annotate validate_temperature_data.py @ 4:9ca3ff390666 draft

Uploaded
author greg
date Tue, 27 Nov 2018 09:55:22 -0500
parents 3de939253e55
children 467305a51829
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
4290854f3af5 Uploaded
greg
parents:
diff changeset
1 #!/usr/bin/env python
4290854f3af5 Uploaded
greg
parents:
diff changeset
2 import argparse
4290854f3af5 Uploaded
greg
parents:
diff changeset
3 import datetime
4290854f3af5 Uploaded
greg
parents:
diff changeset
4 import decimal
4290854f3af5 Uploaded
greg
parents:
diff changeset
5 import re
4290854f3af5 Uploaded
greg
parents:
diff changeset
6 import shutil
4290854f3af5 Uploaded
greg
parents:
diff changeset
7 import sys
4290854f3af5 Uploaded
greg
parents:
diff changeset
8
4290854f3af5 Uploaded
greg
parents:
diff changeset
9 parser = argparse.ArgumentParser()
4290854f3af5 Uploaded
greg
parents:
diff changeset
10 parser.add_argument('--data_type', dest='data_type', default=None, help='Temperature data type, normals or actuals')
4290854f3af5 Uploaded
greg
parents:
diff changeset
11 parser.add_argument('--input_actuals', dest='input_actuals', default=None, help='Daily actuals temperature data')
4290854f3af5 Uploaded
greg
parents:
diff changeset
12 parser.add_argument('--input_normals', dest='input_normals', default=None, help='30 year normals temperature data')
4290854f3af5 Uploaded
greg
parents:
diff changeset
13 parser.add_argument('--output', dest='output', help='Output dataset'),
4290854f3af5 Uploaded
greg
parents:
diff changeset
14 args = parser.parse_args()
4290854f3af5 Uploaded
greg
parents:
diff changeset
15
4290854f3af5 Uploaded
greg
parents:
diff changeset
16 ACTUALS_HEADER = "LATITUDE,LONGITUDE,DATE,DOY,TMIN,TMAX"
4290854f3af5 Uploaded
greg
parents:
diff changeset
17 NORMALS_HEADER = "stationid,latitude,longitude,elev_m,name,st,mmdd,doy,tmin,tmax"
4290854f3af5 Uploaded
greg
parents:
diff changeset
18
4290854f3af5 Uploaded
greg
parents:
diff changeset
19 def add_error_msg(accumulated_msgs, msg):
4290854f3af5 Uploaded
greg
parents:
diff changeset
20 return "%s\n%s" % (accumulated_msgs, msg)
4290854f3af5 Uploaded
greg
parents:
diff changeset
21
4290854f3af5 Uploaded
greg
parents:
diff changeset
22
4290854f3af5 Uploaded
greg
parents:
diff changeset
23 def empty_value(line_no, label, accumulated_msgs):
4290854f3af5 Uploaded
greg
parents:
diff changeset
24 return add_error_msg(accumulated_msgs, "The required %s value is missing on line %d." % (label, line_no))
4290854f3af5 Uploaded
greg
parents:
diff changeset
25
4290854f3af5 Uploaded
greg
parents:
diff changeset
26
4290854f3af5 Uploaded
greg
parents:
diff changeset
27 def stop_error(msg):
4290854f3af5 Uploaded
greg
parents:
diff changeset
28 sys.exit(msg)
4290854f3af5 Uploaded
greg
parents:
diff changeset
29
4290854f3af5 Uploaded
greg
parents:
diff changeset
30
4290854f3af5 Uploaded
greg
parents:
diff changeset
31 def validate_date_string(line_no, date_string, accumulated_msgs):
4290854f3af5 Uploaded
greg
parents:
diff changeset
32 try:
4290854f3af5 Uploaded
greg
parents:
diff changeset
33 datetime.datetime.strptime(date_string, '%Y-%m-%d')
4290854f3af5 Uploaded
greg
parents:
diff changeset
34 return accumulated_msgs
4290854f3af5 Uploaded
greg
parents:
diff changeset
35 except ValueError:
4290854f3af5 Uploaded
greg
parents:
diff changeset
36 return add_error_msg(accumulated_msgs, "Line %d contains an incorrect date format (%s must be YYYY-MM-DD)." % (line_no, date_string))
4290854f3af5 Uploaded
greg
parents:
diff changeset
37
4290854f3af5 Uploaded
greg
parents:
diff changeset
38
4290854f3af5 Uploaded
greg
parents:
diff changeset
39 def validate_decimal(line_no, decimal_string, accumulated_msgs, label):
4290854f3af5 Uploaded
greg
parents:
diff changeset
40 try:
4290854f3af5 Uploaded
greg
parents:
diff changeset
41 decimal.Decimal(decimal_string)
4290854f3af5 Uploaded
greg
parents:
diff changeset
42 return accumulated_msgs
4290854f3af5 Uploaded
greg
parents:
diff changeset
43 except Exception:
4290854f3af5 Uploaded
greg
parents:
diff changeset
44 return add_error_msg(accumulated_msgs, "Line %d contains an incorrect %s decimal value (%s)." % (line_no, label, decimal_string))
4290854f3af5 Uploaded
greg
parents:
diff changeset
45
4290854f3af5 Uploaded
greg
parents:
diff changeset
46
4290854f3af5 Uploaded
greg
parents:
diff changeset
47 def validate_integer(line_no, integer_string, accumulated_msgs, label):
4290854f3af5 Uploaded
greg
parents:
diff changeset
48 if integer_string.isdigit():
4290854f3af5 Uploaded
greg
parents:
diff changeset
49 return accumulated_msgs
4
9ca3ff390666 Uploaded
greg
parents: 3
diff changeset
50 return add_error_msg(accumulated_msgs, "Line %d contains an incorrect %s integer value (%s)." % (line_no, label, integer_string))
0
4290854f3af5 Uploaded
greg
parents:
diff changeset
51
4290854f3af5 Uploaded
greg
parents:
diff changeset
52
4290854f3af5 Uploaded
greg
parents:
diff changeset
53 def validate_mmdd(line_no, mmdd, accumulated_msgs):
4290854f3af5 Uploaded
greg
parents:
diff changeset
54 try:
3
3de939253e55 Uploaded
greg
parents: 2
diff changeset
55 datetime.datetime.strptime(mmdd, '%m-%d')
0
4290854f3af5 Uploaded
greg
parents:
diff changeset
56 return accumulated_msgs
4290854f3af5 Uploaded
greg
parents:
diff changeset
57 except ValueError:
3
3de939253e55 Uploaded
greg
parents: 2
diff changeset
58 return add_error_msg(accumulated_msgs, "Line %d contains an incorrect date format (%s must be mm-dd)." % (line_no, mmdd))
0
4290854f3af5 Uploaded
greg
parents:
diff changeset
59
4290854f3af5 Uploaded
greg
parents:
diff changeset
60
4290854f3af5 Uploaded
greg
parents:
diff changeset
61 accumulated_msgs = ""
4290854f3af5 Uploaded
greg
parents:
diff changeset
62 # Parse the input file, skipping the header, and validating
4290854f3af5 Uploaded
greg
parents:
diff changeset
63 # that each data line consists of 31 comma-separated items.
4290854f3af5 Uploaded
greg
parents:
diff changeset
64 if args.data_type == "normals":
4290854f3af5 Uploaded
greg
parents:
diff changeset
65 input_file = args.input_normals
4290854f3af5 Uploaded
greg
parents:
diff changeset
66 header = NORMALS_HEADER
4290854f3af5 Uploaded
greg
parents:
diff changeset
67 last_doy = 0
4290854f3af5 Uploaded
greg
parents:
diff changeset
68 num_normals_rows = 0
4290854f3af5 Uploaded
greg
parents:
diff changeset
69 else:
4290854f3af5 Uploaded
greg
parents:
diff changeset
70 input_file = args.input_actuals
4290854f3af5 Uploaded
greg
parents:
diff changeset
71 header = ACTUALS_HEADER
4290854f3af5 Uploaded
greg
parents:
diff changeset
72 with open(input_file, "r") as ih:
4290854f3af5 Uploaded
greg
parents:
diff changeset
73 for i, line in enumerate(ih):
4290854f3af5 Uploaded
greg
parents:
diff changeset
74 line = line.rstrip("\r\n")
4290854f3af5 Uploaded
greg
parents:
diff changeset
75 if i == 0:
4290854f3af5 Uploaded
greg
parents:
diff changeset
76 if line != header:
4290854f3af5 Uploaded
greg
parents:
diff changeset
77 accumulated_msgs = add_error_msg(accumulated_msgs, "The header is invalid, must be %s" % header)
4290854f3af5 Uploaded
greg
parents:
diff changeset
78 items = line.split(",")
4290854f3af5 Uploaded
greg
parents:
diff changeset
79 if args.data_type == "normals":
4290854f3af5 Uploaded
greg
parents:
diff changeset
80 if i > 367:
4290854f3af5 Uploaded
greg
parents:
diff changeset
81 accumulated_msgs = add_error_msg(accumulated_msgs, "The input file contains more than 367 lines (must be 1 header line and 366 data lines).")
4290854f3af5 Uploaded
greg
parents:
diff changeset
82 stop_error(accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
83 if len(items) != 10:
4290854f3af5 Uploaded
greg
parents:
diff changeset
84 accumulated_msgs = add_error_msg(accumulated_msgs, "Line %d contains %s columns, (must be 10)." % (i, len(items)))
4290854f3af5 Uploaded
greg
parents:
diff changeset
85 stop_error(accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
86 stationid = items[0]
2
57c418290b61 Uploaded
greg
parents: 1
diff changeset
87 if len(stationid) == 0:
0
4290854f3af5 Uploaded
greg
parents:
diff changeset
88 accumulated_msgs = empty_value(i, "stationid", accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
89 latitude = items[1]
4290854f3af5 Uploaded
greg
parents:
diff changeset
90 accumulated_msgs = validate_decimal(i, latitude, accumulated_msgs, "latitude")
4290854f3af5 Uploaded
greg
parents:
diff changeset
91 longitude = items[2]
4290854f3af5 Uploaded
greg
parents:
diff changeset
92 accumulated_msgs = validate_decimal(i, longitude, accumulated_msgs, "longitude")
4290854f3af5 Uploaded
greg
parents:
diff changeset
93 elev_m = items[3]
4290854f3af5 Uploaded
greg
parents:
diff changeset
94 accumulated_msgs = validate_decimal(i, elev_m, accumulated_msgs, "elev_m")
4290854f3af5 Uploaded
greg
parents:
diff changeset
95 name = items[4]
4290854f3af5 Uploaded
greg
parents:
diff changeset
96 if len(name) == 0:
4290854f3af5 Uploaded
greg
parents:
diff changeset
97 accumulated_msgs = empty_value(i, "name", accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
98 st = items[5]
4290854f3af5 Uploaded
greg
parents:
diff changeset
99 if len(st) == 0:
4290854f3af5 Uploaded
greg
parents:
diff changeset
100 accumulated_msgs = empty_value(i, "st", accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
101 mmdd = items[6]
4290854f3af5 Uploaded
greg
parents:
diff changeset
102 accumulated_msgs = validate_mmdd(i, mmdd, accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
103 doy = items[7]
4290854f3af5 Uploaded
greg
parents:
diff changeset
104 accumulated_msgs = validate_integer(i, doy, accumulated_msgs, "doy")
4290854f3af5 Uploaded
greg
parents:
diff changeset
105 # Make sure the DOY values are consecutive.
4290854f3af5 Uploaded
greg
parents:
diff changeset
106 try:
4290854f3af5 Uploaded
greg
parents:
diff changeset
107 if int(doy) != (last_doy + 1):
4290854f3af5 Uploaded
greg
parents:
diff changeset
108 accumulated_msgs = add_error_msg(accumulated_msgs, "Line %d contains a DOY (%s) that is not conexcutive." % (i, doy))
4290854f3af5 Uploaded
greg
parents:
diff changeset
109 stop_error(accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
110 else:
4290854f3af5 Uploaded
greg
parents:
diff changeset
111 last_doy += 1
4290854f3af5 Uploaded
greg
parents:
diff changeset
112 except Exception:
4290854f3af5 Uploaded
greg
parents:
diff changeset
113 # The error for an invalid integer was captured above.
4290854f3af5 Uploaded
greg
parents:
diff changeset
114 pass
4290854f3af5 Uploaded
greg
parents:
diff changeset
115 tmin = items[8]
4290854f3af5 Uploaded
greg
parents:
diff changeset
116 accumulated_msgs = validate_decimal(i, tmin, accumulated_msgs, "tmin")
4290854f3af5 Uploaded
greg
parents:
diff changeset
117 tmax = items[9]
4290854f3af5 Uploaded
greg
parents:
diff changeset
118 accumulated_msgs = validate_decimal(i, tmax, accumulated_msgs, "tmax")
4290854f3af5 Uploaded
greg
parents:
diff changeset
119 num_normals_rows += 1
4290854f3af5 Uploaded
greg
parents:
diff changeset
120 else:
4290854f3af5 Uploaded
greg
parents:
diff changeset
121 if i > 367:
4290854f3af5 Uploaded
greg
parents:
diff changeset
122 accumulated_msgs = add_error_msg(accumulated_msgs, "The input file contains more than 367 lines (must be 1 header line and no more than 366 data lines).")
4290854f3af5 Uploaded
greg
parents:
diff changeset
123 stop_error(accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
124 if len(items) != 6:
4290854f3af5 Uploaded
greg
parents:
diff changeset
125 accumulated_msgs = add_error_msg(accumulated_msgs, "Line %d contains %s columns, (must be 6)." % (i, len(items)))
4290854f3af5 Uploaded
greg
parents:
diff changeset
126 stop_error(accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
127 latitude = items[0]
4290854f3af5 Uploaded
greg
parents:
diff changeset
128 accumulated_msgs = validate_decimal(i, latitude, accumulated_msgs, "LATITUDE")
4290854f3af5 Uploaded
greg
parents:
diff changeset
129 longitude = items[1]
4290854f3af5 Uploaded
greg
parents:
diff changeset
130 accumulated_msgs = validate_decimal(i, longitude, accumulated_msgs, "LONGITUDE")
4290854f3af5 Uploaded
greg
parents:
diff changeset
131 date_string = items[2]
4290854f3af5 Uploaded
greg
parents:
diff changeset
132 accumulated_msgs = validate_date_string(line_no, date_string, accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
133 doy = items[3]
4290854f3af5 Uploaded
greg
parents:
diff changeset
134 accumulated_msgs = validate_integer(i, doy, accumulated_msgs, "doy")
4290854f3af5 Uploaded
greg
parents:
diff changeset
135 # Make sure the DOY values are consecutive.
4290854f3af5 Uploaded
greg
parents:
diff changeset
136 if i==0:
4290854f3af5 Uploaded
greg
parents:
diff changeset
137 try:
4290854f3af5 Uploaded
greg
parents:
diff changeset
138 last_doy = int(doy)
4290854f3af5 Uploaded
greg
parents:
diff changeset
139 except Exception:
4290854f3af5 Uploaded
greg
parents:
diff changeset
140 # The error for an invalid integer was captured above.
4290854f3af5 Uploaded
greg
parents:
diff changeset
141 pass
4290854f3af5 Uploaded
greg
parents:
diff changeset
142 else:
4290854f3af5 Uploaded
greg
parents:
diff changeset
143 try:
4290854f3af5 Uploaded
greg
parents:
diff changeset
144 if int(doy) != (last_doy + 1):
4290854f3af5 Uploaded
greg
parents:
diff changeset
145 accumulated_msgs = add_error_msg(accumulated_msgs, "Line %d contains a DOY (%s) that is not conexcutive." % (i, doy))
4290854f3af5 Uploaded
greg
parents:
diff changeset
146 stop_error(accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
147 else:
4290854f3af5 Uploaded
greg
parents:
diff changeset
148 last_doy += 1
4290854f3af5 Uploaded
greg
parents:
diff changeset
149 except Exception:
4290854f3af5 Uploaded
greg
parents:
diff changeset
150 # The error for an invalid integer was captured above.
4290854f3af5 Uploaded
greg
parents:
diff changeset
151 pass
4290854f3af5 Uploaded
greg
parents:
diff changeset
152 tmin = items[8]
4290854f3af5 Uploaded
greg
parents:
diff changeset
153 accumulated_msgs = validate_decimal(i, tmin, accumulated_msgs, "tmin")
4290854f3af5 Uploaded
greg
parents:
diff changeset
154 tmax = items[9]
4290854f3af5 Uploaded
greg
parents:
diff changeset
155 accumulated_msgs = validate_decimal(i, tmax, accumulated_msgs, "tmax")
4290854f3af5 Uploaded
greg
parents:
diff changeset
156 if args.data_type == "normals" and num_normals_rows != 367:
1
cfe1ce427aa7 Uploaded
greg
parents: 0
diff changeset
157 accumulated_msgs = add_error_msg(accumulated_msgs, "The input file contains %d rows, (must be 367)." % num_normals_rows)
0
4290854f3af5 Uploaded
greg
parents:
diff changeset
158
4290854f3af5 Uploaded
greg
parents:
diff changeset
159 if len(accumulated_msgs) > 0:
4290854f3af5 Uploaded
greg
parents:
diff changeset
160 stop_error(accumulated_msgs)
4290854f3af5 Uploaded
greg
parents:
diff changeset
161
4290854f3af5 Uploaded
greg
parents:
diff changeset
162 shutil.copyfile(input_file, args.output)