added on local at 2026-07-01 21:22:27
| 1 | #!/usr/bin/perl | |
| 2 | #====================================================================== | |
| 3 | # PTMatrix -- _funnel_discovery_worker.pl | |
| 4 | # | |
| 5 | # Auto-discovers top page sequences from anon_sessions + anon_pageviews | |
| 6 | # and materializes them as funnel rows with source='auto'. The admin can | |
| 7 | # rename, edit steps, or hide these in /admin_funnels.cgi. Hidden rows | |
| 8 | # stay hidden across worker re-runs (worker respects user state via | |
| 9 | # slug-idempotency). | |
| 10 | # | |
| 11 | # Runs hourly via cron. | |
| 12 | #====================================================================== | |
| 13 | use strict; | |
| 14 | use warnings; | |
| 15 | use lib '/var/www/vhosts/3dshawn.com/ptmatrix.3dshawn.com'; | |
| 16 | use MODS::DBConnect; | |
| 17 | use MODS::PTMatrix::Config; | |
| 18 | use Digest::MD5 qw(md5_hex); | |
| 19 | ||
| 20 | my $db = MODS::DBConnect->new; | |
| 21 | my $cfg = MODS::PTMatrix::Config->new; | |
| 22 | my $DB = $cfg->settings('database_name'); | |
| 23 | my $dbh = $db->db_connect; | |
| 24 | die "no dbh\n" unless $dbh; | |
| 25 | ||
| 26 | # Sequence separator (avoid arrows since page paths may contain -> characters) | |
| 27 | my $SEP = "\x1f"; | |
| 28 | ||
| 29 | # Pull last 30 days of anon sessions, materialize their pageview sequences. | |
| 30 | my @rows = $db->db_readwrite_multiple($dbh, qq~ | |
| 31 | SELECT s.id, | |
| 32 | GROUP_CONCAT(p.page_path ORDER BY p.sequence SEPARATOR '${SEP}') AS seq | |
| 33 | FROM `${DB}`.anon_sessions s | |
| 34 | JOIN `${DB}`.anon_pageviews p ON p.session_id = s.id | |
| 35 | WHERE s.first_seen_at >= DATE_SUB(NOW(), INTERVAL 30 DAY) | |
| 36 | GROUP BY s.id | |
| 37 | HAVING seq IS NOT NULL AND seq != '' | |
| 38 | ~, __FILE__, __LINE__); | |
| 39 | ||
| 40 | # Bucket by 2..4-step prefixes so we surface funnels of a few lengths. | |
| 41 | my %counts; | |
| 42 | foreach my $r (@rows) { | |
| 43 | my @pages = split /$SEP/, ($r->{seq} // ''); | |
| 44 | next if scalar(@pages) < 2; | |
| 45 | ||
| 46 | # Collapse consecutive duplicates (F5 spam). | |
| 47 | my @clean; | |
| 48 | foreach my $p (@pages) { push @clean, $p if !@clean || $clean[-1] ne $p; } | |
| 49 | next if scalar(@clean) < 2; | |
| 50 | ||
| 51 | my $max_end = $#clean < 3 ? $#clean : 3; | |
| 52 | for my $end (1 .. $max_end) { | |
| 53 | my @slice = @clean[0..$end]; | |
| 54 | my $key = join('|', @slice); | |
| 55 | my $slug = 'auto_' . substr(md5_hex($key), 0, 16); | |
| 56 | $counts{$slug} ||= { steps => \@slice, name => _name_from_slice(\@slice), n => 0 }; | |
| 57 | $counts{$slug}{n}++; | |
| 58 | } | |
| 59 | } | |
| 60 | ||
| 61 | # Keep the top 8 sequences. | |
| 62 | my @top = sort { $counts{$b}{n} <=> $counts{$a}{n} } keys %counts; | |
| 63 | @top = @top[0..7] if scalar(@top) > 8; | |
| 64 | ||
| 65 | my $inserted = 0; | |
| 66 | my $skipped = 0; | |
| 67 | ||
| 68 | foreach my $slug (@top) { | |
| 69 | my $entry = $counts{$slug}; | |
| 70 | next if $entry->{n} < 3; # need at least 3 sessions to bother | |
| 71 | ||
| 72 | my $name = $entry->{name}; $name =~ s/'/''/g; | |
| 73 | my $steps_j = _steps_json($entry->{steps}); $steps_j =~ s/'/''/g; | |
| 74 | my $slug_q = $slug; $slug_q =~ s/'/''/g; | |
| 75 | ||
| 76 | # Insert-or-noop: if the slug already exists, leave it alone so the | |
| 77 | # user's rename / hide state is preserved. | |
| 78 | my $r = $db->db_readwrite($dbh, qq~ | |
| 79 | SELECT id FROM `${DB}`.funnels WHERE slug='$slug_q' LIMIT 1 | |
| 80 | ~, __FILE__, __LINE__); | |
| 81 | if ($r && $r->{id}) { $skipped++; next; } | |
| 82 | ||
| 83 | $db->db_readwrite($dbh, qq~ | |
| 84 | INSERT INTO `${DB}`.funnels | |
| 85 | (name, steps, is_active, is_hidden, source, slug, created_at) | |
| 86 | VALUES | |
| 87 | ('$name', '$steps_j', 1, 0, 'auto', '$slug_q', NOW()) | |
| 88 | ~, __FILE__, __LINE__); | |
| 89 | $inserted++; | |
| 90 | } | |
| 91 | ||
| 92 | $db->db_disconnect($dbh); | |
| 93 | print "discovery: inserted=$inserted skipped=$skipped\n"; | |
| 94 | ||
| 95 | sub _name_from_slice { | |
| 96 | my ($slice) = @_; | |
| 97 | return join(' -> ', map { | |
| 98 | my $p = $_; $p =~ s{^/}{}; $p =~ s{\.cgi$}{}; $p = 'Home' if !length $p; | |
| 99 | substr($p, 0, 40) | |
| 100 | } @$slice); | |
| 101 | } | |
| 102 | ||
| 103 | sub _steps_json { | |
| 104 | my ($slice) = @_; | |
| 105 | my @parts; | |
| 106 | foreach my $p (@$slice) { | |
| 107 | my $label = $p; $label =~ s{^/}{}; $label = 'Home' if !length $label; | |
| 108 | $label = substr($label, 0, 60); | |
| 109 | my $p_esc = $p; $p_esc =~ s/\\/\\\\/g; $p_esc =~ s/"/\\"/g; | |
| 110 | my $l_esc = $label; $l_esc =~ s/\\/\\\\/g; $l_esc =~ s/"/\\"/g; | |
| 111 | push @parts, qq~{"kind":"path","match":"$p_esc","label":"$l_esc"}~; | |
| 112 | } | |
| 113 | return '[' . join(',', @parts) . ']'; | |
| 114 | } |