From eddd05213730c2d41260c589c97730ec84b67a70 Mon Sep 17 00:00:00 2001
From: Hubert depesz Lubaczewski <depesz@depesz.com>
Date: Sun, 6 Feb 2022 17:00:12 +0100
Subject: [PATCH] change output of find_interesting_plans

---
 scripts/find_interesting_plans | 42 +++++++++++++++++++++++-----------
 1 file changed, 29 insertions(+), 13 deletions(-)

diff --git a/scripts/find_interesting_plans b/scripts/find_interesting_plans
index 14f32b5..d0f22fd 100755
--- a/scripts/find_interesting_plans
+++ b/scripts/find_interesting_plans
@@ -60,6 +60,8 @@ for my $partno ( 0 .. $#{ $partitions } ) {
     $pm->start and next PARTITION;
     printf "Starting partition #%d\n", 1 + $partno;
 
+    open my $out, '>', "${output}/${part}.txt";
+
     my $dbh = get_dbh();
 
     my $seen        = 0;
@@ -83,18 +85,21 @@ for my $partno ( 0 .. $#{ $partitions } ) {
                 next;
             }
             next unless $explain->top_node;
-            next unless is_plan_interesting( $explain );
-            $interesting++;
 
-            my $output_path = sprintf '%s/%s.plan', $output, $row->{ 'id' };
-            open my $fh, '>', $output_path;
-            print $fh $row->{ 'plan' };
-            close $fh;
+            my $is_interesting = 0;
+
+            for my $line ( @{ output_lines( $row->{'id'}, $explain ) } ) {
+                print $out join("\t", @{$line}) . "\n";
+                $is_interesting = 1;
+            }
+
+            $interesting++ if $is_interesting;
         }
     }
 
     $dbh->rollback();
     $dbh->disconnect();
+    close $out;
     printf "Partition %d done. %d plans scanned, %d errored out, %d interesting saved.\n", 1 + $partno, $seen, $errors, $interesting;
     $pm->finish;
 }
@@ -104,18 +109,29 @@ printf "All done, output in %s\n", $output;
 
 exit;
 
-sub is_plan_interesting {
-    my $plan = shift;
+sub output_lines {
+    my ( $id, $plan ) = @_;
+    my $ret = [];
     for my $node ( $plan->top_node, $plan->top_node->all_recursive_subnodes ) {
+
+        next unless $node->estimated_row_width;
         next unless $node->total_rows;
-        next unless $node->total_rows > 200;
         next unless $node->total_rows_removed;
-        next unless $node->total_rows_removed > 9 * $node->total_rows;
         next unless $node->extra_info;
-        next unless any { /^Filter:/ } @{ $node->extra_info };
-        return 1;
+
+        # At least 3 pages worth of data is returned
+        next unless $node->total_rows * $node->estimated_row_width > 3 * 8192;
+
+        # At least 90% of rows were removed
+        next unless $node->total_rows_removed > 9 * $node->total_rows;
+
+        my @filter_lines = grep { /^Filter:/ } @{ $node->extra_info };
+        # There are filter expressions
+        next if 0 == scalar @filter_lines;
+
+        push @{ $ret }, map { [ $id, $node->id, $node->type, $node->estimated_row_width, $node->total_rows, $node->total_rows_removed, $_ ] } @filter_lines;
     }
-    return;
+    return $ret;
 }
 
 sub get_dbh {