WIP: Introduce CompressibleString

third_party/WebKit/Source/wtf/text/CompressibleString.cpp

+// Copyright 2016 The Chromium 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 "wtf/text/CompressibleString.h"
+
+#include "third_party/zlib/google/compression_utils.h"
+#include "wtf/Assertions.h"
+#include "wtf/LinkedStack.h"
+#include "wtf/Partitions.h"
+#include "wtf/PassOwnPtr.h"
+#include "wtf/WTFThreadData.h"
+#include "wtf/text/WTFString.h"
+
+namespace WTF {
+
+class CompressibleStringTable {
+    WTF_MAKE_NONCOPYABLE(CompressibleStringTable);
+public:
+    static CompressibleStringTable* create(WTFThreadData& data)
+    {
+        data.m_compressibleStringTable = new CompressibleStringTable;
+        data.m_compressibleStringTableDestructor = CompressibleStringTable::destroy;
+        return data.m_compressibleStringTable;
+    }
+
+    void add(CompressibleStringImpl* string)
+    {
+        m_table.add(string);
+    }
+
+    bool contains(CompressibleStringImpl* string) const
+    {
+        return m_table.contains(string);
+    }
+
+    void remove(CompressibleStringImpl* string)
+    {
+        m_table.remove(string);
+    }
+
+    void compressAll()
+    {
+        HashSet<CompressibleStringImpl*>::iterator end = m_table.end();
+        for (HashSet<CompressibleStringImpl*>::iterator iter = m_table.begin(); iter != end; ++iter) {
+            CompressibleStringImpl* string = *iter;
+            if (!string->isCompressed())
+                string->compressString();
+        }
+    }
+
+private:
+    CompressibleStringTable() { }
+
+    static void destroy(CompressibleStringTable* table)
+    {
+        delete table;
+    }
+
+    HashSet<CompressibleStringImpl*> m_table;
+};
+
+static inline CompressibleStringTable& compressibleStringTable()
+{
+    // Once possible we should make this non-lazy (constructed in WTFThreadData's constructor).
+    WTFThreadData& data = wtfThreadData();
+    CompressibleStringTable* table = data.compressibleStringTable();
+    if (UNLIKELY(!table))
+        table = CompressibleStringTable::create(data);
+    return *table;
+}
+
+void CompressibleStringImpl::purgeMemory()
+{
+    // TODO(hajimehoshi): Compressing for the current thread is not enough.
+    // Do the same thing on other threads.
+    compressibleStringTable().compressAll();
+}
+
+CompressibleStringImpl::CompressibleStringImpl(PassRefPtr<StringImpl> impl)
+    : m_impl(impl)
+    , m_originalLength(m_impl->length())
+    , m_compressedData(nullptr)
+    , m_compressedDataSize(0)
+    , m_is8Bit(m_impl->is8Bit())
+{
+    ASSERT(m_impl);
+    if (originalContentSizeInBytes() > 100000)
+        compressibleStringTable().add(this);
+}
+
+CompressibleStringImpl::~CompressibleStringImpl()
+{
+    if (compressibleStringTable().contains(this))
+        compressibleStringTable().remove(this);
+    if (m_compressedData)
+        Partitions::fastFree(m_compressedData);
+}
+
+unsigned CompressibleStringImpl::originalContentSizeInBytes() const
+{
+    if (is8Bit())
+        return m_originalLength * sizeof(LChar);
+    return m_originalLength * sizeof(UChar);
+}
+
+unsigned CompressibleStringImpl::currentSizeInBytes() const
+{
+    if (UNLIKELY(isCompressed()))
+        return m_compressedDataSize;
+    return originalContentSizeInBytes();
+}
+
+String CompressibleStringImpl::toString()
+{
+    if (UNLIKELY(isCompressed()))
+        uncompressString();
+    return m_impl.get();
+}
+
+const LChar* CompressibleStringImpl::characters8()
+{
+    return toString().characters8();
+}
+
+const UChar* CompressibleStringImpl::characters16()
+{
+    return toString().characters16();
+}
+
+void CompressibleStringImpl::compressString()
+{
+    // UMA
+    // Count when the tab is background
+
+    ASSERT(m_impl);
+    ASSERT(!isCompressed());
+    ASSERT(!m_compressedDataSize);
+
+    // TODO(hajimehoshi): Now components offers funcitons accepting only
+    // std::strings. This is not efficient. We should offer char* version.
+    std::string in, out;
+    if (m_is8Bit)
+        in = std::string(reinterpret_cast<const char*>(m_impl->characters8()), originalContentSizeInBytes());
+    else
+        in = std::string(reinterpret_cast<const char*>(m_impl->characters16()), originalContentSizeInBytes());
+    compression::GzipCompress(in, &out);
+
+    m_impl = nullptr;
+    m_compressedData = Partitions::fastMalloc(out.size(), "CompressibleString");
+    memcpy(m_compressedData, out.c_str(), out.size());
+    m_compressedDataSize = out.size();
+}
+
+void CompressibleStringImpl::uncompressString()
+{
+    // UMA
+    // Count when the tab is background
+
+    ASSERT(m_compressedData);
+    ASSERT(m_compressedDataSize);
+    ASSERT(!m_impl);
+
+    std::string in(static_cast<const char*>(m_compressedData), m_compressedDataSize);
+    std::string out;
+    compression::GzipUncompress(in, &out);
+
+    if (m_is8Bit) {
+        LChar* data = nullptr;
+        m_impl = StringImpl::createUninitialized(out.size() / sizeof(LChar), data);
+        memcpy(data, out.c_str(), out.size());
+    } else {
+        UChar* data = nullptr;
+        m_impl = StringImpl::createUninitialized(out.size() / sizeof(UChar), data);
+        memcpy(data, out.c_str(), out.size());
+    }
+
+    Partitions::fastFree(m_compressedData);
+    m_compressedData = nullptr;
+    m_compressedDataSize = 0;
+    ASSERT(m_is8Bit == m_impl->is8Bit());
+    ASSERT(m_originalLength == m_impl->length());
+}
+
+CompressibleString::CompressibleString()
+    : m_impl(nullptr)
+{
+}
+
+CompressibleString::CompressibleString(const CompressibleString& rhs)
+    : m_impl(rhs.m_impl)
+{
+}
+
+CompressibleString::CompressibleString(PassRefPtr<StringImpl> impl)
+    : m_impl(impl ? adoptRef(new CompressibleStringImpl(impl)) : nullptr)
+{
+}
+
+void CompressibleString::compressString() const
+{
+    m_impl->compressString();
+}
+
+void CompressibleString::uncompressString() const
+{
+    m_impl->uncompressString();
+}
+
+} // namespace WTF