From 48aeb8836b8e1f73626b3454acf5771a552aaaac Mon Sep 17 00:00:00 2001
From: paulzierep <paul.zierep@googlemail.com>
Date: Tue, 3 Dec 2024 13:27:13 +0100
Subject: [PATCH 1/2] logic for downloading v6

---
 .../data_manager_fetch_mapseq_db.py           | 26 ++++++++++++++++---
 .../data_manager/mapseq_db_fetcher.xml        |  4 +--
 2 files changed, 24 insertions(+), 6 deletions(-)

diff --git a/data_managers/data_manager_mapseq/data_manager/data_manager_fetch_mapseq_db.py b/data_managers/data_manager_mapseq/data_manager/data_manager_fetch_mapseq_db.py
index 6c1d5c41517..34180971d48 100644
--- a/data_managers/data_manager_mapseq/data_manager/data_manager_fetch_mapseq_db.py
+++ b/data_managers/data_manager_mapseq/data_manager/data_manager_fetch_mapseq_db.py
@@ -52,9 +52,21 @@ def download_untar_store(url, tmp_path, dest_path):
     tar = tarfile.open(tarfile_path)
     tar.extractall(extract_path)
 
+    print(f"Content of folder: {extract_path}", os.listdir(extract_path))
+
+    # case for mapseq v6: all DB files are directly in the tar.gz file
+    # remove the VERSION.txt file since the tool can only handle on .txt file in the DB
     if len(list(os.listdir(extract_path))) > 1:
-        print("More then one folder in zipped file, aborting !")
+        print(f"Found multiple files in {extract_path}. Copy the content.")
+        print(f"Copy data to {dest_path}")
+        version_file_path = os.path.join(extract_path, "VERSION.txt")
+        os.remove(version_file_path)
+        shutil.copytree(extract_path, dest_path)
+        print("Done !")
+
+    # case for mapseq v5: all files are in a subfolder in the tar.gz file
     else:
+        print(f"Found a folder in {extract_path}. Copy the content of the folder.")
         for folder in os.listdir(extract_path):
             folder_path = os.path.join(extract_path, folder)
 
@@ -69,8 +81,12 @@ def main():
     # Parse Command Line
     parser = argparse.ArgumentParser(description="Create data manager JSON.")
     parser.add_argument("--out", dest="output", action="store", help="JSON filename")
-    parser.add_argument("--version", dest="version", action="store", help="Version of the DB")
-    parser.add_argument("--database-type", dest="db_type", action="store", help="Db type")
+    parser.add_argument(
+        "--version", dest="version", action="store", help="Version of the DB"
+    )
+    parser.add_argument(
+        "--database-type", dest="db_type", action="store", help="Db type"
+    )
     parser.add_argument(
         "--test",
         action="store_true",
@@ -85,7 +101,7 @@ def main():
     with open(args.output) as fh:
         params = json.load(fh)
 
-    print(params)
+    # print(params)
 
     workdir = params["output_data"][0]["extra_files_path"]
     os.mkdir(workdir)
@@ -97,6 +113,8 @@ def main():
     db_path = os.path.join(workdir, db_value)
     tmp_path = os.path.join(workdir, "tmp")
 
+    print(tmp_path)
+
     # create DB
     if args.test:
         url = DB_paths["test_lsu"]
diff --git a/data_managers/data_manager_mapseq/data_manager/mapseq_db_fetcher.xml b/data_managers/data_manager_mapseq/data_manager/mapseq_db_fetcher.xml
index 0bfaa71bd77..383b3020ce0 100644
--- a/data_managers/data_manager_mapseq/data_manager/mapseq_db_fetcher.xml
+++ b/data_managers/data_manager_mapseq/data_manager/mapseq_db_fetcher.xml
@@ -17,8 +17,8 @@
     </command>
     <inputs>
         <!-- <param name="test_data_manager" type="hidden" /> -->
-        <param name="test_data_manager_v5" type="boolean" truevalue="--test" falsevalue="" checked="False" label="Download minimal test DB and create mock data table entry." />
-        <param name="test_data_manager_v6" type="boolean" truevalue="--test" falsevalue="" checked="False" label="Download minimal test DB and create mock data table entry." />
+        <param name="test_data_manager_v5" type="boolean" truevalue="--test" falsevalue="" checked="False" label="Download minimal test DB and create mock data table entry for v5." />
+        <param name="test_data_manager_v6" type="boolean" truevalue="--test" falsevalue="" checked="False" label="Download minimal test DB and create mock data table entry for v6." />
     
         <param name="database_type" type="select" multiple="false" label="Database Type">
             <option value="mgnify_v6_lsu">MGnify LSU (v6.0)</option>

From 6e78a45810c151401c5d7e27be468d7d60625c20 Mon Sep 17 00:00:00 2001
From: paulzierep <paul.zierep@googlemail.com>
Date: Tue, 3 Dec 2024 13:28:42 +0100
Subject: [PATCH 2/2] rm unused print

---
 .../data_manager/data_manager_fetch_mapseq_db.py              | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/data_managers/data_manager_mapseq/data_manager/data_manager_fetch_mapseq_db.py b/data_managers/data_manager_mapseq/data_manager/data_manager_fetch_mapseq_db.py
index 34180971d48..40be5a9a343 100644
--- a/data_managers/data_manager_mapseq/data_manager/data_manager_fetch_mapseq_db.py
+++ b/data_managers/data_manager_mapseq/data_manager/data_manager_fetch_mapseq_db.py
@@ -101,8 +101,6 @@ def main():
     with open(args.output) as fh:
         params = json.load(fh)
 
-    # print(params)
-
     workdir = params["output_data"][0]["extra_files_path"]
     os.mkdir(workdir)
 
@@ -113,8 +111,6 @@ def main():
     db_path = os.path.join(workdir, db_value)
     tmp_path = os.path.join(workdir, "tmp")
 
-    print(tmp_path)
-
     # create DB
     if args.test:
         url = DB_paths["test_lsu"]