« blib/lib/App/dupfind/Threaded/MapReduce.pm
↑ Summary
blib/lib/App/dupfind/Threaded/MapReduce/Weed.pm »
| 1 | # ABSTRACT: Map-reduce version of digest_dups method and worker thread for it | ||
| 2 | |||
| 3 | 1 | ▶ | use strict; |
|
Subroutine
BEGIN: called CC=1 SCAR=0.0
| |||
| 4 | 1 | ▶ | use warnings; |
|
Subroutine
BEGIN: called CC=1 SCAR=0.0
| |||
| 5 | |||
| 6 | package App::dupfind::Threaded::MapReduce::Digest; | ||
| 7 | { | ||
| 8 | $App::dupfind::Threaded::MapReduce::Digest::VERSION = '0.140230'; # TRIAL | ||
| 9 | } | ||
| 10 | |||
| 11 | 1 | ▶ | use 5.010; |
|
Subroutine
BEGIN: called CC=1 SCAR=0.0
| |||
| 12 | |||
| 13 | 0 | ▶ | use threads; |
|
Subroutine
BEGIN: called CC=1 SCAR=2.6
| |||
| 14 | use threads::shared; | ||
| 15 | |||
| 16 | use Moo::Role; | ||
| 17 | |||
| 18 | use Time::HiRes 'usleep'; | ||
| 19 | use Digest::xxHash 'xxhash_hex'; | ||
| 20 | |||
| 21 | requires 'opts'; | ||
| 22 | |||
| 23 | my $read_semaphore :shared; | ||
| 24 | |||
| 25 | sub digest_dups | ||
| 26 | { | ||
| 27 | my ( $self, $size_dups ) = @_; | ||
| 28 | |||
| 29 | # you have to do this for this threaded version of dupfind, and it has | ||
| 30 | # to happen after you've already pruned out the hardlinks | ||
| 31 | |||
| 32 | # don't bother to hash zero-size files | ||
| 33 | |||
| 34 | my $zero_digest = xxhash_hex '', 0; | ||
| 35 | |||
| 36 | my $zero_files = delete $size_dups->{0}; | ||
| 37 | |||
| 38 | my $map_code = sub { $self->_digest_worker( @_ ) }; | ||
| 39 | |||
| 40 | my $reduced = $self->map_reduce( $size_dups => $map_code ); | ||
| 41 | |||
| 42 | $reduced->{ $zero_digest } = $zero_files if ref $zero_files; | ||
| 43 | |||
| 44 | return $reduced; | ||
| 45 | } | ||
| 46 | |||
| 47 | # !! The code to be mapped MUST increment $self->counter for each item it sees | ||
| 48 | sub _digest_worker | ||
| 49 | { | ||
| 50 | my $self = shift; | ||
| 51 | |||
| 52 | my $digest_cache = {}; | ||
| 53 | my $cache_stop = $self->opts->{cachestop}; | ||
| 54 | my $max_cache = $self->opts->{cachesize}; | ||
| 55 | my $ram_caching = !! $self->opts->{ramcache}; | ||
| 56 | my $cache_size = 0; | ||
| 57 | my $cache_hits = 0; | ||
| 58 | my $cache_misses = 0; | ||
| 59 | |||
| 60 | local $/; | ||
| 61 | |||
| 62 | WORKER: while | ||
| 63 | ( | ||
| 64 | ! $self->term_flag | ||
| 65 | && defined ( my $grouping = $self->work_queue->dequeue ) | ||
| 66 | ) | ||
| 67 | { | ||
| 68 | my $data; | ||
| 69 | |||
| 70 | GROUPING: for my $file ( @$grouping ) | ||
| 71 | { | ||
| 72 | my $size = -s $grouping->[0]; | ||
| 73 | my $digest; | ||
| 74 | |||
| 75 | open my $fh, '<', $file or do | ||
| 76 | { | ||
| 77 | $self->increment_counter; | ||
| 78 | |||
| 79 | next GROUPING; | ||
| 80 | }; | ||
| 81 | |||
| 82 | READ_LOCK: | ||
| 83 | { | ||
| 84 | lock $read_semaphore; | ||
| 85 | |||
| 86 | $data = <$fh>; | ||
| 87 | } | ||
| 88 | |||
| 89 | close $fh; | ||
| 90 | |||
| 91 | if ( $ram_caching ) | ||
| 92 | { | ||
| 93 | if ( $digest = $digest_cache->{ $data } ) | ||
| 94 | { | ||
| 95 | $cache_hits++; | ||
| 96 | } | ||
| 97 | else | ||
| 98 | { | ||
| 99 | if ( $cache_size < $max_cache && $size <= $cache_stop ) | ||
| 100 | { | ||
| 101 | $digest_cache->{ $data } = $digest = xxhash_hex $data, 0; | ||
| 102 | |||
| 103 | $cache_size++; | ||
| 104 | |||
| 105 | $cache_misses++; | ||
| 106 | } | ||
| 107 | else | ||
| 108 | { | ||
| 109 | $digest = xxhash_hex $data, 0; | ||
| 110 | } | ||
| 111 | } | ||
| 112 | } | ||
| 113 | else | ||
| 114 | { | ||
| 115 | $digest = xxhash_hex $data, 0; | ||
| 116 | } | ||
| 117 | |||
| 118 | $self->push_mapped( $digest => $file ); | ||
| 119 | |||
| 120 | $self->increment_counter; | ||
| 121 | } | ||
| 122 | |||
| 123 | $digest_cache = {}; # it's only worthwhile per-size-grouping | ||
| 124 | $cache_size = 0; | ||
| 125 | |||
| 126 | $self->add_stats( cache_hits => $cache_hits ); | ||
| 127 | $self->add_stats( cache_misses => $cache_misses ); | ||
| 128 | } | ||
| 129 | } | ||
| 130 | |||
| 131 | 1; | ||
| 132 | |||
| 133 | __END__ | ||