cimpress-mcp / j2v Goto Github PK

J2V outputs SQL without any column aliases. We want to add it.

Build functionality to surface a column alias
e.g. ROOT_ARRAY_ARRAY_FIELDNAME

So the produced SQL:

• DATA.JSON:"payload":"statuses":"Address":"name"::string AS PAYLOAD_STATUSES_ADDRESS_NAME
• And in the case of flattened arrays
  DATA.JSON:"payload":"statuses".VALUE:"Address":"name"::string AS PAYLOAD_STATUSES_ADDRESS_NAME

https://github.com/Cimpress-MCP/j2v/tree/column_aliases

Feature request:

1. Add option to provide the column to be used for PK
2. Add a couple of generic measures (MIN/MAX/COUNT) on a user specified column

Current functionality of J2V is limited to put field name in double quotes only when space is detected. According to the Snowflake's best practice of querying JSONs we can doublequote everything, and we need to do it.
https://docs.snowflake.net/manuals/user-guide/querying-semistructured.html

An example of path generated now (SQL output & Looker files) might look like:
DATA.payload:name:"Name in English"::string

After the change:
DATA."payload":"name":"Name in English"::string

During the development make sure that Snowflake rules are isolated from the main output generation logic.

Here's an example input statement :

<MODEL BODY> include: "manufacturing_issues"

Looker does not appear to like this syntax. Correct syntax appears to be :

<MODEL BODY> include: "manufacturing_issues.view.lkml"

J2V works well with examples, but examples often do not cover an entire data set. If we have already crafted a schema in another format (in this case, Open API), using that schema definition may be a helpful alternative or addition to JSON examples.

It could also help generate better Looker views by pulling descriptions from the external schema