Sap Service Layer Expert

SAP Service Layer Expert

Quick start

1. Read SESSION.md for the complete ServiceLayer context-manager class.
2. Copy assets/script_template.py as the starting point for every new script.
3. Look up endpoint schemas in assets/spec/paths/{ResourceName}.yaml for field names.
4. Check ENDPOINTS.md for available HTTP methods and action functions per resource.
5. See REFERENCE.md for base URL/auth, OData params, HTTP semantics, document chaining, and schema resolution strategies.

Two-file output contract

Every mutation script ships exactly two files under .temp/ in the repo root:

Naming: snake_case, verb first — patch_business_partners, create_sales_order.

# Example main()
def main(data: list, sl: ServiceLayer) -> None:
    for record in data:
        card_code = record.pop("CardCode")
        sl.patch(f"BusinessPartners('{card_code}')", record)
        log.info("Patched %s", card_code)

Script checklist

• Derive repo root from workspace context; create .temp/ if needed
• _data.json contains only data — no credentials, no boilerplate
• _run.py starts from assets/script_template.py
• Always use PATCH (not PUT) for partial updates
• Log out in finally (handled by ServiceLayer.__exit__)
• main(data, sl) implements the full logic
• Use sl.patch() for updates (not sl.post() or PUT)
• Log each record processed with log.info()
• Verify key field names against assets/spec/paths/{Resource}_id.yaml before patching
• After creating both files, tell the user they can review and ask you to run the script

Running a script

When the user asks to run a script:

1. Run it directly — do not check for .env files or environment variables beforehand. The script validates its own environment on startup and will print a clear error if credentials are missing.
2. Working directory: the main repo root (derived from workspace context — the app folder that owns the current task). Scripts live under .temp/ within that root, but cd to the repo root so that python-dotenv finds the .env file there.
3. Python interpreter: use the bench virtualenv Python at /workspace/development/frappe-bench/env/bin/python.

cd <main-repo-root>
/workspace/development/frappe-bench/env/bin/python .temp/<operation>_run.py

Related skills

See REFERENCE.md for schema resolution strategies and ⚠️ property name / DB column mapping pitfalls.