From f440d7576cc5f9f33debf00f6f90b88a783fed35 Mon Sep 17 00:00:00 2001
From: dakkar <dakkar@thenautilus.net>
Date: Sat, 11 Jan 2020 15:12:23 +0000
Subject: pre-compute more Bayes stuff

when all my mails were indexed, it took ages to guess the mailbox
---
 lib/MaildirIndexer/Index/ByAddresses.rakumod | 21 +++++++++++++--------
 1 file changed, 13 insertions(+), 8 deletions(-)

diff --git a/lib/MaildirIndexer/Index/ByAddresses.rakumod b/lib/MaildirIndexer/Index/ByAddresses.rakumod
index 638740f..40e98f9 100644
--- a/lib/MaildirIndexer/Index/ByAddresses.rakumod
+++ b/lib/MaildirIndexer/Index/ByAddresses.rakumod
@@ -14,6 +14,7 @@ has Array[Str] %!addresses-for-file;
 has BagHash $!count-by-address-and-mailbox .= new;
 has BagHash $!known-addresses .= new;
 has BagHash $!count-by-mailbox .= new;
+has Hash $!p-address-given-mailbox .= new;
 has $!total-count;
 
 method dump() {
@@ -24,8 +25,18 @@ submethod account-for(Str @addresses,Str $mailbox,Int $step) {
     $!count-by-mailbox{$mailbox} += $step;
 
     for @addresses -> Str $addr {
+        my $pair = ( $addr => $mailbox );
         $!known-addresses{$addr} += $step;
-        $!count-by-address-and-mailbox{$addr => $mailbox} += $step;
+        my $count = $!count-by-address-and-mailbox{$pair} += $step;
+
+        if ($count) {
+            my $a = 1e-15 + $count;
+            my $b = 1 + $!count-by-mailbox{$mailbox};
+            $!p-address-given-mailbox{$pair} = $a / $b;
+        }
+        else {
+            $!p-address-given-mailbox{$pair} :delete;
+        }
     }
 }
 
@@ -58,12 +69,6 @@ method del-path(IO:D $file, Str:D $mailbox --> Nil) {
     }
 }
 
-submethod p-address-given-mailbox(Str $addr, Str $mailbox) {
-    my $a = 1e-15 + $!count-by-address-and-mailbox{$addr => $mailbox};
-    my $b = 1 + $!count-by-mailbox{$mailbox};
-    return $a / $b;
-}
-
 submethod predict-mailbox-given-addresses(@addresses) {
     my %prediction;
     my Bag $addr-bag .= new(|@addresses);
@@ -72,7 +77,7 @@ submethod predict-mailbox-given-addresses(@addresses) {
         my $p = $!count-by-mailbox{$mailbox} / $!total-count;
 
         for $!known-addresses.keys -> Str $addr {
-            my $addr-p = self.p-address-given-mailbox($addr,$mailbox);
+            my $addr-p = $!p-address-given-mailbox{$addr => $mailbox} // 1e-15;
             if ($addr-bag{$addr}) {
                 $p *= $addr-p;
             }
-- 
cgit v1.2.3