Implement case mapping using the Unicode default case mapping algorithm.

runtime/vm/object.cc

Index: runtime/vm/object.cc
diff --git a/runtime/vm/object.cc b/runtime/vm/object.cc
index ca04df567c082a3df3ca1581cecfeb6a687f5a59..ce5c537401e8d98a1bccb76bebd1e0d74a6c9324 100644
--- a/runtime/vm/object.cc
+++ b/runtime/vm/object.cc
@@ -5721,6 +5721,48 @@ const char* String::ToCString() const {
 }
 
 
+RawString* String::Transform(int32_t (*mapping)(int32_t ch),
+                             const String& str,
+                             Heap::Space space) {
+  ASSERT(!str.IsNull());
+  bool has_mapping = false;
+  int32_t dst_max = 0;
+  intptr_t len = str.Length();
+  // TODO(cshapiro): assume a transform is required, rollback if not.
+  for (intptr_t i = 0; i < len; ++i) {
+    int32_t src = str.CharAt(i);
+    int32_t dst = mapping(src);
+    if (src != dst) {
+      has_mapping = true;
+    }
+    dst_max = Utils::Maximum(dst_max, dst);
+  }
+  if (!has_mapping) {
+    return str.raw();
+  }
+  if (dst_max <= 0xFF) {
+    return OneByteString::Transform(mapping, str, space);
+  }
+  if (dst_max <= 0xFFFF) {
+    return TwoByteString::Transform(mapping, str, space);
+  }
+  ASSERT(dst_max > 0xFFFF);
+  return FourByteString::Transform(mapping, str, space);
+}
+
+
+RawString* String::ToUpperCase(const String& str, Heap::Space space) {
+  // TODO(cshapiro): create a fast-path for OneByteString instances.
+  return Transform(CaseMapping::ToUpper, str, space);
+}
+
+
+RawString* String::ToLowerCase(const String& str, Heap::Space space) {
+  // TODO(cshapiro): create a fast-path for OneByteString instances.
+  return Transform(CaseMapping::ToLower, str, space);
+}
+
+
 RawOneByteString* OneByteString::New(intptr_t len,
                                      Heap::Space space) {
   Isolate* isolate = Isolate::Current();
@@ -5823,6 +5865,22 @@ RawOneByteString* OneByteString::SubString(const OneByteString& str,
 }
 
 
+RawOneByteString* OneByteString::Transform(int32_t (*mapping)(int32_t ch),
+                                           const String& str,
+                                           Heap::Space space) {
+  ASSERT(!str.IsNull());
+  intptr_t len = str.Length();
+  const OneByteString& onestr =

[Ivan Posva, 2011/10/21 14:03:37]

onestr -> result?

[cshapiro, 2011/10/21 20:29:08]

Yes, that would be more consistent.  Renamed.

+      OneByteString::Handle(OneByteString::New(len, space));
+  for (intptr_t i = 0; i < len; ++i) {
+    int32_t ch = mapping(str.CharAt(i));
+    ASSERT(ch >= 0 && ch <= 0xFF);
+    *onestr.CharAddr(i) = ch;
+  }
+  return onestr.raw();
+}
+
+
 const char* OneByteString::ToCString() const {
   return String::ToCString();
 }
@@ -5922,12 +5980,27 @@ RawTwoByteString* TwoByteString::SubString(const TwoByteString& str,
 }
 
 
+RawTwoByteString* TwoByteString::Transform(int32_t (*mapping)(int32_t ch),
+                                           const String& str,
+                                           Heap::Space space) {
+  ASSERT(!str.IsNull());
+  intptr_t len = str.Length();
+  const TwoByteString& twostr =

[Ivan Posva, 2011/10/21 14:03:37]

ditto

[cshapiro, 2011/10/21 20:29:08]

Done.

+      TwoByteString::Handle(TwoByteString::New(len, space));
+  for (intptr_t i = 0; i < len; ++i) {
+    int32_t ch = mapping(str.CharAt(i));
+    ASSERT(ch >= 0 && ch <= 0xFFFF);
+    *twostr.CharAddr(i) = ch;
+  }
+  return twostr.raw();
+}
+
+
 const char* TwoByteString::ToCString() const {
   return String::ToCString();
 }
 
 
-
 RawFourByteString* FourByteString::New(intptr_t len,
                                        Heap::Space space) {
   Isolate* isolate = Isolate::Current();
@@ -6014,6 +6087,22 @@ RawFourByteString* FourByteString::SubString(const FourByteString& str,
 }
 
 
+RawFourByteString* FourByteString::Transform(int32_t (*mapping)(int32_t ch),
+                                             const String& str,
+                                             Heap::Space space) {
+  ASSERT(!str.IsNull());
+  intptr_t len = str.Length();
+  const FourByteString& fourstr =

[Ivan Posva, 2011/10/21 14:03:37]

ditto

[cshapiro, 2011/10/21 20:29:08]

Done.

+      FourByteString::Handle(FourByteString::New(len, space));
+  for (intptr_t i = 0; i < len; ++i) {
+    int32_t ch = mapping(str.CharAt(i));
+    ASSERT(ch >= 0 && ch <= 0x10FFFF);
+    *fourstr.CharAddr(i) = ch;
+  }
+  return fourstr.raw();
+}
+
+
 const char* FourByteString::ToCString() const {
   return String::ToCString();
 }