fix: Pipeline TypeError: can only concatenate list (not "NoneType") to list Using Sou (#5518)

[aviruthen]

Description

Fix TypeError when SourceCode.requirements is None in pipeline code hash computation

Problem

When creating a TrainingStep in a pipeline with a ModelTrainer whose SourceCode.requirements is None (the default), the pipeline build could fail with:

TypeError: can only concatenate list (not 'NoneType') to list

This occurred because get_processing_code_hash() concatenates the dependencies parameter directly with a list without first checking if it's None.

Root Cause

In get_processing_code_hash(), the dependencies parameter is concatenated with lists (e.g., [source_dir] + dependencies) without a guard against None values. While get_training_code_hash() already has if dependencies: guards, get_processing_code_hash() did not.

Fix

• Added dependencies = dependencies or [] at the top of get_processing_code_hash() to defensively normalize None to an empty list before any list concatenation occurs.
• Added comprehensive test cases covering None dependencies for both get_processing_code_hash() and get_training_code_hash().

Testing

• Added test_get_processing_code_hash_with_none_dependencies - verifies no TypeError with source_dir and None dependencies
• Added test_get_processing_code_hash_code_only_with_none_dependencies - verifies no TypeError with code only and None dependencies
• Added test_get_training_code_hash_with_none_dependencies_source_dir - verifies None dependencies with source_dir
• Added test_get_training_code_hash_with_none_dependencies_entry_point_only - verifies None dependencies with entry_point only

Related Issue

Fixes aws#5518

Changes Made

The bug occurs when SourceCode.requirements is None (not set by the user) and is passed as the dependencies parameter to get_training_code_hash(). While the current code in the repository has if dependencies: guards in get_training_code_hash(), the issue reporter's version (3.3.1) encounters a TypeError: can only concatenate list (not 'NoneType') to list at the hash_files_or_dirs([source_dir] + dependencies) call. The fix should be applied in two places: (1) in get_code_hash() to defensively pass None requirements as empty/no-dependency, and (2) in get_processing_code_hash() which has the same latent bug where dependencies could theoretically be None but is concatenated without a guard.

AI-Generated PR

This PR was automatically generated by the PySDK Issue Agent.

• Confidence score: 90%
• Classification: bug
• SDK version target: V3

Merge Checklist

• Changes are backward compatible
• Commit message follows prefix: description format
• Unit tests added/updated
• Integration tests added (if applicable)
• Documentation updated (if applicable)

[aviruthen]

Can you add a small comment explaining why you are adding a default value of [] if dependencies is None?