diff options
author | Ilya Ganelin <ilya.ganelin@capitalone.com> | 2015-06-29 14:15:15 -0700 |
---|---|---|
committer | Reynold Xin <rxin@databricks.com> | 2015-06-29 14:15:15 -0700 |
commit | f6fc254ec4ce5f103d45da6d007b4066ce751236 (patch) | |
tree | e00615856783d0cdbdf70da13c4b2903ef09803b /python | |
parent | 27ef85451cd237caa7016baa69957a35ab365aa8 (diff) | |
download | spark-f6fc254ec4ce5f103d45da6d007b4066ce751236.tar.gz spark-f6fc254ec4ce5f103d45da6d007b4066ce751236.tar.bz2 spark-f6fc254ec4ce5f103d45da6d007b4066ce751236.zip |
[SPARK-8056][SQL] Design an easier way to construct schema for both Scala and Python
I've added functionality to create new StructType similar to how we add parameters to a new SparkContext.
I've also added tests for this type of creation.
Author: Ilya Ganelin <ilya.ganelin@capitalone.com>
Closes #6686 from ilganeli/SPARK-8056B and squashes the following commits:
27c1de1 [Ilya Ganelin] Rename
467d836 [Ilya Ganelin] Removed from_string in favor of _parse_Datatype_json_value
5fef5a4 [Ilya Ganelin] Updates for type parsing
4085489 [Ilya Ganelin] Style errors
3670cf5 [Ilya Ganelin] added string to DataType conversion
8109e00 [Ilya Ganelin] Fixed error in tests
41ab686 [Ilya Ganelin] Fixed style errors
e7ba7e0 [Ilya Ganelin] Moved some python tests to tests.py. Added cleaner handling of null data type and added test for correctness of input format
15868fa [Ilya Ganelin] Fixed python errors
b79b992 [Ilya Ganelin] Merge remote-tracking branch 'upstream/master' into SPARK-8056B
a3369fc [Ilya Ganelin] Fixing space errors
e240040 [Ilya Ganelin] Style
bab7823 [Ilya Ganelin] Constructor error
73d4677 [Ilya Ganelin] Style
4ed00d9 [Ilya Ganelin] Fixed default arg
67df57a [Ilya Ganelin] Removed Foo
04cbf0c [Ilya Ganelin] Added comments for single object
0484d7a [Ilya Ganelin] Restored second method
6aeb740 [Ilya Ganelin] Style
689e54d [Ilya Ganelin] Style
f497e9e [Ilya Ganelin] Got rid of old code
e3c7a88 [Ilya Ganelin] Fixed doctest failure
a62ccde [Ilya Ganelin] Style
966ac06 [Ilya Ganelin] style checks
dabb7e6 [Ilya Ganelin] Added Python tests
a3f4152 [Ilya Ganelin] added python bindings and better comments
e6e536c [Ilya Ganelin] Added extra space
7529a2e [Ilya Ganelin] Fixed formatting
d388f86 [Ilya Ganelin] Fixed small bug
c4e3bf5 [Ilya Ganelin] Reverted to using parse. Updated parse to support long
d7634b6 [Ilya Ganelin] Reverted to fromString to properly support types
22c39d5 [Ilya Ganelin] replaced FromString with DataTypeParser.parse. Replaced empty constructor initializing a null to have it instead create a new array to allow appends to it.
faca398 [Ilya Ganelin] [SPARK-8056] Replaced default argument usage. Updated usage and code for DataType.fromString
1acf76e [Ilya Ganelin] Scala style
e31c674 [Ilya Ganelin] Fixed bug in test
8dc0795 [Ilya Ganelin] Added tests for creation of StructType object with new methods
fdf7e9f [Ilya Ganelin] [SPARK-8056] Created add methods to facilitate building new StructType objects.
Diffstat (limited to 'python')
-rw-r--r-- | python/pyspark/sql/tests.py | 29 | ||||
-rw-r--r-- | python/pyspark/sql/types.py | 52 |
2 files changed, 77 insertions, 4 deletions
diff --git a/python/pyspark/sql/tests.py b/python/pyspark/sql/tests.py index ffee43a94b..34f397d0ff 100644 --- a/python/pyspark/sql/tests.py +++ b/python/pyspark/sql/tests.py @@ -516,6 +516,35 @@ class SQLTests(ReusedPySparkTestCase): self.assertEqual([Row(a=2, b=1, c=3), Row(a=4, b=1, c=4)], df.filter(df.a.between(df.b, df.c)).collect()) + def test_struct_type(self): + from pyspark.sql.types import StructType, StringType, StructField + struct1 = StructType().add("f1", StringType(), True).add("f2", StringType(), True, None) + struct2 = StructType([StructField("f1", StringType(), True), + StructField("f2", StringType(), True, None)]) + self.assertEqual(struct1, struct2) + + struct1 = StructType().add("f1", StringType(), True).add("f2", StringType(), True, None) + struct2 = StructType([StructField("f1", StringType(), True)]) + self.assertNotEqual(struct1, struct2) + + struct1 = (StructType().add(StructField("f1", StringType(), True)) + .add(StructField("f2", StringType(), True, None))) + struct2 = StructType([StructField("f1", StringType(), True), + StructField("f2", StringType(), True, None)]) + self.assertEqual(struct1, struct2) + + struct1 = (StructType().add(StructField("f1", StringType(), True)) + .add(StructField("f2", StringType(), True, None))) + struct2 = StructType([StructField("f1", StringType(), True)]) + self.assertNotEqual(struct1, struct2) + + # Catch exception raised during improper construction + try: + struct1 = StructType().add("name") + self.assertEqual(1, 0) + except ValueError: + self.assertEqual(1, 1) + def test_save_and_load(self): df = self.df tmpPath = tempfile.mkdtemp() diff --git a/python/pyspark/sql/types.py b/python/pyspark/sql/types.py index 932686e5e4..ae9344e610 100644 --- a/python/pyspark/sql/types.py +++ b/python/pyspark/sql/types.py @@ -355,8 +355,7 @@ class StructType(DataType): This is the data type representing a :class:`Row`. """ - - def __init__(self, fields): + def __init__(self, fields=None): """ >>> struct1 = StructType([StructField("f1", StringType(), True)]) >>> struct2 = StructType([StructField("f1", StringType(), True)]) @@ -368,8 +367,53 @@ class StructType(DataType): >>> struct1 == struct2 False """ - assert all(isinstance(f, DataType) for f in fields), "fields should be a list of DataType" - self.fields = fields + if not fields: + self.fields = [] + else: + self.fields = fields + assert all(isinstance(f, StructField) for f in fields),\ + "fields should be a list of StructField" + + def add(self, field, data_type=None, nullable=True, metadata=None): + """ + Construct a StructType by adding new elements to it to define the schema. The method accepts + either: + a) A single parameter which is a StructField object. + b) Between 2 and 4 parameters as (name, data_type, nullable (optional), + metadata(optional). The data_type parameter may be either a String or a DataType object + + >>> struct1 = StructType().add("f1", StringType(), True).add("f2", StringType(), True, None) + >>> struct2 = StructType([StructField("f1", StringType(), True),\ + StructField("f2", StringType(), True, None)]) + >>> struct1 == struct2 + True + >>> struct1 = StructType().add(StructField("f1", StringType(), True)) + >>> struct2 = StructType([StructField("f1", StringType(), True)]) + >>> struct1 == struct2 + True + >>> struct1 = StructType().add("f1", "string", True) + >>> struct2 = StructType([StructField("f1", StringType(), True)]) + >>> struct1 == struct2 + True + + :param field: Either the name of the field or a StructField object + :param data_type: If present, the DataType of the StructField to create + :param nullable: Whether the field to add should be nullable (default True) + :param metadata: Any additional metadata (default None) + :return: a new updated StructType + """ + if isinstance(field, StructField): + self.fields.append(field) + else: + if isinstance(field, str) and data_type is None: + raise ValueError("Must specify DataType if passing name of struct_field to create.") + + if isinstance(data_type, str): + data_type_f = _parse_datatype_json_value(data_type) + else: + data_type_f = data_type + self.fields.append(StructField(field, data_type_f, nullable, metadata)) + return self def simpleString(self): return 'struct<%s>' % (','.join(f.simpleString() for f in self.fields)) |