我有一个由json文件使用以下模式创建的数据框
root
|-- content: string (nullable = true)
|-- conversationId: long (nullable = true)
|-- date: string (nullable = true)
|-- id: long (nullable = true)
|-- lang: string (nullable = true)
|-- likeCount: long (nullable = true)
|-- media: array (nullable = true)
| |-- element: struct (containsNull = true)
| | |-- duration: double (nullable = true)
| | |-- fullUrl: string (nullable = true)
| | |-- previewUrl: string (nullable = true)
| | |-- thumbnailUrl: string (nullable = true)
| | |-- type: string (nullable = true)
| | |-- variants: array (nullable = true)
| | | |-- element: struct (containsNull = true)
| | | | |-- bitrate: long (nullable = true)
| | | | |-- contentType: string (nullable = true)
| | | | |-- url: string (nullable = true)
我要提取单独的数据框
问题是我无法删除新variants
数据框中的media
列。
在media
新数据框中,我不希望media
列中的所有数据,我希望media
对象成为其数据框中的新根,其字段为column ,并且与variants
新数据框相同
这是我的代码:
tweets_df = spark.read.json(jsonFile)
media_df = tweets_df.select(col('media'))
variants_df = media_df.select(col('media.variants'))
# i tried also media_df = media_df.drop('media.variants')
media_df = media_df.drop('variants')
tweets_df = tweets_df.drop('media')
这是结果:
root
|-- variants: array (nullable = true)
| |-- element: array (containsNull = true)
| | |-- element: struct (containsNull = true)
| | | |-- bitrate: long (nullable = true)
| | | |-- contentType: string (nullable = true)
| | | |-- url: string (nullable = true)
root
|-- media: array (nullable = true)
| |-- element: struct (containsNull = true)
| | |-- duration: double (nullable = true)
| | |-- fullUrl: string (nullable = true)
| | |-- previewUrl: string (nullable = true)
| | |-- thumbnailUrl: string (nullable = true)
| | |-- type: string (nullable = true) **IT'S NOT DELETED**
| | |-- variants: array (nullable = true)
| | | |-- element: struct (containsNull = true)
| | | | |-- bitrate: long (nullable = true)
| | | | |-- contentType: string (nullable = true)
| | | | |-- url: string (nullable = true)
root
|-- content: string (nullable = true)
|-- conversationId: long (nullable = true)
|-- date: string (nullable = true)
|-- idOnSite: long (nullable = true)
|-- lang: string (nullable = true)
|-- likeCount: long (nullable = true)