Repository: incubator-hivemall Updated Branches: refs/heads/master ae0d3c410 -> c40366959
http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/core/src/test/java/hivemall/topicmodel/PLSAUDTFTest.java ---------------------------------------------------------------------- diff --git a/core/src/test/java/hivemall/topicmodel/PLSAUDTFTest.java b/core/src/test/java/hivemall/topicmodel/PLSAUDTFTest.java index 87b2f4c..7f344d1 100644 --- a/core/src/test/java/hivemall/topicmodel/PLSAUDTFTest.java +++ b/core/src/test/java/hivemall/topicmodel/PLSAUDTFTest.java @@ -41,7 +41,8 @@ public class PLSAUDTFTest { PLSAUDTF udtf = new PLSAUDTF(); ObjectInspector[] argOIs = new ObjectInspector[] { - ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector), + ObjectInspectorFactory.getStandardListObjectInspector( + PrimitiveObjectInspectorFactory.javaStringObjectInspector), ObjectInspectorUtils.getConstantObjectInspector( PrimitiveObjectInspectorFactory.javaStringObjectInspector, "-topics 2 -alpha 0.1 -delta 0.00001 -iter 10000")}; @@ -49,8 +50,8 @@ public class PLSAUDTFTest { udtf.initialize(argOIs); String[] doc1 = new String[] {"fruits:1", "healthy:1", "vegetables:1"}; - String[] doc2 = new String[] {"apples:1", "avocados:1", "colds:1", "flu:1", "like:2", - "oranges:1"}; + String[] doc2 = + new String[] {"apples:1", "avocados:1", "colds:1", "flu:1", "like:2", "oranges:1"}; udtf.process(new Object[] {Arrays.asList(doc1)}); udtf.process(new Object[] {Arrays.asList(doc2)}); @@ -92,8 +93,9 @@ public class PLSAUDTFTest { k2 = 0; } - Assert.assertTrue("doc1 is in topic " + k1 + " (" + (topicDistr[k1] * 100) + "%), " - + "and `vegetables` SHOULD be more suitable topic word than `flu` in the topic", + Assert.assertTrue( + "doc1 is in topic " + k1 + " (" + (topicDistr[k1] * 100) + "%), " + + "and `vegetables` SHOULD be more suitable topic word than `flu` in the topic", udtf.getWordScore("vegetables", k1) > udtf.getWordScore("flu", k1)); Assert.assertTrue("doc2 is in topic " + k2 + " (" + (topicDistr[k2] * 100) + "%), " + "and `avocados` SHOULD be more suitable topic word than `healthy` in the topic", @@ -105,7 +107,8 @@ public class PLSAUDTFTest { PLSAUDTF udtf = new PLSAUDTF(); ObjectInspector[] argOIs = new ObjectInspector[] { - ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector), + ObjectInspectorFactory.getStandardListObjectInspector( + PrimitiveObjectInspectorFactory.javaStringObjectInspector), ObjectInspectorUtils.getConstantObjectInspector( PrimitiveObjectInspectorFactory.javaStringObjectInspector, "-topics 2 -alpha 0.1 -delta 0.00001 -iter 10000 -mini_batch_size 1")}; @@ -155,25 +158,26 @@ public class PLSAUDTFTest { k2 = 0; } - Assert.assertTrue("doc1 is in topic " + k1 + " (" + (topicDistr[k1] * 100) + "%), " - + "and `éè` SHOULD be more suitable topic word than `ã¤ã³ãã«ã¨ã³ã¶` in the topic", + Assert.assertTrue( + "doc1 is in topic " + k1 + " (" + (topicDistr[k1] * 100) + "%), " + + "and `éè` SHOULD be more suitable topic word than `ã¤ã³ãã«ã¨ã³ã¶` in the topic", udtf.getWordScore("éè", k1) > udtf.getWordScore("ã¤ã³ãã«ã¨ã³ã¶", k1)); - Assert.assertTrue("doc2 is in topic " + k2 + " (" + (topicDistr[k2] * 100) + "%), " - + "and `ã¢ãã«ã` SHOULD be more suitable topic word than `å¥åº·` in the topic", + Assert.assertTrue( + "doc2 is in topic " + k2 + " (" + (topicDistr[k2] * 100) + "%), " + + "and `ã¢ãã«ã` SHOULD be more suitable topic word than `å¥åº·` in the topic", udtf.getWordScore("ã¢ãã«ã", k2) > udtf.getWordScore("å¥åº·", k2)); } @Test public void testSerialization() throws HiveException { - TestUtils.testGenericUDTFSerialization( - PLSAUDTF.class, + TestUtils.testGenericUDTFSerialization(PLSAUDTF.class, new ObjectInspector[] { - ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector), + ObjectInspectorFactory.getStandardListObjectInspector( + PrimitiveObjectInspectorFactory.javaStringObjectInspector), ObjectInspectorUtils.getConstantObjectInspector( PrimitiveObjectInspectorFactory.javaStringObjectInspector, "-topics 2 -alpha 0.1 -delta 0.00001 -iter 10000")}, - new Object[][] { - {Arrays.asList("fruits:1", "healthy:1", "vegetables:1")}, + new Object[][] {{Arrays.asList("fruits:1", "healthy:1", "vegetables:1")}, {Arrays.asList("apples:1", "avocados:1", "colds:1", "flu:1", "like:2", "oranges:1")}}); } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/core/src/test/java/hivemall/utils/codec/Base91Test.java ---------------------------------------------------------------------- diff --git a/core/src/test/java/hivemall/utils/codec/Base91Test.java b/core/src/test/java/hivemall/utils/codec/Base91Test.java index 809ba97..6ab3ebf 100644 --- a/core/src/test/java/hivemall/utils/codec/Base91Test.java +++ b/core/src/test/java/hivemall/utils/codec/Base91Test.java @@ -44,8 +44,9 @@ public class Base91Test { @Test public void testLongEncodeDecode() throws IOException { - byte[] expected1 = IOUtils.toString( - Base91Test.class.getResourceAsStream("Base91Test.class")).getBytes(); + byte[] expected1 = + IOUtils.toString(Base91Test.class.getResourceAsStream("Base91Test.class")) + .getBytes(); Assert.assertTrue(expected1.length > 1000); byte[] actual1 = Base91.decode(Base91.encode(expected1)); Assert.assertArrayEquals(expected1, actual1); http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/core/src/test/java/hivemall/utils/codec/DeflateCodecTest.java ---------------------------------------------------------------------- diff --git a/core/src/test/java/hivemall/utils/codec/DeflateCodecTest.java b/core/src/test/java/hivemall/utils/codec/DeflateCodecTest.java index 091644e..48c93de 100644 --- a/core/src/test/java/hivemall/utils/codec/DeflateCodecTest.java +++ b/core/src/test/java/hivemall/utils/codec/DeflateCodecTest.java @@ -43,7 +43,8 @@ public class DeflateCodecTest { byte[] compressed1 = codec.compress(original1); byte[] decompressed1 = codec.decompress(compressed1); Assert.assertTrue("compressed size (" + compressed1.length + " bytes) < original size (" - + original1.length + " bytes)", compressed1.length < original1.length); + + original1.length + " bytes)", + compressed1.length < original1.length); Assert.assertArrayEquals(original1, decompressed1); codec.close(); } @@ -51,12 +52,15 @@ public class DeflateCodecTest { @Test public void testNonString() throws IOException { DeflateCodec codec = new DeflateCodec(); - byte[] original1 = IOUtils.toString( - DeflateCodecTest.class.getResourceAsStream("DeflateCodecTest.class")).getBytes(); + byte[] original1 = IOUtils + .toString(DeflateCodecTest.class.getResourceAsStream( + "DeflateCodecTest.class")) + .getBytes(); byte[] compressed1 = codec.compress(original1); byte[] decompressed1 = codec.decompress(compressed1); Assert.assertTrue("compressed size (" + compressed1.length + " bytes) < original size (" - + original1.length + " bytes)", compressed1.length < original1.length); + + original1.length + " bytes)", + compressed1.length < original1.length); Assert.assertArrayEquals(original1, decompressed1); codec.close(); } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/core/src/test/java/hivemall/utils/collections/BoundedPriorityQueueTest.java ---------------------------------------------------------------------- diff --git a/core/src/test/java/hivemall/utils/collections/BoundedPriorityQueueTest.java b/core/src/test/java/hivemall/utils/collections/BoundedPriorityQueueTest.java index 1220d76..b8aa559 100644 --- a/core/src/test/java/hivemall/utils/collections/BoundedPriorityQueueTest.java +++ b/core/src/test/java/hivemall/utils/collections/BoundedPriorityQueueTest.java @@ -31,13 +31,13 @@ public class BoundedPriorityQueueTest { @Test public void testTop3() { - BoundedPriorityQueue<Integer> queue = new BoundedPriorityQueue<Integer>(3, - new Comparator<Integer>() { - @Override - public int compare(Integer o1, Integer o2) { - return Integer.compare(o1, o2); - } - }); + BoundedPriorityQueue<Integer> queue = + new BoundedPriorityQueue<Integer>(3, new Comparator<Integer>() { + @Override + public int compare(Integer o1, Integer o2) { + return Integer.compare(o1, o2); + } + }); Assert.assertTrue(queue.offer(1)); Assert.assertTrue(queue.offer(4)); Assert.assertTrue(queue.offer(3)); @@ -58,8 +58,8 @@ public class BoundedPriorityQueueTest { @Test public void testTail3() { - BoundedPriorityQueue<Integer> queue = new BoundedPriorityQueue<Integer>(3, - Collections.<Integer>reverseOrder()); + BoundedPriorityQueue<Integer> queue = + new BoundedPriorityQueue<Integer>(3, Collections.<Integer>reverseOrder()); Assert.assertTrue(queue.offer(1)); Assert.assertTrue(queue.offer(4)); Assert.assertTrue(queue.offer(3)); @@ -80,13 +80,13 @@ public class BoundedPriorityQueueTest { @Test public void testString1() { - BoundedPriorityQueue<String> queue = new BoundedPriorityQueue<>(3, - new Comparator<String>() { - @Override - public int compare(String o1, String o2) { - return StringUtils.compare(o1, o2); - } - }); + BoundedPriorityQueue<String> queue = + new BoundedPriorityQueue<>(3, new Comparator<String>() { + @Override + public int compare(String o1, String o2) { + return StringUtils.compare(o1, o2); + } + }); queue.offer("B"); queue.offer("A"); queue.offer("C"); @@ -99,8 +99,8 @@ public class BoundedPriorityQueueTest { @Test public void testString2() { - BoundedPriorityQueue<String> queue = new BoundedPriorityQueue<>(3, - NaturalComparator.<String>getInstance()); + BoundedPriorityQueue<String> queue = + new BoundedPriorityQueue<>(3, NaturalComparator.<String>getInstance()); queue.offer("B"); queue.offer("A"); queue.offer("C"); http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/core/src/test/java/hivemall/utils/geospatial/GeoSpatialUtilsTest.java ---------------------------------------------------------------------- diff --git a/core/src/test/java/hivemall/utils/geospatial/GeoSpatialUtilsTest.java b/core/src/test/java/hivemall/utils/geospatial/GeoSpatialUtilsTest.java index 6f551be..4600e2a 100644 --- a/core/src/test/java/hivemall/utils/geospatial/GeoSpatialUtilsTest.java +++ b/core/src/test/java/hivemall/utils/geospatial/GeoSpatialUtilsTest.java @@ -25,14 +25,13 @@ public class GeoSpatialUtilsTest { @Test public void testTile() { - double[] lat_array = new double[] {GeoSpatialUtils.MIN_LATITUDE, 0.d, - GeoSpatialUtils.MAX_LATITUDE}; + double[] lat_array = + new double[] {GeoSpatialUtils.MIN_LATITUDE, 0.d, GeoSpatialUtils.MAX_LATITUDE}; double[] lon_array = new double[] {-180.d, 0.d, 180.d}; for (double lat : lat_array) { for (double lon : lon_array) { - Assert.assertTrue( - String.format("lat=%s, lon=%s, tile=%s", lat, lon, - GeoSpatialUtils.tile(lat, lon, 4)), GeoSpatialUtils.tile(lat, lon, 4) >= 0); + Assert.assertTrue(String.format("lat=%s, lon=%s, tile=%s", lat, lon, + GeoSpatialUtils.tile(lat, lon, 4)), GeoSpatialUtils.tile(lat, lon, 4) >= 0); } } } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/core/src/test/java/hivemall/utils/hadoop/JsonSerdeUtilsTest.java ---------------------------------------------------------------------- diff --git a/core/src/test/java/hivemall/utils/hadoop/JsonSerdeUtilsTest.java b/core/src/test/java/hivemall/utils/hadoop/JsonSerdeUtilsTest.java index 556cc71..a3e81d2 100644 --- a/core/src/test/java/hivemall/utils/hadoop/JsonSerdeUtilsTest.java +++ b/core/src/test/java/hivemall/utils/hadoop/JsonSerdeUtilsTest.java @@ -80,18 +80,17 @@ public class JsonSerdeUtilsTest { @Test public void testLooseJsonReadability() throws Exception { List<String> columnNames = Arrays.asList("s,k".split(",")); - List<TypeInfo> columnTypes = TypeInfoUtils.getTypeInfosFromTypeString("struct<a:int,b:string>,int"); + List<TypeInfo> columnTypes = + TypeInfoUtils.getTypeInfosFromTypeString("struct<a:int,b:string>,int"); - Text jsonText1 = new Text( - "{ \"x\" : \"abc\" , " - + " \"t\" : { \"a\":\"1\", \"b\":\"2\", \"c\":[ { \"x\":2 , \"y\":3 } , { \"x\":3 , \"y\":2 }] } ," - + "\"s\" : { \"a\" : 2 , \"b\" : \"blah\", \"c\": \"woo\" } }"); + Text jsonText1 = new Text("{ \"x\" : \"abc\" , " + + " \"t\" : { \"a\":\"1\", \"b\":\"2\", \"c\":[ { \"x\":2 , \"y\":3 } , { \"x\":3 , \"y\":2 }] } ," + + "\"s\" : { \"a\" : 2 , \"b\" : \"blah\", \"c\": \"woo\" } }"); - Text jsonText2 = new Text( - "{ \"x\" : \"abc\" , " - + " \"t\" : { \"a\":\"1\", \"b\":\"2\", \"c\":[ { \"x\":2 , \"y\":3 } , { \"x\":3 , \"y\":2 }] } ," - + "\"s\" : { \"a\" : 2 , \"b\" : \"blah\", \"c\": \"woo\" } , " - + "\"k\" : 113 " + "}"); + Text jsonText2 = new Text("{ \"x\" : \"abc\" , " + + " \"t\" : { \"a\":\"1\", \"b\":\"2\", \"c\":[ { \"x\":2 , \"y\":3 } , { \"x\":3 , \"y\":2 }] } ," + + "\"s\" : { \"a\" : 2 , \"b\" : \"blah\", \"c\": \"woo\" } , " + "\"k\" : 113 " + + "}"); List<Object> expected1 = Arrays.<Object>asList(Arrays.asList(2, "blah"), null); List<Object> expected2 = Arrays.<Object>asList(Arrays.asList(2, "blah"), 113); @@ -105,7 +104,8 @@ public class JsonSerdeUtilsTest { @Test public void testMapValues() throws SerDeException { List<String> columnNames = Arrays.asList("a,b".split(",")); - List<TypeInfo> columnTypes = TypeInfoUtils.getTypeInfosFromTypeString("array<string>,map<string,int>"); + List<TypeInfo> columnTypes = + TypeInfoUtils.getTypeInfosFromTypeString("array<string>,map<string,int>"); Text text1 = new Text("{ \"a\":[\"aaa\"],\"b\":{\"bbb\":1}} "); Text text2 = new Text("{\"a\":[\"yyy\"],\"b\":{\"zzz\":123}}"); @@ -193,18 +193,22 @@ public class JsonSerdeUtilsTest { DefaultHCatRecord r = new DefaultHCatRecord(rlist); - List<String> columnNames = Arrays.asList("ti,si,i,bi,d,f,s,n,r,l,m,b,c1,bd,hc,hvc,dt,ts,bin".split(",")); - List<TypeInfo> columnTypes = TypeInfoUtils.getTypeInfosFromTypeString("tinyint,smallint,int,bigint,double,float,string,string," - + "struct<a:string,b:string>,array<int>,map<smallint,string>,boolean," - + "array<struct<i1:int,i2:struct<ii1:array<int>,ii2:map<string,struct<iii1:int>>>>>," - + "decimal(5,2),char(10),varchar(20),date,timestamp,binary"); + List<String> columnNames = + Arrays.asList("ti,si,i,bi,d,f,s,n,r,l,m,b,c1,bd,hc,hvc,dt,ts,bin".split(",")); + List<TypeInfo> columnTypes = TypeInfoUtils.getTypeInfosFromTypeString( + "tinyint,smallint,int,bigint,double,float,string,string," + + "struct<a:string,b:string>,array<int>,map<smallint,string>,boolean," + + "array<struct<i1:int,i2:struct<ii1:array<int>,ii2:map<string,struct<iii1:int>>>>>," + + "decimal(5,2),char(10),varchar(20),date,timestamp,binary"); - StructTypeInfo rowTypeInfo = (StructTypeInfo) TypeInfoFactory.getStructTypeInfo( - columnNames, columnTypes); - HCatRecordObjectInspector objInspector = HCatRecordObjectInspectorFactory.getHCatRecordObjectInspector(rowTypeInfo); + StructTypeInfo rowTypeInfo = + (StructTypeInfo) TypeInfoFactory.getStructTypeInfo(columnNames, columnTypes); + HCatRecordObjectInspector objInspector = + HCatRecordObjectInspectorFactory.getHCatRecordObjectInspector(rowTypeInfo); Text serialized = JsonSerdeUtils.serialize(r, objInspector, columnNames); - List<Object> deserialized = JsonSerdeUtils.deserialize(serialized, columnNames, columnTypes); + List<Object> deserialized = + JsonSerdeUtils.deserialize(serialized, columnNames, columnTypes); assertRecordEquals(rlist, deserialized); } @@ -236,18 +240,22 @@ public class JsonSerdeUtilsTest { DefaultHCatRecord r = new DefaultHCatRecord(nlist); - List<String> columnNames = Arrays.asList("ti,si,i,bi,d,f,s,n,r,l,m,b,c1,bd,hc,hvc,dt,ts,bin".split(",")); - List<TypeInfo> columnTypes = TypeInfoUtils.getTypeInfosFromTypeString("tinyint,smallint,int,bigint,double,float,string,string," - + "struct<a:string,b:string>,array<int>,map<smallint,string>,boolean," - + "array<struct<i1:int,i2:struct<ii1:array<int>,ii2:map<string,struct<iii1:int>>>>>," - + "decimal(5,2),char(10),varchar(20),date,timestamp,binary"); + List<String> columnNames = + Arrays.asList("ti,si,i,bi,d,f,s,n,r,l,m,b,c1,bd,hc,hvc,dt,ts,bin".split(",")); + List<TypeInfo> columnTypes = TypeInfoUtils.getTypeInfosFromTypeString( + "tinyint,smallint,int,bigint,double,float,string,string," + + "struct<a:string,b:string>,array<int>,map<smallint,string>,boolean," + + "array<struct<i1:int,i2:struct<ii1:array<int>,ii2:map<string,struct<iii1:int>>>>>," + + "decimal(5,2),char(10),varchar(20),date,timestamp,binary"); - StructTypeInfo rowTypeInfo = (StructTypeInfo) TypeInfoFactory.getStructTypeInfo( - columnNames, columnTypes); - HCatRecordObjectInspector objInspector = HCatRecordObjectInspectorFactory.getHCatRecordObjectInspector(rowTypeInfo); + StructTypeInfo rowTypeInfo = + (StructTypeInfo) TypeInfoFactory.getStructTypeInfo(columnNames, columnTypes); + HCatRecordObjectInspector objInspector = + HCatRecordObjectInspectorFactory.getHCatRecordObjectInspector(rowTypeInfo); Text serialized = JsonSerdeUtils.serialize(r, objInspector, columnNames); - List<Object> deserialized = JsonSerdeUtils.deserialize(serialized, columnNames, columnTypes); + List<Object> deserialized = + JsonSerdeUtils.deserialize(serialized, columnNames, columnTypes); assertRecordEquals(nlist, deserialized); } @@ -258,8 +266,8 @@ public class JsonSerdeUtilsTest { TypeInfo type1 = TypeInfoUtils.getTypeInfoFromTypeString("struct<name:string,age:int>"); List<Object> expected1 = Arrays.<Object>asList("makoto", 37); - List<Object> deserialized1 = JsonSerdeUtils.deserialize(json1, Arrays.asList("person"), - Arrays.asList(type1)); + List<Object> deserialized1 = + JsonSerdeUtils.deserialize(json1, Arrays.asList("person"), Arrays.asList(type1)); assertRecordEquals(expected1, deserialized1); } @@ -329,15 +337,15 @@ public class JsonSerdeUtilsTest { int mySz = first.size(); int urSz = second.size(); if (mySz != urSz) { - throw new RuntimeException("#expected != #actual. #expected=" + mySz + ", #actual=" - + urSz); + throw new RuntimeException( + "#expected != #actual. #expected=" + mySz + ", #actual=" + urSz); } else { for (int i = 0; i < first.size(); i++) { int c = DataType.compare(first.get(i), second.get(i)); if (c != 0) { String msg = "first.get(" + i + "}='" + first.get(i) + "' second.get(" + i - + ")='" + second.get(i) + "' compared as " + c + "\n" - + "Types 1st/2nd=" + DataType.findType(first.get(i)) + "/" + + ")='" + second.get(i) + "' compared as " + c + "\n" + "Types 1st/2nd=" + + DataType.findType(first.get(i)) + "/" + DataType.findType(second.get(i)) + '\n' + "first='" + first.get(i) + "' second='" + second.get(i) + "'"; if (first.get(i) instanceof Date) { http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/core/src/test/java/hivemall/utils/lang/HalfFloatTest.java ---------------------------------------------------------------------- diff --git a/core/src/test/java/hivemall/utils/lang/HalfFloatTest.java b/core/src/test/java/hivemall/utils/lang/HalfFloatTest.java index 42396c2..8b0c8c0 100644 --- a/core/src/test/java/hivemall/utils/lang/HalfFloatTest.java +++ b/core/src/test/java/hivemall/utils/lang/HalfFloatTest.java @@ -97,18 +97,18 @@ public class HalfFloatTest { 33, 1.966953E-6f, 83, 4.947186E-6f, 167, 9.953976E-6f, 335, 1.996756E-5f, 838, 4.994869E-5f, 1677, 9.995699E-05f, 2701, 0.000199914f, 4120, 0.0004997253f, 5144, 0.0009994507f, 6168, 0.001998901f, 7454, 0.004997253f, 8478, 0.009994507f, 9502, - 0.01998901f, 10854, 0.04998779f, 11878, 0.09997559f, 12902, 0.1999512f, 14336, - 0.5f, 15360, 1f, 16384, 2f, 17664, 5f, 18688, 10f, 19712, 20f, 21056, 50f, 22080, - 100f, 23104, 200f, 24528, 500f, 25552, 1000f, 26576, 2000f, 27874, 5000f, 28898, - 10000f, 29922, 20000f, 31258, 49984f, 32769, -5.960464E-08f, 32771, -1.788139E-07f, - 32776, -4.768372E-07f, 32784, -9.536743E-07f, 32801, -1.966953E-06f, 32851, - -4.947186E-06f, 32935, -9.953976E-06f, 33103, -1.996756E-05f, 33606, - -4.994869E-05f, 34445, -9.995699E-05f, 35469, -0.000199914f, 36888, -0.0004997253f, - 37912, -0.0009994507f, 38936, -0.001998901f, 40222, -0.004997253f, 41246, - -0.009994507f, 42270, -0.01998901f, 43622, -0.04998779f, 44646, -0.09997559f, - 45670, -0.1999512f, 47104, -0.5f, 48128, -1f, 49152, -2f, 50432, -5f, 51456, -10f, - 52480, -20f, 53824, -50f, 54848, -100f, 55872, -200f, 57296, -500f, 58320, -1000f, - 59344, -2000f, 60642, -5000f, 61666, -10000f, 62690, -20000f, 64026, -49984f}; + 0.01998901f, 10854, 0.04998779f, 11878, 0.09997559f, 12902, 0.1999512f, 14336, 0.5f, + 15360, 1f, 16384, 2f, 17664, 5f, 18688, 10f, 19712, 20f, 21056, 50f, 22080, 100f, + 23104, 200f, 24528, 500f, 25552, 1000f, 26576, 2000f, 27874, 5000f, 28898, 10000f, + 29922, 20000f, 31258, 49984f, 32769, -5.960464E-08f, 32771, -1.788139E-07f, 32776, + -4.768372E-07f, 32784, -9.536743E-07f, 32801, -1.966953E-06f, 32851, -4.947186E-06f, + 32935, -9.953976E-06f, 33103, -1.996756E-05f, 33606, -4.994869E-05f, 34445, + -9.995699E-05f, 35469, -0.000199914f, 36888, -0.0004997253f, 37912, -0.0009994507f, + 38936, -0.001998901f, 40222, -0.004997253f, 41246, -0.009994507f, 42270, + -0.01998901f, 43622, -0.04998779f, 44646, -0.09997559f, 45670, -0.1999512f, 47104, + -0.5f, 48128, -1f, 49152, -2f, 50432, -5f, 51456, -10f, 52480, -20f, 53824, -50f, + 54848, -100f, 55872, -200f, 57296, -500f, 58320, -1000f, 59344, -2000f, 60642, + -5000f, 61666, -10000f, 62690, -20000f, 64026, -49984f}; for (int i = 0; i < test.length; i += 2) { nonExactCheck((short) test[i], test[i + 1]); } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/core/src/test/java/hivemall/utils/math/MatrixUtilsTest.java ---------------------------------------------------------------------- diff --git a/core/src/test/java/hivemall/utils/math/MatrixUtilsTest.java b/core/src/test/java/hivemall/utils/math/MatrixUtilsTest.java index 8b3053a..0623cca 100644 --- a/core/src/test/java/hivemall/utils/math/MatrixUtilsTest.java +++ b/core/src/test/java/hivemall/utils/math/MatrixUtilsTest.java @@ -82,13 +82,16 @@ public class MatrixUtilsTest { // 3 2 1 Assert.assertArrayEquals(new RealMatrix[] {new Array2DRowRealMatrix(new double[] {1}), new Array2DRowRealMatrix(new double[] {2}), - new Array2DRowRealMatrix(new double[] {3})}, A[0]); + new Array2DRowRealMatrix(new double[] {3})}, + A[0]); Assert.assertArrayEquals(new RealMatrix[] {new Array2DRowRealMatrix(new double[] {2}), new Array2DRowRealMatrix(new double[] {1}), - new Array2DRowRealMatrix(new double[] {2})}, A[1]); + new Array2DRowRealMatrix(new double[] {2})}, + A[1]); Assert.assertArrayEquals(new RealMatrix[] {new Array2DRowRealMatrix(new double[] {3}), new Array2DRowRealMatrix(new double[] {2}), - new Array2DRowRealMatrix(new double[] {1})}, A[2]); + new Array2DRowRealMatrix(new double[] {1})}, + A[2]); } @Test @@ -110,12 +113,12 @@ public class MatrixUtilsTest { @Test public void testCombinedMatrices2D() { RealMatrix[] m1 = new RealMatrix[] { - new Array2DRowRealMatrix(new double[][] {new double[] {1, 2, 3}, - new double[] {4, 5, 6}}), - new Array2DRowRealMatrix(new double[][] {new double[] {7, 8, 9}, - new double[] {10, 11, 12}}), - new Array2DRowRealMatrix(new double[][] {new double[] {13, 14, 15}, - new double[] {16, 17, 18}})}; + new Array2DRowRealMatrix( + new double[][] {new double[] {1, 2, 3}, new double[] {4, 5, 6}}), + new Array2DRowRealMatrix( + new double[][] {new double[] {7, 8, 9}, new double[] {10, 11, 12}}), + new Array2DRowRealMatrix( + new double[][] {new double[] {13, 14, 15}, new double[] {16, 17, 18}})}; RealMatrix flatten1 = MatrixUtils.combinedMatrices(m1); Assert.assertEquals(3, flatten1.getColumnDimension()); Assert.assertEquals(6, flatten1.getRowDimension()); @@ -172,12 +175,12 @@ public class MatrixUtilsTest { @Test public void testFlatten2D() { RealMatrix[] m1 = new RealMatrix[] { - new Array2DRowRealMatrix(new double[][] {new double[] {1, 2, 3}, - new double[] {4, 5, 6}}), - new Array2DRowRealMatrix(new double[][] {new double[] {7, 8, 9}, - new double[] {10, 11, 12}}), - new Array2DRowRealMatrix(new double[][] {new double[] {13, 14, 15}, - new double[] {16, 17, 18}})}; + new Array2DRowRealMatrix( + new double[][] {new double[] {1, 2, 3}, new double[] {4, 5, 6}}), + new Array2DRowRealMatrix( + new double[][] {new double[] {7, 8, 9}, new double[] {10, 11, 12}}), + new Array2DRowRealMatrix( + new double[][] {new double[] {13, 14, 15}, new double[] {16, 17, 18}})}; double[] actual = MatrixUtils.flatten(m1); double[] expected = new double[18]; for (int i = 0; i < expected.length; i++) { @@ -195,22 +198,22 @@ public class MatrixUtilsTest { RealMatrix[] actual = MatrixUtils.unflatten(data, 2, 3, 4); RealMatrix[] expected = new RealMatrix[] { - new Array2DRowRealMatrix(new double[][] {new double[] {1, 2, 3}, - new double[] {4, 5, 6}}), - new Array2DRowRealMatrix(new double[][] {new double[] {7, 8, 9}, - new double[] {10, 11, 12}}), - new Array2DRowRealMatrix(new double[][] {new double[] {13, 14, 15}, - new double[] {16, 17, 18}}), - new Array2DRowRealMatrix(new double[][] {new double[] {19, 20, 21}, - new double[] {22, 23, 24}})}; + new Array2DRowRealMatrix( + new double[][] {new double[] {1, 2, 3}, new double[] {4, 5, 6}}), + new Array2DRowRealMatrix( + new double[][] {new double[] {7, 8, 9}, new double[] {10, 11, 12}}), + new Array2DRowRealMatrix( + new double[][] {new double[] {13, 14, 15}, new double[] {16, 17, 18}}), + new Array2DRowRealMatrix( + new double[][] {new double[] {19, 20, 21}, new double[] {22, 23, 24}})}; Assert.assertArrayEquals(expected, actual); } @Test public void testPower1() { - RealMatrix A = new Array2DRowRealMatrix(new double[][] {new double[] {1, 2, 3}, - new double[] {4, 5, 6}}); + RealMatrix A = new Array2DRowRealMatrix( + new double[][] {new double[] {1, 2, 3}, new double[] {4, 5, 6}}); double[] x = new double[3]; x[0] = Math.random(); @@ -241,9 +244,9 @@ public class MatrixUtilsTest { RealMatrix actual = new Array2DRowRealMatrix(new double[4][4]); MatrixUtils.lanczosTridiagonalization(C, a, actual); - RealMatrix expected = new Array2DRowRealMatrix(new double[][] {new double[] {40, 60, 0, 0}, - new double[] {60, 10, 120, 0}, new double[] {0, 120, 10, 120}, - new double[] {0, 0, 120, 10}}); + RealMatrix expected = new Array2DRowRealMatrix( + new double[][] {new double[] {40, 60, 0, 0}, new double[] {60, 10, 120, 0}, + new double[] {0, 120, 10, 120}, new double[] {0, 0, 120, 10}}); Assert.assertEquals(expected, actual); } @@ -251,9 +254,9 @@ public class MatrixUtilsTest { @Test public void testTridiagonalEigen() { // Tridiagonal Matrix - RealMatrix T = new Array2DRowRealMatrix(new double[][] {new double[] {40, 60, 0, 0}, - new double[] {60, 10, 120, 0}, new double[] {0, 120, 10, 120}, - new double[] {0, 0, 120, 10}}); + RealMatrix T = new Array2DRowRealMatrix( + new double[][] {new double[] {40, 60, 0, 0}, new double[] {60, 10, 120, 0}, + new double[] {0, 120, 10, 120}, new double[] {0, 0, 120, 10}}); double[] eigvals = new double[4]; RealMatrix eigvecs = new Array2DRowRealMatrix(new double[4][4]); http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/mixserv/src/main/java/hivemall/mix/server/MixServer.java ---------------------------------------------------------------------- diff --git a/mixserv/src/main/java/hivemall/mix/server/MixServer.java b/mixserv/src/main/java/hivemall/mix/server/MixServer.java index 67ed25f..c6aa219 100644 --- a/mixserv/src/main/java/hivemall/mix/server/MixServer.java +++ b/mixserv/src/main/java/hivemall/mix/server/MixServer.java @@ -109,8 +109,8 @@ public final class MixServer implements Runnable { public String toString() { return "[port=" + port + ", numWorkers=" + numWorkers + ", ssl=" + ssl + ", scale=" + scale + ", syncThreshold=" + syncThreshold + ", sessionTTLinSec=" + sessionTTLinSec - + ", sweepIntervalInSec=" + sweepIntervalInSec + ", jmx=" + jmx + ", state=" - + state + "]"; + + ", sweepIntervalInSec=" + sweepIntervalInSec + ", jmx=" + jmx + ", state=" + state + + "]"; } public ServerState getState() { @@ -143,7 +143,8 @@ public final class MixServer implements Runnable { // configure metrics ScheduledExecutorService metricCollector = Executors.newScheduledThreadPool(1); MixServerMetrics metrics = new MixServerMetrics(); - ThroughputCounter throughputCounter = new ThroughputCounter(metricCollector, 5000L, metrics); + ThroughputCounter throughputCounter = + new ThroughputCounter(metricCollector, 5000L, metrics); if (jmx) {// register mbean MetricsRegistry.registerMBeans(metrics, port); } @@ -151,8 +152,8 @@ public final class MixServer implements Runnable { // configure initializer SessionStore sessionStore = new SessionStore(); MixServerHandler msgHandler = new MixServerHandler(sessionStore, syncThreshold, scale); - MixServerInitializer initializer = new MixServerInitializer(msgHandler, throughputCounter, - sslCtx); + MixServerInitializer initializer = + new MixServerInitializer(msgHandler, throughputCounter, sslCtx); Runnable cleanSessionTask = new IdleSessionSweeper(sessionStore, sessionTTLinSec * 1000L); ScheduledExecutorService idleSessionChecker = Executors.newScheduledThreadPool(1); http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/mixserv/src/main/java/hivemall/mix/server/MixServerHandler.java ---------------------------------------------------------------------- diff --git a/mixserv/src/main/java/hivemall/mix/server/MixServerHandler.java b/mixserv/src/main/java/hivemall/mix/server/MixServerHandler.java index ae5c27c..bf33c69 100644 --- a/mixserv/src/main/java/hivemall/mix/server/MixServerHandler.java +++ b/mixserv/src/main/java/hivemall/mix/server/MixServerHandler.java @@ -90,7 +90,8 @@ public final class MixServerHandler extends SimpleChannelInboundHandler<MixMessa } @Nonnull - private PartialResult getPartialResult(@Nonnull MixMessage msg, @Nonnull SessionObject session) { + private PartialResult getPartialResult(@Nonnull MixMessage msg, + @Nonnull SessionObject session) { final ConcurrentMap<Object, PartialResult> map = session.get(); Object feature = msg.getFeature(); http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/mixserv/src/main/java/hivemall/mix/store/PartialArgminKLD.java ---------------------------------------------------------------------- diff --git a/mixserv/src/main/java/hivemall/mix/store/PartialArgminKLD.java b/mixserv/src/main/java/hivemall/mix/store/PartialArgminKLD.java index a6ffc4b..85d0936 100644 --- a/mixserv/src/main/java/hivemall/mix/store/PartialArgminKLD.java +++ b/mixserv/src/main/java/hivemall/mix/store/PartialArgminKLD.java @@ -52,7 +52,8 @@ public final class PartialArgminKLD extends PartialResult { } @Override - public void subtract(float localWeight, float covar, @Nonnegative int deltaUpdates, float scale) { + public void subtract(float localWeight, float covar, @Nonnegative int deltaUpdates, + float scale) { this.sum_mean_div_covar -= (localWeight / covar) / scale; this.sum_inv_covar -= (1.f / covar) / scale; } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/mixserv/src/main/java/hivemall/mix/store/PartialAverage.java ---------------------------------------------------------------------- diff --git a/mixserv/src/main/java/hivemall/mix/store/PartialAverage.java b/mixserv/src/main/java/hivemall/mix/store/PartialAverage.java index e03a6d5..0d1156b 100644 --- a/mixserv/src/main/java/hivemall/mix/store/PartialAverage.java +++ b/mixserv/src/main/java/hivemall/mix/store/PartialAverage.java @@ -54,7 +54,8 @@ public final class PartialAverage extends PartialResult { } @Override - public void subtract(float localWeight, float covar, @Nonnegative int deltaUpdates, float scale) { + public void subtract(float localWeight, float covar, @Nonnegative int deltaUpdates, + float scale) { assert (deltaUpdates > 0) : deltaUpdates; scaledSumWeights -= ((localWeight / scale) * deltaUpdates); totalUpdates -= deltaUpdates; // note deltaUpdates is in range (0,127] http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/mixserv/src/main/java/hivemall/mix/store/SessionStore.java ---------------------------------------------------------------------- diff --git a/mixserv/src/main/java/hivemall/mix/store/SessionStore.java b/mixserv/src/main/java/hivemall/mix/store/SessionStore.java index 1a2892e..9ff0a53 100644 --- a/mixserv/src/main/java/hivemall/mix/store/SessionStore.java +++ b/mixserv/src/main/java/hivemall/mix/store/SessionStore.java @@ -51,8 +51,8 @@ public final class SessionStore { public SessionObject get(@Nonnull String groupID) { SessionObject sessionObj = sessions.get(groupID); if (sessionObj == null) { - ConcurrentMap<Object, PartialResult> map = new ConcurrentHashMap<Object, PartialResult>( - EXPECTED_MODEL_SIZE); + ConcurrentMap<Object, PartialResult> map = + new ConcurrentHashMap<Object, PartialResult>(EXPECTED_MODEL_SIZE); sessionObj = new SessionObject(map); SessionObject existing = sessions.putIfAbsent(groupID, sessionObj); if (existing != null) { @@ -76,7 +76,8 @@ public final class SessionStore { private final ConcurrentMap<String, SessionObject> sessions; private final long ttl; - public IdleSessionSweeper(@Nonnull SessionStore sessionStore, @Nonnegative long ttlInMillis) { + public IdleSessionSweeper(@Nonnull SessionStore sessionStore, + @Nonnegative long ttlInMillis) { this.sessions = sessionStore.getSessions(); this.ttl = ttlInMillis; } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/mixserv/src/test/java/hivemall/mix/server/MixServerHandlerTest.java ---------------------------------------------------------------------- diff --git a/mixserv/src/test/java/hivemall/mix/server/MixServerHandlerTest.java b/mixserv/src/test/java/hivemall/mix/server/MixServerHandlerTest.java index fbb9cc1..7d4b2f3 100644 --- a/mixserv/src/test/java/hivemall/mix/server/MixServerHandlerTest.java +++ b/mixserv/src/test/java/hivemall/mix/server/MixServerHandlerTest.java @@ -52,8 +52,9 @@ public final class MixServerHandlerTest extends HivemallTestBase { SessionStore session = new SessionStore(); MixServerHandler handler = new MixServerHandler(session, 4, 1.0f); - Method mixMethod = MixServerHandler.class.getDeclaredMethod("mix", - ChannelHandlerContext.class, MixMessage.class, PartialResult.class, SessionObject.class); + Method mixMethod = + MixServerHandler.class.getDeclaredMethod("mix", ChannelHandlerContext.class, + MixMessage.class, PartialResult.class, SessionObject.class); mixMethod.setAccessible(true); SessionObject sessionObj = session.get("dummy"); @@ -77,8 +78,8 @@ public final class MixServerHandlerTest extends HivemallTestBase { Assert.assertEquals(5.0, acc.getWeight(1.0f), 0.001); // Check expected exceptions - exception.expectCause(new CauseMatcher(IllegalArgumentException.class, - "Illegal deltaUpdates received: 0")); + exception.expectCause( + new CauseMatcher(IllegalArgumentException.class, "Illegal deltaUpdates received: 0")); MixMessage msg4 = new MixMessage(MixEventName.average, dummyFeature, 0.0f, (short) 0, 0); mixMethod.invoke(handler, ctx, msg4, acc, sessionObj); } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/mixserv/src/test/java/hivemall/mix/server/MixServerTest.java ---------------------------------------------------------------------- diff --git a/mixserv/src/test/java/hivemall/mix/server/MixServerTest.java b/mixserv/src/test/java/hivemall/mix/server/MixServerTest.java index f6db5f8..afa201a 100644 --- a/mixserv/src/test/java/hivemall/mix/server/MixServerTest.java +++ b/mixserv/src/test/java/hivemall/mix/server/MixServerTest.java @@ -298,8 +298,8 @@ public class MixServerTest extends HivemallTestBase { private static void invokeClient01(String groupId, int serverPort, boolean denseModel, boolean cancelMix) throws InterruptedException { - PredictionModel model = denseModel ? new NewDenseModel(100) - : new NewSparseModel(100, false); + PredictionModel model = + denseModel ? new NewDenseModel(100) : new NewSparseModel(100, false); model.configureClock(); MixClient client = null; try { http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/mixserv/src/test/java/hivemall/test/HivemallTestBase.java ---------------------------------------------------------------------- diff --git a/mixserv/src/test/java/hivemall/test/HivemallTestBase.java b/mixserv/src/test/java/hivemall/test/HivemallTestBase.java index ee80eba..5eb1879 100644 --- a/mixserv/src/test/java/hivemall/test/HivemallTestBase.java +++ b/mixserv/src/test/java/hivemall/test/HivemallTestBase.java @@ -39,7 +39,7 @@ public abstract class HivemallTestBase { @Before public void afterEachTest() { - logger.info("\n\n===== FINISHED " + packageName + ": '" + testName.getMethodName() - + "' =====\n"); + logger.info( + "\n\n===== FINISHED " + packageName + ": '" + testName.getMethodName() + "' =====\n"); } } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/nlp/src/main/java/hivemall/nlp/tokenizer/KuromojiUDF.java ---------------------------------------------------------------------- diff --git a/nlp/src/main/java/hivemall/nlp/tokenizer/KuromojiUDF.java b/nlp/src/main/java/hivemall/nlp/tokenizer/KuromojiUDF.java index cdf6dab..48b566f 100644 --- a/nlp/src/main/java/hivemall/nlp/tokenizer/KuromojiUDF.java +++ b/nlp/src/main/java/hivemall/nlp/tokenizer/KuromojiUDF.java @@ -59,8 +59,7 @@ import org.apache.lucene.analysis.ja.dict.UserDictionary; import org.apache.lucene.analysis.tokenattributes.CharTermAttribute; import org.apache.lucene.analysis.util.CharArraySet; -@Description( - name = "tokenize_ja", +@Description(name = "tokenize_ja", value = "_FUNC_(String line [, const string mode = \"normal\", const array<string> stopWords, const array<string> stopTags, const array<string> userDict (or string userDictURL)])" + " - returns tokenized strings in array<string>", extended = "select tokenize_ja(\"kuromojiã使ã£ãåãã¡æ¸ãã®ãã¹ãã§ãã第äºå¼æ°ã«ã¯normal/search/extendedãæå®ã§ãã¾ããããã©ã«ãã§ã¯normalã¢ã¼ãã§ãã\");\n" @@ -86,8 +85,8 @@ public final class KuromojiUDF extends GenericUDF { public ObjectInspector initialize(ObjectInspector[] arguments) throws UDFArgumentException { final int arglen = arguments.length; if (arglen < 1 || arglen > 5) { - throw new UDFArgumentException("Invalid number of arguments for `tokenize_ja`: " - + arglen); + throw new UDFArgumentException( + "Invalid number of arguments for `tokenize_ja`: " + arglen); } this._mode = (arglen >= 2) ? tokenizationMode(arguments[1]) : Mode.NORMAL; @@ -96,8 +95,8 @@ public final class KuromojiUDF extends GenericUDF { this._stopWordsArray = HiveUtils.getConstStringArray(arguments[2]); } - this._stopTags = (arglen >= 4) ? stopTags(arguments[3]) - : JapaneseAnalyzer.getDefaultStopTags(); + this._stopTags = + (arglen >= 4) ? stopTags(arguments[3]) : JapaneseAnalyzer.getDefaultStopTags(); if (arglen >= 5) { if (HiveUtils.isConstListOI(arguments[4])) { @@ -112,7 +111,8 @@ public final class KuromojiUDF extends GenericUDF { this._analyzer = null; - return ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.writableStringObjectInspector); + return ObjectInspectorFactory.getStandardListObjectInspector( + PrimitiveObjectInspectorFactory.writableStringObjectInspector); } @Override @@ -271,17 +271,17 @@ public final class KuromojiUDF extends GenericUDF { final InputStream is; try { - is = IOUtils.decodeInputStream(HttpUtils.getLimitedInputStream(conn, - MAX_INPUT_STREAM_SIZE)); + is = IOUtils.decodeInputStream( + HttpUtils.getLimitedInputStream(conn, MAX_INPUT_STREAM_SIZE)); } catch (NullPointerException | IOException e) { throw new UDFArgumentException("Failed to get input stream from the connection: " + userDictURL + '\n' + ExceptionUtils.prettyPrintStackTrace(e)); } - CharsetDecoder decoder = StandardCharsets.UTF_8.newDecoder() - .onMalformedInput(CodingErrorAction.REPORT) - .onUnmappableCharacter( - CodingErrorAction.REPORT); + CharsetDecoder decoder = + StandardCharsets.UTF_8.newDecoder() + .onMalformedInput(CodingErrorAction.REPORT) + .onUnmappableCharacter(CodingErrorAction.REPORT); final Reader reader = new InputStreamReader(is, decoder); try { return UserDictionary.open(reader); // return null if empty http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/nlp/src/main/java/hivemall/nlp/tokenizer/SmartcnUDF.java ---------------------------------------------------------------------- diff --git a/nlp/src/main/java/hivemall/nlp/tokenizer/SmartcnUDF.java b/nlp/src/main/java/hivemall/nlp/tokenizer/SmartcnUDF.java index d185b0d..cf6249f 100644 --- a/nlp/src/main/java/hivemall/nlp/tokenizer/SmartcnUDF.java +++ b/nlp/src/main/java/hivemall/nlp/tokenizer/SmartcnUDF.java @@ -56,14 +56,15 @@ public final class SmartcnUDF extends GenericUDF { public ObjectInspector initialize(ObjectInspector[] arguments) throws UDFArgumentException { final int arglen = arguments.length; if (arglen < 1 || arglen > 2) { - throw new UDFArgumentException("Invalid number of arguments for `tokenize_cn`: " - + arglen); + throw new UDFArgumentException( + "Invalid number of arguments for `tokenize_cn`: " + arglen); } this._stopWordsArray = (arglen >= 2) ? HiveUtils.getConstStringArray(arguments[1]) : null; this._analyzer = null; - return ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.writableStringObjectInspector); + return ObjectInspectorFactory.getStandardListObjectInspector( + PrimitiveObjectInspectorFactory.writableStringObjectInspector); } @Override http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/nlp/src/test/java/hivemall/nlp/tokenizer/KuromojiUDFTest.java ---------------------------------------------------------------------- diff --git a/nlp/src/test/java/hivemall/nlp/tokenizer/KuromojiUDFTest.java b/nlp/src/test/java/hivemall/nlp/tokenizer/KuromojiUDFTest.java index 356507d..eb755e0 100644 --- a/nlp/src/test/java/hivemall/nlp/tokenizer/KuromojiUDFTest.java +++ b/nlp/src/test/java/hivemall/nlp/tokenizer/KuromojiUDFTest.java @@ -351,8 +351,7 @@ public class KuromojiUDFTest { // userDictUrl (Kuromoji official sample user defined dict on GitHub) // e.g., "æ¥æ¬çµæ¸æ°è" will be "æ¥æ¬", "çµæ¸", and "æ°è" argOIs[4] = PrimitiveObjectInspectorFactory.getPrimitiveWritableConstantObjectInspector( - stringType, - new Text( + stringType, new Text( "https://raw.githubusercontent.com/atilika/kuromoji/909fd6b32bf4e9dc86b7599de5c9b50ca8f004a1/kuromoji-core/src/test/resources/userdict.txt";)); udf.initialize(argOIs); http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/spark/common/src/main/java/hivemall/dataset/LogisticRegressionDataGeneratorUDTFWrapper.java ---------------------------------------------------------------------- diff --git a/spark/common/src/main/java/hivemall/dataset/LogisticRegressionDataGeneratorUDTFWrapper.java b/spark/common/src/main/java/hivemall/dataset/LogisticRegressionDataGeneratorUDTFWrapper.java index a6d5468..cf10ed7 100644 --- a/spark/common/src/main/java/hivemall/dataset/LogisticRegressionDataGeneratorUDTFWrapper.java +++ b/spark/common/src/main/java/hivemall/dataset/LogisticRegressionDataGeneratorUDTFWrapper.java @@ -41,7 +41,8 @@ import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector; @Description(name = "lr_datagen", value = "_FUNC_(options string) - Generates a logistic regression dataset") public final class LogisticRegressionDataGeneratorUDTFWrapper extends UDTFWithOptions { - private transient LogisticRegressionDataGeneratorUDTF udtf = new LogisticRegressionDataGeneratorUDTF(); + private transient LogisticRegressionDataGeneratorUDTF udtf = + new LogisticRegressionDataGeneratorUDTF(); @Override protected Options getOptions() { http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/spark/common/src/main/java/hivemall/ftvec/AddBiasUDFWrapper.java ---------------------------------------------------------------------- diff --git a/spark/common/src/main/java/hivemall/ftvec/AddBiasUDFWrapper.java b/spark/common/src/main/java/hivemall/ftvec/AddBiasUDFWrapper.java index b454fd9..e7da7cb 100644 --- a/spark/common/src/main/java/hivemall/ftvec/AddBiasUDFWrapper.java +++ b/spark/common/src/main/java/hivemall/ftvec/AddBiasUDFWrapper.java @@ -65,7 +65,8 @@ public class AddBiasUDFWrapper extends GenericUDF { throw new UDFArgumentTypeException(0, "Type mismatch: features"); } - return ObjectInspectorFactory.getStandardListObjectInspector(argumentOI.getListElementObjectInspector()); + return ObjectInspectorFactory.getStandardListObjectInspector( + argumentOI.getListElementObjectInspector()); } @Override http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/spark/common/src/main/java/hivemall/ftvec/AddFeatureIndexUDFWrapper.java ---------------------------------------------------------------------- diff --git a/spark/common/src/main/java/hivemall/ftvec/AddFeatureIndexUDFWrapper.java b/spark/common/src/main/java/hivemall/ftvec/AddFeatureIndexUDFWrapper.java index 0b687db..6be3a9e 100644 --- a/spark/common/src/main/java/hivemall/ftvec/AddFeatureIndexUDFWrapper.java +++ b/spark/common/src/main/java/hivemall/ftvec/AddFeatureIndexUDFWrapper.java @@ -39,8 +39,7 @@ import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectIn * NOTE: This is needed to avoid the issue of Spark reflection. That is, spark cannot handle List<> * as a return type in Hive UDF. Therefore, the type must be passed via ObjectInspector. */ -@Description( - name = "add_feature_index", +@Description(name = "add_feature_index", value = "_FUNC_(dense features in array<double>) - Returns a feature vector with feature indices") @UDFType(deterministic = true, stateful = false) public class AddFeatureIndexUDFWrapper extends GenericUDF { @@ -67,7 +66,8 @@ public class AddFeatureIndexUDFWrapper extends GenericUDF { throw new UDFArgumentTypeException(0, "Type mismatch: features"); } - return ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector); + return ObjectInspectorFactory.getStandardListObjectInspector( + PrimitiveObjectInspectorFactory.javaStringObjectInspector); } @Override http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/spark/common/src/main/java/hivemall/ftvec/ExtractWeightUDFWrapper.java ---------------------------------------------------------------------- diff --git a/spark/common/src/main/java/hivemall/ftvec/ExtractWeightUDFWrapper.java b/spark/common/src/main/java/hivemall/ftvec/ExtractWeightUDFWrapper.java index 8580247..b5ef807 100644 --- a/spark/common/src/main/java/hivemall/ftvec/ExtractWeightUDFWrapper.java +++ b/spark/common/src/main/java/hivemall/ftvec/ExtractWeightUDFWrapper.java @@ -56,7 +56,8 @@ public class ExtractWeightUDFWrapper extends GenericUDF { throw new UDFArgumentTypeException(0, "Type mismatch: feature"); } - return PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(PrimitiveCategory.DOUBLE); + return PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector( + PrimitiveCategory.DOUBLE); } @Override http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/spark/common/src/main/java/hivemall/ftvec/SortByFeatureUDFWrapper.java ---------------------------------------------------------------------- diff --git a/spark/common/src/main/java/hivemall/ftvec/SortByFeatureUDFWrapper.java b/spark/common/src/main/java/hivemall/ftvec/SortByFeatureUDFWrapper.java index 584be6c..e13e030 100644 --- a/spark/common/src/main/java/hivemall/ftvec/SortByFeatureUDFWrapper.java +++ b/spark/common/src/main/java/hivemall/ftvec/SortByFeatureUDFWrapper.java @@ -61,8 +61,10 @@ public class SortByFeatureUDFWrapper extends GenericUDF { ObjectInspector valueOI = argumentOI.getMapValueObjectInspector(); if (keyOI.getCategory().equals(Category.PRIMITIVE) && valueOI.getCategory().equals(Category.PRIMITIVE)) { - final PrimitiveCategory keyCategory = ((PrimitiveObjectInspector) keyOI).getPrimitiveCategory(); - final PrimitiveCategory valueCategory = ((PrimitiveObjectInspector) valueOI).getPrimitiveCategory(); + final PrimitiveCategory keyCategory = + ((PrimitiveObjectInspector) keyOI).getPrimitiveCategory(); + final PrimitiveCategory valueCategory = + ((PrimitiveObjectInspector) valueOI).getPrimitiveCategory(); if (keyCategory == PrimitiveCategory.INT && valueCategory == PrimitiveCategory.FLOAT) { break; @@ -81,7 +83,8 @@ public class SortByFeatureUDFWrapper extends GenericUDF { public Object evaluate(DeferredObject[] arguments) throws HiveException { assert (arguments.length == 1); @SuppressWarnings("unchecked") - final Map<IntWritable, FloatWritable> input = (Map<IntWritable, FloatWritable>) argumentOI.getMap(arguments[0].get()); + final Map<IntWritable, FloatWritable> input = + (Map<IntWritable, FloatWritable>) argumentOI.getMap(arguments[0].get()); return udf.evaluate(input); } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/spark/common/src/main/java/hivemall/ftvec/scaling/L2NormalizationUDFWrapper.java ---------------------------------------------------------------------- diff --git a/spark/common/src/main/java/hivemall/ftvec/scaling/L2NormalizationUDFWrapper.java b/spark/common/src/main/java/hivemall/ftvec/scaling/L2NormalizationUDFWrapper.java index db533be..dcdba24 100644 --- a/spark/common/src/main/java/hivemall/ftvec/scaling/L2NormalizationUDFWrapper.java +++ b/spark/common/src/main/java/hivemall/ftvec/scaling/L2NormalizationUDFWrapper.java @@ -56,7 +56,8 @@ public class L2NormalizationUDFWrapper extends GenericUDF { switch (arguments[0].getCategory()) { case LIST: - ObjectInspector elmOI = ((ListObjectInspector) arguments[0]).getListElementObjectInspector(); + ObjectInspector elmOI = + ((ListObjectInspector) arguments[0]).getListElementObjectInspector(); if (elmOI.getCategory().equals(Category.PRIMITIVE)) { if (((PrimitiveObjectInspector) elmOI).getPrimitiveCategory() == PrimitiveCategory.STRING) { break; @@ -71,7 +72,8 @@ public class L2NormalizationUDFWrapper extends GenericUDF { // Create a ObjectInspector converter for arguments ObjectInspector outputElemOI = ObjectInspectorFactory.getReflectionObjectInspector( Text.class, ObjectInspectorOptions.JAVA); - ObjectInspector outputOI = ObjectInspectorFactory.getStandardListObjectInspector(outputElemOI); + ObjectInspector outputOI = + ObjectInspectorFactory.getStandardListObjectInspector(outputElemOI); toListText = ObjectInspectorConverters.getConverter(arguments[0], outputOI); ObjectInspector listElemOI = PrimitiveObjectInspectorFactory.javaStringObjectInspector; http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/spark/common/src/main/java/hivemall/knn/lsh/MinHashesUDFWrapper.java ---------------------------------------------------------------------- diff --git a/spark/common/src/main/java/hivemall/knn/lsh/MinHashesUDFWrapper.java b/spark/common/src/main/java/hivemall/knn/lsh/MinHashesUDFWrapper.java index d3bcbe6..3c1fe9b 100644 --- a/spark/common/src/main/java/hivemall/knn/lsh/MinHashesUDFWrapper.java +++ b/spark/common/src/main/java/hivemall/knn/lsh/MinHashesUDFWrapper.java @@ -35,8 +35,7 @@ import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectIn import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorUtils; /** A wrapper of [[hivemall.knn.lsh.MinHashesUDF]]. */ -@Description( - name = "minhashes", +@Description(name = "minhashes", value = "_FUNC_(features in array<string>, noWeight in boolean) - Returns hashed features as array<int>") @UDFType(deterministic = true, stateful = false) public class MinHashesUDFWrapper extends GenericUDF { @@ -70,7 +69,9 @@ public class MinHashesUDFWrapper extends GenericUDF { throw new UDFArgumentException("Type mismatch: noWeight"); } - return ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(PrimitiveCategory.INT)); + return ObjectInspectorFactory.getStandardListObjectInspector( + PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector( + PrimitiveCategory.INT)); } @Override @@ -78,8 +79,8 @@ public class MinHashesUDFWrapper extends GenericUDF { assert (arguments.length == 2); @SuppressWarnings("unchecked") final List<String> features = (List<String>) featuresOI.getList(arguments[0].get()); - final Boolean noWeight = PrimitiveObjectInspectorUtils.getBoolean(arguments[1].get(), - noWeightOI); + final Boolean noWeight = + PrimitiveObjectInspectorUtils.getBoolean(arguments[1].get(), noWeightOI); return udf.evaluate(features, noWeight); } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/spark/common/src/main/java/hivemall/tools/mapred/RowIdUDFWrapper.java ---------------------------------------------------------------------- diff --git a/spark/common/src/main/java/hivemall/tools/mapred/RowIdUDFWrapper.java b/spark/common/src/main/java/hivemall/tools/mapred/RowIdUDFWrapper.java index f386223..e907c38 100644 --- a/spark/common/src/main/java/hivemall/tools/mapred/RowIdUDFWrapper.java +++ b/spark/common/src/main/java/hivemall/tools/mapred/RowIdUDFWrapper.java @@ -30,8 +30,7 @@ import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector; import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory; /** An alternative implementation of [[hivemall.tools.mapred.RowIdUDF]]. */ -@Description( - name = "rowid", +@Description(name = "rowid", value = "_FUNC_() - Returns a generated row id of a form {TASK_ID}-{UUID}-{SEQUENCE_NUMBER}") @UDFType(deterministic = false, stateful = true) public class RowIdUDFWrapper extends GenericUDF { http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/tools/hivemall-docs/src/main/java/hivemall/docs/FuncsListGenerator.java ---------------------------------------------------------------------- diff --git a/tools/hivemall-docs/src/main/java/hivemall/docs/FuncsListGenerator.java b/tools/hivemall-docs/src/main/java/hivemall/docs/FuncsListGenerator.java index 187c752..c940c1d 100644 --- a/tools/hivemall-docs/src/main/java/hivemall/docs/FuncsListGenerator.java +++ b/tools/hivemall-docs/src/main/java/hivemall/docs/FuncsListGenerator.java @@ -113,7 +113,8 @@ public class FuncsListGenerator extends AbstractMojo { Collections.singletonList("hivemall.ftvec.selection")); funcsHeaders.put("## Feature transformation and vectorization", Collections.singletonList("hivemall.ftvec.trans")); - funcsHeaders.put("# Geospatial functions", Collections.singletonList("hivemall.geospatial")); + funcsHeaders.put("# Geospatial functions", + Collections.singletonList("hivemall.geospatial")); funcsHeaders.put("# Distance measures", Collections.singletonList("hivemall.knn.distance")); funcsHeaders.put("# Locality-sensitive hashing", Collections.singletonList("hivemall.knn.lsh")); @@ -138,12 +139,10 @@ public class FuncsListGenerator extends AbstractMojo { return; } - generate( - new File(basedir, pathToGenericFuncs), + generate(new File(basedir, pathToGenericFuncs), "This page describes a list of useful Hivemall generic functions. See also a [list of machine-learning-related functions](./funcs.md).", genericFuncsHeaders); - generate( - new File(basedir, pathToFuncs), + generate(new File(basedir, pathToFuncs), "This page describes a list of Hivemall functions. See also a [list of generic Hivemall functions](./generic_funcs.md) for more general-purpose functions such as array and map UDFs.", funcsHeaders); } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/xgboost/src/main/java/hivemall/xgboost/NativeLibLoader.java ---------------------------------------------------------------------- diff --git a/xgboost/src/main/java/hivemall/xgboost/NativeLibLoader.java b/xgboost/src/main/java/hivemall/xgboost/NativeLibLoader.java index 207376f..bf3ea1f 100644 --- a/xgboost/src/main/java/hivemall/xgboost/NativeLibLoader.java +++ b/xgboost/src/main/java/hivemall/xgboost/NativeLibLoader.java @@ -98,8 +98,8 @@ public final class NativeLibLoader { try { File tempFile = createTempFileFromResource(userDefinedLibFile.getName(), new FileInputStream(userDefinedLibFile.getAbsolutePath())); - logger.info("Copyed the user-defined native library as " - + tempFile.getAbsolutePath()); + logger.info( + "Copyed the user-defined native library as " + tempFile.getAbsolutePath()); addLibraryPath(tempFile.getParent()); } catch (Exception e) { // Simply ignore it here @@ -126,12 +126,12 @@ public final class NativeLibLoader { * @throws IOException * @throws IllegalArgumentException */ - static File createTempFileFromResource(String libName, InputStream is) throws IOException, - IllegalArgumentException { + static File createTempFileFromResource(String libName, InputStream is) + throws IOException, IllegalArgumentException { // Create a temporary folder with a random number for the native lib final String uuid = UUID.randomUUID().toString(); - final File tempFolder = new File(System.getProperty("java.io.tmpdir"), String.format( - "%s-%s", getPrefix(libName), uuid)); + final File tempFolder = new File(System.getProperty("java.io.tmpdir"), + String.format("%s-%s", getPrefix(libName), uuid)); if (!tempFolder.exists()) { boolean created = tempFolder.mkdirs(); if (!created) { http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/xgboost/src/main/java/hivemall/xgboost/XGBoostPredictUDTF.java ---------------------------------------------------------------------- diff --git a/xgboost/src/main/java/hivemall/xgboost/XGBoostPredictUDTF.java b/xgboost/src/main/java/hivemall/xgboost/XGBoostPredictUDTF.java index fd4c0b4..9c3b476 100644 --- a/xgboost/src/main/java/hivemall/xgboost/XGBoostPredictUDTF.java +++ b/xgboost/src/main/java/hivemall/xgboost/XGBoostPredictUDTF.java @@ -87,8 +87,8 @@ public abstract class XGBoostPredictUDTF extends UDTFWithOptions { cl = this.parseOptions(rawArgs); _batch_size = Primitives.parseInt(cl.getOptionValue("_batch_size"), _batch_size); if (_batch_size < 1) { - throw new IllegalArgumentException("batch_size must be greater than 0: " - + _batch_size); + throw new IllegalArgumentException( + "batch_size must be greater than 0: " + _batch_size); } } this.batch_size = _batch_size; @@ -171,8 +171,8 @@ public abstract class XGBoostPredictUDTF extends UDTFWithOptions { } final String modelId = PrimitiveObjectInspectorUtils.getString(args[2], modelIdOI); if (!mapToModel.containsKey(modelId)) { - final byte[] predModel = PrimitiveObjectInspectorUtils.getBinary(args[3], modelOI) - .getBytes(); + final byte[] predModel = + PrimitiveObjectInspectorUtils.getBinary(args[3], modelOI).getBytes(); mapToModel.put(modelId, initXgBooster(predModel)); } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/xgboost/src/main/java/hivemall/xgboost/XGBoostUDTF.java ---------------------------------------------------------------------- diff --git a/xgboost/src/main/java/hivemall/xgboost/XGBoostUDTF.java b/xgboost/src/main/java/hivemall/xgboost/XGBoostUDTF.java index 3f3c08f..272614f 100644 --- a/xgboost/src/main/java/hivemall/xgboost/XGBoostUDTF.java +++ b/xgboost/src/main/java/hivemall/xgboost/XGBoostUDTF.java @@ -135,16 +135,15 @@ public abstract class XGBoostUDTF extends UDTFWithOptions { /** Parameters for Tree Booster */ opts.addOption("eta", true, "Step size shrinkage used in update to prevents overfitting [default: 0.3]"); - opts.addOption( - "gamma", - true, + opts.addOption("gamma", true, "Minimum loss reduction required to make a further partition on a leaf node of the tree [default: 0.0]"); opts.addOption("max_depth", true, "Max depth of decision tree [default: 6]"); opts.addOption("min_child_weight", true, "Minimum sum of instance weight(hessian) needed in a child [default: 1]"); opts.addOption("max_delta_step", true, "Maximum delta step we allow each tree's weight estimation to be [default: 0]"); - opts.addOption("subsample", true, "Subsample ratio of the training instance [default: 1.0]"); + opts.addOption("subsample", true, + "Subsample ratio of the training instance [default: 1.0]"); opts.addOption("colsample_bytree", true, "Subsample ratio of columns when constructing each tree [default: 1.0]"); opts.addOption("colsample_bylevel", true, @@ -218,7 +217,8 @@ public abstract class XGBoostUDTF extends UDTFWithOptions { params.put("subsample", Double.valueOf(cl.getOptionValue("subsample"))); } if (cl.hasOption("colsample_bytree")) { - params.put("colsamle_bytree", Double.valueOf(cl.getOptionValue("colsample_bytree"))); + params.put("colsamle_bytree", + Double.valueOf(cl.getOptionValue("colsample_bytree"))); } if (cl.hasOption("colsample_bylevel")) { params.put("colsamle_bylevel", @@ -309,8 +309,8 @@ public abstract class XGBoostUDTF extends UDTFWithOptions { Class<?>[] args = {Map.class, DMatrix[].class}; Constructor<Booster> ctor = Booster.class.getDeclaredConstructor(args); ctor.setAccessible(true); - return ctor.newInstance(new Object[] {params, - new DMatrix[] {new DMatrix(input.iterator(), "")}}); + return ctor.newInstance( + new Object[] {params, new DMatrix[] {new DMatrix(input.iterator(), "")}}); } @Override http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/xgboost/src/main/java/hivemall/xgboost/XGBoostUtils.java ---------------------------------------------------------------------- diff --git a/xgboost/src/main/java/hivemall/xgboost/XGBoostUtils.java b/xgboost/src/main/java/hivemall/xgboost/XGBoostUtils.java index 0472229..aa3d7fd 100644 --- a/xgboost/src/main/java/hivemall/xgboost/XGBoostUtils.java +++ b/xgboost/src/main/java/hivemall/xgboost/XGBoostUtils.java @@ -29,7 +29,8 @@ public final class XGBoostUtils { /** Transform List<String> inputs into a XGBoost input format */ @Nullable - public static LabeledPoint parseFeatures(final double target, @Nonnull final String[] features) { + public static LabeledPoint parseFeatures(final double target, + @Nonnull final String[] features) { final int size = features.length; if (size == 0) { return null; http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostBinaryClassifierUDTF.java ---------------------------------------------------------------------- diff --git a/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostBinaryClassifierUDTF.java b/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostBinaryClassifierUDTF.java index 6636bc1..1138e73 100644 --- a/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostBinaryClassifierUDTF.java +++ b/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostBinaryClassifierUDTF.java @@ -27,8 +27,7 @@ import org.apache.hadoop.hive.ql.metadata.HiveException; * A XGBoost binary classification and the document is as follows; - * https://github.com/dmlc/xgboost/tree/master/demo/binary_classification */ -@Description( - name = "train_xgboost_classifier", +@Description(name = "train_xgboost_classifier", value = "_FUNC_(string[] features, double target [, string options]) - Returns a relation consisting of <string model_id, array<byte> pred_model>") public final class XGBoostBinaryClassifierUDTF extends XGBoostUDTF { http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostMulticlassClassifierUDTF.java ---------------------------------------------------------------------- diff --git a/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostMulticlassClassifierUDTF.java b/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostMulticlassClassifierUDTF.java index 62ede2c..7bdb16f 100644 --- a/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostMulticlassClassifierUDTF.java +++ b/xgboost/src/main/java/hivemall/xgboost/classification/XGBoostMulticlassClassifierUDTF.java @@ -31,8 +31,7 @@ import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector; * A XGBoost multiclass classification and the document is as follows; - * https://github.com/dmlc/xgboost/tree/master/demo/multiclass_classification */ -@Description( - name = "train_multiclass_xgboost_classifier", +@Description(name = "train_multiclass_xgboost_classifier", value = "_FUNC_(string[] features, double target [, string options]) - Returns a relation consisting of <string model_id, array<byte> pred_model>") public final class XGBoostMulticlassClassifierUDTF extends XGBoostUDTF { @@ -61,8 +60,8 @@ public final class XGBoostMulticlassClassifierUDTF extends XGBoostUDTF { if (cli.hasOption("num_class")) { int _num_class = Integer.valueOf(cli.getOptionValue("num_class")); if (_num_class < 2) { - throw new UDFArgumentException("num_class must be greater than 1: " - + _num_class); + throw new UDFArgumentException( + "num_class must be greater than 1: " + _num_class); } params.put("num_class", _num_class); } http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/xgboost/src/main/java/hivemall/xgboost/regression/XGBoostRegressionUDTF.java ---------------------------------------------------------------------- diff --git a/xgboost/src/main/java/hivemall/xgboost/regression/XGBoostRegressionUDTF.java b/xgboost/src/main/java/hivemall/xgboost/regression/XGBoostRegressionUDTF.java index 3a7aec6..5d38afe 100644 --- a/xgboost/src/main/java/hivemall/xgboost/regression/XGBoostRegressionUDTF.java +++ b/xgboost/src/main/java/hivemall/xgboost/regression/XGBoostRegressionUDTF.java @@ -27,8 +27,7 @@ import hivemall.xgboost.XGBoostUDTF; * A XGBoost regression and the document is as follows; - * https://github.com/dmlc/xgboost/tree/master/demo/regression */ -@Description( - name = "train_xgboost_regr", +@Description(name = "train_xgboost_regr", value = "_FUNC_(string[] features, double target [, string options]) - Returns a relation consisting of <string model_id, array<byte> pred_model>") public final class XGBoostRegressionUDTF extends XGBoostUDTF { http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostMulticlassPredictUDTF.java ---------------------------------------------------------------------- diff --git a/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostMulticlassPredictUDTF.java b/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostMulticlassPredictUDTF.java index d3e1bcb..b80f95a 100644 --- a/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostMulticlassPredictUDTF.java +++ b/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostMulticlassPredictUDTF.java @@ -32,8 +32,7 @@ import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectIn import hivemall.utils.lang.Preconditions; -@Description( - name = "xgboost_multiclass_predict", +@Description(name = "xgboost_multiclass_predict", value = "_FUNC_(string rowid, string[] features, string model_id, array<byte> pred_model [, string options]) " + "- Returns a prediction result as (string rowid, string label, float probability)") public final class XGBoostMulticlassPredictUDTF extends hivemall.xgboost.XGBoostPredictUDTF { http://git-wip-us.apache.org/repos/asf/incubator-hivemall/blob/c4036695/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostPredictUDTF.java ---------------------------------------------------------------------- diff --git a/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostPredictUDTF.java b/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostPredictUDTF.java index df5498d..85d0fb1 100644 --- a/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostPredictUDTF.java +++ b/xgboost/src/main/java/hivemall/xgboost/tools/XGBoostPredictUDTF.java @@ -32,8 +32,7 @@ import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory; import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector; import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory; -@Description( - name = "xgboost_predict", +@Description(name = "xgboost_predict", value = "_FUNC_(string rowid, string[] features, string model_id, array<byte> pred_model [, string options]) " + "- Returns a prediction result as (string rowid, float predicted)") public final class XGBoostPredictUDTF extends hivemall.xgboost.XGBoostPredictUDTF {
