Add initial rough outline

1 files changed, 362 insertions, 0 deletions

diff --git a/tis-as.c b/tis-as.c
new file mode 100644
index 0000000..0d3625c
--- /dev/null
+++ b/tis-as.c
@@ -0,0 +1,362 @@
+#include "tpu.h"
+
+#include <stdarg.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdint.h>
+#include <stdbool.h>
+#include <stdlib.h>
+
+#define WHITESPACE " \t\v\r\n"
+
+enum tok {
+	/* Global */
+	TOK_STDIN, TOK_STDOUT, TOK_TPU, TOK_END,
+
+	/* Operands (order like OP_*) */
+	TOK_ACC, TOK_BAK, TOK_NIL, TOK_LEFT, TOK_RIGHT,
+	TOK_UP, TOK_DOWN, TOK_ANY, TOK_LAST, TOK_LIT,
+
+	/* 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_JLZ, TOK_JRO,
+
+	/* Misc */
+	TOK_COMMENT, TOK_LABEL, TOK_XPOS, TOK_YPOS, TOK_NL, TOK_EOF
+};
+
+struct tokenizer {
+	FILE *file;
+	enum tok tok;
+	char *tokstr;
+	size_t lineno, off;
+	char linebuf[256];
+};
+
+static int stdin_x = 0, stdin_y = 0;
+static int stdout_x = 0, stdout_y = 0;
+
+struct tpu_map map;
+
+static const char *tok_reprs[] = {
+	/* Global */
+	"STDIN", "STDOUT", "BLOCK", "END",
+
+	/* Operands */
+	"ACC", "BAK", "NIL", "LEFT", "RIGHT", "UP",
+	"DOWN", "ANY", "LAST", "<LIT>",
+
+	/* Instructions */
+	"NOP", "MOV", "SWP", "SAV", "ADD", "SUB",
+	"NEG", "JMP", "JEZ", "JNZ", "JLZ", "JRO",
+
+	/* Misc */
+	"#<COMMENT>", "<LABEL>", "X<INT>", "Y<INT>", "<NL>", "<EOF>"
+};
+
+static void
+__attribute__((format(printf, 1, 2)))
+__attribute__((noreturn))
+die(const char *fmt, ...)
+{
+	va_list ap;
+
+	va_start(ap, fmt);
+	fprintf(stderr, "tis: ");
+	vfprintf(stderr, fmt, ap);
+	if (*fmt && fmt[strlen(fmt)-1] == ':') {
+		fputc(' ', stderr);
+		perror(NULL);
+	} else {
+		fputc('\n', stderr);
+	}
+	va_end(ap);
+
+	exit(1);
+}
+
+bool
+is_lit(const char *str)
+{
+	unsigned long v;
+	char *end;
+
+	v = strtoul(str, &end, 10);
+	if (!end || *end) return false;
+
+	return v < 256;
+}
+
+uint8_t
+str2lit(const char *str)
+{
+	return (uint8_t) atoi(str);
+}
+
+enum tpu_inst_type
+tok_to_inst(enum 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 tok tok)
+{
+	if (tok < TOK_ACC || tok >= TOK_LIT) abort();
+	return tok - TOK_ACC + OP_ACC;
+}
+
+enum tok
+tok_next(struct tokenizer *tok)
+{
+	size_t len;
+	char *s;
+
+	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_EOF;
+		if (*s && s[strlen(s)-1] != '\n')
+			die("tokenizer: line %lu too long", tok->lineno);
+		tok->off = 0;
+	}
+
+	s = tok->linebuf + tok->off;
+	len = strspn(s, WHITESPACE);
+	tok->off += len;
+	s += len;
+
+	len = strcspn(s, WHITESPACE);
+	s[len] = '\0';
+	tok->off += len;
+	if (!len) return TOK_NL;
+
+	if (!strcasecmp(s, "stdin")) {
+		return TOK_STDIN;
+	} else if (!strcasecmp(s, "stdout")) {
+		return TOK_STDIN;
+	} else if (!strcasecmp(s, "block")) {
+		return TOK_TPU;
+	} else if (!strcasecmp(s, "end")) {
+		return TOK_END;
+	} else if (!strcasecmp(s, "acc")) {
+		return TOK_ACC;
+	} else if (!strcasecmp(s, "bak")) {
+		return TOK_BAK;
+	} else if (!strcasecmp(s, "nil")) {
+		return TOK_NIL;
+	} else if (!strcasecmp(s, "left")) {
+		return TOK_LEFT;
+	} else if (!strcasecmp(s, "right")) {
+		return TOK_RIGHT;
+	} else if (!strcasecmp(s, "up")) {
+		return TOK_UP;
+	} else if (!strcasecmp(s, "down")) {
+		return TOK_DOWN;
+	} else if (!strcasecmp(s, "any")) {
+		return TOK_ANY;
+	} else if (!strcasecmp(s, "last")) {
+		return TOK_LAST;
+	} else if (is_lit(s)) {
+		return TOK_LIT;
+	} else if (!strcasecmp(s, "nop")) {
+		return TOK_NOP;
+	} else if (!strcasecmp(s, "mov")) {
+		return TOK_MOV;
+	} else if (!strcasecmp(s, "swp")) {
+		return TOK_SWP;
+	} else if (!strcasecmp(s, "sav")) {
+		return TOK_SAV;
+	} else if (!strcasecmp(s, "add")) {
+		return TOK_ADD;
+	} else if (!strcasecmp(s, "sub")) {
+		return TOK_SUB;
+	} else if (!strcasecmp(s, "neg")) {
+		return TOK_NEG;
+	} else if (!strcasecmp(s, "jmp")) {
+		return TOK_JMP;
+	} else if (!strcasecmp(s, "jez")) {
+		return TOK_JEZ;
+	} else if (!strcasecmp(s, "jnz")) {
+		return TOK_JNZ;
+	} else if (!strcasecmp(s, "jlz")) {
+		return TOK_JLZ;
+	} else if (!strcasecmp(s, "jro")) {
+		return TOK_JRO;
+	} else if (*s == '#') {
+		tok->off = strlen(s);
+		return TOK_COMMENT;
+	} else if (len && s[len-1] == ':') {
+		return TOK_LABEL;
+	} else if (*s == 'X' && atoi(s+1) > 0) {
+		return TOK_XPOS;
+	} else if (*s == 'Y' && atoi(s+1) > 0) {
+		return TOK_YPOS;
+	} else {
+		die("tokenizer: line %lu, invalid token '%s'", tok->lineno, s);
+	}
+}
+
+enum tok
+tok_next_in(struct tokenizer *tokenizer, ...)
+{
+	va_list ap, cpy;
+	enum 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: tokenizer: ");
+	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);
+
+	fprintf(stderr, "\n");
+	exit(1);
+}
+
+static void
+load(const char *fname)
+{
+	struct tokenizer tokenizer;
+	enum tpu_inst_op_type op1, op2;
+	enum tok op1_tok, op2_tok;
+	uint8_t op1_lit, op2_lit;
+	enum tpu_inst_type inst;
+	struct tpu *tpu = NULL;
+	enum tok tok;
+	FILE *file;
+
+	file = fopen(fname, "r");
+	if (!file) die("fopen:");
+
+	tokenizer.lineno = 1;
+	tokenizer.file = file;
+	tokenizer.off = 0;
+	tokenizer.tokstr = NULL;
+	tokenizer.linebuf[tokenizer.off] = '\0';
+	while ((tok = tok_next(&tokenizer)) != TOK_EOF) {
+		switch (tok) {
+		case TOK_STDIN:
+			if (tpu) goto disallowed;
+			tok_next_in(&tokenizer, TOK_XPOS, -1);
+			stdin_x = atoi(tokenizer.tokstr + 1);
+			tok_next_in(&tokenizer, TOK_YPOS, -1);
+			stdin_y = atoi(tokenizer.tokstr + 1);
+			tok_next_in(&tokenizer, TOK_NL, -1);
+			break;
+		case TOK_STDOUT:
+			if (tpu) goto disallowed;
+			tok_next_in(&tokenizer, TOK_XPOS, -1);
+			stdout_x = atoi(tokenizer.tokstr + 1);
+			tok_next_in(&tokenizer, TOK_YPOS, -1);
+			stdout_y = atoi(tokenizer.tokstr + 1);
+			tok_next_in(&tokenizer, TOK_NL, -1);
+			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);
+			/* TODO insert tpu */
+			break;
+		case TOK_END:
+			if (!tpu) goto disallowed;
+			tpu = NULL;
+			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_JLZ: case TOK_JRO:
+			if (!tpu) goto disallowed;
+			inst = tok_to_inst(tok);
+
+			op1_tok = tok_next_in(&tokenizer, TOK_ACC,
+				TOK_BAK, TOK_UP, TOK_DOWN,
+				TOK_LEFT, TOK_RIGHT, TOK_NL, -1);
+			if (op1_tok == TOK_NL) {
+				tok_next_in(&tokenizer, TOK_NL, -1);
+				tpu_add_inst(tpu, inst, -1, 0, -1, 0);
+				break;
+			}
+
+			op1 = tok_to_optype(op1_tok);
+			if (op1 == OP_LIT)
+				op1_lit = str2lit(tokenizer.tokstr);
+
+			op2_tok = tok_next_in(&tokenizer, TOK_ACC,
+				TOK_BAK, TOK_UP, TOK_DOWN,
+				TOK_LEFT, TOK_RIGHT, TOK_NL, -1);
+			if (op2_tok == TOK_NL) {
+				tok_next_in(&tokenizer, TOK_NL, -1);
+				tpu_add_inst(tpu, inst,
+					(int) op1, op1_lit, -1, 0);
+				break;
+			}
+
+			op2 = tok_to_optype(op2_tok);
+			if (op2 == OP_LIT)
+				op2_lit = str2lit(tokenizer.tokstr);
+			tok_next_in(&tokenizer, TOK_NL, -1);
+			tpu_add_inst(tpu, inst, (int) op1, op1_lit,
+				(int) op2, op2_lit);
+			break;
+		case TOK_COMMENT:
+			tok_next_in(&tokenizer, TOK_NL, -1);
+			break;
+		case TOK_LABEL:
+			/* TODO */
+			break;
+		default:
+			goto disallowed;
+		}
+	}
+
+	fclose(file);
+
+disallowed:
+	die("tokenizer: line %lu, token %s not allowed here",
+		tokenizer.lineno, tok_reprs[tok]);
+}
+
+void
+run(void)
+{
+
+}
+
+int
+main(int argc, const char **argv)
+{
+	if (argc != 2) {
+		fprintf(stderr, "Usage: tis-as FILE\n");
+		exit(1);
+	}
+
+	load(argv[1]);
+
+	run();
+}