Add a utils/standard library package to Joystick

[rglover]

Just came to mind looking at some functions I've copy/pasted across several projects. Would be great to just organize these into something like @joystick.js/utils. Easy enough to do. Just want to avoid recreating something like Lodash.

Keeping a list...

• [ ] Types (could rely on Node util/types but want it to be isomorphic and thorough)
• [ ] Delay/Debounce
• [ ] Set Query Param
• [ ] Delete Query Param
• [ ] Date Format Helpers (maybe—could end up being @joystick.js/dates)
• [ ] Wait (Promised)
• [ ] Copy to Clipboard
• [ ] Check if Valid JSON
• [ ] Generate ID (technically in Joystick proper)
• [ ] URL encode JSON body for POST
• [ ] Input Validation
• [ ] CLILog
• [ ] loadSettings
• [ ] serializeQueryParameters
• [ ] trackFunctionCall (from joystick.js/test)

[rglover]

See also #49

[rglover]

For types, I want to have a type checker function that can be called anywhere to validate the type/shape of things. So, if I pass an object, I can validate the types like this:

const user = { firstName: 'Ryan' };

utils.types.validate(user, {
  firstName: {
    type: 'string',
    required: true,
  },
}):

I should be able to adapt the existing input validation from the server and just make it globally available via the utils package.

Note: The value passed as the second argument could be an object, or, a type string like this: utils.types.validate(user, 'object') for generic type checking.

[rglover]

For dates, I'd like to have some basic Date-object based helpers like:

• [ ] Add (unit) to timestamp (e.g., add seconds, hours, etc)
• [ ] Exchange one ISO timestamp for another w/ a different offset
• [ ] Get back a date object from an ISO string

Right now I use dayjs for this stuff but it's unnecessary. It'd be best to standardize it and have control over all dates running through the framework so there's zero confusion.

[rglover]

For SQL:

const generate_sql_from_object = {
  insert: (options = {}) => {
    const column_names = Object.keys(options?.data)?.join(',');
    const value_placeholders = Object.keys(options?.data)?.map((_, index) => `$${index + 1}`)?.join(',');

    return {
      statement: `INSERT INTO ${options?.table} (${column_names}) VALUES (${value_placeholders})`,
      column_names,
      value_placeholders,
      values: Object.values(options?.data),
    };
  },
  update: (options = {}) => {
    const whereEntries = Object.entries(options?.where);
    const sets = Object.keys(options?.data).map((key, index) => {
      return `${key} = $${whereEntries.length + index + 1}`;
    })?.join(',');
    const where = whereEntries?.map(([key], index) => {
        return `${key} = $${index + 1}`;
    })?.join(',');

    return {
      statement: `UPDATE ${options?.table} SET ${sets} WHERE ${where}`,
      sets,
      where,
      values: [
        ...(Object.values(options?.where)),
        ...(Object.values(options?.data))
      ],
    };
  },
};

Usage:

const insert = generate_sql_from_object.insert({
  table: 'users',
  data: {
    user_id: 'abc123',
    pancakes: 'good',
    sausage: 'decent',
    pizza: 'oh baby',
  },
});

await process.databases.postgresql.query(insert.statement, insert.values);

const update = generate_sql_from_object.update({
  table: 'users',
  data: {
    pancakes: 'good',
    sausage: 'decent',
    pizza: 'oh baby',
  },
  where: { user_id: 'abc123', role: 'eater' }
});

await process.databases.postgresql.query(update.statement, update.values);

Consider offering the above as a single wrapper on the postgresql object like process.databases.postgresql.insert() or process.databases.postgresql.update() which takes in a table, data, and optional where object and wraps the example code above.