1 |
4591
|
aaronmk
|
#!/usr/bin/env python
|
2 |
|
|
# Translates a spreadsheet column using a dictionary.
|
3 |
|
|
# Unrecognized names are left untouched, permitting successive runs on different
|
4 |
|
|
# dictionaries.
|
5 |
|
|
# Case- and punctuation-sensitive. (Use canon first for case-insensitivity.)
|
6 |
|
|
|
7 |
|
|
import csv
|
8 |
|
|
import sys
|
9 |
|
|
|
10 |
|
|
def main():
|
11 |
|
|
try: _prog_name, col_num, dict_path = sys.argv
|
12 |
|
|
except ValueError: raise SystemExit('Usage: '+sys.argv[0]
|
13 |
|
|
+' <in col# dict [| '+sys.argv[0]+' col# dict_2]... >out')
|
14 |
|
|
col_num = int(col_num)
|
15 |
|
|
|
16 |
|
|
# Get dict
|
17 |
|
|
dict_ = {}
|
18 |
|
|
stream = open(dict_path, 'rb')
|
19 |
|
|
reader = csv.reader(stream)
|
20 |
|
|
reader.next() # skip header
|
21 |
4626
|
aaronmk
|
for row in reader: dict_[row[0]] = row[1]
|
22 |
4591
|
aaronmk
|
stream.close()
|
23 |
|
|
|
24 |
|
|
# Translate input
|
25 |
|
|
reader = csv.reader(sys.stdin)
|
26 |
|
|
writer = csv.writer(sys.stdout)
|
27 |
|
|
writer.writerow(reader.next()) # pass through header
|
28 |
|
|
for row in reader:
|
29 |
|
|
term = row[col_num]
|
30 |
|
|
try: row[col_num] = dict_[term]
|
31 |
|
|
except KeyError: pass
|
32 |
|
|
writer.writerow(row)
|
33 |
|
|
|
34 |
|
|
main()
|