Forget the Arxiv version for better compatibility

PubPeerFoundation · May 24, 2019 · 256f0d8 · 256f0d8
1 parent 262db5c
commit 256f0d8
Show file tree

Hide file tree

Showing 3 changed files with 66 additions and 26 deletions.
diff --git a/src/Resources/Extractors/Arxiv.php b/src/Resources/Extractors/Arxiv.php
@@ -40,6 +40,16 @@ public function extractIdentifiersData(): void
             'value' => '2331-8422',
             'type' => 'issn',
         ];
+
+        $namespace = $this->searchTree->getNamespaces(true);
+        foreach ($this->searchTree->children($namespace['arxiv']) as $key => $child) {
+            if ('doi' === $key) {
+                $this->resourceOutput['identifiers'][] = [
+                    'value' => (string) $child,
+                    'type' => 'doi',
+                ];
+            }
+        }
     }
 
     /**
@@ -87,8 +97,8 @@ public function extractTypesData(): void
      */
     protected function getIdentifier(): string
     {
-        $urlParts = explode('/', $this->searchTree->id[0]);
+        preg_match('/(\d{2}(0|1)[0-9]\.\d{4,5})(v|V)?(\d)?/', $this->searchTree->id[0], $matches);
 
-        return array_pop($urlParts);
+        return $matches[1];
     }
 }
diff --git a/tests/Integration/ArxivDataExtractorTest.php b/tests/Integration/ArxivDataExtractorTest.php
@@ -47,7 +47,7 @@ public function it_can_extract_identifiers_data_from_arxiv_api()
         // Assert
         $this->assertArraySubset([
             [
-                'value' => '1708.03486v2',
+                'value' => '1708.03486',
                 'type' => 'arxiv',
             ],
             [
@@ -82,4 +82,32 @@ public function it_can_extract_authors_from_arxiv_api()
 
         $this->assertArrayIsValid($extracted);
     }
+
+    /** @test */
+    public function it_will_link_a_preprint_to_its_published_version_if_provided()
+    {
+        // Arrange
+        $file = $this->loadXml('arXiv/linked-article');
+
+        // Act
+        $extracted = (new Arxiv($file, new Output()))->extract();
+
+        // Assert
+        $this->assertArraySubset([
+            [
+                'value' => '1406.0755',
+                'type' => 'arxiv',
+            ],
+            [
+                'value' => '2331-8422',
+                'type' => 'issn',
+            ],
+            [
+                'value' => '10.1007/JHEP07(2014)103',
+                'type' => 'doi',
+            ],
+        ], $extracted['identifiers']);
+
+        $this->assertArrayIsValid($extracted);
+    }
 }
diff --git a/tests/Integration/OutputTest.php b/tests/Integration/OutputTest.php
@@ -15,29 +15,31 @@ class OutputTest extends TestCase
     public function test_20_randomly_selected_identifiers()
     {
         $ids = [
-            '10.1038/s41591-018-0058-y',
-            '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',
-            '10.1016/s0195-6663(03)00120-x',
-            '29767272',
+            '24834831',
+            '1406.0755v3',
+//            '10.1038/s41591-018-0058-y',
+//            '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',
+//            '10.1016/s0195-6663(03)00120-x',
+//            '29767272',
         ];
 
         // Act