procurement-ingest/lambdas/api/handler.py

194 lines
7.6 KiB
Python
Raw Normal View History

feat(api): procurement-api read stack + OpenAPI docs (SHOC reconciliation path) (#127) * feat(api): add procurement-api stack - read API + OpenAPI docs page Third CDK stack: API Gateway REST API (IAM SigV4) over both pipelines' tables, replacing SHOC's retired SyncController cross-account DynamoDB scan as the reconciliation/backfill path. - lambdas/api/: handler (healthcheck + docs-token gate + router dispatch), router (single route table), pagination (opaque cursor, hostile -> 400), Decimal-safe serialization, wo_repo/po_repo reads. No VendorReplies. - OpenAPI 3.1 spec as source of truth incl. top-level webhooks section documenting the outbound SHOC feed; phase-2 write endpoints x-planned (router answers 501). Self-contained /docs page, no CDN. - Auth: AWS_IAM on data routes + resource policy scoped to exactly arn:aws:iam::396287094661:role/shoc-backend-dev on GET/*; /docs and /openapi.json carve-out is token-gated in the Lambda via shared web_ui_auth (fail-closed, INFRA-74 posture). - KMS: explicit Decrypt/DescribeKey on the DynamoDB CMK from SSM (name-imported table drops the key association - INFRA-104 class). - Alarms: errors/throttles/duration(p99>=22.5s) + gateway 5xx, ALARM-only to site-alerts. No access logging in v1 (docs ?token= shim stays out of logs); cloud_watch_role=False. - Tests: handler auth-seam + routing + Decimal round-trip; moto cursor pagination incl. hostile cursors; spec<->router drift gate; bundle AST pins for the api command; pytest.ini --cov + loader siblings. - Deploy role: third stack DescribeStacks ARN + procurement-api smoke invoke ARN (re-run create-deploy-role.sh before merge). * harden(api): apply sh-security-review findings to procurement-api Fan-out (6 detectors) + review findings resolved: Correctness / DoS: - pagination: require EXACT key-set match (was subset) so a partial/foreign composite cursor can't reach DynamoDB as an inconsistent ExclusiveStartKey -> ValidationException -> 500; comments Query now pins the cursor's work_order_id to the path entity. - handler: map botocore ValidationException to 400 (defense in depth) so a crafted cursor can't drive the zero-threshold 5xx alarm. - web_ui_auth: compare tokens as bytes; a non-ASCII presented token now fails closed (401) instead of crashing hmac.compare_digest into a 500. Resolves the pre-existing xfail(strict) follow-up test; hardens the web UIs too. Docs page: - typeStr() now escapes the one spec-derived string that reached innerHTML. - spec inlined into the docs <script> block escapes "<" -> < (</script> breakout guard); /openapi.json still served byte-faithful. - Cache-Control: no-store + Referrer-Policy: no-referrer on docs responses so the ?token= URL stays out of caches/Referer. - spec-drift test asserts the committed spec carries no "</" / "<!--". IAM / IaC: - resource policy enumerates the 7 data GET resources instead of GET/* so a future GET route can't silently inherit SHOC cross-account reach. - kms:Decrypt grant gains a kms:ViaService=dynamodb condition. - stage throttling (50 rps / 100 burst) bounds the unauthenticated /docs blast radius below the 10k account default. - corrected the PATCH/POST comment (same-account callers aren't blocked by the resource policy; 501 handler + absent write grant are the gate). - documented the RETAIN log-group first-deploy rollback trap and the resource-policy-needs-redeploy gotcha in-stack. Mandatory GPT-4.1 cross-family review of the full policy surface: no BLOCK/FIX. 675 tests pass, ruff clean, cdk synth green.
2026-07-23 19:32:20 -04:00
"""Procurement API Lambda (API Gateway REST proxy integration).
Auth is split by route class and enforced at two layers:
- Data routes: AWS_IAM at the gateway (SigV4; cross-account callers allowed by
the API resource policy). The handler does NOT re-check a token there --
authorization is API Gateway's job on those routes.
- Docs routes (/docs, /openapi.json): reachable at the gateway (auth NONE +
resource-policy carve-out) but the handler fails closed on the shared
header token via web_ui_auth (same secret + constant-time compare as the
web UIs), so they are never an unauthenticated data path (INFRA-74).
"""
import logging
from pathlib import Path
import po_repo
import wo_repo
from botocore.exceptions import ClientError
from pagination import BadCursor, clamp_limit
from router import DATA_ROUTES, DOCS_ROUTES, PLANNED_ROUTES
from serialization import error_response, json_response
from web_ui_auth import is_authenticated
logger = logging.getLogger()
logger.setLevel(logging.INFO)
_MODULE_DIR = Path(__file__).resolve().parent
_SPEC_PATH = _MODULE_DIR / "openapi.json"
_DOCS_PATH = _MODULE_DIR / "docs.html"
# The docs page template carries this placeholder where the spec JSON is
# inlined, so /docs is a single token-gated request (a browser can't attach
# the auth header to a follow-up asset fetch).
_SPEC_PLACEHOLDER = "__OPENAPI_SPEC_JSON__"
_spec_cache = None
_docs_cache = None
_REPO_FUNCS = {
"list_work_orders": wo_repo.list_work_orders,
"get_work_order": wo_repo.get_work_order,
"list_comments": wo_repo.list_comments,
"list_purchase_orders": po_repo.list_purchase_orders,
"get_purchase_order": po_repo.get_purchase_order,
"list_verified_sites": po_repo.list_verified_sites,
"get_verified_site": po_repo.get_verified_site,
}
_PATH_PARAM_BY_RESOURCE = {
"/work-orders/{workOrderId}": "workOrderId",
"/work-orders/{workOrderId}/comments": "workOrderId",
"/purchase-orders/{poNumber}": "poNumber",
"/verified-sites/{siteCode}": "siteCode",
}
# Headers on the docs responses: the token rides in the ?token= query shim,
# so keep the token-keyed URL and page out of shared/browser caches and out of
# any Referer sent to a followed link.
_DOCS_SECURITY_HEADERS = {
"Cache-Control": "no-store",
"Referrer-Policy": "no-referrer",
}
def _load_spec() -> str:
"""Raw spec bytes, served verbatim at /openapi.json (byte-faithful JSON)."""
global _spec_cache
if _spec_cache is None:
_spec_cache = _SPEC_PATH.read_text(encoding="utf-8")
return _spec_cache
def _load_docs_html() -> str:
global _docs_cache
if _docs_cache is None:
# Neutralize any "</script"/"<!--" a future spec edit could introduce:
# the spec is spliced into a <script type="application/json"> block
# where the HTML parser ends the element at the first literal "<"
# sequence regardless of JSON quoting. "<" is the same JSON
# string value ("<") to JSON.parse but can never close the script tag.
inlined_spec = _load_spec().replace("<", "\\u003c")
_docs_cache = _DOCS_PATH.read_text(encoding="utf-8").replace(
_SPEC_PLACEHOLDER, inlined_spec
)
return _docs_cache
def _with_token_shim(event: dict) -> dict:
"""Copy a ?token= query parameter into an x-auth-token header.
Browsers can't set headers on plain navigation, so /docs accepts the
shared token as a query parameter too. The constant-time compare still
happens inside web_ui_auth -- this only synthesizes the header on a
shallow copy. Acceptable only while API Gateway access logging stays off
(nothing at the gateway records the query string); rotate the token if
access logging is ever enabled.
"""
qs = event.get("queryStringParameters") or {}
token = qs.get("token")
if not token:
return event
shimmed = dict(event)
headers = dict(event.get("headers") or {})
headers["x-auth-token"] = token
shimmed["headers"] = headers
return shimmed
def _docs_response(resource: str, event: dict) -> dict:
# SECURITY INVARIANT: these routes serve ONLY the committed spec and the
# static docs page -- never table data. The gateway resource policy allows
# Principal "*" on exactly these two GETs on the strength of that; serving
# anything dynamic here requires a resource-policy + security re-review.
if not is_authenticated(_with_token_shim(event)):
return error_response(401, "unauthorized")
if resource == "/openapi.json":
return {
"statusCode": 200,
"headers": {"Content-Type": "application/json", **_DOCS_SECURITY_HEADERS},
"body": _load_spec(),
}
return {
"statusCode": 200,
"headers": {
"Content-Type": "text/html; charset=utf-8",
**_DOCS_SECURITY_HEADERS,
},
"body": _load_docs_html(),
}
def _data_response(route_key: tuple, event: dict) -> dict:
method, resource = route_key
func = _REPO_FUNCS[DATA_ROUTES[route_key]]
qs = event.get("queryStringParameters") or {}
path_params = event.get("pathParameters") or {}
if DATA_ROUTES[route_key].startswith("list_"):
limit = clamp_limit(qs.get("limit"))
cursor = qs.get("cursor")
if resource in _PATH_PARAM_BY_RESOURCE:
entity_id = path_params.get(_PATH_PARAM_BY_RESOURCE[resource], "")
items, next_cursor = func(entity_id, limit, cursor)
else:
items, next_cursor = func(limit, cursor)
return json_response(200, {"items": items, "next_cursor": next_cursor})
entity_id = path_params.get(_PATH_PARAM_BY_RESOURCE[resource], "")
item = func(entity_id)
if item is None:
return error_response(404, "not found")
return json_response(200, item)
def _dispatch(route_key: tuple, event: dict) -> dict:
if route_key in PLANNED_ROUTES:
return error_response(501, "planned endpoint - not implemented (phase 2)")
if route_key in DOCS_ROUTES:
return _docs_response(route_key[1], event)
if route_key in DATA_ROUTES:
return _data_response(route_key, event)
return error_response(404, "not found")
def handler(event, context):
# Deploy-guard healthcheck: a direct-invoke {"healthcheck": true} probe
# returns before any routing/auth so the post-deploy smoke gate can verify
# the bundle imports and the runtime boots.
if isinstance(event, dict) and event.get("healthcheck") is True:
return {"healthcheck": "ok"}
method = (event.get("httpMethod") or "").upper()
resource = event.get("resource") or ""
try:
return _dispatch((method, resource), event)
except BadCursor as exc:
return error_response(400, str(exc))
except ClientError as exc:
# A client-supplied cursor that survives validation but is still
# inconsistent at the data layer makes DynamoDB raise
# ValidationException; map it to 400, not 500, so a crafted cursor
# can't drive the 5xx alarm. Any other AWS error is a real 500.
if exc.response.get("Error", {}).get("Code") == "ValidationException":
return error_response(400, "cursor is not valid")
logger.exception("AWS error serving %s %s", method, resource)
return error_response(500, "internal error")
except Exception:
# A raised exception would surface as an opaque 502 from the proxy
# integration; return a clean 500 instead. The API Gateway 5XX alarm
# pages on these; the exception (never the request token) is logged.
logger.exception("Unhandled error serving %s %s", method, resource)
return error_response(500, "internal error")