path typecasting

[kakoni]

I have json data that looks like this;

{
 "attributes": [{
          "attributeName": "Abbreviation",
          "attributeValue": ["Pt-ACTH-R1"]
        }, {
          "attributeName": "ShortName",
          "attributeValue": ["Pt-Adrenokortikotropiini-koe, lyhyt"]
        }, {
          "attributeName": "LongName",
          "attributeValue": ["Pt-Adrenokortikotropiini-koe, lyhyt"]
        }, {
          "attributeName": "Erikoisala",
          "attributeValue": ["K"]
        }, {
          "attributeName": "Langt_namn",
          "attributeValue": ["Pt-Adrenokortikotropin-test, kort"]
        }, {
          "attributeName": "Yksikko",
          "attributeValue": ["Lomake"]
        }]
}

I wish to get attribate name/value pairs into columns;

    columns:
      - {name: "abreviation", type: string, path: "$.attributes[?(@.attributeName== 'Abbreviation')].attributeValue[0]"}

This almost works ok, but gives me string value that has brackets;

["Pt-ACTH-R1"]

Wonder if there is a away to extract just the string value?

[hiroyuki-sato]

I'm the maintainer of this plugin, but I'm not familiar JSONPath 😄

Have you ever tried the following?

in:
  type: file
  path_prefix: ./test.txt
  parser:
    type: jsonpath
    columns:
    - {name: "abreviation", type: string, path: "$.attributeValue[0]"}
    root: "$.attributes[?(@.attributeName== 'Abbreviation')]"
out:
  type: stdout

embulk preview shows the following output.

+--------------------+
| abreviation:string |
+--------------------+
|         Pt-ACTH-R1 |
+--------------------+

[kakoni]

Right, thanks. In my use case, I need to pick all of those values, meaning something like

      - {name: "abreviation", type: string, path: "$.attributes[?(@.attributeName== 'Abbreviation')].attributeValue[0]"}
      - {name: "short_name", type: string, path: "$.attributes[?(@.attributeName== 'ShortName')].attributeValue[0]"}
...

So using root isn't an option.

[hiroyuki-sato]

Hello, @kakoni

embulk-parser-jsonpath expects to specify a root path for input data.
I think it raises an error if you don't use root attributes.
Can you parse your data even incompletely?

I think it requires implementing that feature.

Could you show me your configuration?