From de482e831bd7ad64d84708e42ccdd8f358e70579 Mon Sep 17 00:00:00 2001
From: Hagen Wierstorf <hwierstorf@audeering.com>
Date: Wed, 27 Mar 2024 15:50:00 +0100
Subject: [PATCH] Skip broken datasets in audb.available() (#383)

* Skip broken datasets in audb.available()

* Add broken dataset as test

* Try to test available() with broken dataset

* Expand test
---
 audb/core/api.py  | 25 +++++++++++++++----------
 tests/conftest.py |  2 ++
 tests/test_api.py | 14 ++++++++++++++
 3 files changed, 31 insertions(+), 10 deletions(-)

diff --git a/audb/core/api.py b/audb/core/api.py
index c2cb4d7f..ed5daae3 100644
--- a/audb/core/api.py
+++ b/audb/core/api.py
@@ -51,16 +51,21 @@ def available(
                 # see https://github.com/audeering/audbackend/issues/132
                 for p in backend._repo.path:
                     name = p.name
-                    for version in [str(x).split("/")[-1] for x in p / "db"]:
-                        databases.append(
-                            [
-                                name,
-                                repository.backend,
-                                repository.host,
-                                repository.name,
-                                version,
-                            ]
-                        )
+                    try:
+                        for version in [str(x).split("/")[-1] for x in p / "db"]:
+                            databases.append(
+                                [
+                                    name,
+                                    repository.backend,
+                                    repository.host,
+                                    repository.name,
+                                    version,
+                                ]
+                            )
+                    except FileNotFoundError:
+                        # If the `db` folder does not exist,
+                        # we do not include the dataset
+                        pass
             else:
                 for path, version in backend.ls("/"):
                     if path.endswith(define.HEADER_FILE):
diff --git a/tests/conftest.py b/tests/conftest.py
index d8e5a19f..fdecb8db 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -182,6 +182,7 @@ def private_and_public_repository():
     Configure the following repositories:
     * data-private: repo on public Artifactory without access
     * data-public: repo on public Artifactory with anonymous access
+    * data-public2: repo on public Artifactory with anonymous access
 
     Note, that the order of the repos is important.
     audb will visit the repos in the given order
@@ -194,6 +195,7 @@ def private_and_public_repository():
     audb.config.REPOSITORIES = [
         audb.Repository("data-private", host, backend),
         audb.Repository("data-public", host, backend),
+        audb.Repository("data-public2", host, backend),
     ]
 
     yield repository
diff --git a/tests/test_api.py b/tests/test_api.py
index 12881a00..58fa252d 100644
--- a/tests/test_api.py
+++ b/tests/test_api.py
@@ -29,3 +29,17 @@ def test_available(repository):
     ]
     df = audb.available()
     assert len(df) == 0
+
+
+def test_available_broken_dataset(private_and_public_repository):
+    """Test for listing datasets, including a broken one.
+
+    This uses the public repositories,
+    from which ``data-public2``
+    includes ``broken-dataset``,
+    which has a missing ``db`` folder.
+
+    """
+    df = audb.available(only_latest=True)
+    assert len(df) > 0
+    assert "broken-dataset" not in df