Merge pull request #1040 from mattn/backtick

xquote operator

3 files changed, 74 insertions, 3 deletions

diff --git a/src/codegen.c b/src/codegen.c
index 7a91d597d..8fb1eebe9 100644
--- a/src/codegen.c
+++ b/src/codegen.c
@@ -2031,6 +2031,26 @@ codegen(codegen_scope *s, node *tree, int val)
     gen_literal_array(s, tree, TRUE, val);
     break;
 
+  case NODE_XSTR:
+    if (val) {
+      char *p = (char*)tree->car;
+      size_t len = (intptr_t)tree->cdr;
+      int ai = mrb_gc_arena_save(s->mrb);
+      int sym = new_sym(s, mrb_intern2(s->mrb, "Kernel", 6));
+      int off = new_lit(s, mrb_str_new(s->mrb, p, len));
+
+      genop(s, MKOP_A(OP_OCLASS, cursp()));
+      genop(s, MKOP_ABx(OP_GETMCNST, cursp(), sym));
+      push();
+      genop(s, MKOP_ABx(OP_STRING, cursp(), off));
+      pop();
+      sym = new_sym(s, mrb_intern2(s->mrb, "`", 1));
+      genop(s, MKOP_ABC(OP_SEND, cursp(), sym, 1));
+      mrb_gc_arena_restore(s->mrb, ai);
+      push();
+    }
+    break;
+
   case NODE_REGX:
     if (val) {
       char *p1 = (char*)tree->car;
diff --git a/src/node.h b/src/node.h
index 4f9db8265..df27c431f 100644
--- a/src/node.h
+++ b/src/node.h
@@ -65,6 +65,8 @@ enum node_type {
     NODE_SYM,
     NODE_STR,
     NODE_DSTR,
+    NODE_XSTR,
+    NODE_DXSTR,
     NODE_REGX,
     NODE_DREGX,
     NODE_DREGX_ONCE,
diff --git a/src/parse.y b/src/parse.y
index 8f1241d33..506e6e279 100644
--- a/src/parse.y
+++ b/src/parse.y
@@ -702,6 +702,20 @@ new_dstr(parser_state *p, node *a)
   return cons((node*)NODE_DSTR, a);
 }
 
+// (:str . (s . len))
+static node*
+new_xstr(parser_state *p, const char *s, int len)
+{
+  return cons((node*)NODE_XSTR, cons((node*)strndup(s, len), (node*)(intptr_t)len));
+}
+
+// (:xstr . a)
+static node*
+new_dxstr(parser_state *p, node *a)
+{
+  return cons((node*)NODE_DXSTR, a);
+}
+
 // (:dsym . a)
 static node*
 new_dsym(parser_state *p, node *a)
@@ -974,12 +988,12 @@ heredoc_end(parser_state *p)
 	keyword__ENCODING__
 
 %token <id>  tIDENTIFIER tFID tGVAR tIVAR tCONSTANT tCVAR tLABEL
-%token <nd>  tINTEGER tFLOAT tCHAR tREGEXP
+%token <nd>  tINTEGER tFLOAT tCHAR tXSTRING tREGEXP
 %token <nd>  tSTRING tSTRING_PART tSTRING_MID
 %token <nd>  tNTH_REF tBACK_REF
 %token <num> tREGEXP_END
 
-%type <nd> singleton string string_rep string_interp regexp
+%type <nd> singleton string string_rep string_interp xstring regexp
 %type <nd> literal numeric cpath symbol
 %type <nd> top_compstmt top_stmts top_stmt
 %type <nd> bodystmt compstmt stmts stmt expr arg primary command command_call method_call
@@ -1028,7 +1042,7 @@ heredoc_end(parser_state *p)
 %token tAMPER             /* & */
 %token tLAMBDA            /* -> */
 %token tSYMBEG tREGEXP_BEG tWORDS_BEG tSYMBOLS_BEG
-%token tSTRING_BEG tSTRING_DVAR tLAMBEG
+%token tSTRING_BEG tXSTRING_BEG tSTRING_DVAR tLAMBEG
 %token <nd> tHEREDOC_BEG  /* <<, <<- */
 %token tHEREDOC_END tLITERAL_DELIM
 
@@ -1936,6 +1950,7 @@ mrhs		: args ',' arg_value
 
 primary		: literal
 		| string
+		| xstring
 		| regexp
 		| heredoc
 		| var_ref
@@ -2585,6 +2600,16 @@ string_interp	: tSTRING_MID
 		    }
 		;
 
+xstring		: tXSTRING_BEG tXSTRING
+		    {
+			$$ = $2;
+		    }
+		| tXSTRING_BEG string_rep tXSTRING
+		    {
+		      $$ = new_dxstr(p, push($2, $3));
+		    }
+		;
+
 regexp		: tREGEXP_BEG tREGEXP
 		    {
 			$$ = $2;
@@ -2985,6 +3010,8 @@ singleton	: var_ref
 			switch ((enum node_type)(int)(intptr_t)$3->car) {
 			case NODE_STR:
 			case NODE_DSTR:
+			case NODE_XSTR:
+			case NODE_DXSTR:
 			case NODE_DREGX:
 			case NODE_MATCH:
 			case NODE_FLOAT:
@@ -3641,6 +3668,11 @@ parse_string(parser_state *p)
   p->lstate = EXPR_END;
   end_strterm(p);
 
+  if (type & STR_FUNC_XQUOTE) {
+    yylval.nd = new_xstr(p, tok(p), toklen(p));
+    return tXSTRING;
+  }
+
   if (type & STR_FUNC_REGEXP) {
     int f = 0;
     int c;
@@ -3988,6 +4020,10 @@ parser_yylex(parser_state *p)
     p->lex_strterm = new_strterm(p, str_squote, '\'', 0);
     return parse_string(p);
 
+  case '`':
+    p->lex_strterm = new_strterm(p, str_xquote, '`', 0);
+    return tXSTRING_BEG;
+
   case '?':
     if (IS_END()) {
       p->lstate = EXPR_VALUE;
@@ -4628,6 +4664,10 @@ parser_yylex(parser_state *p)
 	p->lex_strterm = new_strterm(p, str_sword, term, paren);
 	return tWORDS_BEG;
 
+      case 'x':
+	p->lex_strterm = new_strterm(p, str_xquote, term, paren);
+	return tXSTRING_BEG;
+
       case 'r':
 	p->lex_strterm = new_strterm(p, str_regexp, term, paren);
 	return tREGEXP_BEG;
@@ -5693,6 +5733,15 @@ parser_dump(mrb_state *mrb, node *tree, int offset)
     dump_recur(mrb, tree, offset+1);
     break;
 
+  case NODE_XSTR:
+    printf("NODE_XSTR \"%s\" len %d\n", (char*)tree->car, (int)(intptr_t)tree->cdr);
+    break;
+
+  case NODE_DXSTR:
+    printf("NODE_DXSTR\n");
+    dump_recur(mrb, tree, offset+1);
+    break;
+
   case NODE_REGX:
     printf("NODE_REGX /%s/%s\n", (char*)tree->car, (char*)tree->cdr);
     break;