Fix intersecting period logic so proper archives will archive in parallel. (#16946)

2 files changed, 104 insertions, 25 deletions

diff --git a/tests/PHPUnit/Integration/CronArchiveTest.php b/tests/PHPUnit/Integration/CronArchiveTest.php
index 00a8b4867e..4693092589 100644
--- a/tests/PHPUnit/Integration/CronArchiveTest.php
+++ b/tests/PHPUnit/Integration/CronArchiveTest.php
@@ -697,48 +697,31 @@ Checking for queued invalidations...
   Segment "actions>=2" was created or changed recently and will therefore archive today (for site ID = 1)
   Segment "actions>=4" was created or changed recently and will therefore archive today (for site ID = 1)
 Done invalidating
-Found invalidated archive we can skip (no visits): [idinvalidation = 53, idsite = 1, period = day(2020-02-03 - 2020-02-03), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 54, idsite = 1, period = week(2020-02-03 - 2020-02-09), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 52, idsite = 1, period = day(2020-02-02 - 2020-02-02), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 50, idsite = 1, period = day(2020-02-01 - 2020-02-01), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 51, idsite = 1, period = month(2020-02-01 - 2020-02-29), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 46, idsite = 1, period = week(2020-01-27 - 2020-02-02), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 43, idsite = 1, period = day(2020-01-01 - 2020-01-01), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 44, idsite = 1, period = month(2020-01-01 - 2020-01-31), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 45, idsite = 1, period = year(2020-01-01 - 2020-12-31), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 35, idsite = 1, period = day(2019-12-31 - 2019-12-31), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 33, idsite = 1, period = day(2019-12-30 - 2019-12-30), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 34, idsite = 1, period = week(2019-12-30 - 2020-01-05), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 31, idsite = 1, period = day(2019-12-23 - 2019-12-23), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 32, idsite = 1, period = week(2019-12-23 - 2019-12-29), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 29, idsite = 1, period = day(2019-12-16 - 2019-12-16), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Found invalidated archive we can skip (no visits): [idinvalidation = 30, idsite = 1, period = week(2019-12-16 - 2019-12-22), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Processing invalidation: [idinvalidation = 5, idsite = 1, period = day(2019-12-12 - 2019-12-12), name = donee0512c03f7c20af6ef96a8d792c6bb9f].
-Processing invalidation: [idinvalidation = 14, idsite = 1, period = day(2019-12-11 - 2019-12-11), name = donee0512c03f7c20af6ef96a8d792c6bb9f].
-Processing invalidation: [idinvalidation = 17, idsite = 1, period = day(2019-12-10 - 2019-12-10), name = donee0512c03f7c20af6ef96a8d792c6bb9f].
+Processing invalidation: [idinvalidation = 5, idsite = 1, period = day(2019-12-12 - 2019-12-12), name = donee0512c03f7c20af6ef96a8d792c6bb9f, segment = actions>=2].
+Processing invalidation: [idinvalidation = 14, idsite = 1, period = day(2019-12-11 - 2019-12-11), name = donee0512c03f7c20af6ef96a8d792c6bb9f, segment = actions>=2].
+Processing invalidation: [idinvalidation = 17, idsite = 1, period = day(2019-12-10 - 2019-12-10), name = donee0512c03f7c20af6ef96a8d792c6bb9f, segment = actions>=2].
 Starting archiving for ?module=API&method=CoreAdminHome.archiveReports&idSite=1&period=day&date=2019-12-12&format=json&segment=actions%3E%3D2&trigger=archivephp
 Starting archiving for ?module=API&method=CoreAdminHome.archiveReports&idSite=1&period=day&date=2019-12-11&format=json&segment=actions%3E%3D2&trigger=archivephp
 Starting archiving for ?module=API&method=CoreAdminHome.archiveReports&idSite=1&period=day&date=2019-12-10&format=json&segment=actions%3E%3D2&trigger=archivephp
 Archived website id 1, period = day, date = 2019-12-12, segment = 'actions%3E%3D2', 0 visits found. Time elapsed: %fs
 Archived website id 1, period = day, date = 2019-12-11, segment = 'actions%3E%3D2', 0 visits found. Time elapsed: %fs
 Archived website id 1, period = day, date = 2019-12-10, segment = 'actions%3E%3D2', 0 visits found. Time elapsed: %fs
-Found invalidated archive we can skip (no visits): [idinvalidation = 28, idsite = 1, period = day(2019-12-09 - 2019-12-09), name = donee0512c03f7c20af6ef96a8d792c6bb9f]
-Processing invalidation: [idinvalidation = 6, idsite = 1, period = week(2019-12-09 - 2019-12-15), name = donee0512c03f7c20af6ef96a8d792c6bb9f].
-Processing invalidation: [idinvalidation = 22, idsite = 1, period = day(2019-12-02 - 2019-12-02), name = donee0512c03f7c20af6ef96a8d792c6bb9f].
+Processing invalidation: [idinvalidation = 6, idsite = 1, period = week(2019-12-09 - 2019-12-15), name = donee0512c03f7c20af6ef96a8d792c6bb9f, segment = actions>=2].
+Processing invalidation: [idinvalidation = 22, idsite = 1, period = day(2019-12-02 - 2019-12-02), name = donee0512c03f7c20af6ef96a8d792c6bb9f, segment = actions>=2].
 No next invalidated archive.
 Starting archiving for ?module=API&method=CoreAdminHome.archiveReports&idSite=1&period=week&date=2019-12-09&format=json&segment=actions%3E%3D2&trigger=archivephp
 Starting archiving for ?module=API&method=CoreAdminHome.archiveReports&idSite=1&period=day&date=2019-12-02&format=json&segment=actions%3E%3D2&trigger=archivephp
 Archived website id 1, period = week, date = 2019-12-09, segment = 'actions%3E%3D2', 0 visits found. Time elapsed: %fs
 Archived website id 1, period = day, date = 2019-12-02, segment = 'actions%3E%3D2', 0 visits found. Time elapsed: %fs
-Processing invalidation: [idinvalidation = 23, idsite = 1, period = week(2019-12-02 - 2019-12-08), name = donee0512c03f7c20af6ef96a8d792c6bb9f].
+Processing invalidation: [idinvalidation = 23, idsite = 1, period = week(2019-12-02 - 2019-12-08), name = donee0512c03f7c20af6ef96a8d792c6bb9f, segment = actions>=2].
 No next invalidated archive.
 Starting archiving for ?module=API&method=CoreAdminHome.archiveReports&idSite=1&period=week&date=2019-12-02&format=json&segment=actions%3E%3D2&trigger=archivephp
 Archived website id 1, period = week, date = 2019-12-02, segment = 'actions%3E%3D2', 0 visits found. Time elapsed: %fs
-Processing invalidation: [idinvalidation = 7, idsite = 1, period = month(2019-12-01 - 2019-12-31), name = donee0512c03f7c20af6ef96a8d792c6bb9f].
+Processing invalidation: [idinvalidation = 7, idsite = 1, period = month(2019-12-01 - 2019-12-31), name = donee0512c03f7c20af6ef96a8d792c6bb9f, segment = actions>=2].
 No next invalidated archive.
 Starting archiving for ?module=API&method=CoreAdminHome.archiveReports&idSite=1&period=month&date=2019-12-01&format=json&segment=actions%3E%3D2&trigger=archivephp
 Archived website id 1, period = month, date = 2019-12-01, segment = 'actions%3E%3D2', 0 visits found. Time elapsed: %fs
-Processing invalidation: [idinvalidation = 8, idsite = 1, period = year(2019-01-01 - 2019-12-31), name = donee0512c03f7c20af6ef96a8d792c6bb9f].
+Processing invalidation: [idinvalidation = 8, idsite = 1, period = year(2019-01-01 - 2019-12-31), name = donee0512c03f7c20af6ef96a8d792c6bb9f, segment = actions>=2].
 No next invalidated archive.
 Starting archiving for ?module=API&method=CoreAdminHome.archiveReports&idSite=1&period=year&date=2019-01-01&format=json&segment=actions%3E%3D2&trigger=archivephp
 Archived website id 1, period = year, date = 2019-01-01, segment = 'actions%3E%3D2', 0 visits found. Time elapsed: %fs
@@ -767,6 +750,7 @@ LOG;
         $output = array_filter($output, function ($l) { return strpos($l, 'Found archive with intersecting period') === false; });
         $output = array_filter($output, function ($l) { return strpos($l, 'Found duplicate invalidated archive') === false; });
         $output = array_filter($output, function ($l) { return strpos($l, 'No usable archive exists') === false; });
+        $output = array_filter($output, function ($l) { return strpos($l, 'Found invalidated archive we can skip (no visits)') === false; });
         $output = implode("\n", $output);
         return $output;
     }
diff --git a/tests/PHPUnit/Unit/CronArchive/QueueConsumerTest.php b/tests/PHPUnit/Unit/CronArchive/QueueConsumerTest.php
new file mode 100644
index 0000000000..f8353ae554
--- /dev/null
+++ b/tests/PHPUnit/Unit/CronArchive/QueueConsumerTest.php
@@ -0,0 +1,95 @@
+<?php
+/**
+ * Matomo - free/libre analytics platform
+ *
+ * @link https://matomo.org
+ * @license http://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later
+ *
+ */
+
+namespace PHPUnit\Unit\CronArchive;
+
+
+use PHPUnit\Framework\TestCase;
+use Piwik\CronArchive\QueueConsumer;
+use Piwik\Period\Factory;
+use Piwik\Piwik;
+
+class QueueConsumerTest extends TestCase
+{
+    /**
+     * @dataProvider getTestDataForHasIntersectingPeriod
+     */
+    public function test_hasIntersectingPeriod($archivesToProcess, $invalidatedArchive, $expected)
+    {
+        $periods = array_flip(Piwik::$idPeriods);
+        foreach ($archivesToProcess as &$archive) {
+            $periodLabel = $periods[$archive['period']];
+            $archive['periodObj'] = Factory::build($periodLabel, $archive['date1']);
+        }
+
+        $periodLabel = $periods[$invalidatedArchive['period']];
+        $invalidatedArchive['periodObj'] = Factory::build($periodLabel, $invalidatedArchive['date1']);
+
+        $actual = QueueConsumer::hasIntersectingPeriod($archivesToProcess, $invalidatedArchive);
+        $this->assertEquals($expected, $actual);
+    }
+
+    public function getTestDataForHasIntersectingPeriod()
+    {
+        return  [
+            // no intersecting periods
+            [
+                [
+                    ['period' => 1, 'date1' => '2020-03-04', 'date2' => '2020-03-04'],
+                    ['period' => 3, 'date1' => '2020-04-01', 'date2' => '2020-04-30'],
+                    ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==abc'],
+                ],
+                ['period' => 1, 'date1' => '2020-03-05', 'date2' => '2020-03-05'],
+                false,
+            ],
+
+            // intersecting periods
+            [
+                [
+                    ['period' => 1, 'date1' => '2020-03-04', 'date2' => '2020-03-04'],
+                    ['period' => 3, 'date1' => '2020-04-01', 'date2' => '2020-04-30'],
+                    ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==abc'],
+                ],
+                ['period' => 2, 'date1' => '2020-03-02', 'date2' => '2020-03-08'],
+                true,
+            ],
+
+            // all same period, different segments
+            [
+                [
+                    ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==def'],
+                    ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==ghi'],
+                    ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==abc'],
+                ],
+                ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==lmn'],
+                false,
+            ],
+
+            // all same period, all visits in one
+            [
+                [
+                    ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => ''],
+                ],
+                ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==lmn'],
+                true,
+            ],
+
+            // all same period, different segments, all visits in next
+            [
+                [
+                    ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==def'],
+                    ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==ghi'],
+                    ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15', 'segment' => 'pageUrl==abc'],
+                ],
+                ['period' => 1, 'date1' => '2020-03-15', 'date2' => '2020-03-15'],
+                true,
+            ],
+        ];
+    }
+}
\ No newline at end of file