fix: broken test after change in python-bigquery

Following this change:
googleapis/python-bigquery#1408
the SchemaField __init__ and __repr__ method have evolved.

This broke some tests.

Author: FurcyPin

bigquery_frame/data_diff/dataframe_comparator.py

@@ -106,9 +106,9 @@ def _schema_to_string(
         c1 STRING (nullable)
         c2 INTEGER (nullable)
         >>> schema = [
-        ...     SchemaField('id', 'INTEGER', 'NULLABLE', 'An id', (), None),
-        ...     SchemaField('c1', 'STRING', 'REQUIRED', 'A string column', (), None),
-        ...     SchemaField('c2', 'INTEGER', 'NULLABLE', 'An int column', (), None)
+        ...     SchemaField(name='id', field_type='INTEGER', mode='NULLABLE', description='An id'),
+        ...     SchemaField(name='c1', field_type='STRING', mode='REQUIRED', description='A string column'),
+        ...     SchemaField(name='c2', field_type='INTEGER', mode='NULLABLE', description='An int column')
         ... ]
         >>> print('\\n'.join(df_comparator._schema_to_string(schema, include_nullable=True, include_metadata=True)))
         id INTEGER (nullable) An id

bigquery_frame/dataframe.py

@@ -77,7 +77,7 @@ def schema_to_simple_string(schema: List[SchemaField]):
     >>> bq = BigQueryBuilder()
     >>> df = bq.sql('SELECT 1 as id, STRUCT(1 as a, [STRUCT(2 as c, 3 as d)] as b, [4, 5] as e) as s')
     >>> print(df.schema)  # noqa: E501
-    [SchemaField('id', 'INTEGER', 'NULLABLE', None, (), None), SchemaField('s', 'RECORD', 'NULLABLE', None, (SchemaField('a', 'INTEGER', 'NULLABLE', None, (), None), SchemaField('b', 'RECORD', 'REPEATED', None, (SchemaField('c', 'INTEGER', 'NULLABLE', None, (), None), SchemaField('d', 'INTEGER', 'NULLABLE', None, (), None)), None), SchemaField('e', 'INTEGER', 'REPEATED', None, (), None)), None)]
+    [SchemaField('id', 'INTEGER', 'NULLABLE', None, None, (), None), SchemaField('s', 'RECORD', 'NULLABLE', None, None, (SchemaField('a', 'INTEGER', 'NULLABLE', None, None, (), None), SchemaField('b', 'RECORD', 'REPEATED', None, None, (SchemaField('c', 'INTEGER', 'NULLABLE', None, None, (), None), SchemaField('d', 'INTEGER', 'NULLABLE', None, None, (), None)), None), SchemaField('e', 'INTEGER', 'REPEATED', None, None, (), None)), None)]
     >>> schema_to_simple_string(df.schema)
     'id:INTEGER,s:STRUCT<a:INTEGER,b:ARRAY<STRUCT<c:INTEGER,d:INTEGER>>,e:ARRAY<INTEGER>>'
 

bigquery_frame/transformations_impl/flatten_schema.py

@@ -57,11 +57,11 @@ def flatten_schema_field(prefix: str, schema_field: SchemaField, nullable: bool)
                     mode = "REPEATED"
             return [
                 SchemaField(
-                    prefix,
-                    schema_field.field_type,
-                    mode,
-                    schema_field.description,
-                    schema_field.fields,
+                    name=prefix,
+                    field_type=schema_field.field_type,
+                    mode=mode,
+                    description=schema_field.description,
+                    fields=schema_field.fields,
                 )
             ]
 

tests/test_dataframe.py

@@ -14,10 +14,10 @@ def test_createOrReplaceTempView(bq: BigQueryBuilder):
     df2 = bq.sql("""SELECT * FROM pokedex""")
 
     expected = [
-        SchemaField("id", "INTEGER", "NULLABLE", None, (), None),
-        SchemaField("name", "STRING", "NULLABLE", None, (), None),
-        SchemaField("types", "STRING", "REPEATED", None, (), None),
-        SchemaField("other_col", "INTEGER", "NULLABLE", None, (), None),
+        SchemaField(name="id", field_type="INTEGER", mode="NULLABLE"),
+        SchemaField(name="name", field_type="STRING", mode="NULLABLE"),
+        SchemaField(name="types", field_type="STRING", mode="REPEATED"),
+        SchemaField(name="other_col", field_type="INTEGER", mode="NULLABLE"),
     ]
 
     assert df2.schema == expected
@@ -29,7 +29,7 @@ def test_createOrReplaceTempView_with_reserved_keyword_alias(bq: BigQueryBuilder
     bq.sql("""SELECT 1 as id""").createOrReplaceTempView("all")
     df = bq.table("all")
 
-    expected = [SchemaField("id", "INTEGER", "NULLABLE", None, (), None)]
+    expected = [SchemaField(name="id", field_type="INTEGER", mode="NULLABLE")]
 
     assert df.schema == expected
 
@@ -66,10 +66,10 @@ def test_createOrReplaceTempTable(bq: BigQueryBuilder):
     df2 = bq.sql("""SELECT * FROM pokedex""")
 
     expected = [
-        SchemaField("id", "INTEGER", "NULLABLE", None, (), None),
-        SchemaField("name", "STRING", "NULLABLE", None, (), None),
-        SchemaField("types", "STRING", "REPEATED", None, (), None),
-        SchemaField("other_col", "INTEGER", "NULLABLE", None, (), None),
+        SchemaField(name="id", field_type="INTEGER", mode="NULLABLE"),
+        SchemaField(name="name", field_type="STRING", mode="NULLABLE"),
+        SchemaField(name="types", field_type="STRING", mode="REPEATED"),
+        SchemaField(name="other_col", field_type="INTEGER", mode="NULLABLE"),
     ]
 
     assert df2.schema == expected
@@ -81,7 +81,7 @@ def test_createOrReplaceTempTable_with_reserved_keyword_alias(bq: BigQueryBuilde
     bq.sql("""SELECT 1 as id""").createOrReplaceTempTable("all")
     df = bq.table("all")
 
-    expected = [SchemaField("id", "INTEGER", "NULLABLE", None, (), None)]
+    expected = [SchemaField(name="id", field_type="INTEGER", mode="NULLABLE")]
 
     assert df.schema == expected
 
@@ -98,8 +98,8 @@ def test_select(bq: BigQueryBuilder):
     df = bq.sql("""SELECT 1 as c1, 2 as c2""").select("c1", "c2").select(["c1", "c2"])
 
     expected = [
-        SchemaField("c1", "INTEGER", "NULLABLE", None, (), None),
-        SchemaField("c2", "INTEGER", "NULLABLE", None, (), None),
+        SchemaField(name="c1", field_type="INTEGER", mode="NULLABLE"),
+        SchemaField(name="c2", field_type="INTEGER", mode="NULLABLE"),
     ]
 
     assert df.schema == expected
@@ -117,10 +117,10 @@ def test_bare_strings(bq: BigQueryBuilder):
     df5 = df4.withColumn("name", "LOWER(name)", replace=True)
 
     expected = [
-        SchemaField("id", "INTEGER", "NULLABLE", None, (), None),
-        SchemaField("name", "STRING", "NULLABLE", None, (), None),
-        SchemaField("types", "STRING", "REPEATED", None, (), None),
-        SchemaField("nb_types", "INTEGER", "NULLABLE", None, (), None),
+        SchemaField(name="id", field_type="INTEGER", mode="NULLABLE"),
+        SchemaField(name="name", field_type="STRING", mode="NULLABLE"),
+        SchemaField(name="types", field_type="STRING", mode="REPEATED"),
+        SchemaField(name="nb_types", field_type="INTEGER", mode="NULLABLE"),
     ]
 
     assert df5.schema == expected
@@ -176,9 +176,9 @@ def test_drop(bq: BigQueryBuilder):
     df2 = df.drop("other_col", "col_that_does_not_exists")
 
     expected = [
-        SchemaField("id", "INTEGER", "NULLABLE", None, (), None),
-        SchemaField("name", "STRING", "NULLABLE", None, (), None),
-        SchemaField("types", "STRING", "REPEATED", None, (), None),
+        SchemaField(name="id", field_type="INTEGER", mode="NULLABLE"),
+        SchemaField(name="name", field_type="STRING", mode="NULLABLE"),
+        SchemaField(name="types", field_type="STRING", mode="REPEATED"),
     ]
 
     assert df2.schema == expected