Support concat for children node fields

[abdullah2993]

It would be really nice if concat transform could support the ability to concat child columns in parent e.g.

 {
    "database": "postgres",
    "index": "products",
    "nodes": {
      "table": "products",
      "children": [
        {
          "table": "vendors",
          "label": "vendor",
          "columns": [
            "name"
          ],
          "relationship": {
            "variant": "object",
            "type": "one_to_one"
          }
        },
        {
          "table": "line_items",
          "label": "items",
          "columns": [
            "description"
          ],
          "relationship": {
            "variant": "object",
            "type": "one_to_many",
            "foreign_key": {
              "child": [
                "product_id"
              ],
              "parent": [
                "id"
              ]
            }
          }
        }
      ],
      "columns": [
        "number",
        "purchase_order",
        "sku",
        "quantity",
        "stock_date",
        "amount"
      ],
      "transform": {
        "concat": {
          "columns": [
            "number",
            "purchase_order",
            "billing_address",
            "email_addresses",
            "estimate_date",
            "amount",
            "items.description", <------ support this
            "vendor.name" <------ support this
          ],
          "destination": "global_search_field",
          "delimiter": " "
        }
      }
    }
  }

support concating items.description and vendor.name in the given sample

[toluaina]

I get what you mean. That could get complicated in generalised cases though. e.g if the parent had multiple children then you need to specify in index to concat.

[abdullah2993]

I'm not sure if I understand the question.

[carlos-pereira-costoya]

@toluaina It would be very interesting in cases of having a one_by_one relationship type to avoid and filter situations with multiple results. Is this issue still open, or is this other implementation for pgsync in some version snapshot? Is there a repository of plugins where I can find something similar? Thank you very much.

[lxup]

Hi, I wanna ask a similar question why I really need concat for children node fields for translations. I have a table postgres movie which contains :

- id : bigint
- original_title : text

And athis table has a one-to-many relation with movie_translation :

- movieId : bigint
- title : text
- overview : text
- poster_path : url

So I wanna be able to search between title for the corresponding user language and with original_title. Elasticsearch recommend to use one field per language like this title_en, title_fr, overview_en, etc... I need to have this Elasticsearch result :

- id : bigint
- original_title : text
- title_en : text
- title_fr : text
- overview_en : text
- overview_fr : text
- poster_path_en : url
- poster_path_fr : url
-- and for each language available in movie_translation

How can we achieve this with pgsync ? Thx for ur help ! 😁