adding a fallback for charset using http-equiv

jshemas · jshemas · commit 112932ec9400 · 2023-10-22T16:44:55.000-04:00
diff --git a/dist/lib/fallback.js b/dist/lib/fallback.js
@@ -198,6 +198,11 @@ function fallback(ogObject, options, $, body) {
     else if (doesElementExist('head > meta[name="charset"]', 'content', $)) {
         ogObject.charset = $('head > meta[name="charset"]').attr('content');
     }
+    else if (doesElementExist('head > meta[http-equiv="content-type"]', 'content', $)) {
+        const content = $('head > meta[http-equiv="content-type"]').attr('content');
+        const charsetRegEx = /charset=([^()<>@,;:"/[\]?.=\s]*)/i;
+        ogObject.charset = charsetRegEx.test(content) ? charsetRegEx.exec(content)[1] : 'UTF-8';
+    }
     else if (body) {
         ogObject.charset = chardet_1.default.detect(Buffer.from(body)) || '';
     }
diff --git a/lib/fallback.ts b/lib/fallback.ts
@@ -175,6 +175,10 @@ export function fallback(ogObject: OgObjectInteral, options: OpenGraphScraperOpt
     ogObject.charset = $('meta').attr('charset');
   } else if (doesElementExist('head > meta[name="charset"]', 'content', $)) {
     ogObject.charset = $('head > meta[name="charset"]').attr('content');
+  } else if (doesElementExist('head > meta[http-equiv="content-type"]', 'content', $)) {
+    const content = $('head > meta[http-equiv="content-type"]').attr('content');
+    const charsetRegEx = /charset=([^()<>@,;:"/[\]?.=\s]*)/i;
+    ogObject.charset = charsetRegEx.test(content) ? charsetRegEx.exec(content)[1] : 'UTF-8';
   } else if (body) {
     ogObject.charset = chardet.detect(Buffer.from(body)) || '';
   }
diff --git a/tests/unit/fallback.spec.ts b/tests/unit/fallback.spec.ts
@@ -425,6 +425,12 @@ describe('fallback', function () {
       expect(ogObject.charset).to.be.eql('bar');
       expect(ogObject).to.have.all.keys('charset');
     });
+    it('when there is a meta tag with http-equiv charset', function () {
+      const $ = load('<html><head><meta http-equiv="Content-Type" content="text/html; charset=foo_bar"></head></html>');
+      const ogObject = fallback({}, {}, $, '');
+      expect(ogObject.charset).to.be.eql('foo_bar');
+      expect(ogObject).to.have.all.keys('charset');
+    });
     it('when trying to get a charset from the body', function () {
       const body = '<html><head></head></html>';
       const $ = load(body);

Original file line number	Diff line number	Diff line change
`@@ -198,6 +198,11 @@ function fallback(ogObject, options, $, body) {`
`198`	`198`	`else if (doesElementExist('head > meta[name="charset"]', 'content', $)) {`
`199`	`199`	`ogObject.charset = $('head > meta[name="charset"]').attr('content');`
`200`	`200`	`}`
	`201`	`+ else if (doesElementExist('head > meta[http-equiv="content-type"]', 'content', $)) {`
	`202`	`+ const content = $('head > meta[http-equiv="content-type"]').attr('content');`
	`203`	`+ const charsetRegEx = /charset=([^()<>@,;:"/[\]?.=\s]*)/i;`
	`204`	`+ ogObject.charset = charsetRegEx.test(content) ? charsetRegEx.exec(content)[1] : 'UTF-8';`
	`205`	`+ }`
`201`	`206`	`else if (body) {`
`202`	`207`	`ogObject.charset = chardet_1.default.detect(Buffer.from(body)) \|\| '';`
`203`	`208`	`}`