readability: infer lang from page, add fallback language option (#7)

common/options.js

@@ -4,14 +4,21 @@
 }) => {
 
 const model = {
+	defaultLanguage: {
+		title: `Default Book Language`,
+		description: `Default '<dc:language>' to set if none could be detected.`,
+		default: 'en',
+		restrict: { match: (/^[a-z]{2,10}(:?-[A-Z]{2,10})?$/), message: `Must be n ISO language code, e.g. 'en' or 'en-US'`, },
+		input: { type: 'boolean', },
+	},
 	setNavProperty: {
-		title: `Set 'nav' property`,
+		title: `Set 'nav' Property`,
 		description: `Standard compliant when set, but disables the navigation in Sumatra PDF.`,
 		default: false,
 		input: { type: 'boolean', },
 	},
 	collectStyles: {
-		title: `Keep style information`,
+		title: `Keep Style Information`,
 		description: `If enabled, some style information are extracted, but some readers won't be able to change the font settings when styles are set.
 		Only applies to <code>overdrive.com</code>, reader mode always removes all styles.`,
 		default: false,

content/collect.js

@@ -8,6 +8,7 @@
 	const collect = (await require.async('content/collect/'+ collector));
 	const contents = (await collect({ styles: options.collectStyles.value, ...params, }));
 	if (!contents) { return null; }
+	if (!contents.language) { contents.language = options.defaultLanguage.value; }
 
 	const book = new EPub(Object.assign(contents, { markNav: options.setNavProperty.value, }));
 	(await book.loadResources({ allowErrors: true, }));

content/collect/about-reader.js

@@ -38,7 +38,7 @@ return ({
 	}, ],
 	title,
 	description: `Offline reader version of ${url}`,
-	language: null,
+	language: (new URL(url).hostname.match(/[.](.{2})$/) || [ null, null, ])[1],
 	creator: [ { name: author, role: 'author', }, ],
 	resources,
 	cover: false,

content/collect/readability.js

@@ -31,6 +31,10 @@ document.querySelector('.reader-title').textContent = parsed.title || '';
 document.querySelector('.reader-credits').textContent = parsed.byline || '';
 document.querySelector('.container>.content').innerHTML = parsed.content; // this is not a live document, so this should be unproblematic: https://github.com/mozilla/readability/issues/404
 
-return aboutReader({ document, });
+const content = (await aboutReader({ document, }));
+return {
+	...content,
+	language: global.document.documentElement.lang || global.document.body.lang || content.language,
+};
 
 }); })(this);