#define STB_DS_IMPLEMENTATION
#include "stb_ds.h"

#include <stdio.h>
#include <ctype.h>
#include <string.h>
#include <stdint.h>
#include <stdbool.h>

#define STREQ(s1, s2) (strcmp((s1), (s2)) == 0)
#define STRMT(s) ((s)[0] == '\0')
#define LINE_LEN 100

typedef struct {
	char lab[10];
	char opc[4];
	char opr[10];
	bool ind;
	int psu;
} token_t;

token_t tokenize(char *);
int split(char *, char [4][10]);
void extfname(char *, char *);

int main(int argc, char** argv)
{
	struct { char *key; uint16_t value; } *hopc = NULL;
	struct { char *key; uint16_t value; } *hlab = NULL;
	struct { uint8_t key; uint16_t value; } *hopr = NULL;
	token_t tkn;
	FILE *inpf, *outf;
	char *lines, *tlines;
	char line[LINE_LEN], opc[4], out[10], ofname[10];
	uint16_t lc, ins, addr;
	uint8_t opr, i;
	bool unknown, end;

	if (argc < 2) {
		puts("no input file");
		return 1;
	}

	sh_new_strdup(hopc);
	sh_new_strdup(hlab);
	shdefault(hopc, -1);
	shdefault(hlab, -1);

	inpf = fopen("opcodes", "r");
	for (i = 25; i > 0; --i) {
		fscanf(inpf, "%s %x", opc, &addr);
		shput(hopc, opc, addr);
	}

	lines = (char *)malloc(LINE_LEN*100*sizeof(char));
	inpf = fopen(argv[1], "r");
	lc = 0, opr = 0, end = false;
	while (fgets(line, LINE_LEN, inpf) != NULL) {
		tkn = tokenize(line);
		if (STRMT(tkn.opc) && tkn.psu == 0)
			continue;

		if (!STRMT(tkn.lab)) {
			hmput(hopr, shget(hlab, tkn.lab), lc);
			shput(hlab, tkn.lab, lc);
		}

		switch (tkn.psu) {
		case 1:
			lc = strtol(tkn.opr, NULL, 16);
			continue;
		case 2:
			ins = strtol(tkn.opr, NULL, 10);
			break;
		case 3:
			ins = strtol(tkn.opr, NULL, 16);
			break;
		case 4:
			end = true;
		}
		if (end) break;

		if (!STRMT(tkn.opc)) {
			ins = shget(hopc, tkn.opc);
			if (tkn.ind) ins += 8;
			if (!STRMT(tkn.opr)) {
				ins = ins << 12;
				addr = shget(hlab, tkn.opr);
				if (addr == 0xffff) {
					unknown = true;
					addr = opr;
					shput(hlab, tkn.opr, opr++);
				}
				ins = ins | addr;
			}
		}

		sprintf(out, "%03x %04x\n", lc, ins);
		if (unknown) {
			out[5] = '?';
			unknown = false;
		}

		strcat(lines, out);
		lc++;
	}
	fclose(inpf);
	shfree(hopc);
	shfree(hlab);

	tlines = lines;
	while ((tlines = strchr(tlines, '?')) != NULL) {
		opr = strtol(++tlines, NULL, 16);
		addr = hmget(hopr, opr);
		sprintf(out, "%03x", addr);
		strncpy(--tlines, out, 3);
	}

	puts(lines);
	extfname(argv[1], ofname);
	outf = fopen(ofname, "w");
	fputs(lines, outf);

	fclose(outf);
	hmfree(hopr);
	free(lines);
	return 0;
}

token_t tokenize(char *start)
{
	const char pseudo[][4] = { "ORG", "DEC", "HEX", "END" };
	token_t tkn = { '\0', '\0', '\0', false, 0 };
	char parts[4][10];
	int n, pi = 0, i;

	n = split(start, parts);
	if (n == 0) return tkn;

	if (parts[0][strlen(parts[0])-1] == ',') {
		parts[0][strlen(parts[0])-1] = '\0';
		strcpy(tkn.lab, parts[0]);
		pi++; n--;
		if (n <= 0) return tkn;
	}

	for (i = 3; i >= 0; --i) {
		if (STREQ(parts[pi], pseudo[i])) {
			tkn.psu = i + 1; pi++;
			if (n-- > 0)
				strcpy(tkn.opr, parts[pi]);
			return tkn;
		}
	}

	if (parts[pi][0] != '\0') {
		strcpy(tkn.opc, parts[pi++]); n--;
		if (n > 0) {
			strcpy(tkn.opr, parts[pi++]);
			n--;
		}
	}

	if (STREQ(parts[pi], "I"))
		if (n-- > 0) tkn.ind = true;

	return tkn;
}

int split(char *start, char list[4][10])
{
	int si = 0, wi = 0, li = 0;
	char c;
	do {
		c = start[si++];
		if (c == '\n' || c == '/') {
			if (wi)
				list[li++][wi] = '\0';
			break;
		}
		if (c == ' ' || c == '\t') {
			if (wi) {
				list[li++][wi] = '\0';
				wi = 0;
			}
			continue;
		}
		list[li][wi++] = c;
	} while (c != '\0');
	return li;
}

void extfname(char *path, char *fname)
{
	int si = 0, i;
	char c;
	for (i = 0; path[i] != '\0'; ++i) {
		c = path[i];
		fname[si++] = c;
		if (c == '/') si = 0;
		if (c == '.') {
			si--;
			break;
		}
	}
	fname[si] = '\0';
}