从subversion存储库中删除敏感信息

Posted 2021-02-26

I recently had to remove some files from a svn repo.
should work pretty much out of the box, just fill in blatantly exampleish text and read the comments

#!/usr/bin/perl
use strict;
my $svn_repo='local/path/to/repo';
my $f='full_repo_dump';
my $pattern='foldername/otherfoldername/offendingFileOrDirectory';
`svnadmin dump $svnrepo > $f`;
 
open I,'<'.$f;
open O,'>'.$f.'_cleaned';
 
my $c=0;
my $i=1;
while($_=<I>){
  while(/^Node-path: $pattern/){#this is what we want to remove
    $_=<I>;
    while(!/^Node-path/){
      $_=<I>;
      exit unless($_); #we are out of file to read, quittin time(this is likely bad form)
    }
  }
###just a workaround for large repo dumps and status bar type stuff
  if($c eq 0){
    $c=100001;
    print ++$i,"
" ;
  }
  if($i%1000 eq 0){# also if your repo dump>2gb or so it will crash. this is not very precise, but it worked in my case
    close O;          #anyway, you get the point, close and open as append only to flush whatever file buffers perl's using
    open O,'>>'.$f.'_cleaned';
  }
  $c--;
###
  print O $_;
}
 
close I;
close O;
`svnadmin create freshly_cleaned_repo`;
`svnadmin load freshly_cleaned_repo < $f`;
#you should also `svnadmin dump $svnrepo -r0 > hey_whats_my_uuid`, look inside it for your uuid and 
# then `svnadmin setuuid freshly_cleaned_repo this-is-your-really-long-uuid-here`
# pulling and setting the uuid could be automated