Is my PDF stored anywhere?
No. PDFs are processed in memory and discarded when the request ends. There is
no database and no temp file held between requests, and results are sent
Cache-Control: no-store.
Does it phone home / send telemetry? No. The default image emits nothing externally. OpenTelemetry export is an opt-in build feature.
How big is the image? Roughly 8–16 MB — a single static Rust binary on a distroless base.
Can it fill Japanese / CJK form fields? Yes. Field values are passed to the engine verbatim as UTF-8 — no transcoding or normalisation — so CJK text is handled correctly.
Does it do OCR / rendering? Not in the default slim image (those pull heavy native dependencies). v0.1 focuses on extraction, forms, and document ops.
How do I process password-protected PDFs?
v0.1 returns PDF_ENCRYPTED for encrypted documents; password support is on
the roadmap.
How does it scale?
It's stateless — run multiple replicas behind any load balancer. Concurrency is
bounded per instance by PDF_OXIDE_API_MAX_INFLIGHT.
What's the difference between this and the pdf_oxide library?
pdf_oxide is the Rust engine (also available for Python, Go, JS, C#, Java).
pdf_oxide_api is a self-hostable HTTP service that exposes it over REST for
tools that integrate via HTTP.
Which versions am I running?
GET /version reports both the API version and the embedded pdf_oxide
engine version.