cyclone/backend/tests/test_api_batch_export_837.py

"""Tests for POST /api/batches/{batch_id}/export-837.

Reads Claim.raw_json for each requested claim_id and returns a ZIP of
regenerated X12 837 files. No DB state mutation. Mirrors the
X-Cyclone-Serialize-Errors convention from /api/inbox/rejected/resubmit?download=true.
"""
import io
import json
import zipfile
from pathlib import Path

from fastapi.testclient import TestClient

from cyclone.api import app


def _seed_batch(client: TestClient, filename: str = "claim.txt") -> dict:
    """Parse a single 837P file and return a dict with ``batch_id`` and
    ``claims``.

    The parse-837 happy-path response does not currently surface
    ``batch_id`` at the top level (it's a server-side UUID, surfaced in
    409 errors but not in 200s). We look it up from the DB — the Batch
    row is already persisted by the time the parse endpoint returns.
    """
    from cyclone import db
    from cyclone.db import Batch

    fixture = Path("tests/fixtures/co_medicaid_837p.txt").read_text()
    r = client.post(
        "/api/parse-837",
        files={"file": (filename, io.BytesIO(fixture.encode()), "text/plain")},
        headers={"Accept": "application/json"},
    )
    assert r.status_code == 200, r.text
    body = r.json()
    with db.SessionLocal()() as s:
        most_recent = s.query(Batch).order_by(Batch.parsed_at.desc()).first()
        assert most_recent is not None, "expected at least one Batch row after parse"
        batch_id = most_recent.id
    return {"batch_id": batch_id, "claims": body["claims"]}


def _claim_ids_from_seed(seeded: dict) -> list[str]:
    return [c["claim_id"] for c in seeded["claims"]]


# ---------------------------------------------------------------------------
# Happy path
# ---------------------------------------------------------------------------


def test_happy_path_returns_zip_with_one_x12_per_claim():
    with TestClient(app) as client:
        seeded = _seed_batch(client)
        batch_id = seeded["batch_id"]
        claim_ids = _claim_ids_from_seed(seeded)

        r = client.post(
            f"/api/batches/{batch_id}/export-837",
            json={"claim_ids": claim_ids},
        )

    assert r.status_code == 200, r.text
    assert r.headers["content-type"].startswith("application/zip")
    assert "attachment" in r.headers["content-disposition"]
    assert (
        f"batch-{batch_id}-{len(claim_ids)}-claims.zip"
        in r.headers["content-disposition"]
    )
    # Per-claim failures header absent on full success.
    assert "x-cyclone-serialize-errors" not in {k.lower() for k in r.headers.keys()}

    with zipfile.ZipFile(io.BytesIO(r.content)) as zf:
        names = zf.namelist()
        assert len(names) == len(claim_ids)
        # Each entry follows the HCPF outbound naming template
        # "{tpid}-837P-{yyyymmddhhmmssSSS}-1of1.x12" (the seeded
        # clearhouse TPID is "11525703"). Every name must be unique.
        from cyclone.edi.filenames import is_outbound_filename
        seen = set()
        for name in names:
            assert is_outbound_filename(name), (
                f"expected HCPF outbound filename, got {name!r}"
            )
            assert name not in seen, f"duplicate filename: {name}"
            seen.add(name)
            with zf.open(name) as f:
                first_line = f.readline().decode("ascii", errors="replace")
            assert first_line.startswith("ISA*"), f"{name} didn't start with ISA"


def test_each_x12_in_zip_uses_clearhouse_submit_and_payer_receiver():
    """Regression: regenerated 837s used to emit 'CYCLONE' / 'RECEIVER'
    placeholders. The export endpoint must thread the clearhouse
    submitter (dzinesco's TPID 11525703) and payer receiver
    (COMEDASSISTPROG) through to the serializer."""
    with TestClient(app) as client:
        seeded = _seed_batch(client)
        batch_id = seeded["batch_id"]
        claim_ids = _claim_ids_from_seed(seeded)
        r = client.post(
            f"/api/batches/{batch_id}/export-837",
            json={"claim_ids": claim_ids},
        )
    assert r.status_code == 200, r.text

    with zipfile.ZipFile(io.BytesIO(r.content)) as zf:
        for name in zf.namelist():
            with zf.open(name) as f:
                text = f.read().decode("ascii")
            # Submitter block must use the clearhouse TPID + name, not
            # the 'CYCLONE' placeholder.
            assert "CYCLONE" not in text, f"{name} still emits CYCLONE placeholder"
            assert "11525703" in text, f"{name} missing clearhouse TPID"
            assert "Dzinesco" in text, f"{name} missing clearhouse name"
            # Receiver block must use the CO_TXIX payer config
            # (COMEDASSISTPROG), not the 'RECEIVER' placeholder.
            assert "RECEIVER" not in text, f"{name} still emits RECEIVER placeholder"
            assert "COMEDASSISTPROG" in text, f"{name} missing receiver id"
            # Loop 1000A requires PER — must be present, not omitted.
            assert "PER*IC*" in text, f"{name} missing required PER segment"
            # SBR09 should be 'MC' (Medicaid claim filing indicator),
            # not the member id.
            sbr_line = next(seg for seg in text.split("~") if seg.startswith("SBR*"))
            sbr09 = sbr_line.rstrip("~").split("*")[9]
            assert sbr09 == "MC", f"{name} SBR09 expected 'MC', got {sbr09!r}"


def test_each_x12_in_zip_round_trips_through_parser():
    """Fidelity check: each .x12 must parse back to a ClaimOutput deep-equal
    to the source row's raw_json (modulo recomputed validation)."""
    from cyclone.parsers.parse_837 import parse
    from cyclone.parsers.payer import PayerConfig

    with TestClient(app) as client:
        seeded = _seed_batch(client)
        batch_id = seeded["batch_id"]
        claim_ids = _claim_ids_from_seed(seeded)

        r = client.post(
            f"/api/batches/{batch_id}/export-837",
            json={"claim_ids": claim_ids},
        )
    assert r.status_code == 200, r.text

    by_id = {c["claim_id"]: c for c in seeded["claims"]}
    with zipfile.ZipFile(io.BytesIO(r.content)) as zf:
        for name in zf.namelist():
            with zf.open(name) as f:
                text = f.read().decode("ascii")
            result = parse(text, PayerConfig(name="CO_MEDICAID"))
            assert result.claims, f"{name} didn't parse back to any claims"
            # Claim ids must round-trip (proves the serializer didn't drop
            # or rewrite the id).
            assert result.claims[0].claim.claim_id in by_id, (
                f"{name} parsed to an unknown claim_id"
            )


# ---------------------------------------------------------------------------
# Partial-failure surface
# ---------------------------------------------------------------------------


def test_partial_failure_one_claim_with_no_raw_json_returns_zip_and_errors_header():
    """If one claim has raw_json=None, the ZIP still returns for the others
    and the failure is surfaced via X-Cyclone-Serialize-Errors."""
    from cyclone import db
    from cyclone.edi.filenames import is_outbound_filename

    with TestClient(app) as client:
        seeded = _seed_batch(client)
        batch_id = seeded["batch_id"]
        claim_ids = _claim_ids_from_seed(seeded)
        assert len(claim_ids) >= 2, "fixture must produce at least 2 claims"

        # Wipe raw_json on one claim to simulate a corrupted row.
        with db.SessionLocal()() as s:
            from cyclone.db import Claim
            target = s.get(Claim, claim_ids[0])
            target.raw_json = None
            s.commit()

        r = client.post(
            f"/api/batches/{batch_id}/export-837",
            json={"claim_ids": claim_ids},
        )

    assert r.status_code == 200, r.text
    # Filename uses SUCCESS count, not requested count.
    expected_success = len(claim_ids) - 1
    assert (
        f"batch-{batch_id}-{expected_success}-claims.zip"
        in r.headers["content-disposition"]
    )
    err_header = r.headers.get("x-cyclone-serialize-errors")
    assert err_header, "expected X-Cyclone-Serialize-Errors header"
    errs = json.loads(err_header)
    assert len(errs) == 1
    assert errs[0]["claim_id"] == claim_ids[0]
    assert "raw_json" in errs[0]["reason"].lower()

    with zipfile.ZipFile(io.BytesIO(r.content)) as zf:
        names = zf.namelist()
        assert len(names) == expected_success
        # Every entry follows HCPF outbound template; none is the
        # 'claim-CLM-X.x12' placeholder from the old endpoint.
        for name in names:
            assert is_outbound_filename(name), f"non-HCPF name: {name!r}"


def test_all_claims_fail_to_serialize_returns_422():
    from cyclone import db

    with TestClient(app) as client:
        seeded = _seed_batch(client)
        batch_id = seeded["batch_id"]
        claim_ids = _claim_ids_from_seed(seeded)
        assert len(claim_ids) >= 1

        with db.SessionLocal()() as s:
            from cyclone.db import Claim
            for cid in claim_ids:
                c = s.get(Claim, cid)
                c.raw_json = None
            s.commit()

        r = client.post(
            f"/api/batches/{batch_id}/export-837",
            json={"claim_ids": claim_ids},
        )

    assert r.status_code == 422, r.text
    body = r.json()
    # The 422 body surfaces the failure list so the UI can show details
    # without parsing a header.
    errs = body.get("detail", {}).get("serialize_errors") or body.get("serialize_errors")
    assert errs is not None
    assert len(errs) == len(claim_ids)
    for entry in errs:
        assert entry["claim_id"] in claim_ids


# ---------------------------------------------------------------------------
# Error cases
# ---------------------------------------------------------------------------


def test_empty_claim_ids_returns_400():
    with TestClient(app) as client:
        seeded = _seed_batch(client)
        batch_id = seeded["batch_id"]
        r = client.post(
            f"/api/batches/{batch_id}/export-837",
            json={"claim_ids": []},
        )
    assert r.status_code == 400, r.text


def test_missing_claim_ids_key_returns_400():
    with TestClient(app) as client:
        seeded = _seed_batch(client)
        batch_id = seeded["batch_id"]
        r = client.post(
            f"/api/batches/{batch_id}/export-837",
            json={},
        )
    assert r.status_code == 400, r.text


def test_unknown_batch_id_returns_404():
    with TestClient(app) as client:
        r = client.post(
            "/api/batches/BATCH-DOES-NOT-EXIST/export-837",
            json={"claim_ids": ["CLM-1"]},
        )
    assert r.status_code == 404, r.text
    body = r.json()
    assert "BATCH-DOES-NOT-EXIST" in (body.get("detail") or "")


def test_unknown_claim_id_silently_omitted():
    """A claim_id that doesn't exist is omitted from the ZIP and surfaced
    in the errors header — same convention as the resubmit endpoint."""
    with TestClient(app) as client:
        seeded = _seed_batch(client)
        batch_id = seeded["batch_id"]
        real_ids = _claim_ids_from_seed(seeded)
        assert real_ids, "fixture must produce at least 1 claim"

        requested = real_ids + ["CLM-GHOST"]
        r = client.post(
            f"/api/batches/{batch_id}/export-837",
            json={"claim_ids": requested},
        )

    assert r.status_code == 200, r.text
    # Filename uses success count = len(real_ids), not len(requested).
    assert (
        f"batch-{batch_id}-{len(real_ids)}-claims.zip"
        in r.headers["content-disposition"]
    )
    err_header = r.headers.get("x-cyclone-serialize-errors")
    assert err_header
    errs = json.loads(err_header)
    assert any(e["claim_id"] == "CLM-GHOST" for e in errs)

    with zipfile.ZipFile(io.BytesIO(r.content)) as zf:
        names = zf.namelist()
        # HCPF outbound filenames, one per real claim. We don't pin the
        # exact ts (it depends on the wall clock at test time), but the
        # count and the HCPF template must match.
        from cyclone.edi.filenames import is_outbound_filename
        assert len(names) == len(real_ids)
        for name in names:
            assert is_outbound_filename(name), f"non-HCPF name: {name!r}"