gsub has trouble with \0

[mqnc]

txt1 = "\0"
print((txt1:gsub("\0", "o")))
-- lua:o jit:oo

txt2 = "\0\1\2"
print((txt2:gsub("\0\1\2", "o")))
-- lua:o jit:oooo

txt3 = "\0\1"
print((txt3:gsub("[\0]", "o")))
-- lua:o jit: malformed pattern (missing ']')

This is probably an issue with string termination in C but it works fine in lua

[GitSparTV]

LuaJIT is compatible to Lua 5.1 which has the same problem, \0 can be used in patterns since Lua 5.2. Use %z.

[mqnc]

thank you!

[vanc]

LuaJIT is compatible to Lua 5.1 which has the same problem, \0 can be used in patterns since Lua 5.2. Use %z.

That doesn't seem to be the case. Lua 5.1.5 behaves the same as Lua 5.3 for the first two cases.

> txt1 = "\0"
> print((txt1:gsub("\0", "o")))
o
> txt2 = "\0\1\2"
> print((txt2:gsub("\0\1\2", "o")))
o
> txt3 = "\0\1"
> print((txt3:gsub("[\0]", "o")))
stdin:1: malformed pattern (missing ']')
stack traceback:
        [C]: in function 'gsub'
        stdin:1: in main chunk
        [C]: ?

For the 3rd case, Lua 5.3 printed a special character after 'o' which should be the "\1".

$ lua5.3
Lua 5.3.3  Copyright (C) 1994-2016 Lua.org, PUC-Rio
> txt3 = "\0\1"
> print((txt3:gsub("[\0]", "o")))
o�
>

The "%z" does seem work with LuaJIT. By replacing the "\0" in the patterns with "%z", LuaJIT behaves the same as Lua 5.3

[MikePall]

You need to inspect the bytes of the returned string to see what's actually happening. Your console doesn't show some of these characters.

The pattern string is terminated at the first "\0", so the pattern is empty. And a gsub with an empty pattern has weird semantics.

Oh, and string.format("%q") is partially broken in Lua 5.1, but fixed in LuaJIT -- so don't use that one to inspect the bytes.

But all of this is moot, since the Lua 5.1 manual clearly says A pattern cannot contain embedded zeros. So, don't do that.

[GitSparTV]

You can see actual content using this function

function showbytes(str)
    local comma = false
    for i = 1, #str do
        io.write(comma and ", " or "", str:sub(i, i), " (", str:byte(i, i), ")")
        comma = true
    end
    io.write("\n")
    io.flush()
end