[Added] Added PageWeight as a metric in SEOmetrics

[Added] Added the headings to the SEOmetrics display to improve its readability [Improved] SEOmetrics now does a better job displaying Top Keywords on pages with very little text
Pennebaker · Sep 4, 2016 · f3a02d4 · f3a02d4
1 parent 14a5f62
commit f3a02d4
Show file tree

Hide file tree

Showing 7 changed files with 116 additions and 42 deletions.
diff --git a/README.md b/README.md
@@ -61,7 +61,10 @@ Some things to do, and ideas for potential features:
 * [Improved] Do more sanity checking when building the breadcrumbs
 * [Improved] We don't swap in the element->title for mainEntityOfPage anymore
 * [Added] Added a variable for craft.seomatic.getTemplateMeta()
-* [Improved] Protocol-relative URLs are now allowed in  Schema
+* [Improved] Protocol-relative URLs are now allowed in the Schema
+* [Added] Added `PageWeight` as a metric in SEOmetrics
+* [Added] Added the headings to the SEOmetrics display to improve its readability
+* [Improved] SEOmetrics now does a better job displaying Top Keywords on pages with very little text
 * [Improved] Updated the README.md
 
 ### 1.1.32 -- 2016.08.25

diff --git a/controllers/SeomaticController.php b/controllers/SeomaticController.php
@@ -177,11 +177,29 @@ public function actionRenderMetrics()
                     curl_setopt($ch, CURLOPT_URL, $pagespeedDesktopUrl);
                     $pagespeedDesktopResult = curl_exec($ch);
                     curl_close($ch);
+                    $pageSpeedPageStats = array();
                     if ($pagespeedDesktopResult)
                     {
                         $pagespeedJson = json_decode($pagespeedDesktopResult, true);
                         if ($pagespeedJson)
                         {
+                            $pageSpeedPageStats = $pagespeedJson['pageStats'];
+                            if (empty($pageSpeedPageStats['htmlResponseBytes']))
+                                $pageSpeedPageStats['htmlResponseBytes'] = 0;
+                            if (empty($pageSpeedPageStats['cssResponseBytes']))
+                                $pageSpeedPageStats['cssResponseBytes'] = 0;
+                            if (empty($pageSpeedPageStats['imageResponseBytes']))
+                                $pageSpeedPageStats['imageResponseBytes'] = 0;
+                            if (empty($pageSpeedPageStats['javascriptResponseBytes']))
+                                $pageSpeedPageStats['javascriptResponseBytes'] = 0;
+                            if (empty($pageSpeedPageStats['otherResponseBytes']))
+                                $pageSpeedPageStats['otherResponseBytes'] = 0;
+                            $pageSpeedPageStats['totalResponseBytes'] = $pageSpeedPageStats['htmlResponseBytes'] +
+                                $pageSpeedPageStats['cssResponseBytes'] +
+                                $pageSpeedPageStats['imageResponseBytes'] +
+                                $pageSpeedPageStats['javascriptResponseBytes'] +
+                                $pageSpeedPageStats['otherResponseBytes'];
+
                             if (isset($pagespeedJson['responseCode']) && ($pagespeedJson['responseCode'] == "200" || $pagespeedJson['responseCode'] == "301" || $pagespeedJson['responseCode'] == "302"))
                             {
                                 if (isset($pagespeedJson['ruleGroups']['SPEED']['score']))
@@ -237,9 +255,10 @@ public function actionRenderMetrics()
                     foreach($dom->find('script') as $element)
                         $element->outertext = '';
                     $strippedDom = html_entity_decode($dom->plaintext);
-                    $strippedDom = preg_replace('@[^0-9a-z\.\!]+@i', ', ', $strippedDom);
+//                    $strippedDom = preg_replace('@[^0-9a-z\.\!]+@i', ', ', $strippedDom);
+                    $strippedDom = stripslashes($strippedDom);
                     $htmlDom = html_entity_decode($dom->outertext);
-                    $htmlDom = preg_replace('@[^0-9a-z\.\!]+@i', '', $htmlDom);
+//                    $htmlDom = preg_replace('@[^0-9a-z\.\!]+@i', '', $htmlDom);
 
     /* -- SEO statistics */
 
@@ -292,8 +311,16 @@ public function actionRenderMetrics()
 
                     $textToHtmlRatio = (strlen($strippedDom) / (strlen($htmlDom) - strlen($strippedDom))) * 100;
 
+                    $strippedDom = preg_replace('/\s+/', ' ', $strippedDom);
+
+/* -- Extract the page keywords, and clean them up a bit */
+
                     $pageKeywords = craft()->seomatic->extractKeywords($strippedDom);
-                    $pageKeywords = str_replace(",",", ", $pageKeywords);
+
+                    $pageKeywords = str_replace(",,",",", $pageKeywords);
+                    $pageKeywords = str_replace(" ,",",", $pageKeywords);
+                    $pageKeywords = str_replace(" .",".", $pageKeywords);
+                    $pageKeywords = preg_replace('/,+/', ',', $pageKeywords);
 
     /* -- Focus keywords */
 
@@ -357,6 +384,7 @@ public function actionRenderMetrics()
                         'validatorStatus' => $validatorStatus,
                         'validatorErrors' => $validatorErrors,
                         'validatorWarnings' => $validatorWarnings,
+                        'pageSpeedPageStats' => $pageSpeedPageStats,
                         'pagespeedDesktopScore' => $pagespeedDesktopScore,
                         'pagespeedDesktopUrl' => $pagespeedDesktopUrl,
                         'pagespeedMobileScore' => $pagespeedMobileScore,

diff --git a/releases.json b/releases.json
@@ -7,7 +7,10 @@
             "[Improved] Do more sanity checking when building the breadcrumbs",
             "[Improved] We don't swap in the element->title for mainEntityOfPage anymore",
             "[Added] Added a variable for craft.seomatic.getTemplateMeta()",
-            "[Improved] Protocol-relative URLs are now allowed in  Schema",
+            "[Improved] Protocol-relative URLs are now allowed in the Schema",
+            "[Added] Added `PageWeight` as a metric in SEOmetrics",
+            "[Added] Added the headings to the SEOmetrics display to improve its readability",
+            "[Improved] SEOmetrics now does a better job displaying Top Keywords on pages with very little text",
             "[Improved] Updated the README.md"
         ]
     },

diff --git a/services/SeomaticService.php b/services/SeomaticService.php
@@ -2976,6 +2976,24 @@ public function extractSummary($text = null, $limit = null, $withoutStopWords =
         return $summary;
     } /* -- extractSummary */
 
+/* --------------------------------------------------------------------------------
+    Return a human-readable file size
+-------------------------------------------------------------------------------- */
+
+    public function humanFileSize($size)
+    {
+        if ($size >= 1073741824) {
+          $fileSize = round($size / 1024 / 1024 / 1024,1) . 'GB';
+        } elseif ($size >= 1048576) {
+            $fileSize = round($size / 1024 / 1024,1) . 'MB';
+        } elseif($size >= 1024) {
+            $fileSize = round($size / 1024,1) . 'KB';
+        } else {
+            $fileSize = $size . ' bytes';
+        }
+        return $fileSize;
+    } /* -- humanFileSize */
+
 /* --------------------------------------------------------------------------------
     Sanitize the passed in array recursively
 -------------------------------------------------------------------------------- */