| File: | mtx_reader.cc |
| Warning: | line 805, column 3 Value stored to 'val' is never read |
Press '?' to see keyboard shortcuts
Keyboard shortcuts:
| 1 | /* |
| 2 | * Copyright (C) 2005 Universitat d'Alacant / Universidad de Alicante |
| 3 | * |
| 4 | * This program is free software; you can redistribute it and/or |
| 5 | * modify it under the terms of the GNU General Public License as |
| 6 | * published by the Free Software Foundation; either version 2 of the |
| 7 | * License, or (at your option) any later version. |
| 8 | * |
| 9 | * This program is distributed in the hope that it will be useful, but |
| 10 | * WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 12 | * General Public License for more details. |
| 13 | * |
| 14 | * You should have received a copy of the GNU General Public License |
| 15 | * along with this program; if not, see <https://www.gnu.org/licenses/>. |
| 16 | */ |
| 17 | #include <apertium/mtx_reader.h> |
| 18 | #include <lttoolbox/xml_parse_util.h> |
| 19 | #include <lttoolbox/compression.h> |
| 20 | #include <lttoolbox/string_utils.h> |
| 21 | #include <apertium/utils.h> |
| 22 | #include <apertium/tsx_reader.h> |
| 23 | #include <apertium/perceptron_spec.h> |
| 24 | |
| 25 | #include <cstdlib> |
| 26 | #include <iostream> |
| 27 | #include <sstream> |
| 28 | #include <iterator> |
| 29 | |
| 30 | // XML parsing function guideline |
| 31 | // When control is pass to you, you might need to stepToTag |
| 32 | // When delegating or returning control, step beyond yourself |
| 33 | |
| 34 | namespace Apertium { |
| 35 | MTXReader::MTXReader(VM &spec) : |
| 36 | spec(spec), in_global_defn(false), |
| 37 | template_slot_counter(0), cur_feat(NULL__null) {} |
| 38 | |
| 39 | size_t MTXReader::pushSetConst(std::string &val) |
| 40 | { |
| 41 | size_t set_idx = spec.set_consts.size(); |
| 42 | std::stringstream val_ss(val); |
| 43 | spec.set_consts.push_back(set<std::string>( |
| 44 | istream_iterator<std::string>(val_ss), |
| 45 | istream_iterator<std::string>() |
| 46 | )); |
| 47 | return set_idx; |
| 48 | } |
| 49 | |
| 50 | size_t MTXReader::pushStrConst(std::string &val) |
| 51 | { |
| 52 | size_t str_idx = spec.str_consts.size(); |
| 53 | spec.str_consts.push_back(val); |
| 54 | return str_idx; |
| 55 | } |
| 56 | |
| 57 | void MTXReader::emitBytecode(VM::Bytecode bc) |
| 58 | { |
| 59 | cur_feat->push_back(bc.intbyte); |
| 60 | } |
| 61 | |
| 62 | void MTXReader::pokeBytecode(size_t addr, VM::Bytecode bc) |
| 63 | { |
| 64 | (*cur_feat)[addr] = bc.intbyte; |
| 65 | } |
| 66 | |
| 67 | void MTXReader::emitOpcode(VM::Opcode op) |
| 68 | { |
| 69 | emitBytecode((VM::Bytecode){.op = op}); |
| 70 | } |
| 71 | |
| 72 | void MTXReader::emitInt(int val) |
| 73 | { |
| 74 | assert(-128 <= val && val < 128)(static_cast <bool> (-128 <= val && val < 128) ? void (0) : __assert_fail ("-128 <= val && val < 128" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 75 | emitBytecode((VM::Bytecode){.intbyte = (signed char)val}); |
| 76 | } |
| 77 | |
| 78 | void MTXReader::emitUInt(int val) |
| 79 | { |
| 80 | assert(0 <= val && val < 256)(static_cast <bool> (0 <= val && val < 256 ) ? void (0) : __assert_fail ("0 <= val && val < 256" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 81 | emitBytecode((VM::Bytecode){.uintbyte = (unsigned char)val}); |
| 82 | } |
| 83 | |
| 84 | void MTXReader::procCoarseTags() |
| 85 | { |
| 86 | std::string tsx_fn((const char*) xmlTextReaderGetAttribute(reader, (const xmlChar*) "tag")); |
| 87 | bool is_abs = ((tsx_fn.size() >= 1 && tsx_fn[0] == '/') || |
| 88 | (tsx_fn.size() >= 2 && tsx_fn[1] == ':')); |
| 89 | if (!is_abs) { |
| 90 | size_t last_slash_pos = path.rfind("/"); |
| 91 | if (last_slash_pos != string::npos) { |
| 92 | tsx_fn = path.substr(0, last_slash_pos + 1) + tsx_fn; |
| 93 | } |
| 94 | } |
| 95 | TSXReader tsx_reader; |
| 96 | tsx_reader.read(tsx_fn); |
| 97 | spec.coarse_tags = Optional<TaggerDataPercepCoarseTags>( |
| 98 | tsx_reader.getTaggerData()); |
| 99 | stepPastSelfClosingTag("coarse-tags"_u); |
| 100 | } |
| 101 | |
| 102 | void MTXReader::procSetDef() |
| 103 | { |
| 104 | std::string name = attrib_str("name"_u); |
| 105 | stepToNextTag(); |
| 106 | size_t set_idx = spec.set_consts.size(); |
| 107 | spec.set_consts.push_back(VMSet()); |
| 108 | VMSet &vm_set = spec.set_consts.back(); |
| 109 | while (type != XML_READER_TYPE_END_ELEMENT) { |
| 110 | if (name == "set-member") { |
| 111 | std::string tag = attrib_str("tag"_u); |
| 112 | std::string str = attrib_str("str"_u); |
| 113 | vm_set.insert(tag.empty() ? str : tag); |
| 114 | } else { |
| 115 | parseError("Expected set-member"_u); |
| 116 | } |
| 117 | stepToNextTag(); |
| 118 | } |
| 119 | set_names[name] = set_idx; |
| 120 | assert(name == "def-set")(static_cast <bool> (name == "def-set") ? void (0) : __assert_fail ("name == \"def-set\"", __builtin_FILE (), __builtin_LINE () , __extension__ __PRETTY_FUNCTION__)); |
| 121 | stepToNextTag(); |
| 122 | } |
| 123 | |
| 124 | void MTXReader::procStrDef() |
| 125 | { |
| 126 | std::string name = attrib_str("name"_u); |
| 127 | std::string tag = attrib_str("tag"_u); |
| 128 | std::string str = attrib_str("str"_u); |
| 129 | str_names[name] = pushStrConst(tag.empty() ? str : tag); |
| 130 | stepPastSelfClosingTag("def-str"_u); |
| 131 | } |
| 132 | |
| 133 | void |
| 134 | MTXReader::procDefns() |
| 135 | { |
| 136 | stepToNextTag(); |
| 137 | while (type != XML_READER_TYPE_END_ELEMENT) { |
| 138 | if (name == "def-set"_u) { |
| 139 | procSetDef(); |
| 140 | } else if (name == "def-str"_u) { |
| 141 | procStrDef(); |
| 142 | } else if (name == "def-macro"_u) { |
| 143 | procDefMacro(); |
| 144 | } else if (name == "#text"_u || name == "#comment"_u) { |
| 145 | // skip |
| 146 | } else { |
| 147 | unexpectedTag(); |
| 148 | } |
| 149 | } |
| 150 | assert(name == "defns"_u)(static_cast <bool> (name == "defns"_u) ? void (0) : __assert_fail ("name == \"defns\"_u", __builtin_FILE (), __builtin_LINE () , __extension__ __PRETTY_FUNCTION__)); |
| 151 | stepToNextTag(); |
| 152 | } |
| 153 | |
| 154 | void |
| 155 | MTXReader::procGlobalPred() |
| 156 | { |
| 157 | cur_feat = &spec.global_pred; |
| 158 | stepToNextTag(); |
| 159 | procBoolExpr(); |
| 160 | assert(name == "global-pred"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "global-pred"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"global-pred\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 161 | stepToNextTag(); |
| 162 | } |
| 163 | |
| 164 | void |
| 165 | MTXReader::emitSetImmOp(VM::Opcode op) |
| 166 | { |
| 167 | emitOpcode(op); |
| 168 | getAndEmitSetRef(); |
| 169 | } |
| 170 | |
| 171 | void |
| 172 | MTXReader::emitStrImmOp(VM::Opcode op) |
| 173 | { |
| 174 | emitOpcode(op); |
| 175 | getAndEmitStrRef(); |
| 176 | } |
| 177 | |
| 178 | void |
| 179 | MTXReader::procBinCompareOp(VM::Opcode op) |
| 180 | { |
| 181 | procIntExpr(); |
| 182 | procIntExpr(); |
| 183 | emitOpcode(op); |
| 184 | } |
| 185 | |
| 186 | void |
| 187 | MTXReader::procCommBoolOp(VM::Opcode op) |
| 188 | { |
| 189 | int operand_count = 0; |
| 190 | while (type != XML_READER_TYPE_END_ELEMENT) { |
| 191 | procBoolExpr(); |
| 192 | operand_count++; |
| 193 | } |
| 194 | while (--operand_count > 0) { |
| 195 | emitOpcode(op); |
| 196 | } |
| 197 | } |
| 198 | |
| 199 | bool |
| 200 | MTXReader::procIntExpr(bool allow_fail) |
| 201 | { |
| 202 | /* Self-closing tags */ |
| 203 | if (!tryProcArg(INTEXPR, true) |
| 204 | && !tryProcVar(VM::INTVAL)) { |
| 205 | if (name == "sentlen"_u) { |
| 206 | emitOpcode(VM::SENTLENTOK); |
| 207 | stepPastSelfClosingTag("sentlen"_u); |
| 208 | } else if (name == "pathlen"_u) { |
| 209 | emitOpcode(VM::SENTLENWRD); |
| 210 | stepPastSelfClosingTag("pathlen"_u); |
| 211 | } else if (name == "tokaddr"_u) { |
| 212 | emitOpcode(VM::PUSHTOKADDR); |
| 213 | stepPastSelfClosingTag("tokaddr"_u); |
| 214 | } else if (name == "wrdidx"_u) { |
| 215 | emitOpcode(VM::PUSHWRDADDR); |
| 216 | stepPastSelfClosingTag("wrdidx"_u); |
| 217 | } else if (name == "int"_u) { |
| 218 | emitOpcode(VM::PUSHINT); |
| 219 | getAndEmitInt(); |
| 220 | stepPastSelfClosingTag("int"_u); |
| 221 | /* Other tags */ |
| 222 | } else if (name == "add"_u) { |
| 223 | stepToNextTag(); |
| 224 | procIntExpr(); |
| 225 | procIntExpr(); |
| 226 | assert(name == "add"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "add"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"add\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 227 | emitOpcode(VM::ADD); |
| 228 | stepToNextTag(); |
| 229 | } else if (name == "toklen"_u) { |
| 230 | procIntExpr(); |
| 231 | assert(name == "toklen"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "toklen"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"toklen\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 232 | emitOpcode(VM::TOKLENWRD); |
| 233 | stepToNextTag(); |
| 234 | } else if (name == "strlen"_u) { |
| 235 | procStrExpr(); |
| 236 | assert(name == "strlen"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "strlen"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"strlen\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 237 | emitOpcode(VM::STRLEN); |
| 238 | stepToNextTag(); |
| 239 | } else if (name == "arrlen"_u) { |
| 240 | procStrArrExpr(); |
| 241 | assert(name == "arrlen"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "arrlen"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"arrlen\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 242 | procBinCompareOp(VM::ARRLEN); |
| 243 | stepToNextTag(); |
| 244 | } else { |
| 245 | if (allow_fail) { |
| 246 | return false; |
| 247 | } |
| 248 | parseError("Expected an integer expression."_u); |
| 249 | } |
| 250 | } |
| 251 | return true; |
| 252 | } |
| 253 | |
| 254 | bool |
| 255 | MTXReader::procStrArrExpr(bool allow_fail) |
| 256 | { |
| 257 | stepToTag(); |
| 258 | if (!tryProcArg(STRARREXPR, true) |
| 259 | && !tryProcVar(VM::STRARRVAL) |
| 260 | && !tryProcSlice(&MTXReader::procStrArrExpr)) { |
| 261 | if (name == "ex-tags"_u) { |
| 262 | stepToNextTag(); |
| 263 | procWordoidExpr(); |
| 264 | assert(type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (type == XML_READER_TYPE_END_ELEMENT ) ? void (0) : __assert_fail ("type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 265 | emitOpcode(VM::EXTAGS); |
| 266 | } else if (name == "ex-ambgset"_u) { |
| 267 | stepToNextTag(); |
| 268 | procIntExpr(); |
| 269 | emitOpcode(VM::EXAMBGSET); |
| 270 | } else if (name == "for-each"_u) { |
| 271 | procForEach(STREXPR); |
| 272 | } else { |
| 273 | if (allow_fail) { |
| 274 | return false; |
| 275 | } |
| 276 | parseError("Expected a string list expression."_u); |
| 277 | } |
| 278 | stepToNextTag(); |
| 279 | } |
| 280 | return true; |
| 281 | } |
| 282 | |
| 283 | bool MTXReader::tryProcSubscript(bool (MTXReader::*proc_inner)(bool)) |
| 284 | { |
| 285 | if (name == "subscript"_u) { |
| 286 | int idx = getInt("idx"_u); |
| 287 | stepToNextTag(); |
| 288 | (this->*proc_inner)(false); |
| 289 | emitOpcode(VM::SUBSCRIPT); |
| 290 | emitUInt(idx); |
| 291 | assert(name == "subscript"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "subscript"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"subscript\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 292 | stepToNextTag(); |
| 293 | return true; |
| 294 | } |
| 295 | return false; |
| 296 | } |
| 297 | |
| 298 | bool MTXReader::tryProcSlice(bool (MTXReader::*proc_inner)(bool)) |
| 299 | { |
| 300 | if (name == "slice"_u) { |
| 301 | stepToNextTag(); |
| 302 | (this->*proc_inner)(false); |
| 303 | bool exists; |
| 304 | emitOpcode(VM::SLICE); |
| 305 | int start_lit = getInt("start"_u, exists); |
| 306 | if (exists) { |
| 307 | emitInt(start_lit); |
| 308 | } else { |
| 309 | emitInt(0); |
| 310 | } |
| 311 | int end_lit = getInt("end"_u, exists); |
| 312 | if (exists) { |
| 313 | emitInt(end_lit); |
| 314 | } else { |
| 315 | emitInt(0); |
| 316 | } |
| 317 | assert(name == "slice"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "slice"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"slice\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 318 | stepToNextTag(); |
| 319 | return true; |
| 320 | } |
| 321 | return false; |
| 322 | } |
| 323 | |
| 324 | bool MTXReader::tryProcArg(ExprType expr_type, bool allow_fail) |
| 325 | { |
| 326 | if (name == "var"_u) { |
| 327 | std::string var_name = attrib_str("name"_u); |
| 328 | if (in_global_defn) { |
| 329 | VarNVMap::const_iterator arg_name_it = template_arg_names.find(var_name); |
| 330 | if (arg_name_it != template_arg_names.end()) { |
| 331 | cur_replacements->push_back(make_pair(arg_name_it->second, expr_type)); |
| 332 | stepPastSelfClosingTag("var"_u); |
| 333 | return true; |
| 334 | } |
| 335 | if (!allow_fail) { |
| 336 | parseError("No such argument " + var_name); |
| 337 | } |
| 338 | } |
| 339 | } |
| 340 | return false; |
| 341 | } |
| 342 | |
| 343 | bool MTXReader::tryProcVar(VM::StackValueType svt) |
| 344 | { |
| 345 | if (name == "var"_u) { |
| 346 | std::string var_name = attrib_str("name"_u); |
| 347 | |
| 348 | VarNVMap::const_iterator slot_names_it = slot_names.find(var_name); |
| 349 | if (slot_names_it != slot_names.end()) { |
| 350 | if (slot_types[slot_names_it->second] != svt) { |
| 351 | parseError("Variable " + var_name + " has the wrong type"); |
| 352 | } |
| 353 | emitOpcode(VM::GETVAR); |
| 354 | emitUInt(slot_names_it->second); |
| 355 | stepPastSelfClosingTag("var"_u); |
| 356 | return true; |
| 357 | } |
| 358 | |
| 359 | parseError("Variable " + var_name + " has not been set."); |
| 360 | } else if (!in_global_defn && name == "macro"_u) { |
| 361 | // Get template data |
| 362 | std::string var_name = attrib_str("name"_u); |
| 363 | VarNVMap::const_iterator template_name_it = template_slot_names.find(var_name); |
| 364 | if (template_name_it == template_slot_names.end()) { |
| 365 | parseError("No such macro " + var_name); |
| 366 | } |
| 367 | size_t templ_idx = template_name_it->second; |
| 368 | if (template_slot_types[templ_idx] != svt) { |
| 369 | parseError("Macro " + var_name + " returns the wrong type"); |
| 370 | } |
| 371 | std::pair<VM::FeatureDefn, TemplateReplacements> &templ_defn = template_defns[templ_idx]; |
| 372 | // Get arg values |
| 373 | stepToNextTag(); |
| 374 | std::vector<VM::FeatureDefn> arg_values; |
| 375 | VM::FeatureDefn *saved_feat = cur_feat; |
| 376 | TemplateReplacements::const_iterator templ_repl_it = templ_defn.second.begin(); |
| 377 | for (; templ_repl_it != templ_defn.second.end(); templ_repl_it++) { |
| 378 | arg_values.push_back(VM::FeatureDefn()); |
| 379 | cur_feat = &arg_values.back(); |
| 380 | procTypeExpr(templ_repl_it->second); |
| 381 | } |
| 382 | cur_feat = saved_feat; |
| 383 | // Check for previous instanciation... |
| 384 | InstanciationMap::const_iterator templ_instcia_it = |
| 385 | template_instanciations.find(make_pair(templ_idx, arg_values)); |
| 386 | if (templ_instcia_it == template_instanciations.end()) { |
| 387 | // and instanciate if doesn't exist... |
| 388 | // by ordering argument defns correctly |
| 389 | vector<unsigned int> indices(templ_defn.second.size(), 0); |
| 390 | for (size_t i = 0; i < templ_defn.second.size(); i++) { |
| 391 | indices[i] = i; |
| 392 | } |
| 393 | SortByComparer<std::pair<size_t, ExprType>, int> sbc(templ_defn.second); |
| 394 | std::sort(indices.begin(), indices.end(), sbc); |
| 395 | // And then copying the template defn bits and the argument defn bits into the global defn |
| 396 | VM::FeatureDefn::const_iterator src_start_it = templ_defn.first.begin(); |
| 397 | VM::FeatureDefn::const_iterator src_end_it = templ_defn.first.end(); |
| 398 | VM::FeatureDefn::const_iterator src_cur_it = src_start_it; |
| 399 | spec.global_defns.push_back(VM::FeatureDefn()); |
| 400 | VM::FeatureDefn &global_defn = spec.global_defns.back(); |
| 401 | std::back_insert_iterator<VM::FeatureDefn> dest_it(global_defn); |
| 402 | for (size_t i = 0; i < indices.size(); i++) { |
| 403 | size_t repl_i = indices[i]; |
| 404 | size_t offset = templ_defn.second[repl_i].first; |
| 405 | VM::FeatureDefn &arg_src = arg_values[repl_i]; |
| 406 | std::copy(src_cur_it, src_cur_it + offset, dest_it); |
| 407 | src_cur_it = src_start_it + offset; |
| 408 | std::copy(arg_src.begin(), arg_src.end(), dest_it); |
| 409 | } |
| 410 | std::copy(src_cur_it, src_end_it, dest_it); |
| 411 | templ_instcia_it = template_instanciations.insert( |
| 412 | make_pair( |
| 413 | make_pair(templ_idx, arg_values), |
| 414 | spec.global_defns.size() - 1)).first; |
| 415 | } |
| 416 | // Finally emit a reference to the relevant global |
| 417 | emitOpcode(VM::GETGVAR); |
| 418 | emitUInt(templ_instcia_it->second); |
| 419 | // Step past end |
| 420 | assert(name == "macro"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "macro"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"macro\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 421 | stepToNextTag(); |
| 422 | return true; |
| 423 | } |
| 424 | return false; |
| 425 | } |
| 426 | |
| 427 | bool |
| 428 | MTXReader::procStrExpr(bool allow_fail) |
| 429 | { |
| 430 | if (!tryProcArg(STREXPR, true) |
| 431 | && !tryProcVar(VM::STRVAL) |
| 432 | && !tryProcSlice(&MTXReader::procStrExpr) |
| 433 | && !tryProcSubscript(&MTXReader::procStrArrExpr)) { |
| 434 | if (name == "ex-surf"_u) { |
| 435 | stepToNextTag(); |
| 436 | procIntExpr(); |
| 437 | emitOpcode(VM::EXTOKSURF); |
| 438 | } else if (name == "ex-lemma"_u) { |
| 439 | stepToNextTag(); |
| 440 | procWordoidExpr(); |
| 441 | emitOpcode(VM::EXWRDLEMMA); |
| 442 | } else if (name == "ex-coarse"_u) { |
| 443 | stepToNextTag(); |
| 444 | procWordoidExpr(); |
| 445 | emitOpcode(VM::EXWRDCOARSETAG); |
| 446 | } else if (name == "join"_u) { |
| 447 | bool has_attr; |
| 448 | size_t str_idx = getStrRef(has_attr); |
| 449 | if (!has_attr) { |
| 450 | str_idx = 255; |
| 451 | } |
| 452 | stepToNextTag(); |
| 453 | procStrArrExpr(); |
| 454 | emitOpcode(VM::JOIN); |
| 455 | emitUInt(str_idx); |
| 456 | } else { |
| 457 | if (allow_fail) { |
| 458 | return false; |
| 459 | } |
| 460 | parseError("Expected a string expression."_u); |
| 461 | } |
| 462 | assert(type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (type == XML_READER_TYPE_END_ELEMENT ) ? void (0) : __assert_fail ("type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 463 | stepToNextTag(); |
| 464 | } |
| 465 | return true; |
| 466 | } |
| 467 | |
| 468 | bool |
| 469 | MTXReader::procBoolExpr(bool allow_fail) |
| 470 | { |
| 471 | if (!tryProcArg(BEXPR, true) |
| 472 | && !tryProcVar(VM::BVAL)) { |
| 473 | if (name == "and"_u) { |
| 474 | stepToNextTag(); |
| 475 | procCommBoolOp(VM::AND); |
| 476 | assert(name == "and"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "and"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"and\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 477 | stepToNextTag(); |
| 478 | } else if (name == "or"_u) { |
| 479 | stepToNextTag(); |
| 480 | procCommBoolOp(VM::OR); |
| 481 | assert(name == "or"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "or"_u && type == XML_READER_TYPE_END_ELEMENT ) ? void (0) : __assert_fail ("name == \"or\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 482 | stepToNextTag(); |
| 483 | } else if (name == "not"_u) { |
| 484 | stepToNextTag(); |
| 485 | procBoolExpr(); |
| 486 | emitOpcode(VM::NOT); |
| 487 | assert(name == "not"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "not"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"not\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 488 | stepToNextTag(); |
| 489 | } else if (name == "eq"_u) { |
| 490 | stepToNextTag(); |
| 491 | procBinCompareOp(VM::EQ); |
| 492 | assert(name == "eq"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "eq"_u && type == XML_READER_TYPE_END_ELEMENT ) ? void (0) : __assert_fail ("name == \"eq\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 493 | stepToNextTag(); |
| 494 | } else if (name == "neq"_u) { |
| 495 | stepToNextTag(); |
| 496 | procBinCompareOp(VM::NEQ); |
| 497 | assert(name == "neq"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "neq"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"neq\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 498 | stepToNextTag(); |
| 499 | } else if (name == "lt"_u) { |
| 500 | stepToNextTag(); |
| 501 | procBinCompareOp(VM::LT); |
| 502 | assert(name == "lt"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "lt"_u && type == XML_READER_TYPE_END_ELEMENT ) ? void (0) : __assert_fail ("name == \"lt\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 503 | stepToNextTag(); |
| 504 | } else if (name == "lte"_u) { |
| 505 | stepToNextTag(); |
| 506 | procBinCompareOp(VM::LTE); |
| 507 | assert(name == "lte"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "lte"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"lte\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 508 | stepToNextTag(); |
| 509 | } else if (name == "gt"_u) { |
| 510 | stepToNextTag(); |
| 511 | procBinCompareOp(VM::GT); |
| 512 | assert(name == "gt"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "gt"_u && type == XML_READER_TYPE_END_ELEMENT ) ? void (0) : __assert_fail ("name == \"gt\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 513 | stepToNextTag(); |
| 514 | } else if (name == "gte"_u) { |
| 515 | stepToNextTag(); |
| 516 | procBinCompareOp(VM::GTE); |
| 517 | assert(name == "gte"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "gte"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"gte\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 518 | stepToNextTag(); |
| 519 | } else if (name == "streq"_u) { |
| 520 | size_t str_ref = getStrRef(); |
| 521 | stepToNextTag(); |
| 522 | procStrExpr(); |
| 523 | emitOpcode(VM::STREQ); |
| 524 | emitUInt(str_ref); |
| 525 | assert(name == "streq"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "streq"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"streq\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 526 | stepToNextTag(); |
| 527 | } else if (name == "strin"_u) { |
| 528 | size_t set_ref = getSetRef(); |
| 529 | stepToNextTag(); |
| 530 | procStrExpr(); |
| 531 | emitOpcode(VM::STRIN); |
| 532 | emitUInt(set_ref); |
| 533 | assert(name == "strin"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "strin"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"strin\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 534 | stepToNextTag(); |
| 535 | /* Identical to strin? |
| 536 | } else if (name == "sethas"_u) { |
| 537 | stepToNextTag(); |
| 538 | procStrExpr(); |
| 539 | emitSetImmOp(VM::SETHAS); |
| 540 | */ |
| 541 | } else if (name == "sethasany"_u) { |
| 542 | size_t set_ref = getSetRef(); |
| 543 | stepToNextTag(); |
| 544 | procStrArrExpr(); |
| 545 | emitOpcode(VM::SETHASANY); |
| 546 | emitUInt(set_ref); |
| 547 | assert(name == "sethasany"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "sethasany"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"sethasany\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 548 | stepToNextTag(); |
| 549 | } else if (name == "sethasall"_u) { |
| 550 | size_t set_ref = getSetRef(); |
| 551 | stepToNextTag(); |
| 552 | procStrArrExpr(); |
| 553 | emitOpcode(VM::SETHASALL); |
| 554 | emitUInt(set_ref); |
| 555 | assert(name == "sethasall"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "sethasall"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"sethasall\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 556 | stepToNextTag(); |
| 557 | } else { |
| 558 | if (allow_fail) { |
| 559 | return false; |
| 560 | } |
| 561 | parseError("Expected a boolean expression."_u); |
| 562 | } |
| 563 | } |
| 564 | return true; |
| 565 | } |
| 566 | |
| 567 | void |
| 568 | MTXReader::procAddrExpr() |
| 569 | { |
| 570 | stepToTag(); |
| 571 | /* Self-closing tags */ |
| 572 | if (!tryProcArg(ADDREXPR)) { |
| 573 | if (name == "wrdaddr"_u) { |
| 574 | emitOpcode(VM::PUSHADDR); |
| 575 | stepPastSelfClosingTag("wrdaddr"_u); |
| 576 | /* Others */ |
| 577 | } else if (name == "addr-of-ints"_u) { |
| 578 | stepToNextTag(); |
| 579 | procIntExpr(); |
| 580 | procIntExpr(); |
| 581 | assert(name == "addr-of-ints"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "addr-of-ints"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"addr-of-ints\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 582 | stepToNextTag(); |
| 583 | } else if (name == "add"_u) { |
| 584 | stepToNextTag(); |
| 585 | procAddrExpr(); |
| 586 | procAddrExpr(); |
| 587 | assert(name == "add"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "add"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"add\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 588 | emitOpcode(VM::ADD2); |
| 589 | stepToNextTag(); |
| 590 | } else if (name == "adjust"_u) { |
| 591 | stepToNextTag(); |
| 592 | procAddrExpr(); |
| 593 | assert(name == "adjust"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "adjust"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"adjust\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 594 | emitOpcode(VM::ADJADDR); |
| 595 | stepToNextTag(); |
| 596 | } else if (name == "clamp"_u) { |
| 597 | stepToNextTag(); |
| 598 | procAddrExpr(); |
| 599 | assert(name == "clamp"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "clamp"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"clamp\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 600 | emitOpcode(VM::CLAMPADDR); |
| 601 | stepToNextTag(); |
| 602 | } else { |
| 603 | parseError("Expected an address expression."_u); |
| 604 | } |
| 605 | } |
| 606 | } |
| 607 | |
| 608 | bool |
| 609 | MTXReader::procWordoidArrExpr(bool allow_fail) |
| 610 | { |
| 611 | if (!tryProcArg(WRDARREXPR, true) |
| 612 | && !tryProcVar(VM::WRDARRVAL) |
| 613 | && !tryProcSlice(&MTXReader::procWordoidArrExpr)) { |
| 614 | if (name == "ex-wordoids"_u) { |
| 615 | stepToNextTag(); |
| 616 | procIntExpr(); |
| 617 | emitOpcode(VM::EXWRDARR); |
| 618 | assert(name == "ex-wordoids"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "ex-wordoids"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"ex-wordoids\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 619 | } else if (name == "for-each"_u) { |
| 620 | procForEach(WRDEXPR); |
| 621 | } else { |
| 622 | if (allow_fail) { |
| 623 | return false; |
| 624 | } |
| 625 | parseError("Expected a wordoid array expression."_u); |
| 626 | } |
| 627 | stepToNextTag(); |
| 628 | } |
| 629 | return true; |
| 630 | } |
| 631 | |
| 632 | bool |
| 633 | MTXReader::procWordoidExpr(bool allow_fail) |
| 634 | { |
| 635 | stepToTag(); |
| 636 | if (!tryProcArg(WRDEXPR, true) |
| 637 | && !tryProcVar(VM::WRDVAL) |
| 638 | && !tryProcSubscript(&MTXReader::procWordoidArrExpr)) { |
| 639 | if (name == "ex-wordoid"_u) { |
| 640 | stepToNextTag(); |
| 641 | procAddrExpr(); |
| 642 | emitOpcode(VM::GETWRD); |
| 643 | } else { |
| 644 | if (allow_fail) { |
| 645 | return false; |
| 646 | } |
| 647 | parseError("Expected a wordoid expression."_u); |
| 648 | } |
| 649 | assert(type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (type == XML_READER_TYPE_END_ELEMENT ) ? void (0) : __assert_fail ("type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 650 | stepToNextTag(); |
| 651 | } |
| 652 | return true; |
| 653 | } |
| 654 | |
| 655 | void |
| 656 | MTXReader::procPred() |
| 657 | { |
| 658 | stepToNextTag(); |
| 659 | procBoolExpr(); |
| 660 | assert(name == "pred"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "pred"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"pred\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 661 | emitOpcode(VM::DIEIFFALSE); |
| 662 | stepToNextTag(); |
| 663 | } |
| 664 | |
| 665 | size_t |
| 666 | MTXReader::getConstRef( |
| 667 | const UString &ref_attr, |
| 668 | const UString &lit_attr, |
| 669 | const UString &what, |
| 670 | VarNVMap &const_map, |
| 671 | size_t (MTXReader::*push_new)(std::string&), |
| 672 | bool& exists) |
| 673 | { |
| 674 | std::string const_name = attrib_str(ref_attr); |
| 675 | if (!const_name.empty()) { |
| 676 | exists = true; |
| 677 | VarNVMap::iterator sit = const_map.find(const_name); |
| 678 | if (sit == const_map.end()) { |
| 679 | parseError("No "_u + what + " named "_u + to_ustring(const_name.c_str())); |
| 680 | } |
| 681 | return sit->second; |
| 682 | } |
| 683 | std::string const_lit = attrib_str(lit_attr); |
| 684 | if (!const_lit.empty()) { |
| 685 | exists = true; |
| 686 | return (this->*push_new)(const_lit); |
| 687 | } |
| 688 | exists = false; |
| 689 | return 0; |
| 690 | } |
| 691 | |
| 692 | size_t |
| 693 | MTXReader::getSetRef(bool& exists) |
| 694 | { |
| 695 | return getConstRef("name"_u, "val"_u, "set"_u, set_names, &MTXReader::pushSetConst, exists); |
| 696 | } |
| 697 | |
| 698 | size_t |
| 699 | MTXReader::getSetRef() |
| 700 | { |
| 701 | bool has_attr; |
| 702 | size_t set_ref = getSetRef(has_attr); |
| 703 | if (!has_attr) { |
| 704 | parseError("Set required"_u); |
| 705 | } |
| 706 | return set_ref; |
| 707 | } |
| 708 | |
| 709 | size_t |
| 710 | MTXReader::getStrRef(bool& exists) |
| 711 | { |
| 712 | return getConstRef("name"_u, "val"_u, "string"_u, str_names, &MTXReader::pushStrConst, exists); |
| 713 | } |
| 714 | |
| 715 | size_t |
| 716 | MTXReader::getStrRef() |
| 717 | { |
| 718 | bool has_attr; |
| 719 | size_t str_ref = getStrRef(has_attr); |
| 720 | if (!has_attr) { |
| 721 | parseError("String required"_u); |
| 722 | } |
| 723 | return str_ref; |
| 724 | } |
| 725 | |
| 726 | int |
| 727 | MTXReader::getInt(const UString& attr_name, bool& exists) |
| 728 | { |
| 729 | std::string int_lit = attrib_str(attr_name); |
| 730 | if (!int_lit.empty()) { |
| 731 | exists = true; |
| 732 | int int_out; |
| 733 | stringstream int_ss(int_lit); |
| 734 | int_ss >> int_out; |
| 735 | return int_out; |
| 736 | } |
| 737 | exists = false; |
| 738 | return 0; |
| 739 | } |
| 740 | |
| 741 | int |
| 742 | MTXReader::getInt(bool& exists) |
| 743 | { |
| 744 | return getInt("val"_u, exists); |
| 745 | } |
| 746 | |
| 747 | int |
| 748 | MTXReader::getInt(const UString& attr_name) |
| 749 | { |
| 750 | bool has_attr; |
| 751 | int i = getInt(attr_name, has_attr); |
| 752 | if (!has_attr) { |
| 753 | parseError("String required"); |
| 754 | } |
| 755 | return i; |
| 756 | } |
| 757 | |
| 758 | int |
| 759 | MTXReader::getInt() |
| 760 | { |
| 761 | return getInt("val"_u); |
| 762 | } |
| 763 | |
| 764 | template<typename GetT, typename EmitT> |
| 765 | void |
| 766 | MTXReader::emitAttr( |
| 767 | std::string what, GetT (MTXReader::*getter)(bool&), void (MTXReader::*emitter)(EmitT)) |
| 768 | { |
| 769 | bool has_attr = false; |
| 770 | GetT val = (this->*getter)(has_attr); |
| 771 | if (!has_attr) { |
| 772 | parseError(what + " required"); |
| 773 | } |
| 774 | (this->*emitter)(val); |
| 775 | } |
| 776 | |
| 777 | void |
| 778 | MTXReader::getAndEmitStrRef() |
| 779 | { |
| 780 | emitAttr("String", &MTXReader::getStrRef, &MTXReader::emitUInt); |
| 781 | } |
| 782 | |
| 783 | void |
| 784 | MTXReader::getAndEmitSetRef() |
| 785 | { |
| 786 | emitAttr("Set", &MTXReader::getSetRef, &MTXReader::emitUInt); |
| 787 | } |
| 788 | |
| 789 | void |
| 790 | MTXReader::getAndEmitInt() |
| 791 | { |
| 792 | emitAttr("Integer", &MTXReader::getInt, &MTXReader::emitInt); |
| 793 | } |
| 794 | |
| 795 | void |
| 796 | MTXReader::procInst() |
| 797 | { |
| 798 | // XXX: There's no way to tell the difference between an empty and absent |
| 799 | // attribute with the current lttoolbox xml code |
| 800 | std::string op = attrib_str("opcode"_u); |
| 801 | std::transform(op.begin(), op.end(), op.begin(), ::toupper); |
| 802 | emitOpcode(VM::opcode_values[op]); |
| 803 | int val; |
| 804 | bool has_set_ref; |
| 805 | val = getSetRef(has_set_ref); |
Value stored to 'val' is never read | |
| 806 | bool has_str_ref; |
| 807 | val = getStrRef(has_str_ref); |
| 808 | bool has_int_lit; |
| 809 | val = getInt(has_int_lit); |
| 810 | int num_operands = has_set_ref + has_str_ref + has_int_lit; |
| 811 | if (num_operands > 1) { |
| 812 | parseError("Opcodes can have at most one operand."_u); |
| 813 | } else if (num_operands == 1) { |
| 814 | if (has_int_lit) { |
| 815 | emitInt(val); |
| 816 | } else { |
| 817 | emitUInt(val); |
| 818 | } |
| 819 | } |
| 820 | } |
| 821 | |
| 822 | void |
| 823 | MTXReader::procOut() |
| 824 | { |
| 825 | bool has_expr = false; |
| 826 | stepToNextTag(); |
| 827 | if (procStrExpr(true)) { |
| 828 | emitOpcode(VM::FCATSTR); |
| 829 | has_expr = true; |
| 830 | } |
| 831 | if (!has_expr && procBoolExpr(true)) { |
| 832 | emitOpcode(VM::FCATBOOL); |
| 833 | has_expr = true; |
| 834 | } |
| 835 | if (!has_expr && procIntExpr(true)) { |
| 836 | emitOpcode(VM::FCATINT); |
| 837 | has_expr = true; |
| 838 | } |
| 839 | if (!has_expr) { |
| 840 | parseError("Expected a string, bool or int expression."_u); |
| 841 | } |
| 842 | stepToTag(); |
| 843 | assert(name == "out"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "out"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ("name == \"out\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 844 | stepToNextTag(); |
| 845 | } |
| 846 | |
| 847 | void |
| 848 | MTXReader::procOutMany() |
| 849 | { |
| 850 | stepToNextTag(); |
| 851 | procStrArrExpr(); |
| 852 | emitOpcode(VM::FCATSTRARR); |
| 853 | assert(name == "out-many"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "out-many"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"out-many\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 854 | stepToNextTag(); |
| 855 | } |
| 856 | |
| 857 | void |
| 858 | MTXReader::printTmplDefn(const TemplateDefn &tmpl_defn) |
| 859 | { |
| 860 | PerceptronSpec::printFeature(std::cerr, tmpl_defn.first); |
| 861 | if (tmpl_defn.second.size() > 0) { |
| 862 | std::cerr << "Replacements:\n"_u; |
| 863 | TemplateReplacements::const_iterator it = tmpl_defn.second.begin(); |
| 864 | for (; it != tmpl_defn.second.end(); it++) { |
| 865 | std::cerr << "Index: "_u << it->first << " "_u; |
| 866 | printTypeExpr(it->second); |
| 867 | std::cerr << "\n"_u; |
| 868 | } |
| 869 | } |
| 870 | } |
| 871 | |
| 872 | void |
| 873 | MTXReader::printStackValueType(VM::StackValueType svt) |
| 874 | { |
| 875 | switch (svt) { |
| 876 | case VM::INTVAL: |
| 877 | std::cerr << "INT"; |
| 878 | break; |
| 879 | case VM::BVAL: |
| 880 | std::cerr << "BOOL"; |
| 881 | break; |
| 882 | case VM::STRVAL: |
| 883 | std::cerr << "STR"; |
| 884 | break; |
| 885 | case VM::STRARRVAL: |
| 886 | std::cerr << "STRARR"; |
| 887 | break; |
| 888 | case VM::WRDVAL: |
| 889 | std::cerr << "WRD"; |
| 890 | break; |
| 891 | case VM::WRDARRVAL: |
| 892 | std::cerr << "WRDARR"; |
| 893 | break; |
| 894 | default: |
| 895 | throw 1; |
| 896 | } |
| 897 | } |
| 898 | |
| 899 | void |
| 900 | MTXReader::printTypeExpr(ExprType expr_type) |
| 901 | { |
| 902 | switch (expr_type) { |
| 903 | case VOIDEXPR: |
| 904 | std::cerr << "VOID"; |
| 905 | break; |
| 906 | case INTEXPR: |
| 907 | std::cerr << "INT"; |
| 908 | break; |
| 909 | case BEXPR: |
| 910 | std::cerr << "BOOL"; |
| 911 | break; |
| 912 | case STREXPR: |
| 913 | std::cerr << "STR"; |
| 914 | procStrExpr(); |
| 915 | break; |
| 916 | case STRARREXPR: |
| 917 | std::cerr << "STRARR"; |
| 918 | break; |
| 919 | case WRDEXPR: |
| 920 | std::cerr << "WRD"; |
| 921 | break; |
| 922 | case WRDARREXPR: |
| 923 | std::cerr << "WRDARR"; |
| 924 | break; |
| 925 | case ADDREXPR: |
| 926 | std::cerr << "ADDR"; |
| 927 | break; |
| 928 | default: |
| 929 | throw 1; |
| 930 | } |
| 931 | } |
| 932 | |
| 933 | void |
| 934 | MTXReader::procTypeExpr(ExprType expr_type) |
| 935 | { |
| 936 | switch (expr_type) { |
| 937 | case VOIDEXPR: |
| 938 | procVoidExpr(); |
| 939 | break; |
| 940 | case INTEXPR: |
| 941 | procIntExpr(); |
| 942 | break; |
| 943 | case BEXPR: |
| 944 | procBoolExpr(); |
| 945 | break; |
| 946 | case STREXPR: |
| 947 | procStrExpr(); |
| 948 | break; |
| 949 | case STRARREXPR: |
| 950 | procStrArrExpr(); |
| 951 | break; |
| 952 | case WRDEXPR: |
| 953 | procWordoidExpr(); |
| 954 | break; |
| 955 | case WRDARREXPR: |
| 956 | procWordoidArrExpr(); |
| 957 | break; |
| 958 | case ADDREXPR: |
| 959 | procAddrExpr(); |
| 960 | break; |
| 961 | default: |
| 962 | throw 1; |
| 963 | } |
| 964 | } |
| 965 | |
| 966 | void |
| 967 | MTXReader::procForEach(ExprType expr_type) |
| 968 | { |
| 969 | std::string var_name = attrib_str("as"_u); |
| 970 | if (var_name.empty()) { |
| 971 | parseError("'as' attribute required for for-each."_u); |
| 972 | } |
| 973 | size_t slot_idx = slot_counter++; |
| 974 | slot_names[var_name] = slot_idx; |
| 975 | bool has_expr = false; |
| 976 | stepToNextTag(); |
| 977 | if (procStrArrExpr(true)) { |
| 978 | slot_types.push_back(VM::STRVAL); |
| 979 | has_expr = true; |
| 980 | } |
| 981 | if (!has_expr && procWordoidArrExpr(true)) { |
| 982 | slot_types.push_back(VM::WRDVAL); |
| 983 | has_expr = true; |
| 984 | } |
| 985 | if (!has_expr) { |
| 986 | parseError("Expected a string array or wordoid array expression."_u); |
| 987 | } |
| 988 | |
| 989 | emitOpcode(VM::FOREACHINIT); |
| 990 | size_t for_each_begin = cur_feat->size(); |
| 991 | emitOpcode(VM::FOREACH); |
| 992 | emitUInt(slot_idx); |
| 993 | size_t begin_offset_placeholder = cur_feat->size(); |
| 994 | emitUInt(0); // offset placeholder |
| 995 | size_t for_each_body_begin = cur_feat->size(); |
| 996 | |
| 997 | switch (expr_type) { |
| 998 | case VOIDEXPR: |
| 999 | procVoidExpr(); |
| 1000 | break; |
| 1001 | case STREXPR: |
| 1002 | procStrExpr(); |
| 1003 | break; |
| 1004 | case WRDEXPR: |
| 1005 | procWordoidExpr(); |
| 1006 | break; |
| 1007 | default: |
| 1008 | throw 1; |
| 1009 | break; |
| 1010 | } |
| 1011 | assert(type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (type == XML_READER_TYPE_END_ELEMENT ) ? void (0) : __assert_fail ("type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 1012 | |
| 1013 | size_t end_for_each_addr = cur_feat->size(); |
| 1014 | emitOpcode(VM::ENDFOREACH); |
| 1015 | emitInt(end_for_each_addr + 2 - for_each_begin); // offset |
| 1016 | pokeBytecode(begin_offset_placeholder, (VM::Bytecode){ |
| 1017 | .uintbyte=(unsigned char)(end_for_each_addr + 2 - for_each_body_begin)}); |
| 1018 | } |
| 1019 | |
| 1020 | bool |
| 1021 | MTXReader::procVoidExpr(bool allow_fail) |
| 1022 | { |
| 1023 | stepToTag(); |
| 1024 | if (name == "pred"_u) { |
| 1025 | procPred(); |
| 1026 | } else if (name == "out"_u) { |
| 1027 | procOut(); |
| 1028 | } else if (name == "out-many"_u) { |
| 1029 | procOutMany(); |
| 1030 | } else if (name == "for-each"_u) { |
| 1031 | procForEach(VOIDEXPR); |
| 1032 | } else if (name == "inst"_u) { |
| 1033 | procInst(); |
| 1034 | } else { |
| 1035 | if (allow_fail) { |
| 1036 | return false; |
| 1037 | } |
| 1038 | parseError("Expected a void expression."_u); |
| 1039 | } |
| 1040 | return true; |
| 1041 | } |
| 1042 | |
| 1043 | void |
| 1044 | MTXReader::procDefMacro() |
| 1045 | { |
| 1046 | in_global_defn = true; |
| 1047 | slot_counter = 0; |
| 1048 | template_defns.push_back(make_pair(VM::FeatureDefn(), TemplateReplacements())); |
| 1049 | cur_feat = &template_defns.back().first; |
| 1050 | cur_replacements = &template_defns.back().second; |
| 1051 | |
| 1052 | std::string var_name = attrib_str("as"_u); |
| 1053 | if (var_name.empty()) { |
| 1054 | parseError("'as' attribute required for def-macro."_u); |
| 1055 | } |
| 1056 | template_slot_names[var_name] = template_slot_counter; |
| 1057 | |
| 1058 | template_arg_names.clear(); |
| 1059 | std::string args = attrib_str("args"_u); |
| 1060 | std::istringstream args_ss(args); |
| 1061 | size_t arg_i = 0; |
| 1062 | for (; !args_ss.eof(); arg_i++) { |
| 1063 | std::string arg_name; |
| 1064 | args_ss >> arg_name; |
| 1065 | if (arg_name.empty()) { |
| 1066 | break; |
| 1067 | } |
| 1068 | template_arg_names[arg_name] = arg_i; |
| 1069 | } |
| 1070 | stepToNextTag(); |
| 1071 | |
| 1072 | bool has_expr = false; |
| 1073 | if (procIntExpr(true)) { |
| 1074 | template_slot_types.push_back(VM::INTVAL); |
| 1075 | has_expr = true; |
| 1076 | } |
| 1077 | if (!has_expr && procBoolExpr(true)) { |
| 1078 | template_slot_types.push_back(VM::BVAL); |
| 1079 | has_expr = true; |
| 1080 | } |
| 1081 | if (!has_expr && procStrExpr(true)) { |
| 1082 | template_slot_types.push_back(VM::STRVAL); |
| 1083 | has_expr = true; |
| 1084 | } |
| 1085 | if (!has_expr && procStrArrExpr(true)) { |
| 1086 | template_slot_types.push_back(VM::STRARRVAL); |
| 1087 | has_expr = true; |
| 1088 | } |
| 1089 | if (!has_expr && procWordoidArrExpr(true)) { |
| 1090 | template_slot_types.push_back(VM::WRDARRVAL); |
| 1091 | has_expr = true; |
| 1092 | } |
| 1093 | if (!has_expr && procWordoidExpr(true)) { |
| 1094 | template_slot_types.push_back(VM::WRDVAL); |
| 1095 | has_expr = true; |
| 1096 | } |
| 1097 | if (!has_expr) { |
| 1098 | parseError("Expected a non-void expression."_u); |
| 1099 | } |
| 1100 | assert(name == "def-macro"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "def-macro"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"def-macro\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 1101 | stepToNextTag(); |
| 1102 | |
| 1103 | template_slot_counter++; |
| 1104 | in_global_defn = false; |
| 1105 | } |
| 1106 | |
| 1107 | void |
| 1108 | MTXReader::procFeat() |
| 1109 | { |
| 1110 | slot_counter = 0; |
| 1111 | spec.features.push_back(VM::FeatureDefn()); |
| 1112 | cur_feat = &spec.features.back(); |
| 1113 | stepToNextTag(); |
| 1114 | while (type != XML_READER_TYPE_END_ELEMENT) { |
| 1115 | procVoidExpr(); |
| 1116 | } |
| 1117 | assert(name == "feat"_u)(static_cast <bool> (name == "feat"_u) ? void (0) : __assert_fail ("name == \"feat\"_u", __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__)); |
| 1118 | stepToNextTag(); |
| 1119 | } |
| 1120 | |
| 1121 | void |
| 1122 | MTXReader::procFeats() |
| 1123 | { |
| 1124 | stepToNextTag(); |
| 1125 | while (type != XML_READER_TYPE_END_ELEMENT) { |
| 1126 | if (name == "feat"_u) { |
| 1127 | procFeat(); |
| 1128 | } else { |
| 1129 | unexpectedTag(); |
| 1130 | } |
| 1131 | } |
| 1132 | assert(name == "feats"_u)(static_cast <bool> (name == "feats"_u) ? void (0) : __assert_fail ("name == \"feats\"_u", __builtin_FILE (), __builtin_LINE () , __extension__ __PRETTY_FUNCTION__)); |
| 1133 | stepToNextTag(); |
| 1134 | } |
| 1135 | |
| 1136 | void |
| 1137 | MTXReader::printTmplDefns() |
| 1138 | { |
| 1139 | std::vector<TemplateDefn>::const_iterator it = template_defns.begin(); |
| 1140 | for (; it != template_defns.end(); it++) { |
| 1141 | std::cerr << " Macro " << it - template_defns.begin() << "\n"; |
| 1142 | printTmplDefn(*it); |
| 1143 | } |
| 1144 | } |
| 1145 | |
| 1146 | void |
| 1147 | MTXReader::parse() |
| 1148 | { |
| 1149 | xmlTextReaderSetParserProp(reader, XML_PARSER_SUBST_ENTITIES, true); |
| 1150 | stepToNextTag(); |
| 1151 | if (type == XML_READER_TYPE_DOCUMENT_TYPE) { |
| 1152 | stepToNextTag(); |
| 1153 | } |
| 1154 | if (name != "metatag"_u) { |
| 1155 | parseError("expected <metatag> tag"_u); |
| 1156 | } |
| 1157 | stepToNextTag(); |
| 1158 | if (name == "coarse-tags"_u) { |
| 1159 | procCoarseTags(); |
| 1160 | } |
| 1161 | if (name == "beam-width"_u) { |
| 1162 | size_t val; |
| 1163 | std::istringstream val_ss(attrib_str("val"_u)); |
| 1164 | val_ss >> val; |
| 1165 | spec.beam_width = val; |
| 1166 | } else { |
| 1167 | spec.beam_width = 4; |
| 1168 | } |
| 1169 | if (name == "defns"_u) { |
| 1170 | procDefns(); |
| 1171 | } |
| 1172 | if (name == "global-pred"_u) { |
| 1173 | procGlobalPred(); |
| 1174 | } |
| 1175 | if (name == "feats"_u) { |
| 1176 | procFeats(); |
| 1177 | } |
| 1178 | assert(name == "metatag"_u && type == XML_READER_TYPE_END_ELEMENT)(static_cast <bool> (name == "metatag"_u && type == XML_READER_TYPE_END_ELEMENT) ? void (0) : __assert_fail ( "name == \"metatag\"_u && type == XML_READER_TYPE_END_ELEMENT" , __builtin_FILE (), __builtin_LINE (), __extension__ __PRETTY_FUNCTION__ )); |
| 1179 | } |
| 1180 | } |