fluent-plugin-webhdfs 0.7.1 → 1.0.0

data/lib/fluent/plugin/webhdfs_compressor_bzip2.rb

@@ -1,14 +1,10 @@
-module Fluent
-  class WebHDFSOutput < Fluent::TimeSlicedOutput
+module Fluent::Plugin
+  class WebHDFSOutput < Output
     class Bzip2Compressor < Compressor
       WebHDFSOutput.register_compressor('bzip2', self)
 
       def initialize(options = {})
-        begin
-          require "bzip2/ffi"
-        rescue LoadError
-          raise Fluent::ConfigError, "Install bzip2-ffi before use bzip2 compressor"
-        end
+        require "bzip2/ffi"
       end
 
       def ext

data/lib/fluent/plugin/webhdfs_compressor_gzip.rb

@@ -1,5 +1,5 @@
-module Fluent
-  class WebHDFSOutput < Fluent::TimeSlicedOutput
+module Fluent::Plugin
+  class WebHDFSOutput < Output
     class GzipCompressor < Compressor
       WebHDFSOutput.register_compressor('gzip', self)
 
@@ -14,7 +14,7 @@ module Fluent
       def compress(chunk, tmp)
         w = Zlib::GzipWriter.new(tmp)
         chunk.write_to(w)
-        w.finish
+        w.close
       end
     end
   end

data/lib/fluent/plugin/webhdfs_compressor_lzo_command.rb

@@ -1,9 +1,9 @@
-module Fluent
-  class WebHDFSOutput
+module Fluent::Plugin
+  class WebHDFSOutput < Output
     class LZOCommandCompressor < Compressor
       WebHDFSOutput.register_compressor('lzo_command', self)
 
-      config_param :command_parameter, :string, :default => '-qf1'
+      config_param :command_parameter, :string, default: '-qf1'
 
       def configure(conf)
         super

data/lib/fluent/plugin/webhdfs_compressor_snappy.rb

@@ -1,5 +1,5 @@
-module Fluent
-  class WebHDFSOutput < Fluent::TimeSlicedOutput
+module Fluent::Plugin
+  class WebHDFSOutput < Output
     class SnappyCompressor < Compressor
       WebHDFSOutput.register_compressor('snappy', self)
 

data/lib/fluent/plugin/webhdfs_compressor_text.rb

@@ -1,5 +1,5 @@
-module Fluent
-  class WebHDFSOutput < Fluent::TimeSlicedOutput
+module Fluent::Plugin
+  class WebHDFSOutput < Output
     class TextCompressor < Compressor
       WebHDFSOutput.register_compressor('text', self)
 

data/test/helper.rb

@@ -8,10 +8,13 @@ rescue Bundler::BundlerError => e
   exit e.status_code
 end
 require 'test/unit'
+require 'test/unit/rr'
 
 $LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', 'lib'))
 $LOAD_PATH.unshift(File.dirname(__FILE__))
 require 'fluent/test'
+require 'fluent/test/helpers'
+require 'fluent/test/driver/output'
 unless ENV.has_key?('VERBOSE')
   nulllogger = Object.new
   nulllogger.instance_eval {|obj|
@@ -22,6 +25,8 @@ unless ENV.has_key?('VERBOSE')
   $log = nulllogger
 end
 
+include Fluent::Test::Helpers
+
 require 'fluent/plugin/out_webhdfs'
 
 class Test::Unit::TestCase

data/test/plugin/test_compressor.rb

@@ -16,11 +16,11 @@ class CompressorTest < Test::Unit::TestCase
     def setup
       omit unless Object.const_defined?(:Snappy)
       Fluent::Test.setup
-      @compressor = Fluent::WebHDFSOutput::SnappyCompressor.new
+      @compressor = Fluent::Plugin::WebHDFSOutput::SnappyCompressor.new
     end
 
-    def create_driver(conf=CONFIG,tag='test')
-      Fluent::Test::OutputTestDriver.new(Fluent::WebHDFSOutput, tag).configure(conf)
+    def create_driver(conf = CONFIG)
+      Fluent::Test::Driver::Output.new(Fluent::Plugin::WebHDFSOutput).configure(conf)
     end
 
     def test_ext

data/test/plugin/test_out_webhdfs.rb

@@ -1,169 +1,243 @@
 require 'helper'
 
 class WebHDFSOutputTest < Test::Unit::TestCase
-  CONFIG = %[
-host namenode.local
-path /hdfs/path/file.%Y%m%d.log
-  ]
+  CONFIG_DEFAULT = config_element("match", "", {"host" => "namenode.local", "path" => "/hdfs/path/file.%Y%m%d.log"})
+
+  CONFIG_COMPAT = config_element(
+    "ROOT", "", {
+      "output_data_type" => "",
+      "host" => "namenode.local",
+      "path" => "/hdfs/path/file.%Y%m%d.log"
+    })
+
   def setup
     Fluent::Test.setup
   end
 
-  def create_driver(conf=CONFIG,tag='test')
-    Fluent::Test::OutputTestDriver.new(Fluent::WebHDFSOutput, tag).configure(conf)
+  def create_driver(conf)
+    Fluent::Test::Driver::Output.new(Fluent::Plugin::WebHDFSOutput).configure(conf)
   end
 
-  class ConfigureTest < self
-    def test_default
-      d = create_driver
+  sub_test_case "default configuration" do
+    test 'configured with standard out_file format with specified hdfs info' do
+      d = create_driver(CONFIG_DEFAULT)
+      assert_true d.instance.instance_eval{ @using_formatter_config }
+
       assert_equal 'namenode.local', d.instance.instance_eval{ @namenode_host }
       assert_equal 50070, d.instance.instance_eval{ @namenode_port }
       assert_equal '/hdfs/path/file.%Y%m%d.log', d.instance.path
-      assert_equal '%Y%m%d', d.instance.time_slice_format
       assert_equal false, d.instance.httpfs
       assert_nil d.instance.username
       assert_equal false, d.instance.ignore_start_check_error
 
-      assert_equal true, d.instance.output_include_time
-      assert_equal true, d.instance.output_include_tag
-      assert_equal 'json', d.instance.output_data_type
+      assert_equal 'Fluent::Plugin::OutFileFormatter', d.instance.formatter.class.to_s
+      assert_equal true, d.instance.end_with_newline
+
+      # deprecated params
+      assert_nil d.instance.instance_eval{ @output_include_time }
+      assert_nil d.instance.instance_eval{ @output_include_tag }
       assert_nil d.instance.remove_prefix
-      assert_equal 'TAB', d.instance.field_separator
-      assert_equal true, d.instance.add_newline
-      assert_equal 'tag_missing', d.instance.default_tag
+      assert_nil d.instance.instance_eval{ @header_separator }
+      assert_nil d.instance.default_tag
+    end
+  end
+
+  sub_test_case "flat configuration" do
+    def test_default_for_traditional_config
+      d = create_driver(CONFIG_COMPAT)
+      assert_false d.instance.instance_eval{ @using_formatter_config }
+
+      assert_equal 'namenode.local', d.instance.instance_eval{ @namenode_host }
+      assert_equal 50070, d.instance.instance_eval{ @namenode_port }
+      assert_equal '/hdfs/path/file.%Y%m%d.log', d.instance.path
+      assert_equal false, d.instance.httpfs
+      assert_nil d.instance.username
+      assert_equal false, d.instance.ignore_start_check_error
+
+      assert_equal 'Fluent::Plugin::JSONFormatter', d.instance.formatter.class.to_s
+      assert_equal true, d.instance.end_with_newline
+
+      assert_equal true, d.instance.instance_eval{ @output_include_time }
+      assert_equal true, d.instance.instance_eval{ @output_include_tag }
+      assert_nil d.instance.instance_eval{ @remove_prefix }
+      assert_equal "\t", d.instance.instance_eval{ @header_separator }
+      assert_equal 'tag_missing', d.instance.instance_eval{ @default_tag }
     end
 
     def test_httpfs
-      d = create_driver %[
-namenode server.local:14000
-path /hdfs/path/file.%Y%m%d.%H%M.log
-httpfs yes
-username hdfs_user
-]
+      conf = config_element(
+        "ROOT", "", {
+          "namenode" => "server.local:14000",
+          "path" => "/hdfs/path/file.%Y%m%d.%H%M.log",
+          "httpfs" => "yes",
+          "username" => "hdfs_user"
+        })
+      d = create_driver(conf)
+
       assert_equal 'server.local', d.instance.instance_eval{ @namenode_host }
       assert_equal 14000, d.instance.instance_eval{ @namenode_port }
       assert_equal '/hdfs/path/file.%Y%m%d.%H%M.log', d.instance.path
-      assert_equal '%Y%m%d%H%M', d.instance.time_slice_format
       assert_equal true, d.instance.httpfs
       assert_equal 'hdfs_user', d.instance.username
     end
 
     def test_ssl
-      d = create_driver %[
-namenode server.local:14000
-path /hdfs/path/file.%Y%m%d.%H%M.log
-ssl true
-ssl_ca_file /path/to/ca_file.pem
-ssl_verify_mode peer
-kerberos true
-kerberos_keytab /path/to/kerberos.keytab
-]
+      conf = config_element(
+        "ROOT", "", {
+          "namenode" => "server.local:14000",
+          "path" => "/hdfs/path/file.%Y%m%d.%H%M.log",
+          "ssl" => true,
+          "ssl_ca_file" => "/path/to/ca_file.pem",
+          "ssl_verify_mode" => "peer",
+          "kerberos" => true
+        })
+      d = create_driver(conf)
 
       assert_equal 'server.local', d.instance.instance_eval{ @namenode_host }
       assert_equal 14000, d.instance.instance_eval{ @namenode_port }
       assert_equal '/hdfs/path/file.%Y%m%d.%H%M.log', d.instance.path
-      assert_equal '%Y%m%d%H%M', d.instance.time_slice_format
       assert_equal true, d.instance.ssl
       assert_equal '/path/to/ca_file.pem', d.instance.ssl_ca_file
       assert_equal :peer, d.instance.ssl_verify_mode
       assert_equal true, d.instance.kerberos
-      assert_equal '/path/to/kerberos.keytab', d.instance.kerberos_keytab
     end
 
-    data(gzip: ['gzip', Fluent::WebHDFSOutput::GzipCompressor],
-         bzip2: ['bzip2', Fluent::WebHDFSOutput::Bzip2Compressor],
-         snappy: ['snappy', Fluent::WebHDFSOutput::SnappyCompressor],
-         lzo: ['lzo_command', Fluent::WebHDFSOutput::LZOCommandCompressor])
+    data(gzip: [:gzip, Fluent::Plugin::WebHDFSOutput::GzipCompressor],
+         bzip2: [:bzip2, Fluent::Plugin::WebHDFSOutput::Bzip2Compressor],
+         snappy: [:snappy, Fluent::Plugin::WebHDFSOutput::SnappyCompressor],
+         lzo: [:lzo_command, Fluent::Plugin::WebHDFSOutput::LZOCommandCompressor])
     def test_compress(data)
       compress_type, compressor_class = data
       begin
-        d = create_driver %[
-namenode server.local:14000
-path /hdfs/path/file.%Y%m%d.%H%M.log
-compress #{compress_type}
-        ]
+        conf = config_element(
+          "ROOT", "", {
+            "namenode" => "server.local:14000",
+            "path" => "/hdfs/path/file.%Y%m%d.%H%M.log",
+            "compress" => compress_type
+          })
+        d = create_driver(conf)
       rescue Fluent::ConfigError => ex
         omit ex.message
       end
       assert_equal 'server.local', d.instance.instance_eval{ @namenode_host }
       assert_equal 14000, d.instance.instance_eval{ @namenode_port }
       assert_equal '/hdfs/path/file.%Y%m%d.%H%M.log', d.instance.path
-      assert_equal '%Y%m%d%H%M', d.instance.time_slice_format
       assert_equal compress_type, d.instance.compress
       assert_equal compressor_class, d.instance.compressor.class
     end
 
-    def test_placeholders
-      d = create_driver %[
-hostname testing.node.local
-namenode server.local:50070
-path /hdfs/${hostname}/file.%Y%m%d%H.log
-]
+    def test_placeholders_old_style
+      conf = config_element(
+        "ROOT", "", {
+          "hostname" => "testing.node.local",
+          "namenode" => "server.local:50070",
+          "path" => "/hdfs/${hostname}/file.%Y%m%d%H.log"
+        })
+      d = create_driver(conf)
       assert_equal '/hdfs/testing.node.local/file.%Y%m%d%H.log', d.instance.path
     end
 
-    class PathFormatTest < self
-      def test_default
-        d = create_driver
-        assert_equal '/hdfs/path/file.%Y%m%d.log', d.instance.path
-        assert_equal '%Y%m%d', d.instance.time_slice_format
-        assert_equal '/hdfs/path/file.20120718.log', d.instance.path_format('20120718')
-      end
+    data("%Y%m%d" => ["/hdfs/path/file.%Y%m%d.log", "/hdfs/path/file.20120718.log"],
+         "%Y%m%d.%H%M" => ["/hdfs/path/file.%Y%m%d.%H%M.log", "/hdfs/path/file.20120718.1503.log"])
+    test "generate_path" do |(path, expected)|
+      conf = config_element(
+        "ROOT", "", {
+          "namenode" => "server.local:14000",
+          "path" => path
+        })
+      d = create_driver(conf)
+      formatter = Fluent::Timezone.formatter("+0900", path)
+      mock(Fluent::Timezone).formatter(Time.now.strftime("%z"), path) { formatter }
+      time = event_time("2012-07-18 15:03:00 +0900")
+      metadata = d.instance.metadata("test", time, {})
+      chunk = d.instance.buffer.generate_chunk(metadata)
+      assert_equal expected, d.instance.generate_path(chunk)
+    end
 
-      def test_time_slice_format
-        d = create_driver %[
-namenode server.local:14000
-path /hdfs/path/file.%Y%m%d.%H%M.log
-]
-        assert_equal '/hdfs/path/file.%Y%m%d.%H%M.log', d.instance.path
-        assert_equal '%Y%m%d%H%M', d.instance.time_slice_format
-        assert_equal '/hdfs/path/file.20120718.1503.log', d.instance.path_format('201207181503')
+    data(path: { "append" => false },
+         ssl: { "ssl" => true, "ssl_verify_mode" => "invalid" },
+         compress: { "compress" => "invalid" })
+    test "invalid" do |attr|
+      conf = config_element(
+        "ROOT", "", {
+          "namenode" => "server.local:14000",
+          "path" => "/hdfs/path/file.%Y%m%d.%H%M.log"
+        })
+      conf += config_element("", "", attr)
+      assert_raise Fluent::ConfigError do
+        create_driver(conf)
       end
     end
+  end
 
-    class InvalidTest < self
-      def test_path
-        assert_raise Fluent::ConfigError do
-          d = create_driver %[
-            namenode server.local:14000
-            path /hdfs/path/file.%Y%m%d.%H%M.log
-            append false
-          ]
-        end
-      end
+  sub_test_case "sub section configuration" do
+    def test_time_key
+      conf = config_element(
+        "ROOT", "", {
+          "host" => "namenode.local",
+          "path" => "/hdfs/path/file.%Y%m%d.log"
+        }, [
+          config_element(
+            "buffer", "time", {
+              "timekey" => 1
+            })
+        ]
+      )
+      d = create_driver(conf)
+      time = event_time("2012-07-18 15:03:00 +0900")
+      metadata = d.instance.metadata("test", time, {})
+      chunk = d.instance.buffer.generate_chunk(metadata)
+      assert_equal 1, d.instance.buffer_config.timekey
+      assert_equal "/hdfs/path/file.20120718.log", d.instance.generate_path(chunk)
+    end
+  end
 
-      def test_ssl
-        assert_raise Fluent::ConfigError do
-          create_driver %[
-            namenode server.local:14000
-            path /hdfs/path/file.%Y%m%d.%H%M.log
-            ssl true
-            ssl_verify_mode invalid
-          ]
-        end
-      end
+  sub_test_case "using format subsection" do
+    test "blank format means default format 'out_file' with UTC timezone" do
+      format_section = config_element("format", "", {}, [])
+      conf = config_element("match", "", {"host" => "namenode.local", "path" => "/hdfs/path/file.%Y%m%d%H.log"}, [format_section])
+      d = create_driver(conf)
+      time = event_time("2017-01-24 13:10:30 -0700")
+      line = d.instance.format("test.now", time, {"message" => "yay", "name" => "tagomoris"})
+      assert_equal "2017-01-24T20:10:30Z\ttest.now\t{\"message\":\"yay\",\"name\":\"tagomoris\"}\n", line
+    end
 
-      def test_invalid_compress
-        assert_raise Fluent::ConfigError do
-          create_driver %[
-            namenode server.local:14000
-            path /hdfs/path/file.%Y%m%d.%H%M.log
-            compress invalid
-          ]
-        end
-      end
+    test "specifying timezone works well in format section" do
+      format_section = config_element("format", "", {"timezone" => "+0100"}, [])
+      conf = config_element("match", "", {"host" => "namenode.local", "path" => "/hdfs/path/file.%Y%m%d%H.log"}, [format_section])
+      d = create_driver(conf)
+      time = event_time("2017-01-24 13:10:30 -0700")
+      line = d.instance.format("test.now", time, {"message" => "yay", "name" => "tagomoris"})
+      assert_equal "2017-01-24T21:10:30+01:00\ttest.now\t{\"message\":\"yay\",\"name\":\"tagomoris\"}\n", line
+    end
 
-      data("%{uuid:hostname}" => "%{uuid:hostname}",
-           "%{uuid:timestamp}" => "%{uuid:timestamp}")
-      def test_obsoleted_placeholders(placeholder)
-        assert_raise Fluent::ConfigError do
-          create_driver %[
-            namenode server.local:14000
-            path /hdfs/path/#{placeholder}/file.%Y%m%d.%H%M.log
-            append false
-          ]
-        end
-      end
+    test "specifying formatter type LTSV for records, without tag and timezone" do
+      format_section = config_element("format", "", {"@type" => "ltsv"}, [])
+      conf = config_element("match", "", {"host" => "namenode.local", "path" => "/hdfs/path/file.%Y%m%d%H.log"}, [format_section])
+      d = create_driver(conf)
+      time = event_time("2017-01-24 13:10:30 -0700")
+      line = d.instance.format("test.now", time, {"message" => "yay", "name" => "tagomoris"})
+      assert_equal "message:yay\tname:tagomoris\n", line
+    end
+
+    test "specifying formatter type LTSV for records, with inject section to insert tag and time" do
+      inject_section = config_element("inject", "", {"tag_key" => "tag", "time_key" => "time", "time_type" => "string", "localtime" => "false"})
+      format_section = config_element("format", "", {"@type" => "ltsv"}, [])
+      conf = config_element("match", "", {"host" => "namenode.local", "path" => "/hdfs/path/file.%Y%m%d%H.log"}, [inject_section, format_section])
+      d = create_driver(conf)
+      time = event_time("2017-01-24 13:10:30 -0700")
+      line = d.instance.format("test.now", time, {"message" => "yay", "name" => "tagomoris"})
+      assert_equal "message:yay\tname:tagomoris\ttag:test.now\ttime:2017-01-24T20:10:30Z\n", line
+    end
+  end
+
+  sub_test_case "using older configuration" do
+    test "output_data_type json is same with out_file with UTC timezone" do
+      conf = config_element("match", "", {"host" => "namenode.local", "path" => "/hdfs/path/file.%Y%m%d%H.log", "output_data_type" => "json"}, [])
+      d = create_driver(conf)
+      time = event_time("2017-01-24 13:10:30 -0700")
+      line = d.instance.format("test.now", time, {"message" => "yay", "name" => "tagomoris"})
+      assert_equal "2017-01-24T20:10:30Z\ttest.now\t{\"message\":\"yay\",\"name\":\"tagomoris\"}\n", line
     end
   end
 end