ruby/prism/internal/parser.h

#ifndef PRISM_INTERNAL_PARSER_H
#define PRISM_INTERNAL_PARSER_H

#include "prism/compiler/accel.h"

#include "prism/internal/arena.h"
#include "prism/internal/comments.h"
#include "prism/internal/constant_pool.h"
#include "prism/internal/encoding.h"
#include "prism/internal/list.h"
#include "prism/internal/options.h"
#include "prism/internal/static_literals.h"
#include "prism/internal/strpbrk.h"

#include "prism/ast.h"
#include "prism/line_offset_list.h"
#include "prism/parser.h"

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

/*
 * This enum provides various bits that represent different kinds of states that
 * the lexer can track. This is used to determine which kind of token to return
 * based on the context of the parser.
 */
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;

/*
 * This enum combines the various bits from the above enum into individual
 * values that represent the various states of the lexer.
 */
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;

/*
 * The type of quote that a heredoc uses.
 */
typedef enum {
    PM_HEREDOC_QUOTE_NONE,
    PM_HEREDOC_QUOTE_SINGLE = '\'',
    PM_HEREDOC_QUOTE_DOUBLE = '"',
    PM_HEREDOC_QUOTE_BACKTICK = '`',
} pm_heredoc_quote_t;

/*
 * The type of indentation that a heredoc uses.
 */
typedef enum {
    PM_HEREDOC_INDENT_NONE,
    PM_HEREDOC_INDENT_DASH,
    PM_HEREDOC_INDENT_TILDE,
} pm_heredoc_indent_t;

/*
 * All of the information necessary to store to lexing a heredoc.
 */
typedef struct {
    /* A pointer to the start of the heredoc identifier. */
    const uint8_t *ident_start;

    /* The length of the heredoc identifier. */
    size_t ident_length;

    /* The type of quote that the heredoc uses. */
    pm_heredoc_quote_t quote;

    /* The type of indentation that the heredoc uses. */
    pm_heredoc_indent_t indent;
} pm_heredoc_lex_mode_t;

/*
 * When lexing Ruby source, the lexer has a small amount of state to tell which
 * kind of token it is currently lexing. For example, when we find the start of
 * a string, the first token that we return is a TOKEN_STRING_BEGIN token. After
 * that the lexer is now in the PM_LEX_STRING mode, and will return tokens that
 * are found as part of a string.
 */
typedef struct pm_lex_mode {
    /* The type of this lex mode. */
    enum {
        /* This state is used when any given token is being lexed. */
        PM_LEX_DEFAULT,

        /*
         * This state is used when we're lexing as normal but inside an embedded
         * expression of a string.
         */
        PM_LEX_EMBEXPR,

        /*
         * This state is used when we're lexing a variable that is embedded
         * directly inside of a string with the # shorthand.
         */
        PM_LEX_EMBVAR,

        /* This state is used when you are inside the content of a heredoc. */
        PM_LEX_HEREDOC,

        /*
         * This state is used when we are lexing a list of tokens, as in a %w
         * word list literal or a %i symbol list literal.
         */
        PM_LEX_LIST,

        /*
         * This state is used when a regular expression has been begun and we
         * are looking for the terminator.
         */
        PM_LEX_REGEXP,

        /*
         * This state is used when we are lexing a string or a string-like
         * token, as in string content with either quote or an xstring.
         */
        PM_LEX_STRING
    } mode;

    /* The data associated with this type of lex mode. */
    union {
        struct {
            /* This keeps track of the nesting level of the list. */
            size_t nesting;

            /* Whether or not interpolation is allowed in this list. */
            bool interpolation;

            /*
             * Whether any token has been emitted from this list. A word
             * separator delimits the opener from the first word, so one
             * without source characters is emitted when the list does not
             * start with whitespace.
             */
            bool started;

            /*
             * Whether the previously emitted token was a word separator. A
             * word separator delimits the last word from the terminator, so
             * one without source characters is emitted when the list does
             * not end with whitespace.
             */
            bool separated;

            /*
             * When lexing a list, it takes into account balancing the
             * terminator if the terminator is one of (), [], {}, or <>.
             */
            uint8_t incrementor;

            /* This is the terminator of the list literal. */
            uint8_t terminator;

            /*
             * This is the character set that should be used to delimit the
             * tokens within the list.
             */
            uint8_t breakpoints[PM_STRPBRK_CACHE_SIZE];
        } list;

        struct {
            /*
             * This keeps track of the nesting level of the regular expression.
             */
            size_t nesting;

            /*
             * When lexing a regular expression, it takes into account balancing
             * the terminator if the terminator is one of (), [], {}, or <>.
             */
            uint8_t incrementor;

            /* This is the terminator of the regular expression. */
            uint8_t terminator;

            /*
             * This is the character set that should be used to delimit the
             * tokens within the regular expression.
             */
            uint8_t breakpoints[PM_STRPBRK_CACHE_SIZE];
        } regexp;

        struct {
            /* This keeps track of the nesting level of the string. */
            size_t nesting;

            /* Whether or not interpolation is allowed in this string. */
            bool interpolation;

            /*
             * Whether or not at the end of the string we should allow a :,
             * which would indicate this was a dynamic symbol instead of a
             * string.
             */
            bool label_allowed;

            /*
             * When lexing a string, it takes into account balancing the
             * terminator if the terminator is one of (), [], {}, or <>.
             */
            uint8_t incrementor;

            /*
             * This is the terminator of the string. It is typically either a
             * single or double quote.
             */
            uint8_t terminator;

            /*
             * This is the character set that should be used to delimit the
             * tokens within the string.
             */
            uint8_t breakpoints[PM_STRPBRK_CACHE_SIZE];
        } string;

        struct {
            /*
             * All of the data necessary to lex a heredoc.
             */
            pm_heredoc_lex_mode_t base;

            /*
             * This is the pointer to the character where lexing should resume
             * once the heredoc has been completely processed.
             */
            const uint8_t *next_start;

            /*
             * This is used to track the amount of common whitespace on each
             * line so that we know how much to dedent each line in the case of
             * a tilde heredoc.
             */
            size_t *common_whitespace;

            /* True if the previous token ended with a line continuation. */
            bool line_continuation;
        } heredoc;
    } as;

    /* The previous lex state so that it knows how to pop. */
    struct pm_lex_mode *prev;
} pm_lex_mode_t;

/*
 * While parsing, we keep track of a stack of contexts. This is helpful for
 * error recovery so that we can pop back to a previous context when we hit a
 * token that is understood by a parent context but not by the current context.
 */
typedef enum {
    /* a null context, used for returning a value from a function */
    PM_CONTEXT_NONE = 0,

    /* a begin statement */
    PM_CONTEXT_BEGIN,

    /* an ensure statement with an explicit begin */
    PM_CONTEXT_BEGIN_ENSURE,

    /* a rescue else statement with an explicit begin */
    PM_CONTEXT_BEGIN_ELSE,

    /* a rescue statement with an explicit begin */
    PM_CONTEXT_BEGIN_RESCUE,

    /* expressions in block arguments using braces */
    PM_CONTEXT_BLOCK_BRACES,

    /* expressions in block arguments using do..end */
    PM_CONTEXT_BLOCK_KEYWORDS,

    /* an ensure statement within a do..end block */
    PM_CONTEXT_BLOCK_ENSURE,

    /* a rescue else statement within a do..end block */
    PM_CONTEXT_BLOCK_ELSE,

    /* expressions in block parameters `foo do |...| end ` */
    PM_CONTEXT_BLOCK_PARAMETERS,

    /* a rescue statement within a do..end block */
    PM_CONTEXT_BLOCK_RESCUE,

    /* a case when statements */
    PM_CONTEXT_CASE_WHEN,

    /* a case in statements */
    PM_CONTEXT_CASE_IN,

    /* a class declaration */
    PM_CONTEXT_CLASS,

    /* an ensure statement within a class statement */
    PM_CONTEXT_CLASS_ENSURE,

    /* a rescue else statement within a class statement */
    PM_CONTEXT_CLASS_ELSE,

    /* a rescue statement within a class statement */
    PM_CONTEXT_CLASS_RESCUE,

    /* a method definition */
    PM_CONTEXT_DEF,

    /* an ensure statement within a method definition */
    PM_CONTEXT_DEF_ENSURE,

    /* a rescue else statement within a method definition */
    PM_CONTEXT_DEF_ELSE,

    /* a rescue statement within a method definition */
    PM_CONTEXT_DEF_RESCUE,

    /* a method definition's parameters */
    PM_CONTEXT_DEF_PARAMS,

    /* a defined? expression */
    PM_CONTEXT_DEFINED,

    /* a method definition's default parameter */
    PM_CONTEXT_DEFAULT_PARAMS,

    /* an else clause */
    PM_CONTEXT_ELSE,

    /* an elsif clause */
    PM_CONTEXT_ELSIF,

    /* an interpolated expression */
    PM_CONTEXT_EMBEXPR,

    /* a for loop */
    PM_CONTEXT_FOR,

    /* a for loop's index */
    PM_CONTEXT_FOR_INDEX,

    /* an if statement */
    PM_CONTEXT_IF,

    /* a lambda expression with braces */
    PM_CONTEXT_LAMBDA_BRACES,

    /* a lambda expression with do..end */
    PM_CONTEXT_LAMBDA_DO_END,

    /* an ensure statement within a lambda expression */
    PM_CONTEXT_LAMBDA_ENSURE,

    /* a rescue else statement within a lambda expression */
    PM_CONTEXT_LAMBDA_ELSE,

    /* a rescue statement within a lambda expression */
    PM_CONTEXT_LAMBDA_RESCUE,

    /* the predicate clause of a loop statement */
    PM_CONTEXT_LOOP_PREDICATE,

    /* the top level context */
    PM_CONTEXT_MAIN,

    /* a module declaration */
    PM_CONTEXT_MODULE,

    /* an ensure statement within a module statement */
    PM_CONTEXT_MODULE_ENSURE,

    /* a rescue else statement within a module statement */
    PM_CONTEXT_MODULE_ELSE,

    /* a rescue statement within a module statement */
    PM_CONTEXT_MODULE_RESCUE,

    /* a multiple target expression */
    PM_CONTEXT_MULTI_TARGET,

    /* a parenthesized expression */
    PM_CONTEXT_PARENS,

    /* an END block */
    PM_CONTEXT_POSTEXE,

    /* a predicate inside an if/elsif/unless statement */
    PM_CONTEXT_PREDICATE,

    /* a BEGIN block */
    PM_CONTEXT_PREEXE,

    /* a modifier rescue clause */
    PM_CONTEXT_RESCUE_MODIFIER,

    /* a singleton class definition */
    PM_CONTEXT_SCLASS,

    /* an ensure statement with a singleton class */
    PM_CONTEXT_SCLASS_ENSURE,

    /* a rescue else statement with a singleton class */
    PM_CONTEXT_SCLASS_ELSE,

    /* a rescue statement with a singleton class */
    PM_CONTEXT_SCLASS_RESCUE,

    /* a ternary expression */
    PM_CONTEXT_TERNARY,

    /* an unless statement */
    PM_CONTEXT_UNLESS,

    /* an until statement */
    PM_CONTEXT_UNTIL,

    /* a while statement */
    PM_CONTEXT_WHILE,

    /* the number of contexts, which is not itself a context */
    PM_CONTEXT_MAXIMUM,
} pm_context_t;

/* This is a node in a linked list of contexts. */
typedef struct pm_context_node {
    /* The context that this node represents. */
    pm_context_t context;

    /* A pointer to the previous context in the linked list. */
    struct pm_context_node *prev;

    /* One bit set per context in this list, including one for this node. */
    uint64_t mask;
} pm_context_node_t;

/* The type of shareable constant value that can be set. */
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;

/*
 * This tracks an individual local variable in a certain lexical context, as
 * well as the number of times is it read.
 */
typedef struct {
    /* The name of the local variable. */
    pm_constant_id_t name;

    /* The location of the local variable in the source. */
    pm_location_t location;

    /* The index of the local variable in the local table. */
    uint32_t index;

    /* The number of times the local variable is read. */
    uint32_t reads;

    /* The hash of the local variable. */
    uint32_t hash;
} pm_local_t;

/*
 * This is a set of local variables in a certain lexical context (method, class,
 * module, etc.). We need to track how many times these variables are read in
 * order to warn if they only get written.
 */
typedef struct pm_locals {
    /* The number of local variables in the set. */
    uint32_t size;

    /* The capacity of the local variables set. */
    uint32_t capacity;

    /*
     * A bloom filter over constant IDs stored in this set. Used to quickly
     * reject lookups for names that are definitely not present, avoiding the
     * cost of a linear scan or hash probe.
     */
    uint32_t bloom;

    /* The nullable allocated memory for the local variables in the set. */
    pm_local_t *locals;
} pm_locals_t;

/* The flags about scope parameters that can be set. */
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;

/*
 * This struct represents a node in a linked list of scopes. Some scopes can see
 * into their parent scopes, while others cannot.
 */
typedef struct pm_scope {
    /* A pointer to the previous scope in the linked list. */
    struct pm_scope *previous;

    /* The IDs of the locals in the given scope. */
    pm_locals_t locals;

    /*
     * This is a list of the implicit parameters contained within the block.
     * These will be processed after the block is parsed to determine the kind
     * of parameters node that should be used and to check if any errors need to
     * be added.
     */
    pm_node_list_t implicit_parameters;

    /*
     * This is a bitfield that indicates the parameters that are being used in
     * this scope. It is a combination of the PM_SCOPE_PARAMETERS_* constants.
     * There are three different kinds of parameters that can be used in a
     * scope:
     *
     * - Ordinary parameters (e.g., def foo(bar); end)
     * - Numbered parameters (e.g., def foo; _1; end)
     * - The it parameter (e.g., def foo; it; end)
     *
     * If ordinary parameters are being used, then certain parameters can be
     * forwarded to another method/structure. Those are indicated by four
     * additional bits in the params field. For example, some combinations of:
     *
     * - def foo(*); end
     * - def foo(**); end
     * - def foo(&); end
     * - def foo(...); end
     */
    pm_scope_parameters_t parameters;

    /*
     * The current state of constant shareability for this scope. This is
     * changed by magic shareable_constant_value comments.
     */
    pm_shareable_constant_value_t shareable_constant;

    /*
     * A boolean indicating whether or not this scope can see into its parent.
     * If closed is true, then the scope cannot see into its parent.
     */
    bool closed;
} pm_scope_t;

/*
 * A struct that represents a stack of boolean values.
 */
typedef uint32_t pm_state_stack_t;

/*
 * This struct represents the overall parser. It contains a reference to the
 * source file, as well as pointers that indicate where in the source it's
 * currently parsing. It also contains the most recent and current token that
 * it's considering.
 */
struct pm_parser_t {
    /* The arena used for all AST-lifetime allocations. Caller-owned. */
    pm_arena_t *arena;

    /*
     * The arena used for parser metadata (comments, diagnostics, etc.), which
     * lives as long as the parser.
     */
    pm_arena_t metadata_arena;

    /*
     * The arena used for state that is only needed while parsing, such as
     * scopes, locals, and contexts. It is released when pm_parse returns, so
     * nothing allocated in it can be reachable from the tree or from the
     * parser's metadata.
     */
    pm_arena_t scratch_arena;

    /*
     * The next node identifier that will be assigned. This is a unique
     * identifier used to track nodes such that the syntax tree can be dropped
     * but the node can be found through another parse.
     */
    uint32_t node_id;

    /*
     * A single-entry cache for pm_parser_constant_id_raw. Avoids redundant
     * constant pool lookups when the same token is resolved multiple times
     * (e.g., once during lexing for local variable detection, and again
     * during parsing for node creation).
     */
    struct {
        const uint8_t *start;
        const uint8_t *end;
        pm_constant_id_t id;
    } constant_cache;

    /* The current state of the lexer. */
    pm_lex_state_t lex_state;

    /* Tracks the current nesting of (), [], and {}. */
    int enclosure_nesting;

    /*
     * Used to temporarily track the nesting of enclosures to determine if a {
     * is the beginning of a lambda following the parameters of a lambda.
     */
    int lambda_enclosure_nesting;

    /*
     * Used to track the nesting of braces to ensure we get the correct value
     * when we are interpolating blocks with braces.
     */
    int brace_nesting;

    /*
     * The stack used to determine if a do keyword belongs to the predicate of a
     * while, until, or for loop.
     */
    pm_state_stack_t do_loop_stack;

    /*
     * The stack used to determine if a do keyword belongs to the beginning of a
     * block.
     */
    pm_state_stack_t accepts_block_stack;

    /* A stack of lex modes, linked through their prev pointers. */
    struct {
        /* The current mode of the lexer. */
        pm_lex_mode_t *current;

        /*
         * The mode at the bottom of the stack, which is never popped. Every
         * other mode is allocated from the scratch arena.
         */
        pm_lex_mode_t base;

        /*
         * Lex modes that have been popped, kept for reuse by the next push.
         * They are allocated from the scratch arena, so they are never freed
         * individually.
         */
        pm_lex_mode_t *free;
    } lex_modes;

    /* The pointer to the start of the source. */
    const uint8_t *start;

    /* The pointer to the end of the source. */
    const uint8_t *end;

    /* The previous token we were considering. */
    pm_token_t previous;

    /* The current token we're considering. */
    pm_token_t current;

    /*
     * This is a special field set on the parser when we need the parser to jump
     * to a specific location when lexing the next token, as opposed to just
     * using the end of the previous token. Normally this is NULL.
     */
    const uint8_t *next_start;

    /*
     * This field indicates the end of a heredoc whose identifier was found on
     * the current line. If another heredoc is found on the same line, then this
     * will be moved forward to the end of that heredoc. If no heredocs are
     * found on a line then this is NULL.
     */
    const uint8_t *heredoc_end;

    /* The list of comments that have been found while parsing. */
    pm_list_t comment_list;

    /*
     * The attachments of comments to the nodes that own them, populated after
     * parsing when the attach_comments option is set.
     */
    pm_comment_attachments_t comment_attachments;

    /* The list of magic comments that have been found while parsing. */
    pm_list_t magic_comment_list;

    /*
     * An optional location that represents the location of the __END__ marker
     * and the rest of the content of the file. This content is loaded into the
     * DATA constant when the file being parsed is the main file being executed.
     */
    pm_location_t data_loc;

    /* The list of warnings that have been found while parsing. */
    pm_list_t warning_list;

    /* The list of errors that have been found while parsing. */
    pm_list_t error_list;

    /* The current local scope. */
    pm_scope_t *current_scope;

    /*
     * Scopes that have been popped, kept for reuse by the next push. Scopes
     * are allocated from the scratch arena, so they are never freed
     * individually.
     */
    pm_scope_t *free_scopes;

    /* The current parsing context. */
    pm_context_node_t *current_context;

    /*
     * Context nodes that have been popped, kept for reuse by the next push.
     * Context nodes are allocated from the scratch arena, so they are never
     * freed individually.
     */
    pm_context_node_t *free_contexts;

    /*
     * The encoding functions for the current file is attached to the parser as
     * it's parsing so that it can change with a magic comment.
     */
    const pm_encoding_t *encoding;

    /*
     * When the encoding that is being used to parse the source is changed by
     * prism, we provide the ability here to call out to a user-defined
     * function.
     */
    pm_encoding_changed_callback_t encoding_changed_callback;

    /*
     * This pointer indicates where a comment must start if it is to be
     * considered an encoding comment.
     */
    const uint8_t *encoding_comment_start;

    /*
     * When you are lexing through a file, the lexer needs all of the information
     * that the parser additionally provides (for example, the local table). So if
     * you want to properly lex Ruby, you need to actually lex it in the context of
     * the parser. In order to provide this functionality, we optionally allow a
     * struct to be attached to the parser that calls back out to a user-provided
     * callback when each token is lexed.
     */
    struct {
        /*
         * This is the callback that is called when a token is lexed. It is
         * passed the opaque data pointer, the parser, and the token that was
         * lexed.
         */
        pm_lex_callback_t callback;

        /*
         * This opaque pointer is used to provide whatever information the user
         * deemed necessary to the callback. In our case we use it to pass the
         * array that the tokens get appended into.
         */
        void *data;
    } lex_callback;

    /*
     * This is the path of the file being parsed. We use the filepath when
     * constructing SourceFileNodes.
     */
    pm_string_t filepath;

    /*
     * This constant pool keeps all of the constants defined throughout the file
     * so that we can reference them later.
     */
    pm_constant_pool_t constant_pool;

    /* This is the list of line offsets in the source file. */
    pm_line_offset_list_t line_offsets;

    /*
     * State communicated from the lexer to the parser for integer tokens.
     */
    struct {
        /*
         * A flag indicating the base of the integer (binary, octal, decimal,
         * hexadecimal). Set during lexing and read during node creation.
         */
        pm_node_flags_t base;

        /*
         * When lexing a decimal integer that fits in a uint32_t, we compute
         * the value during lexing to avoid re-scanning the digits during
         * parsing. If lexed is true, this holds the result and
         * pm_integer_parse can be skipped.
         */
        uint32_t value;

        /* Whether value holds a valid pre-computed integer. */
        bool lexed;
    } integer;

    /*
     * This string is used to pass information from the lexer to the parser. It
     * is particularly necessary because of escape sequences.
     */
    pm_string_t current_string;

    /*
     * The line number at the start of the parse. This will be used to offset
     * the line numbers of all of the locations.
     */
    int32_t start_line;

    /*
     * When a string-like expression is being lexed, any byte or escape sequence
     * that resolves to a value whose top bit is set (i.e., >= 0x80) will
     * explicitly set the encoding to the same encoding as the source.
     * Alternatively, if a unicode escape sequence is used (e.g., \\u{80}) that
     * resolves to a value whose top bit is set, then the encoding will be
     * explicitly set to UTF-8.
     *
     * The _next_ time this happens, if the encoding that is about to become the
     * explicitly set encoding does not match the previously set explicit
     * encoding, a mixed encoding error will be emitted.
     *
     * When the expression is finished being lexed, the explicit encoding
     * controls the encoding of the expression. For the most part this means
     * that the expression will either be encoded in the source encoding or
     * UTF-8. This holds for all encodings except US-ASCII. If the source is
     * US-ASCII and an explicit encoding was set that was _not_ UTF-8, then the
     * expression will be encoded as ASCII-8BIT.
     *
     * Note that if the expression is a list, different elements within the same
     * list can have different encodings, so this will get reset between each
     * element. Furthermore all of this only applies to lists that support
     * interpolation, because otherwise escapes that could change the encoding
     * are ignored.
     *
     * At first glance, it may make more sense for this to live on the lexer
     * mode, but we need it here to communicate back to the parser for character
     * literals that do not push a new lexer mode.
     */
    const pm_encoding_t *explicit_encoding;

    /*
     * When parsing block exits (e.g., break, next, redo), we need to validate
     * that they are in correct contexts. For the most part we can do this by
     * looking at our parent contexts. However, modifier while and until
     * expressions can change that context to make block exits valid. In these
     * cases, we need to keep track of the block exits and then validate them
     * after the expression has been parsed.
     *
     * We use a pointer here because we don't want to keep a whole list attached
     * since this will only be used in the context of begin/end expressions.
     */
    pm_node_list_t *current_block_exits;

    /* The version of prism that we should use to parse. */
    pm_options_version_t version;

    /* The command line flags given from the options. */
    uint8_t command_line;

    /*
     * Whether or not we have found a frozen_string_literal magic comment with
     * a true or false value.
     * May be:
     *  - PM_OPTIONS_FROZEN_STRING_LITERAL_DISABLED
     *  - PM_OPTIONS_FROZEN_STRING_LITERAL_ENABLED
     *  - PM_OPTIONS_FROZEN_STRING_LITERAL_UNSET
     */
    int8_t frozen_string_literal;

    /*
     * Whether or not we are parsing an eval string. This impacts whether or not
     * we should evaluate if block exits/yields are valid.
     */
    bool parsing_eval;

    /*
     * Whether or not we are parsing a "partial" script, which is a script that
     * will be evaluated in the context of another script, so we should not
     * check jumps (next/break/etc.) for validity.
     */
    bool partial_script;

    /*
     * Whether or not comments should be attached to their owning nodes once
     * parsing has finished.
     */
    bool attach_comments;

    /* Whether or not we're at the beginning of a command. */
    bool command_start;

    /*
     * Whether or not we're currently parsing the body of an endless method
     * definition. In this context, PM_TOKEN_KEYWORD_DO_BLOCK should not be
     * consumed by commands (it should bubble up to the outer context).
     */
    bool in_endless_def_body;

    /* Whether or not we're currently recovering from a syntax error. */
    bool recovering;

    /*
     * Whether or not the source being parsed could become valid if more input
     * were appended. This is set to false when the parser encounters a token
     * that is definitively wrong (e.g., a stray `end` or `]`) as opposed to
     * merely incomplete.
     */
    bool continuable;

    /*
     * This is very specialized behavior for when you want to parse in a context
     * that does not respect encoding comments. Its main use case is translating
     * into the whitequark/parser AST which re-encodes source files in UTF-8
     * before they are parsed and ignores encoding comments.
     */
    bool encoding_locked;

    /*
     * Whether or not the encoding has been changed by a magic comment. We use
     * this to provide a fast path for the lexer instead of going through the
     * function pointer.
     */
    bool encoding_changed;

    /*
     * This flag indicates that we are currently parsing a pattern matching
     * expression and impacts that calculation of newlines.
     */
    bool pattern_matching_newlines;

    /* This flag indicates that we are currently parsing a keyword argument. */
    bool in_keyword_arg;

    /*
     * Whether or not the parser has seen a token that has semantic meaning
     * (i.e., a token that is not a comment or whitespace).
     */
    bool semantic_token_seen;

    /*
     * By default, Ruby always warns about mismatched indentation. This can be
     * toggled with a magic comment.
     */
    bool warn_mismatched_indentation;

#if defined(PRISM_HAS_NEON) || defined(PRISM_HAS_SSE2) || defined(PRISM_HAS_SWAR)
    /*
     * Cached lookup tables for pm_strpbrk's SIMD fast path. Avoids rebuilding
     * the nibble-based tables on every call when the charset hasn't changed
     * (which is the common case during string/regex/list lexing).
     */
    struct {
        /* The cached charset (null-terminated, max 11 chars + NUL). */
        uint8_t charset[12];

        /* Nibble-based low lookup table for SIMD matching. */
        uint8_t low_lut[16];

        /* Nibble-based high lookup table for SIMD matching. */
        uint8_t high_lut[16];

        /* Scalar fallback table (4 x 64-bit bitmasks covering all ASCII). */
        uint64_t table[4];
    } strpbrk_cache;
#endif
};

/*
 * Initialize a parser with the given start and end pointers.
 */
void pm_parser_init(pm_arena_t *arena, pm_parser_t *parser, const uint8_t *source, size_t size, const pm_options_t *options);

/*
 * Free the memory held by the given parser.
 *
 * This does not free the `pm_options_t` object that was used to initialize the
 * parser.
 */
void pm_parser_cleanup(pm_parser_t *parser);

/*
 * Returns true if the bytes between start and end spell a keyword, using the
 * same lookup that the lexer does.
 */
bool pm_keyword_p(const uint8_t *start, const uint8_t *end);

#endif