embulk-output-td 0.1.4 → 0.1.5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.travis.yml +7 -0
- data/CHANGELOG.md +4 -0
- data/README.md +1 -0
- data/build.gradle +5 -1
- data/config/checkstyle/checkstyle.xml +117 -0
- data/embulk-output-td.gemspec +1 -1
- data/gradle/check.gradle +34 -0
- data/src/main/java/com/treasuredata/api/TdApiClient.java +47 -23
- data/src/main/java/com/treasuredata/api/TdApiClientConfig.java +3 -3
- data/src/main/java/com/treasuredata/api/TdApiConstants.java +6 -2
- data/src/main/java/com/treasuredata/api/TdApiExecutionInterruptedException.java +2 -1
- data/src/main/java/com/treasuredata/api/TdApiExecutionTimeoutException.java +2 -1
- data/src/main/java/com/treasuredata/api/model/TDArrayColumnType.java +1 -1
- data/src/main/java/com/treasuredata/api/model/TDBulkImportSession.java +6 -4
- data/src/main/java/com/treasuredata/api/model/TDColumn.java +4 -2
- data/src/main/java/com/treasuredata/api/model/TDColumnTypeDeserializer.java +26 -13
- data/src/main/java/com/treasuredata/api/model/TDDatabase.java +2 -1
- data/src/main/java/com/treasuredata/api/model/TDMapColumnType.java +1 -1
- data/src/main/java/com/treasuredata/api/model/TDTablePermission.java +4 -2
- data/src/main/java/com/treasuredata/api/model/TDTableType.java +2 -1
- data/src/main/java/org/embulk/output/td/FinalizableExecutorService.java +35 -17
- data/src/main/java/org/embulk/output/td/MsgpackGZFileBuilder.java +13 -7
- data/src/main/java/org/embulk/output/td/RecordWriter.java +21 -382
- data/src/main/java/org/embulk/output/td/TdOutputPlugin.java +175 -40
- data/src/main/java/org/embulk/output/td/writer/BooleanFieldWriter.java +23 -0
- data/src/main/java/org/embulk/output/td/writer/DoubleFieldWriter.java +23 -0
- data/src/main/java/org/embulk/output/td/writer/FieldWriter.java +38 -0
- data/src/main/java/org/embulk/output/td/writer/FieldWriterSet.java +206 -0
- data/src/main/java/org/embulk/output/td/writer/LongFieldWriter.java +23 -0
- data/src/main/java/org/embulk/output/td/writer/StringFieldWriter.java +23 -0
- data/src/main/java/org/embulk/output/td/writer/TimestampFieldLongDuplicator.java +28 -0
- data/src/main/java/org/embulk/output/td/writer/TimestampLongFieldWriter.java +23 -0
- data/src/main/java/org/embulk/output/td/writer/TimestampStringFieldWriter.java +27 -0
- data/src/main/java/org/embulk/output/td/writer/UnixTimestampFieldDuplicator.java +27 -0
- data/src/main/java/org/embulk/output/td/writer/UnixTimestampLongFieldWriter.java +26 -0
- data/src/test/java/com/treasuredata/api/TestTdApiClient.java +1 -1
- data/src/test/java/org/embulk/output/td/TestRecordWriter.java +198 -0
- data/src/test/java/org/embulk/output/td/TestTdOutputPlugin.java +529 -0
- data/src/test/java/org/embulk/output/td/writer/TestFieldWriterSet.java +146 -0
- metadata +29 -14
- data/src/test/java/org/embulk/output/td/TestFieldWriter.java +0 -105
@@ -143,13 +143,15 @@ public class TDBulkImportSession
|
|
143
143
|
|
144
144
|
public String getErrorMessage()
|
145
145
|
{
|
146
|
-
if (validRecords == 0)
|
146
|
+
if (validRecords == 0) {
|
147
147
|
return "No record processed";
|
148
|
-
|
148
|
+
}
|
149
|
+
if (errorRecords > 0) {
|
149
150
|
return String.format("%d invalid parts", errorParts);
|
150
|
-
|
151
|
+
}
|
152
|
+
if (errorRecords > 0) {
|
151
153
|
return String.format("%d invalid records", errorRecords);
|
152
|
-
|
154
|
+
}
|
153
155
|
return null;
|
154
156
|
}
|
155
157
|
}
|
@@ -43,13 +43,15 @@ public class TDColumn
|
|
43
43
|
TDColumnTypeDeserializer.parseColumnType(tuple[1]),
|
44
44
|
tuple[0].getBytes());
|
45
45
|
|
46
|
-
}
|
46
|
+
}
|
47
|
+
else if (tuple != null && tuple.length == 3) {
|
47
48
|
return new TDColumn(
|
48
49
|
tuple[0],
|
49
50
|
TDColumnTypeDeserializer.parseColumnType(tuple[1]),
|
50
51
|
tuple[2].getBytes());
|
51
52
|
|
52
|
-
}
|
53
|
+
}
|
54
|
+
else {
|
53
55
|
throw new RuntimeJsonMappingException("Unexpected string tuple to deserialize TDColumn");
|
54
56
|
}
|
55
57
|
}
|
@@ -39,19 +39,24 @@ public class TDColumnTypeDeserializer
|
|
39
39
|
if (p.scan("string")) {
|
40
40
|
return TDPrimitiveColumnType.STRING;
|
41
41
|
|
42
|
-
}
|
42
|
+
}
|
43
|
+
else if (p.scan("int")) {
|
43
44
|
return TDPrimitiveColumnType.INT;
|
44
45
|
|
45
|
-
}
|
46
|
+
}
|
47
|
+
else if (p.scan("long")) {
|
46
48
|
return TDPrimitiveColumnType.LONG;
|
47
49
|
|
48
|
-
}
|
50
|
+
}
|
51
|
+
else if (p.scan("double")) {
|
49
52
|
return TDPrimitiveColumnType.DOUBLE;
|
50
53
|
|
51
|
-
}
|
54
|
+
}
|
55
|
+
else if (p.scan("float")) {
|
52
56
|
return TDPrimitiveColumnType.FLOAT;
|
53
57
|
|
54
|
-
}
|
58
|
+
}
|
59
|
+
else if (p.scan("array")) {
|
55
60
|
if (!p.scan("<")) {
|
56
61
|
throw new IllegalArgumentException("Cannot parse type: expected '<' for array type: " + p.getString());
|
57
62
|
}
|
@@ -61,7 +66,8 @@ public class TDColumnTypeDeserializer
|
|
61
66
|
}
|
62
67
|
return new TDArrayColumnType(elementType);
|
63
68
|
|
64
|
-
}
|
69
|
+
}
|
70
|
+
else if (p.scan("map")) {
|
65
71
|
if (!p.scan("<")) {
|
66
72
|
throw new IllegalArgumentException("Cannot parse type: expected '<' for map type: " + p.getString());
|
67
73
|
}
|
@@ -75,24 +81,29 @@ public class TDColumnTypeDeserializer
|
|
75
81
|
}
|
76
82
|
return new TDMapColumnType(keyType, valueType);
|
77
83
|
|
78
|
-
}
|
84
|
+
}
|
85
|
+
else {
|
79
86
|
throw new IllegalArgumentException("Cannot parse type: " + p.getString());
|
80
87
|
}
|
81
88
|
}
|
82
89
|
|
83
|
-
private static class Parser
|
90
|
+
private static class Parser
|
91
|
+
{
|
84
92
|
private final String string;
|
85
93
|
private int offset;
|
86
94
|
|
87
|
-
public Parser(String string)
|
95
|
+
public Parser(String string)
|
96
|
+
{
|
88
97
|
this.string = string;
|
89
98
|
}
|
90
99
|
|
91
|
-
public String getString()
|
100
|
+
public String getString()
|
101
|
+
{
|
92
102
|
return string;
|
93
103
|
}
|
94
104
|
|
95
|
-
public boolean scan(String s)
|
105
|
+
public boolean scan(String s)
|
106
|
+
{
|
96
107
|
skipSpaces();
|
97
108
|
if (string.startsWith(s, offset)) {
|
98
109
|
offset += s.length();
|
@@ -101,12 +112,14 @@ public class TDColumnTypeDeserializer
|
|
101
112
|
return false;
|
102
113
|
}
|
103
114
|
|
104
|
-
public boolean eof()
|
115
|
+
public boolean eof()
|
116
|
+
{
|
105
117
|
skipSpaces();
|
106
118
|
return string.length() <= offset;
|
107
119
|
}
|
108
120
|
|
109
|
-
private void skipSpaces()
|
121
|
+
private void skipSpaces()
|
122
|
+
{
|
110
123
|
while (string.startsWith(" ", offset)) {
|
111
124
|
offset++;
|
112
125
|
}
|
@@ -17,12 +17,14 @@ public class TDTablePermission
|
|
17
17
|
}
|
18
18
|
|
19
19
|
@JsonProperty("importable")
|
20
|
-
public boolean isImportable()
|
20
|
+
public boolean isImportable()
|
21
|
+
{
|
21
22
|
return importable;
|
22
23
|
}
|
23
24
|
|
24
25
|
@JsonProperty("queryable")
|
25
|
-
public boolean isQueryable()
|
26
|
+
public boolean isQueryable()
|
27
|
+
{
|
26
28
|
return queryable;
|
27
29
|
}
|
28
30
|
|
@@ -21,7 +21,8 @@ public enum TDTableType
|
|
21
21
|
{
|
22
22
|
if ("log".equals(name)) {
|
23
23
|
return LOG;
|
24
|
-
}
|
24
|
+
}
|
25
|
+
else if ("item".equals(name)) {
|
25
26
|
return ITEM;
|
26
27
|
}
|
27
28
|
throw new RuntimeJsonMappingException("Unexpected string tuple to deserialize TDTableType");
|
@@ -17,7 +17,8 @@ public class FinalizableExecutorService
|
|
17
17
|
{
|
18
18
|
@Override
|
19
19
|
public void close()
|
20
|
-
throws IOException
|
20
|
+
throws IOException
|
21
|
+
{
|
21
22
|
// ignore
|
22
23
|
}
|
23
24
|
}
|
@@ -25,58 +26,75 @@ public class FinalizableExecutorService
|
|
25
26
|
protected ExecutorService threads;
|
26
27
|
protected Queue<RunningTask> runningTasks;
|
27
28
|
|
28
|
-
public FinalizableExecutorService()
|
29
|
+
public FinalizableExecutorService()
|
30
|
+
{
|
29
31
|
this.threads = Executors.newCachedThreadPool();
|
30
32
|
this.runningTasks = new LinkedList<>();
|
31
33
|
}
|
32
34
|
|
33
|
-
private static class RunningTask
|
35
|
+
private static class RunningTask
|
36
|
+
{
|
34
37
|
private Future<Void> future;
|
35
38
|
private Closeable finalizer;
|
36
39
|
|
37
|
-
RunningTask(Future<Void> future, Closeable finalizer)
|
40
|
+
RunningTask(Future<Void> future, Closeable finalizer)
|
41
|
+
{
|
38
42
|
this.future = future;
|
39
43
|
this.finalizer = finalizer;
|
40
44
|
}
|
41
45
|
|
42
|
-
public void join()
|
46
|
+
public void join()
|
47
|
+
throws IOException
|
48
|
+
{
|
43
49
|
try {
|
44
50
|
future.get();
|
45
|
-
}
|
46
|
-
|
47
|
-
|
48
|
-
|
51
|
+
}
|
52
|
+
catch (InterruptedException e) {
|
53
|
+
throw new IOException(e);
|
54
|
+
}
|
55
|
+
catch (ExecutionException e) {
|
56
|
+
throw new IOException(e.getCause());
|
49
57
|
}
|
50
58
|
finalizer.close();
|
51
59
|
}
|
52
60
|
|
53
|
-
public void abort()
|
61
|
+
public void abort()
|
62
|
+
throws IOException
|
63
|
+
{
|
54
64
|
finalizer.close();
|
55
65
|
}
|
56
66
|
}
|
57
67
|
|
58
|
-
public void submit(Callable<Void> task, Closeable finalizer)
|
68
|
+
public void submit(Callable<Void> task, Closeable finalizer)
|
69
|
+
{
|
59
70
|
Future<Void> future = threads.submit(task);
|
60
71
|
runningTasks.add(new RunningTask(future, finalizer));
|
61
72
|
}
|
62
73
|
|
63
|
-
public void joinPartial(long upto)
|
64
|
-
|
74
|
+
public void joinPartial(long upto)
|
75
|
+
throws IOException
|
76
|
+
{
|
77
|
+
while (runningTasks.size() > upto) {
|
65
78
|
runningTasks.peek().join();
|
66
79
|
runningTasks.remove();
|
67
80
|
}
|
68
81
|
}
|
69
82
|
|
70
|
-
public void joinAll()
|
83
|
+
public void joinAll()
|
84
|
+
throws IOException
|
85
|
+
{
|
71
86
|
joinPartial(0);
|
72
87
|
}
|
73
88
|
|
74
|
-
public void shutdown()
|
89
|
+
public void shutdown()
|
90
|
+
throws IOException
|
91
|
+
{
|
75
92
|
try {
|
76
93
|
joinAll();
|
77
|
-
}
|
94
|
+
}
|
95
|
+
finally {
|
78
96
|
threads.shutdown();
|
79
|
-
for(RunningTask task : runningTasks) {
|
97
|
+
for (RunningTask task : runningTasks) {
|
80
98
|
task.abort();
|
81
99
|
}
|
82
100
|
}
|
@@ -1,6 +1,5 @@
|
|
1
1
|
package org.embulk.output.td;
|
2
2
|
|
3
|
-
import org.embulk.spi.Exec;
|
4
3
|
import org.msgpack.MessagePack;
|
5
4
|
import org.msgpack.packer.Packer;
|
6
5
|
|
@@ -18,29 +17,35 @@ import static com.google.common.base.Preconditions.checkNotNull;
|
|
18
17
|
public class MsgpackGZFileBuilder
|
19
18
|
implements Closeable
|
20
19
|
{
|
21
|
-
static class DataSizeFilter
|
20
|
+
static class DataSizeFilter
|
21
|
+
extends FilterOutputStream
|
22
|
+
{
|
22
23
|
private long size = 0;
|
23
24
|
|
24
|
-
public DataSizeFilter(OutputStream out)
|
25
|
+
public DataSizeFilter(OutputStream out)
|
26
|
+
{
|
25
27
|
super(out);
|
26
28
|
}
|
27
29
|
|
28
30
|
@Override
|
29
|
-
public void write(int b)
|
31
|
+
public void write(int b)
|
32
|
+
throws IOException
|
30
33
|
{
|
31
34
|
size += 1;
|
32
35
|
super.write(b);
|
33
36
|
}
|
34
37
|
|
35
38
|
@Override
|
36
|
-
public void write(byte[] b, int off, int len)
|
39
|
+
public void write(byte[] b, int off, int len)
|
40
|
+
throws IOException
|
37
41
|
{
|
38
42
|
size += len;
|
39
43
|
super.write(b, off, len);
|
40
44
|
}
|
41
45
|
|
42
46
|
@Override
|
43
|
-
public void close()
|
47
|
+
public void close()
|
48
|
+
throws IOException
|
44
49
|
{
|
45
50
|
super.close();
|
46
51
|
}
|
@@ -94,7 +99,8 @@ public class MsgpackGZFileBuilder
|
|
94
99
|
{
|
95
100
|
try {
|
96
101
|
packer.flush();
|
97
|
-
}
|
102
|
+
}
|
103
|
+
finally {
|
98
104
|
close();
|
99
105
|
}
|
100
106
|
}
|
@@ -1,11 +1,12 @@
|
|
1
1
|
package org.embulk.output.td;
|
2
2
|
|
3
|
-
import com.google.common.
|
3
|
+
import com.google.common.annotations.VisibleForTesting;
|
4
4
|
import com.google.common.base.Stopwatch;
|
5
5
|
import com.google.common.base.Throwables;
|
6
6
|
import com.treasuredata.api.TdApiClient;
|
7
7
|
import org.embulk.config.CommitReport;
|
8
|
-
import org.embulk.
|
8
|
+
import org.embulk.output.td.writer.FieldWriter;
|
9
|
+
import org.embulk.output.td.writer.FieldWriterSet;
|
9
10
|
import org.embulk.spi.Column;
|
10
11
|
import org.embulk.spi.ColumnVisitor;
|
11
12
|
import org.embulk.spi.Exec;
|
@@ -13,16 +14,6 @@ import org.embulk.spi.Page;
|
|
13
14
|
import org.embulk.spi.PageReader;
|
14
15
|
import org.embulk.spi.Schema;
|
15
16
|
import org.embulk.spi.TransactionalPageOutput;
|
16
|
-
import org.embulk.spi.time.TimestampFormatter;
|
17
|
-
import org.embulk.spi.type.BooleanType;
|
18
|
-
import org.embulk.spi.type.DoubleType;
|
19
|
-
import org.embulk.spi.type.LongType;
|
20
|
-
import org.embulk.spi.type.StringType;
|
21
|
-
import org.embulk.spi.type.TimestampType;
|
22
|
-
import org.embulk.spi.type.Type;
|
23
|
-
import org.embulk.spi.util.Timestamps;
|
24
|
-
import org.joda.time.DateTimeZone;
|
25
|
-
import org.jruby.embed.ScriptingContainer;
|
26
17
|
import org.msgpack.MessagePack;
|
27
18
|
import org.slf4j.Logger;
|
28
19
|
|
@@ -91,6 +82,12 @@ public class RecordWriter
|
|
91
82
|
this.builder = new MsgpackGZFileBuilder(msgpack, tempFile);
|
92
83
|
}
|
93
84
|
|
85
|
+
@VisibleForTesting
|
86
|
+
MsgpackGZFileBuilder getBuilder()
|
87
|
+
{
|
88
|
+
return builder;
|
89
|
+
}
|
90
|
+
|
94
91
|
@Override
|
95
92
|
public void add(final Page page)
|
96
93
|
{
|
@@ -136,7 +133,8 @@ public class RecordWriter
|
|
136
133
|
FieldWriter fieldWriter = fieldWriters.getFieldWriter(column.getIndex());
|
137
134
|
try {
|
138
135
|
fieldWriter.writeKeyValue(builder, pageReader, column);
|
139
|
-
}
|
136
|
+
}
|
137
|
+
catch (IOException e) {
|
140
138
|
throw Throwables.propagate(e);
|
141
139
|
}
|
142
140
|
}
|
@@ -150,7 +148,8 @@ public class RecordWriter
|
|
150
148
|
}
|
151
149
|
}
|
152
150
|
|
153
|
-
}
|
151
|
+
}
|
152
|
+
catch (IOException e) {
|
154
153
|
throw Throwables.propagate(e);
|
155
154
|
}
|
156
155
|
}
|
@@ -206,9 +205,11 @@ public class RecordWriter
|
|
206
205
|
{
|
207
206
|
try {
|
208
207
|
flush();
|
209
|
-
}
|
208
|
+
}
|
209
|
+
catch (IOException e) {
|
210
210
|
throw Throwables.propagate(e);
|
211
|
-
}
|
211
|
+
}
|
212
|
+
finally {
|
212
213
|
close();
|
213
214
|
}
|
214
215
|
}
|
@@ -220,7 +221,8 @@ public class RecordWriter
|
|
220
221
|
try {
|
221
222
|
executor.joinAll();
|
222
223
|
executor.shutdown(); // shutdown calls joinAll
|
223
|
-
}
|
224
|
+
}
|
225
|
+
finally {
|
224
226
|
if (builder != null) {
|
225
227
|
builder.close();
|
226
228
|
builder.delete();
|
@@ -231,7 +233,8 @@ public class RecordWriter
|
|
231
233
|
client.close();
|
232
234
|
}
|
233
235
|
}
|
234
|
-
}
|
236
|
+
}
|
237
|
+
catch (IOException e) {
|
235
238
|
throw Throwables.propagate(e);
|
236
239
|
}
|
237
240
|
}
|
@@ -249,368 +252,4 @@ public class RecordWriter
|
|
249
252
|
// TODO
|
250
253
|
return report;
|
251
254
|
}
|
252
|
-
|
253
|
-
static class FieldWriterSet
|
254
|
-
{
|
255
|
-
private enum ColumnWriterMode
|
256
|
-
{
|
257
|
-
PRIMARY_KEY,
|
258
|
-
SIMPLE_VALUE,
|
259
|
-
DUPLICATE_PRIMARY_KEY;
|
260
|
-
}
|
261
|
-
|
262
|
-
private final int fieldCount;
|
263
|
-
private final FieldWriter[] fieldWriters;
|
264
|
-
|
265
|
-
public FieldWriterSet(Logger log, TdOutputPlugin.PluginTask task, Schema schema)
|
266
|
-
{
|
267
|
-
Optional<String> userDefinedPrimaryKeySourceColumnName = task.getTimeColumn();
|
268
|
-
boolean hasPkWriter = false;
|
269
|
-
int duplicatePrimaryKeySourceIndex = -1;
|
270
|
-
int firstTimestampColumnIndex = -1;
|
271
|
-
|
272
|
-
int fc = 0;
|
273
|
-
fieldWriters = new FieldWriter[schema.size()];
|
274
|
-
TimestampFormatter[] timestampFormatters = Timestamps.newTimestampColumnFormatters(task, schema, task.getColumnOptions());
|
275
|
-
|
276
|
-
for (int i = 0; i < schema.size(); i++) {
|
277
|
-
String columnName = schema.getColumnName(i);
|
278
|
-
Type columnType = schema.getColumnType(i);
|
279
|
-
|
280
|
-
// choose the mode
|
281
|
-
final ColumnWriterMode mode;
|
282
|
-
|
283
|
-
if (userDefinedPrimaryKeySourceColumnName.isPresent() &&
|
284
|
-
columnName.equals(userDefinedPrimaryKeySourceColumnName.get())) {
|
285
|
-
// found time_column
|
286
|
-
if ("time".equals(userDefinedPrimaryKeySourceColumnName.get())) {
|
287
|
-
mode = ColumnWriterMode.PRIMARY_KEY;
|
288
|
-
} else {
|
289
|
-
mode = ColumnWriterMode.DUPLICATE_PRIMARY_KEY;
|
290
|
-
}
|
291
|
-
|
292
|
-
} else if ("time".equals(columnName)) {
|
293
|
-
// the column name is same with the primary key name.
|
294
|
-
if (userDefinedPrimaryKeySourceColumnName.isPresent()) {
|
295
|
-
columnName = newColumnUniqueName(columnName, schema);
|
296
|
-
mode = ColumnWriterMode.SIMPLE_VALUE;
|
297
|
-
log.warn("time_column '{}' is set but 'time' column also exists. The existent 'time' column is renamed to {}",
|
298
|
-
userDefinedPrimaryKeySourceColumnName.get(), "time", "time", columnName);
|
299
|
-
} else {
|
300
|
-
mode = ColumnWriterMode.PRIMARY_KEY;
|
301
|
-
}
|
302
|
-
|
303
|
-
} else {
|
304
|
-
mode = ColumnWriterMode.SIMPLE_VALUE;
|
305
|
-
}
|
306
|
-
|
307
|
-
// create the fieldWriters writer depending on the mode
|
308
|
-
final FieldWriter writer;
|
309
|
-
|
310
|
-
switch (mode) {
|
311
|
-
case PRIMARY_KEY:
|
312
|
-
log.info("Using {}:{} column as the data partitioning key", columnName, columnType);
|
313
|
-
if (columnType instanceof LongType) {
|
314
|
-
if (task.getUnixTimestampUnit() != TdOutputPlugin.UnixTimestampUnit.SEC) {
|
315
|
-
log.warn("time column is converted from {} to seconds", task.getUnixTimestampUnit());
|
316
|
-
}
|
317
|
-
writer = new UnixTimestampLongFieldWriter(columnName, task.getUnixTimestampUnit().getFractionUnit());
|
318
|
-
hasPkWriter = true;
|
319
|
-
} else if (columnType instanceof TimestampType) {
|
320
|
-
writer = new TimestampLongFieldWriter(columnName);
|
321
|
-
|
322
|
-
hasPkWriter = true;
|
323
|
-
} else {
|
324
|
-
throw new ConfigException(String.format("Type of '%s' column must be long or timestamp but got %s",
|
325
|
-
columnName, columnType));
|
326
|
-
}
|
327
|
-
break;
|
328
|
-
|
329
|
-
case SIMPLE_VALUE:
|
330
|
-
if (columnType instanceof BooleanType) {
|
331
|
-
writer = new BooleanFieldWriter(columnName);
|
332
|
-
} else if (columnType instanceof LongType) {
|
333
|
-
writer = new LongFieldWriter(columnName);
|
334
|
-
} else if (columnType instanceof DoubleType) {
|
335
|
-
writer = new DoubleFieldWriter(columnName);
|
336
|
-
} else if (columnType instanceof StringType) {
|
337
|
-
writer = new StringFieldWriter(columnName);
|
338
|
-
} else if (columnType instanceof TimestampType) {
|
339
|
-
writer = new TimestampStringFieldWriter(timestampFormatters[i], columnName);
|
340
|
-
if (firstTimestampColumnIndex < 0) {
|
341
|
-
firstTimestampColumnIndex = i;
|
342
|
-
}
|
343
|
-
} else {
|
344
|
-
throw new ConfigException("Unsupported type: " + columnType);
|
345
|
-
}
|
346
|
-
break;
|
347
|
-
|
348
|
-
case DUPLICATE_PRIMARY_KEY:
|
349
|
-
duplicatePrimaryKeySourceIndex = i;
|
350
|
-
writer = null; // handle later
|
351
|
-
break;
|
352
|
-
|
353
|
-
default:
|
354
|
-
throw new AssertionError();
|
355
|
-
}
|
356
|
-
|
357
|
-
fieldWriters[i] = writer;
|
358
|
-
fc += 1;
|
359
|
-
}
|
360
|
-
|
361
|
-
if (!hasPkWriter) {
|
362
|
-
// PRIMARY_KEY was not found.
|
363
|
-
if (duplicatePrimaryKeySourceIndex < 0) {
|
364
|
-
if (userDefinedPrimaryKeySourceColumnName.isPresent()) {
|
365
|
-
throw new ConfigException(String.format("time_column '%s' does not exist", userDefinedPrimaryKeySourceColumnName.get()));
|
366
|
-
} else if (firstTimestampColumnIndex >= 0) {
|
367
|
-
// if time is not found, use the first timestamp column
|
368
|
-
duplicatePrimaryKeySourceIndex = firstTimestampColumnIndex;
|
369
|
-
} else {
|
370
|
-
throw new ConfigException(String.format("TD output plugin requires at least one timestamp column, or a long column named 'time'"));
|
371
|
-
}
|
372
|
-
}
|
373
|
-
|
374
|
-
String columnName = schema.getColumnName(duplicatePrimaryKeySourceIndex);
|
375
|
-
Type columnType = schema.getColumnType(duplicatePrimaryKeySourceIndex);
|
376
|
-
|
377
|
-
FieldWriter writer;
|
378
|
-
if (columnType instanceof LongType) {
|
379
|
-
log.info("Duplicating {}:{} column (unix timestamp {}) to 'time' column as seconds for the data partitioning",
|
380
|
-
columnName, columnType, task.getUnixTimestampUnit());
|
381
|
-
writer = new UnixTimestampFieldDuplicator(columnName, "time", task.getUnixTimestampUnit().getFractionUnit());
|
382
|
-
} else if (columnType instanceof TimestampType) {
|
383
|
-
log.info("Duplicating {}:{} column to 'time' column as seconds for the data partitioning",
|
384
|
-
columnName, columnType);
|
385
|
-
writer = new TimestampFieldLongDuplicator(timestampFormatters[duplicatePrimaryKeySourceIndex], columnName, "time");
|
386
|
-
} else {
|
387
|
-
throw new ConfigException(String.format("Type of '%s' column must be long or timestamp but got %s",
|
388
|
-
columnName, columnType));
|
389
|
-
}
|
390
|
-
|
391
|
-
// replace existint writer
|
392
|
-
fieldWriters[duplicatePrimaryKeySourceIndex] = writer;
|
393
|
-
fc += 1;
|
394
|
-
}
|
395
|
-
|
396
|
-
fieldCount = fc;
|
397
|
-
}
|
398
|
-
|
399
|
-
private static String newColumnUniqueName(String originalName, Schema schema)
|
400
|
-
{
|
401
|
-
String name = originalName;
|
402
|
-
do {
|
403
|
-
name += "_";
|
404
|
-
} while (containsColumnName(schema, name));
|
405
|
-
return name;
|
406
|
-
}
|
407
|
-
|
408
|
-
private static boolean containsColumnName(Schema schema, String name)
|
409
|
-
{
|
410
|
-
for (Column c : schema.getColumns()) {
|
411
|
-
if (c.getName().equals(name)) {
|
412
|
-
return true;
|
413
|
-
}
|
414
|
-
}
|
415
|
-
return false;
|
416
|
-
}
|
417
|
-
|
418
|
-
public FieldWriter getFieldWriter(int index)
|
419
|
-
{
|
420
|
-
return fieldWriters[index];
|
421
|
-
}
|
422
|
-
|
423
|
-
public int getFieldCount()
|
424
|
-
{
|
425
|
-
return fieldCount;
|
426
|
-
}
|
427
|
-
}
|
428
|
-
|
429
|
-
static abstract class FieldWriter
|
430
|
-
{
|
431
|
-
private final String keyName;
|
432
|
-
|
433
|
-
protected FieldWriter(String keyName)
|
434
|
-
{
|
435
|
-
this.keyName = keyName;
|
436
|
-
}
|
437
|
-
|
438
|
-
public void writeKeyValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
439
|
-
throws IOException
|
440
|
-
{
|
441
|
-
writeKey(builder);
|
442
|
-
if (reader.isNull(column)) {
|
443
|
-
builder.writeNil();
|
444
|
-
} else {
|
445
|
-
writeValue(builder, reader, column);
|
446
|
-
}
|
447
|
-
}
|
448
|
-
|
449
|
-
private void writeKey(MsgpackGZFileBuilder builder)
|
450
|
-
throws IOException
|
451
|
-
{
|
452
|
-
builder.writeString(keyName);
|
453
|
-
}
|
454
|
-
|
455
|
-
protected abstract void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
456
|
-
throws IOException;
|
457
|
-
}
|
458
|
-
|
459
|
-
static class DoubleFieldWriter
|
460
|
-
extends FieldWriter
|
461
|
-
{
|
462
|
-
public DoubleFieldWriter(String keyName)
|
463
|
-
{
|
464
|
-
super(keyName);
|
465
|
-
}
|
466
|
-
|
467
|
-
@Override
|
468
|
-
public void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
469
|
-
throws IOException
|
470
|
-
{
|
471
|
-
builder.writeDouble(reader.getDouble(column));
|
472
|
-
}
|
473
|
-
}
|
474
|
-
|
475
|
-
static class BooleanFieldWriter
|
476
|
-
extends FieldWriter
|
477
|
-
{
|
478
|
-
public BooleanFieldWriter(String keyName)
|
479
|
-
{
|
480
|
-
super(keyName);
|
481
|
-
}
|
482
|
-
|
483
|
-
@Override
|
484
|
-
public void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
485
|
-
throws IOException
|
486
|
-
{
|
487
|
-
builder.writeBoolean(reader.getBoolean(column));
|
488
|
-
}
|
489
|
-
}
|
490
|
-
|
491
|
-
static class LongFieldWriter
|
492
|
-
extends FieldWriter
|
493
|
-
{
|
494
|
-
LongFieldWriter(String keyName)
|
495
|
-
{
|
496
|
-
super(keyName);
|
497
|
-
}
|
498
|
-
|
499
|
-
@Override
|
500
|
-
public void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
501
|
-
throws IOException
|
502
|
-
{
|
503
|
-
builder.writeLong(reader.getLong(column));
|
504
|
-
}
|
505
|
-
}
|
506
|
-
|
507
|
-
static class UnixTimestampLongFieldWriter
|
508
|
-
extends FieldWriter
|
509
|
-
{
|
510
|
-
private final int fractionUnit;
|
511
|
-
|
512
|
-
UnixTimestampLongFieldWriter(String keyName, int fractionUnit)
|
513
|
-
{
|
514
|
-
super(keyName);
|
515
|
-
this.fractionUnit = fractionUnit;
|
516
|
-
}
|
517
|
-
|
518
|
-
@Override
|
519
|
-
public void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
520
|
-
throws IOException
|
521
|
-
{
|
522
|
-
builder.writeLong(reader.getLong(column) / fractionUnit);
|
523
|
-
}
|
524
|
-
}
|
525
|
-
|
526
|
-
static class StringFieldWriter
|
527
|
-
extends FieldWriter
|
528
|
-
{
|
529
|
-
public StringFieldWriter(String keyName)
|
530
|
-
{
|
531
|
-
super(keyName);
|
532
|
-
}
|
533
|
-
|
534
|
-
@Override
|
535
|
-
public void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
536
|
-
throws IOException
|
537
|
-
{
|
538
|
-
builder.writeString(reader.getString(column));
|
539
|
-
}
|
540
|
-
}
|
541
|
-
|
542
|
-
static class TimestampStringFieldWriter
|
543
|
-
extends FieldWriter
|
544
|
-
{
|
545
|
-
private final TimestampFormatter formatter;
|
546
|
-
|
547
|
-
public TimestampStringFieldWriter(TimestampFormatter formatter, String keyName)
|
548
|
-
{
|
549
|
-
super(keyName);
|
550
|
-
this.formatter = formatter;
|
551
|
-
}
|
552
|
-
|
553
|
-
@Override
|
554
|
-
public void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
555
|
-
throws IOException
|
556
|
-
{
|
557
|
-
builder.writeString(formatter.format(reader.getTimestamp(column)));
|
558
|
-
}
|
559
|
-
}
|
560
|
-
|
561
|
-
static class TimestampLongFieldWriter
|
562
|
-
extends FieldWriter
|
563
|
-
{
|
564
|
-
public TimestampLongFieldWriter(String keyName)
|
565
|
-
{
|
566
|
-
super(keyName);
|
567
|
-
}
|
568
|
-
|
569
|
-
@Override
|
570
|
-
public void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
571
|
-
throws IOException
|
572
|
-
{
|
573
|
-
builder.writeLong(reader.getTimestamp(column).getEpochSecond());
|
574
|
-
}
|
575
|
-
}
|
576
|
-
|
577
|
-
static class UnixTimestampFieldDuplicator
|
578
|
-
extends LongFieldWriter
|
579
|
-
{
|
580
|
-
private final UnixTimestampLongFieldWriter timeFieldWriter;
|
581
|
-
|
582
|
-
public UnixTimestampFieldDuplicator(String keyName, String duplicateKeyName, int fractionUnit)
|
583
|
-
{
|
584
|
-
super(keyName);
|
585
|
-
timeFieldWriter = new UnixTimestampLongFieldWriter(duplicateKeyName, fractionUnit);
|
586
|
-
}
|
587
|
-
|
588
|
-
@Override
|
589
|
-
public void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
590
|
-
throws IOException
|
591
|
-
{
|
592
|
-
super.writeValue(builder, reader, column);
|
593
|
-
timeFieldWriter.writeKeyValue(builder, reader, column);
|
594
|
-
}
|
595
|
-
}
|
596
|
-
|
597
|
-
static class TimestampFieldLongDuplicator
|
598
|
-
extends TimestampStringFieldWriter
|
599
|
-
{
|
600
|
-
private final TimestampLongFieldWriter timeFieldWriter;
|
601
|
-
|
602
|
-
public TimestampFieldLongDuplicator(TimestampFormatter formatter, String keyName, String longDuplicateKeyName)
|
603
|
-
{
|
604
|
-
super(formatter, keyName);
|
605
|
-
timeFieldWriter = new TimestampLongFieldWriter(longDuplicateKeyName);
|
606
|
-
}
|
607
|
-
|
608
|
-
@Override
|
609
|
-
public void writeValue(MsgpackGZFileBuilder builder, PageReader reader, Column column)
|
610
|
-
throws IOException
|
611
|
-
{
|
612
|
-
super.writeValue(builder, reader, column);
|
613
|
-
timeFieldWriter.writeKeyValue(builder, reader, column);
|
614
|
-
}
|
615
|
-
}
|
616
255
|
}
|