1 files changed, 627 insertions, 0 deletions
diff --git a/test/ruby/ruby.lm b/test/ruby/ruby.lm
new file mode 100644
index 00000000..b9505515
--- /dev/null
+++ b/test/ruby/ruby.lm
@@ -0,0 +1,627 @@
+#
+# Grammar
+#
+
+# The items in this scanner may have newline in front of them.
+lex start
+{
+	# Reserved Words.
+	literal '__LINE__', '__FILE__', '__ENCODING__', 'BEGIN', 'END', 'alias',
+		'and', 'begin', 'break', 'case', 'class', 'def', 'defined?', 'do',
+		'else', 'elsif', 'end', 'ensure', 'false', 'for', 'in', 'module',
+		'next', 'nil', 'not', 'or', 'redo', 'rescue', 'retry', 'return',
+		'self', 'super', 'then', 'true', 'undef', 'when', 'yield', 'if',
+		'unless', 'while', 'until'
+
+	token tNTH_REF /'$' [0-9]+/
+	token tBACK_REF /'$' ( '&' | '`' | '\'' | '+' ) /
+
+	literal ')', ',', ']'
+	literal '{', '}', ':'
+	literal '.', '::'
+	literal '->'
+	
+	# Unary operators.
+	literal '!', '~'
+	token tUPLUS /'+'/
+	token tUMINUS /'-'/
+
+	token tLBRACK /'['/
+	token tLPAREN /'('/
+	token tSTAR /'*'/
+	token tBAR /'|'/
+	token tAMPER /'&'/
+
+	token tIDENTIFIER /[a-z][a-zA-Z_]*/
+	token tFID /[a-z][a-zA-Z_]* ('!'|'?')/
+	token tCONSTANT /[A-Z][a-zA-Z_]*/
+	token tGVAR /'$' [a-zA-Z_]+/
+	token tIVAR /'@' [a-zA-Z_]+/
+	token tCVAR /'@@' [a-zA-Z_]+/
+
+	token tINTEGER /[0-9]+/
+	token tFLOAT /[0-9]+ '.' [0-9]+/
+
+	token tDSTRING_BEG /'"'/
+	token tSSTRING_BEG /'\''/
+	token tXSTRING_BEG /'`'/
+
+	ignore /[ \t\n]+/
+	ignore comment /'#' [^\n]* '\n'/
+}
+
+# These items cannot appear at the beginning of a line (except maybe the first).
+lex expr_cont_ops
+{
+	ignore /[\t ]+/
+
+	literal '+', '-', '*', '**', '/', '%', '^'
+	literal '|', '&', '||', '&&'
+	literal '[', '('
+	literal '='
+	literal '<<', '>>'
+	literal '?'
+	literal '<=>'
+	literal '=>'
+	literal '[]', '[]='
+	literal '=~', '!~'
+	literal '<', '>', '>=', '<='
+	literal '!=', '==', '==='
+	literal '..', '...'
+}
+lex terms
+{
+	ignore /[\t ]+/
+	ignore /'#' [^\n]*/
+	literal ';'
+	literal '\n'
+}
+
+
+lex dstring_contents
+{
+	token dstring_contents /[^"]+/
+	token tDSTRING_END /'"'/
+}
+
+lex sstring_contents
+{
+	token sstring_contents /[^']+/
+	token tSSTRING_END /'\''/
+}
+
+lex xstring_contents
+{
+	token xstring_contents /[^`]+/
+	token tXSTRING_END /'`'/
+}
+
+def ruby
+	[compstmt]
+
+def compstmt
+	[stmts opt_terms]
+
+def bodystmt
+	[compstmt opt_rescue opt_else opt_ensure]
+
+def opt_rescue
+#	['rescue' exc_list exc_var then compstmt opt_rescue] |
+	[]
+
+def then
+	[term]
+|	['then']
+|	[term 'then']
+
+def do
+	[term]
+|	['do']
+
+def if_tail
+	[opt_else]
+|	['elsif' expr_value then compstmt if_tail]
+
+def opt_else
+	['else' compstmt]
+|	[]
+
+def opt_ensure
+	['ensure' compstmt]
+|	[]
+
+def stmts
+	[stmts terms stmt]
+|	[stmt]
+|	[]
+
+def opt_terms
+	[terms]
+|	[]
+
+def terms
+	[term]
+|	[terms ';']
+
+def term
+	[';']
+|	['\n']
+
+def stmt
+	['alias' fitem fitem]
+|	['undef' undef_list]
+|	[stmt 'if' expr_value]
+|	[stmt 'unless' expr_value]
+|	[stmt 'while' expr_value]
+|	[stmt 'until' expr_value]
+|	[stmt 'rescue' stmt]
+|	['BEGIN' '{' compstmt '}']
+|	['END' '{' compstmt '}']
+|	[lhs '=' mrhs]
+|	[mlhs '=' arg_value]
+|	[mlhs '=' mrhs]
+|	[expr]
+
+def mlhs
+	[mlhs_basic]
+|	[tLPAREN mlhs ')']
+
+def mlhs_basic
+	[mlhs_head]
+
+def mlhs_head
+	[mlhs_item ',' mlhs_head]
+|	[mlhs_item]
+
+def mlhs_item
+	[variable]
+|	['*' mlhs_item]
+|	['*']
+|	[primary_value '[' opt_call_args ']']
+|	[primary_value '.' tIDENTIFIER]
+|	[primary_value '.' tCONSTANT]
+|	[primary_value '::' tIDENTIFIER]
+|	[primary_value '::' tCONSTANT]
+|	['::' tCONSTANT]
+|	[backref]
+|	[tLPAREN mlhs ')']
+
+def lhs
+	[variable]
+|	[primary_value '[' opt_call_args ']']
+|	[primary_value '.' tIDENTIFIER]
+|	[primary_value '.' tCONSTANT]
+|	[primary_value '::' tIDENTIFIER]
+|	[primary_value '::' tCONSTANT]
+|	['::' tCONSTANT]
+|	[backref]
+
+def mrhs
+	[args ',' arg_value]
+|	[args ',' '*' arg_value]
+|	['*' arg_value]
+
+def expr
+	[expr 'and' expr]
+|	[expr 'or' expr]
+|	['not' expr]
+|	[arg]
+
+def expr_value
+	[expr]
+
+def opt_brace_block
+	[brace_block]
+|	[]
+
+def block_param_def
+	[tBAR opt_bv_decl tBAR]
+|	[tBAR block_param opt_bv_decl tBAR]
+
+def block_param
+	[block_arg_list]
+|	[]
+
+def block_arg_list
+	[block_arg_list ',' block_arg_item]
+|	[block_arg_item]
+
+def block_arg_item
+	[f_norm_arg]
+|	[f_rest_arg]
+|	[f_block_arg]
+|	['(' f_args ')']
+
+def opt_bv_decl
+	[';' bv_decls]
+|	[]
+
+def bv_decls
+	[bvar]
+|	[bv_decls ',' bvar]
+
+def bvar
+	[tIDENTIFIER]
+
+def opt_block_param
+	[block_param_def]
+|	[]
+
+def operation
+	[tIDENTIFIER]
+|	[tCONSTANT]
+|	[tFID]
+
+def operation2
+	[tIDENTIFIER]
+|	[tCONSTANT]
+|	[tFID]
+|	[op]
+
+def operation3
+	[tIDENTIFIER]
+|	[tFID]
+|	[op]
+
+def op
+	['|'] | ['^'] | ['&'] | ['<=>'] | ['=='] | ['==='] | ['=~'] | ['!~'] |
+	['>'] | ['>='] | ['<'] | ['<='] | ['!='] | ['<<'] | ['>>'] | ['+'] |
+	['-'] | ['*'] | ['/'] | ['%'] | ['**'] | ['!'] | ['~'] | ['[]'] | ['[]='] |
+	[tXSTRING_BEG]
+
+def opt_call_args
+	[call_args]
+|	[]
+
+def call_args
+	[args opt_block_arg]
+|	[assocs opt_block_arg]
+|	[args ',' assocs opt_block_arg]
+|	[block_arg]
+
+def args
+	[arg_value]
+|	['*' arg_value]
+|	[args ',' arg_value]
+|	[args ',' '*' arg_value]
+
+def arg_value
+	[arg]
+
+def opt_block_arg	
+	[',' block_arg]
+|	[]
+
+def block_arg
+	[tAMPER arg_value]
+
+def arg 
+	['defined?' arg]
+|	[arg_assign]
+
+def arg_assign
+	[lhs '=' arg_assign]
+|	[lhs '=' arg_assign 'rescue' arg]
+|	[arg_sel]
+
+def arg_sel
+	[arg_dot '?' arg_sel ':' arg_sel]
+|	[arg_dot]
+
+def arg_dot
+	[arg_logical '..' arg_dot]
+|	[arg_logical '...' arg_dot]
+|	[arg_logical]
+
+def arg_logical
+	[arg_eq '&&' arg_logical]
+|	[arg_eq '||' arg_logical]
+|	[arg_eq]
+
+def arg_eq
+	[arg_cmp '<=>' arg_eq]
+|	[arg_cmp '==' arg_eq]
+|	[arg_cmp '===' arg_eq]
+|	[arg_cmp '!=' arg_eq]
+|	[arg_cmp '=~' arg_eq]
+|	[arg_cmp '!~' arg_eq]
+|	[arg_cmp]
+
+def arg_cmp
+	[arg_bitor '>=' arg_cmp]
+|	[arg_bitor '<=' arg_cmp]
+|	[arg_bitor '>' arg_cmp]
+|	[arg_bitor '<' arg_cmp]
+|	[arg_bitor]
+
+def arg_bitor
+	[arg_bitand '|' arg_bitor]
+|	[arg_bitand '^' arg_bitor]
+|	[arg_bitand]
+
+def arg_bitand
+	[arg_shift '&' arg_bitand]
+|	[arg_shift]
+
+def arg_shift
+	[arg_add '<<' arg_shift]
+|	[arg_add '>>' arg_shift]
+|	[arg_add]
+
+def arg_add
+	[arg_mult '+' arg_add]
+|	[arg_mult '-' arg_add]
+|	[arg_mult]
+
+def arg_mult
+	[arg_pow '*' arg_mult]
+|	[arg_pow '/' arg_mult]
+|	[arg_pow '%' arg_mult]
+|	[arg_pow]
+
+def arg_pow
+	[arg_unary '**' arg_pow]
+|	[arg_unary]
+
+def arg_unary
+	['!' primary]
+|	['~' primary]
+|	[tUPLUS primary]
+|	[tUMINUS primary]
+|	[primary]
+
+def primary_value 
+	[primary]
+
+def primary
+	[pliteral]
+|	[strings]
+|	[xstring]
+#|	[regexp]
+#|	[words]
+#|	[qwords]
+|	[var_ref]
+|	[backref]
+|	[tFID]
+|	['begin' bodystmt 'end']
+|	[tLPAREN compstmt ')']
+|	[primary_value '::' tCONSTANT]
+|	['::' tCONSTANT]
+|	[tLBRACK aref_args ']']
+|	['{' assoc_list '}']
+|	['defined?' '(' expr ')']
+|	[operation brace_block]
+|	[method_call]
+|	[method_call brace_block]
+|	['->' lambda]
+|	['if' expr_value then compstmt if_tail 'end']
+|	['unless' expr_value then compstmt opt_else 'end']
+|	['while' expr_value do compstmt 'end']
+|	['until' expr_value do compstmt 'end']
+#|	['case' expr_value opt_terms case_body 'end']
+#|	['case' opt_terms case_body 'end']
+|	['for' for_var 'in' expr_value do compstmt 'end']
+|	['class' cpath superclass bodystmt 'end']
+|	['class' '<<' expr term bodystmt 'end']
+|	['module' cpath bodystmt 'end']
+|	['def' fname f_arglist bodystmt 'end']
+|	['def' singleton dot_or_colon fname f_arglist bodystmt 'end']
+|	['break']
+|	['next']
+|	['redo']
+|	['retry']
+
+def for_var
+	[lhs]
+|	[mlhs]
+
+def lambda
+	[f_larglist lambda_body]
+
+def f_larglist
+	['(' f_args opt_bv_decl ')']
+|	[f_args opt_bv_decl]
+
+def lambda_body
+	['{' compstmt '}']
+|	['do' compstmt 'end']
+
+def assoc_list
+	[assocs trailer]
+|	[]
+
+def assocs
+	[assocs ',' assoc]
+|	[assoc]
+
+def assoc
+	[arg_value '=>' arg_value]
+|	[':' arg_value]
+
+def singleton
+	[var_ref]
+|	['(' expr ')']
+
+def dot_or_colon
+	['.']
+|	['::']
+
+def aref_args
+	[args trailer]
+|	[args ',' assocs trailer]
+|	[assocs trailer]
+|	[]
+
+def trailer		
+	[',']
+|	[]
+
+def brace_block
+	['{' opt_block_param compstmt '}']
+|	['do' opt_block_param compstmt 'end']
+
+def f_arglist
+	['(' f_args ')']
+|	[f_args term]
+
+def f_args
+	[f_arg_list]
+|	[]
+
+def f_arg_list
+	[f_arg_list ',' f_arg_item]
+|	[f_arg_item]
+
+def f_arg_item
+	[f_norm_arg]
+|	[f_opt]
+|	[f_rest_arg]
+|	[f_block_arg]
+|	['(' f_args ')']
+
+def f_opt
+	[tIDENTIFIER '=' arg_value]
+
+def f_rest_arg
+	['*' tIDENTIFIER]
+|	['*']
+
+def f_block_arg
+	[tAMPER tIDENTIFIER]
+
+def f_norm_arg
+	[tIDENTIFIER]
+
+def backref
+	[tNTH_REF] | [tBACK_REF]
+
+def superclass	
+	[term]
+|	['<' expr_value term]
+
+def cpath
+	['::' cname]
+|	[cname]
+|	[primary_value '::' cname]
+
+def fname
+	[tIDENTIFIER]
+|	[tCONSTANT]
+|	[tFID]
+|	[op]
+|	[reswords]
+
+def reswords
+	['__LINE__'] |	['__FILE__'] |	['__ENCODING__'] |	['BEGIN'] |	['END'] |
+	['alias'] |	['and'] |	['begin'] |	['break'] |	['case'] |	['class'] |
+	['def'] |	['defined?'] |	['do'] |	['else'] |	['elsif'] |	['end'] |
+	['ensure'] |	['false'] |	['for'] |	['in'] |	['module'] |
+	['next'] |	['nil'] |	['not'] |	['or'] |	['redo'] |	['rescue'] |
+	['retry'] |	['return'] |	['self'] |	['super'] |	['then'] |	['true'] |
+	['undef'] |	['when'] |	['yield'] |	['if'] |	['unless'] |	['while'] |
+	['until']
+
+def cname
+	[tIDENTIFIER]
+|	[tCONSTANT]
+
+def pliteral
+	[numeric]
+|	[symbol]
+#|	[dsym]
+
+def strings
+	[string]
+
+def string
+#	[tCHAR]
+	[string1]
+|	[string string1]
+
+def string1
+	[tSSTRING_BEG sstring_contents? tSSTRING_END]
+|	[tDSTRING_BEG dstring_contents? tDSTRING_END]
+
+def xstring
+	[tXSTRING_BEG xstring_contents? tXSTRING_END]
+
+def numeric
+	[tINTEGER]
+|	[tFLOAT]
+
+def symbol
+	[':' sym]
+
+def sym
+	[fname]
+|	[tIVAR]
+|	[tGVAR]
+|	[tCVAR]
+
+def fitem
+	[fsym]
+#|	[dsym]
+
+def undef_list
+	[fitem]
+|	[undef_list ',' fitem]
+
+def fsym
+	[fname]
+|	[symbol]
+
+#def dsym
+#	[':' xstring_contents tDSTRING_END]
+
+def var_ref
+	[variable]
+
+def variable
+	[tIDENTIFIER] | [tIVAR] | [tGVAR] | [tCONSTANT] | [tCVAR] | ['nil'] |
+	['self'] | ['true'] | ['false'] | ['__FILE__'] | ['__LINE__'] |
+	['__ENCODING__']
+
+
+# Required whitespace, but newline is not allowed.
+token ws_no_nl
+	/[ \t]+[^ \t\n]/
+	{
+		send( make_token( typeid ws_no_nl, pull(stdin, match_length-1) ) )
+	}
+
+def method_call
+	[operation paren_args]
+|	[operation ws_no_nl call_args]
+|	[primary_value '.' operation2 opt_paren_args]
+|	[primary_value '.' operation2 ws_no_nl call_args]
+|	[primary_value '::' operation2 opt_paren_args]
+|	[primary_value '::' operation2 ws_no_nl call_args]
+|	[primary_value '.' paren_args]
+|	[primary_value '::' paren_args]
+|	['super' paren_args]
+|	['super' ws_no_nl call_args]
+|	['super']
+|	['yield' paren_args]
+|	['yield' ws_no_nl call_args]
+|	['yield']
+|	['return' call_args]
+|	['return']
+|	[primary_value '[' opt_call_args ']']
+
+def opt_paren_args
+	[paren_args]
+|	[]
+
+def paren_args
+	['(' opt_call_args ')']
+
+#
+# Grammar finished
+#
+
+ruby R = parse ruby(stdin)
+
+print_xml( R )
+
+#for T: primary in R
+#	print_xml( T, '\n\n' )