Mappings to triple letters (e.g. Tamil ன → "nnn", கன்னியாகுமரி [க,ன,்,ன,ி,...] → "knnnnnniyaakumri")

[lionel-rowe]

Selected triple mappings from the exported charmap:

{
  "ऩ": "nnn",
  "ऴ": "lll",
  "ड़": "dddh",
  "ன": "nnn",
  "ழ": "lll",
  "ೞ": "lll",
  "ഴ": "lll",
  "ᄴ": "sss",
}

I don't know the relevant scripts but these surely have to be wrong? I can't even figure out how the sextuple-n in Tamil கன்னியாகுமரி → "knnnnnniyaakumri" would be pronounced. For context, Google Translate transliterates the same word as "Kaṉṉiyākumari", and based on how the corresponding TTS audio sounds that seems reasonable.

It looks like data/charmap.ts is generated from data/data.json via scripts/generate_data.js, but how is data/data.json itself generated? Maybe there's a more up-to-date version of this data with better Tamil mappings?