small_file_cache.diff

diff --git a/swiftonfile/swift/common/utils.py b/swiftonfile/swift/common/utils.py
index c7ad8bf..5cb30ae 100644
--- a/swiftonfile/swift/common/utils.py
+++ b/swiftonfile/swift/common/utils.py
@@ -292,7 +292,7 @@ def _get_etag(path_or_fd):
     return etag
 
 
-def get_object_metadata(obj_path_or_fd, stats=None):
+def get_object_metadata(obj_path_or_fd, stats=None, etag=None):
     """
     Return metadata of object.
     """
@@ -311,6 +311,10 @@ def get_object_metadata(obj_path_or_fd, stats=None):
         metadata = {}
     else:
         is_dir = stat.S_ISDIR(stats.st_mode)
+        if not is_dir:
+            etag = etag or _get_etag(obj_path_or_fd)
+        else:
+            etag = md5().hexdigest()
         metadata = {
             X_TYPE: OBJECT,
             X_TIMESTAMP: normalize_timestamp(stats.st_ctime),
@@ -318,7 +322,7 @@ def get_object_metadata(obj_path_or_fd, stats=None):
             X_OBJECT_TYPE: DIR_NON_OBJECT if is_dir else FILE,
             X_CONTENT_LENGTH: 0 if is_dir else stats.st_size,
             X_MTIME: 0 if is_dir else normalize_timestamp(stats.st_mtime),
-            X_ETAG: md5().hexdigest() if is_dir else _get_etag(obj_path_or_fd)}
+            X_ETAG: etag}
     return metadata
 
 
@@ -333,11 +337,12 @@ def restore_metadata(path, metadata, meta_orig):
     return meta_new
 
 
-def create_object_metadata(obj_path_or_fd, stats=None, existing_meta={}):
+def create_object_metadata(obj_path_or_fd, stats=None, etag=None,
+                           existing_meta={}):
     # We must accept either a path or a file descriptor as an argument to this
     # method, as the diskfile modules uses a file descriptior and the DiskDir
     # module (for container operations) uses a path.
-    metadata_from_stat = get_object_metadata(obj_path_or_fd, stats)
+    metadata_from_stat = get_object_metadata(obj_path_or_fd, stats, etag)
     return restore_metadata(obj_path_or_fd, metadata_from_stat, existing_meta)
 
 
diff --git a/swiftonfile/swift/obj/diskfile.py b/swiftonfile/swift/obj/diskfile.py
index 4d00ef9..23dc149 100644
--- a/swiftonfile/swift/obj/diskfile.py
+++ b/swiftonfile/swift/obj/diskfile.py
@@ -25,6 +25,7 @@ import logging
 import time
 from uuid import uuid4
 from eventlet import sleep
+from hashlib import md5
 from contextlib import contextmanager
 from swiftonfile.swift.common.exceptions import AlreadyExistsAsFile, \
     AlreadyExistsAsDir
@@ -40,13 +41,14 @@ from swiftonfile.swift.common.fs_utils import do_fstat, do_open, do_close, \
     do_fadvise64, do_rename, do_fdatasync, do_lseek, do_mkdir
 from swiftonfile.swift.common.utils import read_metadata, write_metadata, \
     validate_object, create_object_metadata, rmobjdir, dir_is_object, \
-    get_object_metadata, write_pickle
+    get_object_metadata, write_pickle, CHUNK_SIZE
 from swiftonfile.swift.common.utils import X_CONTENT_TYPE, \
     X_TIMESTAMP, X_TYPE, X_OBJECT_TYPE, FILE, OBJECT, DIR_TYPE, \
     FILE_TYPE, DEFAULT_UID, DEFAULT_GID, DIR_NON_OBJECT, DIR_OBJECT, \
     X_ETAG, X_CONTENT_LENGTH, X_MTIME
 from swift.obj.diskfile import DiskFileManager as SwiftDiskFileManager
 from swift.obj.diskfile import get_async_dir
+from six import BytesIO as six_BytesIO
 
 # FIXME: Hopefully we'll be able to move to Python 2.7+ where O_CLOEXEC will
 # be back ported. See http://www.python.org/dev/peps/pep-0433/
@@ -547,6 +549,24 @@ class DiskFileReader(object):
                 do_close(fd)
 
 
+class SmallDiskFileReader(object):
+    def __init__(self, fd, smallfile):
+        self._fd = fd
+        self._smallfile = smallfile
+
+    def __iter__(self):
+        try:
+            yield self._smallfile.getvalue()
+        finally:
+            self.close()
+
+    def close(self):
+        if self._fd is not None:
+            fd, self._fd = self._fd, None
+            if fd > -1:
+                do_close(fd)
+
+
 class DiskFile(object):
     """
     Manage object files on disk.
@@ -602,6 +622,10 @@ class DiskFile(object):
 
         self._data_file = os.path.join(self._put_datadir, self._obj)
 
+        # Small file optimization
+        self._small_file_size = mgr.disk_chunk_size
+        self._small_file = None
+
     def open(self):
         """
         Open the object.
@@ -636,8 +660,16 @@ class DiskFile(object):
 
             self._metadata = read_metadata(self._fd)
             if not validate_object(self._metadata, self._stat):
+                etag = None
+                self._small_file_size
+                if not self._is_dir and \
+                        obj_size <= self._small_file_size:
+                    chunk = do_read(self._fd, CHUNK_SIZE)
+                    etag = md5(chunk).hexdigest()
+                    # https://github.com/tornadoweb/tornado/issues/1110
+                    self._small_file = six_BytesIO(chunk)
                 self._metadata = create_object_metadata(self._fd, self._stat,
-                                                        self._metadata)
+                                                        etag, self._metadata)
             assert self._metadata is not None
             self._filter_metadata()
 
@@ -762,10 +794,13 @@ class DiskFile(object):
         """
         if self._metadata is None:
             raise DiskFileNotOpen()
-        dr = DiskFileReader(
-            self._fd, self._threadpool, self._mgr.disk_chunk_size,
-            self._obj_size, self._mgr.keep_cache_size,
-            iter_hook=iter_hook, keep_cache=keep_cache)
+        if self._small_file:
+            dr = SmallDiskFileReader(self._fd, self._small_file)
+        else:
+            dr = DiskFileReader(
+                self._fd, self._threadpool, self._mgr.disk_chunk_size,
+                self._obj_size, self._mgr.keep_cache_size,
+                iter_hook=iter_hook, keep_cache=keep_cache)
         # At this point the reader object is now responsible for closing
         # the file pointer.
         self._fd = None