Merge pull request #6 from majewan/master

Issue #5: added createSitemapIndex

lib/errors.js

@@ -39,3 +39,12 @@ exports.PriorityInvalidError = function (message) {
   this.message = message || 'priority is invalid';
 }
 exports.PriorityInvalidError.prototype = Error.prototype;
+
+/**
+ * SitemapIndex target Folder does not exists
+ */
+exports.UndefinedTargetFolder = function (message) {
+  this.name = 'UndefinedTargetFolder';
+  this.message = message || 'Target folder must exist';
+}
+exports.UndefinedTargetFolder.prototype = Error.prototype;

lib/sitemap.js

@@ -11,6 +11,7 @@ var ut = require('./utils')
 exports.Sitemap = Sitemap;
 exports.SitemapItem = SitemapItem;
 exports.createSitemap = createSitemap;
+exports.createSitemapIndex = createSitemapIndex;
 
 /**
  * Shortcut for `new Sitemap (...)`.
@@ -199,7 +200,7 @@ Sitemap.prototype.toString = function () {
       smi = {'url': elem};
     }
     // insert domain name
-    if ( self.hostname && smi.url.indexOf('http') === -1 ) {
+    if ( self.hostname && smi.url.indexOf('http:') === -1 ) {
       smi.url = self.hostname + smi.url;
     }
     xml.push( new SitemapItem(smi) );
@@ -211,9 +212,120 @@ Sitemap.prototype.toString = function () {
   return this.cache;
 }
 
+/**
+ * Shortcut for `new Sitemap (...)`.
+ *
+ * @param   {Object}        conf
+ * @param   {String|Array}  conf.urls
+ * @param   {String}        conf.targetFolder
+ * @param   {String}        conf.hostname
+ * @param   {Number}        conf.cacheTime
+ * @param   {String}        conf.sitemapName
+ * @param   {Number}        conf.sitemapSize
+ * @return  {SitemapIndex}
+ */
+function createSitemapIndex(conf) {
+  return new SitemapIndex(conf.urls, conf.targetFolder, conf.hostname, conf.cacheTime, conf.sitemapName, conf.sitemapSize, conf.callback);
+}
+
 /**
  * Sitemap index (for several sitemaps)
+ * @param {String|Array}  urls
+ * @param {String}        targetFolder
+ * @param {String}        hostname      optional
+ * @param {Number}        cacheTime     optional in milliseconds
+ * @param {String}        sitemapName   optionnal
+ * @param {Number}        sitemapSize   optionnal
  */
-function SitemapIndex() {
+function SitemapIndex(urls, targetFolder, hostname, cacheTime, sitemapName, sitemapSize, callback) {
+
+  var self = this;
+
+  self.fs = require('fs');
+
+  // Base domain
+  self.hostname = hostname;
+
+  if(sitemapName === undefined) {
+    self.sitemapName = 'sitemap';
+  }
+  else {
+    self.sitemapName = sitemapName;
+  }
+
+  // This limit is defined by Google. See:
+  // http://sitemaps.org/protocol.php#index
+  self.sitemapSize = sitemapSize;
+
+  self.sitemapId = 0;
+
+  self.sitemaps = [];
+
+  self.targetFolder = '.';
+
+  if(!self.fs.existsSync(targetFolder)) {
+    throw new err.UndefinedTargetFolder();
+  }
+
+  self.targetFolder = targetFolder;
+
+  // URL list for sitemap
+  self.urls = urls || [];
+  if ( !(this.urls instanceof Array) ) {
+    this.urls = [ this.urls ]
+  }
+
+  self.chunks = ut.chunkArray(self.urls, self.sitemapSize);
+
+  self.callback = callback;
+
+  var processesCount = self.chunks.length + 1;
+
+  self.chunks.forEach( function (chunk, index) {
+
+    var filename = self.sitemapName + '-' + self.sitemapId++ + '.xml';
+    self.sitemaps.push(filename);
+
+    var sitemap = createSitemap ({
+      hostname: self.hostname,
+      cacheTime: self.cacheTime,        // 600 sec - cache purge period
+      urls: chunk
+    });
+
+    var stream = self.fs.createWriteStream(targetFolder + '/' + filename);
+    stream.once('open', function(fd) {
+      stream.write(sitemap.toString());
+      stream.end();
+      processesCount--;
+      if(processesCount === 0) {
+        callback(null, true);
+      }
+    });
+
+  });
+
+  var xml = [];
+
+  xml.push('<?xml version="1.0" encoding="UTF-8"?>');
+  xml.push('<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">');
+
+  self.sitemaps.forEach( function (sitemap, index) {
+    xml.push('<sitemap>');
+    xml.push('<loc>' + hostname + '/' + sitemap + '</loc>');
+//    xml.push('<lastmod>' + new Date() + '</lastmod>');
+    xml.push('</sitemap>');
+  });
+
+  xml.push('</sitemapindex>');
+
+  var stream = self.fs.createWriteStream(targetFolder + '/' + self.sitemapName + '-index.xml');
+  stream.once('open', function(fd) {
+    stream.write(xml.join('\n'));
+    stream.end();
+    processesCount--;
+    if(processesCount === 0) {
+      callback(null, true);
+    }
+  });
 
 }

lib/utils.js

@@ -60,3 +60,11 @@ exports.distinctArray = function (arr) {
     }
     return res;
 }
+
+exports.chunkArray = function(arr, chunkSize) {
+  return [].concat.apply([],
+    arr.map(function(elem,i) {
+      return i%chunkSize ? [] : [arr.slice(i,i+chunkSize)];
+    })
+  );
+}

package.json

@@ -1,6 +1,6 @@
 {
   "name": "sitemap",
-  "version": "0.6.0",
+  "version": "0.7.0",
   "description": "Sitemap-generating framework",
   "keywords": ["sitemap", "sitemap.xml"],
   "repository": "git://github.com/ekalinin/sitemap.js.git",

tests/sitemap.test.js

@@ -84,6 +84,43 @@ module.exports = {
                     '<priority>0.5</priority> '+
                 '</url>\n'+
               '</urlset>');
+  },
+  'simple sitemap index': function() {
+    var url1 = 'http://ya.ru';
+    var url2 = 'http://ya2.ru';
+
+    assert.throws(
+      function() {
+        var ssp = new sm.createSitemapIndex({
+          cacheTime: 600000,
+          hostname: 'http://www.sitemap.org',
+          sitemapName: 'sm-test',
+          sitemapSize: 1,
+          targetFolder: '/tmp2',
+          urls: [url1, url2]
+        });
+      },
+      /UndefinedTargetFolder/
+    );
+
+    var ssp = new sm.createSitemapIndex({
+      cacheTime: 600000,
+      hostname: 'http://www.sitemap.org',
+      sitemapName: 'sm-test',
+      sitemapSize: 1,
+      targetFolder: '/tmp',
+      urls: [url1, url2],
+      callback: function(err, result) {
+        assert.eql(err, null);
+        assert.eql(result, true);
+        assert.eql(require('fs').existsSync('/tmp/sm-test-0.xml'), true);
+        assert.eql(require('fs').existsSync('/tmp/sm-test-1.xml'), true);
+        assert.eql(require('fs').existsSync('/tmp/sm-test-index.xml'), true);
+      }
+    });
+
+
+
   },
   'lpad test': function() {
     assert.eql(sm.utils.lpad(5, 2), '05');