Merge branch 'return_lfns' into 'master'

Added command to return LFNs

See merge request !36

setup.cfg

@@ -44,6 +44,7 @@ console_scripts =
     apd-logout = apd.command:cmd_logout
     apd-cache-info = apd.command:cmd_cache_ap_info
     apd-list-pfns = apd.command:cmd_list_pfns
+    apd-list-lfns = apd.command:cmd_list_lfns
     apd-list-samples = apd.command:cmd_list_samples
     apd-summary = apd.command:cmd_summary
     apd-cache-files = apd.command:cmd_cache_ap_files

src/apd/__init__.py

 ###############################################################################
-# (c) Copyright 2021 CERN for the benefit of the LHCb Collaboration           #
+# (c) Copyright 2021-2023 CERN for the benefit of the LHCb Collaboration      #
 #                                                                             #
 # This software is distributed under the terms of the GNU General Public      #
 # Licence version 3 (GPL Version 3), copied verbatim in the file "COPYING".   #

src/apd/analysis_data.py

@@ -351,8 +351,8 @@ class AnalysisData:
     ):
         # pylint: disable-msg=too-many-locals
         """Main method that returns the dataset info.
-        The normal behaviour is to return the PFNs for the samples but setting
-        return_pfns to false returns the SampleCollection"""
+        The normal behaviour is to return the PFNs for the samples, but setting
+        return_type to ApdReturnType.SAMPLE returns the SampleCollection"""
 
         # Establishing the list of samples to run on
         samples = self._samples

src/apd/command.py

 ###############################################################################
-# (c) Copyright 2021-2023 CERN for the benefit of the LHCb Collaboration      #
+# (c) Copyright 2021-203 CERN for the benefit of the LHCb Collaboration       #
 #                                                                             #
 # This software is distributed under the terms of the GNU General Public      #
 # Licence version 3 (GPL Version 3), copied verbatim in the file "COPYING".   #
@@ -21,7 +21,12 @@ import click  # type: ignore[import]
 import click_log  # type: ignore[import]
 import requests
 
-from .analysis_data import APD_DATA_CACHE_DIR, APD_METADATA_CACHE_DIR, get_analysis_data
+from .analysis_data import (
+    APD_DATA_CACHE_DIR,
+    APD_METADATA_CACHE_DIR,
+    ApdReturnType,
+    get_analysis_data,
+)
 from .ap_info import cache_ap_info
 from .authentication import get_auth_headers, logout
 from .data_cache import DataCache
@@ -192,6 +197,67 @@ def cmd_list_pfns(
         click.echo(f)
 
 
+@click.command()
+@click.argument("working_group")
+@click.argument("analysis")
+@click.option(
+    "--cache_dir",
+    default=os.environ.get(APD_METADATA_CACHE_DIR, None),
+    help="Specify location of the cache for the analysis metadata",
+)
+@click.option("--tag", default=None, help="Tag to filter datasets", multiple=True)
+@click.option(
+    "--value",
+    default=None,
+    help="Tag value used if the name is specified",
+    multiple=True,
+)
+@click.option(
+    "--eventtype", default=None, help="eventtype to filter the datasets", multiple=True
+)
+@click.option(
+    "--datatype", default=None, help="datatype to filter the datasets", multiple=True
+)
+@click.option(
+    "--polarity", default=None, help="polarity to filter the datasets", multiple=True
+)
+@click.option(
+    "--config", default=None, help="Config to use (e.g. lhcb or mc)", multiple=True
+)
+@click.option("--name", default=None, help="dataset name")
+@click.option("--version", default=None, help="dataset version")
+@click.option("--date", default=None, help="analysis date in ISO 8601 format")
+@click_log.simple_verbosity_option(logger)
+@common_docstr()
+def cmd_list_lfns(
+    working_group,
+    analysis,
+    cache_dir,
+    tag,
+    value,
+    eventtype,
+    datatype,
+    polarity,
+    config,
+    name,
+    version,
+    date,
+):
+    """List the LFNs for the analysis, matching the tags specified.
+    This command checks that the arguments are not ambiguous."""
+
+    # Loading the data and filtering/displaying
+    datasets = get_analysis_data(
+        working_group, analysis, metadata_cache=cache_dir, ap_date=date
+    )
+    filter_tags = _process_common_tags(
+        eventtype, datatype, polarity, config, name, version
+    )
+    filter_tags |= dict(zip(tag, value))
+    for f in datasets(**filter_tags, return_type=ApdReturnType.LFN):
+        click.echo(f)
+
+
 @click.command()
 @click.argument("working_group")
 @click.argument("analysis")
@@ -249,7 +315,9 @@ def cmd_list_samples(
         eventtype, datatype, polarity, config, name, version
     )
     filter_tags |= dict(zip(tag, value))
-    matching = datasets(check_data=False, return_pfns=False, **filter_tags)
+    matching = datasets(
+        check_data=False, return_type=ApdReturnType.SAMPLE, **filter_tags
+    )
     click.echo(matching)
 
 

src/apd/snakemake.py

@@ -16,7 +16,8 @@ try:
     from snakemake.remote.XRootD import RemoteProvider as XRootDRemoteProvider  # type: ignore[import]
 except Exception as exc:
     raise Exception("apd.snakemake requires snakemake to be available") from exc
-# isort: on
+
+from .analysis_data import ApdReturnType
 from .analysis_data import get_analysis_data as std_get_analysis_data
 from .eos import auth as std_auth
 from .eos import authw as std_authw
@@ -68,8 +69,8 @@ class AnalysisDataWrapper:
 
     def __call__(self, *args, **kwargs):
         results = self.analysisData(*args, **kwargs)
-        return_pfns = kwargs.get("return_pfns", True)
-        if not return_pfns:
+        return_type = kwargs.get("return_type", ApdReturnType.PFN)
+        if return_type != ApdReturnType.PFN:
             return results
         return [remote(f) for f in results]
 

tests/test_b2oc.py

 ###############################################################################
-# (c) Copyright 2021 CERN for the benefit of the LHCb Collaboration           #
+# (c) Copyright 2021-2023 CERN for the benefit of the LHCb Collaboration      #
 #                                                                             #
 # This software is distributed under the terms of the GNU General Public      #
 # Licence version 3 (GPL Version 3), copied verbatim in the file "COPYING".   #