| 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 |
734
|
aaronmk
|
import os.path
|
| 6 |
51
|
aaronmk
|
import sys
|
| 7 |
939
|
aaronmk
|
import warnings
|
| 8 |
51
|
aaronmk
|
|
| 9 |
734
|
aaronmk
|
sys.path.append(os.path.dirname(__file__)+"/../lib")
|
| 10 |
732
|
aaronmk
|
|
| 11 |
734
|
aaronmk
|
import maps
|
| 12 |
1282
|
aaronmk
|
import util
|
| 13 |
734
|
aaronmk
|
|
| 14 |
51
|
aaronmk
|
def main():
|
| 15 |
67
|
aaronmk
|
try: _prog_name, map_1_path = sys.argv
|
| 16 |
51
|
aaronmk
|
except ValueError:
|
| 17 |
180
|
aaronmk
|
raise SystemExit('Usage: '+sys.argv[0]+' <map_0 map_1 [| '+sys.argv[0]
|
| 18 |
|
|
+' map_2]... >joined_map')
|
| 19 |
51
|
aaronmk
|
|
| 20 |
|
|
# Get map 1
|
| 21 |
|
|
map_1 = {}
|
| 22 |
|
|
stream = open(map_1_path, 'rb')
|
| 23 |
59
|
aaronmk
|
reader = csv.reader(stream)
|
| 24 |
737
|
aaronmk
|
map_1_cols = reader.next()
|
| 25 |
51
|
aaronmk
|
for row in reader:
|
| 26 |
732
|
aaronmk
|
if row[0] != '': map_1[row[0]] = row
|
| 27 |
51
|
aaronmk
|
stream.close()
|
| 28 |
|
|
|
| 29 |
|
|
# Join map 1 to map 0
|
| 30 |
59
|
aaronmk
|
reader = csv.reader(sys.stdin)
|
| 31 |
120
|
aaronmk
|
writer = csv.writer(sys.stdout)
|
| 32 |
737
|
aaronmk
|
map_0_cols = reader.next()
|
| 33 |
|
|
if not map_0_cols[1] == map_1_cols[0]: raise SystemExit('Map error: '
|
| 34 |
62
|
aaronmk
|
'Map 0 output column name doesn\'t match map 1 input column name')
|
| 35 |
737
|
aaronmk
|
writer.writerow(maps.merge_mappings(map_0_cols, map_1_cols))
|
| 36 |
51
|
aaronmk
|
for row in reader:
|
| 37 |
730
|
aaronmk
|
if row[1] != '':
|
| 38 |
1283
|
aaronmk
|
out_orig = row[1] # used in "No join mapping" error msg
|
| 39 |
|
|
|
| 40 |
1170
|
aaronmk
|
# Look for a match
|
| 41 |
|
|
out_row = None
|
| 42 |
|
|
suffix = ''
|
| 43 |
|
|
while True:
|
| 44 |
|
|
try:
|
| 45 |
|
|
out_row = map_1[row[1]]
|
| 46 |
|
|
break
|
| 47 |
|
|
except KeyError:
|
| 48 |
|
|
# Heuristically look for a match on a parent path.
|
| 49 |
|
|
# If this produces a syntactically invalid parent path (e.g.
|
| 50 |
|
|
# there is a / within []), it will be ignored since there
|
| 51 |
|
|
# wouldn't be a an entry in map_1.
|
| 52 |
|
|
row[1], sep, new_suffix = row[1].rpartition('/')
|
| 53 |
|
|
if sep == '': break
|
| 54 |
|
|
suffix = sep+new_suffix+suffix # prepend new suffix
|
| 55 |
|
|
|
| 56 |
|
|
# Write new mapping
|
| 57 |
|
|
if out_row != None:
|
| 58 |
|
|
row = maps.merge_mappings(row, out_row)
|
| 59 |
|
|
row[1] += suffix # don't modify out_row!
|
| 60 |
|
|
else:
|
| 61 |
1283
|
aaronmk
|
msg = 'No join mapping for '+out_orig
|
| 62 |
939
|
aaronmk
|
warnings.warn(UserWarning(msg))
|
| 63 |
1282
|
aaronmk
|
row[2] = '** '+msg+' ** '+util.list_setdefault(row, 2, '')
|
| 64 |
730
|
aaronmk
|
row[1] = ''
|
| 65 |
120
|
aaronmk
|
writer.writerow(row)
|
| 66 |
51
|
aaronmk
|
|
| 67 |
|
|
main()
|
