strokedb 0.0.2.1 → 0.0.2.2

data/lib/{util → strokedb/util}/serialization.rb

@@ -1,4 +1,25 @@
 module StrokeDB
+
+  class ::Array
+    def to_raw
+      map do |v|
+        v.respond_to?(:to_raw) ? v.to_raw : v
+      end
+    end
+  end
+
+  class ::Hash
+    def to_raw
+      raw_hash = {}
+      map do |k,v|
+        _k = k.respond_to?(:to_raw) ? k.to_raw : k
+        _v = v.respond_to?(:to_raw) ? v.to_raw : v
+        raw_hash[_k] = _v
+      end
+      raw_hash
+    end
+  end
+  
   module JsonSerializationMethod
     def serialize(x)
       x.to_json

data/lib/strokedb/util/uuid.rb

@@ -0,0 +1,159 @@
+begin
+  module StrokeDB; end
+  class FastUUID
+    require 'rubygems'
+    require 'inline'
+
+    C_FLAGS = `uuid-config --cflags`.chomp
+    LD_FLAGS = `uuid-config --ldflags --libs`.chomp
+    
+    inline(:C) do |builder|
+      builder.add_compile_flags C_FLAGS
+      builder.add_link_flags LD_FLAGS
+      builder.prefix %{
+        #include "dlfcn.h"
+        #include "uuid.h"
+
+        #define PROLOG(size)               \\
+          uuid_t *uuid;                    \\
+          char str[size], *strptr = str;   \\
+          size_t len = sizeof(str);        \\
+          uuid_create(&uuid)
+
+        #define EPILOG(format, size)                 \\
+          uuid_export(uuid, format, &strptr, &len),  \\
+          uuid_destroy(uuid),                        \\
+          rb_str_new(str,size)
+
+        #define CHARS  EPILOG(UUID_FMT_STR, 36)
+        #define BINARY EPILOG(UUID_FMT_BIN, 16)
+      }
+      builder.c %{
+        VALUE random_uuid() 
+        {
+          PROLOG(40);
+          uuid_make(uuid, UUID_MAKE_V4);
+          return CHARS;
+        }
+      }
+      builder.c %{
+        VALUE random_uuid_raw()
+        {
+          PROLOG(20);
+          uuid_make(uuid, UUID_MAKE_V4);
+          return BINARY;
+        }
+      }
+      builder.c %{
+        VALUE sha1_uuid(VALUE oid)
+        {
+          uuid_t *uuid_ns;
+          PROLOG(40);
+          
+          uuid_create(&uuid_ns);
+          uuid_load(uuid_ns, "ns:OID");
+          uuid_make(uuid, UUID_MAKE_V5, uuid_ns, StringValuePtr(oid));
+
+          uuid_destroy(uuid_ns);
+          
+          return CHARS;
+        }
+      }
+      
+      builder.c %{
+        VALUE uuid_to_raw(VALUE r_uuid)
+        {
+          PROLOG(20);
+          uuid_import(uuid, UUID_FMT_STR, StringValuePtr(r_uuid), 36);
+          return BINARY;
+        }
+      }
+
+      builder.c %{
+        VALUE uuid_to_formatted(VALUE r_uuid)
+        {
+          PROLOG(40);
+          uuid_import(uuid, UUID_FMT_BIN, StringValuePtr(r_uuid), 36);
+          return CHARS;
+        }
+      }
+
+    end
+  end
+  
+  FAST_UUID = FastUUID.new
+
+  class ::String
+    # Convert to raw (16 bytes) string (self can be already raw or formatted).
+    def to_raw_uuid
+      if size == 16 
+        self.freeze 
+      else
+        FAST_UUID.uuid_to_raw(self)
+      end 
+    end
+    # Convert to formatted string (self can be raw or already formatted).
+    def to_formatted_uuid
+      if size == 16 
+        FAST_UUID.uuid_to_formatted(self)
+      else
+        self.freeze 
+      end 
+    end
+  end
+  
+  module StrokeDB::Util
+
+    def self.random_uuid
+      ::FAST_UUID.random_uuid
+    end
+    def self.random_uuid_raw
+      ::FAST_UUID.random_uuid_raw
+    end
+    def self.sha1_uuid(oid)
+      ::FAST_UUID.sha1_uuid(oid)
+    end
+    
+  end
+
+
+
+rescue NotImplementedError, CompilationError
+  if $!.is_a?(CompilationError)
+    puts "# Can't compile C code, make sure you have ossp-uuid installed"
+  else
+    puts "# Error: #{$!.message}"
+  end
+  puts "# Falling back to uuidtools gem"
+  require 'uuidtools'
+  module StrokeDB::Util
+
+    def self.random_uuid
+      ::UUID.random_create.to_s
+    end
+    def self.random_uuid_raw
+      ::UUID.random_create.raw
+    end
+    def self.sha1_uuid(oid)
+      ::UUID.sha1_create(UUID_OID_NAMESPACE, oid).to_s
+    end
+
+    class ::String
+      # Convert to raw (16 bytes) string (self can be already raw or formatted).
+      def to_raw_uuid
+        size == 16 ? self.freeze : ::UUID.parse(self).raw
+      end
+      # Convert to formatted string (self can be raw or already formatted).
+      def to_formatted_uuid
+        size == 16 ? ::UUID.parse_raw(self).to_s : self.freeze
+      end
+    end
+
+  end
+end
+
+module StrokeDB::Util
+  RAW_UUID_SIZE       = random_uuid_raw.size
+  FORMATTED_UUID_SIZE = random_uuid.size
+end
+

data/lib/{view → strokedb}/view.rb

@@ -1,5 +1,5 @@
 module StrokeDB
-  View = Meta.new(:uuid => VIEW_UUID) do
+  View = Meta.new do
     attr_accessor :map_with_proc
     attr_reader :reduce_with_proc
 
@@ -22,7 +22,7 @@ module StrokeDB
     end
 
   end
-  ViewCut = Meta.new(:uuid => VIEWCUT_UUID) do
+  ViewCut = Meta.new do
 
     on_new_document do |cut|
       cut.instance_eval do

data/lib/strokedb/volumes.rb

@@ -0,0 +1,5 @@
+require 'volumes/archive_volume'
+require 'volumes/block_volume'
+require 'volumes/distributed_pointer'
+require 'volumes/map_volume'
+require 'volumes/fixed_length_skiplist_volume'

data/lib/strokedb/volumes/archive_volume.rb

@@ -0,0 +1,165 @@
+require 'readbytes'
+module StrokeDB
+  class ArchiveVolume
+    attr_reader :file_path, :tail
+    
+    if defined?(::Spec)
+      DEFAULT_SIZE = 512*1024
+    else 
+      DEFAULT_SIZE = 64*1024*1024
+    end
+    
+    DEFAULT_PATH = "."
+    
+    # Open a volume in a directory +:path+, with UUID +:uuid+
+    # and a specified +:size+. If the file does not exist, it is created 
+    # and filled with zero bytes up to the specified size. 
+    # Otherwise, it is just opened and ready for reads and writes.
+    #
+    # Defaults:
+    #   :path => "."
+    #   :size => 64 Mb
+    #
+    # Example:
+    #   DataVolume.new(:uuid => uuid, :path => "/var/dir", :size => 1024)
+    #
+    def initialize(options = {})
+      @options = options.stringify_keys.reverse_merge('size' => DEFAULT_SIZE, 'path' => DEFAULT_PATH)
+      initialize_file
+    end
+    
+    # Read a record sitting in a +position+ in the volume file.
+    # Record length is stored in a first 4 bytes before the record.
+    # 
+    def read(position)
+      @file.seek(position)
+      size = @file.readbytes(4).unpack('N').first
+      @file.readbytes(size)
+    end
+    
+    # Write some data to the end of the file.
+    # Returns record position.
+    #
+    def insert(data)
+      raise VolumeCapacityExceeded if (new_tail = @tail + 4 + data.size) > size
+      @file.seek(@tail)
+      @file.write([data.size].pack('N') + data)
+      t = @tail
+      @tail = new_tail
+      write_tail(@file, @tail)
+      t
+    end
+    
+    # Close the volume file. You cannot read/insert after that operation.
+    # In such case, VolumeClosedException is raised. 
+    # Call DataVolume.new to open volume again.
+    #
+    def close!
+      safe_close
+    end
+    
+    # Close and delete the volume file. You cannot read/insert after that 
+    # operation. In such case, VolumeClosedException is raised.
+    #
+    def delete!
+      safe_close
+      File.delete(@file_path)
+    end
+
+    def path
+      @options['path']
+    end
+    
+    def size
+      @options['size']
+    end
+    
+    def uuid
+      case @options['uuid'] 
+      when /^#{UUID_RE}$/
+        @options['uuid']
+      when nil
+        @options['uuid'] = Util.random_uuid
+      else
+        @options['uuid'] = @options['uuid'].to_formatted_uuid
+      end
+    end
+    
+    def raw_uuid
+      @raw_uuid ||= uuid.to_raw_uuid
+    end
+    
+    # VolumeClosedException is thrown when you call +read+ or +insert+
+    # method on a closed or deleted volume.
+    #
+    class VolumeClosedException < Exception; end
+    
+    # VolumeCapacityExceeded is thrown when you are trying to +insert+ data
+    # that is larger than available capacity.
+    class VolumeCapacityExceeded < Exception; end
+        
+  private
+
+    def initialize_file
+      @file_path = File.join(path, hierarchify(uuid) + ".av")
+      unless File.exist?(@file_path)
+        create_file(@file_path, size) 
+      else
+        @file = File.open(@file_path, File::RDWR)
+      end
+      @tail = read_tail(@file)
+    end
+    
+    # Create file skeleton filled with zeros with a prefix 
+    # containing current file tail.
+    #
+    def create_file(path, size)
+      FileUtils.mkdir_p(File.dirname(path))
+      @file = File.open(path, File::CREAT | File::EXCL | File::RDWR) 
+      @file.truncate(size)
+      write_tail(@file, 4) # 4 is a size of long type.
+    end
+    
+    # Close the file if it is opened and remove
+    # +read+ and +write+ methods from the instance.
+    #
+    def safe_close
+      @file.close if @file
+      @file = nil
+      class <<self
+        alias :read  :raise_volume_closed
+        alias :insert :raise_volume_closed
+      end
+    end
+    
+    # +read+ and +write+ methods are aliased to this
+    # when file is closed or deleted.
+    #
+    def raise_volume_closed(*args)
+      raise VolumeClosedException, "Throw this object away and instantiate another one."
+    end
+    public :raise_volume_closed
+    
+    # Transform filename "aabbccdd" into "aa/bb/aabbccdd"
+    # for faster access to a bunch of datavolumes.
+    #
+    def hierarchify(filename)
+      File.join(filename[0,2], filename[2,2], filename)
+    end
+    
+    # Read current file end position ("tail") from the file header.
+    # 
+    def read_tail(f)
+      f.seek(0)
+      f.readbytes(4).unpack('N').first
+    end
+    
+    # Update file's end position.
+    #
+    def write_tail(f, pos)
+      f.seek(0)
+      f.write([pos].pack('N'))
+      pos
+    end
+  end
+end

data/lib/strokedb/volumes/block_volume.rb

@@ -0,0 +1,169 @@
+require 'readbytes'
+module StrokeDB
+  
+  # TODO: inherit from the common AbstractVolume
+  class BlockVolume
+    attr_reader :file_path, :blocks_count
+    
+    HEADER_LENGTH = 8 # block_size, blocks_count
+    DEFAULT_BLOCKS_COUNT = 1024
+    DEFAULT_PATH = "."
+        
+    # Open a volume in a directory +:path+, with UUID +:uuid+
+    # and a specified +:block_size+. If the file does not exist, it is created 
+    # and filled with zero bytes up to the specified size. 
+    # Otherwise, it is just opened and ready for reads and writes.
+    # File contains +block_count+ blocks of :block_size: bytes size.
+    # When insertion is done to a new position, file is autoextended.
+    #
+    # Required params: +:block_size+ and +:uuid+
+    # Default +:path+ is ".", +:blocks_count+ is 1024
+    #
+    # Example:
+    #   DataVolume.new(:uuid => uuid, :path => "/var/dir", :block_size => 1024)
+    #
+    def initialize(options = {})
+      @options = options.stringify_keys.reverse_merge(
+        'path' => DEFAULT_PATH, 
+        'blocks_count' => DEFAULT_BLOCKS_COUNT)
+      initialize_file
+    end
+    
+    # Read a record sitting in a +position+ in the volume file.
+    # Record length is stored in a first 4 bytes before the record.
+    # 
+    def read(index)
+      csize = @block_size
+      @file.seek(HEADER_LENGTH + index * csize)
+      @file.readbytes(csize)
+    end
+    
+    # Write some data to the end of the file.
+    # Returns record position.
+    #
+    def insert(index, data)
+      extend_volume! if index >= @blocks_count
+      csize = @block_size
+      @file.seek(HEADER_LENGTH + index * csize)
+      @file.write(data)
+      self
+    end
+    
+    # Close the volume file. You cannot read/insert after that operation.
+    # In such case, VolumeClosedException is raised. 
+    # Call DataVolume.new to open volume again.
+    #
+    def close!
+      safe_close
+    end
+    
+    # Close and delete the volume file. You cannot read/insert after that 
+    # operation. In such case, VolumeClosedException is raised.
+    #
+    def delete!
+      safe_close
+      File.delete(@file_path)
+    end
+
+    def path
+      @options['path']
+    end
+    
+    def block_size
+      @options['block_size']
+    end
+    
+    def blocks_count
+      @options['blocks_count']
+    end
+    
+    def uuid
+      case @options['uuid'] 
+      when /^#{UUID_RE}$/
+        @options['uuid']
+      when nil
+        @options['uuid'] = Util.random_uuid
+      else
+        @options['uuid'] = @options['uuid'].to_formatted_uuid
+      end
+    end
+    
+    # VolumeClosedException is thrown when you call +read+ or +insert+
+    # method on a closed or deleted volume.
+    #
+    class VolumeClosedException < Exception; end
+      
+  private
+
+    def initialize_file
+      @file_path = File.join(path, hierarchify(uuid) + ".blocks")
+      create_file(@file_path, block_size, blocks_count) unless File.exist?(@file_path)
+      @file = File.open(@file_path, File::RDWR)
+      @block_size, @blocks_count = read_header(@file)
+      if @block_size != block_size && block_size != nil
+        raise "Block size collision! Declared #{block_size} bytes, but actually file is formatted by #{@block_size}-byte blocks."
+      end
+    end
+    
+    # Extends volume to a double size.
+    #
+    def extend_volume!
+      @file.close
+      @blocks_count *= 2
+      create_file(@file_path, @block_size, @blocks_count)
+      @file = File.open(@file_path, File::RDWR)
+    end
+    
+    # Create file skeleton filled with zeros with a prefix 
+    # containing current file tail.
+    #
+    def create_file(path, block_size, blocks)
+      FileUtils.mkdir_p(File.dirname(path))
+      File.open(path, File::CREAT | File::WRONLY) do |f|
+        f.truncate(HEADER_LENGTH + block_size*blocks)
+        write_header(f, block_size, blocks)
+      end
+    end
+    
+    # Close the file if it is opened and remove
+    # +read+ and +write+ methods from the instance.
+    #
+    def safe_close
+      @file.close if @file
+      @file = nil
+      class <<self
+        alias :read  :raise_volume_closed
+        alias :insert :raise_volume_closed
+      end
+    end
+    
+    # +read+ and +insert+ methods are aliased to this
+    # when file is closed or deleted.
+    #
+    def raise_volume_closed(*args)
+      raise VolumeClosedException, "Throw this object away and instantiate another one."
+    end
+    public :raise_volume_closed
+    
+    # Transform filename "aabbccdd" into "aa/bb/aabbccdd"
+    # for faster access to a bunch of datavolumes.
+    #
+    def hierarchify(filename)
+      File.join(filename[0,2], filename[2,2], filename)
+    end
+    
+    # Read current file block size and blocks count.
+    # 
+    def read_header(f)
+      f.seek(0)
+      f.readbytes(8).unpack('N2')
+    end
+    
+    # Update file's end position.
+    #
+    def write_header(f, block_size, blocks)
+      f.seek(0)
+      f.write([block_size, blocks].pack('N2'))
+    end
+  end
+end