Re: PATCH 1/3: Fix buffer overflow and wrong behavior in url_encode

[Mikael Magnusson <mikachu@xxxxxxxxx>]

On Thu, May 7, 2026 at 11:35 AM Philippe Altherr
<philippe.altherr@xxxxxxxxx> wrote:
>
> "const unsigned char *in" looks very weird to me. Wouldn't it be possible to cast "*in" in the call to sprintf:
>
> out += sprintf(out, "%%%02X", (unsigned char)*in);
>
> or maybe
>
> out += sprintf(out, "%%%02X", *in & 0xFF);

It is indeed another possibility, which is why I asked. But if we do
that, we'd also need to cast it for the isalnum() call. However, and I
didn't notice this before, base64_encode already looks like this
(surely mangled by gmail):

static char*
base64_encode(const char *src, size_t len) {
    static const char* base64_table =
"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";

    const unsigned char *end = (unsigned char *)src + len;
    const unsigned char *in = (unsigned char *)src;

So I think I'll just use that pattern. (I wasn't sure if the cast
needed a const too, guess not).

-- 
Mikael Magnusson