fixed read_soft to deal with stupid Soft format feature

git-svn-id: http://biopieces.googlecode.com/svn/trunk@219 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/code_perl/Maasha/Biopieces.pm b/code_perl/Maasha/Biopieces.pm
index 11c8020346a80201b4a139d375d6cf1765cf1b20..ce308a3ebcba15cd00ea1c9ef68bfb098e88e0e3 100644 (file)
--- a/code_perl/Maasha/Biopieces.pm
+++ b/code_perl/Maasha/Biopieces.pm
@@ -1015,7 +1015,7 @@ sub get_options
                 Maasha::Common::error( qq(Genome $options{ $opt } not found in "$ENV{ 'BP_DATA' }/genomes/") );
             }
         }
-        elsif ( $opt eq "terminal" and not $options{ $opt } =~ /^(svg|post|dumb)/ )
+        elsif ( $opt eq "terminal" and not $options{ $opt } =~ /^(svg|post|dumb|x11)/ )
         {
             Maasha::Common::error( qq(Bad --$opt argument "$options{ $opt }") );
         }
@@ -1736,7 +1736,7 @@ sub script_read_soft
 
     # Returns nothing.
 
-    my ( $data_in, $file, $num, $records, $record, $soft_index, $fh, @platforms, $plat_table, @samples, $sample, $old_end );
+    my ( $data_in, $file, $num, $records, $record, $soft_index, $fh, @platforms, $plat_table, @samples, $sample, $old_end, $skip );
 
     while ( $record = get_record( $in ) ) {
         put_record( $record, $out );
@@ -1746,25 +1746,30 @@ sub script_read_soft
 
     foreach $file ( @{ $options->{ "files" } } )
     {
+        print STDERR "Creating index for file: $file\n" if $options->{ "verbose" };
+
         $soft_index = Maasha::NCBI::soft_index_file( $file );
 
         $fh         = Maasha::Common::read_open( $file );
 
-        @platforms  = grep { $_->[ 0 ] =~ /PLATFORM/ } @{ $soft_index };
+        @platforms  = grep { $_->{ "SECTION" } =~ /PLATFORM/ } @{ $soft_index };
+
+        print STDERR "Getting platform tables for file: $file\n" if $options->{ "verbose" };
 
-        $plat_table = Maasha::NCBI::soft_get_platform( $fh, $platforms[ 0 ]->[ 1 ], $platforms[ -1 ]->[ 2 ] );
+        $plat_table = Maasha::NCBI::soft_get_platform( $fh, $platforms[ 0 ]->{ "LINE_BEG" }, $platforms[ -1 ]->{ "LINE_END" } );
 
-        @samples    = grep { $_->[ 0 ] =~ /SAMPLE/ } @{ $soft_index };
+        @samples    = grep { $_->{ "SECTION" } =~ /SAMPLE/ } @{ $soft_index };
 
-        $old_end    = $platforms[ -1 ]->[ 2 ];
+        $old_end    = $platforms[ -1 ]->{ "LINE_END" };
 
         foreach $sample ( @samples )
         {
-            if ( $options->{ "samples" } ) {
-                next if grep { $sample->[ 0 ] !~ $_ } @{ $options->{ "samples" } };
-            }
+            $skip = 0;
+            $skip = 1 if ( $options->{ "samples" } and grep { $sample->{ "SECTION" } !~ /$_/ } @{ $options->{ "samples" } } );
+
+            print STDERR "Getting samples for dataset: $sample->{ 'SECTION' }\n" if $options->{ "verbose" } and not $skip;
 
-            $records = Maasha::NCBI::soft_get_sample( $fh, $plat_table, $sample->[ 1 ] - $old_end - 1, $sample->[ 2 ] - $old_end - 1 );
+            $records = Maasha::NCBI::soft_get_sample( $fh, $plat_table, $sample->{ "LINE_BEG" } - $old_end - 1, $sample->{ "LINE_END" } - $old_end - 1, $skip );
 
             foreach $record ( @{ $records } )
             {
@@ -1775,7 +1780,7 @@ sub script_read_soft
                 $num++;
             }
 
-            $old_end = $sample->[ 2 ];
+            $old_end = $sample->{ "LINE_END" };
         }
 
         close $fh;

diff --git a/code_perl/Maasha/NCBI.pm b/code_perl/Maasha/NCBI.pm
index e6b7407e90031004cc875592ec31c2b9e532d631..a94dc4a88550236db861eb83630198ed6240e36c 100644 (file)
--- a/code_perl/Maasha/NCBI.pm
+++ b/code_perl/Maasha/NCBI.pm
@@ -155,7 +155,7 @@ sub soft_parse
                 @fields = split "\t", $lines[ $c ];
 
                 $id_hash{ $fields[ $key_hash{ "ID" } ] } = $fields[ $key_hash{ "SEQUENCE" } ];
-            
+
                 $c++;
             }
 
@@ -233,8 +233,8 @@ sub soft_index_file
 
     $first = 1;
 
-    $i = 0;
-    $c = 0;
+    $i      = 0;
+    $c      = 0;
 
     while ( $line = <$fh> )
     {
@@ -242,11 +242,14 @@ sub soft_index_file
 
         if ( $line =~ /^\^/ )
         {
-            push @index, [ $line, $i ];
+            push @index, {
+                SECTION  => $line, 
+                LINE_BEG => $i,
+            };
 
             if ( not $first )
             {
-                push @{ $index[ $c - 1 ] }, $i - 1;
+                $index[ $c - 1 ]{ "LINE_END" } = $i - 1;
             }
             else
             {
@@ -259,7 +262,7 @@ sub soft_index_file
         $i++; 
     }
 
-    push @{ $index[ $c - 1 ] }, $i - 1;
+    $index[ $c - 1 ]{ "LINE_END" } = $i - 1;
 
     close $fh;
 
@@ -271,7 +274,7 @@ sub soft_get_platform
 {
     # Martin A. Hansen, June 2008.
 
-    # Given a filehandle to a SOFT file parses the platform table
+    # Given a filehandle to a SOFT file parses all the platform tables
     # which is returned.
 
     my ( $fh,    # filehandle
@@ -314,7 +317,7 @@ sub soft_get_platform
             {
                 @fields = split "\t", $lines[ $c ];
 
-                $id_hash{ $fields[ $key_hash{ "ID" } ] } = $fields[ $key_hash{ "SEQUENCE" } ];
+                $id_hash{ $fields[ $key_hash{ "ID" } ] } = $fields[ $key_hash{ "SEQUENCE" } ] || $fields[ $key_hash{ "ID" } ];
             
                 $c++;
             }
@@ -333,16 +336,17 @@ sub soft_get_sample
 {
     # Martin A. Hansen, June 2008.
 
-    # Given a filehandle to a SOFT file parses the platform table
-    # which is returned.
+    # Given a filehandle to a SOFT file and a platform table 
+    # parses sample records which are returned.
 
-    my ( $fh,         # filehandle
-         $plat_table,  # hashref with platform tables
-         $beg,        # line number where sample table begin
-         $end,        # line number where sample table end
+    my ( $fh,           # filehandle
+         $plat_table,   # hashref with platform table
+         $beg,          # line number where sample table begin
+         $end,          # line number where sample table end
+         $skip,         # flag indicating that this sample table should not be parsed.
        ) = @_;
 
-    # Returns hashref
+    # Returns list of hashref
 
     my ( $line, @lines, $i, $c, $platform_id, @fields, %key_hash, $num, $sample_id, $sample_title, $id, $seq, $count, @records, $record );
 
@@ -350,15 +354,22 @@ sub soft_get_sample
 
     while ( $line = <$fh> )
     {
-        chomp $line;
+        if ( not $skip )
+        {
+            chomp $line;
+
+            push @lines, $line if $i >= $beg;
+        }
 
-        push @lines, $line if $i >= $beg;
-    
         last if $i == $end;
 
         $i++;
     }
 
+    if ( $skip ) {
+        return wantarray ? () : [];
+    }
+
     $i = 0;
 
     $num = 1;