From f4ffeb37539a142aea763a92d1122cdbcd8c6c9b Mon Sep 17 00:00:00 2001
From: Tadas Juozapaitis <kasp3rito@gmail.com>
Date: Mon, 7 Apr 2014 16:58:08 +0300
Subject: [PATCH] refactoring and adding youtube parser

---
 README.md                                  |  34 +++++
 src/LinkPreview/LinkPreview.php            |  72 ++---------
 src/LinkPreview/Model/VideoLink.php        |  77 +++++++++++
 src/LinkPreview/Parser/GeneralParser.php   |  52 +++++++-
 src/LinkPreview/Parser/ParserInterface.php |  27 +++-
 src/LinkPreview/Parser/YoutubeParser.php   | 141 +++++++++++++++++++++
 src/LinkPreview/Reader/GeneralReader.php   |   2 +-
 src/LinkPreview/Reader/ReaderInterface.php |   1 -
 8 files changed, 340 insertions(+), 66 deletions(-)
 create mode 100644 src/LinkPreview/Model/VideoLink.php
 create mode 100644 src/LinkPreview/Parser/YoutubeParser.php

diff --git a/README.md b/README.md
index b107628..0818775 100644
--- a/README.md
+++ b/README.md
@@ -62,9 +62,43 @@ GitHub is the best place to build software together. Over 4 million people use G
 https://github.global.ssl.fastly.net/images/modules/open_graph/github-octocat.png
 ```
 
+```php
+$linkPreview = new LinkPreview('https://www.youtube.com/watch?v=8ZcmTl_1ER8');
+$parsed = $linkPreview->getParsed();
+foreach ($parsed as $parserName => $link) {
+    echo $parserName . PHP_EOL . PHP_EOL;
+
+    echo $link->getUrl() . PHP_EOL;
+    echo $link->getRealUrl() . PHP_EOL;
+    echo $link->getTitle() . PHP_EOL;
+    echo $link->getDescription() . PHP_EOL;
+    echo $link->getImage() . PHP_EOL;
+    echo $link->getVideoId() . PHP_EOL;
+    echo $link->getEmbedCode() . PHP_EOL;
+}
+```
+
+
+**Output**
+
+```
+youtube
+
+https://www.youtube.com/watch?v=8ZcmTl_1ER8
+http://gdata.youtube.com/feeds/api/videos/8ZcmTl_1ER8?v=2&alt=jsonc
+Epic sax guy 10 hours
+I had to remove my original one so I reuploaded this with much better quality.
+(If you want it sound like previous one, try setting quality to 240p)
+Yeah, I know that video sucks compared to original but no can do :(
+http://i1.ytimg.com/vi/8ZcmTl_1ER8/hqdefault.jpg
+8ZcmTl_1ER8
+<iframe id="ytplayer" type="text/html" width="640" height="390" src="http://www.youtube.com/embed/8ZcmTl_1ER8" frameborder="0"/>
+```
+
 ## Todo
 1. Add more unit tests
 2. Update documentation
+3. Add more parsers
 
 ## License
 
diff --git a/src/LinkPreview/LinkPreview.php b/src/LinkPreview/LinkPreview.php
index 6111187..41ae42c 100644
--- a/src/LinkPreview/LinkPreview.php
+++ b/src/LinkPreview/LinkPreview.php
@@ -2,24 +2,17 @@
 
 namespace LinkPreview;
 
-use LinkPreview\Model\Link;
 use LinkPreview\Model\LinkInterface;
 use LinkPreview\Parser\GeneralParser;
 use LinkPreview\Parser\ParserInterface;
-use LinkPreview\Reader\GeneralReader;
-use LinkPreview\Reader\ReaderInterface;
+use LinkPreview\Parser\YoutubeParser;
 
 class LinkPreview
 {
     /**
-     * @var LinkInterface $link
+     * @var string $url
      */
-    private $link;
-
-    /**
-     * @var ReaderInterface $reader
-     */
-    private $reader;
+    private $url;
 
     /**
      * @var ParserInterface[]
@@ -49,55 +42,17 @@ public function __construct($url = null)
      */
     public function setUrl($url)
     {
-        $this->setLink(new Link($url));
-
-        return $this;
-    }
-
-    /**
-     * Set model
-     *
-     * @param LinkInterface $link Link model
-     * @return $this
-     */
-    public function setLink(LinkInterface $link)
-    {
-        $this->link = $link;
+        $this->url = $url;
 
         return $this;
     }
 
     /**
-     * Get model
-     *
-     * @return LinkInterface
+     * @return string
      */
-    public function getLink()
+    public function getUrl()
     {
-        return $this->link;
-    }
-
-    /**
-     * Get reader
-     *
-     * @return ReaderInterface
-     */
-    public function getReader()
-    {
-        return $this->reader;
-    }
-
-    /**
-     * Set reader
-     *
-     * @param ReaderInterface $reader
-     * @return $this
-     */
-    public function setReader($reader)
-    {
-        $this->reader = $reader;
-
-        return $this;
+        return $this->url;
     }
 
     /**
@@ -190,19 +145,11 @@ public function getParsed()
             $this->addDefaultParsers();
         }
 
-        if (null === $this->getReader()) {
-            $this->setReader(new GeneralReader());
-        }
-
-        $reader = $this->getReader()->setLink($this->getLink());
-        $link = $reader->readLink()->getLink();
-        $this->setLink($link);
-
         foreach ($this->getParsers() as $name => $parser) {
-            $parser->setLink($this->getLink());
+            $parser->getLink()->setUrl($this->getUrl());
 
             if ($parser->isValidParser()) {
-                $parsed[$name] = $parser->parseLink()->getLink();
+                $parsed[$name] = $parser->parseLink();
 
                 if (!$this->getPropagation()) {
                     break;
@@ -219,5 +166,6 @@ public function getParsed()
     protected function addDefaultParsers()
     {
         $this->addParser(new GeneralParser());
+        $this->addParser(new YoutubeParser());
     }
 }
\ No newline at end of file
diff --git a/src/LinkPreview/Model/VideoLink.php b/src/LinkPreview/Model/VideoLink.php
new file mode 100644
index 0000000..45691df
--- /dev/null
+++ b/src/LinkPreview/Model/VideoLink.php
@@ -0,0 +1,77 @@
+<?php
+
+namespace LinkPreview\Model;
+
+class VideoLink extends Link
+{
+    /**
+     * @var string $video Url to video
+     */
+    private $video;
+
+    /**
+     * @var string $embedCode Video embed code
+     */
+    private $embedCode;
+
+    /**
+     * @var string $videoId Video identification code
+     */
+    private $videoId;
+
+    /**
+     * @return string
+     */
+    public function getVideo()
+    {
+        return $this->video;
+    }
+
+    /**
+     * @param string $video
+     * @return $this
+     */
+    public function setVideo($video)
+    {
+        $this->video = $video;
+
+        return $this;
+    }
+
+    /**
+     * @return string
+     */
+    public function getEmbedCode()
+    {
+        return $this->embedCode;
+    }
+
+    /**
+     * @param string $embedCode
+     * @return $this
+     */
+    public function setEmbedCode($embedCode)
+    {
+        $this->embedCode = $embedCode;
+
+        return $this;
+    }
+
+    /**
+     * @return string
+     */
+    public function getVideoId()
+    {
+        return $this->videoId;
+    }
+
+    /**
+     * @param string $videoId
+     */
+    public function setVideoId($videoId)
+    {
+        $this->videoId = $videoId;
+    }
+
+
+}
\ No newline at end of file
diff --git a/src/LinkPreview/Parser/GeneralParser.php b/src/LinkPreview/Parser/GeneralParser.php
index 23c0dfc..adbc2bd 100644
--- a/src/LinkPreview/Parser/GeneralParser.php
+++ b/src/LinkPreview/Parser/GeneralParser.php
@@ -2,7 +2,10 @@
 
 namespace LinkPreview\Parser;
 
+use LinkPreview\Model\Link;
 use LinkPreview\Model\LinkInterface;
+use LinkPreview\Reader\GeneralReader;
+use LinkPreview\Reader\ReaderInterface;
 
 class GeneralParser implements ParserInterface
 {
@@ -29,6 +32,26 @@ class GeneralParser implements ParserInterface
      */
     private $link;
 
+    /**
+     * @var ReaderInterface $reader
+     */
+    private $reader;
+
+    public function __construct(ReaderInterface $reader = null, LinkInterface $link = null)
+    {
+        if (null !== $reader) {
+            $this->setReader($reader);
+        } else {
+            $this->setReader(new GeneralReader());
+        }
+
+        if (null !== $link) {
+            $this->setLink($link);
+        } else {
+            $this->setLink(new Link());
+        }
+    }
+
     /**
      * @inheritdoc
      */
@@ -37,6 +60,25 @@ public function __toString()
         return 'general';
     }
 
+    /**
+     * @return ReaderInterface
+     */
+    public function getReader()
+    {
+        return $this->reader;
+    }
+
+    /**
+     * @param ReaderInterface $reader
+     * @return $this
+     */
+    public function setReader(ReaderInterface $reader)
+    {
+        $this->reader = $reader;
+
+        return $this;
+    }
+
     /**
      * @inheritdoc
      */
@@ -71,11 +113,19 @@ public function isValidParser()
         return $isValid;
     }
 
+    private function readLink()
+    {
+        $reader = $this->getReader()->setLink($this->getLink());
+        $this->setLink($reader->readLink());
+    }
+
     /**
      * @inheritdoc
      */
     public function parseLink()
     {
+        $this->readLink();
+
         $link = $this->getLink();
 
         if (!strncmp($link->getContentType(), 'text/', strlen('text/'))) {
@@ -88,7 +138,7 @@ public function parseLink()
             $link->setImage($link->getRealUrl());
         }
 
-        return $this;
+        return $link;
     }
 
     /**
diff --git a/src/LinkPreview/Parser/ParserInterface.php b/src/LinkPreview/Parser/ParserInterface.php
index af47011..106c17d 100644
--- a/src/LinkPreview/Parser/ParserInterface.php
+++ b/src/LinkPreview/Parser/ParserInterface.php
@@ -3,9 +3,19 @@
 namespace LinkPreview\Parser;
 
 use LinkPreview\Model\LinkInterface;
+use LinkPreview\Reader\ReaderInterface;
 
 interface ParserInterface
 {
+    /**
+     * Set default reader and model
+     *
+     *
+     * @param ReaderInterface $reader
+     * @param LinkInterface $link
+     */
+    public function __construct(ReaderInterface $reader = null, LinkInterface $link = null);
+
     /**
      * Parser name
      *
@@ -13,6 +23,21 @@ interface ParserInterface
      */
     public function __toString();
 
+    /**
+     * Set reader
+     *
+     * @param ReaderInterface $reader
+     * @return $this
+     */
+    public function setReader(ReaderInterface $reader);
+
+    /**
+     * Get reader
+     *
+     * @return ReaderInterface
+     */
+    public function getReader();
+
     /**
      * Set model
      *
@@ -38,7 +63,7 @@ public function isValidParser();
     /**
      * Parse link
      *
-     * @return $this
+     * @return LinkInterface
      */
     public function parseLink();
 } 
\ No newline at end of file
diff --git a/src/LinkPreview/Parser/YoutubeParser.php b/src/LinkPreview/Parser/YoutubeParser.php
new file mode 100644
index 0000000..e8f6ac8
--- /dev/null
+++ b/src/LinkPreview/Parser/YoutubeParser.php
@@ -0,0 +1,141 @@
+<?php
+
+namespace LinkPreview\Parser;
+
+use LinkPreview\Model\LinkInterface;
+use LinkPreview\Model\VideoLink;
+use LinkPreview\Reader\GeneralReader;
+use LinkPreview\Reader\ReaderInterface;
+
+class YoutubeParser implements ParserInterface
+{
+    /**
+     * Url validation pattern taken from http://stackoverflow.com/questions/2964678/jquery-youtube-url-validation-with-regex
+     */
+    const PATTERN = '/^(?:https?:\/\/)?(?:www\.)?(?:youtu\.be\/|youtube\.com\/(?:embed\/|v\/|watch\?v=|watch\?.+&v=))((\w|-){11})(?:\S+)?$/';
+
+    const YOUTUBE_API_URL = 'http://gdata.youtube.com/feeds/api/videos/[id]?v=2&alt=jsonc';
+
+    /**
+     * @var VideoLink $link
+     */
+    private $link;
+
+    /**
+     * @var ReaderInterface $reader
+     */
+    private $reader;
+
+    public function __construct(ReaderInterface $reader = null, LinkInterface $link = null)
+    {
+        if (null !== $reader) {
+            $this->setReader($reader);
+        } else {
+            $this->setReader(new GeneralReader());
+        }
+
+        if (null !== $link) {
+            $this->setLink($link);
+        } else {
+            $this->setLink(new VideoLink());
+        }
+    }
+
+    /**
+     * @inheritdoc
+     */
+    public function __toString()
+    {
+        return 'youtube';
+    }
+
+    /**
+     * @return ReaderInterface
+     */
+    public function getReader()
+    {
+        return $this->reader;
+    }
+
+    /**
+     * @param ReaderInterface $reader
+     * @return $this
+     */
+    public function setReader(ReaderInterface $reader)
+    {
+        $this->reader = $reader;
+
+        return $this;
+    }
+
+    /**
+     * @inheritdoc
+     */
+    public function setLink(LinkInterface $link)
+    {
+        $this->link = $link;
+
+        return $this;
+    }
+
+    /**
+     * @inheritdoc
+     */
+    public function getLink()
+    {
+        return $this->link;
+    }
+
+    /**
+     * @inheritdoc
+     */
+    public function isValidParser()
+    {
+        $isValid = false;
+
+        $link = $this->getLink();
+        $url = $link->getUrl();
+
+        if (is_string($url) && preg_match(static::PATTERN, $url, $matches)) {
+            $link->setVideoId($matches[1]);
+            $isValid = true;
+        }
+
+        return $isValid;
+    }
+
+    private function readLink()
+    {
+        $link = $this->getLink();
+        $originalUrl = $link->getUrl();
+        // change url to api link
+        $link->setUrl(str_replace('[id]', $link->getVideoId(), self::YOUTUBE_API_URL));
+
+        $reader = $this->getReader()->setLink($link);
+        $link = $reader->readLink();
+        // change to original url after reading
+        $link->setUrl($originalUrl);
+
+        $this->setLink($link);
+    }
+
+    /**
+     * @inheritdoc
+     */
+    public function parseLink()
+    {
+        $this->readLink();
+
+        $link = $this->getLink();
+        $data = json_decode($link->getContent());
+
+        $link->setTitle($data->data->title)
+            ->setDescription($data->data->description)
+            ->setImage($data->data->thumbnail->hqDefault)
+            ->setEmbedCode(
+                '<iframe id="ytplayer" type="text/html" width="640" height="390" src="http://www.youtube.com/embed/' . $link->getVideoId() . '" frameborder="0"/>'
+            );
+
+        return $link;
+    }
+}
\ No newline at end of file
diff --git a/src/LinkPreview/Reader/GeneralReader.php b/src/LinkPreview/Reader/GeneralReader.php
index 5052426..003e073 100644
--- a/src/LinkPreview/Reader/GeneralReader.php
+++ b/src/LinkPreview/Reader/GeneralReader.php
@@ -44,6 +44,6 @@ public function readLink()
             ->setContentType($response->getContentType())
             ->setRealUrl($response->getEffectiveUrl());
 
-        return $this;
+        return $link;
     }
 } 
\ No newline at end of file
diff --git a/src/LinkPreview/Reader/ReaderInterface.php b/src/LinkPreview/Reader/ReaderInterface.php
index 23ee79c..ab41d7b 100644
--- a/src/LinkPreview/Reader/ReaderInterface.php
+++ b/src/LinkPreview/Reader/ReaderInterface.php
@@ -25,7 +25,6 @@ public function getLink();
      * Read and update model
      *
      * @return LinkInterface
-     * @return $this
      */
     public function readLink();
 } 
\ No newline at end of file