Project

General

Profile

1 123 aaronmk
#!/usr/bin/env python
2
# Converts a map spreadsheet to human-readable (but machine unusable) form
3 3807 aaronmk
# Usage: self [col_num...] <in_map >out_map
4 123 aaronmk
5
import csv
6
import re
7
import sys
8
9
def sub_nested(regex, repl, str_):
10
    while True:
11
        str_, n = re.subn(regex, repl, str_)
12
        if n == 0: return str_
13
14
def cleanup(xpath):
15
    truncated = False
16
17
    # Remove attrs
18
    xpath = sub_nested(r':\[[^\[\]]*?\]', r'', xpath)
19
20
    # Remove lookahead assertions
21
    xpath = sub_nested(r'\((/[^\)]*?)\)(?=/)', r'\1', xpath)
22
23
    # Remove pointers
24
    xpath, n = re.subn(r'^.*->', r'', xpath)
25
    if n > 0: truncated = True
26
27
    # Remove part of path before first key list, XML function, or path end
28
    # Leave enough to include the table of a user-defined value
29
    xpath, n = re.subn(r'^(?:/(?!_)[\w*]+)*(?=(?:/(?!_)[\w*]+){2}(?:\[|/_|$))',
30
        r'', xpath)
31
    # Prepend / to show truncation
32
    if n > 0: truncated = True
33
34
    # Remove backward (child-to-parent) pointer's target ID attr
35
    xpath = re.sub(r'\[[\w*]+\]|(?<=\[)[\w*]+,', r'', xpath)
36
37
    # Remove negative keys
38
    xpath = re.sub(r',?!(?:[\w*]+/)*@?[\w*]+', r'', xpath)
39
40
    # Remove path before key
41
    xpath = re.sub(r'(?:[\w*]+/)*(@?[\w*]+)(?==)', r'\1', xpath)
42
43
    # Prepend / to show truncation
44
    if truncated: xpath = '/'+xpath
45
46
    return xpath
47
48
def main():
49 3807 aaronmk
    col_nums = sys.argv[1:]
50
    if col_nums == []: col_nums = range(2)
51
    col_nums = map(int, col_nums) # 0-based
52
53 123 aaronmk
    # Convert map
54
    reader = csv.reader(sys.stdin)
55
    writer = csv.writer(sys.stdout)
56
    writer.writerow(reader.next())
57 125 aaronmk
    for row in reader:
58 3807 aaronmk
        for i in col_nums: row[i] = cleanup(row[i])
59 125 aaronmk
        writer.writerow(row)
60 123 aaronmk
61
main()