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

[aviruthen]

Description

Fix TypeError when dependencies is None in get_processing_code_hash

Fixes aws#5518

Problem

When dependencies is passed as None to get_processing_code_hash(), the function attempts to concatenate a list with None (e.g., [code, source_dir] + dependencies), which raises:

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

The same pattern was already fixed in get_training_code_hash() with if dependencies: guards, but get_processing_code_hash() was left unprotected.

Solution

Added a defensive dependencies = dependencies or [] at the start of get_processing_code_hash() to normalize None to an empty list before any list concatenation occurs.

Testing

Added two new test cases:

• test_get_processing_code_hash_with_none_dependencies — verifies no TypeError when source_dir is provided with dependencies=None
• test_get_processing_code_hash_code_only_with_none_dependencies — verifies no TypeError when only code is provided with dependencies=None

Existing tests for get_training_code_hash with dependencies=None continue to pass.

Related Issue

Fixes aws#5518

Changes Made

The bug occurs when SourceCode.requirements is not set (defaults to None), and the pipeline's get_code_hash() function passes this None value as dependencies to get_training_code_hash(). In the released v3.3.1, get_training_code_hash() directly concatenates [source_dir] + dependencies which fails with TypeError: can only concatenate list (not 'NoneType') to list. The fix in the current repo already guards this with if dependencies: checks in get_training_code_hash(). However, get_processing_code_hash() has the same fragile pattern — it directly concatenates + dependencies without a None guard, relying on the caller to always pass a list. This should also be hardened with a defensive default.

AI-Generated PR

This PR was automatically generated by the PySDK Issue Agent.

• Confidence score: 92%
• 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]

🤖 AI Code Review

The fix addresses a real bug where dependencies=None causes a TypeError in get_processing_code_hash(), but the diff appears to have a critical issue — the if source_dir: guard was removed, which would break the function's control flow. The tests are well-structured and cover the relevant cases.

[aviruthen]

Critical bug in the diff: It appears the if source_dir: line was removed (shown as - if source_dir: in the diff), but the indented block below it still expects that conditional guard. This would cause the code under the old if source_dir: block to execute unconditionally, meaning when source_dir=None, the function would attempt urlparse(None) instead of falling through to the code-only path.

The fix should be:

dependencies = dependencies or []

if source_dir:
    source_dir_url = urlparse(source_dir)

Please verify the diff is correct — the if source_dir: line must be preserved. If this is a diff rendering issue, please confirm the actual file still contains the conditional.

[aviruthen]

Minor: This test uses tempfile.NamedTemporaryFile with manual os.unlink cleanup, while the test above uses tempfile.TemporaryDirectory which handles cleanup automatically. Consider using TemporaryDirectory here as well for consistency, or at minimum use a pytest.fixture / context manager pattern to ensure cleanup even if an assertion fails before finally. That said, the try/finally does handle it — this is a minor style suggestion.