Merge pull request #297 from 0xc0d3c5/execute-scrape-tests-within-nodejs

Execute scrape.test.js within a Node.js environment

src/methods/artoo.methods.scrape.js

@@ -163,11 +163,11 @@
 
     params = params || {};
 
-    var sel = typeof root !== 'string' ? root.selector : root,
+    var sel = root,
         headers;
 
     if (!params.headers) {
-      return artoo.scrape(sel + ' tr:has(td)', {
+      return artoo.scrape($(sel).find('tr:has(td)'), {
         scrape: {
           iterator: 'td',
           data: params.data || 'text'
@@ -182,12 +182,12 @@
 
       if (headerType === 'th') {
         headers = artoo.scrape(
-          sel + ' th', headerFn || 'text'
+          $(sel).find('th'), headerFn || 'text'
         );
       }
       else if (headerType === 'first') {
         headers = artoo.scrape(
-          sel + ' tr:has(td):first-of-type td',
+          $(sel).find(' tr:has(td):first-of-type td'),
           headerFn || 'text'
         );
       }
@@ -200,8 +200,8 @@
 
       // Scraping
       return artoo.scrape(
-        sel + ' tr:has(td)' +
-        (headerType === 'first' ? ':not(:first-of-type)' : ''), function() {
+        $(sel).find('tr:has(td)' +
+        (headerType === 'first' ? ':not(:first-of-type)' : '')), function() {
           var o = {};
 
           headers.forEach(function(h, i) {

test/endpoint.js

@@ -6,3 +6,4 @@
  */
 
 require('./suites/node.test.js');
+require('./suites/node.scrape.test.js');

test/suites/node.scrape.test.js

@@ -0,0 +1,30 @@
+/**
+ * artoo scrape method node unit tests
+ * ===================================
+ *
+ */
+var cheerio = require('cheerio'),
+    artoo = require('../../build/artoo.node.js');
+
+function readFile(path) {
+    return fs.readFileSync(path, 'utf-8');
+}
+
+// Monkey patch - begin
+var assert = require('assert')
+var fs = require('fs')
+var $
+
+helpers = {
+    // Define a nodejs compliant fetchHTMLResource
+    fetchHTMLResource: function (name, cb) {
+        var $newDom = cheerio.load('<div id="' + name + '"></div>');
+        artoo.setContext($newDom)
+        $newDom('#' + name).append(readFile(__dirname + '/../resources/' + name + '.html'));
+        $ = $newDom
+        cb('#' + name);
+    }
+};
+// Monkey patch - end
+
+eval(readFile(__dirname + '/scrape.test.js'))

test/suites/scrape.test.js

@@ -57,7 +57,7 @@
         assert.deepEqual(
           artoo.scrape(
             function($) {
-              return $(id + ' li > a:first').add(id + ' li > a:last');
+              return $(id + ' li:first-of-type > a').add(id + ' li:last-of-type > a');
             }
           ),
           [titleList[0], titleList[3]],
@@ -365,7 +365,7 @@
         assert.deepEqual(
           artoo.scrapeOne(id + ' > ul', {
             title: {
-              sel: 'li:first > span'
+              sel: 'li:first-of-type > span'
             },
             items: {
               sel: 'li:nth-child(2)',
@@ -403,7 +403,7 @@
         });
 
         assert.deepEqual(
-          artoo.scrape(id + ' .reference tr:not(:first)', {
+          artoo.scrape(id + ' .reference tr:not(:first-of-type)', {
             scrape: {
               iterator: 'td',
               data: 'text'
@@ -426,10 +426,10 @@
         );
 
         assert.deepEqual(
-          artoo.scrape(id + ' .reference tr:not(:first)', {
-            firstname: {sel: 'td:first'},
-            lastname: {sel: 'td:eq(1)'},
-            points: {sel: 'td:eq(2)'}
+          artoo.scrape(id + ' .reference tr:not(:first-of-type)', {
+            firstname: {sel: 'td:first-of-type'},
+            lastname: {sel: 'td:nth-of-type(2)'},
+            points: {sel: 'td:nth-of-type(3)'}
           }),
           objects,
           'Scraping the list more easily should return the correct array of objects'