Reland sticky regexps https://codereview.chromium.org/567313003/

src/jsregexp.cc

 // Copyright 2012 the V8 project authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "src/v8.h"
 
 #include "src/ast.h"
 #include "src/base/platform/platform.h"
 #include "src/compilation-cache.h"
 #include "src/compiler.h"
@@ skipping 52 matching lines @@
     switch (str->Get(i)) {
       case 'i':
         flags |= JSRegExp::IGNORE_CASE;
         break;
       case 'g':
         flags |= JSRegExp::GLOBAL;
         break;
       case 'm':
         flags |= JSRegExp::MULTILINE;
         break;
+      case 'y':
+        if (FLAG_harmony_regexps) flags |= JSRegExp::STICKY;
+        break;
     }
   }
   return JSRegExp::Flags(flags);
 }
 
 
 MUST_USE_RESULT
 static inline MaybeHandle<Object> ThrowRegExpException(
     Handle<JSRegExp> re,
     Handle<String> pattern,
@@ skipping 95 matching lines @@
     return ThrowRegExpException(re,
                                 pattern,
                                 parse_result.error,
                                 "malformed_regexp");
   }
 
   bool has_been_compiled = false;
 
   if (parse_result.simple &&
       !flags.is_ignore_case() &&
+      !flags.is_sticky() &&
       !HasFewDifferentCharacters(pattern)) {
     // Parse-tree is a single atom that is equal to the pattern.
     AtomCompile(re, pattern, flags, pattern);
     has_been_compiled = true;
   } else if (parse_result.tree->IsAtom() &&
       !flags.is_ignore_case() &&
+      !flags.is_sticky() &&
       parse_result.capture_count == 0) {
     RegExpAtom* atom = parse_result.tree->AsAtom();
     Vector<const uc16> atom_pattern = atom->data();
     Handle<String> atom_string;
     ASSIGN_RETURN_ON_EXCEPTION(
         isolate, atom_string,
         isolate->factory()->NewStringFromTwoByte(atom_pattern),
         Object);
     if (!HasFewDifferentCharacters(atom_string)) {
       AtomCompile(re, pattern, flags, atom_string);
@@ skipping 219 matching lines @@
     // Throw an exception if we fail to parse the pattern.
     // THIS SHOULD NOT HAPPEN. We already pre-parsed it successfully once.
     USE(ThrowRegExpException(re,
                              pattern,
                              compile_data.error,
                              "malformed_regexp"));
     return false;
   }
   RegExpEngine::CompilationResult result = RegExpEngine::Compile(
       &compile_data, flags.is_ignore_case(), flags.is_global(),
-      flags.is_multiline(), pattern, sample_subject, is_one_byte, &zone);
+      flags.is_multiline(), flags.is_sticky(), pattern, sample_subject,
+      is_one_byte, &zone);
   if (result.error_message != NULL) {
     // Unable to compile regexp.
     Handle<String> error_message = isolate->factory()->NewStringFromUtf8(
         CStrVector(result.error_message)).ToHandleChecked();
     CreateRegExpErrorObjectAndThrow(re, error_message, isolate);
     return false;
   }
 
   Handle<FixedArray> data = Handle<FixedArray>(FixedArray::cast(re->data()));
   data->set(JSRegExp::code_index(is_one_byte), result.code);
@@ skipping 5576 matching lines @@
 
 
 void DispatchTableConstructor::VisitAction(ActionNode* that) {
   RegExpNode* target = that->on_success();
   target->Accept(this);
 }
 
 
 RegExpEngine::CompilationResult RegExpEngine::Compile(
     RegExpCompileData* data, bool ignore_case, bool is_global,
-    bool is_multiline, Handle<String> pattern, Handle<String> sample_subject,
-    bool is_one_byte, Zone* zone) {
+    bool is_multiline, bool is_sticky, Handle<String> pattern,
+    Handle<String> sample_subject, bool is_one_byte, Zone* zone) {
   if ((data->capture_count + 1) * 2 - 1 > RegExpMacroAssembler::kMaxRegister) {
     return IrregexpRegExpTooBig(zone->isolate());
   }
   RegExpCompiler compiler(data->capture_count, ignore_case, is_one_byte, zone);
 
   // Sample some characters from the middle of the string.
   static const int kSampleSize = 128;
 
   sample_subject = String::Flatten(sample_subject);
   int chars_sampled = 0;
   int half_way = (sample_subject->length() - kSampleSize) / 2;
   for (int i = Max(0, half_way);
        i < sample_subject->length() && chars_sampled < kSampleSize;
        i++, chars_sampled++) {
     compiler.frequency_collator()->CountCharacter(sample_subject->Get(i));
   }
 
   // Wrap the body of the regexp in capture #0.
   RegExpNode* captured_body = RegExpCapture::ToNode(data->tree,
                                                     0,
                                                     &compiler,
                                                     compiler.accept());
   RegExpNode* node = captured_body;
   bool is_end_anchored = data->tree->IsAnchoredAtEnd();
   bool is_start_anchored = data->tree->IsAnchoredAtStart();
   int max_length = data->tree->max_match();
-  if (!is_start_anchored) {
+  if (!is_start_anchored && !is_sticky) {
     // Add a .*? at the beginning, outside the body capture, unless
-    // this expression is anchored at the beginning.
+    // this expression is anchored at the beginning or sticky.
     RegExpNode* loop_node =
         RegExpQuantifier::ToNode(0,
                                  RegExpTree::kInfinity,
                                  false,
                                  new(zone) RegExpCharacterClass('*'),
                                  &compiler,
                                  captured_body,
                                  data->contains_anchor);
 
     if (data->contains_anchor) {
@@ skipping 82 matching lines @@
   }
 
   return compiler.Assemble(&macro_assembler,
                            node,
                            data->capture_count,
                            pattern);
 }
 
 
 }}  // namespace v8::internal