Refactor BaseFile class to support sparse files

content/browser/download/base_file.h

Index: content/browser/download/base_file.h
diff --git a/content/browser/download/base_file.h b/content/browser/download/base_file.h
index 8351c49be340654f99fce3e0f41477ecb77f65a6..39470a36043c46f70d1b2bd4b6387f427c38c66f 100644
--- a/content/browser/download/base_file.h
+++ b/content/browser/download/base_file.h
@@ -33,6 +33,13 @@ namespace content {
 // Detach().
 class CONTENT_EXPORT BaseFile {
  public:
+  // Enum to indicate whether the file is exclusived owned, or shared among
+  // multiple writers.
+  enum AccessMode {
+    EXCLUSIVE = 0,
+    SHARED,

[asanka, 2017/02/16 18:46:26]

I'd argue that the distinction is that the file is sparse or not.

[qinmin, 2017/02/16 21:35:50]

Done. Changed this to a boolean variable is_sparse_file.

+  };
+
   // May be constructed on any thread.  All other routines (including
   // destruction) must occur on the FILE thread.
   BaseFile(const net::NetLogWithSource& net_log);
@@ -61,36 +68,52 @@ class CONTENT_EXPORT BaseFile {
   //     will not attempt to determine the |full_path|.
   //
   // |bytes_so_far|: If a file is provided (via |full_path| or |file|), then
-  //     this argument specifies the size of the file to expect. It is legal for
-  //     the file to be larger, in which case the file will be truncated down to
-  //     this size. However, if the file is shorter, then the operation will
+  //     this argument specifies the amount of data that has been written to
+  //     the file. If |access_mode| is EXCLUSIVE, this value should be the size
+  //     of the file to expect. It is legal for the file to be larger, in which
+  //     case the file will be truncated down to this size if |access_mode| is
+  //     EXCLUSIVE. However, if the file is shorter, then the operation will
   //     fail with DOWNLOAD_INTERRUPT_REASON_FILE_TOO_SHORT.
+
   //
-  // |hash_so_far|: If |bytes_so_far| is non-zero, this specifies the SHA-256
-  //     hash of the first |bytes_so_far| bytes of the target file. If
-  //     specified, BaseFile will read the first |bytes_so_far| of the target
-  //     file in order to calculate the hash and verify that the file matches.
-  //     If there's a mismatch, then the operation fails with
-  //     DOWNLOAD_INTERRUPT_REASON_FILE_HASH_MISMATCH. Not used if |hash_state|
-  //     is also specified.
+  // |hash_so_far|: If |bytes_so_far| is non-zero and |access_mode| is
+  //     EXCLUSIVE, this specifies the SHA-256 hash of the first |bytes_so_far|
+  //     bytes of the target file. If specified, BaseFile will read the first
+  //     |bytes_so_far| of the target file in order to calculate the hash and
+  //     verify that the file matches. If there's a mismatch, then the operation
+  //     fails with DOWNLOAD_INTERRUPT_REASON_FILE_HASH_MISMATCH. Not used if
+  //     |hash_state| is also specified.
   //
   // |hash_state|: The partial hash object to use. Only meaningful if there's a
   //     preexisting target file and it is non-empty (i.e. bytes_so_far is
   //     non-zero). If specified, BaseFile will assume that the bytes up to
   //     |bytes_so_far| has been accurately hashed into |hash_state| and will
-  //     ignore |hash_so_far|.
+  //     ignore |hash_so_far|. Not used if |access_mode| is SHARED.
+  //
+  // |access_mode|: Specifies whether the file is going to be accessed by
+  //     multiple writers. If so, it is possible that one writer can request
+  //     writing to an offset that is larger than the file size.
   DownloadInterruptReason Initialize(
       const base::FilePath& full_path,
       const base::FilePath& default_directory,
       base::File file,
       int64_t bytes_so_far,
       const std::string& hash_so_far,
-      std::unique_ptr<crypto::SecureHash> hash_state);
+      std::unique_ptr<crypto::SecureHash> hash_state,
+      AccessMode access_mode);
 
   // Write a new chunk of data to the file. Returns a DownloadInterruptReason
-  // indicating the result of the operation.
+  // indicating the result of the operation. Works only if |access_mode| is
+  // EXCLUSIVE.
   DownloadInterruptReason AppendDataToFile(const char* data, size_t data_len);
 
+  // Write a new chunk of data to the file. Returns a DownloadInterruptReason
+  // indicating the result of the operation.
+  DownloadInterruptReason WriteDataToFile(
+      int64_t offset,
+      const char* data,
+      size_t data_len);
+
   // Rename the download file. Returns a DownloadInterruptReason indicating the
   // result of the operation. A return code of NONE indicates that the rename
   // was successful. After a failure, the full_path() and in_progress() can be
@@ -137,7 +160,10 @@ class CONTENT_EXPORT BaseFile {
   // renamed.
   bool in_progress() const { return file_.IsValid(); }
 
-  // Returns the number of bytes in the file pointed to by full_path().
+  // Returns the number of bytes that has been written so far. If |access_mode_|
+  // is EXCLUSIVE, this should always be equal to the file size. If
+  // |access_mode_| is SHARED, this should not be larger than the file size
+  // as the file may contain holes in it.
   int64_t bytes_so_far() const { return bytes_so_far_; }
 
   std::string DebugString() const;
@@ -148,13 +174,15 @@ class CONTENT_EXPORT BaseFile {
 
   // Creates and opens the file_ if it is invalid.
   //
-  // If |hash_so_far| is not empty, then it must match the SHA-256 hash of the
-  // first |bytes_so_far_| bytes of |file_|. If there's a hash mismatch, Open()
-  // fails with DOWNLOAD_INTERRUPT_REASON_FILE_HASH_MISMATCH.
+  // If |access_mode_| is EXCLUSIVE and |hash_so_far| is not empty, then it must
+  // match the SHA-256 hash of the first |bytes_so_far_| bytes of |file_|. If
+  // there's a hash mismatch, Open() fails with
+  // DOWNLOAD_INTERRUPT_REASON_FILE_HASH_MISMATCH.
   //
   // If the opened file is shorter than |bytes_so_far_| bytes, then Open() fails
-  // with DOWNLOAD_INTERRUPT_REASON_FILE_TOO_SHORT. If the opened file is
-  // longer, then the file is truncated to |bytes_so_far_|.
+  // with DOWNLOAD_INTERRUPT_REASON_FILE_TOO_SHORT. If the opened file is longer
+  // and |access_mode_| is EXCLUSIVE, then the file is truncated to
+  // |bytes_so_far_|.
   //
   // Open() can fail for other reasons as well. In that case, it returns a
   // relevant interrupt reason. Unless Open() return
@@ -168,6 +196,9 @@ class CONTENT_EXPORT BaseFile {
   // Resets file_.
   void ClearFile();
 
+  // Writes data at the current position.
+  DownloadInterruptReason WriteAtCurrentPos(const char* data, size_t data_len);
+
   // Platform specific method that moves a file to a new path and adjusts the
   // security descriptor / permissions on the file to match the defaults for the
   // new directory.
@@ -177,7 +208,8 @@ class CONTENT_EXPORT BaseFile {
   // Split out from CurrentSpeed to enable testing.
   int64_t CurrentSpeedAtTime(base::TimeTicks current_time) const;
 
-  // Verifies that:
+  // This function is only useful when |access_mode_| is EXCLUSIVE. It verifies

[asanka, 2017/02/16 18:46:26]

Once the download is complete, we'd still need use this to calculate the hash of
the entire file. Otherwise we'll break SafeBrowsing.

[qinmin, 2017/02/16 21:35:50]

Added the code to calculate the hash when finish() is called.

+  // that:
   // * Size of the file represented by |file_| is at least |bytes_so_far_|.
   //
   // * If |hash_to_expect| is not empty, then the result of hashing the first
@@ -222,6 +254,8 @@ class CONTENT_EXPORT BaseFile {
   // won't delete it on destruction.
   bool detached_ = false;
 
+  AccessMode access_mode_ = EXCLUSIVE;
+
   net::NetLogWithSource net_log_;
 
   DISALLOW_COPY_AND_ASSIGN(BaseFile);