From 6858e18675086feb7cf0261494f738719d44818b Mon Sep 17 00:00:00 2001
From: grourk <grourk@gmail.com>
Date: Fri, 17 Jun 2011 01:35:52 -0700
Subject: [PATCH 1/2] Allow skipping tombstones when iterating over entries. 
 Also, adding method to iterate over keys with more efficient checking for
 tombstone.

---
 lib/bitcask.rb           | 16 ++++++++++++++--
 lib/bitcask/data_file.rb | 27 +++++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 2 deletions(-)

diff --git a/lib/bitcask.rb b/lib/bitcask.rb
index 33f540f..c18b8b1 100644
--- a/lib/bitcask.rb
+++ b/lib/bitcask.rb
@@ -44,11 +44,23 @@ def data_files
     end
   end
 
+  # Iterates over all keys in keydir. Yields key.
+  def each_key(skip_tombstones=false)
+    @keydir.each do |key, index|
+      if skip_tombstones
+        key, is_tombstone = @keydir.data_files[index.file_id].read_key(index.value_pos, index.value_sz)
+        yield key unless is_tombstone
+      else
+        yield key
+      end
+    end
+  end
+
   # Iterates over all keys in keydir. Yields key, value pairs.
-  def each
+  def each(skip_tombstones=false)
     @keydir.each do |key, index|
       entry = @keydir.data_files[index.file_id][index.value_pos, index.value_sz]
-      yield [entry.key, entry.value]
+      yield [entry.key, entry.value] unless skip_tombstones and entry.value == TOMBSTONE
     end
   end
 
diff --git a/lib/bitcask/data_file.rb b/lib/bitcask/data_file.rb
index 2496bce..b6140bd 100644
--- a/lib/bitcask/data_file.rb
+++ b/lib/bitcask/data_file.rb
@@ -63,6 +63,33 @@ def pos
   end
   alias tell pos
 
+  def read_key(offset, size = nil)
+    seek offset
+
+    if size
+      f = StringIO.new @file.read(size)
+    else
+      f = @file
+    end
+
+    # Parse header
+    header = f.read(14) or return
+    crc, tstamp, ksz, value_sz = header.unpack "NNnN"
+
+    # Read key
+    key = f.read ksz
+
+    if value_sz != Bitcask::TOMBSTONE.size
+      is_tombstone = false
+    else
+      value = f.read value_sz
+      raise Bitcask::ChecksumError unless crc == Zlib.crc32(header[4..-1] + key + value)
+      is_tombstone = value == Bitcask::TOMBSTONE
+    end
+
+    return key, is_tombstone
+  end
+
   # Returns a single Entry read from the current offset, and advances to the
   # next.
   #

From 68844499deb1164814a99c1c21a5c592cadfd0c2 Mon Sep 17 00:00:00 2001
From: grourk <grourk@gmail.com>
Date: Fri, 17 Jun 2011 02:16:22 -0700
Subject: [PATCH 2/2] Woops, don't read value from file in read_key if we don't
 have to.

---
 lib/bitcask.rb           | 2 +-
 lib/bitcask/data_file.rb | 8 ++------
 2 files changed, 3 insertions(+), 7 deletions(-)

diff --git a/lib/bitcask.rb b/lib/bitcask.rb
index c18b8b1..b363bc2 100644
--- a/lib/bitcask.rb
+++ b/lib/bitcask.rb
@@ -48,7 +48,7 @@ def data_files
   def each_key(skip_tombstones=false)
     @keydir.each do |key, index|
       if skip_tombstones
-        key, is_tombstone = @keydir.data_files[index.file_id].read_key(index.value_pos, index.value_sz)
+        key, is_tombstone = @keydir.data_files[index.file_id].read_key(index.value_pos)
         yield key unless is_tombstone
       else
         yield key
diff --git a/lib/bitcask/data_file.rb b/lib/bitcask/data_file.rb
index b6140bd..970f907 100644
--- a/lib/bitcask/data_file.rb
+++ b/lib/bitcask/data_file.rb
@@ -63,14 +63,10 @@ def pos
   end
   alias tell pos
 
-  def read_key(offset, size = nil)
+  def read_key(offset)
     seek offset
 
-    if size
-      f = StringIO.new @file.read(size)
-    else
-      f = @file
-    end
+    f = @file
 
     # Parse header
     header = f.read(14) or return