parse.c

// Parses a dies irae "script" extracted from memory and dumps the content text

#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>

#define and &&
#define or ||

#ifndef NULL
#define NULL 0
#endif

/*
begin audio association      # 07 00 08 00
audio file name              # 76 00 5F 00 73 00 79 00 30 00 30 00 ........
between audio and text       # 00 00
text (a begin quote)         # 0C 30
begin ruby text association  # 07 00 01 00
text (latin)                 # 50 00 61 00 74 00 65 00 72 00 20 00 ........
between text and ruby text   # 0A 00
ruby text (japanese)         # 29 59 6B 30 7E 30 57 30 7E 30 59 30 ........
end of ruby text association # 00 00
text (an end quote)          # 0D 30
end audio association        # 07 00 09 00
pause for input              # 07 00 06 00
pagefeed                     # 00 00

text (japanese)              # 00 30 15 20 15 20 11 5C 74 5E 02 30
pause for input              # 07 00 06 00
text (japanese)              # DE 30 EB 30 B3 30 88 30 8A 30 82 30 ........
begin ruby text association  # 07 00 01 00
text (kanji)                 # EF 83 62 59
between text and ruby text   # 0A 00
ruby text (hiragana)         # 4D 30 83 30 57 30 83 30
end ruby text association    # 00 00
text (japanese)              # 6A 30 53 4F AF 8E 6E 30 11 5C 74 5E ........
pause for input              # 07 00 06 00
newline                      # 0A 00
*/

/*
begin audio association      # 07 00 08 00
end audio association        # 07 00 09 00
pause for input              # 07 00 06 00
begin ruby text association  # 07 00 01 00
*/

void fread_or_die(void* a, size_t b, size_t c, void* d)
{
    int got = fread(a, b, c, d);
    if(feof(d)) {puts("feof"); exit(0);}
    if(ferror(d)) {puts("ferror"); exit(0);}
    if(got != c) {exit(0);}
}


FILE * script;
FILE * text;
uint16_t data;

void pull()
{
    fread_or_die(&data, sizeof(uint16_t), 1, script);
}
void push(uint16_t X)
{
    uint16_t data = X;
    fwrite(&data, sizeof(uint16_t), 1, text);
}

int main()
{
    script = fopen("test.bin", "rb");
    if(!script) return puts("failed to open test.bin"), 0;
    text   = fopen("test.txt", "wb+");
    if(!text)   return puts("failed to open test.txt"), 0;
    
    push(0xFEFF);
    
    uint8_t header[20]; // header? initial commands?
    fread(header, 0x20, 1, script);
    
    uint8_t commands[0x10000/8]; // one byte per 8 commands
    int audio_state = 0; // 0 = off, higher numbers = later stages
    int ruby_state = 0;
    int command_state = 0;
    while(1)
    {
        pull();
        
        if(command_state == 0 and data == 0x0007)
            command_state = 1;
        else if(command_state == 1)
        {
            if(data == 0x0001) // furigana
            {
                // 3008 〈
                // 3009 〉
                // 300A 《
                // 300B 》
                push(0x3008);
                while(1)
                {
                    pull();
                    if(data == 0x000A) break;
                    push(data);
                }
                push(0x3009);
                push(0x300A);
                while(1)
                {
                    pull();
                    if(data == 0x0000) break;
                    push(data);
                }
                push(0x300B);
            }
            if(data == 0x0008) // audio filename is not content text, skip
            {
                while(1)
                {
                    pull();
                    if(data == 0x0000) break;
                }
            }
            command_state = 0;
        }
        else
        {
            if(data < 0x0020)
            {
                if(data == 0x0000)
                {
                    push(0x000A);
                    push(0x000A);
                }
                else if(data == 0x000A)
                {
                    push(data);
                }
            }
            else
            {
                push(data);
            }
        }
    }
}