DaveM@23: #!/usr/bin/env python DaveM@23: DaveM@23: import csv DaveM@23: import time DaveM@23: import unicodedata DaveM@23: from geopy.geocoders import Nominatim DaveM@23: from geopy.exc import GeocoderTimedOut DaveM@23: import random DaveM@23: import pdb DaveM@23: DaveM@23: DEFAULT_TIME = None DaveM@23: # DEFAULT_TIME_M = None DaveM@23: DEAULT_LOCATION = None DaveM@23: DaveM@23: def regulateData(dataDict): DaveM@23: print("Parse %s"%(str(dataDict['ID']))) DaveM@23: p_DOBQ = "What is your partner's date of birth? Please use the format DD/MM/YYYY (for example, 29/03/1981)." DaveM@23: p_TOBQ = "At what exact time were your partner born? Please use the format HHMM (for example, 2204)." DaveM@23: DOB_DQ = "Which day (numeric) have you been born?" DaveM@23: DOB_MQ = "Which month have you been born?" DaveM@23: DOB_YQ = "Year Of Birth" DaveM@23: TOB_Q = "At what exact time were you born? Please use the format HHMM (for example, 2204)." DaveM@23: COB = "What is your place of birth? Please specify city and country (for example, San Francisco, USA)." DaveM@23: p_COB = "What is your partner's place of birth? Please specify city and country (for example, San Francisco, USA)." DaveM@23: dataDict['DOB'] = parseDOB(dataDict[DOB_DQ],dataDict[DOB_MQ],dataDict[DOB_YQ]) DaveM@23: # print (dataDict[DOB_DQ],dataDict[DOB_MQ],dataDict[DOB_YQ]) DaveM@23: # print dataDict['DOB'] DaveM@23: dataDict['TOB'] = parseTOB(dataDict[TOB_Q]) DaveM@23: dataDict['pDOB'] = parsePartnerDOB(dataDict[p_DOBQ]) DaveM@23: dataDict['pTOB'] = parseTOB(dataDict[p_TOBQ]) DaveM@23: # MAKE RANDOM PLACE DaveM@23: # dataDict['COB'] = (random.uniform(-90, 90),random.uniform(-90, 90)) DaveM@23: # dataDict['pCOB'] = (random.uniform(-90, 90),random.uniform(-90, 90)) DaveM@23: dataDict['COB'] = parseBirthTown(dataDict[COB]) DaveM@23: dataDict['pCOB'] = parseBirthTown(dataDict[p_COB]) DaveM@23: return dataDict DaveM@23: DaveM@23: DaveM@23: def parseBirthTown(s): DaveM@23: try: DaveM@23: s = s.encode('ascii') DaveM@23: except UnicodeDecodeError: DaveM@23: # pdb.set_trace() DaveM@23: s = s.decode('latin-1') DaveM@23: # s = unicodedata.normalize('NFKD',s.decode('utf-8')).encode('ascii','ignore') DaveM@23: timeoutTime = 2 DaveM@23: geolocator = Nominatim(timeout=timeoutTime) DaveM@23: while s is not [] and timeoutTime < 60: DaveM@23: try: DaveM@23: location = geolocator.geocode(s) DaveM@23: if location is not None: DaveM@23: # print(location.raw) DaveM@23: # print (location.latitude, location.longitude) DaveM@23: return (location.latitude, location.longitude, location.raw) DaveM@23: else: DaveM@23: s = s.split(' ',1) DaveM@23: if len(s) == 2: DaveM@23: s = s[1] DaveM@23: # print s DaveM@23: else: DaveM@23: s = DEAULT_LOCATION DaveM@23: except: DaveM@23: timeoutTime += 1 DaveM@23: print("Error: geocode failed on input %s, incrementing timeout time to %d"%(s,timeoutTime)) DaveM@23: time.sleep(5) DaveM@23: geolocator = Nominatim(timeout=timeoutTime) DaveM@23: # places = geograpy.get_place_context(text=s) DaveM@23: DaveM@23: def parsePartnerDOB(dob): DaveM@23: # print dob DaveM@23: # pdb.set_trace() DaveM@23: dob = dob.strip() DaveM@23: if(dob.count('-') == 2): DaveM@23: dob = dob.replace('-','/') DaveM@23: if(dob.count(' ') == 2): DaveM@23: dob = dob.replace(' ','/') DaveM@23: dob_ = dob.split('/') DaveM@23: if(len(dob_) != 3): DaveM@23: dob = dob.replace('/','').strip() DaveM@23: dob_ = [] DaveM@23: # print dob DaveM@23: if len(dob) == 8: # ddmmyyyy DaveM@23: dob_.append(dob[:2]) DaveM@23: dob_.append(dob[2:4]) DaveM@23: dob_.append(dob[4:]) DaveM@23: elif len(dob) == 7 and dob[1] == '1' and (dob[2] == '0' or dob[2] == '1' or dob[2] == '2'): # dmmyyyy DaveM@23: dob_.append(dob[0]) DaveM@23: dob_.append(dob[1:3]) DaveM@23: dob_.append(dob[3:]) DaveM@23: elif(len(dob) == 7): DaveM@23: if int(dob[:2]) > 31:# dmmyyyy DaveM@23: dob_.append(dob[0]) DaveM@23: dob_.append(dob[1:3]) DaveM@23: dob_.append(dob[3:]) DaveM@23: elif len(dob) == 7: # ddmyyyy DaveM@23: dob_.append(dob[0:2]) DaveM@23: dob_.append(dob[2]) DaveM@23: dob_.append(dob[3:]) DaveM@23: elif len(dob) == 6 and dob[3:4] != '19': # ddmmyy DaveM@23: dob_.append(dob[:2]) DaveM@23: dob_.append(dob[2:4]) DaveM@23: dob_.append(dob[4:]) DaveM@23: elif len(dob) == 5 and dob[1] == '1' and (dob[2] == '0' or dob[2] == '1' or dob[2] == '2'): # dmmyy DaveM@23: dob_.append(dob[0]) DaveM@23: dob_.append(dob[1:3]) DaveM@23: dob_.append(dob[3:]) DaveM@23: elif len(dob) == 5: # ddmyy DaveM@23: dob_.append(dob[:2]) DaveM@23: dob_.append(dob[2]) DaveM@23: dob_.append(dob[3:]) DaveM@23: elif len(dob) == 4: # dmyy DaveM@23: dob_.append(dob[0]) DaveM@23: dob_.append(dob[1]) DaveM@23: dob_.append(dob[2:]) DaveM@23: else: DaveM@23: if(len(dob) < 4): DaveM@23: return None DaveM@23: # print dob DaveM@23: # print filter(lambda x: x.isdigit(),dob) DaveM@23: print 'no / partnerDOB issue' DaveM@23: # deal with no /'s DaveM@23: try: DaveM@23: d = int(filter(lambda x: x.isdigit(),dob_[0])) DaveM@23: m = int(filter(lambda x: x.isdigit(),dob_[1])) DaveM@23: y = int(filter(lambda x: x.isdigit(),dob_[2])) DaveM@23: if y < 100: DaveM@23: y = y + 1900 DaveM@23: if (m > 12 and d <= 12): DaveM@23: temp = d DaveM@23: d = m DaveM@23: m = temp DaveM@23: if(d > 31 or d < 1 or m > 12 or m < 1 or y > 2017 or y < 1900): DaveM@23: print 'error with DOB '+d+'/'+m+'/'+y DaveM@23: pdb.set_trace() DaveM@23: except TypeError: DaveM@23: return None DaveM@23: # print (d,m,y) DaveM@23: return (d,m,y) DaveM@23: DaveM@23: def monthStringToNum(s): DaveM@23: # print 'inMonthStringToNum' DaveM@23: m = {'jan':1,'feb':2,'mar':3, DaveM@23: 'apr':4,'may':5,'jun':6,'jul':7,'aug':8, DaveM@23: 'sep':9,'oct':10,'nov':11,'dec':12} DaveM@23: s_ = s.strip()[:3].lower() DaveM@23: try: DaveM@23: out = m[s_] DaveM@23: return out DaveM@23: except: DaveM@23: raise ValueError('Not a month') DaveM@23: DaveM@23: def checkMonthDay(d,m): DaveM@23: if d > 31: # take first two digits of day DaveM@23: d = int(str(d)[:2]) DaveM@23: if d > 31: DaveM@23: d = int(str(d)[1]) DaveM@23: if m > 12 and d < 12: # Day and month wrong way round - American DaveM@23: temp = m DaveM@23: m = d DaveM@23: d = temp DaveM@23: if(m == 2): DaveM@23: if d <= 29: DaveM@23: return (True,d,m) DaveM@23: else: DaveM@23: return (False,d,m) DaveM@23: elif m in [4,6,9,11]: DaveM@23: if d <= 30: DaveM@23: return (True,d,m) DaveM@23: else: DaveM@23: return (False,d,m) DaveM@23: elif m <= 12 and d <= 31: DaveM@23: return (True,d,m) DaveM@23: else: DaveM@23: return (False,d,m) DaveM@23: DaveM@23: def parseDOB(d,m,y): DaveM@23: d = int(filter(lambda x: x.isdigit(),d)) DaveM@23: y = int(filter(lambda x: x.isdigit(),y)) DaveM@23: try: DaveM@23: # print m DaveM@23: m = monthStringToNum(m.strip()) DaveM@23: except ValueError: DaveM@23: m = int(m.strip()) DaveM@23: if(y < 100): DaveM@23: y = y + 1900 DaveM@23: (r,d,m) = checkMonthDay(d,m) DaveM@23: if not r: DaveM@23: print 'error with day month' DaveM@23: print (r,d,m) DaveM@23: return (d,m,y) DaveM@23: DaveM@23: def parseTOB(T): DaveM@23: timeFlag = None DaveM@23: T_ = T.replace('.','').lower().strip() DaveM@23: if 'am' in T_: DaveM@23: timeFlag = 0 DaveM@23: T = T_.replace('am','') DaveM@23: if 'pm' in T_: DaveM@23: timeFlag = 1 DaveM@23: T = T_.replace('pm','') DaveM@23: T = T.strip() DaveM@23: if T.count('.') == 1: DaveM@23: T = T.replace('.',':') DaveM@23: try: DaveM@23: if ':' in T: DaveM@23: T_ = T.split(':') DaveM@23: DaveM@23: H = int(T_[0]) DaveM@23: M = int(T_[1]) DaveM@23: else: DaveM@23: if len(T) == 4: DaveM@23: H = int(T[:2]) DaveM@23: M = int(T[2:]) DaveM@23: elif int(T) <= 24 : DaveM@23: H = int(T) DaveM@23: M = 0 DaveM@23: elif int(T) > 100: DaveM@23: H = int(T)/100 DaveM@23: M = int(T)%100 DaveM@23: else: DaveM@23: return None DaveM@23: if timeFlag is not None: DaveM@23: if timeFlag == 0: DaveM@23: H = H%12 DaveM@23: else: DaveM@23: H = H%12 + 12 DaveM@23: except ValueError: DaveM@23: return None DaveM@23: if H > 24 or M > 60: DaveM@23: return None DaveM@23: else: DaveM@23: return (H,M) DaveM@23: DaveM@23: DaveM@23: