Move CXFA_FileRead out of XFA and rename

This code doesn't depend on XFA, but is only used by it. Moving it
out, so it will be easier for me to write some tools for extracting
data out of PDFs.

Bug:
Change-Id: Ic18613b46abed5124c47f539833b01b12c1c6e56
Reviewed-on: https://pdfium-review.googlesource.com/18410
Reviewed-by: dsinclair <dsinclair@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>

3 files changed, 206 insertions, 0 deletions

diff --git a/core/fxcrt/cfx_seekablemultistream.cpp b/core/fxcrt/cfx_seekablemultistream.cpp
new file mode 100644
index 0000000000..12682d7678
--- /dev/null
+++ b/core/fxcrt/cfx_seekablemultistream.cpp
@@ -0,0 +1,85 @@
+// Copyright 2017 PDFium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
+
+#include "core/fxcrt/cfx_seekablemultistream.h"
+
+#include <algorithm>
+
+#include "core/fpdfapi/parser/cpdf_stream_acc.h"
+#include "third_party/base/logging.h"
+#include "third_party/base/stl_util.h"
+
+CFX_SeekableMultiStream::CFX_SeekableMultiStream(
+    const std::vector<CPDF_Stream*>& streams) {
+  for (CPDF_Stream* pStream : streams) {
+    m_Data.push_back(pdfium::MakeRetain<CPDF_StreamAcc>(pStream));
+    m_Data.back()->LoadAllData();
+  }
+}
+
+CFX_SeekableMultiStream::~CFX_SeekableMultiStream() {}
+
+FX_FILESIZE CFX_SeekableMultiStream::GetSize() {
+  uint32_t dwSize = 0;
+  for (const auto& acc : m_Data)
+    dwSize += acc->GetSize();
+  return dwSize;
+}
+
+bool CFX_SeekableMultiStream::ReadBlock(void* buffer,
+                                        FX_FILESIZE offset,
+                                        size_t size) {
+  int32_t iCount = pdfium::CollectionSize<int32_t>(m_Data);
+  int32_t index = 0;
+  while (index < iCount) {
+    const auto& acc = m_Data[index];
+    FX_FILESIZE dwSize = acc->GetSize();
+    if (offset < dwSize)
+      break;
+
+    offset -= dwSize;
+    index++;
+  }
+  while (index < iCount) {
+    const auto& acc = m_Data[index];
+    uint32_t dwSize = acc->GetSize();
+    size_t dwRead = std::min(size, static_cast<size_t>(dwSize - offset));
+    memcpy(buffer, acc->GetData() + offset, dwRead);
+    size -= dwRead;
+    if (size == 0)
+      return true;
+
+    buffer = static_cast<uint8_t*>(buffer) + dwRead;
+    offset = 0;
+    index++;
+  }
+  return false;
+}
+
+size_t CFX_SeekableMultiStream::ReadBlock(void* buffer, size_t size) {
+  NOTREACHED();
+  return 0;
+}
+
+FX_FILESIZE CFX_SeekableMultiStream::GetPosition() {
+  return 0;
+}
+
+bool CFX_SeekableMultiStream::IsEOF() {
+  return false;
+}
+
+bool CFX_SeekableMultiStream::Flush() {
+  NOTREACHED();
+  return false;
+}
+
+bool CFX_SeekableMultiStream::WriteBlock(const void* pData,
+                                         FX_FILESIZE offset,
+                                         size_t size) {
+  NOTREACHED();
+  return false;
+}
diff --git a/core/fxcrt/cfx_seekablemultistream.h b/core/fxcrt/cfx_seekablemultistream.h
new file mode 100644
index 0000000000..9138d7c321
--- /dev/null
+++ b/core/fxcrt/cfx_seekablemultistream.h
@@ -0,0 +1,36 @@
+// Copyright 2017 PDFium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
+
+#ifndef CORE_FXCRT_CFX_SEEKABLEMULTISTREAM_H_
+#define CORE_FXCRT_CFX_SEEKABLEMULTISTREAM_H_
+
+#include <vector>
+
+#include "core/fxcrt/fx_stream.h"
+#include "core/fxcrt/retain_ptr.h"
+
+class CPDF_Stream;
+class CPDF_StreamAcc;
+
+class CFX_SeekableMultiStream : public IFX_SeekableStream {
+ public:
+  explicit CFX_SeekableMultiStream(const std::vector<CPDF_Stream*>& streams);
+  ~CFX_SeekableMultiStream() override;
+
+  // IFX_SeekableReadStream
+  FX_FILESIZE GetPosition() override;
+  FX_FILESIZE GetSize() override;
+  bool ReadBlock(void* buffer, FX_FILESIZE offset, size_t size) override;
+  size_t ReadBlock(void* buffer, size_t size) override;
+  bool IsEOF() override;
+  bool Flush() override;
+  bool WriteBlock(const void* pData, FX_FILESIZE offset, size_t size) override;
+
+ private:
+  std::vector<RetainPtr<CPDF_StreamAcc>> m_Data;
+};
+
+#endif  // CORE_FXCRT_CFX_SEEKABLEMULTISTREAM_H_
diff --git a/core/fxcrt/cfx_seekablemultistream_unittest.cpp b/core/fxcrt/cfx_seekablemultistream_unittest.cpp
new file mode 100644
index 0000000000..89213b13af
--- /dev/null
+++ b/core/fxcrt/cfx_seekablemultistream_unittest.cpp
@@ -0,0 +1,85 @@
+// Copyright 2016 PDFium 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 "core/fxcrt/cfx_seekablemultistream.h"
+
+#include <memory>
+#include <vector>
+
+#include "core/fpdfapi/parser/cpdf_dictionary.h"
+#include "core/fpdfapi/parser/cpdf_stream.h"
+#include "core/fxcrt/fx_memory.h"
+#include "testing/gtest/include/gtest/gtest.h"
+#include "testing/test_support.h"
+#include "third_party/base/ptr_util.h"
+
+TEST(CFX_SeekableMultiStreamTest, NoStreams) {
+  std::vector<CPDF_Stream*> streams;
+  auto fileread = pdfium::MakeRetain<CFX_SeekableMultiStream>(streams);
+
+  uint8_t output_buffer[16];
+  memset(output_buffer, 0xbd, sizeof(output_buffer));
+  EXPECT_FALSE(fileread->ReadBlock(output_buffer, 0, 0));
+  EXPECT_EQ(0xbd, output_buffer[0]);
+}
+
+TEST(CXFAFileReadTest, EmptyStreams) {
+  std::vector<CPDF_Stream*> streams;
+  auto stream1 = pdfium::MakeUnique<CPDF_Stream>();
+  streams.push_back(stream1.get());
+  auto fileread = pdfium::MakeRetain<CFX_SeekableMultiStream>(streams);
+
+  uint8_t output_buffer[16];
+  memset(output_buffer, 0xbd, sizeof(output_buffer));
+  EXPECT_FALSE(fileread->ReadBlock(output_buffer, 0, 0));
+  EXPECT_EQ(0xbd, output_buffer[0]);
+}
+
+TEST(CXFAFileReadTest, NormalStreams) {
+  std::vector<CPDF_Stream*> streams;
+  auto stream1 = pdfium::MakeUnique<CPDF_Stream>();
+  auto stream2 = pdfium::MakeUnique<CPDF_Stream>();
+  auto stream3 = pdfium::MakeUnique<CPDF_Stream>();
+
+  // 16 chars total.
+  stream1->InitStream(reinterpret_cast<const uint8_t*>("one t"), 5,
+                      pdfium::MakeUnique<CPDF_Dictionary>());
+  stream2->InitStream(reinterpret_cast<const uint8_t*>("wo "), 3,
+                      pdfium::MakeUnique<CPDF_Dictionary>());
+  stream3->InitStream(reinterpret_cast<const uint8_t*>("three!!!"), 8,
+                      pdfium::MakeUnique<CPDF_Dictionary>());
+
+  streams.push_back(stream1.get());
+  streams.push_back(stream2.get());
+  streams.push_back(stream3.get());
+  auto fileread = pdfium::MakeRetain<CFX_SeekableMultiStream>(streams);
+
+  uint8_t output_buffer[16];
+  memset(output_buffer, 0xbd, sizeof(output_buffer));
+  EXPECT_TRUE(fileread->ReadBlock(output_buffer, 0, 0));
+  EXPECT_EQ(0xbd, output_buffer[0]);
+
+  memset(output_buffer, 0xbd, sizeof(output_buffer));
+  EXPECT_TRUE(fileread->ReadBlock(output_buffer, 1, 0));
+  EXPECT_EQ(0xbd, output_buffer[0]);
+
+  memset(output_buffer, 0xbd, sizeof(output_buffer));
+  EXPECT_TRUE(fileread->ReadBlock(output_buffer, 0, 1));
+  EXPECT_EQ(0, memcmp(output_buffer, "o", 1));
+  EXPECT_EQ(0xbd, output_buffer[1]);
+
+  memset(output_buffer, 0xbd, sizeof(output_buffer));
+  EXPECT_TRUE(fileread->ReadBlock(output_buffer, 0, sizeof(output_buffer)));
+  EXPECT_EQ(0, memcmp(output_buffer, "one two three!!!", 16));
+
+  memset(output_buffer, 0xbd, sizeof(output_buffer));
+  EXPECT_TRUE(fileread->ReadBlock(output_buffer, 2, 10));
+  EXPECT_EQ(0, memcmp(output_buffer, "e two thre", 10));
+  EXPECT_EQ(0xbd, output_buffer[11]);
+
+  memset(output_buffer, 0xbd, sizeof(output_buffer));
+  EXPECT_FALSE(fileread->ReadBlock(output_buffer, 1, sizeof(output_buffer)));
+  EXPECT_EQ(0, memcmp(output_buffer, "ne two three!!!", 15));
+  EXPECT_EQ(0xbd, output_buffer[15]);
+}