File Coverage

blib/lib/MCE/Grep.pm

Criterion	Covered	Total	%
statement	175	229	76.4
branch	80	164	48.7
condition	24	55	43.6
subroutine	15	18	83.3
pod	5	5	100.0
total	299	471	63.4

###############################################################################
## ----------------------------------------------------------------------------
## Parallel grep model similar to the native grep function.
##
###############################################################################

package MCE::Grep;

use strict;
use warnings;

no warnings qw( threads recursion uninitialized );

our $VERSION = '1.889';

## no critic (BuiltinFunctions::ProhibitStringyEval)
## no critic (Subroutines::ProhibitSubroutinePrototypes)
## no critic (TestingAndDebugging::ProhibitNoStrict)

use Scalar::Util qw( looks_like_number weaken );
use MCE;

our @CARP_NOT = qw( MCE );

my $_tid = $INC{'threads.pm'} ? threads->tid() : 0;

sub CLONE {
   $_tid = threads->tid() if $INC{'threads.pm'};
}

###############################################################################
## ----------------------------------------------------------------------------
## Import routine.
##
###############################################################################

my ($_MCE, $_def, $_params, $_prev_c, $_tag) = ({}, {}, {}, {}, 'MCE::Grep');

sub import {
   my ($_class, $_pkg) = (shift, caller);

   my $_p = $_def->{$_pkg} = {
      MAX_WORKERS => 'auto',
      CHUNK_SIZE  => 'auto',
   };

   ## Import functions.
   if ($_pkg !~ /^MCE::/) {
      no strict 'refs'; no warnings 'redefine';
      *{ $_pkg.'::mce_grep_f' } = \&run_file;
      *{ $_pkg.'::mce_grep_s' } = \&run_seq;
      *{ $_pkg.'::mce_grep'   } = \&run;
   }

   ## Process module arguments.
   while ( my $_argument = shift ) {
      my $_arg = lc $_argument;

      $_p->{MAX_WORKERS} = shift, next if ( $_arg eq 'max_workers' );
      $_p->{CHUNK_SIZE}  = shift, next if ( $_arg eq 'chunk_size' );
      $_p->{TMP_DIR}     = shift, next if ( $_arg eq 'tmp_dir' );
      $_p->{FREEZE}      = shift, next if ( $_arg eq 'freeze' );
      $_p->{THAW}        = shift, next if ( $_arg eq 'thaw' );
      $_p->{INIT_RELAY}  = shift, next if ( $_arg eq 'init_relay' );
      $_p->{USE_THREADS} = shift, next if ( $_arg eq 'use_threads' );

      ## Sereal 3.015+, if available, is used automatically by MCE 1.8+.
      if ( $_arg eq 'sereal' ) {
         if ( shift eq '0' ) {
            require Storable;
            $_p->{FREEZE} = \&Storable::freeze;
            $_p->{THAW}   = \&Storable::thaw;
         }
         next;
      }

      _croak("Error: ($_argument) invalid module option");
   }

   $_p->{MAX_WORKERS} = MCE::_parse_max_workers($_p->{MAX_WORKERS});

   MCE::_validate_number($_p->{MAX_WORKERS}, 'MAX_WORKERS', $_tag);
   MCE::_validate_number($_p->{CHUNK_SIZE}, 'CHUNK_SIZE', $_tag)
      unless ($_p->{CHUNK_SIZE} eq 'auto');

   return;
}

###############################################################################
## ----------------------------------------------------------------------------
## Gather callback for storing by chunk_id => chunk_ref into a hash.
##
###############################################################################

my ($_total_chunks, %_tmp);

sub _gather {

   my ($_chunk_id, $_data_ref) = @_;

   $_tmp{$_chunk_id} = $_data_ref;
   $_total_chunks++;

   return;
}

###############################################################################
## ----------------------------------------------------------------------------
## Init and finish routines.
##
###############################################################################

sub init (@) {

   shift if (defined $_[0] && $_[0] eq 'MCE::Grep');
   my $_pkg = "$$.$_tid.".caller();

   $_params->{$_pkg} = (ref $_[0] eq 'HASH') ? shift : { @_ };

   _croak("$_tag: (HASH) not allowed as input by this MCE model")
      if ( ref $_params->{$_pkg}{input_data} eq 'HASH' );

   @_ = ();

   return;
}

sub finish (@) {

   shift if (defined $_[0] && $_[0] eq 'MCE::Grep');
   my $_pkg = (defined $_[0]) ? shift : "$$.$_tid.".caller();

   if ( $_pkg eq 'MCE' ) {
      for my $_k ( keys %{ $_MCE } ) { MCE::Grep->finish($_k, 1); }
   }
   elsif ( $_MCE->{$_pkg} && $_MCE->{$_pkg}{_init_pid} eq "$$.$_tid" ) {
      $_MCE->{$_pkg}->shutdown(@_) if $_MCE->{$_pkg}{_spawned};
      $_total_chunks = undef, undef %_tmp;

      delete $_prev_c->{$_pkg};
      delete $_MCE->{$_pkg};
   }

   @_ = ();

   return;
}

###############################################################################
## ----------------------------------------------------------------------------
## Parallel grep with MCE -- file.
##
###############################################################################

sub run_file (&@) {

   shift if (defined $_[0] && $_[0] eq 'MCE::Grep');

   my $_code = shift; my $_file = shift;
   my $_pid  = "$$.$_tid.".caller();

   if (defined (my $_p = $_params->{$_pid})) {
      delete $_p->{input_data} if (exists $_p->{input_data});
      delete $_p->{sequence}   if (exists $_p->{sequence});
   }
   else {
      $_params->{$_pid} = {};
   }

   if (defined $_file && ref $_file eq '' && $_file ne '') {
      _croak("$_tag: ($_file) does not exist")      unless (-e $_file);
      _croak("$_tag: ($_file) is not readable")     unless (-r $_file);
      _croak("$_tag: ($_file) is not a plain file") unless (-f $_file);
      $_params->{$_pid}{_file} = $_file;
   }
   elsif (ref $_file eq 'SCALAR' || ref($_file) =~ /^(?:GLOB|FileHandle|IO::)/) {
      $_params->{$_pid}{_file} = $_file;
   }
   else {
      _croak("$_tag: (file) is not specified or valid");
   }

   @_ = ();

   return run($_code);
}

###############################################################################
## ----------------------------------------------------------------------------
## Parallel grep with MCE -- sequence.
##
###############################################################################

sub run_seq (&@) {

   shift if (defined $_[0] && $_[0] eq 'MCE::Grep');

   my $_code = shift;
   my $_pid  = "$$.$_tid.".caller();

   if (defined (my $_p = $_params->{$_pid})) {
      delete $_p->{input_data} if (exists $_p->{input_data});
      delete $_p->{_file}      if (exists $_p->{_file});
   }
   else {
      $_params->{$_pid} = {};
   }

   my ($_begin, $_end);

   if (ref $_[0] eq 'HASH') {
      $_begin = $_[0]->{begin}, $_end = $_[0]->{end};
      $_params->{$_pid}{sequence} = $_[0];
   }
   elsif (ref $_[0] eq 'ARRAY') {
      if (@{ $_[0] } > 3 && $_[0]->[3] =~ /\d$/) {
         $_begin = $_[0]->[0], $_end = $_[0]->[-1];
         $_params->{$_pid}{sequence} = [ $_[0]->[0], $_[0]->[-1] ];
      }
      else {
         $_begin = $_[0]->[0], $_end = $_[0]->[1];
         $_params->{$_pid}{sequence} = $_[0];
      }
   }
   elsif (ref $_[0] eq '' || ref($_[0]) =~ /^Math::/) {
      if (@_ > 3 && $_[3] =~ /\d$/) {
         $_begin = $_[0], $_end = $_[-1];
         $_params->{$_pid}{sequence} = [ $_[0], $_[-1] ];
      }
      else {
         $_begin = $_[0], $_end = $_[1];
         $_params->{$_pid}{sequence} = [ @_ ];
      }
   }
   else {
      _croak("$_tag: (sequence) is not specified or valid");
   }

   _croak("$_tag: (begin) is not specified for sequence")
      unless (defined $_begin);
   _croak("$_tag: (end) is not specified for sequence")
      unless (defined $_end);

   $_params->{$_pid}{sequence_run} = undef;

   @_ = ();

   return run($_code);
}

###############################################################################
## ----------------------------------------------------------------------------
## Parallel grep with MCE.
##
###############################################################################

sub run (&@) {

   shift if (defined $_[0] && $_[0] eq 'MCE::Grep');

   my $_code = shift;  $_total_chunks = 0; undef %_tmp;
   my $_pkg  = caller() eq 'MCE::Grep' ? caller(1) : caller();
   my $_pid  = "$$.$_tid.$_pkg";

   my $_input_data; my $_max_workers = $_def->{$_pkg}{MAX_WORKERS};
   my $_r = ref $_[0];

   if (@_ == 1 && $_r =~ /^(?:ARRAY|HASH|SCALAR|CODE|GLOB|FileHandle|IO::)/) {
      _croak("$_tag: (HASH) not allowed as input by this MCE model")
         if $_r eq 'HASH';
      $_input_data = shift;
   }

   if (defined (my $_p = $_params->{$_pid})) {
      $_max_workers = MCE::_parse_max_workers($_p->{max_workers})
         if (exists $_p->{max_workers});

      delete $_p->{sequence}    if (defined $_input_data || scalar @_);
      delete $_p->{user_func}   if (exists $_p->{user_func});
      delete $_p->{user_tasks}  if (exists $_p->{user_tasks});
      delete $_p->{use_slurpio} if (exists $_p->{use_slurpio});
      delete $_p->{bounds_only} if (exists $_p->{bounds_only});
      delete $_p->{gather}      if (exists $_p->{gather});
   }

   my $_chunk_size = do {
      my $_p = $_params->{$_pid} || {};
      (defined $_p->{init_relay} || defined $_def->{$_pkg}{INIT_RELAY}) ? 1 :
      MCE::_parse_chunk_size(
         $_def->{$_pkg}{CHUNK_SIZE}, $_max_workers, $_params->{$_pid},
         $_input_data, scalar @_
      );
   };

   if (defined (my $_p = $_params->{$_pid})) {
      if (exists $_p->{_file}) {
         $_input_data = delete $_p->{_file};
      } else {
         $_input_data = $_p->{input_data} if exists $_p->{input_data};
      }
   }

   ## -------------------------------------------------------------------------

   MCE::_save_state($_MCE->{$_pid});

   if (!defined $_prev_c->{$_pid} || $_prev_c->{$_pid} != $_code) {
      $_MCE->{$_pid}->shutdown() if (defined $_MCE->{$_pid});
      $_prev_c->{$_pid} = $_code;

      my %_opts = (
         max_workers => $_max_workers, task_name => $_tag,
         user_func => sub {

            my ($_mce, $_chunk_ref, $_chunk_id) = @_;
            my $_wantarray = $_mce->{user_args}[0];

            if ($_wantarray) {
               my @_a;

               if (ref $_chunk_ref eq 'SCALAR') {
                  local $/ = $_mce->{RS} if defined $_mce->{RS};
                  open my $_MEM_FH, '<', $_chunk_ref;
                  binmode $_MEM_FH, ':raw';
                  while (<$_MEM_FH>) { push (@_a, $_) if &{ $_code }; }
                  close   $_MEM_FH;
                  weaken  $_MEM_FH;
               }
               else {
                  if (ref $_chunk_ref) {
                     push @_a, grep { &{ $_code } } @{ $_chunk_ref };
                  } else {
                     push @_a, grep { &{ $_code } } $_chunk_ref;
                  }
               }

               MCE->gather($_chunk_id, \@_a);
            }
            else {
               my $_cnt = 0;

               if (ref $_chunk_ref eq 'SCALAR') {
                  local $/ = $_mce->{RS} if defined $_mce->{RS};
                  open my $_MEM_FH, '<', $_chunk_ref;
                  binmode $_MEM_FH, ':raw';
                  while (<$_MEM_FH>) { $_cnt++ if &{ $_code }; }
                  close   $_MEM_FH;
                  weaken  $_MEM_FH;
               }
               else {
                  if (ref $_chunk_ref) {
                     $_cnt += grep { &{ $_code } } @{ $_chunk_ref };
                  } else {
                     $_cnt += grep { &{ $_code } } $_chunk_ref;
                  }
               }

               MCE->gather($_cnt) if defined $_wantarray;
            }
         },
      );

      if (defined (my $_p = $_params->{$_pid})) {
         for my $_k (keys %{ $_p }) {
            next if ($_k eq 'sequence_run');
            next if ($_k eq 'input_data');
            next if ($_k eq 'chunk_size');

            _croak("$_tag: ($_k) is not a valid constructor argument")
               unless (exists $MCE::_valid_fields_new{$_k});

            $_opts{$_k} = $_p->{$_k};
         }
      }

      for my $_k (qw/ tmp_dir freeze thaw init_relay use_threads /) {
         $_opts{$_k} = $_def->{$_pkg}{uc($_k)}
            if (exists $_def->{$_pkg}{uc($_k)} && !exists $_opts{$_k});
      }

      $_MCE->{$_pid} = MCE->new(pkg => $_pkg, %_opts);
   }

   ## -------------------------------------------------------------------------

   my $_cnt = 0; my $_wantarray = wantarray;

   $_MCE->{$_pid}{use_slurpio} = ($_chunk_size > &MCE::MAX_RECS_SIZE) ? 1 : 0;
   $_MCE->{$_pid}{user_args}   = [ $_wantarray ];

   $_MCE->{$_pid}{gather} = $_wantarray
      ? \&_gather : sub { $_cnt += $_[0]; return; };

   if (defined $_input_data) {
      @_ = ();
      $_MCE->{$_pid}->process({ chunk_size => $_chunk_size }, $_input_data);
      delete $_MCE->{$_pid}{input_data};
   }
   elsif (scalar @_) {
      $_MCE->{$_pid}->process({ chunk_size => $_chunk_size }, \@_);
      delete $_MCE->{$_pid}{input_data};
   }
   else {
      if (defined $_params->{$_pid} && exists $_params->{$_pid}{sequence}) {
         $_MCE->{$_pid}->run({
             chunk_size => $_chunk_size,
             sequence   => $_params->{$_pid}{sequence}
         }, 0);
         if (exists $_params->{$_pid}{sequence_run}) {
             delete $_params->{$_pid}{sequence_run};
             delete $_params->{$_pid}{sequence};
         }
         delete $_MCE->{$_pid}{sequence};
      }
   }

   MCE::_restore_state();

   if ($_wantarray) {
      return map { @{ $_ } } delete @_tmp{ 1 .. $_total_chunks };
   }
   elsif (defined $_wantarray) {
      return $_cnt;
   }

   return;
}

###############################################################################
## ----------------------------------------------------------------------------
## Private methods.
##
###############################################################################

sub _croak {

   goto &MCE::_croak;
}

1;

__END__