Print MSX-BASIC tokenized binary file to stdout in plain texts. Fixed messy indents, typos and bugs in original version. Tested with Turbo C++1/DJGPP GCC 7.3 and some BAS files of opensource Japanese homebrew games.

msxlist.c

/****************************************************************************\
 list v0.3

 This programm will convert a MSX-BASIC tokenized file into an ascii file.
 syntax: list [inputfile]
 To save converted texts, just pipe output with '> outputfile'.

 Vincent van Dam
 (vandam@ronix.ptf.hro.nl)

 //Please notify me in case of modifications.
 Original modified: 26/6/97
 Last modified by sin_sin on 2022-5-5
\****************************************************************************/

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

/* Uncomment this to escape double-byte GRAPH chars in \xXX form */
/* #define ESCGRAPH 1 */

/* MSX-BASIC tokens */
const char *tokens_1[128] = {
    "","END","FOR","NEXT","DATA","INPUT","DIM","READ","LET","GOTO","RUN","IF",
       "RESTORE","GOSUB","RETURN","REM",                           /* 80-8F */
    "STOP","PRINT","CLEAR","LIST","NEW","ON","WAIT","DEF","POKE","CONT",
       "CSAVE","CLOAD","OUT","LPRINT","LLIST","CLS",               /* 90-9F */
    "WIDTH","ELSE","TRON","TROFF","SWAP","ERASE","ERROR","RESUME","DELETE",
       "AUTO","RENUM","DEFSTR","DEFINT","DEFSNG","DEFDBL","LINE",  /* A0-AF */
    "OPEN","FIELD","GET","PUT","CLOSE","LOAD","MERGE","FILES","LSET","RSET",
       "SAVE","LFILES","CIRCLE","COLOR","DRAW","PAINT",            /* B0-BF */
    "BEEP","PLAY","PSET","PRESET","SOUND","SCREEN","VPOKE","SPRITE","VDP",
       "BASE","CALL","TIME","KEY","MAX","MOTOR","BLOAD",           /* C0-CF */
    "BSAVE","DSKO$","SET","NAME","KILL","IPL","COPY","CMD","LOCATE","TO",
       "THEN","TAB(","STEP","USR","FN","SPC(",                     /* D0-DF */
    "NOT","ERL","ERR","STRING$","USING","INSTR","'","VARPTR","CSRLIN","ATTR$",
       "DSKI$","OFF","INKEY$","POINT",">","=",                     /* E0-EF */
    "<","+","-","*","/","^","AND","OR","XOR","EQV","IMP","MOD","\\","","",
       "'doublebyte-token'"                                        /* F0-FF */

};

/* double bytes tokens */
const char *tokens_2[128] = {
    "","LEFT$","RIGHT$","MID$","SGN","INT","ABS","SQR","RND","SIN","LOG",
       "EXP","COS","TAN","ATN","FRE",                             /* 80-8F */
    "INP","POS","LEN","STR$","VAL","ASC","CHR$","PEEK","VPEEK","SPACE$",
       "OCT$","HEX$","LPOS","BIN$","CINT","CSNG",                 /* 90-9F */
    "CDBL","FIX","STICK","STRIG","PDL","PAD","DSKF","FPOS","CVI","CVS","CVD",
       "EOF","LOC","LOF","MKI$","MKS$",                           /* A0-AF */
    "MKD$","","","","","","","","","","","","","","","",          /* B0-BF */
    "","","","","","","","","","","","","","","","",              /* C0-CF */
    "","","","","","","","","","","","","","","","",              /* D0-DF */
    "","","","","","","","","","","","","","","","",              /* E0-EF */
    "","","","","","","","","","","","","","","",""               /* F0-FF */
};

/* read a 2-byte word */
unsigned int readword(FILE *fd) {
    return getc(fd)+256*getc(fd);
}

/* print quote/remark/data char */
int printchar(FILE *fd) {
    unsigned char ch=getc(fd);
    if (ch==0) return -1;
    if (ch!=1) printf("%c",ch);
    /* printable GRAPH char on MSX only */
    if (ch==1)
#ifdef ESCGRAPH
        printf("\\x%X",getc(fd));
#else
        printf("%c",ch);
#endif
    return 0;
}

/* get exponent of floating point number */
int getexp(FILE *fd) {
    int exp;
    exp=getc(fd);
    if (exp>127) { exp=exp-128; printf("-"); }
    return exp-64;
}

/* print single/double precission float point */
void printfloat(FILE *fd, int digits, char suffix) {
    unsigned char lo,hi,str[21];
    int exp,i,j,flag,isint;
    exp=getexp(fd);
    lo=0;
    memset(str,'0',20);
    str[20]=0;
    isint=0;

    /* get digits */
    for (i=j=0; i<digits*2; i++) {
        if (i%2==0) {
            hi=getc(fd);
            lo=hi&0x0f; hi=hi>>4;
            str[j++]='0'+hi;
        } else
            str[j++]='0'+lo;
    }
    str[j]=0;

    /* insert dot */
    if (exp<=-64) {
        str[0]='0';
        str[1]='.';
        exp=0;
    } else if (exp>digits*2 || exp<-1) {
        for (i=(digits*2+2);i>1;i--) str[i]=str[i-1];
        str[1]='.';
        exp--;
    } else if (exp<0) {
        for (i=(digits*2+3);i>1;i--) str[i]=str[i-2];
        str[0]='.';
        str[1]='0';
    } else {
        for (i=(digits*2+2);i>exp;i--) str[i]=str[i-1];
        str[exp]='.';
    }

    /* delete extra zero's */
    for (i=flag=0; i<digits*2+2; i++)
        if (str[i]=='0') {
            if (flag==0) flag=i;
        } else {
            if (str[i]==0) break; else flag=0;
        }
    if (flag!=0) {
        if (str[flag-1]=='.') {
            str[flag-1]=0;
            isint=1;
        } else str[flag]=0;
    }

    printf("%s",str);
    /* scientific notation */
    if (exp>digits*2 || exp<-1) {
        printf("E%+02d",exp);
    /* print literal suffix if not E-notation */
    } else if (isint || digits>3) {
        printf("%c",suffix);
    }
}

int main(int argc, char *argv[]) {
    FILE *fd;
    unsigned char ch;
    unsigned char *token;
    unsigned int linenumber;
    unsigned int lineaddress;
    int goon;
    int quoted;
    int datastr;

    /* check argument */
    if (argc!=2) {
        puts("Usage: LIST filename.BAS");
        return 1;
    }
    /* open file */
    if ((fd=fopen(argv[1],"rb"))==NULL) {
        fprintf(stderr,"Cannot open file - %s\n",argv[1]);
        exit(1);
    }

    /* read header and check if it is a BASIC program */
    ch=getc(fd);
    if (ch!=0xff) {
        fprintf(stderr,"File not a MSX-BASIC program - %s\n",argv[1]);
        exit(1);
    }

    /* read line number and address */
    /* address tells where the next line begins, high 7th bit indicates 'loaded', always 1 */
    lineaddress=readword(fd);
    while (lineaddress!=0) {
        linenumber=readword(fd);
        /* MSX-BASIC line number 0~65529 */
        printf("%u ",linenumber);
        /* read a line of BASIC program */
        goon=1;
        quoted=0;
        datastr=0;
        while(goon) {
            ch=getc(fd);
            if (ch==0) goon=0;
            else if (ch==0x22) {                 /* double quotes " */
                printf("%c",ch);
                quoted=1-quoted;
            } else if (quoted==1) {              /* keep output quoted chars */
                ungetc(ch,fd);
                printchar(fd);
            } else if (datastr==1) {             /* keep output data until : */
                if (ch==0x3a) datastr=0;
                ungetc(ch,fd);
                printchar(fd);
            } else if (ch>=0x80 && ch<0xff) {    /* single byte token */
                token=(unsigned char *)tokens_1[ch-0x80];
                if (*token==0)
                    printf("'0x%x?'",ch);
                else
                    printf("%s",token);
                if (ch==0x84) datastr=1;         /* DATA found */
                if (ch==0x8f) {                  /* REM found, output remaining */
                    while (!printchar(fd));
                    goon=0;
                }
            } else if (ch==0xff) {               /* double bytes token */
                ch=getc(fd);
                token=(unsigned char *)tokens_2[ch-0x80];
                if (*token==0)
                    printf("'0xff%x?'",ch);
                else
                    printf("%s",token);
            } else if (ch==0x3a) {               /* colon : */
                ch=getc(fd);
                if (ch==0xa1)
                    printf("ELSE");
                else if (ch==0x8f) {             /* REM variant */
                    ch=getc(fd);
                    if (ch==0xe6) {              /* single quote mark */
                        printf("'");
                        while (!printchar(fd));  /* output remaining chars */
                        goon=0;
                    } else if (ch==0) {          /* unknown case */
                        printf(":REM");
                        while (!printchar(fd));
                        goon=0;
                    } else {
                        printf("':0x%x?'",ch);
                    }
                } else {
                    printf(":");
                    ungetc(ch,fd);
                }
            } else if (ch==0x0b)                 /* octal number */
                printf("&O%o",readword(fd));
            else if (ch==0x0c)                   /* hexadecimal number */
                printf("&H%X",readword(fd));
            else if (ch==0x0d)                   /* line number: address */
                printf("'unsupport line ref'");
            else if (ch==0x0e)                   /* line number */
                printf("%u",readword(fd));
            else if (ch==0x0f)                   /* integer: byte */
                printf("%u",getc(fd));
            else if (ch==0x1c)                   /* integer: word */
                printf("%u",readword(fd));
            else if (ch==0x1d) {                 /* single precission float */
                printfloat(fd,3,'!');
            } else if (ch==0x1f) {               /* double precission float */
                printfloat(fd,7,'#');
            } else if (ch>=0x11 && ch<=0x1a)
                printf("%u",ch-0x11);
            else                                 /* literal ascii char, including &B number */
                printf("%c",ch);
        }
        printf("\n");
        lineaddress=readword(fd);
    }

    fclose(fd);
    return 0;
}