From ad6a662616e3f56880d64e924d932497e458063c Mon Sep 17 00:00:00 2001
From: Bill Erickson <berick@esilibrary.com>
Date: Thu, 9 Aug 2012 10:26:02 -0400
Subject: [PATCH] URLVerify.pm; settings, tmp caching, cleanup / misc

Signed-off-by: Bill Erickson <berick@esilibrary.com>
---
 .../perlmods/lib/OpenILS/Application/URLVerify.pm  | 87 ++++++++++++++++------
 1 file changed, 66 insertions(+), 21 deletions(-)

diff --git a/Open-ILS/src/perlmods/lib/OpenILS/Application/URLVerify.pm b/Open-ILS/src/perlmods/lib/OpenILS/Application/URLVerify.pm
index 6f6fa9f235..c3f1dd5340 100644
--- a/Open-ILS/src/perlmods/lib/OpenILS/Application/URLVerify.pm
+++ b/Open-ILS/src/perlmods/lib/OpenILS/Application/URLVerify.pm
@@ -275,26 +275,17 @@ sub verify_url {
 
     my $url = $e->retrieve_url_verify_url($url_id) or return $e->event;
 
-    my $attempt = $e->retrieve_url_verify_verification_attempt([
-        $attempt_id, {
-            flesh => 1,
-            flesh_fields => {uvva => ['session']}
-        }
-    ]) or return $e->event;
-
-    my $session = $attempt->session;
+    my ($attempt, $delay, $max_redirects, $timeout) =
+        collect_verify_attempt_and_settings($attempt_id);
 
-    return $e->event unless $e->allowed('VERIFY_URL');
+    return $e->event unless $e->allowed(
+        'VERIFY_URL', $attempt->session->owning_lib);
 
-    my $delay = 2; # TODO: org setting
-    my $max_redirects = 20; # TODO: org setting
-    my $timeout = 5; # TODO: org setting
-
-    my $depth = 0;
     my $cur_url = $url;
     my $loop_detected = 0;
+    my $redir_count = 0;
 
-    while ($depth++ < $max_redirects) {
+    while ($redir_count++ < $max_redirects) {
 
         if ($seen_urls{$cur_url->full_url}) {
             $loop_detected = 1;
@@ -315,7 +306,7 @@ sub verify_url {
         $cur_url = $url_resp->{redirect_url} or last;
     }
 
-    if ($loop_detected or $depth == $max_redirects) {
+    if ($loop_detected or $redir_count == $max_redirects) {
 
         my $vcation = Fieldmapper::url_verify::url_verification->new;
         $vcation->url($cur_url->id);
@@ -347,6 +338,65 @@ sub verify_url {
     return undef;
 }
 
+# temporarily cache some data to avoid a pile 
+# of data lookups on every URL processed.
+my %cache;
+sub collect_verify_attempt_and_settings {
+    my ($e, $attempt_id) = @_;
+    my $attempt;
+
+    if (!(keys %cache) or $cache{age} > 20) { # configurable?
+        %cache = (
+            age => 0,
+            attempt => {},
+            delay => {},
+            redirects => {},
+            timeout => {},
+        );
+    }
+
+    if ( !($attempt = $cache{attempt}{$attempt_id}) ) {
+
+        # attempt may have just been created, so 
+        # we need to guarantee a write-DB read.
+        $e->xact_begin;
+
+        $$attempt = 
+            $e->retrieve_url_verify_verification_attempt([
+                $attempt_id, {
+                    flesh => 1,
+                    flesh_fields => {uvva => ['session']}
+                }
+            ]) or return $e->die_event;
+
+        $e->rollback;
+
+        $cache{attempt}{$attempt_id} = $attempt;
+    }
+
+    my $org = $attempt->session->owning_lib;
+
+    if (!$cache{delay}{$org}) {
+        $cache{delay}{$org} = $U->ou_ancestor_setting_value(
+            $org, 'url_verify.url_verification_delay', $e) || 2;
+
+        $cache{redirects}{$org} = $U->ou_ancestor_setting_value(
+            $org, 'url_verify.url_verification_max_redirects', $e) || 20;
+
+        $cache{timeout}{$org} = $U->ou_ancestor_setting_value(
+            $org, 'url_verify.url_verification_max_wait', $e) || 5;
+    }
+
+    $cache{age}++;
+
+    return (
+        $cache{attempt}{$attempt_id},
+        $cache{delay}{$org},
+        $cache{redirects}{$org},
+        $cache{timeout}{$org}
+    );
+}
+
 =head comment
 
 1. create the verification object and commit.
@@ -423,11 +473,6 @@ sub verify_one_url {
                         $redir_url = Fieldmapper::url_verify::url->new;
                         $redir_url->redirect_from($url->id);
                         $redir_url->full_url($loc);
-                        $redir_url->item($url->item);
-                        $redir_url->url_selector($url->url_selector);
-                        $redir_url->tag($url->tag);
-                        $redir_url->subfield($url->subfield);
-                        $redir_url->ord($url->ord);
 
                         $logger->info("url: redirect found $url_text => $loc");
 
-- 
2.11.0