X-Git-Url: https://git.donarmstrong.com/?p=function2gene.git;a=blobdiff_plain;f=bin%2Ffunction2gene;h=f3bd3ddfd7364a89afac2c4d391e43b3787e8b8d;hp=ed9a38ca23a3b948c7c07913c62efde34d1949a5;hb=364cd5a25c2bb985b587a6b0b5d2e23de18d9c10;hpb=a4b9214b7939fbe990f2373684dfd7aa1f9e77e2

diff --git a/bin/function2gene b/bin/function2gene
index ed9a38c..f3bd3dd 100755
--- a/bin/function2gene
+++ b/bin/function2gene
@@ -28,6 +28,7 @@ use Storable;
   --results directory to store results in
   --database databases to search
   --restart-at mode to start searching at
+  --invalidate-state state to invalidate
   --debug, -d debugging level (Default 0)
   --help, -h display this help
   --man, -m display manual
@@ -40,7 +41,9 @@ use Storable;
 
 A file which contains a newline delinated list of keywords to search
 for. Can be specified multiple times. Lines starting with # or ; are
-ignored.
+ignored. An optional weight can be specified after the keyword, which
+is separated from the keyword by a tab. (If not specified, 1 is
+assumed.)
 
 =item B<--results>
 
@@ -59,6 +62,14 @@ If you need to restart the process at a particular state (which has
 already been completed) specify this option. Valid values are get,
 parse, or combine.
 
+=item B<--invalidate-state>
+
+This is a more powerful version of --restart-at, which can
+specifically invalidate a certain method,database,keyword combination.
+
+For example, you can request that the keyword foo be retreived again
+from ncbi using --invalidate-state 'get,ncbi,foo'
+
 =item B<--debug, -d>
 
 Debug verbosity. (Default 0)
@@ -99,10 +110,12 @@ my %options = (databases       => [],
 	       help            => 0,
 	       man             => 0,
 	       results         => '',
+	       invalidate_state => [],
 	       );
 
 GetOptions(\%options,'keywords=s@','databases=s@',
 	   'restart_at|restart-at=s','results=s',
+	   'invalidate_state|invalidate-state=s@',
 	   'debug|d+','help|h|?','man|m');
 
 pod2usage() if $options{help};
@@ -147,9 +160,9 @@ $options{keywords} = [map {abs_path($_)} @{$options{keywords}}];
 
 chdir $options{results} or die "Unable to chdir to $options{results}";
 
-if (-e "do_it_all_state") {
+if (-e "function2gene_state") {
      ADVISE("Using existing state information");
-     my $state_fh = IO::File->new("do_it_all_state",'r') or die
+     my $state_fh = IO::File->new("function2gene_state",'r') or die
 	  "Unable to open state file for reading: $!";
      local $/;
      my $state_file = <$state_fh>;
@@ -179,6 +192,9 @@ if (@{$options{keywords}}) {
 	       next if /^\s*[#;]/;
 	       next unless /\w+/;
 	       chomp;
+	       my ($keyword,$weight) = split /\t/, $_;
+	       $weight = 1 if not defined $weight;
+	       $state{keyword_weight}{$keyword} = $weight;
 	       if (not $old_keywords{$_}) {
 		    DEBUG("Adding new keyword '$_'");
 		    push @new_keywords, $_;
@@ -206,6 +222,47 @@ if (exists $options{restart_at} and length $options{restart_at}) {
      }
 }
 
+if (exists $options{invalidate_state}) {
+     for my $invalidate_state (@{$options{invalidate_state}}) {
+	  my ($method,$database,$keyword) = split /,/, $invalidate_state;
+	  if (grep {not defined $_ } ($method,$database,$keyword) ) {
+	       print STDERR "The invalidate state option '$invalidate_state' is invalid.\n";
+	       next;
+	  }
+	  if (not exists $state{done_keywords}{$method}) {
+	       print STDERR "Method '$method' does not exist, and cannot be invalidated\n";
+	       next;
+	  }
+	  if (not exists $state{done_keywords}{$method}{$database}) {
+	       print STDERR "Database '$database' does not exist for method '$method', and cannot be invalidated\n";
+	       next;
+	  }
+	  if (not length $keyword) {
+	       delete $state{done_keywords}{$method}{$database};
+	       if ($method eq 'get') {
+		    delete $state{done_keywords}{parse}{$database};
+		    delete $state{done_keywords}{combine}{$database};
+	       }
+	       if ($method eq 'parse') {
+		    delete $state{done_keywords}{combine}{$database};
+	       }
+	       next;
+	  }
+	  if (not exists $state{done_keywords}{$method}{$database}{$keyword}) {
+	       print STDERR "Keyword '$keyword' does not exist for database '$database' and method '$method', and cannot be invalidated\n";
+	       next;
+	  }
+	  delete $state{done_keywords}{$method}{$database}{$keyword};
+	  if ($method eq 'get') {
+	       delete $state{done_keywords}{parse}{$database}{$keyword};
+	       delete $state{done_keywords}{combine}{$database}{$keyword};
+	  }
+	  if ($method eq 'parse') {
+	       delete $state{done_keywords}{combine}{$database}{$keyword};
+	  }
+     }
+}
+
 # now we need to figure out what has to happen
 # for each keyword, we check to see if we've got results, parsed
 # results, and combined it. If not, we queue up those actions.
@@ -267,7 +324,7 @@ for my $state (qw(get parse)) {
      }
      save_state(\%state);
      if ($ERRORS) {
-	  WARN("Stoping, as there are errors");
+	  WARN("Stoping, asthere are errors");
 	  exit 1;
      }
 }
@@ -281,8 +338,12 @@ if ($actions{combine}) {
 				} keys %{$state{done_keywords}{parse}{$db}}
 			   } keys %{$state{done_keywords}{parse}};
 
+     # create temporary file to store keyword weights
+
      write_command_to_file('combined_results.txt',
 			   "$base_dir/combine_results",
+			   '--keywords',
+			   
 			   @parsed_results,
 			  );
      for my $result (@parsed_results) {
@@ -354,7 +415,7 @@ sub handle_action{
 
 sub save_state{
      my ($state) = @_;
-     my $state_fh = IO::File->new("do_it_all_state",'w') or die
+     my $state_fh = IO::File->new("function2gene_state",'w') or die
 	  "Unable to open state file for writing: $!";
      print {$state_fh} freeze($state) or die "Unable to freeze state file";
      close $state_fh or die "Unable to close state file: $!";