PDF Creator App

A service and library for turning infopack IRNs into PDFs or raw HTML. You can run the bundled Express server or import the library directly.

HTTP usage

1. Provide an IRN as GET /single?irn=<IRN>.
2. Infopack files must be .partial.html.

Library usage

Import and use the expander and PDF helpers without starting the server:

import { InfopackCollectionExpander, createPdfFromCollection } from 'pdf-creator-app'

const expander = new InfopackCollectionExpander(collectionDefinition)
await expander.expand()
const expanded = expander.get()

// Render HTML (set raw:false to get a PDF buffer)
const html = await createPdfFromCollection(collectionDefinition, {
  raw: true,
  templateOverrides: {
    paths: { collection: './my-templates/collection.template.html' }
  }
})

Templates can be overridden by changing file paths (templateOverrides.paths) or by passing template strings/functions directly on templateOverrides.

Templates and their roles

• standard: Wrapper for a single infopack file (createPdfFromIrn). Should include room for content and QR code ({{qrCode}}).
• collection: Outer HTML for a collection; receives title, name, version, toc, content.
• toc: Builds the table of contents for a collection (gets files from the expander).
• file: Renders each file in a collection (receives the full content object from the expander).
• header / footer: Raw HTML strings printed on every PDF page. These go straight to Puppeteer’s headerTemplate/footerTemplate, so follow Puppeteer’s rules: use <span class="pageNumber"></span> and <span class="totalPages"></span> for page numbers, and keep the total height within Puppeteer’s header/footer limits. They are not compiled with Handlebars unless you do it yourself before passing in the string.

Data available inside each template

• standard (single IRN): gets everything from infopackUrlResolver plus qrCode. Typical fields: content (HTML), meta (from .meta.json), valid, notValid, messages, url, fetchedDate, classes (if set), and qrCode (data URI for registry link).
• collection: receives { title, name, version, toc, content } where toc and content are rendered strings produced by the templates below.
• toc: receives { files } where files is the array from the collection expander. Each file item can include: kind ("infopack-file" or "html-file"), collectionPath, packagePath, filePath, irn (if an infopack file), url, fetchedDate, content (HTML if returnContent is true), meta, valid, notValid, messages, and optional classes.
• file: receives the same file item objects as toc, so you can access any of the fields above per file.
• header / footer: no data injection; treated as literal HTML passed to Puppeteer. If you need Handlebars data here, compile it yourself before passing the string.

Images and assets (e.g. logos)

When you add images in templates (including header/footer), use either:

• Absolute URLs reachable from the rendering environment (e.g. https://.../logo.png), or
• Inline data URIs (<img src="data:image/png;base64,..." />) if you need to bundle the asset.

Relative paths will not resolve from page.setContent; prefer absolute URLs or data URIs for reliable rendering.

To inline local relative images in templates (e.g. logo in header/footer), set assetsBasePath so the library can convert <img src="./..."> into data URIs in templates. Content HTML is not inlined automatically:

await createPdfFromCollection(collectionDefinition, {
  assetsBasePath: './assets', // only files under this dir will be read
  templateOverrides: { /* ... */ }
})

Supported formats for inlining: png, jpg, jpeg, gif, svg, webp. Files outside assetsBasePath are ignored; missing files are left unchanged.

Caching

• Enabled by default: fetched infopack content and metadata are cached under .cache/infopack (git-ignored).
• TTL: 24h for URLs/IRNs containing latest; 90 days for all other versions.
• Disable or customize via cache options:

// Disable cache
await createPdfFromIrn('ns:name:version:path', { cache: { enabled: false } })

// Custom cache dir/TTL via expander in collection rendering
await createPdfFromCollection(collectionDefinition, {
  expanderOptions: {
    cache: { dir: './tmp/cache', ttlMs: 7 * 24 * 60 * 60 * 1000 } // 7 days for non-latest
  }
})

// Render cache (final HTML) can also be cached on disk (default .cache/render)
await createPdfFromCollection(collectionDefinition, {
  renderCache: { enabled: true, dir: './.cache/render' }
})

Cache keys

This library uses a cache key concept to map each fetched resource to a single file on disk:

• For infopack fetches, the cache key is the full URL. The cache file name is sha1(url).json.
• For render caching, the cache key is a string built from the collection content and template overrides. That string is then hashed the same way.
• The hash keeps filenames short and filesystem-safe; no plain URLs are used as filenames.
• Each cache file stores a small JSON object with metadata (version, fetchedAt, ttlMs, type) plus the body payload.

Default paths live in lib/templates.js (defaultTemplatePaths). You can repoint files or supply inline templates:

import { createPdfFromCollection, createPdfFromIrn } from 'pdf-creator-app'

// Point to custom files
await createPdfFromCollection(collectionDefinition, {
  templateOverrides: {
    paths: {
      collection: './my-templates/collection.template.html',
      file: './my-templates/file.template.html'
    }
  }
})

// Inline string/function (compiled with Handlebars); header/footer are raw strings
await createPdfFromCollection(collectionDefinition, {
  raw: true,
  templateOverrides: {
    collection: '<html><body>{{{toc}}}{{{content}}}</body></html>',
    header: '<span>My header</span>',
    footer: '<span>My footer</span>'
  }
})

// Single IRN with custom template
await createPdfFromIrn('namespace:name:version:path', {
  templateOverrides: {
    standard: '<html><body>{{{content}}} {{qrCode}}</body></html>'
  }
})

Build

docker build . -t registry.gitlab.com/infopack/pdf-creator-app:latest
docker push registry.gitlab.com/infopack/pdf-creator-app:latest

Installation

Docker

The app is packaged as a Docker image; releases are available here.

docker run --rm --name pdf-creator-app -p 9000:9000 registry.gitlab.com/infopack/pdf-creator-app:<latest version>

Release

GitLab builds and publishes the image on every tag on main. Latest images are in the container registry linked above.

Helm

To update the Helm chart, ensure helm is installed and configured, then:

cd helm/pdf-creator-app
helm --namespace infopack-pdf-creator-app upgrade pdf-creator-app .