Add lexer for names and identifiers

3 files changed, 138 insertions, 0 deletions

diff --git a/include/lex.h b/include/lex.h
new file mode 100644
index 0000000..23619d5
--- /dev/null
+++ b/include/lex.h
@@ -0,0 +1,114 @@
+#ifndef CERC_LEX_H
+#define CERC_LEX_H
+#include <stdio.h>
+#include "utf8.h"
+
+#define C_EOF UTF8_INVALID
+
+enum lexical_token {
+	// Keywords
+	T_BREAK,
+	T_CONST,
+	T_CONTINUE,
+	T_DEFER,
+	T_ELSE,
+	T_ENUM,
+	T_EXPORT,
+	T_EXTERN,
+	T_FOR,
+	T_FUNC,
+	T_IF,
+	T_IMPORT,
+	T_LET,
+	T_MUT,
+	T_PUB,
+	T_RETURN,
+	T_STRUCT,
+	T_UNION,
+	T_LAST_KEYWORD,
+
+	// Builtin types
+	T_F32,
+	T_F64,
+	T_I8,
+	T_I16,
+	T_I32,
+	T_I64,
+	T_ISIZE,
+	T_U8,
+	T_U16,
+	T_U32,
+	T_U64,
+	T_USIZE,
+	T_LAST_BUILTIN_TYPE,
+
+	// Operators
+	T_ADD,
+	T_AND,
+	T_BAND,
+	T_BANG,
+	T_BNOT,
+	T_BOR,
+	T_BSHL,
+	T_BSHR,
+	T_BXOR,
+	T_COLON,
+	T_COMMA,
+	T_DIV,
+	T_DOT,
+	T_EQ,
+	T_GE,
+	T_GT,
+	T_LBRACE,
+	T_LBRACKET,
+	T_LE,
+	T_LPAREN,
+	T_LT,
+	T_MINUS,
+	T_MODULO,
+	T_NEQ,
+	T_OR,
+	T_QUESTION,
+	T_RBRACE,
+	T_RBRACKET,
+	T_RPAREN,
+	T_SEMICOLON,
+	T_STAR,
+	T_UNDERSCORE,
+	T_LAST_OPERATOR,
+
+	// Tokens with additional information
+	T_IDENT,
+	T_NAME,
+	T_NUMBER,
+	T_STRING,
+
+	// Magic values
+	T_EOF,
+	T_NONE,
+};
+
+extern const char *tokens[];
+
+struct token {
+	enum lexical_token token;
+	union {
+		const char *str;
+	} info;
+};
+
+struct lexer {
+	FILE *in;
+	uint32_t c[2];
+	char *buf;
+	size_t bufsz, buflen;
+	struct token un;
+};
+
+void lex_init(struct lexer *lexer, FILE *f);
+void lex_finish(struct lexer *lexer);
+
+enum lexical_token lex(struct lexer *lexer, struct token *out);
+void unlex(struct lexer *lexer, const struct token *in);
+
+#endif
diff --git a/include/utf8.h b/include/utf8.h
new file mode 100644
index 0000000..f29dcd3
--- /dev/null
+++ b/include/utf8.h
@@ -0,0 +1,15 @@
+#ifndef CERC_UTF8_H
+#include <limits.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <stdio.h>
+
+#define CERC_UTF8_H
+#define UTF8_MAX_SIZE 4
+#define UTF8_INVALID UINT32_MAX
+
+uint32_t utf8_decode(const char **s);
+size_t utf8_encode(char *s, uint32_t c);
+uint32_t utf8_get(FILE *f);
+
+#endif
diff --git a/include/util.h b/include/util.h
new file mode 100644
index 0000000..2a1e72a
--- /dev/null
+++ b/include/util.h
@@ -0,0 +1,9 @@
+#ifndef CERC_UTIL_H
+#define CERC_UTIL_H
+enum exit_status {
+	/* EXIT_SUCCESS = 0 (defined in stdlib.h) */
+	EXIT_USER = 1,
+	EXIT_ABNORMAL = 255,
+};
+
+#endif