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 10348 aaronmk
quote_re = '[\'"`]'
18 12746 aaronmk
excluded_prefix_re = (
19 12749 aaronmk
     '(?<! )' # not if it's a word in a sentence
20 12750 aaronmk
    +'(?<!-)' # not if it's part of a '-'-separated identifier
21 12749 aaronmk
    +'(?<!\*)' # don't double leading *
22
    )
23
excluded_suffix_re = (
24
     '(?! )' # not if it's a word in a sentence
25 12750 aaronmk
    +'(?!-)' # not if it's part of a '-'-separated identifier
26 12749 aaronmk
    )
27 10348 aaronmk
28 9785 aaronmk
def unescape_html(str_): return HTMLParser.HTMLParser().unescape(str_)
29
30
def repl_unescape_html(match): return unescape_html(match.group(0))
31
32 205 aaronmk
def main():
33 4357 aaronmk
    env_names = []
34
    def usage_err():
35
        raise SystemExit('Usage: '+opts.env_usage(env_names, True)+' '
36
            +sys.argv[0]+' <map repl [col_num] [| '+sys.argv[0]
37
            +' repl_1 [col_num_1]]... >new_map')
38
39
    text = opts.env_flag('text', False, env_names) # all patterns are plain text
40 205 aaronmk
    try: _prog_name, repl_path = sys.argv[:2]
41 4357 aaronmk
    except ValueError: usage_err()
42 205 aaronmk
    col_num = None
43
    try: col_num = sys.argv[2]
44 211 aaronmk
    except IndexError: pass
45 205 aaronmk
    if col_num != None: col_num = int(col_num) # 0-based
46
47
    # Get replacements
48
    repls = []
49
    stream = open(repl_path, 'rb')
50
    reader = csv.reader(stream)
51 8740 aaronmk
    reader.next() # skip header
52 205 aaronmk
    for row in reader:
53 210 aaronmk
        in_, out = row[:2]
54 216 aaronmk
        if in_ != '':
55 10348 aaronmk
            is_word = re.match(r'^\w+$', in_)
56
            if text or is_word: # match as whole-word text (like SQL identifier)
57
                in_str_re = re.escape(in_)
58
                q = quote_re
59
                in_ = '(?<='+q+')'+in_str_re+'(?='+q+')' # require quotes
60 12742 aaronmk
                if is_word: # also match with quotes optional
61 12748 aaronmk
                    # don't try to match word w/ suffix, because there are cases
62
                    # where a mapping adds a suffix which would cause the same
63
                    # replacement to be performed repeatedly
64 12754 aaronmk
                    in_word_re = r'\b'+in_str_re+r'\b'
65
66
                    # only use excluded_prefix_re/excluded_suffix_re in text
67
                    # mode (used in renaming columns in SQL scripts), to prevent
68
                    # the special coding for column renames from also affecting
69
                    # regular regexp/word replacements
70
                    if text: in_word_re = (excluded_prefix_re+in_word_re
71 12749 aaronmk
                        +excluded_suffix_re)
72 12754 aaronmk
73 10348 aaronmk
                    in_ = '(?:'+in_+'|'+in_word_re+')'
74 6713 aaronmk
            repls.append((r'(?m)'+in_, out))
75 205 aaronmk
    stream.close()
76
    def repl_all(str_):
77 9784 aaronmk
        str_ = strings.ustr(str_)
78 9785 aaronmk
        for repl, with_ in repls:
79
            if with_ == 'unescape_html()': with_ = repl_unescape_html
80
            str_ = re.sub(repl, with_, str_)
81 205 aaronmk
        return str_
82
83
    # Modify map or file
84
    if col_num != None:
85
        reader = csv.reader(sys.stdin)
86
        writer = csv.writer(sys.stdout)
87
        cols = reader.next()
88
        writer.writerow(cols)
89
        for row in reader:
90
            row[col_num] = repl_all(row[col_num])
91
            writer.writerow(row)
92 9784 aaronmk
    else: sys.stdout.write(strings.to_raw_str(repl_all(sys.stdin.read())))
93 205 aaronmk
94
main()