(egs/wsj/s5) Adding a new script summarize_logs.pl

The script functions just like the summarize_warnings.pl, but greps for more
interresting events (ATM, warnings _and_ errors) -- in some cases it is good
to know that some (non-fatal) errors occured. These errors _were_ non-fatal 
in the sense that the script continued on running -- but that does not mean
everything was fine.


git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@2974 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

egs/wsj/s5/utils/summarize_logs.pl

+#!/usr/bin/perl
+
+# Copyright 2012 Johns Hopkins University (Author: Daniel Povey).  Apache 2.0.
+
+ @ARGV != 1 && print STDERR "Usage: summarize_warnings.pl <log-dir>\n" && exit 1;
+
+$dir = $ARGV[0];
+
+! -d $dir && print STDERR "summarize_warnings.pl: no such directory $dir\n" && exit 1;
+
+$dir =~ s:/$::; # Remove trailing slash.
+
+
+# Group the files into categories where all have the same base-name.
+foreach $f (glob ("$dir/*.log")) {
+  $f_category = $f;
+  # do next expression twice; s///g doesn't work as they overlap.
+  $f_category =~ s:\.\d+\.:.*.:;
+  $f_category =~ s:\.\d+\.:.*.:;
+  $fmap{$f_category} .= " $f";
+}
+
+sub split_hundreds { # split list of filenames into groups of 100.
+  my $names = shift @_;
+  my @A = split(" ", $names);
+  my @ans = ();
+  while (@A > 0) {
+    my $group = "";
+    for ($x = 0; $x < 100 && @A>0; $x++) {
+      $fname = pop @A;
+      $group .= "$fname ";
+    }
+    push @ans, $group;
+  }
+  return @ans;
+}
+
+foreach $c (keys %fmap) {
+  $n = 0;
+  foreach $fgroup (split_hundreds($fmap{$c})) {
+    $n += `grep -w WARNING $fgroup | wc -l`;
+  }
+  if ($n != 0) {
+    print "$n warnings in $c\n"
+  }
+}
+foreach $c (keys %fmap) {
+  $n = 0;
+  foreach $fgroup (split_hundreds($fmap{$c})) {
+    $n += `grep -w ERROR $fgroup | wc -l`;
+  }
+  if ($n != 0) {
+    print "$n errors in $c\n"
+  }
+}