Merge pull request apache#1533 from metamx/extraCheckGroupByDimExtrac…

…tion Add more unit tests for group by
mistercrunch · Jul 16, 2015 · 3a0793a · 3a0793a
2 parents 4b1bd29 + 7d0b77c
commit 3a0793a
Showing 1 changed file with 229 additions and 0 deletions.
diff --git a/processing/src/test/java/io/druid/query/groupby/GroupByQueryRunnerTest.java b/processing/src/test/java/io/druid/query/groupby/GroupByQueryRunnerTest.java
@@ -318,6 +318,235 @@ public void testGroupByWithRebucketRename()
     TestHelper.assertExpectedObjects(expectedResults, results, "");
   }
 
+
+
+  @Test
+  public void testGroupByWithSimpleRenameRetainMissingNonInjective()
+  {
+    Map<String, String> map = new HashMap<>();
+    map.put("automotive", "automotive0");
+    map.put("business", "business0");
+    map.put("entertainment", "entertainment0");
+    map.put("health", "health0");
+    map.put("mezzanine", "mezzanine0");
+    map.put("news", "news0");
+    map.put("premium", "premium0");
+    map.put("technology", "technology0");
+    map.put("travel", "travel0");
+    GroupByQuery query = GroupByQuery
+        .builder()
+        .setDataSource(QueryRunnerTestHelper.dataSource)
+        .setQuerySegmentSpec(QueryRunnerTestHelper.firstToThird)
+        .setDimensions(
+            Lists.<DimensionSpec>newArrayList(
+                new ExtractionDimensionSpec(
+                    "quality", "alias", new LookupExtractionFn(new MapLookupExtractor(map), true, null, false), null
+                )
+            )
+        )
+        .setAggregatorSpecs(
+            Arrays.asList(
+                QueryRunnerTestHelper.rowsCount,
+                new LongSumAggregatorFactory("idx", "index")
+            )
+        )
+        .setGranularity(QueryRunnerTestHelper.dayGran)
+        .build();
+
+    List<Row> expectedResults = Arrays.asList(
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "automotive0", "rows", 1L, "idx", 135L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "business0", "rows", 1L, "idx", 118L),
+        GroupByQueryRunnerTestHelper.createExpectedRow(
+            "2011-04-01",
+            "alias",
+            "entertainment0",
+            "rows",
+            1L,
+            "idx",
+            158L
+        ),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "health0", "rows", 1L, "idx", 120L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "mezzanine0", "rows", 3L, "idx", 2870L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "news0", "rows", 1L, "idx", 121L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "premium0", "rows", 3L, "idx", 2900L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "technology0", "rows", 1L, "idx", 78L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "travel0", "rows", 1L, "idx", 119L),
+
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "automotive0", "rows", 1L, "idx", 147L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "business0", "rows", 1L, "idx", 112L),
+        GroupByQueryRunnerTestHelper.createExpectedRow(
+            "2011-04-02",
+            "alias",
+            "entertainment0",
+            "rows",
+            1L,
+            "idx",
+            166L
+        ),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "health0", "rows", 1L, "idx", 113L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "mezzanine0", "rows", 3L, "idx", 2447L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "news0", "rows", 1L, "idx", 114L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "premium0", "rows", 3L, "idx", 2505L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "technology0", "rows", 1L, "idx", 97L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "travel0", "rows", 1L, "idx", 126L)
+    );
+
+    Iterable<Row> results = GroupByQueryRunnerTestHelper.runQuery(factory, runner, query);
+    TestHelper.assertExpectedObjects(expectedResults, results, "");
+  }
+
+
+  @Test
+  public void testGroupByWithSimpleRenameRetainMissing()
+  {
+    Map<String, String> map = new HashMap<>();
+    map.put("automotive", "automotive0");
+    map.put("business", "business0");
+    map.put("entertainment", "entertainment0");
+    map.put("health", "health0");
+    map.put("mezzanine", "mezzanine0");
+    map.put("news", "news0");
+    map.put("premium", "premium0");
+    map.put("technology", "technology0");
+    map.put("travel", "travel0");
+    GroupByQuery query = GroupByQuery
+        .builder()
+        .setDataSource(QueryRunnerTestHelper.dataSource)
+        .setQuerySegmentSpec(QueryRunnerTestHelper.firstToThird)
+        .setDimensions(
+            Lists.<DimensionSpec>newArrayList(
+                new ExtractionDimensionSpec(
+                    "quality", "alias", new LookupExtractionFn(new MapLookupExtractor(map), true, null, true), null
+                )
+            )
+        )
+        .setAggregatorSpecs(
+            Arrays.asList(
+                QueryRunnerTestHelper.rowsCount,
+                new LongSumAggregatorFactory("idx", "index")
+            )
+        )
+        .setGranularity(QueryRunnerTestHelper.dayGran)
+        .build();
+
+    List<Row> expectedResults = Arrays.asList(
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "automotive0", "rows", 1L, "idx", 135L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "business0", "rows", 1L, "idx", 118L),
+        GroupByQueryRunnerTestHelper.createExpectedRow(
+            "2011-04-01",
+            "alias",
+            "entertainment0",
+            "rows",
+            1L,
+            "idx",
+            158L
+        ),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "health0", "rows", 1L, "idx", 120L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "mezzanine0", "rows", 3L, "idx", 2870L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "news0", "rows", 1L, "idx", 121L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "premium0", "rows", 3L, "idx", 2900L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "technology0", "rows", 1L, "idx", 78L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "travel0", "rows", 1L, "idx", 119L),
+
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "automotive0", "rows", 1L, "idx", 147L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "business0", "rows", 1L, "idx", 112L),
+        GroupByQueryRunnerTestHelper.createExpectedRow(
+            "2011-04-02",
+            "alias",
+            "entertainment0",
+            "rows",
+            1L,
+            "idx",
+            166L
+        ),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "health0", "rows", 1L, "idx", 113L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "mezzanine0", "rows", 3L, "idx", 2447L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "news0", "rows", 1L, "idx", 114L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "premium0", "rows", 3L, "idx", 2505L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "technology0", "rows", 1L, "idx", 97L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "travel0", "rows", 1L, "idx", 126L)
+    );
+
+    Iterable<Row> results = GroupByQueryRunnerTestHelper.runQuery(factory, runner, query);
+    TestHelper.assertExpectedObjects(expectedResults, results, "");
+  }
+
+
+  @Test
+  public void testGroupByWithSimpleRenameAndMissingString()
+  {
+    Map<String, String> map = new HashMap<>();
+    map.put("automotive", "automotive0");
+    map.put("business", "business0");
+    map.put("entertainment", "entertainment0");
+    map.put("health", "health0");
+    map.put("mezzanine", "mezzanine0");
+    map.put("news", "news0");
+    map.put("premium", "premium0");
+    map.put("technology", "technology0");
+    map.put("travel", "travel0");
+    GroupByQuery query = GroupByQuery
+        .builder()
+        .setDataSource(QueryRunnerTestHelper.dataSource)
+        .setQuerySegmentSpec(QueryRunnerTestHelper.firstToThird)
+        .setDimensions(
+            Lists.<DimensionSpec>newArrayList(
+                new ExtractionDimensionSpec(
+                    "quality", "alias", new LookupExtractionFn(new MapLookupExtractor(map), false, "MISSING", true), null
+                )
+            )
+        )
+        .setAggregatorSpecs(
+            Arrays.asList(
+                QueryRunnerTestHelper.rowsCount,
+                new LongSumAggregatorFactory("idx", "index")
+            )
+        )
+        .setGranularity(QueryRunnerTestHelper.dayGran)
+        .build();
+
+    List<Row> expectedResults = Arrays.asList(
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "automotive0", "rows", 1L, "idx", 135L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "business0", "rows", 1L, "idx", 118L),
+        GroupByQueryRunnerTestHelper.createExpectedRow(
+            "2011-04-01",
+            "alias",
+            "entertainment0",
+            "rows",
+            1L,
+            "idx",
+            158L
+        ),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "health0", "rows", 1L, "idx", 120L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "mezzanine0", "rows", 3L, "idx", 2870L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "news0", "rows", 1L, "idx", 121L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "premium0", "rows", 3L, "idx", 2900L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "technology0", "rows", 1L, "idx", 78L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "travel0", "rows", 1L, "idx", 119L),
+
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "automotive0", "rows", 1L, "idx", 147L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "business0", "rows", 1L, "idx", 112L),
+        GroupByQueryRunnerTestHelper.createExpectedRow(
+            "2011-04-02",
+            "alias",
+            "entertainment0",
+            "rows",
+            1L,
+            "idx",
+            166L
+        ),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "health0", "rows", 1L, "idx", 113L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "mezzanine0", "rows", 3L, "idx", 2447L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "news0", "rows", 1L, "idx", 114L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "premium0", "rows", 3L, "idx", 2505L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "technology0", "rows", 1L, "idx", 97L),
+        GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "travel0", "rows", 1L, "idx", 126L)
+    );
+
+    Iterable<Row> results = GroupByQueryRunnerTestHelper.runQuery(factory, runner, query);
+    TestHelper.assertExpectedObjects(expectedResults, results, "");
+  }
+
   @Test
   public void testGroupByWithSimpleRename()
   {