Use mime types

lib/config/resource-ext-by-type.js

@@ -0,0 +1,8 @@
+var types = require('./resource-types');
+var defaultExtensions = {};
+
+// should contain same data as ./resource-type-by-ext
+defaultExtensions[types.html] = [ '.html', '.htm' ];
+defaultExtensions[types.css] = [ '.css' ];
+
+module.exports = defaultExtensions;

lib/config/resource-type-by-ext.js

@@ -0,0 +1,8 @@
+var types = require('./resource-types');
+
+// should contain same data as ./resource-ext-by-type
+module.exports = {
+	'.html': types.html,
+	'.htm': types.html,
+	'.css': types.css
+};

lib/config/resource-type-by-mime.js

@@ -0,0 +1,6 @@
+var types = require('./resource-types');
+
+module.exports = {
+	'text/html': types.html,
+	'text/css': types.css
+};

lib/config/resource-types.js

@@ -1,7 +1,6 @@
 var types = {
 	css: 'css',
-	html: 'html',
-	other: 'other'
+	html: 'html'
 };
 
 module.exports = types;

lib/filename-generator/by-site-structure.js

@@ -2,7 +2,7 @@ var _ = require('lodash');
 var path = require('path');
 var utils = require('../utils');
 var resourceTypes = require('../config/resource-types');
-var resourceTypeExtensions = require('../config/resource-extensions-by-type');
+var resourceTypeExtensions = require('../config/resource-ext-by-type');
 
 module.exports = function generateFilename (resource, options) {
 	var resourceUrl = resource.getUrl();
@@ -11,7 +11,7 @@ module.exports = function generateFilename (resource, options) {
 
 	// If we have HTML from 'http://example.com/path' => set 'path/index.html' as filepath
 	if (resource.isHtml()) {
-		var htmlExtensions = resourceTypeExtensions[resourceTypes.html].possibleExtensions;
+		var htmlExtensions = resourceTypeExtensions[resourceTypes.html];
 		var resourceHasHtmlExtension = _.includes(htmlExtensions, extension);
 		// add index.html only if filepath has ext != html '/path/test.com' => '/path/test.com/index.html'
 		if (!resourceHasHtmlExtension) {

lib/filename-generator/by-type.js

@@ -1,7 +1,7 @@
 var _ = require('lodash');
 var path = require('path');
 var utils = require('../utils.js');
-var typeExtensions = require('../config/resource-extensions-by-type');
+var typeExtensions = require('../config/resource-ext-by-type');
 
 module.exports = function generateFilename (resource, options, occupiedFileNames) {
 	var occupiedNames = getSubDirectoryNames(options).concat(occupiedFileNames);
@@ -33,7 +33,7 @@ function getFilenameForResource (resource, options) {
 	var extension = utils.getFilenameExtension(filename);
 
 	if (!extension && typeExtensions[resourceType]) {
-		extension = typeExtensions[resourceType].defaultExtension;
+		extension = typeExtensions[resourceType][0];
 		filename += extension;
 	}
 

lib/request.js

@@ -3,13 +3,18 @@ var Promise = require('bluebird');
 var request = require('request');
 var get = Promise.promisify(request.get);
 
+function getMimeType (contentType) {
+	return contentType ? contentType.split(';')[0] : null;
+}
+
 function makeRequest (options, url) {
 	var requestOptions = _.clone(options);
 	requestOptions.url = url;
 
 	return get(requestOptions).then(function handleResponse (data) {
 		return {
 			url: data.request.href,
+			mimeType: getMimeType(data.headers['content-type']),
 			body: data.body
 		};
 	});

lib/resource-handler/html/index.js

@@ -19,15 +19,15 @@ HtmlResourceHandler.prototype.handle = function handle (resource) {
 		});
 };
 
-HtmlResourceHandler.prototype.loadResourcesForRule = function loadResourcesForRule ($, resource, rule) {
+HtmlResourceHandler.prototype.loadResourcesForRule = function loadResourcesForRule ($, parentResource, rule) {
 	var self = this;
 	var promises = $(rule.selector).map(function loadForElement () {
 		var el = new HtmlSourceElement($(this), rule);
 		var pathContainer = el.getPathContainer();
 		if (!pathContainer) {
 			return Promise.resolve();
 		}
-		return self.handleChildrenPaths(pathContainer, resource, createHtmlData(el)).then(el.setData.bind(el));
+		return self.handleChildrenPaths(pathContainer, parentResource).then(el.setData.bind(el));
 	}).get();
 
 	return utils.waitAllFulfilled(promises);
@@ -51,16 +51,4 @@ function loadTextToCheerio (text) {
 	});
 }
 
-/**
- * @param {HtmlSourceElement} htmlSourceEl
- * @returns {HtmlData}
- */
-function createHtmlData (htmlSourceEl) {
-	return {
-		tagName: htmlSourceEl.el[0].name,
-		attributeName: htmlSourceEl.rule.attr,
-		attributeValue: htmlSourceEl.el.attr(htmlSourceEl.rule.attr)
-	};
-}
-
 module.exports = HtmlResourceHandler;

lib/resource-handler/index.js

@@ -33,22 +33,19 @@ ResourceHandler.prototype.getResourceHandler = function getResourceHandler (reso
 
 /**
  * Request all resources from pathContainers paths
- * @param pathContainer - instance of imgSrcsetTag or CommonTag or CssText, contains original paths for resources
+ * @param pathContainer - instance of ImgSrcsetTag or CommonTag or CssText, contains original paths for resources
  * @param {Resource} parentResource
- * @param {HtmlData} [childResourceHtmlData]
  * @returns {Promise} - resolved when all resources from pathContainer were requested
  * and original paths in parentResource were updated with local paths for children resources
  */
-ResourceHandler.prototype.handleChildrenResources = function handleChildrenResources (pathContainer, parentResource, childResourceHtmlData) {
+ResourceHandler.prototype.handleChildrenResources = function handleChildrenResources (pathContainer, parentResource) {
 	var self = this;
 	var childrenPaths = pathContainer.getPaths();
 	var pathsToUpdate = [];
 
 	var childrenPromises = childrenPaths.map(function loadChildPath (childPath) {
 		var childResourceUrl = utils.getUrl(parentResource.getUrl(), childPath);
-
 		var childResource = parentResource.createChild(childResourceUrl);
-		childResource.setHtmlData(childResourceHtmlData);
 
 		return self.context.requestResource(childResource).then(function updateChildPath (respondedResource) {
 			if (respondedResource) {

lib/resource.js

@@ -1,36 +1,30 @@
-var _ = require('lodash');
-var utils = require('./utils');
 var types = require('./config/resource-types');
-var typesByHtmlData = require('./config/resource-types-by-tag');
-
-function getTypeByHtmlData (htmlData) {
-	var type = _.findKey(typesByHtmlData, function containsHtmlData (rules) {
-		return _.find(rules, htmlData);
-	});
-	return type || types.other;
-}
 
 function Resource (url, filename) {
 	this.url = url;
 	this.filename = filename;
+
 	this.assets = [];
+	this.type = null;
+	this.depth = 0;
+	this.parent = null;
+	this.saved = false;
 }
 
 Resource.prototype.createChild = function createChild (url, filename) {
 	var child = new Resource(url, filename);
-
 	var currentDepth = this.getDepth();
 
-	child.setParent(this);
-	child.setDepth(++currentDepth);
+	child.parent = this;
+	child.depth = ++currentDepth;
 
 	this.assets.push(child);
 
 	return child;
 };
 
 Resource.prototype.updateChild = function updateChild (oldChild, newChild) {
-	var index = _.indexOf(this.assets, oldChild);
+	var index = this.assets.indexOf(oldChild);
 	if (index >= 0) {
 		this.assets[index] = newChild;
 	}
@@ -60,58 +54,16 @@ Resource.prototype.setText = function setText (text) {
 	this.text = text;
 };
 
-Resource.prototype.setParent = function setParent (parent) {
-	this.parent = parent;
-};
-
 Resource.prototype.getDepth = function getDepth () {
-	return this.depth || 0;
+	return this.depth;
 };
 
-Resource.prototype.setDepth = function setDepth (depth) {
-	this.depth = depth;
-};
-
-/**
- * Html Data for resource, represents html element where resource was found
- *
- * @typedef {Object} HtmlData
- * @property {string} tagName - tag of element
- * @property {string} attributeName - attribute in tag where resource was found
- * @property {string} attributeValue - attribute value, contains url of resources
- *
- * Example: for resource in <img src="/images/foo.png"> it will be
- *  {
- *   tagName: 'img',
- *   attributeName: 'src',
- *   attributeValue: '/images/foo.png'
- *  }
- */
-
-/**
- *
- * @param {HtmlData} data
- */
-Resource.prototype.setHtmlData = function setHtmlData (data) {
-	this.htmlData = _.pick(data, ['tagName', 'attributeName']);
+Resource.prototype.setType = function setType (type) {
+	this.type = type;
 };
 
 Resource.prototype.getType = function getType () {
-	var ext = utils.getFilenameExtension(this.filename);
-	var parent = this.parent;
-	var hasHtmlData = !_.isEmpty(this.htmlData);
-
-	switch (true) {
-		case ext === '.html' || ext === '.htm':
-			return types.html;
-		case ext === '.css':
-		case !ext && parent && parent.isCss():
-			return types.css;
-		case !ext && parent && parent.isHtml() && hasHtmlData:
-			return getTypeByHtmlData(this.htmlData);
-		default:
-			return types.other;
-	}
+	return this.type;
 };
 
 Resource.prototype.isHtml = function isHtml () {
@@ -127,11 +79,11 @@ Resource.prototype.toString = function toString () {
 };
 
 Resource.prototype.isSaved = function isSaved () {
-	return this.saved || 0;
+	return this.saved;
 };
 
 Resource.prototype.setSaved = function setSaved () {
-	this.saved = 1;
+	this.saved = true;
 };
 
 module.exports = Resource;

lib/scraper.js

@@ -126,8 +126,17 @@ Scraper.prototype.requestResource = function requestResource (resource) {
 				self.addRespondedResourcePromise(responseData.url, respondedResourcePromise);
 			}
 
+			resource.setType(utils.getTypeByMime(responseData.mimeType));
+
 			var filename = self.filenameGenerator.generateFilename(resource);
 			resource.setFilename(filename);
+
+			// if type was not determined by mime
+			// we can try to get it from filename after it was generated
+			if (!resource.getType()) {
+				resource.setType(utils.getTypeByFilename(filename));
+			}
+
 			resource.setText(responseData.body);
 
 			logger.debug('finish request for ' + resource);

lib/utils.js

@@ -2,6 +2,8 @@ var url = require('url');
 var path = require('path');
 var Promise = require('bluebird');
 var normalizeUrl = require('normalize-url');
+var typeByMime = require('./config/resource-type-by-mime');
+var typeByExt = require('./config/resource-type-by-ext');
 
 var logger = require('./logger');
 
@@ -91,6 +93,15 @@ function isUriSchemaSupported (path) {
 	return !protocol || protocol && isUrl(path);
 }
 
+function getTypeByMime (mimeType) {
+	return typeByMime[mimeType];
+}
+
+function getTypeByFilename (filename) {
+	var ext = getFilenameExtension(filename);
+	return typeByExt[ext];
+}
+
 module.exports = {
 	isUrl: isUrl,
 	getUrl: getUrl,
@@ -104,5 +115,7 @@ module.exports = {
 	waitAllFulfilled: waitAllFulfilled,
 	normalizeUrl: normalizeUrl,
 	urlsEqual: urlsEqual,
-	isUriSchemaSupported: isUriSchemaSupported
+	isUriSchemaSupported: isUriSchemaSupported,
+	getTypeByMime: getTypeByMime,
+	getTypeByFilename: getTypeByFilename
 };

test/functional/resource-without-ext/resource-without-ext.test.js

@@ -35,12 +35,16 @@ describe('Functional resources without extensions', function() {
 		nock('http://example.com/').get('/').replyWithFile(200, mockDirname + '/index.html');
 
 		// mock for css fonts
-		nock('http://fonts.googleapis.com/').get('/css?family=Lato').replyWithFile(200, mockDirname + '/fonts.css');
+		nock('http://fonts.googleapis.com/').get('/css?family=Lato').replyWithFile(200, mockDirname + '/fonts.css', {
+			'content-type': 'text/css'
+		});
 		nock('http://fonts.gstatic.com/').get('/s/lato/v11/UyBMtLsHKBKXelqf4x7VRQ.woff2').reply(200, 'OK');
 		nock('http://fonts.gstatic.com/').get('/s/lato/v11/1YwB1sO8YE1Lyjf12WNiUA.woff2').reply(200, 'OK');
 
 		// mock for iframe
-		nock('http://example.com/').get('/iframe').replyWithFile(200, mockDirname + '/iframe.html');
+		nock('http://example.com/').get('/iframe').replyWithFile(200, mockDirname + '/iframe.html', {
+			'Content-Type': 'text/html'
+		});
 		nock('http://example.com/').get('/cat.png').reply(200, 'OK');
 
 		// mock for anchor

test/unit/request-test.js

@@ -21,7 +21,7 @@ describe('Request', function () {
 	describe('#makeRequest', function () {
 
 		it('should call request with correct params', function(done) {
-			var responseMock = { request: {href: ''}, body: '' };
+			var responseMock = { request: {href: ''}, body: '', headers: {} };
 			var requestStub = sinon.stub().yields(null, responseMock);
 
 			var customRequest = proxyquire('../../lib/request', {