use blob instead of identifier

catalystneuro · Aug 16, 2024 · 788e86a · 788e86a
1 parent 1fb8a09
commit 788e86a
Show file tree

Hide file tree

Showing 7 changed files with 10 additions and 9 deletions.
diff --git a/src/dandi_s3_log_parser/_dandiset_mapper.py b/src/dandi_s3_log_parser/_dandiset_mapper.py
@@ -74,12 +74,13 @@ def _map_reduced_logs_to_dandiset(
 
         all_reduced_logs = []
         for asset in dandiset_version.get_assets():
-            asset_id = asset.identifier.replace("-", "_")
             asset_suffixes = pathlib.Path(asset.path).suffixes
+            is_asset_zarr = ".zarr" in asset_suffixes
 
-            blob_or_zarr = "blobs" if ".zarr" not in asset_suffixes else "zarr"
+            blob_id = asset.identifier if not is_asset_zarr else asset.zarr
+            blobs_or_zarr = "blobs" if not is_asset_zarr else "zarr"
 
-            reduced_log_file_path = reduced_s3_logs_folder_path / f"{blob_or_zarr}_{asset_id}.tsv"
+            reduced_log_file_path = reduced_s3_logs_folder_path / f"{blobs_or_zarr}_{blob_id}.tsv"
 
             if not reduced_log_file_path.exists():
                 continue  # No reduced logs found (possible asset was never accessed); skip to next asset

diff --git a/tests/examples/mapped_to_dandiset_example_0/expected_output/000003/0.210812.1448.tsv b/tests/examples/mapped_to_dandiset_example_0/expected_output/000003/0.210812.1448.tsv
@@ -1,3 +1,3 @@
 	filename	timestamp	bytes_sent	region
-0	5e9e92e1-f044-4aa0-ab47-1cfcb8899348	2022-03-16 02:21:12	512	unknown
-1	5e9e92e1-f044-4aa0-ab47-1cfcb8899348	2022-05-04 05:06:35	512	unknown
+0	sub-YutaMouse20/sub-YutaMouse20_ses-YutaMouse20-140327_behavior+ecephys.nwb	2022-03-16 02:21:12	512	unknown
+1	sub-YutaMouse20/sub-YutaMouse20_ses-YutaMouse20-140327_behavior+ecephys.nwb	2022-05-04 05:06:35	512	unknown
diff --git a/tests/examples/mapped_to_dandiset_example_0/expected_output/000003/0.230629.1955.tsv b/tests/examples/mapped_to_dandiset_example_0/expected_output/000003/0.230629.1955.tsv
@@ -1,3 +1,3 @@
 	filename	timestamp	bytes_sent	region
-0	5e9e92e1-f044-4aa0-ab47-1cfcb8899348	2022-03-16 02:21:12	512	unknown
-1	5e9e92e1-f044-4aa0-ab47-1cfcb8899348	2022-05-04 05:06:35	512	unknown
+0	sub-YutaMouse20/sub-YutaMouse20_ses-YutaMouse20-140327_behavior+ecephys.nwb	2022-03-16 02:21:12	512	unknown
+1	sub-YutaMouse20/sub-YutaMouse20_ses-YutaMouse20-140327_behavior+ecephys.nwb	2022-05-04 05:06:35	512	unknown
diff --git a/tests/examples/mapped_to_dandiset_example_0/expected_output/000003/draft.tsv b/tests/examples/mapped_to_dandiset_example_0/expected_output/000003/draft.tsv
@@ -1,3 +1,3 @@
 	filename	timestamp	bytes_sent	region
-0	5e9e92e1-f044-4aa0-ab47-1cfcb8899348	2022-03-16 02:21:12	512	unknown
-1	5e9e92e1-f044-4aa0-ab47-1cfcb8899348	2022-05-04 05:06:35	512	unknown
+0	sub-YutaMouse20/sub-YutaMouse20_ses-YutaMouse20-140327_behavior+ecephys.nwb	2022-03-16 02:21:12	512	unknown
+1	sub-YutaMouse20/sub-YutaMouse20_ses-YutaMouse20-140327_behavior+ecephys.nwb	2022-05-04 05:06:35	512	unknown
diff --git a/..._5e9e92e1_f044_4aa0_ab47_1cfcb8899348.tsv → ..._5e9e92e1-f044-4aa0-ab47-1cfcb8899348.tsv b/..._5e9e92e1_f044_4aa0_ab47_1cfcb8899348.tsv → ..._5e9e92e1-f044-4aa0-ab47-1cfcb8899348.tsv
diff --git a/..._a7b032b8_1e31-429f_975f_52a28cec6629.tsv → ..._a7b032b8-1e31-429f-975f-52a28cec6629.tsv b/..._a7b032b8_1e31-429f_975f_52a28cec6629.tsv → ..._a7b032b8-1e31-429f-975f-52a28cec6629.tsv
diff --git a/..._cbcf1d6d_7f64_4d1f_8692_75e09e177ca6.tsv → ..._cbcf1d6d-7f64-4d1f-8692-75e09e177ca6.tsv b/..._cbcf1d6d_7f64_4d1f_8692_75e09e177ca6.tsv → ..._cbcf1d6d-7f64-4d1f-8692-75e09e177ca6.tsv