knowledge_share/pySpark点滴.md

+# 将json字符串转化成结构构型
+
+```python
+schema_changes = ArrayType(StructType([
+    StructField('change_date', StringType( ), nullable=True),
+    StructField('change_field', StringType( ), nullable=True),
+    StructField('change_no', IntegerType( ), nullable=True),
+    StructField('change_after', StringType( ), nullable=True),
+    StructField('change_before', StringType( ), nullable=True),
+]))
+
+df_changes = df_ic.filter("changes is not null and changes != ''").select("company_name_digest", from_json("changes", schema_changes).alias('changes'))
+df_changes_e = df_changes.select('company_name_digest', explode('changes'))
+df_changes_e.filter("company_name_digest == 'b6dc6bc9632a20e62c118bd5f3b204d7'").select("company_name_digest", 'col.change_date').show(truncate=False)
+
+```
\ No newline at end of file