procurement-ingest/tests/test_api_pagination_moto.py

"""Cursor pagination against real DynamoDB semantics (moto).

The offline handler tests fake Dynamo, which can't exercise real
LastEvaluatedKey behavior. Here moto-backed tables with the production key
schemas prove: pages are disjoint and complete, the final page's next_cursor
is null, comment Query pagination respects the composite key, and hostile
cursors (garbage, wrong key attrs, huge) map to 400 -- never 500, never a raw
ExclusiveStartKey error.

The repo modules build their boto3 resource at import time; a resource created
before mock_aws starts is NOT intercepted, so each test rebinds the modules'
``dynamodb`` to a fresh resource inside the mock (same trap as test_po_merge's
moto-before-handler ordering).
"""

import base64
import json

import boto3
import pytest
from moto import mock_aws

from tests.support import load_lambda_module


def _event(method, resource, path_params=None, qs=None):
    return {
        "httpMethod": method,
        "resource": resource,
        "pathParameters": path_params or {},
        "queryStringParameters": qs or {},
        "headers": {},
    }


@pytest.fixture()
def api(monkeypatch):
    with mock_aws():
        mod = load_lambda_module("api", "handler")
        resource = boto3.resource("dynamodb", region_name="us-east-1")
        monkeypatch.setattr(mod.wo_repo, "dynamodb", resource)
        monkeypatch.setattr(mod.po_repo, "dynamodb", resource)

        resource.create_table(
            TableName=mod.wo_repo.WORK_ORDERS_TABLE,
            KeySchema=[{"AttributeName": "work_order_id", "KeyType": "HASH"}],
            AttributeDefinitions=[
                {"AttributeName": "work_order_id", "AttributeType": "S"}
            ],
            BillingMode="PAY_PER_REQUEST",
        )
        resource.create_table(
            TableName=mod.wo_repo.COMMENTS_TABLE,
            KeySchema=[
                {"AttributeName": "work_order_id", "KeyType": "HASH"},
                {"AttributeName": "comment_id", "KeyType": "RANGE"},
            ],
            AttributeDefinitions=[
                {"AttributeName": "work_order_id", "AttributeType": "S"},
                {"AttributeName": "comment_id", "AttributeType": "S"},
            ],
            BillingMode="PAY_PER_REQUEST",
        )
        resource.create_table(
            TableName=mod.po_repo.PO_TABLE,
            KeySchema=[{"AttributeName": "po_number", "KeyType": "HASH"}],
            AttributeDefinitions=[{"AttributeName": "po_number", "AttributeType": "S"}],
            BillingMode="PAY_PER_REQUEST",
        )
        resource.create_table(
            TableName=mod.po_repo.VERIFIED_SITES_TABLE,
            KeySchema=[{"AttributeName": "siteCode", "KeyType": "HASH"}],
            AttributeDefinitions=[{"AttributeName": "siteCode", "AttributeType": "S"}],
            BillingMode="PAY_PER_REQUEST",
        )
        yield mod, resource


def _walk_pages(mod, resource_path, qs_extra=None, path_params=None, limit=5):
    seen, pages = [], 0
    cursor = None
    while True:
        qs = {"limit": str(limit), **(qs_extra or {})}
        if cursor:
            qs["cursor"] = cursor
        response = mod.handler(
            _event("GET", resource_path, path_params=path_params, qs=qs), None
        )
        assert response["statusCode"] == 200
        body = json.loads(response["body"])
        seen.extend(body["items"])
        pages += 1
        cursor = body["next_cursor"]
        assert pages < 50, "pagination failed to terminate"
        if cursor is None:
            return seen, pages


def test_work_order_scan_pages_are_disjoint_and_complete(api):
    mod, resource = api
    table = resource.Table(mod.wo_repo.WORK_ORDERS_TABLE)
    for i in range(12):
        table.put_item(Item={"work_order_id": str(10000 + i), "wo_status": "new"})

    items, pages = _walk_pages(mod, "/work-orders", limit=5)
    ids = [item["work_order_id"] for item in items]
    assert len(ids) == 12
    assert len(set(ids)) == 12
    assert pages >= 3


def test_comment_query_pagination_composite_key(api):
    mod, resource = api
    table = resource.Table(mod.wo_repo.COMMENTS_TABLE)
    for i in range(7):
        table.put_item(
            Item={
                "work_order_id": "555",
                "comment_id": f"555#2026-01-0{i + 1}T00:00:00#{i:012d}",
                "text": f"comment {i}",
            }
        )
    # A second work order's comments must never bleed into the page.
    table.put_item(
        Item={"work_order_id": "666", "comment_id": "666#x#0", "text": "other"}
    )

    items, _ = _walk_pages(
        mod,
        "/work-orders/{workOrderId}/comments",
        path_params={"workOrderId": "555"},
        limit=3,
    )
    assert len(items) == 7
    assert {item["work_order_id"] for item in items} == {"555"}


def test_purchase_order_and_site_pagination(api):
    mod, resource = api
    po_table = resource.Table(mod.po_repo.PO_TABLE)
    for i in range(6):
        po_table.put_item(Item={"po_number": f"2D-{i:08d}"})
    sites_table = resource.Table(mod.po_repo.VERIFIED_SITES_TABLE)
    for code in ("JFK8", "WNY2", "UVA5"):
        sites_table.put_item(Item={"siteCode": code})

    po_items, _ = _walk_pages(mod, "/purchase-orders", limit=4)
    assert len(po_items) == 6
    site_items, _ = _walk_pages(mod, "/verified-sites", limit=2)
    assert {item["siteCode"] for item in site_items} == {"JFK8", "WNY2", "UVA5"}


@pytest.mark.parametrize(
    "cursor",
    [
        "not-base64!!!",
        "aGVsbG8=",  # base64("hello") -- not JSON-dict
        "e30=",  # base64("{}") -- empty dict
        # base64 of {"evil_attr": "x"} -- key attr not allowed for this table
        "eyJldmlsX2F0dHIiOiAieCJ9",
        "A" * 3000,  # oversized
    ],
)
def test_hostile_cursor_is_400_not_500(api, cursor):
    mod, resource = api
    resource.Table(mod.wo_repo.WORK_ORDERS_TABLE).put_item(Item={"work_order_id": "1"})
    response = mod.handler(_event("GET", "/work-orders", qs={"cursor": cursor}), None)
    assert response["statusCode"] == 400
    assert "error" in json.loads(response["body"])


def _cursor(payload):
    return base64.urlsafe_b64encode(json.dumps(payload).encode()).decode()


def test_partial_composite_cursor_is_400_not_500(api):
    # A work-orders-shaped cursor {work_order_id} is a partial key for the
    # comments Query (needs work_order_id+comment_id). Exact-key match rejects
    # it as 400 rather than letting DynamoDB ValidationException -> 500.
    mod, resource = api
    resource.Table(mod.wo_repo.COMMENTS_TABLE).put_item(
        Item={"work_order_id": "555", "comment_id": "555#x#0"}
    )
    response = mod.handler(
        _event(
            "GET",
            "/work-orders/{workOrderId}/comments",
            path_params={"workOrderId": "555"},
            qs={"cursor": _cursor({"work_order_id": "555"})},
        ),
        None,
    )
    assert response["statusCode"] == 400


def test_cross_work_order_comment_cursor_is_400(api):
    # A full comments cursor minted for WO A must not resume WO B's Query.
    mod, resource = api
    response = mod.handler(
        _event(
            "GET",
            "/work-orders/{workOrderId}/comments",
            path_params={"workOrderId": "B"},
            qs={"cursor": _cursor({"work_order_id": "A", "comment_id": "A#x#0"})},
        ),
        None,
    )
    assert response["statusCode"] == 400


def test_dynamodb_validation_error_maps_to_400(api, monkeypatch):
    # Defense in depth: even if some crafted-but-valid cursor reached DynamoDB
    # and raised ValidationException, the handler maps it to 400, not a 500
    # that would page the zero-threshold 5xx alarm.
    from botocore.exceptions import ClientError

    mod, _ = api

    class _RaisingTable:
        def scan(self, **kwargs):
            raise ClientError(
                {"Error": {"Code": "ValidationException", "Message": "bad key"}},
                "Scan",
            )

    monkeypatch.setattr(
        mod.wo_repo,
        "dynamodb",
        type("D", (), {"Table": lambda self, n: _RaisingTable()})(),
    )
    response = mod.handler(_event("GET", "/work-orders"), None)
    assert response["statusCode"] == 400