Bug `std::string::NullStringBase` does not work within a section.

[bwrsandman]

I'm hitting a limitation of NullString and by extension std::mem::read_unsigned when parsing data in a section.

The issue is when a NullString is encoded, it will try to measure the length based on the $ offset but using std::mem::read_unsigned but this function does not read in the current section.

DataType data[while(std::mem::read_unsigned($, sizeof(DataType)) != 0x00)];

Expected outcome: NullString's detected length should be based on the data in the current section
Actual outcome: NullString uses the offset in the main data to compute the length of string in a section.

hello.x.zip

Data:

00000000  57 68 6f 6f 70 73 21 00  00 00 00 00 00 00 00 00  |Whoops!.........|
00000010  48 65 6c 6c 6f 20 57 6f  72 6c 64 21 21 21 21 00  |Hello World!!!!.|
00000020  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
00000030

Pattern:

import std.mem;
import std.string;

std::mem::Section section = std::mem::create_section("MySection");
std::mem::copy_section_to_section(null, 0x10, section, 0, 0x20);
 
std::string::NullString nullstr @ 0x00 in section;
std::print("nullstr: {}", nullstr); // Expected "Hello World!", Actual "Hello W" which is the length of "Whoops"
char chararray[0x10] @ 0x00 in section;
std::print("chararr: {}", chararray); // Prints "Hello World!"

struct InSection {
    std::print("{}", $); // Prints 15
};
InSection a @ 15 in section;

[bwrsandman]

Using this as a workaround at the moment but it's limited to 1024 length strings

fn null_string_len(auto buf) {
    u32 i = 0;
    while (buf[i] != 0) {
        i += 1;
    }
    return i + 1;
};

struct WorkAroundNullString {
    u8 buf[1024] @ $ [[hidden]];
    char data[null_string_len(buf) - 1];
    u8 null_chr [[hidden]];
} [[sealed, format("std::string::impl::format_string"), transform("std::string::impl::format_string")]];