|
| 1 | +from typing import Any, Optional |
| 2 | + |
| 3 | +import structlog |
| 4 | +from django.conf import settings |
| 5 | +from django.db.models.signals import pre_save |
| 6 | +from django.dispatch import receiver |
| 7 | + |
| 8 | +from api.managers.dvc_manager import DVCManager |
| 9 | +from api.models.Dataset import Dataset |
| 10 | +from api.models.Resource import Resource, ResourceVersion, _increment_version |
| 11 | +from api.utils.enums import DatasetStatus |
| 12 | + |
| 13 | +logger = structlog.getLogger(__name__) |
| 14 | + |
| 15 | + |
| 16 | +@receiver(pre_save, sender=Dataset) |
| 17 | +def handle_dataset_publication(sender: Any, instance: Dataset, **kwargs: Any) -> None: |
| 18 | + """ |
| 19 | + Signal handler to increment resource versions when a dataset is published. |
| 20 | +
|
| 21 | + This will trigger a minor version increment for all resources in the dataset |
| 22 | + when the dataset status changes from DRAFT to PUBLISHED. |
| 23 | + """ |
| 24 | + try: |
| 25 | + # Check if this is an existing dataset (not a new one) |
| 26 | + if instance.pk: |
| 27 | + # Get the original dataset from the database |
| 28 | + original = Dataset.objects.get(pk=instance.pk) |
| 29 | + |
| 30 | + # Check if status is changing to PUBLISHED |
| 31 | + if ( |
| 32 | + original.status != DatasetStatus.PUBLISHED |
| 33 | + and instance.status == DatasetStatus.PUBLISHED |
| 34 | + ): |
| 35 | + logger.info( |
| 36 | + f"Dataset {instance.title} is being published, incrementing resource versions" |
| 37 | + ) |
| 38 | + |
| 39 | + # Get all resources for this dataset |
| 40 | + resources = instance.resources.all() |
| 41 | + |
| 42 | + # Initialize DVC manager |
| 43 | + dvc = DVCManager(settings.DVC_REPO_PATH) |
| 44 | + |
| 45 | + # Increment version for each resource |
| 46 | + for resource in resources: # type: Resource |
| 47 | + try: |
| 48 | + # Get the latest version |
| 49 | + last_version: Optional[ResourceVersion] = ( |
| 50 | + resource.versions.order_by("-created_at").first() |
| 51 | + ) |
| 52 | + |
| 53 | + if last_version: |
| 54 | + # Increment minor version (publication is a significant change) |
| 55 | + new_version = _increment_version( |
| 56 | + last_version.version_number, increment_type="minor" |
| 57 | + ) |
| 58 | + |
| 59 | + # Get the resource file path |
| 60 | + file_path: str = resource.resourcefiledetails.file.path |
| 61 | + |
| 62 | + # Track with DVC |
| 63 | + dvc_file = dvc.track_resource(file_path) |
| 64 | + message: str = ( |
| 65 | + f"Publishing resource: {resource.name} to version {new_version}" |
| 66 | + ) |
| 67 | + dvc.commit_version(dvc_file, message) |
| 68 | + dvc.tag_version(f"{resource.name}-{new_version}") |
| 69 | + |
| 70 | + # Create version record |
| 71 | + ResourceVersion.objects.create( |
| 72 | + resource=resource, |
| 73 | + version_number=new_version, |
| 74 | + change_description=f"Version created due to dataset publication", |
| 75 | + ) |
| 76 | + |
| 77 | + # Update resource version field |
| 78 | + resource.version = new_version |
| 79 | + resource.save(update_fields=["version"]) |
| 80 | + |
| 81 | + logger.info( |
| 82 | + f"Incremented version for resource {resource.name} to {new_version}" |
| 83 | + ) |
| 84 | + else: |
| 85 | + logger.warning( |
| 86 | + f"No previous version found for resource {resource.name}" |
| 87 | + ) |
| 88 | + except Exception as e: |
| 89 | + logger.error( |
| 90 | + f"Failed to increment version for resource {resource.name}: {str(e)}" |
| 91 | + ) |
| 92 | + except Exception as e: |
| 93 | + logger.error(f"Error in dataset publication signal handler: {str(e)}") |
| 94 | + # Don't raise the exception to avoid blocking the save operation |
0 commit comments