DaveM@8: #!/usr/bin/env python DaveM@8: DaveM@8: import csv DaveM@8: import time DaveM@8: import unicodedata DaveM@8: from geopy.geocoders import Nominatim DaveM@8: from geopy.exc import GeocoderTimedOut DaveM@8: import random DaveM@8: import pdb DaveM@8: DaveM@8: DEFAULT_TIME_H = 12 DaveM@8: DEFAULT_TIME_M = 00 DaveM@8: DEAULT_LOCATION = 'USA' DaveM@8: DaveM@8: def parseCSV(filename): DaveM@8: stream = csv.DictReader(open(filename,'rb')) DaveM@8: dictList = [] DaveM@8: for line in stream: DaveM@8: dictList.append(regulateData(line)) DaveM@8: return dictList DaveM@8: DaveM@8: def regulateData(dataDict): DaveM@8: print("Parse %s"%(str(dataDict['ID']))) DaveM@8: p_DOBQ = "What is your partner's date of birth? Please use the format DD/MM/YYYY (for example, 29/03/1981)." DaveM@8: p_TOBQ = "At what exact time were your partner born? Please use the format HHMM (for example, 2204)." DaveM@8: DOB_DQ = "Which day (numeric) have you been born?" DaveM@8: DOB_MQ = "Which month have you been born?" DaveM@8: DOB_YQ = "Year Of Birth" DaveM@8: TOB_Q = "At what exact time were you born? Please use the format HHMM (for example, 2204)." DaveM@8: COB = "What is your place of birth? Please specify city and country (for example, San Francisco, USA)." DaveM@8: p_COB = "What is your partner's place of birth? Please specify city and country (for example, San Francisco, USA)." DaveM@8: dataDict['DOB'] = parseDOB(dataDict[DOB_DQ],dataDict[DOB_MQ],dataDict[DOB_YQ]) DaveM@8: # print (dataDict[DOB_DQ],dataDict[DOB_MQ],dataDict[DOB_YQ]) DaveM@8: # print dataDict['DOB'] DaveM@8: dataDict['TOB'] = parseTOB(dataDict[TOB_Q]) DaveM@8: dataDict['pDOB'] = parsePartnerDOB(dataDict[p_DOBQ]) DaveM@8: dataDict['pTOB'] = parseTOB(dataDict[p_TOBQ]) DaveM@8: # MAKE RANDOM PLACE DaveM@8: # dataDict['COB'] = (random.uniform(-90, 90),random.uniform(-90, 90)) DaveM@8: # dataDict['pCOB'] = (random.uniform(-90, 90),random.uniform(-90, 90)) DaveM@8: dataDict['COB'] = parseBirthTown(dataDict[COB]) DaveM@8: dataDict['pCOB'] = parseBirthTown(dataDict[p_COB]) DaveM@8: return dataDict DaveM@8: DaveM@8: DaveM@8: def parseBirthTown(s): DaveM@8: try: DaveM@8: s = s.encode('ascii') DaveM@8: except UnicodeDecodeError: DaveM@8: # pdb.set_trace() DaveM@8: s = s.decode('latin-1') DaveM@8: # s = unicodedata.normalize('NFKD',s.decode('utf-8')).encode('ascii','ignore') DaveM@8: timeoutTime = 2 DaveM@8: geolocator = Nominatim(timeout=timeoutTime) DaveM@8: while s is not [] and timeoutTime < 60: DaveM@8: try: DaveM@8: location = geolocator.geocode(s) DaveM@8: if location is not None: DaveM@8: # print(location.raw) DaveM@8: # print (location.latitude, location.longitude) DaveM@8: return (location.latitude, location.longitude, location.raw) DaveM@8: else: DaveM@8: s = s.split(' ',1) DaveM@8: if len(s) == 2: DaveM@8: s = s[1] DaveM@8: # print s DaveM@8: else: DaveM@8: s = DEAULT_LOCATION DaveM@8: except: DaveM@8: timeoutTime += 1 DaveM@8: print("Error: geocode failed on input %s, incrementing timeout time to %d"%(s,timeoutTime)) DaveM@8: time.sleep(5) DaveM@8: geolocator = Nominatim(timeout=timeoutTime) DaveM@8: # places = geograpy.get_place_context(text=s) DaveM@8: DaveM@8: def parsePartnerDOB(dob): DaveM@8: # print dob DaveM@8: # pdb.set_trace() DaveM@8: dob = dob.strip() DaveM@8: if(dob.count('-') == 2): DaveM@8: dob = dob.replace('-','/') DaveM@8: if(dob.count(' ') == 2): DaveM@8: dob = dob.replace(' ','/') DaveM@8: dob_ = dob.split('/') DaveM@8: if(len(dob_) != 3): DaveM@8: dob = dob.replace('/','').strip() DaveM@8: dob_ = [] DaveM@8: # print dob DaveM@8: if len(dob) == 8: # ddmmyyyy DaveM@8: dob_.append(dob[:2]) DaveM@8: dob_.append(dob[2:4]) DaveM@8: dob_.append(dob[4:]) DaveM@8: elif len(dob) == 7 and dob[1] == '1' and (dob[2] == '0' or dob[2] == '1' or dob[2] == '2'): # dmmyyyy DaveM@8: dob_.append(dob[0]) DaveM@8: dob_.append(dob[1:3]) DaveM@8: dob_.append(dob[3:]) DaveM@8: elif(len(dob) == 7): DaveM@8: if int(dob[:2]) > 31:# dmmyyyy DaveM@8: dob_.append(dob[0]) DaveM@8: dob_.append(dob[1:3]) DaveM@8: dob_.append(dob[3:]) DaveM@8: elif len(dob) == 7: # ddmyyyy DaveM@8: dob_.append(dob[0:2]) DaveM@8: dob_.append(dob[2]) DaveM@8: dob_.append(dob[3:]) DaveM@8: elif len(dob) == 6 and dob[3:4] != '19': # ddmmyy DaveM@8: dob_.append(dob[:2]) DaveM@8: dob_.append(dob[2:4]) DaveM@8: dob_.append(dob[4:]) DaveM@8: elif len(dob) == 5 and dob[1] == '1' and (dob[2] == '0' or dob[2] == '1' or dob[2] == '2'): # dmmyy DaveM@8: dob_.append(dob[0]) DaveM@8: dob_.append(dob[1:3]) DaveM@8: dob_.append(dob[3:]) DaveM@8: elif len(dob) == 5: # ddmyy DaveM@8: dob_.append(dob[:2]) DaveM@8: dob_.append(dob[2]) DaveM@8: dob_.append(dob[3:]) DaveM@8: elif len(dob) == 4: # dmyy DaveM@8: dob_.append(dob[0]) DaveM@8: dob_.append(dob[1]) DaveM@8: dob_.append(dob[2:]) DaveM@8: else: DaveM@8: if(len(dob) < 4): DaveM@8: return None DaveM@8: # print dob DaveM@8: # print filter(lambda x: x.isdigit(),dob) DaveM@8: print 'no / partnerDOB issue' DaveM@8: # deal with no /'s DaveM@8: try: DaveM@8: d = int(filter(lambda x: x.isdigit(),dob_[0])) DaveM@8: m = int(filter(lambda x: x.isdigit(),dob_[1])) DaveM@8: y = int(filter(lambda x: x.isdigit(),dob_[2])) DaveM@8: if y < 100: DaveM@8: y = y + 1900 DaveM@8: if (m > 12 and d <= 12): DaveM@8: temp = d DaveM@8: d = m DaveM@8: m = temp DaveM@8: if(d > 31 or d < 1 or m > 12 or m < 1 or y > 2017 or y < 1900): DaveM@8: print 'error with DOB '+d+'/'+m+'/'+y DaveM@8: pdb.set_trace() DaveM@8: except TypeError: DaveM@8: return None DaveM@8: # print (d,m,y) DaveM@8: return (d,m,y) DaveM@8: DaveM@8: def monthStringToNum(s): DaveM@8: # print 'inMonthStringToNum' DaveM@8: m = {'jan':1,'feb':2,'mar':3, DaveM@8: 'apr':4,'may':5,'jun':6,'jul':7,'aug':8, DaveM@8: 'sep':9,'oct':10,'nov':11,'dec':12} DaveM@8: s_ = s.strip()[:3].lower() DaveM@8: try: DaveM@8: out = m[s_] DaveM@8: return out DaveM@8: except: DaveM@8: raise ValueError('Not a month') DaveM@8: DaveM@8: def checkMonthDay(d,m): DaveM@8: if d > 31: # take first two digits of day DaveM@8: d = int(str(d)[:2]) DaveM@8: if d > 31: DaveM@8: d = int(str(d)[1]) DaveM@8: if m > 12 and d < 12: # Day and month wrong way round - American DaveM@8: temp = m DaveM@8: m = d DaveM@8: d = temp DaveM@8: if(m == 2): DaveM@8: if d <= 29: DaveM@8: return (True,d,m) DaveM@8: else: DaveM@8: return (False,d,m) DaveM@8: elif m in [4,6,9,11]: DaveM@8: if d <= 30: DaveM@8: return (True,d,m) DaveM@8: else: DaveM@8: return (False,d,m) DaveM@8: elif m <= 12 and d <= 31: DaveM@8: return (True,d,m) DaveM@8: else: DaveM@8: return (False,d,m) DaveM@8: DaveM@8: def parseDOB(d,m,y): DaveM@8: d = int(filter(lambda x: x.isdigit(),d)) DaveM@8: y = int(filter(lambda x: x.isdigit(),y)) DaveM@8: try: DaveM@8: # print m DaveM@8: m = monthStringToNum(m.strip()) DaveM@8: except ValueError: DaveM@8: m = int(m.strip()) DaveM@8: if(y < 100): DaveM@8: y = y + 1900 DaveM@8: (r,d,m) = checkMonthDay(d,m) DaveM@8: if not r: DaveM@8: print 'error with day month' DaveM@8: print (r,d,m) DaveM@8: return (d,m,y) DaveM@8: DaveM@8: def parseTOB(T): DaveM@8: # pdb.set_trace() DaveM@8: timeFlag = None DaveM@8: T_ = T.replace('.','').lower().strip() DaveM@8: if 'am' in T_: DaveM@8: timeFlag = 0 DaveM@8: T = T_.replace('am','') DaveM@8: if 'pm' in T_: DaveM@8: timeFlag = 1 DaveM@8: T = T_.replace('pm','') DaveM@8: T = T.strip() DaveM@8: if T.count('.') == 1: DaveM@8: T = T.replace('.',':') DaveM@8: try: DaveM@8: if ':' in T: DaveM@8: T_ = T.split(':') DaveM@8: DaveM@8: H = int(T_[0]) DaveM@8: M = int(T_[1]) DaveM@8: else: DaveM@8: if len(T) == 4: DaveM@8: H = int(T[:2]) DaveM@8: M = int(T[2:]) DaveM@8: elif int(T) <= 24 : DaveM@8: H = int(T) DaveM@8: M = 0 DaveM@8: elif int(T) > 100: DaveM@8: H = int(T)/100 DaveM@8: M = int(T)%100 DaveM@8: if timeFlag is not None: DaveM@8: if timeFlag == 0: DaveM@8: H = H%12 DaveM@8: else: DaveM@8: H = H%12 + 12 DaveM@8: except ValueError: DaveM@8: H = DEFAULT_TIME_H DaveM@8: M = DEFAULT_TIME_M DaveM@8: return (H,M) DaveM@8: DaveM@8: def makePayload(dataDict): DaveM@8: if type(dataDict['COB']) is str: DaveM@8: cob_0 = float(dataDict['COB'].split(',')[0][1:]) DaveM@8: cob_1 = float(dataDict['COB'].split(',')[1]) DaveM@8: dataDict['COB'] = (cob_0,cob_1) DaveM@8: if type(dataDict['pCOB']) is str: DaveM@8: pcob_0 = float(dataDict['pCOB'].split(',')[0][1:]) DaveM@8: pcob_1 = float(dataDict['pCOB'].split(',')[1]) DaveM@8: dataDict['pCOB'] = (pcob_0,pcob_1) DaveM@8: if type(dataDict['DOB']) is str: DaveM@8: dataDict['DOB'] = dataDict['DOB'][1:-1].split(',') DaveM@8: if type(dataDict['pDOB']) is str: DaveM@8: dataDict['pDOB'] = dataDict['pDOB'][1:-1].split(',') DaveM@8: if type(dataDict['TOB']) is str: DaveM@8: dataDict['TOB'] = dataDict['TOB'][1:-1].split(',') DaveM@8: if type(dataDict['pTOB']) is str: DaveM@8: dataDict['pTOB'] = dataDict['pTOB'][1:-1].split(',') DaveM@8: # pdb.set_trace() DaveM@8: DaveM@8: print dataDict['pDOB'] DaveM@8: DaveM@8: R = {'send_calculation':'1', #Req DaveM@8: 'muz_narozeni_den':dataDict['DOB'][0], DaveM@8: 'muz_narozeni_mesic':dataDict['DOB'][1], DaveM@8: 'muz_narozeni_rok':dataDict['DOB'][2], DaveM@8: 'muz_narozeni_hodina':dataDict['TOB'][0], DaveM@8: 'muz_narozeni_minuta':dataDict['TOB'][1], DaveM@8: 'muz_narozeni_city':'', DaveM@8: 'muz_narozeni_mesto_hidden':'Manually+place%3A+%C2%B0%27N%2C+%C2%B0%27E',#auto DaveM@8: 'muz_narozeni_stat_hidden':'XX', DaveM@8: 'muz_narozeni_podstat_kratky_hidden':'', DaveM@8: 'muz_narozeni_podstat_hidden':'', DaveM@8: 'muz_narozeni_podstat2_kratky_hidden':'', DaveM@8: 'muz_narozeni_podstat3_kratky_hidden':'', DaveM@8: 'muz_narozeni_input_hidden':'', DaveM@8: 'muz_narozeni_sirka_stupne':str(abs(dataDict['COB'][0])).split('.')[0], DaveM@8: 'muz_narozeni_sirka_minuty':str(float('0.'+str(dataDict['COB'][0]).split('.')[1])*60).split('.')[0], DaveM@8: 'muz_narozeni_sirka_smer': '1' if dataDict['COB'][0]<0 else '0', #address N Dir (0':'N',1':'S) DaveM@8: 'muz_narozeni_delka_stupne':str(abs(dataDict['COB'][1])).split('.')[0], #address E - Main DaveM@8: 'muz_narozeni_delka_minuty':str(float('0.'+str(dataDict['COB'][1]).split('.')[1])*60).split('.')[0], DaveM@8: 'muz_narozeni_delka_smer': '1' if dataDict['COB'][1]<0 else '0', #address E Dir (0':'E',1':'W) DaveM@8: 'muz_narozeni_timezone_form':'auto', DaveM@8: 'muz_narozeni_timezone_dst_form':'auto', DaveM@8: 'send_calculation':'1', DaveM@8: 'zena_narozeni_den':dataDict['pDOB'][0], DaveM@8: 'zena_narozeni_mesic':dataDict['pDOB'][1], DaveM@8: 'zena_narozeni_rok':dataDict['pDOB'][2], DaveM@8: 'zena_narozeni_hodina':dataDict['pTOB'][0], DaveM@8: 'zena_narozeni_minuta':dataDict['pTOB'][1], DaveM@8: 'zena_narozeni_city':'', DaveM@8: 'zena_narozeni_mesto_hidden':'Manually+place%3A+%C2%B0%27N%2C+%C2%B0%27E', DaveM@8: 'zena_narozeni_stat_hidden':'XX', DaveM@8: 'zena_narozeni_podstat_kratky_hidden':'', DaveM@8: 'zena_narozeni_podstat_hidden':'', DaveM@8: 'zena_narozeni_podstat2_kratky_hidden':'', DaveM@8: 'zena_narozeni_podstat3_kratky_hidden':'', DaveM@8: 'zena_narozeni_input_hidden':'', DaveM@8: 'zena_narozeni_sirka_stupne':str(abs(dataDict['pCOB'][0])).split('.')[0], DaveM@8: 'zena_narozeni_sirka_minuty':str(float('0.'+str(dataDict['pCOB'][0]).split('.')[1])*60).split('.')[0], DaveM@8: 'zena_narozeni_sirka_smer': '1' if dataDict['pCOB'][0]<0 else '0', DaveM@8: 'zena_narozeni_delka_stupne':str(abs(dataDict['pCOB'][1])).split('.')[0], DaveM@8: 'zena_narozeni_delka_minuty':str(float('0.'+str(dataDict['pCOB'][1]).split('.')[1])*60).split('.')[0], DaveM@8: 'zena_narozeni_delka_smer': '1' if dataDict['pCOB'][1]<0 else '0', DaveM@8: 'zena_narozeni_timezone_form':'auto', DaveM@8: 'zena_narozeni_timezone_dst_form':'auto', DaveM@8: 'switch_interpretations':'0', DaveM@8: 'house_system':'placidus', DaveM@8: 'uhel_orbis':'#tabs_redraw'} DaveM@8: return R DaveM@8: DaveM@8: DaveM@8: DaveM@8: