Merge pull request #16 from mobify/WEB-2162

WEB-2162: Remove plaintext from SourceDoc upon Capture.init
mobify · May 1, 2019 · 5240c4d · 5240c4d
2 parents 179ae11 + dbb803c
commit 5240c4d
Show file tree

Hide file tree

Showing 4 changed files with 105 additions and 28 deletions.
diff --git a/CHANGELOG b/CHANGELOG
@@ -1,3 +1,12 @@
+1.0.7
+    - Upgrade dependencies (grunt-conbtrib-qunit and grunt-saucelabs).
+    - Make bower, browserify, and grunt-cli dependencies explicit.
+    - Update list of browsers to run saucelabs on.
+    - Remove plaintext from sourceDoc after captured doc elements have been
+      saved into the capturedDoc to prevent googlebot from picking up the
+      lingering plaintext as metadata.
+    - Add checks to tests to confirm that plaintext was added before
+      Capture.init and removed after the callback in Capture.init.
 1.0.6
     - Fixed issue where <!-- --> comments containing '<head>' created false
       positives when looking for the <head> section in HTML

diff --git a/package.json b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "mobify-capturejs",
-  "version": "1.0.6",
+  "version": "1.0.7",
   "devDependencies": {
     "bower": "~1.8.8",
     "browserify": "~3.46.1",

diff --git a/src/capture.js b/src/capture.js
@@ -128,7 +128,16 @@ Capture.init = Capture.initCapture = function(callback, doc, prefix) {
         Utils.extend(capture, capturedStringFragments);
         var capturedDOMFragments = capture.createDocumentFragments();
         Utils.extend(capture, capturedDOMFragments);
-        callback(capture);
+        try {
+            callback(capture);
+        } finally {
+            // Remove plaintext element after Capture.initCapture() callback
+            // has been called, and be robust in case of exceptions
+            var plaintextEl = capture.sourceDoc.querySelector('plaintext')
+            if (plaintextEl) {
+                plaintextEl.parentNode.removeChild(plaintextEl)
+            }
+        }
     };
 
     if (Utils.domIsReady(doc)) {

diff --git a/tests/capture-tests.js b/tests/capture-tests.js
@@ -79,26 +79,48 @@ require(["saucelabs-config", "qunit", "mobifyjs/utils", "capture"], function(_,
             // We remove the webdriver attribute set when running tests on selenium (typically done through SauceLabs)
             var htmlEl = doc.getElementsByTagName("html")[0].removeAttribute("webdriver")
 
+            var expectedPlaintext =
+                "    <link rel=stylesheet href=/path/to/stylesheet.css>" +
+                "</head>" +
+                "<body bar=baz>" +
+                "  <!-- comment with <head> -->" +
+                "  <p>Plaintext example page!</p>" +
+                "  <script src=/path/to/script.js><\/script>" +
+                "</body>" +
+                "</html>";
+
+            var plaintextEl = doc.querySelector('plaintext')
+            // Firefox 4.0 does not support access to innerText
+            var plaintextText = plaintextEl.innerText || plaintextEl.innerHTML
+            ok(compareHTMLStrings(plaintextText, expectedPlaintext), "plaintext is added")
+
             Capture.init(function(capture) {
-                var capturedDoc = capture.capturedDoc;
-
-                var expectedHtml =
-                    "<!DOCTYPE HTML>" +
-                    "<html class=testclass>" +
-                    "<head foo=bar>" +
-                    "    <link rel=stylesheet href=/path/to/stylesheet.css>" +
-                    "</head>" +
-                    "<body bar=baz>" +
-                    "  <!-- comment with <head> -->" +
-                    "  <p>Plaintext example page!</p>" +
-                    "  <script src=/path/to/script.js><\/script>" +
-                    "</body>" +
-                    "</html>";
-
-                var html = capture.enabledHTMLString(capturedDoc);
-                ok(compareHTMLStrings(html, expectedHtml), "Passed!");
-                start();
+                // wait for a turn of the event loop to see the removed 
+                // <plaintext> element
+                setTimeout(function() {
+                    var capturedDoc = capture.capturedDoc;
+
+                    var expectedHtml =
+                        "<!DOCTYPE HTML>" +
+                        "<html class=testclass>" +
+                        "<head foo=bar>" +
+                        "    <link rel=stylesheet href=/path/to/stylesheet.css>" +
+                        "</head>" +
+                        "<body bar=baz>" +
+                        "  <!-- comment with <head> -->" +
+                        "  <p>Plaintext example page!</p>" +
+                        "  <script src=/path/to/script.js><\/script>" +
+                        "</body>" +
+                        "</html>";
+
+                    var html = capture.enabledHTMLString(capturedDoc);
+                    ok(doc.querySelector('plaintext') === null, "plaintext is removed")
+                    ok(compareHTMLStrings(html, expectedHtml), "Passed!");
+                    start();
+                }, 0);
             }, doc);
+
+            ok(doc.querySelector('plaintext') === null, "plaintext is removed")
         });
         $("#qunit-fixture").append(iframe);
     });
@@ -492,6 +514,15 @@ require(["saucelabs-config", "qunit", "mobifyjs/utils", "capture"], function(_,
         });
         var el = $iframe[0];
 
+        var expectedPlaintext =
+            '</head>' +
+            '<body bar="baz">' +
+            '  <!-- comment with <head> -->' +
+            '  <p>Plaintext example page!</p>' +
+            '  <script src="/path/to/script.js"><\/script>' +
+            '</body>' +
+            '</html>';
+
         var expectedHTML =
             '<!DOCTYPE HTML>' +
             '<html class="testclass">' +
@@ -523,9 +554,19 @@ require(["saucelabs-config", "qunit", "mobifyjs/utils", "capture"], function(_,
             // We remove the webdriver attribute set when running tests on selenium (typically done through SauceLabs)
             var htmlEl = doc.getElementsByTagName("html")[0].removeAttribute("webdriver")
 
+            var plaintextEl = doc.querySelector('plaintext')
+            // Firefox 4.0 does not support access to innerText
+            var plaintextText = plaintextEl.innerText || plaintextEl.innerHTML
+            ok(compareHTMLStrings(plaintextText, expectedPlaintext), 'plaintext added')
             Capture.init(function(capture) {
-                capture.restore('<script>parent.postMessage("done!", "*");<\/script>');
+                // wait for a turn of the event loop to see the removed 
+                // <plaintext> element
+                setTimeout(function() {
+                    ok(doc.querySelector('plaintext') === null, "plaintext is removed")
+                    capture.restore('<script>parent.postMessage("done!", "*");<\/script>');
+                }, 0);
             }, doc);
+            ok(doc.querySelector('plaintext') === null, 'plaintext is removed')
         });
         $("#qunit-fixture").append($iframe);
     });
@@ -542,15 +583,33 @@ require(["saucelabs-config", "qunit", "mobifyjs/utils", "capture"], function(_,
         iframe.one('load', function() {
             var doc = this.contentDocument;
 
+            var expectedPlaintext =
+                '</head>' +
+                '<body>' +
+                '  <form><\/form>' +
+                '  <form><\/form>' +
+                '</body>' +
+                '</html>';
+
+            var plaintextEl = doc.querySelector('plaintext')
+            // Firefox 4.0 does not support access to innerText
+            var plaintextText = plaintextEl.innerText || plaintextEl.innerHTML
+            ok(compareHTMLStrings(plaintextText, expectedPlaintext), "plaintext added")
             Capture.init(function(capture) {
-                var capturedDoc = capture.capturedDoc;
-
-                var bodyChildren = capturedDoc.body.children;
-                equal(bodyChildren.length, 2);
-                equal(bodyChildren[0].nodeName, 'FORM', 'first child element is a form');
-                equal(bodyChildren[1].nodeName, 'FORM', 'second child element is a form');
-                start();
+                // wait for a turn of the event loop to see the removed 
+                // <plaintext> element
+                setTimeout(function() {
+                    var capturedDoc = capture.capturedDoc;
+
+                    var bodyChildren = capturedDoc.body.children;
+                    ok(doc.querySelector('plaintext') === null, 'plaintext is removed')
+                    equal(bodyChildren.length, 2);
+                    equal(bodyChildren[0].nodeName, 'FORM', 'first child element is a form');
+                    equal(bodyChildren[1].nodeName, 'FORM', 'second child element is a form');
+                    start();
+                }, 0);
             }, doc);
+            ok(doc.querySelector('plaintext') === null, 'plaintext is removed')
         });
         $("#qunit-fixture").append(iframe);
     });