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

#include "amsbtok.h"

/* this should always be defined in <stdio.h>, but just in case... */
#ifndef BUFSIZ
#define BUFSIZ 4096
#endif

/* range for one-byte tokens */
#define MIN_STD_TOK 0x80 /* END */
#define MAX_STD_TOK 0xf8 /* <   */

/* range for 2nd byte of two-byte tokens (1st is always 0xff) */
#define MIN_EXT_TOK 0xa3 /* SGN */
#define MAX_EXT_TOK 0xc5 /* STACK */

/* AMSB's tokens for "!" and "'", used to introduce comments */
#define TOK_SQUOTE 0x9a
#define TOK_BANG   0x9b

/* good old Atari EOL character */
#define EOL 0x9b

/* every MS BASIC I ever saw had the same line number limit. it's
   kind of arbitrary: why not allow the full range, up to 65535? */
#define MAX_LINENO 63999

/* a program bigger than this can't possibly fit into memory,
   even with cart-based AMSB2 and no DOS loaded. */
#define MAX_PROGLEN 30000

/* there should never be a valid line of BASIC longer than MAX_LINE_LEN
   bytes, since there would be no way to enter it in the editor. AMSB
   uses the standard E: device, which limits you to 4 screen lines, or
   max 120 bytes. there *cannot* be a line longer than MAX_LINE_LEN_HARD,
   because AMSB would crash when you try to LOAD such a program. */
#define MAX_LINE_LEN 0x80
#define MAX_LINE_LEN_HARD 0xff

/* a program whose header has a length less than MIN_PROGLEN can't be
   a real AMSB program. EMPTY_PROGLEN is what you get if you
   SAVE when there's no program in memory (right after boot or
   a NEW). The minimum size for a program that actually contains
   code seems to be 5 (for 10 PRINT) */
#define MIN_PROGLEN 5
#define EMPTY_PROGLEN 2

/* an EOL address below this has to be an error, since
   this is the lowest MEMLO can ever be. */
#define MIN_PTR 0x0700

/* an EOL address higher than this has to be an error, since
	it would overlap the GR.0 display list or the ROMs at $c000 */
#define MAX_PTR 0xbc1f

const char *self;

char pipe_command[BUFSIZ + 1] = { "a8cat" };

int verbose =     0;           /* -v */
int raw_output =  0;           /* -a */
int check_only =  0;           /* -c */
int startline =   0;           /* -r */
int endline   =   MAX_LINENO;  /* -r */


int need_pclose = 0;
int bytes_read = 0;
int warnings = 0;
int proglen = 0;

FILE *infile;
FILE *outfile;

void set_self(const char *argv0) {
   char *p;

   self = argv0;
   p = strrchr(self, '/');
   if(p) self = p + 1;
}

void die_with(const char *msg, int status) {
	fprintf(stderr, "%s: %s\n", self, msg);
	exit(status);
}

#define die(x) die_with(x,1)
#define die2(x) die_with(x,2)

unsigned char read_byte(void) {
	int c;

	c = fgetc(infile);

	if(c < 0)
		die2("unexpected EOF, file truncated?");

	bytes_read++;
	return (unsigned char)c;
}

int read_word(void) {
	int w;

	w = read_byte();
	w |= (read_byte() << 8);

	return w;
}

void read_header(void) {
	unsigned char b;

	b = read_byte();
	if(b) die2("not an AMSB file: first byte not $00");

	proglen = read_word();

	if(verbose)
		fprintf(stderr, "proglen == %d (%04x)\n", proglen, proglen);

	if(proglen > MAX_PROGLEN) {
		fprintf(stderr, "%s: not an AMSB file: too big (%d bytes), won't fit in Atari memory\n",
				self, proglen);
		exit(2);
	}

	if(proglen == EMPTY_PROGLEN) {
		fprintf(stderr, "%s: program length is 2, no code in file (SAVE after NEW)\n", self);
		warnings++;
	} else {
		if(proglen < MIN_PROGLEN) {
			fprintf(stderr, "%s: not an AMSB file: program size too small (%d). Atari BASIC file?\n",
					self, proglen);
			exit(2);
		}
	}
}

void unknown_token(int lineno, unsigned char byte, int ext) {
	fprintf(outfile, "<unknown %stoken ", (ext ? "extended " : ""));
	fprintf(outfile, "%s%02x>", (ext ? "$ff ": ""), byte);
}

int next_line(void) {
	static int last_lineno = -1;
	static int last_ptr = -1;
	int ptr, lineno, was_ff, in_string, offset, len;
	int printing;
	unsigned char byte;

	offset = bytes_read;

	/* pointer to last token on the line, offset by whatever MEMLO
		happened to be when the file was SAVEd. 0 means this is the
		last line. */
	ptr = read_word();
	if(!ptr) {
		if(verbose)
			fprintf(stderr, "end of program\n");
		return 0;
	}

	lineno = read_word();
	if(verbose)
		fprintf(stderr, "found line %d, offset %d, end-of-line %d\n", lineno, offset, ptr);

	printing = (lineno >= startline) && (lineno <= endline);

	if(ptr < MIN_PTR) {
		fprintf(stderr, "%s: line %d: EOL address $%04x too low (<$%04x)\n",
			self, lineno, ptr, MIN_PTR);
		warnings++;
	} else if(ptr >= MAX_PTR) {
		fprintf(stderr, "%s: line %d: EOL address $%04x too high (>$%04x)\n",
			self, lineno, ptr, MAX_PTR);
		warnings++;
	}

	if(last_ptr != -1) {
		if(ptr <= last_ptr) {
			fprintf(stderr, "%s: line %d: EOL address $%04x <= previous $%04x\n",
				self, lineno, ptr, last_ptr);
			warnings++;
		}
	}

	if(lineno <= last_lineno) {
		fprintf(stderr, "%s: line number out of order (%d <= %d)\n",
				self, lineno, last_lineno);
		warnings++;
	}

	if(lineno > MAX_LINENO) {
		fprintf(stderr, "%s: line number out range (%d > %d)\n",
				self, lineno, MAX_LINENO);
		warnings++;
	}

	last_lineno = lineno;

	/* note that AMSB always puts a space after the line number in LIST */
	if(printing) fprintf(outfile, "%d ", lineno);

	was_ff = 0;
	in_string = 0;

	/* walk and print the tokens. when we hit a null byte, we're done. */
	while(1) {
		byte = read_byte();

		if(in_string) {
			if(byte == 0x00 || byte == '|') {
				/* end of string */
				in_string = 0;
				if(printing) putc('"', outfile);

				/* if we read a null, that means the line ends with a string
					that's missing its closing double-quote. */
				if(byte == 0x00) {
					break;
				} else {
					continue;
				}
			}
			if(printing) putc(byte, outfile);
		} else if(byte == ':') {
			/* don't print the colon if the next token is a ! or ' for a comment */
			unsigned char next = read_byte();
			if( !(next == TOK_SQUOTE || next == TOK_BANG) )
				if(printing) putc(byte, outfile);
			ungetc(next, infile);
			bytes_read--;
		} else if(byte == '"') {
			/* strings start but *don't end* with a double-quote */
			in_string = 1;
			if(printing) putc(byte, outfile);
		} else if(was_ff) {
			if(byte >= MIN_EXT_TOK && byte <= MAX_EXT_TOK) {
				if(printing) fputs(ext_tokens[byte - MIN_EXT_TOK], outfile);
			} else {
				if(printing) unknown_token(lineno, byte, 1);
				warnings++;
			}
			was_ff = 0;
		} else if(byte == 0xff) {
			was_ff = 1;
		} else if(byte >= MIN_STD_TOK && byte <= MAX_STD_TOK) {
			if(printing) fputs(std_tokens[byte - MIN_STD_TOK], outfile);
		} else if(byte >= 0x80) {
			if(printing) unknown_token(lineno, byte, 0);
			warnings++;
		} else {
			if(!byte) break;
			if(byte < 0x20) {
				/* ATASCII graphics outside of a string */
				fprintf(stderr, "%s: line %d has character %d outside of a string, "
						"maybe Atari BASIC?\n",
						self, lineno, byte);
				warnings++;
			}
			if(printing) putc(byte, outfile);
		}
	}

	len = bytes_read - offset;

	if(verbose) {
		fprintf(stderr, "      line %d length: %d\n", lineno, len);
	}

	if(len > MAX_LINE_LEN) {
		int hard = len > MAX_LINE_LEN_HARD;
		fprintf(stderr, "%s: line %d is %s long (length %d > %d)\n",
			self,
			lineno,
			hard ? "impossibly" : "supiciously",
			len,
			hard ? MAX_LINE_LEN : MAX_LINE_LEN_HARD);
	}

	if(last_ptr != -1) {
		int plen = ptr - last_ptr;
		if(len != plen) {
			fprintf(stderr, "%s: line %d: EOL address doesn't match actual line length %d\n",
					self, lineno, len);
			warnings++;
		}
	}

	last_ptr = ptr;

	if(printing) putc(EOL, outfile);

	return 1;
}

void print_help(void) {
	printf("%s v" VERSION " - detokenize Atari Microsoft BASIC files\n", self);
	puts("By B. Watson <urchlay@slackware.uk>, released under the WTFPL");
	printf("Usage: %s [-a] [-v] [-h] [-i] [-u] [-t] [-m] [-s] [-r *start,end*] [file]\n", self);
	puts("  -a: raw ATASCII output");
	puts("  -c: check only (no listing)");
	puts("  -v: verbose");
	puts("  -r: only list lines numbered from *start* to *end*");
	puts("  --help, -h: print this help and exit");
	puts("  --version:  print version number and exit");
	puts("  -i -u -t -m -s: passed to a8cat (try 'a8cat -h')");
	puts("file must be a tokenized (SAVEd) AMSB file. if not given, reads from stdin.");
}

void version(void) {
	printf("%s " VERSION "\n", self);
}

void get_line_range(const char *arg) {
	int val = 0, comma = 0;
	const char *p = arg;

	while(*p) {
		if(*p >= '0' && *p <= '9') {
			val *= 10;
			val += *p - '0';
			if(val > MAX_LINENO) {
				fprintf(stderr, "invalid line number for -r (range is 0-%d)\n", MAX_LINENO);
				exit(1);
			}
		} else if(*p == ',' || *p == '-') {
			if(comma) die("invalid argument for -r (too many commas)");
			comma++;
			startline = val;
			val = 0;
		} else {
			if(comma) die("invalid argument for -r (only digits and comma allowed)");
		}
		p++;
	}

	if(comma)
		endline = val ? val : MAX_LINENO;
	else
		startline = endline = val;

	if(endline < startline)
		die("invalid argument for -r (start > end)");
}

void parse_args(int argc, char **argv) {
	char tmp[10];
	int opt;

	if(argc >= 2) {
		if(strcmp(argv[1], "--help") == 0) {
			print_help();
			exit(0);
		} else if(strcmp(argv[1], "--version") == 0) {
			version();
			exit(0);
		}
	}

	while( (opt = getopt(argc, argv, "r:cvaiutmsh")) != -1) {
		switch(opt) {
			case 'c': check_only = 1; break;
			case 'a': raw_output = 1; break;
			case 'h': print_help(); exit(0);
			case 'v': verbose = 1; break;
			case 'r': get_line_range(optarg); break;
			case 'i':
			case 'u':
			case 't':
			case 'm':
			case 's':
				if(strlen(pipe_command) > (BUFSIZ - 10))
					die("too many a8cat options");
				sprintf(tmp, " -%c", opt);
				strcat(pipe_command, tmp);
				break;
			default: print_help(); exit(1);
		}
	}

	if(optind >= argc) {
		if(isatty(fileno(stdin))) {
			fprintf(stderr, "%s: can't read binary data from a terminal\n", self);
			print_help();
			exit(1);
		}
		freopen(NULL, "rb", stdin);
		infile = stdin;
	} else {
		infile = fopen(argv[optind], "rb");
		if(!infile) {
			fprintf(stderr, "%s: ", self);
			perror(argv[optind]);
			exit(1);
		}
	}
}

void open_output() {
	if(check_only) {
		outfile = freopen("/dev/null", "wb", stdout);
		if(!outfile) {
			fprintf(stderr, "%s: ", self);
			perror("/dev/null");
			exit(1);
		}
		if(verbose)
			fprintf(stderr, "using /dev/null for output (check_only)\n");
	} else if(raw_output) {
		if(isatty(fileno(stdout)))
			die("refusing to write raw ATASCII to a terminal");
		outfile = stdout;
		if(verbose)
			fprintf(stderr, "using stdout for output\n");
	} else {
		if(verbose)
			fprintf(stderr, "using pipe for output: %s\n", pipe_command);
		outfile = popen(pipe_command, "w");
		if(!outfile) {
			perror(pipe_command);
			exit(1);
		}
		need_pclose = 1;
	}
}

int main(int argc, char **argv) {
	int rv = 0, linecount = 0;

	set_self(argv[0]);

	parse_args(argc, argv);

	open_output();

	read_header();

	while(next_line())
		linecount++;

	if(verbose) {
		fprintf(stderr, "read %d bytes\n", bytes_read);
		fprintf(stderr, "listed %d lines\n", linecount);
	}

	if(!linecount) {
		fprintf(stderr, "%s: no lines of code in program\n", self);
		warnings++;
	}

	if(proglen == (bytes_read - 3)) {
		if(verbose)
			fprintf(stderr, "file size matches proglen\n");
	} else {
		fprintf(stderr, "%s: actual program size doesn't match program size in header\n", self);
		warnings++;
	}

	if(fgetc(infile) != EOF) {
		fprintf(stderr, "%s: trailing garbage at end of file\n", self);
		warnings++;
	}

	if(warnings) {
		fprintf(stderr, "%s: file has %d warnings\n", self, warnings);
		rv = 2;
	}

	if(need_pclose) {
		int got = pclose(outfile);
		if(verbose)
			fprintf(stderr, "return value from pipe is %d\n", got);
		if(got != 0) {
			die("a8cat child process failed, do you have a8cat on your PATH?");
		}
	}

	exit(rv);
}