SoftFileLock._try_break_stale_lock() wraps its body in suppress(OSError) to handle I/O errors gracefully. However, if the lock file content is corrupted (e.g. non-numeric PID line), int(pid_str) raises ValueError which is not suppressed, crashing the lock acquisition.

This can happen when:

• The lock file was partially written (process killed mid-write)
• The file was corrupted by an external tool
• A different application wrote to the same path

import socket
from filelock import SoftFileLock

# Simulate a corrupted lock file with non-numeric PID
with open("test.lock", "w") as f:
    f.write(f"not_a_pid\n{socket.gethostname()}\n")

lock = SoftFileLock("test.lock")
lock.acquire()  # ValueError: invalid literal for int() with base 10: 'not_a_pid'

The fix adds ValueError to the suppress() call, so corrupted lock files are treated the same as unreadable ones — the stale lock check is skipped and acquisition proceeds normally.