procurement-ingest/lambdas/wo/email_processor/tests/test_bedrock_fallback.py

"""End-to-end dispatch tests: the Bedrock AI extractor is invoked ONLY when the
deterministic parse misses, EMF metrics are emitted on both paths, and the
Bedrock response is decoded through the same contract."""

import json
import os

import pytest

import handler
from _wo_parser_support import FIXTURES


def _raw(subdir, stem):
    with open(os.path.join(FIXTURES, subdir, f"{stem}.eml"), "rb") as fh:
        return fh.read()


class FakeBody:
    def __init__(self, data):
        self._data = data

    def read(self):
        return self._data


class FakeS3:
    def __init__(self, raw):
        self._raw = raw

    def get_object(self, Bucket, Key):  # noqa: N803
        return {"Body": FakeBody(self._raw)}


class FakeBedrock:
    def __init__(self, payload):
        self.calls = []
        self._payload = payload

    def invoke_model(self, modelId, body):  # noqa: N803
        self.calls.append({"modelId": modelId, "body": body})
        text = json.dumps(self._payload)
        return {"body": FakeBody(json.dumps({"content": [{"text": text}]}).encode())}


AI_17_KEY = {
    "email_type": "comment",
    "work_order_id": "77777777777",
    "description": None,
    "status": None,
    "site_code": None,
    "building": None,
    "address": None,
    "severity": None,
    "priority": None,
    "date_reported": None,
    "scheduled_start": None,
    "due_date": None,
    "assigned_to": None,
    "commenter": None,
    "comment_text": "ai extracted",
    "comment_time": None,
}


def _event():
    return {
        "Records": [
            {
                "s3": {
                    "bucket": {"name": "workorder-ingest-emails-x"},
                    "object": {"key": "inbound/o1"},
                }
            }
        ]
    }


@pytest.fixture
def metric_spy(monkeypatch):
    calls = []
    monkeypatch.setattr(
        handler,
        "emit_parse_metric",
        lambda *a: calls.append(a),
    )
    return calls


def test_template_path_skips_bedrock(fake_dynamo, metric_spy, monkeypatch):
    fake_bedrock = FakeBedrock(AI_17_KEY)
    monkeypatch.setattr(
        handler, "s3", FakeS3(_raw("update-plaintext", "update-plaintext-01"))
    )
    monkeypatch.setattr(handler, "bedrock", fake_bedrock)
    # This suite exercises parse dispatch, not the fail-closed SES sender-auth
    # gate (INFRA-107) that now runs first in handler(); the scrubbed .eml
    # fixtures carry no SES-stamped Authentication-Results header, so bypass it
    # here. Authentication itself is covered by tests/test_ses_auth.py.
    monkeypatch.setattr(handler, "authenticate_inbound_email", lambda *a: True)

    handler.handler(_event(), None)

    assert fake_bedrock.calls == []  # deterministic parse handled it
    method, template_id, reason, wo = metric_spy[0]
    assert method == "template"
    assert template_id == "update_plaintext"
    assert reason == "ok"
    assert wo == "11144580730"


def test_fallback_path_invokes_bedrock(fake_dynamo, metric_spy, monkeypatch):
    fake_bedrock = FakeBedrock(AI_17_KEY)
    monkeypatch.setattr(handler, "s3", FakeS3(_raw("ai-fallback", "unknown-subject")))
    monkeypatch.setattr(handler, "bedrock", fake_bedrock)
    # See note in test_template_path_skips_bedrock: bypass the INFRA-107 sender
    # auth gate so this dispatch test reaches the parse path.
    monkeypatch.setattr(handler, "authenticate_inbound_email", lambda *a: True)

    handler.handler(_event(), None)

    assert len(fake_bedrock.calls) == 1
    # Uses the configured inference profile and the bedrock message contract.
    call = fake_bedrock.calls[0]
    assert call["modelId"] == handler.BEDROCK_MODEL_ID
    body = json.loads(call["body"])
    assert body["anthropic_version"] == "bedrock-2023-05-31"
    assert body["max_tokens"] == 1024
    method, _template_id, _reason, wo = metric_spy[0]
    assert method == "ai_fallback"
    assert wo == "77777777777"
    # The AI result was written through to DynamoDB.
    wo_table = fake_dynamo.tables[handler.WORK_ORDERS_TABLE]
    assert wo_table.updates, "expected a work-order upsert from the AI path"


def test_extract_with_bedrock_returns_full_contract(monkeypatch):
    fake_bedrock = FakeBedrock(AI_17_KEY)
    monkeypatch.setattr(handler, "bedrock", fake_bedrock)
    email_data = {
        "subject": "x",
        "sender": "a",
        "to": "b",
        "cc": "",
        "date": "d",
        "body": "body",
    }
    out = handler.extract_with_bedrock(email_data)
    assert set(out.keys()) == set(AI_17_KEY.keys())
    assert len(fake_bedrock.calls) == 1


def test_extract_with_bedrock_strips_markdown_fence(monkeypatch):
    class FenceBedrock:
        def invoke_model(self, modelId, body):  # noqa: N803
            fenced = "```json\n" + json.dumps(AI_17_KEY) + "\n```"
            return {
                "body": FakeBody(json.dumps({"content": [{"text": fenced}]}).encode())
            }

    monkeypatch.setattr(handler, "bedrock", FenceBedrock())
    out = handler.extract_with_bedrock(
        {"subject": "", "sender": "", "to": "", "cc": "", "date": "", "body": ""}
    )
    assert out["work_order_id"] == "77777777777"


def test_emit_parse_metric_writes_emf(capsys):
    handler.emit_parse_metric("template", "update_plaintext", "ok", "123")
    line = capsys.readouterr().out.strip()
    emf = json.loads(line)
    assert emf["ParseMethod"] == "template"
    assert emf["TemplateId"] == "update_plaintext"
    assert emf["ReasonCode"] == "ok"
    assert emf["ParseOutcome"] == 1
    dims = emf["_aws"]["CloudWatchMetrics"][0]["Dimensions"]
    # Two dimension sets must be published: the ParseMethod-only aggregate that
    # the fallback-rate alarm queries, AND the per-template breakdown. Without
    # the ["ParseMethod"] set the alarm's single-dimension series never receives
    # data and can never fire (regression guard for the coverage-collapse alarm).
    assert ["ParseMethod"] in dims
    assert ["ParseMethod", "TemplateId"] in dims
    assert (
        emf["_aws"]["CloudWatchMetrics"][0]["Namespace"] == "Seahaven/WorkorderIngest"
    )