From 785995b4730c3e9bf556fd4e9d2f772d457d918c Mon Sep 17 00:00:00 2001
From: yackob03 <jacob.moshenko@gmail.com>
Date: Thu, 10 Oct 2013 20:53:14 -0400
Subject: [PATCH] First stab at trying to pre-render content for search
 crawlers.

---
 requirements-nover.txt            |  3 ++-
 seo-snapshots/make_snapshot.py    | 29 +++++++++++++++++++++++++++++
 seo-snapshots/phantomjs-runner.js | 23 +++++++++++++++++++++++
 static/js/controllers.js          |  5 ++++-
 static/partials/guide.html        |  2 +-
 static/partials/landing.html      |  2 +-
 6 files changed, 60 insertions(+), 4 deletions(-)
 create mode 100644 seo-snapshots/make_snapshot.py
 create mode 100644 seo-snapshots/phantomjs-runner.js

diff --git a/requirements-nover.txt b/requirements-nover.txt
index 69d73fb1b..cd844d0cb 100644
--- a/requirements-nover.txt
+++ b/requirements-nover.txt
@@ -10,4 +10,5 @@ pymysql
 stripe
 gunicorn
 eventlet
-mixpanel-py
\ No newline at end of file
+mixpanel-py
+beautifulsoup4
\ No newline at end of file
diff --git a/seo-snapshots/make_snapshot.py b/seo-snapshots/make_snapshot.py
new file mode 100644
index 000000000..75ab0be9d
--- /dev/null
+++ b/seo-snapshots/make_snapshot.py
@@ -0,0 +1,29 @@
+import subprocess
+import urllib
+
+from BeautifulSoup import BeautifulSoup
+
+
+BASE_URL = 'http://localhost:5000'
+OUTPUT_PATH = 'snapshots/'
+
+URLS = [
+  ('/', 'index.html')
+]
+
+for url, output in URLS:
+  final_url = BASE_URL + url
+
+  out_html = subprocess.check_output(['phantomjs', 'phantomjs-runner.js',
+                                      final_url])
+
+  # Remove script tags
+  soup = BeautifulSoup(out_html)
+  to_extract = soup.findAll('script')
+  for item in to_extract:
+      item.extract()
+
+  to_write = OUTPUT_PATH + output
+
+  with open(to_write, 'w') as output_file:
+    output_file.write(soup.prettify())
\ No newline at end of file
diff --git a/seo-snapshots/phantomjs-runner.js b/seo-snapshots/phantomjs-runner.js
new file mode 100644
index 000000000..e001b73c8
--- /dev/null
+++ b/seo-snapshots/phantomjs-runner.js
@@ -0,0 +1,23 @@
+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 ready = document.getElementsByClassName('ready-indicator')[0];
+          if(ready.getAttribute('data-status') == 'ready') {
+            return document.getElementsByTagName('html')[0].outerHTML;
+          }
+        });
+        if(html) {
+          clearTimeout(delay);
+          console.log(html);
+          phantom.exit();
+        }
+      });
+      delay = setInterval(checker, 100);
+    }
+  });
+}
diff --git a/static/js/controllers.js b/static/js/controllers.js
index 57eacb213..f297c9282 100644
--- a/static/js/controllers.js
+++ b/static/js/controllers.js
@@ -93,7 +93,8 @@ function PlansCtrl($scope, UserService, PlanService) {
   };
 }
 
-function GuideCtrl($scope, Restangular) {
+function GuideCtrl($scope) {
+  $scope.status = 'ready';
 }
 
 function RepoListCtrl($scope, Restangular, UserService) {
@@ -194,6 +195,8 @@ function LandingCtrl($scope, $timeout, Restangular, UserService, KeyService) {
       $scope.loadingmyrepos = false;
     });
   };
+
+  $scope.status = 'ready';
 }
 
 function RepoCtrl($scope, Restangular, $routeParams, $rootScope) {
diff --git a/static/partials/guide.html b/static/partials/guide.html
index 4020cb287..5ee657c1d 100644
--- a/static/partials/guide.html
+++ b/static/partials/guide.html
@@ -1,4 +1,4 @@
-<div class="container">
+<div class="container ready-indicator" data-status="{{ status }}">
   <div class="alert alert-warning">Warning: Quay requires docker version 0.6.2 or higher to work</div>
 
   <h2>Getting started guide</h2>
diff --git a/static/partials/landing.html b/static/partials/landing.html
index b7e2d6248..2baf6baec 100644
--- a/static/partials/landing.html
+++ b/static/partials/landing.html
@@ -1,4 +1,4 @@
-<div class="jumbotron landing">
+<div class="jumbotron landing ready-indicator" data-status="{{ status }}">
   <div class="container">
     <div class="row messages">
       <div class="col-md-7">