ODT Parsing API

Native ODT, Not "DOCX-ish"

An .odt file is a zip archive containing XML — specifically, the OpenDocument Format. Paragraphs live in content.xml as text:p elements. Tables use the table: namespace. Headers and footers sit inside style:master-page in styles.xml. None of this maps cleanly to Word's w: schema.

Most parsers handle ODT by shelling out to LibreOffice to convert to DOCX or PDF first — losing whatever ODF-specific structure won't survive the trip. AILANG Parse reads ODF XML directly, with the same Block ADT output you get from DOCX. No subprocess, no rendering, no conversion.

Raw ODF XML vs Structured Output

<office:body>
 <office:text>
  <text:h text:outline-level="1">
    Q1 Plan
  </text:h>
  <text:p text:style-name="Normal">
    Revenue target: 2.4M EUR
  </text:p>
  <table:table table:name="Targets">
   <table:table-row>
    <table:table-cell
      table:number-columns-spanned="2">
     <text:p>Region</text:p>
    </table:table-cell>
    <table:table-cell>
     <text:p>Q1</text:p>
    </table:table-cell>
   </table:table-row>
  </table:table>
  <text:list>
    <text:list-item>
      <text:p>EMEA</text:p>
    </text:list-item>
  </text:list>
 </office:text>
</office:body>

{
  "metadata": {
    "title": "Q1 Plan",
    "author": "Alice Chen",
    "created": "2026-03-15T09:00:00Z"
  },
  "blocks": [
    {
      "type": "heading",
      "level": 1,
      "text": "Q1 Plan"
    },
    {
      "type": "text",
      "text": "Revenue target: 2.4M EUR",
      "style": "normal"
    },
    {
      "type": "table",
      "headers": [
        {"text": "Region", "colSpan": 2},
        {"text": "Q1"}
      ],
      "rows": [...]
    },
    {
      "type": "list",
      "ordered": false,
      "items": ["EMEA"]
    }
  ]
}

What Gets Extracted

Headers & Footers

Unlike DOCX, ODT puts headers and footers inside the style definitions, not the body. They live in style:master-page elements inside styles.xml. AILANG Parse reads styles.xml as a second pass, walks every master-page, and emits any header/footer content as a SectionBlock with kind: "header" or "footer".

{
  "type": "section",
  "kind": "header",
  "blocks": [
    {"type": "text", "text": "Confidential — Internal Only", "style": "normal"}
  ]
}

Tables & Merged Cells

ODF tables encode horizontal merges with table:number-columns-spanned on the cell that owns the merge. AILANG Parse resolves these into the same colSpan/rowSpan shape used by the DOCX parser, so downstream code can treat ODT and DOCX tables identically.

{
  "type": "table",
  "headers": [
    {"text": "Region", "colSpan": 2, "rowSpan": 1, "merged": false},
    {"text": "Total", "colSpan": 1, "rowSpan": 1, "merged": false}
  ],
  "rows": [
    [{"text": "EMEA"}, {"text": "UK"}, {"text": "€125K"}],
    [{"text": "EMEA"}, {"text": "DE"}, {"text": "€98K"}]
  ]
}

Use Cases

Try It

Workbench (in-browser)

Drop an ODT file into the Workbench — ODT parsing runs in WebAssembly with the same AILANG parser used by the CLI. No upload, no signup.

CLI

# Parse an ODT file
./bin/docparse report.odt

# Convert ODT to Markdown
./bin/docparse report.odt --convert output.md

# Convert ODT to DOCX
./bin/docparse report.odt --convert output.docx

API

curl -X POST https://docparse.ailang.sunholo.com/api/v1/parse \
  -H "Content-Type: application/json" \
  -d '{"filepath":"sample.odt","outputFormat":"json","apiKey":"YOUR_API_KEY"}'

Python SDK

from ailang_parse import DocParse

client = DocParse(api_key="YOUR_API_KEY")
result = client.parse_file("report.odt")

print(result.metadata.title)
for block in result.blocks:
    if block.type == "heading":
        print("H" + str(block.level), block.text)
    elif block.type == "table":
        print("Table:", len(block.rows), "rows")

Parse in Browser    API Reference