extension.c

#include "prism/extension.h"
 
#ifdef _WIN32
#include <ruby/win32.h>
#endif
 
// NOTE: this file should contain only bindings. All non-trivial logic should be
// in libprism so it can be shared its the various callers.
 
VALUE rb_cPrism;
VALUE rb_cPrismNode;
VALUE rb_cPrismSource;
VALUE rb_cPrismToken;
VALUE rb_cPrismLocation;
 
VALUE rb_cPrismComment;
VALUE rb_cPrismInlineComment;
VALUE rb_cPrismEmbDocComment;
VALUE rb_cPrismMagicComment;
VALUE rb_cPrismParseError;
VALUE rb_cPrismParseWarning;
VALUE rb_cPrismResult;
VALUE rb_cPrismParseResult;
VALUE rb_cPrismLexResult;
VALUE rb_cPrismParseLexResult;
VALUE rb_cPrismStringQuery;
 
VALUE rb_cPrismDebugEncoding;
 
ID rb_id_option_command_line;
ID rb_id_option_encoding;
ID rb_id_option_filepath;
ID rb_id_option_frozen_string_literal;
ID rb_id_option_line;
ID rb_id_option_main_script;
ID rb_id_option_partial_script;
ID rb_id_option_scopes;
ID rb_id_option_version;
ID rb_id_source_for;
 
/******************************************************************************/
/* IO of Ruby code                                                            */
/******************************************************************************/
 
static const char *
check_string(VALUE value) {
    // Check if the value is a string. If it's not, then raise a type error.
    if (!RB_TYPE_P(value, T_STRING)) {
        rb_raise(rb_eTypeError, "wrong argument type %" PRIsVALUE " (expected String)", rb_obj_class(value));
    }
 
    // Otherwise, return the value as a C string.
    return RSTRING_PTR(value);
}
 
static void
input_load_string(pm_string_t *input, VALUE string) {
    // Check if the string is a string. If it's not, then raise a type error.
    if (!RB_TYPE_P(string, T_STRING)) {
        rb_raise(rb_eTypeError, "wrong argument type %" PRIsVALUE " (expected String)", rb_obj_class(string));
    }
 
    pm_string_constant_init(input, RSTRING_PTR(string), RSTRING_LEN(string));
}
 
/******************************************************************************/
/* Building C options from Ruby options                                       */
/******************************************************************************/
 
static void
build_options_scopes(pm_options_t *options, VALUE scopes) {
    // Check if the value is an array. If it's not, then raise a type error.
    if (!RB_TYPE_P(scopes, T_ARRAY)) {
        rb_raise(rb_eTypeError, "wrong argument type %"PRIsVALUE" (expected Array)", rb_obj_class(scopes));
    }
 
    // Initialize the scopes array.
    size_t scopes_count = RARRAY_LEN(scopes);
    if (!pm_options_scopes_init(options, scopes_count)) {
        rb_raise(rb_eNoMemError, "failed to allocate memory");
    }
 
    // Iterate over the scopes and add them to the options.
    for (size_t scope_index = 0; scope_index < scopes_count; scope_index++) {
        VALUE scope = rb_ary_entry(scopes, scope_index);
 
        // Check that the scope is an array. If it's not, then raise a type
        // error.
        if (!RB_TYPE_P(scope, T_ARRAY)) {
            rb_raise(rb_eTypeError, "wrong argument type %"PRIsVALUE" (expected Array)", rb_obj_class(scope));
        }
 
        // Initialize the scope array.
        size_t locals_count = RARRAY_LEN(scope);
        pm_options_scope_t *options_scope = &options->scopes[scope_index];
        if (!pm_options_scope_init(options_scope, locals_count)) {
            rb_raise(rb_eNoMemError, "failed to allocate memory");
        }
 
        // Iterate over the locals and add them to the scope.
        for (size_t local_index = 0; local_index < locals_count; local_index++) {
            VALUE local = rb_ary_entry(scope, local_index);
 
            // Check that the local is a symbol. If it's not, then raise a
            // type error.
            if (!RB_TYPE_P(local, T_SYMBOL)) {
                rb_raise(rb_eTypeError, "wrong argument type %"PRIsVALUE" (expected Symbol)", rb_obj_class(local));
            }
 
            // Add the local to the scope.
            pm_string_t *scope_local = &options_scope->locals[local_index];
            const char *name = rb_id2name(SYM2ID(local));
            pm_string_constant_init(scope_local, name, strlen(name));
        }
    }
}
 
static int
build_options_i(VALUE key, VALUE value, VALUE argument) {
    pm_options_t *options = (pm_options_t *) argument;
    ID key_id = SYM2ID(key);
 
    if (key_id == rb_id_option_filepath) {
        if (!NIL_P(value)) pm_options_filepath_set(options, check_string(value));
    } else if (key_id == rb_id_option_encoding) {
        if (!NIL_P(value)) {
            if (value == Qfalse) {
                pm_options_encoding_locked_set(options, true);
            } else {
                pm_options_encoding_set(options, rb_enc_name(rb_to_encoding(value)));
            }
        }
    } else if (key_id == rb_id_option_line) {
        if (!NIL_P(value)) pm_options_line_set(options, NUM2INT(value));
    } else if (key_id == rb_id_option_frozen_string_literal) {
        if (!NIL_P(value)) pm_options_frozen_string_literal_set(options, RTEST(value));
    } else if (key_id == rb_id_option_version) {
        if (!NIL_P(value)) {
            const char *version = check_string(value);
 
            if (!pm_options_version_set(options, version, RSTRING_LEN(value))) {
                rb_raise(rb_eArgError, "invalid version: %" PRIsVALUE, value);
            }
        }
    } else if (key_id == rb_id_option_scopes) {
        if (!NIL_P(value)) build_options_scopes(options, value);
    } else if (key_id == rb_id_option_command_line) {
        if (!NIL_P(value)) {
            const char *string = check_string(value);
            uint8_t command_line = 0;
 
            for (size_t index = 0; index < strlen(string); index++) {
                switch (string[index]) {
                    case 'a': command_line |= PM_OPTIONS_COMMAND_LINE_A; break;
                    case 'e': command_line |= PM_OPTIONS_COMMAND_LINE_E; break;
                    case 'l': command_line |= PM_OPTIONS_COMMAND_LINE_L; break;
                    case 'n': command_line |= PM_OPTIONS_COMMAND_LINE_N; break;
                    case 'p': command_line |= PM_OPTIONS_COMMAND_LINE_P; break;
                    case 'x': command_line |= PM_OPTIONS_COMMAND_LINE_X; break;
                    default: rb_raise(rb_eArgError, "invalid command line flag: '%c'", string[index]); break;
                }
            }
 
            pm_options_command_line_set(options, command_line);
        }
    } else if (key_id == rb_id_option_main_script) {
        if (!NIL_P(value)) pm_options_main_script_set(options, RTEST(value));
    } else if (key_id == rb_id_option_partial_script) {
        if (!NIL_P(value)) pm_options_partial_script_set(options, RTEST(value));
    } else {
        rb_raise(rb_eArgError, "unknown keyword: %" PRIsVALUE, key);
    }
 
    return ST_CONTINUE;
}
 
struct build_options_data {
    pm_options_t *options;
    VALUE keywords;
};
 
static VALUE
build_options(VALUE argument) {
    struct build_options_data *data = (struct build_options_data *) argument;
    rb_hash_foreach(data->keywords, build_options_i, (VALUE) data->options);
    return Qnil;
}
 
static void
extract_options(pm_options_t *options, VALUE filepath, VALUE keywords) {
    options->line = 1; // default
 
    if (!NIL_P(keywords)) {
        struct build_options_data data = { .options = options, .keywords = keywords };
        struct build_options_data *argument = &data;
 
        int state = 0;
        rb_protect(build_options, (VALUE) argument, &state);
 
        if (state != 0) {
            pm_options_free(options);
            rb_jump_tag(state);
        }
    }
 
    if (!NIL_P(filepath)) {
        if (!RB_TYPE_P(filepath, T_STRING)) {
            pm_options_free(options);
            rb_raise(rb_eTypeError, "wrong argument type %"PRIsVALUE" (expected String)", rb_obj_class(filepath));
        }
 
        pm_options_filepath_set(options, RSTRING_PTR(filepath));
    }
}
 
static void
string_options(int argc, VALUE *argv, pm_string_t *input, pm_options_t *options) {
    VALUE string;
    VALUE keywords;
    rb_scan_args(argc, argv, "1:", &string, &keywords);
 
    extract_options(options, Qnil, keywords);
    input_load_string(input, string);
}
 
static void
file_options(int argc, VALUE *argv, pm_string_t *input, pm_options_t *options, VALUE *encoded_filepath) {
    VALUE filepath;
    VALUE keywords;
    rb_scan_args(argc, argv, "1:", &filepath, &keywords);
 
    Check_Type(filepath, T_STRING);
    *encoded_filepath = rb_str_encode_ospath(filepath);
    extract_options(options, *encoded_filepath, keywords);
 
    const char *source = (const char *) pm_string_source(&options->filepath);
    pm_string_init_result_t result;
 
    switch (result = pm_string_file_init(input, source)) {
        case PM_STRING_INIT_SUCCESS:
            break;
        case PM_STRING_INIT_ERROR_GENERIC: {
            pm_options_free(options);
 
#ifdef _WIN32
            int e = rb_w32_map_errno(GetLastError());
#else
            int e = errno;
#endif
 
            rb_syserr_fail(e, source);
            break;
        }
        case PM_STRING_INIT_ERROR_DIRECTORY:
            pm_options_free(options);
            rb_syserr_fail(EISDIR, source);
            break;
        default:
            pm_options_free(options);
            rb_raise(rb_eRuntimeError, "Unknown error (%d) initializing file: %s", result, source);
            break;
    }
}
 
#ifndef PRISM_EXCLUDE_SERIALIZATION
 
/******************************************************************************/
/* Serializing the AST                                                        */
/******************************************************************************/
 
static VALUE
dump_input(pm_string_t *input, const pm_options_t *options) {
    pm_buffer_t buffer;
    if (!pm_buffer_init(&buffer)) {
        rb_raise(rb_eNoMemError, "failed to allocate memory");
    }
 
    pm_parser_t parser;
    pm_parser_init(&parser, pm_string_source(input), pm_string_length(input), options);
 
    pm_node_t *node = pm_parse(&parser);
    pm_serialize(&parser, node, &buffer);
 
    VALUE result = rb_str_new(pm_buffer_value(&buffer), pm_buffer_length(&buffer));
    pm_node_destroy(&parser, node);
    pm_buffer_free(&buffer);
    pm_parser_free(&parser);
 
    return result;
}
 
static VALUE
dump(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
    string_options(argc, argv, &input, &options);
 
#ifdef PRISM_BUILD_DEBUG
    size_t length = pm_string_length(&input);
    char* dup = xmalloc(length);
    memcpy(dup, pm_string_source(&input), length);
    pm_string_constant_init(&input, dup, length);
#endif
 
    VALUE value = dump_input(&input, &options);
 
#ifdef PRISM_BUILD_DEBUG
    xfree(dup);
#endif
 
    pm_string_free(&input);
    pm_options_free(&options);
 
    return value;
}
 
static VALUE
dump_file(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
 
    VALUE encoded_filepath;
    file_options(argc, argv, &input, &options, &encoded_filepath);
 
    VALUE value = dump_input(&input, &options);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return value;
}
 
#endif
 
/******************************************************************************/
/* Extracting values for the parse result                                     */
/******************************************************************************/
 
static VALUE
parser_comments(pm_parser_t *parser, VALUE source) {
    VALUE comments = rb_ary_new_capa(parser->comment_list.size);
 
    for (pm_comment_t *comment = (pm_comment_t *) parser->comment_list.head; comment != NULL; comment = (pm_comment_t *) comment->node.next) {
        VALUE location_argv[] = {
            source,
            LONG2FIX(comment->location.start - parser->start),
            LONG2FIX(comment->location.end - comment->location.start)
        };
 
        VALUE type = (comment->type == PM_COMMENT_EMBDOC) ? rb_cPrismEmbDocComment : rb_cPrismInlineComment;
        VALUE comment_argv[] = { rb_class_new_instance(3, location_argv, rb_cPrismLocation) };
        rb_ary_push(comments, rb_class_new_instance(1, comment_argv, type));
    }
 
    return comments;
}
 
static VALUE
parser_magic_comments(pm_parser_t *parser, VALUE source) {
    VALUE magic_comments = rb_ary_new_capa(parser->magic_comment_list.size);
 
    for (pm_magic_comment_t *magic_comment = (pm_magic_comment_t *) parser->magic_comment_list.head; magic_comment != NULL; magic_comment = (pm_magic_comment_t *) magic_comment->node.next) {
        VALUE key_loc_argv[] = {
            source,
            LONG2FIX(magic_comment->key_start - parser->start),
            LONG2FIX(magic_comment->key_length)
        };
 
        VALUE value_loc_argv[] = {
            source,
            LONG2FIX(magic_comment->value_start - parser->start),
            LONG2FIX(magic_comment->value_length)
        };
 
        VALUE magic_comment_argv[] = {
            rb_class_new_instance(3, key_loc_argv, rb_cPrismLocation),
            rb_class_new_instance(3, value_loc_argv, rb_cPrismLocation)
        };
 
        rb_ary_push(magic_comments, rb_class_new_instance(2, magic_comment_argv, rb_cPrismMagicComment));
    }
 
    return magic_comments;
}
 
static VALUE
parser_data_loc(const pm_parser_t *parser, VALUE source) {
    if (parser->data_loc.end == NULL) {
        return Qnil;
    } else {
        VALUE argv[] = {
            source,
            LONG2FIX(parser->data_loc.start - parser->start),
            LONG2FIX(parser->data_loc.end - parser->data_loc.start)
        };
 
        return rb_class_new_instance(3, argv, rb_cPrismLocation);
    }
}
 
static VALUE
parser_errors(pm_parser_t *parser, rb_encoding *encoding, VALUE source) {
    VALUE errors = rb_ary_new_capa(parser->error_list.size);
    pm_diagnostic_t *error;
 
    for (error = (pm_diagnostic_t *) parser->error_list.head; error != NULL; error = (pm_diagnostic_t *) error->node.next) {
        VALUE location_argv[] = {
            source,
            LONG2FIX(error->location.start - parser->start),
            LONG2FIX(error->location.end - error->location.start)
        };
 
        VALUE level = Qnil;
        switch (error->level) {
            case PM_ERROR_LEVEL_SYNTAX:
                level = ID2SYM(rb_intern("syntax"));
                break;
            case PM_ERROR_LEVEL_ARGUMENT:
                level = ID2SYM(rb_intern("argument"));
                break;
            case PM_ERROR_LEVEL_LOAD:
                level = ID2SYM(rb_intern("load"));
                break;
            default:
                rb_raise(rb_eRuntimeError, "Unknown level: %" PRIu8, error->level);
        }
 
        VALUE error_argv[] = {
            ID2SYM(rb_intern(pm_diagnostic_id_human(error->diag_id))),
            rb_enc_str_new_cstr(error->message, encoding),
            rb_class_new_instance(3, location_argv, rb_cPrismLocation),
            level
        };
 
        rb_ary_push(errors, rb_class_new_instance(4, error_argv, rb_cPrismParseError));
    }
 
    return errors;
}
 
static VALUE
parser_warnings(pm_parser_t *parser, rb_encoding *encoding, VALUE source) {
    VALUE warnings = rb_ary_new_capa(parser->warning_list.size);
    pm_diagnostic_t *warning;
 
    for (warning = (pm_diagnostic_t *) parser->warning_list.head; warning != NULL; warning = (pm_diagnostic_t *) warning->node.next) {
        VALUE location_argv[] = {
            source,
            LONG2FIX(warning->location.start - parser->start),
            LONG2FIX(warning->location.end - warning->location.start)
        };
 
        VALUE level = Qnil;
        switch (warning->level) {
            case PM_WARNING_LEVEL_DEFAULT:
                level = ID2SYM(rb_intern("default"));
                break;
            case PM_WARNING_LEVEL_VERBOSE:
                level = ID2SYM(rb_intern("verbose"));
                break;
            default:
                rb_raise(rb_eRuntimeError, "Unknown level: %" PRIu8, warning->level);
        }
 
        VALUE warning_argv[] = {
            ID2SYM(rb_intern(pm_diagnostic_id_human(warning->diag_id))),
            rb_enc_str_new_cstr(warning->message, encoding),
            rb_class_new_instance(3, location_argv, rb_cPrismLocation),
            level
        };
 
        rb_ary_push(warnings, rb_class_new_instance(4, warning_argv, rb_cPrismParseWarning));
    }
 
    return warnings;
}
 
static VALUE
parse_result_create(VALUE class, pm_parser_t *parser, VALUE value, rb_encoding *encoding, VALUE source) {
    VALUE result_argv[] = {
        value,
        parser_comments(parser, source),
        parser_magic_comments(parser, source),
        parser_data_loc(parser, source),
        parser_errors(parser, encoding, source),
        parser_warnings(parser, encoding, source),
        source
    };
 
    return rb_class_new_instance(7, result_argv, class);
}
 
/******************************************************************************/
/* Lexing Ruby code                                                           */
/******************************************************************************/
 
typedef struct {
    VALUE source;
    VALUE tokens;
    rb_encoding *encoding;
} parse_lex_data_t;
 
static void
parse_lex_token(void *data, pm_parser_t *parser, pm_token_t *token) {
    parse_lex_data_t *parse_lex_data = (parse_lex_data_t *) parser->lex_callback->data;
 
    VALUE yields = rb_assoc_new(
        pm_token_new(parser, token, parse_lex_data->encoding, parse_lex_data->source),
        INT2FIX(parser->lex_state)
    );
 
    rb_ary_push(parse_lex_data->tokens, yields);
}
 
static void
parse_lex_encoding_changed_callback(pm_parser_t *parser) {
    parse_lex_data_t *parse_lex_data = (parse_lex_data_t *) parser->lex_callback->data;
    parse_lex_data->encoding = rb_enc_find(parser->encoding->name);
 
    // Since the encoding changed, we need to go back and change the encoding of
    // the tokens that were already lexed. This is only going to end up being
    // one or two tokens, since the encoding can only change at the top of the
    // file.
    VALUE tokens = parse_lex_data->tokens;
    for (long index = 0; index < RARRAY_LEN(tokens); index++) {
        VALUE yields = rb_ary_entry(tokens, index);
        VALUE token = rb_ary_entry(yields, 0);
 
        VALUE value = rb_ivar_get(token, rb_intern("@value"));
        rb_enc_associate(value, parse_lex_data->encoding);
        ENC_CODERANGE_CLEAR(value);
    }
}
 
static VALUE
parse_lex_input(pm_string_t *input, const pm_options_t *options, bool return_nodes) {
    pm_parser_t parser;
    pm_parser_init(&parser, pm_string_source(input), pm_string_length(input), options);
    pm_parser_register_encoding_changed_callback(&parser, parse_lex_encoding_changed_callback);
 
    VALUE source_string = rb_str_new((const char *) pm_string_source(input), pm_string_length(input));
    VALUE offsets = rb_ary_new_capa(parser.newline_list.size);
    VALUE source = rb_funcall(rb_cPrismSource, rb_id_source_for, 3, source_string, LONG2NUM(parser.start_line), offsets);
 
    parse_lex_data_t parse_lex_data = {
        .source = source,
        .tokens = rb_ary_new(),
        .encoding = rb_utf8_encoding()
    };
 
    parse_lex_data_t *data = &parse_lex_data;
    pm_lex_callback_t lex_callback = (pm_lex_callback_t) {
        .data = (void *) data,
        .callback = parse_lex_token,
    };
 
    parser.lex_callback = &lex_callback;
    pm_node_t *node = pm_parse(&parser);
 
    // Here we need to update the Source object to have the correct
    // encoding for the source string and the correct newline offsets.
    // We do it here because we've already created the Source object and given
    // it over to all of the tokens, and both of these are only set after pm_parse().
    rb_encoding *encoding = rb_enc_find(parser.encoding->name);
    rb_enc_associate(source_string, encoding);
 
    for (size_t index = 0; index < parser.newline_list.size; index++) {
        rb_ary_push(offsets, ULONG2NUM(parser.newline_list.offsets[index]));
    }
 
    VALUE result;
    if (return_nodes) {
        VALUE value = rb_ary_new_capa(2);
        rb_ary_push(value, pm_ast_new(&parser, node, parse_lex_data.encoding, source));
        rb_ary_push(value, parse_lex_data.tokens);
        result = parse_result_create(rb_cPrismParseLexResult, &parser, value, parse_lex_data.encoding, source);
    } else {
        result = parse_result_create(rb_cPrismLexResult, &parser, parse_lex_data.tokens, parse_lex_data.encoding, source);
    }
 
    pm_node_destroy(&parser, node);
    pm_parser_free(&parser);
 
    return result;
}
 
static VALUE
lex(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
    string_options(argc, argv, &input, &options);
 
    VALUE result = parse_lex_input(&input, &options, false);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return result;
}
 
static VALUE
lex_file(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
 
    VALUE encoded_filepath;
    file_options(argc, argv, &input, &options, &encoded_filepath);
 
    VALUE value = parse_lex_input(&input, &options, false);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return value;
}
 
/******************************************************************************/
/* Parsing Ruby code                                                          */
/******************************************************************************/
 
static VALUE
parse_input(pm_string_t *input, const pm_options_t *options) {
    pm_parser_t parser;
    pm_parser_init(&parser, pm_string_source(input), pm_string_length(input), options);
 
    pm_node_t *node = pm_parse(&parser);
    rb_encoding *encoding = rb_enc_find(parser.encoding->name);
 
    VALUE source = pm_source_new(&parser, encoding);
    VALUE value = pm_ast_new(&parser, node, encoding, source);
    VALUE result = parse_result_create(rb_cPrismParseResult, &parser, value, encoding, source) ;
 
    pm_node_destroy(&parser, node);
    pm_parser_free(&parser);
 
    return result;
}
 
static VALUE
parse(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
    string_options(argc, argv, &input, &options);
 
#ifdef PRISM_BUILD_DEBUG
    size_t length = pm_string_length(&input);
    char* dup = xmalloc(length);
    memcpy(dup, pm_string_source(&input), length);
    pm_string_constant_init(&input, dup, length);
#endif
 
    VALUE value = parse_input(&input, &options);
 
#ifdef PRISM_BUILD_DEBUG
    xfree(dup);
#endif
 
    pm_string_free(&input);
    pm_options_free(&options);
    return value;
}
 
static VALUE
parse_file(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
 
    VALUE encoded_filepath;
    file_options(argc, argv, &input, &options, &encoded_filepath);
 
    VALUE value = parse_input(&input, &options);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return value;
}
 
static void
profile_input(pm_string_t *input, const pm_options_t *options) {
    pm_parser_t parser;
    pm_parser_init(&parser, pm_string_source(input), pm_string_length(input), options);
 
    pm_node_t *node = pm_parse(&parser);
    pm_node_destroy(&parser, node);
    pm_parser_free(&parser);
}
 
static VALUE
profile(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
 
    string_options(argc, argv, &input, &options);
    profile_input(&input, &options);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return Qnil;
}
 
static VALUE
profile_file(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
 
    VALUE encoded_filepath;
    file_options(argc, argv, &input, &options, &encoded_filepath);
 
    profile_input(&input, &options);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return Qnil;
}
 
static char *
parse_stream_fgets(char *string, int size, void *stream) {
    RUBY_ASSERT(size > 0);
 
    VALUE line = rb_funcall((VALUE) stream, rb_intern("gets"), 1, INT2FIX(size - 1));
    if (NIL_P(line)) {
        return NULL;
    }
 
    const char *cstr = RSTRING_PTR(line);
    long length = RSTRING_LEN(line);
 
    memcpy(string, cstr, length);
    string[length] = '\0';
 
    return string;
}
 
static VALUE
parse_stream(int argc, VALUE *argv, VALUE self) {
    VALUE stream;
    VALUE keywords;
    rb_scan_args(argc, argv, "1:", &stream, &keywords);
 
    pm_options_t options = { 0 };
    extract_options(&options, Qnil, keywords);
 
    pm_parser_t parser;
    pm_buffer_t buffer;
 
    pm_node_t *node = pm_parse_stream(&parser, &buffer, (void *) stream, parse_stream_fgets, &options);
    rb_encoding *encoding = rb_enc_find(parser.encoding->name);
 
    VALUE source = pm_source_new(&parser, encoding);
    VALUE value = pm_ast_new(&parser, node, encoding, source);
    VALUE result = parse_result_create(rb_cPrismParseResult, &parser, value, encoding, source);
 
    pm_node_destroy(&parser, node);
    pm_buffer_free(&buffer);
    pm_parser_free(&parser);
 
    return result;
}
 
static VALUE
parse_input_comments(pm_string_t *input, const pm_options_t *options) {
    pm_parser_t parser;
    pm_parser_init(&parser, pm_string_source(input), pm_string_length(input), options);
 
    pm_node_t *node = pm_parse(&parser);
    rb_encoding *encoding = rb_enc_find(parser.encoding->name);
 
    VALUE source = pm_source_new(&parser, encoding);
    VALUE comments = parser_comments(&parser, source);
 
    pm_node_destroy(&parser, node);
    pm_parser_free(&parser);
 
    return comments;
}
 
static VALUE
parse_comments(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
    string_options(argc, argv, &input, &options);
 
    VALUE result = parse_input_comments(&input, &options);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return result;
}
 
static VALUE
parse_file_comments(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
 
    VALUE encoded_filepath;
    file_options(argc, argv, &input, &options, &encoded_filepath);
 
    VALUE value = parse_input_comments(&input, &options);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return value;
}
 
static VALUE
parse_lex(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
    string_options(argc, argv, &input, &options);
 
    VALUE value = parse_lex_input(&input, &options, true);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return value;
}
 
static VALUE
parse_lex_file(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
 
    VALUE encoded_filepath;
    file_options(argc, argv, &input, &options, &encoded_filepath);
 
    VALUE value = parse_lex_input(&input, &options, true);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return value;
}
 
static VALUE
parse_input_success_p(pm_string_t *input, const pm_options_t *options) {
    pm_parser_t parser;
    pm_parser_init(&parser, pm_string_source(input), pm_string_length(input), options);
 
    pm_node_t *node = pm_parse(&parser);
    pm_node_destroy(&parser, node);
 
    VALUE result = parser.error_list.size == 0 ? Qtrue : Qfalse;
    pm_parser_free(&parser);
 
    return result;
}
 
static VALUE
parse_success_p(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
    string_options(argc, argv, &input, &options);
 
    VALUE result = parse_input_success_p(&input, &options);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return result;
}
 
static VALUE
parse_failure_p(int argc, VALUE *argv, VALUE self) {
    return RTEST(parse_success_p(argc, argv, self)) ? Qfalse : Qtrue;
}
 
static VALUE
parse_file_success_p(int argc, VALUE *argv, VALUE self) {
    pm_string_t input;
    pm_options_t options = { 0 };
 
    VALUE encoded_filepath;
    file_options(argc, argv, &input, &options, &encoded_filepath);
 
    VALUE result = parse_input_success_p(&input, &options);
    pm_string_free(&input);
    pm_options_free(&options);
 
    return result;
}
 
static VALUE
parse_file_failure_p(int argc, VALUE *argv, VALUE self) {
    return RTEST(parse_file_success_p(argc, argv, self)) ? Qfalse : Qtrue;
}
 
/******************************************************************************/
/* String query methods                                                       */
/******************************************************************************/
 
static VALUE
string_query(pm_string_query_t result) {
    switch (result) {
        case PM_STRING_QUERY_ERROR:
            rb_raise(rb_eArgError, "Invalid or non ascii-compatible encoding");
            return Qfalse;
        case PM_STRING_QUERY_FALSE:
            return Qfalse;
        case PM_STRING_QUERY_TRUE:
            return Qtrue;
    }
    return Qfalse;
}
 
static VALUE
string_query_local_p(VALUE self, VALUE string) {
    const uint8_t *source = (const uint8_t *) check_string(string);
    return string_query(pm_string_query_local(source, RSTRING_LEN(string), rb_enc_get(string)->name));
}
 
static VALUE
string_query_constant_p(VALUE self, VALUE string) {
    const uint8_t *source = (const uint8_t *) check_string(string);
    return string_query(pm_string_query_constant(source, RSTRING_LEN(string), rb_enc_get(string)->name));
}
 
static VALUE
string_query_method_name_p(VALUE self, VALUE string) {
    const uint8_t *source = (const uint8_t *) check_string(string);
    return string_query(pm_string_query_method_name(source, RSTRING_LEN(string), rb_enc_get(string)->name));
}
 
/******************************************************************************/
/* Initialization of the extension                                            */
/******************************************************************************/
 
RUBY_FUNC_EXPORTED void
Init_prism(void) {
    // Make sure that the prism library version matches the expected version.
    // Otherwise something was compiled incorrectly.
    if (strcmp(pm_version(), EXPECTED_PRISM_VERSION) != 0) {
        rb_raise(
            rb_eRuntimeError,
            "The prism library version (%s) does not match the expected version (%s)",
            pm_version(),
            EXPECTED_PRISM_VERSION
        );
    }
 
    // Grab up references to all of the constants that we're going to need to
    // reference throughout this extension.
    rb_cPrism = rb_define_module("Prism");
    rb_cPrismNode = rb_define_class_under(rb_cPrism, "Node", rb_cObject);
    rb_cPrismSource = rb_define_class_under(rb_cPrism, "Source", rb_cObject);
    rb_cPrismToken = rb_define_class_under(rb_cPrism, "Token", rb_cObject);
    rb_cPrismLocation = rb_define_class_under(rb_cPrism, "Location", rb_cObject);
    rb_cPrismComment = rb_define_class_under(rb_cPrism, "Comment", rb_cObject);
    rb_cPrismInlineComment = rb_define_class_under(rb_cPrism, "InlineComment", rb_cPrismComment);
    rb_cPrismEmbDocComment = rb_define_class_under(rb_cPrism, "EmbDocComment", rb_cPrismComment);
    rb_cPrismMagicComment = rb_define_class_under(rb_cPrism, "MagicComment", rb_cObject);
    rb_cPrismParseError = rb_define_class_under(rb_cPrism, "ParseError", rb_cObject);
    rb_cPrismParseWarning = rb_define_class_under(rb_cPrism, "ParseWarning", rb_cObject);
    rb_cPrismResult = rb_define_class_under(rb_cPrism, "Result", rb_cObject);
    rb_cPrismParseResult = rb_define_class_under(rb_cPrism, "ParseResult", rb_cPrismResult);
    rb_cPrismLexResult = rb_define_class_under(rb_cPrism, "LexResult", rb_cPrismResult);
    rb_cPrismParseLexResult = rb_define_class_under(rb_cPrism, "ParseLexResult", rb_cPrismResult);
    rb_cPrismStringQuery = rb_define_class_under(rb_cPrism, "StringQuery", rb_cObject);
 
    // Intern all of the IDs eagerly that we support so that we don't have to do
    // it every time we parse.
    rb_id_option_command_line = rb_intern_const("command_line");
    rb_id_option_encoding = rb_intern_const("encoding");
    rb_id_option_filepath = rb_intern_const("filepath");
    rb_id_option_frozen_string_literal = rb_intern_const("frozen_string_literal");
    rb_id_option_line = rb_intern_const("line");
    rb_id_option_main_script = rb_intern_const("main_script");
    rb_id_option_partial_script = rb_intern_const("partial_script");
    rb_id_option_scopes = rb_intern_const("scopes");
    rb_id_option_version = rb_intern_const("version");
    rb_id_source_for = rb_intern("for");
 
    rb_define_const(rb_cPrism, "VERSION", rb_str_new2(EXPECTED_PRISM_VERSION));
 
    // First, the functions that have to do with lexing and parsing.
    rb_define_singleton_method(rb_cPrism, "lex", lex, -1);
    rb_define_singleton_method(rb_cPrism, "lex_file", lex_file, -1);
    rb_define_singleton_method(rb_cPrism, "parse", parse, -1);
    rb_define_singleton_method(rb_cPrism, "parse_file", parse_file, -1);
    rb_define_singleton_method(rb_cPrism, "profile", profile, -1);
    rb_define_singleton_method(rb_cPrism, "profile_file", profile_file, -1);
    rb_define_singleton_method(rb_cPrism, "parse_stream", parse_stream, -1);
    rb_define_singleton_method(rb_cPrism, "parse_comments", parse_comments, -1);
    rb_define_singleton_method(rb_cPrism, "parse_file_comments", parse_file_comments, -1);
    rb_define_singleton_method(rb_cPrism, "parse_lex", parse_lex, -1);
    rb_define_singleton_method(rb_cPrism, "parse_lex_file", parse_lex_file, -1);
    rb_define_singleton_method(rb_cPrism, "parse_success?", parse_success_p, -1);
    rb_define_singleton_method(rb_cPrism, "parse_failure?", parse_failure_p, -1);
    rb_define_singleton_method(rb_cPrism, "parse_file_success?", parse_file_success_p, -1);
    rb_define_singleton_method(rb_cPrism, "parse_file_failure?", parse_file_failure_p, -1);
 
#ifndef PRISM_EXCLUDE_SERIALIZATION
    rb_define_singleton_method(rb_cPrism, "dump", dump, -1);
    rb_define_singleton_method(rb_cPrism, "dump_file", dump_file, -1);
#endif
 
    rb_define_singleton_method(rb_cPrismStringQuery, "local?", string_query_local_p, 1);
    rb_define_singleton_method(rb_cPrismStringQuery, "constant?", string_query_constant_p, 1);
    rb_define_singleton_method(rb_cPrismStringQuery, "method_name?", string_query_method_name_p, 1);
 
    // Next, initialize the other APIs.
    Init_prism_api_node();
    Init_prism_pack();
}