sitelinks_helpers: fix support for multi-part language codes

maxlath · Aug 24, 2022 · ccb2bae · ccb2bae
1 parent f9e13c7
commit ccb2bae
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 1 deletion.
diff --git a/lib/helpers/sitelinks.js b/lib/helpers/sitelinks.js
@@ -53,6 +53,8 @@ const getSitelinkData = site => {
       lang = 'en'
       project = key = 'commons'
     } else {
+      // Support multi-parts language codes, such as be_x_old
+      lang = lang.replace(/-/g, '_')
       key = `${lang}${project}`.replace('wikipedia', 'wiki')
     }
     return { lang, project, key, title, url }
@@ -61,7 +63,7 @@ const getSitelinkData = site => {
     const specialProjectName = specialSites[key]
     if (specialProjectName) return { lang: 'en', project: specialProjectName, key }
 
-    const [ lang, projectSuffix, rest ] = key.split('wik')
+    let [ lang, projectSuffix, rest ] = key.split('wik')
 
     // Detecting cases like 'frwikiwiki' that would return [ 'fr', 'i', 'i' ]
     if (rest != null) throw new Error(`invalid sitelink key: ${key}`)
@@ -70,6 +72,9 @@ const getSitelinkData = site => {
       throw new Error(`sitelink lang not found: ${lang}. Updating wikibase-sdk to a more recent version might fix the issue.`)
     }
 
+    // Support keys such as be_x_oldwiki, which refers to be-x-old.wikipedia.org
+    lang = lang.replace(/_/g, '-')
+
     const project = projectsBySuffix[projectSuffix]
     if (!project) throw new Error(`sitelink project not found: ${project}`)
 

diff --git a/test/sitelinks_helpers.js b/test/sitelinks_helpers.js
@@ -90,6 +90,13 @@ describe('getSitelinkUrl', () => {
     getSitelinkUrl.bind(null, 'frperlinpinpin', 'Lyon').should.throw()
     getSitelinkUrl.bind(null, 'frwikiwiki', 'Lyon').should.throw()
   })
+
+  it('should support multi-part language codes', () => {
+    getSitelinkUrl({ site: 'zh_classicalwiki', title: '編訂名詞館' })
+    .should.startWith('https://zh-classical.wikipedia.org/')
+    getSitelinkUrl({ site: 'be_x_oldwiki', title: 'Віктор_Юго' })
+    .should.startWith('https://be-x-old.wikipedia.org/')
+  })
 })
 
 describe('getSitelinkData', () => {
@@ -132,11 +139,20 @@ describe('getSitelinkData', () => {
     getSitelinkData('https://de.wikipedia.org/wiki/The_Score_%282001%29').project.should.equal('wikipedia')
     getSitelinkData('https://de.wikipedia.org/wiki/The_Score_%282001%29').key.should.equal('dewiki')
   })
+
+  it('should support multi-part language codes', () => {
+    const data = getSitelinkData('https://be-x-old.wikipedia.org/wiki/Беларускі_клясычны_правапіс')
+    data.title.should.equal('Беларускі_клясычны_правапіс')
+    data.lang.should.equal('be_x_old')
+    data.project.should.equal('wikipedia')
+    data.key.should.equal('be_x_oldwiki')
+  })
 })
 
 describe('isSitelinkKey', () => {
   it('should return true for valid sitelink keys', () => {
     isSitelinkKey('frwiki').should.be.true()
+    isSitelinkKey('be_x_oldwiki').should.be.true()
     isSitelinkKey('commonswiki').should.be.true()
     isSitelinkKey('wikidatawiki').should.be.true()
     isSitelinkKey('commons').should.be.false()
@@ -148,5 +164,6 @@ describe('isSitelinkKey', () => {
     isSitelinkKey('frwikilinpinpin').should.be.false()
     isSitelinkKey('imaginarylangwiki').should.be.false()
     isSitelinkKey('frwikiwiki').should.be.false()
+    isSitelinkKey('be-x-oldwiki').should.be.false()
   })
 })