LbAPCommon 0.18.2__tar.gz → 0.18.3__tar.gz

{lbapcommon-0.18.2 → lbapcommon-0.18.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: LbAPCommon
-Version: 0.18.2
+Version: 0.18.3
 Summary: Common utilities used by LHCb DPA WP2 related software
 Author: LHCb
 Project-URL: Homepage, https://gitlab.cern.ch/lhcb-dpa/analysis-productions/LbAPCommon

{lbapcommon-0.18.2 → lbapcommon-0.18.3}/src/LbAPCommon/data/query_dirac.py

@@ -212,41 +212,62 @@ def parse_data_period(bk_query):
 
 
 def mc_dstformat(bk_query):
-    """dstformat for Run 3 MC, from the data-taking year/period in the BK path.
+    """dstformat for Run 3 *pp* MC, from the data-taking period in the BK path.
 
-    Mirrors LbMCSubmit's MC sprucing rule:
+    dstformat applies to MC only (this is never called for data). The value can
+    only be determined with certainty for pp data-taking:
 
-      * 2024 MC and preTS-2025 MC (2025.W21, 2025.W22.25) -> ``uDST``
-      * postTS-2025 MC and all 2026 MC -> ``DST``
+      * 2024 and preTS-2025 (2025.W21, 2025.W22.25) -> ``uDST``
+      * postTS-2025 and 2026 -> ``DST``
 
-    The year is taken from the ``YYYY.Wxx`` period token when present, else from
-    a bare year in the path (2026 ``expected-2026`` samples carry no week token).
-    Returns None when the year can't be determined or isn't a year we have a
-    rule for (so nothing is assumed); real data never gets a dstformat and is
-    not handled here.
+    preTS-2025 is taken directly from LbMCSubmit
+    (``SPRUCING_UDST_DSTFORMAT_DATA_TYPES``); the 2024 / postTS-2025 / 2026
+    mapping is the agreed pp rule. The pp period comes from the ``YYYY.Wxx``
+    token, or a pp ``expected-YYYY`` anticipated sample (e.g. 2026, no week
+    token).
+
+    For anything else -- heavy-ion (``Lead<YY>``), 5TeV reference runs
+    (``YYYY-5TeV``), fixed-target, an unknown year, or an unrecognised path --
+    the dstformat cannot be determined from the bookkeeping path (the real
+    trigger, the sample's ``file_format``, is not in it), so this RAISES rather
+    than guess. The job is then blocked until the user sets it by hand.
     """
     period = parse_data_period(bk_query)
     if period is not None:
         year = period.split(".", 1)[0]
+        if year == "2024":
+            return "uDST"
+        if year == "2025":
+            # preTS 2025 is uDST; everything later (postTS) is DST.
+            return "uDST" if period in _UDST_2025_PERIODS else "DST"
+        if year == "2026":
+            return "DST"
     else:
-        m = re.search(r"\b(202\d)\b", bk_query)
-        year = m.group(1) if m else None
-    if year == "2024":
-        return "uDST"
-    if year == "2025":
-        # preTS 2025 is uDST; everything later (postTS) is DST.
-        return "uDST" if period in _UDST_2025_PERIODS else "DST"
-    if year == "2026":
-        return "DST"
-    return None
+        # No pp week token: only pp ``expected-YYYY`` anticipated samples can be
+        # classified by year (2025 expected has no pre/post-TS info -> falls
+        # through to the error below).
+        m = re.search(r"\bexpected-(20\d\d)\b", bk_query)
+        if m:
+            year = m.group(1)
+            if year == "2024":
+                return "uDST"
+            if year == "2026":
+                return "DST"
+    raise AutoConfError(
+        f"Cannot determine the MC dstformat for {bk_query!r} with certainty. "
+        "Automatic dstformat is only supported for pp data-taking (2024-2026); "
+        "heavy-ion, 5TeV and other configurations are not. Set the required "
+        "extra_options (including dstformat) by hand and use "
+        "automatically_configure: false."
+    )
 
 
 def run3_dstformat(bk_query, simulation):
-    """dstformat for a Run 3 job's extra_options, or None to leave it unset.
+    """dstformat for a Run 3 job's extra_options.
 
-    Only Run 3 MC carries a dstformat, following the data-taking period
-    (:func:`mc_dstformat`) regardless of whether the job reads spruced or HLT2
-    output. Real data never gets a dstformat.
+    dstformat applies to MC only -- real data never gets one (returns None). For
+    MC it is :func:`mc_dstformat`, which raises ``AutoConfError`` when the value
+    cannot be determined with certainty (so the job is blocked, not guessed).
     """
     if not simulation:
         return None

{lbapcommon-0.18.2 → lbapcommon-0.18.3}/src/LbAPCommon.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: LbAPCommon
-Version: 0.18.2
+Version: 0.18.3
 Summary: Common utilities used by LHCb DPA WP2 related software
 Author: LHCb
 Project-URL: Homepage, https://gitlab.cern.ch/lhcb-dpa/analysis-productions/LbAPCommon

{lbapcommon-0.18.2 → lbapcommon-0.18.3}/src/LbAPCommon.egg-info/SOURCES.txt

@@ -21,6 +21,8 @@ src/LbAPCommon.egg-info/SOURCES.txt
 src/LbAPCommon.egg-info/dependency_links.txt
 src/LbAPCommon.egg-info/entry_points.txt
 src/LbAPCommon.egg-info/requires.txt
+src/LbAPCommon.egg-info/scm_file_list.json
+src/LbAPCommon.egg-info/scm_version.json
 src/LbAPCommon.egg-info/top_level.txt
 src/LbAPCommon/data/__init__.py
 src/LbAPCommon/data/ap_merger.py

lbapcommon-0.18.3/src/LbAPCommon.egg-info/scm_file_list.json

@@ -0,0 +1,141 @@
+{
+  "files": [
+    ".gitignore",
+    ".gitlab-ci.yml",
+    ".pre-commit-config.yaml",
+    "LICENSE",
+    "README.md",
+    "environment.yaml",
+    "pixi.lock",
+    "pixi.toml",
+    "pyproject.toml",
+    "src/LbAPCommon/__init__.py",
+    "src/LbAPCommon/__main__.py",
+    "src/LbAPCommon/app_catalog.py",
+    "src/LbAPCommon/cern_sso.py",
+    "src/LbAPCommon/config.py",
+    "src/LbAPCommon/dirac_conversion.py",
+    "src/LbAPCommon/hacks.py",
+    "src/LbAPCommon/parsing.py",
+    "src/LbAPCommon/workflow.py",
+    "src/LbAPCommon/data/__init__.py",
+    "src/LbAPCommon/data/ap_merger.py",
+    "src/LbAPCommon/data/query_dirac.py",
+    "src/LbAPCommon/data/templates/configure-legacy-application.py.j2",
+    "src/LbAPCommon/linting/__init__.py",
+    "src/LbAPCommon/linting/bk_paths.py",
+    "src/LbAPCommon/models/__init__.py",
+    "src/LbAPCommon/models/recipes.py",
+    "src/LbAPCommon/models/validators.py",
+    "src/LbAPCommon/options_parsing/__init__.py",
+    "src/LbAPCommon/options_parsing/gaudi_pickle_to_json.py",
+    "src/LbAPCommon/prod_request_to_cwl/__init__.py",
+    "src/LbAPCommon/prod_request_to_cwl/__main__.py",
+    "src/LbAPCommon/prod_request_to_cwl/anaprod.py",
+    "src/LbAPCommon/prod_request_to_cwl/common.py",
+    "src/LbAPCommon/prod_request_to_cwl/simulation.py",
+    "src/LbAPCommon/transforms/__init__.py",
+    "src/LbAPCommon/transforms/runner.py",
+    "src/LbAPCommon/transforms/yaml-transform.js",
+    "src/LbAPCommon/transforms/vendor/LICENSE-yaml",
+    "src/LbAPCommon/transforms/vendor/yaml.bundle.js",
+    "src/LbAPCommon/validators/__init__.py",
+    "src/LbAPCommon/validators/bookkeeping_xml.py",
+    "src/LbAPCommon/validators/check_gen_fsr.py",
+    "src/LbAPCommon/validators/counters.py",
+    "src/LbAPCommon/validators/logs/__init__.py",
+    "src/LbAPCommon/validators/logs/data/known_messages.yaml",
+    "tests/conftest.py",
+    "tests/test_app_catalog.py",
+    "tests/test_auto_configure.py",
+    "tests/test_auto_configure_dstformat.py",
+    "tests/test_auto_configure_streams.py",
+    "tests/test_autoconfig_logging.py",
+    "tests/test_dirac_conversion.py",
+    "tests/test_good_parsing.py",
+    "tests/test_hacks.py",
+    "tests/test_log_parsing.py",
+    "tests/test_log_splitting.py",
+    "tests/test_performance.py",
+    "tests/test_platform_selection.py",
+    "tests/test_transforms.py",
+    "tests/example-logs/error-failed-to-read-file.log",
+    "tests/example-logs/error-fill-dependent-scale-factor.log",
+    "tests/example-logs/error-illegal-instruction.log",
+    "tests/example-logs/error-missing-shared-library.log",
+    "tests/example-logs/error-platform-unsupported.log",
+    "tests/example-logs/error-related-info-missing.log",
+    "tests/example-logs/error-rootioalg.log",
+    "tests/example-logs/error-unpack-raw-event.log",
+    "tests/example-logs/good-DaVinci_00110296_00000038_1.log",
+    "tests/example-logs/good-DaVinci_00110296_00000194_1.log",
+    "tests/example-logs/good-Gauss_00104988_00000011_1.log",
+    "tests/example_workflows/complex_workflow_with_filtering.in.yaml",
+    "tests/example_workflows/complex_workflow_with_filtering.out.yaml",
+    "tests/example_workflows/complex_workflow_with_filtering_two_groups.in.yaml",
+    "tests/example_workflows/complex_workflow_with_filtering_two_groups.out.yaml",
+    "tests/example_workflows/complex_workflow_with_layers_of_splitting.in.yaml",
+    "tests/example_workflows/complex_workflow_with_layers_of_splitting.out.yaml",
+    "tests/example_workflows/mcreweight_and_pidgen2.in.yaml",
+    "tests/example_workflows/mcreweight_and_pidgen2.out.yaml",
+    "tests/example_workflows/complex_workflow_with_filtering/AnaProd_example_workflows_job_tuple,split,filter_B0.cwl",
+    "tests/example_workflows/complex_workflow_with_filtering_two_groups/AnaProd_example_workflows_job2_tuple,split,filter_B0,S0.cwl",
+    "tests/example_workflows/complex_workflow_with_filtering_two_groups/AnaProd_example_workflows_job_tuple,split,filter_B0,S0.cwl",
+    "tests/example_workflows/complex_workflow_with_layers_of_splitting/AnaProd_example_workflows_xicpststtoxicppipi_data_2018_magdown_split,_xicpstst,0st,pststwspmm.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_b0_dstdsx_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_b0_dstdx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_b0_dstp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_b0_dstp_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_b0_dststp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_b0_dststp_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_b0_dststx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_bp_dstdx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_bp_dstdx_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_bp_dstst_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_bp_dstst_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_bp_dststx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block5_bs_dsststp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_b0_dstdsx_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_b0_dstdx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_b0_dstp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_b0_dstp_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_b0_dststp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_b0_dststp_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_b0_dststx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_bp_dstdx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_bp_dstdx_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_bp_dstst_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_bp_dstst_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_bp_dststx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block6_bs_dsststp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_b0_dstdsx_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_b0_dstdx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_b0_dstp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_b0_dstp_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_b0_dststp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_b0_dststp_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_b0_dststx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_bp_dstdx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_bp_dstdx_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_bp_dstst_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_bp_dstst_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_bp_dststx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block7_bs_dsststp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_b0_dstdsx_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_b0_dstdx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_b0_dstp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_b0_dstp_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_b0_dststp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_b0_dststp_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_b0_dststx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_bp_dstdx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_bp_dstdx_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_bp_dstst_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_bp_dstst_taue_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_bp_dststx_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/example_workflows/mcreweight_and_pidgen2/AnaProd_example_workflows_noPID_block8_bs_dsststp_e_SPRUCE,DV,mva_rank,cmatch,PIDGen,mcreweight.cwl",
+    "tests/linting/test_bk_paths.py",
+    "tests/linting/test_processing_pass.py",
+    "tests/test_performance/example1.yaml"
+  ]
+}

lbapcommon-0.18.3/src/LbAPCommon.egg-info/scm_version.json

@@ -0,0 +1,8 @@
+{
+  "tag": "0.18.3",
+  "distance": 0,
+  "node": "g0ab0755cf8c2ecbd729005dcd63b9707b819c238",
+  "dirty": false,
+  "branch": "HEAD",
+  "node_date": "2026-06-25"
+}

{lbapcommon-0.18.2 → lbapcommon-0.18.3}/tests/test_auto_configure_dstformat.py

@@ -80,22 +80,44 @@ class TestMcDstformat:
                 "HLT2_FILTERED-2025.W28/10000000/HLT2_FILTERED.DST",
                 "DST",
             ),
-            # 2026 -> DST (incl. the bare-year 'expected-2026' samples)
+            # 2026 -> DST (the pp 'expected-2026' samples carry no week token)
             ("/MC/2026/Beam-2026.W10-MagUp/Sim11/Spruce/1/SPRUCE.DST", "DST"),
             (
                 "/MC/Dev/Beam6800GeV-expected-2026-MagUp-Nu7.6-25ns-Pythia8/"
                 "Sim10g/HLT2_pp_expected-2026/10000000/HLT2.DST",
                 "DST",
             ),
-            # 2024 with no week token still resolves by year -> uDST
-            ("/MC/2024/x/Sim10f/HLT2-2024/11884651/HLT2.DST", "uDST"),
-            # genuinely no data-taking year -> None (assume nothing)
-            ("/MC/Dev/Beam-MagUp/Sim10g/HLT2/1/HLT2.DST", None),
+            # pp 'expected-2024' anticipated sample -> uDST
+            (
+                "/MC/Dev/Beam6800GeV-expected-2024-MagUp/Sim10g/"
+                "HLT2_pp_expected-2024/10000000/HLT2.DST",
+                "uDST",
+            ),
         ],
     )
     def test_mc_dstformat(self, bk_query, expected):
         assert mc_dstformat(bk_query) == expected
 
+    @pytest.mark.parametrize(
+        "bk_query",
+        [
+            # 5TeV reference run (configVersion '2024-5TeV') -- NOT pp
+            "/MC/2024-5TeV/Beam2510GeV-2024-5TeV-MagUp/Sim10g/HLT2-2024/1/HLT2.DST",
+            # heavy-ion (configVersion 'Lead26')
+            "/MC/Lead26/Beam2680GeV-Lead26-MagUp-PbPb/Sim10g/HLT2-Lead26/1/HLT2.DST",
+            # year digits in path but no week token and not 'expected-'
+            "/MC/2024/x/Sim10f/HLT2-2024/11884651/HLT2.DST",
+            # 'expected-2025' cannot be classified pre/post-TS
+            "/MC/Dev/Beam6800GeV-expected-2025-MagUp/Sim10g/HLT2_pp_expected-2025/1/HLT2.DST",
+            # genuinely no data-taking year
+            "/MC/Dev/Beam-MagUp/Sim10g/HLT2/1/HLT2.DST",
+        ],
+    )
+    def test_mc_dstformat_uncertain_raises(self, bk_query):
+        # If we cannot be certain, we must throw -- never guess / silently unset.
+        with pytest.raises(AutoConfError, match="dstformat"):
+            mc_dstformat(bk_query)
+
 
 class TestRun3Dstformat:
     """dstformat applies to MC (spruced OR HLT2 input); data never gets one."""
@@ -125,6 +147,17 @@ class TestRun3Dstformat:
         bk = "/LHCb/Collision25/Beam-2025.W21/Real Data/Sprucing25c1/9/CHARM.DST"
         assert run3_dstformat(bk, simulation=False) is None
 
+    def test_non_pp_mc_raises(self):
+        # Non-pp MC cannot be configured with certainty -> must raise, not unset.
+        bk = "/MC/Lead26/Beam2680GeV-Lead26-MagUp-PbPb/Sim10g/HLT2-Lead26/1/HLT2.DST"
+        with pytest.raises(AutoConfError):
+            run3_dstformat(bk, simulation=True)
+
+    def test_non_pp_data_returns_none(self):
+        # dstformat is MC-only: even a non-pp DATA path gets None (never raises).
+        bk = "/LHCb/Lead26/Beam2680GeV-VeloClosed-MagUp/Real Data/Spruce/9/HI.DST"
+        assert run3_dstformat(bk, simulation=False) is None
+
 
 DETDESC_MC_SPRUCED_FACTS = {
     "simulation": True,