Sitemap rewrite first commit (experimental)

cli.js

@@ -27,8 +27,9 @@ const commands = {
 	'clean': createCommand(runner.clean, ['dest']),
 	'clone-assets': createCommand(runner.clone_assets, ['baseurl', 'dest']),
 	'reseed': createCommand(runner.build, ['baseurl', 'dest']),
-	'rewrite_css': createCommand(runner.rewrite_css, ['baseurl', 'dest']),
+	'rewrite-css': createCommand(runner.rewrite_css, ['baseurl', 'dest']),
 	'rewrite-html': createCommand(runner.rewrite_html, ['baseurl', 'dest']),
+	'rewrite-sitemap': createCommand(runner.rewrite_sitemap, ['baseurl', 'dest']),
 	'serve': createCommand(runner.buildAndServe, ['baseurl', 'dest']),
 	'watch': createCommand(runner.watch, ['baseurl', 'dest'])
 	/* eslint-enable quote-props */
@@ -93,6 +94,7 @@ module.exports = {
 		const source = flags.source || defaultSrc;
 		const destination = flags.dest;
 		const baseurl = flags.baseurl || '';
+		const sitemap = flags.sitemap || 'sitemap.xml';
 		const port = this.checkPortNumber(flags.port) || defaultPort;
 		const split = flags.split || 1;
 		const partition = flags.partition || 1;
@@ -104,7 +106,8 @@ module.exports = {
 			paths: {
 				src: source,
 				dest: destination,
-				baseurl: baseurl
+				baseurl: baseurl,
+				sitemap: sitemap
 			},
 			serve: {
 				port: port,

index.js

@@ -11,6 +11,7 @@ Commands:
     clone-assets      Clones non CSS and HTML files from src to dest
     rewrite-css       Clones CSS files from src to dest and rewrites urls to include baseurl
     rewrite-html      Clones HTML files from src to dest and rewrites attributes to include baseurl
+	rewrite-sitemap   Finds all sitemap files based on index sitemap, and rewrites links to include baseurl
     serve             Runs 'build' then a local webserver on the dest folder
     watch             Watches the src folder and triggers builds
 
@@ -23,6 +24,7 @@ Options:
     -b | --baseurl    The base-URL to prepend to the files once copied
     -p | --port       The portnumber to serve the cloned site on
     -e | --extrasrc   A list of extra src attributes to be rewritten
+	-m | --sitemap    A path to a valid sitemap file
     -o | --overwrite  When cleaning --dest, don't prompt for confirmation
          --split      The number of partitions to divide files into
          --partition  The partition number to process
@@ -57,6 +59,10 @@ const inputs = meow(
 				alias: 'e',
 				isMultiple: true
 			},
+			sitemap: {
+				type: 'string',
+				alias: 'm'
+			},
 			overwrite: {
 				type: 'boolean',
 				alias: 'o'

lib/processors/sitemap.js

@@ -0,0 +1,76 @@
+const fs = require('fs-extra');
+const cheerio = require('cheerio');
+const log = require('fancy-log');
+const { URL } = require('url');
+
+let extraPaths = [];
+
+function rewritePath(baseurl, link) {
+	try {
+		const url = new URL(link);
+		const newPath = `${baseurl}/${url.pathname}`.replace('//', '/');
+		return `${url.origin}/${newPath}`;
+	} catch (urlError) {
+		return link;
+	}
+}
+
+function rewriteXML(xml, baseurl) {
+	const $ = cheerio.load(xml, {
+		xmlMode: true
+	});
+
+	const rootNode = $.root().children()[0];
+	const isIndex = rootNode && rootNode.name === 'sitemapindex';
+
+	$('loc').each(function () {
+		const $el = $(this);
+		const originalValue = $el.text();
+		const updated = rewritePath(baseurl, originalValue);
+		$el.text(updated);
+
+		if (isIndex && $el.parent()[0].name === 'sitemap') {
+			extraPaths.push(originalValue);
+		}
+	});
+
+	$('xhtml\\:link').each(function () {
+		const $el = $(this);
+		const originalValue = $el.attr('href');
+		const updated = rewritePath(baseurl, originalValue);
+		$el.attr('href', updated);
+	});
+	return $.xml();
+}
+
+module.exports = {
+	rewrite: rewriteXML,
+
+	/** Handles rewriting urls in sitemap(s)
+	 *
+	 * @param {string} file the absolute path to the sitemap file.
+	 * @param {string} destination the absolute path to the destination directory.
+	 * @param {string} baseurl the baseurl to prepend to the source files.
+	 */
+	plugin: function (file, destination, baseurl) {
+		extraPaths = [];
+		if (!file) {
+			log.error('Error rewriting XML: Invalid file specified.');
+			return 1;
+		}
+		if (!destination || !baseurl) {
+			log.error('Error rewriting XML: No destination specified.');
+			return 1;
+		}
+		const contents = fs.readFileSync(file);
+		const xml = contents.toString('utf-8');
+
+		if (!xml) {
+			return 0;
+		}
+		const rewritten = rewriteXML(xml, baseurl);
+
+		fs.writeFileSync(file, rewritten);
+		return extraPaths.length ? extraPaths : 0;
+	}
+};