Merge pull request #18 from d3b-center/feature/as/dewrangle_list

Feature/as/dewrangle list

README.md

@@ -64,14 +64,15 @@ To perform dewrangling tasks, use the dewrangle command with subcommands:
 ```bash
 d3b dewrangle
 Subparser 'dewrangle'
-usage: d3b dewrangle [-h] {hash,list_jobs,download} ...
+usage: d3b dewrangle [-h] {hash,list_volume,list_jobs,download} ...
 
 optional arguments:
   -h, --help            show this help message and exit
 
 Dewrangle Subcommands:
   {hash,list_jobs,download}
     hash                Hash volume in Dewrangle
+    list_volume         List volume in Dewrangle
     list_jobs           List volume jobs in Dewrangle
     download            Download job results from Dewrangle
 ```

d3b_dff_cli/cli.py

@@ -5,6 +5,7 @@
 from .modules.validation.check_readgroup import main as check_readgroup
 from .modules.validation.check_url import main as check_url
 from .modules.dewrangle.volume import main as hash_volume
+from .modules.dewrangle.volume import run_list as list_volume
 from .modules.dewrangle.list_jobs import main as list_jobs
 from .modules.dewrangle.download_job import main as download_dewrangle_job
 from .modules.jira.create_ticket import main as create_ticket
@@ -99,6 +100,7 @@ def create_parser():
 
     # Dewrangle commands
     # hash: load a bucket to Dewrangle and hash it
+    # list: load a bucket to Dewrangle and list files in it
     # list_jobs: list jobs run on a bucket
     # download: download the results of a job
     dewrangle_parser = subparsers.add_parser("dewrangle", help="Dewrangle commands")
@@ -113,6 +115,13 @@ def create_parser():
     hash_parser = add_dewrangle_arguments(hash_parser)
     hash_parser.set_defaults(func=hash_volume)
 
+    # list subcommand
+    list_vol_parser = dewrangle_subparsers.add_parser(
+        "list_volume", help="List volume in Dewrangle"
+    )
+    list_vol_parser = add_dewrangle_arguments(list_vol_parser)
+    list_vol_parser.set_defaults(func=list_volume)
+
     # list_jobs subcommand
     list_parser = dewrangle_subparsers.add_parser(
         "list_jobs", help="List volume jobs in Dewrangle"

d3b_dff_cli/modules/dewrangle/volume.py

@@ -105,17 +105,52 @@ def list_and_hash_volume(client, volume_id, billing_id):
     return job_id
 
 
-def load_and_hash_volume(
+def list_volume(client, volume_id):
+    """Run Dewrangle list volume mutation."""
+
+    # prepare mutation
+    mutation = gql(
+        """
+        mutation VolumeListMutation($id: ID!) {
+            volumeList(id: $id) {
+                errors {
+                    ... on MutationError {
+                        message
+                        field
+                    }
+                }
+                job {
+                    id
+                }    
+            }
+        }
+        """
+    )
+
+    params = {"id": volume_id}
+
+    # run mutation
+    result = client.execute(mutation, variable_values=params)
+
+    hf.check_mutation_result(result)
+
+    job_id = result["volumeList"]["job"]["id"]
+
+    return job_id
+
+
+def load_and_run_job(
     bucket_name,
     study_name,
     region,
+    job_type,
     prefix=None,
     billing=None,
     aws_cred=None,
     token=None,
 ):
     """
-    Wrapper function that checks if a volume is loaded, and hashes it.
+    Wrapper function that checks if a volume is loaded, and either hashes or lists it.
     Inputs: AWS bucket name, study name, aws region, and optional volume prefix.
     Output: job id of parent job created when volume is hashed.
     """
@@ -129,9 +164,6 @@ def load_and_hash_volume(
         study_id = hf.get_study_id(client, study_name)
         org_id = hf.get_org_id_from_study(client, study_id)
 
-        # get billing group id
-        billing_group_id = hf.get_billing_id(client, org_id, billing)
-
         # check if volume loaded to study
         study_volumes = hf.get_study_volumes(client, study_id)
         volume_id = hf.process_volumes(
@@ -147,8 +179,13 @@ def load_and_hash_volume(
                 client, study_id, prefix, region, bucket_name, aws_cred_id
             )
 
-        # hash
-        job_id = list_and_hash_volume(client, volume_id, billing_group_id)
+        if job_type == "hash":
+            # get billing group id
+            billing_group_id = hf.get_billing_id(client, org_id, billing)
+            job_id = list_and_hash_volume(client, volume_id, billing_group_id)
+
+        elif job_type == "list":
+            job_id = list_volume(client, volume_id)
 
     except Exception:
         print(
@@ -161,12 +198,27 @@ def load_and_hash_volume(
     return job_id
 
 
+def run_list(args):
+    """Other main function to load and list a volume."""
+    job_id = load_and_run_job(
+        args.bucket,
+        args.study,
+        args.region,
+        "list",
+        args.prefix,
+        args.billing,
+        args.credential,
+    )
+    print(job_id)
+
+
 def main(args):
     """Main function. Call load_and_hash and output job_id."""
-    job_id = load_and_hash_volume(
+    job_id = load_and_run_job(
         args.bucket,
         args.study,
         args.region,
+        "hash",
         args.prefix,
         args.billing,
         args.credential,