populatetools / csvzip Goto Github PK

Love the idea.

Just a quick suggestion. In the example you end up with a JSON mapping like

{
  "census": {
    "DESC_DISTRITO": {
      "CENTRO": "0",
      "ARGANZUELA": "1",
      "RETIRO": "2",
      "SALAMANCA": "3",
      ...
    },
    "DESC_BARRIO": {
      "PALACIO": "0",
      "EMBAJADORES": "1",
      "UNIVERSIDAD": "2",
      "CHOPERA": "3",
      "PACIFICO": "4",
      ...
    }
  }
}

If I were using this I'd probably generate a lookup table for each key of census, so any lookups would be one operation, instead of traversing object keys and checking the values for a match.

Basically, flip them this way.

{
  "census": {
    "DESC_DISTRITO": {
      "0": "CENTRO",
      "1": "ARGANZUELA",
      ...
    },
...
}

Let's say we're looping over the rows, or we're looking at the info for a single row.

const row = "139,1".split(','); // Fake row with two columns 

This would always be fast

const name = mapping['census']['DESC_DISTRITO'][row[1]];

With the current mapping, lookups would get slower with the number of keys in the DESC_DISTRITO object

Maybe I'm overlooking/missing something.