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 +1,111 @@
|
|
1
|
+
package org.embulk.spi.util;
|
2
|
+
|
3
|
+
import java.util.Arrays;
|
4
|
+
import java.util.Iterator;
|
5
|
+
import java.io.InputStream;
|
6
|
+
import java.io.Closeable;
|
7
|
+
import java.io.IOException;
|
8
|
+
import org.embulk.spi.Buffer;
|
9
|
+
import org.embulk.spi.FileInput;
|
10
|
+
import org.embulk.spi.BufferAllocator;
|
11
|
+
|
12
|
+
public class InputStreamFileInput
|
13
|
+
implements FileInput
|
14
|
+
{
|
15
|
+
public interface Provider extends Closeable
|
16
|
+
{
|
17
|
+
public InputStream openNext() throws IOException;
|
18
|
+
|
19
|
+
public void close() throws IOException;
|
20
|
+
}
|
21
|
+
|
22
|
+
public static class IteratorProvider implements Provider {
|
23
|
+
private Iterator<InputStream> iterator;
|
24
|
+
|
25
|
+
public IteratorProvider(Iterable<InputStream> iterable)
|
26
|
+
{
|
27
|
+
this.iterator = iterable.iterator();
|
28
|
+
}
|
29
|
+
|
30
|
+
public IteratorProvider(Iterator<InputStream> iterator)
|
31
|
+
{
|
32
|
+
this.iterator = iterator;
|
33
|
+
}
|
34
|
+
|
35
|
+
public InputStream openNext() throws IOException
|
36
|
+
{
|
37
|
+
if (!iterator.hasNext()) {
|
38
|
+
return null;
|
39
|
+
}
|
40
|
+
return iterator.next();
|
41
|
+
}
|
42
|
+
|
43
|
+
@Override
|
44
|
+
public void close() throws IOException
|
45
|
+
{
|
46
|
+
while (iterator.hasNext()) {
|
47
|
+
iterator.next().close();
|
48
|
+
}
|
49
|
+
}
|
50
|
+
}
|
51
|
+
|
52
|
+
private final BufferAllocator allocator;
|
53
|
+
private final Provider provider;
|
54
|
+
private InputStream current;
|
55
|
+
|
56
|
+
public InputStreamFileInput(BufferAllocator allocator, Provider provider)
|
57
|
+
{
|
58
|
+
this.allocator = allocator;
|
59
|
+
this.provider = provider;
|
60
|
+
this.current = null;
|
61
|
+
}
|
62
|
+
|
63
|
+
public Buffer poll()
|
64
|
+
{
|
65
|
+
// TODO check current != null and throw Illegal State - file is not opened
|
66
|
+
if (current == null) {
|
67
|
+
throw new IllegalStateException("openNext must be called before poll()");
|
68
|
+
}
|
69
|
+
Buffer buffer = allocator.allocate();
|
70
|
+
try {
|
71
|
+
int n = current.read(buffer.array(), buffer.offset(), buffer.capacity());
|
72
|
+
if (n < 0) {
|
73
|
+
return null;
|
74
|
+
}
|
75
|
+
buffer.limit(n);
|
76
|
+
Buffer b = buffer;
|
77
|
+
buffer = null;
|
78
|
+
return b;
|
79
|
+
} catch (IOException ex) {
|
80
|
+
throw new RuntimeException(ex);
|
81
|
+
} finally {
|
82
|
+
if (buffer != null) {
|
83
|
+
buffer.release();
|
84
|
+
buffer = null;
|
85
|
+
}
|
86
|
+
}
|
87
|
+
}
|
88
|
+
|
89
|
+
public boolean nextFile()
|
90
|
+
{
|
91
|
+
try {
|
92
|
+
if (current != null) {
|
93
|
+
current.close();
|
94
|
+
current = null;
|
95
|
+
}
|
96
|
+
current = provider.openNext();
|
97
|
+
return current != null;
|
98
|
+
} catch (IOException ex) {
|
99
|
+
throw new RuntimeException(ex);
|
100
|
+
}
|
101
|
+
}
|
102
|
+
|
103
|
+
public void close()
|
104
|
+
{
|
105
|
+
try {
|
106
|
+
provider.close();
|
107
|
+
} catch (IOException ex) {
|
108
|
+
throw new RuntimeException(ex);
|
109
|
+
}
|
110
|
+
}
|
111
|
+
}
|
@@ -0,0 +1,74 @@
|
|
1
|
+
package org.embulk.spi.util;
|
2
|
+
|
3
|
+
import java.util.Iterator;
|
4
|
+
import java.util.NoSuchElementException;
|
5
|
+
|
6
|
+
import org.embulk.spi.Exec;
|
7
|
+
import org.embulk.spi.FileInput;
|
8
|
+
import org.embulk.spi.Buffer;
|
9
|
+
import org.embulk.spi.time.Timestamp;
|
10
|
+
import org.embulk.spi.time.TimestampFormatter;
|
11
|
+
import org.joda.time.DateTimeZone;
|
12
|
+
|
13
|
+
public class Inputs
|
14
|
+
{
|
15
|
+
private static abstract class AbstractPollIterator <E>
|
16
|
+
implements Iterator<E>
|
17
|
+
{
|
18
|
+
private E next;
|
19
|
+
|
20
|
+
protected abstract E poll();
|
21
|
+
|
22
|
+
@Override
|
23
|
+
public boolean hasNext()
|
24
|
+
{
|
25
|
+
if (next != null) {
|
26
|
+
return true;
|
27
|
+
} else {
|
28
|
+
next = poll();
|
29
|
+
return next != null;
|
30
|
+
}
|
31
|
+
}
|
32
|
+
|
33
|
+
@Override
|
34
|
+
public E next()
|
35
|
+
{
|
36
|
+
if (!hasNext()) {
|
37
|
+
throw new NoSuchElementException();
|
38
|
+
}
|
39
|
+
E l = next;
|
40
|
+
next = null;
|
41
|
+
return l;
|
42
|
+
}
|
43
|
+
|
44
|
+
@Override
|
45
|
+
public void remove()
|
46
|
+
{
|
47
|
+
throw new UnsupportedOperationException();
|
48
|
+
}
|
49
|
+
}
|
50
|
+
|
51
|
+
public static Iterable<Buffer> each(final FileInput input)
|
52
|
+
{
|
53
|
+
return new Iterable<Buffer>() {
|
54
|
+
public Iterator<Buffer> iterator()
|
55
|
+
{
|
56
|
+
return new AbstractPollIterator<Buffer>() {
|
57
|
+
public Buffer poll()
|
58
|
+
{
|
59
|
+
return input.poll();
|
60
|
+
}
|
61
|
+
};
|
62
|
+
}
|
63
|
+
};
|
64
|
+
}
|
65
|
+
|
66
|
+
public static String formatPath(String pathFormat)
|
67
|
+
{
|
68
|
+
Timestamp timestamp = Exec.session().getTransactionTime();
|
69
|
+
DateTimeZone timezone = Exec.session().getTransactionTimeZone();
|
70
|
+
// newTimestampFormatter (eventually calls org.jruby.util.RubyDateFormat.<init> doesn't throw exceptions
|
71
|
+
TimestampFormatter formatter = Exec.session().newTimestampFormatter(pathFormat, timezone);
|
72
|
+
return formatter.format(timestamp); // TimestampFormatter.format doesn't throw exceptions
|
73
|
+
}
|
74
|
+
}
|
@@ -0,0 +1,118 @@
|
|
1
|
+
package org.embulk.spi.util;
|
2
|
+
|
3
|
+
import java.util.Iterator;
|
4
|
+
import java.util.NoSuchElementException;
|
5
|
+
import java.io.Reader;
|
6
|
+
import java.io.BufferedReader;
|
7
|
+
import java.io.InputStreamReader;
|
8
|
+
import java.io.IOException;
|
9
|
+
import java.nio.charset.Charset;
|
10
|
+
import java.nio.charset.CharsetDecoder;
|
11
|
+
import java.nio.charset.CodingErrorAction;
|
12
|
+
import org.embulk.config.Task;
|
13
|
+
import org.embulk.config.Config;
|
14
|
+
import org.embulk.config.ConfigDefault;
|
15
|
+
import org.embulk.spi.FileInput;
|
16
|
+
|
17
|
+
public class LineDecoder
|
18
|
+
implements AutoCloseable, Iterable<String>
|
19
|
+
{
|
20
|
+
// TODO optimize
|
21
|
+
|
22
|
+
public static interface DecoderTask
|
23
|
+
extends Task
|
24
|
+
{
|
25
|
+
@Config("charset")
|
26
|
+
@ConfigDefault("\"utf-8\"")
|
27
|
+
public Charset getCharset();
|
28
|
+
|
29
|
+
@Config("newline")
|
30
|
+
@ConfigDefault("\"CRLF\"")
|
31
|
+
public Newline getNewline();
|
32
|
+
}
|
33
|
+
|
34
|
+
private final FileInputInputStream inputStream;
|
35
|
+
private final BufferedReader reader;
|
36
|
+
|
37
|
+
public LineDecoder(FileInput in, DecoderTask task)
|
38
|
+
{
|
39
|
+
CharsetDecoder decoder = task.getCharset()
|
40
|
+
.newDecoder()
|
41
|
+
.onMalformedInput(CodingErrorAction.REPLACE) // TODO configurable?
|
42
|
+
.onUnmappableCharacter(CodingErrorAction.REPLACE); // TODO configurable?
|
43
|
+
this.inputStream = new FileInputInputStream(in);
|
44
|
+
this.reader = new BufferedReader(new InputStreamReader(inputStream, decoder));
|
45
|
+
}
|
46
|
+
|
47
|
+
public boolean nextFile()
|
48
|
+
{
|
49
|
+
return inputStream.nextFile();
|
50
|
+
}
|
51
|
+
|
52
|
+
public String poll()
|
53
|
+
{
|
54
|
+
try {
|
55
|
+
return reader.readLine();
|
56
|
+
} catch (IOException ex) {
|
57
|
+
// unexpected
|
58
|
+
throw new RuntimeException(ex);
|
59
|
+
}
|
60
|
+
}
|
61
|
+
|
62
|
+
public void close()
|
63
|
+
{
|
64
|
+
try {
|
65
|
+
reader.close();
|
66
|
+
} catch (IOException ex) {
|
67
|
+
// unexpected
|
68
|
+
throw new RuntimeException(ex);
|
69
|
+
}
|
70
|
+
}
|
71
|
+
|
72
|
+
public Iterator<String> iterator()
|
73
|
+
{
|
74
|
+
return new Ite(this);
|
75
|
+
}
|
76
|
+
|
77
|
+
private String nextLine;
|
78
|
+
|
79
|
+
private static class Ite
|
80
|
+
implements Iterator<String>
|
81
|
+
{
|
82
|
+
private LineDecoder self;
|
83
|
+
|
84
|
+
public Ite(LineDecoder self)
|
85
|
+
{
|
86
|
+
// TODO non-static inner class causes a problem with JRuby
|
87
|
+
this.self = self;
|
88
|
+
}
|
89
|
+
|
90
|
+
@Override
|
91
|
+
public boolean hasNext()
|
92
|
+
{
|
93
|
+
if (self.nextLine != null) {
|
94
|
+
return true;
|
95
|
+
} else {
|
96
|
+
self.nextLine = self.poll();
|
97
|
+
return self.nextLine != null;
|
98
|
+
}
|
99
|
+
}
|
100
|
+
|
101
|
+
@Override
|
102
|
+
public String next()
|
103
|
+
{
|
104
|
+
if (!hasNext()) {
|
105
|
+
throw new NoSuchElementException();
|
106
|
+
}
|
107
|
+
String l = self.nextLine;
|
108
|
+
self.nextLine = null;
|
109
|
+
return l;
|
110
|
+
}
|
111
|
+
|
112
|
+
@Override
|
113
|
+
public void remove()
|
114
|
+
{
|
115
|
+
throw new UnsupportedOperationException();
|
116
|
+
}
|
117
|
+
}
|
118
|
+
}
|
@@ -0,0 +1,109 @@
|
|
1
|
+
package org.embulk.spi.util;
|
2
|
+
|
3
|
+
import java.io.Writer;
|
4
|
+
import java.io.OutputStreamWriter;
|
5
|
+
import java.io.IOException;
|
6
|
+
import java.nio.charset.Charset;
|
7
|
+
import java.nio.charset.CharsetEncoder;
|
8
|
+
import java.nio.charset.CodingErrorAction;
|
9
|
+
import com.fasterxml.jackson.annotation.JacksonInject;
|
10
|
+
import org.embulk.config.Task;
|
11
|
+
import org.embulk.config.Config;
|
12
|
+
import org.embulk.config.ConfigDefault;
|
13
|
+
import org.embulk.spi.FileOutput;
|
14
|
+
import org.embulk.spi.BufferAllocator;
|
15
|
+
|
16
|
+
public class LineEncoder
|
17
|
+
implements AutoCloseable
|
18
|
+
{
|
19
|
+
// TODO optimize
|
20
|
+
|
21
|
+
public interface EncoderTask
|
22
|
+
extends Task
|
23
|
+
{
|
24
|
+
@Config("charset")
|
25
|
+
@ConfigDefault("\"utf-8\"")
|
26
|
+
public Charset getCharset();
|
27
|
+
|
28
|
+
@Config("newline")
|
29
|
+
@ConfigDefault("\"CRLF\"")
|
30
|
+
public Newline getNewline();
|
31
|
+
|
32
|
+
@JacksonInject
|
33
|
+
public BufferAllocator getBufferAllocator();
|
34
|
+
}
|
35
|
+
|
36
|
+
private final String newline;
|
37
|
+
private final FileOutputOutputStream outputStream;
|
38
|
+
private final Writer writer;
|
39
|
+
|
40
|
+
public LineEncoder(FileOutput out, EncoderTask task)
|
41
|
+
{
|
42
|
+
CharsetEncoder encoder = task.getCharset()
|
43
|
+
.newEncoder()
|
44
|
+
.onMalformedInput(CodingErrorAction.REPLACE) // TODO configurable?
|
45
|
+
.onUnmappableCharacter(CodingErrorAction.REPLACE); // TODO configurable?
|
46
|
+
this.newline = task.getNewline().getString();
|
47
|
+
this.outputStream = new FileOutputOutputStream(out, task.getBufferAllocator());
|
48
|
+
this.writer = new OutputStreamWriter(outputStream, encoder);
|
49
|
+
}
|
50
|
+
|
51
|
+
public void addNewLine()
|
52
|
+
{
|
53
|
+
try {
|
54
|
+
writer.append(newline);
|
55
|
+
} catch (IOException ex) {
|
56
|
+
// unexpected
|
57
|
+
throw new RuntimeException(ex);
|
58
|
+
}
|
59
|
+
}
|
60
|
+
|
61
|
+
public void addLine(String line)
|
62
|
+
{
|
63
|
+
try {
|
64
|
+
writer.append(line);
|
65
|
+
} catch (IOException ex) {
|
66
|
+
// unexpected
|
67
|
+
throw new RuntimeException(ex);
|
68
|
+
}
|
69
|
+
addNewLine();
|
70
|
+
}
|
71
|
+
|
72
|
+
public void addText(String text)
|
73
|
+
{
|
74
|
+
try {
|
75
|
+
writer.append(text);
|
76
|
+
} catch (IOException ex) {
|
77
|
+
// unexpected
|
78
|
+
throw new RuntimeException(ex);
|
79
|
+
}
|
80
|
+
}
|
81
|
+
|
82
|
+
public void nextFile()
|
83
|
+
{
|
84
|
+
try {
|
85
|
+
writer.flush();
|
86
|
+
} catch (IOException ex) {
|
87
|
+
// unexpected
|
88
|
+
throw new RuntimeException(ex);
|
89
|
+
}
|
90
|
+
outputStream.nextFile();
|
91
|
+
}
|
92
|
+
|
93
|
+
public void finish()
|
94
|
+
{
|
95
|
+
close(); // flush all remaining buffer in writer
|
96
|
+
outputStream.finish();
|
97
|
+
}
|
98
|
+
|
99
|
+
@Override
|
100
|
+
public void close()
|
101
|
+
{
|
102
|
+
try {
|
103
|
+
writer.close();
|
104
|
+
} catch (IOException ex) {
|
105
|
+
// unexpected
|
106
|
+
throw new RuntimeException(ex);
|
107
|
+
}
|
108
|
+
}
|
109
|
+
}
|
@@ -0,0 +1,52 @@
|
|
1
|
+
package org.embulk.spi.util;
|
2
|
+
|
3
|
+
import java.util.List;
|
4
|
+
import java.util.ArrayList;
|
5
|
+
import java.util.Iterator;
|
6
|
+
import org.embulk.spi.Buffer;
|
7
|
+
import org.embulk.spi.FileInput;
|
8
|
+
|
9
|
+
public class ListFileInput
|
10
|
+
implements FileInput
|
11
|
+
{
|
12
|
+
private Iterator<? extends Iterable<Buffer>> files;
|
13
|
+
private Iterator<Buffer> currentBuffers;
|
14
|
+
|
15
|
+
public ListFileInput(Iterable<? extends Iterable<Buffer>> files)
|
16
|
+
{
|
17
|
+
this.files = files.iterator();
|
18
|
+
}
|
19
|
+
|
20
|
+
public boolean nextFile()
|
21
|
+
{
|
22
|
+
if (!files.hasNext()) {
|
23
|
+
return false;
|
24
|
+
}
|
25
|
+
currentBuffers = files.next().iterator();
|
26
|
+
return true;
|
27
|
+
}
|
28
|
+
|
29
|
+
public Buffer poll()
|
30
|
+
{
|
31
|
+
if (currentBuffers == null) {
|
32
|
+
throw new IllegalStateException("FileInput.nextFile is not called");
|
33
|
+
}
|
34
|
+
if (!currentBuffers.hasNext()) {
|
35
|
+
return null;
|
36
|
+
}
|
37
|
+
return currentBuffers.next();
|
38
|
+
}
|
39
|
+
|
40
|
+
public void close()
|
41
|
+
{
|
42
|
+
do {
|
43
|
+
while (true) {
|
44
|
+
Buffer b = poll();
|
45
|
+
if (b == null) {
|
46
|
+
break;
|
47
|
+
}
|
48
|
+
b.release();
|
49
|
+
}
|
50
|
+
} while (nextFile());
|
51
|
+
}
|
52
|
+
}
|