asm: implement asembly lexer

asm/CMakeLists.txt

@@ -2,5 +2,5 @@ file(GLOB_RECURSE asm_sources *.c *.h include/ivy/asm/*.h)
 
 add_library(ivy-asm SHARED ${asm_sources})
 target_include_directories(ivy-asm PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/include/)
-target_link_libraries(ivy-asm ivy-common)
+target_link_libraries(ivy-asm ivy-common Bluelib::Core Bluelib::Object)
 target_compile_definitions(ivy-asm PRIVATE IVY_EXPORT=1)

asm/include/ivy/asm/lex.h

@@ -1,8 +1,89 @@
 #ifndef IVY_ASM_LEX_H_
 #define IVY_ASM_LEX_H_
 
+#include <ivy/line-source.h>
 #include <ivy/misc.h>
+#include <ivy/status.h>
+#include <stdbool.h>
 
-IVY_API void placeholder12(void);
+enum ivy_asm_token_type {
+	IVY_ASM_TOK_NONE = 0,
+	IVY_ASM_TOK_KEYWORD,
+	IVY_ASM_TOK_SYMBOL,
+	IVY_ASM_TOK_INT,
+	IVY_ASM_TOK_DOUBLE,
+	IVY_ASM_TOK_LABEL,
+	IVY_ASM_TOK_IDENT,
+	IVY_ASM_TOK_STRING,
+	IVY_ASM_TOK_LINEFEED,
+};
 
-#endif
+enum ivy_asm_keyword {
+	IVY_ASM_KW_NONE = 0,
+	IVY_ASM_KW_USE,
+	IVY_ASM_KW_IDENT,
+	IVY_ASM_KW_SELECTOR,
+	IVY_ASM_KW_ATOM,
+	IVY_ASM_KW_LAMBDA,
+	IVY_ASM_KW_CONSTPOOL,
+	IVY_ASM_KW_CLASS,
+	IVY_ASM_KW_MSGH,
+	IVY_ASM_KW_END,
+};
+
+enum ivy_asm_symbol {
+	IVY_ASM_SYM_NONE = 0,
+	IVY_ASM_SYM_DOT,
+	IVY_ASM_SYM_COMMA,
+	IVY_ASM_SYM_LEFT_PAREN,
+	IVY_ASM_SYM_RIGHT_PAREN,
+	IVY_ASM_SYM_LEFT_BRACKET,
+	IVY_ASM_SYM_RIGHT_BRACKET,
+	IVY_ASM_SYM_COLON,
+	IVY_ASM_SYM_SEMICOLON,
+	IVY_ASM_SYM_DOLLAR,
+	IVY_ASM_SYM_HYPHEN,
+	IVY_ASM_SYM_SQUOTE,
+	IVY_ASM_SYM_DQUOTE,
+	IVY_ASM_SYM_FORWARD_SLASH_ASTERISK,
+	IVY_ASM_SYM_ASTERISK_FORWARD_SLASH,
+};
+
+struct ivy_asm_token {
+	enum ivy_asm_token_type t_type;
+	struct ivy_asm_token *t_next;
+
+	union {
+		enum ivy_asm_keyword t_keyword;
+		enum ivy_asm_symbol t_symbol;
+		struct {
+			long long v;
+			unsigned long long uv;
+			bool sign;
+		} t_int;
+		double t_double;
+		char *t_str;
+	};
+};
+
+struct ivy_asm_lexer_symbol_node;
+struct ivy_asm_lexer_state;
+struct ivy_asm_lexer;
+
+IVY_API enum ivy_status ivy_asm_lexer_create(struct ivy_asm_lexer **lex);
+IVY_API void ivy_asm_lexer_destroy(struct ivy_asm_lexer *lex);
+
+IVY_API void ivy_asm_lexer_set_source(
+	struct ivy_asm_lexer *lex, struct ivy_line_source *src);
+IVY_API enum ivy_status ivy_asm_lexer_get_status(struct ivy_asm_lexer *lex);
+
+IVY_API struct ivy_asm_token *ivy_asm_lexer_peek(struct ivy_asm_lexer *lex);
+IVY_API struct ivy_asm_token *ivy_asm_lexer_read(struct ivy_asm_lexer *lex);
+
+IVY_API void ivy_asm_token_destroy(struct ivy_asm_token *tok);
+
+IVY_API const char *ivy_asm_token_type_to_string(enum ivy_asm_token_type type);
+IVY_API const char *ivy_asm_keyword_to_string(enum ivy_asm_keyword keyword);
+IVY_API const char *ivy_asm_symbol_to_string(enum ivy_asm_symbol sym);
+
+#endif

asm/lex.h

@@ -0,0 +1,53 @@
+#ifndef _LEX_H_
+#define _LEX_H_
+
+#include <blue/core/queue.h>
+#include <blue/object/string.h>
+#include <blue/object/dict.h>
+#include <stdint.h>
+
+enum lexer_state_type {
+	STATE_NORMAL,
+	STATE_STRING,
+	STATE_DSTRING,
+};
+
+struct lexer_state {
+	enum lexer_state_type s_type;
+	b_queue_entry s_entry;
+};
+
+struct ivy_asm_lexer_symbol_node {
+	char s_char;
+	enum ivy_symbol s_id;
+
+	b_queue_entry s_entry;
+	b_queue s_children;
+};
+
+struct lex_token_def {
+	int id;
+	const char *name;
+	uint64_t name_hash;
+};
+
+struct ivy_asm_lexer {
+	struct ivy_asm_lexer_symbol_node *lex_sym_tree;
+	struct ivy_line_source *lex_source;
+	b_dict *lex_keywords;
+	enum ivy_status lex_status;
+
+	struct ivy_asm_token *lex_queue;
+	enum ivy_asm_token_type lex_prev_token;
+
+	b_string *lex_temp;
+	b_queue lex_state;
+	unsigned int lex_brace_depth;
+
+	char *lex_linebuf;
+	size_t lex_linebuf_len;
+	size_t lex_linebuf_cap;
+	size_t lex_linebuf_ptr;
+};
+
+#endif