#!/usr/bin/env python3
# NB Depends on dict preserving the insert order (CPython >= 3.6, PyPy all)

from csv           import DictReader
from datetime      import datetime as DateTime
from os            import scandir
from simplejsonseq import dump
from sys           import argv, stderr, stdout

import ietfcsv  # ietf-tab CSV dialect

NAME = '%Y%m%dT%H%MZ.tsv'
HEAD = ('region №', 'region', 'ТВО №', 'Центр ТВО', '№ ВД')

def without(dict, *keys):
	keys = set(keys)
	return {k: v for k, v in dict.items() if k not in keys}

rows = {}  # aggregated data
keys = []  # final row order

if len(argv) > 1 and argv[1] == '-v':
	def trace(*args, **named):
		print(*args, **named, file=stderr)
else:
	def trace(*args, **named):
		pass

for entry in sorted(scandir(), key=lambda e: e.name):
	if not entry.name.endswith('.tsv'): continue
	trace(entry.name, end=': ')
	with open(entry.name, newline='\r\n') as tsv:
		keys = []
		time = DateTime.strptime(entry.name, NAME)
		prevlen, updated = len(rows), 0

		for line in DictReader(tsv, dialect='ietf-tab'):
			del line['Кількість виборчих дільниць в окрузі']
			del line['Кількість виборчих дільниць щодо яких '
			         'надійшли відомості']
			assert 'timestamp' not in line
			line['timestamp'] = time.strftime('%Y-%m-%dT%H:%MZ')

			row = {k: line.pop(k) for k in HEAD}
			row['history'] = [line]
			key = (row['ТВО №'], int(row['№ ВД']))
			keys.append(key)

			row = rows.setdefault(key, row)
			if (without(row['history'][-1], 'timestamp') !=
			    without(line, 'timestamp')):
				row['history'].append(line)
				updated += 1

		trace('{} lines, {} inserted, {} updated'
		      .format(len(keys), len(rows)-prevlen, updated))
		if len(rows)-prevlen == 0 and updated == 0:
			print('warning: {}: no changes'.format(entry.name),
			      file=stderr)
		assert set(keys) <= set(rows)
		if len(keys) < len(rows):
			for k, row in rows.items():
				if k in keys: continue
				print('warning: {}: ТВО {}, ВД {} missing'
				      .format(entry.name,
				              row['ТВО №'],
				              row['№ ВД']),
				      file=stderr)

for k, row in rows.items():
	# Not {r.pop('timestamp'): r for r ...} because CPython evaluates
	# dictionary comprehensions in the wrong order (#29652)
	row['history'] = {r['timestamp']: without(r, 'timestamp')
	                  for r in row['history']}
stdout.reconfigure(newline='\r\n')
dump((rows[k] for k in keys), stdout, ensure_ascii=False, indent='\t')