procurement-ingest/tests/test_reprocess_contract.py

"""Synthetic-event-shape contract test for scripts/reprocess.py (refactor §4.7).

Pins the exact S3 event shape reprocess emits and asserts the object key is
the RAW list_objects_v2 key — reprocess applies NO URL-encoding or decoding.

Why this matters: a real S3 event notification URL-encodes the object key, and
the handler is where any decode would live. reprocess builds its synthetic
event from the RAW list_objects_v2 obj["Key"] (or the raw --key arg), so it
must emit that key byte-for-byte. Replaying a pre-decoded (or pre-encoded) key
would not match what s3.get_object(Bucket, Key) expects downstream.

Recon note (honored narrowly): neither the PO nor the WO handler imports
urllib.parse or calls unquote/unquote_plus today — the raw event key goes
straight into s3.get_object. So the "handler decodes -> replaying a decoded
key double-decodes" failure mode is NOT reproducible against today's handlers
(they never decode once). This test therefore guards ONLY reprocess's own
contract: emit the raw list_objects_v2 key with no transformation. That keeps
replay single-decode-correct if a future handler ever adds unquote_plus.
"""

import importlib.util
import os
import pathlib
import sys

import pytest

# reprocess.py builds its boto3 clients inside main()/functions, so importing
# it is side-effect-free (no AWS calls, no clients at module scope). Set dummy
# region/creds defensively anyway so import stays offline-safe even if that
# ever changes. build_s3_event itself makes no AWS calls.
os.environ.setdefault("AWS_DEFAULT_REGION", "us-east-1")
os.environ.setdefault("AWS_ACCESS_KEY_ID", "testing")
os.environ.setdefault("AWS_SECRET_ACCESS_KEY", "testing")
os.environ.setdefault("AWS_SESSION_TOKEN", "testing")

# scripts/ is not a package (no __init__.py) and is not on sys.path, and
# tests/conftest.py only wires up per-handler dirs — so load reprocess.py by
# file path with importlib rather than `import scripts.reprocess`.
_p = pathlib.Path(__file__).resolve().parents[1] / "scripts" / "reprocess.py"
_spec = importlib.util.spec_from_file_location("reprocess", _p)
reprocess = importlib.util.module_from_spec(_spec)
_spec.loader.exec_module(reprocess)


def test_build_s3_event_shape_and_raw_key():
    bucket = "po-ingest-emails-328440206208"
    # Deliberately contains a SPACE, a literal '+', and a literal '%41'.
    # A real S3 event notification would deliver this key encoded as
    #   "inbound/2026/AB+12%2B34+%2541.eml"
    # (space->'+', '+'->'%2B', '%'->'%25'); the handler is where any decode
    # would live. reprocess builds from the RAW list_objects_v2 key, so it
    # must emit "inbound/2026/AB 12+34 %41.eml" unchanged.
    raw_key = "inbound/2026/AB 12+34 %41.eml"

    event = reprocess.build_s3_event(bucket, raw_key)

    # 1. Exact top-level shape: a single Records entry.
    assert list(event.keys()) == ["Records"]
    assert len(event["Records"]) == 1

    # 2. Bucket name is nested at Records[0].s3.bucket.name.
    assert event["Records"][0]["s3"]["bucket"]["name"] == bucket

    # 3. Object key is nested at Records[0].s3.object.key, byte-for-byte.
    assert event["Records"][0]["s3"]["object"]["key"] == raw_key

    # 4. RAW / no-URL-encoding, no-URL-decoding on the emitted key.
    emitted = event["Records"][0]["s3"]["object"]["key"]
    assert emitted == raw_key  # reprocess applies NO transformation
    # space NOT percent-encoded (a real S3 notification would send %20):
    assert " " in emitted and "%20" not in emitted
    # literal '+' preserved, not turned into a space:
    assert "+" in emitted
    # '%41' NOT decoded to 'A':
    assert "%41" in emitted and "A .eml" not in emitted


# --- Selector mutual-exclusion: --all is a MODE, not a modifier ---------------
# reprocess.main() reaches its first AWS call (sts get_caller_identity) only
# AFTER the argparse mutual-exclusion gate, so an invalid selector combination
# raises SystemExit(2) with no AWS access needed. This pins that `--all` cannot
# be silently combined with a narrower --prefix/--since (which the args.all
# clobber branch would otherwise discard -> unintended whole-corpus sweep).
def _expect_argparse_reject(monkeypatch, capsys, argv):
    monkeypatch.setattr(sys, "argv", ["reprocess.py", *argv])
    with pytest.raises(SystemExit) as exc:
        reprocess.main()
    assert exc.value.code == 2  # argparse error exit code
    return capsys.readouterr().err


def test_all_rejects_prefix(monkeypatch, capsys):
    err = _expect_argparse_reject(
        monkeypatch,
        capsys,
        ["--pipeline", "po", "--all", "--prefix", "inbound/2026/07/"],
    )
    assert "--all cannot be combined with --prefix or --since" in err


def test_all_rejects_since(monkeypatch, capsys):
    err = _expect_argparse_reject(
        monkeypatch,
        capsys,
        ["--pipeline", "po", "--all", "--since", "2026-07-01T00:00:00Z"],
    )
    assert "--all cannot be combined with --prefix or --since" in err


def test_all_rejects_prefix_and_since_together(monkeypatch, capsys):
    _expect_argparse_reject(
        monkeypatch,
        capsys,
        [
            "--pipeline",
            "po",
            "--all",
            "--prefix",
            "inbound/2026/",
            "--since",
            "2026-07-01T00:00:00Z",
        ],
    )