semos/compiler.h

#ifndef COMPILER_H
#define COMPILER_H

#include <stdbool.h>
#include <stddef.h>

typedef struct {
    size_t index;
    int line;
    int col;
} Pos;

void print_error(const char* message, Pos pos);

typedef enum {
    TokenType_Error,
    TokenType_EOF,
    TokenType_Id,
    TokenType_Int,
    TokenType_Not,
    TokenType_And,
    TokenType_Or,
    TokenType_If,
    TokenType_Else,
    TokenType_Loop,
    TokenType_Let,
    TokenType_Fn,
    TokenType_Return,
    TokenType_Break,
    TokenType_LParen,
    TokenType_RParen,
    TokenType_LBrace,
    TokenType_RBrace,
    TokenType_LBracket,
    TokenType_RBracket,
    TokenType_Comma,
    TokenType_Semicolon,
    TokenType_Plus,
    TokenType_PlusEqual,
    TokenType_Minus,
    TokenType_MinusEqual,
    TokenType_MinusGT,
    TokenType_Asterisk,
    TokenType_AsteriskEqual,
    TokenType_Equal,
    TokenType_EqualEqual,
    TokenType_Exclamation,
    TokenType_ExclamationEqual,
    TokenType_LT,
    TokenType_LTEqual,
    TokenType_GT,
    TokenType_GTEqual,
    TokenType_Pipe,
    TokenType_PipeGT,
} TokenType;

typedef struct {
    TokenType token_type;
    Pos pos;
    size_t length;
} Token;

typedef struct {
    const char* text;
    size_t text_length;
    size_t index;
    int line;
    int col;
    bool failed;
} Lexer;

void lexer_construct(Lexer* lexer, const char* text, size_t text_length);
Token lexer_next(Lexer* lexer);
bool lexer_failed(const Lexer* lexer);
Token lexer_token(Lexer* lexer, TokenType token_type, Pos pos);
void lexer_step(Lexer* lexer);
bool lexer_done(const Lexer* lexer);
char lexer_current(const Lexer* lexer);
Pos lexer_pos(const Lexer* lexer);

typedef enum {
    ASTNodeType_Error,
    ASTNodeType_Id,
    ASTNodeType_Int,
    ASTNodeType_Group,
    ASTNodeType_Block,
    ASTNodeType_If,
    ASTNodeType_Loop,
    ASTNodeType_Call,
    ASTNodeType_Index,
    ASTNodeType_Unary,
    ASTNodeType_Binary,
    ASTNodeType_Assign,
    ASTNodeType_Let,
    ASTNodeType_Break,
    ASTNodeType_Fn,
    ASTNodeType_Return,
    ASTNodeType_Statements,
} ASTNodeType;

typedef struct ASTNode ASTNode;

typedef struct {
    ASTNode** data;
    size_t length;
    size_t capacity;
} ASTNodeVec;

int ast_node_vec_construct(ASTNodeVec* vec);
void ast_node_vec_destroy(ASTNodeVec* vec);
int ast_node_vec_push(ASTNodeVec* vec, ASTNode* item);

typedef struct {
    ASTNode* condition;
    ASTNode* truthy;
    ASTNode* falsy;
} ASTIfNode;

typedef struct {
    ASTNode* body;
} ASTLoopNode;

typedef struct {
    ASTNode* subject;
    ASTNodeVec args;
} ASTCallNode;

typedef struct {
    ASTNode* subject;
    ASTNode* value;
} ASTIndexNode;

typedef enum {
    UnaryType_Not,
    UnaryType_Negate,
} UnaryType;

typedef struct {
    UnaryType unary_type;
    ASTNode* subject;
} ASTUnaryNode;

typedef enum {
    BinaryType_And,
    BinaryType_Or,
    BinaryType_EE,
    BinaryType_NE,
    BinaryType_LT,
    BinaryType_GT,
    BinaryType_LTE,
    BinaryType_GTE,
    BinaryType_Add,
    BinaryType_Subtract,
    BinaryType_Multiply,
} BinaryType;

typedef struct {
    BinaryType binary_type;
    ASTNode* left;
    ASTNode* right;
} ASTBinaryNode;

typedef enum {
    AssignType_Assign,
    AssignType_Add,
    AssignType_Subtract,
    AssignType_Multiply,
} AssignType;

typedef struct {
    AssignType assign_type;
    ASTNode* subject;
    ASTNode* value;
} ASTAssignNode;

typedef struct {
    ASTNode* id;
    ASTNode* value;
} ASTLetNode;

typedef struct {
    ASTNode* id;
    ASTNodeVec params;
    ASTNode* body;
} ASTFnNode;

typedef struct {
    ASTNode* value;
} ASTReturnNode;

typedef struct Symbol Symbol;

struct ASTNode {
    ASTNodeType node_type;
    Pos pos;
    union {
        struct {
            char* id_value;
            Symbol* id_symbol;
        };
        int int_value;
        ASTNode* group_value;
        ASTNodeVec statements;
        ASTIfNode if_node;
        ASTLoopNode loop_node;
        ASTCallNode call_node;
        ASTIndexNode index_node;
        ASTUnaryNode unary_node;
        ASTBinaryNode binary_node;
        ASTAssignNode assign_node;
        ASTLetNode let_node;
        ASTFnNode fn_node;
        ASTReturnNode return_node;
    };
};

ASTNode* ast_node_new(ASTNodeType node_type, Pos pos, ASTNode spec_init);
void ast_node_free(ASTNode* node);

typedef struct {
    const char* text;
    size_t text_length;
    Lexer lexer;
    Token current;
    bool failed;
} Parser;

void parser_construct(Parser* parser, const char* text, size_t text_length);
bool parser_failed(const Parser* parser);
void parser_step(Parser* parser);
bool parser_done(const Parser* parser);
ASTNode* parser_parse_statements(Parser* parser);
ASTNode* parser_parse_statement(Parser* parser);
ASTNode* parser_parse_fn(Parser* parser);
ASTNode* parser_parse_single_line_statement(Parser* parser);
ASTNode* parser_parse_let(Parser* parser);
ASTNode* parser_parse_return(Parser* parser);
ASTNode* parser_parse_break(Parser* parser);
ASTNode* parser_parse_assign(Parser* parser);
ASTNode* parser_parse_expr(Parser* parser);
ASTNode* parser_parse_or(Parser* parser);
ASTNode* parser_parse_and(Parser* parser);
ASTNode* parser_parse_equality(Parser* parser);
ASTNode* parser_parse_comparison(Parser* parser);
ASTNode* parser_parse_term(Parser* parser);
ASTNode* parser_parse_factor(Parser* parser);
ASTNode* parser_parse_unary(Parser* parser);
ASTNode* parser_parse_index_call(Parser* parser);
ASTNode* parser_parse_operand(Parser* parser);
ASTNode* parser_parse_id(Parser* parser);
ASTNode* parser_parse_int(Parser* parser);
ASTNode* parser_parse_group(Parser* parser);
ASTNode* parser_parse_block(Parser* parser);
ASTNode* parser_parse_if(Parser* parser);
ASTNode* parser_parse_loop(Parser* parser);

typedef enum {
    SymbolType_Local,
    SymbolType_Global,
    SymbolType_Builtin,
} SymbolType;

struct Symbol {
    SymbolType type;
    union {
        int local_location;
    };
};

size_t string_hash_djb2(const char* value);

typedef struct {
    size_t key_hash;
    Symbol value;
} StringSymbolMapEntry;

typedef struct {
    StringSymbolMapEntry* data;
    size_t length;
    size_t capacity;
} StringSymbolMap;

int string_symbol_map_construct(StringSymbolMap* map);
void string_symbol_map_destroy(StringSymbolMap* map);
Symbol* string_symbol_map_get(const StringSymbolMap* map, size_t key_hash);
int string_symbol_map_set(StringSymbolMap* map, size_t key_hash, Symbol value);

typedef struct SymbolTable SymbolTable;

struct SymbolTable {
    SymbolTable* parent;
    StringSymbolMap map;
};

void symbol_table_construct(SymbolTable* table, SymbolTable* parent);
void symbol_table_destroy(SymbolTable* table);
Symbol* symbol_table_resolve(SymbolTable* table, const char* id);
Symbol* symbol_table_resolve_hash(SymbolTable* table, size_t hash);
Symbol* symbol_table_resolve_local(SymbolTable* table, const char* id);
void symbol_table_define(SymbolTable* table, const char* id, Symbol symbol);

typedef struct {
    ASTNode* statements;
    SymbolTable table;
} StatementsSymbols;

typedef struct {
    StatementsSymbols* data;
    size_t length;
    size_t capacity;
} StatementsSymbolsVec;

int statements_symbols_vec_construct(StatementsSymbolsVec* vec);
void statements_symbols_vec_destroy(StatementsSymbolsVec* vec);
int statements_symbols_vec_push(
    StatementsSymbolsVec* vec, StatementsSymbols pair);

typedef struct {
    bool failed;
    SymbolTable head_table;
    StatementsSymbolsVec statements_symbols;
} Checker;

typedef struct {
    SymbolTable head_table;
    StatementsSymbolsVec statements_symbols;
} CheckerResult;

void checker_construct(Checker* checker);
void checker_destroy(Checker* checker);
bool checker_failed(const Checker* checker);
CheckerResult checker_result(Checker* checker);
void checker_check_statements(
    Checker* checker, SymbolTable* table, ASTNode* node);
void checker_check_statement(
    Checker* checker, SymbolTable* table, ASTNode* node);
void checker_check_assign(Checker* checker, SymbolTable* table, ASTNode* node);
void checker_check_let(Checker* checker, SymbolTable* table, ASTNode* node);
void checker_check_fn(Checker* checker, SymbolTable* table, ASTNode* node);
void checker_check_expr(Checker* checker, SymbolTable* table, ASTNode* node);
void checker_check_id(Checker* checker, SymbolTable* table, ASTNode* node);

#endif