20160708-dojo-date-validation-using-regex-and-try-except.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This focus of this notebook is using regular expressions (regexes).\n",
    "\n",
    "It explores a variety of ways of using them accomplish a task.\n",
    "\n",
    "This notebook is based on\n",
    "cohpy-20160706-lunch-iso8601-regex.ipynb\n",
    "which is based on\n",
    "[Travis Risner's presentation](https://mail.python.org/pipermail/centraloh/2016-July/002888.html) about regexes\n",
    "from the 2016-06-27 COhPy meeting."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "The code converts input strings to datetime.date objects.\n",
    "\n",
    "There are many ways to do this.\n",
    "The code uses a mixed approach of combining\n",
    "regexes, standard library functions, \n",
    "try/except statements and ordinary code.\n",
    "Each technique has something that it is better at\n",
    "than other techniques.\n",
    "Each technique is used for what it is better \n",
    "than the other techniques at.\n",
    "For example,\n",
    "it uses regexes for the part of validation and parsing that is easy\n",
    "for regexes to do and tedious to do with ordinary code.\n",
    "It uses datetime.date() and min_date <= date <= max_date\n",
    "to do the rest of the validation,\n",
    "which would be very difficult with regexes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import re\n",
    "import datetime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "date_pattern = re.compile(r'''\n",
    "    # Expects dates to be in YYYY?MM?DD format\n",
    "    # where ? is separator.\n",
    "    # The separator may be one of '-', ',', '.', '/', or ' '.\n",
    "    # Both separators must be the same.\n",
    "    # Month and day may each have either one or two digits.\n",
    "    \n",
    "    # The (?P<name>...) stuff is fantastic for allowing one\n",
    "    # to pull things out of a match object by names.\n",
    "    \n",
    "    ^                         # matches beginning of string\n",
    "    (?P<year>\\d{4})           # year must have exactly four digits\n",
    "    (?P<separator>[-,./ ])    # separator must be one of\n",
    "                              # '-', ',', '.', '/', or ' '.\n",
    "    (?P<month>\\d{1,2})        # month must have one or two digits\n",
    "    (?P=separator)            # must match earlier separator\n",
    "    (?P<day>\\d{1,2})          # day must have one or two digits\n",
    "    $                         # matches end of string\n",
    "''', re.VERBOSE)\n",
    "\n",
    "min_date = datetime.date(1900, 1, 1)\n",
    "max_date = datetime.date(2100, 1, 1) - datetime.timedelta(days=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sample_input = [\n",
    "    '2016-7-6',\n",
    "    '2016-07-06',\n",
    "    '2016 07-06',\n",
    "    '2016 07 06',\n",
    "    '2016/07 06',\n",
    "    '2016/07/06',\n",
    "    '2016-07/06',\n",
    "    '2016-02-29',\n",
    "    '2015-02-29',\n",
    "    '1899-12-31',\n",
    "    '1900-01-01',\n",
    "    '2099-12-31',\n",
    "    '2100-01-01',\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "'2016-7-6' year='2016', month='7', day='6' separator='-'\n",
      "'2016-07-06' year='2016', month='07', day='06' separator='-'\n",
      "'2016 07-06' no match\n",
      "'2016 07 06' year='2016', month='07', day='06' separator=' '\n",
      "'2016/07 06' no match\n",
      "'2016/07/06' year='2016', month='07', day='06' separator='/'\n",
      "'2016-07/06' no match\n",
      "'2016-02-29' year='2016', month='02', day='29' separator='-'\n",
      "'2015-02-29' year='2015', month='02', day='29' separator='-'\n",
      "'1899-12-31' year='1899', month='12', day='31' separator='-'\n",
      "'1900-01-01' year='1900', month='01', day='01' separator='-'\n",
      "'2099-12-31' year='2099', month='12', day='31' separator='-'\n",
      "'2100-01-01' year='2100', month='01', day='01' separator='-'\n"
     ]
    }
   ],
   "source": [
    "# This shows the input and output\n",
    "# for applying the date_pattern regex to sample_input.\n",
    "# Note that 2015-02-29 passes the regex,\n",
    "# although it is not a valid date.\n",
    "\n",
    "for s in sample_input:\n",
    "    print('%r ' % s, end='')\n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        print('no match')\n",
    "        continue\n",
    "    print(\n",
    "        'year=%r, month=%r, day=%r separator=%r' % (\n",
    "        m.group('year'),\n",
    "        m.group('month'),\n",
    "        m.group('day'),\n",
    "        m.group('separator'),\n",
    "    ))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, the regex is applied in a function that does the whole parsing and validation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    try:\n",
    "        date = datetime.date(\n",
    "            int(m.group('year')),\n",
    "            int(m.group('month')),\n",
    "            int(m.group('day')),\n",
    "        )\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "'2016-7-6' -> 2016-07-06\n",
      "'2016-07-06' -> 2016-07-06\n",
      "'2016 07-06' -> None\n",
      "'2016 07 06' -> 2016-07-06\n",
      "'2016/07 06' -> None\n",
      "'2016/07/06' -> 2016-07-06\n",
      "'2016-07/06' -> None\n",
      "'2016-02-29' -> 2016-02-29\n",
      "'2015-02-29' -> None\n",
      "'1899-12-31' -> None\n",
      "'1900-01-01' -> 1900-01-01\n",
      "'2099-12-31' -> 2099-12-31\n",
      "'2100-01-01' -> None\n"
     ]
    }
   ],
   "source": [
    "for s in sample_input:\n",
    "    date = convert_to_date(s)\n",
    "    print('%r -> %s' % (s, date))\n",
    "\n",
    "presumed_good_output = [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "#\n",
    "# It minimizes the code inside the try clause\n",
    "# by getting the year, month, and day strings from the match object\n",
    "# outside the try clause.\n",
    "# It is also easy to read.\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    year  = int(m.group('year'))\n",
    "    month = int(m.group('month'))\n",
    "    day   = int(m.group('day'))\n",
    "    try:\n",
    "        date = datetime.date(year, month, day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "#\n",
    "# The year, month, and day are gotten as a tuple\n",
    "# from the match object in a single expression.\n",
    "# It is correct but hard to read,\n",
    "# so the previous version of the function is better.\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    year_month_day = map(\n",
    "        lambda name: int(m.group(name)),\n",
    "        ('year', 'month', 'day')\n",
    "    )\n",
    "    try:\n",
    "        date = datetime.date(*year_month_day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "#\n",
    "# match.group() can handle multiple names, returning a tuple.\n",
    "# The year_month_day stuff is now easier to read.\n",
    "# How does this code compare to two cells ago?\n",
    "# Which is easier to understand?\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    year_month_day = map(int, m.group('year', 'month', 'day'))\n",
    "    try:\n",
    "        date = datetime.date(*year_month_day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "#\n",
    "# Went back to separate year, month, and day variables.\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    year, month, day = map(int, m.group('year', 'month', 'day'))\n",
    "    try:\n",
    "        date = datetime.date(year, month, day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Of cells 7 through 10 inclusive, which function was easier to understand? How? Why?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "Now play with anonymous grouping in the regex.\n",
    "\n",
    "It works, but it is a nuisance to keep track of the indexes.\n",
    "If one inserts a group in or deletes a group from the regex,\n",
    "one has to adjust the index references.\n",
    "\n",
    "I really like the named groups much more."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "date_pattern = re.compile(r'''\n",
    "    # Expects dates to be in YYYY?MM?DD format\n",
    "    # where ? is separator.\n",
    "    # The separator may be one of '-', ',', '.', '/', or ' '.\n",
    "    # Both separators must be the same.\n",
    "    # Month and day may each have either one or two digits.\n",
    "    \n",
    "    # Stuff in parentheses can be gotten from the match object\n",
    "    # by index.\n",
    "    \n",
    "    ^            # beginning of string\n",
    "    (\\d{4})      # group 1: year must have exactly four digits\n",
    "    ([-,./ ])    # group 2: separator must be one of\n",
    "                 # '-', ',', '.', '/', or ' '.\n",
    "    (\\d{1,2})    # group 3: month must have one or two digits\n",
    "    \\2           # must match earlier separator (of group 2)\n",
    "    (\\d{1,2})    # group 4: day must have one or two digits\n",
    "    $            # end of string\n",
    "''', re.VERBOSE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    year  = int(m.group(1))\n",
    "    month = int(m.group(3))\n",
    "    day   = int(m.group(4))\n",
    "    try:\n",
    "        date = datetime.date(year, month, day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "# It is a minor variation on the above,\n",
    "# consolidating three statements:\n",
    "#\n",
    "#     year  = int(m.group(1))\n",
    "#     month = int(m.group(3))\n",
    "#     day   = int(m.group(4))\n",
    "#\n",
    "# into one statement:\n",
    "#\n",
    "#     year, month, day = map(int, m.group(1, 3, 4))\n",
    "#\n",
    "# Which way is is easier to read?\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    year, month, day = map(int, m.group(1, 3, 4))\n",
    "    try:\n",
    "        date = datetime.date(year, month, day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "#\n",
    "# It is a minor variation on the above,\n",
    "# getting all four groups at once.\n",
    "# The name _ is used for the unwanted group.\n",
    "# Using the name _ for unwanted data, comes from Ruby.\n",
    "# Converting the strings to integers has to be done\n",
    "# in a separate step to avoid crashing while \n",
    "# converting a separator to an int.\n",
    "#\n",
    "# What benefits does this technique have?\n",
    "# What drawbacks does this technique have?\n",
    "#\n",
    "# Is this easier to read than the previous cell?\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    year, _, month, day = m.groups()\n",
    "    year, month, day = map(int, (year, month, day))\n",
    "    try:\n",
    "        date = datetime.date(year, month, day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "# It is a minor variation on the function two cells above,\n",
    "# consolidating three variables:\n",
    "#\n",
    "#     year, month, day\n",
    "#\n",
    "# into a single sequence:\n",
    "#\n",
    "#     year_month_day\n",
    "#\n",
    "# Note the *year_month_day argument to datetime.date().\n",
    "#\n",
    "# Which way is more readable?\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    year_month_day = map(int, m.group(1, 3, 4))\n",
    "    try:\n",
    "        date = datetime.date(*year_month_day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, the regex is used mostly for ensuring that the string is a valid,\n",
    "and parses only the separator."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "date_pattern = re.compile(r'''\n",
    "    # Expects dates to be in YYYY?MM?DD format\n",
    "    # where ? is separator.\n",
    "    # The separator may be one of '-', ',', '.', '/', or ' '.\n",
    "    # Both separators must be the same.\n",
    "    # Month and day may each have either one or two digits.\n",
    "    \n",
    "    ^            # beginning of string\n",
    "    \\d{4}        # year must have exactly four digits\n",
    "    ([-,./ ])    # group 1: separator must be one of\n",
    "                 # '-', ',', '.', '/', or ' '.\n",
    "    \\d{1,2}      # month must have one or two digits\n",
    "    \\1           # must match earlier separator (of group 1)\n",
    "    \\d{1,2}      # day must have one or two digits\n",
    "    $            # end of string\n",
    "''', re.VERBOSE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "# It gets the separator from the match object,\n",
    "# then uses that to split the string into year_month_day list.\n",
    "#\n",
    "# Which way is more readable?\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    separator = m.group(1)\n",
    "    year_month_day = map(int, s.split(separator))\n",
    "    try:\n",
    "        date = datetime.date(*year_month_day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "# It does not use the regex for parsing.\n",
    "#\n",
    "# Which way is more readable?\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    # Convert the non-digits to spaces, for easy splitting.\n",
    "    s = ''.join([\n",
    "        c if c.isdigit() else ' '\n",
    "        for c in s\n",
    "    ])\n",
    "    year_month_day = map(int, s.split())\n",
    "    try:\n",
    "        date = datetime.date(*year_month_day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The grouping was handy for making sure that the delimiters were the same. Now we remove that group matching from the regex."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "date_pattern = re.compile(r'''\n",
    "    # Expects dates to be in YYYY?MM?DD format\n",
    "    # where ? is separator.\n",
    "    # The separator may be one of '-', ',', '.', '/', or ' '.\n",
    "    # Both separators must be the same.\n",
    "    # Month and day may each have either one or two digits.\n",
    "    \n",
    "    ^            # beginning of string\n",
    "    (\\d{4}-\\d{1,2}-\\d{1,2})\n",
    "    |\n",
    "    (\\d{4},\\d{1,2},\\d{1,2})\n",
    "    |\n",
    "    (\\d{4}\\.\\d{1,2}\\.\\d{1,2})\n",
    "    |\n",
    "    (\\d{4}/\\d{1,2}/\\d{1,2})\n",
    "    |\n",
    "    (\\d{4}\\ \\d{1,2}\\ \\d{1,2})\n",
    "    $            # end of string\n",
    "''', re.VERBOSE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The function in this cell works the same as the earlier version.\n",
    "# It does not use the regex for parsing.\n",
    "#\n",
    "# Which way is more readable? This way or the previous way?\n",
    "\n",
    "def convert_to_date(s):\n",
    "    '''Converts input string s to a datetime.date object.\n",
    "    \n",
    "    Returns datetime.date object if s is valid date.\n",
    "    Date must be between min_date and max_date inclusive.\n",
    "    Otherwise returns None.'''\n",
    "    \n",
    "    m = re.match(date_pattern, s)\n",
    "    if not m:\n",
    "        return None\n",
    "    \n",
    "    # Convert the non-digits to spaces, for easy splitting.\n",
    "    s = ''.join([\n",
    "        c if c.isdigit() else ' '\n",
    "        for c in s\n",
    "    ])\n",
    "    year_month_day = map(int, s.split())\n",
    "    try:\n",
    "        date = datetime.date(*year_month_day)\n",
    "    except ValueError:\n",
    "        return None\n",
    "\n",
    "    if min_date <= date <= max_date:\n",
    "        return date\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "assert presumed_good_output == [convert_to_date(s) for s in sample_input]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Review:\n",
    "\n",
    "- Which regex did you find to be the easiest to understand? How? Why?\n",
    "\n",
    "- Which function did you find to be the easiest to understand? How? Why?"
   ]
  }
 ],
 "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": "3.4.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}