From 552ac8efc13da3d02dfdd2f095a5ddc040eb299c Mon Sep 17 00:00:00 2001
From: Nicolas Hohm <nickel7152@gmail.com>
Date: Sun, 14 Feb 2016 12:38:56 +0100
Subject: [PATCH] Implement content checker

---
 composer.json                     |  2 +-
 config.php.dist                   | 14 ++---
 console                           |  2 +
 src/Checker/Content.php           | 86 +++++++++++++++++++++++++++++++
 src/Checker/TextOrHtmlPresent.php | 18 -------
 5 files changed, 97 insertions(+), 25 deletions(-)
 create mode 100644 src/Checker/Content.php
 delete mode 100644 src/Checker/TextOrHtmlPresent.php

diff --git a/composer.json b/composer.json
index bd90989..1c71498 100644
--- a/composer.json
+++ b/composer.json
@@ -1,5 +1,5 @@
 {
-    "name": "nickel715/websiteMonitoring",
+    "name": "nickel715/website-monitoring",
     "description": "",
     "license": "MIT",
     "authors": [
diff --git a/config.php.dist b/config.php.dist
index 3ebf2cf..b791d78 100644
--- a/config.php.dist
+++ b/config.php.dist
@@ -10,20 +10,22 @@ return [
         [
             'url'     => '',
             'checker' => [
-                'page-exists'          => [
+                'page-exists'       => [
                     'pages' => $privatePages,
                 ],
-                'robots-txt'           => [
+                'robots-txt'        => [
                     'disallows' => $privatePages,
                 ],
-                'sitemap-blacklist'    => [
+                'sitemap-blacklist' => [
                     'path'      => 'sitemap.xml',
                     'blacklist' => $privatePages,
                 ],
-                'text-or-html-present' => [
+                'content'           => [
                     [
-                        'page' => '/',
-                        'text' => 'copyright',
+                        'page'             => '/',
+                        'text_present'     => 'copyright',
+                        'html_present'     => ['<h1>', '<h2>'],
+                        'text_not_present' => 'debug',
                     ],
                 ],
             ],
diff --git a/console b/console
index d97240c..c1ebc5b 100755
--- a/console
+++ b/console
@@ -4,6 +4,7 @@
 require_once 'vendor/autoload.php';
 
 use Interop\Container\ContainerInterface;
+use WebsiteMonitoring\Checker\Content;
 use WebsiteMonitoring\Checker\PageExists;
 use WebsiteMonitoring\Checker\RobotsTxt;
 use WebsiteMonitoring\Checker\SitemapBlacklist;
@@ -20,6 +21,7 @@ $serviceManager->setFactory(CheckerPluginManager::class, function (ContainerInte
             'page-exists'       => PageExists::class,
             'robots-txt'        => RobotsTxt::class,
             'sitemap-blacklist' => SitemapBlacklist::class,
+            'content'           => Content::class,
         ],
     ]);
 });
diff --git a/src/Checker/Content.php b/src/Checker/Content.php
new file mode 100644
index 0000000..fa99aa8
--- /dev/null
+++ b/src/Checker/Content.php
@@ -0,0 +1,86 @@
+<?php
+
+namespace WebsiteMonitoring\Checker;
+
+use WebsiteMonitoring\WebsiteConfig;
+use Zend\Http\Client;
+
+class Content implements CheckerInterface
+{
+    public function check(WebsiteConfig $config, array $checkerConfig = [])
+    {
+        $failures = [];
+        foreach ($checkerConfig as $page) {
+            if (empty($page['page'])) {
+                $failures[] = 'Config error, no page defined';
+                continue;
+            }
+            $link = $config->getUrl() . $page['page'];
+            $response = (new Client($link))->send();
+            if (!$response->isOk()) {
+                $failures = array_merge($failures, [
+                    'Can not request the sitemap at ' . $link . '.',
+                    'The following error occurs: ' . $response->getStatusCode() . ' ' . $response->getReasonPhrase(),
+                ]);
+            }
+            $failures = array_merge($failures, $this->checkPage($response->getBody(), $page));
+        }
+        return $failures;
+    }
+
+    private function checkPage($html, array $pageConfig)
+    {
+        $failures = [];
+        $text = strip_tags(html_entity_decode($html));
+        if (!empty($pageConfig['text_present'])) {
+            foreach ((array)$pageConfig['text_present'] as $item) {
+                strpos($text, $item) === false
+                    && $failures[] = 'Can not find ' . $item . ' on page ' . $pageConfig['page'];
+            }
+        }
+        if (!empty($pageConfig['text_not_present'])) {
+            foreach ((array)$pageConfig['text_not_present'] as $item) {
+                strpos($text, $item) !== false
+                    && $failures[] = 'Found ' . $item . ' on page ' . $pageConfig['page'];
+            }
+        }
+        if (!empty($pageConfig['html_present'])) {
+            foreach ((array)$pageConfig['html_present'] as $item) {
+                strpos($html, $item) === false
+                    && $failures[] = 'Can not find  ' . $item . ' on page ' . $pageConfig['page'];
+            }
+        }
+        if (!empty($pageConfig['html_not_present'])) {
+            foreach ((array)$pageConfig['html_not_present'] as $item) {
+                strpos($html, $item) !== false
+                && $failures[] = 'Found ' . $item . ' on page ' . $pageConfig['page'];
+            }
+        }
+        return $failures;
+    }
+
+    public function parse(WebsiteConfig $config, array $checkerConfig = [])
+    {
+        $result = [];
+
+        foreach ($checkerConfig as $page) {
+            if (empty($page['page'])) {
+                return 'Config error, no page defined';
+            }
+            $result[] = 'Check page ' . $page['page'];
+            if (!empty($page['text_present'])) {
+                $result[] = 'For following text(s) are present: ' . implode(', ', (array)$page['text_present']);
+            }
+            if (!empty($page['text_not_present'])) {
+                $result[] = 'For following text(s) are not present: ' . implode(', ', (array)$page['text_not_present']);
+            }
+            if (!empty($page['html_present'])) {
+                $result[] = 'For following html are present: ' . implode(', ', (array)$page['html_present']);
+            }
+            if (!empty($page['html_not_present'])) {
+                $result[] = 'For following html are not present: ' . implode(', ', (array)$page['html_not_present']);
+            }
+            return $result;
+        }
+    }
+}
diff --git a/src/Checker/TextOrHtmlPresent.php b/src/Checker/TextOrHtmlPresent.php
deleted file mode 100644
index b0ca6b8..0000000
--- a/src/Checker/TextOrHtmlPresent.php
+++ /dev/null
@@ -1,18 +0,0 @@
-<?php
-
-namespace WebsiteMonitoring\Checker;
-
-use WebsiteMonitoring\WebsiteConfig;
-
-class TextOrHtmlPresent implements CheckerInterface
-{
-    public function check(WebsiteConfig $config, array $checkerConfig = [])
-    {
-        // TODO: Implement check() method.
-    }
-
-    public function parse(WebsiteConfig $config, array $checkerConfig = [])
-    {
-        // TODO: Implement parse() method.
-    }
-}