Fix off-by-two padding bug in set_string and empty string edge case

[gijzelaerr]

Summary

Fixes #479

set_string() left the last 2 character positions unpadded when writing a shorter string over a longer one, leaving stale data in the buffer. Additionally, both set_string() and set_fstring() failed to clear the first character position when writing an empty string.

Root cause analysis

Bug 1: Off-by-two in set_string padding (line 247)

# BEFORE (buggy)
for r in range(i + 1, bytearray_[byte_index] - 2):
    bytearray_[byte_index + 2 + r] = ord(" ")

The - 2 was an incorrect attempt to account for the 2-byte string header (max_size byte + length byte). However, the header offset is already handled by the byte_index + 2 + r indexing expression. The variable r is a relative index into the data area (positions 0 through max_size-1), so the range should go up to max_size, not max_size - 2.

Impact: When writing "ab" into a buffer with max_size=10, positions 8 and 9 (relative to the data area) were never cleared:

Expected: [10, 2, 'a', 'b', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ']
Actual:   [10, 2, 'a', 'b', ' ', ' ', ' ', ' ', ' ', ' ',  ?,  ?]
                                                             ^^  ^^
                                                       stale data!

Bug 2: Empty string edge case (both set_string and set_fstring)

i = 0
for i, c in enumerate(value):  # doesn't execute for empty string
    ...
for r in range(i + 1, ...):    # starts at 1, skipping position 0
    ...

When value is empty, the enumerate loop doesn't execute, so i stays at 0. The padding loop then starts from range(0 + 1, ...), skipping position 0 entirely.

For set_string, this bug was masked by get_string() which reads the length byte (0 for empty) and returns "" regardless. But for set_fstring, the stale character at position 0 would be visible through get_fstring().

Fix

1. Changed range(i + 1, bytearray_[byte_index] - 2) to range(len(value), bytearray_[byte_index]) in set_string
2. Changed range(i + 1, max_length) to range(len(value), max_length) in set_fstring

Using len(value) instead of i + 1 correctly handles both the normal case and the empty string edge case, and removing - 2 fixes the off-by-two.

Test plan

• All 326 existing tests pass
• Manual verification that padding is correct for "ab" with max_size=10
• Manual verification that empty string clears all positions
• Add unit tests that verify buffer contents after set_string (not just round-trip)

🤖 Generated with Claude Code