cassandra 0.9.0 → 0.9.1

data/CHANGELOG

@@ -1,3 +1,7 @@
+v0.9.1 Support for secondary indexing. (jhermes)
+Fix bug in mock where we didn't support range queries. (therealadam)
+Support deletes in batch mutations. [blanquer]
+
 v0.9.0 cassandra 0.7 compat
 
 v0.8.2 Renamed :thrift_client_class option, UUID fix, 0.6 update, fixing cassanda install, much Mock fixes

data/README.rdoc

@@ -54,6 +54,10 @@ Insert into a column family. You can insert a `Cassandra::OrderedHash`, or a reg
 
   client.insert(:Users, "5", {'screen_name' => "buttonscat"})
 
+The 0.7 API insert() includes support for TTL on columns.  The following example inserts into a comlumn family with a time to live of 30 seconds.
+
+  client.insert(:Users, "5", {'screen_name' => "buttonscat"}, {:ttl=>30})
+  
 Insert into a super column family:
 
   client.insert(:UserRelationships, "5", {"user_timeline" => {UUID.new => "1"}})
@@ -64,6 +68,17 @@ Query a super column:
 
 The returned result will always be a Cassandra::OrderedHash.
 
+Create and delete a 2ary index:
+
+  client.create_index("Twitter", "Users", "revenue_generating_units", "LongType")
+  client.delete_index("Twitter", "Users", "revenue_generating_units"
+
+Create an index clause and query an indexed column family:
+
+  expr   = client.create_idx_expr("revenue_generating_units", 100, ">")
+  clause = client.create_idx_clause([expr])
+  client.get_indexed_slices(:Users, clause)
+
 See Cassandra for more methods.
 
 == Configuration

data/Rakefile

@@ -10,14 +10,12 @@ unless ENV['FROM_BIN_CASSANDRA_HELPER']
     p.dependencies = ['thrift_client >=0.6.0', 'json', 'rake', 'simple_uuid >=0.1.0']
     p.ignore_pattern = /^(data|vendor\/cassandra|cassandra|vendor\/thrift)/
     p.rdoc_pattern = /^(lib|bin|tasks|ext)|^README|^CHANGELOG|^TODO|^LICENSE|^COPYING$/
-    p.url = "http://blog.evanweaver.com/files/doc/fauna/cassandra/"
-    p.docs_host = "blog.evanweaver.com:~/www/bax/public/files/doc/"
   end
 end
 
 CASSANDRA_HOME = ENV['CASSANDRA_HOME'] || "#{ENV['HOME']}/cassandra"
 DOWNLOAD_DIR = "/tmp"
-DIST_URL = "http://www.takeyellow.com/apachemirror//cassandra/0.6.8/apache-cassandra-0.6.8-bin.tar.gz"
+DIST_URL = "http://www.fightrice.com/mirrors/apache/cassandra/0.6.12/apache-cassandra-0.6.12-bin.tar.gz"
 DIST_FILE = DIST_URL.split('/').last
 
 directory CASSANDRA_HOME

data/cassandra.gemspec

@@ -2,18 +2,18 @@
 
 Gem::Specification.new do |s|
   s.name = %q{cassandra}
-  s.version = "0.9.0"
+  s.version = "0.9.1"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0.8") if s.respond_to? :required_rubygems_version=
   s.authors = ["Evan Weaver, Ryan King"]
-  s.date = %q{2010-12-08}
+  s.date = %q{2011-04-06}
   s.default_executable = %q{cassandra_helper}
   s.description = %q{A Ruby client for the Cassandra distributed database.}
   s.email = %q{}
   s.executables = ["cassandra_helper"]
   s.extra_rdoc_files = ["CHANGELOG", "LICENSE", "README.rdoc", "bin/cassandra_helper", "lib/cassandra.rb", "lib/cassandra/0.6.rb", "lib/cassandra/0.6/cassandra.rb", "lib/cassandra/0.6/columns.rb", "lib/cassandra/0.6/protocol.rb", "lib/cassandra/0.7.rb", "lib/cassandra/0.7/cassandra.rb", "lib/cassandra/0.7/column_family.rb", "lib/cassandra/0.7/columns.rb", "lib/cassandra/0.7/keyspace.rb", "lib/cassandra/0.7/protocol.rb", "lib/cassandra/array.rb", "lib/cassandra/cassandra.rb", "lib/cassandra/columns.rb", "lib/cassandra/comparable.rb", "lib/cassandra/constants.rb", "lib/cassandra/debug.rb", "lib/cassandra/helpers.rb", "lib/cassandra/long.rb", "lib/cassandra/mock.rb", "lib/cassandra/ordered_hash.rb", "lib/cassandra/time.rb"]
   s.files = ["CHANGELOG", "LICENSE", "Manifest", "README.rdoc", "Rakefile", "bin/cassandra_helper", "conf/cassandra.in.sh", "conf/cassandra.yaml", "conf/log4j.properties", "conf/storage-conf.xml", "lib/cassandra.rb", "lib/cassandra/0.6.rb", "lib/cassandra/0.6/cassandra.rb", "lib/cassandra/0.6/columns.rb", "lib/cassandra/0.6/protocol.rb", "lib/cassandra/0.7.rb", "lib/cassandra/0.7/cassandra.rb", "lib/cassandra/0.7/column_family.rb", "lib/cassandra/0.7/columns.rb", "lib/cassandra/0.7/keyspace.rb", "lib/cassandra/0.7/protocol.rb", "lib/cassandra/array.rb", "lib/cassandra/cassandra.rb", "lib/cassandra/columns.rb", "lib/cassandra/comparable.rb", "lib/cassandra/constants.rb", "lib/cassandra/debug.rb", "lib/cassandra/helpers.rb", "lib/cassandra/long.rb", "lib/cassandra/mock.rb", "lib/cassandra/ordered_hash.rb", "lib/cassandra/time.rb", "test/cassandra_client_test.rb", "test/cassandra_mock_test.rb", "test/cassandra_test.rb", "test/comparable_types_test.rb", "test/eventmachine_test.rb", "test/ordered_hash_test.rb", "test/test_helper.rb", "vendor/0.6/gen-rb/cassandra.rb", "vendor/0.6/gen-rb/cassandra_constants.rb", "vendor/0.6/gen-rb/cassandra_types.rb", "vendor/0.7/gen-rb/cassandra.rb", "vendor/0.7/gen-rb/cassandra_constants.rb", "vendor/0.7/gen-rb/cassandra_types.rb", "cassandra.gemspec"]
-  s.homepage = %q{http://blog.evanweaver.com/files/doc/fauna/cassandra/}
+  s.homepage = %q{}
   s.rdoc_options = ["--line-numbers", "--inline-source", "--title", "Cassandra", "--main", "README.rdoc"]
   s.require_paths = ["lib"]
   s.rubyforge_project = %q{fauna}

data/conf/storage-conf.xml

@@ -162,6 +162,7 @@
   </DataFileDirectories>
   <CalloutLocation>data/cassandra/callouts</CalloutLocation>
   <StagingFileDirectory>data/cassandra/staging</StagingFileDirectory>
+  <SavedCachesDirectory>data/cassandra/saved_caches</SavedCachesDirectory>
 
 
   <!--

data/lib/cassandra/0.6/cassandra.rb

@@ -41,23 +41,21 @@ class Cassandra
   def reconnect!
     @servers = all_nodes
     @client = new_client
-    check_keyspace
-  end
-
-  def check_keyspace
-    unless (keyspaces = client.get_string_list_property("keyspaces")).include?(@keyspace)
-      raise AccessError, "Keyspace #{@keyspace.inspect} not found. Available: #{keyspaces.inspect}"
-    end
   end
 
   def all_nodes
     if @auto_discover_nodes
-      ips = ::JSON.parse(new_client.get_string_property('token map')).values
-      port = @servers.first.split(':').last
-      ips.map{|ip| "#{ip}:#{port}" }
+      temp_client = new_client
+      begin
+        ips = ::JSON.parse(temp_client.get_string_property('token map')).values
+        port = @servers.first.split(':').last
+        ips.map{|ip| "#{ip}:#{port}" }
+      ensure 
+        temp_client.disconnect!
+      end
     else
       @servers
     end
   end
 
-end
+end

data/lib/cassandra/0.6/columns.rb

@@ -31,5 +31,31 @@ class Cassandra
         )
       )
     end
+
+    # General info about a deletion object within a mutation
+    # timestamp - required. If this is the only param, it will cause deletion of the whole key at that TS
+    # supercolumn - opt. If passed, the deletes will only occur within that supercolumn (only subcolumns 
+    #               will be deleted). Otherwise the normal columns will be deleted.
+    # predicate - opt. Defines how to match the columns to delete. if supercolumn passed, the slice will 
+    #               be scoped to subcolumns of that supercolumn.
+    
+    # Deletes a single column from the containing key/CF (and possibly supercolumn), at a given timestamp. 
+    # Although mutations (as opposed to 'remove' calls) support deleting slices and lists of columns in one shot, this is not implemented here.
+    # The main reason being that the batch function takes removes, but removes don't have that capability...so we'd need to change the remove
+    # methods to use delete mutation calls...although that might have performance implications. We'll leave that refactoring for later.
+    def _delete_mutation(cf, column, subcolumn, timestamp, options={})
+      
+      deletion_hash = {:timestamp => timestamp}
+      if is_super(cf)
+         deletion_hash[:super_column] = column if column
+         deletion_hash[:predicate] = CassandraThrift::SlicePredicate.new(:column_names => [subcolumn]) if subcolumn
+     else
+         deletion_hash[:predicate] = CassandraThrift::SlicePredicate.new(:column_names => [column]) if column
+      end
+      CassandraThrift::Mutation.new(
+        :deletion => CassandraThrift::Deletion.new(deletion_hash)
+        )      
+    end
+    
   end
 end

data/lib/cassandra/0.7/cassandra.rb

@@ -1,4 +1,5 @@
 class Cassandra
+
   def self.DEFAULT_TRANSPORT_WRAPPER
     Thrift::FramedTransport
   end
@@ -16,7 +17,7 @@ class Cassandra
   end
 
   def keyspace=(ks)
-    client.set_keyspace(ks) if check_keyspace(ks)
+    client.set_keyspace(ks)
     @schema = nil; @keyspace = ks
   end
 
@@ -99,6 +100,16 @@ class Cassandra
     res
   end
 
+  def update_column_family(cf_def)
+    begin
+      res = client.system_update_column_family(cf_def)
+    rescue CassandraThrift::TimedOutException => te
+      puts "Timed out: #{te.inspect}"
+    end
+    @schema = nil
+    res
+  end
+
   def add_keyspace(ks_def)
     begin
       res = client.system_add_keyspace(ks_def)
@@ -137,6 +148,18 @@ class Cassandra
     res
   end
 
+  def update_keyspace(ks_def)
+    begin
+      res = client.system_update_keyspace(ks_def)
+    rescue CassandraThrift::TimedOutException => toe
+      puts "Timed out: #{toe.inspect}"
+    rescue Thrift::TransportException => te
+      puts "Timed out: #{te.inspect}"
+    end
+    @keyspaces = nil
+    res
+  end
+
   # Open a batch operation and yield self. Inserts and deletes will be queued
   # until the block closes, and then sent atomically to the server.  Supports
   # the <tt>:consistency</tt> option, which overrides the consistency set in
@@ -145,28 +168,83 @@ class Cassandra
     _, _, _, options =
       extract_and_validate_params(schema.cf_defs.first.name, "", [options], WRITE_DEFAULTS)
 
-    @batch = []
-    yield(self)
-    compact_mutations!
+      @batch = []
+      yield(self)
+      compacted_map,seen_clevels = compact_mutations!
+      clevel = if options[:consistency] != nil # Override any clevel from individual mutations if 
+                 options[:consistency]
+               elsif seen_clevels.length > 1 # Cannot choose which CLevel to use if there are several ones
+                 raise "Multiple consistency levels used in the batch, and no override...cannot pick one" 
+               else # if no consistency override has been provided but all the clevels in the batch are the same: use that one
+                 seen_clevels.first
+               end
 
-    @batch.each do |mutation|
-      case mutation.first
-      when :remove
-        _remove(*mutation[1])
-      else
-        _mutate(*mutation)
-      end
-    end
+      _mutate(compacted_map,clevel)
   ensure
     @batch = nil
   end
 
+### 2ary Indexing
+
+  def create_index(ks_name, cf_name, c_name, v_class)
+    cf_def = client.describe_keyspace(ks_name).cf_defs.find{|x| x.name == cf_name}
+    if !cf_def.nil? and !cf_def.column_metadata.find{|x| x.name == c_name}
+      c_def  = CassandraThrift::ColumnDef.new do |cd|
+        cd.name             = c_name
+        cd.validation_class = "org.apache.cassandra.db.marshal."+v_class
+        cd.index_type       = CassandraThrift::IndexType::KEYS
+      end
+      cf_def.column_metadata.push(c_def)
+      update_column_family(cf_def)
+    end
+  end
+
+  def drop_index(ks_name, cf_name, c_name)
+    cf_def = client.describe_keyspace(ks_name).cf_defs.find{|x| x.name == cf_name}
+    if !cf_def.nil? and cf_def.column_metadata.find{|x| x.name == c_name}
+      cf_def.column_metadata.delete_if{|x| x.name == c_name}
+      update_column_family(cf_def)
+    end
+  end
+
+  def create_idx_expr(c_name, value, op)
+    CassandraThrift::IndexExpression.new(
+      :column_name => c_name,
+      :value => value,
+      :op => (case op
+                when nil, "EQ", "eq", "=="
+                  CassandraThrift::IndexOperator::EQ
+                when "GTE", "gte", ">="
+                  CassandraThrift::IndexOperator::GTE
+                when "GT", "gt", ">"
+                  CassandraThrift::IndexOperator::GT
+                when "LTE", "lte", "<="
+                  CassandraThrift::IndexOperator::LTE
+                when "LT", "lt", "<"
+                  CassandraThrift::IndexOperator::LT
+              end ))
+  end
+
+  def create_idx_clause(idx_expressions, start = "")
+    CassandraThrift::IndexClause.new(
+      :start_key => start,
+      :expressions => idx_expressions)
+  end
+
+  # TODO: Supercolumn support.
+  def get_indexed_slices(column_family, idx_clause, *columns_and_options)
+    column_family, columns, _, options =
+      extract_and_validate_params(column_family, [], columns_and_options, READ_DEFAULTS)
+    _get_indexed_slices(column_family, idx_clause, columns, options[:count], options[:start],
+      options[:finish], options[:reversed], options[:consistency])
+  end
+
   protected
 
   def client
     if @client.nil? || @client.current_server.nil?
       reconnect!
-      @client.set_keyspace(@keyspace) if check_keyspace
+      @client.set_keyspace(@keyspace)
     end
     @client
   end
@@ -176,17 +254,16 @@ class Cassandra
     @client = new_client
   end
 
-  def check_keyspace(ks = @keyspace)
-    !(unless (_keyspaces = keyspaces()).include?(ks)
-      raise AccessError, "Keyspace #{ks.inspect} not found. Available: #{_keyspaces.inspect}"
-    end)
-  end
-
   def all_nodes
     if @auto_discover_nodes && !@keyspace.eql?("system")
-      ips = (new_client.describe_ring(@keyspace).map {|range| range.endpoints}).flatten.uniq
-      port = @servers.first.split(':').last
-      ips.map{|ip| "#{ip}:#{port}" }
+      temp_client = new_client
+      begin
+        ips = (temp_client.describe_ring(@keyspace).map {|range| range.endpoints}).flatten.uniq
+        port = @servers.first.split(':').last
+        ips.map{|ip| "#{ip}:#{port}" }
+      ensure
+        temp_client.disconnect!
+      end
     else
       @servers
     end

data/lib/cassandra/0.7/columns.rb

@@ -55,5 +55,31 @@ class Cassandra
         )
       )
     end
+
+    # General info about a deletion object within a mutation
+    # timestamp - required. If this is the only param, it will cause deletion of the whole key at that TS
+    # supercolumn - opt. If passed, the deletes will only occur within that supercolumn (only subcolumns 
+    #               will be deleted). Otherwise the normal columns will be deleted.
+    # predicate - opt. Defines how to match the columns to delete. if supercolumn passed, the slice will 
+    #               be scoped to subcolumns of that supercolumn.
+    
+    # Deletes a single column from the containing key/CF (and possibly supercolumn), at a given timestamp. 
+    # Although mutations (as opposed to 'remove' calls) support deleting slices and lists of columns in one shot, this is not implemented here.
+    # The main reason being that the batch function takes removes, but removes don't have that capability...so we'd need to change the remove
+    # methods to use delete mutation calls...although that might have performance implications. We'll leave that refactoring for later.
+    def _delete_mutation(cf, column, subcolumn, timestamp, options={})
+      
+      deletion_hash = {:timestamp => timestamp}
+      if is_super(cf)
+         deletion_hash[:super_column] = column if column
+         deletion_hash[:predicate] = CassandraThrift::SlicePredicate.new(:column_names => [subcolumn]) if subcolumn
+     else
+         deletion_hash[:predicate] = CassandraThrift::SlicePredicate.new(:column_names => [column]) if column
+      end
+      CassandraThrift::Mutation.new(
+        :deletion => CassandraThrift::Deletion.new(deletion_hash)
+        )      
+    end
+   
   end
 end

data/lib/cassandra/0.7/protocol.rb

@@ -85,6 +85,22 @@ class Cassandra
       _get_range(column_family, start, finish, count, consistency).collect{|i| i.key }
     end
 
+    # TODO: Supercolumn support
+    def _get_indexed_slices(column_family, idx_clause, column, count, start, finish, reversed, consistency)
+      column_parent = CassandraThrift::ColumnParent.new(:column_family => column_family)
+      if column
+        predicate = CassandraThrift::SlicePredicate.new(:column_names => [column])
+      else
+        predicate = CassandraThrift::SlicePredicate.new(:slice_range =>
+          CassandraThrift::SliceRange.new(
+            :reversed => reversed,
+            :count => count,
+            :start => start,
+            :finish => finish))
+      end
+      client.get_indexed_slices(column_parent, idx_clause, predicate, consistency)
+    end
+
     def each_key(column_family)
       column_parent = CassandraThrift::ColumnParent.new(:column_family => column_family.to_s)
       predicate = CassandraThrift::SlicePredicate.new(:column_names => [])

data/lib/cassandra/cassandra.rb

@@ -84,8 +84,10 @@ class Cassandra
   end
 
   def disconnect!
-    @client.disconnect!
-    @client = nil
+    if @client
+      @client.disconnect!
+      @client = nil
+    end
   end
 
   def keyspaces
@@ -136,16 +138,26 @@ class Cassandra
   # _mutate the element at the column_family:key:[column]:[sub_column]
   # path you request. Supports the <tt>:consistency</tt> and <tt>:timestamp</tt>
   # options.
+  # TODO: we could change this function or add another that support multi-column removal (by list or predicate)
   def remove(column_family, key, *columns_and_options)
     column_family, column, sub_column, options = extract_and_validate_params(column_family, key, columns_and_options, WRITE_DEFAULTS)
 
-    args = {:column_family => column_family}
-    columns = is_super(column_family) ? {:super_column => column, :column => sub_column} : {:column => column}
-    column_path = CassandraThrift::ColumnPath.new(args.merge(columns))
-
-    mutation = [:remove, [key, column_path, options[:timestamp] || Time.stamp, options[:consistency]]]
-
-    @batch ? @batch << mutation : _remove(*mutation[1])
+    if @batch
+      mutation_map = 
+        {
+          key => {
+            column_family => [ _delete_mutation(column_family , is_super(column_family)? column : nil , sub_column , options[:timestamp]|| Time.stamp) ]
+          }
+        }  
+      @batch << [mutation_map, options[:consistency]]
+    else 
+      # Let's continue using the 'remove' thrift method...not sure about the implications/performance of using the mutate instead
+      # Otherwise we coul get use the mutation_map above, and do _mutate(mutation_map, options[:consistency])
+      args = {:column_family => column_family}
+      columns = is_super(column_family) ? {:super_column => column, :column => sub_column} : {:column => column}
+      column_path = CassandraThrift::ColumnPath.new(args.merge(columns))
+      _remove(key, column_path, options[:timestamp] || Time.stamp, options[:consistency])
+    end
   end
 
 ### Read
@@ -214,8 +226,8 @@ class Cassandra
     end
   end
 
-  # Return a list of keys in the column_family you request. Requires the
-  # table to be partitioned with OrderPreservingHash. Supports the
+  # Return a list of keys in the column_family you request. Only works well if
+  # the table is partitioned with OrderPreservingPartitioner. Supports the
   # <tt>:count</tt>, <tt>:start</tt>, <tt>:finish</tt>, and <tt>:consistency</tt>
   # options.
   def get_range(column_family, options = {})
@@ -241,16 +253,16 @@ class Cassandra
 
     @batch = []
     yield(self)
-    compact_mutations!
-
-    @batch.each do |mutation|
-      case mutation.first
-      when :remove
-        _remove(*mutation[1])
-      else
-        _mutate(*mutation)
-      end
-    end
+    compacted_map,seen_clevels = compact_mutations!
+    clevel = if options[:consistency] != nil # Override any clevel from individual mutations if 
+               options[:consistency]
+             elsif seen_clevels.length > 1 # Cannot choose which CLevel to use if there are several ones
+               raise "Multiple consistency levels used in the batch, and no override...cannot pick one" 
+             else # if no consistency override has been provided but all the clevels in the batch are the same: use that one
+               seen_clevels.first
+             end
+        
+    _mutate(compacted_map,clevel)
   ensure
     @batch = nil
   end
@@ -261,13 +273,45 @@ class Cassandra
     "#{self.class}##{caller[0].split('`').last[0..-3]}"
   end
 
-  # Roll up queued mutations, to improve atomicity.
+  # Roll up queued mutations, to improve atomicity (and performance).
   def compact_mutations!
-    #TODO re-do this rollup
+    used_clevels = {} # hash that lists the consistency levels seen in the batch array. key is the clevel, value is true
+    by_key = {}
+    # @batch is an array of mutation_ops.
+    # A mutation op is a 2-item array containing [mutationmap, consistency_number]
+    # a mutation map is a hash, by key (string) that has a hash by CF name, containing a list of column_mutations)
+    @batch.each do |mutation_op|
+      # A single mutation op looks like:
+      # For an insert/update
+      #[ { key1 => 
+      #            { CF1 => [several of CassThrift:Mutation(colname,value,TS,ttl)]
+      #              CF2 => [several mutations]
+      #            },
+      #    key2 => {...} # Not sure if they can come batched like this...so there might only be a single key (and CF)
+      #      }, # [0]
+      #  consistency # [1] 
+      #]
+      # For a remove:
+      # [  :remove,  # [0]
+      #    [key, CassThrift:ColPath, timestamp, consistency ]  # [1]
+      # ]
+      mmap = mutation_op[0] # :remove OR a hash like {"key"=> {"CF"=>[mutationclass1,...] } }
+      used_clevels[mutation_op[1]]=true #save the clevel required for this operation
+
+      mmap.keys.each do |k|
+        by_key[k] = {} unless by_key.has_key? k #make sure the key exists
+        mmap[k].keys.each do |cf| # For each CF in that key
+          by_key[k][cf] = [] unless by_key[k][cf] != nil
+          by_key[k][cf].concat mmap[k][cf] # Append the list of mutations for that key and CF
+        end
+      end
+    end
+    # Returns the batch mutations map, and an array with the consistency levels 'seen' in the batch
+    [by_key, used_clevels.keys]
   end
 
   def new_client
     thrift_client_class.new(CassandraThrift::Cassandra::Client, @servers, @thrift_client_options)
   end
   
-end
+end

data/lib/cassandra/mock.rb

@@ -89,6 +89,7 @@ class Cassandra
       if column
         row[column]
       else
+        row = apply_range(row, column_family, options[:start], options[:finish])
         apply_count(row, options[:count], options[:reversed])
       end
     end
@@ -107,17 +108,7 @@ class Cassandra
           row = cf(column_family)[key] && cf(column_family)[key][column] ?
             cf(column_family)[key][column] :
             OrderedHash.new
-          if options[:start] || options[:finish]
-            start  = to_compare_with_type(options[:start],  column_family, false)
-            finish = to_compare_with_type(options[:finish], column_family, false)
-            ret = OrderedHash.new
-            row.keys.each do |key|
-              if (start.nil? || key >= start) && (finish.nil? || key <= finish)
-                ret[key] = row[key]
-              end
-            end
-            row = ret
-          end
+          row = apply_range(row, column_family, options[:start], options[:finish], false)
           apply_count(row, options[:count], options[:reversed])
         end
       elsif cf(column_family)[key]
@@ -142,7 +133,7 @@ class Cassandra
     def remove(column_family, key, *columns_and_options)
       column_family, column, sub_column, options = extract_and_validate_params_for_real(column_family, key, columns_and_options, WRITE_DEFAULTS)
       if @batch
-        @batch << [:remove, column_family, key, column]
+        @batch << [:remove, column_family, key, column, sub_column]
       else
         if column
           if sub_column
@@ -318,5 +309,18 @@ class Cassandra
         row
       end
     end
+
+    def apply_range(row, column_family, strt, fin, standard=true)
+      start  = to_compare_with_type(strt, column_family, standard)
+      finish = to_compare_with_type(fin,  column_family, standard)
+      ret = OrderedHash.new
+      row.keys.each do |key|
+        if (start.nil? || key >= start) && (finish.nil? || key <= finish)
+          ret[key] = row[key]
+        end
+      end
+      ret
+    end
+
   end
 end

data/lib/cassandra/ordered_hash.rb

@@ -156,13 +156,13 @@ class Cassandra
       super
     end
 
-    def delete_if
-      @timestamps.delete_if
+    def delete_if(&block)
+      @timestamps.delete_if(&block)
       super
     end
 
-    def reject!
-      @timestamps.reject!
+    def reject!(&block)
+      @timestamps.reject!(&block)
       super
     end
 
@@ -189,12 +189,5 @@ class Cassandra
     def inspect
       "#<OrderedHash #{super}\n#{@timestamps.inspect}>"
     end
-
-  private
-
-    def sync_keys!
-      @timestamps.delete_if {|k,v| !has_key?(k)}
-      super
-    end
   end
 end

data/test/cassandra_test.rb

@@ -1,5 +1,9 @@
 require File.expand_path(File.dirname(__FILE__) + '/test_helper')
 
+def cassandra07?
+  CassandraThrift::VERSION != '2.1.0'
+end
+
 class CassandraTest < Test::Unit::TestCase
   include Cassandra::Constants
 
@@ -7,10 +11,10 @@ class CassandraTest < Test::Unit::TestCase
     @twitter = Cassandra.new('Twitter', "127.0.0.1:9160", :retries => 2, :exception_classes => [])
     @twitter.clear_keyspace!
 
-    @blogs = Cassandra.new('Multiblog')
+    @blogs = Cassandra.new('Multiblog', "127.0.0.1:9160",:retries => 2, :exception_classes => [])
     @blogs.clear_keyspace!
 
-    @blogs_long = Cassandra.new('MultiblogLong')
+    @blogs_long = Cassandra.new('MultiblogLong', "127.0.0.1:9160",:retries => 2, :exception_classes => [])
     @blogs_long.clear_keyspace!
 
     @uuids = (0..6).map {|i| SimpleUUID::UUID.new(Time.at(2**(24+i))) }
@@ -102,6 +106,18 @@ class CassandraTest < Test::Unit::TestCase
     assert !@twitter.exists?(:Statuses, 'bogus', 'body')
   end
 
+  def test_get_value_with_range
+    10.times do |i|
+      @twitter.insert(:Statuses, key, {"body-#{i}" => 'v'})
+    end
+
+    assert_equal 5, @twitter.get(:Statuses, key, :count => 5).length
+    assert_equal 5, @twitter.get(:Statuses, key, :start => "body-5").length
+    assert_equal 5, @twitter.get(:Statuses, key, :finish => "body-4").length
+    assert_equal 5, @twitter.get(:Statuses, key, :start => "body-1", :count => 5).length
+    assert_equal 5, @twitter.get(:Statuses, key, :start => "body-0", :finish => "body-4", :count => 7).length
+  end
+
   def test_exists_with_only_key
     @twitter.insert(:Statuses, key, {'body' => 'v'})
     assert @twitter.exists?(:Statuses, key)
@@ -240,11 +256,12 @@ class CassandraTest < Test::Unit::TestCase
   end
 
   def test_remove_super_value
-    columns = {@uuids[1] => 'v1'}
+    columns = {@uuids[1] => 'v1', @uuids[2] => 'v2'}
+    column_name_to_remove = @uuids[2]
     @twitter.insert(:StatusRelationships, key, {'user_timelines' => columns})
-    @twitter.remove(:StatusRelationships, key, 'user_timelines', columns.keys.first)
-    assert_nil @twitter.get(:StatusRelationships, key, 'user_timelines', columns.keys.first)
-    assert_nil @twitter.get(:StatusRelationships, key, 'user_timelines').timestamps[columns.keys.first]
+    @twitter.remove(:StatusRelationships, key, 'user_timelines', column_name_to_remove)
+    assert_equal( columns.reject{|k,v| k == column_name_to_remove}, @twitter.get(:StatusRelationships, key, 'user_timelines') )
+    assert_nil @twitter.get(:StatusRelationships, key, 'user_timelines').timestamps[column_name_to_remove]
   end
 
   def test_clear_column_family
@@ -344,6 +361,12 @@ class CassandraTest < Test::Unit::TestCase
     k = key
 
     @twitter.insert(:Users, k + '1', {'body' => 'v1', 'user' => 'v1'})
+    initial_subcolumns = {@uuids[1] => 'v1', @uuids[2] => 'v2'}
+    @twitter.insert(:StatusRelationships, k, {'user_timelines' => initial_subcolumns, 'dummy_supercolumn' => {@uuids[5] => 'value'}})
+    assert_equal(initial_subcolumns, @twitter.get(:StatusRelationships, k, 'user_timelines'))
+    assert_equal({@uuids[5] => 'value'}, @twitter.get(:StatusRelationships, k, 'dummy_supercolumn'))
+    new_subcolumns = {@uuids[3] => 'v3', @uuids[4] => 'v4'}
+    subcolumn_to_delete = initial_subcolumns.keys.first # the first column of the initial set
 
     @twitter.batch do
       @twitter.insert(:Users, k + '2', {'body' => 'v2', 'user' => 'v2'})
@@ -361,6 +384,15 @@ class CassandraTest < Test::Unit::TestCase
       @twitter.remove(:Users, k + '4')
       @twitter.insert(:Users, k + '4', {'body' => 'v4', 'user' => 'v4'})
       assert_equal({}, @twitter.get(:Users, k + '4')) # Not yet written
+
+      # SuperColumns
+      # Add and delete new sub columns to the user timeline supercolumn
+      @twitter.insert(:StatusRelationships, k, {'user_timelines' => new_subcolumns }) 
+      @twitter.remove(:StatusRelationships, k, 'user_timelines' , subcolumn_to_delete ) # Delete the first of the initial_subcolumns from the user_timeline supercolumn
+      assert_equal(initial_subcolumns, @twitter.get(:StatusRelationships, k, 'user_timelines')) # No additions or deletes reflected yet
+      # Delete a complete supercolumn 
+      @twitter.remove(:StatusRelationships, k, 'dummy_supercolumn' ) # Delete the full dummy supercolumn
+      assert_equal({@uuids[5] => 'value'}, @twitter.get(:StatusRelationships, k, 'dummy_supercolumn')) # dummy supercolumn not yet deleted 
     end
 
     assert_equal({'body' => 'v2', 'user' => 'v2'}, @twitter.get(:Users, k + '2')) # Written
@@ -373,6 +405,12 @@ class CassandraTest < Test::Unit::TestCase
     assert_equal({'body' => 'v3', 'user' => 'v3', 'location' => 'v3'}.keys.sort, @twitter.get(:Users, k + '3').timestamps.keys.sort) # Written and compacted
     assert_equal({'body' => 'v4', 'user' => 'v4'}.keys.sort, @twitter.get(:Users, k + '4').timestamps.keys.sort) # Written
     assert_equal({'body' => 'v'}.keys.sort, @twitter.get(:Statuses, k + '3').timestamps.keys.sort) # Written
+
+    # Final result: initial_subcolumns - initial_subcolumns.first + new_subcolumns
+    resulting_subcolumns = initial_subcolumns.merge(new_subcolumns).reject{|k,v| k == subcolumn_to_delete }
+    assert_equal(resulting_subcolumns, @twitter.get(:StatusRelationships, key, 'user_timelines'))
+    assert_equal({}, @twitter.get(:StatusRelationships, key, 'dummy_supercolumn')) # dummy supercolumn deleted 
+
   end
 
   def test_complain_about_nil_key
@@ -381,13 +419,6 @@ class CassandraTest < Test::Unit::TestCase
     end
   end
 
-  def test_raise_access_error_on_nonexistent_keyspace
-    nonexistent = Cassandra.new('Nonexistent')
-    assert_raises(Cassandra::AccessError) do
-      nonexistent.get "foo", "bar"
-    end
-  end
-
   def test_nil_sub_column_value
     @twitter.insert(:Index, 'asdf', {"thing" => {'jkl' => ''} })
   end
@@ -397,6 +428,13 @@ class CassandraTest < Test::Unit::TestCase
     assert_nil @twitter.instance_variable_get(:@client)
   end
 
+  def test_disconnect_when_not_connected!
+    assert_nothing_raised do
+      @twitter = Cassandra.new('Twitter', "127.0.0.1:9160", :retries => 2, :exception_classes => [])
+      @twitter.disconnect!
+    end
+  end
+
   def test_super_allows_for_non_string_values_while_normal_does_not
     columns = {'user_timelines' => {@uuids[4] => '4', @uuids[5] => '5'}}
 
@@ -404,6 +442,34 @@ class CassandraTest < Test::Unit::TestCase
     @twitter.insert(:Statuses, key, { 'body' => '1' })
   end
 
+  if cassandra07?
+    def test_creating_and_dropping_new_index
+      @twitter.create_index('Twitter', 'Statuses', 'column_name', 'LongType')
+      assert_nil @twitter.create_index('Twitter', 'Statuses', 'column_name', 'LongType')
+
+      @twitter.drop_index('Twitter', 'Statuses', 'column_name')
+      assert_nil @twitter.drop_index('Twitter', 'Statuses', 'column_name')
+
+      # Recreating and redropping the same index should not error either.
+      @twitter.create_index('Twitter', 'Statuses', 'column_name', 'LongType')
+      @twitter.drop_index('Twitter', 'Statuses', 'column_name')
+    end
+
+    def test_get_indexed_slices
+      @twitter.create_index('Twitter', 'Statuses', 'x', 'LongType')
+
+      @twitter.insert(:Statuses, 'row1', { 'x' => [0,10].pack("NN")  })
+      @twitter.insert(:Statuses, 'row2', { 'x' => [0,20].pack("NN")  })
+
+      idx_expr   = @twitter.create_idx_expr('x', [0,20].pack("NN"), "==")
+      idx_clause = @twitter.create_idx_clause([idx_expr])
+
+      indexed_row = @twitter.get_indexed_slices(:Statuses, idx_clause)
+      assert_equal(1,      indexed_row.length)
+      assert_equal('row2', indexed_row.first.key)
+    end
+  end
+
   private
 
   def key

data/test/ordered_hash_test.rb

@@ -321,6 +321,7 @@ class OrderedHashTest < Test::Unit::TestCase
     @ordered_hash.reject! { |k, _| k == 'pink' }
     assert_equal copy, @ordered_hash
     assert !@ordered_hash.keys.include?('pink')
+    assert !@ordered_hash.timestamps.keys.include?('pink')
   end
 
   def test_reject

data/vendor/0.6/gen-rb/cassandra.rb

@@ -6,6 +6,7 @@
 
 require 'thrift'
 require 'cassandra_types'
+require 'cassandra_constants'
 
     module CassandraThrift
       module Cassandra

data/vendor/0.7/gen-rb/cassandra.rb

@@ -6,6 +6,7 @@
 
 require 'thrift'
 require 'cassandra_types'
+require 'cassandra_constants'
 
     module CassandraThrift
       module Cassandra

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: cassandra
 version: !ruby/object:Gem::Version 
-  hash: 59
+  hash: 57
   prerelease: false
   segments: 
   - 0
   - 9
-  - 0
-  version: 0.9.0
+  - 1
+  version: 0.9.1
 platform: ruby
 authors: 
 - Evan Weaver, Ryan King
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-12-08 00:00:00 -08:00
+date: 2011-04-06 00:00:00 -07:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -159,7 +159,7 @@ files:
 - vendor/0.7/gen-rb/cassandra_types.rb
 - cassandra.gemspec
 has_rdoc: true
-homepage: http://blog.evanweaver.com/files/doc/fauna/cassandra/
+homepage: ""
 licenses: []
 
 post_install_message: