lib/rhino/interface/base.rb lib/rhino/interface/hbase-thrift/Hbase.rb lib/rhino/interface/hbase-thrift/base.rb lib/rhino/interface/hbase-thrift/gen/Hbase.rb lib/rhino/interface/hbase-thrift/gen/Hbase_constants.rb lib/rhino/interface/hbase-thrift/gen/Hbase_types.rb lib/rhino/interface/hbase-thrift/scanner.rb lib/rhino/interface/hbase-thrift/table.rb lib/rhino/interface/hbase/hbase.rb lib/rhino/interface/hbase/htable.rb lib/rhino/interface/hbase/scanner.rb lib/rhino/interface/table.rb @@ -9,16 +9,16 @@ require 'active_support/core_ext/class/attribute_accessors.rb' class Array; include ActiveSupport::CoreExtensions::Array::ExtractOptions; end -require 'rhino/interface/hbase' -require 'rhino/interface/htable' +require 'rhino/interface/base' +require 'rhino/interface/table' require 'rhino/interface/scanner' require 'thrift/transport/tsocket.rb' require 'thrift/protocol/tbinaryprotocol.rb' -require 'rhino/thrift/gen/Hbase' -require 'rhino/thrift/hbase' -require 'rhino/thrift/htable' -require 'rhino/thrift/scanner' +require 'rhino/interface/hbase-thrift/gen/Hbase' +require 'rhino/interface/hbase-thrift/base' +require 'rhino/interface/hbase-thrift/table' +require 'rhino/interface/hbase-thrift/scanner' require 'rhino/debug' lib/rhino.rb @@ -46,7 +46,7 @@ module Rhino # Writes this cell's data to the row and saves only this cell. def save(timestamp=nil) write - row.class.htable.put(row.key, {attr_name=>contents}, timestamp) + row.class.table.put(row.key, {attr_name=>contents}, timestamp) end # TODO: update to destroy the cell without re-saving the row lib/rhino/cell.rb @@ -92,7 +92,7 @@ module Rhino # so this accomplishes both data.delete('timestamp') - self.class.htable.put(key, data, with_timestamp) + self.class.table.put(key, data, with_timestamp) if new_record? @opts[:new_record] = false @opts[:was_new_record] = true @@ -102,7 +102,7 @@ module Rhino def destroy debug("Model#destroy() [key=#{key.inspect}]") - self.class.htable.delete_row(key) + self.class.table.delete_row(key) end def data @@ -208,33 +208,36 @@ module Rhino # CLASS METHODS # ################# - # Specifies the endpoint URL for the HBase REST API. This URL should end in a slash (e.g., "http://localhost:60010/api"). - # The connection is not actually "established", however, until +connection+ is called. - def Model.connect(host, port) - debug("Model.connect(#{host.inspect}, #{port.inspect})") - raise "already connected" if hbase - @hbase = Rhino::ThriftInterface::HBase.new(host, port) + def Model.connect(host, port, adapter=Rhino::HBaseThriftInterface) + debug("Model.connect(#{host.inspect}, #{port.inspect}, #{adapter.inspect})") + raise "already connected" if connection + @adapter = adapter + @conn = adapter::Base.new(host, port) + end + + def Model.adapter + @adapter end # Returns true if connected to the database, and false otherwise. def Model.connected? - @hbase != nil + @conn != nil end - # Returns the connection to HBase. The connection to HBase is shared across all models and is stored in Rhino::Model, + # Returns the connection. The connection is shared across all models and is stored in Rhino::Model, # so models retrieve it from Rhino::Table. - def Model.hbase + def Model.connection # uses self.name instead of self.class because in class methods, self.class==Object and self.name=="Rhino::Model" if self.name == "Rhino::Model" - @hbase + @conn else - Rhino::Model.hbase + Rhino::Model.connection end end - # Returns the HTable interface. - def Model.htable - @htable ||= Rhino::ThriftInterface::HTable.new(hbase, table_name) + # Returns the table interface. + def Model.table + @table ||= Rhino::Model.adapter::Table.new(connection, table_name) end cattr_accessor :column_families @@ -324,16 +327,16 @@ module Rhino # get the row begin - data = htable.get(key, :timestamp=>timestamp) + data = table.get(key, :timestamp=>timestamp) debug("-> found [key=#{key.inspect}, data=#{data.inspect}]") load(key, data) - rescue Rhino::Interface::HTable::RowNotFound + rescue Rhino::Interface::Table::RowNotFound return nil end end def Model.delete_all - htable.delete_all_rows + table.delete_all_rows end end end \ No newline at end of file lib/rhino/model.rb @@ -6,11 +6,11 @@ module Rhino def initialize(model, opts) @model = model @opts = opts - @hscanner = Rhino::ThriftInterface::Scanner.new(@model.htable, self.opts) + @scanner = Rhino::Model.adapter::Scanner.new(@model.table, self.opts) end def each - while row_data = @hscanner.next_row() + while row_data = @scanner.next_row() row_key = row_data.delete('key') row = @model.load(row_key, row_data) yield(row) lib/rhino/scanner.rb @@ -1,12 +1,12 @@ require 'spec/spec_helper.rb' -describe Rhino::Interface::HBase do +describe Rhino::Interface::Base do describe "when working with tables" do before do # TODO: this is a hardcoded list of tables in my db - should get real fixtures!! @table_names = %w(users pages) # TODO: there should be a Rhino-namespace way of getting the hbase object, not just through the model - @hbase = Rhino::Model.hbase + @hbase = Rhino::Model.connection end it "should return a list of tables" do spec/interface/hbase_spec.rb @@ -1,42 +1,42 @@ require 'spec/spec_helper.rb' # TODO these tests should not be dependent on Rhino::Model (Page is a subclass of Model) -describe Rhino::Interface::HTable do +describe Rhino::Interface::Table do before do - @page_htable = Page.htable + @page_table = Page.table end after do - @page_htable.delete_all_rows + @page_table.delete_all_rows end describe "when getting rows" do it "should raise RowNotFound for nonexistent rows" do - lambda { @page_htable.get('this row key does not exist') }.should raise_error(Rhino::Interface::HTable::RowNotFound) + lambda { @page_table.get('this row key does not exist') }.should raise_error(Rhino::Interface::Table::RowNotFound) end it "should raise ArgumentError if nil or blank key is given" do - lambda { @page_htable.get('') }.should raise_error(ArgumentError) - lambda { @page_htable.get(nil) }.should raise_error(ArgumentError) + lambda { @page_table.get('') }.should raise_error(ArgumentError) + lambda { @page_table.get(nil) }.should raise_error(ArgumentError) end it "should get the latest timestamp" do ts = (Time.now.to_f * 1000).to_i - @page_htable.put('abc', {'title:'=>'hello1', 'contents:'=>'hello there'}, ts) - @page_htable.put('abc', {'title:'=>'hello2'}, ts+1000) - @page_htable.put('abc', {'title:'=>'hello3'}, ts+5000) - @page_htable.get('abc')['timestamp'].should == ts+5000 + @page_table.put('abc', {'title:'=>'hello1', 'contents:'=>'hello there'}, ts) + @page_table.put('abc', {'title:'=>'hello2'}, ts+1000) + @page_table.put('abc', {'title:'=>'hello3'}, ts+5000) + @page_table.get('abc')['timestamp'].should == ts+5000 end it "should get the timestamp" do - @page_htable.put('a99', {'title:'=>'hello2'}) - @page_htable.get('a99')['timestamp'].should be_close((Time.now.to_f * 1000).to_i, 100) + @page_table.put('a99', {'title:'=>'hello2'}) + @page_table.get('a99')['timestamp'].should be_close((Time.now.to_f * 1000).to_i, 100) end it "should retrieve the row" do key = 'hello.com' - @page_htable.put(key, {'title:'=>'howdy'}) - row = @page_htable.get(key) + @page_table.put(key, {'title:'=>'howdy'}) + row = @page_table.get(key) row.keys.sort.should == %w(timestamp title:) row['title:'].should == 'howdy' end @@ -44,46 +44,46 @@ describe Rhino::Interface::HTable do describe "when deleting all rows" do before do - @page_htable.put('a', {'title:'=>'abc'}) - @page_htable.put('b', {'title:'=>'bcd'}) - @page_htable.put('c', {'title:'=>'cde'}) + @page_table.put('a', {'title:'=>'abc'}) + @page_table.put('b', {'title:'=>'bcd'}) + @page_table.put('c', {'title:'=>'cde'}) end it "should remove all rows" do - @page_htable.get('a').should_not == nil - @page_htable.delete_all_rows - lambda { @page_htable.get('a') }.should raise_error(Rhino::Interface::HTable::RowNotFound) - @page_htable.scan.collect.should == [] + @page_table.get('a').should_not == nil + @page_table.delete_all_rows + lambda { @page_table.get('a') }.should raise_error(Rhino::Interface::Table::RowNotFound) + @page_table.scan.collect.should == [] end end describe "when deleting entire rows" do before do @some_key = 'some-key' - @page_htable.put(@some_key, {'title:'=>'abc'}) + @page_table.put(@some_key, {'title:'=>'abc'}) end it "should delete the row" do - @page_htable.get(@some_key).should_not == nil - @page_htable.delete_row(@some_key) - lambda { @page_htable.get(@some_key) }.should raise_error(Rhino::Interface::HTable::RowNotFound) + @page_table.get(@some_key).should_not == nil + @page_table.delete_row(@some_key) + lambda { @page_table.get(@some_key) }.should raise_error(Rhino::Interface::Table::RowNotFound) end end describe "when putting rows" do it "should require that column values be strings" do - lambda { @page_htable.put('a', {'title:'=>Object}) }.should raise_error(ArgumentError) + lambda { @page_table.put('a', {'title:'=>Object}) }.should raise_error(ArgumentError) end end describe "when putting existing rows" do it "should delete cells that previously existed if their value is changed to nil" do key = 'example.com' - @page_htable.put(key, {'title:'=>'howdy', 'links:com.google'=>'Google'}) - @page_htable.get(key).keys.include?('links:com.google').should == true + @page_table.put(key, {'title:'=>'howdy', 'links:com.google'=>'Google'}) + @page_table.get(key).keys.include?('links:com.google').should == true # the cell has been deleted - @page_htable.put(key, {'title:'=>'howdy', 'links:com.google'=>nil}) - @page_htable.get(key).keys.include?('links:com.google').should == false + @page_table.put(key, {'title:'=>'howdy', 'links:com.google'=>nil}) + @page_table.get(key).keys.include?('links:com.google').should == false end end @@ -94,10 +94,10 @@ describe Rhino::Interface::HTable do it "should update the values" do key = 'hi.example.com' - @page_htable.put(key, {'title:'=>'howdy'}) - @page_htable.get(key)['title:'].should == 'howdy' - @page_htable.put(key, {'title:'=>'goodbye'}) - @page_htable.get(key)['title:'].should == 'goodbye' + @page_table.put(key, {'title:'=>'howdy'}) + @page_table.get(key)['title:'].should == 'howdy' + @page_table.put(key, {'title:'=>'goodbye'}) + @page_table.get(key)['title:'].should == 'goodbye' end end end \ No newline at end of file spec/interface/htable_spec.rb @@ -2,20 +2,20 @@ require 'spec/spec_helper.rb' describe Rhino::Interface::Scanner do before do - @page_htable = Page.htable - @page_htable.put('com.apple', {'title:'=>'apple'}) - @page_htable.put('com.google', {'title:'=>'google'}) - @page_htable.put('com.microsoft', {'title:'=>'microsoft'}) - @page_htable.put('com.yahoo', {'title:'=>'yahoo'}) + @page_table = Page.table + @page_table.put('com.apple', {'title:'=>'apple'}) + @page_table.put('com.google', {'title:'=>'google'}) + @page_table.put('com.microsoft', {'title:'=>'microsoft'}) + @page_table.put('com.yahoo', {'title:'=>'yahoo'}) end after do - @page_htable.delete_all_rows + @page_table.delete_all_rows end describe "scanning all rows" do it "should return all rows" do - rows = @page_htable.scan.collect + rows = @page_table.scan.collect rows.collect { |row| row['title:'] }.should == %w(apple google microsoft yahoo) rows.collect { |row| row['key'] }.should == %w(com.apple com.google com.microsoft com.yahoo) end @@ -23,36 +23,36 @@ describe Rhino::Interface::Scanner do describe "when scanning with only a start row specified" do it "should return all rows including and after the start row" do - rows = @page_htable.scan(:start_row=>'com.google') + rows = @page_table.scan(:start_row=>'com.google') rows.collect { |row| row['key'] }.should == %w(com.google com.microsoft com.yahoo) end end describe "when scanning with a start row and a stop row specified" do it "should return all rows between the start row (inclusive) and stop row (exclusive)" do - rows = @page_htable.scan(:start_row=>'com.google', :stop_row=>'com.yahoo') + rows = @page_table.scan(:start_row=>'com.google', :stop_row=>'com.yahoo') rows.collect { |row| row['key'] }.should == %w(com.google com.microsoft) end end describe "when scanning with only a stop row specified" do it "should return all rows up to but not including the stop row" do - rows = @page_htable.scan(:stop_row=>'com.microsoft') + rows = @page_table.scan(:stop_row=>'com.microsoft') rows.collect { |row| row['key'] }.should == %w(com.apple com.google) end end describe "when no rows in the table exist" do before do - @page_htable.delete_all_rows + @page_table.delete_all_rows end it "should not raise an error" do - lambda { @page_htable.scan }.should_not raise_error + lambda { @page_table.scan }.should_not raise_error end it "should return an empty array" do - @page_htable.scan.collect.should == [] + @page_table.scan.collect.should == [] end end @@ -60,12 +60,12 @@ describe Rhino::Interface::Scanner do describe "when many rows exist" do before do (1..45000).each do |n| - @page_htable.put("item#{n}", {'title:'=>"title#{n}"}) + @page_table.put("item#{n}", {'title:'=>"title#{n}"}) end end it "should scan all rows without an error" do - lambda { @page_htable.scan.each do |row| + lambda { @page_table.scan.each do |row| puts row.title end }.should_not raise_error end spec/interface/scanner_spec.rb lib/rhino/interface/hbase.rb lib/rhino/interface/htable.rb lib/rhino/thrift/Hbase.rb lib/rhino/thrift/gen/Hbase.rb lib/rhino/thrift/gen/Hbase_constants.rb lib/rhino/thrift/gen/Hbase_types.rb lib/rhino/thrift/hbase.rb lib/rhino/thrift/htable.rb lib/rhino/thrift/scanner.rb 3602835f373e43cf557398a0bde323d7bef005d5 Quinn Slack qslack@qslack.com http://github.com/sqs/rhino/commit/3b5b60c8ee15242fbd07be32c3366bc10d6a51dd 3b5b60c8ee15242fbd07be32c3366bc10d6a51dd 2008-11-09T22:30:00-08:00 2008-11-09T22:30:00-08:00 genericized adapter to get ready for possible cassandra support 0b931ca496ce9a81543999451995c500704f22de Quinn Slack qslack@qslack.com