[SPARK-8738] [SQL] [PYSPARK] capture SQL AnalysisException in Python API

Capture the AnalysisException in SQL, hide the long java stack trace, only show the error message.

cc rxin

Author: Davies Liu <davies@databricks.com>

Closes #7135 from davies/ananylis and squashes the following commits:

dad7ae7 [Davies Liu] add comment
ec0c0e8 [Davies Liu] Update utils.py
cdd7edd [Davies Liu] add doc
7b044c2 [Davies Liu] fix python 3
f84d3bd [Davies Liu] capture SQL AnalysisException in Python API

1 files changed, 7 insertions, 0 deletions

diff --git a/python/pyspark/sql/tests.py b/python/pyspark/sql/tests.py
index 34f397d0ff..5af2ce09bc 100644
--- a/python/pyspark/sql/tests.py
+++ b/python/pyspark/sql/tests.py
@@ -46,6 +46,7 @@ from pyspark.sql.types import UserDefinedType, _infer_type
 from pyspark.tests import ReusedPySparkTestCase
 from pyspark.sql.functions import UserDefinedFunction
 from pyspark.sql.window import Window
+from pyspark.sql.utils import AnalysisException
 
 
 class UTC(datetime.tzinfo):
@@ -847,6 +848,12 @@ class SQLTests(ReusedPySparkTestCase):
         self.assertEqual(row.age, 10)
         self.assertEqual(row.height, None)
 
+    def test_capture_analysis_exception(self):
+        self.assertRaises(AnalysisException, lambda: self.sqlCtx.sql("select abc"))
+        self.assertRaises(AnalysisException, lambda: self.df.selectExpr("a + b"))
+        # RuntimeException should not be captured
+        self.assertRaises(py4j.protocol.Py4JJavaError, lambda: self.sqlCtx.sql("abc"))
+
 
 class HiveContextSQLTests(ReusedPySparkTestCase):