fix(sirv): append charset=utf-8 to HTML content-type (#122)

* fix: append charset=utf-8 to text/html

Closes #106

I considered fixing this upstream in "mime" or "mime-db" packages,
but then I came across these issues, which indicated that it was
not a viable path.

broofa/mime#174
jshttp/mime-db#94

* test: check for charset=utf-8 in .html requests

* fix(test): prevent undefined `full` access

* chore: add `charset=utf-8` to test expectants

* chore: light code style/spacing

Co-authored-by: Luke Edwards <luke.edwards05@gmail.com>

packages/sirv/index.js

@@ -88,20 +88,26 @@ function send(req, res, file, stats, headers) {
 	fs.createReadStream(file, opts).pipe(res);
 }
 
-function isEncoding(name, type, headers) {
-	headers['Content-Encoding'] = type;
-	headers['Content-Type'] = mime.getType(name.replace(/\.([^.]*)$/, '')) || '';
-}
+const ENCODING = {
+	'.br': 'br',
+	'.gz': 'gzip',
+};
 
 function toHeaders(name, stats, isEtag) {
+	let enc = ENCODING[name.slice(-3)];
+
+	let ctype = mime.getType(name.slice(0, enc && -3)) || '';
+	if (ctype === 'text/html') ctype += ';charset=utf-8';
+
 	let headers = {
 		'Content-Length': stats.size,
-		'Content-Type': mime.getType(name) || '',
+		'Content-Type': ctype,
 		'Last-Modified': stats.mtime.toUTCString(),
 	};
+
+	if (enc) headers['Content-Encoding'] = enc;
 	if (isEtag) headers['ETag'] = `W/"${stats.size}-${stats.mtime.getTime()}"`;
-	if (/\.br$/.test(name)) isEncoding(name, 'br', headers);
-	if (/\.gz$/.test(name)) isEncoding(name, 'gzip', headers);
+
 	return headers;
 }
 

tests/helpers.js

@@ -76,10 +76,14 @@ export async function lookup(filepath, enc) {
 	let full = join(www, filepath);
 	let stats = await statfile(full);
 	filedata = await readfile(full, enc);
+
+	let ctype = mime.getType(full) || '';
+	if (ctype === 'text/html') ctype += ';charset=utf-8';
+
 	return CACHE[filepath] = {
 		data: filedata,
 		size: stats.size,
-		type: mime.getType(full) || '',
+		type: ctype,
 		mtime: stats.mtime.getTime(),
 	};
 }

tests/sirv.js

@@ -733,7 +733,7 @@ brotli('should serve prepared `.br` file of any asset, if found', async () => {
 
 	try {
 		let res1 = await server.send('GET', '/', { headers });
-		assert.is(res1.headers['content-type'], 'text/html');
+		assert.is(res1.headers['content-type'], 'text/html;charset=utf-8');
 		assert.is(res1.headers['content-encoding'], 'br');
 		assert.is(res1.data, 'brotli html\n');
 		assert.is(res1.statusCode, 200);
@@ -751,7 +751,7 @@ brotli('should be preferred when "Accept-Encoding" allows both', async () => {
 
 	try {
 		let res1 = await server.send('GET', '/', { headers });
-		assert.is(res1.headers['content-type'], 'text/html');
+		assert.is(res1.headers['content-type'], 'text/html;charset=utf-8');
 		assert.is(res1.headers['content-encoding'], 'br');
 		assert.is(res1.data, 'brotli html\n');
 		assert.is(res1.statusCode, 200);
@@ -798,7 +798,7 @@ gzip('should serve prepared `.gz` file of any asset, if found', async () => {
 
 	try {
 		let res1 = await server.send('GET', '/', { headers });
-		assert.is(res1.headers['content-type'], 'text/html');
+		assert.is(res1.headers['content-type'], 'text/html;charset=utf-8');
 		assert.is(res1.headers['content-encoding'], 'gzip');
 		assert.is(res1.data, 'gzip html\n');
 		assert.is(res1.statusCode, 200);
@@ -816,7 +816,7 @@ gzip('should defer to brotli when "Accept-Encoding" allows both', async () => {
 
 	try {
 		let res1 = await server.send('GET', '/', { headers });
-		assert.is(res1.headers['content-type'], 'text/html');
+		assert.is(res1.headers['content-type'], 'text/html;charset=utf-8');
 		assert.is(res1.headers['content-encoding'], 'br');
 		assert.is(res1.data, 'brotli html\n');
 		assert.is(res1.statusCode, 200);