parser.h

Go to the documentation of this file.

 
#ifndef PRISM_PARSER_H
#define PRISM_PARSER_H
 
#include "prism/defines.h"
#include "prism/ast.h"
#include "prism/encoding.h"
#include "prism/options.h"
#include "prism/static_literals.h"
#include "prism/util/pm_constant_pool.h"
#include "prism/util/pm_list.h"
#include "prism/util/pm_newline_list.h"
#include "prism/util/pm_string.h"
 
#include <stdbool.h>
 
typedef enum {
    PM_LEX_STATE_BIT_BEG,
    PM_LEX_STATE_BIT_END,
    PM_LEX_STATE_BIT_ENDARG,
    PM_LEX_STATE_BIT_ENDFN,
    PM_LEX_STATE_BIT_ARG,
    PM_LEX_STATE_BIT_CMDARG,
    PM_LEX_STATE_BIT_MID,
    PM_LEX_STATE_BIT_FNAME,
    PM_LEX_STATE_BIT_DOT,
    PM_LEX_STATE_BIT_CLASS,
    PM_LEX_STATE_BIT_LABEL,
    PM_LEX_STATE_BIT_LABELED,
    PM_LEX_STATE_BIT_FITEM
} pm_lex_state_bit_t;
 
typedef enum {
    PM_LEX_STATE_NONE = 0,
    PM_LEX_STATE_BEG = (1 << PM_LEX_STATE_BIT_BEG),
    PM_LEX_STATE_END = (1 << PM_LEX_STATE_BIT_END),
    PM_LEX_STATE_ENDARG = (1 << PM_LEX_STATE_BIT_ENDARG),
    PM_LEX_STATE_ENDFN = (1 << PM_LEX_STATE_BIT_ENDFN),
    PM_LEX_STATE_ARG = (1 << PM_LEX_STATE_BIT_ARG),
    PM_LEX_STATE_CMDARG = (1 << PM_LEX_STATE_BIT_CMDARG),
    PM_LEX_STATE_MID = (1 << PM_LEX_STATE_BIT_MID),
    PM_LEX_STATE_FNAME = (1 << PM_LEX_STATE_BIT_FNAME),
    PM_LEX_STATE_DOT = (1 << PM_LEX_STATE_BIT_DOT),
    PM_LEX_STATE_CLASS = (1 << PM_LEX_STATE_BIT_CLASS),
    PM_LEX_STATE_LABEL = (1 << PM_LEX_STATE_BIT_LABEL),
    PM_LEX_STATE_LABELED = (1 << PM_LEX_STATE_BIT_LABELED),
    PM_LEX_STATE_FITEM = (1 << PM_LEX_STATE_BIT_FITEM),
    PM_LEX_STATE_BEG_ANY = PM_LEX_STATE_BEG | PM_LEX_STATE_MID | PM_LEX_STATE_CLASS,
    PM_LEX_STATE_ARG_ANY = PM_LEX_STATE_ARG | PM_LEX_STATE_CMDARG,
    PM_LEX_STATE_END_ANY = PM_LEX_STATE_END | PM_LEX_STATE_ENDARG | PM_LEX_STATE_ENDFN
} pm_lex_state_t;
 
typedef enum {
    PM_HEREDOC_QUOTE_NONE,
    PM_HEREDOC_QUOTE_SINGLE = '\'',
    PM_HEREDOC_QUOTE_DOUBLE = '"',
    PM_HEREDOC_QUOTE_BACKTICK = '`',
} pm_heredoc_quote_t;
 
typedef enum {
    PM_HEREDOC_INDENT_NONE,
    PM_HEREDOC_INDENT_DASH,
    PM_HEREDOC_INDENT_TILDE,
} pm_heredoc_indent_t;
 
typedef struct {
    const uint8_t *ident_start;
 
    size_t ident_length;
 
    pm_heredoc_quote_t quote;
 
    pm_heredoc_indent_t indent;
} pm_heredoc_lex_mode_t;
 
typedef struct pm_lex_mode {
    enum {
        PM_LEX_DEFAULT,
 
        PM_LEX_EMBEXPR,
 
        PM_LEX_EMBVAR,
 
        PM_LEX_HEREDOC,
 
        PM_LEX_LIST,
 
        PM_LEX_REGEXP,
 
        PM_LEX_STRING
    } mode;
 
    union {
        struct {
            size_t nesting;
 
            bool interpolation;
 
            uint8_t incrementor;
 
            uint8_t terminator;
 
            uint8_t breakpoints[11];
        } list;
 
        struct {
            size_t nesting;
 
            uint8_t incrementor;
 
            uint8_t terminator;
 
            uint8_t breakpoints[7];
        } regexp;
 
        struct {
            size_t nesting;
 
            bool interpolation;
 
            bool label_allowed;
 
            uint8_t incrementor;
 
            uint8_t terminator;
 
            uint8_t breakpoints[7];
        } string;
 
        struct {
            pm_heredoc_lex_mode_t base;
 
            const uint8_t *next_start;
 
            size_t *common_whitespace;
 
            bool line_continuation;
        } heredoc;
    } as;
 
    struct pm_lex_mode *prev;
} pm_lex_mode_t;
 
#define PM_LEX_STACK_SIZE 4
 
typedef struct pm_parser pm_parser_t;
 
typedef enum {
    PM_CONTEXT_NONE = 0,
 
    PM_CONTEXT_BEGIN,
 
    PM_CONTEXT_BEGIN_ENSURE,
 
    PM_CONTEXT_BEGIN_ELSE,
 
    PM_CONTEXT_BEGIN_RESCUE,
 
    PM_CONTEXT_BLOCK_BRACES,
 
    PM_CONTEXT_BLOCK_KEYWORDS,
 
    PM_CONTEXT_BLOCK_ENSURE,
 
    PM_CONTEXT_BLOCK_ELSE,
 
    PM_CONTEXT_BLOCK_PARAMETERS,
 
    PM_CONTEXT_BLOCK_RESCUE,
 
    PM_CONTEXT_CASE_WHEN,
 
    PM_CONTEXT_CASE_IN,
 
    PM_CONTEXT_CLASS,
 
    PM_CONTEXT_CLASS_ENSURE,
 
    PM_CONTEXT_CLASS_ELSE,
 
    PM_CONTEXT_CLASS_RESCUE,
 
    PM_CONTEXT_DEF,
 
    PM_CONTEXT_DEF_ENSURE,
 
    PM_CONTEXT_DEF_ELSE,
 
    PM_CONTEXT_DEF_RESCUE,
 
    PM_CONTEXT_DEF_PARAMS,
 
    PM_CONTEXT_DEFINED,
 
    PM_CONTEXT_DEFAULT_PARAMS,
 
    PM_CONTEXT_ELSE,
 
    PM_CONTEXT_ELSIF,
 
    PM_CONTEXT_EMBEXPR,
 
    PM_CONTEXT_FOR,
 
    PM_CONTEXT_FOR_INDEX,
 
    PM_CONTEXT_IF,
 
    PM_CONTEXT_LAMBDA_BRACES,
 
    PM_CONTEXT_LAMBDA_DO_END,
 
    PM_CONTEXT_LAMBDA_ENSURE,
 
    PM_CONTEXT_LAMBDA_ELSE,
 
    PM_CONTEXT_LAMBDA_RESCUE,
 
    PM_CONTEXT_LOOP_PREDICATE,
 
    PM_CONTEXT_MAIN,
 
    PM_CONTEXT_MODULE,
 
    PM_CONTEXT_MODULE_ENSURE,
 
    PM_CONTEXT_MODULE_ELSE,
 
    PM_CONTEXT_MODULE_RESCUE,
 
    PM_CONTEXT_MULTI_TARGET,
 
    PM_CONTEXT_PARENS,
 
    PM_CONTEXT_POSTEXE,
 
    PM_CONTEXT_PREDICATE,
 
    PM_CONTEXT_PREEXE,
 
    PM_CONTEXT_RESCUE_MODIFIER,
 
    PM_CONTEXT_SCLASS,
 
    PM_CONTEXT_SCLASS_ENSURE,
 
    PM_CONTEXT_SCLASS_ELSE,
 
    PM_CONTEXT_SCLASS_RESCUE,
 
    PM_CONTEXT_TERNARY,
 
    PM_CONTEXT_UNLESS,
 
    PM_CONTEXT_UNTIL,
 
    PM_CONTEXT_WHILE,
} pm_context_t;
 
typedef struct pm_context_node {
    pm_context_t context;
 
    struct pm_context_node *prev;
} pm_context_node_t;
 
typedef enum {
    PM_COMMENT_INLINE,
    PM_COMMENT_EMBDOC
} pm_comment_type_t;
 
typedef struct pm_comment {
    pm_list_node_t node;
 
    pm_location_t location;
 
    pm_comment_type_t type;
} pm_comment_t;
 
typedef struct {
    pm_list_node_t node;
 
    const uint8_t *key_start;
 
    const uint8_t *value_start;
 
    uint32_t key_length;
 
    uint32_t value_length;
} pm_magic_comment_t;
 
typedef void (*pm_encoding_changed_callback_t)(pm_parser_t *parser);
 
typedef struct {
    void *data;
 
    void (*callback)(void *data, pm_parser_t *parser, pm_token_t *token);
} pm_lex_callback_t;
 
typedef uint8_t pm_shareable_constant_value_t;
static const pm_shareable_constant_value_t PM_SCOPE_SHAREABLE_CONSTANT_NONE = 0x0;
static const pm_shareable_constant_value_t PM_SCOPE_SHAREABLE_CONSTANT_LITERAL = PM_SHAREABLE_CONSTANT_NODE_FLAGS_LITERAL;
static const pm_shareable_constant_value_t PM_SCOPE_SHAREABLE_CONSTANT_EXPERIMENTAL_EVERYTHING = PM_SHAREABLE_CONSTANT_NODE_FLAGS_EXPERIMENTAL_EVERYTHING;
static const pm_shareable_constant_value_t PM_SCOPE_SHAREABLE_CONSTANT_EXPERIMENTAL_COPY = PM_SHAREABLE_CONSTANT_NODE_FLAGS_EXPERIMENTAL_COPY;
 
typedef struct {
    pm_constant_id_t name;
 
    pm_location_t location;
 
    uint32_t index;
 
    uint32_t reads;
 
    uint32_t hash;
} pm_local_t;
 
typedef struct pm_locals {
    uint32_t size;
 
    uint32_t capacity;
 
    pm_local_t *locals;
} pm_locals_t;
 
typedef uint8_t pm_scope_parameters_t;
static const pm_scope_parameters_t PM_SCOPE_PARAMETERS_NONE = 0x0;
static const pm_scope_parameters_t PM_SCOPE_PARAMETERS_FORWARDING_POSITIONALS = 0x1;
static const pm_scope_parameters_t PM_SCOPE_PARAMETERS_FORWARDING_KEYWORDS = 0x2;
static const pm_scope_parameters_t PM_SCOPE_PARAMETERS_FORWARDING_BLOCK = 0x4;
static const pm_scope_parameters_t PM_SCOPE_PARAMETERS_FORWARDING_ALL = 0x8;
static const pm_scope_parameters_t PM_SCOPE_PARAMETERS_IMPLICIT_DISALLOWED = 0x10;
static const pm_scope_parameters_t PM_SCOPE_PARAMETERS_NUMBERED_INNER = 0x20;
static const pm_scope_parameters_t PM_SCOPE_PARAMETERS_NUMBERED_FOUND = 0x40;
 
typedef struct pm_scope {
    struct pm_scope *previous;
 
    pm_locals_t locals;
 
    pm_node_list_t implicit_parameters;
 
    pm_scope_parameters_t parameters;
 
    pm_shareable_constant_value_t shareable_constant;
 
    bool closed;
} pm_scope_t;
 
typedef uint32_t pm_state_stack_t;
 
struct pm_parser {
    uint32_t node_id;
 
    pm_lex_state_t lex_state;
 
    int enclosure_nesting;
 
    int lambda_enclosure_nesting;
 
    int brace_nesting;
 
    pm_state_stack_t do_loop_stack;
 
    pm_state_stack_t accepts_block_stack;
 
    struct {
        pm_lex_mode_t *current;
 
        pm_lex_mode_t stack[PM_LEX_STACK_SIZE];
 
        size_t index;
    } lex_modes;
 
    const uint8_t *start;
 
    const uint8_t *end;
 
    pm_token_t previous;
 
    pm_token_t current;
 
    const uint8_t *next_start;
 
    const uint8_t *heredoc_end;
 
    pm_list_t comment_list;
 
    pm_list_t magic_comment_list;
 
    pm_location_t data_loc;
 
    pm_list_t warning_list;
 
    pm_list_t error_list;
 
    pm_scope_t *current_scope;
 
    pm_context_node_t *current_context;
 
    pm_static_literals_t *current_hash_keys;
 
    const pm_encoding_t *encoding;
 
    pm_encoding_changed_callback_t encoding_changed_callback;
 
    const uint8_t *encoding_comment_start;
 
    pm_lex_callback_t *lex_callback;
 
    pm_string_t filepath;
 
    pm_constant_pool_t constant_pool;
 
    pm_newline_list_t newline_list;
 
    pm_node_flags_t integer_base;
 
    pm_string_t current_string;
 
    int32_t start_line;
 
    const pm_encoding_t *explicit_encoding;
 
    pm_node_list_t *current_block_exits;
 
    pm_options_version_t version;
 
    uint8_t command_line;
 
    int8_t frozen_string_literal;
 
    bool parsing_eval;
 
    bool partial_script;
 
    bool command_start;
 
    bool recovering;
 
    bool encoding_locked;
 
    bool encoding_changed;
 
    bool pattern_matching_newlines;
 
    bool in_keyword_arg;
 
    bool semantic_token_seen;
 
    bool current_regular_expression_ascii_only;
 
    bool warn_mismatched_indentation;
};
 
#endif