sap_export.py

"""VynFi SAP Integration Pack — generate → pull → foreign-key check (Scale+).

Runs a generation job with ``exportFormat: "sap"`` and the S/4HANA CDS
dialect, lists the SAP tables that landed in the archive, parses BKPF +
BSEG into DataFrames, and asserts the document-header / line-item
foreign key holds (``BSEG.BELNR ⊆ BKPF.BELNR``).

Uses the portal's default 8-table set — pass ``tables=[...]`` on
``SapExportConfig`` to request any subset of the 27-table superset
(see ``vynfi.SAP_ALL_TABLES``). A 0.25× credit multiplier applies when
``exportFormat ∈ {sap, saft}``.

Requires pandas: ``pip install vynfi[pandas]``.
"""

from __future__ import annotations

import os

import vynfi
from vynfi import SapExportConfig

client = vynfi.VynFi(api_key=os.environ["VYNFI_API_KEY"], timeout=180.0)

sap = SapExportConfig(
    dialect="hana",  # S/4HANA CDS (UTF-8 BOM, semicolon, decimal comma)
    client="200",
    ledger="0L",
    source_system="DATASYNTH",
    local_currency="EUR",
    # tables omitted → portal's default 8-table set
    # (bkpf, bseg, acdoca, lfa1, kna1, mara, csks, cepc)
)

print("Submitting SAP export job …")
job = client.jobs.generate_config(
    config={
        "sector": "retail",
        "country": "DE",
        "accountingFramework": "ifrs",
        "rows": 500,
        "companies": 1,
        "periods": 2,
        "periodLength": "monthly",
        "processModels": ["o2c", "p2p"],
        "exportFormat": "sap",
        "output": {"sap": sap.to_dict()},
    },
)
done = client.jobs.wait(job.id, timeout=420)
print(f"  job {done.id}  → {done.status}")
if done.status != "completed":
    raise SystemExit(done.error_detail or "Generation did not complete.")

archive = client.jobs.download_archive(done.id)
print(f"\n{archive}")

tables = archive.sap_tables()
print(f"\nSAP tables in sap_export/: {tables}")

bkpf = archive.sap_table_dataframe("bkpf")
bseg = archive.sap_table_dataframe("bseg")

if bkpf is None or bseg is None:
    raise SystemExit("Expected BKPF + BSEG in the archive — check tier and config.")

print(f"\nBKPF: {len(bkpf)} document headers, columns: {list(bkpf.columns[:8])}")
print(f"BSEG: {len(bseg)} line items,         columns: {list(bseg.columns[:8])}")

# Foreign-key sanity — the same assertion the DS upstream smoke test makes.
bkpf_ids = set(bkpf["BELNR"])
bseg_ids = set(bseg["BELNR"])
orphans = bseg_ids - bkpf_ids
if orphans:
    print(f"\n✗ {len(orphans)} BSEG.BELNR values missing from BKPF — foreign key broken")
else:
    print("\n✓ Every BSEG.BELNR resolves to a BKPF.BELNR (FK integrity holds)")

# Show the per-company distribution — useful for multi-company BODS loads.
if "BUKRS" in bkpf.columns:
    print("\nBKPF by company code (BUKRS):")
    for bukrs, n in bkpf["BUKRS"].value_counts().items():
        print(f"  {bukrs}: {n}")

archive.close()

print(
    "\nTo load these into your SAP S/4HANA test system:"
    "\n  - UTF-8 BOM + semicolon + decimal comma = BODS/HANA-native format"
    "\n  - BKPF → document headers, BSEG → line items, ACDOCA → Universal Journal"
    "\n  - LFA1 / KNA1 / MARA / CSKS / CEPC are master data for referential joins"
)