Optimize PDF API
Lossless structural PDF optimisation using the secure task lifecycle, with optional metadata sanitisation.
Endpoint and task semantics#
Operation code: optimize
Start this operation with POST /v1/tasks and the tool code shown above. Upload and attach source PDFs (or use task-scoped upload endpoints), then call POST /v1/tasks/{task_id}/process with operation settings.
Reduce redundant PDF object structure before archiving and delivery without replacing selectable text.
Inputs: Exactly one sealed PDF, an operation ceiling of 100 MiB and 500 pages; configured upload/job and plan limits may be lower.
Configuration and extra parameters#
| Setting | Type | Behaviour |
|---|---|---|
| profile | balanced | lossless | Balanced cleans content streams; lossless retains original stream structure. Default: balanced. |
| strip_metadata | boolean | Also remove metadata, JavaScript, hidden text and attachments. Default: false. |
Options are validated by the API. Unknown settings and unsupported values return an error instead of being ignored.
Working request sequence#
Programming language: cURL
Choose a language · scroll horizontally for all 10 SDK languages
# Server-side API calls. Save the returned task_id.
curl -fsS -X POST 'https://api.pdfhq.io/v1/tasks' \
-H 'x-api-key: YOUR_API_KEY' -H 'Content-Type: application/json' \
-d '{"tool":"optimize"}'
# Upload and verify 1 PDF(s), then obtain and approve a credit quote:
curl -fsS 'https://api.pdfhq.io/v1/tasks/TASK_ID/quote' -H 'x-api-key: YOUR_API_KEY'
# ONLY after explicit customer approval:
test "$PDFHQ_APPROVE_QUOTE" = yes || { echo "Approval required" >&2; exit 1; }
curl -fsS -X POST 'https://api.pdfhq.io/v1/tasks/TASK_ID/process' \
-H 'x-api-key: YOUR_API_KEY' -H 'Content-Type: application/json' \
-d '{"options":{"profile":"balanced","strip_metadata":false}}'
curl -fsS 'https://api.pdfhq.io/v1/tasks/TASK_ID' -H 'x-api-key: YOUR_API_KEY'After starting a task, upload and verify PDFs, review the credit quote, and get explicit approval before processing. API keys belong on your server, never on signed storage uploads.
Generated downloadable packages cover the task workflow only. Full-service SDKs require a verified runtime OpenAPI specification; the Python, JavaScript and TypeScript task SDKs stream file uploads; the other languages require signed multipart uploads implemented by the caller.
The returned task ID is used for both status polling and retrieval. After the task succeeds, request a signed output URL rather than embedding a download secret.
Output and important details#
New PDF with compacted object references and deflated streams; page count preserved.
This is structural optimisation, not lossy image compression. Output may be larger. Forms or signatures must be checked after rewriting; cryptographic signatures will not be preserved.
Errors, retries and billing