embulk 0.1.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (204) hide show
  1. checksums.yaml +15 -0
  2. data/.gitignore +13 -0
  3. data/Gemfile +3 -0
  4. data/Gemfile.lock +33 -0
  5. data/README.md +117 -0
  6. data/Rakefile +58 -0
  7. data/bin/embulk +63 -0
  8. data/build.gradle +149 -0
  9. data/embulk-cli/build.gradle +6 -0
  10. data/embulk-cli/pom.xml +94 -0
  11. data/embulk-cli/src/main/java/org/embulk/cli/Main.java +15 -0
  12. data/embulk-core/build.gradle +6 -0
  13. data/embulk-core/pom.xml +143 -0
  14. data/embulk-core/src/main/java/org/embulk/EmbulkService.java +39 -0
  15. data/embulk-core/src/main/java/org/embulk/command/Runner.java +199 -0
  16. data/embulk-core/src/main/java/org/embulk/command/TablePrinter.java +119 -0
  17. data/embulk-core/src/main/java/org/embulk/config/CommitReport.java +26 -0
  18. data/embulk-core/src/main/java/org/embulk/config/Config.java +15 -0
  19. data/embulk-core/src/main/java/org/embulk/config/ConfigDefault.java +15 -0
  20. data/embulk-core/src/main/java/org/embulk/config/ConfigException.java +20 -0
  21. data/embulk-core/src/main/java/org/embulk/config/ConfigLoader.java +83 -0
  22. data/embulk-core/src/main/java/org/embulk/config/ConfigSource.java +28 -0
  23. data/embulk-core/src/main/java/org/embulk/config/DataSource.java +35 -0
  24. data/embulk-core/src/main/java/org/embulk/config/DataSourceImpl.java +208 -0
  25. data/embulk-core/src/main/java/org/embulk/config/DataSourceSerDe.java +80 -0
  26. data/embulk-core/src/main/java/org/embulk/config/GenericTypeReference.java +20 -0
  27. data/embulk-core/src/main/java/org/embulk/config/ModelManager.java +125 -0
  28. data/embulk-core/src/main/java/org/embulk/config/NextConfig.java +26 -0
  29. data/embulk-core/src/main/java/org/embulk/config/Task.java +10 -0
  30. data/embulk-core/src/main/java/org/embulk/config/TaskInvocationHandler.java +180 -0
  31. data/embulk-core/src/main/java/org/embulk/config/TaskSerDe.java +343 -0
  32. data/embulk-core/src/main/java/org/embulk/config/TaskSource.java +28 -0
  33. data/embulk-core/src/main/java/org/embulk/config/TaskValidationException.java +37 -0
  34. data/embulk-core/src/main/java/org/embulk/config/TaskValidator.java +24 -0
  35. data/embulk-core/src/main/java/org/embulk/exec/ExecModule.java +45 -0
  36. data/embulk-core/src/main/java/org/embulk/exec/ExecuteInterruptedException.java +10 -0
  37. data/embulk-core/src/main/java/org/embulk/exec/ExecuteResult.java +19 -0
  38. data/embulk-core/src/main/java/org/embulk/exec/ExtensionServiceLoaderModule.java +43 -0
  39. data/embulk-core/src/main/java/org/embulk/exec/ForSystemConfig.java +16 -0
  40. data/embulk-core/src/main/java/org/embulk/exec/GuessExecutor.java +307 -0
  41. data/embulk-core/src/main/java/org/embulk/exec/LocalExecutor.java +274 -0
  42. data/embulk-core/src/main/java/org/embulk/exec/LoggerProvider.java +30 -0
  43. data/embulk-core/src/main/java/org/embulk/exec/NoSampleException.java +10 -0
  44. data/embulk-core/src/main/java/org/embulk/exec/PooledBufferAllocator.java +58 -0
  45. data/embulk-core/src/main/java/org/embulk/exec/PreviewExecutor.java +138 -0
  46. data/embulk-core/src/main/java/org/embulk/exec/PreviewResult.java +27 -0
  47. data/embulk-core/src/main/java/org/embulk/exec/PreviewedNoticeError.java +17 -0
  48. data/embulk-core/src/main/java/org/embulk/exec/SamplingParserPlugin.java +116 -0
  49. data/embulk-core/src/main/java/org/embulk/exec/SystemConfigModule.java +24 -0
  50. data/embulk-core/src/main/java/org/embulk/jruby/JRubyPluginSource.java +69 -0
  51. data/embulk-core/src/main/java/org/embulk/jruby/JRubyScriptingModule.java +100 -0
  52. data/embulk-core/src/main/java/org/embulk/plugin/BuiltinPluginSourceModule.java +17 -0
  53. data/embulk-core/src/main/java/org/embulk/plugin/InjectedPluginSource.java +92 -0
  54. data/embulk-core/src/main/java/org/embulk/plugin/PluginManager.java +34 -0
  55. data/embulk-core/src/main/java/org/embulk/plugin/PluginSource.java +6 -0
  56. data/embulk-core/src/main/java/org/embulk/plugin/PluginSourceNotMatchException.java +19 -0
  57. data/embulk-core/src/main/java/org/embulk/plugin/PluginType.java +47 -0
  58. data/embulk-core/src/main/java/org/embulk/plugin/SetThreadContextClassLoader.java +19 -0
  59. data/embulk-core/src/main/java/org/embulk/spi/Buffer.java +113 -0
  60. data/embulk-core/src/main/java/org/embulk/spi/BufferAllocator.java +8 -0
  61. data/embulk-core/src/main/java/org/embulk/spi/Column.java +92 -0
  62. data/embulk-core/src/main/java/org/embulk/spi/ColumnConfig.java +79 -0
  63. data/embulk-core/src/main/java/org/embulk/spi/DecoderPlugin.java +16 -0
  64. data/embulk-core/src/main/java/org/embulk/spi/EncoderPlugin.java +16 -0
  65. data/embulk-core/src/main/java/org/embulk/spi/Exec.java +76 -0
  66. data/embulk-core/src/main/java/org/embulk/spi/ExecAction.java +6 -0
  67. data/embulk-core/src/main/java/org/embulk/spi/ExecSession.java +105 -0
  68. data/embulk-core/src/main/java/org/embulk/spi/Extension.java +42 -0
  69. data/embulk-core/src/main/java/org/embulk/spi/FileInput.java +11 -0
  70. data/embulk-core/src/main/java/org/embulk/spi/FileInputPlugin.java +19 -0
  71. data/embulk-core/src/main/java/org/embulk/spi/FileInputRunner.java +113 -0
  72. data/embulk-core/src/main/java/org/embulk/spi/FileOutput.java +13 -0
  73. data/embulk-core/src/main/java/org/embulk/spi/FileOutputPlugin.java +20 -0
  74. data/embulk-core/src/main/java/org/embulk/spi/FileOutputRunner.java +167 -0
  75. data/embulk-core/src/main/java/org/embulk/spi/FormatterPlugin.java +18 -0
  76. data/embulk-core/src/main/java/org/embulk/spi/GuessPlugin.java +9 -0
  77. data/embulk-core/src/main/java/org/embulk/spi/InputPlugin.java +20 -0
  78. data/embulk-core/src/main/java/org/embulk/spi/OutputPlugin.java +21 -0
  79. data/embulk-core/src/main/java/org/embulk/spi/Page.java +45 -0
  80. data/embulk-core/src/main/java/org/embulk/spi/PageBuilder.java +327 -0
  81. data/embulk-core/src/main/java/org/embulk/spi/PageFormat.java +47 -0
  82. data/embulk-core/src/main/java/org/embulk/spi/PageOutput.java +11 -0
  83. data/embulk-core/src/main/java/org/embulk/spi/PageReader.java +227 -0
  84. data/embulk-core/src/main/java/org/embulk/spi/ParserPlugin.java +17 -0
  85. data/embulk-core/src/main/java/org/embulk/spi/Schema.java +101 -0
  86. data/embulk-core/src/main/java/org/embulk/spi/SchemaConfig.java +52 -0
  87. data/embulk-core/src/main/java/org/embulk/spi/SchemaVisitor.java +14 -0
  88. data/embulk-core/src/main/java/org/embulk/spi/Transactional.java +10 -0
  89. data/embulk-core/src/main/java/org/embulk/spi/TransactionalFileInput.java +17 -0
  90. data/embulk-core/src/main/java/org/embulk/spi/TransactionalFileOutput.java +19 -0
  91. data/embulk-core/src/main/java/org/embulk/spi/TransactionalPageOutput.java +17 -0
  92. data/embulk-core/src/main/java/org/embulk/spi/time/DateTimeZoneSerDe.java +57 -0
  93. data/embulk-core/src/main/java/org/embulk/spi/time/JRubyTimeParserHelper.java +8 -0
  94. data/embulk-core/src/main/java/org/embulk/spi/time/JRubyTimeParserHelperFactory.java +6 -0
  95. data/embulk-core/src/main/java/org/embulk/spi/time/Timestamp.java +159 -0
  96. data/embulk-core/src/main/java/org/embulk/spi/time/TimestampFormat.java +98 -0
  97. data/embulk-core/src/main/java/org/embulk/spi/time/TimestampFormatter.java +55 -0
  98. data/embulk-core/src/main/java/org/embulk/spi/time/TimestampParseException.java +6 -0
  99. data/embulk-core/src/main/java/org/embulk/spi/time/TimestampParser.java +60 -0
  100. data/embulk-core/src/main/java/org/embulk/spi/time/TimestampSerDe.java +50 -0
  101. data/embulk-core/src/main/java/org/embulk/spi/type/AbstractType.java +55 -0
  102. data/embulk-core/src/main/java/org/embulk/spi/type/BooleanType.java +12 -0
  103. data/embulk-core/src/main/java/org/embulk/spi/type/DoubleType.java +12 -0
  104. data/embulk-core/src/main/java/org/embulk/spi/type/LongType.java +12 -0
  105. data/embulk-core/src/main/java/org/embulk/spi/type/StringType.java +12 -0
  106. data/embulk-core/src/main/java/org/embulk/spi/type/TimestampType.java +39 -0
  107. data/embulk-core/src/main/java/org/embulk/spi/type/Type.java +15 -0
  108. data/embulk-core/src/main/java/org/embulk/spi/type/TypeDeserializer.java +47 -0
  109. data/embulk-core/src/main/java/org/embulk/spi/type/Types.java +14 -0
  110. data/embulk-core/src/main/java/org/embulk/spi/util/CharsetSerDe.java +55 -0
  111. data/embulk-core/src/main/java/org/embulk/spi/util/Decoders.java +81 -0
  112. data/embulk-core/src/main/java/org/embulk/spi/util/Encoders.java +81 -0
  113. data/embulk-core/src/main/java/org/embulk/spi/util/FileInputInputStream.java +110 -0
  114. data/embulk-core/src/main/java/org/embulk/spi/util/FileOutputOutputStream.java +94 -0
  115. data/embulk-core/src/main/java/org/embulk/spi/util/InputStreamFileInput.java +111 -0
  116. data/embulk-core/src/main/java/org/embulk/spi/util/Inputs.java +74 -0
  117. data/embulk-core/src/main/java/org/embulk/spi/util/LineDecoder.java +118 -0
  118. data/embulk-core/src/main/java/org/embulk/spi/util/LineEncoder.java +109 -0
  119. data/embulk-core/src/main/java/org/embulk/spi/util/ListFileInput.java +52 -0
  120. data/embulk-core/src/main/java/org/embulk/spi/util/Newline.java +38 -0
  121. data/embulk-core/src/main/java/org/embulk/spi/util/PagePrinter.java +102 -0
  122. data/embulk-core/src/main/java/org/embulk/spi/util/Pages.java +139 -0
  123. data/embulk-core/src/test/java/org/embulk/EmbulkTestRuntime.java +110 -0
  124. data/embulk-core/src/test/java/org/embulk/GuiceBinder.java +72 -0
  125. data/embulk-core/src/test/java/org/embulk/RandomManager.java +53 -0
  126. data/embulk-core/src/test/java/org/embulk/TestPluginSourceModule.java +23 -0
  127. data/embulk-core/src/test/java/org/embulk/TestUtilityModule.java +17 -0
  128. data/embulk-core/src/test/java/org/embulk/config/TestConfigSource.java +114 -0
  129. data/embulk-core/src/test/java/org/embulk/config/TestTaskSource.java +70 -0
  130. data/embulk-core/src/test/java/org/embulk/plugin/MockPluginSource.java +57 -0
  131. data/embulk-core/src/test/java/org/embulk/plugin/TestPluginType.java +18 -0
  132. data/embulk-core/src/test/java/org/embulk/spi/MockFileOutput.java +63 -0
  133. data/embulk-core/src/test/java/org/embulk/spi/MockFormatterPlugin.java +101 -0
  134. data/embulk-core/src/test/java/org/embulk/spi/MockParserPlugin.java +73 -0
  135. data/embulk-core/src/test/java/org/embulk/spi/PageTestUtils.java +78 -0
  136. data/embulk-core/src/test/java/org/embulk/spi/TestFileInputInputStream.java +67 -0
  137. data/embulk-core/src/test/java/org/embulk/spi/TestFileInputRunner.java +180 -0
  138. data/embulk-core/src/test/java/org/embulk/spi/TestFileOutputRunner.java +192 -0
  139. data/embulk-core/src/test/java/org/embulk/spi/TestInputStreamFileInput.java +188 -0
  140. data/embulk-core/src/test/java/org/embulk/spi/TestPageBuilderReader.java +301 -0
  141. data/embulk-core/src/test/java/org/embulk/spi/time/TestTimestamp.java +116 -0
  142. data/embulk-core/src/test/java/org/embulk/spi/time/TestTimestampFormatterParser.java +52 -0
  143. data/embulk-core/src/test/java/org/embulk/spi/type/TestTypeSerDe.java +45 -0
  144. data/embulk-core/src/test/java/org/embulk/spi/util/TestLineDecoder.java +132 -0
  145. data/embulk-core/src/test/java/org/embulk/spi/util/TestLineEncoder.java +123 -0
  146. data/embulk-standards/build.gradle +6 -0
  147. data/embulk-standards/pom.xml +68 -0
  148. data/embulk-standards/src/main/java/org/embulk/standards/CsvFormatterPlugin.java +158 -0
  149. data/embulk-standards/src/main/java/org/embulk/standards/CsvParserPlugin.java +233 -0
  150. data/embulk-standards/src/main/java/org/embulk/standards/CsvTokenizer.java +355 -0
  151. data/embulk-standards/src/main/java/org/embulk/standards/GzipFileDecoderPlugin.java +55 -0
  152. data/embulk-standards/src/main/java/org/embulk/standards/GzipFileEncoderPlugin.java +39 -0
  153. data/embulk-standards/src/main/java/org/embulk/standards/LocalFileInputPlugin.java +138 -0
  154. data/embulk-standards/src/main/java/org/embulk/standards/LocalFileOutputPlugin.java +128 -0
  155. data/embulk-standards/src/main/java/org/embulk/standards/NullOutputPlugin.java +46 -0
  156. data/embulk-standards/src/main/java/org/embulk/standards/S3FileInputPlugin.java +238 -0
  157. data/embulk-standards/src/main/java/org/embulk/standards/StandardPluginExtension.java +16 -0
  158. data/embulk-standards/src/main/java/org/embulk/standards/StandardPluginModule.java +44 -0
  159. data/embulk-standards/src/main/java/org/embulk/standards/StdoutOutputPlugin.java +71 -0
  160. data/embulk-standards/src/main/resources/META-INF/services/org.embulk.spi.Extension +1 -0
  161. data/embulk-standards/src/test/java/org/embulk/standards/TestCsvParserPlugin.java +69 -0
  162. data/embulk-standards/src/test/java/org/embulk/standards/TestCsvTokenizer.java +291 -0
  163. data/embulk-standards/src/test/java/org/embulk/standards/TestS3FileInputPlugin.java +43 -0
  164. data/embulk.gemspec +27 -0
  165. data/examples/config.yml +34 -0
  166. data/examples/csv/sample.csv.gz +0 -0
  167. data/gradle/wrapper/gradle-wrapper.jar +0 -0
  168. data/gradle/wrapper/gradle-wrapper.properties +6 -0
  169. data/gradlew +164 -0
  170. data/gradlew.bat +90 -0
  171. data/lib/embulk.rb +16 -0
  172. data/lib/embulk/buffer.rb +17 -0
  173. data/lib/embulk/column.rb +47 -0
  174. data/lib/embulk/command/embulk.rb +39 -0
  175. data/lib/embulk/command/embulk_example.rb +32 -0
  176. data/lib/embulk/command/embulk_generate_bin.rb +62 -0
  177. data/lib/embulk/command/embulk_run.rb +243 -0
  178. data/lib/embulk/data/bundle/.bundle/config +3 -0
  179. data/lib/embulk/data/bundle/Gemfile +31 -0
  180. data/lib/embulk/data/bundle/Gemfile.lock +8 -0
  181. data/lib/embulk/data/bundle/embulk/input_example.rb +40 -0
  182. data/lib/embulk/data/bundle/embulk/output_example.rb +51 -0
  183. data/lib/embulk/data_source.rb +66 -0
  184. data/lib/embulk/error.rb +5 -0
  185. data/lib/embulk/guess_charset.rb +26 -0
  186. data/lib/embulk/guess_csv.rb +195 -0
  187. data/lib/embulk/guess_gzip.rb +18 -0
  188. data/lib/embulk/guess_newline.rb +20 -0
  189. data/lib/embulk/guess_plugin.rb +113 -0
  190. data/lib/embulk/input_plugin.rb +53 -0
  191. data/lib/embulk/java/bootstrap.rb +12 -0
  192. data/lib/embulk/java/imports.rb +26 -0
  193. data/lib/embulk/java/time_helper.rb +77 -0
  194. data/lib/embulk/output_plugin.rb +104 -0
  195. data/lib/embulk/page.rb +28 -0
  196. data/lib/embulk/page_builder.rb +22 -0
  197. data/lib/embulk/plugin.rb +152 -0
  198. data/lib/embulk/plugin_registry.rb +70 -0
  199. data/lib/embulk/schema.rb +85 -0
  200. data/lib/embulk/time_format_guess.rb +331 -0
  201. data/lib/embulk/version.rb +3 -0
  202. data/pom.xml +533 -0
  203. data/settings.gradle +5 -0
  204. metadata +370 -0
@@ -0,0 +1,98 @@
1
+ package org.embulk.spi.time;
2
+
3
+ import java.util.Set;
4
+ import java.util.Date;
5
+ import java.text.SimpleDateFormat;
6
+ import java.text.ParsePosition;
7
+ import com.google.common.collect.ImmutableSet;
8
+ import com.fasterxml.jackson.annotation.JsonValue;
9
+ import com.fasterxml.jackson.annotation.JsonCreator;
10
+ import org.joda.time.DateTimeZone;
11
+ import org.joda.time.format.DateTimeFormat;
12
+
13
+ public class TimestampFormat
14
+ {
15
+ private final String format;
16
+
17
+ @JsonCreator
18
+ public TimestampFormat(String format)
19
+ {
20
+ this.format = format;
21
+ }
22
+
23
+ @JsonValue
24
+ public String getFormat()
25
+ {
26
+ return format;
27
+ }
28
+
29
+ public TimestampFormatter newFormatter(TimestampFormatter.FormatterTask task)
30
+ {
31
+ return new TimestampFormatter(format, task);
32
+ }
33
+
34
+ public TimestampParser newParser(TimestampParser.ParserTask task)
35
+ {
36
+ return new TimestampParser(format, task);
37
+ }
38
+
39
+ private static Set<String> availableTimeZoneNames = ImmutableSet.copyOf(DateTimeZone.getAvailableIDs());
40
+
41
+ public static DateTimeZone parseDateTimeZone(String s)
42
+ {
43
+ if(s.startsWith("+") || s.startsWith("-")) {
44
+ return DateTimeZone.forID("GMT"+s);
45
+
46
+ } else {
47
+ try {
48
+ int rawOffset = (int) DateTimeFormat.forPattern("z").parseMillis(s);
49
+ if(rawOffset == 0) {
50
+ return DateTimeZone.UTC;
51
+ }
52
+ int offset = rawOffset / -1000;
53
+ int h = offset / 3600;
54
+ int m = offset % 3600;
55
+ return DateTimeZone.forOffsetHoursMinutes(h, m);
56
+ } catch (IllegalArgumentException ex) {
57
+ // parseMillis failed
58
+ }
59
+
60
+ // TimeZone.getTimeZone returns GMT zone if given timezone id is not found
61
+ // we want to only return timezone if exact match, otherwise exception
62
+ if (availableTimeZoneNames.contains(s)) {
63
+ //return TimeZone.getTimeZone(s);
64
+ return DateTimeZone.forID(s);
65
+ }
66
+ return null;
67
+ }
68
+ }
69
+
70
+ //// Java standard TimeZone
71
+ //static TimeZone parseDateTimeZone(String s)
72
+ //{
73
+ // if(s.startsWith("+") || s.startsWith("-")) {
74
+ // return TimeZone.getTimeZone("GMT"+s);
75
+ //
76
+ // } else {
77
+ // ParsePosition pp = new ParsePosition(0);
78
+ // Date off = new SimpleDateFormat("z").parse(s, pp);
79
+ // if(off != null && pp.getErrorIndex() == -1) {
80
+ // int rawOffset = (int) off.getTime();
81
+ // if(rawOffset == 0) {
82
+ // return TimeZone.UTC;
83
+ // }
84
+ // int offset = rawOffset / -1000;
85
+ // int h = offset / 3600;
86
+ // int m = offset % 3600;
87
+ // return DateTimeZone.getTimeZone(String.format("GMT%+02d%02d", h, m));
88
+ // }
89
+ //
90
+ // // TimeZone.getTimeZone returns GMT zone if given timezone id is not found
91
+ // // we want to only return timezone if exact match, otherwise exception
92
+ // if (availableTimeZoneNames.contains(s)) {
93
+ // return TimeZone.getTimeZone(s);
94
+ // }
95
+ // return null;
96
+ // }
97
+ //}
98
+ }
@@ -0,0 +1,55 @@
1
+ package org.embulk.spi.time;
2
+
3
+ import java.util.Locale;
4
+ import org.joda.time.DateTime;
5
+ import org.joda.time.DateTimeZone;
6
+ import org.jruby.embed.ScriptingContainer;
7
+ import org.jruby.util.RubyDateFormat;
8
+ import com.fasterxml.jackson.annotation.JacksonInject;
9
+ import org.embulk.config.Task;
10
+ import org.embulk.config.Config;
11
+ import org.embulk.config.ConfigDefault;
12
+ import org.embulk.config.ConfigException;
13
+ import org.embulk.spi.util.LineEncoder;
14
+
15
+ public class TimestampFormatter
16
+ {
17
+ public interface FormatterTask
18
+ extends Task
19
+ {
20
+ @Config("timezone")
21
+ @ConfigDefault("\"UTC\"")
22
+ public DateTimeZone getTimeZone();
23
+
24
+ @JacksonInject
25
+ public ScriptingContainer getJRuby();
26
+ }
27
+
28
+ private final RubyDateFormat dateFormat;
29
+ private final DateTimeZone timeZone;
30
+
31
+ public TimestampFormatter(String format, FormatterTask task)
32
+ {
33
+ this(task.getJRuby(), format, task.getTimeZone());
34
+ }
35
+
36
+ public TimestampFormatter(ScriptingContainer jruby, String format, DateTimeZone timeZone)
37
+ {
38
+ this.timeZone = timeZone;
39
+ this.dateFormat = new RubyDateFormat(format, Locale.ENGLISH, true);
40
+ }
41
+
42
+ public void format(Timestamp value, LineEncoder encoder)
43
+ {
44
+ // TODO optimize by directly appending to internal buffer
45
+ encoder.addText(format(value));
46
+ }
47
+
48
+ public String format(Timestamp value)
49
+ {
50
+ // TODO optimize by using reused StringBuilder
51
+ dateFormat.setDateTime(new DateTime(value.toEpochMilli(), timeZone));
52
+ dateFormat.setNSec(value.getNano());
53
+ return dateFormat.format(null);
54
+ }
55
+ }
@@ -0,0 +1,6 @@
1
+ package org.embulk.spi.time;
2
+
3
+ public class TimestampParseException
4
+ extends Exception
5
+ {
6
+ }
@@ -0,0 +1,60 @@
1
+ package org.embulk.spi.time;
2
+
3
+ import org.joda.time.DateTimeZone;
4
+ import org.jruby.embed.ScriptingContainer;
5
+ import com.fasterxml.jackson.annotation.JacksonInject;
6
+ import org.embulk.config.Task;
7
+ import org.embulk.config.Config;
8
+ import org.embulk.config.ConfigDefault;
9
+ import org.embulk.config.ConfigException;
10
+ import static org.embulk.spi.time.TimestampFormat.parseDateTimeZone;
11
+
12
+ public class TimestampParser
13
+ {
14
+ public interface ParserTask
15
+ extends Task
16
+ {
17
+ @Config("default_timezone")
18
+ @ConfigDefault("\"UTC\"")
19
+ public DateTimeZone getDefaultTimeZone();
20
+
21
+ @JacksonInject
22
+ public ScriptingContainer getJRuby();
23
+ }
24
+
25
+ private final JRubyTimeParserHelper helper;
26
+ private final DateTimeZone defaultTimeZone;
27
+
28
+ public TimestampParser(String format, ParserTask task)
29
+ {
30
+ this(task.getJRuby(), format, task.getDefaultTimeZone());
31
+ }
32
+
33
+ // TODO this is still private because this might need current time
34
+ private TimestampParser(ScriptingContainer jruby, String format, DateTimeZone defaultTimeZone)
35
+ {
36
+ JRubyTimeParserHelperFactory helperFactory = (JRubyTimeParserHelperFactory) jruby.runScriptlet("Embulk::Java::TimeParserHelper::Factory.new");
37
+ // TODO get default current time from ExecTask.getExecTimestamp
38
+ this.helper = (JRubyTimeParserHelper) helperFactory.newInstance(format, 1970, 1, 1, 0, 0, 0, 0); // TODO default time zone
39
+ this.defaultTimeZone = defaultTimeZone;
40
+ }
41
+
42
+ public Timestamp parse(String text) throws TimestampParseException
43
+ {
44
+ long localMillis = helper.strptime(text);
45
+ String zone = helper.getZone();
46
+
47
+ DateTimeZone timeZone = defaultTimeZone;
48
+ if (zone != null) {
49
+ // TODO cache parsed zone?
50
+ timeZone = parseDateTimeZone(zone);
51
+ if (timeZone == null) {
52
+ throw new TimestampParseException();
53
+ }
54
+ }
55
+
56
+ long milli = timeZone.convertLocalToUTC(localMillis, false);
57
+
58
+ return Timestamp.ofEpochMilli(milli);
59
+ }
60
+ }
@@ -0,0 +1,50 @@
1
+ package org.embulk.spi.time;
2
+
3
+ import java.io.IOException;
4
+ import com.fasterxml.jackson.core.JsonGenerator;
5
+ import com.fasterxml.jackson.databind.Module;
6
+ import com.fasterxml.jackson.databind.module.SimpleModule;
7
+ import com.fasterxml.jackson.databind.SerializerProvider;
8
+ import com.fasterxml.jackson.databind.DeserializationContext;
9
+ import com.fasterxml.jackson.databind.JsonSerializer;
10
+ import com.fasterxml.jackson.databind.JsonMappingException;
11
+ import com.fasterxml.jackson.databind.deser.std.FromStringDeserializer;
12
+ import com.fasterxml.jackson.module.guice.ObjectMapperModule;
13
+
14
+ public class TimestampSerDe
15
+ {
16
+ public static void configure(ObjectMapperModule mapper)
17
+ {
18
+ SimpleModule module = new SimpleModule();
19
+ module.addSerializer(Timestamp.class, new DateTimeZoneSerializer());
20
+ module.addDeserializer(Timestamp.class, new DateTimeZoneDeserializer());
21
+ mapper.registerModule(module);
22
+ }
23
+
24
+ public static class DateTimeZoneSerializer
25
+ extends JsonSerializer<Timestamp>
26
+ {
27
+ @Override
28
+ public void serialize(Timestamp value, JsonGenerator jgen, SerializerProvider provider)
29
+ throws IOException
30
+ {
31
+ jgen.writeString(value.toString());
32
+ }
33
+ }
34
+
35
+ public static class DateTimeZoneDeserializer
36
+ extends FromStringDeserializer<Timestamp>
37
+ {
38
+ public DateTimeZoneDeserializer()
39
+ {
40
+ super(Timestamp.class);
41
+ }
42
+
43
+ @Override
44
+ protected Timestamp _deserialize(String value, DeserializationContext context)
45
+ throws JsonMappingException
46
+ {
47
+ return Timestamp.fromString(value);
48
+ }
49
+ }
50
+ }
@@ -0,0 +1,55 @@
1
+ package org.embulk.spi.type;
2
+
3
+ public abstract class AbstractType
4
+ implements Type
5
+ {
6
+ private final String name;
7
+ private final Class<?> javaType;
8
+ private byte fixedStorageSize;
9
+
10
+ protected AbstractType(String name, Class<?> javaType, int fixedStorageSize)
11
+ {
12
+ this.name = name;
13
+ this.javaType = javaType;
14
+ this.fixedStorageSize = (byte) fixedStorageSize;
15
+ }
16
+
17
+ @Override
18
+ public String getName()
19
+ {
20
+ return name;
21
+ }
22
+
23
+ @Override
24
+ public Class<?> getJavaType()
25
+ {
26
+ return javaType;
27
+ }
28
+
29
+ @Override
30
+ public byte getFixedStorageSize()
31
+ {
32
+ return fixedStorageSize;
33
+ }
34
+
35
+ @Override
36
+ public boolean equals(Object o)
37
+ {
38
+ if (o == null) {
39
+ return false;
40
+ }
41
+ return o.getClass().isAssignableFrom(getClass());
42
+ }
43
+
44
+ @Override
45
+ public int hashCode()
46
+ {
47
+ return getClass().hashCode();
48
+ }
49
+
50
+ @Override
51
+ public String toString()
52
+ {
53
+ return name;
54
+ }
55
+ }
@@ -0,0 +1,12 @@
1
+ package org.embulk.spi.type;
2
+
3
+ public class BooleanType
4
+ extends AbstractType
5
+ {
6
+ static final BooleanType BOOLEAN = new BooleanType();
7
+
8
+ private BooleanType()
9
+ {
10
+ super("boolean", boolean.class, 1);
11
+ }
12
+ }
@@ -0,0 +1,12 @@
1
+ package org.embulk.spi.type;
2
+
3
+ public class DoubleType
4
+ extends AbstractType
5
+ {
6
+ static final DoubleType DOUBLE = new DoubleType();
7
+
8
+ private DoubleType()
9
+ {
10
+ super("double", double.class, 8);
11
+ }
12
+ }
@@ -0,0 +1,12 @@
1
+ package org.embulk.spi.type;
2
+
3
+ public class LongType
4
+ extends AbstractType
5
+ {
6
+ static final LongType LONG = new LongType();
7
+
8
+ private LongType()
9
+ {
10
+ super("long", long.class, 8);
11
+ }
12
+ }
@@ -0,0 +1,12 @@
1
+ package org.embulk.spi.type;
2
+
3
+ public class StringType
4
+ extends AbstractType
5
+ {
6
+ static final StringType STRING = new StringType();
7
+
8
+ private StringType()
9
+ {
10
+ super("string", String.class, 4);
11
+ }
12
+ }
@@ -0,0 +1,39 @@
1
+ package org.embulk.spi.type;
2
+
3
+ import org.embulk.spi.time.Timestamp;
4
+
5
+ public class TimestampType
6
+ extends AbstractType
7
+ {
8
+ static final TimestampType TIMESTAMP = new TimestampType();
9
+
10
+ private static final String DEFAULT_FORMAT = "%Y-%m-%d %H-%M-%S,%N %z";
11
+
12
+ private final String format;
13
+
14
+ private TimestampType()
15
+ {
16
+ this(null);
17
+ }
18
+
19
+ private TimestampType(String format)
20
+ {
21
+ super("timestamp", Timestamp.class, 12); // long msec + int nsec
22
+ this.format = format;
23
+ }
24
+
25
+ public TimestampType withFormat(String format)
26
+ {
27
+ // TODO is this correct design...?
28
+ return new TimestampType(format);
29
+ }
30
+
31
+ public String getFormat()
32
+ {
33
+ if (format == null) {
34
+ return DEFAULT_FORMAT;
35
+ } else {
36
+ return format;
37
+ }
38
+ }
39
+ }
@@ -0,0 +1,15 @@
1
+ package org.embulk.spi.type;
2
+
3
+ import com.fasterxml.jackson.annotation.JsonValue;
4
+ import com.fasterxml.jackson.databind.annotation.JsonDeserialize;
5
+
6
+ @JsonDeserialize(using=TypeDeserializer.class)
7
+ public interface Type
8
+ {
9
+ @JsonValue
10
+ public String getName();
11
+
12
+ public Class<?> getJavaType();
13
+
14
+ public byte getFixedStorageSize();
15
+ }