Optimize phishing page term feature extraction.

base/string_piece.h

Index: base/string_piece.h
diff --git a/base/string_piece.h b/base/string_piece.h
index 1077ac145018831022624289de851cded7981d77..278c7b692ae4750673dee32abcc2cc69c1631d3b 100644
--- a/base/string_piece.h
+++ b/base/string_piece.h
@@ -14,6 +14,11 @@
 // Systematic usage of StringPiece is encouraged as it will reduce unnecessary
 // conversions from "const char*" to "string" and back again.
 //
+// StringPiece16 is similar to StringPiece but for base::string16 instead of
+// std::string. We do not define as large of a subset of the STL functions
+// from basic_string as in StringPiece, but this can be changed if these
+// functions (find, find_first_of, etc.) are found to be useful in this context.
+//
 
 #ifndef BASE_STRING_PIECE_H_
 #define BASE_STRING_PIECE_H_
@@ -23,6 +28,8 @@
 
 #include "base/base_export.h"
 #include "base/basictypes.h"
+#include "base/hash_tables.h"
+#include "base/string16.h"
 
 namespace base {
 
@@ -164,6 +171,84 @@ class BASE_EXPORT StringPiece {
   size_type     length_;
 };
 
+class BASE_EXPORT StringPiece16 {
+ public:
+  // standard STL container boilerplate
+  typedef size_t size_type;
+  typedef char16 value_type;
+  typedef const char16* pointer;
+  typedef const char16& reference;
+  typedef const char16& const_reference;
+  typedef ptrdiff_t difference_type;
+  typedef const char16* const_iterator;
+  typedef const char16* iterator;
+  typedef std::reverse_iterator<const_iterator> const_reverse_iterator;
+  typedef std::reverse_iterator<iterator> reverse_iterator;
+
+ public:
+  // We provide non-explicit singleton constructors so users can pass
+  // in a "const char16*" or a "string16" wherever a "StringPiece16" is
+  // expected.
+  StringPiece16() : ptr_(NULL), length_(0) { }
+  StringPiece16(const char16* str)
+      : ptr_(str),
+        length_((str == NULL) ? 0 : string16::traits_type::length(str)) { }
+  StringPiece16(const string16& str)
+      : ptr_(str.data()), length_(str.size()) { }
+  StringPiece16(const char16* offset, size_type len)
+      : ptr_(offset), length_(len) { }
+
+  // data() may return a pointer to a buffer with embedded NULs, and the
+  // returned buffer may or may not be null terminated.  Therefore it is
+  // typically a mistake to pass data() to a routine that expects a NUL
+  // terminated string.
+  const char16* data() const { return ptr_; }
+  size_type size() const { return length_; }
+  size_type length() const { return length_; }
+  bool empty() const { return length_ == 0; }
+
+  void clear() {
+    ptr_ = NULL;
+    length_ = 0;
+  }
+  void set(const char16* data, size_type len) {
+    ptr_ = data;
+    length_ = len;
+  }
+  void set(const char16* str) {
+    ptr_ = str;
+    length_ = str ? string16::traits_type::length(str) : 0;
+  }
+
+  char16 operator[](size_type i) const { return ptr_[i]; }
+
+  string16 as_string16() const {
+    // StringPiece claims that this is bad when data() is NULL, but unittesting
+    // seems to say otherwise.
+    return string16(data(), size());
+  }
+
+  iterator begin() const { return ptr_; }
+  iterator end() const { return ptr_ + length_; }
+  const_reverse_iterator rbegin() const {
+    return const_reverse_iterator(ptr_ + length_);
+  }
+  const_reverse_iterator rend() const {
+    return const_reverse_iterator(ptr_);
+  }
+
+  size_type max_size() const { return length_; }
+  size_type capacity() const { return length_; }
+
+  static int wordmemcmp(const char16* p, const char16* p2, size_type N) {
+    return string16::traits_type::compare(p, p2, N);
+  }
+
+ private:
+  const char16*   ptr_;
+  size_type       length_;
+};
+
 BASE_EXPORT bool operator==(const StringPiece& x, const StringPiece& y);
 
 inline bool operator!=(const StringPiece& x, const StringPiece& y) {
@@ -188,6 +273,77 @@ inline bool operator>=(const StringPiece& x, const StringPiece& y) {
   return !(x < y);
 }
 
+inline bool operator==(const StringPiece16& x, const StringPiece16& y) {
+  if (x.size() != y.size())
+    return false;
+
+  return StringPiece16::wordmemcmp(x.data(), y.data(), x.size()) == 0;
+}
+
+inline bool operator!=(const StringPiece16& x, const StringPiece16& y) {
+  return !(x == y);
+}
+
+inline bool operator<(const StringPiece16& x, const StringPiece16& y) {
+  const int r = StringPiece16::wordmemcmp(
+      x.data(), y.data(), (x.size() < y.size() ? x.size() : y.size()));
+  return ((r < 0) || ((r == 0) && (x.size() < y.size())));
+}
+
+inline bool operator>(const StringPiece16& x, const StringPiece16& y) {
+  return y < x;
+}
+
+inline bool operator<=(const StringPiece16& x, const StringPiece16& y) {
+  return !(x > y);
+}
+
+inline bool operator>=(const StringPiece16& x, const StringPiece16& y) {
+  return !(x < y);
+}
+
 }  // namespace base
 
+// We provide appropriate hash functions so StringPiece and StringPiece16 can
+// be used as keys in hash sets and maps.
+
+// This hash function is copied from base/hash_tables.h. We don't use the
+// ones already defined for string and string16 directly because it would
+// require the string constructors to be called, which we don't want.
+#define HASH_STRING_PIECE(StringPieceType, string_piece)                \
+  std::size_t result = 0;                                               \
+  for (StringPieceType::const_iterator i = string_piece.begin();        \
+       i != string_piece.end(); ++i)                                    \
+    result = (result * 131) + *i;                                       \
+  return result;                                                        \
+
+namespace BASE_HASH_NAMESPACE {
+#if defined(COMPILER_GCC)
+
+template<>
+struct hash<base::StringPiece> {
+  std::size_t operator()(const base::StringPiece& sp) const {
+    HASH_STRING_PIECE(base::StringPiece, sp);
+  }
+};
+template<>
+struct hash<base::StringPiece16> {
+  std::size_t operator()(const base::StringPiece16& sp16) const {
+    HASH_STRING_PIECE(base::StringPiece16, sp16);
+  }
+};
+
+#elif defined(COMPILER_MSVC)
+
+inline size_t hash_value(const base::StringPiece& sp) {
+  HASH_STRING_PIECE(base::StringPiece, sp);
+}
+inline size_t hash_value(const base::StringPiece16& sp16) {
+  HASH_STRING_PIECE(base::StringPiece16, sp16);
+}
+
+#endif  // COMPILER
+
+}  // namespace BASE_HASH_NAMESPACE
+
 #endif  // BASE_STRING_PIECE_H_