Files.app: Implement md5 hashing over file streams.

chrome/browser/drive/drive_api_util.cc

Index: chrome/browser/drive/drive_api_util.cc
diff --git a/chrome/browser/drive/drive_api_util.cc b/chrome/browser/drive/drive_api_util.cc
index 064f4a2d8740af5dc150982e12f84a2570fe56ca..622430b042392eab136e54c790b26e876b13f618 100644
--- a/chrome/browser/drive/drive_api_util.cc
+++ b/chrome/browser/drive/drive_api_util.cc
@@ -16,6 +16,9 @@
 #include "base/values.h"
 #include "google_apis/drive/drive_api_parser.h"
 #include "net/base/escape.h"
+#include "net/base/io_buffer.h"
+#include "net/base/net_errors.h"
+#include "storage/browser/blob/file_stream_reader.h"
 #include "third_party/re2/re2/re2.h"
 #include "url/gurl.h"
 
@@ -163,6 +166,54 @@ std::string GetMd5Digest(const base::FilePath& file_path) {
   return MD5DigestToBase16(digest);
 }
 
+FileStreamMd5Digester::FileStreamMd5Digester()
+    : buffer_(new net::IOBuffer(kBufferSize_)), weak_ptr_factory_(this) {
+}
+
+FileStreamMd5Digester::~FileStreamMd5Digester() {
+}
+
+void FileStreamMd5Digester::GetMd5Digest(
+    scoped_ptr<storage::FileStreamReader> streamReader,
+    const base::Callback<void(const std::string&)>& callback) {
+  reader_ = streamReader.Pass();
+  callback_ = callback;
+  base::MD5Init(&md5Context_);
+
+  // Start the read/hash.
+  readNextChunk();
+}
+
+void FileStreamMd5Digester::readNextChunk() {
+  int result = reader_->Read(buffer_.get(), kBufferSize_,

[mtomasz, 2015/01/31 04:53:12]

IIRC reader may have to be called on IO thread only. But it's not documented
whether GetMd5Digest() should be called on IO thread or UI. Could you double
check?

[mtomasz, 2015/01/31 04:53:12]

nit: const int

[Ben Kwa, 2015/02/02 22:22:11]

Hm, you're right; I misread the FileStreamReader::Read doc.  I reworked this to
be on the IO thread.  Salient points:
- FileStreamMd5Digester is now a ref-counted object.  This enables it to be used
properly with refptr and bind.
- The extension function no longer has to hold on to a FileStreamMd5Digester as
a class member; it just instantiates one in RunAsync, and relies on refcounting
to manage the lifetime of that object.

[Ben Kwa, 2015/02/02 22:22:11]

Done.

+                             base::Bind(&FileStreamMd5Digester::OnChunkRead,
+                                        weak_ptr_factory_.GetWeakPtr()));
+  if (result != net::ERR_IO_PENDING) {
+    OnChunkRead(result);
+  }
+}
+
+void FileStreamMd5Digester::OnChunkRead(int result) {
+  if (result < 0) {
+    // Error - just return empty string

[mtomasz, 2015/01/31 04:53:12]

nit: period at the end of comments.

[Ben Kwa, 2015/02/02 22:22:11]

Done.

+    callback_.Run("");
+    return;
+  } else if (result == 0) {
+    // EOF
+    base::MD5Digest digest;
+    base::MD5Final(&digest, &md5Context_);
+    std::string result = MD5DigestToBase16(digest);
+    callback_.Run(result);
+    return;
+  }
+
+  // Read data and digest it.
+  base::MD5Update(&md5Context_, base::StringPiece(buffer_->data(), result));
+
+  // Kick off the next read.
+  readNextChunk();
+}
+
 std::string GetHostedDocumentExtension(const std::string& mime_type) {
   for (size_t i = 0; i < arraysize(kHostedDocumentKinds); ++i) {
     if (mime_type == kHostedDocumentKinds[i].mime_type)