From 12376a8761d89686f9e11cfb07b3704e90be5072 Mon Sep 17 00:00:00 2001 From: Christophe Haen Date: Fri, 26 Jul 2024 17:33:18 +0200 Subject: [PATCH] feat (consistency): check for duplicated lines --- consistency_check/consistency.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/consistency_check/consistency.py b/consistency_check/consistency.py index 527999687c8..3485304a29b 100755 --- a/consistency_check/consistency.py +++ b/consistency_check/consistency.py @@ -27,6 +27,8 @@ def load_se_dump(se_dump_path): se_dump = pd.read_csv(se_dump_path, names=["pfn", "se_cks"], delimiter="|", index_col="pfn") se_dump["se_cks"] = se_dump["se_cks"].str.lower().str.pad(8, fillchar="0") se_dump["version"] = "se_dump" + assert not se_dump.index.duplicated().any(), f"Duplicated entries in SE dump {se_dump[se_dump.index.duplicated()]}" + return se_dump