Document Conversions
Convert documents to accessible formats with real-time pipeline tracking.
Overview
Document conversion transforms your uploaded documents into accessible formats. The conversion pipeline parses your document, analyzes its structure, applies automatic remediation, validates the results, and generates output in your chosen format.
Starting a Conversion
From the Document Detail Page
- Open a document
- Click Convert Document
- The pipeline visualization shows real-time progress
Via the API
curl -X POST https://api.adaline.ink/api/v1/conversions/ \
-H "Authorization: Bearer YOUR_TOKEN" \
-H "Content-Type: application/json" \
-d '{"document_id": "YOUR_DOC_ID"}'
Returns a job_id that you can poll for status.
Pipeline Steps
The conversion pipeline has 6 stages:
- Upload -- document received and queued
- Parse -- structure extraction (headings, images, tables, lists, math)
- Analyze -- accessibility evaluation against WCAG 2.1 AA
- Remediate -- automatic fixes for common issues
- Validate -- verify all remediations meet standards
- Complete -- accessible document ready
Real-Time Tracking
The pipeline stepper shows each step's status:
- Gray -- pending
- Cyan pulse -- currently processing
- Green check -- completed
- Red X -- failed
Hover over any step for a description and duration.
Auto-Remediation
During the Remediate step, the engine automatically fixes:
- Missing document language -- sets to English if unspecified
- Missing title -- infers from the first heading
- Heading hierarchy -- fixes skipped levels (H1 to H3 becomes H1 to H2)
- Missing alt text -- injects AI-generated descriptions (if AI is enabled)
- Figure alt text -- copies caption to alt text when missing
- Missing table headers -- marks first row as header
- Missing form labels -- infers from field names
- Untyped lists -- defaults to unordered
All changes are tracked in a change log so you can see exactly what was modified.
Output Formats
- Accessible HTML -- semantic markup, screen reader optimized. See PDF to HTML
- Tagged PDF -- PDF with proper structure tags
- Accessible DOCX -- Word document with styles and formatting preserved
- LaTeX -- accessible LaTeX with hyperref and accessibility packages
Polling Job Status
curl https://api.adaline.ink/api/v1/conversions/JOB_ID \
-H "Authorization: Bearer YOUR_TOKEN"
Response includes status and steps array with per-step timing:
{
"id": "job-id",
"status": "completed",
"steps": [
{ "step_name": "upload", "status": "completed", "duration_ms": 120 },
{ "step_name": "parse", "status": "completed", "duration_ms": 2340 },
{ "step_name": "analyze", "status": "completed", "duration_ms": 1500 }
]
}
Limitations
- Conversions process one document at a time
- Very large documents (500+ pages) may take several minutes
- AI-generated alt text requires AI processing to be enabled in your organization settings
- Visual layout fidelity varies by output format (HTML simplifies multi-column layouts)