fixing a bug for when a user gives spaces in filenames

lskatz · Aug 28, 2023 · b74ae20 · b74ae20
1 parent 6268a5c
commit b74ae20
Show file tree

Hide file tree

Showing 4 changed files with 40 additions and 6 deletions.
diff --git a/bin/mashtree b/bin/mashtree
@@ -317,7 +317,7 @@ sub mashSketch{
       logmsg "WARNING: $fastq is a zero byte file. Skipping.";
       next;
     } else {
-      my $sketchCommand="mash sketch -S $$settings{seed} -k $$settings{kmerlength} -s $$settings{'sketch-size'} $sketchXopts -o $outPrefix $fastq 2>&1";
+      my $sketchCommand="mash sketch -S $$settings{seed} -k $$settings{kmerlength} -s $$settings{'sketch-size'} $sketchXopts -o \Q$outPrefix\E \Q$fastq\E 2>&1";
       my $stdout = `$sketchCommand`;
       if ($?){
         logmsg "ERROR running $sketchCommand!\n  $stdout";
@@ -421,8 +421,9 @@ sub mashDist{
   for my $msh(@$mshArr){
     #my $outfile="$outdir/".basename($msh).".tsv";
     logmsg "Distances for $msh";
-    my @distRes =`mash dist -t $msh -l $mshList`;
-    die "ERROR with 'mash dist -t $msh -l $mshList'" if $?;
+    my $cmd = "mash dist -t \Q$msh\E -l $mshList";
+    my @distRes = `$cmd`;
+    die "ERROR with $cmd" if $?;
     chomp(@distRes);
 
     my $query="";

diff --git a/lib/Mashtree.pm b/lib/Mashtree.pm
@@ -86,7 +86,7 @@ Used to mark whether a file is being read, so that Mashtree limits disk I/O
 ######
 # CONSTANTS
 
-our $VERSION = "1.4.3";
+our $VERSION = "1.4.4";
 our $MASHTREE_VERSION=$VERSION;
 our @fastqExt=qw(.fastq.gz .fastq .fq .fq.gz);
 our @fastaExt=qw(.fasta .fna .faa .mfa .fas .fsa .fa);

diff --git a/lib/Mashtree/Db.pm b/lib/Mashtree/Db.pm
@@ -434,7 +434,12 @@ sub toString_phylip{
 
   $str.=(" " x 4) . "$numGenomes\n";
   for(my $i=0;$i<$numGenomes;$i++){
-    $str.=$name[$i];
+
+    # sanitize sample names by removing spaces
+    my $sanitizedName = $name[$i];
+    $sanitizedName =~ s/\s/_/g;
+
+    $str.=$sanitizedName;
     $str.=" " x ($maxGenomeLength - length($name[$i]) + 2);
     my $distanceHash = $dist{$name[$i]};
 

diff --git a/t/02_lambda.t b/t/02_lambda.t
@@ -7,15 +7,43 @@ use FindBin qw/$RealBin/;
 
 use lib "$RealBin/../lib";
 use File::Basename qw/dirname basename/;
+use File::Temp qw/tempdir/;
+use File::Copy qw/cp/;
 use Digest::MD5 qw/md5_hex/;
 
-use Test::More tests => 6;
+use Test::More tests => 7;
 
 use_ok 'Mashtree';
 use Mashtree qw/treeDist mashDist raw_mash_distance mashHashes/;
 
 $ENV{PATH}="./bin:$ENV{PATH}";
 
+my $tempdir = tempdir(basename($0).".XXXXXX", TMP=>1, CLEANUP=>1);
+
+subtest 'space in filename' => sub {
+  my $wd = "$tempdir/spaces";
+  mkdir $wd;
+  my @target;
+  for my $filename(glob("$RealBin/lambda/*.fastq.gz")){
+    my $target = "$wd/".basename($filename);
+    $target =~ s/sample(\d)/sample $1/; # add in a space for funsies
+    cp($filename, $target) or die "ERROR: could not copy $filename to $target: $!";
+
+    push(@target, $target);
+  }
+
+  # e.g., '02_lambda.tmp.tiF_bn/spaces/sample 1.fastq.gz' '02_lambda.tmp.tiF_bn/spaces/sample 2.fastq.gz' '02_lambda.tmp.tiF_bn/spaces/sample 3.fastq.gz' '02_lambda.tmp.tiF_bn/spaces/sample 4.fastq.gz'
+  my $targets = "'" . join("' '", @target) . "'";
+  my $cmd = "mashtree --outmatrix lambdadist.tsv --genomesize 40000 --numcpus 1 $targets 2>$0.log";
+  note $cmd;
+  system($cmd);
+  my $exit_code = $? >> 8;
+  if($exit_code){
+    note `cat $0.log`;
+  }
+  is($exit_code, 0, "Ran mashtree with exit code $exit_code");
+};
+
 my $correctMashtree="(sample3:0.00195,sample4:0.00205,(sample1:0.00205,sample2:0.00205):0.00010);";
 $correctMashtree=~s/(\d+\.)(\d+)/$1 . substr($2,0,4)/ge; # global and expression