feat(kernel): preprocessing recipe domain and driver port

docs/content/api.md

@@ -607,13 +607,13 @@ argument for branching on `code`.
 | Status | Codes |
 | --- | --- |
 | **401** | `UNAUTHORIZED` — with a `WWW-Authenticate: Bearer` challenge |
-| **404** | `PROJECT_NOT_FOUND` · `SCHEMA_NOT_FOUND` · `SCHEMA_DRAFT_NOT_FOUND` · `BATCH_NOT_FOUND` · `JOB_NOT_FOUND` · `INGEST_JOB_NOT_FOUND` · `BACKGROUND_JOB_NOT_FOUND` · `ASSET_NOT_FOUND` · `SOURCE_NOT_FOUND` · `DATASET_NOT_FOUND` · `ANNOTATION_NOT_FOUND` · `RELEASE_NOT_FOUND` · `TOKEN_NOT_FOUND` · `INFERENCE_CONNECTION_NOT_FOUND` · `ASSET_NOT_IN_JOB` · `ASSET_NOT_IN_DATASET` · `NO_SPLIT_RECIPE` · `EXPORT_FORMAT_NOT_FOUND` · `THUMBNAIL_NOT_CACHED` · `NOT_FOUND` (no such route) |
+| **404** | `PROJECT_NOT_FOUND` · `SCHEMA_NOT_FOUND` · `SCHEMA_DRAFT_NOT_FOUND` · `BATCH_NOT_FOUND` · `JOB_NOT_FOUND` · `INGEST_JOB_NOT_FOUND` · `BACKGROUND_JOB_NOT_FOUND` · `ASSET_NOT_FOUND` · `SOURCE_NOT_FOUND` · `DATASET_NOT_FOUND` · `ANNOTATION_NOT_FOUND` · `RELEASE_NOT_FOUND` · `TOKEN_NOT_FOUND` · `INFERENCE_CONNECTION_NOT_FOUND` · `ASSET_NOT_IN_JOB` · `ASSET_NOT_IN_DATASET` · `NO_SPLIT_RECIPE` · `EXPORT_FORMAT_NOT_FOUND` · `EXPORT_TARGET_NOT_FOUND` · `THUMBNAIL_NOT_CACHED` · `NOT_FOUND` (no such route) |
 | **405** | `METHOD_NOT_ALLOWED` |
-| **409** | `PROJECT_NAME_TAKEN` · `RELEASE_TAG_TAKEN` · `TOKEN_NAME_TAKEN` · `INFERENCE_CONNECTION_NAME_TAKEN` · `WORKSPACE_ALREADY_EXISTS` · `WORKSPACE_NOT_EMPTY` · `SCHEMA_VERSION_CONFLICT` · `INVALID_TRANSITION` · `STALE_WRITE` · `BATCH_NOT_EDITABLE` · `BATCH_IMMUTABLE` · `BATCH_NOT_IN_ANNOTATION` · `ASSET_NOT_WRITABLE` · `JOB_FINISHED` · `BATCH_NOT_COMPLETE` · `JOB_NOT_COMPLETE` · `EMPTY_BATCH` · `EMPTY_RELEASE` · `RELEASE_CONTENT_WOULD_VIOLATE_SCHEMA` · `CONFIRMATION_REQUIRED` · `DESTRUCTIVE_SCHEMA_CHANGE` · `SCHEMA_CHANGE_WOULD_ORPHAN` · `SCHEMA_HAS_NO_DETECTABLE_CLASS` · `UNSERIALIZABLE_MANIFEST` · `LOSSY_EXPORT_NOT_CONSENTED` · `EXPORT_SOURCE_UNREADABLE` · `INFERENCE_CONNECTION_NOT_DOWNLOADABLE` · `INFERENCE_CONNECTION_NOT_CHECKABLE` · `INFERENCE_CONNECTION_NOT_TESTABLE` · `INFERENCE_CONNECTION_MODEL_FIXED` · `WEIGHTS_DAMAGED` · `INFERENCE_CONNECTION_NOT_SET_UP` |
+| **409** | `PROJECT_NAME_TAKEN` · `RELEASE_TAG_TAKEN` · `TOKEN_NAME_TAKEN` · `INFERENCE_CONNECTION_NAME_TAKEN` · `WORKSPACE_ALREADY_EXISTS` · `WORKSPACE_NOT_EMPTY` · `SCHEMA_VERSION_CONFLICT` · `INVALID_TRANSITION` · `STALE_WRITE` · `BATCH_NOT_EDITABLE` · `BATCH_IMMUTABLE` · `BATCH_NOT_IN_ANNOTATION` · `ASSET_NOT_WRITABLE` · `JOB_FINISHED` · `BATCH_NOT_COMPLETE` · `JOB_NOT_COMPLETE` · `EMPTY_BATCH` · `EMPTY_RELEASE` · `RELEASE_CONTENT_WOULD_VIOLATE_SCHEMA` · `CONFIRMATION_REQUIRED` · `DESTRUCTIVE_SCHEMA_CHANGE` · `SCHEMA_CHANGE_WOULD_ORPHAN` · `SCHEMA_HAS_NO_DETECTABLE_CLASS` · `UNSERIALIZABLE_MANIFEST` · `LOSSY_EXPORT_NOT_CONSENTED` · `EXPORT_SOURCE_UNREADABLE` · `INFERENCE_CONNECTION_NOT_DOWNLOADABLE` · `INFERENCE_CONNECTION_NOT_CHECKABLE` · `INFERENCE_CONNECTION_NOT_TESTABLE` · `INFERENCE_CONNECTION_MODEL_FIXED` · `WEIGHTS_DAMAGED` · `INFERENCE_CONNECTION_NOT_SET_UP` · `AUGMENTATION_REQUIRES_SPLIT` · `PREPROCESSING_STEP_UNSUPPORTED_GEOMETRY` |
 | **422** | `VALIDATION_ERROR` · `ASSET_NOT_IN_BATCH` · `ANNOTATION_NOT_FROM_MODEL` · `INVALID_NAME` · `INFERENCE_CONNECTION_INVALID` · `INVALID_SCHEMA` · `UNSUPPORTED_GEOMETRY` · `INVALID_ANNOTATION` · `LABEL_CLASS_NOT_IN_SCHEMA` · `DISALLOWED_GEOMETRY` · `ANNOTATION_GEOMETRY_OUT_OF_BOUNDS` · `DUPLICATE_CLASSIFICATION_TAG` · `MISSING_REQUIRED_ATTRIBUTE` · `UNKNOWN_ATTRIBUTE` · `INVALID_ATTRIBUTE_VALUE` · `INVALID_PARTITION` · `UNKNOWN_JOB_TYPE` · `MEDIA_ERROR` · `UNSUPPORTED_MEDIA` · `CORRUPT_MEDIA` · `UNSUPPORTED_PROMPT` · `PROMPT_POINT_OUT_OF_BOUNDS` · `GEOMETRY_NOT_PRODUCED` |
 | **502** | `INFERENCE_ENDPOINT_UNAVAILABLE` |
 | **503** | `WORKSPACE_BUSY` |
-| **500** | `WORKSPACE_CORRUPT` · `NOT_A_WORKSPACE` · `WORKSPACE_FORMAT_TOO_NEW` · `WORKSPACE_SCHEMA_MISMATCH` · `ENTITY_NOT_FOUND` · `ENTITY_ALREADY_EXISTS` · `CONSTRAINT_VIOLATED` · `MEDIA_TOOL_UNAVAILABLE` · `LOCAL_INFERENCE_UNAVAILABLE` · `INFERENCE_CONNECTION_NOT_RUNNABLE` · `INFERENCE_OUT_OF_MEMORY` · `INTERNAL_ERROR` |
+| **500** | `WORKSPACE_CORRUPT` · `NOT_A_WORKSPACE` · `WORKSPACE_FORMAT_TOO_NEW` · `WORKSPACE_SCHEMA_MISMATCH` · `ENTITY_NOT_FOUND` · `ENTITY_ALREADY_EXISTS` · `CONSTRAINT_VIOLATED` · `MEDIA_TOOL_UNAVAILABLE` · `LOCAL_INFERENCE_UNAVAILABLE` · `INFERENCE_CONNECTION_NOT_RUNNABLE` · `INFERENCE_OUT_OF_MEMORY` · `EXPORT_TARGET_CONFLICT` · `INVALID_EXPORT_TARGET` · `INTERNAL_ERROR` |
 
 Every row but `VALIDATION_ERROR`, `NOT_FOUND`, `METHOD_NOT_ALLOWED`, `UNAUTHORIZED` and
 `INTERNAL_ERROR` — the five the framework and the auth guard raise — comes from `ERROR_RULES`

src/visionset/formats/_dummy.py

@@ -7,6 +7,7 @@
 
 from pathlib import Path
 
+from visionset.formats._targets import self_target
 from visionset.kernel.domain import GeometryType, Manifest, Release
 from visionset.kernel.ports import ContentReader
 
@@ -40,6 +41,8 @@ class DummyExporter:
     #: that named it would have to grow when the domain does.
     supported_modalities = frozenset({"image", "video", "point_cloud"})
 
+    targets = self_target(format_name, supported_geometries)
+
     def export(
         self,
         release: Release,

src/visionset/formats/_targets.py

@@ -0,0 +1,43 @@
+# usage: from visionset.formats._targets import self_target
+"""The one-target declaration every non-YOLO exporter shares.
+
+An exporter that is not a trainer's format still declares exactly one target,
+named after itself, family ``other``, with no trainer tasks — so a surface
+renders one control for every export rather than a target select beside a
+format select. Spelled once here for the same reason ``_layout`` exists:
+the day two spellings of the rule disagree, the catalog and the format list
+stop describing the same thing.
+
+Private to :mod:`visionset.formats`, like ``_layout``: importable, but not part
+of the ``Exporter`` contract.
+"""
+
+from __future__ import annotations
+
+from visionset.kernel.domain import (
+    ExportTarget,
+    GeometryType,
+    PreprocessingHints,
+    TargetFamily,
+)
+
+
+def self_target(format_name: str, geometries: frozenset[GeometryType]) -> frozenset[ExportTarget]:
+    """The whole ``targets`` declaration for a format that is its own target."""
+    return frozenset(
+        {
+            ExportTarget(
+                name=format_name,
+                label=format_name,
+                family=TargetFamily.OTHER,
+                tasks=frozenset(),
+                supported_geometries=geometries,
+                hints=PreprocessingHints(
+                    recommended_size=None,
+                    recommended_strategy=None,
+                    trainer_resizes=True,
+                    augmentation_common=False,
+                ),
+            )
+        }
+    )

src/visionset/formats/classification/__init__.py

@@ -58,6 +58,7 @@ class index on the *from-the-schema-not-the-data* half of its rule — deriving
 from typing import Final
 
 from visionset.formats._layout import IMAGES_DIRNAME, folds_of, write_image
+from visionset.formats._targets import self_target
 from visionset.kernel.domain import (
     ClassificationGeometry,
     GeometryType,
@@ -107,6 +108,8 @@ class ClassificationExporter:
     #: A classification dataset is a directory of pictures.
     supported_modalities = frozenset({"image"})
 
+    targets = self_target(format_name, supported_geometries)
+
     def export(
         self,
         release: Release,

src/visionset/formats/coco/__init__.py

@@ -54,6 +54,7 @@
     folds_of,
     write_image,
 )
+from visionset.formats._targets import self_target
 from visionset.kernel.domain import (
     BboxGeometry,
     GeometryType,
@@ -115,6 +116,8 @@ class CocoExporter:
 
     supported_modalities = frozenset({"image"})
 
+    targets = self_target(format_name, supported_geometries)
+
     def export(
         self,
         release: Release,

src/visionset/formats/lanes/__init__.py

@@ -52,6 +52,7 @@
     folds_of,
     write_image,
 )
+from visionset.formats._targets import self_target
 from visionset.formats.lanes._core import (
     BDD100K_CATEGORIES,
     CULANE_SLOTS,
@@ -106,6 +107,11 @@ class TuSimpleExporter:
 
     supported_modalities = frozenset({"image"})
 
+    #: The one self-named target whose geometries are not the supported set:
+    #: that set is empty here, everything arriving degraded, and a target must
+    #: carry at least one geometry — so it names the polyline this format writes.
+    targets = self_target(format_name, frozenset({GeometryType.POLYLINE}))
+
     def export(
         self,
         release: Release,
@@ -138,6 +144,8 @@ class CurveLanesExporter:
     degraded_geometries: frozenset[GeometryType] = frozenset()
     supported_modalities = frozenset({"image"})
 
+    targets = self_target(format_name, supported_geometries)
+
     def export(
         self,
         release: Release,
@@ -164,6 +172,8 @@ class Bdd100kLaneExporter:
     degraded_geometries: frozenset[GeometryType] = frozenset()
     supported_modalities = frozenset({"image"})
 
+    targets = self_target(format_name, supported_geometries)
+
     def export(
         self,
         release: Release,
@@ -196,6 +206,8 @@ class CuLaneExporter:
     degraded_geometries: frozenset[GeometryType] = frozenset()
     supported_modalities = frozenset({"image"})
 
+    targets = self_target(format_name, supported_geometries)
+
     def export(
         self,
         release: Release,
@@ -229,6 +241,8 @@ class OpenLane2dExporter:
     degraded_geometries: frozenset[GeometryType] = frozenset()
     supported_modalities = frozenset({"image"})
 
+    targets = self_target(format_name, supported_geometries)
+
     def export(
         self,
         release: Release,

src/visionset/formats/voc/__init__.py

@@ -45,6 +45,7 @@
     folds_of,
     write_image,
 )
+from visionset.formats._targets import self_target
 from visionset.kernel.domain import (
     BboxGeometry,
     GeometryType,
@@ -101,6 +102,8 @@ class VocExporter:
 
     supported_modalities = frozenset({"image"})
 
+    targets = self_target(format_name, supported_geometries)
+
     def export(
         self,
         release: Release,

src/visionset/formats/yolo/__init__.py

@@ -62,6 +62,7 @@
     folds_of,
     write_image,
 )
+from visionset.formats._targets import self_target
 from visionset.kernel.domain import (
     BboxGeometry,
     Geometry,
@@ -118,6 +119,8 @@ class YoloDetectionExporter:
     #: A YOLO dataset is a directory of pictures.
     supported_modalities = frozenset({"image"})
 
+    targets = self_target(format_name, supported_geometries)
+
     def export(
         self,
         release: Release,

src/visionset/kernel/__init__.py

@@ -15,6 +15,7 @@
     AssetNotInDataset,
     AssetNotInJob,
     AssetNotWritable,
+    AugmentationRequiresSplit,
     BackgroundJobNotFound,
     BatchImmutable,
     BatchNotComplete,
@@ -34,6 +35,8 @@
     EntityNotFound,
     ExportFormatNotFound,
     ExportSourceUnreadable,
+    ExportTargetConflict,
+    ExportTargetNotFound,
     GeometryNotProduced,
     InferenceConnectionInvalid,
     InferenceConnectionModelFixed,
@@ -49,6 +52,7 @@
     IngestJobNotFound,
     InvalidAnnotation,
     InvalidAttributeValue,
+    InvalidExportTarget,
     InvalidName,
     InvalidPartition,
     InvalidSchema,
@@ -64,6 +68,7 @@
     MissingRequiredAttribute,
     NoSplitRecipe,
     NotAWorkspace,
+    PreprocessingStepUnsupportedGeometry,
     ProjectNameTaken,
     ProjectNotFound,
     PromptPointOutOfBounds,
@@ -105,6 +110,7 @@
     "AssetNotInDataset",
     "AssetNotInJob",
     "AssetNotWritable",
+    "AugmentationRequiresSplit",
     "BackgroundJobNotFound",
     "BatchImmutable",
     "BatchNotComplete",
@@ -124,10 +130,13 @@
     "EntityNotFound",
     "ExportFormatNotFound",
     "ExportSourceUnreadable",
+    "ExportTargetConflict",
+    "ExportTargetNotFound",
     "GeometryNotProduced",
     "IngestJobNotFound",
     "InvalidAnnotation",
     "InvalidAttributeValue",
+    "InvalidExportTarget",
     "InvalidName",
     "InvalidPartition",
     "InvalidSchema",
@@ -143,6 +152,7 @@
     "MissingRequiredAttribute",
     "NoSplitRecipe",
     "NotAWorkspace",
+    "PreprocessingStepUnsupportedGeometry",
     "ProjectNameTaken",
     "ProjectNotFound",
     "PromptPointOutOfBounds",

src/visionset/kernel/domain/__init__.py

@@ -66,6 +66,14 @@
     IngestCompleted,
     ReleasePublished,
 )
+from visionset.kernel.domain.export_target import (
+    TARGET_NAME_PATTERN,
+    ExportTarget,
+    PreprocessingHints,
+    ResizeStrategy,
+    TargetFamily,
+    Task,
+)
 from visionset.kernel.domain.geometry import (
     IMPLEMENTED_GEOMETRIES,
     BboxGeometry,
@@ -162,6 +170,28 @@
     TextPrompt,
     require_points_on_asset,
 )
+from visionset.kernel.domain.preprocessing import (
+    AugmentOp,
+    AugmentStep,
+    PreprocessingRecipe,
+    RecipeSpec,
+    ResizeStep,
+    Step,
+    brightness_contrast_factors,
+    hflip_applied,
+    recipe_hash,
+    rot90_quarter_turns,
+    variant_seed,
+)
+from visionset.kernel.domain.preprocessing_transform import (
+    Fold,
+    LetterboxFit,
+    TransformedAnnotation,
+    TransformedFile,
+    TransformedView,
+    letterbox_fit,
+    transform_manifest,
+)
 from visionset.kernel.domain.project import Project, ProjectPreview, ProjectStats
 from visionset.kernel.domain.release import (
     MANIFEST_VERSION,
@@ -379,6 +409,30 @@
     "ClassExportStatus",
     "ExportCompatibility",
     "ExportResult",
+    "ExportTarget",
+    "PreprocessingHints",
+    "ResizeStrategy",
+    "TargetFamily",
+    "Task",
+    "TARGET_NAME_PATTERN",
+    "AugmentOp",
+    "AugmentStep",
+    "Fold",
+    "LetterboxFit",
+    "PreprocessingRecipe",
+    "RecipeSpec",
+    "ResizeStep",
+    "Step",
+    "TransformedAnnotation",
+    "TransformedFile",
+    "TransformedView",
+    "brightness_contrast_factors",
+    "hflip_applied",
+    "letterbox_fit",
+    "recipe_hash",
+    "rot90_quarter_turns",
+    "transform_manifest",
+    "variant_seed",
     "Geometry",
     "GeometryType",
     "ImageFormat",