Parcel/core-parser/inc/ast.h

#ifndef PARCEL_AST_H
#define PARCEL_AST_H

#include <parcel/utility/utility.h>
#include <logger.h>
#include <tokenizer.h>

// pac_ast_set_e: An enumeration of all sets known in Parcel's AST.
//
//  Sets are descriptions which describe a rough format of token, like
//  with variable names; the format is known, but the actual name isn't.
typedef enum
{
    PAC_AST_SET_RUNE,
    PAC_AST_SET_WORD,
    PAC_AST_SET_INTEGER,
    PAC_AST_SET_FLOAT
    
} pac_ast_set_e;

typedef enum
{
    PAC_AST_ITEM_INVALID            = 0x00,
    PAC_AST_ITEM_REFERENCE,
    PAC_AST_ITEM_LITERAL,
    PAC_AST_ITEM_SET

} pac_ast_item_e;

// pac_ast_recovery_level_e: How much higher in the call stack the program
//  flow has to go to be able to recover.
typedef enum
{
    PAC_AST_STATUS_SUCCESS,
    PAC_AST_STATUS_ERROR_HANDLED,
    PAC_AST_STATUS_STOP_ESCALATING_IN_ITEM_PARSER,
    PAC_AST_STATUS_STOP_ESCALATING_IN_VARIANT_PARSER,
    PAC_AST_STATUS_STOP_ESCALATING_IN_RULE_PARSER,
    PAC_AST_STATUS_CONTINUE_AFTER_FINDING_NEXT_VARIANT,
    PAC_AST_STATUS_CONTINUE_AFTER_FINDING_NEXT_RULE,
    PAC_AST_STATUS_NOT_RECOVERABLE,
    PAC_AST_STATUS_UNEXPECTED_FILE_END
    
} pac_ast_status_e;

typedef struct pac_ast                  pac_ast_s;
typedef struct pac_ast_rule             pac_ast_rule_s;
typedef struct pac_ast_variant          pac_ast_variant_s;
typedef struct pac_ast_string_literal   pac_ast_string_literal_s;
typedef struct pac_ast_reference        pac_ast_reference_s;
typedef struct pac_ast_item             pac_ast_item_s;

struct pac_ast
{
    usz_t                       num_rules;
    pac_ast_rule_s             *rules;
    pac_arena_s                 string_arena;
};

struct pac_ast_rule
{
    char                       *name;

    usz_t                       num_variants;
    pac_ast_variant_s          *variants;
};

struct pac_ast_variant
{
    usz_t                       num_items;
    pac_ast_item_s             *items;
};

// pac_ast_reference: Also called non-terminal, a reference is an item
//  which represents all the contents of another rule.
struct pac_ast_reference
{
    usz_t                       len_name;
    char                       *name;
};

struct pac_ast_string_literal
{
    usz_t                       length;
    char                       *string;
};

struct pac_ast_item
{
    pac_ast_item_e              type;
    union pac_ast_item_data
    {
        pac_ast_string_literal_s    string_literal;
        pac_ast_set_e               set;
        pac_ast_reference_s         reference;
    } data;
};

typedef struct pac_ast_builder
{
    usz_t                       cursor;
    pac_tlist_s                *token_list;
    bool_t                      failed;
    
    pac_logger_s               *logger;
    pac_arena_s                 string_arena;
    
} pac_ast_builder_s;

pac_ast_s               pac_build_ast                                           (pac_tlist_s tokens, pac_logger_s *logger);
void                    pac_delete_ast                                          (pac_ast_s ast);

pac_ast_status_e        pac_ast_handle_invalid_reference_name_token             (pac_ast_builder_s *builder);
pac_ast_status_e        pac_ast_handle_missing_reference_close_tag              (pac_ast_builder_s *builder);
pac_ast_status_e        pac_ast_handle_missing_rule_header_closing_sign         (pac_ast_builder_s *builder);
pac_ast_status_e        pac_ast_handle_reference_with_equals_sign               (pac_ast_builder_s *builder);
pac_ast_status_e        pac_ast_handle_missing_item_separator                   (pac_ast_builder_s *builder, char *rule_name, usz_t variant_index, usz_t item_index);
pac_ast_status_e        pac_ast_handle_missing_equals_sign_after_rule_header    (pac_ast_builder_s *builder);
pac_ast_status_e        pac_ast_handle_unknown_item_type                        (pac_ast_builder_s *builder, char *rule_name, usz_t variant_index, usz_t item_index);

char *                  pac_ast_stringify_status                                (pac_ast_status_e status);

#endif // PARCEL_AST_H