Replace private data in RCS file

rcsfile-replace-private-data.py

#! /usr/bin/python

import os
import re
import sys


_rcsdiff_re = re.compile(r"""
    \n
    ([^\n]+)\n
    log\n
    @(?:[^@]+|@@)*@\n
    text\n
    @((?:[^@]+|@@)*)@\n
""".encode('ascii'), re.VERBOSE)

_line_re = re.compile(b'[^\n]*(?:\n|[^\n]\Z)')

_diff_re = re.compile(b'[ad][0-9]+ ([0-9]+)')

_author_re = re.compile(b';\tauthor [^;]+')

def _replace_rcsfile(f):
    state = {'first': True, 'count': 0}

    def content_repl(tag, count, match):
        line = b'<%s/%d/%d>' % count
        if match.endswith(b'\n'):
            line += b'\n'
        return line

    def body_repl(match):
        tag = match.group(1)
        log = b'(Log %s)\n' % tag
        text = match.group(2)
        if state['first']:
            state['first'] = False
            state['count'] += 1
            count = state['count']
            text = b''.join(b'<%s/%d/%d>' % (tag, count, idx) +
                            (b'\n' if m.group(0).endswith(b'\n') else b'')
                            for idx, m in enumerate(_line_re.finditer(text)))
        else:
            lines = [m.group(0) for m in _line_re.finditer(text)]
            new_lines = []
            iter_lines = iter(lines)
            readline = lambda: next(iter_lines)
            while 1:
                try:
                    line = readline()
                except StopIteration:
                    line = None
                    break
                m = _diff_re.match(line)
                assert m, repr(line)
                new_lines.append(line)
                if line.startswith(b'd'):
                    continue
                state['count'] += 1
                count = state['count']
                n = int(m.group(1))
                while n > 0:
                    line = readline()
                    new_line = b'<%s/%d/%d>' % (tag, count, n)
                    if line.endswith(b'\n'):
                        new_line += b'\n'
                    new_lines.append(new_line)
                    n -= 1
            text = b''.join(new_lines)
        return b'\n%s\nlog\n@%s@\ntext\n@%s@\n' % (tag, log, text)

    def author_repl(match):
        return b';\tauthor anonymous'

    content = f.read()
    header, body = content.split(b'\ndesc\n', 1)
    header = re.sub(b'\nsymbols[^;]*', b'\nsymbols', header, re.M)
    header = _author_re.sub(author_repl, header)
    os.write(1, header)
    os.write(1, b'\ndesc\n')
    os.write(1, _rcsdiff_re.sub(body_repl, body))


def main(args):
    for arg in args:
        with open(arg, 'rb') as f:
            _replace_rcsfile(f)


if __name__ == '__main__':
    main(sys.argv[1:])