report_pdf_renderer (rust)

Standalone PDF renderer for persisted report templates.

Goals

• Fast standalone binary (Rust)
• Render report sections with explicit spacing/layout rules
• Embed frame images (single image or grid)
• Render template sentence sections with variable substitution
• Callable from Python (subprocess) without importing heavy PDF libs

Quick Start (with devenv)

This repo includes a devenv.nix with Rust tooling.

cd lx-report-generator
devenv shell

Then build and run:

cargo build --release
./target/release/report_pdf_renderer \
  --input examples/report_payload.json \
  --output /tmp/report_example.pdf

Open the generated file:

xdg-open /tmp/report_example.pdf

Install From PyPI

Once published, install the renderer with:

python -m pip install report-pdf-renderer

The wheel installs the Rust binary as report_pdf_renderer on your Python environment's PATH:

report_pdf_renderer \
  --input examples/report_payload.json \
  --output /tmp/report_example.pdf

Build (separate install, no devenv)

cargo build --release
# binary: target/release/report_pdf_renderer

CLI

report_pdf_renderer --input payload.json --output report.pdf

Run From Anywhere

Either:

./target/release/report_pdf_renderer --input payload.json --output report.pdf

or install/copy the binary to your PATH:

install -m755 target/release/report_pdf_renderer ~/.local/bin/report_pdf_renderer
report_pdf_renderer --input payload.json --output report.pdf

Backend integration can also point to a custom path via:

export ENDOREG_REPORT_PDF_RENDERER_BIN=/absolute/path/to/report_pdf_renderer

When installed from PyPI in the same runtime environment as the backend, Python callers can either rely on report_pdf_renderer being available on PATH or set the integration variable explicitly:

export ENDOREG_REPORT_PDF_RENDERER_BIN="$(command -v report_pdf_renderer)"

Input JSON (snake_case)

Top-level keys:

• title: string
• subtitle: optional string
• header: optional object
• layout: optional object (page, margins_mm, spacing values)
• assets_root: optional path for relative image lookups
• blocks: ordered content blocks

Supported block types

• heading
• paragraph
• spacer
• sentence_group (template sentences + variables)
• image
• image_grid

sentence_group

{
  "type": "sentence_group",
  "section_title": "Findings",
  "variables": {"segment": "sigma", "size_mm": "8"},
  "sentences": [
    {"template": "Polyp in {segment}.", "enabled": true},
    {"template": "Estimated size {size_mm} mm.", "enabled": true}
  ]
}

image_grid

{
  "type": "image_grid",
  "title": "Frames",
  "columns": 3,
  "image_paths": ["frames/f_001.png", "frames/f_002.png"],
  "captions": ["frame 1", "frame 2"]
}

Notes

• This renderer uses a lightweight layout engine (text flow + page breaks).
• For now it uses built-in PDF fonts (Helvetica).
• If an image fails to load, the renderer logs and continues.
• Integrate from Python by writing a JSON payload to a temp file and invoking the binary.
• assets_root is used to resolve relative image paths (e.g. frame images in image_grid).