/* carray.c : PUBLIC DOMAIN - Jon Mayo - August 30, 2006
 * - You may remove any comments you wish, modify this code any way you wish,
 *   and distribute any way you wish.*/
/* dumps binary or text into a fully escaped C array. 
 * useful for creating UTF-8 and UTF-16 string constants */

/* TODO:
 * accept a list of arguments to dump.
 * dump multiple files to multiple arrays, for a simple filesystem
 * optionally output a directory index holding all the file information */

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

enum { METHOD_STRING, METHOD_DECIMAL, METHOD_HEXDUMP, METHOD_CHARHEX };

static char *name = "data";
static int method = METHOD_STRING;
static int max_width = 78;

static void usage(const char *argv0) {
	printf("usage: %s\n", argv0);
	printf(
		"\t-s\t\tstring mode\n"
		"\t-d\t\tdecimal mode\n"
		"\t-x\t\thex mode\n"
		"\t-c\t\tchar and hex mode\n"
		"\t-w <width>\tselects the width to use (default=78, minimum=20)\n"
	, max_width
	);
	exit(EXIT_FAILURE);
}

/* I am not proud of this code, I prefer to use getopt() */
static int parse_args(int argc, char **argv) {
	int i,j;
	char ch;

	for(i=1;i<argc;i++) {
		if(argv[i][0]=='-') {
			for(j=1;argv[i][j];j++) {
				switch(ch=argv[i][j]) {
					case 'w': /* width */
						if(argv[i][j+1]) {
							max_width=atoi(argv[i]+(j+1));
							goto next_arg;
						} else if(i+1<argc) {
							max_width=atoi(argv[i+1]);
							goto next_arg;
						} else {
							usage(argv[0]);
						}
						break;
					case 'n': /* name */
						if(argv[i][j+1]) {
							name=argv[i]+(j+1);
							goto next_arg;
						} else if(i+1<argc) {
							name=argv[i+1];
							goto next_arg;
						} else {
							usage(argv[0]);
						}
					case 's': method=METHOD_STRING; break;
					case 'x': method=METHOD_HEXDUMP; break;
					case 'd': method=METHOD_DECIMAL; break;
					case 'c': method=METHOD_CHARHEX; break;
					case 'h':
					default:
						usage(argv[0]);
				}
			}
		}
	next_arg:
		;
	}
	return i;
}

/* style 1 - strings */
static void check_width(int *width, int adj) {
	*width+=adj;
	if(*width>=max_width) { printf("\"\n\t\""); *width=8+adj; }
}

static void dump_string(FILE *f) {
	int len;
	int width;
	int ch;
	len=0;
	
	printf("\t\"");
	width=8;
	while((ch=fgetc(f))!=EOF) {
		len++;
		if(isprint(ch)) {
			if(ch=='\\') {
				check_width(&width, 2);
				printf("\\\\");
			} else if(ch=='"') {
				check_width(&width, 2);
				printf("\\\"");
			} else {
				check_width(&width, 1);
				putc(ch, stdout);
			}
		} else {
			if(ch=='\0') {
				check_width(&width, 2);
				printf("\\0");
			} else if(ch=='\n') {
				check_width(&width, 2);
				printf("\\n");
			} else if(ch=='\r') {
				check_width(&width, 2);
				printf("\\r");
			} else if(ch=='\t') {
				check_width(&width, 2);
				printf("\\t");
			} else if(ch=='\a') {
				check_width(&width, 2);
				printf("\\a");
			} else if(ch=='\v') {
				check_width(&width, 2);
				printf("\\v");
			} else {
				check_width(&width, ch<010 ? 3 : ch<0100 ? 4 : 5);
				printf("\\0%o", ch);
			}
		}
	}
	printf("\";\n");

	printf("const int %s_len=%d;\n", name, len);
}

/* style 2 - decimal */
void dump_decimal(FILE *f) {
	int len;
	int ch;
	int width;
	int cw;
	printf("{");
	width=max_width;
	len=0;
	while((ch=fgetc(stdin))!=EOF) {
		len++;
		cw=ch<10 ? 2 : ch<100 ? 3 : 4; /* width of this entry */
		if(width+cw+1>=max_width) {
			printf("\n\t");
			width=8+cw;
		} else {
			printf(" ");
			width+=cw+1;
		}
		printf("%u,", ch);
	}
	printf("\n};\n");
	printf("const int %s_len=%d;\n", name, len);
}


/* style 3 - hexdump */
void dump_hex(FILE *f) {
	int len;
	int ch;
	int width;
	printf("{");
	width=max_width;
	len=0;
	while((ch=fgetc(stdin))!=EOF) {
		len++;
		width+=6;
		if(width>=max_width) {
			printf("\n\t");
			width=8+5;
		} else {
			printf(" ");
		}
		printf("0x%02x,", ch);
	}
	printf("\n};");
	printf("const int %s_len=%d;\n", name, len);
}

/* style 4 - charhex */
void dump_charhex(FILE *f) {
	int len;
	int ch;
	int width;
	int cw;
	printf("{");
	width=max_width;
	len=0;
	while((ch=fgetc(stdin))!=EOF) {
		len++;

		if(ch=='\'' || ch=='\\') {
			cw=5;
		} else if(isprint(ch)) {
			cw=4;
		} else {
			cw=5;
		}
		if(width+cw+1>=max_width) {
			printf("\n\t");
			width=8+cw;
		} else {
			printf(" ");
			width+=cw+1;
		}

		if(ch=='\'' || ch=='\\') {
			printf("'\\%c',", ch);
		} else if(isprint(ch)) {
			printf("'%c',", ch);
		} else {
			printf("0x%02x,", ch);
		}
	}
	printf("\n};\n");
	printf("const int %s_len=%d;\n", name, len);
}

int main(int argc, char **argv) {
	int optind;
	optind=parse_args(argc, argv);
	printf("const unsigned char %s[] =\n", name);

	if(max_width<20) {
		usage(argv[0]);
	}

	switch(method) {
		case METHOD_STRING:
			dump_string(stdin);
			break;
		case METHOD_DECIMAL:
			dump_decimal(stdin);
			break;
		case METHOD_HEXDUMP:
			dump_hex(stdin);
			break;
		case METHOD_CHARHEX:
			dump_charhex(stdin);
			break;
		default:
			usage(argv[0]);
	}
	return 0;
}