added infrastructure code

data_store/repositories/__init__.py

@@ -0,0 +1,14 @@
+"""
+Repositories package for visual-semiotic-ai-analysis.
+Exposes repository implementations for use throughout the project.
+"""
+
+from .annotation_repository import AnnotationRepository
+from .image_repository import ImageRepository
+from .job_repository import JobRepository
+
+__all__ = [
+    "AnnotationRepository",
+    "ImageRepository",
+    "JobRepository",
+]

data_store/repositories/annotation_repository.py

@@ -0,0 +1,106 @@
+"""Definition of AnnotationRepository."""
+
+from __future__ import annotations
+from uuid import UUID
+import json
+from pathlib import Path
+from io import BytesIO
+
+from python_repositories.adapters import MinioAdapter
+
+from data_store.interfaces import AnnotationRepositoryInterface
+from data_store.dto import (
+    Annotation,
+    DistanceEnum,
+    AngleEnum,
+    PointOfViewEnum,
+    ContactEnum,
+)
+
+
+class AnnotationRepository(MinioAdapter, AnnotationRepositoryInterface):
+    """AnnotationRepository implementation using MinIO as the backend."""
+
+    encoding: str = "utf-8"
+    data_format: str = "json"
+    folder_name: str = "annotations"
+
+    @classmethod
+    def _object_name(cls, job_id: UUID) -> str:
+        """Generate the object name for a given job ID."""
+        # Check input
+        if not isinstance(job_id, UUID):
+            raise ValueError("job_id must be a valid UUID.")
+        # Return object name
+        return f"{cls.folder_name}/{job_id}.{cls.data_format}"
+
+    def __enter__(self) -> AnnotationRepository:
+        """Enter the runtime context related to this object."""
+        super().__enter__()
+        return self
+
+    def get(self, job_id: UUID) -> Annotation | None:
+        """Retrieve an Annotation by its job ID."""
+        # Check input
+        if not isinstance(job_id, UUID):
+            raise ValueError("job_id must be a valid UUID.")
+        # Get object
+        object_name = self._object_name(job_id)
+        buffer = self._get(object_name)
+        if buffer is None:
+            return None
+        # Convert buffer to Annotation
+        try:
+            data_str = buffer.read().decode(self.encoding)
+            data = json.loads(data_str)
+            annotation = Annotation(
+                distance=DistanceEnum(data["distance"]),
+                angle=AngleEnum(data["angle"]),
+                point_of_view=PointOfViewEnum(data["point_of_view"]),
+                contact=ContactEnum(data["contact"]),
+            )
+        except (KeyError, ValueError) as e:
+            self.logger.error(
+                "Failed to parse Annotation from data",
+                object_name=object_name,
+                error=str(e),
+            )
+            return None
+        return annotation
+
+    def put(self, annotation: Annotation, job_id: UUID) -> None:
+        """Update or create an Annotation."""
+        # Check inputs
+        if not isinstance(annotation, Annotation):
+            raise ValueError("annotation must be an Annotation instance.")
+        if not isinstance(job_id, UUID):
+            raise ValueError("job_id must be a valid UUID.")
+        # Convert Annotation to json
+        data_json = annotation.model_dump(mode="json")
+        data_bytes = json.dumps(data_json).encode(self.encoding)
+        buffer = BytesIO(data_bytes)
+        # Put object
+        object_name = self._object_name(job_id)
+        self._put(object_name, buffer)
+
+    def delete(self, job_id: UUID) -> None:
+        """Delete an Annotation by its job ID."""
+        # Check input
+        if not isinstance(job_id, UUID):
+            raise ValueError("job_id must be a valid UUID.")
+        # Delete object
+        object_name = self._object_name(job_id)
+        self._delete(object_name)
+
+    def list_all(self) -> list[UUID]:
+        """List all job IDs in the repository."""
+        # List objects in the bucket
+        objects = self._list_objects(prefix=self.folder_name + "/")
+        # Stop early if no objects
+        if len(objects) == 0:
+            return []
+        # Remove folder prefix and file extension
+        object_stems = [Path(obj).stem for obj in objects]
+        # Convert to list of UUIDs
+        job_ids = [UUID(name) for name in object_stems]
+        return job_ids

data_store/repositories/image_repository.py

@@ -0,0 +1,82 @@
+"""Definition of ImageRepository class."""
+
+from __future__ import annotations
+from pathlib import Path
+from io import BytesIO
+from uuid import UUID
+from PIL import Image
+
+from python_repositories.adapters import MinioAdapter
+
+from data_store.interfaces import ImageRepositoryInterface
+
+
+class ImageRepository(MinioAdapter, ImageRepositoryInterface):
+    """ImageRepository implementation using MinIO as the backend."""
+
+    image_format: str = "PNG"
+    folder_name: str = "images"
+
+    @classmethod
+    def _object_name(cls, image_id: UUID) -> str:
+        """Generate the object name for a given image ID."""
+        # Check input
+        if not isinstance(image_id, UUID):
+            raise ValueError("image_id must be a valid UUID.")
+        # Return object name
+        return f"{cls.folder_name}/{image_id}.{cls.image_format.lower()}"
+
+    def __enter__(self) -> ImageRepository:
+        """Enter the runtime context related to this object."""
+        super().__enter__()
+        return self
+
+    def get(self, image_id: UUID) -> Image.Image | None:
+        """Retrieve an Image by its ID."""
+        # Check input
+        if not isinstance(image_id, UUID):
+            raise ValueError("image_id must be a valid UUID.")
+        # Get object
+        object_name = self._object_name(image_id)
+        buffer = self._get(object_name)
+        if buffer is None:
+            return None
+        # Convert bytes back to PIL Image
+        image: Image.Image = Image.open(buffer, formats=[self.image_format])
+        return image
+
+    def put(self, image: Image.Image, image_id: UUID) -> None:
+        """Update or create an Image."""
+        # Check inputs
+        if not isinstance(image, Image.Image):
+            raise ValueError("image must be a PIL Image instance.")
+        if not isinstance(image_id, UUID):
+            raise ValueError("image_id must be a valid UUID.")
+        # Convert image to bytes
+        buffer = BytesIO()
+        image.save(buffer, self.image_format)
+        # Put object
+        object_name = self._object_name(image_id)
+        self._put(object_name, buffer)
+
+    def delete(self, image_id: UUID) -> None:
+        """Delete an Image by its ID."""
+        # Check input
+        if not isinstance(image_id, UUID):
+            raise ValueError("image_id must be a valid UUID.")
+        # Delete object
+        object_name = self._object_name(image_id)
+        self._delete(object_name)
+
+    def list_all(self) -> list[UUID]:
+        """List all image IDs in the repository."""
+        # List objects in the bucket
+        objects = self._list_objects(prefix=self.folder_name + "/")
+        # Stop early if no objects
+        if len(objects) == 0:
+            return []
+        # Remove folder prefix and file extension
+        object_stems = [Path(obj).stem for obj in objects]
+        # Convert to list of UUIDs
+        image_ids = [UUID(name) for name in object_stems]
+        return image_ids

data_store/repositories/job_repository.py

@@ -0,0 +1,78 @@
+"""Definition of JobRepository class."""
+
+from __future__ import annotations
+from uuid import UUID
+from python_repositories.adapters import MinioAdapter
+from data_store.interfaces import JobRepositoryInterface
+from data_store.repositories import AnnotationRepository, ImageRepository
+from data_store.dto import Job
+
+
+class JobRepository(MinioAdapter, JobRepositoryInterface):
+    """JobRepository implementation using MinIO as the backend."""
+
+    def __enter__(self) -> JobRepository:
+        """Enter the runtime context related to this object."""
+        super().__enter__()
+        return self
+
+    def get(self, job_id: UUID) -> Job | None:
+        """Retrieve a Job by its ID."""
+        # Check input
+        if not isinstance(job_id, UUID):
+            raise ValueError("job_id must be a valid UUID.")
+        # Get image
+        with ImageRepository() as image_repo:
+            image = image_repo.get(job_id)
+        if image is None:
+            self.logger.warning(f"Image for job ID {job_id} not found.")
+            return None
+        # Get annotation
+        with AnnotationRepository() as annotation_repo:
+            annotation = annotation_repo.get(job_id)
+        if annotation is None:
+            self.logger.warning(f"Annotation for job ID {job_id} not found.")
+            return None
+        # Create Job DTO
+        job = Job(
+            job_id=job_id,
+            image=image,
+            annotation=annotation,
+        )
+        return job
+
+    def put(self, job: Job) -> None:
+        """Update or create a Job."""
+        # Check input
+        if not isinstance(job, Job):
+            raise ValueError("job must be a Job instance.")
+        # Store image
+        with ImageRepository() as image_repo:
+            image_repo.put(job.image, job.job_id)
+        # Store annotation
+        with AnnotationRepository() as annotation_repo:
+            annotation_repo.put(job.annotation, job.job_id)
+
+    def delete(self, job_id: UUID) -> None:
+        """Delete a Job by its ID."""
+        # Check input
+        if not isinstance(job_id, UUID):
+            raise ValueError("job_id must be a valid UUID.")
+        # Delete image
+        with ImageRepository() as image_repo:
+            image_repo.delete(job_id)
+        # Delete annotation
+        with AnnotationRepository() as annotation_repo:
+            annotation_repo.delete(job_id)
+
+    def list_all(self) -> list[UUID]:
+        """List all Job IDs."""
+        # List image IDs
+        with ImageRepository() as image_repo:
+            image_ids = set(image_repo.list_all())
+        # List annotation IDs
+        with AnnotationRepository() as annotation_repo:
+            annotation_ids = set(annotation_repo.list_all())
+        # Find intersection of IDs
+        job_ids = list(image_ids.intersection(annotation_ids))
+        return job_ids