Merge pull request #63 from uhh-cms/feature/broken_files_query

update get_das_info to include empty and broken files
uhh-cms · Oct 14, 2024 · f1193f5 · f1193f5
2 parents 83feb6d + 7491b38
commit f1193f5
Show file tree

Hide file tree

Showing 3 changed files with 179 additions and 88 deletions.
diff --git a/cmsdb/campaigns/run3_2022_postEE_nano_v12/data.py b/cmsdb/campaigns/run3_2022_postEE_nano_v12/data.py
@@ -4,6 +4,8 @@
 CMS datasets from the 2022 post-EE data-taking campaign
 """
 
+from order import DatasetInfo
+
 import cmsdb.processes as procs
 from cmsdb.campaigns.run3_2022_postEE_nano_v12 import campaign_run3_2022_postEE_nano_v12 as cpn
 
@@ -120,11 +122,18 @@
     id=14783435,
     is_data=True,
     processes=[procs.data_muoneg],
-    keys=[
-        "/MuonEG/Run2022E-22Sep2023-v1/NANOAOD",  # noqa
-    ],
-    n_files=29,
-    n_events=12873327,
+    info=dict(
+        nominal=DatasetInfo(
+            keys=[
+                "/MuonEG/Run2022E-22Sep2023-v1/NANOAOD",  # noqa: E501
+            ],
+            aux={
+                "broken_files": [],
+            },
+            n_files=29,  # 29-0
+            n_events=12873327,
+        ),
+    ),
     aux={
         "era": "E",
     },
@@ -135,11 +144,21 @@
     id=14784482,
     is_data=True,
     processes=[procs.data_muoneg],
-    keys=[
-        "/MuonEG/Run2022F-22Sep2023-v1/NANOAOD",  # noqa
-    ],
-    n_files=95,
-    n_events=38219969,
+    info=dict(
+        nominal=DatasetInfo(
+            keys=[
+                "/MuonEG/Run2022F-22Sep2023-v1/NANOAOD",  # noqa: E501
+            ],
+            aux={
+                "broken_files": [
+                    "/store/data/Run2022F/MuonEG/NANOAOD/22Sep2023-v1/50000/4d76213a-ef14-411a-9558-559a6df3f978.root",  # empty  # noqa: E501
+                    "/store/data/Run2022F/MuonEG/NANOAOD/22Sep2023-v1/50000/4fb72196-3b02-4499-8f6c-a54e15692b32.root",  # empty  # noqa: E501
+                ],
+            },
+            n_files=93,  # 95-2
+            n_events=38219969,
+        ),
+    ),
     aux={
         "era": "F",
     },
@@ -150,11 +169,20 @@
     id=14784485,
     is_data=True,
     processes=[procs.data_muoneg],
-    keys=[
-        "/MuonEG/Run2022G-22Sep2023-v1/NANOAOD",  # noqa
-    ],
-    n_files=27,
-    n_events=6238527,
+    info=dict(
+        nominal=DatasetInfo(
+            keys=[
+                "/MuonEG/Run2022G-22Sep2023-v1/NANOAOD",  # noqa: E501
+            ],
+            aux={
+                "broken_files": [
+                    "/store/data/Run2022G/MuonEG/NANOAOD/22Sep2023-v1/2520000/cd404eb6-8218-4787-b5ed-af6cd9fe3750.root",  # empty  # noqa: E501
+                ],
+            },
+            n_files=26,  # 27-1
+            n_events=6238527,
+        ),
+    ),
     aux={
         "era": "G",
     },

diff --git a/cmsdb/campaigns/run3_2022_preEE_nano_v12/data.py b/cmsdb/campaigns/run3_2022_preEE_nano_v12/data.py
@@ -4,6 +4,8 @@
 CMS datasets from the 2022 pre-EE data-taking campaign
 """
 
+from order import DatasetInfo
+
 import cmsdb.processes as procs
 from cmsdb.campaigns.run3_2022_preEE_nano_v12 import campaign_run3_2022_preEE_nano_v12 as cpn
 
@@ -120,11 +122,20 @@
     id=14783289,
     is_data=True,
     processes=[procs.data_muoneg],
-    keys=[
-        "/MuonEG/Run2022A-22Sep2023-v1/NANOAOD",  # noqa
-    ],
-    n_files=5,
-    n_events=12,
+    info=dict(
+        nominal=DatasetInfo(
+            keys=[
+                "/MuonEG/Run2022A-22Sep2023-v1/NANOAOD",  # noqa: E501
+            ],
+            aux={
+                "broken_files": [
+                    "/store/data/Run2022A/MuonEG/NANOAOD/22Sep2023-v1/50000/9a127bdb-9522-4f49-b754-67bb9152c0b3.root",  # empty  # noqa: E501
+                ],
+            },
+            n_files=4,  # 5-1
+            n_events=12,
+        ),
+    ),
     aux={
         "era": "A",
     },
@@ -135,11 +146,20 @@
     id=14784076,
     is_data=True,
     processes=[procs.data_muoneg],
-    keys=[
-        "/MuonEG/Run2022B-22Sep2023-v1/NANOAOD",  # noqa
-    ],
-    n_files=7,
-    n_events=254803,
+    info=dict(
+        nominal=DatasetInfo(
+            keys=[
+                "/MuonEG/Run2022B-22Sep2023-v1/NANOAOD",  # noqa: E501
+            ],
+            aux={
+                "broken_files": [
+                    "/store/data/Run2022B/MuonEG/NANOAOD/22Sep2023-v1/50000/947809ff-822e-4a3a-84a2-d3fe84fc2573.root",  # empty  # noqa: E501
+                ],
+            },
+            n_files=6,  # 7-1
+            n_events=254803,
+        ),
+    ),
     aux={
         "era": "B",
     },
@@ -150,11 +170,18 @@
     id=14784125,
     is_data=True,
     processes=[procs.data_muoneg],
-    keys=[
-        "/MuonEG/Run2022C-22Sep2023-v1/NANOAOD",  # noqa
-    ],
-    n_files=28,
-    n_events=15768439,
+    info=dict(
+        nominal=DatasetInfo(
+            keys=[
+                "/MuonEG/Run2022C-22Sep2023-v1/NANOAOD",  # noqa: E501
+            ],
+            aux={
+                "broken_files": [],
+            },
+            n_files=28,  # 28-0
+            n_events=15768439,
+        ),
+    ),
     aux={
         "era": "C",
     },
@@ -165,11 +192,18 @@
     id=14784209,
     is_data=True,
     processes=[procs.data_muoneg],
-    keys=[
-        "/MuonEG/Run2022D-22Sep2023-v1/NANOAOD",  # noqa
-    ],
-    n_files=16,
-    n_events=8007031,
+    info=dict(
+        nominal=DatasetInfo(
+            keys=[
+                "/MuonEG/Run2022D-22Sep2023-v1/NANOAOD",  # noqa: E501
+            ],
+            aux={
+                "broken_files": [],
+            },
+            n_files=16,  # 16-0
+            n_events=8007031,
+        ),
+    ),
     aux={
         "era": "D",
     },