codex-pdf
Structured PDF extraction API that turns complex files into consistent JSON.
Compile is the only writer in the Print With Synergy stack — it applies trapping, printer's marks, imposition, soft-proofing, and white-underbase as declarative, schema-versioned producers and emits byte-identical, audit-logged PDFs. Same inputs, same bytes, every time.
AGPL-3.0 · trap · marks · impose · cjd · soft-proof · white-underbase · content-addressed lineage
How it works
Compile turns declarative production plans into deterministic, audit-logged PDFs — built on codex's facts, governed by the policy you hand it, and traceable end to end.
Compile consumes codex's read-only primitives — geometry, colour, spot intents, dieline — and never re-extracts them. The writer builds on the facts; it doesn't recompute them.
Hand Compile a declarative plan — a trap policy, a marks template, an impose layout — and the matching producer writes it into the PDF. Trap, marks, impose, soft-proof, white-underbase: one job each.
Same inputs, same bytes. Every apply is content-addressed by a SHA-256 of the input, the producer schema version, the engine fingerprint, and the config — so a result is reproducible and cacheable.
Each producer call writes an HMAC-chained lineage record — input/output hashes, engine fingerprint, timings. The chain proves no record was silently dropped or rewritten.
Compile is reactivated as Platform's production add-on — trap, marks, impose, CJD, soft-proof, and white-underbase, each a schema-versioned, deterministic producer that writes into the PDF and logs its lineage.
Ink-pair spread / choke trapping with selectable engines (pure-Python or Ghostscript) to compensate for press registration and eliminate white seams at ink boundaries.
Printer's marks for production and proofing — registration, crop, bleed, colour bars, fold lines, slug text — plus PDF / SVG / PNG template ingestion for the shop floor.
Sheet step-and-repeat with work-and-turn / work-and-tumble backing, bleed expansion, and multi-cell placement — driven by sift via the sift.solve → compile.impose chain.
Bundle multiple producers into one job envelope in canonical order — rewrite → marks → impose → trap — with lineage tracking and a single audited result.
ICC simulation that renders a PDF under a destination profile and reports ΔE summaries (max / mean / p95), so colour drift is caught before a hardcopy proof.
White-ink, underbase, varnish, or foil plate generation for labels, garments, and dark substrates — derives the named separation with coverage + choke policy.
Why Compile
Compile is the stack's only writer — and it treats production as declarative schemas with content-addressed caching and an immutable lineage chain: the determinism and auditability ad-hoc PDF tooling lacks.
Same job description + input PDF → identical bytes. Each apply is content-addressed by a SHA-256 of the input, producer schema version, engine fingerprint, and config — so there's no “it worked yesterday”. Reproducible, cacheable, diffable.
Every producer call writes an HMAC-chained lineage record — input/output sha256, engine fingerprint, codex section versions, timings. Chain validation proves no record was silently dropped or rewritten.
Compile consumes codex's facts and applies the policy it's handed — a trap policy, a marks template, an impose plan. It doesn't decide whether the result is acceptable; that's lint. Each engine owns one job.
COMPILE_PRODUCER gates per-producer deployments: ship a trap-only image, a marks-only image, etc. Synergy's gateway routes accordingly. Run it all-in-one, or split the writer across sidecars.
Pricing
Run the writer yourself, or enable the production add-on in your Platform workspace — same engine, you pick who runs it.
AGPL-3.0 · your infrastructure
Run the whole writer yourself — every producer, the CLI, and the HTTP API. No quotas, no per-document fees.
Get the source →Pros
Cons
enabled in your Platform workspace
Enable the production add-on in Platform and call the writer through the production module routes — managed, metered, supported.
Add on PlatformPros
Cons
The open-source edition is AGPL-3.0 and free forever. The production add-on is enabled and metered through your Platform workspace.
Open source · managed hosting
A toolkit of focused, standalone PDF utilities — extraction, preflight, viewing, assembly, imposition planning, and an asset store. Each one plugs into the prepress workflow you already run. Use the open source yourself, or let us host any single tool for you on host.withsynergy.io.
Structured PDF extraction API that turns complex files into consistent JSON.
Objective file-comparison engine — measured coverage and geometry differences for plate ↔ 1-up and document ↔ document, never a verdict.
Detection-only PDF preflight engine — 500+ checks plus the PDF/X-4 conformance suite.
Embeddable PDF viewer with separations, TAC, layers, and annotation overlays.
GWG 2022 conformance assay — benchmark a preflight engine against the spec.
Content-addressed digital-asset plane — versioned blobs, a presigned data plane, and on-prem agent recall.
The print-data integration hub — canonical jobs, orders, and customers kept in sync across your MIS, ERP, and prepress tools.
The deterministic PDF writer — trapping, marks, imposition, soft-proof, and white-underbase as audit-logged producers.