Add --progress option

tseemann · Apr 1, 2018 · 107d504 · 107d504
1 parent fd45ec4
commit 107d504
Show file tree

Hide file tree

Showing 3 changed files with 19 additions and 9 deletions.
diff --git a/.gitignore b/.gitignore
@@ -33,3 +33,7 @@ inc/
 /MANIFEST.bak
 /pm_to_blib
 /*.zip
+
+# samclip
+*~
+/empty.fai
diff --git a/.travis.yml b/.travis.yml
@@ -26,3 +26,4 @@ script:
     - "samclip --ref test.fna < test.sam 2>&1 > /dev/null | grep 'Header contained 5 lines'"
     - "samclip --max 0 --ref test.fna test.sam 2>&1 > /dev/null | grep Done"
     - "samclip --invert --ref test.fna test.sam 2>&1 > /dev/null | grep Done"
+    - "samclip --progress 50 --ref test.fna test.sam 2>&1 > /dev/null | grep 'Processed 150 records'"
diff --git a/samclip b/samclip
@@ -26,10 +26,11 @@ use constant {
 #----------------------------------------------------------------------
 # command line parameters
 
-my $max    = 5;
-my $ref    = '';
-my $invert = 0;
-my $debug  = 0;
+my $max      = 5;
+my $ref      = '';
+my $invert   = 0;
+my $debug    = 0;
+my $progress = 100_000;
 
 #----------------------------------------------------------------------
 sub usage {
@@ -43,11 +44,13 @@ sub usage {
     "  % samclip --ref ref.fa < in.sam > out.sam\n",
     "  % minimap2 ref.fa R1.fq R2.fq | samclip --ref ref.fa | samtools sort > out.bam\n",
     "OPTIONS\n",
-    "  --help        This help.\n",
-    "  --version     Print version and exit.\n",
-    "  --ref FASTA   Reference genome - needs FASTA.fai index.\n",
-    "  --invert      Output rejected SAM lines and ignore good ones.\n",
-    "  --debug       Print verbose debug info to stderr.\n",
+    "  --help         This help\n",
+    "  --version      Print version and exit\n",
+    "  --ref FASTA    Reference genome - needs FASTA.fai index\n",
+    "  --max NUM      Maximum clip length to allow (default=$max)\n",
+    "  --invert       Output rejected SAM lines and ignore good ones\n",
+    "  --debug        Print verbose debug info to stderr\n",
+    "  --progress N   Print progress every NUM records (default=$progress,none=0)\n",
     "HOMEPAGE\n  $HOMEPAGE\n",
     "";
   exit($exitcode);
@@ -65,6 +68,7 @@ GetOptions(
   "max=i"      => \$max,
   "invert"     => \$invert,
   "debug"      => \$debug,
+  "progress=i" => \$progress,
 ) or usage(1);
 
 $ref or err("Please supply reference genome with --ref");
@@ -97,6 +101,7 @@ while (my $line = <ARGV>) {
     next;
   }
   $total++;
+  msg("Processed $total records...") if $progress and $total % $progress == 0;
   my @sam = split m/\t/, $line;
   # do a quick 'clipped?' check before heavyweight parsing
   if ($sam[SAM_CIGAR] =~ /\d[SH]/) {