Create JOURNAL_ARCHIVES_SITES

ms609 · Dec 27, 2024 · 8bc221a · 8bc221a
2 parents 5b8ded1 + cf93aea
commit 8bc221a
Show file tree

Hide file tree

Showing 4 changed files with 56 additions and 5 deletions.
diff --git a/Zotero.php b/Zotero.php
@@ -945,7 +945,7 @@ public static function process_zotero_response(string $zotero_response, Template
                     (stripos((string) $result->publicationTitle, ' edition') === false)) { // Do not add if "journal" includes "edition"
                     if (str_replace(NON_JOURNALS, '', (string) $result->publicationTitle) === (string) $result->publicationTitle) {
                         if (str_ireplace(NON_JOURNAL_WEBSITES, '', $url) === $url || $template->wikiname() === 'cite journal') {
-                            if (str_ireplace(CANONICAL_PUBLISHER_URLS, '', $url) === $url) {
+                            if (str_ireplace(CANONICAL_PUBLISHER_URLS, '', $url) === $url && str_ireplace(JOURNAL_ARCHIVES_SITES, '', $url) === $url) {
                                 bot_debug_log('Possible journal URL: ' . $url);
                                 $template->add_if_new('work', (string) $result->publicationTitle);
                             } else {

diff --git a/constants/bad_data.php b/constants/bad_data.php
@@ -643,6 +643,12 @@
     'doi.library.ubc.ca',
 ];
 
+const JOURNAL_ARCHIVES_SITES = [
+    'zaguan.unizar.es',
+    'hal.science',
+    'shs.cairn.info',
+];
+
 const PROXY_HOSTS_TO_ALWAYS_DROP = [
     'proxy.libraries',
     'proxy.lib.',

diff --git a/tests/phpunit/constantsTest.php b/tests/phpunit/constantsTest.php
@@ -422,6 +422,29 @@ public function testNonJournalListIsNotBad() {
                  echo "\n\n CANONICAL_PUBLISHER_URLS damaged by NON_JOURNAL_WEBSITES: " . $journal . ' ' . $check . "\n\n";
                  $this->flush();
             }
+            $check = $journal;
+            foreach (JOURNAL_ARCHIVES_SITES as $bad) {
+                $check = str_ireplace($bad, '', $check);
+            }
+            if ($check !== $journal) {
+                 $failed = true;
+                 $this->flush();
+                 echo "\n\n JOURNAL_ARCHIVES_SITES damaged by NON_JOURNAL_WEBSITES: " . $journal . ' ' . $check . "\n\n";
+                 $this->flush();
+            }
+        }
+        foreach (NON_JOURNAL_WEBSITES as $journal) {
+            $journal = $journal . '/';
+            $check = $journal;
+            foreach (JOURNAL_ARCHIVES_SITES as $bad) {
+                $check = str_ireplace($bad, '', $check);
+            }
+            if ($check !== $journal) {
+                 $failed = true;
+                 $this->flush();
+                 echo "\n\n NON_JOURNAL_WEBSITES damaged by JOURNAL_ARCHIVES_SITES: " . $journal . ' ' . $check . "\n\n";
+                 $this->flush();
+            }
         }
         $this->assertFalse($failed);
     }

diff --git a/tests/phpunit/zoteroTest.php b/tests/phpunit/zoteroTest.php
@@ -615,16 +615,38 @@ public function testZoteroResponse35(): void {
     }
 
     public function testZoteroResponse36(): void {
-        $text = '{{cite web|id=}}';
-        $template = $this->make_citation($text);
-        $access_date = 0;
-        $url = '';
         $zotero_data[0] = (object) ['title' => 'Billy', 'itemType' => 'journalArticle', 'publicationTitle' => "X"];
         $zotero_response = json_encode($zotero_data);
+        $access_date = 0;
+        $text = '{{cite web|id=}}';
+
+        $template = $this->make_citation($text);
+        $url = 'zaguan.unizar.es';
+        Zotero::process_zotero_response($zotero_response, $template, $url, $access_date);
+        $this->assertSame('cite journal', $template->wikiname());
+        $this->assertSame('Billy', $template->get2('title'));
+        $this->assertSame('X', $template->get2('journal'));
+
+        $template = $this->make_citation($text);
+        $url = 'bmj.com/cgi/pmidlookup';
         Zotero::process_zotero_response($zotero_response, $template, $url, $access_date);
         $this->assertSame('cite journal', $template->wikiname());
         $this->assertSame('Billy', $template->get2('title'));
         $this->assertSame('X', $template->get2('journal'));
+
+        $template = $this->make_citation($text);
+        $url = 'www.nsw.gov.au/sss';
+        Zotero::process_zotero_response($zotero_response, $template, $url, $access_date);
+        $this->assertSame('cite journal', $template->wikiname());
+        $this->assertSame('Billy', $template->get2('title'));
+        $this->assertSame('X', $template->get2('work'));
+
+        $template = $this->make_citation($text);
+        $url = 'NotOnAnyList';
+        Zotero::process_zotero_response($zotero_response, $template, $url, $access_date);
+        $this->assertSame('cite journal', $template->wikiname());
+        $this->assertSame('Billy', $template->get2('title'));
+        $this->assertSame('X', $template->get2('work'));
     }
 
     public function testZoteroResponse37(): void {