regexp_predef_range[parser_type & tree, syntac::terminal_range & _range] ::=
{ std::string _char_class;/* bool _invert = false; }*/
(
"." { _char_class.assign("any");/* }*/
| "\\"
( "d" { _char_class.assign("digit");/* }*/
| "D" { _char_class.assign("digit");/* _invert = true; }*/
| "w" { _char_class.assign("word");/* }*/
| "W" { _char_class.assign("word");/* _invert = true; }*/
| "s" { _char_class.assign("space");/* }*/
| "S" { _char_class.assign("space");/* _invert = true; }*/
| "p{" '[A-Za-z_0-9&]+' "}"
{ assign(_char_class, $2);/* }*/
| "P{" '[A-Za-z_0-9&]+' "}"
{ assign(_char_class, $2);/* _invert = true; }*/
)
| "[:"
( "L" { _char_class.assign("Letter");/* }*/
| "Lu"{ _char_class.assign("Uppercase_Letter");/* }*/
| "Ll"{ _char_class.assign("Lowercase_Letter");/* }*/
| "Lt"{ _char_class.assign("Titlecase_Letter");/* }*/
| "L&"{ _char_class.assign("LetterAmp");/* }*/
| "Lm"{ _char_class.assign("Modifier_Letter");/* }*/
| "Lo"{ _char_class.assign("Other_Letter");/* }*/
| "M" { _char_class.assign("Mark");/* }*/
| "Mn"{ _char_class.assign("Nonspacing_Mark");/* }*/
| "Me"{ _char_class.assign("Enclosing_Mark");/* }*/
| "Mc"{ _char_class.assign("Spacing_Mark");/* }*/
| "N" { _char_class.assign("Number");/* }*/
| "Nd"{ _char_class.assign("Decimal_Number");/* }*/
| "Nl"{ _char_class.assign("Letter_Number");/* }*/
| "No"{ _char_class.assign("Other_Number");/* }*/
| "Z" { _char_class.assign("Separator");/* }*/
| "Zs"{ _char_class.assign("Space_Separator");/* }*/
| "Zl"{ _char_class.assign("Line_Separator");/* }*/
| "Zp"{ _char_class.assign("Paragraph_Separator");/* }*/
| "C" { _char_class.assign("Other");/* }*/
| "Cc"{ _char_class.assign("Control");/* }*/
| "Cf"{ _char_class.assign("Format");/* }*/
| "Co"{ _char_class.assign("Private_Use");/* }*/
| "Cs"{ _char_class.assign("Surrogate");/* }*/
| "Cn"{ _char_class.assign("Unassigned");/* }*/
| "P" { _char_class.assign("Punctuation");/* }*/
| "Pd"{ _char_class.assign("Dash_Punctuation");/* }*/
| "Ps"{ _char_class.assign("Open_Punctuation");/* }*/
| "Pe"{ _char_class.assign("Close_Punctuation");/* }*/
| "Pc"{ _char_class.assign("Connector_Punctuation");/* }*/
| "Pi"{ _char_class.assign("Initial_Punctuation");/* }*/
| "Pf"{ _char_class.assign("Final_Punctuation");/* }*/
| "Po"{ _char_class.assign("Other_Punctuation");/* }*/
| "S" { _char_class.assign("Symbol");/* }*/
| "Sm"{ _char_class.assign("Math_Symbol");/* }*/
| "Sc"{ _char_class.assign("Currency_Symbol");/* }*/
| "Sk"{ _char_class.assign("Modifier_Symbol");/* }*/
| "So"{ _char_class.assign("Other_Symbol");/* }*/
| ('[A-Za-z_0-9&]+' { assign(_char_class, $0);/* })*/
)
":]"
)
{
if ( _char_class.empty() || !tree.is_symbol_defined(_char_class.c_str()))
{
std::cerr << $file << ':' << $line << ": warning: attempt to use undefined character class: " << _char_class << std::endl;
}
else
{
vertex_token class_symbol = tree.get_symbol_id(_char_class.c_str());
arc_token _childx = tree.out_arcs_begin(class_symbol);
if (_childx != tree.out_arcs_end(class_symbol))
{
vertex_token child_node = parser_type::arc_to(* _childx);
if (tree.get_vertex_data(child_node).get_type() != syntac::parse_vertex::terminal_vertex_type)
{
std::cerr << $file << ':' << $line << ": warning: attempt to use non-simple symbol as character class: " << _char_class << std::endl;
}
else
{
_range.add_data(tree.get_vertex_data(child_node).get_name());
if (_invert)
{
_range.invert();
}
}
}
}
} |
