#!/usr/bin/env ruby

require 'base32' # you will have to `gem install` this
require 'uri/ni' # this too (i wrote it)

# i had claude (sonnet 4.6) initially provide me the basics for these
# two functions and then i cleaned up its silliness

# note that's "variable integer" not a typo of "variant"
def decode_varint string
  # don't raise here just return nil
  return if string.empty?

  # first make sure the string is bytes and duplicate it so we don't
  # hurt the original
  string = string.b.dup

  out   = 0 # the return value
  shift = 0 # 7-bit increments to shift left

  loop do
    # do raise here though because this is actually an error
    raise ArgumentError, 'malformed varint' if string.empty?

    # peel off the first byte and get its ordinal value
    byte = string.slice!(0).ord
    # mask, shift left and OR it
    out |= (byte & 0x7f) << (7 * shift)

    # continue only if the high-order bit is set
    break if (byte & 0x80).zero?

    shift += 1 # the other 7 bits are data
  end

  # return the varint and remaining string
  [out, string]
end

def decode_cidv1_base32 cid
  # first character is the encoding
  raw = Base32.decode(cid[1..].upcase)

  # this will clip the varints off the front of the string
  version, raw = decode_varint raw
  codec,   raw = decode_varint raw
  algo,    raw = decode_varint raw
  len,     raw = decode_varint raw

  # note:
  # * `version` would dispatch but we only deal with version 1 here
  # * `codec`   tells us something about the thing that was hashed
  # * `algo`    0x12 is sha-256 so that's hard coded for now
  # * `len`     is the payload length which we hope is fully present
  #
  # (see https://github.com/multiformats/multicodec/blob/master/table.csv)
  #
  # so we could just imagine mapping that table to hash algorithms and
  # mime types or whatever

  # here have some diagnostics if you like, lol
  #
  # warn (["0x%02x (%d)"] * 4).join(', ') %
  #   [version, codec, algo, len].map { |x| [x, x] }.flatten

  # oh yeah ruby has base64 built into `pack`
  b64 = [raw[0, len]].pack('m0').tr('+/', '-_').delete(?=)

  # the content-type query parameter accounts for the `codec` component
  URI("ni:///sha-256;#{b64}?ct=application/vnd.ipld.dag-cbor")
end

if $0 == __FILE__
  uri = decode_cidv1_base32 'bafyreiemgv3douzgbjnytcp5tbdb4iguk34l3ik2w5ahwnenzehmexdidm'
  puts uri           # ni:///sha-256;jDV2N1MmCluJif2YRh4g1Fb4vaFat0B7NI3JDsJcaBs?ct=application/vnd.ipld.dag-cbor
  puts uri.hexdigest # 8c35763753260a5b8989fd98461e20d456f8bda15ab7407b348dc90ec25c681b
end