Merge pull request #5122 from shuujii/change-name-and-usage-of-presym-macros

Change name and usage of presym macros

Rakefile

@@ -148,8 +148,21 @@ op_table = {
   "||" => "oror",
   "~" => "neg",
 }
+macro_to_symbol = {
+#      Macro               Symbol
+# [prefix, suffix] => [prefix, suffix]
+  ["CV"  , ""    ] => ["@@"  , ""    ],
+  ["IV"  , ""    ] => ["@"   , ""    ],
+  [""    , "_B"  ] => [""    , "!"   ],
+  [""    , "_Q"  ] => [""    , "?"   ],
+  [""    , "_E"  ] => [""    , "="   ],
+  [""    , ""    ] => [""    , ""    ],
+}
 
 file presym_file => cfiles+rbfiles+psfiles+[__FILE__] do
+  prefix_re = Regexp.union(*macro_to_symbol.keys.uniq.map(&:first))
+  suffix_re = Regexp.union(*macro_to_symbol.keys.uniq.map(&:last))
+  macro_re = /MRB_(#{prefix_re})SYM(#{suffix_re})\((\w+)\)/o
   csymbols = cfiles.map do |f|
     src = File.read(f)
     src.gsub!(/\/\/.+(\n|$)/, "\n")
@@ -161,10 +174,9 @@ file presym_file => cfiles+rbfiles+psfiles+[__FILE__] do
      src.scan(/mrb_define_module_function\([^\n"]*"([^\n"]*)"/),
      src.scan(/mrb_define_const\([^\n"]*"([^\n"]*)"/),
      src.scan(/mrb_define_global_const\([^\n"]*"([^\n"]*)"/),
-     src.scan(/MRB_SYM\((\w+)\)/),
-     src.scan(/MRB_QSYM\((\w+)\)/).map{|x,|
-       x.sub!(/_p$/, "?") || x.sub!(/_b$/, "!") || x.sub!(/_e$/, "=") || x.sub!(/^0_/, "@")  || x.sub!(/^00_/, "@@")
-     }.compact]
+     src.scan(macro_re).map{|prefix, suffix, name|
+       macro_to_symbol[[prefix, suffix]] * name
+     }]
   end
   rbsymbols = rbfiles.map do |f|
     src = File.read(f)
@@ -197,34 +209,20 @@ end
 task presym_inc do
   presyms = File.readlines(presym_file)
   presyms.each{|x| x.chomp!}
+  symbol_to_macro = macro_to_symbol.invert
+  prefix_re = Regexp.union(*symbol_to_macro.keys.uniq.map(&:first))
+  suffix_re = Regexp.union(*symbol_to_macro.keys.uniq.map(&:last))
+  macro_re = /\A(#{prefix_re})?([\w&&\D]\w*)(#{suffix_re})?\z/o
   File.open(presym_inc, "w") do |f|
-    f.print "/* MRB_PRESYM_CSYM(sym, num) - symbol which is valid C id name */\n"
-    f.print "/* MRB_PRESYM_QSYM(name, sym, num) - symbol with alias name */\n"
-    f.print "/* MRB_PRESYM_SYM(name, num) - symbol which is not valid C id */\n"
-    presyms.each.with_index do |sym,i|
-      if sym.bytes.detect{|x|x>0x80} || /\A\$/ =~ sym
-        f.print "MRB_PRESYM_SYM(\"#{sym}\", #{i+1})\n"
-      elsif /\A\w+\Z/ =~ sym
-        f.print "MRB_PRESYM_CSYM(#{sym}, #{i+1})\n"
-      elsif op_table.key?(sym)
-        f.print "MRB_PRESYM_QSYM(\"#{sym}\", #{op_table[sym]}, #{i+1})\n"
-      elsif /\A[A-Za-z_]\w*\?\Z/ =~ sym
-        s = sym.dup; s[-1] = "_p"
-        f.print "MRB_PRESYM_QSYM(\"#{sym}\", #{s}, #{i+1})\n"
-      elsif /\A[A-Za-z_]\w*\!\Z/ =~ sym
-        s = sym.dup; s[-1] = "_b"
-        f.print "MRB_PRESYM_QSYM(\"#{sym}\", #{s}, #{i+1})\n"
-      elsif /\A[A-Za-z_]\w*\=\Z/ =~ sym
-        s = sym.dup; s[-1] = "_e"
-        f.print "MRB_PRESYM_QSYM(\"#{sym}\", #{s}, #{i+1})\n"
-      elsif /\A@@/ =~ sym
-        s = sym.dup; s[0,2] = "00_"
-        f.print "MRB_PRESYM_QSYM(\"#{sym}\", #{s}, #{i+1})\n"
-      elsif /\A@/ =~ sym
-        s = sym.dup; s[0] = "0_"
-        f.print "MRB_PRESYM_QSYM(\"#{sym}\", #{s}, #{i+1})\n"
-      else
-        f.print "MRB_PRESYM_SYM(\"#{sym}\", #{i+1})\n"
+    f.puts "/* MRB_PRESYM_NAMED(lit, num, type, name) */"
+    f.puts "/* MRB_PRESYM_UNNAMED(lit, num) */"
+    presyms.each.with_index(1) do |sym, num|
+      if macro_re =~ sym && (affixes = symbol_to_macro[[$1, $3]])
+        f.puts %|MRB_PRESYM_NAMED("#{sym}", #{num}, #{affixes * 'SYM'}, #{$2})|
+      elsif name = op_table[sym]
+        f.puts %|MRB_PRESYM_NAMED("#{sym}", #{num}, OPSYM, #{name})|
+      elsif
+        f.puts %|MRB_PRESYM_UNNAMED("#{sym}", #{num})|
       end
     end
     f.print "#define MRB_PRESYM_MAX #{presyms.size}"

include/mruby.h

@@ -1080,7 +1080,8 @@ MRB_API mrb_value mrb_funcall_argv(mrb_state *mrb, mrb_value val, mrb_sym name,
  */
 MRB_API mrb_value mrb_funcall_with_block(mrb_state *mrb, mrb_value val, mrb_sym name, mrb_int argc, const mrb_value *argv, mrb_value block);
 /**
- * Create a symbol from C string. But usually it's better to use MRB_SYM(sym) and MRB_QSYM(qsym).
+ * Create a symbol from C string. But usually it's better to use MRB_SYM,
+ * MRB_OPSYM, MRB_CVSYM, MRB_IVSYM, MRB_SYM_B, MRB_SYM_Q, MRB_SYM_E macros.
  *
  * Example:
  *
@@ -1090,7 +1091,7 @@ MRB_API mrb_value mrb_funcall_with_block(mrb_state *mrb, mrb_value val, mrb_sym
  *     // C style:
  *     mrb_sym sym1 = mrb_intern_lit(mrb, "pizza"); //  => :pizza
  *     mrb_sym sym2 = MRB_SYM(pizza);               //  => :pizza
- *     mrb_sym sym3 = MRB_SYM(pizza_p);             //  => :pizza?
+ *     mrb_sym sym3 = MRB_SYM_Q(pizza);             //  => :pizza?
  *
  * @param mrb The current mruby state.
  * @param str The string to be symbolized

include/mruby/presym.h

@@ -8,18 +8,37 @@
 #define MRUBY_PRESYM_H
 
 #undef MRB_PRESYM_MAX
-#define MRB_PRESYM_CSYM(sym, num) MRB_PRESYM__##sym = (num<<1),
-#define MRB_PRESYM_QSYM(str, sym, num) MRB_PRESYM_q_##sym = (num<<1),
-#define MRB_PRESYM_SYM(sym, num) 
+#define MRB_PRESYM_NAMED(lit, num, type, name) MRB_##type##__##name = (num<<1),
+#define MRB_PRESYM_UNNAMED(lit, num)
 
 enum mruby_presym {
 #include <../build/presym.inc>
 };
 
-#undef MRB_PRESYM_CSYM
-#undef MRB_PRESYM_QSYM
-#undef MRB_PRESYM_SYM
+#undef MRB_PRESYM_NAMED
+#undef MRB_PRESYM_UNNAMED
+
+/*
+ * For `MRB_OPSYM`, specify the names corresponding to operators (refer to
+ * `op_table` in `Rakefile` for the names that can be specified for it).
+ * Other than that, describe only word characters excluding leading and
+ * ending punctuations.
+ *
+ * Example:
+ *   MRB_OPSYM(and)  //=> &
+ *   MRB_CVSYM(foo)  //=> @@foo
+ *   MRB_IVSYM(foo)  //=> @foo
+ *   MRB_SYM_B(foo)  //=> foo!
+ *   MRB_SYM_Q(foo)  //=> foo?
+ *   MRB_SYM_E(foo)  //=> foo=
+ *   MRB_SYM(foo)    //=> foo
+ */
+#define MRB_OPSYM(name) MRB_OPSYM__##name  /* Operator */
+#define MRB_CVSYM(name) MRB_CVSYM__##name  /* Class Variable */
+#define MRB_IVSYM(name) MRB_IVSYM__##name  /* Instance Variable */
+#define MRB_SYM_B(name) MRB_SYM_B__##name  /* Method with Bang */
+#define MRB_SYM_Q(name) MRB_SYM_Q__##name  /* Method with Question mark */
+#define MRB_SYM_E(name) MRB_SYM_E__##name  /* Method with Equal */
+#define MRB_SYM(name)   MRB_SYM__##name    /* Word characters */
 
-#define MRB_SYM(sym) MRB_PRESYM__##sym
-#define MRB_QSYM(sym) MRB_PRESYM_q_##sym
 #endif  /* MRUBY_PRESYM_H */

mrbgems/mruby-compiler/core/codegen.c

@@ -1649,7 +1649,7 @@ codegen(codegen_scope *s, node *tree, int val)
         {
           node *n = tree->car->cdr;
           mrb_sym mid = nsym(n->cdr->car);
-          mrb_sym mnil = MRB_QSYM(nil_p);
+          mrb_sym mnil = MRB_SYM_Q(nil);
           if (mid == mnil && n->cdr->cdr->car == NULL) {
             nil_p = TRUE;
             codegen(s, n->car, VAL);
@@ -1788,7 +1788,7 @@ codegen(codegen_scope *s, node *tree, int val)
               genop_3(s, OP_SEND, cursp(), new_sym(s, MRB_SYM(__case_eqq)), 1);
             }
             else {
-              genop_3(s, OP_SEND, cursp(), new_sym(s, MRB_QSYM(eqq)), 1);
+              genop_3(s, OP_SEND, cursp(), new_sym(s, MRB_OPSYM(eqq)), 1);
             }
           }
           else {
@@ -2576,7 +2576,7 @@ codegen(codegen_scope *s, node *tree, int val)
 
       default:
         if (val) {
-          int sym = new_sym(s, MRB_QSYM(minus));
+          int sym = new_sym(s, MRB_OPSYM(minus));
           codegen(s, tree, VAL);
           pop();
           genop_3(s, OP_SEND, cursp(), sym, 0);
@@ -2668,7 +2668,7 @@ codegen(codegen_scope *s, node *tree, int val)
       }
       push();                   /* for block */
       pop_n(3);
-      sym = new_sym(s, MRB_QSYM(tick)); /* ` */
+      sym = new_sym(s, MRB_OPSYM(tick)); /* ` */
       genop_3(s, OP_SEND, cursp(), sym, 1);
       if (val) push();
       mrb_gc_arena_restore(s->mrb, ai);
@@ -2688,7 +2688,7 @@ codegen(codegen_scope *s, node *tree, int val)
       genop_bs(s, OP_STRING, cursp(), off);
       push(); push();
       pop_n(3);
-      sym = new_sym(s, MRB_QSYM(tick)); /* ` */
+      sym = new_sym(s, MRB_OPSYM(tick)); /* ` */
       genop_3(s, OP_SEND, cursp(), sym, 1);
       if (val) push();
       mrb_gc_arena_restore(s->mrb, ai);

mrbgems/mruby-io/src/io.c

@@ -113,7 +113,7 @@ io_set_process_status(mrb_state *mrb, pid_t pid, int status)
   } else {
     v = mrb_fixnum_value(WEXITSTATUS(status));
   }
-  mrb_gv_set(mrb, mrb_intern_cstr(mrb, "$?"), v);
+  mrb_gv_set(mrb, mrb_intern_lit(mrb, "$?"), v);
 }
 
 static int

mrbgems/mruby-kernel-ext/src/kernel.c

@@ -57,7 +57,7 @@ mrb_f_caller(mrb_state *mrb, mrb_value self)
     return mrb_ary_new(mrb);
   }
 
-  return mrb_funcall_id(mrb, bt, MRB_QSYM(aref), 2, mrb_fixnum_value(lev), mrb_fixnum_value(n));
+  return mrb_funcall_id(mrb, bt, MRB_OPSYM(aref), 2, mrb_fixnum_value(lev), mrb_fixnum_value(n));
 }
 
 /*

mrbgems/mruby-method/src/method.c

@@ -324,10 +324,10 @@ mrb_search_method_owner(mrb_state *mrb, struct RClass *c, mrb_value obj, mrb_sym
     if (unbound) {
       goto name_error;
     }
-    if (!mrb_respond_to(mrb, obj, MRB_QSYM(respond_to_missing_p))) {
+    if (!mrb_respond_to(mrb, obj, MRB_SYM_Q(respond_to_missing))) {
       goto name_error;
     }
-    ret = mrb_funcall_id(mrb, obj, MRB_QSYM(respond_to_missing_p), 2, mrb_symbol_value(name), mrb_true_value());
+    ret = mrb_funcall_id(mrb, obj, MRB_SYM_Q(respond_to_missing), 2, mrb_symbol_value(name), mrb_true_value());
     if (!mrb_test(ret)) {
       goto name_error;
     }

mrbgems/mruby-socket/src/socket.c

@@ -200,7 +200,7 @@ mrb_addrinfo_getnameinfo(mrb_state *mrb, mrb_value self)
   host = mrb_str_buf_new(mrb, NI_MAXHOST);
   serv = mrb_str_buf_new(mrb, NI_MAXSERV);
 
-  sastr = mrb_iv_get(mrb, self, MRB_QSYM(0_sockaddr));
+  sastr = mrb_iv_get(mrb, self, MRB_IVSYM(sockaddr));
   if (!mrb_string_p(sastr)) {
     mrb_raise(mrb, E_SOCKET_ERROR, "invalid sockaddr");
   }
@@ -222,7 +222,7 @@ mrb_addrinfo_unix_path(mrb_state *mrb, mrb_value self)
 {
   mrb_value sastr;
 
-  sastr = mrb_iv_get(mrb, self, MRB_QSYM(0_sockaddr));
+  sastr = mrb_iv_get(mrb, self, MRB_IVSYM(sockaddr));
   if (((struct sockaddr *)RSTRING_PTR(sastr))->sa_family != AF_UNIX)
     mrb_raise(mrb, E_SOCKET_ERROR, "need AF_UNIX address");
   if (RSTRING_LEN(sastr) < (mrb_int)offsetof(struct sockaddr_un, sun_path) + 1) {

src/array.c

@@ -1314,7 +1314,7 @@ static const mrb_code each_iseq[] = {
 static const mrb_sym each_syms[] = {
   MRB_SYM(each),
   MRB_SYM(to_enum),
-  MRB_QSYM(aref),
+  MRB_OPSYM(aref),
   MRB_SYM(call),
   MRB_SYM(length),
 };

src/kernel.c

@@ -564,7 +564,7 @@ obj_respond_to(mrb_state *mrb, mrb_value self)
   mrb_get_args(mrb, "n|b", &id, &priv);
   respond_to_p = basic_obj_respond_to(mrb, self, id, !priv);
   if (!respond_to_p) {
-    rtm_id = MRB_QSYM(respond_to_missing_p);
+    rtm_id = MRB_SYM_Q(respond_to_missing);
     if (basic_obj_respond_to(mrb, self, rtm_id, !priv)) {
       mrb_value args[2], v;
       args[0] = mrb_symbol_value(id);
@@ -581,7 +581,7 @@ mrb_obj_ceqq(mrb_state *mrb, mrb_value self)
 {
   mrb_value v = mrb_get_arg1(mrb);
   mrb_int i, len;
-  mrb_sym eqq = MRB_QSYM(eqq);
+  mrb_sym eqq = MRB_OPSYM(eqq);
   mrb_value ary;
 
   if (mrb_array_p(self)) {

src/object.c

@@ -58,7 +58,7 @@ mrb_equal(mrb_state *mrb, mrb_value obj1, mrb_value obj2)
       return TRUE;
   }
 #endif
-  result = mrb_funcall_id(mrb, obj1, MRB_QSYM(eq), 1, obj2);
+  result = mrb_funcall_id(mrb, obj1, MRB_OPSYM(eq), 1, obj2);
   if (mrb_test(result)) return TRUE;
   return FALSE;
 }
@@ -660,5 +660,5 @@ MRB_API mrb_bool
 mrb_eql(mrb_state *mrb, mrb_value obj1, mrb_value obj2)
 {
   if (mrb_obj_eq(mrb, obj1, obj2)) return TRUE;
-  return mrb_test(mrb_funcall_id(mrb, obj1, MRB_QSYM(eql_p), 1, obj2));
+  return mrb_test(mrb_funcall_id(mrb, obj1, MRB_SYM_Q(eql), 1, obj2));
 }

src/proc.c

@@ -315,7 +315,7 @@ mrb_init_proc(mrb_state *mrb)
   p = mrb_proc_new(mrb, &call_irep);
   MRB_METHOD_FROM_PROC(m, p);
   mrb_define_method_raw(mrb, mrb->proc_class, MRB_SYM(call), m);
-  mrb_define_method_raw(mrb, mrb->proc_class, MRB_QSYM(aref), m);
+  mrb_define_method_raw(mrb, mrb->proc_class, MRB_OPSYM(aref), m);
 
   mrb_define_class_method(mrb, mrb->kernel_module, "lambda", proc_lambda, MRB_ARGS_NONE()|MRB_ARGS_BLOCK()); /* 15.3.1.2.6  */
   mrb_define_method(mrb, mrb->kernel_module,       "lambda", proc_lambda, MRB_ARGS_NONE()|MRB_ARGS_BLOCK()); /* 15.3.1.3.27 */

src/symbol.c

@@ -13,12 +13,8 @@
 #include <mruby/class.h>
 
 #undef MRB_PRESYM_MAX
-#undef MRB_PRESYM_CSYM
-#undef MRB_PRESYM_QSYM
-#undef MRB_PRESYM_SYM
-#define MRB_PRESYM_CSYM(sym, num) {#sym,sizeof(#sym)-1},
-#define MRB_PRESYM_QSYM(str, name, num) {str,sizeof(str)-1},
-#define MRB_PRESYM_SYM(str, num) {str,sizeof(str)-1},
+#define MRB_PRESYM_NAMED(lit, num, type, name) {lit, sizeof(lit)-1},
+#define MRB_PRESYM_UNNAMED(lit, num) {lit, sizeof(lit)-1},
 
 static const struct {
   const char *name;

src/vm.c

@@ -2261,7 +2261,7 @@ RETRY_TRY_BLOCK:
     OP_MATH_CASE_STRING_##op_name();                                        \
     default:                                                                \
       c = 1;                                                                \
-      mid = MRB_QSYM(op_name);                                              \
+      mid = MRB_OPSYM(op_name);                                              \
       goto L_SEND_SYM;                                                      \
   }                                                                         \
   NEXT;
@@ -2347,7 +2347,7 @@ RETRY_TRY_BLOCK:
 #endif
       default:
         c = 1;
-        mid = MRB_QSYM(div);
+        mid = MRB_OPSYM(div);
         goto L_SEND_SYM;
       }
 
@@ -2366,7 +2366,7 @@ RETRY_TRY_BLOCK:
     default:                                                                \
       SET_INT_VALUE(mrb,regs[a+1], b);                                      \
       c = 1;                                                                \
-      mid = MRB_QSYM(op_name);                                              \
+      mid = MRB_OPSYM(op_name);                                              \
       goto L_SEND_SYM;                                                      \
   }                                                                         \
   NEXT;
@@ -2412,7 +2412,7 @@ RETRY_TRY_BLOCK:
     break;\
   default:\
     c = 1;\
-    mid = MRB_QSYM(sym);\
+    mid = MRB_OPSYM(sym);\
     goto L_SEND_SYM;\
   }\
   if (result) {\
@@ -2441,7 +2441,7 @@ RETRY_TRY_BLOCK:
     break;\
   default:\
     c = 1;\
-    mid = MRB_QSYM(sym);\
+    mid = MRB_OPSYM(sym);\
     goto L_SEND_SYM;\
   }\
   if (result) {\