Start refactoring to reduce executable type knowledge.

courgette/versioning_unittest.cc

+// Copyright (c) 2011 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 <string>
+
+#include "base/path_service.h"
+#include "base/file_util.h"
+#include "base/string_util.h"
+
+#include "courgette/courgette.h"
+#include "courgette/streams.h"
+
+#include "testing/gtest/include/gtest/gtest.h"
+
+class VersioningTest : public testing::Test {
+ public:
+  void TestApplyingOldPatch(const char* src_file,
+                         const char* patch_file,
+                         const char* expected_file) const;
+
+ private:
+  void SetUp() {
+    PathService::Get(base::DIR_SOURCE_ROOT, &testdata_dir_);
+    testdata_dir_ = testdata_dir_.AppendASCII("courgette");
+    testdata_dir_ = testdata_dir_.AppendASCII("testdata");
+  }
+
+  void TearDown() { }
+
+  // Returns contents of |file_name| as uninterprested bytes stored in a string.
+  std::string FileContents(const char* file_name) const;
+
+  FilePath testdata_dir_;  // Full path name of testdata directory
+};
+
+//  Reads a test file into a string.
+std::string VersioningTest::FileContents(const char* file_name) const {
+  FilePath file_path = testdata_dir_;
+  file_path = file_path.AppendASCII(file_name);
+  std::string file_contents;
+  if (!file_util::ReadFileToString(file_path, &file_contents)) {
+    EXPECT_TRUE(!"Could not read test data");

[Paweł Hajdan Jr., 2011/09/16 18:34:34]

Huh? Why not EXPECT_TRUE(file_util::ReadFileToString(...)) above?

[dgarrett, 2011/09/22 02:20:46]

Honestly, because I copied that block from another similar unittest. There are a
couple of helper methods like that which should probably be pulled into a single
location and shared.

I'll probably get to that later on in this refactor process.

[sra1, 2011/09/22 02:58:17]

This is an old habit of mine.  Fix it if you want to, but I find that when the
test is failing, the string expression is more meaningful that the boolean
expression.  !"string" is always false.

[Paweł Hajdan Jr., 2011/09/26 17:08:21]

Please note you can EXPECT_TRUE(ReadFileToString(...)) << "Could not read test
data";

if you want. GTest is quite smarter than the traditional "assert" solution.

Could you please fix all usages of that? I'd like this pattern not to
proliferate throughout our codebase.

+  }
+  return file_contents;
+}
+
+void VersioningTest::TestApplyingOldPatch(const char* src_file,
+                                          const char* patch_file,
+                                          const char* expected_file) const {
+
+  std::string old_buffer = FileContents(src_file);
+  std::string new_buffer = FileContents(patch_file);
+  std::string expected_buffer = FileContents(expected_file);
+
+  courgette::SourceStream old_stream;
+  courgette::SourceStream patch_stream;
+  old_stream.Init(old_buffer);
+  patch_stream.Init(new_buffer);
+
+  courgette::SinkStream generated_stream;
+
+  courgette::Status status =
+      courgette::ApplyEnsemblePatch(&old_stream,
+                                    &patch_stream,
+                                    &generated_stream);
+
+  EXPECT_EQ(status, courgette::C_OK);
+
+  size_t expected_length = expected_buffer.size();
+  size_t generated_length = generated_stream.Length();
+
+  EXPECT_EQ(generated_length, expected_length);
+  EXPECT_EQ(0, memcmp(generated_stream.Buffer(),
+                      expected_buffer.c_str(),
+                      expected_length));
+}
+
+
+TEST_F(VersioningTest, All) {
+  TestApplyingOldPatch("setup1.exe", "setup1-setup2.v1.patch", "setup2.exe");
+
+  // We also need a way to test that newly generated patches are appropriately
+  // applicable by older clients... not sure of the best way to do that.
+}