| 1 |
205
|
aaronmk
|
#!/usr/bin/env python
|
| 2 |
|
|
# Modifies a map spreadsheet A->B or any file using a replacements spreadsheet
|
| 3 |
|
|
# A->C or B->C
|
| 4 |
|
|
|
| 5 |
|
|
import csv
|
| 6 |
1705
|
aaronmk
|
import os.path
|
| 7 |
205
|
aaronmk
|
import re
|
| 8 |
|
|
import sys
|
| 9 |
|
|
|
| 10 |
1705
|
aaronmk
|
sys.path.append(os.path.dirname(__file__)+"/../lib")
|
| 11 |
|
|
|
| 12 |
|
|
import maps
|
| 13 |
|
|
|
| 14 |
205
|
aaronmk
|
def main():
|
| 15 |
|
|
try: _prog_name, repl_path = sys.argv[:2]
|
| 16 |
|
|
except ValueError:
|
| 17 |
|
|
raise SystemExit('Usage: '+sys.argv[0]+' <map repl [col_num] [| '+
|
| 18 |
|
|
sys.argv[0]+' repl_1 [col_num_1]]... >new_map')
|
| 19 |
|
|
col_num = None
|
| 20 |
|
|
try: col_num = sys.argv[2]
|
| 21 |
211
|
aaronmk
|
except IndexError: pass
|
| 22 |
205
|
aaronmk
|
if col_num != None: col_num = int(col_num) # 0-based
|
| 23 |
|
|
|
| 24 |
|
|
# Get replacements
|
| 25 |
|
|
repls = []
|
| 26 |
|
|
stream = open(repl_path, 'rb')
|
| 27 |
|
|
reader = csv.reader(stream)
|
| 28 |
|
|
repl_in, repl_out = reader.next()[:2]
|
| 29 |
|
|
for row in reader:
|
| 30 |
210
|
aaronmk
|
in_, out = row[:2]
|
| 31 |
216
|
aaronmk
|
if in_ != '':
|
| 32 |
564
|
aaronmk
|
if re.match(r'^\w+$', in_):
|
| 33 |
686
|
aaronmk
|
in_ = r'(?<![^\W_])'+in_+r'(?![^\W_])' # match whole word
|
| 34 |
1229
|
aaronmk
|
repls.append((r'(?m)'+in_, out))
|
| 35 |
205
|
aaronmk
|
stream.close()
|
| 36 |
|
|
def repl_all(str_):
|
| 37 |
|
|
for repl, with_ in repls: str_ = re.sub(repl, with_, str_)
|
| 38 |
|
|
return str_
|
| 39 |
|
|
|
| 40 |
|
|
# Modify map or file
|
| 41 |
|
|
if col_num != None:
|
| 42 |
|
|
reader = csv.reader(sys.stdin)
|
| 43 |
|
|
writer = csv.writer(sys.stdout)
|
| 44 |
|
|
cols = reader.next()
|
| 45 |
1705
|
aaronmk
|
label, root = maps.col_info(cols[col_num])[:2]
|
| 46 |
205
|
aaronmk
|
if label != repl_in: raise SystemExit('Map error: Map column '+
|
| 47 |
|
|
str(col_num)+' label "'+label+'" doesn\'t match replacements input '
|
| 48 |
|
|
'column label "'+repl_in+'"')
|
| 49 |
|
|
cols[col_num] = repl_out+sep+repl_all(root)
|
| 50 |
|
|
writer.writerow(cols)
|
| 51 |
|
|
for row in reader:
|
| 52 |
|
|
row[col_num] = repl_all(row[col_num])
|
| 53 |
|
|
writer.writerow(row)
|
| 54 |
358
|
aaronmk
|
else: sys.stdout.write(repl_all(sys.stdin.read()))
|
| 55 |
205
|
aaronmk
|
|
| 56 |
|
|
main()
|
