ATLAS-294 Select queries(ex: from DB select DB.name) response contains column…

addons/hive-bridge/src/test/java/org/apache/atlas/hive/hook/HiveHookIT.java

@@ -297,7 +297,7 @@ public class HiveHookIT {
         String query = String.format(
                 "%s as t where tableName = '%s', db where name = '%s' and clusterName = '%s'" + " select t",
                 HiveDataTypes.HIVE_TABLE.getName(), tableName.toLowerCase(), dbName.toLowerCase(), CLUSTER_NAME);
-        return assertEntityIsRegistered(query);
+        return assertEntityIsRegistered(query, "t");
     }
 
     private String assertDatabaseIsRegistered(String dbName) throws Exception {
@@ -322,10 +322,10 @@ public class HiveHookIT {
                                + "db where name = '%s' and clusterName = '%s' select p", typeName, value,
                             tableName.toLowerCase(), dbName.toLowerCase(), CLUSTER_NAME);
 
-        assertEntityIsRegistered(dslQuery);
+        assertEntityIsRegistered(dslQuery, "p");
     }
 
-    private String assertEntityIsRegistered(final String query) throws Exception {
+    private String assertEntityIsRegistered(final String query, String... arg) throws Exception {
         waitFor(2000, new Predicate() {
             @Override
             public boolean evaluate() throws Exception {
@@ -334,6 +334,8 @@ public class HiveHookIT {
             }
         });
 
+        String column = (arg.length > 0) ? arg[0] : "_col_0";
+
         JSONArray results = dgiCLient.search(query);
         JSONObject row = results.getJSONObject(0);
         if (row.has("__guid")) {
@@ -341,7 +343,7 @@ public class HiveHookIT {
         } else if (row.has("$id$")) {
             return row.getJSONObject("$id$").getString("id");
         } else {
-            return row.getJSONObject("_col_0").getString("id");
+            return row.getJSONObject(column).getString("id");
         }
     }
 

release-log.txt

@@ -9,6 +9,7 @@ ATLAS-54 Rename configs in hive hook (shwethags)
 ATLAS-3 Mixed Index creation fails with Date types (sumasai via shwethags)
 
 ALL CHANGES:
+ATLAS-294 Select queries(ex: from DB select DB.name) response contains column names as "_col_x" instead of the actual names requested in the query.(thiyag via sumasai)
 ATLAS-297 KafkaNotificationTest.testSendReceiveMessage fails when atlas-server is running on the same machine (yhemanth via shwethags)
 ATLAS-306 change javadoc generation from 'package' to 'site' phase(jspeidel via sumasai)
 ATLAS-289 updateEntity does not remove existing edge for multiplicity-one reference (dkantor via shwethags)

repository/src/main/scala/org/apache/atlas/query/Expressions.scala

@@ -676,13 +676,11 @@ object Expressions {
         override def toString = s"$child where $condExpr"
     }
 
-    val GEN_COL_ALIAS_PREFIX = "_col"
-
     case class SelectExpression(child: Expression, selectList: List[Expression]) extends Expression {
         val children = List(child) ::: selectList
         lazy val selectListWithAlias = selectList.zipWithIndex map {
             case (s: AliasExpression, _) => s
-            case (x, i) => new AliasExpression(x, s"${GEN_COL_ALIAS_PREFIX}_$i")
+            case (x, i) => new AliasExpression(x, s"${x}")
         }
 
         lazy val dataType = {

repository/src/main/scala/org/apache/atlas/query/QueryParser.scala

@@ -78,7 +78,7 @@ trait ExpressionUtils {
     def select(input: Expression, s: List[(Expression, Option[String])]) = {
         val selList = s.map { t =>
             t._2 match {
-                case None => t._1
+                case None => t._1.as(s"${t._1}")
                 case _ => t._1.as(t._2.get)
             }
         }

repository/src/test/scala/org/apache/atlas/query/GremlinTest.scala

@@ -313,12 +313,12 @@ class GremlinTest extends BaseGremlinTest {
     val r = QueryProcessor.evaluate(_class("DB").where(id("name").`=`(string("Reporting"))).
       select(id("name"), id("owner")), g, gp)
     validateJson(r, """{
-                      |    "query": "DB where (name = \"Reporting\") as _src1 select _src1.name as _col_0, _src1.owner as _col_1",
+                      |    "query": "DB where (name = \"Reporting\") as _src1 select _src1.name as _src1.name, _src1.owner as _src1.owner",
                       |    "dataType": {
                       |        "typeName": "__tempQueryResultStruct1",
                       |        "attributeDefinitions": [
                       |            {
-                      |                "name": "_col_0",
+                      |                "name": "_src1.name",
                       |                "dataTypeName": "string",
                       |                "multiplicity": {
                       |                    "lower": 0,
@@ -331,7 +331,7 @@ class GremlinTest extends BaseGremlinTest {
                       |                "reverseAttributeName": null
                       |            },
                       |            {
-                      |                "name": "_col_1",
+                      |                "name": "_src1.owner",
                       |                "dataTypeName": "string",
                       |                "multiplicity": {
                       |                    "lower": 0,
@@ -348,8 +348,8 @@ class GremlinTest extends BaseGremlinTest {
                       |    "rows": [
                       |        {
                       |            "$typeName$": "__tempQueryResultStruct1",
-                      |            "_col_1": "Jane BI",
-                      |            "_col_0": "Reporting"
+                      |            "_src1.owner": "Jane BI",
+                      |            "_src1.name": "Reporting"
                       |        }
                       |    ]
                       |}""".stripMargin);
@@ -777,7 +777,7 @@ class GremlinTest extends BaseGremlinTest {
   @Test def testArith {
     val r = QueryProcessor.evaluate(_class("DB").where(id("name").`=`(string("Reporting"))).
       select(id("name"), id("createTime") + int(1)), g, gp)
-    validateJson(r, "{\n  \"query\":\"DB where (name = \\\"Reporting\\\") as _src1 select _src1.name as _col_0, (_src1.createTime + 1) as _col_1\",\n  \"dataType\":{\n    \"typeName\":\"__tempQueryResultStruct3\",\n    \"attributeDefinitions\":[\n      {\n        \"name\":\"_col_0\",\n        \"dataTypeName\":\"string\",\n        \"multiplicity\":{\n          \"lower\":0,\n          \"upper\":1,\n          \"isUnique\":false\n        },\n        \"isComposite\":false,\n        \"isUnique\":false,\n        \"isIndexable\":true,\n        \"reverseAttributeName\":null\n      },\n      {\n        \"name\":\"_col_1\",\n        \"dataTypeName\":\"int\",\n        \"multiplicity\":{\n          \"lower\":0,\n          \"upper\":1,\n          \"isUnique\":false\n        },\n        \"isComposite\":false,\n        \"isUnique\":false,\n        \"isIndexable\":true,\n        \"reverseAttributeName\":null\n      }\n    ]\n  },\n  \"rows\":[\n    {\n      \"$typeName$\":\"__tempQueryResultStruct3\",\n      \"_col_1\":1501,\n      \"_col_0\":\"Reporting\"\n    }\n  ]\n}")
+    validateJson(r, "{\n  \"query\":\"DB where (name = \\\"Reporting\\\") as _src1 select _src1.name as _src1.name, (_src1.createTime + 1) as (_src1.createTime + 1)\",\n  \"dataType\":{\n    \"typeName\":\"__tempQueryResultStruct3\",\n    \"attributeDefinitions\":[\n      {\n        \"name\":\"_src1.name\",\n        \"dataTypeName\":\"string\",\n        \"multiplicity\":{\n          \"lower\":0,\n          \"upper\":1,\n          \"isUnique\":false\n        },\n        \"isComposite\":false,\n        \"isUnique\":false,\n        \"isIndexable\":true,\n        \"reverseAttributeName\":null\n      },\n      {\n        \"name\":\"(_src1.createTime + 1)\",\n        \"dataTypeName\":\"int\",\n        \"multiplicity\":{\n          \"lower\":0,\n          \"upper\":1,\n          \"isUnique\":false\n        },\n        \"isComposite\":false,\n        \"isUnique\":false,\n        \"isIndexable\":true,\n        \"reverseAttributeName\":null\n      }\n    ]\n  },\n  \"rows\":[\n    {\n      \"$typeName$\":\"__tempQueryResultStruct3\",\n      \"(_src1.createTime + 1)\":1501,\n      \"_src1.name\":\"Reporting\"\n    }\n  ]\n}")
   }
 
   @Test def testComparisonLogical {
@@ -906,12 +906,12 @@ class GremlinTest extends BaseGremlinTest {
       " db where name = 'Reporting' and clusterName = 'test' select p").right.get
     val r = QueryProcessor.evaluate(e, g, gp)
     validateJson(r, """{
-                      |  "query":"Partition as p where (values = [\"2015-01-01\"]) table where (name = \"sales_fact_daily_mv\") db where (name = \"Reporting\") and (clusterName = \"test\") as _src1 select p as _col_0",
+                      |  "query":"Partition as p where (values = [\"2015-01-01\"]) table where (name = \"sales_fact_daily_mv\") db where (name = \"Reporting\") and (clusterName = \"test\") as _src1 select p as p",
                       |  "dataType":{
                       |    "typeName":"__tempQueryResultStruct2",
                       |    "attributeDefinitions":[
                       |      {
-                      |        "name":"_col_0",
+                      |        "name":"p",
                       |        "dataTypeName":"Partition",
                       |        "multiplicity":{
                       |          "lower":0,
@@ -928,7 +928,7 @@ class GremlinTest extends BaseGremlinTest {
                       |  "rows":[
                       |    {
                       |      "$typeName$":"__tempQueryResultStruct2",
-                      |      "_col_0":{
+                      |      "p":{
                       |        "$typeName$":"Partition",
                       |        "version":0
                       |      }
@@ -945,12 +945,12 @@ class GremlinTest extends BaseGremlinTest {
     val r = QueryProcessor.evaluate(e, g, gp)
     validateJson(r,
       """{
-        |  "query":"Partition as p where (values = [\"2015-01-01\"]) table where (name = \"sales_fact_daily_mv\") db where (name = \"Reporting\") and (clusterName = \"test\") as _src1 select p.values as _col_0",
+        |  "query":"Partition as p where (values = [\"2015-01-01\"]) table where (name = \"sales_fact_daily_mv\") db where (name = \"Reporting\") and (clusterName = \"test\") as _src1 select p.values as p.values",
         |  "dataType":{
         |    "typeName":"__tempQueryResultStruct2",
         |    "attributeDefinitions":[
         |  {
-        |    "name":"_col_0",
+        |    "name":"p.values",
         |    "dataTypeName":"array<string>",
         |    "multiplicity":{
         |    "lower":0,
@@ -967,7 +967,7 @@ class GremlinTest extends BaseGremlinTest {
         |  "rows":[
         |  {
         |    "$typeName$":"__tempQueryResultStruct2",
-        |    "_col_0":[
+        |    "p.values":[
         |    "2015-01-01"
         |    ]
         |  }