Skip to content

spark_frame.schema_utils

This module contains methods useful for manipulating DataFrame schemas.

schema_from_json(json_string: str) -> StructType

Parses the given json string representing a Spark :class:StructType.

Only schema representing StructTypes can be parsed, this means that schema_from_json(schema_to_json(data_type)) will crash if data_type is not a StructType.

Parameters:

Name Type Description Default
json_string str

A string representation of a DataFrame schema.

 required

Returns:

Type Description
StructType

A StructType object representing the DataFrame schema

Examples:

>>> schema_from_json('''{"fields":[
...     {"metadata":{},"name":"a","nullable":true,"type":"byte"},
...     {"metadata":{},"name":"b","nullable":true,"type":"decimal(16,8)"}
... ],"type":"struct"}''')
StructType([StructField('a', ByteType(), True), StructField('b', DecimalType(16,8), True)])
>>> schema_from_json('''{"fields":[
...     {"metadata":{},"name":"a","nullable":true,"type":"double"},
...     {"metadata":{},"name":"b","nullable":true,"type":"string"}
... ],"type":"struct"}''')
StructType([StructField('a', DoubleType(), True), StructField('b', StringType(), True)])
>>> schema_from_json('''{"fields":[
... {"metadata":{},"name":"a","nullable":true,"type":{
...     "containsNull":true,"elementType":"short","type":"array"
... }}
... ],"type":"struct"}''')
StructType([StructField('a', ArrayType(ShortType(), True), True)])

Error cases:

>>> schema_from_json('"integer"')
Traceback (most recent call last):
  ...
TypeError: string indices must be integers
>>> schema_from_json('''{"keyType":"string","type":"map",
... "valueContainsNull":true,"valueType":"string"}''')
Traceback (most recent call last):
  ...
KeyError: 'fields'
Source code in spark_frame/schema_utils.py 
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
def schema_from_json(json_string: str) -> StructType:
    """Parses the given json string representing a Spark :class:`StructType`.

    Only schema representing StructTypes can be parsed, this means that
    `schema_from_json(schema_to_json(data_type))` will crash if `data_type` is not a StructType.

    Args:
        json_string: A string representation of a DataFrame schema.

    Returns:
        A StructType object representing the DataFrame schema

    Examples:
        >>> schema_from_json('''{"fields":[
        ...     {"metadata":{},"name":"a","nullable":true,"type":"byte"},
        ...     {"metadata":{},"name":"b","nullable":true,"type":"decimal(16,8)"}
        ... ],"type":"struct"}''')
        StructType([StructField('a', ByteType(), True), StructField('b', DecimalType(16,8), True)])
        >>> schema_from_json('''{"fields":[
        ...     {"metadata":{},"name":"a","nullable":true,"type":"double"},
        ...     {"metadata":{},"name":"b","nullable":true,"type":"string"}
        ... ],"type":"struct"}''')
        StructType([StructField('a', DoubleType(), True), StructField('b', StringType(), True)])
        >>> schema_from_json('''{"fields":[
        ... {"metadata":{},"name":"a","nullable":true,"type":{
        ...     "containsNull":true,"elementType":"short","type":"array"
        ... }}
        ... ],"type":"struct"}''')
        StructType([StructField('a', ArrayType(ShortType(), True), True)])

        **Error cases:**

        >>> schema_from_json('"integer"') # doctest: +IGNORE_EXCEPTION_DETAIL
        Traceback (most recent call last):
          ...
        TypeError: string indices must be integers
        >>> schema_from_json('''{"keyType":"string","type":"map",
        ... "valueContainsNull":true,"valueType":"string"}''') # doctest: +IGNORE_EXCEPTION_DETAIL
        Traceback (most recent call last):
          ...
        KeyError: 'fields'

    """
    return StructType.fromJson(json.loads(json_string))

schema_from_simple_string(schema_string: str) -> DataType

Parses the given data type string to a :class:DataType. The data type string format equals pyspark.sql.types.DataType.simpleString, except that the top level struct type can omit the struct<>. This method requires the SparkSession to have already been instantiated.

Parameters:

Name Type Description Default
schema_string str

A simpleString representing a DataFrame schema.

 required

Returns:

Type Description
DataType

A DataType object representing the DataFrame schema.

Raises:

Type Description
AssertionError

If no SparkContext has been instantiated first.

Examples:

>>> from pyspark.sql import SparkSession
>>> spark = SparkSession.builder.appName("doctest").getOrCreate()
>>> schema_from_simple_string("int ")
IntegerType()
>>> schema_from_simple_string("INT ")
IntegerType()
>>> schema_from_simple_string("a: byte, b: decimal(  16 , 8   ) ")
StructType([StructField('a', ByteType(), True), StructField('b', DecimalType(16,8), True)])
>>> schema_from_simple_string("a DOUBLE, b STRING")
StructType([StructField('a', DoubleType(), True), StructField('b', StringType(), True)])
>>> schema_from_simple_string("a: array< short>")
StructType([StructField('a', ArrayType(ShortType(), True), True)])
>>> schema_from_simple_string(" map<string , string > ")
MapType(StringType(), StringType(), True)

Error cases:

>>> schema_from_simple_string("blabla")
Traceback (most recent call last):
  ...
pyspark.sql.utils.ParseException:...
>>> schema_from_simple_string("a: int,")
Traceback (most recent call last):
  ...
pyspark.sql.utils.ParseException:...
>>> schema_from_simple_string("array<int")
Traceback (most recent call last):
  ...
pyspark.sql.utils.ParseException:...
>>> schema_from_simple_string("map<int, boolean>>")
Traceback (most recent call last):
  ...
pyspark.sql.utils.ParseException:...
Source code in spark_frame/schema_utils.py 
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
def schema_from_simple_string(schema_string: str) -> DataType:
    """Parses the given data type string to a :class:`DataType`. The data type string format equals
    [pyspark.sql.types.DataType.simpleString][], except that the top level struct type can omit
    the ``struct<>``.
    This method requires the SparkSession to have already been instantiated.

    Args:
        schema_string: A simpleString representing a DataFrame schema.

    Returns:
        A DataType object representing the DataFrame schema.

    Raises:
        AssertionError: If no SparkContext has been instantiated first.

    Examples:
        >>> from pyspark.sql import SparkSession
        >>> spark = SparkSession.builder.appName("doctest").getOrCreate()
        >>> schema_from_simple_string("int ")
        IntegerType()
        >>> schema_from_simple_string("INT ")
        IntegerType()
        >>> schema_from_simple_string("a: byte, b: decimal(  16 , 8   ) ")
        StructType([StructField('a', ByteType(), True), StructField('b', DecimalType(16,8), True)])
        >>> schema_from_simple_string("a DOUBLE, b STRING")
        StructType([StructField('a', DoubleType(), True), StructField('b', StringType(), True)])
        >>> schema_from_simple_string("a: array< short>")
        StructType([StructField('a', ArrayType(ShortType(), True), True)])
        >>> schema_from_simple_string(" map<string , string > ")
        MapType(StringType(), StringType(), True)

        **Error cases:**

        >>> schema_from_simple_string("blabla") # doctest: +IGNORE_EXCEPTION_DETAIL
        Traceback (most recent call last):
          ...
        pyspark.sql.utils.ParseException:...
        >>> schema_from_simple_string("a: int,") # doctest: +IGNORE_EXCEPTION_DETAIL
        Traceback (most recent call last):
          ...
        pyspark.sql.utils.ParseException:...
        >>> schema_from_simple_string("array<int") # doctest: +IGNORE_EXCEPTION_DETAIL
        Traceback (most recent call last):
          ...
        pyspark.sql.utils.ParseException:...
        >>> schema_from_simple_string("map<int, boolean>>") # doctest: +IGNORE_EXCEPTION_DETAIL
        Traceback (most recent call last):
          ...
        pyspark.sql.utils.ParseException:...

    """
    sc = SparkContext._active_spark_context  # noqa: SLF001
    assert_true(sc is not None, "No SparkContext has been instantiated yet")
    return _parse_datatype_string(schema_string)

schema_to_json(schema: DataType) -> str

Convert the given datatype into a json string.

Parameters:

Name Type Description Default
schema DataType

A DataFrame schema.

 required

Returns:

Type Description
str

A single-line json string representing the DataFrame schema.

Examples:

>>> from pyspark.sql.types import *
>>> schema_to_json(IntegerType())
'"integer"'
>>> schema_to_json(StructType([StructField('a', ByteType(), True), StructField('b', DecimalType(16,8), True)]))
'{"fields":[{"metadata":{},"name":"a","nullable":true,"type":"byte"},{"metadata":{},"name":"b","nullable":true,"type":"decimal(16,8)"}],"type":"struct"}'
>>> schema_to_json(StructType([StructField('a', DoubleType(), True), StructField('b', StringType(), True)]))
'{"fields":[{"metadata":{},"name":"a","nullable":true,"type":"double"},{"metadata":{},"name":"b","nullable":true,"type":"string"}],"type":"struct"}'
>>> schema_to_json(StructType([StructField('a', ArrayType(ShortType(), True), True)]))
'{"fields":[{"metadata":{},"name":"a","nullable":true,"type":{"containsNull":true,"elementType":"short","type":"array"}}],"type":"struct"}'
>>> schema_to_json(MapType(StringType(), StringType(), True))
'{"keyType":"string","type":"map","valueContainsNull":true,"valueType":"string"}'
Source code in spark_frame/schema_utils.py 
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
def schema_to_json(schema: DataType) -> str:
    """Convert the given datatype into a json string.

    Args:
        schema: A DataFrame schema.

    Returns:
        A single-line json string representing the DataFrame schema.

    Examples:
        >>> from pyspark.sql.types import *
        >>> schema_to_json(IntegerType())
        '"integer"'
        >>> schema_to_json(StructType([StructField('a', ByteType(), True), StructField('b', DecimalType(16,8), True)]))
        '{"fields":[{"metadata":{},"name":"a","nullable":true,"type":"byte"},{"metadata":{},"name":"b","nullable":true,"type":"decimal(16,8)"}],"type":"struct"}'
        >>> schema_to_json(StructType([StructField('a', DoubleType(), True), StructField('b', StringType(), True)]))
        '{"fields":[{"metadata":{},"name":"a","nullable":true,"type":"double"},{"metadata":{},"name":"b","nullable":true,"type":"string"}],"type":"struct"}'
        >>> schema_to_json(StructType([StructField('a', ArrayType(ShortType(), True), True)]))
        '{"fields":[{"metadata":{},"name":"a","nullable":true,"type":{"containsNull":true,"elementType":"short","type":"array"}}],"type":"struct"}'
        >>> schema_to_json(MapType(StringType(), StringType(), True))
        '{"keyType":"string","type":"map","valueContainsNull":true,"valueType":"string"}'

    """
    return schema.json()

schema_to_pretty_json(schema: DataType) -> str

Convert the given datatype into a pretty (indented) json string.

Parameters:

Name Type Description Default
schema DataType

A DataFrame schema.

 required

Returns:

Type Description
str

A multi-line indented json string representing the DataFrame schema.

Examples:

>>> from pyspark.sql.types import *
>>> print(schema_to_pretty_json(IntegerType()))
"integer"
>>> print(schema_to_pretty_json(StructType([StructField('a', ArrayType(ShortType(), True), True)])))
{
  "fields": [
    {
      "metadata": {},
      "name": "a",
      "nullable": true,
      "type": {
        "containsNull": true,
        "elementType": "short",
        "type": "array"
      }
    }
  ],
  "type": "struct"
}
>>> print(schema_to_pretty_json(MapType(StringType(), StringType(), True)))
{
  "keyType": "string",
  "type": "map",
  "valueContainsNull": true,
  "valueType": "string"
}

:param schema: :return:

Source code in spark_frame/schema_utils.py 
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
def schema_to_pretty_json(schema: DataType) -> str:
    """Convert the given datatype into a pretty (indented) json string.

    Args:
        schema: A DataFrame schema.

    Returns:
        A multi-line indented json string representing the DataFrame schema.

    Examples:
        >>> from pyspark.sql.types import *
        >>> print(schema_to_pretty_json(IntegerType()))
        "integer"
        >>> print(schema_to_pretty_json(StructType([StructField('a', ArrayType(ShortType(), True), True)])))
        {
          "fields": [
            {
              "metadata": {},
              "name": "a",
              "nullable": true,
              "type": {
                "containsNull": true,
                "elementType": "short",
                "type": "array"
              }
            }
          ],
          "type": "struct"
        }
        >>> print(schema_to_pretty_json(MapType(StringType(), StringType(), True)))
        {
          "keyType": "string",
          "type": "map",
          "valueContainsNull": true,
          "valueType": "string"
        }

    :param schema:
    :return:
    """
    schema_dict = json.loads(schema.json())
    return json.dumps(schema_dict, indent=2, sort_keys=True)

schema_to_simple_string(schema: DataType) -> str

Convert the given datatype into a simple sql string. This method is equivalent to calling schema.simpleString() directly.

Parameters:

Name Type Description Default
schema DataType

A DataFrame schema.

 required

Returns:

Type Description
str

A simpleString representing the DataFrame schema.

Examples:

>>> from pyspark.sql.types import *
>>> schema_to_simple_string(IntegerType())
'int'
>>> schema_to_simple_string(StructType([
...     StructField('a', ByteType(), True),
...     StructField('b', DecimalType(16,8), True)
... ]))
'struct<a:tinyint,b:decimal(16,8)>'
>>> schema_to_simple_string(StructType([
...     StructField('a', DoubleType(), True),
...     StructField('b', StringType(), True)
... ]))
'struct<a:double,b:string>'
>>> schema_to_simple_string(StructType([StructField('a', ArrayType(ShortType(), True), True)]))
'struct<a:array<smallint>>'
>>> schema_to_simple_string(MapType(StringType(), StringType(), True))
'map<string,string>'
Source code in spark_frame/schema_utils.py 
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
def schema_to_simple_string(schema: DataType) -> str:
    """Convert the given datatype into a simple sql string.
    This method is equivalent to calling [`schema.simpleString()`][pyspark.sql.types.DataType.simpleString] directly.

    Args:
        schema: A DataFrame schema.

    Returns:
        A simpleString representing the DataFrame schema.

    Examples:
        >>> from pyspark.sql.types import *
        >>> schema_to_simple_string(IntegerType())
        'int'
        >>> schema_to_simple_string(StructType([
        ...     StructField('a', ByteType(), True),
        ...     StructField('b', DecimalType(16,8), True)
        ... ]))
        'struct<a:tinyint,b:decimal(16,8)>'
        >>> schema_to_simple_string(StructType([
        ...     StructField('a', DoubleType(), True),
        ...     StructField('b', StringType(), True)
        ... ]))
        'struct<a:double,b:string>'
        >>> schema_to_simple_string(StructType([StructField('a', ArrayType(ShortType(), True), True)]))
        'struct<a:array<smallint>>'
        >>> schema_to_simple_string(MapType(StringType(), StringType(), True))
        'map<string,string>'
    """
    return schema.simpleString()