Add recursive and maxDepth options

README.md

@@ -19,7 +19,7 @@ npm install website-scraper
 
 ## Usage
 ```javascript
-var scraper = require('website-scraper'); 
+var scraper = require('website-scraper');
 var options = {
   urls: ['http://nodejs.org/'],
   directory: '/path/to/save/',
@@ -38,7 +38,7 @@ scraper.scrape(options).then(function (result) {
 
 ## API
 ### scrape(options, callback)
-Makes requests to `urls` and saves all files found with `sources` to `directory`. 
+Makes requests to `urls` and saves all files found with `sources` to `directory`.
 
 **options** - object containing next options:
 
@@ -48,31 +48,34 @@ Makes requests to `urls` and saves all files found with `sources` to `directory`
  - `sources:` array of objects to load, specifies selectors and attribute values to select files for loading *(optional, see default value in `lib/config/defaults.js`)*
  - `subdirectories:` array of objects, specifies subdirectories for file extensions. If `null` all files will be saved to `directory` *(optional, see example below)*
  - `request`: object, custom options for [request](https://github.com/request/request#requestoptions-callback) *(optional, see example below)*
-
-
+ - `recursive`: boolean, if `true` scraper will follow anchors in html files. Don't forget to set `maxDepth` to avoid infinite downloading *(optional, see example below)*
+ - `maxDepth`: positive number, maximum allowed depth for dependencies *(optional, see example below)*
+
+
 **callback** - callback function *(optional)*, includes following parameters:
- 
+
   - `error:` if error - `Error` object, if success - `null`
   - `result:` if error - `null`, if success - array if objects containing:
     - `url:` url of loaded page
     - `filename:` filename where page was saved (relative to `directory`)
 
 
 ## Examples
-Let's scrape some pages from [http://nodejs.org/](http://nodejs.org/) with images, css, js files and save them to `/path/to/save/`. 
+#### Example 1
+Let's scrape some pages from [http://nodejs.org/](http://nodejs.org/) with images, css, js files and save them to `/path/to/save/`.
 Imagine we want to load:
   - [Home page](http://nodejs.org/) to `index.html`
   - [About page](http://nodejs.org/about/) to `about.html`
   - [Blog](http://blog.nodejs.org/) to `blog.html`
-  
+
 and separate files into directories:
 
-  - `img` for .jpg, .png, .svg (full path `/path/to/save/img`) 
+  - `img` for .jpg, .png, .svg (full path `/path/to/save/img`)
   - `js` for .js (full path `/path/to/save/js`)
   - `css` for .css (full path `/path/to/save/css`)
 
 ```javascript
-var scraper = require('website-scraper'); 
+var scraper = require('website-scraper');
 scraper.scrape({
   urls: [
     'http://nodejs.org/',	// Will be saved with default filename 'index.html'
@@ -101,3 +104,16 @@ scraper.scrape({
   console.log(err);
 });
 ```
+
+#### Example 2. Recursive downloading
+```javascript
+// Links from example.com will be followed
+// Links from links will be ignored because theirs depth = 2 is greater than maxDepth
+var scraper = require('website-scraper');
+scraper.scrape({
+  urls: ['http://example.com/'],
+  directory: '/path/to/save',
+  recursive: true,
+  maxDepth: 1
+}).then(console.log).catch(console.log);
+```

lib/config/recursive-sources.js

@@ -0,0 +1,3 @@
+module.exports = [
+	{ selector: 'a', attr: 'href' }
+];

lib/file-handlers/css.js

@@ -1,7 +1,6 @@
 var _ = require('underscore');
 var Promise = require('bluebird');
 var getCssUrls = require('css-url-parser');
-var Resource = require('../resource');
 var utils = require('../utils');
 
 function loadCss (context, resource) {
@@ -12,8 +11,7 @@ function loadCss (context, resource) {
 
 	var promises = _.map(cssUrls, function loadResourceFromCssUrl (cssUrl) {
 		var resourceUrl = utils.getUrl(url, cssUrl);
-		var cssResource = new Resource(resourceUrl);
-		cssResource.setParent(resource);
+		var cssResource = resource.createChild(resourceUrl);
 
 		return context.loadResource(cssResource).then(function handleLoadedSource (loadedResource) {
 			var relativePath = utils.getRelativePath(filename, loadedResource.getFilename());

lib/file-handlers/html.js

@@ -1,7 +1,6 @@
 var cheerio = require('cheerio');
 var Promise = require('bluebird');
 var utils = require('../utils');
-var Resource = require('../resource');
 
 function loadHtml (context, resource) {
 	var sources = context.getHtmlSources();
@@ -50,8 +49,7 @@ function loadResources (context, resource, source) {
 
 		if (attr) {
 			var resourceUrl = utils.getUrl(url, attr);
-			var htmlResource = new Resource(resourceUrl);
-			htmlResource.setParent(resource);
+			var htmlResource = resource.createChild(resourceUrl);
 			htmlResource.setHtmlData({ tagName: el[0].name, attributeName: source.attr });
 
 			return context.loadResource(htmlResource).then(function handleLoadedSource (loadedResource) {

lib/resource.js

@@ -15,6 +15,17 @@ function Resource (url, filename) {
 	this.filename = filename;
 }
 
+Resource.prototype.createChild = function createChild (url, filename) {
+	var child = new Resource(url, filename);
+
+	var currentDepth = this.getDepth();
+
+	child.setParent(this);
+	child.setDepth(++currentDepth);
+
+	return child;
+};
+
 Resource.prototype.getUrl = function getUrl () {
 	return this.url;
 };
@@ -43,6 +54,14 @@ Resource.prototype.setParent = function setParent (parent) {
 	this.parent = parent;
 };
 
+Resource.prototype.getDepth = function getDepth () {
+	return this.depth || 0;
+};
+
+Resource.prototype.setDepth = function setDepth (depth) {
+	this.depth = depth;
+};
+
 /**
  *
  * @param {Object} data - html element data

lib/scraper.js

@@ -1,28 +1,27 @@
 var Promise = require('bluebird');
-var fs = Promise.promisifyAll(require('fs-extra'));
+
+var fs = require('fs-extra');
+var existsAsync = Promise.promisify(fs.stat);
+var outputFileAsync = Promise.promisify(fs.outputFile);
+var ensureDirAsync = Promise.promisify(fs.ensureDir);
+
 var path = require('path');
 var _ = require('underscore');
 
 var defaults = require('./config/defaults');
 var types = require('./config/resource-types');
+var recursiveSources = require('./config/recursive-sources');
 var utils = require('./utils.js');
 var request = require('./request');
 var Resource = require('./resource');
-var loadHtml = require('./file-handlers/html');
-var loadCss = require('./file-handlers/css');
 var compareUrls = require('compare-urls');
 
-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;
-	}
+var loadHtml = require('./file-handlers/html');
+var loadCss = require('./file-handlers/css');
+function loadHtmlAndCss (context, po) {
+	return loadHtml(context, po).then(function (loaded) {
+		return loadCss(context, loaded);
+	});
 }
 
 function Scraper (options) {
@@ -83,6 +82,20 @@ Scraper.prototype.getDirectoryByExtension = function getDirectoryByExtension (ex
 		.value() || '';
 };
 
+Scraper.prototype.getResourceHandler = function getHandler (resource) {
+	var self = this;
+	var type = resource.getType();
+	var depth = resource.getDepth();
+	var depthGreaterThanMax = self.options.maxDepth && depth >= self.options.maxDepth;
+
+	switch (true) {
+		case depthGreaterThanMax: return _.noop;
+		case type == types.css: return loadCss;
+		case type == types.html: return loadHtmlAndCss;
+		default: return _.noop;
+	}
+};
+
 Scraper.prototype.loadResource = function loadResource (resource) {
 	var self = this;
 
@@ -102,12 +115,12 @@ Scraper.prototype.loadResource = function loadResource (resource) {
 		return self.makeRequest(url).then(function requestCompleted(data) {
 			resource.setUrl(data.url);  // Url may be changed in redirects
 			resource.setText(data.body);
-			handleFile = getHandleFunction(resource);
+			handleFile = self.getResourceHandler(resource);
 			return handleFile(self, resource);
 		}).then(function fileHandled() {
 			var filename = path.join(self.options.directory, resource.getFilename());
 			var text = resource.getText();
-			return fs.outputFileAsync(filename, text, { encoding: 'binary' });
+			return outputFileAsync(filename, text, { encoding: 'binary' });
 		}).then(function fileSaved() {
 			return Promise.resolve(resource);
 		});
@@ -116,15 +129,16 @@ Scraper.prototype.loadResource = function loadResource (resource) {
 };
 
 Scraper.prototype.validate = function validate () {
-	if (fs.existsSync(this.options.directory)) {
-		return Promise.reject(new Error('Path ' + this.options.directory + ' exists'));
-	}
-	return Promise.resolve();
+	var dir = this.options.directory;
+	return existsAsync(dir).then(function handleDirectoryExist () {
+		return Promise.reject(new Error('Path ' + dir + ' exists'));
+	}, function handleDirectoryNotExist () {
+		return Promise.resolve();
+	});
 };
 
 Scraper.prototype.prepare = function prepare () {
 	var self = this;
-	fs.ensureDirSync(self.options.directory);
 
 	// Create makeRequest function with custom request params
 	self.makeRequest = request.makeRequest.bind(null, self.options.request);
@@ -136,7 +150,12 @@ Scraper.prototype.prepare = function prepare () {
 		var filename = _.isObject(obj) && _.has(obj, 'filename') ? obj.filename : self.options.defaultFilename;
 		return new Resource(url, filename);
 	});
-	return Promise.resolve();
+
+	if (self.options.recursive) {
+		self.options.sources = _.union(self.options.sources, recursiveSources);
+	}
+
+	return ensureDirAsync(self.options.directory);
 };
 
 Scraper.prototype.load = function load () {

package.json

@@ -44,7 +44,9 @@
     "istanbul": "^0.4.0",
     "mocha": "^2.2.5",
     "nock": "^2.9.1",
+    "proxyquire": "^1.7.3",
     "should": "^7.0.2",
-    "sinon": "^1.15.4"
+    "sinon": "^1.15.4",
+    "sinon-as-promised": "^4.0.0"
   }
 }

test/functional/mocks/recursive/about.html

@@ -0,0 +1,13 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <title>Title</title>
+</head>
+<body>
+<a href="/link1.html"></a>
+<a href="/link2.html"></a>
+<a href="/link3.html"></a>
+
+</body>
+</html>

test/functional/mocks/recursive/index.html

@@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <title>Title</title>
+</head>
+<body>
+<a href="/about.html"></a>
+
+</body>
+</html>