Define all scraper methods in prototype

Author: s0ph1e

lib/scraper.js

@@ -12,34 +12,55 @@ var loadHtml = require('./file-handlers/html');
 var loadCss = require('./file-handlers/css');
 var compareUrls = require('compare-urls');
 
-function getLoadedResource (resource) {
+function getHandleFunction (resource) {
+	var type = resource.getType();
+	switch (type) {
+		case types.css: return loadCss;
+		case types.html: return function loadHtmlAndCss (context, po) {
+			return loadHtml(context, po).then(function (loaded) {
+				return loadCss(context, loaded);
+			});
+		};
+		default: return _.noop;
+	}
+}
+
+function Scraper (options) {
+	this.originalResources = [];
+	this.loadedResources = [];
+
+	this.options = _.extend({}, defaults, options);
+	this.options.directory = path.resolve(process.cwd(), this.options.directory || '');
+}
+
+Scraper.prototype.getLoadedResource = function getLoadedResource (resource) {
 	return _.find(this.loadedResources, function(lr) {
 		return compareUrls(resource.getUrl(), lr.getUrl());
 	});
-}
+};
 
-function addLoadedResource (resource) {
+Scraper.prototype.addLoadedResource = function addLoadedResource (resource) {
 	this.loadedResources.push(resource);
-}
+};
 
-function getOccupiedFilenames () {
+Scraper.prototype.getOccupiedFilenames = function getOccupiedFilenames () {
 	var subdirectories = _.map(this.options.subdirectories, function (dir) { return dir.directory; });
 	var loadedFiles = _.map(this.loadedResources, function(r) { return r.getFilename(); });
 	return subdirectories.concat(loadedFiles);
-}
+};
 
-function getHtmlSources () {
+Scraper.prototype.getHtmlSources = function getHtmlSources () {
 	return this.options.sources;
-}
+};
 
-function generateFilename (resource) {
+Scraper.prototype.generateFilename = function generateFilename (resource) {
 	var self = this;
 
 	var occupiedFilenames = self.getOccupiedFilenames();
 
-	var preferedFilename = resource.getFilename();       // which was set in options
+	var preferredFilename = resource.getFilename();       // which was set in options
 	var urlFilename = utils.getFilenameFromUrl(resource.getUrl());  // try to get filename from url
-	var filename = preferedFilename || urlFilename || self.options.defaultFilename;
+	var filename = preferredFilename || urlFilename || self.options.defaultFilename;
 
 	var ext = path.extname(filename);
 	var dir = self.getDirectoryByExtension(ext);
@@ -52,30 +73,17 @@ function generateFilename (resource) {
 		index++;
 	}
 	return currentFilename;
-}
+};
 
-function getDirectoryByExtension (ext) {
+Scraper.prototype.getDirectoryByExtension = function getDirectoryByExtension (ext) {
 	return _.chain(this.options.subdirectories)
 		.filter(function (dir) { return _.contains(dir.extensions, ext); })
 		.map(function (dir) { return dir.directory; })
 		.first()
 		.value() || '';
-}
-
-function getHandleFunction (resource) {
-	var type = resource.getType();
-	switch (type) {
-		case types.css: return loadCss;
-		case types.html: return function loadHtmlAndCss (context, po) {
-			return loadHtml(context, po).then(function (loaded) {
-				return loadCss(context, loaded);
-			});
-		};
-		default: return _.noop;
-	}
-}
+};
 
-function loadResource (resource) {
+Scraper.prototype.loadResource = function loadResource (resource) {
 	var self = this;
 
 	var loaded = self.getLoadedResource(resource); 	// try to find already loaded
@@ -104,18 +112,17 @@ function loadResource (resource) {
 			return Promise.resolve(resource);
 		});
 	}
-
 	return Promise.resolve(loaded);
-}
+};
 
-function validate () {
+Scraper.prototype.validate = function validate () {
 	if (fs.existsSync(this.options.directory)) {
 		return Promise.reject(new Error('Path ' + this.options.directory + ' exists'));
 	}
 	return Promise.resolve();
-}
+};
 
-function prepare () {
+Scraper.prototype.prepare = function prepare () {
 	var self = this;
 	fs.ensureDirSync(self.options.directory);
 
@@ -130,9 +137,9 @@ function prepare () {
 		return new Resource(url, filename);
 	});
 	return Promise.resolve();
-}
+};
 
-function load () {
+Scraper.prototype.load = function load () {
 	var self = this;
 	return Promise.map(self.originalResources, function loadPage (po) {
 		return self.loadResource(po).then(function pageLoaded (loaded) {
@@ -142,39 +149,19 @@ function load () {
 			});
 		});
 	});
-}
+};
 
-function errorCleanup (error) {
+Scraper.prototype.errorCleanup = function errorCleanup (error) {
 	if (!_.isEmpty(this.loadedResources)) {
 		fs.removeSync(this.options.directory);
 	}
 	throw error;
-}
-
-function Scraper (options) {
-	this.originalResources = [];
-	this.loadedResources = [];
-
-	this.options = _.extend({}, defaults, options);
-	this.options.directory = path.resolve(process.cwd(), this.options.directory || '');
-
-	this.validate = validate.bind(this);
-	this.prepare = prepare.bind(this);
-	this.load = load.bind(this);
-	this.errorCleanup = errorCleanup.bind(this);
-
-	this.loadResource = loadResource.bind(this);
-	this.getLoadedResource = getLoadedResource.bind(this);
-	this.addLoadedResource = addLoadedResource.bind(this);
-	this.getOccupiedFilenames = getOccupiedFilenames.bind(this);
-	this.generateFilename = generateFilename.bind(this);
-	this.getDirectoryByExtension = getDirectoryByExtension.bind(this);
-	this.getHtmlSources = getHtmlSources.bind(this);
-}
+};
 
 Scraper.prototype.scrape = function scrape(callback) {
 	var self = this;
-	return self.validate()
+	return Promise.bind(self)
+		.then(self.validate)
 		.then(self.prepare)
 		.then(self.load)
 		.catch(self.errorCleanup)

test/unit/scraper-test.js

@@ -157,7 +157,7 @@ describe('Scraper', function () {
 			});
 
 			var loadResourceSpy = sinon.spy(s, 'loadResource');
-			s.prepare().then(s.load).then(function() {
+			s.prepare().bind(s).then(s.load).then(function() {
 				loadResourceSpy.calledTwice.should.be.eql(true);
 				done();
 			}).catch(done);
@@ -175,7 +175,7 @@ describe('Scraper', function () {
 				directory: testDirname
 			});
 
-			s.prepare().then(s.load).then(function(res) {
+			s.prepare().bind(s).then(s.load).then(function(res) {
 				res.should.be.instanceOf(Array);
 				res.should.have.length(2);
 				res[0].should.have.properties(['url', 'filename']);
@@ -487,7 +487,7 @@ describe('Scraper', function () {
 		});
 	});
 
-	describe('#load', function() {
+	describe('#loadResource', function() {
 		it('should load resource', function(done) {
 			nock('http://example.com').get('/a.png').reply(200, 'OK');