Should not require {{}} for static values.

Suggestions:

"type": "'video'",
"active": true (or "true"),
"order": 0 (or "0"),
"name": "_.path + '/' + _.name",
"duration": "_.duration / 60"

Create a regex for identifying the pattern table.column. Should probably trim fields before working with them.

Example object:

"bpi": {
  "USD": {
    "code": "USD",
    "symbol": "$",
    "rate": "29,911.4932",
    "description": "United States Dollar",
    "rate_float": 29911.4932
  },
  "GBP": {
    "code": "GBP",
    "symbol": "£",
    "rate": "23,800.9640",
    "description": "British Pound Sterling",
    "rate_float": 23800.964
  },
  "EUR": {
    "code": "EUR",
    "symbol": "€",
    "rate": "27,997.2174",
    "description": "Euro",
    "rate_float": 27997.2174
  }
}

We should be able to iterate through each key as a row and get their values.

Suggestion:

"alias": {
  "_currency": "_.bpi__*__"
},
"tables": {
  "currency": {
    "set": {
      "name": "_currency.code",
      "label": "_currency.description",
      "rate": "_currency.label"
    }
  }
}

It is not necessary to create a temporary table. We can instead directly select values from a CTE.

Example:

with "cte__" as (
  values (1, 'one'),
  (2, 'two'),
  (3, 'three')
) s("id", "name")

Some data parsing might be necessary to insert into tables that are not of type text (not yet tested). Ideally, RDM should be able to run a database introspection to determine the correct column types. Focus on getting text, boolean, integer, float and date working for this issue.

SELECT
    column_name,
    data_type
FROM
    information_schema.columns
WHERE
    table_name = 'table_name';

Developer should be able to add data from each dataset into different databases.

Suggestion:

...
"database": {
  "url": "{{env.SOUNTR_DB_URL}}"
},
...

Commands suggestion:

• Create a new project: rdm init [--name <name> | -n <name>] [--input-type <type> | -it <type>]
• Execute data transfer: rdm apply

Initializing a RDM project creates a new directory with a RDM file.

Allow to set headers from RDM Object. Suggestion:

...
"headers": {
  "Authentication": "'Bearer ' + env.API_AUTH"
},
...

• One for upserting data based on unique identifiers from table
• One for only updating existing data
  • Option to fail if it doesn't exist
• One for only inserting data if it doesn't exist
  • Option to fail if it exists
  • Option to clear all before inserting

Going to be useful when mapping data from json files.

Suggestion:

...
alias: {
  "$track": "_.track__items__track"
},
fields: {
  ...
  "track.name": "$track.name",
  ...
},
...

• Create type RdmField (props: name, entity, isTemplate)
• Parse field strings from RDM file to RdmFields
• Remove util functions to get entity/field from string

Suggestion:

rdm apply --args="id=x"

Or

rdm apply --id x

Code:

"type": "http",
"url": "http://localhost:3000/wh",
"params": {
  "id": "args.id"
}

Should not try to insert multiple rows for same record.

Example, for the following object:

{
  "track": "Track",
  "artist": ["Artist 1", "Artist 2"]
}

Return the following from flattenObjectIntoArrayOfRows:

----
track: Track
artist: Artist 1
----
artist: Artist 2
----

Use case: someone has a dataset, for example a json file, and wants to export it to a database in order to execute queries.

Person should configure the RDM File as usual, and the CLI asks them if the missing tables/columns/database/unique constraint should be created.

Use some database introspection to identify the unique keys from each table. When inserting/upserting, we should be able to identify which of them are being used.

Make sure it also works for partial unique indexes.

Allow the following:

...
"album": {
  "set": {
    "is_single": "_.type == 'single'",
    "is_album": "_.tracks_count > 4"
  }
}
...

There are scenarios where we have one endpoint to list data and another to get more detailed data for one specific id.

Suggestion:

• [x] Multiple inputs
• [ ] Join responses.