Fix lldb string formatter.

Tools/lldb/lldb_webkit.py

Index: Tools/lldb/lldb_webkit.py
diff --git a/Tools/lldb/lldb_webkit.py b/Tools/lldb/lldb_webkit.py
index 4adc4388350a6eb99aa4673a925519c020fc3831..1420a7a13780031e93223ec3dd651f53368ed6d4 100644
--- a/Tools/lldb/lldb_webkit.py
+++ b/Tools/lldb/lldb_webkit.py
@@ -30,6 +30,7 @@
 """
 
 import lldb
+import struct
 
 
 def __lldb_init_module(debugger, dict):
@@ -74,45 +75,6 @@ def WTFHashTable_SummaryProvider(valobj, dict):
 # def JSCJSString_SummaryProvider(valobj, dict):
 
 
-def guess_string_length(valobj, error):
-    if not valobj.GetValue():
-        return 0
-
-    for i in xrange(0, 2048):
-        if valobj.GetPointeeData(i, 1).GetUnsignedInt16(error, 0) == 0:
-            return i
-
-    return 256
-
-
-def ustring_to_string(valobj, error, length=None):
-    if length is None:
-        length = guess_string_length(valobj, error)
-    else:
-        length = int(length)
-
-    out_string = u""
-    for i in xrange(0, length):
-        char_value = valobj.GetPointeeData(i, 1).GetUnsignedInt16(error, 0)
-        out_string = out_string + unichr(char_value)
-
-    return out_string.encode('utf-8')
-
-
-def lstring_to_string(valobj, error, length=None):
-    if length is None:
-        length = guess_string_length(valobj, error)
-    else:
-        length = int(length)
-
-    out_string = u""
-    for i in xrange(0, length):
-        char_value = valobj.GetPointeeData(i, 1).GetUnsignedInt8(error, 0)
-        out_string = out_string + unichr(char_value)
-
-    return out_string.encode('utf-8')
-
-
 class WTFStringImplProvider:
     def __init__(self, valobj, dict):
         self.valobj = valobj
@@ -120,22 +82,37 @@ class WTFStringImplProvider:
     def get_length(self):
         return self.valobj.GetChildMemberWithName('m_length').GetValueAsUnsigned(0)
 
+    def get_data(self, length):
+        offset = 12
+        # Try to get the address of the object.
+        addr = self.valobj.GetAddress()
+        # If address == 0 then the object is a pointer so we can use it's
+        # value instead. We need to offset the pointer so it points to
+        # raw string data.
+        ptr = (int(str(addr), 16) if addr else self.valobj.GetValueAsUnsigned(0)) + offset
+        error = lldb.SBError()
+        # Read the raw data from memory.
+        return self.valobj.GetTarget().GetProcess().ReadMemory(ptr, length, error)
+
     def get_data8(self):
-        return self.valobj.GetChildAtIndex(2).GetChildMemberWithName('m_data8')
+        return self.get_data(self.get_length())
 
     def get_data16(self):
-        return self.valobj.GetChildAtIndex(2).GetChildMemberWithName('m_data16')
+        length = self.get_length()
+        # Read raw string data.
+        data = self.get_data(2 * length)
+        # Create a list of length 2-byte unsigned integers from byte array.
+        ushort_list = struct.unpack(length * 'H', data)
+        # Return the string.
+        return u"".join(map(unichr, ushort_list))
 
     def to_string(self):
         error = lldb.SBError()
-        if self.is_8bit():
-            return lstring_to_string(self.get_data8(), error, self.get_length())
-        return ustring_to_string(self.get_data16(), error, self.get_length())
+        data = self.get_data8() if self.is_8bit() else self.get_data16()
+        return data.encode('utf-8')
 
     def is_8bit(self):
-        # FIXME: find a way to access WTF::StringImpl::s_hashFlag8BitBuffer
-        return bool(self.valobj.GetChildMemberWithName('m_hashAndFlags').GetValueAsUnsigned(0) \
-            & 1 << 6)
+        return bool(self.valobj.GetChildMemberWithName('m_is8Bit').GetValueAsUnsigned(0))
 
 
 class WTFStringProvider: