1 | #!/usr/bin/env python3 |
---|
2 | # -*- coding: utf-8 -*- |
---|
3 | """Comparison of resulting Grib files of two flex_extract versions. |
---|
4 | |
---|
5 | |
---|
6 | |
---|
7 | The script should be called like: |
---|
8 | |
---|
9 | python test_cmp_grib_files.py -r <path-to-reference-files> -n <path-to-new-files> -p <file-pattern> |
---|
10 | |
---|
11 | Note |
---|
12 | ---- |
---|
13 | |
---|
14 | |
---|
15 | Example |
---|
16 | ------- |
---|
17 | python test_cmp_grib_file.py -r 7.0.4/EA5/ -n 7.1/EA5/ -p 'EA*' |
---|
18 | """ |
---|
19 | |
---|
20 | # ------------------------------------------------------------------------------ |
---|
21 | # MODULES |
---|
22 | # ------------------------------------------------------------------------------ |
---|
23 | import os |
---|
24 | import sys |
---|
25 | from eccodes import GribFile, GribMessage |
---|
26 | |
---|
27 | # ------------------------------------------------------------------------------ |
---|
28 | # FUNCTION |
---|
29 | # ------------------------------------------------------------------------------ |
---|
30 | def get_cmdline_params(parlist, debug=True): |
---|
31 | |
---|
32 | import getopt |
---|
33 | |
---|
34 | iref_path = '' # e.g. Reference_files |
---|
35 | inew_path = '' # e.g. New_files |
---|
36 | smatch = '' # e.g. files matching pattern |
---|
37 | |
---|
38 | try: |
---|
39 | opts, pars = getopt.getopt(parlist, |
---|
40 | "hr:n:p:", |
---|
41 | ["ipref=", "ipnew=", "pattern="]) |
---|
42 | except getopt.GetoptError: |
---|
43 | print('test_cmp_grib_file.py -r <ipref> -n <ipnew> -p <pattern>') |
---|
44 | sys.exit(2) |
---|
45 | |
---|
46 | for opt, par in opts: |
---|
47 | if opt == '-h': |
---|
48 | print('test_cmp_grib_file.py -r <ipref> -n <ipnew> -p <pattern>') |
---|
49 | sys.exit() |
---|
50 | elif opt in ("-r", "--ipref"): |
---|
51 | iref_path = par |
---|
52 | elif opt in ("-n", "--ipnew"): |
---|
53 | inew_path = par |
---|
54 | elif opt in ("-p", "--pattern"): |
---|
55 | smatch = par |
---|
56 | |
---|
57 | if iref_path == '': |
---|
58 | sys.exit('NO REFERENCE INPUT PATH SET!') |
---|
59 | if inew_path == '': |
---|
60 | sys.exit('NO "NEW" COMPARISON INPUT PATH SET!') |
---|
61 | if smatch == '': |
---|
62 | sys.exit('NO MATCHING PATTERN FOR FILES SET!') |
---|
63 | |
---|
64 | if debug: |
---|
65 | print("\n\nWelcome!") |
---|
66 | print('Reference path is: ', iref_path) |
---|
67 | print('New path is: ', inew_path) |
---|
68 | print('Filepattern is: ', smatch) |
---|
69 | |
---|
70 | return iref_path, inew_path, smatch |
---|
71 | |
---|
72 | def get_files(ipath, matchingstring, debug=True): |
---|
73 | """ |
---|
74 | @Description: |
---|
75 | Get filenames from input path matching the |
---|
76 | string or regular expression and return it. |
---|
77 | |
---|
78 | @Input: |
---|
79 | ipath: string |
---|
80 | Path to the files. |
---|
81 | |
---|
82 | matchingstring: string |
---|
83 | A string defining the filenames, |
---|
84 | with or without regular exprssion. |
---|
85 | |
---|
86 | @Return |
---|
87 | filename: list of strings |
---|
88 | A list of all files matching the pattern of |
---|
89 | matchingstring. |
---|
90 | """ |
---|
91 | import fnmatch |
---|
92 | |
---|
93 | files = [] |
---|
94 | |
---|
95 | for fn in os.listdir(ipath): |
---|
96 | if fnmatch.fnmatch(fn, matchingstring): |
---|
97 | files.append(fn) |
---|
98 | |
---|
99 | filelist = sorted(files) |
---|
100 | if debug: |
---|
101 | print('The input files are: %s' %(filelist)) |
---|
102 | |
---|
103 | return filelist |
---|
104 | |
---|
105 | def cmp_files_list(flist1, flist2): |
---|
106 | ''' |
---|
107 | ''' |
---|
108 | |
---|
109 | # 1. same length? |
---|
110 | length = len(flist1) == len(flist2) |
---|
111 | if not length: |
---|
112 | print('There are not the same amount of files.') |
---|
113 | sys.exit('Message 1') |
---|
114 | |
---|
115 | # 2. same content? |
---|
116 | content = [True for i, j in zip(flist1, flist2) if i == j] |
---|
117 | if not len(content) == len(flist1): |
---|
118 | print('Not the same file list') |
---|
119 | sys.exit('Message 2') |
---|
120 | |
---|
121 | return True |
---|
122 | |
---|
123 | |
---|
124 | def cmp_number_messages(iref, flist1, inew, flist2): |
---|
125 | |
---|
126 | ref_dict = {} |
---|
127 | new_dict = {} |
---|
128 | res_dict = {} |
---|
129 | for file in flist1: |
---|
130 | with GribFile(os.path.join(iref,file)) as grib: |
---|
131 | ref_dict[file] = len(grib) |
---|
132 | with GribFile(os.path.join(inew,file)) as grib: |
---|
133 | new_dict[file] = len(grib) |
---|
134 | |
---|
135 | res_dict[file] = ref_dict[file] == new_dict[file] |
---|
136 | |
---|
137 | for k, res in res_dict.items(): |
---|
138 | if not res == True: |
---|
139 | print('LOG: Amount of messages in files {} are not the same!'.format(k)) |
---|
140 | |
---|
141 | return True |
---|
142 | |
---|
143 | |
---|
144 | def cmp_grib_msg_header(ipath_ref, ipath_new, filelist): |
---|
145 | from subprocess import Popen, PIPE |
---|
146 | # ref_dict = {} |
---|
147 | # new_dict = {} |
---|
148 | # for file in flist1: |
---|
149 | # with GribFile(os.path.join(iref,file)) as grib: |
---|
150 | # for i in range(len(grib)): |
---|
151 | # msg = GribMessage(grib) |
---|
152 | # ref_dict[file] = {} |
---|
153 | # ref_dict[file][i] = [msg['shortName'],msg['level'], |
---|
154 | # msg['editionNumber'],msg['dataDate'], |
---|
155 | # msg['dataTime'],msg['marsClass'], |
---|
156 | # msg['type'], msg['gridType'], |
---|
157 | # msg['stepRange']] |
---|
158 | error_flag = False |
---|
159 | cmp_flag = False |
---|
160 | for file in filelist: |
---|
161 | try: |
---|
162 | res = Popen(['grib_compare', '-H', |
---|
163 | ipath_ref + '/' + file, |
---|
164 | ipath_new + '/' + file], |
---|
165 | stdin=PIPE, stdout=PIPE, stderr=PIPE) |
---|
166 | output, error = res.communicate()#.decode("utf-8")) |
---|
167 | if error: |
---|
168 | print('... ERROR: \n\t{}'.format(error.decode())) |
---|
169 | error_flag = True |
---|
170 | if output: |
---|
171 | print('{}'.format(output.decode())) |
---|
172 | cmp_flag = True |
---|
173 | except ValueError as e: |
---|
174 | print('... ERROR CODE: ' + str(e.returncode)) |
---|
175 | print('... ERROR MESSAGE:\n \t ' + str(res)) |
---|
176 | error_flag = True |
---|
177 | except OSError as e: |
---|
178 | print('... ERROR CODE: ' + str(e.errno)) |
---|
179 | print('... ERROR MESSAGE:\n \t ' + str(e.strerror)) |
---|
180 | error_flag = True |
---|
181 | |
---|
182 | if error_flag: |
---|
183 | sys.exit('... ERROR IN GRIB MESSAGE COMPARISON!') |
---|
184 | if cmp_flag: |
---|
185 | sys.exit('... FILES HAVE DIFFERENCES IN GRIB MESSAGES!') |
---|
186 | |
---|
187 | return True |
---|
188 | |
---|
189 | |
---|
190 | def cmp_grib_msg_statistics(ipath_ref, ipath_new, filelist): |
---|
191 | |
---|
192 | from subprocess import Popen, PIPE |
---|
193 | |
---|
194 | error_flag = False |
---|
195 | cmp_flag = False |
---|
196 | for file in filelist: |
---|
197 | try: |
---|
198 | res = Popen(['grib_compare', '-c', 'statistics:n', |
---|
199 | ipath_ref + '/' + file, |
---|
200 | ipath_new + '/' + file], |
---|
201 | stdin=PIPE, stdout=PIPE, stderr=PIPE) |
---|
202 | output, error = res.communicate()#.decode("utf-8")) |
---|
203 | if error: |
---|
204 | print('... ERROR: \n\t{}'.format(error.decode())) |
---|
205 | error_flag = True |
---|
206 | if output: |
---|
207 | print('\nIn File: {}'.format(file)) |
---|
208 | print('{}'.format(output.decode())) |
---|
209 | cmp_flag = True |
---|
210 | except ValueError as e: |
---|
211 | print('... ERROR CODE: ' + str(e.returncode)) |
---|
212 | print('... ERROR MESSAGE:\n \t ' + str(res)) |
---|
213 | error_flag = True |
---|
214 | except OSError as e: |
---|
215 | print('... ERROR CODE: ' + str(e.errno)) |
---|
216 | print('... ERROR MESSAGE:\n \t ' + str(e.strerror)) |
---|
217 | error_flag = True |
---|
218 | |
---|
219 | if error_flag: |
---|
220 | sys.exit('... ERROR IN GRIB MESSAGE COMPARISON!') |
---|
221 | if cmp_flag: |
---|
222 | sys.exit('... FILES HAVE DIFFERENT STATISTICS!') |
---|
223 | return True |
---|
224 | |
---|
225 | if __name__ == '__main__': |
---|
226 | |
---|
227 | # get the parameter list of program call |
---|
228 | ref_path, new_path, fmatch = get_cmdline_params(sys.argv[1:]) |
---|
229 | |
---|
230 | # get the list of files of both cases |
---|
231 | ref_files = get_files(ref_path, fmatch) |
---|
232 | new_files = get_files(new_path, fmatch) |
---|
233 | |
---|
234 | # flag to store successfull tests |
---|
235 | suc = True |
---|
236 | |
---|
237 | # 1. Does the 2 cases contain the same list of files? |
---|
238 | suc = True if suc and cmp_files_list(ref_files, new_files) else False |
---|
239 | |
---|
240 | # 2. Does each file in both cases contain the same amount of messages? |
---|
241 | suc = True if suc and cmp_number_messages(ref_path, ref_files, new_path, new_files) else False |
---|
242 | |
---|
243 | # 3. Does each file has the same parameters (in Header)? |
---|
244 | # Since we can be sure that both cases have the same files, |
---|
245 | # we just use 1 filelist |
---|
246 | suc = True if suc and cmp_grib_msg_header(ref_path, new_path, new_files) else False |
---|
247 | |
---|
248 | # 4. Are the statistics of each message the same? |
---|
249 | # Since we can be sure that both cases have the same files, |
---|
250 | # we just use 1 filelist |
---|
251 | suc = True if suc and cmp_grib_msg_statistics(ref_path, new_path, new_files) else False |
---|
252 | |
---|
253 | # If the program comes this far and flag "suc" = True, |
---|
254 | # all tests were successful |
---|
255 | if suc: |
---|
256 | exit('GRIB_COMPARISON: SUCCESSFULL!') |
---|
257 | else: |
---|
258 | exit('GRIB_COMPARISON: FAILURE!') |
---|