Uniword converts OOXML documents to MHTML format for Word 2003+ compatibility. The conversion pipeline transforms the OOXML object model into a MIME multipart HTML document.

1. Conversion Steps

  1. Extract content — The OOXML document model is traversed to extract paragraphs, tables, images, and formatting

  2. Generate HTML — Document elements are converted to HTML with Word-compatible markup

  3. Generate CSS — OOXML styles and themes are translated to CSS rules

  4. Encode images — Binary image data from the DOCX package is encoded as Base64

  5. Build MIME structure — HTML, CSS, and embedded resources are wrapped in a MIME multipart/related structure

2. MIME Structure

The MHTML output follows this structure:

MIME-Version: 1.0
Content-Type: multipart/related; boundary="----=_boundary"

------=_boundary
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: quoted-printable

<html>
  <head>
    <style><!-- CSS rules --></style>
  </head>
  <body>
    <!-- HTML content -->
    <img src="cid:image001">
  </body>
</html>

------=_boundary
Content-Type: image/png
Content-Transfer-Encoding: base64
Content-ID: <image001>

iVBORw0KGgo...base64 data...
------=_boundary--

3. CSS Generation

OOXML formatting is translated to CSS:

Paragraph formatting

spacing, indentation, alignment become CSS margin, padding, and text-align properties.

Character formatting

bold, italic, font-size, color become CSS font and color properties.

Table formatting

borders, shading, cell-padding become CSS table properties.

Theme colors

Resolved to RGB hex values before CSS generation.

4. Binary Data Handling

Images and other binary resources are handled as follows:

  • Each image in the DOCX package is extracted from its ZIP part

  • The image binary data is Base64-encoded

  • A MIME part is created with the appropriate Content-Type and Content-ID

  • The HTML references the image via cid: URI

5. Usage 

# Convert DOCX to MHTML
doc = Uniword.load('document.docx')
doc.save('document.mhtml')

# Or explicitly specify the format
doc.save('document.mhtml', format: :mhtml)