DaveM@13: #!/usr/bin/env python
DaveM@13: 
DaveM@13: import csv
DaveM@13: import time
DaveM@13: import unicodedata
DaveM@13: from geopy.geocoders import Nominatim
DaveM@13: from geopy.exc import GeocoderTimedOut
DaveM@13: import random
DaveM@13: import pdb
DaveM@13: 
DaveM@19: DEFAULT_TIME = None
DaveM@19: # DEFAULT_TIME_M = None
DaveM@15: DEAULT_LOCATION = None
DaveM@13: 
DaveM@13: def regulateData(dataDict):
DaveM@13: 	print("Parse %s"%(str(dataDict['ID'])))
DaveM@13: 	p_DOBQ = "What is your partner's date of birth? Please use the format DD/MM/YYYY (for example, 29/03/1981)."
DaveM@13: 	p_TOBQ = "At what exact time were your partner born? Please use the format HHMM (for example, 2204)."
DaveM@13: 	DOB_DQ = "Which day (numeric) have you been born?"
DaveM@13: 	DOB_MQ = "Which month have you been born?"
DaveM@13: 	DOB_YQ = "Year Of Birth"
DaveM@13: 	TOB_Q = "At what exact time were you born? Please use the format HHMM (for example, 2204)."
DaveM@13: 	COB = "What is your place of birth? Please specify city and country (for example, San Francisco, USA)."
DaveM@13: 	p_COB = "What is your partner's place of birth? Please specify city and country (for example, San Francisco, USA)."
DaveM@13: 	dataDict['DOB'] = parseDOB(dataDict[DOB_DQ],dataDict[DOB_MQ],dataDict[DOB_YQ])
DaveM@13: 	# print (dataDict[DOB_DQ],dataDict[DOB_MQ],dataDict[DOB_YQ])
DaveM@13: 	# print dataDict['DOB']
DaveM@13: 	dataDict['TOB'] = parseTOB(dataDict[TOB_Q])
DaveM@13: 	dataDict['pDOB'] = parsePartnerDOB(dataDict[p_DOBQ])
DaveM@13: 	dataDict['pTOB'] = parseTOB(dataDict[p_TOBQ])
DaveM@13: 	# MAKE RANDOM PLACE
DaveM@13: 	# dataDict['COB'] = (random.uniform(-90, 90),random.uniform(-90, 90))
DaveM@13: 	# dataDict['pCOB'] = (random.uniform(-90, 90),random.uniform(-90, 90))
DaveM@13: 	dataDict['COB'] = parseBirthTown(dataDict[COB])
DaveM@13: 	dataDict['pCOB'] = parseBirthTown(dataDict[p_COB])
DaveM@13: 	return dataDict
DaveM@13: 
DaveM@13: 
DaveM@13: def parseBirthTown(s):
DaveM@13: 	try:
DaveM@13: 		s = s.encode('ascii')
DaveM@13: 	except UnicodeDecodeError:
DaveM@13: 		# pdb.set_trace()
DaveM@13: 		s = s.decode('latin-1')
DaveM@13: 		# s = unicodedata.normalize('NFKD',s.decode('utf-8')).encode('ascii','ignore')
DaveM@13: 	timeoutTime = 2
DaveM@13: 	geolocator = Nominatim(timeout=timeoutTime)
DaveM@13: 	while s is not [] and timeoutTime < 60:
DaveM@13: 		try:
DaveM@13: 			location = geolocator.geocode(s)
DaveM@13: 			if location is not None:
DaveM@13: 				# print(location.raw)
DaveM@13: 				# print (location.latitude, location.longitude)
DaveM@13: 				return (location.latitude, location.longitude, location.raw)
DaveM@13: 			else:
DaveM@13: 				s = s.split(' ',1)
DaveM@13: 				if len(s) == 2:
DaveM@13: 					s = s[1]
DaveM@13: 					# print s
DaveM@13: 				else:
DaveM@13: 					s = DEAULT_LOCATION
DaveM@13: 		except:
DaveM@13: 			timeoutTime += 1
DaveM@13: 			print("Error: geocode failed on input %s, incrementing timeout time to %d"%(s,timeoutTime))
DaveM@13: 			time.sleep(5)
DaveM@13: 			geolocator = Nominatim(timeout=timeoutTime)
DaveM@13: 	# places = geograpy.get_place_context(text=s)
DaveM@13: 
DaveM@13: def parsePartnerDOB(dob):
DaveM@13: 	# print dob
DaveM@13: 	# pdb.set_trace()
DaveM@13: 	dob = dob.strip()
DaveM@13: 	if(dob.count('-') == 2):
DaveM@13: 		dob = dob.replace('-','/')
DaveM@13: 	if(dob.count(' ') == 2):
DaveM@13: 		dob = dob.replace(' ','/')
DaveM@13: 	dob_ = dob.split('/')
DaveM@13: 	if(len(dob_) != 3):
DaveM@13: 		dob = dob.replace('/','').strip()
DaveM@13: 		dob_ = []
DaveM@13: 		# print dob
DaveM@13: 		if len(dob) == 8: # ddmmyyyy
DaveM@13: 			dob_.append(dob[:2])
DaveM@13: 			dob_.append(dob[2:4])
DaveM@13: 			dob_.append(dob[4:])
DaveM@13: 		elif len(dob) == 7 and dob[1] == '1' and (dob[2] == '0' or dob[2] == '1' or dob[2] == '2'): # dmmyyyy
DaveM@13: 			dob_.append(dob[0])
DaveM@13: 			dob_.append(dob[1:3])
DaveM@13: 			dob_.append(dob[3:])
DaveM@13: 		elif(len(dob) == 7):
DaveM@13: 			if int(dob[:2]) > 31:# dmmyyyy
DaveM@13: 				dob_.append(dob[0])
DaveM@13: 				dob_.append(dob[1:3])
DaveM@13: 				dob_.append(dob[3:])
DaveM@13: 		elif len(dob) == 7: # ddmyyyy
DaveM@13: 			dob_.append(dob[0:2])
DaveM@13: 			dob_.append(dob[2])
DaveM@13: 			dob_.append(dob[3:])
DaveM@13: 		elif len(dob) == 6 and dob[3:4] != '19': # ddmmyy
DaveM@13: 			dob_.append(dob[:2])
DaveM@13: 			dob_.append(dob[2:4])
DaveM@13: 			dob_.append(dob[4:])
DaveM@13: 		elif len(dob) == 5 and dob[1] == '1' and (dob[2] == '0' or dob[2] == '1' or dob[2] == '2'): # dmmyy
DaveM@13: 			dob_.append(dob[0])
DaveM@13: 			dob_.append(dob[1:3])
DaveM@13: 			dob_.append(dob[3:])
DaveM@13: 		elif len(dob) == 5: # ddmyy
DaveM@13: 			dob_.append(dob[:2])
DaveM@13: 			dob_.append(dob[2])
DaveM@13: 			dob_.append(dob[3:])
DaveM@13: 		elif len(dob) == 4: # dmyy
DaveM@13: 			dob_.append(dob[0])
DaveM@13: 			dob_.append(dob[1])
DaveM@13: 			dob_.append(dob[2:])
DaveM@13: 		else:
DaveM@13: 			if(len(dob) < 4):
DaveM@13: 				return None
DaveM@13: 			# print dob
DaveM@13: 			# print filter(lambda x: x.isdigit(),dob)
DaveM@13: 			print 'no / partnerDOB issue'
DaveM@13: 		# deal with no /'s
DaveM@13: 	try:
DaveM@13: 		d = int(filter(lambda x: x.isdigit(),dob_[0]))
DaveM@13: 		m = int(filter(lambda x: x.isdigit(),dob_[1]))
DaveM@13: 		y = int(filter(lambda x: x.isdigit(),dob_[2]))
DaveM@13: 		if y < 100:
DaveM@13: 			y = y + 1900
DaveM@13: 		if (m > 12 and d <= 12):
DaveM@13: 			temp = d
DaveM@13: 			d = m
DaveM@13: 			m = temp 
DaveM@13: 		if(d > 31 or d < 1 or m > 12 or m < 1 or y > 2017 or y < 1900):
DaveM@13: 			print 'error with DOB '+d+'/'+m+'/'+y
DaveM@13: 			pdb.set_trace()
DaveM@13: 	except TypeError:
DaveM@13: 		return None
DaveM@13: 	# print  (d,m,y)
DaveM@13: 	return (d,m,y)
DaveM@13: 
DaveM@13: def monthStringToNum(s):
DaveM@13: 	# print 'inMonthStringToNum'
DaveM@13: 	m = {'jan':1,'feb':2,'mar':3,
DaveM@13: 	'apr':4,'may':5,'jun':6,'jul':7,'aug':8,
DaveM@13: 	'sep':9,'oct':10,'nov':11,'dec':12}
DaveM@13: 	s_ = s.strip()[:3].lower()
DaveM@13: 	try:
DaveM@13: 		out = m[s_]
DaveM@13: 		return out
DaveM@13: 	except:
DaveM@13: 		raise ValueError('Not a month')
DaveM@13: 
DaveM@13: def checkMonthDay(d,m):
DaveM@13: 	if d > 31: # take first two digits of day
DaveM@13: 		d = int(str(d)[:2])
DaveM@13: 		if d > 31:
DaveM@13: 			d = int(str(d)[1])
DaveM@13: 	if m > 12 and d < 12: # Day and month wrong way round - American
DaveM@13: 		temp = m
DaveM@13: 		m = d
DaveM@13: 		d = temp
DaveM@13: 	if(m == 2):
DaveM@13: 		if d <= 29:
DaveM@13: 			return (True,d,m)
DaveM@13: 		else:
DaveM@13: 			return (False,d,m)
DaveM@13: 	elif m in [4,6,9,11]:
DaveM@13: 		if d <= 30:
DaveM@13: 			return (True,d,m)
DaveM@13: 		else:
DaveM@13: 			return (False,d,m)
DaveM@13: 	elif m <= 12 and d <= 31:
DaveM@13: 		return (True,d,m)
DaveM@13: 	else:
DaveM@13: 		return (False,d,m)
DaveM@13: 
DaveM@13: def parseDOB(d,m,y):
DaveM@13: 	d = int(filter(lambda x: x.isdigit(),d))
DaveM@13: 	y = int(filter(lambda x: x.isdigit(),y))
DaveM@13: 	try:
DaveM@13: 		# print m
DaveM@13: 		m = monthStringToNum(m.strip())
DaveM@13: 	except ValueError:
DaveM@13: 		m = int(m.strip())
DaveM@13: 	if(y < 100):
DaveM@13: 		y = y + 1900
DaveM@13: 	(r,d,m) = checkMonthDay(d,m)
DaveM@13: 	if not r:
DaveM@13: 		print 'error with day month'
DaveM@13: 		print (r,d,m)
DaveM@13: 	return (d,m,y)
DaveM@13: 
DaveM@13: def parseTOB(T):
DaveM@13: 	timeFlag = None
DaveM@13: 	T_ = T.replace('.','').lower().strip()
DaveM@13: 	if 'am' in T_:
DaveM@13: 		timeFlag = 0
DaveM@13: 		T = T_.replace('am','')
DaveM@13: 	if 'pm' in T_:
DaveM@13: 		timeFlag = 1
DaveM@13: 		T = T_.replace('pm','')
DaveM@13: 	T = T.strip()
DaveM@13: 	if T.count('.') == 1:
DaveM@13: 		T = T.replace('.',':')
DaveM@13: 	try:
DaveM@13: 		if ':' in T:
DaveM@13: 			T_ = T.split(':')
DaveM@13: 			
DaveM@13: 			H = int(T_[0])
DaveM@13: 			M = int(T_[1])
DaveM@13: 		else:
DaveM@13: 			if len(T) == 4:
DaveM@13: 				H = int(T[:2])
DaveM@13: 				M = int(T[2:])
DaveM@13: 			elif int(T) <= 24 :
DaveM@13: 				H = int(T)
DaveM@13: 				M = 0
DaveM@13: 			elif int(T) > 100:
DaveM@13: 				H = int(T)/100
DaveM@13: 				M = int(T)%100
DaveM@19: 			else:
DaveM@19: 				return None
DaveM@13: 		if timeFlag is not None:
DaveM@13: 			if timeFlag == 0:
DaveM@13: 				H = H%12
DaveM@13: 			else:
DaveM@13: 				H = H%12 + 12
DaveM@13: 	except ValueError:
DaveM@19: 		return None
DaveM@21: 	if H > 24 or M > 60:
DaveM@21: 		return None
DaveM@21: 	else:
DaveM@21: 		return (H,M)
DaveM@19: 	
DaveM@13: 
DaveM@13: