From 738973cf3957c30285aca4b6e42d2b45673ff27e Mon Sep 17 00:00:00 2001
From: Joseph Schorr <jschorr@gmail.com>
Date: Mon, 18 Nov 2013 17:11:06 -0500
Subject: [PATCH] Add the snapshot endpoint to web.py and have the phantomjs
 running only load the page's HTML once there are no further pending XHR
 requests

---
 endpoints/web.py         | 12 ++++++++++++
 static/js/app.js         | 12 +++++++++++-
 util/phantomjs-runner.js | 37 +++++++++++++++++++++++++++++++++++++
 util/seo.py              | 27 +++++++++++++++++++++++++++
 4 files changed, 87 insertions(+), 1 deletion(-)
 create mode 100644 util/phantomjs-runner.js
 create mode 100644 util/seo.py

diff --git a/endpoints/web.py b/endpoints/web.py
index ff6f1790a..c76d2f60b 100644
--- a/endpoints/web.py
+++ b/endpoints/web.py
@@ -11,6 +11,7 @@ from data import model
 from app import app, login_manager, mixpanel
 from auth.permissions import QuayDeferredPermissionUser, AdministerOrganizationPermission
 from util.invoice import renderInvoiceToPdf
+from util.seo import renderSnapshot
 
 logger = logging.getLogger(__name__)
 
@@ -49,6 +50,17 @@ def index(path):
   return render_template('index.html')
 
 
+@app.route('/snapshot', methods=['GET'])
+@app.route('/snapshot/', methods=['GET'])
+@app.route('/snapshot/<path:path>', methods=['GET'])
+def snapshot(path = ''):
+  result = renderSnapshot(path)
+  if result:
+    return result
+
+  abort(404)
+
+
 @app.route('/plans/')
 def plans():
   return index('')
diff --git a/static/js/app.js b/static/js/app.js
index 26d7aa5da..13b63a5b6 100644
--- a/static/js/app.js
+++ b/static/js/app.js
@@ -1053,7 +1053,8 @@ quayApp.directive('ngBlur', function() {
   };
 });
 
-quayApp.run(['$location', '$rootScope', 'Restangular', 'UserService', function($location, $rootScope, Restangular, UserService) {
+quayApp.run(['$location', '$rootScope', 'Restangular', 'UserService', '$http',
+   function($location, $rootScope, Restangular, UserService, $http) {
   Restangular.setErrorInterceptor(function(response) {
     if (response.status == 401) {
       $('#sessionexpiredModal').modal({});
@@ -1068,4 +1069,13 @@ quayApp.run(['$location', '$rootScope', 'Restangular', 'UserService', function($
       $rootScope.title = current.$$route.title;
     }
   });
+
+  var initallyChecked = false;
+  window.__isLoading = function() {
+    if (!initallyChecked) {
+      initallyChecked = true;
+      return true;
+    }
+    return $http.pendingRequests.length > 0;
+  };
 }]);
diff --git a/util/phantomjs-runner.js b/util/phantomjs-runner.js
new file mode 100644
index 000000000..30b0439fa
--- /dev/null
+++ b/util/phantomjs-runner.js
@@ -0,0 +1,37 @@
+var system = require('system');
+var url = system.args[1] || '';
+if(url.length > 0) {
+  var page = require('webpage').create();  
+  page.open(url, function (status) {
+    if (status == 'success') {     
+      var delay, checker = (function() {
+        var html = page.evaluate(function () {
+          var found = document.getElementsByTagName('html')[0].outerHTML || '';
+          if (window.__isLoading && !window.__isLoading()) {
+            return found;
+          }
+          if (found.indexOf('404 Not Found') > 0) {
+            return found;
+          }
+          return null;
+        });
+
+        if (html) {
+          if (html.indexOf('404 Not Found') > 0) {
+            console.log('Not Found');
+            phantom.exit();
+            return;
+          }
+
+          clearTimeout(delay);
+          console.log(html);
+          phantom.exit();
+        }
+      });
+      delay = setInterval(checker, 100);
+    } else {
+      console.log('Not Found');
+      phantom.exit();
+    }
+  });
+}
\ No newline at end of file
diff --git a/util/seo.py b/util/seo.py
new file mode 100644
index 000000000..6958e2c67
--- /dev/null
+++ b/util/seo.py
@@ -0,0 +1,27 @@
+import subprocess
+import urllib
+import os
+import logging
+import codecs
+
+from bs4 import BeautifulSoup
+
+logger = logging.getLogger(__name__)
+logging.basicConfig(level=logging.DEBUG)
+
+def renderSnapshot(path):
+    final_url = 'http://localhost:5000/' + path
+    logger.info('Snapshotting url: %s -> %s' % (path, final_url))
+    out_html = subprocess.check_output(['phantomjs', '--ignore-ssl-errors=yes',
+                                        'util/phantomjs-runner.js', final_url])
+
+    if not out_html or out_html.strip() == 'Not Found':
+      return None
+
+    # Remove script tags
+    soup = BeautifulSoup(out_html)
+    to_extract = soup.findAll('script')
+    for item in to_extract:
+        item.extract()
+
+    return soup.prettify()