Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 1 | // Copyright 2016 the V8 project authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style license that can be |
| 3 | // found in the LICENSE file. |
| 4 | |
| 5 | #include "src/ostreams.h" |
| 6 | #include "src/regexp/regexp-ast.h" |
| 7 | |
| 8 | namespace v8 { |
| 9 | namespace internal { |
| 10 | |
| 11 | #define MAKE_ACCEPT(Name) \ |
| 12 | void* RegExp##Name::Accept(RegExpVisitor* visitor, void* data) { \ |
| 13 | return visitor->Visit##Name(this, data); \ |
| 14 | } |
| 15 | FOR_EACH_REG_EXP_TREE_TYPE(MAKE_ACCEPT) |
| 16 | #undef MAKE_ACCEPT |
| 17 | |
| 18 | #define MAKE_TYPE_CASE(Name) \ |
| 19 | RegExp##Name* RegExpTree::As##Name() { return NULL; } \ |
| 20 | bool RegExpTree::Is##Name() { return false; } |
| 21 | FOR_EACH_REG_EXP_TREE_TYPE(MAKE_TYPE_CASE) |
| 22 | #undef MAKE_TYPE_CASE |
| 23 | |
| 24 | #define MAKE_TYPE_CASE(Name) \ |
| 25 | RegExp##Name* RegExp##Name::As##Name() { return this; } \ |
| 26 | bool RegExp##Name::Is##Name() { return true; } |
| 27 | FOR_EACH_REG_EXP_TREE_TYPE(MAKE_TYPE_CASE) |
| 28 | #undef MAKE_TYPE_CASE |
| 29 | |
| 30 | |
| 31 | static Interval ListCaptureRegisters(ZoneList<RegExpTree*>* children) { |
| 32 | Interval result = Interval::Empty(); |
| 33 | for (int i = 0; i < children->length(); i++) |
| 34 | result = result.Union(children->at(i)->CaptureRegisters()); |
| 35 | return result; |
| 36 | } |
| 37 | |
| 38 | |
| 39 | Interval RegExpAlternative::CaptureRegisters() { |
| 40 | return ListCaptureRegisters(nodes()); |
| 41 | } |
| 42 | |
| 43 | |
| 44 | Interval RegExpDisjunction::CaptureRegisters() { |
| 45 | return ListCaptureRegisters(alternatives()); |
| 46 | } |
| 47 | |
| 48 | |
| 49 | Interval RegExpLookaround::CaptureRegisters() { |
| 50 | return body()->CaptureRegisters(); |
| 51 | } |
| 52 | |
| 53 | |
| 54 | Interval RegExpCapture::CaptureRegisters() { |
| 55 | Interval self(StartRegister(index()), EndRegister(index())); |
| 56 | return self.Union(body()->CaptureRegisters()); |
| 57 | } |
| 58 | |
| 59 | |
| 60 | Interval RegExpQuantifier::CaptureRegisters() { |
| 61 | return body()->CaptureRegisters(); |
| 62 | } |
| 63 | |
| 64 | |
| 65 | bool RegExpAssertion::IsAnchoredAtStart() { |
| 66 | return assertion_type() == RegExpAssertion::START_OF_INPUT; |
| 67 | } |
| 68 | |
| 69 | |
| 70 | bool RegExpAssertion::IsAnchoredAtEnd() { |
| 71 | return assertion_type() == RegExpAssertion::END_OF_INPUT; |
| 72 | } |
| 73 | |
| 74 | |
| 75 | bool RegExpAlternative::IsAnchoredAtStart() { |
| 76 | ZoneList<RegExpTree*>* nodes = this->nodes(); |
| 77 | for (int i = 0; i < nodes->length(); i++) { |
| 78 | RegExpTree* node = nodes->at(i); |
| 79 | if (node->IsAnchoredAtStart()) { |
| 80 | return true; |
| 81 | } |
| 82 | if (node->max_match() > 0) { |
| 83 | return false; |
| 84 | } |
| 85 | } |
| 86 | return false; |
| 87 | } |
| 88 | |
| 89 | |
| 90 | bool RegExpAlternative::IsAnchoredAtEnd() { |
| 91 | ZoneList<RegExpTree*>* nodes = this->nodes(); |
| 92 | for (int i = nodes->length() - 1; i >= 0; i--) { |
| 93 | RegExpTree* node = nodes->at(i); |
| 94 | if (node->IsAnchoredAtEnd()) { |
| 95 | return true; |
| 96 | } |
| 97 | if (node->max_match() > 0) { |
| 98 | return false; |
| 99 | } |
| 100 | } |
| 101 | return false; |
| 102 | } |
| 103 | |
| 104 | |
| 105 | bool RegExpDisjunction::IsAnchoredAtStart() { |
| 106 | ZoneList<RegExpTree*>* alternatives = this->alternatives(); |
| 107 | for (int i = 0; i < alternatives->length(); i++) { |
| 108 | if (!alternatives->at(i)->IsAnchoredAtStart()) return false; |
| 109 | } |
| 110 | return true; |
| 111 | } |
| 112 | |
| 113 | |
| 114 | bool RegExpDisjunction::IsAnchoredAtEnd() { |
| 115 | ZoneList<RegExpTree*>* alternatives = this->alternatives(); |
| 116 | for (int i = 0; i < alternatives->length(); i++) { |
| 117 | if (!alternatives->at(i)->IsAnchoredAtEnd()) return false; |
| 118 | } |
| 119 | return true; |
| 120 | } |
| 121 | |
| 122 | |
| 123 | bool RegExpLookaround::IsAnchoredAtStart() { |
| 124 | return is_positive() && type() == LOOKAHEAD && body()->IsAnchoredAtStart(); |
| 125 | } |
| 126 | |
| 127 | |
| 128 | bool RegExpCapture::IsAnchoredAtStart() { return body()->IsAnchoredAtStart(); } |
| 129 | |
| 130 | |
| 131 | bool RegExpCapture::IsAnchoredAtEnd() { return body()->IsAnchoredAtEnd(); } |
| 132 | |
| 133 | |
| 134 | // Convert regular expression trees to a simple sexp representation. |
| 135 | // This representation should be different from the input grammar |
| 136 | // in as many cases as possible, to make it more difficult for incorrect |
| 137 | // parses to look as correct ones which is likely if the input and |
| 138 | // output formats are alike. |
| 139 | class RegExpUnparser final : public RegExpVisitor { |
| 140 | public: |
| 141 | RegExpUnparser(std::ostream& os, Zone* zone) : os_(os), zone_(zone) {} |
| 142 | void VisitCharacterRange(CharacterRange that); |
| 143 | #define MAKE_CASE(Name) void* Visit##Name(RegExp##Name*, void* data) override; |
| 144 | FOR_EACH_REG_EXP_TREE_TYPE(MAKE_CASE) |
| 145 | #undef MAKE_CASE |
| 146 | private: |
| 147 | std::ostream& os_; |
| 148 | Zone* zone_; |
| 149 | }; |
| 150 | |
| 151 | |
| 152 | void* RegExpUnparser::VisitDisjunction(RegExpDisjunction* that, void* data) { |
| 153 | os_ << "(|"; |
| 154 | for (int i = 0; i < that->alternatives()->length(); i++) { |
| 155 | os_ << " "; |
| 156 | that->alternatives()->at(i)->Accept(this, data); |
| 157 | } |
| 158 | os_ << ")"; |
| 159 | return NULL; |
| 160 | } |
| 161 | |
| 162 | |
| 163 | void* RegExpUnparser::VisitAlternative(RegExpAlternative* that, void* data) { |
| 164 | os_ << "(:"; |
| 165 | for (int i = 0; i < that->nodes()->length(); i++) { |
| 166 | os_ << " "; |
| 167 | that->nodes()->at(i)->Accept(this, data); |
| 168 | } |
| 169 | os_ << ")"; |
| 170 | return NULL; |
| 171 | } |
| 172 | |
| 173 | |
| 174 | void RegExpUnparser::VisitCharacterRange(CharacterRange that) { |
Ben Murdoch | 097c5b2 | 2016-05-18 11:27:45 +0100 | [diff] [blame] | 175 | os_ << AsUC32(that.from()); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 176 | if (!that.IsSingleton()) { |
Ben Murdoch | 097c5b2 | 2016-05-18 11:27:45 +0100 | [diff] [blame] | 177 | os_ << "-" << AsUC32(that.to()); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 178 | } |
| 179 | } |
| 180 | |
| 181 | |
| 182 | void* RegExpUnparser::VisitCharacterClass(RegExpCharacterClass* that, |
| 183 | void* data) { |
| 184 | if (that->is_negated()) os_ << "^"; |
| 185 | os_ << "["; |
| 186 | for (int i = 0; i < that->ranges(zone_)->length(); i++) { |
| 187 | if (i > 0) os_ << " "; |
| 188 | VisitCharacterRange(that->ranges(zone_)->at(i)); |
| 189 | } |
| 190 | os_ << "]"; |
| 191 | return NULL; |
| 192 | } |
| 193 | |
| 194 | |
| 195 | void* RegExpUnparser::VisitAssertion(RegExpAssertion* that, void* data) { |
| 196 | switch (that->assertion_type()) { |
| 197 | case RegExpAssertion::START_OF_INPUT: |
| 198 | os_ << "@^i"; |
| 199 | break; |
| 200 | case RegExpAssertion::END_OF_INPUT: |
| 201 | os_ << "@$i"; |
| 202 | break; |
| 203 | case RegExpAssertion::START_OF_LINE: |
| 204 | os_ << "@^l"; |
| 205 | break; |
| 206 | case RegExpAssertion::END_OF_LINE: |
| 207 | os_ << "@$l"; |
| 208 | break; |
| 209 | case RegExpAssertion::BOUNDARY: |
| 210 | os_ << "@b"; |
| 211 | break; |
| 212 | case RegExpAssertion::NON_BOUNDARY: |
| 213 | os_ << "@B"; |
| 214 | break; |
| 215 | } |
| 216 | return NULL; |
| 217 | } |
| 218 | |
| 219 | |
| 220 | void* RegExpUnparser::VisitAtom(RegExpAtom* that, void* data) { |
| 221 | os_ << "'"; |
| 222 | Vector<const uc16> chardata = that->data(); |
| 223 | for (int i = 0; i < chardata.length(); i++) { |
| 224 | os_ << AsUC16(chardata[i]); |
| 225 | } |
| 226 | os_ << "'"; |
| 227 | return NULL; |
| 228 | } |
| 229 | |
| 230 | |
| 231 | void* RegExpUnparser::VisitText(RegExpText* that, void* data) { |
| 232 | if (that->elements()->length() == 1) { |
| 233 | that->elements()->at(0).tree()->Accept(this, data); |
| 234 | } else { |
| 235 | os_ << "(!"; |
| 236 | for (int i = 0; i < that->elements()->length(); i++) { |
| 237 | os_ << " "; |
| 238 | that->elements()->at(i).tree()->Accept(this, data); |
| 239 | } |
| 240 | os_ << ")"; |
| 241 | } |
| 242 | return NULL; |
| 243 | } |
| 244 | |
| 245 | |
| 246 | void* RegExpUnparser::VisitQuantifier(RegExpQuantifier* that, void* data) { |
| 247 | os_ << "(# " << that->min() << " "; |
| 248 | if (that->max() == RegExpTree::kInfinity) { |
| 249 | os_ << "- "; |
| 250 | } else { |
| 251 | os_ << that->max() << " "; |
| 252 | } |
| 253 | os_ << (that->is_greedy() ? "g " : that->is_possessive() ? "p " : "n "); |
| 254 | that->body()->Accept(this, data); |
| 255 | os_ << ")"; |
| 256 | return NULL; |
| 257 | } |
| 258 | |
| 259 | |
| 260 | void* RegExpUnparser::VisitCapture(RegExpCapture* that, void* data) { |
| 261 | os_ << "(^ "; |
| 262 | that->body()->Accept(this, data); |
| 263 | os_ << ")"; |
| 264 | return NULL; |
| 265 | } |
| 266 | |
| 267 | |
| 268 | void* RegExpUnparser::VisitLookaround(RegExpLookaround* that, void* data) { |
| 269 | os_ << "("; |
| 270 | os_ << (that->type() == RegExpLookaround::LOOKAHEAD ? "->" : "<-"); |
| 271 | os_ << (that->is_positive() ? " + " : " - "); |
| 272 | that->body()->Accept(this, data); |
| 273 | os_ << ")"; |
| 274 | return NULL; |
| 275 | } |
| 276 | |
| 277 | |
| 278 | void* RegExpUnparser::VisitBackReference(RegExpBackReference* that, |
| 279 | void* data) { |
| 280 | os_ << "(<- " << that->index() << ")"; |
| 281 | return NULL; |
| 282 | } |
| 283 | |
| 284 | |
| 285 | void* RegExpUnparser::VisitEmpty(RegExpEmpty* that, void* data) { |
| 286 | os_ << '%'; |
| 287 | return NULL; |
| 288 | } |
| 289 | |
| 290 | |
| 291 | std::ostream& RegExpTree::Print(std::ostream& os, Zone* zone) { // NOLINT |
| 292 | RegExpUnparser unparser(os, zone); |
| 293 | Accept(&unparser, NULL); |
| 294 | return os; |
| 295 | } |
| 296 | |
| 297 | |
| 298 | RegExpDisjunction::RegExpDisjunction(ZoneList<RegExpTree*>* alternatives) |
| 299 | : alternatives_(alternatives) { |
| 300 | DCHECK(alternatives->length() > 1); |
| 301 | RegExpTree* first_alternative = alternatives->at(0); |
| 302 | min_match_ = first_alternative->min_match(); |
| 303 | max_match_ = first_alternative->max_match(); |
| 304 | for (int i = 1; i < alternatives->length(); i++) { |
| 305 | RegExpTree* alternative = alternatives->at(i); |
| 306 | min_match_ = Min(min_match_, alternative->min_match()); |
| 307 | max_match_ = Max(max_match_, alternative->max_match()); |
| 308 | } |
| 309 | } |
| 310 | |
| 311 | |
| 312 | static int IncreaseBy(int previous, int increase) { |
| 313 | if (RegExpTree::kInfinity - previous < increase) { |
| 314 | return RegExpTree::kInfinity; |
| 315 | } else { |
| 316 | return previous + increase; |
| 317 | } |
| 318 | } |
| 319 | |
| 320 | |
| 321 | RegExpAlternative::RegExpAlternative(ZoneList<RegExpTree*>* nodes) |
| 322 | : nodes_(nodes) { |
| 323 | DCHECK(nodes->length() > 1); |
| 324 | min_match_ = 0; |
| 325 | max_match_ = 0; |
| 326 | for (int i = 0; i < nodes->length(); i++) { |
| 327 | RegExpTree* node = nodes->at(i); |
| 328 | int node_min_match = node->min_match(); |
| 329 | min_match_ = IncreaseBy(min_match_, node_min_match); |
| 330 | int node_max_match = node->max_match(); |
| 331 | max_match_ = IncreaseBy(max_match_, node_max_match); |
| 332 | } |
| 333 | } |
| 334 | |
| 335 | |
| 336 | } // namespace internal |
| 337 | } // namespace v8 |