bug: stat size() uses character length instead of byte length for wide characters #408

Summary

size() uses Perl's length() to compute stat[7] (st_size). For strings containing characters above U+00FF (where the UTF8 flag is set), length() returns the character count, which is less than the byte count. Since stat(2) st_size is defined in bytes, this produces incorrect values.

Reproduction

use Test::MockFile qw(nostrict);
no warnings 'utf8';

my $content = "Hello \x{263A}";  # 7 chars, 9 bytes (U+263A = 3 bytes UTF-8)
my $f = Test::MockFile->file('/tmp/test', $content);

my @st = stat('/tmp/test');
print "stat size: $st[7]\n";       # prints 7 (wrong — should be 9)
use bytes;
print "byte length: ", length($f->contents), "\n";  # prints 9 (correct)

Expected behavior

stat[7] should report byte count (9), matching what a real filesystem returns.

Root cause

size() at line 2487:

return length $self->contents;

Should use bytes::length():

return bytes::length( $self->contents );

Same issue affects symlink size (byte length of target path).

Fix

PR #408

🤖 Created by Kōan from autonomous session