initial tokenization

1 files changed, 447 insertions, 0 deletions

diff --git a/src/fwdscript.c b/src/fwdscript.c
new file mode 100644
index 0000000..c1eda21
--- /dev/null
+++ b/src/fwdscript.c
@@ -0,0 +1,447 @@
+#include <err.h>
+#include <stdio.h>
+#include <ctype.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <string.h>
+
+#define TODO() errx(EXIT_FAILURE, "%s:%d: todo", __FILE__,  __LINE__)
+
+struct fwdval;
+struct fwdvals;
+
+struct fwdcnt {
+	size_t pc;
+	struct fwdvalues *ctx;
+};
+
+enum fwdtype {
+	FWDSCRIPT_NUM,
+	FWDSCRIPT_STR,
+	FWDSCRIPT_CNT,
+	FWDSCRIPT_TBL,
+	FWDSCRIPT_OPQ
+};
+
+struct fwdval {
+	enum fwdtype kind;
+
+	union {
+		double num;
+		char *str;
+		struct fwdcnt cnt;
+		struct fwdtbl *tbl;
+		void *opq;
+	};
+};
+
+#define FWDSTR(s) (struct fwdval){.kind = FWDSCRIPT_STR, .str = (s)}
+
+static int fwdcmp(const struct fwdval *a, const struct fwdval *b)
+{
+	if (a->kind != b->kind)
+		return a->kind - b->kind;
+
+	switch (a->kind) {
+	case FWDSCRIPT_NUM: return a->num - b->num;
+	case FWDSCRIPT_STR: return strcmp(a->str, b->str);
+	case FWDSCRIPT_CNT: return a->cnt.pc - b->cnt.pc;
+	case FWDSCRIPT_TBL: return (intptr_t)a->tbl - (intptr_t)b->tbl;
+	case FWDSCRIPT_OPQ: return (intptr_t)a->opq - (intptr_t)b->opq;
+	default: abort();
+	}
+
+	return 0;
+}
+
+static int fwdhash(const struct fwdval *a)
+{
+	switch (a->kind) {
+	/* trash hashes, lol */
+	case FWDSCRIPT_NUM: return a->num * 15647839;
+	case FWDSCRIPT_STR: return a->str[0];
+	case FWDSCRIPT_CNT: return a->cnt.pc * 15647839;
+	case FWDSCRIPT_TBL: return 10;
+	case FWDSCRIPT_OPQ: return (intptr_t)a->opq;
+	default: abort();
+	}
+
+	return 0;
+}
+
+#define MAP_KEY struct fwdval
+#define MAP_TYPE struct fwdval
+#define MAP_CMP(A, B) fwdcmp(&(A), &(B))
+#define MAP_HASH(A) fwdhash(&(A))
+#define MAP_NAME fwdtbl
+#include <conts/map.h>
+
+#define MAP_KEY const char *
+#define MAP_TYPE size_t
+#define MAP_CMP(A, B) strcmp((A), (B))
+#define MAP_HASH(A) (A)[0]
+#define MAP_NAME fwdfuncs
+#include <conts/map.h>
+
+#define VEC_TYPE struct fwdval
+#define VEC_NAME fwdvals
+#include <conts/vec.h>
+
+#define VEC_TYPE enum fwdtype
+#define VEC_NAME fwdtypes
+#include <conts/vec.h>
+
+struct fwdinsn {
+	/* use args[0] as destination, if /cnt then call the
+	 * continuation, if /opq call builtin, otherwise abort */
+	struct fwdvals args;
+};
+
+#define VEC_TYPE struct fwdinsn
+#define VEC_NAME fwdinsns
+#include <conts/vec.h>
+
+struct fwdscript;
+
+typedef int (*fwd_typecheck_t)(struct fwdscript *s, struct fwdtypes *types);
+typedef int (*fwd_callback_t)(struct fwdscript *s, struct fwdvals *args);
+
+struct fwdbuiltin {
+	fwd_callback_t cb;
+	fwd_typecheck_t tp;
+};
+
+#define MAP_KEY const char *
+#define MAP_TYPE struct fwdbuiltin
+#define MAP_CMP(A, B) strcmp((A), (B))
+#define MAP_HASH(A) (A)[0]
+#define MAP_NAME fwdbuiltins
+#include <conts/map.h>
+
+struct fwdscript {
+	struct fwdfuncs funcs;
+	struct fwdinsns insns;
+	struct fwdbuiltins builtins;
+};
+
+static struct fwdscript fwdscript_create()
+{
+	struct fwdscript inst = {
+		.funcs = fwdfuncs_create(1),
+		.insns = fwdinsns_create(1),
+		.builtins = fwdbuiltins_create(1)
+	};
+
+	return inst;
+}
+
+static int fwdscript_register(struct fwdscript *s, const char *name,
+		fwd_callback_t cb, fwd_typecheck_t tp)
+{
+	struct fwdbuiltin builtin = {
+		.cb = cb,
+		.tp = tp
+	};
+
+	struct fwdbuiltin *b = fwdbuiltins_insert(&s->builtins,
+			name, builtin);
+
+	if (b == NULL) {
+		fprintf(stderr, "failed registering %s\n", name);
+		return -1;
+	}
+
+	return 0;
+}
+
+static int fwdscript_get(struct fwdscript *s, struct fwdvals *args)
+{
+	TODO();
+}
+
+static int fwdscript_get_typecheck(struct fwdscript *s, struct fwdtypes *types)
+{
+	TODO();
+}
+
+static int fwdscript_abort(struct fwdscript *s, struct fwdvals *args)
+{
+	TODO();
+}
+
+static int fwdscript_abort_typecheck(struct fwdscript *s, struct fwdtypes *types)
+{
+	TODO();
+}
+
+static int fwdscript_print(struct fwdscript *s, struct fwdvals *args)
+{
+	TODO();
+}
+
+static int fwdscript_print_typecheck(struct fwdscript *s, struct fwdtypes *args)
+{
+	TODO();
+}
+
+static int fwdscript_register_builtins(struct fwdscript *s)
+{
+	if (fwdscript_register(s, "get", fwdscript_get, fwdscript_get_typecheck))
+		return -1;
+
+	if (fwdscript_register(s, "abort", fwdscript_abort, fwdscript_abort_typecheck))
+		return -1;
+
+	if (fwdscript_register(s, "print", fwdscript_print, fwdscript_print_typecheck))
+		return -1;
+
+	return 0;
+}
+
+static const char *skip_ignored(const char *buf)
+{
+	while (*buf && isspace(*buf))
+		buf++;
+
+	if (*buf == '#')
+	while (*buf && *buf != '\n')
+		buf++;
+
+	return buf;
+}
+
+static int separator(int c)
+{
+	return c == ';' || c == '{' || c == '}'
+	    || c == ';' || c == '=' || c == '"'
+	    || c == '\n';
+}
+
+static const char *next_separator(const char *buf)
+{
+	while (*buf && !isspace(*buf) && !separator(*buf))
+		buf++;
+
+	return buf;
+}
+
+struct fwdtoken {
+	enum {
+		FWDTOKEN_ID,
+		FWDTOKEN_STR,
+		FWDTOKEN_OPEN,
+		FWDTOKEN_CLOSE,
+		FWDTOKEN_TO,
+		FWDTOKEN_END
+	} kind;
+
+	size_t line;
+	char *s;
+};
+
+#define VEC_TYPE struct fwdtoken
+#define VEC_NAME fwdtokens
+#include <conts/vec.h>
+
+static int fwdscript_tokenize(struct fwdtokens *tokens, const char *buf)
+{
+	size_t line = 0;
+
+	const char *start = buf;
+	while (*start) {
+		const char *tok = skip_ignored(start);
+		if (!*tok)
+			break;
+
+		if (*tok == '\n') {
+			line += 1;
+			start = tok + 1;
+			continue;
+		}
+
+		if (*tok == ';') {
+			struct fwdtoken end = {
+				.kind = FWDTOKEN_END,
+				.line = line,
+				.s    = NULL
+			};
+
+			if (!fwdtokens_append(tokens, end)) {
+				fprintf(stderr, "failed appending end token\n");
+				return -1;
+			}
+
+			start = tok + 1;
+			continue;
+		}
+
+		if (*tok == '{') {
+			struct fwdtoken open = {
+				.kind = FWDTOKEN_OPEN,
+				.line = line,
+				.s    = NULL
+			};
+
+			if (!fwdtokens_append(tokens, open)) {
+				fprintf(stderr, "failed appending open token\n");
+				return -1;
+			}
+
+			start = tok + 1;
+			continue;
+		}
+
+		if (*tok == '}') {
+			struct fwdtoken close = {
+				.kind = FWDTOKEN_CLOSE,
+				.line = line,
+				.s    = NULL
+			};
+
+			if (!fwdtokens_append(tokens, close)) {
+				fprintf(stderr, "failed appending close token\n");
+				return -1;
+			}
+
+			start = tok + 1;
+			continue;
+		}
+
+		if (*tok == '=' && *(tok + 1) == '>') {
+			struct fwdtoken to = {
+				.kind = FWDTOKEN_TO,
+				.line = line,
+				.s    = NULL
+			};
+
+			if (!fwdtokens_append(tokens, to)) {
+				fprintf(stderr, "failed appending to token\n");
+				return -1;
+			}
+
+			start = tok + 2;
+			continue;
+		}
+
+		if (*tok == '"') {
+			size_t start_line = line;
+			const char *end = tok + 1;
+			while (*end && *end != '"') {
+				if (*end == '\n') {
+					line++;
+
+					while (isblank(*end))
+						end++;
+				}
+
+				if (*end == '\'')
+					end++;
+
+				end++;
+			}
+
+			if (!*end) {
+				fprintf(stderr, "unterminated string at %zu\n", start_line);
+				return -1;
+			}
+
+			char *s = strndup(tok + 1, end - tok - 2);
+			if (!s) {
+				fprintf(stderr, "failed duping str token\n");
+				return -1;
+			}
+
+			struct fwdtoken str = {
+				.kind = FWDTOKEN_STR,
+				.line = start_line,
+				.s    = s
+			};
+
+			if (!fwdtokens_append(tokens, str)) {
+				fprintf(stderr, "failed appending str token\n");
+				free(s);
+				return -1;
+			}
+
+			start = end + 1;
+			continue;
+		}
+
+		/* otherwise, read until next separation */
+		const char *end = next_separator(tok);
+		char *s = strndup(tok, end - tok);
+		if (!s) {
+			fprintf(stderr, "failed duping id token\n");
+			return -1;
+		}
+
+		struct fwdtoken id = {
+			.kind = FWDTOKEN_ID,
+			.line = line,
+			.s    = s
+		};
+
+		if (!fwdtokens_append(tokens, id)) {
+			fprintf(stderr, "failed appending id token\n");
+			free(s);
+			return -1;
+		}
+
+		start = end;
+	}
+
+	return 0;
+}
+
+static int fwdscript_parse(struct fwdscript *s, const char *buf)
+{
+	struct fwdtokens tokens = fwdtokens_create(0);
+
+	if (fwdscript_tokenize(&tokens, buf)) {
+		fwdtokens_destroy(&tokens);
+		return -1;
+	}
+
+	/* convert tokens into some kind of structure */
+	todo();
+	fwdtokens_destroy(&tokens);
+	return 0;
+}
+
+static int fwdscript_run(struct fwdscript *s, const char *name, struct fwdvals *args)
+{
+	TODO();
+}
+
+static void fwdscript_destroy(struct fwdscript *s)
+{
+	fwdfuncs_destroy(&s->funcs);
+	fwdinsns_destroy(&s->insns);
+	fwdbuiltins_destroy(&s->builtins);
+}
+
+int main()
+{
+	int ret = 0;
+	struct fwdscript inst = fwdscript_create();
+
+	ret = fwdscript_register_builtins(&inst);
+	assert(ret == 0);
+
+	ret = fwdscript_parse(&inst,
+			"main /tbl args {\n"
+			"	get /str args 0 => {abort \"impossible\"}\n"
+			"	=> /str line;\n"
+			"	print line\n"
+			"}\n");
+
+	assert(ret == 0);
+
+	struct fwdvals args = fwdvals_create(1);
+	fwdvals_append(&args, FWDSTR(strdup("Hello world!\n")));
+
+	fwdscript_run(&inst, "main", &args);
+
+	fwdvals_destroy(&args);
+	fwdscript_destroy(&inst);
+}