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..e1714c6ca397ef8513096d9ed9efe85828603fcf 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,
+  };
+
   // May be constructed on any thread.  All other routines (including
   // destruction) must occur on the FILE thread.
   BaseFile(const net::NetLogWithSource& net_log);
@@ -60,36 +67,45 @@ class CONTENT_EXPORT BaseFile {
   //     perfect way to come up with a canonical path for a file. So 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
-  //     fail with DOWNLOAD_INTERRUPT_REASON_FILE_TOO_SHORT.
+  // |offset|: If a file is provided (via |full_path| or |file|), then this
+  //     argument specifies the starting position to write 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. However, if the file is
+  //     shorter, then the operation will fail with
+  //     DOWNLOAD_INTERRUPT_REASON_FILE_TOO_SHORT. If |access_mode_| is SHARED,
+  //     file size check is ignored as the writer is expecting other writers to
+  //     fill the gaps or append more data.
   //
-  // |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
+  // |hash_so_far|: If |offset| is non-zero, this specifies the SHA-256
+  //     hash of the first |offset| bytes of the target file. If
+  //     specified, BaseFile will read the first |offset| 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.
+  //     is also specified or |access_mode| is equal to SHARED.
   //
   // |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
+  //     preexisting target file and it is non-empty (i.e. offset 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|.
+  //     |offset| has been accurately hashed into |hash_state| and will
+  //     ignore |hash_so_far|. Not used if |access_mode| is equal to 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 a |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,
+      int64_t offset,
       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.
-  DownloadInterruptReason AppendDataToFile(const char* data, size_t data_len);
+  // Write a new chunk of data to the current |offset_| inside the file. Returns
+  // a DownloadInterruptReason indicating the result of the operation.
+  DownloadInterruptReason WriteDataToFile(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
@@ -137,9 +153,16 @@ 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 only records the byte that is written by
+  // the current writer. So it is smaller than the file size as there are other
+  // writers around.
   int64_t bytes_so_far() const { return bytes_so_far_; }
 
+  // Returns the current offset that data will be written to.
+  int64_t offset() const { return offset_; }
+
   std::string DebugString() const;
 
  private:
@@ -148,13 +171,15 @@ 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 |offset_| 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_|.
+  // When |access_mode_| is EXCLUSIVE, If the opened file is shorter than
+  // |offset_| bytes, then Open() fails with
+  // DOWNLOAD_INTERRUPT_REASON_FILE_TOO_SHORT. If the opened file is
+  // longer, then the file is truncated to |offset_|.
   //
   // Open() can fail for other reasons as well. In that case, it returns a
   // relevant interrupt reason. Unless Open() return
@@ -177,14 +202,15 @@ class CONTENT_EXPORT BaseFile {
   // Split out from CurrentSpeed to enable testing.
   int64_t CurrentSpeedAtTime(base::TimeTicks current_time) const;
 
-  // Verifies that:
-  // * Size of the file represented by |file_| is at least |bytes_so_far_|.
+  // This function is only used when |access_mode_| is EXCLUSIVE. It verifies
+  // that:
+  // * Size of the file represented by |file_| is at least |offset_|.
   //
   // * If |hash_to_expect| is not empty, then the result of hashing the first
-  //   |bytes_so_far_| bytes of |file_| matches |hash_to_expect|.
+  //   |offset_| bytes of |file_| matches |hash_to_expect|.
   //
   // If the result is REASON_NONE, then on return |secure_hash_| is valid and
-  // is ready to hash bytes from offset |bytes_so_far_| + 1.
+  // is ready to hash bytes from offset |offset_| + 1.
   DownloadInterruptReason CalculatePartialHash(
       const std::string& hash_to_expect);
 
@@ -209,7 +235,12 @@ class CONTENT_EXPORT BaseFile {
   // OS file for writing
   base::File file_;
 
-  // Amount of data received up so far, in bytes.
+  // Current byte offset to write the data.
+  int64_t offset_ = 0;
+
+  // Amount of data received up so far, in bytes. This should be equal to
+  // |offset_| when |access_mode_| is EXCLUSIVE. If |access_mode_| is SHARED,
+  // this only tracks the bytes received by the current writer.
   int64_t bytes_so_far_ = 0;
 
   // Used to calculate hash for the file when calculate_hash_ is set.
@@ -222,6 +253,8 @@ 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);