[EDPDEV-2479][EDPDEV-2789][EDPDEV-2598] (#481)

EDPDEV-2749 - Fix upload command for nested folders when dry-run flag is active
EDPDEV-2789 - Fix subjects retrieval in GlobalSearch module
EDPDEV-2598 - Enhance solvebio download command to support one file download even if --recursive flag is active

Author: nrasulicpfm

solvebio/cli/data.py

@@ -195,9 +195,17 @@ def _upload_folder(
     # be populated
     all_folder_parts = sorted(all_folder_parts, key=lambda x: len(x.split("/")))
     for folder in all_folder_parts:
-        print("{}Creating folder {}".format(
-            "[Dry Run] " if dry_run else "", folder))
-        Object.create_folder(vault, folder)
+        if dry_run:
+            try:
+                f = Object.get_by_full_path(folder)
+                if not f.is_folder:
+                    print("[Dry Run] {} is not a folder - this will cause an error on upload.".format(folder))
+                else:
+                    print("[Dry Run] Folder {} already exists - skipping creation".format(folder))
+            except NotFoundError:
+                print("[Dry Run] Creating folder {}".format(folder))
+        else:
+            Object.create_folder(vault, folder)
 
     # Create files in parallel
     # Signal handling allows for graceful exit upon KeyboardInterrupt
@@ -234,18 +242,32 @@ def _create_file_job(args):
     try:
         local_file_path, remote_folder_full_path, vault_path, dry_run, archive_folder, client_auth, follow_shortcuts \
             = args
-        if dry_run:
-            print("[Dry Run] Uploading {} to {}".format(
-                local_file_path, remote_folder_full_path))
-            return
+
         # Provides the global host, token, token_type
         client = SolveClient(*client_auth)
-        remote_parent = Object.get_by_full_path(
-            remote_folder_full_path,
-            assert_type="folder",
-            follow_shortcuts=follow_shortcuts,
-            client=client
-        )
+
+        remote_parent = None
+        try: 
+            remote_parent = Object.get_by_full_path(
+                remote_folder_full_path,
+                assert_type="folder",
+                follow_shortcuts=follow_shortcuts,
+                client=client
+            )
+        except NotFoundError as e:
+            if not dry_run:
+                raise e
+
+        if dry_run:
+            if not _object_exists(remote_parent, local_file_path, client):
+                print("[Dry Run] Uploading {} to {}".format(
+                    local_file_path, remote_folder_full_path))
+                return
+            else:
+                print("[Dry Run] File {} already exists at {} - skipping upload".format(
+                    local_file_path, remote_folder_full_path))
+                return
+
         Object.upload_file(
             local_file_path,
             remote_parent.path,
@@ -260,6 +282,26 @@ def _create_file_job(args):
     except Exception as e:
         return e
 
+def _object_exists(remote_parent, local_path, _client):
+    if remote_parent is None:
+        return False
+    full_path, path_dict = Object.validate_full_path(
+            os.path.join('{}:{}'.format(remote_parent.vault.full_path, remote_parent.path),
+                         os.path.basename(local_path)), client=_client)
+    try:
+        obj = Object.get_by_full_path(full_path, client=_client)
+        if not obj.is_file:
+            return False
+        else:
+            # Check if the md5sum matches
+            local_md5 = md5sum(local_path)[0]
+            remote_md5 = obj.get("md5")
+            if remote_md5 and remote_md5 == local_md5:
+                return True
+            else:
+                return False
+    except NotFoundError:
+        return False
 
 def _create_template_from_file(template_file, dry_run=False):
     mode = "r"
@@ -752,9 +794,11 @@ def _download_recursive(
 
     min_depth = min([x.depth for x in remote_objects])
     num_at_min_depth = len([x for x in remote_objects if x.depth == min_depth])
-    if num_at_min_depth == 1:
+    if num_at_min_depth == 1 and not _is_single_file(remote_objects):
+        # when downloading from folder
         base_folder_depth = min_depth
     else:
+        # when downloading from vault root or singular file
         base_folder_depth = min_depth - 1
 
     downloaded_files = set()
@@ -995,6 +1039,10 @@ def _ls(full_path, recursive=False, follow_shortcuts=False):
     return files
 
 
+def _is_single_file(objects):
+    return len(objects) == 1 and objects[0].get("object_type") == "file"
+
+
 def should_tag_by_object_type(args, object_):
     """Returns True if object matches object type requirements"""
     valid = True

solvebio/global_search.py

@@ -202,7 +202,7 @@ def subjects(self):
 
         # Executes a query to get a full API response which contains subjects list
         gs = self.limit(0)
-        gs.execute(include_subjects=True)
+        gs.execute(include_subjects=True, include_all_subjects=True)
 
         return gs._response.get('subjects')
 

solvebio/resource/object.py

@@ -370,6 +370,7 @@ def create_folder(cls, vault, full_path, tags=None, **kwargs):
                 parent_object_id = parent.id
 
             # Make the API call
+            print("Creating folder {}".format(full_path))
             new_obj = Object.create(
                 vault_id=vault.id,
                 parent_object_id=parent_object_id,