| 1 |
51
|
aaronmk
|
#!/usr/bin/env python
|
| 2 |
|
|
# Inner-joins two map spreadsheets A->B and B->C to A->C
|
| 3 |
|
|
|
| 4 |
|
|
import csv
|
| 5 |
|
|
import sys
|
| 6 |
|
|
|
| 7 |
732
|
aaronmk
|
def merge_rows(*rows):
|
| 8 |
|
|
'''e.g. ['a','b'] + ['','y','z'] = ['a','b; y','z']'''
|
| 9 |
|
|
def get(row, i):
|
| 10 |
|
|
try: return row[i]
|
| 11 |
|
|
except IndexError: return ''
|
| 12 |
|
|
return ['; '.join(filter(lambda v: v != '', [get(row, i) for row in rows]))
|
| 13 |
|
|
for i in xrange(max(map(len, rows)))]
|
| 14 |
|
|
|
| 15 |
51
|
aaronmk
|
def main():
|
| 16 |
67
|
aaronmk
|
try: _prog_name, map_1_path = sys.argv
|
| 17 |
51
|
aaronmk
|
except ValueError:
|
| 18 |
180
|
aaronmk
|
raise SystemExit('Usage: '+sys.argv[0]+' <map_0 map_1 [| '+sys.argv[0]
|
| 19 |
|
|
+' map_2]... >joined_map')
|
| 20 |
51
|
aaronmk
|
|
| 21 |
|
|
# Get map 1
|
| 22 |
|
|
map_1 = {}
|
| 23 |
|
|
stream = open(map_1_path, 'rb')
|
| 24 |
59
|
aaronmk
|
reader = csv.reader(stream)
|
| 25 |
51
|
aaronmk
|
map_1_in, map_1_out = reader.next()[:2]
|
| 26 |
|
|
for row in reader:
|
| 27 |
732
|
aaronmk
|
if row[0] != '': map_1[row[0]] = row
|
| 28 |
51
|
aaronmk
|
stream.close()
|
| 29 |
|
|
|
| 30 |
|
|
# Join map 1 to map 0
|
| 31 |
59
|
aaronmk
|
reader = csv.reader(sys.stdin)
|
| 32 |
120
|
aaronmk
|
writer = csv.writer(sys.stdout)
|
| 33 |
|
|
cols = reader.next()
|
| 34 |
177
|
aaronmk
|
if not cols[1] == map_1_in: raise SystemExit('Map error: '
|
| 35 |
62
|
aaronmk
|
'Map 0 output column name doesn\'t match map 1 input column name')
|
| 36 |
120
|
aaronmk
|
cols[1] = map_1_out
|
| 37 |
|
|
writer.writerow(cols)
|
| 38 |
51
|
aaronmk
|
for row in reader:
|
| 39 |
730
|
aaronmk
|
if row[1] != '':
|
| 40 |
732
|
aaronmk
|
try:
|
| 41 |
|
|
new = map_1[row[1]]
|
| 42 |
|
|
row[1] = new[1] # mapping
|
| 43 |
|
|
row[2:] = merge_rows(row[2:], new[2:]) # comments
|
| 44 |
730
|
aaronmk
|
except KeyError:
|
| 45 |
|
|
row[2] = '** No join mapping for '+row[1]+' ** '+row[2]
|
| 46 |
|
|
row[1] = ''
|
| 47 |
120
|
aaronmk
|
writer.writerow(row)
|
| 48 |
51
|
aaronmk
|
|
| 49 |
|
|
main()
|
