fix: improve error handling in similarity tools

[lxcxjxhx]

Description

This PR fixes several error handling issues in the similarity tools (slither-simil) that could cause crashes or misleading error messages.

Problems Fixed

1. Path Concatenation Bug (encode.py:96)

• Issue: Incorrect path construction using string concatenation: x + '/'.join(y) + '/' + f
• Fix: Use os.path.join(x, f) for proper cross-platform path handling
• Impact: Prevents path errors on Windows and ensures consistent path separators

2. Unsafe Attribute Access (train.py:52, test.py:50, plot.py:82)

• Issue: Direct access to �rgs.filename in exception handlers could raise AttributeError
• Fix: Use getattr(args, 'filename', '') for safe attribute access
• Impact: Prevents secondary crashes when reporting errors

3. Potential KeyError (encode.py:228)

• Issue: Direct dictionary access kwargs['solc'] could raise KeyError
• Fix: Use kwargs.get('solc', 'unknown') for safe access
• Impact: Provides graceful error reporting when solc parameter is missing

Testing

• Verified syntax correctness
• No functional behavior changes for valid inputs
• Error messages now more robust for edge cases

Note

These are defensive programming improvements that enhance tool stability. All changes are backward compatible.

---

I apologize for any inconvenience caused by previous submissions. This is a focused fix for error handling in similarity tools.

Local environment limitations: Dependency constraints (missing wn module), relying on CI/CD automated testing.

cc @crytic maintainers

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}