capo-lambda 0.14.0__tar.gz → 0.15.0__tar.gz

{capo_lambda-0.14.0/src/capo_lambda.egg-info → capo_lambda-0.15.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: capo-lambda
-Version: 0.14.0
+Version: 0.15.0
 Summary: Python SDK for AWS Lambda.
 Classifier: Development Status :: 4 - Beta
 Classifier: Intended Audience :: Developers

{capo_lambda-0.14.0 → capo_lambda-0.15.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "capo-lambda"
-version = "0.14.0"
+version = "0.15.0"
 description = "Python SDK for AWS Lambda."
 readme = "README.md"
 requires-python = ">=3.10"

{capo_lambda-0.14.0 → capo_lambda-0.15.0}/src/capo_lambda/_auth/_signers.py

@@ -7,7 +7,7 @@ from zapros import Request
 
 from capo_lambda._auth._identity import Credentials, Identity
 from capo_lambda._auth._providers import IdentityProvider
-from capo_lambda._auth._sigv4 import SigV4AuthContext, sign_sigv4
+from capo_lambda._auth._sigv4 import S3_SIGNING_NAMES, SigV4AuthContext, sign_sigv4
 
 IdentityT = TypeVar("IdentityT", bound="Identity")
 
@@ -30,14 +30,25 @@ class SigV4Signer(Signer[Credentials]):
     The full auth scheme (``name`` variant, ``signingName``, ``signingRegion``,
     encoding/normalization flags) is provided by the caller — either from the
     endpoint rule-set's ``authSchemes`` property or built by the generated
-    ``get_signer`` from operation defaults.
+    ``get_signer`` from operation defaults. ``unsigned_payload`` mirrors the
+    operation's ``aws.auth#unsignedPayload`` trait: the body is sent but left
+    out of the signature. ``event_stream`` marks a request event stream: the
+    request is signed with the ``STREAMING-AWS4-HMAC-SHA256-EVENTS`` payload
+    marker and every event with a signature chained from it.
     """
 
     def __init__(
-        self, provider: IdentityProvider[Credentials], *, auth_scheme: dict[str, Any]
+        self,
+        provider: IdentityProvider[Credentials],
+        *,
+        auth_scheme: dict[str, Any],
+        unsigned_payload: bool = False,
+        event_stream: bool = False,
     ) -> None:
         super().__init__(provider)
         self._auth_scheme = auth_scheme
+        self._unsigned_payload = unsigned_payload
+        self._event_stream = event_stream
 
     async def asign(self, req: Request) -> Request:
         creds = await self.provider.aresolve_identity()
@@ -55,13 +66,24 @@ class SigV4Signer(Signer[Credentials]):
                 "disableNormalizePath", False
             ),
         }
-        if req.body is None:
-            body: bytes | None = b""
-        elif isinstance(req.body, bytes):
-            body = req.body
-        else:
-            body = None
-        return sign_sigv4(req, ctx, body)
+        unsigned_payload = self._unsigned_payload
+        if (
+            not unsigned_payload
+            and not self._event_stream
+            and req.body is not None
+            and not isinstance(req.body, bytes)
+        ):
+            if self._auth_scheme["signingName"] in S3_SIGNING_NAMES:
+                # S3 accepts UNSIGNED-PAYLOAD for any operation; streamed bodies rely on it.
+                unsigned_payload = True
+            else:
+                raise NotImplementedError(
+                    "Currently we don't support signed chunked payloads, so buffer the body and "
+                    "pass bytes as a workaround; chunked signed implementation coming soon"
+                )
+        return sign_sigv4(
+            req, ctx, unsigned_payload=unsigned_payload, event_stream=self._event_stream
+        )
 
     def sign(self, req: Request) -> Request:
         creds = self.provider.resolve_identity()
@@ -79,10 +101,21 @@ class SigV4Signer(Signer[Credentials]):
                 "disableNormalizePath", False
             ),
         }
-        if req.body is None:
-            body: bytes | None = b""
-        elif isinstance(req.body, bytes):
-            body = req.body
-        else:
-            body = None
-        return sign_sigv4(req, ctx, body)
+        unsigned_payload = self._unsigned_payload
+        if (
+            not unsigned_payload
+            and not self._event_stream
+            and req.body is not None
+            and not isinstance(req.body, bytes)
+        ):
+            if self._auth_scheme["signingName"] in S3_SIGNING_NAMES:
+                # S3 accepts UNSIGNED-PAYLOAD for any operation; streamed bodies rely on it.
+                unsigned_payload = True
+            else:
+                raise NotImplementedError(
+                    "Currently we don't support signed chunked payloads, so buffer the body and "
+                    "pass bytes as a workaround; chunked signed implementation coming soon"
+                )
+        return sign_sigv4(
+            req, ctx, unsigned_payload=unsigned_payload, event_stream=self._event_stream
+        )

{capo_lambda-0.14.0 → capo_lambda-0.15.0}/src/capo_lambda/_auth/_sigv4.py

@@ -1,4 +1,4 @@
-"""AWS Signature Version 4 — single-chunk signing.
+"""AWS Signature Version 4 — single-chunk and event-stream signing.
 
 Reference:
     https://docs.aws.amazon.com/IAM/latest/UserGuide/reference_sigv-create-signed-request.html
@@ -15,7 +15,8 @@ import functools
 import hashlib
 import hmac
 import re
-from typing import Any, Literal, TypedDict
+from collections.abc import AsyncIterator, Iterator
+from typing import Any, Literal, TypedDict, cast
 from urllib.parse import quote, unquote
 
 import zapros
@@ -23,6 +24,9 @@ from pywhatwgurl import URLSearchParams
 from zapros import Headers, Request
 from zapros._utils import get_host_header_value
 
+from .._iter import AnyIterator
+from .._protocol.eventstream import HeaderValue, Message, encode_headers
+
 
 def build_sigv4_auth_scheme(
     signing_name: str, region: str | None, endpoint_scheme: dict[str, Any] | None = None
@@ -65,6 +69,10 @@ class SigV4AuthContext(TypedDict):
 
 
 _SIGV4_ALGORITHM = "AWS4-HMAC-SHA256"
+# Per-event algorithm and request payload marker of a signed event stream.
+_EVENT_ALGORITHM = "AWS4-HMAC-SHA256-PAYLOAD"
+_EVENT_STREAM_PAYLOAD = "STREAMING-AWS4-HMAC-SHA256-EVENTS"
+_AMZ_DATE_FORMAT = "%Y%m%dT%H%M%SZ"
 _EMPTY_PAYLOAD_SHA256 = hashlib.sha256(b"").hexdigest()
 
 # Headers excluded from the signed-headers set. Mirrors botocore's denylist:
@@ -95,9 +103,11 @@ _UNSIGNED_HEADERS = frozenset(
 
 _MULTI_SPACE = re.compile(r" +")
 
-# Services that require the payload hash to travel in ``x-amz-content-sha256``.
-# Other services sign the hash into the canonical request without sending it.
-_S3_SIGNING_NAMES = frozenset({"s3", "s3express", "s3-outposts", "s3-object-lambda"})
+# Services that require the payload hash to travel in ``x-amz-content-sha256``
+# on every request, and that accept ``UNSIGNED-PAYLOAD`` for any operation.
+# Other services sign the hash into the canonical request without sending it,
+# and only see the header when the payload is left unsigned.
+S3_SIGNING_NAMES = frozenset({"s3", "s3express", "s3-outposts", "s3-object-lambda"})
 
 
 def _uri_encode(value: str) -> str:
@@ -231,12 +241,20 @@ def _canonical_query_from_pairs(pairs: list[tuple[str, str]]) -> str:
 def sign_sigv4(
     request: Request,
     ctx: SigV4AuthContext,
-    body: bytes | None,
+    *,
+    unsigned_payload: bool = False,
+    event_stream: bool = False,
 ) -> Request:
     """Return a new ``Request`` carrying SigV4 single-chunk auth headers.
 
-    Pass ``body=None`` to sign with ``UNSIGNED-PAYLOAD`` (streaming requests).
-    The original ``request.body`` is forwarded unchanged in that case.
+    The payload hash covers ``request.body``, which must be ``bytes`` or
+    ``None``. ``unsigned_payload`` signs ``UNSIGNED-PAYLOAD`` instead and
+    forwards any body unchanged (streaming S3 requests and operations carrying
+    ``aws.auth#unsignedPayload``). ``event_stream`` signs a request event
+    stream: the ``STREAMING-AWS4-HMAC-SHA256-EVENTS`` marker stands in for the
+    payload hash and the body — an iterator of encoded event messages — is
+    wrapped in an :class:`EventStreamIterator` that signs every event with a
+    signature chained from this request's.
     """
     service = ctx["signing_name"]
     region = ctx["signing_region"]
@@ -250,21 +268,31 @@ def sign_sigv4(
         date_stamp = amz_date[:8]
     else:
         now = _amz_now()
-        amz_date = now.strftime("%Y%m%dT%H%M%SZ")
+        amz_date = now.strftime(_AMZ_DATE_FORMAT)
         date_stamp = now.strftime("%Y%m%d")
         headers["X-Amz-Date"] = amz_date
 
-    # Payload hash. For S3-family services, x-amz-content-sha256 is mandatory
-    # and must be set BEFORE computing the canonical request (it gets signed).
+    # Payload hash. For S3-family services, x-amz-content-sha256 is mandatory;
+    # for every service it is the only way to announce an unsigned payload or
+    # an event stream.
+    # Either way it must be set BEFORE computing the canonical request (it
+    # gets signed).
     payload_hash = headers.get("X-Amz-Content-SHA256")
     if payload_hash is None:
-        if body is None:
+        if event_stream:
+            payload_hash = _EVENT_STREAM_PAYLOAD
+        elif unsigned_payload:
             payload_hash = "UNSIGNED-PAYLOAD"
+        elif request.body is None:
+            payload_hash = _EMPTY_PAYLOAD_SHA256
+        elif isinstance(request.body, bytes):
+            payload_hash = hashlib.sha256(request.body).hexdigest()
         else:
-            payload_hash = (
-                hashlib.sha256(body).hexdigest() if body else _EMPTY_PAYLOAD_SHA256
+            raise TypeError(
+                "sign_sigv4 hashes a bytes body only; pass unsigned_payload=True "
+                "to send a streamed body without a payload hash"
             )
-    if service in _S3_SIGNING_NAMES:
+    if service in S3_SIGNING_NAMES or unsigned_payload or event_stream:
         headers["X-Amz-Content-SHA256"] = payload_hash
 
     # Session token (STS / assumed-role credentials).
@@ -308,18 +336,135 @@ def sign_sigv4(
         f"Signature={signature}"
     )
 
-    effective_body = body if body is not None else request.body
-    if effective_body is not None:
+    body = request.body
+    if event_stream:
+        if not isinstance(body, (Iterator, AsyncIterator)):
+            raise TypeError(
+                "event_stream requires an iterator of encoded events as the body"
+            )
+        body = EventStreamIterator(
+            cast("Iterator[bytes] | AsyncIterator[bytes]", body), ctx, signature
+        )
+    if body is not None:
         return Request(
             request.url,
             request.method,
             headers,
-            body=effective_body,
+            body=body,
             context=request.context,
         )
     return Request(request.url, request.method, headers, context=request.context)
 
 
+def _sign_event(
+    ctx: SigV4AuthContext, prior_signature: str, payload: bytes
+) -> tuple[bytes, str]:
+    """Wrap one encoded event in a signed outer message.
+
+    Returns ``(frame, signature)``; the hex signature seeds the next event.
+    The string to sign chains every frame to its predecessor — the first one
+    to the request's ``Authorization`` signature — so the service can verify
+    the stream incrementally::
+
+        AWS4-HMAC-SHA256-PAYLOAD
+        <date, YYYYMMDDTHHMMSSZ>
+        <credential scope>
+        <prior signature, hex>
+        <hex sha256 of the encoded ``:date`` header>
+        <hex sha256 of the payload>
+
+    The AWS SDKs (and the service) use the ``-PAYLOAD`` algorithm name; the
+    Transcribe developer guide's pseudocode omits the suffix.
+    """
+    service = ctx["signing_name"]
+    region = ctx["signing_region"]
+    now = _amz_now()
+    amz_date = now.strftime(_AMZ_DATE_FORMAT)
+    date_stamp = amz_date[:8]
+    credential_scope = f"{date_stamp}/{region}/{service}/aws4_request"
+    date_header: dict[str, HeaderValue] = {":date": now}
+    string_to_sign = "\n".join(
+        (
+            _EVENT_ALGORITHM,
+            amz_date,
+            credential_scope,
+            prior_signature,
+            hashlib.sha256(encode_headers(date_header)).hexdigest(),
+            hashlib.sha256(payload).hexdigest(),
+        )
+    )
+    signing_key = _derive_signing_key(
+        ctx["secret_access_key"], date_stamp, region, service
+    )
+    signature = hmac.new(
+        signing_key, string_to_sign.encode("utf-8"), hashlib.sha256
+    ).hexdigest()
+    frame = Message(
+        {**date_header, ":chunk-signature": bytes.fromhex(signature)}, payload
+    ).encode()
+    return frame, signature
+
+
+class EventStreamIterator(AnyIterator[bytes]):
+    """Sign a request event stream, one frame per event.
+
+    Wraps an iterator of encoded event messages (as produced by the generated
+    ``serialize_event_*`` functions). Each event becomes the payload of an
+    outer message carrying ``:date`` and ``:chunk-signature`` headers, the
+    signature chained from the previous frame's and seeded by the request's
+    ``Authorization`` signature. Once the events are exhausted, one empty
+    signed frame is yielded to mark the end of the stream.
+
+    Like :class:`~capo._body.Body`, it is both a sync and an async iterator:
+    iterate it the way the wrapped source is iterable (``next`` over a sync
+    source, ``async for`` over an async one).
+
+    Reference:
+        https://docs.aws.amazon.com/transcribe/latest/dg/streaming-setting-up.html
+    """
+
+    def __init__(
+        self,
+        events: Iterator[bytes] | AsyncIterator[bytes],
+        ctx: SigV4AuthContext,
+        seed_signature: str,
+    ) -> None:
+        self._events = events
+        self._ctx = ctx
+        self._prior_signature = seed_signature
+        self._ended = False
+
+    def _frame(self, payload: bytes) -> bytes:
+        frame, self._prior_signature = _sign_event(
+            self._ctx, self._prior_signature, payload
+        )
+        return frame
+
+    def __next__(self) -> bytes:
+        if self._ended:
+            raise StopIteration
+        if not isinstance(self._events, Iterator):
+            raise TypeError("this event stream wraps an async source; use `async for`")
+        try:
+            payload = next(cast(Iterator[bytes], self._events))
+        except StopIteration:
+            self._ended = True
+            payload = b""
+        return self._frame(payload)
+
+    async def __anext__(self) -> bytes:
+        if self._ended:
+            raise StopAsyncIteration
+        if not isinstance(self._events, AsyncIterator):
+            raise TypeError("this event stream wraps a sync source; use `for`")
+        try:
+            payload = await cast(AsyncIterator[bytes], self._events).__anext__()
+        except StopAsyncIteration:
+            self._ended = True
+            payload = b""
+        return self._frame(payload)
+
+
 def presign_sigv4(
     request: Request,
     ctx: SigV4AuthContext,
@@ -346,7 +491,7 @@ def presign_sigv4(
     service = ctx["signing_name"]
     region = ctx["signing_region"]
 
-    amz_date = sign_time.strftime("%Y%m%dT%H%M%SZ")
+    amz_date = sign_time.strftime(_AMZ_DATE_FORMAT)
     date_stamp = sign_time.strftime("%Y%m%d")
     credential_scope = f"{date_stamp}/{region}/{service}/aws4_request"
 

{capo_lambda-0.14.0 → capo_lambda-0.15.0}/src/capo_lambda/_protocol/eventstream.py

@@ -89,7 +89,7 @@ def _decode_header_value(
     raise ValueError(f"unsupported eventstream header type: {type_byte:#x}")
 
 
-def _encode_headers(headers: dict[str, HeaderValue]) -> bytes:
+def encode_headers(headers: dict[str, HeaderValue]) -> bytes:
     parts: list[bytes] = []
     for name, value in headers.items():
         name_bytes = name.encode("utf-8")
@@ -124,10 +124,11 @@ class Message:
 
     def encode(self) -> bytes:
         """Serialize this message into eventstream framing bytes."""
-        headers_bytes = _encode_headers(self.headers)
+        headers_bytes = encode_headers(self.headers)
         headers_length = len(headers_bytes)
         payload = self.payload
-        total_length = 12 + headers_length + len(payload)
+        # total_length covers the whole frame, trailing message CRC included.
+        total_length = 12 + headers_length + len(payload) + 4
 
         prelude = total_length.to_bytes(4, "big") + headers_length.to_bytes(4, "big")
         prelude_crc = zlib.crc32(prelude).to_bytes(4, "big")
@@ -158,31 +159,31 @@ class MessageDecoder:
             prelude = bytes(self._buffer[:8])
             prelude_crc = int.from_bytes(self._buffer[8:12], "big")
 
-            if total_length < 12 + headers_length:
+            # total_length covers the whole frame: 12-byte prelude, headers,
+            # payload, and the trailing 4-byte message CRC.
+            if total_length < 16 + headers_length:
                 raise ValueError(
                     f"invalid eventstream message: total_length={total_length} "
-                    f"is smaller than minimum {12 + headers_length}"
+                    f"is smaller than minimum {16 + headers_length}"
                 )
 
-            message_end = total_length + 4
-            if len(self._buffer) < message_end:
+            if len(self._buffer) < total_length:
                 return
 
             computed_prelude_crc = zlib.crc32(prelude) & 0xFFFFFFFF
             if computed_prelude_crc != prelude_crc:
                 raise ValueError("eventstream prelude checksum mismatch")
 
-            message_bytes = bytes(self._buffer[:total_length])
-            message_crc = int.from_bytes(
-                self._buffer[total_length : total_length + 4], "big"
-            )
+            payload_end = total_length - 4
+            message_bytes = bytes(self._buffer[:payload_end])
+            message_crc = int.from_bytes(self._buffer[payload_end:total_length], "big")
             computed_message_crc = zlib.crc32(message_bytes) & 0xFFFFFFFF
             if computed_message_crc != message_crc:
                 raise ValueError("eventstream message checksum mismatch")
 
             headers = _decode_headers(self._buffer[12 : 12 + headers_length])
-            payload = bytes(self._buffer[12 + headers_length : total_length])
-            del self._buffer[:message_end]
+            payload = bytes(self._buffer[12 + headers_length : payload_end])
+            del self._buffer[:total_length]
             yield Message(headers=headers, payload=payload)
 
 
@@ -233,6 +234,7 @@ __all__ = [
     "MessageDecoder",
     "async_raw_stream_to_events",
     "async_read_messages",
+    "encode_headers",
     "raw_stream_to_events",
     "read_messages",
 ]

{capo_lambda-0.14.0 → capo_lambda-0.15.0}/src/capo_lambda/types/invoke_response_stream_update.py

@@ -27,7 +27,11 @@ def deserialize_json(data: dict) -> InvokeResponseStreamUpdate:
 
 
 def serialize_event_json(value: InvokeResponseStreamUpdate) -> bytes:
-    headers: dict[str, HeaderValue] = {":event-type": "PayloadChunk"}
+    headers: dict[str, HeaderValue] = {
+        ":message-type": "event",
+        ":event-type": "PayloadChunk",
+        ":content-type": "application/octet-stream",
+    }
     payload = b""
     payload = value["payload"]
     return Message(headers=headers, payload=payload).encode()

{capo_lambda-0.14.0 → capo_lambda-0.15.0}/src/capo_lambda/types/invoke_with_response_stream_complete_event.py

@@ -1,5 +1,6 @@
 """Generated from Smithy shape ``com.amazonaws.lambda#InvokeWithResponseStreamCompleteEvent``."""
 
+import json
 from typing import TYPE_CHECKING
 
 from typing_extensions import NotRequired, TypedDict
@@ -43,8 +44,13 @@ def deserialize_json(data: dict) -> InvokeWithResponseStreamCompleteEvent:
 
 
 def serialize_event_json(value: InvokeWithResponseStreamCompleteEvent) -> bytes:
-    headers: dict[str, HeaderValue] = {":event-type": "InvokeComplete"}
+    headers: dict[str, HeaderValue] = {
+        ":message-type": "event",
+        ":event-type": "InvokeComplete",
+        ":content-type": "application/json",
+    }
     payload = b""
+    payload = json.dumps(serialize_json(value)).encode("utf-8")
     return Message(headers=headers, payload=payload).encode()
 
 
@@ -52,4 +58,6 @@ def deserialize_event_json(message: Message) -> InvokeWithResponseStreamComplete
     headers = message.headers  # noqa: F841
     payload = message.payload  # noqa: F841
     out: InvokeWithResponseStreamCompleteEvent = {}  # type: ignore[typeddict-item]
+    if payload:
+        out = deserialize_json(json.loads(payload))
     return out

{capo_lambda-0.14.0 → capo_lambda-0.15.0}/src/capo_lambda/types/invoke_with_response_stream_response_event.py

@@ -6,6 +6,9 @@ from typing_extensions import TypedDict
 
 from capo_lambda._iter import AnyIterator
 from capo_lambda._protocol.eventstream import Message
+from capo_lambda.errors import (
+    UnknownServiceError,
+)
 
 if TYPE_CHECKING:
     import capo_lambda.types.invoke_response_stream_update
@@ -53,7 +56,20 @@ def serialize_event_json(value: _InvokeWithResponseStreamResponseEvent) -> bytes
 
 def deserialize_event_json(message: Message) -> _InvokeWithResponseStreamResponseEvent:
     headers = message.headers
-    message_type = headers.get(":message-type", "event")  # noqa: F841
+    message_type = headers.get(":message-type", "event")
+    if message_type == "exception":
+        exception_type = headers.get(":exception-type")
+        raise UnknownServiceError(
+            code=str(exception_type), message=None, response=message
+        )
+    if message_type == "error":
+        error_code = headers.get(":error-code")
+        error_message = headers.get(":error-message")
+        raise UnknownServiceError(
+            code=None if error_code is None else str(error_code),
+            message=None if error_message is None else str(error_message),
+            response=message,
+        )
     event_type = headers.get(":event-type")
     match event_type:
         case "PayloadChunk":

{capo_lambda-0.14.0 → capo_lambda-0.15.0/src/capo_lambda.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: capo-lambda
-Version: 0.14.0
+Version: 0.15.0
 Summary: Python SDK for AWS Lambda.
 Classifier: Development Status :: 4 - Beta
 Classifier: Intended Audience :: Developers