Parse arXiv URLs

nasa-gcn · Sep 19, 2023 · 11c6a02 · 11c6a02
1 parent 57083a7
commit 11c6a02
Show file tree

Hide file tree

Showing 4 changed files with 36 additions and 1 deletion.
diff --git a/src/replacements/arXiv/index.ts b/src/replacements/arXiv/index.ts
@@ -1,7 +1,10 @@
 import type { FindAndReplace } from '../index.js'
 
+const marker = /(?:arXiv:|https?:\/\/(?:www\.)?arxiv\.org\/abs\/)/.source
+const articleId = /(?:\d{4}\.\d{5}|[a-z-]+\/\d{7})(?:v\d+)?/.source
+
 export default {
-  find: /arXiv:((?:\d{4}\.\d{5}|[a-z-]+\/\d{7})(?:v\d+)?)/g,
+  find: new RegExp(`${marker}(${articleId})`, 'g'),
   replace(data, text, value) {
     return data('arXiv', text, value)
   },

diff --git a/src/replacements/arXiv/test.html b/src/replacements/arXiv/test.html
@@ -1,3 +1,4 @@
 <p>Old style arXiv reference: <data class="arXiv" value="astro-ph/1234567">arXiv:astro-ph/1234567</data></p>
 <p>New style arXiv reference: <data class="arXiv" value="1234.56789">arXiv:1234.56789</data></p>
+<p>URL references: <data class="arXiv" value="1234.56789">http://www.arxiv.org/abs/1234.56789</data>, <data class="arXiv" value="1234.56789">https://arxiv.org/abs/1234.56789</data></p>
 <p>Versioned arXiv reference: <data class="arXiv" value="1234.56789v2">arXiv:1234.56789v2</data></p>
diff --git a/src/replacements/arXiv/test.json b/src/replacements/arXiv/test.json
@@ -35,6 +35,35 @@
         }
       ]
     },
+    {
+      "type": "paragraph",
+      "children": [
+        {
+          "type": "text",
+          "value": "URL references: "
+        },
+        {
+          "type": "text",
+          "value": "http://www.arxiv.org/abs/1234.56789",
+          "data": {
+            "class": "arXiv",
+            "value": "1234.56789"
+          }
+        },
+        {
+          "type": "text",
+          "value": ", "
+        },
+        {
+          "type": "text",
+          "value": "https://arxiv.org/abs/1234.56789",
+          "data": {
+            "class": "arXiv",
+            "value": "1234.56789"
+          }
+        }
+      ]
+    },
     {
       "type": "paragraph",
       "children": [

diff --git a/src/replacements/arXiv/test.md b/src/replacements/arXiv/test.md
@@ -2,4 +2,6 @@ Old style arXiv reference: arXiv:astro-ph/1234567
 
 New style arXiv reference: arXiv:1234.56789
 
+URL references: http://www.arxiv.org/abs/1234.56789, https://arxiv.org/abs/1234.56789
+
 Versioned arXiv reference: arXiv:1234.56789v2