>>66575417
You're overcomplicating it. I tried to solve it in my head, and this is what I came up with. It should be good enough. Haven't tested this code, since I can't be bothered with the pregeneration, but it should work.
static[] is a big ass const char array, it should look something like this
"LATIN CAPITAL LETTER A\0\x94B\0\x94C\0" {...} "Z\0LEFT SQUARE BRACKET\0REVERSE SOLIDUS\0\x81IGHT SQUARE BRACKET\0CIRCUMFLEX ACCENT\0LOW LINE\0GRAVE ACCENT\0LATIN SMALL LETTER A\0\x82B\0\x82C\0" {...}

my notes:
[code]
type - full or continue
if continue, last string, overwrite from n, include null term

type = full (0000 0000) or continue (1xxx xxxx) -- write from n-bin(10000000)

[/code]
code:
[code]
#define HIGH 0x80
#define HARDCODED_VALUE (1024*1024)
char* data_start = malloc(HARDCODED_VALUE);
char* data = data_start;
const char* static = static_start;
if (data[0] > HIGH)
{	unsigned char n = data[0] - HIGH;
	strcpy(data, static);
	static += strlen(static);
	strcpy(data+n, static);
} else if (data [0] != '\0')
{	static += strlen(static);
	strcpy(data, static);
} else //null terminator
{	break;
}
size_t len = data-data_start; //replace HARDCODED_VALUE with this+1
[/code]
To index, first transform the codepoint so it's contiguous (e.g. equals line number in https://unicode.org/Public/UNIDATA/UnicodeData.txt), then do
[code]
const char* index(const char* data, unsigned int n)
{
    while (--codepoint, data += strlen(data))
        ;
    return data;
}
[/code]