Files.app: Implement md5 hashing over file streams.

chrome/browser/drive/drive_api_util.h

Index: chrome/browser/drive/drive_api_util.h
diff --git a/chrome/browser/drive/drive_api_util.h b/chrome/browser/drive/drive_api_util.h
index 95771ca266ac150ee548a6af9e007f8fafef2f7f..977d6bbbcabff7d2451e45b450db902394614687 100644
--- a/chrome/browser/drive/drive_api_util.h
+++ b/chrome/browser/drive/drive_api_util.h
@@ -7,6 +7,7 @@
 
 #include <string>
 
+#include "base/md5.h"
 #include "base/memory/scoped_ptr.h"
 #include "google_apis/drive/drive_api_error_codes.h"
 #include "google_apis/drive/drive_common_callbacks.h"
@@ -26,6 +27,14 @@ class FileResource;
 class ResourceEntry;
 }  // namespace google_apis
 
+namespace net {
+class IOBuffer;
+}  // namespace net
+
+namespace storage {
+class FileStreamReader;
+}  // namespace storage
+
 namespace drive {
 namespace util {
 
@@ -64,6 +73,43 @@ std::string CanonicalizeResourceId(const std::string& resource_id);
 // or an empty string if an error is found.
 std::string GetMd5Digest(const base::FilePath& file_path);
 
+// Computes the (base-16 encoded) MD5 digest of data extracted from a file
+// stream.
+class FileStreamMd5Digester
+    : public base::RefCountedThreadSafe<FileStreamMd5Digester> {
+ public:
+  typedef base::Callback<void(const std::string&)> ResultCallback;
+
+  FileStreamMd5Digester();
+
+  // Computes an MD5 digest of data read from the given |streamReader|.  The
+  // work occurs asynchronously, and the resulting hash is returned via the
+  // |callback|.  If an error occurs, |callback| is called with an empty string.
+  // Only one stream can be processed at a time by each digester.  Do not call
+  // GetMd5Digest before the results of a previous call have been returned.
+  void GetMd5Digest(scoped_ptr<storage::FileStreamReader> stream_reader,
+                    const ResultCallback& callback);
+
+ private:
+  // Private for RefCountedThreadSafe.
+  ~FileStreamMd5Digester();
+
+  // Kicks off a read of the next chunk from the stream.
+  void ReadNextChunk();
+  // Handles the incoming chunk of data from a stream read.
+  void OnChunkRead(int result);
+
+  // Maximum chunk size for read operations.
+  static const int kBufferSize_ = 512 * 1024;  // 512 kB.

[mtomasz, 2015/02/02 23:47:08]

nit: Yeah, I'd suggest to move it to an anonymous namespace in .cc file. We
always do that and try to make headers as small as possible (because we already
have tons of them). We usually move everything what's possible from private
section to anonymous namespace.

[Ben Kwa, 2015/02/03 16:32:19]

Done.

+  scoped_ptr<storage::FileStreamReader> reader_;
+  ResultCallback callback_;
+  scoped_refptr<net::IOBuffer> buffer_;
+  base::MD5Context md5_context_;
+
+  friend class base::RefCountedThreadSafe<FileStreamMd5Digester>;
+  DISALLOW_COPY_AND_ASSIGN(FileStreamMd5Digester);
+};
+
 // Returns preferred file extension for hosted documents which have given mime
 // type.
 std::string GetHostedDocumentExtension(const std::string& mime_type);