embulk 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +15 -0
- data/.gitignore +13 -0
- data/Gemfile +3 -0
- data/Gemfile.lock +33 -0
- data/README.md +117 -0
- data/Rakefile +58 -0
- data/bin/embulk +63 -0
- data/build.gradle +149 -0
- data/embulk-cli/build.gradle +6 -0
- data/embulk-cli/pom.xml +94 -0
- data/embulk-cli/src/main/java/org/embulk/cli/Main.java +15 -0
- data/embulk-core/build.gradle +6 -0
- data/embulk-core/pom.xml +143 -0
- data/embulk-core/src/main/java/org/embulk/EmbulkService.java +39 -0
- data/embulk-core/src/main/java/org/embulk/command/Runner.java +199 -0
- data/embulk-core/src/main/java/org/embulk/command/TablePrinter.java +119 -0
- data/embulk-core/src/main/java/org/embulk/config/CommitReport.java +26 -0
- data/embulk-core/src/main/java/org/embulk/config/Config.java +15 -0
- data/embulk-core/src/main/java/org/embulk/config/ConfigDefault.java +15 -0
- data/embulk-core/src/main/java/org/embulk/config/ConfigException.java +20 -0
- data/embulk-core/src/main/java/org/embulk/config/ConfigLoader.java +83 -0
- data/embulk-core/src/main/java/org/embulk/config/ConfigSource.java +28 -0
- data/embulk-core/src/main/java/org/embulk/config/DataSource.java +35 -0
- data/embulk-core/src/main/java/org/embulk/config/DataSourceImpl.java +208 -0
- data/embulk-core/src/main/java/org/embulk/config/DataSourceSerDe.java +80 -0
- data/embulk-core/src/main/java/org/embulk/config/GenericTypeReference.java +20 -0
- data/embulk-core/src/main/java/org/embulk/config/ModelManager.java +125 -0
- data/embulk-core/src/main/java/org/embulk/config/NextConfig.java +26 -0
- data/embulk-core/src/main/java/org/embulk/config/Task.java +10 -0
- data/embulk-core/src/main/java/org/embulk/config/TaskInvocationHandler.java +180 -0
- data/embulk-core/src/main/java/org/embulk/config/TaskSerDe.java +343 -0
- data/embulk-core/src/main/java/org/embulk/config/TaskSource.java +28 -0
- data/embulk-core/src/main/java/org/embulk/config/TaskValidationException.java +37 -0
- data/embulk-core/src/main/java/org/embulk/config/TaskValidator.java +24 -0
- data/embulk-core/src/main/java/org/embulk/exec/ExecModule.java +45 -0
- data/embulk-core/src/main/java/org/embulk/exec/ExecuteInterruptedException.java +10 -0
- data/embulk-core/src/main/java/org/embulk/exec/ExecuteResult.java +19 -0
- data/embulk-core/src/main/java/org/embulk/exec/ExtensionServiceLoaderModule.java +43 -0
- data/embulk-core/src/main/java/org/embulk/exec/ForSystemConfig.java +16 -0
- data/embulk-core/src/main/java/org/embulk/exec/GuessExecutor.java +307 -0
- data/embulk-core/src/main/java/org/embulk/exec/LocalExecutor.java +274 -0
- data/embulk-core/src/main/java/org/embulk/exec/LoggerProvider.java +30 -0
- data/embulk-core/src/main/java/org/embulk/exec/NoSampleException.java +10 -0
- data/embulk-core/src/main/java/org/embulk/exec/PooledBufferAllocator.java +58 -0
- data/embulk-core/src/main/java/org/embulk/exec/PreviewExecutor.java +138 -0
- data/embulk-core/src/main/java/org/embulk/exec/PreviewResult.java +27 -0
- data/embulk-core/src/main/java/org/embulk/exec/PreviewedNoticeError.java +17 -0
- data/embulk-core/src/main/java/org/embulk/exec/SamplingParserPlugin.java +116 -0
- data/embulk-core/src/main/java/org/embulk/exec/SystemConfigModule.java +24 -0
- data/embulk-core/src/main/java/org/embulk/jruby/JRubyPluginSource.java +69 -0
- data/embulk-core/src/main/java/org/embulk/jruby/JRubyScriptingModule.java +100 -0
- data/embulk-core/src/main/java/org/embulk/plugin/BuiltinPluginSourceModule.java +17 -0
- data/embulk-core/src/main/java/org/embulk/plugin/InjectedPluginSource.java +92 -0
- data/embulk-core/src/main/java/org/embulk/plugin/PluginManager.java +34 -0
- data/embulk-core/src/main/java/org/embulk/plugin/PluginSource.java +6 -0
- data/embulk-core/src/main/java/org/embulk/plugin/PluginSourceNotMatchException.java +19 -0
- data/embulk-core/src/main/java/org/embulk/plugin/PluginType.java +47 -0
- data/embulk-core/src/main/java/org/embulk/plugin/SetThreadContextClassLoader.java +19 -0
- data/embulk-core/src/main/java/org/embulk/spi/Buffer.java +113 -0
- data/embulk-core/src/main/java/org/embulk/spi/BufferAllocator.java +8 -0
- data/embulk-core/src/main/java/org/embulk/spi/Column.java +92 -0
- data/embulk-core/src/main/java/org/embulk/spi/ColumnConfig.java +79 -0
- data/embulk-core/src/main/java/org/embulk/spi/DecoderPlugin.java +16 -0
- data/embulk-core/src/main/java/org/embulk/spi/EncoderPlugin.java +16 -0
- data/embulk-core/src/main/java/org/embulk/spi/Exec.java +76 -0
- data/embulk-core/src/main/java/org/embulk/spi/ExecAction.java +6 -0
- data/embulk-core/src/main/java/org/embulk/spi/ExecSession.java +105 -0
- data/embulk-core/src/main/java/org/embulk/spi/Extension.java +42 -0
- data/embulk-core/src/main/java/org/embulk/spi/FileInput.java +11 -0
- data/embulk-core/src/main/java/org/embulk/spi/FileInputPlugin.java +19 -0
- data/embulk-core/src/main/java/org/embulk/spi/FileInputRunner.java +113 -0
- data/embulk-core/src/main/java/org/embulk/spi/FileOutput.java +13 -0
- data/embulk-core/src/main/java/org/embulk/spi/FileOutputPlugin.java +20 -0
- data/embulk-core/src/main/java/org/embulk/spi/FileOutputRunner.java +167 -0
- data/embulk-core/src/main/java/org/embulk/spi/FormatterPlugin.java +18 -0
- data/embulk-core/src/main/java/org/embulk/spi/GuessPlugin.java +9 -0
- data/embulk-core/src/main/java/org/embulk/spi/InputPlugin.java +20 -0
- data/embulk-core/src/main/java/org/embulk/spi/OutputPlugin.java +21 -0
- data/embulk-core/src/main/java/org/embulk/spi/Page.java +45 -0
- data/embulk-core/src/main/java/org/embulk/spi/PageBuilder.java +327 -0
- data/embulk-core/src/main/java/org/embulk/spi/PageFormat.java +47 -0
- data/embulk-core/src/main/java/org/embulk/spi/PageOutput.java +11 -0
- data/embulk-core/src/main/java/org/embulk/spi/PageReader.java +227 -0
- data/embulk-core/src/main/java/org/embulk/spi/ParserPlugin.java +17 -0
- data/embulk-core/src/main/java/org/embulk/spi/Schema.java +101 -0
- data/embulk-core/src/main/java/org/embulk/spi/SchemaConfig.java +52 -0
- data/embulk-core/src/main/java/org/embulk/spi/SchemaVisitor.java +14 -0
- data/embulk-core/src/main/java/org/embulk/spi/Transactional.java +10 -0
- data/embulk-core/src/main/java/org/embulk/spi/TransactionalFileInput.java +17 -0
- data/embulk-core/src/main/java/org/embulk/spi/TransactionalFileOutput.java +19 -0
- data/embulk-core/src/main/java/org/embulk/spi/TransactionalPageOutput.java +17 -0
- data/embulk-core/src/main/java/org/embulk/spi/time/DateTimeZoneSerDe.java +57 -0
- data/embulk-core/src/main/java/org/embulk/spi/time/JRubyTimeParserHelper.java +8 -0
- data/embulk-core/src/main/java/org/embulk/spi/time/JRubyTimeParserHelperFactory.java +6 -0
- data/embulk-core/src/main/java/org/embulk/spi/time/Timestamp.java +159 -0
- data/embulk-core/src/main/java/org/embulk/spi/time/TimestampFormat.java +98 -0
- data/embulk-core/src/main/java/org/embulk/spi/time/TimestampFormatter.java +55 -0
- data/embulk-core/src/main/java/org/embulk/spi/time/TimestampParseException.java +6 -0
- data/embulk-core/src/main/java/org/embulk/spi/time/TimestampParser.java +60 -0
- data/embulk-core/src/main/java/org/embulk/spi/time/TimestampSerDe.java +50 -0
- data/embulk-core/src/main/java/org/embulk/spi/type/AbstractType.java +55 -0
- data/embulk-core/src/main/java/org/embulk/spi/type/BooleanType.java +12 -0
- data/embulk-core/src/main/java/org/embulk/spi/type/DoubleType.java +12 -0
- data/embulk-core/src/main/java/org/embulk/spi/type/LongType.java +12 -0
- data/embulk-core/src/main/java/org/embulk/spi/type/StringType.java +12 -0
- data/embulk-core/src/main/java/org/embulk/spi/type/TimestampType.java +39 -0
- data/embulk-core/src/main/java/org/embulk/spi/type/Type.java +15 -0
- data/embulk-core/src/main/java/org/embulk/spi/type/TypeDeserializer.java +47 -0
- data/embulk-core/src/main/java/org/embulk/spi/type/Types.java +14 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/CharsetSerDe.java +55 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/Decoders.java +81 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/Encoders.java +81 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/FileInputInputStream.java +110 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/FileOutputOutputStream.java +94 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/InputStreamFileInput.java +111 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/Inputs.java +74 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/LineDecoder.java +118 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/LineEncoder.java +109 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/ListFileInput.java +52 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/Newline.java +38 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/PagePrinter.java +102 -0
- data/embulk-core/src/main/java/org/embulk/spi/util/Pages.java +139 -0
- data/embulk-core/src/test/java/org/embulk/EmbulkTestRuntime.java +110 -0
- data/embulk-core/src/test/java/org/embulk/GuiceBinder.java +72 -0
- data/embulk-core/src/test/java/org/embulk/RandomManager.java +53 -0
- data/embulk-core/src/test/java/org/embulk/TestPluginSourceModule.java +23 -0
- data/embulk-core/src/test/java/org/embulk/TestUtilityModule.java +17 -0
- data/embulk-core/src/test/java/org/embulk/config/TestConfigSource.java +114 -0
- data/embulk-core/src/test/java/org/embulk/config/TestTaskSource.java +70 -0
- data/embulk-core/src/test/java/org/embulk/plugin/MockPluginSource.java +57 -0
- data/embulk-core/src/test/java/org/embulk/plugin/TestPluginType.java +18 -0
- data/embulk-core/src/test/java/org/embulk/spi/MockFileOutput.java +63 -0
- data/embulk-core/src/test/java/org/embulk/spi/MockFormatterPlugin.java +101 -0
- data/embulk-core/src/test/java/org/embulk/spi/MockParserPlugin.java +73 -0
- data/embulk-core/src/test/java/org/embulk/spi/PageTestUtils.java +78 -0
- data/embulk-core/src/test/java/org/embulk/spi/TestFileInputInputStream.java +67 -0
- data/embulk-core/src/test/java/org/embulk/spi/TestFileInputRunner.java +180 -0
- data/embulk-core/src/test/java/org/embulk/spi/TestFileOutputRunner.java +192 -0
- data/embulk-core/src/test/java/org/embulk/spi/TestInputStreamFileInput.java +188 -0
- data/embulk-core/src/test/java/org/embulk/spi/TestPageBuilderReader.java +301 -0
- data/embulk-core/src/test/java/org/embulk/spi/time/TestTimestamp.java +116 -0
- data/embulk-core/src/test/java/org/embulk/spi/time/TestTimestampFormatterParser.java +52 -0
- data/embulk-core/src/test/java/org/embulk/spi/type/TestTypeSerDe.java +45 -0
- data/embulk-core/src/test/java/org/embulk/spi/util/TestLineDecoder.java +132 -0
- data/embulk-core/src/test/java/org/embulk/spi/util/TestLineEncoder.java +123 -0
- data/embulk-standards/build.gradle +6 -0
- data/embulk-standards/pom.xml +68 -0
- data/embulk-standards/src/main/java/org/embulk/standards/CsvFormatterPlugin.java +158 -0
- data/embulk-standards/src/main/java/org/embulk/standards/CsvParserPlugin.java +233 -0
- data/embulk-standards/src/main/java/org/embulk/standards/CsvTokenizer.java +355 -0
- data/embulk-standards/src/main/java/org/embulk/standards/GzipFileDecoderPlugin.java +55 -0
- data/embulk-standards/src/main/java/org/embulk/standards/GzipFileEncoderPlugin.java +39 -0
- data/embulk-standards/src/main/java/org/embulk/standards/LocalFileInputPlugin.java +138 -0
- data/embulk-standards/src/main/java/org/embulk/standards/LocalFileOutputPlugin.java +128 -0
- data/embulk-standards/src/main/java/org/embulk/standards/NullOutputPlugin.java +46 -0
- data/embulk-standards/src/main/java/org/embulk/standards/S3FileInputPlugin.java +238 -0
- data/embulk-standards/src/main/java/org/embulk/standards/StandardPluginExtension.java +16 -0
- data/embulk-standards/src/main/java/org/embulk/standards/StandardPluginModule.java +44 -0
- data/embulk-standards/src/main/java/org/embulk/standards/StdoutOutputPlugin.java +71 -0
- data/embulk-standards/src/main/resources/META-INF/services/org.embulk.spi.Extension +1 -0
- data/embulk-standards/src/test/java/org/embulk/standards/TestCsvParserPlugin.java +69 -0
- data/embulk-standards/src/test/java/org/embulk/standards/TestCsvTokenizer.java +291 -0
- data/embulk-standards/src/test/java/org/embulk/standards/TestS3FileInputPlugin.java +43 -0
- data/embulk.gemspec +27 -0
- data/examples/config.yml +34 -0
- data/examples/csv/sample.csv.gz +0 -0
- data/gradle/wrapper/gradle-wrapper.jar +0 -0
- data/gradle/wrapper/gradle-wrapper.properties +6 -0
- data/gradlew +164 -0
- data/gradlew.bat +90 -0
- data/lib/embulk.rb +16 -0
- data/lib/embulk/buffer.rb +17 -0
- data/lib/embulk/column.rb +47 -0
- data/lib/embulk/command/embulk.rb +39 -0
- data/lib/embulk/command/embulk_example.rb +32 -0
- data/lib/embulk/command/embulk_generate_bin.rb +62 -0
- data/lib/embulk/command/embulk_run.rb +243 -0
- data/lib/embulk/data/bundle/.bundle/config +3 -0
- data/lib/embulk/data/bundle/Gemfile +31 -0
- data/lib/embulk/data/bundle/Gemfile.lock +8 -0
- data/lib/embulk/data/bundle/embulk/input_example.rb +40 -0
- data/lib/embulk/data/bundle/embulk/output_example.rb +51 -0
- data/lib/embulk/data_source.rb +66 -0
- data/lib/embulk/error.rb +5 -0
- data/lib/embulk/guess_charset.rb +26 -0
- data/lib/embulk/guess_csv.rb +195 -0
- data/lib/embulk/guess_gzip.rb +18 -0
- data/lib/embulk/guess_newline.rb +20 -0
- data/lib/embulk/guess_plugin.rb +113 -0
- data/lib/embulk/input_plugin.rb +53 -0
- data/lib/embulk/java/bootstrap.rb +12 -0
- data/lib/embulk/java/imports.rb +26 -0
- data/lib/embulk/java/time_helper.rb +77 -0
- data/lib/embulk/output_plugin.rb +104 -0
- data/lib/embulk/page.rb +28 -0
- data/lib/embulk/page_builder.rb +22 -0
- data/lib/embulk/plugin.rb +152 -0
- data/lib/embulk/plugin_registry.rb +70 -0
- data/lib/embulk/schema.rb +85 -0
- data/lib/embulk/time_format_guess.rb +331 -0
- data/lib/embulk/version.rb +3 -0
- data/pom.xml +533 -0
- data/settings.gradle +5 -0
- metadata +370 -0
@@ -0,0 +1,53 @@
|
|
1
|
+
module Embulk
|
2
|
+
|
3
|
+
require 'embulk/data_source'
|
4
|
+
require 'embulk/schema'
|
5
|
+
require 'embulk/page_builder'
|
6
|
+
|
7
|
+
class InputPlugin
|
8
|
+
def self.transaction(config, &control)
|
9
|
+
raise NotImplementedError, "InputPlugin.transaction(config, &control) must be implemented"
|
10
|
+
end
|
11
|
+
|
12
|
+
def self.run(task, schema, index, page_builder)
|
13
|
+
raise NotImplementedError, "InputPlugin#run(task, schema, index, page_builder) must be implemented"
|
14
|
+
end
|
15
|
+
|
16
|
+
if Embulk.java?
|
17
|
+
def self.java_object
|
18
|
+
JavaAdapter.new(self)
|
19
|
+
end
|
20
|
+
|
21
|
+
class JavaAdapter
|
22
|
+
include Java::InputPlugin
|
23
|
+
|
24
|
+
def initialize(ruby_class)
|
25
|
+
@ruby_class = ruby_class
|
26
|
+
end
|
27
|
+
|
28
|
+
def transaction(java_config, java_control)
|
29
|
+
config = DataSource.from_java_object(java_config)
|
30
|
+
next_config_hash = @ruby_class.transaction(config) do |task_source_hash,columns,processor_count|
|
31
|
+
java_task_source = DataSource.from_ruby_hash(task_source_hash).java_object
|
32
|
+
java_schema = Schema.new(columns).java_object
|
33
|
+
java_commit_reports = java_control.run(java_task_source, java_schema, processor_count)
|
34
|
+
java_commit_reports.map {|java_commit_report|
|
35
|
+
DataSource.from_java_object(java_commit_report)
|
36
|
+
}
|
37
|
+
end
|
38
|
+
# TODO check return type of #transaction
|
39
|
+
return DataSource.from_ruby_hash(next_config_hash).java_object
|
40
|
+
end
|
41
|
+
|
42
|
+
def run(java_task_source, java_schema, processor_index, java_output)
|
43
|
+
task_source = DataSource.from_java_object(java_task_source)
|
44
|
+
schema = Schema.from_java_object(java_schema)
|
45
|
+
page_builder = PageBuilder.new(schema, java_output)
|
46
|
+
commit_report_hash = @ruby_class.run(task_source, schema, processor_index, page_builder)
|
47
|
+
return DataSource.from_ruby_hash(commit_report_hash).java_object
|
48
|
+
end
|
49
|
+
end
|
50
|
+
end
|
51
|
+
end
|
52
|
+
|
53
|
+
end
|
@@ -0,0 +1,26 @@
|
|
1
|
+
require 'java'
|
2
|
+
|
3
|
+
#
|
4
|
+
# this file is loaded by embulk/java/bootstrap.rb
|
5
|
+
#
|
6
|
+
|
7
|
+
module Embulk::Java
|
8
|
+
java_import 'org.embulk.spi.Buffer'
|
9
|
+
|
10
|
+
java_import 'org.embulk.config.DataSourceImpl'
|
11
|
+
java_import 'org.embulk.spi.time.Timestamp'
|
12
|
+
java_import 'org.embulk.spi.GuessPlugin'
|
13
|
+
java_import 'org.embulk.spi.OutputPlugin'
|
14
|
+
java_import 'org.embulk.spi.InputPlugin'
|
15
|
+
java_import 'org.embulk.spi.TransactionalPageOutput'
|
16
|
+
java_import 'org.embulk.spi.PageReader'
|
17
|
+
java_import 'org.embulk.spi.PageBuilder'
|
18
|
+
java_import 'org.embulk.spi.util.LineDecoder'
|
19
|
+
java_import 'org.embulk.spi.util.ListFileInput'
|
20
|
+
java_import 'org.embulk.spi.Schema'
|
21
|
+
java_import 'org.embulk.spi.Column'
|
22
|
+
java_import 'org.embulk.spi.type.Type'
|
23
|
+
java_import 'org.embulk.spi.type.Types'
|
24
|
+
|
25
|
+
# TODO
|
26
|
+
end
|
@@ -0,0 +1,77 @@
|
|
1
|
+
module Embulk
|
2
|
+
module Java
|
3
|
+
#require 'embulk/java/imports'
|
4
|
+
require 'time' # Date._strptime
|
5
|
+
|
6
|
+
class TimeParserHelper
|
7
|
+
#include Java::JRubyTimeParserHelper
|
8
|
+
include org.embulk.spi.time.JRubyTimeParserHelper
|
9
|
+
|
10
|
+
class Factory
|
11
|
+
#include Java::JRubyTimeParserHelperFactory
|
12
|
+
include org.embulk.spi.time.JRubyTimeParserHelperFactory
|
13
|
+
|
14
|
+
def newInstance(format_string, year, mon, day, hour, min, sec, usec)
|
15
|
+
default_time = Time.utc(year, mon, day, hour, min, sec, usec)
|
16
|
+
TimeParserHelper.new(format_string, default_time)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
|
20
|
+
def initialize(format_string, default_time)
|
21
|
+
@format_string = format_string
|
22
|
+
@default_time = default_time
|
23
|
+
end
|
24
|
+
|
25
|
+
# Override
|
26
|
+
def strptime(text)
|
27
|
+
hash = Date._strptime(text, @format_string)
|
28
|
+
unless hash
|
29
|
+
raise Java::TimestampParseException.new
|
30
|
+
end
|
31
|
+
|
32
|
+
if seconds = hash[:seconds]
|
33
|
+
return seconds * 1_000_000
|
34
|
+
|
35
|
+
else
|
36
|
+
year = hash[:year]
|
37
|
+
mon = hash[:mon]
|
38
|
+
day = hash[:mday]
|
39
|
+
hour = hash[:hour]
|
40
|
+
min = hash[:min]
|
41
|
+
sec = hash[:sec]
|
42
|
+
sec_fraction = hash[:sec_fraction]
|
43
|
+
zone = hash[:zone]
|
44
|
+
usec = hash[:sec_fraction] ? hash[:sec_fraction] * 1000000 : nil
|
45
|
+
|
46
|
+
now = @default_time
|
47
|
+
begin
|
48
|
+
break if year; year = now.year
|
49
|
+
break if mon; mon = now.mon
|
50
|
+
break if day; day = now.day
|
51
|
+
break if hour; hour = now.hour
|
52
|
+
break if min; min = now.min
|
53
|
+
break if sec; sec = now.sec
|
54
|
+
break if sec_fraction; usec = now.tv_usec
|
55
|
+
end until true
|
56
|
+
|
57
|
+
year ||= 1970
|
58
|
+
mon ||= 1
|
59
|
+
day ||= 1
|
60
|
+
hour ||= 0
|
61
|
+
min ||= 0
|
62
|
+
sec ||= 0
|
63
|
+
usec ||= 0
|
64
|
+
|
65
|
+
@zone = zone
|
66
|
+
time = Time.utc(year, mon, day, hour, min, sec, usec)
|
67
|
+
return time.tv_sec * 1000 + time.tv_usec / 1000
|
68
|
+
end
|
69
|
+
end
|
70
|
+
|
71
|
+
# Override
|
72
|
+
def getZone
|
73
|
+
@zone
|
74
|
+
end
|
75
|
+
end
|
76
|
+
end
|
77
|
+
end
|
@@ -0,0 +1,104 @@
|
|
1
|
+
module Embulk
|
2
|
+
|
3
|
+
require 'embulk/data_source'
|
4
|
+
require 'embulk/schema'
|
5
|
+
require 'embulk/page'
|
6
|
+
|
7
|
+
class OutputPlugin
|
8
|
+
def self.transaction(config, schema, processor_count, &control)
|
9
|
+
yield(config)
|
10
|
+
return {}
|
11
|
+
end
|
12
|
+
|
13
|
+
def initialize(task, schema, index)
|
14
|
+
@task = task
|
15
|
+
@schema = schema
|
16
|
+
@index = index
|
17
|
+
end
|
18
|
+
|
19
|
+
attr_reader :task, :schema, :index
|
20
|
+
|
21
|
+
def add(page)
|
22
|
+
raise NotImplementedError, "OutputPlugin#add(page) must be implemented"
|
23
|
+
end
|
24
|
+
|
25
|
+
def finish
|
26
|
+
end
|
27
|
+
|
28
|
+
def close
|
29
|
+
end
|
30
|
+
|
31
|
+
def abort
|
32
|
+
end
|
33
|
+
|
34
|
+
def commit
|
35
|
+
{}
|
36
|
+
end
|
37
|
+
|
38
|
+
if Embulk.java?
|
39
|
+
def self.java_object
|
40
|
+
JavaAdapter.new(self)
|
41
|
+
end
|
42
|
+
|
43
|
+
class JavaAdapter
|
44
|
+
include Java::OutputPlugin
|
45
|
+
|
46
|
+
def initialize(ruby_class)
|
47
|
+
@ruby_class = ruby_class
|
48
|
+
end
|
49
|
+
|
50
|
+
def transaction(java_config, java_schema, processor_count, java_control)
|
51
|
+
config = DataSource.from_java_object(java_config)
|
52
|
+
schema = Schema.from_java_object(java_schema)
|
53
|
+
next_config_hash = @ruby_class.transaction(config, schema, processor_count) do |task_source_hash|
|
54
|
+
java_task_source = DataSource.from_ruby_hash(task_source_hash).java_object
|
55
|
+
java_commit_reports = java_control.run(java_task_source)
|
56
|
+
java_commit_reports.map {|java_commit_report|
|
57
|
+
DataSource.from_java_object(java_commit_report)
|
58
|
+
}
|
59
|
+
end
|
60
|
+
# TODO check return type of #transaction
|
61
|
+
return DataSource.from_ruby_hash(next_config_hash).java_object
|
62
|
+
end
|
63
|
+
|
64
|
+
def open(java_task_source, java_schema, processor_index)
|
65
|
+
task_source = DataSource.from_java_object(java_task_source)
|
66
|
+
schema = Schema.from_java_object(java_schema)
|
67
|
+
ruby_object = @ruby_class.new(task_source, schema, processor_index)
|
68
|
+
return OutputAdapter.new(ruby_object, schema)
|
69
|
+
end
|
70
|
+
|
71
|
+
class OutputAdapter
|
72
|
+
include Java::TransactionalPageOutput
|
73
|
+
|
74
|
+
def initialize(ruby_object, schema)
|
75
|
+
@ruby_object = ruby_object
|
76
|
+
@schema = schema
|
77
|
+
end
|
78
|
+
|
79
|
+
def add(java_page)
|
80
|
+
@ruby_object.add Page.new(java_page, @schema)
|
81
|
+
end
|
82
|
+
|
83
|
+
def finish
|
84
|
+
@ruby_object.finish
|
85
|
+
end
|
86
|
+
|
87
|
+
def close
|
88
|
+
@ruby_object.close
|
89
|
+
end
|
90
|
+
|
91
|
+
def abort
|
92
|
+
@ruby_object.abort
|
93
|
+
end
|
94
|
+
|
95
|
+
def commit
|
96
|
+
commit_report_hash = @ruby_object.commit
|
97
|
+
return DataSource.from_ruby_hash(commit_report_hash).java_object
|
98
|
+
end
|
99
|
+
end
|
100
|
+
end
|
101
|
+
end
|
102
|
+
end
|
103
|
+
|
104
|
+
end
|
data/lib/embulk/page.rb
ADDED
@@ -0,0 +1,28 @@
|
|
1
|
+
|
2
|
+
module Embulk
|
3
|
+
|
4
|
+
# TODO pure-ruby page reader
|
5
|
+
|
6
|
+
class Page
|
7
|
+
def initialize(java_page, schema)
|
8
|
+
@java_page = java_page
|
9
|
+
@schema = schema
|
10
|
+
end
|
11
|
+
|
12
|
+
attr_reader :schema
|
13
|
+
|
14
|
+
def each
|
15
|
+
schema = @schema
|
16
|
+
reader = Java::PageReader.new(schema.java_object)
|
17
|
+
begin
|
18
|
+
reader.setPage(@java_page)
|
19
|
+
while reader.nextRecord
|
20
|
+
yield schema.read_record(reader)
|
21
|
+
end
|
22
|
+
ensure
|
23
|
+
reader.close
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
27
|
+
|
28
|
+
end
|
@@ -0,0 +1,22 @@
|
|
1
|
+
module Embulk
|
2
|
+
|
3
|
+
class PageBuilder
|
4
|
+
def initialize(schema, java_page_output)
|
5
|
+
@page_builder = Java::PageBuilder.new(Java::Injected::BufferAllocator, schema.java_object, java_page_output)
|
6
|
+
@schema = schema
|
7
|
+
end
|
8
|
+
|
9
|
+
def add(record)
|
10
|
+
@schema.write_record(@page_builder, record)
|
11
|
+
end
|
12
|
+
|
13
|
+
def finish
|
14
|
+
@page_builder.finish
|
15
|
+
end
|
16
|
+
|
17
|
+
def close
|
18
|
+
@page_builder.close
|
19
|
+
end
|
20
|
+
end
|
21
|
+
|
22
|
+
end
|
@@ -0,0 +1,152 @@
|
|
1
|
+
|
2
|
+
module Embulk
|
3
|
+
require 'forwardable'
|
4
|
+
require 'embulk/error'
|
5
|
+
require 'embulk/plugin_registry'
|
6
|
+
require 'embulk/input_plugin'
|
7
|
+
require 'embulk/output_plugin'
|
8
|
+
#require 'embulk/parser_plugin'
|
9
|
+
#require 'embulk/formatter_plugin'
|
10
|
+
#require 'embulk/decoder_plugin'
|
11
|
+
#require 'embulk/encoder_plugin'
|
12
|
+
require 'embulk/guess_plugin'
|
13
|
+
|
14
|
+
class PluginManager
|
15
|
+
def initialize
|
16
|
+
@registries = {}
|
17
|
+
%w[input output parser formatter decoder encoder line_filter guess].each do |category|
|
18
|
+
@registries[category.to_sym] = PluginRegistry.new(category, "embulk/#{category}_")
|
19
|
+
end
|
20
|
+
end
|
21
|
+
|
22
|
+
def register_input(type, klass)
|
23
|
+
register_plugin(:input, type, klass, InputPlugin)
|
24
|
+
end
|
25
|
+
|
26
|
+
def register_output(type, klass)
|
27
|
+
register_plugin(:output, type, klass, OutputPlugin,
|
28
|
+
"Output plugin #{klass} must extend OutputPlugin")
|
29
|
+
end
|
30
|
+
|
31
|
+
def register_parser(type, klass)
|
32
|
+
register_plugin(:parser, type, klass, ParserPlugin)
|
33
|
+
end
|
34
|
+
|
35
|
+
def register_formatter(type, klass)
|
36
|
+
register_plugin(:formatter, type, klass, FormatterPlugin)
|
37
|
+
end
|
38
|
+
|
39
|
+
def register_decoder(type, klass)
|
40
|
+
register_plugin(:decoder, type, klass, DecoderPlugin)
|
41
|
+
end
|
42
|
+
|
43
|
+
def register_encoder(type, klass)
|
44
|
+
register_plugin(:encoder, type, klass, EncoderPlugin)
|
45
|
+
end
|
46
|
+
|
47
|
+
def register_guess(type, klass)
|
48
|
+
register_plugin(:guess, type, klass, GuessPlugin,
|
49
|
+
"Guess plugin #{klass} must extend GuessPlugin, LineGuessPlugin, or TextGuessPlugin class")
|
50
|
+
end
|
51
|
+
|
52
|
+
def get_input(type)
|
53
|
+
# TODO not implemented yet
|
54
|
+
lookup(:guess, type)
|
55
|
+
end
|
56
|
+
|
57
|
+
def get_output(type)
|
58
|
+
# TODO not implemented yet
|
59
|
+
lookup(:guess, type)
|
60
|
+
end
|
61
|
+
|
62
|
+
def get_parser(type)
|
63
|
+
# TODO not implemented yet
|
64
|
+
lookup(:guess, type)
|
65
|
+
end
|
66
|
+
|
67
|
+
def get_formatter(type)
|
68
|
+
# TODO not implemented yet
|
69
|
+
lookup(:guess, type)
|
70
|
+
end
|
71
|
+
|
72
|
+
def get_decoder(type)
|
73
|
+
# TODO not implemented yet
|
74
|
+
lookup(:guess, type)
|
75
|
+
end
|
76
|
+
|
77
|
+
def get_encoder(type)
|
78
|
+
# TODO not implemented yet
|
79
|
+
lookup(:guess, type)
|
80
|
+
end
|
81
|
+
|
82
|
+
def get_guess(type)
|
83
|
+
# TODO not implemented yet
|
84
|
+
lookup(:guess, type)
|
85
|
+
end
|
86
|
+
|
87
|
+
def new_java_input(type)
|
88
|
+
lookup(:input, type).java_object
|
89
|
+
end
|
90
|
+
|
91
|
+
def new_java_output(type)
|
92
|
+
lookup(:output, type).java_object
|
93
|
+
end
|
94
|
+
|
95
|
+
def new_java_parser(type)
|
96
|
+
lookup(:parser, type).java_object
|
97
|
+
end
|
98
|
+
|
99
|
+
def new_java_formatter(type)
|
100
|
+
lookup(:formatter, type).java_object
|
101
|
+
end
|
102
|
+
|
103
|
+
def new_java_decoder(type)
|
104
|
+
lookup(:decoder, type).java_object
|
105
|
+
end
|
106
|
+
|
107
|
+
def new_java_encoder(type)
|
108
|
+
lookup(:encoder, type).java_object
|
109
|
+
end
|
110
|
+
|
111
|
+
def new_java_guess(type)
|
112
|
+
lookup(:guess, type).java_object
|
113
|
+
end
|
114
|
+
|
115
|
+
private
|
116
|
+
|
117
|
+
# TODO lookup should fallback to Java PluginSource
|
118
|
+
# if not found so that ruby plugins can call java plugins.
|
119
|
+
# call injector.newPlugin and wrap the instance in a reverse bridge object.
|
120
|
+
|
121
|
+
def lookup(category, type)
|
122
|
+
@registries[category].lookup(type)
|
123
|
+
end
|
124
|
+
|
125
|
+
def register_plugin(category, type, klass, iface, message=nil)
|
126
|
+
unless klass < iface
|
127
|
+
message ||= "Plugin #{klass} must implement #{iface}"
|
128
|
+
raise message
|
129
|
+
end
|
130
|
+
@registries[category].register(type, klass)
|
131
|
+
end
|
132
|
+
end
|
133
|
+
|
134
|
+
Plugin = PluginManager.new
|
135
|
+
|
136
|
+
#module Plugin
|
137
|
+
# class <<self
|
138
|
+
# INSTANCE = PluginManager.new
|
139
|
+
#
|
140
|
+
# extend Forwardable
|
141
|
+
#
|
142
|
+
# def_delegators 'INSTANCE',
|
143
|
+
# :register_input, :new_input, :new_java_input,
|
144
|
+
# :register_output, :new_output, :new_java_output,
|
145
|
+
# :register_parser, :new_parser, :new_java_parser,
|
146
|
+
# :register_formatter, :new_formatter, :new_java_formatter,
|
147
|
+
# :register_decoder, :new_decoder, :new_java_decoder,
|
148
|
+
# :register_encoder, :new_encoder, :new_java_encoder,
|
149
|
+
# :register_guess, :new_guess, :new_java_guess
|
150
|
+
# end
|
151
|
+
#end
|
152
|
+
end
|