Regex patterns for kk's date 😂

py-lots-regex.ipynb

{
 "cells": [{
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5/1/1017 : 5/1/2017\n",
      "a-001 : ---- REALLY NO IDEA, GO WITH 5/1/2017? ----\n",
      "10/12017 : 10/1/2017\n",
      "131/2017 : 13/1/2017\n",
      "19/1//2017 : 19/1/2017\n",
      "25/1/217 : 25/1/2017\n",
      "27/12017 : 27/1/2017\n",
      "30/1/201 : ---- THE 201X ----\n",
      "29/1/201 : ---- THE 201X ----\n",
      "30/1/25017 : 30/1/2017\n",
      "22/2/017 : 22/2/2017\n",
      "23/2/201 : ---- THE 201X ----\n",
      "5/2/018 : 5/2/2018\n",
      "7/2/20108 : 7/2/2018\n",
      "10/2/20188 : 10/2/2018\n",
      "12/2/018 : 12/2/2018\n",
      "14/2/22018 : 14/2/2018\n",
      "f : ---- REALLY NO IDEA, GO WITH 14/2/2018? ----\n",
      "22/2/20188 : 22/2/2018\n",
      "6/3/22018 : 6/3/2018\n",
      "12/3/20118 : 12/3/2018\n",
      "12/3/118 : 12/3/2018\n",
      "25/3/20181 : 25/3/2018\n",
      "27/3/20018 : 27/3/2018\n",
      "29/3/20188 : 29/3/2018\n",
      "3/3/2/2017 : 3/2/2017\n",
      "9/3//2017 : 9/3/2017\n",
      "13/22/2017 : ---- OVER 12 -> 13/22/2017 ----\n",
      "ฟ-001 : ---- REALLY NO IDEA, GO WITH 9/3/2017? ----\n",
      "36/3/17 : ---- OVER 31 -> 36/3/2017 ----\n",
      "2/13/17 : ---- OVER 12 -> 2/13/2017 ----\n",
      "7/4/217 : 7/4/2017\n",
      "10/4/20107 : 10/4/2017\n",
      "21/4/20117 : 21/4/2017\n",
      "23/4/20174 : 23/4/2017\n",
      "13/4/20107 : 13/4/2017\n",
      "254/4/17 : ---- OVER 31 -> 254/4/17 ----\n",
      "27/4/20117 : 27/4/2017\n",
      "14/15/2017 : ---- OVER 12 -> 14/15/2017 ----\n",
      "18/05/2017 : 18/05/2017\n",
      "19/5/2017000 : 19/5/2017\n",
      "00:00:00 : ---- REALLY NO IDEA, GO WITH 19/5/2017? ----\n",
      "25/05/2017 : 25/05/2017\n",
      "29/05/2017 : 29/05/2017\n",
      "31/05/2017 : 31/05/2017\n",
      "1/6/22017 : 1/6/2017\n",
      "12/6/201 : ---- THE 201X ----\n",
      "7/72017 : 7/7/2017\n",
      "87/72017 : ---- OVER 31 -> 87/7/2017 ----\n",
      "11/7//2017 : 11/7/2017\n",
      "12/72017 : 12/7/2017\n",
      "12/7/22017 : 12/7/2017\n",
      "21/7/20147 : 21/7/2017\n",
      "26/72017 : 26/7/2017\n",
      "29/7/20147 : 29/7/2017\n",
      "1/8/20170 : 1/8/2017\n",
      "12/8/20017 : 12/8/2017\n",
      "3/9/20170 : 3/9/2017\n",
      "10/9/25017 : 10/9/2017\n",
      "19/9/201 : ---- THE 201X ----\n",
      "22/9/20147 : 22/9/2017\n",
      "274/9/2017 : ---- OVER 31 -> 274/9/2017 ----\n",
      "27/9/20107 : 27/9/2017\n",
      "2/10/201 : ---- THE 201X ----\n",
      "อ : ---- REALLY NO IDEA, GO WITH 27/9/2017? ----\n",
      "8/101/2017 : ---- REALLY NO IDEA, GO WITH 27/9/2017? ----\n",
      "12/10/017 : 12/10/2017\n",
      "16/10/20147 : 16/10/2017\n",
      " 17/10/2017 : 17/10/2017\n",
      "1/11/117 : 1/11/2017\n",
      "11/11/22017 : 11/11/2017\n",
      " 17/11/2017 : 17/11/2017\n",
      "11/12/201 : ---- THE 201X ----\n",
      "14/12/201 : ---- THE 201X ----\n",
      "19/12/201 : ---- THE 201X ----\n",
      "21/12/20147 : 21/12/2017\n",
      "22/12/22017 : 22/12/2017\n",
      "27/12/22017 : 27/12/2017\n",
      "29/17/2017 : ---- OVER 12 -> 29/17/2017 ----\n",
      "11/12018 : 11/1/2018\n",
      "16/1/22018 : 16/12/2018\n",
      "18/1/018 : 18/10/2018\n",
      "19/1/22018 : 19/12/2018\n",
      "26/1/201/8 : 26/1/2018\n"
     ]
    }
   ],
   "source": [
    "tmpdate = [i for i in df_cleaned['วันที่'].unique() if not isinstance(i, datetime.date)]\n",
    "\n",
    "# multiple patterns as a list\n",
    "# https://stackoverflow.com/a/46328646/4010864 \n",
    "\n",
    "last = ''\n",
    "\n",
    "dateRegexPatterns = [\n",
    "    # remove starting whitespace\n",
    "    (r'^\\s', r''),\n",
    "    # 30/12017 > 30/1/2017\n",
    "    (r'(^\\d{1,2}/)(\\d{1})([2017]{4})$', r'\\1\\2/\\3'),\n",
    "    # 131/2017 > 13/1/2017\n",
    "    (r'(^[1-3][0-9])([0-9])/([20178]{4})$', r'\\1/\\2/\\3'),\n",
    "    # 30/1/25017 > 30/1/2017, 19/1//2017 > 19/1/2017, 5/1/1017 > 5/1/2017\n",
    "    (r'(^\\d{1,2}/\\d+/)(/?\\d*7\\d*)$', r'\\g<1>2017'),\n",
    "    # 16/1/22018 > 16/12/2018\n",
    "    (r'(^\\d{1,2}/)([1])/([0-2])(\\d*8\\d*)$', r'\\1\\2\\3/\\4'),\n",
    "    # 30/1/25018 > 30/1/2018\n",
    "    (r'(^\\d{1,2}/\\d+/)(/?\\d*8\\d*)$', r'\\g<1>2018'),\n",
    "    # 10/12018 > 10/1/2018, 27/12018 > 27/1/2018\n",
    "    (r'^(\\d+\\/)(\\d+)(\\d{3}8)$', r'\\1\\2/\\3'),\n",
    "    # 3/3/2/2017 > 3/2/2017\n",
    "    (r'^(\\d+)/(\\d+)/(\\d+)/(\\d*7\\d*)$', r'\\2/\\3/\\4'),\n",
    "    # 26/1/201/8 > 26/1/2018\n",
    "    (r'^(\\d+/)(\\d+/)(\\d*)/(\\d*)', r'\\1\\2\\3\\4'), \n",
    "    # remove ending whitespace \n",
    "    (r'$|\\s', r''),\n",
    "]\n",
    "\n",
    "# apply all patterns\n",
    "def regexDate(dateString):\n",
    "    for old, new in dateRegexPatterns:\n",
    "        dateString = re.sub(old, new, dateString)\n",
    "    return dateString\n",
    "\n",
    "# loop through each date\n",
    "for date in tmpdate:\n",
    "    passRegex = str(date)\n",
    "    passRegex = regexDate(passRegex)\n",
    "    \n",
    "    # filter normies and non-normies\n",
    "    normies = re.match(r'(^[0-9]|[0-2][0-9]|[3][0-1])/([1-9]|0[1-9]|1[0-2])/(201[7-8]$)', passRegex)\n",
    "    the201 = re.match(r'.*/201$', passRegex)\n",
    "    over31 = re.match(r'[^0-2][2-9]|[0-9]{3}/.*', passRegex)\n",
    "    over12 = re.match(r'.*/(1[^0-2]|[^0-1][0-9])/.*', passRegex)\n",
    "        \n",
    "    # debugging -- to see what is not in pattern yet\n",
    "    if passRegex is not '' and not normies:\n",
    "#         passRegex = '---- PASS REGEX BUT STILL NOT PARSEABLE -> {0} ----'.format(passRegex)\n",
    "\n",
    "        if over12:\n",
    "            passRegex = '---- OVER 12 -> {0} ----'.format(passRegex)\n",
    "    \n",
    "        if over31:\n",
    "            passRegex = '---- OVER 31 -> {0} ----'.format(passRegex)\n",
    "    \n",
    "    if the201:\n",
    "        passRegex = '---- THE 201X ----'\n",
    "\n",
    "    # if it pass no filter, the value is still the same\n",
    "    # if really not possible to get pattern, use last iter?\n",
    "    if passRegex == str(date) and not normies:\n",
    "        passRegex = '---- REALLY NO IDEA, GO WITH {0}? ----'.format(last)\n",
    "    elif not normies:\n",
    "        pass\n",
    "    else:\n",
    "        last = passRegex\n",
    "\n",
    "    print(\"{0} : {1}\".format(str(date), passRegex))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}