limit all pagination to 10000 results

Paged results are expensive to compute on Elasticsearch. It will refuse
to give results if paged forward far enough.

Limit our pagination and give empty results if paged too deep.

Author: haarg

lib/MetaCPAN/Query/Favorite.pm

@@ -3,7 +3,7 @@ package MetaCPAN::Query::Favorite;
 use MetaCPAN::Moose;
 
 use MetaCPAN::ESConfig qw( es_doc_path );
-use MetaCPAN::Util     qw(hit_total);
+use MetaCPAN::Util     qw( MAX_RESULT_WINDOW hit_total );
 
 with 'MetaCPAN::Query::Role::Common';
 
@@ -148,6 +148,14 @@ sub recent {
     $page //= 1;
     $size //= 100;
 
+    if ( $page * $size >= MAX_RESULT_WINDOW ) {
+        return +{
+            favorites => [],
+            took      => 0,
+            total     => 0,
+        };
+    }
+
     my $favs = $self->es->search(
         es_doc_path('favorite'),
         body => {

lib/MetaCPAN/Query/Release.pm

@@ -4,7 +4,7 @@ use MetaCPAN::Moose;
 
 use MetaCPAN::ESConfig qw( es_doc_path );
 use MetaCPAN::Util
-    qw( hit_total single_valued_arrayref_to_scalar true false );
+    qw( MAX_RESULT_WINDOW hit_total single_valued_arrayref_to_scalar true false );
 
 with 'MetaCPAN::Query::Role::Common';
 
@@ -399,6 +399,14 @@ sub by_author {
     $size //= 1000;
     $page //= 1;
 
+    if ( $page * $size >= MAX_RESULT_WINDOW ) {
+        return {
+            releases => [],
+            took     => 0,
+            total    => 0,
+        };
+    }
+
     my $body = {
         query => {
             bool => {
@@ -491,6 +499,14 @@ sub all_by_author {
     $size //= 100;
     $page //= 1;
 
+    if ( $page * $size >= MAX_RESULT_WINDOW ) {
+        return {
+            releases => [],
+            took     => 0,
+            total    => 0,
+        };
+    }
+
     my $body = {
         query   => { term => { author => uc($author) } },
         sort    => [ { date => 'desc' } ],
@@ -713,6 +729,14 @@ sub _get_depended_releases {
     $page      //= 1;
     $page_size //= 50;
 
+    if ( $page * $page_size >= MAX_RESULT_WINDOW ) {
+        return +{
+            data  => [],
+            took  => 0,
+            total => 0,
+        };
+    }
+
     $sort = _fix_sort_value($sort);
 
     my $dependency_filter = {
@@ -773,17 +797,15 @@ sub recent {
     $page_size //= 10000;
     $type      //= '';
 
-    my $query;
-    my $from = ( $page - 1 ) * $page_size;
-
-    if ( $from + $page_size > 10000 ) {
-        return {
+    if ( $page * $page_size >= MAX_RESULT_WINDOW ) {
+        return +{
             releases => [],
-            total    => 0,
             took     => 0,
+            total    => 0,
         };
     }
 
+    my $query;
     if ( $type eq 'n' ) {
         $query = {
             constant_score => {

lib/MetaCPAN/Query/Search.pm

@@ -8,7 +8,7 @@ use List::Util                qw( min uniq );
 use Log::Contextual           qw( :log :dlog );
 use MetaCPAN::ESConfig        qw( es_doc_path );
 use MetaCPAN::Types::TypeTiny qw( Object Str );
-use MetaCPAN::Util            qw( hit_total true false );
+use MetaCPAN::Util            qw( MAX_RESULT_WINDOW hit_total true false );
 use MooseX::StrictConstructor;
 
 with 'MetaCPAN::Query::Role::Common';
@@ -60,6 +60,15 @@ sub search_web {
     $page_size //= 20;
     $page      //= 1;
 
+    if ( $page * $page_size >= MAX_RESULT_WINDOW ) {
+        return {
+            results  => [],
+            total    => 0,
+            tool     => 0,
+            colapsed => $collapsed ? true : false,
+        };
+    }
+
     $search_term =~ s{([+=><!&|\(\)\{\}[\]\^"~*?\\/])}{\\$1}g;
 
     # munge the search_term

lib/MetaCPAN/Util.pm

@@ -37,9 +37,12 @@ use Sub::Exporter -setup => {
         true
         false
         is_bool
+        MAX_RESULT_WINDOW
     ) ]
 };
 
+use constant MAX_RESULT_WINDOW => 10000;
+
 *true    = \&Cpanel::JSON::XS::true;
 *false   = \&Cpanel::JSON::XS::false;
 *is_bool = \&Cpanel::JSON::XS::is_bool;