| 1 |
927
|
aaronmk
|
#!/usr/bin/env python
|
| 2 |
|
|
# Intersects two map spreadsheets A0->B and A1->C to A->B, with B overwriting C
|
| 3 |
3760
|
aaronmk
|
# Multi-safe (supports an input appearing multiple times).
|
| 4 |
4504
|
aaronmk
|
# Case- and punctuation-insensitive.
|
| 5 |
927
|
aaronmk
|
|
| 6 |
|
|
import csv
|
| 7 |
|
|
import os.path
|
| 8 |
|
|
import sys
|
| 9 |
|
|
|
| 10 |
|
|
sys.path.append(os.path.dirname(__file__)+"/../lib")
|
| 11 |
|
|
|
| 12 |
|
|
import maps
|
| 13 |
1525
|
aaronmk
|
import util
|
| 14 |
927
|
aaronmk
|
|
| 15 |
|
|
def main():
|
| 16 |
1525
|
aaronmk
|
try: _prog_name, map_1_path = sys.argv[:2]
|
| 17 |
927
|
aaronmk
|
except ValueError:
|
| 18 |
1525
|
aaronmk
|
raise SystemExit('Usage: '+sys.argv[0]+' <map_0 map_1 '
|
| 19 |
|
|
'[compare_col_num...] [| '+sys.argv[0]
|
| 20 |
927
|
aaronmk
|
+' map_2]... >intersect_map')
|
| 21 |
1525
|
aaronmk
|
compare_col_nums = map(int, sys.argv[2:]) # 0-based
|
| 22 |
|
|
if compare_col_nums == []:
|
| 23 |
|
|
compare_col_nums = None # list_subset() value for all columns
|
| 24 |
927
|
aaronmk
|
|
| 25 |
4504
|
aaronmk
|
def compare_on(row): return tuple(map(maps.simplify,
|
| 26 |
|
|
util.list_subset(row, compare_col_nums)))
|
| 27 |
1525
|
aaronmk
|
|
| 28 |
|
|
headers = [None]*2
|
| 29 |
|
|
|
| 30 |
927
|
aaronmk
|
# Get map 1
|
| 31 |
1525
|
aaronmk
|
compare_cols = set()
|
| 32 |
927
|
aaronmk
|
stream = open(map_1_path, 'rb')
|
| 33 |
|
|
reader = csv.reader(stream)
|
| 34 |
1525
|
aaronmk
|
headers[1] = reader.next()
|
| 35 |
927
|
aaronmk
|
for row in reader:
|
| 36 |
1525
|
aaronmk
|
if row[0] != '':
|
| 37 |
|
|
compare_cols.add(compare_on(row))
|
| 38 |
927
|
aaronmk
|
stream.close()
|
| 39 |
|
|
|
| 40 |
|
|
# Open map 0
|
| 41 |
|
|
reader = csv.reader(sys.stdin)
|
| 42 |
1525
|
aaronmk
|
headers[0] = reader.next()
|
| 43 |
927
|
aaronmk
|
|
| 44 |
1525
|
aaronmk
|
# Check col labels
|
| 45 |
|
|
combinable = maps.combinable(*headers)
|
| 46 |
4084
|
aaronmk
|
if not combinable:
|
| 47 |
1525
|
aaronmk
|
raise SystemExit('Map error: '
|
| 48 |
|
|
'Map 0 column 0 label doesn\'t contain map 1 column 0 label')
|
| 49 |
|
|
|
| 50 |
927
|
aaronmk
|
# Add map 0 to map 1, overwriting existing entries
|
| 51 |
|
|
writer = csv.writer(sys.stdout)
|
| 52 |
1525
|
aaronmk
|
writer.writerow(headers[0])
|
| 53 |
927
|
aaronmk
|
for row in reader:
|
| 54 |
1525
|
aaronmk
|
if not combinable or compare_on(row) in compare_cols:
|
| 55 |
|
|
# not combinable, or in map 1
|
| 56 |
|
|
writer.writerow(row)
|
| 57 |
927
|
aaronmk
|
|
| 58 |
|
|
main()
|
