SAP Digital Manufacturing Cloud — API Mirror

Mirrors all REST API specs from the SAP Digital Manufacturing Cloud package on SAP Business Accelerator Hub.

Downloads complete Swagger 2.0 / OpenAPI specs for every API, including all endpoints, request/response schemas, and metadata.

Browse the APIs online: https://endogen.github.io/sap-dmc-api/ — search, Swagger UI, changelog, and collection downloads, rebuilt automatically after every mirror update.

What's Included

output/
  README.md          # Full API reference (table of all APIs + endpoints)
  summary.json       # Machine-readable index of all APIs & endpoints
  artifacts.json     # Raw artifact catalog from SAP
  changelog.json     # Combined changelog of all detected spec changes
  history/           # One diff report per mirror run that found changes
  specs/             # 88 Swagger JSON files (one per API)
  metadata/          # 88 API metadata files from OData catalog
  collections/       # Postman & Insomnia collection files

Current stats: 90 APIs · 505 endpoints · 1,329 schemas

Setup

Requires Python 3.10+ and a Chromium browser (installed by Playwright).

# Install dependencies
pip install -r requirements.txt

# Install browser for Playwright
playwright install chromium

# Configure credentials
cp .env.example .env
# Edit .env with your SAP credentials

Credentials

When catalog changes require protected specifications to be downloaded, you need an SAP account with access to the SAP Digital Manufacturing Cloud package. Set these in .env:

SAP_USER=your.email@company.com
SAP_PASS=your-password
SAP_ACCOUNT=S00xxxxxxxx

The account ID (SAP_ACCOUNT) is the S-number shown on the SAP ID account selection screen after login.

Mirroring

# Incremental mirror (checks the public catalog first)
python3 mirror.py

# Custom output directory
python3 mirror.py --output-dir my-output

# Abort without saving if the resulting mirror has fewer than N specs
# (guards against half-failed sessions; used by CI)
python3 mirror.py --min-specs 80

# Check whether updates are needed without credentials or Playwright
# Exit status: 0 = current, 3 = update needed
python3 mirror.py --check

# Force all specifications to be downloaded again
python3 mirror.py --force

# Regenerate summary/README/collections from existing specs
# (no login or Playwright needed)
python3 mirror.py --summary-only

The mirror separates public discovery from authenticated downloads:

1. Fetches the artifact list from the public OData catalog using ordinary HTTP
2. Compares each API's version, state, subtype, and modification timestamp with the existing mirror
3. Exits immediately without a browser when nothing changed
4. When needed, logs in once via Playwright and downloads changed metadata and specifications through Playwright's authenticated HTTP request context
5. Prunes removed APIs, regenerates summaries/collections, and updates the changelog

On a failed login the mirror saves login_failure.png showing what the browser saw. The old python3 scrape.py command remains as a compatibility entry point.

Automation (GitHub Actions + Pages)

Everything also runs unattended on GitHub — no local machine needed:

Workflow	Trigger	What it does
mirror.yml	daily at 05:17 UTC + manual	Checks SAP publicly, mirrors changed specs, opens a deduplicated issue assigned to the repository owner when structural API changes are detected, commits updates to main, and triggers the Pages deploy
deploy-pages.yml	push to main (output/templates/static) + manual	Assembles the static site and publishes it to GitHub Pages
test.yml	push / PR	Runs the pytest suite

The live site is at https://endogen.github.io/sap-dmc-api/.

Required repository secrets (Settings → Secrets and variables → Actions): SAP_USER, SAP_PASS, SAP_ACCOUNT. They are only used when the public catalog check reports changes.

API-change issues use the workflow's built-in GITHUB_TOKEN; no additional secret is required. Metadata-only updates do not create an issue, and rerunning a change report does not create a duplicate.

The mirror job runs with --min-specs 80, so a half-failed SAP session aborts instead of committing a gutted dataset. Chromium is installed only when the public catalog changed.

A single API that SAP serves badly costs that API rather than the run: its previous spec is kept, the name is listed as a workflow warning, and the other APIs are still mirrored and committed. A lost session is different — every remaining API would fail the same way, so the run stops. Failures in the changelog step are not caught at all; they fail the run rather than leaving the changelog silently stale.

If login fails, the run uploads a login-failure artifact with a screenshot of what SAP showed the headless browser. Form fields are blanked out first, because artifacts on a public repository are downloadable by anyone and the login page would otherwise show the address in SAP_USER. If that redaction cannot be applied, no screenshot is written.

Runtime dependencies are pinned exactly in requirements.txt. An unattended 05:17 UTC job driving a headless browser is the last place a surprise release should land, so bump them deliberately.

Browsing APIs

The same interface that runs on GitHub Pages can be served locally — no build step or extra dependencies required.

python3 serve.py
# Opens http://127.0.0.1:8080

The landing page lists all APIs with search/filter, endpoint counts, and links to view each spec in Swagger UI or download the raw JSON.

The server binds to localhost by default; use the HOST/PORT environment variables to change that:

PORT=9090 python3 serve.py
HOST=0.0.0.0 python3 serve.py   # expose on the network

Postman & Insomnia Collections

Collections are regenerated automatically after every mirror update. To rebuild them manually from existing specs:

python3 generate_collections.py

This creates two files in output/collections/:

File	Format	Import via
postman_collection.json	Postman v2.1	Postman → Import → File
insomnia_collection.json	Insomnia Export v4	Insomnia → Import → From File

Both collections include:

• One folder per API, one request per endpoint
• Method, path, description, and request body examples (generated from schemas)
• {{base_url}} / {{token}} variables for host and auth configuration

Refreshing the Mirror

The daily GitHub Actions run takes care of this. A manual run checks the public catalog and only downloads specifications that changed; --force performs a complete refresh. Successful updates regenerate the summary and collections and record spec changes in output/changelog.json.

python3 mirror.py
git diff --stat output/
git add -A && git commit -m "Update specs $(date +%Y-%m-%d)"

Tests

pip install -r requirements-dev.txt
python -m pytest

Covers the diff tracker (change detection and breaking-change flags), collection generation, and the web server's routing/path-traversal protection.