Add autodetection of log format (syslog or stderr) if none is given with option -f

diff --git a/pgbadger b/pgbadger
index 8999242292f31a4bda6ddca0672cddc885ac7698..1fb08367ecd690130cd7725f55f8c07c67b59f85 100755 (executable)
--- a/pgbadger
+++ b/pgbadger
@@ -84,7 +84,7 @@ if (!$logfile) {
 }
 
 # Set default format
-$format ||= 'stderr';
+$format ||= &autodetect_format();
 # Set default syslog ident name
 $ident ||= 'postgres';
 # Set default top query
@@ -1847,4 +1847,41 @@ sub create_graph_twoaxes
 
 }
 
+sub autodetect_format
+{
+
+       # Open log file for reading
+       my $nfound = 0;
+       my $nline = 0;
+       my $fmt = '';
+       my $tfile = new IO::File;
+       if ($logfile !~ /\.gz/) {
+               $tfile->open($logfile) || die "FATAL: cannot read logfile $logfile. $!\n";
+       } else {
+               # Open a pipe to zcat program for compressed log
+               $tfile->open("$ZCAT_PROG $logfile |") || die "FATAL: cannot read from pipe to $ZCAT_PROG $logfile. $!\n";
+       }
+       while (my $line = <$tfile>) {
+               chomp($line);
+               $line =~ s/\r//;
+               next if (!$line);
+               $nline++;
+               # Is syslog lines ?
+               if ($line =~ /^...\s+\d+\s+\d+:\d+:\d+\s+[^\s]+\s+[^\[]+\[\d+\]:\s+\[[0-9\-]+\]\s+[^:]+:\s+duration:/) {
+                       $fmt = 'syslog';
+                       $nfound++;
+               # Is stderr lines
+               } elsif ($line =~ /\d+-\d+-\d+\s+\d+:\d+:\d+\s+[^\s]+\s+\[\d+\]:\s+\[[0-9\-]+\]\s+[^:]+:\s+duration:/) {
+                       $fmt = 'stderr';
+                       $nfound++;
+               }
+               last if (($nfound > 10) || ($nline > 5000));
+       }
+       $tfile->close();
+       if (!$fmt || ($nfound < 10)) {
+               die "FATAL: unable to detect log file format, please use -f option.\n";
+       }
+
+       return $fmt;
+}