diff options
Diffstat (limited to 'lib/App/BorgRestore/PathTimeTable/DB.pm')
| -rw-r--r-- | lib/App/BorgRestore/PathTimeTable/DB.pm | 42 |
1 files changed, 39 insertions, 3 deletions
diff --git a/lib/App/BorgRestore/PathTimeTable/DB.pm b/lib/App/BorgRestore/PathTimeTable/DB.pm index 221063c..685c7b9 100644 --- a/lib/App/BorgRestore/PathTimeTable/DB.pm +++ b/lib/App/BorgRestore/PathTimeTable/DB.pm @@ -1,6 +1,7 @@ package App::BorgRestore::PathTimeTable::DB; use strictures 2; +use Carp::Assert; use Function::Parameters; BEGIN { use Log::Any qw($log); @@ -50,18 +51,46 @@ method set_archive_id($archive_id) { } method add_path($path, $time) { + $log->tracef("Adding path to cache: %s", $path) if TRACE; $self->{stats}->{total_paths}++; my $old_cache_path = $self->{current_path_in_cache}; + # Check if the new path requires us to (partially) invalidate our cache and + # add any files/directories to the database. If the new path is a subpath + # (substring actually) of the cached path, we can keep it only in the cache + # and no flush is needed. Otherwise we need to flush all parts of the path + # that are no longer contained in the new path. + # + # We start by checking the currently cached path ($old_cache_path) against + # the new $path. Then we remove one part from the path at a time, until we + # reach a parent path (directory) of $path. + $log->tracef("Checking if cache invalidation is required") if TRACE; while ((my $slash_index = rindex($old_cache_path, "/")) != -1) { $self->{stats}->{cache_invalidation_loop_iterations}++; - if ($old_cache_path eq substr($path, 0, length($old_cache_path))) { + # Directories in the borg output cannot be differentiated by their + # path, since their path looks just like a file path. I.e. there is no + # directory separator (/) at the end of a directory path. + # + # Since we want to keep any directory in our cache, if it contains + # $path, we can treat any cached path as a directory path. If the + # cached path was really a directory, the new $path will also contain a + # directory separator (/) between the old cached path (the parent + # directory) and the new path (a subdirectory or a file in the + # directory). If the cached path was not actually a directory, + # but a file, the new path cannot match the old one because a file name + # cannot contain a directory separator. + my $cache_check_path = $old_cache_path.'/'; + $log->tracef("Checking if cached path '%s' contains '%s'", $cache_check_path, $path) if TRACE; + if ($cache_check_path eq substr($path, 0, length($cache_check_path))) { + $log->tracef("Cache path '%s' is a parent directory of new path '%s'", $old_cache_path, $path) if TRACE; # keep the cache content for the part of the path that stays the same last; } + $log->tracef("Cached path '%s' requires flush to database", $old_cache_path) if TRACE; my $removed_time = delete $self->{cache}->{$old_cache_path}; $self->_add_path_to_db($self->{archive_id}, $old_cache_path, $removed_time); # strip last part of path $old_cache_path = substr($old_cache_path, 0, $slash_index); + $log->tracef("Changed cache path to parent directory: %s", $old_cache_path) if TRACE; # update parent timestamp my $cached = $self->{cache}->{$old_cache_path}; @@ -70,10 +99,12 @@ method add_path($path, $time) { $self->{cache}->{$old_cache_path} = $removed_time; } } + $log->tracef("Cache invalidation complete") if TRACE; - if ($old_cache_path ne substr($path, 0, length($old_cache_path))) { + my $cache_check_path = $old_cache_path.'/'; + if ($cache_check_path ne substr($path, 0, length($cache_check_path))) { # ensure that top level directory is also written - $self->_add_path_to_db($self->{archive_id}, $old_cache_path, $self->{cache}->{$old_cache_path}) unless $old_cache_path eq "."; + $self->_add_path_to_db($self->{archive_id}, $old_cache_path, $self->{cache}->{$old_cache_path}) unless ($old_cache_path eq "." or $old_cache_path eq ''); } my $cached = $self->{cache}->{$path}; @@ -100,6 +131,11 @@ method save_nodes() { for my $key (keys %{$self->{stats}}) { $log->debugf("Performance counter %s = %s", $key, $self->{stats}->{$key}); } + + # +2 because: + # - borg list gives us `.` as the first path and we essentially skip it + # - we call `add_path` with `.` at the beginning of this method + assert($self->{stats}->{real_calls_to_db_class} + 2 == $self->{stats}->{total_paths}, "All files were actually added to the database"); } 1; |