler

  • .mhtml or .mht files are processed by the MHTML handler

  • Content inspection confirms the detected format

# Format is detected automatically
doc = Uniword.load('document.mhtml')  # Uses MHTML handler
doc.save('output.docx')               # Writes as DOCX