From 29fa5ffc7563f746644c36979147342e8530e2b2 Mon Sep 17 00:00:00 2001
From: Xavier Roussel <xavier.roussel@pubpeer.org>
Date: Thu, 21 Jun 2018 17:48:25 +0200
Subject: [PATCH] Delete DataMerger and introduce Output objects + models

---
 src/ApiDataChecker.php                   |   5 +
 src/ApiDataFetcher.php                   |   5 +-
 src/ApiDataMerger.php                    |  31 ----
 src/Identifiers/Doi.php                  |   2 +-
 src/Models/Affiliations.php              |  30 ++++
 src/Models/Authors.php                   |  48 ++++++
 src/Models/Identifiers.php               |  30 ++++
 src/Models/Journal.php                   |  30 ++++
 src/Models/Model.php                     |  90 ++++++++++
 src/Models/Output.php                    |  51 ++++++
 src/Models/Publication.php               |  30 ++++
 src/Models/Tags.php                      |  30 ++++
 src/Models/Types.php                     |  30 ++++
 src/Models/Updates.php                   |  26 +++
 src/Resources/Extractors/Extractor.php   |  40 +++--
 src/Support/helpers.php                  |   4 +-
 tests/Integration/ApiDataFetcherTest.php | 195 +---------------------
 tests/Integration/ApiDataMergerTest.php  |  27 ---
 tests/Integration/OutputTest.php         | 204 +++++++++++++++++++++++
 19 files changed, 647 insertions(+), 261 deletions(-)
 delete mode 100644 src/ApiDataMerger.php
 create mode 100644 src/Models/Affiliations.php
 create mode 100644 src/Models/Authors.php
 create mode 100644 src/Models/Identifiers.php
 create mode 100644 src/Models/Journal.php
 create mode 100644 src/Models/Model.php
 create mode 100644 src/Models/Output.php
 create mode 100644 src/Models/Publication.php
 create mode 100644 src/Models/Tags.php
 create mode 100644 src/Models/Types.php
 create mode 100644 src/Models/Updates.php
 delete mode 100644 tests/Integration/ApiDataMergerTest.php
 create mode 100644 tests/Integration/OutputTest.php

diff --git a/src/ApiDataChecker.php b/src/ApiDataChecker.php
index ac693d5..07ac55a 100644
--- a/src/ApiDataChecker.php
+++ b/src/ApiDataChecker.php
@@ -118,4 +118,9 @@ public static function check($data)
 
         return $result;
     }
+
+    public static function getDataTypes()
+    {
+        return array_slice(array_keys(ApiDataChecker::SCHEMA['root']), 1);
+    }
 }
diff --git a/src/ApiDataFetcher.php b/src/ApiDataFetcher.php
index 24e8f20..32dc311 100644
--- a/src/ApiDataFetcher.php
+++ b/src/ApiDataFetcher.php
@@ -8,6 +8,7 @@
 use Psr\Http\Message\ResponseInterface;
 use GuzzleHttp\Exception\RequestException;
 use GrahamCampbell\GuzzleFactory\GuzzleFactory;
+use PubPeerFoundation\PublicationDataExtractor\Models\Output;
 use PubPeerFoundation\PublicationDataExtractor\Resources\Resource;
 use PubPeerFoundation\PublicationDataExtractor\Identifiers\Identifier;
 
@@ -106,9 +107,7 @@ public function getData(): array
 
         $this->fetchComplementaryData();
 
-        return array_values(
-            array_filter($this->apiData)
-        );
+        return Output::getInstance()->format();
     }
 
     /**
diff --git a/src/ApiDataMerger.php b/src/ApiDataMerger.php
deleted file mode 100644
index fc8d2e3..0000000
--- a/src/ApiDataMerger.php
+++ /dev/null
@@ -1,31 +0,0 @@
-<?php
-
-namespace PubPeerFoundation\PublicationDataExtractor;
-
-class ApiDataMerger
-{
-    /**
-     * @param  array $data
-     * @return array
-     */
-    public static function handle(array $data)
-    {
-        return self::mergeData($data);
-    }
-
-    /**
-     * @param $data
-     * @return array
-     */
-    protected static function mergeData($data)
-    {
-        $tmp = [];
-        foreach ($data as $resourceData) {
-            foreach ($resourceData as $key => $value) {
-                $tmp[$key][] = $value;
-            }
-        }
-
-        return $tmp;
-    }
-}
diff --git a/src/Identifiers/Doi.php b/src/Identifiers/Doi.php
index 3fee042..09de173 100644
--- a/src/Identifiers/Doi.php
+++ b/src/Identifiers/Doi.php
@@ -12,7 +12,7 @@ class Doi extends Identifier
     protected $resources = [
         \PubPeerFoundation\PublicationDataExtractor\Resources\Doi::class,
         \PubPeerFoundation\PublicationDataExtractor\Resources\Crossref::class,
-        \PubPeerFoundation\PublicationDataExtractor\Resources\PubmedWebsite::class,
+//        \PubPeerFoundation\PublicationDataExtractor\Resources\PubmedWebsite::class,
         \PubPeerFoundation\PublicationDataExtractor\Resources\IdConverter::class,
         \PubPeerFoundation\PublicationDataExtractor\Resources\EutilsEsearch::class,
     ];
diff --git a/src/Models/Affiliations.php b/src/Models/Affiliations.php
new file mode 100644
index 0000000..b572450
--- /dev/null
+++ b/src/Models/Affiliations.php
@@ -0,0 +1,30 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+class Affiliations extends Model
+{
+    /**
+     * Hold cherry picked list of affiliations.
+     *
+     * @var array
+     */
+    protected $list = [];
+
+    /**
+     * Add unknown affiliations to the current list.
+     *
+     * @param $affiliations
+     * @return array
+     */
+    public function add(array $affiliations): array
+    {
+        foreach ($affiliations as $affiliation) {
+            if (! in_array(strtolower($affiliation['name']), $this->knownIdentifierValues('name'))) {
+                $this->list[] = $affiliation;
+            }
+        }
+
+        return $this->list;
+    }
+}
diff --git a/src/Models/Authors.php b/src/Models/Authors.php
new file mode 100644
index 0000000..5b4b211
--- /dev/null
+++ b/src/Models/Authors.php
@@ -0,0 +1,48 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+class Authors extends Model
+{
+    /**
+     * Hold cherry picked list of authors.
+     *
+     * @var array
+     */
+    protected $list = [];
+
+    /**
+     * Add unknown authors to the current list.
+     *
+     * @param $authors
+     * @return array
+     */
+    public function add(array $authors): array
+    {
+        if (($count = count($authors)) !== ($listCount = count($this->list))) {
+            if ($count > $listCount) {
+                return $this->list = $authors;
+            }
+
+            return $this->list;
+        }
+
+        for ($i = 0; $i < $count; ++$i) {
+            $this->addUnknownAttributes($authors, $i);
+        }
+
+        return $this->list;
+    }
+
+    protected function addUnknownAttributes($authors, $i)
+    {
+        foreach ($authors[$i] as $key => $value) {
+            if (empty($value)) {
+                continue;
+            }
+            if (isset($this->list[$i][$key]) && ! empty($this->list[$i][$key])) {
+                $this->list[$i][$key] = $value;
+            }
+        }
+    }
+}
diff --git a/src/Models/Identifiers.php b/src/Models/Identifiers.php
new file mode 100644
index 0000000..cd7d95e
--- /dev/null
+++ b/src/Models/Identifiers.php
@@ -0,0 +1,30 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+class Identifiers extends Model
+{
+    /**
+     * Hold cherry picked list of identifiers.
+     *
+     * @var array
+     */
+    protected $list = [];
+
+    /**
+     * Add unknown identifiers to the current list.
+     *
+     * @param $identifiers
+     * @return array
+     */
+    public function add(array $identifiers): array
+    {
+        foreach ($identifiers as $identifier) {
+            if (! in_array(strtolower($identifier['value']), $this->knownIdentifierValues('value'))) {
+                $this->list[] = $identifier;
+            }
+        }
+
+        return $this->list;
+    }
+}
diff --git a/src/Models/Journal.php b/src/Models/Journal.php
new file mode 100644
index 0000000..e6c5e8f
--- /dev/null
+++ b/src/Models/Journal.php
@@ -0,0 +1,30 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+class Journal extends Model
+{
+    /**
+     * Hold the journal details.
+     *
+     * @var array
+     */
+    protected $list = [];
+
+    /**
+     * Add unknown attributes to the current details array.
+     *
+     * @param $journal
+     * @return array
+     */
+    public function add(array $journal): array
+    {
+        foreach ($journal as $key => $value) {
+            if ($this->shouldKeepAttribute($key, $value)) {
+                $this->list[$key] = $value;
+            }
+        }
+
+        return $this->list;
+    }
+}
diff --git a/src/Models/Model.php b/src/Models/Model.php
new file mode 100644
index 0000000..efeb9f0
--- /dev/null
+++ b/src/Models/Model.php
@@ -0,0 +1,90 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+use Exception;
+use Tightenco\Collect\Support\Arr;
+
+class Model
+{
+    /**
+     * This var should never be used, in favor of Model's internal var.
+     *
+     * @var array
+     */
+    protected $list = [];
+
+    /**
+     * List of statically created instances.
+     *
+     * @var array
+     */
+    private static $instances = [];
+
+    protected function __construct()
+    {
+    }
+
+    protected function __clone()
+    {
+    }
+
+    /**
+     * @throws Exception
+     */
+    public function __wakeup()
+    {
+        throw new Exception('Cannot unserialize singleton');
+    }
+
+    /**
+     * Return the current Model's instance.
+     *
+     * @return Model
+     */
+    public static function getInstance()
+    {
+        $cls = get_called_class();
+        if (! isset(self::$instances[$cls])) {
+            self::$instances[$cls] = new static();
+        }
+
+        return self::$instances[$cls];
+    }
+
+    /**
+     * Figure out which values are already known.
+     *
+     * @param $key
+     * @return array
+     */
+    protected function knownIdentifierValues($key): array
+    {
+        return array_map(function ($value) {
+            return strtolower($value);
+        }, Arr::pluck($this->list, $key));
+    }
+
+    /**
+     * Should the attribute be kept?
+     *
+     * @param $key
+     * @param $value
+     * @return bool
+     */
+    protected function shouldKeepAttribute($key, $value): bool
+    {
+        $function = is_array($value) ? 'count' : 'strlen';
+
+        if (! isset($this->list[$key])) {
+            return true;
+        }
+
+        return $function($this->list[$key]) > $function($value);
+    }
+
+    protected function reset()
+    {
+        $this->list = [];
+    }
+}
diff --git a/src/Models/Output.php b/src/Models/Output.php
new file mode 100644
index 0000000..2d1a30a
--- /dev/null
+++ b/src/Models/Output.php
@@ -0,0 +1,51 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+use PubPeerFoundation\PublicationDataExtractor\ApiDataChecker;
+
+class Output extends Model
+{
+    /**
+     * Global content array.
+     *
+     * @var array
+     */
+    protected $content = [];
+
+    /**
+     * Dynamically call the add method on the desired model.
+     *
+     * @param string $name
+     * @param array  $resourceData
+     */
+    public function __call($name, array $resourceData): void
+    {
+        $name = strtolower(substr($name, 3));
+
+        if (in_array($name, ApiDataChecker::getDataTypes())) {
+            $className = __NAMESPACE__.'\\'.ucfirst($name);
+            $this->content[$name] = $className::getInstance()->add($resourceData[0]);
+        }
+    }
+
+    /**
+     * Output the formatted content array.
+     *
+     * @return array
+     */
+    public function format()
+    {
+        $this->resetLists();
+
+        return $this->content;
+    }
+
+    protected function resetLists()
+    {
+        foreach (ApiDataChecker::getDataTypes() as $type) {
+            $className = __NAMESPACE__.'\\'.ucfirst($type);
+            $className::getInstance()->reset();
+        }
+    }
+}
diff --git a/src/Models/Publication.php b/src/Models/Publication.php
new file mode 100644
index 0000000..4737c0a
--- /dev/null
+++ b/src/Models/Publication.php
@@ -0,0 +1,30 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+class Publication extends Model
+{
+    /**
+     * Hold the publication details.
+     *
+     * @var array
+     */
+    protected $list = [];
+
+    /**
+     * Add unknown attributes to the current details array.
+     *
+     * @param $publication
+     * @return array
+     */
+    public function add(array $publication): array
+    {
+        foreach ($publication as $key => $value) {
+            if ($this->shouldKeepAttribute($key, $value)) {
+                $this->list[$key] = $value;
+            }
+        }
+
+        return $this->list;
+    }
+}
diff --git a/src/Models/Tags.php b/src/Models/Tags.php
new file mode 100644
index 0000000..9ad2d5e
--- /dev/null
+++ b/src/Models/Tags.php
@@ -0,0 +1,30 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+class Tags extends Model
+{
+    /**
+     * Hold cherry picked list of tags.
+     *
+     * @var array
+     */
+    protected $list = [];
+
+    /**
+     * Add unknown tags to the current list.
+     *
+     * @param $tags
+     * @return array
+     */
+    public function add(array $tags): array
+    {
+        foreach ($tags as $tag) {
+            if (! in_array(strtolower($tag['name']), $this->knownIdentifierValues('name'))) {
+                $this->list[] = $tag;
+            }
+        }
+
+        return $this->list;
+    }
+}
diff --git a/src/Models/Types.php b/src/Models/Types.php
new file mode 100644
index 0000000..2d9d02a
--- /dev/null
+++ b/src/Models/Types.php
@@ -0,0 +1,30 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+class Types extends Model
+{
+    /**
+     * Hold cherry picked list of types.
+     *
+     * @var array
+     */
+    protected $list = [];
+
+    /**
+     * Add unknown types to the current list.
+     *
+     * @param $types
+     * @return array
+     */
+    public function add(array $types): array
+    {
+        foreach ($types as $type) {
+            if (! in_array(strtolower($type['name']), $this->knownIdentifierValues('name'))) {
+                $this->list[] = $type;
+            }
+        }
+
+        return $this->list;
+    }
+}
diff --git a/src/Models/Updates.php b/src/Models/Updates.php
new file mode 100644
index 0000000..58a7dd3
--- /dev/null
+++ b/src/Models/Updates.php
@@ -0,0 +1,26 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Models;
+
+class Updates extends Model
+{
+    /**
+     * Hold cherry picked list of types.
+     *
+     * @var array
+     */
+    protected $list = [];
+
+    /**
+     * Add unknown updates to the current list.
+     *
+     * @param $updates
+     * @return array
+     */
+    public function add(array $updates): array
+    {
+        $this->list = $updates;
+
+        return $this->list;
+    }
+}
diff --git a/src/Resources/Extractors/Extractor.php b/src/Resources/Extractors/Extractor.php
index d0c623b..3214b08 100644
--- a/src/Resources/Extractors/Extractor.php
+++ b/src/Resources/Extractors/Extractor.php
@@ -2,6 +2,7 @@
 
 namespace PubPeerFoundation\PublicationDataExtractor\Resources\Extractors;
 
+use PubPeerFoundation\PublicationDataExtractor\Models\Output;
 use PubPeerFoundation\PublicationDataExtractor\ApiDataChecker;
 use PubPeerFoundation\PublicationDataExtractor\Exceptions\JournalTitleNotFoundException;
 
@@ -41,7 +42,7 @@ public function extract(): array
     {
         $this->getDataFromDocument();
 
-        foreach ($this->dataTypes() as $type) {
+        foreach (ApiDataChecker::getDataTypes() as $type) {
             $class = __NAMESPACE__.'\\Provides'.ucfirst($type).'Data';
 
             if ($this instanceof $class) {
@@ -53,22 +54,39 @@ public function extract(): array
                 }
             }
         }
+//        if (isset($this->output['identifiers'])) {
+//            Output::getInstance()->addIdentifiers($this->output['identifiers']);
+//        }
+//        if (isset($this->output['publication'])) {
+//            Output::getInstance()->addPublication($this->output['publication']);
+//        }
+//        if (isset($this->output['journal'])) {
+//            Output::getInstance()->addJournal($this->output['journal']);
+//        }
+//        if (isset($this->output['affiliations'])) {
+//            Output::getInstance()->addAffiliations($this->output['affiliations']);
+//        }
+//        if (isset($this->output['types'])) {
+//            Output::getInstance()->addTypes($this->output['types']);
+//        }
+//        if (isset($this->output['tags'])) {
+//            Output::getInstance()->addTags($this->output['tags']);
+//        }
+//        if (isset($this->output['authors'])) {
+//            Output::getInstance()->addAuthors($this->output['authors']);
+//        }
+        foreach (ApiDataChecker::getDataTypes() as $type) {
+            if (isset($this->output[$type])) {
+                $method = 'add'.ucfirst($type);
+                Output::getInstance()->$method($this->output[$type]);
+            }
+        }
 
         $this->addOutputSource();
 
         return $this->output;
     }
 
-    /**
-     * Get a list of data Types from Schema.
-     *
-     * @return array
-     */
-    protected function dataTypes(): array
-    {
-        return array_slice(array_keys(ApiDataChecker::SCHEMA['root']), 1);
-    }
-
     /**
      * Prepare each data document.
      *
diff --git a/src/Support/helpers.php b/src/Support/helpers.php
index b2d7110..c132311 100644
--- a/src/Support/helpers.php
+++ b/src/Support/helpers.php
@@ -102,7 +102,9 @@ function stringify($data)
         }
 
         if (is_array($data)) {
-            return stringify($data[0]);
+            $data = array_filter($data);
+
+            return stringify(array_shift($data));
         }
 
         return (string) trim($data);
diff --git a/tests/Integration/ApiDataFetcherTest.php b/tests/Integration/ApiDataFetcherTest.php
index dd852a7..5559642 100644
--- a/tests/Integration/ApiDataFetcherTest.php
+++ b/tests/Integration/ApiDataFetcherTest.php
@@ -2,7 +2,6 @@
 
 namespace PubPeerFoundation\PublicationDataExtractor\Test\Integration;
 
-use PubPeerFoundation\PublicationDataExtractor\ApiDataMerger;
 use PubPeerFoundation\PublicationDataExtractor\Test\TestCase;
 use PubPeerFoundation\PublicationDataExtractor\ApiDataFetcher;
 use PubPeerFoundation\PublicationDataExtractor\IdentifierResolver;
@@ -23,7 +22,7 @@ public function it_fetches_data_from_a_doi_identifier()
         $dataFetcher->fetch();
 
         // Assert
-        $this->assertCount(6, $dataFetcher->getData());
+        $this->assertArrayIsValid($dataFetcher->getData());
     }
 
     /**
@@ -40,8 +39,7 @@ public function it_fetches_data_from_a_pubmed_doi_identifier()
         $dataFetcher->fetch();
 
         // Assert
-        $this->assertCount(5, $extracted = $dataFetcher->getData());
-        $this->assertArrayIsValid($extracted);
+        $this->assertArrayIsValid($dataFetcher->getData());
     }
 
     /**
@@ -58,7 +56,7 @@ public function it_fetches_data_from_a_arxiv_identifier()
         $dataFetcher->fetch();
 
         // Assert
-        $this->assertCount(1, $dataFetcher->getData());
+        $this->assertArrayIsValid($dataFetcher->getData());
     }
 
     /**
@@ -75,7 +73,7 @@ public function it_fetches_data_from_a_pubmed_identifier()
         $dataFetcher->fetch();
 
         // Assert
-        $this->assertCount(1, $dataFetcher->getData());
+        $this->assertArrayIsValid($dataFetcher->getData());
     }
 
     /**
@@ -92,7 +90,7 @@ public function it_fetches_data_from_a_doi_book_identifier()
         $dataFetcher->fetch();
 
         // Assert
-        $this->assertCount(2, $dataFetcher->getData());
+        $this->assertArrayIsValid($dataFetcher->getData());
     }
 
     /**
@@ -125,187 +123,10 @@ public function it_can_extract_pubmed_ids_from_id_converter()
         $dataFetcher->fetch();
 
         // Assert
-        $this->assertCount(5, $extracted = $dataFetcher->getData());
+        $extracted = $dataFetcher->getData();
 
-        $merged = ApiDataMerger::handle($extracted);
-        $identifiers = array_merge(...$merged['identifiers']);
-        $this->assertTrue(count(array_filter($identifiers, function ($identifier) {
+        $this->assertCount(1, array_filter($extracted['identifiers'], function ($identifier) {
             return 'pubmed' === $identifier['type'];
-        })) > 0);
-    }
-
-    /**
-     * @test
-     * @group forever
-     */
-    public function test_first_150_entries_of_pubpeer_identifiers()
-    {
-        // Took 3.65 minutes to complete, average .6 seconds per fetch.
-
-        $ids = [
-            '10.1016/j.molcel.2017.01.013',
-            '28157506',
-            '10.1093/humrep/dex008',
-            '28166330',
-            '10.1111/apt.14044',
-            '28318052',
-            '10.1073/pnas.1523936113',
-            '10.1016/j.neuropharm.2017.04.043',
-            '10.1016/j.biomaterials.2015.07.055',
-            '26318819',
-            '10.1038/546033a',
-            '10.1038/leu.2011.62',
-            '21494253',
-            '10.1016/j.canlet.2013.12.005',
-            '10.1080/10408390802064347',
-            '10.1016/j.jss.2013.12.028',
-            '10.1093/embo-reports/kvf213',
-            '12393750',
-            '10.1371/journal.pone.0093386',
-            '10.1016/j.cell.2015.01.009',
-            '11624264',
-            '10.1152/ajpheart.00005.2017',
-            '28130332',
-            '10.1177/1078390316668478',
-            '10.3390/vision1010009',
-            '10.1016/j.cell.2009.11.030',
-            '19962179',
-            '10.1074/jbc.M115.693200',
-            '26634277',
-            '10.1530/REP-17-0063',
-            '28420801',
-            '10.1002/jum.14272',
-            '28586113',
-            '10.1007/s00424-017-2007-x',
-            '28597189',
-            '10.1101/gr.153551.112',
-            '23934932',
-            '10.1080/0142159X.2017.1324137',
-            '10.1016/j.meegid.2017.03.023',
-            '28342885',
-            '10.1002/hep.26505',
-            '10.1152/ajpheart.00281.2016',
-            '27986658',
-            '10.1016/j.gdata.2015.07.008',
-            '26697316',
-            '10.1007/s00299-016-1960-8',
-            '10.1128/MCB.25.8.3338-3347.2005',
-            '15798217',
-            '7958844',
-            '10.1074/jbc.M400881200',
-            '14970224',
-            '10.1002/hep.28927',
-            '28114741',
-            '10.1093/nar/gkw557',
-            '27325741',
-            '10.1038/ncomms7087',
-            '25608663',
-            '10.1104/pp.109.141911',
-            '10.1158/0008-5472.CAN-09-3114',
-            '20197467',
-            '10.12688/f1000research.11119.1',
-            '28529709',
-            '16796559',
-            '10.1167/17.5.16',
-            '10.1371/journal.pgen.1006810',
-            '8386381',
-            '10.1136/bmj.b4144',
-            '10.1083/jcb.201504117',
-            '10.1021/acs.nanolett.7b01464',
-            '10.1523/JNEUROSCI.0175-11.2012',
-            '10.1111/anae.13938',
-            '10.1016/j.ccell.2017.02.017',
-            '10.1073/pnas.1221733110',
-            '23812746',
-            '10.3389/fphar.2017.00275',
-            '28559847',
-            '10.1080/10408398.2016.1246414',
-            '27736161',
-            '10.1167/17.6.1',
-            '10.1136/bmj.b4330',
-            '10.1016/j.devcel.2015.01.012',
-            '10.1016/j.meegid.2017.03.023',
-            '10.1093/biosci/biw010',
-            '10.1016/j.cell.2015.07.047',
-            '26317470',
-            '10.1002/hep.28927',
-            '27831662',
-            '10.1167/17.6.2',
-            '10.1109/TPWRS.2015.2499700',
-            '10.1093/nar/gkw350',
-            '27131367',
-            '10.1126/science.aai7984',
-            '28360327',
-            '10.1093/nar/gkv644',
-            '26130711',
-            '10.1038/srep35422',
-            '10.1016/j.heares.2016.12.006',
-            '28011083',
-            '10.1517/13543784.17.11.1769',
-            '18922112',
-            '10.1016/j.neuron.2017.05.020',
-            '10.1128/MCB.00199-17',
-            '10.1016/j.msec.2017.03.207',
-            '10.1105/tpc.111.094904',
-            '22427335',
-            '10.1074/jbc.M610819200',
-            '17389604',
-            '10.1016/j.bbi.2013.05.005',
-            '10.1007/s11948-017-9915-1',
-            '10.1038/sj.onc.1204553',
-            '11439350',
-            '10.1056/NEJMsb1616595',
-            '28402238',
-            '10.1101/142554',
-            '10.1056/NEJMra1612008',
-            '28514605',
-            '14975242',
-            '10.1016/j.clinthera.2017.02.007',
-            '28291580',
-            '10.4049/jimmunol.0903617',
-            '12509522',
-            '10.1179/016164107X204693',
-            '17626733',
-            '10.1074/jbc.M117.794487',
-            '28512129',
-            '10.4103/0019-509X.175593',
-            '10.3758/s13414-016-1160-1',
-            '10.1096/fj.06-7947com',
-            '17442731',
-            '10.1038/ncomms15160',
-            '28513586',
-            '10.1016/j.amepre.2013.08.019',
-            '24355667',
-            '6191871',
-            '10.1111/mpp.12417',
-            '10.1126/scisignal.2002790',
-            '22692423',
-            '10.1016/j.bbapap.2011.11.003',
-            '22155276',
-            '10.1038/nmeth.4293',
-            '10.1038/nature22314',
-            '10.1523/JNEUROSCI.2086-05.2005',
-            '16107656',
-            '10.4049/jimmunol.1201651',
-            '23028059',
-            '10.1016/j.jesp.2010.01.007',
-            '10.4161/auto.19653',
-            '22576015',
-            '10.1002/1873-3468.12657',
-        ];
-
-        // Act
-        foreach ($ids as $id) {
-            $identifier = new IdentifierResolver($id);
-            $dataFetcher = new ApiDataFetcher($identifier->handle());
-
-            $dataFetcher->fetch();
-
-            $extracted = $dataFetcher->getData();
-
-            foreach ($extracted as $resource) {
-                $this->assertArrayIsValid($resource);
-            }
-        }
+        }));
     }
 }
diff --git a/tests/Integration/ApiDataMergerTest.php b/tests/Integration/ApiDataMergerTest.php
deleted file mode 100644
index 7ff4436..0000000
--- a/tests/Integration/ApiDataMergerTest.php
+++ /dev/null
@@ -1,27 +0,0 @@
-<?php
-
-namespace PubPeerFoundation\PublicationDataExtractor\Test\Integration;
-
-use PubPeerFoundation\PublicationDataExtractor\ApiDataMerger;
-use PubPeerFoundation\PublicationDataExtractor\Test\TestCase;
-
-class ApiDataMergerTest extends TestCase
-{
-    /** @test */
-    public function it_merges_data_from_fetcher()
-    {
-        // Arrange
-        $fetchedData = [
-            ['publication' => ['title' => 'blabla']],
-            ['publication' => ['title' => 'blibli']],
-        ];
-
-        // Act
-        $mergedData = ApiDataMerger::handle($fetchedData);
-
-        // Assert
-        $this->assertCount(1, $mergedData);
-        $this->assertArrayHasKey('publication', $mergedData);
-        $this->assertCount(2, $mergedData['publication']);
-    }
-}
diff --git a/tests/Integration/OutputTest.php b/tests/Integration/OutputTest.php
new file mode 100644
index 0000000..2603aa6
--- /dev/null
+++ b/tests/Integration/OutputTest.php
@@ -0,0 +1,204 @@
+<?php
+
+namespace PubPeerFoundation\PublicationDataExtractor\Test\Integration;
+
+use PubPeerFoundation\PublicationDataExtractor\Models\Output;
+use PubPeerFoundation\PublicationDataExtractor\Test\TestCase;
+use PubPeerFoundation\PublicationDataExtractor\ApiDataFetcher;
+use PubPeerFoundation\PublicationDataExtractor\IdentifierResolver;
+
+class OutputTest extends TestCase
+{
+    /**
+     * @test
+     * @group forever
+     */
+    public function test_first_150_entries_of_pubpeer_identifiers()
+    {
+        $ids = [
+            '10.1016/j.molcel.2017.01.013',
+            '28157506',
+            '10.1093/humrep/dex008',
+            '28166330',
+            '10.1111/apt.14044',
+            '28318052',
+            '10.1073/pnas.1523936113',
+            '10.1016/j.neuropharm.2017.04.043',
+            '10.1016/j.biomaterials.2015.07.055',
+            '26318819',
+            '10.1038/546033a',
+            '10.1038/leu.2011.62',
+            '21494253',
+            '10.1016/j.canlet.2013.12.005',
+            '10.1080/10408390802064347',
+            '10.1016/j.jss.2013.12.028',
+            '10.1093/embo-reports/kvf213',
+            '12393750',
+            '10.1371/journal.pone.0093386',
+            '10.1016/j.cell.2015.01.009',
+            '11624264',
+            '10.1152/ajpheart.00005.2017',
+            '28130332',
+            '10.1177/1078390316668478',
+            '10.3390/vision1010009',
+            '10.1016/j.cell.2009.11.030',
+            '19962179',
+            '10.1074/jbc.M115.693200',
+            '26634277',
+            '10.1530/REP-17-0063',
+            '28420801',
+            '10.1002/jum.14272',
+            '28586113',
+            '10.1007/s00424-017-2007-x',
+            '28597189',
+            '10.1101/gr.153551.112',
+            '23934932',
+            '10.1080/0142159X.2017.1324137',
+            '10.1016/j.meegid.2017.03.023',
+            '28342885',
+            '10.1002/hep.26505',
+            '10.1152/ajpheart.00281.2016',
+            '27986658',
+            '10.1016/j.gdata.2015.07.008',
+            '26697316',
+            '10.1007/s00299-016-1960-8',
+            '10.1128/MCB.25.8.3338-3347.2005',
+            '15798217',
+            '7958844',
+            '10.1074/jbc.M400881200',
+            '14970224',
+            '10.1002/hep.28927',
+            '28114741',
+            '10.1093/nar/gkw557',
+            '27325741',
+            '10.1038/ncomms7087',
+            '25608663',
+            '10.1104/pp.109.141911',
+            '10.1158/0008-5472.CAN-09-3114',
+            '20197467',
+            '10.12688/f1000research.11119.1',
+            '28529709',
+            '16796559',
+            '10.1167/17.5.16',
+            '10.1371/journal.pgen.1006810',
+            '8386381',
+            '10.1136/bmj.b4144',
+            '10.1083/jcb.201504117',
+            '10.1021/acs.nanolett.7b01464',
+            '10.1523/JNEUROSCI.0175-11.2012',
+            '10.1111/anae.13938',
+            '10.1016/j.ccell.2017.02.017',
+            '10.1073/pnas.1221733110',
+            '23812746',
+            '10.3389/fphar.2017.00275',
+            '28559847',
+            '10.1080/10408398.2016.1246414',
+            '27736161',
+            '10.1167/17.6.1',
+            '10.1136/bmj.b4330',
+            '10.1016/j.devcel.2015.01.012',
+            '10.1016/j.meegid.2017.03.023',
+            '10.1093/biosci/biw010',
+            '10.1016/j.cell.2015.07.047',
+            '26317470',
+            '10.1002/hep.28927',
+            '27831662',
+            '10.1167/17.6.2',
+            '10.1109/TPWRS.2015.2499700',
+            '10.1093/nar/gkw350',
+            '27131367',
+            '10.1126/science.aai7984',
+            '28360327',
+            '10.1093/nar/gkv644',
+            '26130711',
+            '10.1038/srep35422',
+            '10.1016/j.heares.2016.12.006',
+            '28011083',
+            '10.1517/13543784.17.11.1769',
+            '18922112',
+            '10.1016/j.neuron.2017.05.020',
+            '10.1128/MCB.00199-17',
+            '10.1016/j.msec.2017.03.207',
+            '10.1105/tpc.111.094904',
+            '22427335',
+            '10.1074/jbc.M610819200',
+            '17389604',
+            '10.1016/j.bbi.2013.05.005',
+            '10.1007/s11948-017-9915-1',
+            '10.1038/sj.onc.1204553',
+            '11439350',
+            '10.1056/NEJMsb1616595',
+            '28402238',
+            '10.1101/142554',
+            '10.1056/NEJMra1612008',
+            '28514605',
+            '14975242',
+            '10.1016/j.clinthera.2017.02.007',
+            '28291580',
+            '10.4049/jimmunol.0903617',
+            '12509522',
+            '10.1179/016164107X204693',
+            '17626733',
+            '10.1074/jbc.M117.794487',
+            '28512129',
+            '10.4103/0019-509X.175593',
+            '10.3758/s13414-016-1160-1',
+            '10.1096/fj.06-7947com',
+            '17442731',
+            '10.1038/ncomms15160',
+            '28513586',
+            '10.1016/j.amepre.2013.08.019',
+            '24355667',
+            '6191871',
+            '10.1111/mpp.12417',
+            '10.1126/scisignal.2002790',
+            '22692423',
+            '10.1016/j.bbapap.2011.11.003',
+            '22155276',
+            '10.1038/nmeth.4293',
+            '10.1038/nature22314',
+            '10.1523/JNEUROSCI.2086-05.2005',
+            '16107656',
+            '10.4049/jimmunol.1201651',
+            '23028059',
+            '10.1016/j.jesp.2010.01.007',
+            '10.4161/auto.19653',
+            '22576015',
+            '10.1002/1873-3468.12657',
+        ];
+        $ids = [
+            '10.1016/j.biomaterials.2013.04.055',
+            '23702148',
+            '21035459',
+            '10.1016/j.jmb.2010.10.030',
+            '28430426',
+            '10.1021/acs.jctc.7b00125',
+            '12610298',
+            '10.1126/science.1079731',
+            '10.1126/science.1236077',
+            '23744776',
+            '10.1038/ncomms12562',
+            '27558844',
+            '10.1021/acsnano.7b05503',
+            '29019651',
+            '10.1016/s0001-4575(99)00042-1',
+            '212431',
+            '10.1101/212431',
+            '10.1177/1090820x12472902',
+            '23335644',
+            '28776029',
+        ];
+
+        // Act
+        foreach ($ids as $id) {
+            $identifier = new IdentifierResolver($id);
+            $dataFetcher = new ApiDataFetcher($identifier->handle());
+
+            $dataFetcher->fetch();
+
+            $dataFetcher->getData();
+            $formatted = Output::getInstance()->format();
+            $this->assertArrayIsValid($formatted);
+        }
+    }
+}