Handle hash in urls for html resources

s0ph1e · s0ph1e · commit 448514fe61f9 · 2016-02-21T13:39:26.000+02:00
diff --git a/lib/file-handlers/html.js b/lib/file-handlers/html.js
@@ -1,6 +1,7 @@
 var cheerio = require('cheerio');
 var Promise = require('bluebird');
 var utils = require('../utils');
+var types = require('../config/resource-types');
 
 function loadHtml (context, resource) {
 	var sources = context.getHtmlSources();
@@ -54,6 +55,12 @@ function loadResources (context, resource, source) {
 
 			return context.loadResource(htmlResource).then(function handleLoadedSource (loadedResource) {
 				var relativePath = utils.getRelativePath(filename, loadedResource.getFilename());
+				var hash = utils.getHashFromUrl(attr);
+
+				if (hash && loadedResource.getType() === types.html) {
+					relativePath = relativePath.concat(hash);
+				}
+
 				el.attr(source.attr, relativePath);
 				return Promise.resolve();
 			});
diff --git a/lib/utils.js b/lib/utils.js
@@ -30,6 +30,10 @@ function getFilenameFromUrl (u) {
 	return path.basename(url.parse(u).pathname);
 }
 
+function getHashFromUrl (u) {
+	return url.parse(u).hash || '';
+}
+
 function waitAllFulfilled(promises) {
 	return Promise.all(promises.map(function(promise) {
 		return promise.reflect();
@@ -42,5 +46,6 @@ module.exports = {
 	getUnixPath: getUnixPath,
 	getRelativePath: getRelativePath,
 	getFilenameFromUrl: getFilenameFromUrl,
+	getHashFromUrl: getHashFromUrl,
 	waitAllFulfilled: waitAllFulfilled
 };
diff --git a/test/unit/file-handlers/html-test.js b/test/unit/file-handlers/html-test.js
@@ -14,7 +14,8 @@ var defaultScraperOpts = {
 	sources: [
 		{ selector: 'img', attr: 'src' },
 		{ selector: 'link[rel="stylesheet"]', attr: 'href' },
-		{ selector: 'script', attr: 'src' }
+		{ selector: 'script', attr: 'src'},
+		{ selector: 'a', attr: 'href' }
 	]
 };
 var scraper;
@@ -213,5 +214,57 @@ describe('Html handler', function () {
 				done();
 			}).catch(done);
 		});
+
+		it('should keep hash in url for html resources', function (done) {
+			nock('http://example.com').get('/page1.html').reply(200, 'OK');
+
+			var resourceStub = new Resource('http://example.com/page1.html', 'local/page1.html');
+			sinon.stub(resourceStub, 'getType').returns('html');
+			sinon.stub(scraper, 'loadResource').returns(Promise.resolve(resourceStub));
+
+			var html = '\
+				<html> \
+				<body> \
+					<a href="http://example.com/page1.html#hash">link</a> \
+				</body> \
+				</html>\
+			';
+
+			var po = new Resource('http://example.com', 'index.html');
+			po.setText(html);
+
+			return loadHtml(scraper, po).then(function(){
+				var text = po.getText();
+				text.should.containEql('local/page1.html#hash');
+				done();
+			}).catch(done);
+		});
+
+		it('should remove hash from url for not-html resources', function (done) {
+			nock('http://example.com').get('/page1.html').reply(200, 'OK');
+
+			var resourceStub = new Resource('http://example.com/page1.html', 'local/page1.html');
+			sinon.stub(resourceStub, 'getType').returns('other');
+			sinon.stub(scraper, 'loadResource').returns(Promise.resolve(resourceStub));
+
+			var html = '\
+				<html> \
+				<body> \
+					<a href="http://example.com/page1.html#hash">link</a> \
+				</body> \
+				</html>\
+			';
+
+			var po = new Resource('http://example.com', 'index.html');
+			po.setText(html);
+
+			return loadHtml(scraper, po).then(function(){
+				var text = po.getText();
+				text.should.not.containEql('local/page1.html#hash');
+				text.should.containEql('local/page1.html');
+				done();
+			}).catch(done);
+		});
+
 	});
 });
diff --git a/test/unit/utils-test.js b/test/unit/utils-test.js
@@ -56,6 +56,21 @@ describe('Common utils', function () {
 		});
 	});
 
+	describe('#getHashFromUrl', function () {
+		it('should return hash from url', function () {
+			utils.getHashFromUrl('#').should.be.equal('#');
+			utils.getHashFromUrl('#hash').should.be.equal('#hash');
+			utils.getHashFromUrl('page.html#hash').should.be.equal('#hash');
+			utils.getHashFromUrl('http://example.com/page.html#hash').should.be.equal('#hash');
+		});
+
+		it('should return empty string if url doesn\'t contain hash', function () {
+			utils.getHashFromUrl('').should.be.equal('');
+			utils.getHashFromUrl('page.html?a=b').should.be.equal('');
+			utils.getHashFromUrl('http://example.com/page.html?a=b').should.be.equal('');
+		});
+	});
+
 	describe('#getRelativePath', function () {
 		it('should return relative path', function () {
 			utils.getRelativePath('css/1.css', 'img/1.png').should.be.equal('../img/1.png');
