Added mysql-based deduping of received SHA1

This is an attempt to remove duplicate emails.  These emails happen when (for
instance) a merge occurs, and somehow, github pushes parts of both parents at
us even though we already saw one of them before.  This has been annoying a
number of us for a while, and this is my first shot at curing the problem.

Also split the config out into a separate file

Author: Beirdo

git_hooks/email_hook.conf renamed to git_hooks/email_hook.apache.conf

@@ -0,0 +1,8 @@
+debug = 1
+ignoreregexp = '^(?:master|fixes\/)'
+<db>
+  database = email_hook
+  host     = localhost
+  user     = email_hook
+  password = email_hook
+</db>

git_hooks/email_hook.cfg.example

@@ -1,13 +1,17 @@
 #! /usr/bin/perl
 # vim:ts=4:sw=4:ai:et:si:sts=4
+
 use strict;
 use warnings;
 use Apache2::Const -compile => qw(M_POST HTTP_METHOD_NOT_ALLOWED);
 use CGI;
 use JSON;
 use Mail::Send;
-
-my $debug = 1;
+use Config::General;
+use DBI;
+use File::Basename;
+use English;
+use Cwd 'abs_path';
 
 my $r = shift;
 
@@ -17,6 +21,12 @@
     return;
 }
 
+my $conffile = dirname(abs_path($0 or $PROGRAM_NAME)) . "email_hook.cfg"
+my $conf = new Config::General($conffile);
+my %config = $conf->getall;
+
+my $debug = $config{'debug'} or 0;
+
 $r->content_type('text/html');
 $r->print();
 
@@ -35,10 +45,23 @@
 my $branch = $payload->{"ref"};
 $branch =~ s/^refs\/.*?\///;
 
-if ($branch !~ /^(?:master|fixes\/)/) {
+$regexp = qr($config{'ignoreregexp'});
+if ($branch !~ $regexp) {
     exit 0;
 }
 
+my $dbh = DBI->connect("dbi:mysql:database=".$config{'db'}{'database'}.
+                       ":host=".$config{'db'}{'host'},
+                       $config{'db'}{'user'}, $config{'db'}{'password'})
+            or die "Cannot connect to database: " . DBI::errstr . "\n";
+
+my $q = "SELECT sha1 FROM seen WHERE sha1 = ?";
+my $select_h = $dbh->prepare($q);
+
+$q = "INSERT INTO seen (sha1, lastseen) VALUES (?, NULL)";
+my $insert_h = $dbh->prepare($q);
+
+
 # These maybe should go into a config file later
 my %headers = (
     "From"          => 'MythTV <noreply@mythtv.org>',
@@ -50,6 +73,10 @@
 
 foreach my $commit ( @{$payload->{"commits"}} ) {
     my $longsha = $commit->{"id"};
+    $select_h->execute($longsha);
+    my ($resultsha) = $select_h->fetchrow_array;
+    next if defined $resultsha;
+
     my $shortsha = substr $longsha, 0, 9;
     my $changeurl = $commit->{"url"};
     $changeurl =~ s/$longsha$/$shortsha/;
@@ -103,5 +130,7 @@
     my $fh = $msg->open;
     print $fh $email;
     $fh->close;
+
+    $insert_h->execute($longsha);
 }
 

git_hooks/email_hook.pl

@@ -0,0 +1,8 @@
+create database email_hook;
+grant all privileges on email_hook.* to email_hook@localhost identified by "email_hook";
+use email_hook;
+CREATE TABLE  `seen` (
+ `sha1` VARCHAR( 40 ) NOT NULL ,
+ `lastseen` TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP ,
+ PRIMARY KEY (  `sha1` )
+);