local utf8 = require 'utf8'

local my_utf8 = {}

function my_utf8.offset(s, pos)
  if pos == 1 then return 1 end
  local result = utf8.offset(s, pos)
  if result == nil then
    assert(false, ('my_utf8.offset(%d) called on a string of length %d (byte size %d); this is likely a failure to handle utf8\n\n^%s$\n'):format(pos, utf8.len(s), #s, s))
  end
  return result
end

function my_utf8.codepoint(s, pos)
  return my_utf8.sub(s, pos, pos+1)
end

-- end_pos is exclusive
function my_utf8.sub(s, start_pos, end_pos)
  local start_offset = my_utf8.offset(s, start_pos)
  local end_offset = my_utf8.offset(s, end_pos)
  return s:sub(start_offset, end_offset-1)
end

function my_utf8.match_at(s, pos, pat)
  return my_utf8.codepoint(s, pos):match(pat)
end

-- create a new iterator for s which provides the index and UTF-8 bytes corresponding to each codepoint
function my_utf8.chars(s, startpos)
  local next_pos = startpos or 1  -- in code points
  local next_offset = utf8.offset(s, next_pos)  -- in bytes
  return function()
    assert(next_offset)  -- never call the iterator after it returns nil
    local curr_pos = next_pos
    next_pos = next_pos+1
    local curr_offset = next_offset
    if curr_offset > #s then return end
    local codepoint = utf8.codepoint(s, curr_offset)
    next_offset = utf8.offset(s, 2, next_offset)
    assert(next_offset)
    local curr_char = s:sub(curr_offset, next_offset-1)
    return curr_pos, codepoint, curr_char
  end
end

return my_utf8