How to read double-quoted string values using terraform csvdecode?

August 21, 2023

I have a CSV file that contains the following content:

id,value
123,{"M":{"name_1":{"S":"value_1"}, "name_2":{"S":"value_2"}}}

I’m trying to read this CSV file and create records in DynamoDB the following way:

locals {
  custom_data = csvdecode(file("${path.module}/../custom_data.csv"))
}

resource "aws_dynamodb_table_item" "custom_table_item" {
  for_each = {for row in local.custom_data : row.id => row}

  table_name = aws_dynamodb_table.custom_table.name
  hash_key   = aws_dynamodb_table.custom_table.hash_key

  item = jsonencode({
    "id" : { "S" : each.value.id },
    "value" : jsondecode(each.value.value)
  })

  lifecycle {
    ignore_changes = [item]
  }
}

However, this code doesn’t work and I can’t find any example on how to read the double-quoted values from the CSV file in a way that jsondecode can create the appropriate JSON structure.
Does anyone know how to do that?

>Solution :

Sanitize your csv file:

id,value
123,"{""M"":{""name_1"":{""S"":""value_1""}, ""name_2"":{""S"":""value_2""}}}"

then:

$ terraform-repl
> csvdecode(file("${path.module}/custom_data.csv"))
tolist([
  {
    "id" = "123"
    "value" = "{\"M\":{\"name_1\":{\"S\":\"value_1\"}, \"name_2\":{\"S\":\"value_2\"}}}"
  },
])

> jsondecode(csvdecode(file("${path.module}/custom_data.csv"))[0].value)
{
  "M" = {
    "name_1" = {
      "S" = "value_1"
    }
    "name_2" = {
      "S" = "value_2"
    }
  }
}