added on local at 2026-07-01 21:14:30
| 1 | #!/usr/bin/perl | |
| 2 | #====================================================================== | |
| 3 | # ABForge -- _funnel_discovery_worker.pl | |
| 4 | # | |
| 5 | # Runs periodically (cron: hourly). For every active site with tracking | |
| 6 | # data, looks at the top-N most common page sequences in the last 30 | |
| 7 | # days and materializes each as a funnel row with source='auto'. The | |
| 8 | # customer can then rename, edit, or hide these in /funnels.cgi. | |
| 9 | # | |
| 10 | # Idempotency: keyed by (site_id, slug). A slug is a normalized digest | |
| 11 | # of the step sequence, so the same sequence never gets duplicated even | |
| 12 | # across runs. | |
| 13 | # | |
| 14 | # Hide-persistence: if a user hides an auto funnel, is_hidden=1 stays | |
| 15 | # set even when this worker sees the same slug again -- we use | |
| 16 | # INSERT ... ON DUPLICATE KEY UPDATE name=VALUES(name) and never touch | |
| 17 | # is_hidden after the initial insert. | |
| 18 | #====================================================================== | |
| 19 | use strict; | |
| 20 | use warnings; | |
| 21 | use lib '/var/www/vhosts/3dshawn.com/abforge.3dshawn.com'; | |
| 22 | use MODS::DBConnect; | |
| 23 | use MODS::ABForge::Config; | |
| 24 | use Digest::MD5 qw(md5_hex); | |
| 25 | ||
| 26 | my $db = MODS::DBConnect->new; | |
| 27 | my $cfg = MODS::ABForge::Config->new; | |
| 28 | my $DB = $cfg->settings('database_name'); | |
| 29 | my $dbh = $db->db_connect; | |
| 30 | die "no dbh\n" unless $dbh; | |
| 31 | ||
| 32 | my @sites = $db->db_readwrite_multiple($dbh, qq~ | |
| 33 | SELECT id FROM `${DB}`.sites WHERE status IN ('active','paused') | |
| 34 | ~, __FILE__, __LINE__); | |
| 35 | ||
| 36 | my $inserted = 0; | |
| 37 | my $skipped = 0; | |
| 38 | ||
| 39 | foreach my $s (@sites) { | |
| 40 | my $sid = int($s->{id}); | |
| 41 | ||
| 42 | # Pull the last 30 days of sessions, materialize their page paths. | |
| 43 | my @sessions = $db->db_readwrite_multiple($dbh, qq~ | |
| 44 | SELECT s.id, | |
| 45 | GROUP_CONCAT(e.page_path ORDER BY e.occurred_at SEPARATOR '\x1f') AS seq | |
| 46 | FROM `${DB}`.mkt_tracking_sessions s | |
| 47 | JOIN `${DB}`.mkt_tracking_events e ON e.session_id = s.id | |
| 48 | WHERE s.storefront_id = $sid | |
| 49 | AND e.event_type = 'page_view' | |
| 50 | AND e.occurred_at >= DATE_SUB(NOW(), INTERVAL 30 DAY) | |
| 51 | GROUP BY s.id | |
| 52 | HAVING seq IS NOT NULL AND seq != '' | |
| 53 | ~, __FILE__, __LINE__); | |
| 54 | ||
| 55 | # Bucket sequences by 2..4-step prefixes so we surface both short | |
| 56 | # and medium funnels. Skip single-page paths (bounces) -- they're | |
| 57 | # not funnels. | |
| 58 | my %counts; # slug => { steps=>[...], name=>'...', n=>N } | |
| 59 | foreach my $r (@sessions) { | |
| 60 | my @pages = split /\x1f/, ($r->{seq} // ''); | |
| 61 | next if scalar(@pages) < 2; | |
| 62 | ||
| 63 | # Deduplicate consecutive identical pages (F5 refresh spam) | |
| 64 | my @clean; | |
| 65 | foreach my $p (@pages) { push @clean, $p if !@clean || $clean[-1] ne $p; } | |
| 66 | next if scalar(@clean) < 2; | |
| 67 | ||
| 68 | for my $len (2..($#clean < 3 ? $#clean : 3)) { | |
| 69 | last if $len > scalar(@clean) - 1; | |
| 70 | my @slice = @clean[0..$len]; | |
| 71 | my $key = join('|', @slice); | |
| 72 | my $slug = 'auto_' . substr(md5_hex($sid . ':' . $key), 0, 16); | |
| 73 | $counts{$slug} ||= { steps => \@slice, name => _name_from_slice(\@slice), n => 0 }; | |
| 74 | $counts{$slug}{n}++; | |
| 75 | } | |
| 76 | } | |
| 77 | ||
| 78 | # Keep only the top 8 sequences per site so the funnel list stays useful | |
| 79 | my @top = sort { $counts{$b}{n} <=> $counts{$a}{n} } keys %counts; | |
| 80 | @top = @top[0..7] if scalar(@top) > 8; | |
| 81 | ||
| 82 | foreach my $slug (@top) { | |
| 83 | my $entry = $counts{$slug}; | |
| 84 | next if $entry->{n} < 3; # need at least 3 sessions to bother | |
| 85 | ||
| 86 | my $name = $entry->{name}; $name =~ s/'/''/g; | |
| 87 | my $steps_j = _steps_json($entry->{steps}); $steps_j =~ s/'/''/g; | |
| 88 | my $slug_q = $slug; $slug_q =~ s/'/''/g; | |
| 89 | ||
| 90 | # Insert-or-noop: if a funnel with this (site_id, slug) already | |
| 91 | # exists, leave it alone (respecting user's hide + rename). | |
| 92 | my $r = $db->db_readwrite($dbh, qq~ | |
| 93 | SELECT id, is_hidden FROM `${DB}`.funnels | |
| 94 | WHERE site_id='$sid' AND slug='$slug_q' LIMIT 1 | |
| 95 | ~, __FILE__, __LINE__); | |
| 96 | if ($r && $r->{id}) { $skipped++; next; } | |
| 97 | ||
| 98 | $db->db_readwrite($dbh, qq~ | |
| 99 | INSERT INTO `${DB}`.funnels | |
| 100 | (site_id, name, steps, is_active, is_hidden, source, slug, created_at) | |
| 101 | VALUES | |
| 102 | ('$sid', '$name', '$steps_j', 1, 0, 'auto', '$slug_q', NOW()) | |
| 103 | ~, __FILE__, __LINE__); | |
| 104 | $inserted++; | |
| 105 | } | |
| 106 | } | |
| 107 | ||
| 108 | $db->db_disconnect($dbh); | |
| 109 | ||
| 110 | print "discovery: inserted=$inserted skipped=$skipped\n"; | |
| 111 | ||
| 112 | sub _name_from_slice { | |
| 113 | my ($slice) = @_; | |
| 114 | return join(' -> ', map { my $p = $_; $p =~ s{^/}{}; $p = 'Home' if !length $p; substr($p, 0, 40) } @$slice); | |
| 115 | } | |
| 116 | ||
| 117 | sub _steps_json { | |
| 118 | my ($slice) = @_; | |
| 119 | my @parts; | |
| 120 | my $idx = 0; | |
| 121 | foreach my $p (@$slice) { | |
| 122 | $idx++; | |
| 123 | my $label = $p; $label =~ s{^/}{}; $label = 'Home' if !length $label; | |
| 124 | $label = substr($label, 0, 60); | |
| 125 | # Minimal JSON escape: backslash + quote | |
| 126 | my $p_esc = $p; $p_esc =~ s/\\/\\\\/g; $p_esc =~ s/"/\\"/g; | |
| 127 | my $l_esc = $label; $l_esc =~ s/\\/\\\\/g; $l_esc =~ s/"/\\"/g; | |
| 128 | push @parts, qq~{"kind":"path","match":"$p_esc","label":"$l_esc"}~; | |
| 129 | } | |
| 130 | return '[' . join(',', @parts) . ']'; | |
| 131 | } |