FileRead() treats count parameter as bytes instead of characters for UTF-8 files

[miraged]

When reading from UTF-8 files (with or without a BOM) the count parameter is treated as the number of bytes rather than the number of characters. UTF-16 and ANSI work as expected. Best case you get less characters than expected, worst case you get partial bytes in the returned string. Tested on 3.3.10.2, 3.3.14.2 and 3.3.15.0.
Example is attached.

[miraged]

Repro

[BrewManNH]

Running that test file shows me something different.

It looks like StringLen is at fault here and not FileRead. If you do a ConsoleWrite right after the FileRead, and output the @extended you will see that it's always reading 7 characters/bytes like it's supposed to, but stringlen reports the wrong information. Look at the Starting and Ending offsets, and they're identical between the first and second tests.

[jchd18]

Furthermore, the poster uses files with BOM, so that shifts the byteread count.

[jchd18]

Current release/beta versions of AutoIt work correctly; the "repro" code is wrong.

This simple code

Local $f = "len.txt"
FileWrite($f, "€€€")
Local $s = FileRead($f)
ConsoleWrite(@extended & @TAB & StringLen($s) & @LF)
FileDelete($f)

correctly yields
9 3
since '€' uses 3 bytes in UTF8.