parser 1.0.1 → 1.1.0

data/lib/parser/lexer.rl

@@ -158,14 +158,15 @@ class Parser::Lexer
 
   LEX_STATES = {
     :line_begin  => lex_en_line_begin,
-    :expr_beg    => lex_en_expr_beg,
-    :expr_value  => lex_en_expr_value,
-    :expr_mid    => lex_en_expr_mid,
     :expr_dot    => lex_en_expr_dot,
     :expr_fname  => lex_en_expr_fname,
-    :expr_end    => lex_en_expr_end,
+    :expr_value  => lex_en_expr_value,
+    :expr_beg    => lex_en_expr_beg,
+    :expr_mid    => lex_en_expr_mid,
     :expr_arg    => lex_en_expr_arg,
+    :expr_end    => lex_en_expr_end,
     :expr_endarg => lex_en_expr_endarg,
+    :expr_endfn  => lex_en_expr_endfn,
   }
 
   def state
@@ -371,8 +372,8 @@ class Parser::Lexer
   c_line     = c_any - c_nl;
 
   c_unicode  = c_any - 0x00..0x7f;
+  c_upper    = [A-Z];
   c_lower    = [a-z_]  | c_unicode;
-  c_upper    = [A-Z]   | c_unicode;
   c_alpha    = c_lower | c_upper;
   c_alnum    = c_alpha | [0-9];
 
@@ -451,7 +452,7 @@ class Parser::Lexer
                         keyword_with_end   | keyword_with_arg |
                         keyword_with_fname | keyword_modifier ;
 
-  constant       = [A-Z] c_alnum*;
+  constant       = c_upper c_alnum*;
   bareword       = c_alpha c_alnum*;
 
   call_or_var    = c_lower c_alnum*;
@@ -984,30 +985,33 @@ class Parser::Lexer
   # statements `alias $a $b`. Symbols are returned verbatim; this
   # is used in `alias :a :"b#{foo}"` and `undef :a`.
   #
-  # Transitions to `expr_end` afterwards.
+  # Transitions to `expr_endfn` afterwards.
   #
   expr_fname := |*
       keyword
       => { emit(KEYWORDS[tok]);
-           fnext expr_end; fbreak; };
+           fnext expr_endfn; fbreak; };
 
       bareword [?=!]?
       => { emit(:tIDENTIFIER)
-           fnext expr_end; fbreak; };
+           fnext expr_endfn; fbreak; };
 
+      global_var
+      => { p = @ts - 1
+           fcall expr_variable; };
+
+      # If the handling was to be delegated to expr_end,
+      # these cases would transition to something else than
+      # expr_end, which is undesirable.
       operator_fname      |
       operator_arithmetic |
       operator_rest
       => { emit_table(PUNCTUATION)
-           fnext expr_end; fbreak; };
+           fnext expr_endfn; fbreak; };
 
       ':'
       => { fhold; fgoto expr_beg; };
 
-      global_var
-      => { p = @ts - 1
-           fcall expr_variable; };
-
       # TODO whitespace rule
       c_space;
       e_heredoc_nl;
@@ -1018,6 +1022,25 @@ class Parser::Lexer
       c_eof => do_eof;
   *|;
 
+  # After literal function name in definition. Behaves like `expr_end`,
+  # but allows a tLABEL.
+  #
+  # Transitions to `expr_end` afterwards.
+  #
+  expr_endfn := |*
+      bareword ':'
+      => { emit(:tLABEL, tok(@ts, @te - 1))
+           fnext expr_end; fbreak; };
+
+      # TODO whitespace rule
+      c_space;
+
+      c_any
+      => { fhold; fgoto expr_end; };
+
+      c_eof => do_eof;
+  *|;
+
   # Literal function name in method call (e.g. `a.class`).
   #
   # Transitions to `expr_arg` afterwards.
@@ -1035,6 +1058,7 @@ class Parser::Lexer
       => { emit(:tFID, tok(@ts, tm), @ts, tm)
            fnext expr_arg; p = tm - 1; fbreak; };
 
+      # See the comment in `expr_fname`.
       operator_fname      |
       operator_arithmetic |
       operator_rest