| 18 |
// along with this program; if not, write to the Free Software |
// along with this program; if not, write to the Free Software |
| 19 |
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
| 20 |
|
|
| 21 |
|
// This is a basic nonrecursive, table driven, top-down parser. |
| 22 |
|
// |
| 23 |
|
// Nonterminals: |
| 24 |
|
// ============ |
| 25 |
|
// STRING: a string, either delimited by '"' and '"', or '[' and ']' |
| 26 |
|
// COMMAND_BEGIN: '%' |
| 27 |
|
// BLOCK_BEGIN: '{' |
| 28 |
|
// BLOCK_END: '}' |
| 29 |
|
// IDENTIFIER: [A-Za-z_][A-Za-z0-9_]* |
| 30 |
|
// EQUALS: '=' |
| 31 |
|
// SEMICOLON: ';' |
| 32 |
|
// ENDL: '\n' |
| 33 |
|
// EOF: end-of-file |
| 34 |
|
// |
| 35 |
|
// |
| 36 |
|
// Productions: |
| 37 |
|
// =========== |
| 38 |
|
// 1. start -> statement_list EOF |
| 39 |
|
// 2. statement_list -> statement statement_list |
| 40 |
|
// 3. | EMPTY |
| 41 |
|
// 4. statement -> command statement_end |
| 42 |
|
// 5. | assignment statement_end |
| 43 |
|
// 6. | statement_end |
| 44 |
|
// 7. statement_end -> SEMICOLON |
| 45 |
|
// 8. | ENDL |
| 46 |
|
// 9. command -> COMMAND_BEGIN IDENTIFIER argument_list |
| 47 |
|
// 10. argument_list -> argument argument_list |
| 48 |
|
// 11. | EMPTY |
| 49 |
|
// 12. argument -> STRING |
| 50 |
|
// 13. | block |
| 51 |
|
// 14. block -> BLOCK_BEGIN statement_list BLOCK_END |
| 52 |
|
// 15. assignment -> IDENTIFIER EQUALS STRING |
| 53 |
|
// |
| 54 |
|
// |
| 55 |
|
// SYMBOL FIRST FOLLOW |
| 56 |
|
// ====== ===== ====== |
| 57 |
|
// start: IDENTIFIER EOF |
| 58 |
|
// COMMAND_BEGIN |
| 59 |
|
// SEMICOLON |
| 60 |
|
// ENDL |
| 61 |
|
// EMPTY |
| 62 |
|
// |
| 63 |
|
// statement_list: COMMAND_BEGIN BLOCK_END |
| 64 |
|
// IDENTIFIER EOF |
| 65 |
|
// SEMICOLON |
| 66 |
|
// ENDL |
| 67 |
|
// EMPTY |
| 68 |
|
// |
| 69 |
|
// statement: COMMAND_BEGIN SEMICOLON |
| 70 |
|
// IDENTIFIER ENDL |
| 71 |
|
// SEMICOLON |
| 72 |
|
// ENDL |
| 73 |
|
// |
| 74 |
|
// statement_end: SEMICOLON COMMAND_BEGIN |
| 75 |
|
// ENDL IDENTIFIER |
| 76 |
|
// SEMICOLON |
| 77 |
|
// ENDL |
| 78 |
|
// BLOCK_END |
| 79 |
|
// EOF |
| 80 |
|
// |
| 81 |
|
// command: COMMAND_BEGIN SEMICOLON |
| 82 |
|
// ENDL |
| 83 |
|
// |
| 84 |
|
// argument_list: STRING SEMICOLON |
| 85 |
|
// BLOCK_BEGIN ENDL |
| 86 |
|
// EMPTY |
| 87 |
|
// |
| 88 |
|
// argument: STRING STRING |
| 89 |
|
// BLOCK_BEGIN BLOCK_BEGIN |
| 90 |
|
// SEMICOLON |
| 91 |
|
// ENDL |
| 92 |
|
// |
| 93 |
|
// block: BLOCK_BEGIN STRING |
| 94 |
|
// BLOCK_BEGIN |
| 95 |
|
// SEMICOLON |
| 96 |
|
// ENDL |
| 97 |
|
// BLOCK_END |
| 98 |
|
// |
| 99 |
|
// assignment: IDENTIFIER SEMICOLON |
| 100 |
|
// ENDL |
| 101 |
|
// BLOCK_END |
| 102 |
|
// |
| 103 |
|
// Parse Table: |
| 104 |
|
// =========== |
| 105 |
|
// (Data are the productions to be used.) |
| 106 |
|
// |
| 107 |
|
// COMMAND_BEGIN IDENTIFIER STRING BLOCK_BEGIN BLOCK_END EQUALS SEMICOLON ENDL EOF |
| 108 |
|
// start: 1 1 - - - - 1 1 1 |
| 109 |
|
// statement_list: 2 2 - - 3 - 2 2 3 |
| 110 |
|
// statement: 4 5 - - - - 6 6 - |
| 111 |
|
// statement_end: - - - - - - 7 8 - |
| 112 |
|
// command: 9 - - - - - - - - |
| 113 |
|
// argument_list: - - 10 10 - - 11 11 - |
| 114 |
|
// argument: - - 12 13 - - - - - |
| 115 |
|
// block: - - - 14 - - - - - |
| 116 |
|
// assignment: - 15 - - - - - - - |
| 117 |
|
// |
| 118 |
|
|
| 119 |
#ifndef __MPAK__PARSER_TCC__ |
#ifndef __MPAK__PARSER_TCC__ |
| 120 |
#define __MPAK__PARSER_TCC__ |
#define __MPAK__PARSER_TCC__ |
| 121 |
|
|
| 122 |
#include <config.h> |
#include <config.h> |
| 123 |
|
|
| 124 |
#include <list> |
#include <stdexcept> |
| 125 |
|
#include <stack> |
| 126 |
|
|
| 127 |
#include <ecstacy/util/macros.hh> |
#include <ecstacy/util/macros.hh> |
| 128 |
|
|
| 130 |
{ |
{ |
| 131 |
namespace spec |
namespace spec |
| 132 |
{ |
{ |
| 133 |
template<typename char_type_, typename traits_type_ = std::char_traits<_char_type> > |
template<typename char_type_, typename traits_type_> |
| 134 |
basic_parser<char_type_, traits_type_>::~basic_parser (void) |
typename basic_parser<char_type_,traits_type_>::parse_tree |
| 135 |
ECSTACY_THROW_SPEC (()) |
basic_parser<char_type_, traits_type_>::do_parse_ (basic_parser<char_type_,traits_type_>::lexer_type &lexer) |
| 136 |
{ |
ECSTACY_THROW_SPEC ((basic_parser<char_type_, traits_type_>::failure)) |
|
} |
|
|
|
|
|
template<typename char_type_, typename traits_type_ = std::char_traits<_char_type> > |
|
|
void |
|
|
basic_parser<char_type_, traits_type_>::parse (void) |
|
|
ECSTACY_THROW_SPEC ((basic_parser::failure)) |
|
| 137 |
{ |
{ |
| 138 |
|
std::stack<symbol> symbol_stack; |
| 139 |
|
symbol_stack.push (symbol (token_type (token_type::type_eof))); |
| 140 |
|
symbol_stack.push (symbol (symbol::type_start)); |
| 141 |
|
token_type token (lexer.next_token ()); |
| 142 |
|
|
| 143 |
|
do { |
| 144 |
|
symbol top = symbol_stack.top (); |
| 145 |
|
|
| 146 |
|
switch (top.type) { |
| 147 |
|
case symbol::type_token: |
| 148 |
|
if (top.token.type == token.type) { |
| 149 |
|
symbol_stack.pop (); |
| 150 |
|
token = lexer.next_token (); |
| 151 |
|
} else { |
| 152 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 153 |
|
} |
| 154 |
|
break; |
| 155 |
|
|
| 156 |
|
case symbol::type_start: |
| 157 |
|
switch (token.type) { |
| 158 |
|
case token_type::type_command_begin: |
| 159 |
|
case token_type::type_identifier: |
| 160 |
|
case token_type::type_semicolon: |
| 161 |
|
case token_type::type_endl: |
| 162 |
|
case token_type::type_eof: |
| 163 |
|
// start -> statement_list |
| 164 |
|
symbol_stack.pop (); |
| 165 |
|
symbol_stack.push (symbol::type_statement_list); |
| 166 |
|
break; |
| 167 |
|
default: |
| 168 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 169 |
|
} |
| 170 |
|
break; |
| 171 |
|
|
| 172 |
|
case symbol::type_statement_list: |
| 173 |
|
switch (token.type) { |
| 174 |
|
case token_type::type_command_begin: |
| 175 |
|
case token_type::type_identifier: |
| 176 |
|
case token_type::type_semicolon: |
| 177 |
|
case token_type::type_endl: |
| 178 |
|
// statement_list -> statement statement_list |
| 179 |
|
symbol_stack.pop (); |
| 180 |
|
symbol_stack.push (symbol (symbol::type_statement_list)); |
| 181 |
|
symbol_stack.push (symbol (symbol::type_statement)); |
| 182 |
|
break; |
| 183 |
|
case token_type::type_block_end: |
| 184 |
|
case token_type::type_eof: |
| 185 |
|
// statement_list -> EMPTY |
| 186 |
|
symbol_stack.pop (); |
| 187 |
|
break; |
| 188 |
|
default: |
| 189 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 190 |
|
} |
| 191 |
|
break; |
| 192 |
|
|
| 193 |
|
case symbol::type_statement: |
| 194 |
|
switch (token.type) { |
| 195 |
|
case token_type::type_command_begin: |
| 196 |
|
// statement -> command statement_end |
| 197 |
|
symbol_stack.pop (); |
| 198 |
|
symbol_stack.push (symbol (symbol::type_statement_end)); |
| 199 |
|
symbol_stack.push (symbol (symbol::type_command)); |
| 200 |
|
break; |
| 201 |
|
case token_type::type_identifier: |
| 202 |
|
// statement -> assignment statement_end |
| 203 |
|
symbol_stack.pop (); |
| 204 |
|
symbol_stack.push (symbol (symbol::type_statement_end)); |
| 205 |
|
symbol_stack.push (symbol (symbol::type_assignment)); |
| 206 |
|
break; |
| 207 |
|
case token_type::type_semicolon: |
| 208 |
|
case token_type::type_endl: |
| 209 |
|
// statement -> statement_end |
| 210 |
|
symbol_stack.pop (); |
| 211 |
|
symbol_stack.push (symbol (symbol::type_statement_end)); |
| 212 |
|
break; |
| 213 |
|
default: |
| 214 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 215 |
|
} |
| 216 |
|
break; |
| 217 |
|
|
| 218 |
|
case symbol::type_statement_end: |
| 219 |
|
switch (token.type) { |
| 220 |
|
case token_type::type_semicolon: |
| 221 |
|
// statement_end -> SEMICOLON |
| 222 |
|
symbol_stack.pop (); |
| 223 |
|
symbol_stack.push (symbol (token_type (token_type::type_semicolon))); |
| 224 |
|
break; |
| 225 |
|
case token_type::type_endl: |
| 226 |
|
// statement_end -> ENDL |
| 227 |
|
symbol_stack.pop (); |
| 228 |
|
symbol_stack.push (symbol (token_type (token_type::type_endl))); |
| 229 |
|
break; |
| 230 |
|
default: |
| 231 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 232 |
|
} |
| 233 |
|
|
| 234 |
|
case symbol::type_command: |
| 235 |
|
switch (token.type) { |
| 236 |
|
case token_type::type_command_begin: |
| 237 |
|
// command -> COMMAND_BEGIN IDENTIFIER argument_list |
| 238 |
|
symbol_stack.pop (); |
| 239 |
|
symbol_stack.push (symbol (symbol::type_argument_list)); |
| 240 |
|
symbol_stack.push (symbol (token_type (token_type::type_identifier))); |
| 241 |
|
symbol_stack.push (symbol (token_type (token_type::type_command_begin))); |
| 242 |
|
break; |
| 243 |
|
default: |
| 244 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 245 |
|
} |
| 246 |
|
break; |
| 247 |
|
|
| 248 |
|
case symbol::type_argument_list: |
| 249 |
|
switch (token.type) { |
| 250 |
|
case token_type::type_string: |
| 251 |
|
case token_type::type_block_begin: |
| 252 |
|
// argument_list -> argument argument_list |
| 253 |
|
symbol_stack.pop (); |
| 254 |
|
symbol_stack.push (symbol (symbol::type_argument_list)); |
| 255 |
|
symbol_stack.push (symbol (symbol::type_argument)); |
| 256 |
|
break; |
| 257 |
|
case token_type::type_semicolon: |
| 258 |
|
case token_type::type_endl: |
| 259 |
|
// argument_list -> EMPTY |
| 260 |
|
symbol_stack.pop (); |
| 261 |
|
break; |
| 262 |
|
default: |
| 263 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 264 |
|
} |
| 265 |
|
break; |
| 266 |
|
|
| 267 |
|
case symbol::type_argument: |
| 268 |
|
switch (token.type) { |
| 269 |
|
case token_type::type_string: |
| 270 |
|
// argument -> STRING |
| 271 |
|
symbol_stack.pop (); |
| 272 |
|
symbol_stack.push (symbol (token_type (token_type::type_string))); |
| 273 |
|
break; |
| 274 |
|
case token_type::type_block_begin: |
| 275 |
|
// argument -> block |
| 276 |
|
symbol_stack.pop (); |
| 277 |
|
symbol_stack.push (symbol (symbol::type_block)); |
| 278 |
|
break; |
| 279 |
|
default: |
| 280 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 281 |
|
} |
| 282 |
|
break; |
| 283 |
|
|
| 284 |
|
case symbol::type_block: |
| 285 |
|
switch (token.type) { |
| 286 |
|
case token_type::type_block_begin: |
| 287 |
|
// block -> BLOCK_BEGIN statement_list BLOCK_END |
| 288 |
|
symbol_stack.pop (); |
| 289 |
|
symbol_stack.push (symbol (token_type (token_type::type_block_end))); |
| 290 |
|
symbol_stack.push (symbol (symbol::type_start)); |
| 291 |
|
symbol_stack.push (symbol (token_type (token_type::type_block_begin))); |
| 292 |
|
default: |
| 293 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 294 |
|
} |
| 295 |
|
break; |
| 296 |
|
|
| 297 |
|
case symbol::type_assignment: |
| 298 |
|
switch (token.type) { |
| 299 |
|
case token_type::type_identifier: |
| 300 |
|
// assignment -> IDENTIFIER EQUALS STRING |
| 301 |
|
symbol_stack.pop (); |
| 302 |
|
symbol_stack.push (symbol (token_type (token_type::type_string))); |
| 303 |
|
symbol_stack.push (symbol (token_type (token_type::type_equals))); |
| 304 |
|
symbol_stack.push (symbol (token_type (token_type::type_identifier))); |
| 305 |
|
default: |
| 306 |
|
ecstacy::util::throw_exception (failure ("unrecognized token")); |
| 307 |
|
} |
| 308 |
|
default: |
| 309 |
|
ecstacy::util::throw_exception (std::logic_error ("what the fuck?")); |
| 310 |
|
} |
| 311 |
|
} while (1); |
| 312 |
|
return parse_tree (); |
| 313 |
} |
} |
| 314 |
} |
} |
| 315 |
} |
} |