Use a more concise and human readable format.

sql/core/src/test/resources/sql-tests/results/number-format.sql.out

@@ -0,0 +1,50 @@
+-- Automatically generated by org.apache.spark.sql.SQLQueryTestSuite
+-- Number of queries: 4
+
+
+-- !query 0
+select 1, -1
+-- !query 0 schema
+int, int
+-- !query 0 output
++---+----+
+|  1|(-1)|
++---+----+
+|  1|  -1|
++---+----+
+
+
+-- !query 1
+select 2147483648, -2147483649
+-- !query 1 schema
+bigint, bigint
+-- !query 1 output
++----------+-------------+
+|2147483648|(-2147483649)|
++----------+-------------+
+|2147483648|  -2147483649|
++----------+-------------+
+
+
+-- !query 2
+select 9223372036854775808, -9223372036854775809
+-- !query 2 schema
+decimal(19,0), decimal(19,0)
+-- !query 2 output
++-------------------+----------------------+
+|9223372036854775808|(-9223372036854775809)|
++-------------------+----------------------+
+|9223372036854775808|  -9223372036854775809|
++-------------------+----------------------+
+
+
+-- !query 3
+select 0.3, -0.8, .5, -.18
+-- !query 3 schema
+decimal(1,1), decimal(1,1), decimal(1,1), decimal(2,2)
+-- !query 3 output
++---+------+---+-------+
+|0.3|(-0.8)|0.5|(-0.18)|
++---+------+---+-------+
+|0.3|  -0.8|0.5|  -0.18|
++---+------+---+-------+

sql/core/src/test/scala/org/apache/spark/sql/SQLQueryTestSuite.scala

@@ -42,11 +42,28 @@ import org.apache.spark.sql.test.SharedSQLContext
  * For example:
  * {{{
  *   -- this is a comment
- *   select 1 + 2;
+ *   select 1, -1;
  *   select current_date;
  * }}}
  *
- * Result files are encoded as XMLs.
+ * The format for golden result files look roughly like:
+ * {{{
+ *   -- some header information
+ *
+ *   -- !query 0
+ *   select 1, -1
+ *   -- !query 0 schema
+ *   int, int
+ *   -- !query 0 output
+ *   +---+----+
+ *   |  1|(-1)|
+ *   +---+----+
+ *   |  1|  -1|
+ *   +---+----+
+ *
+ *   -- !query 1
+ *   ...
+ * }}}
  */
 class SQLQueryTestSuite extends QueryTest with SharedSQLContext {
 
@@ -68,15 +85,14 @@ class SQLQueryTestSuite extends QueryTest with SharedSQLContext {
 
   /** A single SQL query's output. */
   private case class QueryOutput(sql: String, schema: String, output: String) {
-    def toXML: String = {
-      // We are explicitly not using multi-line string due to stripMargin removing |s,
-      // and not using XML interpolation because there is no simple way to indent outputs nicely
-      // (scala.xml.PrettyPrinter has issue with tabs).
-      "<query>\n" +
-      s"  <sql><![CDATA[$sql]]></sql>\n" +
-      s"  <schema><![CDATA[$schema]]></schema>\n" +
-      s"  <output><![CDATA[\n$output]]></output>\n" +
-      s"</query>"
+    def toString(queryIndex: Int): String = {
+      // We are explicitly not using multi-line string due to stripMargin removing "|" in output.
+      s"-- !query $queryIndex\n" +
+        sql + "\n" +
+        s"-- !query $queryIndex schema\n" +
+        schema + "\n" +
+         s"-- !query $queryIndex output\n" +
+        output
     }
   }
 
@@ -96,7 +112,9 @@ class SQLQueryTestSuite extends QueryTest with SharedSQLContext {
 
     // List of SQL queries to run
     val queries: Seq[String] = {
-      val cleaned = input.split("\n").filterNot(_.matches("--.*(?<=[^\\\\]);")).mkString("\n")
+      // val cleaned = input.split("\n").filterNot(_.matches("--.*(?<=[^\\\\]);")).mkString("\n")
+      val cleaned = input.split("\n").filterNot(_.startsWith("--")).mkString("\n")
+      // note: this is not a robust way to split queries using semicolon, but works for now.
       cleaned.split("(?<=[^\\\\]);").map(_.trim).filterNot(q => q == "").toSeq
     }
 
@@ -111,27 +129,26 @@ class SQLQueryTestSuite extends QueryTest with SharedSQLContext {
     }
 
     if (regenerateGoldenFiles) {
-      // If generate golden file flag is on, create the golden file.
-      // Again, we are explicitly not using multi-line string due to stripMargin removing |s,
-      // and not using XML interpolation because there is no simple way to indent outputs nicely
-      // (scala.xml.PrettyPrinter has issue with tabs).
-      val xmlOutput = {
-        "<testcase>\n" +
-        "<!-- Automatically generated by ${getClass.getName} -->\n" +
-        outputs.map(_.toXML).mkString("\n") +
-        "\n</testcase>\n"
+      // Again, we are explicitly not using multi-line string due to stripMargin removing "|".
+      val goldenOutput = {
+        s"-- Automatically generated by ${getClass.getName}\n" +
+        s"-- Number of queries: ${outputs.size}\n\n\n" +
+        outputs.zipWithIndex.map{case (qr, i) => qr.toString(i)}.mkString("\n\n\n") + "\n"
       }
-      stringToFile(new File(testCase.resultFile), xmlOutput)
+      stringToFile(new File(testCase.resultFile), goldenOutput)
     }
 
     // Read back the golden file.
     val expectedOutputs: Seq[QueryOutput] = {
-      val xml = scala.xml.XML.loadString(fileToString(new File(testCase.resultFile)))
-      (xml \ "query").map { q =>
+      val goldenOutput = fileToString(new File(testCase.resultFile))
+      val segments = goldenOutput.split("-- !query.+\n")
+      assert(segments.size == outputs.size * 3 + 1)  // each query has 3 segments, plus the header
+      Seq.tabulate(outputs.size) { i =>
         QueryOutput(
-          sql = (q \ "sql").text,
-          schema = (q \ "schema").text,
-          output = (q \ "output").text.trim)
+          sql = segments(i * 3 + 1).trim,
+          schema = segments(i * 3 + 2).trim,
+          output = segments(i * 3 + 3).trim
+        )
       }
     }
 
@@ -149,7 +166,7 @@ class SQLQueryTestSuite extends QueryTest with SharedSQLContext {
 
   private def listTestCases(): Seq[TestCase] = {
     listFilesRecursively(new File(inputFilePath)).map { file =>
-      val resultFile = file.getAbsolutePath.replace(inputFilePath, goldenFilePath) + ".xml"
+      val resultFile = file.getAbsolutePath.replace(inputFilePath, goldenFilePath) + ".out"
       TestCase(file.getName, file.getAbsolutePath, resultFile)
     }
   }