From: Don Armstrong <don@donarmstrong.com>
Date: Thu, 13 Dec 2007 02:31:32 +0000 (+0000)
Subject:  * Start supporting keyword weights
X-Git-Url: https://git.donarmstrong.com/?p=function2gene.git;a=commitdiff_plain;h=364cd5a25c2bb985b587a6b0b5d2e23de18d9c10

 * Start supporting keyword weights
 * Transition from do_it_all_state to function2gene_state
 * Fix invalidate_state function to invalidate state dependencies


git-svn-id: file:///srv/svn/function2gene/trunk@23 a0738b58-4706-0410-8799-fb830574a030
---

diff --git a/bin/function2gene b/bin/function2gene
index c407047..f3bd3dd 100755
--- a/bin/function2gene
+++ b/bin/function2gene
@@ -41,7 +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>
 
@@ -158,9 +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>;
@@ -190,6 +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, $_;
@@ -220,17 +225,40 @@ 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;
 	  }
-	  elsif (not exists $state{done_keywords}{$method}{$database}) {
+	  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;
 	  }
-	  elsif (not exists $state{done_keywords}{$method}{$database}{$keyword}) {
+	  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;
 	  }
-	  else {
-	       delete $state{done_keywords}{$method}{$database}{$keyword};
+	  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};
 	  }
      }
 }
@@ -296,7 +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;
      }
 }
@@ -310,8 +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) {
@@ -383,7 +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: $!";