Add management command to re-extract for asset metadata

dandiapi/api/management/commands/extract_metadata.py

@@ -0,0 +1,93 @@
+from __future__ import annotations
+
+import logging
+from pathlib import Path
+
+from dandi.dandiapi import RemoteReadableAsset
+from dandi.metadata.nwb import nwb2asset
+from dandischema.models import get_schema_version
+from django.contrib.auth.models import User
+import djclick as click
+from tqdm import tqdm
+
+from dandiapi.api.models import Asset, Dandiset, Version
+from dandiapi.api.services.asset import change_asset
+
+logger = logging.getLogger(__name__)
+
+
+@click.group()
+def group():
+    pass
+
+
+def extract_asset_metadata(asset: Asset, draft_version: Version):
+    readable_asset = RemoteReadableAsset(
+        asset.s3_url, size=asset.size, mtime=asset.modified, name=Path(asset.path).name
+    )
+
+    if not asset.path.lower().endswith('.nwb'):
+        logger.info('Asset %s: Not an NWB file, skipping...', asset.path)
+        return
+
+    new_metadata = nwb2asset(readable_asset, schema_version=get_schema_version()).json_dict()
+
+    # Use dandiset owner, default to some admin user
+    user = (
+        draft_version.dandiset.owners.first()
+        or User.objects.filter(is_superuser=True, is_staff=True).first()
+    )
+
+    # Replace old asset with new asset containing updated metadata
+    change_asset(
+        user=user,
+        asset=asset,
+        version=draft_version,
+        new_asset_blob=asset.blob,
+        new_zarr_archive=asset.zarr,
+        new_metadata=new_metadata,
+    )
+
+
+def extract_dandiset_assets(dandiset: Dandiset):
+    # Only update assets which do not belong to a published version
+    assets = dandiset.draft_version.assets.filter(
+        published=False, path__iendswith='.nwb'
+    ).select_related('blob', 'zarr')
+    if not assets:
+        logger.info('No draft NWB assets found in dandiset %s. Skipping...', dandiset)
+        return
+
+    for asset in tqdm(assets):
+        extract_asset_metadata(asset=asset, draft_version=dandiset.draft_version)
+
+
+@group.command(help='Re-extracts the metadata of this asset')
+@click.argument('asset_id')
+def asset(asset_id: str):
+    asset = Asset.objects.get(asset_id=asset_id)
+    draft_versions = asset.versions.filter(version='draft')
+    if not draft_versions.exists():
+        raise click.ClickException(
+            'Cannot re-extract metadata of asset that has no associated draft versions.'
+        )
+
+    # Re-extract for every draft version
+    for version in draft_versions:
+        extract_asset_metadata(asset=asset, draft_version=version)
+
+
+@group.command(
+    help='Re-extracts the metadata of all assets in the draft version of the provided dandiset'
+)
+@click.argument('dandiset_id')
+def dandiset(dandiset_id: str):
+    dandiset = Dandiset.objects.get(id=int(dandiset_id))
+    extract_dandiset_assets(dandiset)
+
+
+@group.command(name='all', help='Re-extracts the metadata of all assets in all draft versions')
+def all_dandisets():
+    for dandiset in Dandiset.objects.all():
+        logger.info('DANDISET: %s', dandiset.identifier)
+        extract_dandiset_assets(dandiset)

setup.py

@@ -40,6 +40,7 @@
     include_package_data=True,
     install_requires=[
         'celery',
+        'dandi',
         # Pin dandischema to exact version to make explicit which schema version is being used
         'dandischema==0.10.2',  # schema version 0.6.8
         'django~=4.1.0',
@@ -59,6 +60,7 @@
         'djangorestframework-yaml',
         'drf-extensions',
         'drf-yasg',
+        'fsspec[http]',
         'jsonschema',
         'boto3[s3]',
         'more_itertools',