getlang.py

from collections import defaultdict

UNICODE_BLOCKS = {
	'en': range(0x0000, 0x02AF),
	'si': range(0x0D80, 0x0DFF),
	'ta': range(0x0B80, 0x0BFF),
	'dv': range(0x0780, 0x07BF)
}

def getlang(text):
	"""Get language via Unicode range. Partially based on:
		https://github.com/kent37/guess-language/blob/master/guess_language/guess_language.py#L344
	"""

	run_types = defaultdict(int)
	for c in text:
		if(c.isalpha()):
			for block in UNICODE_BLOCKS:
				if(ord(c) in UNICODE_BLOCKS[block]):
					run_types[block] += 1

	return max(run_types, key=run_types.get)