Generate PDF Job

Queue a PDF generation job and retrieve its status and result.

Each job request contains the structured data, HTML (or Markdown), and CSS needed to render one document. The optional filename fields are descriptive metadata only; the content fields carry the actual inline HTML and CSS.

To send image files with the template, use the multipart /pdf/generate/bundle endpoint described in Bundled images and scaling. It uses this same request object plus named binary image uploads, with no base64 URLs required.

Authentication

Influxion uses OAuth2 Bearer tokens issued by its /token endpoint. First obtain a token:

POST /token HTTP/1.1
Content-Type: application/json
{
    "client_id": "<CLIENT_ID>",
    "client_secret": "<CLIENT_SECRET>",
    "scope": "api://d88ae758-e67d-492d-bcc6-63f560ee1f04/.default"
}

Note

The scope is optional. A default scope will be assigned if not provided, and it will allow access to all API endpoints.

Response:

{
  "access_token": "<YOUR_JWT>",
  "token_type": "bearer"
}

All subsequent API calls must include the Authorization header with value Bearer <YOUR_JWT>.

Create a job

Request:

POST /pdf/generate HTTP/1.1
Authorization: Bearer <YOUR_JWT>
Content-Type: application/json
{
  "data": { /* structured data for the Jinja2 template */ },
  "document_id": "my-doc-123",
  "pdf_standard": "pdf/a-1b",
  "sign_key_id": "sign-key-id",
  "html_template": {
    "filename": null,
    "content": "<html>...{{ data }}...</html>"
  },
  "css_template": {
    "filename": null,
    "content": "body { font-family: 'Futura PT', 'DejaVu Sans', sans-serif; }"
  }
}

Response:

{
  "job_id": 1,
  "document_id": "my-doc-123"
}

Check job status

GET /pdf/1 HTTP/1.1
Authorization: Bearer <YOUR_JWT>

Response:

{
  "id": 1,
  "status": "Received",
  "created_at": "2023-01-01T12:00:00Z",
  "started_at": null,
  "completed_at": null,
  "pdf_expires_at": null,
  "size": null,
  "error": null
}

The status can be Received, InProgress, Done, or Error.

Download PDF

GET /pdf/1/download HTTP/1.1
Authorization: Bearer <YOUR_JWT>

Response: Binary PDF (Content-Type: application/pdf).

Input types: HTML and Markdown

/pdf/generate accepts two kinds of input, selected with the optional input_type field:

input_type

Document field

CSS

html (default)

html_template (required, Jinja2)

css_template (required)

markdown

markdown_template (required)

css_template (optional, added to the bundled stylesheet)

Requests without input_type behave exactly as before. Sending the field for the other input type (for example html_template together with "input_type": "markdown") is rejected with 422. The job, status, download, signing and PDF standard handling is the same for both.

Markdown to PDF

Minimal request:

{
  "input_type": "markdown",
  "markdown_template": { "filename": "notes.md", "content": "# Title\n\nSome *text*." }
}

All Markdown options:

{
  "input_type": "markdown",
  "markdown_template": { "filename": "notes.md", "content": "# Title\n\nSome *text*." },
  "css_template": { "filename": "style.css", "content": "h1 { color: #0b5cad; }" },
  "include_default_css": true,
  "title": null,
  "language": "en",
  "data": {},
  "document_id": "my-doc-123",
  "pdf_standard": "pdf/ua-1",
  "sign_key_id": null
}
  • Syntax: CommonMark-style Markdown plus tables, fenced code blocks with syntax highlighting, footnotes, definition lists, attribute lists ({: .class }) and [TOC]. Raw HTML is passed through.

  • Title and language: title sets the document title and defaults to the first level-one heading. language sets the lang attribute, which PDF/UA requires.

  • Templating: if data is non-empty, the Markdown is first rendered as a Jinja2 template with data available (for example {{ data.customer }}). Without data the Markdown is converted untouched, so code samples containing {{ are safe.

Styling Markdown documents

Markdown documents are styled by a bundled stylesheet: A4 pages with page numbers in the footer, Metropolis (with DejaVu Sans as fallback) at 10.5 pt, bordered tables, shaded code blocks and Pygments syntax highlighting. See Fonts for the available font families.

css_template is appended after the bundled stylesheet, so you only need to write what differs:

@page { margin: 3cm; @bottom-center { content: none; } }
html { font-family: 'Sentinel', 'DejaVu Serif', serif; }
h1 { color: #0b5cad; border-bottom: none; }

Set include_default_css to false to use only your own CSS. Then css_template is required.

The Markdown is wrapped in <main class="markdown-body"> and code blocks in <div class="codehilite">, which can be used as CSS selectors. Attribute lists add classes to individual elements: a line {: .note } directly after a paragraph gives it class="note".

A ready-made client is available in Examples.

Short-lived PDF storage

Completed PDFs are written to the runtime directory configured by INFLUXION_PDF_RUNTIME_DIR. The default is /run/influxion/pdfs. The file name is generated by Influxion and includes a random token so it cannot be guessed from the job ID alone.

The database stores job metadata such as status, document_id, generated file size, the runtime PDF filename, and pdf_expires_at. The PDF bytes are not stored in the database, and the submitted HTML/CSS are used only for that generation request.

Completed PDFs remain downloadable for INFLUXION_PDF_RETENTION_SECONDS seconds, with a default of 300 seconds. After that timestamp, or if the runtime file is no longer present, the download endpoint returns 410 Gone. The application also clears leftover runtime PDF files during startup, so completed downloads should be treated as temporary handoff files.

Using Celery background workers (Optional)

When setting up Influxion, one can connect it to a RabbitMQ and use Celery for background PDF generation:

pip install -e .
export INFLUXION_USE_CELERY=1
export INFLUXION_CELERY_BROKER_URL=amqp://guest@localhost//
export INFLUXION_CELERY_RESULT_BACKEND=rpc://
celery -A influxion.celery_app worker --loglevel=info

This enables asynchronous PDF generation, allowing the API to respond immediately and several jobs to be processed concurrently.