light ETL to reformat webdata from http://www.hydroinfo.hu/Html/archivum/archiv_tabla.html to a TSV with date and value columns

waterlevel_etl.py

fi = open ('vizallas.txt', 'r')
fo = open ('vizallas.tsv', 'w')

year = ''

for li in fi:
# fixup days not existing in a given month
  it = li.strip().replace('         ','   ...   ').split('   ')
	
  if len(it) < 2:
  # not a data line
    if it[0].strip().isdigit():
      if int(it[0]) > 2000:
      # catched a year
        year = it[0].strip()
      else:
      	# data line with all month for a day
      	day = '0' + it[0].strip()
      	day = day[-2:]
      	for mo in range(1,13)::
      	  if it[mo].strip() <> '...':
      	    month = '0'+str(mo)
      	    month = month[-2:]
      	    print >>fo, year + '-' + month + '-' + day + '\t' + it[mo].strip()

fo.close()
fi.close()