tis100

asm.c (14642B)

#include "asm.h"
#include "util.h"
#include "tpu.h"

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

#define TEXTALPH "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_"
#define NUMALPH "0123456789"
#define PORTALPH "abcdefghijklmnopqrstuvwxyz0123456789"
#define NAMEALPH TEXTALPH NUMALPH
#define WHITESPACE " \t\v\r\n,"

enum asm_tok {
	/* Missing */
	TOK_NONE = -1,

	/* Global */
	TOK_IN, TOK_OUT, TOK_TPU, TOK_END,

	/* Operands (order like OP_*) */
	TOK_ACC, TOK_NIL, TOK_LEFT, TOK_RIGHT, TOK_UP, TOK_DOWN,
	TOK_ANY, TOK_LAST, TOK_LIT, TOK_NAME,

	/* Instructions (order like INST_*) */
	TOK_NOP, TOK_MOV, TOK_SWP, TOK_SAV, TOK_ADD, TOK_SUB,
	TOK_NEG, TOK_JMP, TOK_JEZ, TOK_JNZ, TOK_JGZ, TOK_JLZ, TOK_JRO,

	/* Misc */
	TOK_COMMENT, TOK_LABEL, TOK_XPOS, TOK_YPOS, TOK_NL, TOK_EOF,
};

struct asm_tokenizer {
	const char *filepath;
	FILE *file;
	enum asm_tok next;
	char *tokstr;
	size_t lineno, off;
	char linebuf[256];
};

static const char *tok_strs[] = {
	/* Global */
	NULL, NULL, "tpu", "end",

	/* Operands (order like OP_*) */
	"acc", "nil", "left", "right", "up", "down",
	"any", "last", NULL, NULL,

	/* Instructions (order like INST_*) */
	"nop", "mov", "swp", "sav", "add", "sub",
	"neg", "jmp", "jez", "jnz", "jgz", "jlz", "jro",

	/* Misc */
	NULL, NULL, NULL, NULL, NULL, NULL
};

static const char *tok_reprs[] = {
	/* Global */
	"IN.<C>", "OUT.<C>", "TPU", "END",

	/* Operands (order like OP_*) */
	"ACC", "NIL", "LEFT", "RIGHT", "UP", "DOWN",
	"ANY", "LAST", "<LIT>", "<NAME>",

	/* Instructions (order like INST_*) */
	"NOP", "MOV", "SWP", "SAV", "ADD", "SUB",
	"NEG", "JMP", "JEZ", "JNZ", "JGZ", "JLZ", "JRO",

	/* Misc */
	"#<COMMENT>", "<LABEL>:", "X<INT>", "Y<INT>", "<NL>", "<EOF>"
};

static bool
is_int(const char *str)
{
	char *end;
	long v;

	v = strtol(str, &end, 10);
	if (!end || *end) return false;
	if (v < INT32_MIN || v > INT32_MAX) return false;

	return true;
}

static int
asm_port_char_to_index(char c)
{
	char *p;

	p = strchr(PORTALPH, tolower(c));
	if (!p) die("invalid io port char '%c'", c);
	return (int) (p - PORTALPH);
}

static bool
asm_is_lit(const char *str)
{
	const char *c;

	c = str;
	if (*c == '-') c++;
	else if (*c == '+') c++;
	for (; *c && isdigit(*c); c++);
	return !*c;
}

static int
asm_str_to_lit(const char *str)
{
	int m, v, b, i, o;

	if (*str == '-') {
		m = -1;
		o = 1;
	} else if (*str == '+') {
		m = 1;
		o = 1;
	} else {
		m = 1;
		o = 0;
	}

	for (b = 1, i = o; str[i]; i++)
		b *= 10;

	if (i >= 4 + o) {
		v = 1000;
	} else {
		for (v = 0, i = o; str[i]; i++) {
			b /= 10;
			v += b * (str[i] - '0');
		}
	}

	return MIN(MAX(m * v, -999), 999);
}

enum tpu_inst_type
tok_to_inst(enum asm_tok tok)
{
	if (tok < TOK_NOP || tok > TOK_JRO) abort();
	return tok - TOK_NOP + INST_NOP;
}

enum tpu_inst_op_type
tok_to_optype(enum asm_tok tok)
{
	if (tok < TOK_ACC || tok > TOK_NAME) abort();
	return tok - TOK_ACC + OP_ACC;
}

struct tpu_inst_op
tok_to_op(struct asm_tokenizer *tokenizer, enum asm_tok tok)
{
	struct tpu_inst_op op;

	op.type = tok_to_optype(tok);
	if (op.type == OP_LIT) {
		op.val.lit = asm_str_to_lit(tokenizer->tokstr);
	} else if (op.type == OP_LABEL) {
		op.val.label = strdup(tokenizer->tokstr);
		if (!op.val.label) die("strdup:");
	}

	return op;
}

static size_t
strlcat_op_name(char *buf, struct tpu_inst_op *op, size_t n)
{
	char tmp[5];

	if (op->type == OP_LIT) {
		snprintf(tmp, 5, "%i", op->val.lit);
		return strdcat(buf, tmp, n);
	} else if (op->type == OP_LABEL) {
		return strdcat(buf, op->val.label, n);
	} else {
		return strdcat(buf, op_reprs[op->type], n);
	}
}

size_t
asm_print_inst(char *buf, size_t n, struct tpu_inst *inst, size_t max)
{
	size_t len, op;

	len = strdcpy(buf, inst_reprs[inst->type], n);
	if (inst->opcnt >= 1) {
		len += strdcat(buf, " ", n);
		len += strlcat_op_name(buf, &inst->ops[0], n);
	}
	if (inst->opcnt >= 2) {
		op = strdcat(buf, ", ", n);
		op += strlcat_op_name(buf, &inst->ops[1], n);
		if (len + op > max && len < n) {
			buf[len] = '\0';
			op = strdcat(buf, ",", n);
			op += strlcat_op_name(buf, &inst->ops[1], n);
		}
		len += op;
	}

	return len;
}

static enum asm_tok
tok_next(struct asm_tokenizer *tok)
{
	enum asm_tok v;
	size_t len;
	char *s;
	int i;

	if (tok->next != TOK_NONE) {
		v = tok->next;
		tok->next = TOK_NONE;
		return v;
	}

	if (!tok->linebuf[tok->off]) {
		if (feof(tok->file)) return TOK_EOF;
		s = fgets(tok->linebuf, sizeof(tok->linebuf), tok->file);
		if (!s && !feof(tok->file)) die("fgets:");
		if (!s) return TOK_NL;

		len = strlen(s);
		if (len && s[len-1] != '\n' && !feof(tok->file))
			die("load: line %lu too long", tok->lineno);
		if (len && s[len-1] == '\n') s[len-1] = '\0';

		tok->lineno += 1;
		tok->tokstr = s;
		tok->off = 0;
		return TOK_NL;
	}

	s = tok->linebuf + tok->off;
	len = strspn(s, WHITESPACE);
	tok->off += len;
	if (!s[len]) return TOK_NL;
	tok->tokstr = (s += len);

	len = strcspn(s, WHITESPACE);
	tok->off += len;
	if (s[len]) {
		s[len] = '\0';
		tok->off += 1;
	}

	for (i = 0; i <= TOK_EOF; i++) {
		if (tok_strs[i] && !strcasecmp(s, tok_strs[i]))
			return (enum asm_tok) i;
	}

	if (!strncasecmp(s, "in.", 3) && len == 4) {
		return TOK_IN;
	} else if (!strncasecmp(s, "out.", 4) && len == 5) {
		return TOK_OUT;
	} else if (asm_is_lit(s)) {
		return TOK_LIT;
	} else if (*s == '#') {
		tok->tokstr = tok->linebuf + tok->off;
		tok->off += strlen(tok->linebuf + tok->off);
		return TOK_COMMENT;
	} else if (len && strchr(TEXTALPH, *s)
			&& strspn(s, NAMEALPH) == len-1 && s[len-1] == ':') {
		s[len-1] = '\0';
		return TOK_LABEL;
	} else if (*s == 'X' && is_int(s+1)) {
		return TOK_XPOS;
	} else if (*s == 'Y' && is_int(s+1)) {
		return TOK_YPOS;
	} else if (strchr(TEXTALPH, *s)
			&& strspn(s, NAMEALPH) == strlen(s)) {
		return TOK_NAME;
	} else {
		die("load: line %lu, invalid token '%s'", tok->lineno, s);
	}
}

static enum asm_tok
tok_next_in(struct asm_tokenizer *tokenizer, ...)
{
	va_list ap, cpy;
	enum asm_tok tok;
	bool first;
	int arg;

	tok = tok_next(tokenizer);

	va_copy(cpy, ap);

	va_start(cpy, tokenizer);
	while ((arg = va_arg(cpy, int)) > 0) {
		if (tok == arg) return tok;
	}
	va_end(cpy);

	fprintf(stderr, "tis-as: load: ");
	fprintf(stderr, "line %lu, got tok %s, expected one of (",
		tokenizer->lineno, tok_reprs[tok]);
	first = true;
	va_start(ap, tokenizer);
	while ((arg = va_arg(ap, int)) > 0) {
		if (!first) fputc(',', stderr);
		fputs(tok_reprs[arg], stderr);
		first = false;
	}
	va_end(ap);
	fputs(")\n", stderr);

	exit(1);
}

static void
tpu_validate(struct tpu *tpu)
{
	int dst, i;

	for (i = 0; i < tpu->inst_cnt; i++) {
		if (tpu->insts[i].opcnt >= 1
				&& tpu->insts[i].ops[0].type == OP_LABEL) {
			dst = label_map_get(&tpu->label_map,
				tpu->insts[i].ops[0].val.label);
			if (dst < 0)
				die("load: tpu X%i Y%i, label '%s' not defined",
					tpu->x, tpu->y,
					tpu->insts[i].ops[0].val.label);
		}
	}
}

void
tis_load_asm(struct tis *tis, const char *filepath)
{
	struct tpu_io_port *io_port;
	struct asm_tokenizer tokenizer;
	struct tpu_inst_op op1, op2;
	enum tpu_inst_type inst_type;
	struct tpu_inst *inst;
	struct tpu *tpu = NULL;
	struct tpu_map_link *link;
	struct tpu_port *port;
	enum asm_tok tok, optok;
	char rowbuf[TPU_MAX_COLS+1];
	size_t colsused;
	int io_x, io_y;
	ssize_t i, k;
	size_t len;
	char c;

	tis->asm_filepath = strdup(filepath);

	tokenizer.filepath = filepath;
	tokenizer.file = fopen(filepath, "r");
	if (!tokenizer.file) die("load: fopen '%s':", filepath);

	tokenizer.next = TOK_NONE;
	tokenizer.lineno = 0;
	tokenizer.off = 0;
	tokenizer.tokstr = NULL;
	tokenizer.linebuf[tokenizer.off] = '\0';

	colsused = 0;
	while ((tok = tok_next(&tokenizer)) != TOK_EOF) {
		switch (tok) {
		case TOK_IN:
		case TOK_OUT:
			if (tpu) goto disallowed;

			io_port = malloc(sizeof(struct tpu_io_port));
			if (!io_port) die("malloc:");

			len = strlen(tokenizer.tokstr);
			c = tokenizer.tokstr[len-1];
			k = asm_port_char_to_index(c);

			if (tok == TOK_IN) {
				if (tis->in_ports[k])
					die("load: line %lu, duplicate in.%c",
						tokenizer.lineno, c);
				tis->in_ports[k] = io_port;
			} else {
				if (tis->out_ports[k])
					die("load: loute %lu, duplicate out.%c",
						tokenizer.lineno, c);
				tis->out_ports[k] = io_port;
			}

			tok_next_in(&tokenizer, TOK_XPOS, -1);
			io_x = atoi(tokenizer.tokstr + 1);
			tok_next_in(&tokenizer, TOK_YPOS, -1);
			io_y = atoi(tokenizer.tokstr + 1);
			tok_next_in(&tokenizer, TOK_NL, -1);

			if (tok == TOK_IN) {
				tpu_io_port_init(io_port, c, DIR_UP,
					PORT_IN, io_x, io_y);
			} else {
				tpu_io_port_init(io_port, c, DIR_DOWN,
					PORT_OUT, io_x, io_y);
			}

			colsused = 0;
			break;
		case TOK_TPU:
			if (tpu) goto disallowed;
			tpu = malloc(sizeof(struct tpu));
			if (!tpu) die("malloc:");
			tpu_init(tpu);
			tok_next_in(&tokenizer, TOK_XPOS, -1);
			tpu->x = atoi(tokenizer.tokstr + 1);
			tok_next_in(&tokenizer, TOK_YPOS, -1);
			tpu->y = atoi(tokenizer.tokstr + 1);
			tok_next_in(&tokenizer, TOK_NL, -1);
			if (!tpu_map_add(&tis->tpu_map, tpu))
				die("load: duplicate tpu location X%i Y%i",
					tpu->x, tpu->y);
			colsused = 0;
			break;
		case TOK_END:
			if (!tpu) goto disallowed;
			tpu_validate(tpu);
			tpu = NULL;
			tok_next_in(&tokenizer, TOK_NL, -1);
			colsused = 0;
			break;
		case TOK_NOP: case TOK_MOV: case TOK_SWP: case TOK_SAV:
		case TOK_ADD: case TOK_SUB: case TOK_NEG: case TOK_JMP:
		case TOK_JEZ: case TOK_JNZ: case TOK_JGZ: case TOK_JLZ:
		case TOK_JRO:
			if (!tpu) goto disallowed;
			inst_type = tok_to_inst(tok);

			optok = tok_next_in(&tokenizer, TOK_ACC,
				TOK_NIL, TOK_LEFT, TOK_RIGHT, TOK_UP, TOK_DOWN,
				TOK_ANY, TOK_LAST, TOK_LIT, TOK_NAME, TOK_NL, -1);
			if (optok == TOK_NL) {
				inst = tpu_add_inst(tpu, inst_type, 0, op1, op2);
				if (!inst) {
					die("load: line %lu, invalid instruction",
						tokenizer.lineno-1);
				}
				goto inst_check;
			}
			op1 = tok_to_op(&tokenizer, optok);

			optok = tok_next_in(&tokenizer, TOK_ACC,
				TOK_NIL, TOK_LEFT, TOK_RIGHT, TOK_UP, TOK_DOWN,
				TOK_ANY, TOK_LAST, TOK_LIT, TOK_NAME, TOK_NL, -1);
			if (optok == TOK_NL) {
				inst = tpu_add_inst(tpu, inst_type, 1, op1, op2);
				if (!inst) {
					die("load: line %lu, invalid instruction",
						tokenizer.lineno-1);
				}
				goto inst_check;
			}
			op2 = tok_to_op(&tokenizer, optok);

			tok_next_in(&tokenizer, TOK_NL, -1);

			inst = tpu_add_inst(tpu, inst_type, 2, op1, op2);
			if (!inst) die("load: line %lu, invalid instruction",
				tokenizer.lineno-1);

inst_check:
			tpu->rows += 1;
			if (tpu->inst_cnt > TPU_MAX_INST_CNT || tpu->rows > TPU_MAX_ROWS)
				die("load: line %lu, tpu has too many rows",
					tokenizer.lineno-1);

			len = asm_print_inst(rowbuf, sizeof(rowbuf),
				inst, TPU_MAX_COLS - colsused);
			if (colsused + len > TPU_MAX_COLS)
				die("load: line %lu, tpu row is too long (%zu,%zu)",
					tokenizer.lineno-1, colsused, len);

			colsused = 0;
			break;
		case TOK_COMMENT:
			if (tpu && !strcmp(tokenizer.tokstr, "DISABLED"))
				tpu->disabled = true;
			tok_next_in(&tokenizer, TOK_NL, -1);
			break;
		case TOK_LABEL:
			len = strlen(tokenizer.tokstr);
			strncpy(rowbuf, tokenizer.tokstr, TPU_MAX_COLS);
			optok = tok_next(&tokenizer);
			if (!label_map_add(&tpu->label_map, rowbuf,
					(int) tpu->inst_cnt, optok != TOK_NL))
				die("load: line %lu, duplicate label %s (pos)",
					tokenizer.lineno, rowbuf);
			if (optok == TOK_NL) {
				tpu->rows += 1;
				if (tpu->rows > TPU_MAX_ROWS)
					die("load: line %lu, line does not fit",
						tokenizer.lineno);
				colsused = 0;
			} else {
				tokenizer.next = optok;
				colsused = len + 1;
			}
			if (len + 1 > TPU_MAX_COLS)
				die("load: line %lu, label too long",
					tokenizer.lineno);
			break;
		case TOK_NL:
			colsused = 0;
			break;
		default:
			goto disallowed;
		}
	}

	for (i = 0; i < TPU_MAP_BUCKETS; i++) {
		for (link = tis->tpu_map.buckets[i]; link; link = link->next) {
			tpu_init_ports(link->tpu, &tis->tpu_map);

			for (k = -TIS_MAX_IO_PORTS; k < TIS_MAX_IO_PORTS; k++) {
				io_port = k < 0 ? tis->in_ports[-k-1]
					: tis->out_ports[k];
				if (!io_port || io_port->port.attached) continue;
				if (link->tpu->x != io_port->x) continue;
				if (link->tpu->y != io_port->y) continue;
				port = &link->tpu->ports[io_port->dir];
				if (port->attached) {
					die("load: io_port X%i Y%i (%s) busy",
						io_port->x, io_port->y,
						dir_reprs[io_port->dir]);
				}
				port->attached = true;
				port->dst_port = &io_port->port;
				port->type = io_port->type;
				io_port->port.attached = true;
				io_port->port.dst_port = port;
			}
		}
	}

	for (i = -TIS_MAX_IO_PORTS; i < TIS_MAX_IO_PORTS; i++) {
		io_port = i < 0 ? tis->in_ports[-i-1] : tis->out_ports[i];
		if (io_port && !io_port->port.attached) {
			die("load: io_port X%i Y%i (%s) not found",
				io_port->x, io_port->y,
				dir_reprs[io_port->dir]);
		}
	}

	fclose(tokenizer.file);

	return;

disallowed:
	if (tok == TOK_NAME) {
		die("load: line %lu, unexpected token '%s'",
			tokenizer.lineno, tokenizer.tokstr);
	} else {
		die("load: line %lu, token %s not allowed here",
			tokenizer.lineno, tok_reprs[tok]);
	}
}

void
tis_load(struct tis *tis, const char **argv)
{
	const char **arg;
	int n, i;
	char c;

	if (!*argv) die("missing argv[0]");

	for (arg = argv + 1; *arg; arg++) {
		if (!strcmp(*arg, "-h") || !strcmp(*arg, "--help")) {
			fprintf(stderr, "Usage: %s [-h] [-s] "
				"[--in.X IN].. [--out.X OUT].. ASM\n", argv[0]);
		} else if (!strcmp(*arg, "-s") || !strcmp(*arg, "--stats")) {
			tis->show_stats = true;
		} else if (sscanf(*arg, "--in.%c%n", &c, &n) == 1 && n == 6) {
			/* after tis_load_asm.. */
			arg++;
		} else if (sscanf(*arg, "--out.%c%n", &c, &n) == 1 && n == 7) {
			/* after tis_load_asm.. */
			arg++;
		} else {
			if (tis->asm_filepath)
				die("multiple asm files");
			tis_load_asm(tis, *arg);
		}
	}

	for (arg = argv + 1; *arg; arg++) {
		if (sscanf(*arg, "--in.%c%n", &c, &n) == 1 && n == 6) {
			i = asm_port_char_to_index(c);
			if (!tis->in_ports[i])
				die("in.%c io port not initialized", c);
			tis->in_ports[i]->file = fopen(*++arg, "r");
			if (!tis->in_ports[i]->file)
				die("fopen '%s':", *arg);
			setvbuf(tis->in_ports[i]->file, NULL, _IONBF, 0);
		} else if (sscanf(*arg, "--out.%c%n", &c, &n) == 1 && n == 7) {
			i = asm_port_char_to_index(c);
			if (!tis->out_ports[i])
				die("out.%c io port not initialized", c);
			tis->out_ports[i]->file = fopen(*++arg, "w+");
			if (!tis->out_ports[i]->file)
				die("fopen '%s':", *arg);
			setvbuf(tis->out_ports[i]->file, NULL, _IONBF, 0);
		}
	}
}