Project

General

Profile

1 187 aaronmk
#!/usr/bin/env python
2 926 aaronmk
# Combines two map spreadsheets A0->B and A1->C to A->B, with B overwriting C
3 3759 aaronmk
# Multi-safe (supports an input appearing multiple times). Note that if there is
4 4158 aaronmk
# *any* non-empty mapping for an input in map_0, all mappings for that input in
5
# map_1 will be excluded.
6 4504 aaronmk
# Case- and punctuation-insensitive.
7 187 aaronmk
8
import csv
9 738 aaronmk
import os.path
10 187 aaronmk
import sys
11
12 738 aaronmk
sys.path.append(os.path.dirname(__file__)+"/../lib")
13
14
import maps
15 1123 aaronmk
import opts
16 1769 aaronmk
import util
17 738 aaronmk
18 4164 aaronmk
def non_empty(row): return row[0] != '' and row[1] != ''
19
20 187 aaronmk
def main():
21 1123 aaronmk
    ignore = opts.env_flag('ignore')
22 1769 aaronmk
    header_num = util.cast(int, opts.get_env_var('header_num'))
23 1127 aaronmk
        # selects which map's header to use as the output header
24 3902 aaronmk
    all_ = opts.env_flag('all') # doesn't merge mappings that are in both maps
25 187 aaronmk
    try: _prog_name, map_1_path = sys.argv
26
    except ValueError:
27 3902 aaronmk
        raise SystemExit('Usage: env [ignore=1] [header_num={0|1}] [all=1] '
28
            +sys.argv[0]+' <map_0 map_1 [| '+sys.argv[0]+' map_2]... >out_map')
29 187 aaronmk
30 1125 aaronmk
    headers = [None]*2
31 187 aaronmk
32 1125 aaronmk
    # Open map 0
33
    map_0_reader = csv.reader(sys.stdin)
34
    headers[0] = map_0_reader.next()
35
36
    # Open map 1
37 187 aaronmk
    stream = open(map_1_path, 'rb')
38 1125 aaronmk
    map_1_reader = csv.reader(stream)
39
    headers[1] = map_1_reader.next()
40 187 aaronmk
41 1125 aaronmk
    # Check col labels
42 1507 aaronmk
    combinable = maps.combinable(*headers)
43
    if not combinable and not ignore:
44 1502 aaronmk
        raise SystemExit('Map error: '
45 1125 aaronmk
        'Map 0 column 0 label doesn\'t contain map 1 column 0 label')
46 187 aaronmk
47 2023 aaronmk
    # Pass through map 0, storing which mappings it defines
48 187 aaronmk
    writer = csv.writer(sys.stdout)
49 1796 aaronmk
    writer.writerow(maps.merge_headers(*headers, **dict(prefer=header_num)))
50 2023 aaronmk
    mappings = set()
51 1125 aaronmk
    for row in map_0_reader:
52 4164 aaronmk
        if all_ or non_empty(row):
53 4504 aaronmk
            if not all_: mappings.add(maps.simplify(row[0]))
54 4164 aaronmk
            writer.writerow(row)
55 1125 aaronmk
56
    if combinable:
57 2023 aaronmk
        # Add mappings in map 1 that weren't already defined or mapped to
58 1125 aaronmk
        for row in map_1_reader:
59 4504 aaronmk
            if non_empty(row) and maps.simplify(row[0]) not in mappings:
60
                writer.writerow(row)
61 1125 aaronmk
62
    stream.close()
63 187 aaronmk
64
main()