localstack · bentsku · Aug 15, 2023 · Aug 9, 2023 · Aug 10, 2023 · Aug 10, 2023
diff --git a/localstack/aws/api/s3/__init__.py b/localstack/aws/api/s3/__init__.py
@@ -853,6 +853,14 @@ class ObjectLockConfigurationNotFoundError(ServiceException):
     BucketName: Optional[BucketName]
 
 
+class InvalidPartNumber(ServiceException):
+    code: str = "InvalidPartNumber"
+    sender_fault: bool = False
+    status_code: int = 416
+    PartNumberRequested: Optional[PartNumber]
+    ActualPartCount: Optional[PartNumber]
+
+
 AbortDate = datetime
 
 
@@ -2233,6 +2241,7 @@ class HeadObjectOutput(TypedDict, total=False):
     ObjectLockMode: Optional[ObjectLockMode]
     ObjectLockRetainUntilDate: Optional[ObjectLockRetainUntilDate]
     ObjectLockLegalHoldStatus: Optional[ObjectLockLegalHoldStatus]
+    StatusCode: Optional[GetObjectResponseStatusCode]
 
 
 class HeadObjectRequest(ServiceRequest):

diff --git a/localstack/aws/spec-patches.json b/localstack/aws/spec-patches.json
@@ -120,6 +120,14 @@
         "location": "statusCode"
       }
     },
+    {
+      "op": "add",
+      "path": "/shapes/HeadObjectOutput/members/StatusCode",
+      "value": {
+        "shape": "GetObjectResponseStatusCode",
+        "location": "statusCode"
+      }
+    },
     {
       "op": "add",
       "path": "/shapes/NoSuchKey/members/Key",
@@ -1048,6 +1056,26 @@
         "documentation": "<p>Object Lock configuration does not exist for this bucket</p>",
         "exception": true
       }
+    },
+    {
+      "op": "add",
+      "path": "/shapes/InvalidPartNumber",
+      "value": {
+        "type": "structure",
+        "members": {
+          "PartNumberRequested": {
+            "shape": "PartNumber"
+          },
+          "ActualPartCount": {
+            "shape": "PartNumber"
+          }
+        },
+        "error": {
+          "httpStatusCode": 416
+        },
+        "documentation": "<p>The requested partnumber is not satisfiable</p>",
+        "exception": true
+      }
     }
   ]
 }
diff --git a/localstack/services/s3/provider.py b/localstack/services/s3/provider.py
@@ -166,6 +166,7 @@
     capitalize_header_name_from_snake_case,
     extract_bucket_key_version_id_from_copy_source,
     get_bucket_from_moto,
+    get_failed_precondition_copy_source,
     get_key_from_moto_bucket,
     get_lifecycle_rule_from_object,
     get_object_checksum_for_algorithm,
@@ -599,36 +600,15 @@ def copy_object(
         )
 
         # see https://docs.aws.amazon.com/AmazonS3/latest/API/API_CopyObject.html
-        condition = None
         source_object_last_modified = source_key_object.last_modified.replace(
             tzinfo=ZoneInfo("GMT")
         )
-        if (cs_if_match := request.get("CopySourceIfMatch")) and source_key_object.etag.strip(
-            '"'
-        ) != cs_if_match.strip('"'):
-            condition = "x-amz-copy-source-If-Match"
-
-        elif (
-            cs_id_unmodified_since := request.get("CopySourceIfUnmodifiedSince")
-        ) and source_object_last_modified > cs_id_unmodified_since:
-            condition = "x-amz-copy-source-If-Unmodified-Since"
-
-        elif (
-            cs_if_none_match := request.get("CopySourceIfNoneMatch")
-        ) and source_key_object.etag.strip('"') == cs_if_none_match.strip('"'):
-            condition = "x-amz-copy-source-If-None-Match"
-
-        elif (
-            cs_id_modified_since := request.get("CopySourceIfModifiedSince")
-        ) and source_object_last_modified < cs_id_modified_since < datetime.datetime.now(
-            tz=ZoneInfo("GMT")
+        if failed_condition := get_failed_precondition_copy_source(
+            request, source_object_last_modified, source_key_object.etag
         ):
-            condition = "x-amz-copy-source-If-Modified-Since"
-
-        if condition:
             raise PreconditionFailed(
                 "At least one of the pre-conditions you specified did not hold",
-                Condition=condition,
+                Condition=failed_condition,
             )
 
         response: CopyObjectOutput = call_moto(context)

diff --git a/localstack/services/s3/utils.py b/localstack/services/s3/utils.py
@@ -19,7 +19,11 @@
 from localstack.aws.api.s3 import (
     BucketName,
     ChecksumAlgorithm,
+    CopyObjectRequest,
     CopySource,
+    ETag,
+    GetObjectRequest,
+    HeadObjectRequest,
     InvalidArgument,
     InvalidRange,
     InvalidTag,
@@ -34,6 +38,7 @@
     ObjectSize,
     ObjectVersionId,
     Owner,
+    PreconditionFailed,
     SSEKMSKeyId,
     TaggingHeader,
     TagSet,
@@ -79,6 +84,7 @@
 
 
 RFC1123 = "%a, %d %b %Y %H:%M:%S GMT"
+_gmt_zone_info = ZoneInfo("GMT")
 
 
 def get_owner_for_account_id(account_id: str):
@@ -164,7 +170,7 @@ def digest(self) -> bytes:
         return self.checksum.to_bytes(4, "big")
 
 
-class ParsedRange(NamedTuple):
+class ObjectRange(NamedTuple):
     """
     NamedTuple representing a parsed Range header with the requested S3 object size
     https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Range
@@ -176,13 +182,13 @@ class ParsedRange(NamedTuple):
     end: int  # the end of the end
 
 
-def parse_range_header(range_header: str, object_size: int) -> ParsedRange:
+def parse_range_header(range_header: str, object_size: int) -> ObjectRange:
     """
     Takes a Range header, and returns a dataclass containing the necessary information to return only a slice of an
     S3 object
     :param range_header: a Range header
     :param object_size: the requested S3 object total size
-    :return: ParsedRange
+    :return: ObjectRange
     """
     last = object_size - 1
     _, rspec = range_header.split("=")
@@ -206,7 +212,7 @@ def parse_range_header(range_header: str, object_size: int) -> ParsedRange:
             RangeRequested=range_header,
         )
 
-    return ParsedRange(
+    return ObjectRange(
         content_range=f"bytes {begin}-{end}/{object_size}",
         content_length=end - begin + 1,
         begin=begin,
@@ -585,7 +591,7 @@ def rfc_1123_datetime(src: datetime.datetime) -> str:
 
 
 def str_to_rfc_1123_datetime(value: str) -> datetime.datetime:
-    return datetime.datetime.strptime(value, RFC1123).replace(tzinfo=ZoneInfo("GMT"))
+    return datetime.datetime.strptime(value, RFC1123).replace(tzinfo=_gmt_zone_info)
 
 
 def iso_8601_datetime_without_milliseconds_s3(
@@ -776,10 +782,81 @@ def get_retention_from_now(days: int = None, years: int = None) -> datetime.date
     """
     if not days and not years:
         raise ValueError("Either 'days' or 'years' needs to be provided")
-    now = datetime.datetime.now(tz=ZoneInfo("GMT"))
+    now = datetime.datetime.now(tz=_gmt_zone_info)
     if days:
         retention = now + datetime.timedelta(days=days)
     else:
         retention = now.replace(year=now.year + years)
 
     return retention
+
+
+def get_failed_precondition_copy_source(
+    request: CopyObjectRequest, last_modified: datetime.datetime, etag: ETag
+) -> Optional[str]:
+    """
+    Validate if the source object LastModified and ETag matches a precondition, and if it does, return the failed
+    precondition
+    # see https://docs.aws.amazon.com/AmazonS3/latest/API/API_CopyObject.html
+    :param request: the CopyObjectRequest
+    :param last_modified: source object LastModified
+    :param etag: source object ETag
+    :return str: the failed precondition to raise
+    """
+    if (cs_if_match := request.get("CopySourceIfMatch")) and etag.strip('"') != cs_if_match.strip(
+        '"'
+    ):
+        return "x-amz-copy-source-If-Match"
+
+    elif (
+        cs_if_unmodified_since := request.get("CopySourceIfUnmodifiedSince")
+    ) and last_modified > cs_if_unmodified_since:
+        return "x-amz-copy-source-If-Unmodified-Since"
+
+    elif (cs_if_none_match := request.get("CopySourceIfNoneMatch")) and etag.strip(
+        '"'
+    ) == cs_if_none_match.strip('"'):
+        return "x-amz-copy-source-If-None-Match"
+
+    elif (
+        cs_if_modified_since := request.get("CopySourceIfModifiedSince")
+    ) and last_modified < cs_if_modified_since < datetime.datetime.now(tz=_gmt_zone_info):
+        return "x-amz-copy-source-If-Modified-Since"
+
+
+def validate_failed_precondition(
+    request: GetObjectRequest | HeadObjectRequest, last_modified: datetime.datetime, etag: ETag
+) -> None:
+    """
+    Validate if the object LastModified and ETag matches a precondition, and if it does, return the failed
+    precondition
+    :param request: the GetObjectRequest or HeadObjectRequest
+    :param last_modified: S3 object LastModified
+    :param etag: S3 object ETag
+    :raises PreconditionFailed
+    :raises NotModified, 304 with an empty body
+    """
+    precondition_failed = None
+    if (if_match := request.get("IfMatch")) and etag != if_match.strip('"'):
+        precondition_failed = "If-Match"
+
+    elif (
+        if_unmodified_since := request.get("IfUnmodifiedSince")
+    ) and last_modified > if_unmodified_since:
+        precondition_failed = "If-Unmodified-Since"
+
+    if precondition_failed:
+        raise PreconditionFailed(
+            "At least one of the pre-conditions you specified did not hold",
+            Condition=precondition_failed,
+        )
+
+    if ((if_none_match := request.get("IfNoneMatch")) and etag == if_none_match.strip('"')) or (
+        (if_modified_since := request.get("IfModifiedSince"))
+        and last_modified < if_modified_since < datetime.datetime.now(tz=_gmt_zone_info)
+    ):
+        raise CommonServiceException(
+            message="Not Modified",
+            code="NotModified",
+            status_code=304,
+        )
diff --git a/localstack/services/s3/v3/models.py b/localstack/services/s3/v3/models.py
@@ -82,7 +82,7 @@
 
 LOG = logging.getLogger(__name__)
 
-gmt_zone_info = ZoneInfo("GMT")
+_gmt_zone_info = ZoneInfo("GMT")
 
 
 # note: not really a need to use a dataclass here, as it has a lot of fields, but only a few are set at creation
@@ -133,7 +133,8 @@ def __init__(
         self.object_ownership = object_ownership
         self.object_lock_enabled = object_lock_enabled_for_bucket
         self.encryption_rule = DEFAULT_BUCKET_ENCRYPTION
-        self.creation_date = datetime.now(tz=gmt_zone_info)
+        self.creation_date = datetime.now(tz=_gmt_zone_info)
+        self.payer = Payer.BucketOwner
         self.multiparts = {}
         self.notification_configuration = {}
         self.cors_rules = None
@@ -249,7 +250,7 @@ class S3Object:
     website_redirect_location: Optional[WebsiteRedirectLocation]
     acl: Optional[str]  # TODO: we need to change something here, how it's done?
     is_current: bool
-    parts: Optional[list[tuple[int, int]]]
+    parts: Optional[dict[int, tuple[int, int]]]
     restore: Optional[Restore]
 
     def __init__(
@@ -296,8 +297,8 @@ def __init__(
         self.expiration = expiration
         self.website_redirect_location = website_redirect_location
         self.is_current = True
-        self.last_modified = datetime.now(tz=gmt_zone_info)
-        self.parts = []
+        self.last_modified = datetime.now(tz=_gmt_zone_info)
+        self.parts = {}
         self.restore = None
 
     def get_system_metadata_fields(self) -> dict:
@@ -349,7 +350,7 @@ def is_locked(self, bypass_governance: bool = False) -> bool:
             return False
 
         if self.lock_until:
-            return self.lock_until > datetime.now(tz=gmt_zone_info)
+            return self.lock_until > datetime.now(tz=_gmt_zone_info)
 
         return False
 
@@ -364,7 +365,7 @@ class S3DeleteMarker:
     def __init__(self, key: ObjectKey, version_id: ObjectVersionId):
         self.key = key
         self.version_id = version_id
-        self.last_modified = datetime.now(tz=gmt_zone_info)
+        self.last_modified = datetime.now(tz=_gmt_zone_info)
         self.is_current = True
 
     @staticmethod
@@ -390,7 +391,7 @@ def __init__(
         checksum_algorithm: Optional[ChecksumAlgorithm] = None,
         checksum_value: Optional[str] = None,
     ):
-        self.last_modified = datetime.now(tz=gmt_zone_info)
+        self.last_modified = datetime.now(tz=_gmt_zone_info)
         self.part_number = part_number
         self.size = size
         self.etag = etag
@@ -430,7 +431,7 @@ def __init__(
         tagging: Optional[dict[str, str]] = None,
     ):
         self.id = token_urlsafe(96)  # MultipartUploadId is 128 characters long
-        self.initiated = datetime.now(tz=gmt_zone_info)
+        self.initiated = datetime.now(tz=_gmt_zone_info)
         self.parts = {}
         self.initiator = initiator
         self.tagging = tagging
@@ -488,8 +489,9 @@ def complete_multipart(self, parts: CompletedPartList):
                 )
 
             object_etag.update(bytes.fromhex(s3_part.etag))
-            # TODO verify this, it seems wrong
-            self.object.parts.append((pos, s3_part.size))
+            # keep track of the parts size, as it can be queried afterward on the object as a Range
+            self.object.parts[part_number] = (pos, s3_part.size)
+            pos += s3_part.size
 
             multipart_etag = f"{object_etag.hexdigest()}-{len(parts)}"
             self.object.etag = multipart_etag