Automatically export workflow traces to tilebox

Author: lukasbindreiter

.gitignore

@@ -17,3 +17,5 @@ coverage.xml
 test-report.xml
 
 *.egg-info/
+
+.amp/

tilebox-datasets/tests/data/datapoint.py

@@ -32,8 +32,6 @@
 from tests.query.pagination import paginations
 from tests.query.time_interval import i64_datetimes
 from tilebox.datasets.data.datapoint import AnyMessage, IngestResponse, QueryResultPage, RepeatedAny
-from tilebox.datasets.datasets.v1 import core_pb2
-from tilebox.datasets.query.time_interval import datetime_to_timestamp
 
 
 @composite
@@ -138,8 +136,7 @@ def example_pandas_datapoints(draw: DrawFn) -> pd.DataFrame:
 @composite
 def anys(draw: DrawFn, generated_fields: bool = False, missing_fields: bool = False) -> AnyMessage:
     """A hypothesis strategy for generating random Any messages"""
-    # we need a random byte string here, but let's actually use a valid protobuf message, in this
-    # case because its easy to generate let's use a DatapointMetadata message
+    # we need a random byte string here, but let's actually use a valid protobuf message
     datapoint = draw(example_datapoints(generated_fields, missing_fields))
     return AnyMessage(example_dataset_type_url(), datapoint.SerializeToString())
 
@@ -158,14 +155,6 @@ def repeated_anys(
     return RepeatedAny(example_dataset_type_url(), [dp.SerializeToString() for dp in datapoints])
 
 
-@composite
-def datapoint_metadata_messages(draw: DrawFn) -> core_pb2.DatapointMetadata:
-    event_time = datetime_to_timestamp(draw(i64_datetimes))
-    ingestion_time = datetime_to_timestamp(draw(i64_datetimes))
-    data_point_id = str(draw(uuids(version=4)))
-    return core_pb2.DatapointMetadata(event_time=event_time, ingestion_time=ingestion_time, id=data_point_id)
-
-
 @composite
 def query_result_pages(
     draw: DrawFn, empty_next_page: bool | None = None, generated_fields: bool = True, missing_fields: bool = False

tilebox-datasets/tilebox/datasets/buf/validate/validate_pb2.py

@@ -388,7 +388,7 @@ class BoolRules(_message.Message):
     def __init__(self, const: bool = ..., example: _Optional[_Iterable[bool]] = ...) -> None: ...
 
 class StringRules(_message.Message):
-    __slots__ = ("const", "len", "min_len", "max_len", "len_bytes", "min_bytes", "max_bytes", "pattern", "prefix", "suffix", "contains", "not_contains", "not_in", "email", "hostname", "ip", "ipv4", "ipv6", "uri", "uri_ref", "address", "uuid", "tuuid", "ip_with_prefixlen", "ipv4_with_prefixlen", "ipv6_with_prefixlen", "ip_prefix", "ipv4_prefix", "ipv6_prefix", "host_and_port", "ulid", "well_known_regex", "strict", "example")
+    __slots__ = ("const", "len", "min_len", "max_len", "len_bytes", "min_bytes", "max_bytes", "pattern", "prefix", "suffix", "contains", "not_contains", "not_in", "email", "hostname", "ip", "ipv4", "ipv6", "uri", "uri_ref", "address", "uuid", "tuuid", "ip_with_prefixlen", "ipv4_with_prefixlen", "ipv6_with_prefixlen", "ip_prefix", "ipv4_prefix", "ipv6_prefix", "host_and_port", "ulid", "protobuf_fqn", "protobuf_dot_fqn", "well_known_regex", "strict", "example")
     Extensions: _python_message._ExtensionDict
     CONST_FIELD_NUMBER: _ClassVar[int]
     LEN_FIELD_NUMBER: _ClassVar[int]
@@ -422,6 +422,8 @@ class StringRules(_message.Message):
     IPV6_PREFIX_FIELD_NUMBER: _ClassVar[int]
     HOST_AND_PORT_FIELD_NUMBER: _ClassVar[int]
     ULID_FIELD_NUMBER: _ClassVar[int]
+    PROTOBUF_FQN_FIELD_NUMBER: _ClassVar[int]
+    PROTOBUF_DOT_FQN_FIELD_NUMBER: _ClassVar[int]
     WELL_KNOWN_REGEX_FIELD_NUMBER: _ClassVar[int]
     STRICT_FIELD_NUMBER: _ClassVar[int]
     EXAMPLE_FIELD_NUMBER: _ClassVar[int]
@@ -456,10 +458,12 @@ class StringRules(_message.Message):
     ipv6_prefix: bool
     host_and_port: bool
     ulid: bool
+    protobuf_fqn: bool
+    protobuf_dot_fqn: bool
     well_known_regex: KnownRegex
     strict: bool
     example: _containers.RepeatedScalarFieldContainer[str]
-    def __init__(self, const: _Optional[str] = ..., len: _Optional[int] = ..., min_len: _Optional[int] = ..., max_len: _Optional[int] = ..., len_bytes: _Optional[int] = ..., min_bytes: _Optional[int] = ..., max_bytes: _Optional[int] = ..., pattern: _Optional[str] = ..., prefix: _Optional[str] = ..., suffix: _Optional[str] = ..., contains: _Optional[str] = ..., not_contains: _Optional[str] = ..., not_in: _Optional[_Iterable[str]] = ..., email: bool = ..., hostname: bool = ..., ip: bool = ..., ipv4: bool = ..., ipv6: bool = ..., uri: bool = ..., uri_ref: bool = ..., address: bool = ..., uuid: bool = ..., tuuid: bool = ..., ip_with_prefixlen: bool = ..., ipv4_with_prefixlen: bool = ..., ipv6_with_prefixlen: bool = ..., ip_prefix: bool = ..., ipv4_prefix: bool = ..., ipv6_prefix: bool = ..., host_and_port: bool = ..., ulid: bool = ..., well_known_regex: _Optional[_Union[KnownRegex, str]] = ..., strict: bool = ..., example: _Optional[_Iterable[str]] = ..., **kwargs) -> None: ...
+    def __init__(self, const: _Optional[str] = ..., len: _Optional[int] = ..., min_len: _Optional[int] = ..., max_len: _Optional[int] = ..., len_bytes: _Optional[int] = ..., min_bytes: _Optional[int] = ..., max_bytes: _Optional[int] = ..., pattern: _Optional[str] = ..., prefix: _Optional[str] = ..., suffix: _Optional[str] = ..., contains: _Optional[str] = ..., not_contains: _Optional[str] = ..., not_in: _Optional[_Iterable[str]] = ..., email: bool = ..., hostname: bool = ..., ip: bool = ..., ipv4: bool = ..., ipv6: bool = ..., uri: bool = ..., uri_ref: bool = ..., address: bool = ..., uuid: bool = ..., tuuid: bool = ..., ip_with_prefixlen: bool = ..., ipv4_with_prefixlen: bool = ..., ipv6_with_prefixlen: bool = ..., ip_prefix: bool = ..., ipv4_prefix: bool = ..., ipv6_prefix: bool = ..., host_and_port: bool = ..., ulid: bool = ..., protobuf_fqn: bool = ..., protobuf_dot_fqn: bool = ..., well_known_regex: _Optional[_Union[KnownRegex, str]] = ..., strict: bool = ..., example: _Optional[_Iterable[str]] = ..., **kwargs) -> None: ...
 
 class BytesRules(_message.Message):
     __slots__ = ("const", "len", "min_len", "max_len", "pattern", "prefix", "suffix", "contains", "not_in", "ip", "ipv4", "ipv6", "uuid", "example")

tilebox-datasets/tilebox/datasets/buf/validate/validate_pb2.pyi

@@ -1,5 +1,4 @@
 from tilebox.datasets.datasets.v1 import dataset_type_pb2 as _dataset_type_pb2
-from google.protobuf import timestamp_pb2 as _timestamp_pb2
 from tilebox.datasets.tilebox.v1 import id_pb2 as _id_pb2
 from tilebox.datasets.tilebox.v1 import query_pb2 as _query_pb2
 from google.protobuf.internal import containers as _containers
@@ -33,14 +32,6 @@ VISIBILITY_PRIVATE: Visibility
 VISIBILITY_SHARED_WITH_ME: Visibility
 VISIBILITY_PUBLIC: Visibility
 
-class LegacyPagination(_message.Message):
-    __slots__ = ("limit", "starting_after")
-    LIMIT_FIELD_NUMBER: _ClassVar[int]
-    STARTING_AFTER_FIELD_NUMBER: _ClassVar[int]
-    limit: int
-    starting_after: str
-    def __init__(self, limit: _Optional[int] = ..., starting_after: _Optional[str] = ...) -> None: ...
-
 class Any(_message.Message):
     __slots__ = ("type_url", "value")
     TYPE_URL_FIELD_NUMBER: _ClassVar[int]
@@ -57,43 +48,13 @@ class RepeatedAny(_message.Message):
     value: _containers.RepeatedScalarFieldContainer[bytes]
     def __init__(self, type_url: _Optional[str] = ..., value: _Optional[_Iterable[bytes]] = ...) -> None: ...
 
-class DatapointMetadata(_message.Message):
-    __slots__ = ("event_time", "ingestion_time", "id")
-    EVENT_TIME_FIELD_NUMBER: _ClassVar[int]
-    INGESTION_TIME_FIELD_NUMBER: _ClassVar[int]
-    ID_FIELD_NUMBER: _ClassVar[int]
-    event_time: _timestamp_pb2.Timestamp
-    ingestion_time: _timestamp_pb2.Timestamp
-    id: str
-    def __init__(self, event_time: _Optional[_Union[_timestamp_pb2.Timestamp, _Mapping]] = ..., ingestion_time: _Optional[_Union[_timestamp_pb2.Timestamp, _Mapping]] = ..., id: _Optional[str] = ...) -> None: ...
-
-class DatapointPage(_message.Message):
-    __slots__ = ("meta", "data", "next_page")
-    META_FIELD_NUMBER: _ClassVar[int]
-    DATA_FIELD_NUMBER: _ClassVar[int]
-    NEXT_PAGE_FIELD_NUMBER: _ClassVar[int]
-    meta: _containers.RepeatedCompositeFieldContainer[DatapointMetadata]
-    data: RepeatedAny
-    next_page: LegacyPagination
-    def __init__(self, meta: _Optional[_Iterable[_Union[DatapointMetadata, _Mapping]]] = ..., data: _Optional[_Union[RepeatedAny, _Mapping]] = ..., next_page: _Optional[_Union[LegacyPagination, _Mapping]] = ...) -> None: ...
-
-class Datapoint(_message.Message):
-    __slots__ = ("meta", "data")
-    META_FIELD_NUMBER: _ClassVar[int]
-    DATA_FIELD_NUMBER: _ClassVar[int]
-    meta: DatapointMetadata
-    data: Any
-    def __init__(self, meta: _Optional[_Union[DatapointMetadata, _Mapping]] = ..., data: _Optional[_Union[Any, _Mapping]] = ...) -> None: ...
-
 class Collection(_message.Message):
-    __slots__ = ("legacy_id", "name", "id")
-    LEGACY_ID_FIELD_NUMBER: _ClassVar[int]
+    __slots__ = ("name", "id")
     NAME_FIELD_NUMBER: _ClassVar[int]
     ID_FIELD_NUMBER: _ClassVar[int]
-    legacy_id: str
     name: str
     id: _id_pb2.ID
-    def __init__(self, legacy_id: _Optional[str] = ..., name: _Optional[str] = ..., id: _Optional[_Union[_id_pb2.ID, _Mapping]] = ...) -> None: ...
+    def __init__(self, name: _Optional[str] = ..., id: _Optional[_Union[_id_pb2.ID, _Mapping]] = ...) -> None: ...
 
 class CollectionInfo(_message.Message):
     __slots__ = ("collection", "availability", "count")