initial commit

src/helper.c

@@ -0,0 +1,22 @@
+#include "helper.h"
+#include "types.h"
+
+int noom_startswith(const char* str, char* compare) {
+	noom_uint_t i = 0;
+	while (1) {
+		if (compare[i] == '\0') return 1; // we did it
+		if (str[i] != compare[i]) return 0;
+		i++;
+	}
+	return 0; // unreachable but whatevs
+}
+
+int noom_streql(const char* stra, noom_uint_t lena, const char* strb, noom_uint_t lenb) {
+	if (lena != lenb) return 0;
+
+	for (noom_uint_t i = 0; i < lena; i++) {
+		if (stra[i] != strb[i]) return 0;
+	}
+
+	return 1;
+}

src/helper.h

@@ -0,0 +1,4 @@
+#include "types.h"
+
+int noom_startswith(const char* str, char* compare);
+int noom_streql(const char* stra, noom_uint_t lena, const char* strb, noom_uint_t lenb); // rename to something better?

src/lexer.c

@@ -0,0 +1,158 @@
+#include "types.h"
+#include "helper.h"
+#include "lexer.h"
+
+int noomL_isalpha(char c) {
+	return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z');
+}
+
+int noomL_isnumber(char c) {
+	return (c >= '0' && c <= '9');
+}
+
+int noomL_isalphanum(char c) {
+	return noomL_isalpha(c) || noomL_isnumber(c);
+}
+
+int noomL_iswhitespace(char c) {
+	return c == ' ' || c == '\r' || c == '\n' || c == '\t';
+}
+
+noom_uint_t noomL_getsymbol(const char* s) { // TODO: maybe find some less shit crap holy crap
+	if (noom_startswith(s, "...")) return 3;
+	
+	if (noom_startswith(s, "==")) return 2;
+	if (noom_startswith(s, "~=")) return 2;
+	if (noom_startswith(s, "<=")) return 2;
+	if (noom_startswith(s, ">=")) return 2;
+	if (noom_startswith(s, "..")) return 2;
+	
+	if (noom_startswith(s, "+")) return 1;
+	if (noom_startswith(s, "-")) return 1;
+	if (noom_startswith(s, "*")) return 1;
+	if (noom_startswith(s, "/")) return 1;
+	if (noom_startswith(s, "%")) return 1;
+	if (noom_startswith(s, "^")) return 1;
+	if (noom_startswith(s, "#")) return 1;
+	if (noom_startswith(s, "<")) return 1;
+	if (noom_startswith(s, ">")) return 1;
+	
+	if (noom_startswith(s, "=")) return 1;
+	if (noom_startswith(s, ",")) return 1;
+	
+	if (noom_startswith(s, "(")) return 1;
+	if (noom_startswith(s, ")")) return 1;
+	if (noom_startswith(s, "{")) return 1;
+	if (noom_startswith(s, "}")) return 1;
+	if (noom_startswith(s, "[")) return 1;
+	if (noom_startswith(s, "]")) return 1;
+
+	if (noom_startswith(s, ":")) return 1;
+	if (noom_startswith(s, ".")) return 1;
+	
+	if (noom_startswith(s, ";")) return 1;
+
+	return 0; // no symbol
+}
+
+noom_uint_t noomL_getnumber(const char* s) { // TODO: more number kinds idk
+	// lazy af rn
+	noom_uint_t len = 0;
+
+	while (noomL_isnumber(s[len])) len++;
+
+	return len;
+}
+
+int noomL_iskeyword(const char* s, noom_uint_t len) {
+	if (noom_streql(s, len, "true", 4)) return 1;
+	if (noom_streql(s, len, "false", 5)) return 1;
+	if (noom_streql(s, len, "nil", 3)) return 1;
+	
+	if (noom_streql(s, len, "if", 2)) return 1;
+	if (noom_streql(s, len, "then", 4)) return 1;
+	if (noom_streql(s, len, "else", 4)) return 1;
+	if (noom_streql(s, len, "elseif", 6)) return 1;
+	
+	if (noom_streql(s, len, "and", 3)) return 1;
+	if (noom_streql(s, len, "or", 2)) return 1;
+	if (noom_streql(s, len, "not", 3)) return 1;
+	
+	if (noom_streql(s, len, "local", 5)) return 1;
+	
+	if (noom_streql(s, len, "for", 3)) return 1;
+	if (noom_streql(s, len, "function", 8)) return 1;
+	if (noom_streql(s, len, "do", 2)) return 1;
+	if (noom_streql(s, len, "until", 5)) return 1;
+	if (noom_streql(s, len, "goto", 4)) return 1;
+	if (noom_streql(s, len, "while", 5)) return 1;
+	if (noom_streql(s, len, "repeat", 6)) return 1;
+	if (noom_streql(s, len, "end", 3)) return 1;
+	if (noom_streql(s, len, "in", 2)) return 1;
+	if (noom_streql(s, len, "return", 6)) return 1;
+	if (noom_streql(s, len, "break", 5)) return 1;
+
+	return 0;
+}
+
+noomL_ErrorType noomL_lex(const char* s, noom_uint_t start, noomL_Token* token) {
+	const char* str = s + start;
+
+	if (str[0] == '\0') {
+		token->type = NOOML_TOKEN_EOF;
+		token->offset = start;
+		token->length = 0;
+
+		return NOOML_ERROR_NONE;
+	}
+
+	if (noomL_iswhitespace(str[0])) {
+		token->type = NOOML_TOKEN_WHITESPACE; // TODO: maybe handle multiple at once? for now it should be okay
+		token->offset = start;
+		token->length = 1;
+
+		return NOOML_ERROR_NONE;
+	}
+
+	if (str[0] == '_' || noomL_isalpha(str[0])) { // TODO: maybe abstract into function for "can start ident"?
+		noom_uint_t len = 1;
+		while (str[len] == '_' || noomL_isalphanum(str[len])) // same here
+			len++;
+
+		token->type = NOOML_TOKEN_IDENTIFIER;
+		if (noomL_iskeyword(str, len)) token->type = NOOML_TOKEN_KEYWORD;
+		
+		token->offset = start;
+		token->length = len;
+
+		return NOOML_ERROR_NONE;
+	}
+
+	{
+		noom_uint_t symbolLen = noomL_getsymbol(str);
+
+		if (symbolLen) {
+			token->type = NOOML_TOKEN_SYMBOL;
+			token->offset = start;
+			token->length = symbolLen;
+
+			return NOOML_ERROR_NONE;
+		}
+	}
+
+	{
+		noom_uint_t numberLen = noomL_getnumber(str);
+
+		if (numberLen) {
+			token->type = NOOML_TOKEN_NUMBER;
+			token->offset = start;
+			token->length = numberLen;
+
+			return NOOML_ERROR_NONE;
+		}
+	}
+
+	// god damn it we errorrreed
+	return NOOML_ERROR_UNKNOWN;
+}
+

src/lexer.h

@@ -0,0 +1,36 @@
+#include "types.h"
+
+typedef enum noomL_TokenType {
+	NOOML_TOKEN_EOF = 0,
+	NOOML_TOKEN_KEYWORD,
+	NOOML_TOKEN_WHITESPACE,
+	NOOML_TOKEN_IDENTIFIER,
+	NOOML_TOKEN_STRING,
+	NOOML_TOKEN_SYMBOL,
+	NOOML_TOKEN_NUMBER,
+
+	NOOML_TOKEN_TCOUNT,
+} noomL_TokenType;
+
+typedef enum noomL_ErrorType {
+	NOOML_ERROR_NONE = 0,
+	NOOML_ERROR_UNKNOWN = 1,
+} noomL_ErrorType;
+
+typedef struct noomL_Token {
+	noomL_TokenType type;
+	noom_uint_t offset;
+	noom_uint_t length;
+} noomL_Token;
+
+
+int noomL_isalpha(char c);
+int noomL_isnumber(char c);
+int noomL_isalphanum(char c);
+int noomL_iswhitespace(char c);
+
+noom_uint_t noomL_getsymbol(const char* s);
+noom_uint_t noomL_getnumber(const char* s);
+
+noomL_ErrorType noomL_lex(const char* s, noom_uint_t start, noomL_Token* token); // TODO: add more error data
+

src/main.c

@@ -0,0 +1,24 @@
+#include <stdio.h> // for now
+#include "lexer.h"
+
+int main(int argc, char** argv) {
+	// uhh uhhh uhhhhh
+	const char* code = "local a = 5;";
+	noom_uint_t pos = 0;
+
+	noomL_Token token;
+	while (1) {
+		noomL_lex(code, pos, &token);
+
+		printf("%d ", token.type);
+		for (noom_uint_t i = 0; i < token.length; i++) putchar((code + token.offset)[i]);
+		putchar('\n');
+
+		pos += token.length;
+
+		if (token.type == NOOML_TOKEN_EOF) break;
+	}
+	
+	
+	return 0;
+}

src/types.h

@@ -0,0 +1 @@
+typedef unsigned long int noom_uint_t;