Added possibility to list LFNs

src/apd/__init__.py

@@ -24,9 +24,10 @@ __all__ = [
     "cache_ap_info",
     "auth",
     "authw",
+    "ApdReturnType",
 ]
 
-from .analysis_data import AnalysisData, get_analysis_data
+from .analysis_data import AnalysisData, ApdReturnType, get_analysis_data
 from .ap_info import (
     SampleCollection,
     cache_ap_info,

src/apd/analysis_data.py

@@ -20,6 +20,7 @@ import copy
 import itertools
 import logging
 import os
+from enum import Enum
 from pathlib import Path
 
 from apd.ap_info import (
@@ -41,6 +42,12 @@ APD_METADATA_LIFETIME_DEFAULT = 600
 APD_DATA_CACHE_DIR = "APD_DATA_CACHE_DIR"
 
 
+class ApdReturnType(Enum):
+    PFN = 0
+    LFN = 1
+    SAMPLE = 2
+
+
 def _load_and_setup_cache(
     cache_dir, working_group, analysis, ap_date=None, api_url="https://lbap.app.cern.ch"
 ):
@@ -335,7 +342,7 @@ class AnalysisData:
     def __call__(
         self,
         *,
-        return_pfns=True,
+        return_type=ApdReturnType.PFN,
         check_data=True,
         use_local_cache=True,
         showmax=10,
@@ -390,12 +397,17 @@ class AnalysisData:
                 logger.debug("Error loading data: %s", error_txt)
                 raise ValueError("Error loading data: " + error_txt)
 
-        if return_pfns:
-            if use_local_cache:
-                return self._transform_pfns(samples.PFNs())
-            return samples.PFNs()
+        if return_type == ApdReturnType.SAMPLE:
+            return samples
+
+        if return_type == ApdReturnType.LFN:
+            print("Returning lfns")
+            return samples.LFNs()
 
-        return samples
+        # by default we return the PFns
+        if use_local_cache:
+            return self._transform_pfns(samples.PFNs())
+        return samples.PFNs()
 
     def _transform_pfns(self, pfns):
         """Method to return PFNs, useful as it can be overriden in inheriting classes"""

src/apd/ap_info.py

@@ -371,6 +371,13 @@ class SampleCollection:
                 pfns.append(auth(pfnlist[0]))
         return pfns
 
+    def LFNs(self):
+        """Collects the LFNs"""
+        lfns = []
+        for sample in self.info:
+            lfns += sample["lfns"].keys()
+        return lfns
+
     def byte_count(self):
         """Collects the number of files from all the samples"""
         count = 0

tests/test_b2oc.py

@@ -11,7 +11,7 @@
 
 import pytest
 
-from apd import AnalysisData
+from apd import AnalysisData, ApdReturnType
 
 
 @pytest.mark.parametrize("with_tokens", [[False], [True]])
@@ -189,6 +189,7 @@ def test_name_case_sensitivity(apd_cache):
 
 
 def test_unknown_tag_value(apd_cache):
+    """CHeck that we throw a ValueError when a value does not exist for a given tag."""
     datasets = AnalysisData("b2oc", "b02dkpi")
 
     with pytest.raises(ValueError, match="No sample for tag datatype=2032.*"):
@@ -198,3 +199,19 @@ def test_unknown_tag_value(apd_cache):
             eventtype="11164047",
             mc=True,
         )
+
+
+def test_lfn(apd_cache):
+    """Check that the method to return the LFNs is functional."""
+    datasets = AnalysisData("b2oc", "b02dkpi")
+    lfns = datasets(
+        datatype="2011",
+        eventtype="11164047",
+        polarity="magdown",
+        return_type=ApdReturnType.LFN,
+    )
+    assert (
+        len(lfns) == 5
+        and all("00128098_0000" in x for x in lfns)
+        and all(x.startswith("/lhcb") for x in lfns)
+    )