Project

General

Profile

1 205 aaronmk
#!/usr/bin/env python
2
# Modifies a map spreadsheet A->B or any file using a replacements spreadsheet
3
# A->C or B->C
4
5
import csv
6 9785 aaronmk
import HTMLParser
7 1705 aaronmk
import os.path
8 205 aaronmk
import re
9
import sys
10
11 1705 aaronmk
sys.path.append(os.path.dirname(__file__)+"/../lib")
12
13
import maps
14 4357 aaronmk
import opts
15 9784 aaronmk
import strings
16 1705 aaronmk
17 9785 aaronmk
def unescape_html(str_): return HTMLParser.HTMLParser().unescape(str_)
18
19
def repl_unescape_html(match): return unescape_html(match.group(0))
20
21 205 aaronmk
def main():
22 4357 aaronmk
    env_names = []
23
    def usage_err():
24
        raise SystemExit('Usage: '+opts.env_usage(env_names, True)+' '
25
            +sys.argv[0]+' <map repl [col_num] [| '+sys.argv[0]
26
            +' repl_1 [col_num_1]]... >new_map')
27
28
    text = opts.env_flag('text', False, env_names) # all patterns are plain text
29 205 aaronmk
    try: _prog_name, repl_path = sys.argv[:2]
30 4357 aaronmk
    except ValueError: usage_err()
31 205 aaronmk
    col_num = None
32
    try: col_num = sys.argv[2]
33 211 aaronmk
    except IndexError: pass
34 205 aaronmk
    if col_num != None: col_num = int(col_num) # 0-based
35
36
    # Get replacements
37
    repls = []
38
    stream = open(repl_path, 'rb')
39
    reader = csv.reader(stream)
40 8740 aaronmk
    reader.next() # skip header
41 205 aaronmk
    for row in reader:
42 210 aaronmk
        in_, out = row[:2]
43 216 aaronmk
        if in_ != '':
44 6702 aaronmk
            if text or re.match(r'^\w+$', in_): # match word
45 6713 aaronmk
                in_ = (r'(?:^|(?<=[\s,"])|^:|(?<=[\s,"]):)'+re.escape(in_)
46 10283 aaronmk
                    +r'(?![a-zA-Z0-9])')
47 6713 aaronmk
            repls.append((r'(?m)'+in_, out))
48 205 aaronmk
    stream.close()
49
    def repl_all(str_):
50 9784 aaronmk
        str_ = strings.ustr(str_)
51 9785 aaronmk
        for repl, with_ in repls:
52
            if with_ == 'unescape_html()': with_ = repl_unescape_html
53
            str_ = re.sub(repl, with_, str_)
54 205 aaronmk
        return str_
55
56
    # Modify map or file
57
    if col_num != None:
58
        reader = csv.reader(sys.stdin)
59
        writer = csv.writer(sys.stdout)
60
        cols = reader.next()
61
        writer.writerow(cols)
62
        for row in reader:
63
            row[col_num] = repl_all(row[col_num])
64
            writer.writerow(row)
65 9784 aaronmk
    else: sys.stdout.write(strings.to_raw_str(repl_all(sys.stdin.read())))
66 205 aaronmk
67
main()