使用 Scala/Spark 读取 Json 文件

Question

I am trying to read a json, example:

{
  "id1": {
    "a": "7",
    "b": "3",
    "c": "10",
    "d": "10",
    "e": "15",
    "f": "11",
    "g": "2",
    "h": "7",
    "i": "5",
    "j": "14"
  },
  "id2": {
    "a": "3",
    "b": "7",
    "c": "12",
    "d": "4",
    "e": "10",
    "f": "4",
    "g": "13",
    "h": "4",
    "i": "1",
    "j": "13"
  },
  "id3": {
    "a": "10",
    "b": "6",
    "c": "1",
    "d": "1",
    "e": "13",
    "f": "12",
    "g": "9",
    "h": "6",
    "i": "7",
    "j": "4"
  }
}

when I process it with spark.read.json ("file.json") it returns a single record with this format:

+-----------------------------------+---------------------------------------+---------------------------------+
|id1                                |id2                                    |id3                              |
+-----------------------------------+---------------------------------------+---------------------------------+
|{7, 3, 10, 10, 15, 11, 2, 7, 5, 14}|{322539, 7, 12, 4, 10, 4, 13, 4, 1, 13}|{10, 6, 1, 1, 13, 12, 9, 6, 7, 4}|
+-----------------------------------+---------------------------------------+---------------------------------+

I would like the result to be something like this when processing the file:

+-----+------+------+------+------+------+------+------+------+------+-------+
| id  | col1 | col2 | col3 | col4 | col5 | col6 | col7 | col8 | col9 | col10 |
+-----+------+------+------+------+------+------+------+------+------+-------+
| id1 |  7   |  3   |  10  |  10  |  15  |  11  |  2   |  7   |  5   |  14   |
+-----+------+------+------+------+------+------+------+------+------+-------+
| id2 |  3   |  7   |  12  |  4   |  10  |  4   |  13  |  4   |  1   |  13   |
+-----+------+------+------+------+------+------+------+------+------+-------+
| id3 |  10  |  6   |  1   |  1   |  13  |  12  |  9   |  6   |  7   |  4    |
+-----+------+------+------+------+------+------+------+------+------+-------+

some simple and fast way?

thanks

Answer 1

Are you able to modify your input JSON file ? If so, making it an array of JS Objects would do the job with spark.read.json() :

[    
    {
      "id": "id1",  
      "a": "7",
      "b": "3",
      "c": "10",
      "d": "10",
      "e": "15",
      "f": "11",
      "g": "2",
      "h": "7",
      "i": "5",
      "j": "14"
    },
    ...
]