comparison V5/dParse.py @ 23:11d4e438045e

make version 5
author DaveM
date Mon, 09 Apr 2018 15:07:21 +0100
parents
children
comparison
equal deleted inserted replaced
22:a5b8e2b91d8f 23:11d4e438045e
1 #!/usr/bin/env python
2
3 import csv
4 import time
5 import unicodedata
6 from geopy.geocoders import Nominatim
7 from geopy.exc import GeocoderTimedOut
8 import random
9 import pdb
10
11 DEFAULT_TIME = None
12 # DEFAULT_TIME_M = None
13 DEAULT_LOCATION = None
14
15 def regulateData(dataDict):
16 print("Parse %s"%(str(dataDict['ID'])))
17 p_DOBQ = "What is your partner's date of birth? Please use the format DD/MM/YYYY (for example, 29/03/1981)."
18 p_TOBQ = "At what exact time were your partner born? Please use the format HHMM (for example, 2204)."
19 DOB_DQ = "Which day (numeric) have you been born?"
20 DOB_MQ = "Which month have you been born?"
21 DOB_YQ = "Year Of Birth"
22 TOB_Q = "At what exact time were you born? Please use the format HHMM (for example, 2204)."
23 COB = "What is your place of birth? Please specify city and country (for example, San Francisco, USA)."
24 p_COB = "What is your partner's place of birth? Please specify city and country (for example, San Francisco, USA)."
25 dataDict['DOB'] = parseDOB(dataDict[DOB_DQ],dataDict[DOB_MQ],dataDict[DOB_YQ])
26 # print (dataDict[DOB_DQ],dataDict[DOB_MQ],dataDict[DOB_YQ])
27 # print dataDict['DOB']
28 dataDict['TOB'] = parseTOB(dataDict[TOB_Q])
29 dataDict['pDOB'] = parsePartnerDOB(dataDict[p_DOBQ])
30 dataDict['pTOB'] = parseTOB(dataDict[p_TOBQ])
31 # MAKE RANDOM PLACE
32 # dataDict['COB'] = (random.uniform(-90, 90),random.uniform(-90, 90))
33 # dataDict['pCOB'] = (random.uniform(-90, 90),random.uniform(-90, 90))
34 dataDict['COB'] = parseBirthTown(dataDict[COB])
35 dataDict['pCOB'] = parseBirthTown(dataDict[p_COB])
36 return dataDict
37
38
39 def parseBirthTown(s):
40 try:
41 s = s.encode('ascii')
42 except UnicodeDecodeError:
43 # pdb.set_trace()
44 s = s.decode('latin-1')
45 # s = unicodedata.normalize('NFKD',s.decode('utf-8')).encode('ascii','ignore')
46 timeoutTime = 2
47 geolocator = Nominatim(timeout=timeoutTime)
48 while s is not [] and timeoutTime < 60:
49 try:
50 location = geolocator.geocode(s)
51 if location is not None:
52 # print(location.raw)
53 # print (location.latitude, location.longitude)
54 return (location.latitude, location.longitude, location.raw)
55 else:
56 s = s.split(' ',1)
57 if len(s) == 2:
58 s = s[1]
59 # print s
60 else:
61 s = DEAULT_LOCATION
62 except:
63 timeoutTime += 1
64 print("Error: geocode failed on input %s, incrementing timeout time to %d"%(s,timeoutTime))
65 time.sleep(5)
66 geolocator = Nominatim(timeout=timeoutTime)
67 # places = geograpy.get_place_context(text=s)
68
69 def parsePartnerDOB(dob):
70 # print dob
71 # pdb.set_trace()
72 dob = dob.strip()
73 if(dob.count('-') == 2):
74 dob = dob.replace('-','/')
75 if(dob.count(' ') == 2):
76 dob = dob.replace(' ','/')
77 dob_ = dob.split('/')
78 if(len(dob_) != 3):
79 dob = dob.replace('/','').strip()
80 dob_ = []
81 # print dob
82 if len(dob) == 8: # ddmmyyyy
83 dob_.append(dob[:2])
84 dob_.append(dob[2:4])
85 dob_.append(dob[4:])
86 elif len(dob) == 7 and dob[1] == '1' and (dob[2] == '0' or dob[2] == '1' or dob[2] == '2'): # dmmyyyy
87 dob_.append(dob[0])
88 dob_.append(dob[1:3])
89 dob_.append(dob[3:])
90 elif(len(dob) == 7):
91 if int(dob[:2]) > 31:# dmmyyyy
92 dob_.append(dob[0])
93 dob_.append(dob[1:3])
94 dob_.append(dob[3:])
95 elif len(dob) == 7: # ddmyyyy
96 dob_.append(dob[0:2])
97 dob_.append(dob[2])
98 dob_.append(dob[3:])
99 elif len(dob) == 6 and dob[3:4] != '19': # ddmmyy
100 dob_.append(dob[:2])
101 dob_.append(dob[2:4])
102 dob_.append(dob[4:])
103 elif len(dob) == 5 and dob[1] == '1' and (dob[2] == '0' or dob[2] == '1' or dob[2] == '2'): # dmmyy
104 dob_.append(dob[0])
105 dob_.append(dob[1:3])
106 dob_.append(dob[3:])
107 elif len(dob) == 5: # ddmyy
108 dob_.append(dob[:2])
109 dob_.append(dob[2])
110 dob_.append(dob[3:])
111 elif len(dob) == 4: # dmyy
112 dob_.append(dob[0])
113 dob_.append(dob[1])
114 dob_.append(dob[2:])
115 else:
116 if(len(dob) < 4):
117 return None
118 # print dob
119 # print filter(lambda x: x.isdigit(),dob)
120 print 'no / partnerDOB issue'
121 # deal with no /'s
122 try:
123 d = int(filter(lambda x: x.isdigit(),dob_[0]))
124 m = int(filter(lambda x: x.isdigit(),dob_[1]))
125 y = int(filter(lambda x: x.isdigit(),dob_[2]))
126 if y < 100:
127 y = y + 1900
128 if (m > 12 and d <= 12):
129 temp = d
130 d = m
131 m = temp
132 if(d > 31 or d < 1 or m > 12 or m < 1 or y > 2017 or y < 1900):
133 print 'error with DOB '+d+'/'+m+'/'+y
134 pdb.set_trace()
135 except TypeError:
136 return None
137 # print (d,m,y)
138 return (d,m,y)
139
140 def monthStringToNum(s):
141 # print 'inMonthStringToNum'
142 m = {'jan':1,'feb':2,'mar':3,
143 'apr':4,'may':5,'jun':6,'jul':7,'aug':8,
144 'sep':9,'oct':10,'nov':11,'dec':12}
145 s_ = s.strip()[:3].lower()
146 try:
147 out = m[s_]
148 return out
149 except:
150 raise ValueError('Not a month')
151
152 def checkMonthDay(d,m):
153 if d > 31: # take first two digits of day
154 d = int(str(d)[:2])
155 if d > 31:
156 d = int(str(d)[1])
157 if m > 12 and d < 12: # Day and month wrong way round - American
158 temp = m
159 m = d
160 d = temp
161 if(m == 2):
162 if d <= 29:
163 return (True,d,m)
164 else:
165 return (False,d,m)
166 elif m in [4,6,9,11]:
167 if d <= 30:
168 return (True,d,m)
169 else:
170 return (False,d,m)
171 elif m <= 12 and d <= 31:
172 return (True,d,m)
173 else:
174 return (False,d,m)
175
176 def parseDOB(d,m,y):
177 d = int(filter(lambda x: x.isdigit(),d))
178 y = int(filter(lambda x: x.isdigit(),y))
179 try:
180 # print m
181 m = monthStringToNum(m.strip())
182 except ValueError:
183 m = int(m.strip())
184 if(y < 100):
185 y = y + 1900
186 (r,d,m) = checkMonthDay(d,m)
187 if not r:
188 print 'error with day month'
189 print (r,d,m)
190 return (d,m,y)
191
192 def parseTOB(T):
193 timeFlag = None
194 T_ = T.replace('.','').lower().strip()
195 if 'am' in T_:
196 timeFlag = 0
197 T = T_.replace('am','')
198 if 'pm' in T_:
199 timeFlag = 1
200 T = T_.replace('pm','')
201 T = T.strip()
202 if T.count('.') == 1:
203 T = T.replace('.',':')
204 try:
205 if ':' in T:
206 T_ = T.split(':')
207
208 H = int(T_[0])
209 M = int(T_[1])
210 else:
211 if len(T) == 4:
212 H = int(T[:2])
213 M = int(T[2:])
214 elif int(T) <= 24 :
215 H = int(T)
216 M = 0
217 elif int(T) > 100:
218 H = int(T)/100
219 M = int(T)%100
220 else:
221 return None
222 if timeFlag is not None:
223 if timeFlag == 0:
224 H = H%12
225 else:
226 H = H%12 + 12
227 except ValueError:
228 return None
229 if H > 24 or M > 60:
230 return None
231 else:
232 return (H,M)
233
234
235