Empty archive deps entry for PARQUET tables

audeering · Jun 21, 2024 · d0279f7 · d0279f7
1 parent 86d2c37
commit d0279f7
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 8 deletions.
diff --git a/audb/core/dependencies.py b/audb/core/dependencies.py
@@ -482,19 +482,21 @@ def _add_meta(
         self,
         file: str,
         version: str,
-        archive: str,
         checksum: str,
     ):
         r"""Add or update table file.
 
         Args:
             file: relative file path
-            archive: archive name without extension
             checksum: checksum of file
             version: version string
 
         """
         format = audeer.file_extension(file).lower()
+        if format == "parquet":
+            archive = ""
+        else:
+            archive = os.path.splitext(file[3:])[0]
 
         self._df.loc[file] = [
             archive,  # archive

diff --git a/audb/core/publish.py b/audb/core/publish.py
@@ -281,7 +281,7 @@ def _find_tables(
     ):
         checksum = utils.md5(os.path.join(db_root, file))
         if file not in deps or checksum != deps.checksum(file):
-            deps._add_meta(file, version, table, checksum)
+            deps._add_meta(file, version, checksum)
             tables.append(table)
 
     return tables

diff --git a/tests/test_dependencies.py b/tests/test_dependencies.py
@@ -532,16 +532,17 @@ def test_add_media(deps, values):
 
 
 @pytest.mark.parametrize(
-    "file, version, archive, checksum",
+    "file, version, checksum, expected_archive",
     [
-        ("db.table1.csv", "2.1.0", "table1", "asddfnfpork45rgfl"),
+        ("db.table1.csv", "2.1.0", "asddfnfpork45rgfl", "table1"),
+        ("db.table1.parquet", "2.1.0", "asddfnfpork45rgfl", ""),
     ],
 )
-def test_add_meta(deps, file, version, archive, checksum):
-    deps._add_meta(file, version, archive, checksum)
+def test_add_meta(deps, file, version, checksum, expected_archive):
+    deps._add_meta(file, version, checksum)
     assert len(deps) == len(ROWS) + 1
     assert deps.version(file) == version
-    assert deps.archive(file) == archive
+    assert deps.archive(file) == expected_archive
     assert deps.checksum(file) == checksum
 
 

diff --git a/tests/test_publish.py b/tests/test_publish.py
@@ -1162,6 +1162,10 @@ def expected_table_checksum(path: str) -> str:
     assert deps.checksum(table_file) == expected_table_checksum(
         audeer.path(build_dir, table_file)
     )
+    if storage_format == "csv":
+        assert deps.archive(table_file) == table
+    elif storage_format == "parquet":
+        assert deps.archive(table_file) == ""
 
     # Load database to cache
     db = audb.load(name, version=version, verbose=False, full_path=False)