Feature #4276: Allow use of quotes in symbol syntactic sugar for hashes

• Status: Closed
• Priority: Normal
• Assignee: matz (Yukihiro Matsumoto)
• Target version: 2.6

[tbenst (Tyler Benster)]

Current syntactic sugar allows this:

hash = {Alabama: "AL"}

This feature request is to also allow symbols delimited by quotes (and thus able to contain a whitespace) to use an equivalent syntactic sugar:

hash2 = {"Rhode Island": "RI"}

[nobu (Nobuyoshi Nakada)]

Hi,

At Thu, 13 Jan 2011 10:17:23 +0900,
Tyler Benster wrote in :

Current syntactic sugar allows this:

hash = {Alabama: "AL"}

This feature request is to also allow symbols delimited by quotes (and thus able to contain a whitespace) to use an equivalent syntactic sugar:

hash2 = {"Rhode Island": "RI"}

I've forgotten to post the patch.

diff --git i/parse.y w/parse.y
index 06f96ce..0cacdd5 100644
--- i/parse.y
+++ w/parse.y
@@ -380,6 +380,8 @@ static NODE *ret_args_gen(struct parser_params*,NODE*);
 static NODE *arg_blk_pass(NODE*,NODE*);
 static NODE *new_yield_gen(struct parser_params*,NODE*);
 #define new_yield(node) new_yield_gen(parser, (node))
+static NODE *dsym_node_gen(struct parser_params*,NODE*);
+#define dsym_node(node) dsym_node_gen(parser, (node))

 static NODE *gettable_gen(struct parser_params*,ID);
 #define gettable(id) gettable_gen(parser,(id))
@@ -678,12 +680,12 @@ static void token_info_pop(struct parser_params*, const char *token);
    keyword__FILE__
    keyword__ENCODING__

-%token <id>   tIDENTIFIER tFID tGVAR tIVAR tCONSTANT tCVAR tLABEL
+%token <id>   tIDENTIFIER tFID tGVAR tIVAR tCONSTANT tCVAR tLABEL tLABEL_END
 %token <node> tINTEGER tFLOAT tSTRING_CONTENT tCHAR
 %token <node> tNTH_REF tBACK_REF
 %token <num>  tREGEXP_END

-%type <node> singleton strings string string1 xstring regexp
+%type <node> singleton strings string string1 xstring regexp string_body
 %type <node> string_contents xstring_contents regexp_contents string_content
 %type <node> words qwords word_list qword_list word
 %type <node> literal numeric dsym cpath
@@ -3841,12 +3843,18 @@ string      : tCHAR
            }
        ;

-string1        : tSTRING_BEG string_contents tSTRING_END
+string_body    : tSTRING_BEG string_contents
            {
-           /*%%%*/
            $$ = $2;
+           }
+       ;
+
+string1        : string_body tSTRING_END
+           {
+           /*%%%*/
+           $$ = $1;
            /*%
-           $$ = dispatch1(string_literal, $2);
+           $$ = dispatch1(string_literal, $1);
            %*/
            }
        ;
@@ -4199,26 +4207,7 @@ dsym     : tSYMBEG xstring_contents tSTRING_END
            {
            lex_state = EXPR_END;
            /*%%%*/
-           if (!($$ = $2)) {
-               $$ = NEW_LIT(ID2SYM(rb_intern("")));
-           }
-           else {
-               VALUE lit;
-
-               switch (nd_type($$)) {
-                 case NODE_DSTR:
-               nd_set_type($$, NODE_DSYM);
-               break;
-                 case NODE_STR:
-               lit = $$->nd_lit;
-               $$->nd_lit = ID2SYM(rb_intern_str(lit));
-               nd_set_type($$, NODE_LIT);
-               break;
-                 default:
-               $$ = NEW_NODE(NODE_DSYM, Qnil, 1, NEW_LIST($$));
-               break;
-               }
-           }
+           $$ = dsym_node($2);
            /*%
            $$ = dispatch1(dyna_symbol, $2);
            %*/
@@ -4761,6 +4750,15 @@ assoc        : arg_value tASSOC arg_value
            $$ = dispatch2(assoc_new, $1, $2);
            %*/
            }
+       | string_body tLABEL_END arg_value
+           {
+           /*%%%*/
+           $$ = list_append(NEW_LIST(dsym_node($1)), $3);
+           /*%
+           $$ = dispatch1(dyna_symbol, $1);
+           $$ = dispatch2(assoc_new, $$, $3);
+           %*/
+           }
        ;

 operation  : tIDENTIFIER
@@ -5334,6 +5332,7 @@ rb_parser_compile_file(volatile VALUE vparser, const char *f, VALUE file, int st
 #define STR_FUNC_QWORDS 0x08
 #define STR_FUNC_SYMBOL 0x10
 #define STR_FUNC_INDENT 0x20
+#define STR_FUNC_LABEL  0x40

 enum string_type {
     str_squote = (0),
@@ -5925,6 +5924,8 @@ parser_tokadd_string(struct parser_params *parser,

 #define NEW_STRTERM(func, term, paren) \
    rb_node_newnode(NODE_STRTERM, (func), (term) | ((paren) << (CHAR_BIT * 2)), 0)
+#define IS_LABEL_SUFFIX(n) (peek_n(':',(n)) && !peek_n(':', (n)+1))
+#define MAYBE_LABEL() (IS_LABEL_POSSIBLE() ? STR_FUNC_LABEL : 0)

 static int
 parser_parse_string(struct parser_params *parser, NODE *quote)
@@ -5946,6 +5947,10 @@ parser_parse_string(struct parser_params *parser, NODE *quote)
        quote->nd_func = -1;
        return ' ';
    }
+   if ((func & STR_FUNC_LABEL) && IS_LABEL_SUFFIX(0)) {
+       lex_state = EXPR_BEG;
+       return tLABEL_END;
+   }
    if (!(func & STR_FUNC_REGEXP)) return tSTRING_END;
         set_yylval_num(regx_options());
    return tREGEXP_END;
@@ -6533,7 +6538,6 @@ parser_prepare(struct parser_params *parser)
 #define IS_BEG() (lex_state == EXPR_BEG || lex_state == EXPR_MID || lex_state == EXPR_VALUE || lex_state == EXPR_CLASS)
 #define IS_SPCARG(c) (IS_ARG() && space_seen && !ISSPACE(c))
 #define IS_LABEL_POSSIBLE() ((lex_state == EXPR_BEG && !cmd_state) || IS_ARG())
-#define IS_LABEL_SUFFIX(n) (peek_n(':',(n)) && !peek_n(':', (n)+1))

 #ifndef RIPPER
 #define ambiguous_operator(op, syn) ( \
@@ -6849,7 +6853,7 @@ parser_yylex(struct parser_params *parser)
    return '>';

       case '"':
-   lex_strterm = NEW_STRTERM(str_dquote, '"', 0);
+   lex_strterm = NEW_STRTERM(str_dquote | MAYBE_LABEL(), '"', 0);
    return tSTRING_BEG;

       case '`':
@@ -6868,7 +6872,7 @@ parser_yylex(struct parser_params *parser)
    return tXSTRING_BEG;

       case '\'':
-   lex_strterm = NEW_STRTERM(str_squote, '\'', 0);
+   lex_strterm = NEW_STRTERM(str_squote | MAYBE_LABEL(), '\'', 0);
    return tSTRING_BEG;

       case '?':
@@ -8987,6 +8991,32 @@ new_args_gen(struct parser_params *parser, NODE *m, NODE *o, ID r, NODE *p, ID b
     ruby_sourceline = saved_line;
     return node;
 }
+
+static NODE*
+dsym_node_gen(struct parser_params *parser, NODE *node)
+{
+    if (!node) {
+   node = NEW_LIT(ID2SYM(rb_intern("")));
+    }
+    else {
+   VALUE lit;
+
+   switch (nd_type(node)) {
+     case NODE_DSTR:
+       nd_set_type(node, NODE_DSYM);
+       break;
+     case NODE_STR:
+       lit = node->nd_lit;
+       node->nd_lit = ID2SYM(rb_intern_str(lit));
+       nd_set_type(node, NODE_LIT);
+       break;
+     default:
+       node = NEW_NODE(NODE_DSYM, Qnil, 1, NEW_LIST(node));
+       break;
+   }
+    }
+    return node;
+}
 #endif /* !RIPPER */

 static void

---

Nobu Nakada

[nahi (Hiroshi Nakamura)]

• Description updated (diff)
• Category set to core
• Status changed from Open to Assigned
• Assignee set to matz (Yukihiro Matsumoto)

[mame (Yusuke Endoh)]

• Target version set to 2.6

[bitsweat (Jeremy Daer)]

This would improve a lot of my code that's punctuated with { foo: bar, :'hoge' => piyo }. Switching syntax is a mental and visual interruption.

I hope Ruby 2 supports this quoted-symbol syntax from the first day it's released, so everyone can rely on it. Please consider this patch for preview2!

[drbrain (Eric Hodel)]

Sorry, this feature is too late for 2.0.0

[nobu (Nobuyoshi Nakada)]

One of the reason I didn't introduce this was, a few bundled scripts had compatibility issue when once I tried.

[nobu (Nobuyoshi Nakada)]

• Has duplicate Feature #9047: Alternate hash key syntax for symbols added

[nobu (Nobuyoshi Nakada)]

• Description updated (diff)

[nobu (Nobuyoshi Nakada)]

Updated with the test in [Feature #4935]: https://github.com/ruby/ruby/pull/684

[rosenfeld (Rodrigo Rosenfeld Rosas)]

Great, from the test it seems to allow interpolation. Could you please confirm I understood it correctly?

[matz (Yukihiro Matsumoto)]

• Status changed from Assigned to Feedback

I am not against the idea, but I want to make sure that key will be symbol, since some may expect

{"foo bar": 12}

to be a hash with string key, especially who is familiar with JSON.

Matz.

[nobu (Nobuyoshi Nakada)]

Rodrigo Rosenfeld Rosas wrote:

Great, from the test it seems to allow interpolation. Could you please confirm I understood it correctly?

Yes, it is allowed inside double quotes, but not single quotes.
Other terminators (e.g., %(...)) cause syntax errors.

[Ajedi32 (Andrew M)]

Yukihiro Matsumoto wrote:

I am not against the idea, but I want to make sure that key will be symbol,

Yes, I believe that was the intention. E.g.

# Currently, Hashes have syntactic sugar which allows:
{:key => "Value"} == {key: "Value"} #=> true

# This feature request is to allow:
{:"some key" => "Value"} == {"some key": "Value"} #=> true

[nobu (Nobuyoshi Nakada)]

OK, I commit the previous patch.
Yell if you don't like it.

[nobu (Nobuyoshi Nakada)]

• Status changed from Feedback to Closed
• % Done changed from 0 to 100

Applied in changeset r47649.

---

parse.y: quoted ID key

• parse.y (assoc): allow quoted ID as a key of a hash literal. [Feature #4276]

[nobu (Nobuyoshi Nakada)]

Just a note: last evening at RubyKaigi 2014, talked to matz about this issue, and got his approval to introduce this to see if anyone rants.

[etienne (Étienne Barrié)]

Hi, any reason it doesn't work with inner hashes or arrays?

{ foo: {} } # => {:foo=>{}}
{ 'foo': {} }
# SyntaxError: (irb):2: syntax error, unexpected '{'
{ 'foo': [] }
SyntaxError: (irb):1: syntax error, unexpected '['

[nobu (Nobuyoshi Nakada)]

• Related to Bug #10653: do-end block in ternary operator is syntax error added