1 | #!/usr/bin/env python
|
---|
2 |
|
---|
3 | from __future__ import print_function
|
---|
4 |
|
---|
5 | import os
|
---|
6 | import re
|
---|
7 | import sys
|
---|
8 | import json
|
---|
9 |
|
---|
10 | import argparse
|
---|
11 | import openpyxl
|
---|
12 |
|
---|
13 | import xlsxutil
|
---|
14 |
|
---|
15 | def eprint(*args, **kwargs):
|
---|
16 | print(*args, file=sys.stderr, **kwargs)
|
---|
17 |
|
---|
18 | def fileset_voting_for_esc_country_in_year(data_hashmap):
|
---|
19 |
|
---|
20 | # Only generating a single fileset record here, however
|
---|
21 | # the Greenstone format allows for this to be an array
|
---|
22 | # of fileset entries => return [ fileset ]
|
---|
23 |
|
---|
24 | fileset = {}
|
---|
25 |
|
---|
26 | return [ fileset ]
|
---|
27 |
|
---|
28 |
|
---|
29 | def filter_finalist_votes(entry):
|
---|
30 | return entry.get('(semi-) final') == "f" and entry.get('Duplicate') == None
|
---|
31 |
|
---|
32 | def sortkey_year_to_country(entry):
|
---|
33 | return str(entry.get('Year'))+entry.get('To country')
|
---|
34 |
|
---|
35 |
|
---|
36 | def create_to_country_voting_groups(data_hashmap_array):
|
---|
37 |
|
---|
38 | # Example values for header-names
|
---|
39 | # (semi-) final: f, sf
|
---|
40 | # Jury or Televoting: J, T
|
---|
41 | # Year: 1975
|
---|
42 | # To country: Belgium
|
---|
43 | # From country: Belgium
|
---|
44 | # Edition: 1975f, 1975sf
|
---|
45 |
|
---|
46 |
|
---|
47 | # Filter down to just the voting results concerning finals
|
---|
48 | data_hashmap_array_finals = list(filter(filter_finalist_votes, data_hashmap_array))
|
---|
49 |
|
---|
50 | # Sort so array entries are grouped by the country receiving the votes in a given year
|
---|
51 | data_hashmap_array_finals.sort(key=sortkey_year_to_country)
|
---|
52 |
|
---|
53 |
|
---|
54 | # Debug output
|
---|
55 | #
|
---|
56 | # for data_hashmap in data_hashmap_array_finals:
|
---|
57 | # print data_hashmap.get('To country'), data_hashmap.get('Year'), data_hashmap.get('Points'), "(Points from " + data_hashmap.get('From country') + ")"
|
---|
58 |
|
---|
59 |
|
---|
60 | # Build array of country groups
|
---|
61 | # A country grouping includes all the votes that country receive that year
|
---|
62 | country_groups = []
|
---|
63 |
|
---|
64 | prev_data_hashmap = data_hashmap_array_finals[0]
|
---|
65 | country_group = [ ]
|
---|
66 |
|
---|
67 | i = 1
|
---|
68 | num_finals = len(data_hashmap_array_finals)
|
---|
69 |
|
---|
70 | while (i < num_finals):
|
---|
71 | country_group.append(prev_data_hashmap)
|
---|
72 | data_hashmap = data_hashmap_array_finals[i]
|
---|
73 |
|
---|
74 | if (data_hashmap.get('To country') != prev_data_hashmap.get('To country')):
|
---|
75 | # moving on to a new country group
|
---|
76 | country_groups.append(country_group)
|
---|
77 | country_group = [ ]
|
---|
78 |
|
---|
79 | prev_data_hashmap = data_hashmap
|
---|
80 | i = i + 1
|
---|
81 |
|
---|
82 | country_group.append(prev_data_hashmap)
|
---|
83 | country_groups.append(country_group)
|
---|
84 |
|
---|
85 | return country_groups
|
---|
86 |
|
---|
87 |
|
---|
88 | def fileset_voting_for_esc_country_in_year(to_country_year_votes):
|
---|
89 |
|
---|
90 | # Looking to build data-structure (for output as JSON) in the form
|
---|
91 | # { "FileSet":
|
---|
92 | # [
|
---|
93 | # { "FileName": "France1991\.nul" },
|
---|
94 | # { "Description":
|
---|
95 | # {
|
---|
96 | # "Metadata":
|
---|
97 | # [
|
---|
98 | # { "name": "Germany-J", "content": 12 }, # J = Jury Vote
|
---|
99 | # { "name": "Germany-T", "content": 6 }, # T = Televote (if present)
|
---|
100 | # ...
|
---|
101 | # ]
|
---|
102 | # }
|
---|
103 | # }
|
---|
104 | # ]
|
---|
105 | # }
|
---|
106 |
|
---|
107 | fileset_array = []
|
---|
108 |
|
---|
109 | metadata_array = []
|
---|
110 |
|
---|
111 | jury_metadata_vals = []
|
---|
112 | tele_metadata_vals = []
|
---|
113 |
|
---|
114 | for to_country_year_vote in to_country_year_votes:
|
---|
115 | to_country = to_country_year_vote.get('To country')
|
---|
116 | year = to_country_year_vote.get('Year')
|
---|
117 | from_country = to_country_year_vote.get('From country')
|
---|
118 | vote_type = to_country_year_vote.get('Jury or Televoting')
|
---|
119 | points = to_country_year_vote.get('Points')
|
---|
120 |
|
---|
121 | voting_rec = { "name": from_country+"-"+vote_type, "content": points }
|
---|
122 |
|
---|
123 | metadata_array.append(voting_rec)
|
---|
124 |
|
---|
125 | if (vote_type == "J"):
|
---|
126 | jury_metadata_vals.append(from_country+"-J")
|
---|
127 | elif (vote_type == "T"):
|
---|
128 | tele_metadata_vals.append(from_country+"-T")
|
---|
129 | else:
|
---|
130 | eprint("Warning: Unrecognized voting type: " + vote_type)
|
---|
131 |
|
---|
132 | if (len(jury_metadata_vals)>0):
|
---|
133 | metadata_array.append({ "name": "JuryVotesJSON", "content": json.dumps(jury_metadata_vals) })
|
---|
134 | if (len(tele_metadata_vals)>0):
|
---|
135 | metadata_array.append({ "name": "TeleVotesJSON", "content": json.dumps(tele_metadata_vals) })
|
---|
136 |
|
---|
137 | id_to_country = to_country_year_votes[0].get('To country')
|
---|
138 | id_to_country = re.sub(r'\s+', '', id_to_country)
|
---|
139 | id_year = to_country_year_votes[0].get('Year');
|
---|
140 | id = id_to_country + str(id_year);
|
---|
141 | filename_id = id + "\\.nul"
|
---|
142 |
|
---|
143 | fileset = {
|
---|
144 | "FileSet" : [
|
---|
145 | { "FileName": filename_id, },
|
---|
146 | { "Description" : { "Metadata" : metadata_array } }
|
---|
147 | ]
|
---|
148 | }
|
---|
149 |
|
---|
150 | return fileset
|
---|
151 |
|
---|
152 |
|
---|
153 | if __name__ == "__main__":
|
---|
154 |
|
---|
155 | parser = argparse.ArgumentParser()
|
---|
156 | parser.add_argument('input-file.xlsx')
|
---|
157 | parser.add_argument('output-file.json', nargs='?')
|
---|
158 | parser.add_argument('--sheetname')
|
---|
159 |
|
---|
160 | args = parser.parse_args()
|
---|
161 |
|
---|
162 | excel_input_filename = getattr(args,'input-file.xlsx');
|
---|
163 | json_output_filename = getattr(args,'output-file.json');
|
---|
164 | sheetname = getattr(args,'sheetname');
|
---|
165 |
|
---|
166 | if (json_output_filename == None):
|
---|
167 | json_output_filename = os.path.splitext(excel_input_filename)[0]+'.json'
|
---|
168 |
|
---|
169 | worksheet = xlsxutil.load_xslx_sheet(excel_input_filename,sheetname)
|
---|
170 |
|
---|
171 | data_hashmap_array = xlsxutil.convert_worksheet_to_hashmaps(worksheet)
|
---|
172 |
|
---|
173 | print("Number of data rows: " + str(len(data_hashmap_array)))
|
---|
174 |
|
---|
175 |
|
---|
176 | to_country_year_voting_groups = create_to_country_voting_groups(data_hashmap_array)
|
---|
177 |
|
---|
178 | # Debug output
|
---|
179 | #
|
---|
180 | # print to_country_year_voting_groups
|
---|
181 |
|
---|
182 | # Next step is to express the grouped by-country voting data
|
---|
183 | # in the Greenstone JSON metadata format:
|
---|
184 |
|
---|
185 | # { "DirectoryMetadata":
|
---|
186 | # [
|
---|
187 | # { "FileSet":
|
---|
188 | # [
|
---|
189 | # { "FileName": "France1991\.nul" },
|
---|
190 | # { "Description":
|
---|
191 | # {
|
---|
192 | # "Metadata":
|
---|
193 | # [
|
---|
194 | # { "name": "Germany-J", "content": "12" }, # J = Jury Vote
|
---|
195 | # ...
|
---|
196 | # ]
|
---|
197 | # }
|
---|
198 | # }
|
---|
199 | # ]
|
---|
200 | # }
|
---|
201 | # ...
|
---|
202 | # ]
|
---|
203 | # }
|
---|
204 |
|
---|
205 | directory_metadata = []
|
---|
206 |
|
---|
207 | print("Creating Greenstone JSON voting metadata for:")
|
---|
208 | for to_country_year_votes in to_country_year_voting_groups:
|
---|
209 |
|
---|
210 | fileset = fileset_voting_for_esc_country_in_year(to_country_year_votes)
|
---|
211 | directory_metadata.append(fileset)
|
---|
212 |
|
---|
213 | filename_id = fileset.get('FileSet')[0].get('FileName')
|
---|
214 | num_countries_voting_data = len(fileset.get('FileSet')[1].get('Description').get('Metadata'))
|
---|
215 |
|
---|
216 | print(" " + filename_id.ljust(28) + ": " + str(num_countries_voting_data) + " votes")
|
---|
217 |
|
---|
218 | greenstone_metadata_json = { "DirectoryMetadata": directory_metadata }
|
---|
219 |
|
---|
220 |
|
---|
221 | with open(json_output_filename, 'w') as outfile:
|
---|
222 | json.dump(greenstone_metadata_json, outfile, indent=2)
|
---|