[GitHub] [spark] itholic commented on a diff in pull request #41711: [SPARK-44155] Adding a dev utility to improve error messages based on LLM

["itholic (via GitHub)" <gi...@apache.org>]

itholic commented on code in PR #41711:
URL: https://github.com/apache/spark/pull/41711#discussion_r1260479005


##########
dev/error_message_refiner.py:
##########
@@ -0,0 +1,265 @@
+#!/usr/bin/env python3
+
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements.  See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License.  You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+"""
+Utility for refining error messages based on LLM.
+
+Usage:
+    python error_message_refiner.py <error_class> [--gpt_version=<version>]
+
+Arguments:
+    <error_class>           Required.
+                            The name of the error class to refine the messages for.
+                            The list of error classes is located in
+                            `core/src/main/resources/error/error-classes.json`.
+
+Options:
+    --gpt_version=<version> Optional.
+                            The version of Chat GPT to use for refining the error messages.
+                            If not provided, the default version("gpt-3.5-turbo") will be used.
+
+Example usage:
+    python error_message_refiner.py CANNOT_DECODE_URL --gpt_version=gpt-4
+
+Description:
+    This script refines error messages using the LLM based approach.
+    It takes the name of the error class as a required argument and, optionally,
+    allows specifying the version of Chat GPT to use for refining the messages.
+
+    Options:
+        --gpt_version: Specifies the version of Chat GPT.
+                       If not provided, the default version("gpt-3.5-turbo") will be used.
+
+    Note:
+    - Ensure that the necessary dependencies are installed before running the script.
+    - Ensure that the valid API key is entered in the `api-key.txt`.
+    - The refined error messages will be displayed in the console output.
+    - To use the gpt-4 model, you need to join the waitlist. Please refer to
+      https://help.openai.com/en/articles/7102672-how-can-i-access-gpt-4 for more details.
+"""
+
+import argparse
+import json
+import openai
+import re
+import subprocess
+import random
+import os
+from typing import Tuple, Optional
+
+from sparktestsupport import SPARK_HOME
+
+PATH_TO_ERROR_CLASS = f"{SPARK_HOME}/core/src/main/resources/error/error-classes.json"

Review Comment:
   Thanks for the comment!
   
   Yes, we can support PySpark error classes as well, but I believe we need separate script since the internal structure & usage of error classes is a bit different from JVM side. Let me made a follow-up for Python script if this PR get approved.
   
   For `SPARK_HOME`, since the script is already using the `SPARK_HOME` environment variable internally, I believe it would be more appropriate to set the value of Spark home through the `SPARK_HOME` rather than configuring it here when executing the script.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org