| 1 |
4591
|
aaronmk
|
#!/usr/bin/env python
|
| 2 |
7448
|
aaronmk
|
# Translates a spreadsheet column using a thesaurus.
|
| 3 |
4650
|
aaronmk
|
# The column header is also translated. CSVs without a header are supported.
|
| 4 |
4591
|
aaronmk
|
# Unrecognized names are left untouched, permitting successive runs on different
|
| 5 |
7448
|
aaronmk
|
# thesauruses.
|
| 6 |
4591
|
aaronmk
|
# Case- and punctuation-sensitive. (Use canon first for case-insensitivity.)
|
| 7 |
|
|
|
| 8 |
|
|
import csv
|
| 9 |
|
|
import sys
|
| 10 |
|
|
|
| 11 |
|
|
def main():
|
| 12 |
|
|
try: _prog_name, col_num, dict_path = sys.argv
|
| 13 |
|
|
except ValueError: raise SystemExit('Usage: '+sys.argv[0]
|
| 14 |
7448
|
aaronmk
|
+' <in col# thesaurus [| '+sys.argv[0]+' col# thesaurus_2]... >out')
|
| 15 |
4591
|
aaronmk
|
col_num = int(col_num)
|
| 16 |
|
|
|
| 17 |
7448
|
aaronmk
|
# Get thesaurus
|
| 18 |
4591
|
aaronmk
|
dict_ = {}
|
| 19 |
|
|
stream = open(dict_path, 'rb')
|
| 20 |
|
|
reader = csv.reader(stream)
|
| 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 |
|
|
for row in reader:
|
| 28 |
|
|
term = row[col_num]
|
| 29 |
|
|
try: row[col_num] = dict_[term]
|
| 30 |
|
|
except KeyError: pass
|
| 31 |
|
|
writer.writerow(row)
|
| 32 |
|
|
|
| 33 |
|
|
main()
|
