Index: src/scanner.h |
diff --git a/src/scanner.h b/src/scanner.h |
index d7328085b79245d97b807ff6f7e248d7a6ed207d..3cefc833ac3323c536ca98015fdfbee1522060d5 100644 |
--- a/src/scanner.h |
+++ b/src/scanner.h |
@@ -34,6 +34,8 @@ |
#include "char-predicates.h" |
#include "checks.h" |
#include "globals.h" |
+#include "hashmap.h" |
+#include "list.h" |
#include "token.h" |
#include "unicode-inl.h" |
#include "utils.h" |
@@ -121,9 +123,10 @@ class Utf16CharacterStream { |
}; |
-class UnicodeCache { |
// --------------------------------------------------------------------- |
// Caching predicates used by scanners. |
+ |
+class UnicodeCache { |
public: |
UnicodeCache() {} |
typedef unibrow::Utf8Decoder<512> Utf8Decoder; |
@@ -148,6 +151,56 @@ class UnicodeCache { |
}; |
+// --------------------------------------------------------------------- |
+// DuplicateFinder discovers duplicate symbols. |
+ |
+class DuplicateFinder { |
+ public: |
+ explicit DuplicateFinder(UnicodeCache* constants) |
+ : unicode_constants_(constants), |
+ backing_store_(16), |
+ map_(&Match) { } |
+ |
+ int AddAsciiSymbol(Vector<const char> key, int value); |
+ int AddUtf16Symbol(Vector<const uint16_t> key, int value); |
+ // Add a a number literal by converting it (if necessary) |
+ // to the string that ToString(ToNumber(literal)) would generate. |
+ // and then adding that string with AddAsciiSymbol. |
+ // This string is the actual value used as key in an object literal, |
+ // and the one that must be different from the other keys. |
+ int AddNumber(Vector<const char> key, int value); |
+ |
+ private: |
+ int AddSymbol(Vector<const byte> key, bool is_ascii, int value); |
+ // Backs up the key and its length in the backing store. |
+ // The backup is stored with a base 127 encoding of the |
+ // length (plus a bit saying whether the string is ASCII), |
+ // followed by the bytes of the key. |
+ byte* BackupKey(Vector<const byte> key, bool is_ascii); |
+ |
+ // Compare two encoded keys (both pointing into the backing store) |
+ // for having the same base-127 encoded lengths and ASCII-ness, |
+ // and then having the same 'length' bytes following. |
+ static bool Match(void* first, void* second); |
+ // Creates a hash from a sequence of bytes. |
+ static uint32_t Hash(Vector<const byte> key, bool is_ascii); |
+ // Checks whether a string containing a JS number is its canonical |
+ // form. |
+ static bool IsNumberCanonical(Vector<const char> key); |
+ |
+ // Size of buffer. Sufficient for using it to call DoubleToCString in |
+ // from conversions.h. |
+ static const int kBufferSize = 100; |
+ |
+ UnicodeCache* unicode_constants_; |
+ // Backing store used to store strings used as hashmap keys. |
+ SequenceCollector<unsigned char> backing_store_; |
+ HashMap map_; |
+ // Buffer used for string->number->canonical string conversions. |
+ char number_buffer_[kBufferSize]; |
+}; |
+ |
+ |
// ---------------------------------------------------------------------------- |
// LiteralBuffer - Collector of chars of literals. |