Try to detect user/database/host from connection strings if

log_connection is enabled and log_line_prefix doesn't include
them.

Extend the regex to autodetect database name, user name, client
ip address and application name. The regex now are the foolowing:

	db => qr/(?:db|database)=([^,]*)/;
	user => qr/(?:user|usr)=([^,]*)/;
	client => qr/(?:client|remote|ip|host)=([^,]*)/;
	appname => qr/(?:app|application)=([^,]*)/;

I hope this patch will limit the number of RTFM support answer.

diff --git a/pgbadger b/pgbadger
index aa5a8421dbb78c60cfcf8b47f6348c9d48a245ef..af1c4989571dc5dedd2bfb4e92892545265a4ad1 100644 (file)
--- a/pgbadger
+++ b/pgbadger
@@ -781,10 +781,10 @@ if ($error_only && $disable_error) {
 }
 
 # Set default search pattern for database, user name, application name and host in log_line_prefix
-my $regex_prefix_dbname = qr/db=([^,]*)/;
-my $regex_prefix_dbuser = qr/user=([^,]*)/;
-my $regex_prefix_dbclient = qr/(?:client|remote)=([^,]*)/;
-my $regex_prefix_dbappname = qr/app=([^,]*)/;
+my $regex_prefix_dbname = qr/(?:db|database)=([^,]*)/;
+my $regex_prefix_dbuser = qr/(?:user|usr)=([^,]*)/;
+my $regex_prefix_dbclient = qr/(?:client|remote|ip|host)=([^,]*)/;
+my $regex_prefix_dbappname = qr/(?:app|application)=([^,]*)/;
 
 # Set pattern to look for query type
 my $action_regex = qr/^[\s\(]*(DELETE|INSERT|UPDATE|SELECT|COPY|WITH|CREATE|DROP|ALTER|TRUNCATE|BEGIN|COMMIT|ROLLBACK|START|END|SAVEPOINT)/is;
@@ -3818,6 +3818,7 @@ Average queries per sessions: $avg_queries
 Average queries duration per sessions: $avg_duration
 };
                foreach (sort {$overall_stat{'peak'}{$b}{session} <=> $overall_stat{'peak'}{$a}{session}} keys %{$overall_stat{'peak'}}) {
+                       next if (!$session_info{count});
                        print $fh "Session peak: ", &comma_numbers($overall_stat{'peak'}{$_}{session}), " sessions at $_";
                        last;
                }
@@ -4938,7 +4939,7 @@ sub print_overall_statistics
        my $query_peak_date = '';
        foreach (sort {$overall_stat{'peak'}{$b}{query} <=> $overall_stat{'peak'}{$a}{query}} keys %{$overall_stat{'peak'}}) {
                $query_peak = &comma_numbers($overall_stat{'peak'}{$_}{query});
-               $query_peak_date = $_;
+               $query_peak_date = $_ if ($query_peak);
                last;
        }
        my $avg_queries = &comma_numbers(int($overall_stat{'queries_number'}/($session_info{count} || 1)));
@@ -4961,12 +4962,13 @@ sub print_overall_statistics
        my $session_peak_date = '';
        foreach (sort {$overall_stat{'peak'}{$b}{connection} <=> $overall_stat{'peak'}{$a}{connection}} keys %{$overall_stat{'peak'}}) {
                $connection_peak = &comma_numbers($overall_stat{'peak'}{$_}{connection});
-               $connection_peak_date = $_;
+               $connection_peak_date = $_ if ($connection_peak);
                last;
        }
        foreach (sort {$overall_stat{'peak'}{$b}{session} <=> $overall_stat{'peak'}{$a}{session}} keys %{$overall_stat{'peak'}}) {
+               next if (!$session_count);
                $session_peak = &comma_numbers($overall_stat{'peak'}{$_}{session});
-               $session_peak_date = $_;
+               $session_peak_date = $_ if ($session_peak);
                last;
        }
         my $main_error = 0;
@@ -5359,15 +5361,14 @@ sub print_sql_traffic
        my $query_peak_date = '';
        foreach (sort {$overall_stat{'peak'}{$b}{query} <=> $overall_stat{'peak'}{$a}{query}} keys %{$overall_stat{'peak'}}) {
                $query_peak = &comma_numbers($overall_stat{'peak'}{$_}{query});
-               $query_peak_date = $_;
+               $query_peak_date = $_ if ($query_peak);
                last;
        }
-
        my $select_peak = 0;
        my $select_peak_date = '';
        foreach (sort {$overall_stat{'peak'}{$b}{select} <=> $overall_stat{'peak'}{$a}{select}} keys %{$overall_stat{'peak'}}) {
                $select_peak = &comma_numbers($overall_stat{'peak'}{$_}{select});
-               $select_peak_date = $_;
+               $select_peak_date = $_ if ($select_peak);
                last;
        }
 
@@ -5375,9 +5376,10 @@ sub print_sql_traffic
        my $write_peak_date = '';
        foreach (sort {$overall_stat{'peak'}{$b}{write} <=> $overall_stat{'peak'}{$a}{write}} keys %{$overall_stat{'peak'}}) {
                $write_peak = &comma_numbers($overall_stat{'peak'}{$_}{write});
-               $write_peak_date = $_;
+               $write_peak_date = $_ if ($write_peak);
                last;
        }
+
        my $fmt_duration = &convert_time($overall_stat{'queries_duration'});
 
        print $fh qq{
@@ -5925,7 +5927,7 @@ sub print_established_connection
        my $connection_peak_date = '';
        foreach (sort {$overall_stat{'peak'}{$b}{connection} <=> $overall_stat{'peak'}{$a}{connection}} keys %{$overall_stat{'peak'}}) {
                $connection_peak = &comma_numbers($overall_stat{'peak'}{$_}{connection});
-               $connection_peak_date = $_;
+               $connection_peak_date = $_ if ($connection_peak);
                last;
        }
 
@@ -6459,7 +6461,7 @@ sub print_simultaneous_session
        my $session_peak_date = '';
        foreach (sort {$overall_stat{'peak'}{$b}{session} <=> $overall_stat{'peak'}{$a}{session}} keys %{$overall_stat{'peak'}}) {
                $session_peak = &comma_numbers($overall_stat{'peak'}{$_}{session});
-               $session_peak_date = $_;
+               $session_peak_date = $_ if ($session_peak);
                last;
        }
 
@@ -7562,7 +7564,7 @@ sub print_temporary_file
        my $tempfile_size_peak_date = '';
        foreach (sort {$overall_stat{'peak'}{$b}{tempfile_size} <=> $overall_stat{'peak'}{$a}{tempfile_size}} keys %{$overall_stat{'peak'}}) {
                $tempfile_size_peak = &pretty_print_size($overall_stat{'peak'}{$_}{tempfile_size});
-               $tempfile_size_peak_date = $_;
+               $tempfile_size_peak_date = $_ if ($tempfile_size_peak);
                last;
        }
        print $fh qq{
@@ -7574,7 +7576,7 @@ sub print_temporary_file
                <h3 class="">Key values</h3>
                <div class="well key-figures">
                        <ul>
-                               <li><span class="figure">$tempfile_size_peak</span> <span class="figure-label">Temp Files Peak</span></li>
+                               <li><span class="figure">$tempfile_size_peak</span> <span class="figure-label">Temp Files size Peak</span></li>
                                <li><span class="figure">$tempfile_size_peak_date</span> <span class="figure-label">Date</span></li>
                        </ul>
                </div>
@@ -7590,7 +7592,7 @@ $drawn_graphs{temporarydata_graph}
        my $tempfile_count_peak_date = '';
        foreach (sort {$overall_stat{'peak'}{$b}{tempfile_count} <=> $overall_stat{'peak'}{$a}{tempfile_count}} keys %{$overall_stat{'peak'}}) {
                $tempfile_count_peak = &comma_numbers($overall_stat{'peak'}{$_}{tempfile_count});
-               $tempfile_count_peak_date = $_;
+               $tempfile_count_peak_date = $_ if ($tempfile_count_peak);
                last;
        }
        print $fh qq{
@@ -12026,6 +12028,26 @@ sub validate_log_line
 {
        my ($t_pid) = @_;
 
+       # Set details about connection if prefix doesn't included them
+       # and log_connection is enabled
+       if ($prefix_vars{'t_loglevel'} eq 'LOG') {
+               if ( !$prefix_vars{'t_client'} && ($prefix_vars{'t_query'} =~ /connection received: host=([^\s]+)(?: port=(\d+))?/) ) {
+                       $current_sessions{$prefix_vars{'t_pid'}}{host} = $1;
+                       return 1;
+               } elsif ( !$prefix_vars{'t_dbname'} && ($prefix_vars{'t_query'} =~ /connection authorized: user=([^\s]+)(?: database=([^\s]+))?/) ) {
+                       $current_sessions{$prefix_vars{'t_pid'}}{user} = $1;
+                       $current_sessions{$prefix_vars{'t_pid'}}{database} = $2;
+               }
+       }
+
+       # Set details from previous connection line
+       $prefix_vars{'t_dbname'} = $current_sessions{$t_pid}{database}
+               if (!$prefix_vars{'t_dbname'} && exists $current_sessions{$t_pid}{database});
+       $prefix_vars{'t_dbuser'} = $current_sessions{$t_pid}{user}
+               if (!$prefix_vars{'t_user'} && exists $current_sessions{$t_pid}{user});
+       $prefix_vars{'t_client'} = $current_sessions{$t_pid}{host}
+               if (!$prefix_vars{'t_client'} && exists $current_sessions{$t_pid}{host});
+
        # Look at particular cases of vacuum/analyze that have the database
        # name inside the log message so that they could be associated
        if ($prefix_vars{'t_query'} =~ / of table "([^\.]+)\.[^\.]+\.[^\.]+":/) {
@@ -12312,6 +12334,7 @@ sub parse_query
 
        # Stores pre-connection activity
        if (($prefix_vars{'t_loglevel'} eq 'LOG') && ($prefix_vars{'t_query'} =~ /connection received: host=([^\s]+)(?: port=(\d+))?/)) {
+               $current_sessions{$prefix_vars{'t_pid'}}{host} = $1;
                return if ($disable_connection);
                $conn_received{$t_pid} = $1;
                $conn_received{$t_pid} = _gethostbyaddr($conn_received{$t_pid}) if ($dns_resolv);
@@ -12320,9 +12343,10 @@ sub parse_query
 
        # Stores connection activity
        if (   ($prefix_vars{'t_loglevel'} eq 'LOG')
-               && ($prefix_vars{'t_query'} =~ /connection authorized: user=([^\s]+) /))
+               && ($prefix_vars{'t_query'} =~ /connection authorized: user=([^\s]+)(?: database=([^\s]+))?/))
        {
-               $current_sessions{$prefix_vars{'t_pid'}} = 1 if (!$disable_session);
+               $current_sessions{$prefix_vars{'t_pid'}}{user} = $1;
+               $current_sessions{$prefix_vars{'t_pid'}}{database} = $2;
 
                return if ($disable_connection);