From b481b673d753339204290d7582dbb91a6e14447a Mon Sep 17 00:00:00 2001 From: yui-knk Date: Sun, 28 May 2023 20:00:20 +0900 Subject: [Feature #19719] Universal Parser Introduce Universal Parser mode for the parser. This commit includes these changes: * Introduce `UNIVERSAL_PARSER` macro. All of CRuby related functions are passed via `struct rb_parser_config_struct` when this macro is enabled. * Add CI task with 'cppflags=-DUNIVERSAL_PARSER' for ubuntu. --- node.c | 1275 +++++++--------------------------------------------------------- 1 file changed, 122 insertions(+), 1153 deletions(-) (limited to 'node.c') diff --git a/node.c b/node.c index 027b00ed06..1d3b7b75e4 100644 --- a/node.c +++ b/node.c @@ -9,1119 +9,40 @@ **********************************************************************/ +#ifdef UNIVERSAL_PARSER + +#include +#include "node.h" +#include "rubyparser.h" +#include "internal/parse.h" +#define T_NODE 0x1b + +#define ruby_xmalloc ast->node_buffer->config->malloc +#undef xfree +#define xfree ast->node_buffer->config->free +#define rb_ident_hash_new ast->node_buffer->config->ident_hash_new +#define rb_xmalloc_mul_add ast->node_buffer->config->xmalloc_mul_add +#define ruby_xrealloc(var,size) (ast->node_buffer->config->realloc_n((void *)var, 1, size)) +#define rb_gc_mark ast->node_buffer->config->gc_mark +#define rb_gc_location ast->node_buffer->config->gc_location +#define rb_gc_mark_movable ast->node_buffer->config->gc_mark_movable +#define Qnil ast->node_buffer->config->qnil +#define Qtrue ast->node_buffer->config->qtrue +#define NIL_P ast->node_buffer->config->nil_p +#define rb_hash_aset ast->node_buffer->config->hash_aset +#define RB_OBJ_WRITE(old, slot, young) ast->node_buffer->config->obj_write((VALUE)(old), (VALUE *)(slot), (VALUE)(young)) + +#else + #include "internal.h" #include "internal/hash.h" #include "internal/variable.h" #include "ruby/ruby.h" #include "vm_core.h" -#define NODE_BUF_DEFAULT_LEN 16 - -#define A(str) rb_str_cat2(buf, (str)) -#define AR(str) rb_str_concat(buf, (str)) - -#define A_INDENT add_indent(buf, indent) -#define D_INDENT rb_str_cat2(indent, next_indent) -#define D_DEDENT rb_str_resize(indent, RSTRING_LEN(indent) - 4) -#define A_ID(id) add_id(buf, (id)) -#define A_INT(val) rb_str_catf(buf, "%d", (val)) -#define A_LONG(val) rb_str_catf(buf, "%ld", (val)) -#define A_LIT(lit) AR(rb_dump_literal(lit)) -#define A_NODE_HEADER(node, term) \ - rb_str_catf(buf, "@ %s (id: %d, line: %d, location: (%d,%d)-(%d,%d))%s"term, \ - ruby_node_name(nd_type(node)), nd_node_id(node), nd_line(node), \ - nd_first_lineno(node), nd_first_column(node), \ - nd_last_lineno(node), nd_last_column(node), \ - (node->flags & NODE_FL_NEWLINE ? "*" : "")) -#define A_FIELD_HEADER(len, name, term) \ - rb_str_catf(buf, "+- %.*s:"term, (len), (name)) -#define D_FIELD_HEADER(len, name, term) (A_INDENT, A_FIELD_HEADER(len, name, term)) - -#define D_NULL_NODE (A_INDENT, A("(null node)\n")) -#define D_NODE_HEADER(node) (A_INDENT, A_NODE_HEADER(node, "\n")) - -#define COMPOUND_FIELD(len, name) \ - FIELD_BLOCK((D_FIELD_HEADER((len), (name), "\n"), D_INDENT), D_DEDENT) - -#define COMPOUND_FIELD1(name, ann) \ - COMPOUND_FIELD(FIELD_NAME_LEN(name, ann), \ - FIELD_NAME_DESC(name, ann)) - -#define FIELD_NAME_DESC(name, ann) name " (" ann ")" -#define FIELD_NAME_LEN(name, ann) (int)( \ - comment ? \ - rb_strlen_lit(FIELD_NAME_DESC(name, ann)) : \ - rb_strlen_lit(name)) -#define SIMPLE_FIELD(len, name) \ - FIELD_BLOCK(D_FIELD_HEADER((len), (name), " "), A("\n")) - -#define FIELD_BLOCK(init, reset) \ - for (init, field_flag = 1; \ - field_flag; /* should be optimized away */ \ - reset, field_flag = 0) - -#define SIMPLE_FIELD1(name, ann) SIMPLE_FIELD(FIELD_NAME_LEN(name, ann), FIELD_NAME_DESC(name, ann)) -#define F_CUSTOM1(name, ann) SIMPLE_FIELD1(#name, ann) -#define F_ID(name, ann) SIMPLE_FIELD1(#name, ann) A_ID(node->name) -#define F_GENTRY(name, ann) SIMPLE_FIELD1(#name, ann) A_ID(node->name) -#define F_INT(name, ann) SIMPLE_FIELD1(#name, ann) A_INT(node->name) -#define F_LONG(name, ann) SIMPLE_FIELD1(#name, ann) A_LONG(node->name) -#define F_LIT(name, ann) SIMPLE_FIELD1(#name, ann) A_LIT(node->name) -#define F_MSG(name, ann, desc) SIMPLE_FIELD1(#name, ann) A(desc) - -#define F_NODE(name, ann) \ - COMPOUND_FIELD1(#name, ann) {dump_node(buf, indent, comment, node->name);} - -#define ANN(ann) \ - if (comment) { \ - A_INDENT; A("| # " ann "\n"); \ - } - -#define LAST_NODE (next_indent = " ") - -VALUE -rb_dump_literal(VALUE lit) -{ - if (!RB_SPECIAL_CONST_P(lit)) { - VALUE str; - switch (RB_BUILTIN_TYPE(lit)) { - case T_CLASS: case T_MODULE: case T_ICLASS: - str = rb_class_path(lit); - if (FL_TEST(lit, FL_SINGLETON)) { - str = rb_sprintf("<%"PRIsVALUE">", str); - } - return str; - default: - break; - } - } - return rb_inspect(lit); -} - -static void -add_indent(VALUE buf, VALUE indent) -{ - AR(indent); -} - -static void -add_id(VALUE buf, ID id) -{ - if (id == 0) { - A("(null)"); - } - else { - VALUE str = rb_id2str(id); - if (str) { - A(":"); AR(str); - } - else { - rb_str_catf(buf, "(internal variable: 0x%"PRIsVALUE")", id); - } - } -} - -struct add_option_arg { - VALUE buf, indent; - st_index_t count; -}; - -static void dump_node(VALUE, VALUE, int, const NODE *); -static const char default_indent[] = "| "; - -static void -dump_array(VALUE buf, VALUE indent, int comment, const NODE *node) -{ - int field_flag; - const char *next_indent = default_indent; - F_LONG(nd_alen, "length"); - F_NODE(nd_head, "element"); - while (node->nd_next && nd_type_p(node->nd_next, NODE_LIST)) { - node = node->nd_next; - F_NODE(nd_head, "element"); - } - LAST_NODE; - F_NODE(nd_next, "next element"); -} - -static void -dump_node(VALUE buf, VALUE indent, int comment, const NODE * node) -{ - int field_flag; - int i; - const char *next_indent = default_indent; - enum node_type type; - - if (!node) { - D_NULL_NODE; - return; - } - - D_NODE_HEADER(node); - - type = nd_type(node); - switch (type) { - case NODE_BLOCK: - ANN("statement sequence"); - ANN("format: [nd_head]; ...; [nd_next]"); - ANN("example: foo; bar"); - i = 0; - do { - A_INDENT; - rb_str_catf(buf, "+- nd_head (%s%d):\n", - comment ? "statement #" : "", ++i); - if (!node->nd_next) LAST_NODE; - D_INDENT; - dump_node(buf, indent, comment, node->nd_head); - D_DEDENT; - } while (node->nd_next && - nd_type_p(node->nd_next, NODE_BLOCK) && - (node = node->nd_next, 1)); - if (node->nd_next) { - LAST_NODE; - F_NODE(nd_next, "next block"); - } - return; - - case NODE_IF: - ANN("if statement"); - ANN("format: if [nd_cond] then [nd_body] else [nd_else] end"); - ANN("example: if x == 1 then foo else bar end"); - F_NODE(nd_cond, "condition expr"); - F_NODE(nd_body, "then clause"); - LAST_NODE; - F_NODE(nd_else, "else clause"); - return; - - case NODE_UNLESS: - ANN("unless statement"); - ANN("format: unless [nd_cond] then [nd_body] else [nd_else] end"); - ANN("example: unless x == 1 then foo else bar end"); - F_NODE(nd_cond, "condition expr"); - F_NODE(nd_body, "then clause"); - LAST_NODE; - F_NODE(nd_else, "else clause"); - return; - - case NODE_CASE: - ANN("case statement"); - ANN("format: case [nd_head]; [nd_body]; end"); - ANN("example: case x; when 1; foo; when 2; bar; else baz; end"); - F_NODE(nd_head, "case expr"); - LAST_NODE; - F_NODE(nd_body, "when clauses"); - return; - case NODE_CASE2: - ANN("case statement with no head"); - ANN("format: case; [nd_body]; end"); - ANN("example: case; when 1; foo; when 2; bar; else baz; end"); - F_NODE(nd_head, "case expr"); - LAST_NODE; - F_NODE(nd_body, "when clauses"); - return; - case NODE_CASE3: - ANN("case statement (pattern matching)"); - ANN("format: case [nd_head]; [nd_body]; end"); - ANN("example: case x; in 1; foo; in 2; bar; else baz; end"); - F_NODE(nd_head, "case expr"); - LAST_NODE; - F_NODE(nd_body, "in clauses"); - return; - - case NODE_WHEN: - ANN("when clause"); - ANN("format: when [nd_head]; [nd_body]; (when or else) [nd_next]"); - ANN("example: case x; when 1; foo; when 2; bar; else baz; end"); - F_NODE(nd_head, "when value"); - F_NODE(nd_body, "when body"); - LAST_NODE; - F_NODE(nd_next, "next when clause"); - return; - - case NODE_IN: - ANN("in clause"); - ANN("format: in [nd_head]; [nd_body]; (in or else) [nd_next]"); - ANN("example: case x; in 1; foo; in 2; bar; else baz; end"); - F_NODE(nd_head, "in pattern"); - F_NODE(nd_body, "in body"); - LAST_NODE; - F_NODE(nd_next, "next in clause"); - return; - - case NODE_WHILE: - ANN("while statement"); - ANN("format: while [nd_cond]; [nd_body]; end"); - ANN("example: while x == 1; foo; end"); - goto loop; - case NODE_UNTIL: - ANN("until statement"); - ANN("format: until [nd_cond]; [nd_body]; end"); - ANN("example: until x == 1; foo; end"); - loop: - F_CUSTOM1(nd_state, "begin-end-while?") { - A_INT((int)node->nd_state); - A((node->nd_state == 1) ? " (while-end)" : " (begin-end-while)"); - } - F_NODE(nd_cond, "condition"); - LAST_NODE; - F_NODE(nd_body, "body"); - return; - - case NODE_ITER: - ANN("method call with block"); - ANN("format: [nd_iter] { [nd_body] }"); - ANN("example: 3.times { foo }"); - goto iter; - case NODE_FOR: - ANN("for statement"); - ANN("format: for * in [nd_iter] do [nd_body] end"); - ANN("example: for i in 1..3 do foo end"); - iter: - F_NODE(nd_iter, "iteration receiver"); - LAST_NODE; - F_NODE(nd_body, "body"); - return; - - case NODE_FOR_MASGN: - ANN("vars of for statement with masgn"); - ANN("format: for [nd_var] in ... do ... end"); - ANN("example: for x, y in 1..3 do foo end"); - LAST_NODE; - F_NODE(nd_var, "var"); - return; - - case NODE_BREAK: - ANN("break statement"); - ANN("format: break [nd_stts]"); - ANN("example: break 1"); - goto jump; - case NODE_NEXT: - ANN("next statement"); - ANN("format: next [nd_stts]"); - ANN("example: next 1"); - goto jump; - case NODE_RETURN: - ANN("return statement"); - ANN("format: return [nd_stts]"); - ANN("example: return 1"); - jump: - LAST_NODE; - F_NODE(nd_stts, "value"); - return; - - case NODE_REDO: - ANN("redo statement"); - ANN("format: redo"); - ANN("example: redo"); - return; - - case NODE_RETRY: - ANN("retry statement"); - ANN("format: retry"); - ANN("example: retry"); - return; - - case NODE_BEGIN: - ANN("begin statement"); - ANN("format: begin; [nd_body]; end"); - ANN("example: begin; 1; end"); - LAST_NODE; - F_NODE(nd_body, "body"); - return; - - case NODE_RESCUE: - ANN("rescue clause"); - ANN("format: begin; [nd_body]; (rescue) [nd_resq]; else [nd_else]; end"); - ANN("example: begin; foo; rescue; bar; else; baz; end"); - F_NODE(nd_head, "body"); - F_NODE(nd_resq, "rescue clause list"); - LAST_NODE; - F_NODE(nd_else, "rescue else clause"); - return; - - case NODE_RESBODY: - ANN("rescue clause (cont'd)"); - ANN("format: rescue [nd_args]; [nd_body]; (rescue) [nd_head]"); - ANN("example: begin; foo; rescue; bar; else; baz; end"); - F_NODE(nd_args, "rescue exceptions"); - F_NODE(nd_body, "rescue clause"); - LAST_NODE; - F_NODE(nd_head, "next rescue clause"); - return; - - case NODE_ENSURE: - ANN("ensure clause"); - ANN("format: begin; [nd_head]; ensure; [nd_ensr]; end"); - ANN("example: begin; foo; ensure; bar; end"); - F_NODE(nd_head, "body"); - LAST_NODE; - F_NODE(nd_ensr, "ensure clause"); - return; - - case NODE_AND: - ANN("&& operator"); - ANN("format: [nd_1st] && [nd_2nd]"); - ANN("example: foo && bar"); - goto andor; - case NODE_OR: - ANN("|| operator"); - ANN("format: [nd_1st] || [nd_2nd]"); - ANN("example: foo || bar"); - andor: - while (1) { - F_NODE(nd_1st, "left expr"); - if (!node->nd_2nd || !nd_type_p(node->nd_2nd, type)) - break; - node = node->nd_2nd; - } - LAST_NODE; - F_NODE(nd_2nd, "right expr"); - return; - - case NODE_MASGN: - ANN("multiple assignment"); - ANN("format: [nd_head], [nd_args] = [nd_value]"); - ANN("example: a, b = foo"); - F_NODE(nd_value, "rhsn"); - F_NODE(nd_head, "lhsn"); - if (NODE_NAMED_REST_P(node->nd_args)) { - LAST_NODE; - F_NODE(nd_args, "splatn"); - } - else { - F_MSG(nd_args, "splatn", "NODE_SPECIAL_NO_NAME_REST (rest argument without name)"); - } - return; - - case NODE_LASGN: - ANN("local variable assignment"); - ANN("format: [nd_vid](lvar) = [nd_value]"); - ANN("example: x = foo"); - F_ID(nd_vid, "local variable"); - if (NODE_REQUIRED_KEYWORD_P(node)) { - F_MSG(nd_value, "rvalue", "NODE_SPECIAL_REQUIRED_KEYWORD (required keyword argument)"); - } - else { - LAST_NODE; - F_NODE(nd_value, "rvalue"); - } - return; - case NODE_DASGN: - ANN("dynamic variable assignment"); - ANN("format: [nd_vid](dvar) = [nd_value]"); - ANN("example: x = nil; 1.times { x = foo }"); - ANN("example: 1.times { x = foo }"); - F_ID(nd_vid, "local variable"); - if (NODE_REQUIRED_KEYWORD_P(node)) { - F_MSG(nd_value, "rvalue", "NODE_SPECIAL_REQUIRED_KEYWORD (required keyword argument)"); - } - else { - LAST_NODE; - F_NODE(nd_value, "rvalue"); - } - return; - case NODE_IASGN: - ANN("instance variable assignment"); - ANN("format: [nd_vid](ivar) = [nd_value]"); - ANN("example: @x = foo"); - F_ID(nd_vid, "instance variable"); - LAST_NODE; - F_NODE(nd_value, "rvalue"); - return; - case NODE_CVASGN: - ANN("class variable assignment"); - ANN("format: [nd_vid](cvar) = [nd_value]"); - ANN("example: @@x = foo"); - F_ID(nd_vid, "class variable"); - LAST_NODE; - F_NODE(nd_value, "rvalue"); - return; - case NODE_GASGN: - ANN("global variable assignment"); - ANN("format: [nd_entry](gvar) = [nd_value]"); - ANN("example: $x = foo"); - F_GENTRY(nd_entry, "global variable"); - LAST_NODE; - F_NODE(nd_value, "rvalue"); - return; - - case NODE_CDECL: - ANN("constant declaration"); - ANN("format: [nd_else]::[nd_vid](constant) = [nd_value]"); - ANN("example: X = foo"); - if (node->nd_vid) { - F_ID(nd_vid, "constant"); - F_MSG(nd_else, "extension", "not used"); - } - else { - F_MSG(nd_vid, "constant", "0 (see extension field)"); - F_NODE(nd_else, "extension"); - } - LAST_NODE; - F_NODE(nd_value, "rvalue"); - return; - - case NODE_OP_ASGN1: - ANN("array assignment with operator"); - ANN("format: [nd_recv] [ [nd_args->nd_head] ] [nd_mid]= [nd_args->nd_body]"); - ANN("example: ary[1] += foo"); - F_NODE(nd_recv, "receiver"); - F_ID(nd_mid, "operator"); - F_NODE(nd_args->nd_head, "index"); - LAST_NODE; - F_NODE(nd_args->nd_body, "rvalue"); - return; - - case NODE_OP_ASGN2: - ANN("attr assignment with operator"); - ANN("format: [nd_recv].[attr] [nd_next->nd_mid]= [nd_value]"); - ANN(" where [attr]: [nd_next->nd_vid]"); - ANN("example: struct.field += foo"); - F_NODE(nd_recv, "receiver"); - F_CUSTOM1(nd_next->nd_vid, "attr") { - if (node->nd_next->nd_aid) A("? "); - A_ID(node->nd_next->nd_vid); - } - F_ID(nd_next->nd_mid, "operator"); - LAST_NODE; - F_NODE(nd_value, "rvalue"); - return; - - case NODE_OP_ASGN_AND: - ANN("assignment with && operator"); - ANN("format: [nd_head] &&= [nd_value]"); - ANN("example: foo &&= bar"); - goto asgn_andor; - case NODE_OP_ASGN_OR: - ANN("assignment with || operator"); - ANN("format: [nd_head] ||= [nd_value]"); - ANN("example: foo ||= bar"); - asgn_andor: - F_NODE(nd_head, "variable"); - LAST_NODE; - F_NODE(nd_value, "rvalue"); - return; - - case NODE_OP_CDECL: - ANN("constant declaration with operator"); - ANN("format: [nd_head](constant) [nd_aid]= [nd_value]"); - ANN("example: A::B ||= 1"); - F_NODE(nd_head, "constant"); - F_ID(nd_aid, "operator"); - LAST_NODE; - F_NODE(nd_value, "rvalue"); - return; - - case NODE_CALL: - ANN("method invocation"); - ANN("format: [nd_recv].[nd_mid]([nd_args])"); - ANN("example: obj.foo(1)"); - F_ID(nd_mid, "method id"); - F_NODE(nd_recv, "receiver"); - LAST_NODE; - F_NODE(nd_args, "arguments"); - return; - - case NODE_OPCALL: - ANN("method invocation"); - ANN("format: [nd_recv] [nd_mid] [nd_args]"); - ANN("example: foo + bar"); - F_ID(nd_mid, "method id"); - F_NODE(nd_recv, "receiver"); - LAST_NODE; - F_NODE(nd_args, "arguments"); - return; - - case NODE_FCALL: - ANN("function call"); - ANN("format: [nd_mid]([nd_args])"); - ANN("example: foo(1)"); - F_ID(nd_mid, "method id"); - LAST_NODE; - F_NODE(nd_args, "arguments"); - return; - - case NODE_VCALL: - ANN("function call with no argument"); - ANN("format: [nd_mid]"); - ANN("example: foo"); - F_ID(nd_mid, "method id"); - return; - - case NODE_QCALL: - ANN("safe method invocation"); - ANN("format: [nd_recv]&.[nd_mid]([nd_args])"); - ANN("example: obj&.foo(1)"); - F_ID(nd_mid, "method id"); - F_NODE(nd_recv, "receiver"); - LAST_NODE; - F_NODE(nd_args, "arguments"); - return; - - case NODE_SUPER: - ANN("super invocation"); - ANN("format: super [nd_args]"); - ANN("example: super 1"); - LAST_NODE; - F_NODE(nd_args, "arguments"); - return; - - case NODE_ZSUPER: - ANN("super invocation with no argument"); - ANN("format: super"); - ANN("example: super"); - return; - - case NODE_LIST: - ANN("list constructor"); - ANN("format: [ [nd_head], [nd_next].. ] (length: [nd_alen])"); - ANN("example: [1, 2, 3]"); - goto ary; - case NODE_VALUES: - ANN("return arguments"); - ANN("format: [ [nd_head], [nd_next].. ] (length: [nd_alen])"); - ANN("example: return 1, 2, 3"); - ary: - dump_array(buf, indent, comment, node); - return; - - case NODE_ZLIST: - ANN("empty list constructor"); - ANN("format: []"); - ANN("example: []"); - return; - - case NODE_HASH: - if (!node->nd_brace) { - ANN("keyword arguments"); - ANN("format: nd_head"); - ANN("example: a: 1, b: 2"); - } - else { - ANN("hash constructor"); - ANN("format: { [nd_head] }"); - ANN("example: { 1 => 2, 3 => 4 }"); - } - F_CUSTOM1(nd_brace, "keyword arguments or hash literal") { - switch (node->nd_brace) { - case 0: A("0 (keyword argument)"); break; - case 1: A("1 (hash literal)"); break; - } - } - LAST_NODE; - F_NODE(nd_head, "contents"); - return; - - case NODE_YIELD: - ANN("yield invocation"); - ANN("format: yield [nd_head]"); - ANN("example: yield 1"); - LAST_NODE; - F_NODE(nd_head, "arguments"); - return; - - case NODE_LVAR: - ANN("local variable reference"); - ANN("format: [nd_vid](lvar)"); - ANN("example: x"); - F_ID(nd_vid, "local variable"); - return; - case NODE_DVAR: - ANN("dynamic variable reference"); - ANN("format: [nd_vid](dvar)"); - ANN("example: 1.times { x = 1; x }"); - F_ID(nd_vid, "local variable"); - return; - case NODE_IVAR: - ANN("instance variable reference"); - ANN("format: [nd_vid](ivar)"); - ANN("example: @x"); - F_ID(nd_vid, "instance variable"); - return; - case NODE_CONST: - ANN("constant reference"); - ANN("format: [nd_vid](constant)"); - ANN("example: X"); - F_ID(nd_vid, "constant"); - return; - case NODE_CVAR: - ANN("class variable reference"); - ANN("format: [nd_vid](cvar)"); - ANN("example: @@x"); - F_ID(nd_vid, "class variable"); - return; - - case NODE_GVAR: - ANN("global variable reference"); - ANN("format: [nd_entry](gvar)"); - ANN("example: $x"); - F_GENTRY(nd_entry, "global variable"); - return; - - case NODE_NTH_REF: - ANN("nth special variable reference"); - ANN("format: $[nd_nth]"); - ANN("example: $1, $2, .."); - F_CUSTOM1(nd_nth, "variable") { A("$"); A_LONG(node->nd_nth); } - return; - - case NODE_BACK_REF: - ANN("back special variable reference"); - ANN("format: $[nd_nth]"); - ANN("example: $&, $`, $', $+"); - F_CUSTOM1(nd_nth, "variable") { - char name[3] = "$ "; - name[1] = (char)node->nd_nth; - A(name); - } - return; - - case NODE_MATCH: - ANN("match expression (against $_ implicitly)"); - ANN("format: [nd_lit] (in condition)"); - ANN("example: if /foo/; foo; end"); - F_LIT(nd_lit, "regexp"); - return; - - case NODE_MATCH2: - ANN("match expression (regexp first)"); - ANN("format: [nd_recv] =~ [nd_value]"); - ANN("example: /foo/ =~ 'foo'"); - F_NODE(nd_recv, "regexp (receiver)"); - if (!node->nd_args) LAST_NODE; - F_NODE(nd_value, "string (argument)"); - if (node->nd_args) { - LAST_NODE; - F_NODE(nd_args, "named captures"); - } - return; +#endif - case NODE_MATCH3: - ANN("match expression (regexp second)"); - ANN("format: [nd_recv] =~ [nd_value]"); - ANN("example: 'foo' =~ /foo/"); - F_NODE(nd_recv, "string (receiver)"); - LAST_NODE; - F_NODE(nd_value, "regexp (argument)"); - return; - - case NODE_LIT: - ANN("literal"); - ANN("format: [nd_lit]"); - ANN("example: 1, /foo/"); - goto lit; - case NODE_STR: - ANN("string literal"); - ANN("format: [nd_lit]"); - ANN("example: 'foo'"); - goto lit; - case NODE_XSTR: - ANN("xstring literal"); - ANN("format: [nd_lit]"); - ANN("example: `foo`"); - lit: - F_LIT(nd_lit, "literal"); - return; - - case NODE_ONCE: - ANN("once evaluation"); - ANN("format: [nd_body]"); - ANN("example: /foo#{ bar }baz/o"); - LAST_NODE; - F_NODE(nd_body, "body"); - return; - case NODE_DSTR: - ANN("string literal with interpolation"); - ANN("format: [nd_lit]"); - ANN("example: \"foo#{ bar }baz\""); - goto dlit; - case NODE_DXSTR: - ANN("xstring literal with interpolation"); - ANN("format: [nd_lit]"); - ANN("example: `foo#{ bar }baz`"); - goto dlit; - case NODE_DREGX: - ANN("regexp literal with interpolation"); - ANN("format: [nd_lit]"); - ANN("example: /foo#{ bar }baz/"); - goto dlit; - case NODE_DSYM: - ANN("symbol literal with interpolation"); - ANN("format: [nd_lit]"); - ANN("example: :\"foo#{ bar }baz\""); - dlit: - F_LIT(nd_lit, "preceding string"); - if (!node->nd_next) return; - F_NODE(nd_next->nd_head, "interpolation"); - LAST_NODE; - F_NODE(nd_next->nd_next, "tailing strings"); - return; - - case NODE_EVSTR: - ANN("interpolation expression"); - ANN("format: \"..#{ [nd_lit] }..\""); - ANN("example: \"foo#{ bar }baz\""); - LAST_NODE; - F_NODE(nd_body, "body"); - return; - - case NODE_ARGSCAT: - ANN("splat argument following arguments"); - ANN("format: ..(*[nd_head], [nd_body..])"); - ANN("example: foo(*ary, post_arg1, post_arg2)"); - F_NODE(nd_head, "preceding array"); - LAST_NODE; - F_NODE(nd_body, "following array"); - return; - - case NODE_ARGSPUSH: - ANN("splat argument following one argument"); - ANN("format: ..(*[nd_head], [nd_body])"); - ANN("example: foo(*ary, post_arg)"); - F_NODE(nd_head, "preceding array"); - LAST_NODE; - F_NODE(nd_body, "following element"); - return; - - case NODE_SPLAT: - ANN("splat argument"); - ANN("format: *[nd_head]"); - ANN("example: foo(*ary)"); - LAST_NODE; - F_NODE(nd_head, "splat'ed array"); - return; - - case NODE_BLOCK_PASS: - ANN("arguments with block argument"); - ANN("format: ..([nd_head], &[nd_body])"); - ANN("example: foo(x, &blk)"); - F_NODE(nd_head, "other arguments"); - LAST_NODE; - F_NODE(nd_body, "block argument"); - return; - - case NODE_DEFN: - ANN("method definition"); - ANN("format: def [nd_mid] [nd_defn]; end"); - ANN("example: def foo; bar; end"); - F_ID(nd_mid, "method name"); - LAST_NODE; - F_NODE(nd_defn, "method definition"); - return; - - case NODE_DEFS: - ANN("singleton method definition"); - ANN("format: def [nd_recv].[nd_mid] [nd_defn]; end"); - ANN("example: def obj.foo; bar; end"); - F_NODE(nd_recv, "receiver"); - F_ID(nd_mid, "method name"); - LAST_NODE; - F_NODE(nd_defn, "method definition"); - return; - - case NODE_ALIAS: - ANN("method alias statement"); - ANN("format: alias [nd_1st] [nd_2nd]"); - ANN("example: alias bar foo"); - F_NODE(nd_1st, "new name"); - LAST_NODE; - F_NODE(nd_2nd, "old name"); - return; - - case NODE_VALIAS: - ANN("global variable alias statement"); - ANN("format: alias [nd_alias](gvar) [nd_orig](gvar)"); - ANN("example: alias $y $x"); - F_ID(nd_alias, "new name"); - F_ID(nd_orig, "old name"); - return; - - case NODE_UNDEF: - ANN("method undef statement"); - ANN("format: undef [nd_undef]"); - ANN("example: undef foo"); - LAST_NODE; - F_NODE(nd_undef, "old name"); - return; - - case NODE_CLASS: - ANN("class definition"); - ANN("format: class [nd_cpath] < [nd_super]; [nd_body]; end"); - ANN("example: class C2 < C; ..; end"); - F_NODE(nd_cpath, "class path"); - F_NODE(nd_super, "superclass"); - LAST_NODE; - F_NODE(nd_body, "class definition"); - return; - - case NODE_MODULE: - ANN("module definition"); - ANN("format: module [nd_cpath]; [nd_body]; end"); - ANN("example: module M; ..; end"); - F_NODE(nd_cpath, "module path"); - LAST_NODE; - F_NODE(nd_body, "module definition"); - return; - - case NODE_SCLASS: - ANN("singleton class definition"); - ANN("format: class << [nd_recv]; [nd_body]; end"); - ANN("example: class << obj; ..; end"); - F_NODE(nd_recv, "receiver"); - LAST_NODE; - F_NODE(nd_body, "singleton class definition"); - return; - - case NODE_COLON2: - ANN("scoped constant reference"); - ANN("format: [nd_head]::[nd_mid]"); - ANN("example: M::C"); - F_ID(nd_mid, "constant name"); - LAST_NODE; - F_NODE(nd_head, "receiver"); - return; - - case NODE_COLON3: - ANN("top-level constant reference"); - ANN("format: ::[nd_mid]"); - ANN("example: ::Object"); - F_ID(nd_mid, "constant name"); - return; - - case NODE_DOT2: - ANN("range constructor (incl.)"); - ANN("format: [nd_beg]..[nd_end]"); - ANN("example: 1..5"); - goto dot; - case NODE_DOT3: - ANN("range constructor (excl.)"); - ANN("format: [nd_beg]...[nd_end]"); - ANN("example: 1...5"); - goto dot; - case NODE_FLIP2: - ANN("flip-flop condition (incl.)"); - ANN("format: [nd_beg]..[nd_end]"); - ANN("example: if (x==1)..(x==5); foo; end"); - goto dot; - case NODE_FLIP3: - ANN("flip-flop condition (excl.)"); - ANN("format: [nd_beg]...[nd_end]"); - ANN("example: if (x==1)...(x==5); foo; end"); - dot: - F_NODE(nd_beg, "begin"); - LAST_NODE; - F_NODE(nd_end, "end"); - return; - - case NODE_SELF: - ANN("self"); - ANN("format: self"); - ANN("example: self"); - return; - - case NODE_NIL: - ANN("nil"); - ANN("format: nil"); - ANN("example: nil"); - return; - - case NODE_TRUE: - ANN("true"); - ANN("format: true"); - ANN("example: true"); - return; - - case NODE_FALSE: - ANN("false"); - ANN("format: false"); - ANN("example: false"); - return; - - case NODE_ERRINFO: - ANN("virtual reference to $!"); - ANN("format: rescue => id"); - ANN("example: rescue => id"); - return; - - case NODE_DEFINED: - ANN("defined? expression"); - ANN("format: defined?([nd_head])"); - ANN("example: defined?(foo)"); - F_NODE(nd_head, "expr"); - return; - - case NODE_POSTEXE: - ANN("post-execution"); - ANN("format: END { [nd_body] }"); - ANN("example: END { foo }"); - LAST_NODE; - F_NODE(nd_body, "END clause"); - return; - - case NODE_ATTRASGN: - ANN("attr assignment"); - ANN("format: [nd_recv].[nd_mid] = [nd_args]"); - ANN("example: struct.field = foo"); - F_NODE(nd_recv, "receiver"); - F_ID(nd_mid, "method name"); - LAST_NODE; - F_NODE(nd_args, "arguments"); - return; - - case NODE_LAMBDA: - ANN("lambda expression"); - ANN("format: -> [nd_body]"); - ANN("example: -> { foo }"); - LAST_NODE; - F_NODE(nd_body, "lambda clause"); - return; - - case NODE_OPT_ARG: - ANN("optional arguments"); - ANN("format: def method_name([nd_body=some], [nd_next..])"); - ANN("example: def foo(a, b=1, c); end"); - F_NODE(nd_body, "body"); - LAST_NODE; - F_NODE(nd_next, "next"); - return; - - case NODE_KW_ARG: - ANN("keyword arguments"); - ANN("format: def method_name([nd_body=some], [nd_next..])"); - ANN("example: def foo(a:1, b:2); end"); - F_NODE(nd_body, "body"); - LAST_NODE; - F_NODE(nd_next, "next"); - return; - - case NODE_POSTARG: - ANN("post arguments"); - ANN("format: *[nd_1st], [nd_2nd..] = .."); - ANN("example: a, *rest, z = foo"); - if (NODE_NAMED_REST_P(node->nd_1st)) { - F_NODE(nd_1st, "rest argument"); - } - else { - F_MSG(nd_1st, "rest argument", "NODE_SPECIAL_NO_NAME_REST (rest argument without name)"); - } - LAST_NODE; - F_NODE(nd_2nd, "post arguments"); - return; - - case NODE_ARGS: - ANN("method parameters"); - ANN("format: def method_name(.., [nd_ainfo->nd_optargs], *[nd_ainfo->rest_arg], [nd_ainfo->first_post_arg], .., [nd_ainfo->kw_args], **[nd_ainfo->kw_rest_arg], &[nd_ainfo->block_arg])"); - ANN("example: def foo(a, b, opt1=1, opt2=2, *rest, y, z, kw: 1, **kwrest, &blk); end"); - F_INT(nd_ainfo->pre_args_num, "count of mandatory (pre-)arguments"); - F_NODE(nd_ainfo->pre_init, "initialization of (pre-)arguments"); - F_INT(nd_ainfo->post_args_num, "count of mandatory post-arguments"); - F_NODE(nd_ainfo->post_init, "initialization of post-arguments"); - F_ID(nd_ainfo->first_post_arg, "first post argument"); - F_CUSTOM1(nd_ainfo->rest_arg, "rest argument") { - if (node->nd_ainfo->rest_arg == NODE_SPECIAL_EXCESSIVE_COMMA) { - A("1 (excessed comma)"); - } - else { - A_ID(node->nd_ainfo->rest_arg); - } - } - F_ID(nd_ainfo->block_arg, "block argument"); - F_NODE(nd_ainfo->opt_args, "optional arguments"); - F_NODE(nd_ainfo->kw_args, "keyword arguments"); - LAST_NODE; - F_NODE(nd_ainfo->kw_rest_arg, "keyword rest argument"); - return; - - case NODE_SCOPE: - ANN("new scope"); - ANN("format: [nd_tbl]: local table, [nd_args]: arguments, [nd_body]: body"); - F_CUSTOM1(nd_tbl, "local table") { - rb_ast_id_table_t *tbl = node->nd_tbl; - int i; - int size = tbl ? tbl->size : 0; - if (size == 0) A("(empty)"); - for (i = 0; i < size; i++) { - A_ID(tbl->ids[i]); if (i < size - 1) A(","); - } - } - F_NODE(nd_args, "arguments"); - LAST_NODE; - F_NODE(nd_body, "body"); - return; - - case NODE_ARYPTN: - ANN("array pattern"); - ANN("format: [nd_pconst]([pre_args], ..., *[rest_arg], [post_args], ...)"); - F_NODE(nd_pconst, "constant"); - F_NODE(nd_apinfo->pre_args, "pre arguments"); - if (NODE_NAMED_REST_P(node->nd_apinfo->rest_arg)) { - F_NODE(nd_apinfo->rest_arg, "rest argument"); - } - else { - F_MSG(nd_apinfo->rest_arg, "rest argument", "NODE_SPECIAL_NO_NAME_REST (rest argument without name)"); - } - LAST_NODE; - F_NODE(nd_apinfo->post_args, "post arguments"); - return; - - case NODE_FNDPTN: - ANN("find pattern"); - ANN("format: [nd_pconst](*[pre_rest_arg], args, ..., *[post_rest_arg])"); - F_NODE(nd_pconst, "constant"); - if (NODE_NAMED_REST_P(node->nd_fpinfo->pre_rest_arg)) { - F_NODE(nd_fpinfo->pre_rest_arg, "pre rest argument"); - } - else { - F_MSG(nd_fpinfo->pre_rest_arg, "pre rest argument", "NODE_SPECIAL_NO_NAME_REST (rest argument without name)"); - } - F_NODE(nd_fpinfo->args, "arguments"); - - LAST_NODE; - if (NODE_NAMED_REST_P(node->nd_fpinfo->post_rest_arg)) { - F_NODE(nd_fpinfo->post_rest_arg, "post rest argument"); - } - else { - F_MSG(nd_fpinfo->post_rest_arg, "post rest argument", "NODE_SPECIAL_NO_NAME_REST (rest argument without name)"); - } - return; - - case NODE_HSHPTN: - ANN("hash pattern"); - ANN("format: [nd_pconst]([nd_pkwargs], ..., **[nd_pkwrestarg])"); - F_NODE(nd_pconst, "constant"); - F_NODE(nd_pkwargs, "keyword arguments"); - LAST_NODE; - if (node->nd_pkwrestarg == NODE_SPECIAL_NO_REST_KEYWORD) { - F_MSG(nd_pkwrestarg, "keyword rest argument", "NODE_SPECIAL_NO_REST_KEYWORD (**nil)"); - } - else { - F_NODE(nd_pkwrestarg, "keyword rest argument"); - } - return; - case NODE_ERROR: - ANN("Broken input recovered by Error Tolerant mode"); - return; - - case NODE_ARGS_AUX: - case NODE_LAST: - break; - } - - rb_bug("dump_node: unknown node: %s", ruby_node_name(nd_type(node))); -} - -VALUE -rb_parser_dump_tree(const NODE *node, int comment) -{ - VALUE buf = rb_str_new_cstr( - "###########################################################\n" - "## Do NOT use this node dump for any purpose other than ##\n" - "## debug and research. Compatibility is not guaranteed. ##\n" - "###########################################################\n\n" - ); - dump_node(buf, rb_str_new_cstr("# "), comment, node); - return buf; -} +#define NODE_BUF_DEFAULT_LEN 16 /* Setup NODE structure. * NODE is not an object managed by GC, but it imitates an object @@ -1145,40 +66,31 @@ rb_node_init(NODE *n, enum node_type type, VALUE a0, VALUE a1, VALUE a2) } const char * -ruby_node_name(int node) +rb_node_name(int node) { switch (node) { #include "node_name.inc" default: - rb_bug("unknown node: %d", node); return 0; } } -typedef struct node_buffer_elem_struct { - struct node_buffer_elem_struct *next; - long len; - NODE buf[FLEX_ARY_LEN]; -} node_buffer_elem_t; - -typedef struct { - long idx, len; - node_buffer_elem_t *head; - node_buffer_elem_t *last; -} node_buffer_list_t; +#ifdef UNIVERSAL_PARSER +const char * +ruby_node_name(int node) +{ + return rb_node_name(node); +} +#else +const char * +ruby_node_name(int node) +{ + const char *name = rb_node_name(node); -struct node_buffer_struct { - node_buffer_list_t unmarkable; - node_buffer_list_t markable; - struct rb_ast_local_table_link *local_tables; - VALUE mark_hash; - // - id (sequence number) - // - token_type - // - text of token - // - location info - // Array, whose entry is array - VALUE tokens; -}; + if (!name) rb_bug("unknown node: %d", node); + return name; +} +#endif static void init_node_buffer_list(node_buffer_list_t * nb, node_buffer_elem_t *head) @@ -1190,6 +102,26 @@ init_node_buffer_list(node_buffer_list_t * nb, node_buffer_elem_t *head) nb->head->next = NULL; } +#ifdef UNIVERSAL_PARSER +static node_buffer_t * +rb_node_buffer_new(rb_parser_config_t *config) +{ + const size_t bucket_size = offsetof(node_buffer_elem_t, buf) + NODE_BUF_DEFAULT_LEN * sizeof(NODE); + const size_t alloc_size = sizeof(node_buffer_t) + (bucket_size * 2); + STATIC_ASSERT( + integer_overflow, + offsetof(node_buffer_elem_t, buf) + NODE_BUF_DEFAULT_LEN * sizeof(NODE) + > sizeof(node_buffer_t) + 2 * sizeof(node_buffer_elem_t)); + node_buffer_t *nb = config->malloc(alloc_size); + init_node_buffer_list(&nb->unmarkable, (node_buffer_elem_t*)&nb[1]); + init_node_buffer_list(&nb->markable, (node_buffer_elem_t*)((size_t)nb->unmarkable.head + bucket_size)); + nb->local_tables = 0; + nb->mark_hash = config->qnil; + nb->tokens = config->qnil; + nb->config = config; + return nb; +} +#else static node_buffer_t * rb_node_buffer_new(void) { @@ -1207,9 +139,10 @@ rb_node_buffer_new(void) nb->tokens = Qnil; return nb; } +#endif static void -node_buffer_list_free(node_buffer_list_t * nb) +node_buffer_list_free(rb_ast_t *ast, node_buffer_list_t * nb) { node_buffer_elem_t *nbe = nb->head; @@ -1229,10 +162,10 @@ struct rb_ast_local_table_link { }; static void -rb_node_buffer_free(node_buffer_t *nb) +rb_node_buffer_free(rb_ast_t *ast, node_buffer_t *nb) { - node_buffer_list_free(&nb->unmarkable); - node_buffer_list_free(&nb->markable); + node_buffer_list_free(ast, &nb->unmarkable); + node_buffer_list_free(ast, &nb->markable); struct rb_ast_local_table_link *local_table = nb->local_tables; while (local_table) { struct rb_ast_local_table_link *next_table = local_table->next; @@ -1243,7 +176,7 @@ rb_node_buffer_free(node_buffer_t *nb) } static NODE * -ast_newnode_in_bucket(node_buffer_list_t *nb) +ast_newnode_in_bucket(rb_ast_t *ast, node_buffer_list_t *nb) { if (nb->idx >= nb->len) { long n = nb->len * 2; @@ -1286,9 +219,10 @@ rb_ast_newnode(rb_ast_t *ast, enum node_type type) node_buffer_t *nb = ast->node_buffer; node_buffer_list_t *bucket = (nodetype_markable_p(type) ? &nb->markable : &nb->unmarkable); - return ast_newnode_in_bucket(bucket); + return ast_newnode_in_bucket(ast, bucket); } +#if RUBY_DEBUG void rb_ast_node_type_change(NODE *n, enum node_type type) { @@ -1298,6 +232,7 @@ rb_ast_node_type_change(NODE *n, enum node_type type) ruby_node_name(old_type), ruby_node_name(type)); } } +#endif rb_ast_id_table_t * rb_ast_new_local_table(rb_ast_t *ast, int size) @@ -1331,6 +266,15 @@ rb_ast_delete_node(rb_ast_t *ast, NODE *n) /* should we implement freelist? */ } +#ifdef UNIVERSAL_PARSER +rb_ast_t * +rb_ast_new(rb_parser_config_t *config) +{ + node_buffer_t *nb = rb_node_buffer_new(config); + config->counter++; + return config->ast_new((VALUE)nb); +} +#else rb_ast_t * rb_ast_new(void) { @@ -1338,36 +282,41 @@ rb_ast_new(void) rb_ast_t *ast = (rb_ast_t *)rb_imemo_new(imemo_ast, 0, 0, 0, (VALUE)nb); return ast; } +#endif -typedef void node_itr_t(void *ctx, NODE * node); +typedef void node_itr_t(rb_ast_t *ast, void *ctx, NODE * node); static void -iterate_buffer_elements(node_buffer_elem_t *nbe, long len, node_itr_t *func, void *ctx) +iterate_buffer_elements(rb_ast_t *ast, node_buffer_elem_t *nbe, long len, node_itr_t *func, void *ctx) { long cursor; for (cursor = 0; cursor < len; cursor++) { - func(ctx, &nbe->buf[cursor]); + func(ast, ctx, &nbe->buf[cursor]); } } static void -iterate_node_values(node_buffer_list_t *nb, node_itr_t * func, void *ctx) +iterate_node_values(rb_ast_t *ast, node_buffer_list_t *nb, node_itr_t * func, void *ctx) { node_buffer_elem_t *nbe = nb->head; /* iterate over the head first because it's not full */ - iterate_buffer_elements(nbe, nb->idx, func, ctx); + iterate_buffer_elements(ast, nbe, nb->idx, func, ctx); nbe = nbe->next; while (nbe) { - iterate_buffer_elements(nbe, nbe->len, func, ctx); + iterate_buffer_elements(ast, nbe, nbe->len, func, ctx); nbe = nbe->next; } } static void -mark_ast_value(void *ctx, NODE * node) +mark_ast_value(rb_ast_t *ast, void *ctx, NODE * node) { +#ifdef UNIVERSAL_PARSER + bug_report_func rb_bug = ast->node_buffer->config->bug; +#endif + switch (nd_type(node)) { case NODE_ARGS: { @@ -1395,8 +344,12 @@ mark_ast_value(void *ctx, NODE * node) } static void -update_ast_value(void *ctx, NODE * node) +update_ast_value(rb_ast_t *ast, void *ctx, NODE * node) { +#ifdef UNIVERSAL_PARSER + bug_report_func rb_bug = ast->node_buffer->config->bug; +#endif + switch (nd_type(node)) { case NODE_ARGS: { @@ -1429,7 +382,7 @@ rb_ast_update_references(rb_ast_t *ast) if (ast->node_buffer) { node_buffer_t *nb = ast->node_buffer; - iterate_node_values(&nb->markable, update_ast_value, NULL); + iterate_node_values(ast, &nb->markable, update_ast_value, NULL); } } @@ -1439,22 +392,29 @@ rb_ast_mark(rb_ast_t *ast) if (ast->node_buffer) { rb_gc_mark(ast->node_buffer->mark_hash); rb_gc_mark(ast->node_buffer->tokens); - } - if (ast->body.compile_option) rb_gc_mark(ast->body.compile_option); - if (ast->node_buffer) { + if (ast->body.compile_option) rb_gc_mark(ast->body.compile_option); node_buffer_t *nb = ast->node_buffer; - - iterate_node_values(&nb->markable, mark_ast_value, NULL); + iterate_node_values(ast, &nb->markable, mark_ast_value, NULL); + if (ast->body.script_lines) rb_gc_mark(ast->body.script_lines); } - if (ast->body.script_lines) rb_gc_mark(ast->body.script_lines); } void rb_ast_free(rb_ast_t *ast) { if (ast->node_buffer) { - rb_node_buffer_free(ast->node_buffer); +#ifdef UNIVERSAL_PARSER + rb_parser_config_t *config = ast->node_buffer->config; +#endif + + rb_node_buffer_free(ast, ast->node_buffer); ast->node_buffer = 0; +#ifdef UNIVERSAL_PARSER + config->counter--; + if (config->counter <= 0) { + rb_ruby_parser_config_free(config); + } +#endif } } @@ -1510,3 +470,12 @@ rb_ast_set_tokens(rb_ast_t *ast, VALUE tokens) { RB_OBJ_WRITE(ast, &ast->node_buffer->tokens, tokens); } + +VALUE +rb_node_set_type(NODE *n, enum node_type t) +{ +#if RUBY_DEBUG + rb_ast_node_type_change(n, t); +#endif + return nd_init_type(n, t); +} -- cgit v1.2.3