Check if workspace is in the cloud for locating cached json files

Author: Nalini Ganapati

genomicsdb/scripts/genomicsdb_cache.py

@@ -48,7 +48,7 @@ def get_arrays(interval, contigs_map, partitions):
 def main():
     parser = argparse.ArgumentParser(
         prog="cache",
-        description="Cache GenomicsDB metadata and generated callset/vidmap/loader json artifacts for workspace cloud URLs",  # noqa
+        description="Cache GenomicsDB metadata and generated callset/vidmap/loader json artifacts for workspace cloud URLs. The metadata is copied to TMPDIR and the json files to the current working directory",  # noqa
         formatter_class=argparse.RawTextHelpFormatter,
         usage="%(prog)s [options]",
     )

genomicsdb/scripts/genomicsdb_query.py

@@ -407,17 +407,18 @@ def setup():
     args = parser.parse_args()
 
     workspace = genomicsdb_common.normalize_path(args.workspace)
+    is_cloud_workspace = True if "://" in workspace else False
     if not genomicsdb.workspace_exists(workspace):
         raise RuntimeError(f"workspace({workspace}) not found")
     callset_file = args.callset
     if not callset_file:
-        if not args.no_cache and genomicsdb.is_file("callset.json"):
+        if is_cloud_workspace and not args.no_cache and genomicsdb.is_file("callset.json"):
             callset_file = "callset.json"
         else:
             callset_file = genomicsdb_common.join_paths(workspace, "callset.json")
     vidmap_file = args.vidmap
     if not vidmap_file:
-        if not args.no_cache and genomicsdb.is_file("vidmap.json"):
+        if is_cloud_workspace and not args.no_cache and genomicsdb.is_file("vidmap.json"):
             vidmap_file = "vidmap.json"
         else:
             vidmap_file = genomicsdb_common.join_paths(workspace, "vidmap.json")

test/scripts/test.sh

@@ -232,7 +232,7 @@ run_command "genomicsdb_query -w $WORKSPACE -i 4 --chunk-size=4 -b -o $OUTPUT -d
 # Duplicates
 check_command_with_duplicates "genomicsdb_query -w $WORKSPACE -i 1 -i 1 --chunk-size=2 -o $OUTPUT" 2 "1 1_1"
 check_command_with_duplicates "genomicsdb_query -w $WORKSPACE -i 1 -i 1 --chunk-size=2 -s HG00141 -s HG00141 -o $OUTPUT" 1 "1"
-
+run_command "genomicsdb_query -w $WORKSPACE -i 4 --chunk-size=4 -b -o $OUTPUT --no-cache"
 
 OLDSTYLE_JSONS="-l $OLDSTYLE_DIR/loader.json -c $OLDSTYLE_DIR/callset_t0_1_2.json -v $OLDSTYLE_DIR/vid.json"
 run_command "genomicsdb_cache -w $WORKSPACE $OLDSTYLE_JSONS $INTERVAL_ARGS"
@@ -247,4 +247,5 @@ run_command "genomicsdb_query -w $WORKSPACE $OLDSTYLE_JSONS $INTERVAL_ARGS -S $T
 rm $WORKSPACE/vcfheader.vcf
 run_command "genomicsdb_query -w $WORKSPACE $OLDSTYLE_JSONS --list-fields"
 
+
 cleanup