fix filetype issue

Author: RyanJField

data_pipeline_api/fdp_utils.py

@@ -229,6 +229,24 @@ def post_storage_root(
         data["root"] = data["root"] + "/"
     return post_entry(url, "storage_root", data, token, api_version)
 
+def post_file_type(
+    url: str, data: dict, token: str, api_version: str = "1.0.0"
+) -> dict:
+    """
+    Internal wrapper function to return check if a file_type already exists and return it.
+    """
+    if not "extension" in data and data["extension"]:
+        raise ValueError("error file_type name not specified")
+    file_type_exists = get_entry(url= url, 
+        endpoint= "file_type", 
+        query= {"extension": data["extension"]}, 
+        api_version= api_version)
+    if file_type_exists : return file_type_exists[0]
+    return post_entry(url= url,
+        endpoint= "file_type",
+        data= data,
+        token= token,
+        api_version= api_version)
 
 def remove_local_from_root(root: str) -> str:
     """

data_pipeline_api/pipeline.py

@@ -82,11 +82,10 @@ def initialise(token: str, config: str, script: str) -> dict:
     config_location_url = config_location_response["url"]
 
     # Configure Yaml File Type
-    config_filetype_response = fdp_utils.post_entry(
+    config_filetype_response = fdp_utils.post_file_type(
         token=token,
         url=registry_url,
-        endpoint="file_type",
-        data={"name": "yaml", "extension": "yaml"},
+        data={"name": "YAML Document", "extension": "yaml"},
         api_version=api_version,
     )
     config_filetype_url = config_filetype_response["url"]
@@ -174,14 +173,14 @@ def initialise(token: str, config: str, script: str) -> dict:
     )
 
     script_location_url = script_location_response["url"]
+    script_file_type = os.path.basename(script).split(".")[-1]
 
     # TODO: Change to Batch?
     # Create Script File Type
-    script_filetype_response = fdp_utils.post_entry(
+    script_filetype_response = fdp_utils.post_file_type(
         token=token,
         url=registry_url,
-        endpoint="file_type",
-        data={"name": "py", "extension": "py"},
+        data={"name": "python submission script", "extension": script_file_type},
         api_version=api_version,
     )
 
@@ -478,23 +477,11 @@ def finalise(token: str, handle: dict) -> None:
 
             file_type = os.path.basename(new_path).split(".")[-1]
 
-            file_type_exists = fdp_utils.get_entry(
+            file_type_url = fdp_utils.post_file_type(
+                token=token,
                 url=registry_url,
-                endpoint="file_type",
-                query={"extension": file_type},
+                data={"name": file_type, "extension": file_type},
                 api_version=api_version,
-            )
-
-            if file_type_exists:
-                entry = fdp_utils.get_first_entry(file_type_exists)
-                file_type_url = entry["url"]
-            else:
-                file_type_url = fdp_utils.post_entry(
-                    token=token,
-                    url=registry_url,
-                    endpoint="file_type",
-                    data={"name": file_type, "extension": file_type},
-                    api_version=api_version,
                 )["url"]
 
             data_product_exists = fdp_utils.get_entry(

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "data_pipeline_api"
-version = "0.7.5"
+version = "0.7.6"
 description = "Python api to interact with the Fair Data Pipeline"
 authors = [
     "Ryan J Field <[email protected]>",