🐛 correctly parse URLs in gdoc front matter

owid · Apr 22, 2024 · 74f9b24 · 74f9b24
1 parent aff9e76
commit 74f9b24
Showing 1 changed file with 9 additions and 1 deletion.
diff --git a/db/model/Gdoc/archieToEnriched.ts b/db/model/Gdoc/archieToEnriched.ts
@@ -26,7 +26,7 @@ import {
 import { convertHeadingTextToId } from "@ourworldindata/components"
 import { parseRawBlocksToEnrichedBlocks, parseRefs } from "./rawToEnriched.js"
 import urlSlug from "url-slug"
-import { parseAuthors, spansToSimpleString } from "./gdocUtils.js"
+import { extractUrl, parseAuthors, spansToSimpleString } from "./gdocUtils.js"
 import { htmlToSimpleTextBlock } from "./htmlToEnriched.js"
 import { RESEARCH_AND_WRITING_DEFAULT_HEADING } from "@ourworldindata/types"
 
@@ -305,6 +305,14 @@ export const archieToEnriched = (
         if (parsed[key] === "false") parsed[key] = false
     }
 
+    // Convert URL front-matter properties to URLs
+    for (const key of Object.keys(parsed)) {
+        const value = parsed[key]
+        if (typeof value === "string") {
+            parsed[key] = extractUrl(value)
+        }
+    }
+
     // Parse elements of the ArchieML into enrichedBlocks
     parsed.body = compact(parsed.body.map(parseRawBlocksToEnrichedBlocks))