prism 0.29.0 → 1.3.0

data/ext/prism/extension.c

@@ -21,38 +21,36 @@ 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_option_id_command_line;
-ID rb_option_id_encoding;
-ID rb_option_id_filepath;
-ID rb_option_id_frozen_string_literal;
-ID rb_option_id_line;
-ID rb_option_id_scopes;
-ID rb_option_id_version;
-ID rb_prism_source_id_for;
+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                                                            */
 /******************************************************************************/
 
 /**
- * Check if the given VALUE is a string. If it's nil, then return NULL. If it's
- * not a string, then raise a type error. Otherwise return the VALUE as a C
- * string.
+ * Check if the given VALUE is a string. If it's not a string, then raise a
+ * TypeError. Otherwise return the VALUE as a C string.
  */
 static const char *
 check_string(VALUE value) {
-    // If the value is nil, then we don't need to do anything.
-    if (NIL_P(value)) {
-        return NULL;
-    }
-
     // 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));
+        rb_raise(rb_eTypeError, "wrong argument type %" PRIsVALUE " (expected String)", rb_obj_class(value));
     }
 
     // Otherwise, return the value as a C string.
@@ -66,7 +64,7 @@ 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));
+        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));
@@ -135,15 +133,21 @@ 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_option_id_filepath) {
+    if (key_id == rb_id_option_filepath) {
         if (!NIL_P(value)) pm_options_filepath_set(options, check_string(value));
-    } else if (key_id == rb_option_id_encoding) {
-        if (!NIL_P(value)) pm_options_encoding_set(options, rb_enc_name(rb_to_encoding(value)));
-    } else if (key_id == rb_option_id_line) {
+    } 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_option_id_frozen_string_literal) {
+    } 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_option_id_version) {
+    } else if (key_id == rb_id_option_version) {
         if (!NIL_P(value)) {
             const char *version = check_string(value);
 
@@ -151,9 +155,9 @@ build_options_i(VALUE key, VALUE value, VALUE argument) {
                 rb_raise(rb_eArgError, "invalid version: %" PRIsVALUE, value);
             }
         }
-    } else if (key_id == rb_option_id_scopes) {
+    } else if (key_id == rb_id_option_scopes) {
         if (!NIL_P(value)) build_options_scopes(options, value);
-    } else if (key_id == rb_option_id_command_line) {
+    } else if (key_id == rb_id_option_command_line) {
         if (!NIL_P(value)) {
             const char *string = check_string(value);
             uint8_t command_line = 0;
@@ -172,6 +176,10 @@ build_options_i(VALUE key, VALUE value, VALUE argument) {
 
             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);
     }
@@ -206,6 +214,7 @@ build_options(VALUE argument) {
 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;
@@ -246,27 +255,41 @@ string_options(int argc, VALUE *argv, pm_string_t *input, pm_options_t *options)
  * Read options for methods that look like (filepath, **options).
  */
 static void
-file_options(int argc, VALUE *argv, pm_string_t *input, pm_options_t *options) {
+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);
 
-    extract_options(options, filepath, keywords);
-
-    const char * string_source = (const char *) pm_string_source(&options->filepath);
+    const char *source = (const char *) pm_string_source(&options->filepath);
+    pm_string_init_result_t result;
 
-    if (!pm_string_file_init(input, string_source)) {
-        pm_options_free(options);
+    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());
+            int e = rb_w32_map_errno(GetLastError());
 #else
-        int e = errno;
+            int e = errno;
 #endif
 
-        rb_syserr_fail(e, string_source);
+            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;
     }
 }
 
@@ -344,7 +367,8 @@ dump_file(int argc, VALUE *argv, VALUE self) {
     pm_string_t input;
     pm_options_t options = { 0 };
 
-    file_options(argc, argv, &input, &options);
+    VALUE encoded_filepath;
+    file_options(argc, argv, &input, &options, &encoded_filepath);
 
     VALUE value = dump_input(&input, &options);
     pm_string_free(&input);
@@ -364,7 +388,7 @@ dump_file(int argc, VALUE *argv, VALUE self) {
  */
 static VALUE
 parser_comments(pm_parser_t *parser, VALUE source) {
-    VALUE comments = rb_ary_new();
+    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[] = {
@@ -386,7 +410,7 @@ parser_comments(pm_parser_t *parser, VALUE source) {
  */
 static VALUE
 parser_magic_comments(pm_parser_t *parser, VALUE source) {
-    VALUE magic_comments = rb_ary_new();
+    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[] = {
@@ -436,7 +460,7 @@ parser_data_loc(const pm_parser_t *parser, VALUE source) {
  */
 static VALUE
 parser_errors(pm_parser_t *parser, rb_encoding *encoding, VALUE source) {
-    VALUE errors = rb_ary_new();
+    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) {
@@ -479,7 +503,7 @@ parser_errors(pm_parser_t *parser, rb_encoding *encoding, VALUE source) {
  */
 static VALUE
 parser_warnings(pm_parser_t *parser, rb_encoding *encoding, VALUE source) {
-    VALUE warnings = rb_ary_new();
+    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) {
@@ -556,9 +580,10 @@ 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_ary_new_capa(2);
-    rb_ary_push(yields, pm_token_new(parser, token, parse_lex_data->encoding, parse_lex_data->source));
-    rb_ary_push(yields, INT2FIX(parser->lex_state));
+    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);
 }
@@ -599,8 +624,8 @@ parse_lex_input(pm_string_t *input, const pm_options_t *options, bool return_nod
     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();
-    VALUE source = rb_funcall(rb_cPrismSource, rb_prism_source_id_for, 3, source_string, LONG2NUM(parser.start_line), offsets);
+    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,
@@ -628,16 +653,16 @@ parse_lex_input(pm_string_t *input, const pm_options_t *options, bool return_nod
         rb_ary_push(offsets, ULONG2NUM(parser.newline_list.offsets[index]));
     }
 
-    VALUE value;
+    VALUE result;
     if (return_nodes) {
-        value = rb_ary_new_capa(2);
+        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 {
-        value = parse_lex_data.tokens;
+        result = parse_result_create(rb_cPrismLexResult, &parser, parse_lex_data.tokens, parse_lex_data.encoding, source);
     }
 
-    VALUE result = parse_result_create(rb_cPrismParseLexResult, &parser, value, parse_lex_data.encoding, source);
     pm_node_destroy(&parser, node);
     pm_parser_free(&parser);
 
@@ -646,10 +671,10 @@ parse_lex_input(pm_string_t *input, const pm_options_t *options, bool return_nod
 
 /**
  * call-seq:
- *   Prism::lex(source, **options) -> Array
+ *   Prism::lex(source, **options) -> LexResult
  *
- * Return an array of Token instances corresponding to the given string. For
- * supported options, see Prism::parse.
+ * Return a LexResult instance that contains an array of Token instances
+ * corresponding to the given string. For supported options, see Prism::parse.
  */
 static VALUE
 lex(int argc, VALUE *argv, VALUE self) {
@@ -666,17 +691,18 @@ lex(int argc, VALUE *argv, VALUE self) {
 
 /**
  * call-seq:
- *   Prism::lex_file(filepath, **options) -> Array
+ *   Prism::lex_file(filepath, **options) -> LexResult
  *
- * Return an array of Token instances corresponding to the given file. For
- * supported options, see Prism::parse.
+ * Return a LexResult instance that contains an array of Token instances
+ * corresponding to the given file. For supported options, see Prism::parse.
  */
 static VALUE
 lex_file(int argc, VALUE *argv, VALUE self) {
     pm_string_t input;
     pm_options_t options = { 0 };
 
-    file_options(argc, argv, &input, &options);
+    VALUE encoded_filepath;
+    file_options(argc, argv, &input, &options, &encoded_filepath);
 
     VALUE value = parse_lex_input(&input, &options, false);
     pm_string_free(&input);
@@ -728,14 +754,27 @@ parse_input(pm_string_t *input, const pm_options_t *options) {
  *       has been set. This should be a boolean or nil.
  * * `line` - the line number that the parse starts on. This should be an
  *       integer or nil. Note that this is 1-indexed.
+ * * `main_script` - a boolean indicating whether or not the source being parsed
+ *       is the main script being run by the interpreter. This controls whether
+ *       or not shebangs are parsed for additional flags and whether or not the
+ *       parser will attempt to find a matching shebang if the first one does
+ *       not contain the word "ruby".
+ * * `partial_script` - when the file being parsed is considered a "partial"
+ *       script, jumps will not be marked as errors if they are not contained
+ *       within loops/blocks. This is used in the case that you're parsing a
+ *       script that you know will be embedded inside another script later, but
+ *       you do not have that context yet. For example, when parsing an ERB
+ *       template that will be evaluated inside another script.
  * * `scopes` - the locals that are in scope surrounding the code that is being
  *       parsed. This should be an array of arrays of symbols or nil. Scopes are
  *       ordered from the outermost scope to the innermost one.
  * * `version` - the version of Ruby syntax that prism should used to parse Ruby
  *       code. By default prism assumes you want to parse with the latest version
  *       of Ruby syntax (which you can trigger with `nil` or `"latest"`). You
- *       may also restrict the syntax to a specific version of Ruby. The
- *       supported values are `"3.3.0"` and `"3.4.0"`.
+ *       may also restrict the syntax to a specific version of Ruby, e.g., with `"3.3.0"`.
+ *       To parse with the same syntax version that the current Ruby is running
+ *       use `version: RUBY_VERSION`. Raises ArgumentError if the version is not
+ *       currently supported by Prism.
  */
 static VALUE
 parse(int argc, VALUE *argv, VALUE self) {
@@ -761,6 +800,85 @@ parse(int argc, VALUE *argv, VALUE self) {
     return value;
 }
 
+/**
+ * call-seq:
+ *   Prism::parse_file(filepath, **options) -> ParseResult
+ *
+ * Parse the given file and return a ParseResult instance. For supported
+ * options, see Prism::parse.
+ */
+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;
+}
+
+/**
+ * Parse the given input and return nothing.
+ */
+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);
+}
+
+/**
+ * call-seq:
+ *   Prism::profile(source, **options) -> nil
+ *
+ * Parse the given string and return nothing. This method is meant to allow
+ * profilers to avoid the overhead of reifying the AST to Ruby. For supported
+ * options, see Prism::parse.
+ */
+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;
+}
+
+/**
+ * call-seq:
+ *   Prism::profile_file(filepath, **options) -> nil
+ *
+ * Parse the given file and return nothing. This method is meant to allow
+ * profilers to avoid the overhead of reifying the AST to Ruby. For supported
+ * options, see Prism::parse.
+ */
+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;
+}
+
 /**
  * An implementation of fgets that is suitable for use with Ruby IO objects.
  */
@@ -773,8 +891,8 @@ parse_stream_fgets(char *string, int size, void *stream) {
         return NULL;
     }
 
-    const char *cstr = StringValueCStr(line);
-    size_t length = strlen(cstr);
+    const char *cstr = RSTRING_PTR(line);
+    long length = RSTRING_LEN(line);
 
     memcpy(string, cstr, length);
     string[length] = '\0';
@@ -815,27 +933,6 @@ parse_stream(int argc, VALUE *argv, VALUE self) {
     return result;
 }
 
-/**
- * call-seq:
- *   Prism::parse_file(filepath, **options) -> ParseResult
- *
- * Parse the given file and return a ParseResult instance. For supported
- * options, see Prism::parse.
- */
-static VALUE
-parse_file(int argc, VALUE *argv, VALUE self) {
-    pm_string_t input;
-    pm_options_t options = { 0 };
-
-    file_options(argc, argv, &input, &options);
-
-    VALUE value = parse_input(&input, &options);
-    pm_string_free(&input);
-    pm_options_free(&options);
-
-    return value;
-}
-
 /**
  * Parse the given input and return an array of Comment objects.
  */
@@ -888,7 +985,8 @@ parse_file_comments(int argc, VALUE *argv, VALUE self) {
     pm_string_t input;
     pm_options_t options = { 0 };
 
-    file_options(argc, argv, &input, &options);
+    VALUE encoded_filepath;
+    file_options(argc, argv, &input, &options, &encoded_filepath);
 
     VALUE value = parse_input_comments(&input, &options);
     pm_string_free(&input);
@@ -899,9 +997,9 @@ parse_file_comments(int argc, VALUE *argv, VALUE self) {
 
 /**
  * call-seq:
- *   Prism::parse_lex(source, **options) -> ParseResult
+ *   Prism::parse_lex(source, **options) -> ParseLexResult
  *
- * Parse the given string and return a ParseResult instance that contains a
+ * Parse the given string and return a ParseLexResult instance that contains a
  * 2-element array, where the first element is the AST and the second element is
  * an array of Token instances.
  *
@@ -926,9 +1024,9 @@ parse_lex(int argc, VALUE *argv, VALUE self) {
 
 /**
  * call-seq:
- *   Prism::parse_lex_file(filepath, **options) -> ParseResult
+ *   Prism::parse_lex_file(filepath, **options) -> ParseLexResult
  *
- * Parse the given file and return a ParseResult instance that contains a
+ * Parse the given file and return a ParseLexResult instance that contains a
  * 2-element array, where the first element is the AST and the second element is
  * an array of Token instances.
  *
@@ -943,7 +1041,8 @@ parse_lex_file(int argc, VALUE *argv, VALUE self) {
     pm_string_t input;
     pm_options_t options = { 0 };
 
-    file_options(argc, argv, &input, &options);
+    VALUE encoded_filepath;
+    file_options(argc, argv, &input, &options, &encoded_filepath);
 
     VALUE value = parse_lex_input(&input, &options, true);
     pm_string_free(&input);
@@ -1013,7 +1112,8 @@ parse_file_success_p(int argc, VALUE *argv, VALUE self) {
     pm_string_t input;
     pm_options_t options = { 0 };
 
-    file_options(argc, argv, &input, &options);
+    VALUE encoded_filepath;
+    file_options(argc, argv, &input, &options, &encoded_filepath);
 
     VALUE result = parse_input_success_p(&input, &options);
     pm_string_free(&input);
@@ -1035,300 +1135,65 @@ parse_file_failure_p(int argc, VALUE *argv, VALUE self) {
 }
 
 /******************************************************************************/
-/* Utility functions exposed to make testing easier                           */
+/* String query methods                                                       */
 /******************************************************************************/
 
 /**
- * call-seq:
- *   Debug::named_captures(source) -> Array
- *
- * Returns an array of strings corresponding to the named capture groups in the
- * given source string. If prism was unable to parse the regular expression,
- * this function returns nil.
- */
-static VALUE
-named_captures(VALUE self, VALUE source) {
-    pm_string_list_t string_list = { 0 };
-
-    if (!pm_regexp_named_capture_group_names((const uint8_t *) RSTRING_PTR(source), RSTRING_LEN(source), &string_list, false, PM_ENCODING_UTF_8_ENTRY)) {
-        pm_string_list_free(&string_list);
-        return Qnil;
-    }
-
-    VALUE names = rb_ary_new();
-    for (size_t index = 0; index < string_list.length; index++) {
-        const pm_string_t *string = &string_list.strings[index];
-        rb_ary_push(names, rb_str_new((const char *) pm_string_source(string), pm_string_length(string)));
-    }
-
-    pm_string_list_free(&string_list);
-    return names;
-}
-
-/**
- * call-seq:
- *   Debug::integer_parse(source) -> [Integer, String]
- *
- * Parses the given source string and returns the integer it represents, as well
- * as a decimal string representation.
- */
-static VALUE
-integer_parse(VALUE self, VALUE source) {
-    const uint8_t *start = (const uint8_t *) RSTRING_PTR(source);
-    size_t length = RSTRING_LEN(source);
-
-    pm_integer_t integer = { 0 };
-    pm_integer_parse(&integer, PM_INTEGER_BASE_UNKNOWN, start, start + length);
-
-    pm_buffer_t buffer = { 0 };
-    pm_integer_string(&buffer, &integer);
-
-    VALUE string = rb_str_new(pm_buffer_value(&buffer), pm_buffer_length(&buffer));
-    pm_buffer_free(&buffer);
-
-    VALUE result = rb_ary_new_capa(2);
-    rb_ary_push(result, pm_integer_new(&integer));
-    rb_ary_push(result, string);
-    pm_integer_free(&integer);
-
-    return result;
-}
-
-/**
- * call-seq:
- *   Debug::memsize(source) -> { length: xx, memsize: xx, node_count: xx }
- *
- * Return a hash of information about the given source string's memory usage.
- */
-static VALUE
-memsize(VALUE self, VALUE string) {
-    pm_parser_t parser;
-    size_t length = RSTRING_LEN(string);
-    pm_parser_init(&parser, (const uint8_t *) RSTRING_PTR(string), length, NULL);
-
-    pm_node_t *node = pm_parse(&parser);
-    pm_memsize_t memsize;
-    pm_node_memsize(node, &memsize);
-
-    pm_node_destroy(&parser, node);
-    pm_parser_free(&parser);
-
-    VALUE result = rb_hash_new();
-    rb_hash_aset(result, ID2SYM(rb_intern("length")), INT2FIX(length));
-    rb_hash_aset(result, ID2SYM(rb_intern("memsize")), INT2FIX(memsize.memsize));
-    rb_hash_aset(result, ID2SYM(rb_intern("node_count")), INT2FIX(memsize.node_count));
-    return result;
-}
-
-/**
- * call-seq:
- *   Debug::profile_file(filepath) -> nil
- *
- * Parse the file, but do nothing with the result. This is used to profile the
- * parser for memory and speed.
+ * Process the result of a call to a string query method and return an
+ * appropriate value.
  */
 static VALUE
-profile_file(VALUE self, VALUE filepath) {
-    pm_string_t input;
-
-    const char *checked = check_string(filepath);
-    Check_Type(filepath, T_STRING);
-
-    if (!pm_string_mapped_init(&input, checked)) {
-#ifdef _WIN32
-        int e = rb_w32_map_errno(GetLastError());
-#else
-        int e = errno;
-#endif
-
-        rb_syserr_fail(e, checked);
+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;
     }
-
-    pm_options_t options = { 0 };
-    pm_options_filepath_set(&options, checked);
-
-    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);
-    pm_options_free(&options);
-    pm_string_free(&input);
-
-    return Qnil;
+    return Qfalse;
 }
 
-#ifndef PRISM_EXCLUDE_PRETTYPRINT
-
 /**
  * call-seq:
- *   Debug::inspect_node(source) -> inspected
+ *   Prism::StringQuery::local?(string) -> bool
  *
- * Inspect the AST that represents the given source using the prism pretty print
- * as opposed to the Ruby implementation.
+ * Returns true if the string constitutes a valid local variable name. Note that
+ * this means the names that can be set through Binding#local_variable_set, not
+ * necessarily the ones that can be set through a local variable assignment.
  */
 static VALUE
-inspect_node(VALUE self, VALUE source) {
-    pm_string_t input;
-    input_load_string(&input, source);
-
-    pm_parser_t parser;
-    pm_parser_init(&parser, pm_string_source(&input), pm_string_length(&input), NULL);
-
-    pm_node_t *node = pm_parse(&parser);
-    pm_buffer_t buffer = { 0 };
-
-    pm_prettyprint(&buffer, &parser, node);
-
-    rb_encoding *encoding = rb_enc_find(parser.encoding->name);
-    VALUE string = rb_enc_str_new(pm_buffer_value(&buffer), pm_buffer_length(&buffer), encoding);
-
-    pm_buffer_free(&buffer);
-    pm_node_destroy(&parser, node);
-    pm_parser_free(&parser);
-
-    return string;
+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));
 }
 
-#endif
-
 /**
  * call-seq:
- *   Debug::format_errors(source, colorize) -> String
+ *   Prism::StringQuery::constant?(string) -> bool
  *
- * Format the errors that are found when parsing the given source string.
+ * Returns true if the string constitutes a valid constant name. Note that this
+ * means the names that can be set through Module#const_set, not necessarily the
+ * ones that can be set through a constant assignment.
  */
 static VALUE
-format_errors(VALUE self, VALUE source, VALUE colorize) {
-    pm_string_t input;
-    input_load_string(&input, source);
-
-    pm_parser_t parser;
-    pm_parser_init(&parser, pm_string_source(&input), pm_string_length(&input), NULL);
-
-    pm_node_t *node = pm_parse(&parser);
-    pm_buffer_t buffer = { 0 };
-
-    pm_parser_errors_format(&parser, &parser.error_list, &buffer, RTEST(colorize), true);
-
-    rb_encoding *encoding = rb_enc_find(parser.encoding->name);
-    VALUE result = rb_enc_str_new(pm_buffer_value(&buffer), pm_buffer_length(&buffer), encoding);
-
-    pm_buffer_free(&buffer);
-    pm_node_destroy(&parser, node);
-    pm_parser_free(&parser);
-    pm_string_free(&input);
-
-    return result;
+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));
 }
 
 /**
  * call-seq:
- *   Debug::static_inspect(source) -> String
- *
- * Inspect the node as it would be inspected by the warnings used in static
- * literal sets.
- */
-static VALUE
-static_inspect(int argc, VALUE *argv, VALUE self) {
-    pm_string_t input;
-    pm_options_t options = { 0 };
-    string_options(argc, argv, &input, &options);
-
-    pm_parser_t parser;
-    pm_parser_init(&parser, pm_string_source(&input), pm_string_length(&input), &options);
-
-    pm_node_t *program = pm_parse(&parser);
-    pm_node_t *node = ((pm_program_node_t *) program)->statements->body.nodes[0];
-
-    pm_buffer_t buffer = { 0 };
-    pm_static_literal_inspect(&buffer, &parser.newline_list, parser.start_line, parser.encoding->name, node);
-
-    rb_encoding *encoding = rb_enc_find(parser.encoding->name);
-    VALUE result = rb_enc_str_new(pm_buffer_value(&buffer), pm_buffer_length(&buffer), encoding);
-
-    pm_buffer_free(&buffer);
-    pm_node_destroy(&parser, program);
-    pm_parser_free(&parser);
-    pm_string_free(&input);
-    pm_options_free(&options);
-
-    return result;
-}
-
-/**
- * call-seq: Debug::Encoding.all -> Array[Debug::Encoding]
+ *   Prism::StringQuery::method_name?(string) -> bool
  *
- * Return an array of all of the encodings that prism knows about.
+ * Returns true if the string constitutes a valid method name.
  */
 static VALUE
-encoding_all(VALUE self) {
-    VALUE encodings = rb_ary_new();
-
-    for (size_t index = 0; index < PM_ENCODING_MAXIMUM; index++) {
-        const pm_encoding_t *encoding = &pm_encodings[index];
-
-        VALUE encoding_argv[] = { rb_str_new_cstr(encoding->name), encoding->multibyte ? Qtrue : Qfalse };
-        rb_ary_push(encodings, rb_class_new_instance(2, encoding_argv, rb_cPrismDebugEncoding));
-    }
-
-    return encodings;
-}
-
-static const pm_encoding_t *
-encoding_find(VALUE name) {
-    const uint8_t *source = (const uint8_t *) RSTRING_PTR(name);
-    size_t length = RSTRING_LEN(name);
-
-    const pm_encoding_t *encoding = pm_encoding_find(source, source + length);
-    if (encoding == NULL) { rb_raise(rb_eArgError, "Unknown encoding: %s", source); }
-
-    return encoding;
-}
-
-/**
- * call-seq: Debug::Encoding.width(source) -> Integer
- *
- * Returns the width of the first character in the given string if it is valid
- * in the encoding. If it is not, this function returns 0.
- */
-static VALUE
-encoding_char_width(VALUE self, VALUE name, VALUE value) {
-    return ULONG2NUM(encoding_find(name)->char_width((const uint8_t *) RSTRING_PTR(value), RSTRING_LEN(value)));
-}
-
-/**
- * call-seq: Debug::Encoding.alnum?(source) -> true | false
- *
- * Returns true if the first character in the given string is an alphanumeric
- * character in the encoding.
- */
-static VALUE
-encoding_alnum_char(VALUE self, VALUE name, VALUE value) {
-    return encoding_find(name)->alnum_char((const uint8_t *) RSTRING_PTR(value), RSTRING_LEN(value)) > 0 ? Qtrue : Qfalse;
-}
-
-/**
- * call-seq: Debug::Encoding.alpha?(source) -> true | false
- *
- * Returns true if the first character in the given string is an alphabetic
- * character in the encoding.
- */
-static VALUE
-encoding_alpha_char(VALUE self, VALUE name, VALUE value) {
-    return encoding_find(name)->alpha_char((const uint8_t *) RSTRING_PTR(value), RSTRING_LEN(value)) > 0 ? Qtrue : Qfalse;
-}
-
-/**
- * call-seq: Debug::Encoding.upper?(source) -> true | false
- *
- * Returns true if the first character in the given string is an uppercase
- * character in the encoding.
- */
-static VALUE
-encoding_isupper_char(VALUE self, VALUE name, VALUE value) {
-    return encoding_find(name)->isupper_char((const uint8_t *) RSTRING_PTR(value), RSTRING_LEN(value)) ? Qtrue : Qfalse;
+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));
 }
 
 /******************************************************************************/
@@ -1364,22 +1229,24 @@ Init_prism(void) {
     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);
-
-    // Intern all of the options that we support so that we don't have to do it
-    // every time we parse.
-    rb_option_id_command_line = rb_intern_const("command_line");
-    rb_option_id_encoding = rb_intern_const("encoding");
-    rb_option_id_filepath = rb_intern_const("filepath");
-    rb_option_id_frozen_string_literal = rb_intern_const("frozen_string_literal");
-    rb_option_id_line = rb_intern_const("line");
-    rb_option_id_scopes = rb_intern_const("scopes");
-    rb_option_id_version = rb_intern_const("version");
-
-    rb_prism_source_id_for = rb_intern("for");
+    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");
 
     /**
      * The version of the prism library.
@@ -1390,8 +1257,10 @@ Init_prism(void) {
     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_stream", parse_stream, -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);
@@ -1406,28 +1275,9 @@ Init_prism(void) {
     rb_define_singleton_method(rb_cPrism, "dump_file", dump_file, -1);
 #endif
 
-    // Next, the functions that will be called by the parser to perform various
-    // internal tasks. We expose these to make them easier to test.
-    VALUE rb_cPrismDebug = rb_define_module_under(rb_cPrism, "Debug");
-    rb_define_singleton_method(rb_cPrismDebug, "named_captures", named_captures, 1);
-    rb_define_singleton_method(rb_cPrismDebug, "integer_parse", integer_parse, 1);
-    rb_define_singleton_method(rb_cPrismDebug, "memsize", memsize, 1);
-    rb_define_singleton_method(rb_cPrismDebug, "profile_file", profile_file, 1);
-    rb_define_singleton_method(rb_cPrismDebug, "format_errors", format_errors, 2);
-    rb_define_singleton_method(rb_cPrismDebug, "static_inspect", static_inspect, -1);
-
-#ifndef PRISM_EXCLUDE_PRETTYPRINT
-    rb_define_singleton_method(rb_cPrismDebug, "inspect_node", inspect_node, 1);
-#endif
-
-    // Next, define the functions that are exposed through the private
-    // Debug::Encoding class.
-    rb_cPrismDebugEncoding = rb_define_class_under(rb_cPrismDebug, "Encoding", rb_cObject);
-    rb_define_singleton_method(rb_cPrismDebugEncoding, "all", encoding_all, 0);
-    rb_define_singleton_method(rb_cPrismDebugEncoding, "_width", encoding_char_width, 2);
-    rb_define_singleton_method(rb_cPrismDebugEncoding, "_alnum?", encoding_alnum_char, 2);
-    rb_define_singleton_method(rb_cPrismDebugEncoding, "_alpha?", encoding_alpha_char, 2);
-    rb_define_singleton_method(rb_cPrismDebugEncoding, "_upper?", encoding_isupper_char, 2);
+    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();