DRILL-8296: Possible type mismatch bug in SplunkBatchReader (#2700)
diff --git a/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkBatchReader.java b/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkBatchReader.java
index cba9e3d..11a0ce9 100644
--- a/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkBatchReader.java
+++ b/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkBatchReader.java
@@ -35,7 +35,6 @@
import org.apache.drill.exec.record.metadata.SchemaBuilder;
import org.apache.drill.exec.record.metadata.TupleMetadata;
import org.apache.drill.exec.store.base.filter.ExprNode;
-import org.apache.drill.exec.util.Utilities;
import org.apache.drill.exec.vector.ValueVector;
import org.apache.drill.exec.vector.accessor.ScalarWriter;
import org.apache.drill.shaded.guava.com.google.common.base.Stopwatch;
@@ -214,28 +213,6 @@
}
/**
- * Checks to see whether the query is a star query. For our purposes, the star query is
- * anything that contains only the ** and the SPECIAL_COLUMNS which are not projected.
- * @return true if it is a star query, false if not.
- */
- private boolean isStarQuery() {
- if (Utilities.isStarQuery(projectedColumns)) {
- return true;
- }
-
- List<SplunkUtils.SPECIAL_FIELDS> specialFields = Arrays.asList(SplunkUtils.SPECIAL_FIELDS.values());
-
- for (SchemaPath path: projectedColumns) {
- if (path.nameEquals("**")) {
- return true;
- } else {
- return specialFields.contains(path.getAsNamePart());
- }
- }
- return false;
- }
-
- /**
* Determines whether a field is a Splunk multifield.
* @param fieldValue The field to be tested
* @return True if a multifield, false if not.
@@ -312,9 +289,9 @@
filters.remove("sourcetype");
}
- // Pushdown the selected fields for non star queries.
- if (! isStarQuery()) {
- builder.addField(projectedColumns);
+ // Add projected columns, skipping star and specials.
+ for (SchemaPath projectedColumn: projectedColumns) {
+ builder.addField(projectedColumn.getAsUnescapedPath());
}
// Apply filters
diff --git a/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkQueryBuilder.java b/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkQueryBuilder.java
index c13c08e..0d83fdd 100644
--- a/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkQueryBuilder.java
+++ b/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkQueryBuilder.java
@@ -23,7 +23,6 @@
import org.apache.drill.exec.store.base.filter.RelOp;
import org.apache.drill.shaded.guava.com.google.common.base.Strings;
-import java.util.List;
import java.util.Map;
public class SplunkQueryBuilder {
@@ -36,7 +35,7 @@
private String query;
private String sourceTypes;
- private String fieldList;
+ private String concatenatedFields;
private String filters;
private int sourcetypeCount;
private int limit;
@@ -69,35 +68,21 @@
* Splunk accepts arguments in the format | fields foo, bar, car. This function adds these fields to the query.
* As an error preventative measure, this function will ignore ** from Drill.
* @param field The field to be added to the query
+ * @return true if the field was added, false if it was skipped
*/
- public void addField (String field) {
+ public boolean addField(String field) {
// Double Star fields cause errors and we will not add to the field list
- if (field.equalsIgnoreCase("**") || Strings.isNullOrEmpty(field)) {
- return;
+ if (SchemaPath.DYNAMIC_STAR.equals(field) || SplunkUtils.SPECIAL_FIELDS.includes(field)) {
+ return false;
}
// Case for first field
- if (fieldList == null) {
- this.fieldList = field;
+ if (concatenatedFields == null) {
+ this.concatenatedFields = field;
} else {
- this.fieldList += "," + field;
+ this.concatenatedFields += "," + field;
}
- }
-
- /**
- * Creates the field list of r
- * As an error preventative measure, this function will ignore ** from Drill.
- * @param columnList SchemaPath of columns to be added to the field list
- */
- public void addField (List<SchemaPath> columnList) {
- for (SchemaPath column : columnList) {
- String columnName = column.getAsUnescapedPath();
- if (columnName.equalsIgnoreCase("**") || Strings.isNullOrEmpty(columnName)) {
- continue;
- } else {
- addField(columnName);
- }
- }
+ return true;
}
/**
@@ -153,7 +138,7 @@
String value = ((ExprNode.ColRelOpConstNode)filter.getValue()).value.value.toString();
// Ignore special cases
- if (SplunkUtils.isSpecialField(fieldName)) {
+ if (SplunkUtils.SPECIAL_FIELDS.includes(fieldName)) {
// Sourcetypes are a special case and can be added via filter pushdown
if (fieldName.equalsIgnoreCase("sourcetype")) {
addSourceType(value);
@@ -226,8 +211,8 @@
}
// Add fields
- if (! Strings.isNullOrEmpty(fieldList)) {
- query += " | fields " + fieldList;
+ if (! Strings.isNullOrEmpty(concatenatedFields)) {
+ query += " | fields " + concatenatedFields;
}
// Add limit
@@ -236,10 +221,10 @@
}
// Add table logic. This tells Splunk to return the data in tabular form rather than the mess that it usually generates
- if ( Strings.isNullOrEmpty(fieldList)) {
- fieldList = "*";
+ if ( Strings.isNullOrEmpty(concatenatedFields)) {
+ concatenatedFields = "*";
}
- query += " | table " + fieldList;
+ query += " | table " + concatenatedFields;
return query;
}
diff --git a/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkUtils.java b/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkUtils.java
index f34f64d..a859f2c 100644
--- a/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkUtils.java
+++ b/contrib/storage-splunk/src/main/java/org/apache/drill/exec/store/splunk/SplunkUtils.java
@@ -18,6 +18,8 @@
package org.apache.drill.exec.store.splunk;
+import java.util.Arrays;
+
public class SplunkUtils {
/**
* These are special fields that alter the queries sent to Splunk.
@@ -46,19 +48,15 @@
this.field = field;
}
- }
-
- /**
- * Indicates whether the field in question is a special field and should be pushed down to the query or not.
- * @param unknownField The field to be pushed down
- * @return true if the field is a special field, false if not.
- */
- public static boolean isSpecialField (String unknownField) {
- for (SPECIAL_FIELDS specialFieldName : SPECIAL_FIELDS.values()) {
- if (specialFieldName.field.equalsIgnoreCase(unknownField)) {
- return true;
- }
+ /**
+ * Indicates whether the field in question is a special field and should be
+ * pushed down to the query or not.
+ * @param unknownField The field to be pushed down
+ * @return true if the field is a special field, false if not.
+ */
+ public static boolean includes(String field) {
+ return Arrays.stream(SplunkUtils.SPECIAL_FIELDS.values())
+ .anyMatch(special -> field.equals(special.name()));
}
- return false;
}
}
diff --git a/contrib/storage-splunk/src/test/java/org/apache/drill/exec/store/splunk/SplunkTestSplunkUtils.java b/contrib/storage-splunk/src/test/java/org/apache/drill/exec/store/splunk/SplunkTestSplunkUtils.java
index f60dafb..a4a2e64 100644
--- a/contrib/storage-splunk/src/test/java/org/apache/drill/exec/store/splunk/SplunkTestSplunkUtils.java
+++ b/contrib/storage-splunk/src/test/java/org/apache/drill/exec/store/splunk/SplunkTestSplunkUtils.java
@@ -31,17 +31,17 @@
@Test
public void testIsSpecialField() {
- assertTrue(SplunkUtils.isSpecialField("sourcetype"));
- assertTrue(SplunkUtils.isSpecialField("earliestTime"));
- assertTrue(SplunkUtils.isSpecialField("latestTime"));
- assertTrue(SplunkUtils.isSpecialField("spl"));
+ assertTrue(SplunkUtils.SPECIAL_FIELDS.includes("sourcetype"));
+ assertTrue(SplunkUtils.SPECIAL_FIELDS.includes("earliestTime"));
+ assertTrue(SplunkUtils.SPECIAL_FIELDS.includes("latestTime"));
+ assertTrue(SplunkUtils.SPECIAL_FIELDS.includes("spl"));
}
@Test
public void testIsNotSpecialField() {
- assertFalse(SplunkUtils.isSpecialField("bob"));
- assertFalse(SplunkUtils.isSpecialField("ip_address"));
- assertFalse(SplunkUtils.isSpecialField("mac_address"));
- assertFalse(SplunkUtils.isSpecialField("latest_Time"));
+ assertFalse(SplunkUtils.SPECIAL_FIELDS.includes("bob"));
+ assertFalse(SplunkUtils.SPECIAL_FIELDS.includes("ip_address"));
+ assertFalse(SplunkUtils.SPECIAL_FIELDS.includes("mac_address"));
+ assertFalse(SplunkUtils.SPECIAL_FIELDS.includes("latest_Time"));
}
}
