From 3640f163fca24fc37cc801645abace8e8b779afd Mon Sep 17 00:00:00 2001
From: Justus Kuhlmann <justus@kuhl-mann.de>
Date: Wed, 6 May 2026 19:06:50 +0200
Subject: [PATCH] Give user a sense of the severity, add basic tests

---
 corrlib/integrity.py    |  5 +++-
 tests/integrity_test.py | 55 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 59 insertions(+), 1 deletion(-)
 create mode 100644 tests/integrity_test.py

diff --git a/corrlib/integrity.py b/corrlib/integrity.py
index 4c35c4b..2fb520d 100644
--- a/corrlib/integrity.py
+++ b/corrlib/integrity.py
@@ -60,7 +60,10 @@ def are_keys_unique(db: Path, table: str, col: str) -> bool:
     c.execute(f"SELECT COUNT( DISTINCT CAST({col} AS nvarchar(4000))), COUNT({col}) FROM {table};")
     results = c.fetchall()[0]
     conn.close()
-    return bool(results[0] == results[1])
+    res = bool(results[0] == results[1])
+    if not res:
+        print("Unique:", results[0], "All:", results[1])
+    return res
 
 
 def _list_projects(path: Path) -> list[tuple[str, str]]:
diff --git a/tests/integrity_test.py b/tests/integrity_test.py
new file mode 100644
index 0000000..2cf8308
--- /dev/null
+++ b/tests/integrity_test.py
@@ -0,0 +1,55 @@
+import corrlib.integrity as integ
+import corrlib.find as find
+import datalad.api as dl
+import corrlib.initialization as cinit
+import sqlite3
+from pathlib import Path
+import os
+
+
+def test_list_ensembles(tmp_path: Path) -> None:
+    """
+    Check against the implementation in find to check if they are the same.
+    """
+    os.mkdir(tmp_path / 'archive')
+    os.mkdir(tmp_path / 'archive' / 'A')
+    os.mkdir(tmp_path / 'archive' / 'B')
+    os.mkdir(tmp_path / 'archive' / 'C')
+    integ_results = integ._list_ensembles(tmp_path)
+    assert len(integ_results) == 3
+    find_results = find.list_ensembles(tmp_path)
+    assert len(find_results) == 3
+    for f,i in zip(find_results, integ_results):
+        assert f == i
+
+
+def test_list_projects(tmp_path: Path) -> None:
+    cinit.create(tmp_path)
+    db = tmp_path / "backlogger.db"
+    dl.unlock(str(db), dataset=str(tmp_path))
+    conn = sqlite3.connect(db)
+    c = conn.cursor()
+
+    customTags = ""
+    owner = "owner"
+    code = "sfcf"
+    created_at = "today"
+    updated_at = "today"
+
+    id = "asdf1"
+    aliases = "a1,s1,d1,f1"
+    c.execute("INSERT INTO projects (id, aliases, customTags, owner, code, created_at, updated_at) VALUES (?,?,?,?,?,?,?)", (id, aliases, customTags, owner, code , created_at, updated_at))
+    id = "asdf2"
+    aliases = "a2,s2,d2,f2"
+    c.execute("INSERT INTO projects (id, aliases, customTags, owner, code, created_at, updated_at) VALUES (?,?,?,?,?,?,?)", (id, aliases, customTags, owner, code , created_at, updated_at))
+    id = "asdf3"
+    aliases = "a3,s3,d3,f3"
+    c.execute("INSERT INTO projects (id, aliases, customTags, owner, code, created_at, updated_at) VALUES (?,?,?,?,?,?,?)", (id, aliases, customTags, owner, code , created_at, updated_at))
+    conn.commit()
+    conn.close
+    integ_results = integ._list_projects(tmp_path)
+    assert len(integ_results) == 3
+    find_results = find.list_projects(tmp_path)
+    assert len(find_results) == 3
+    for f,i in zip(find_results, integ_results):
+        assert f == i