Merge pull request #588 from TomCan/url-transformer

Add UrlTransformerService and integrate in LinkifyExtension

src/Service/UrlTransformerService.php

@@ -0,0 +1,153 @@
+<?php
+
+namespace App\Service;
+
+class UrlTransformerService
+{
+    /*
+     * Known hostformats.
+     *  key is regex to match host portion of url (parse_url)
+     *  value are <type>[,parameters...] where type denotes the affiliate program later used for the transformation logic
+     */
+    private array $hostFormats = [
+        '/^(www\.)?amazon\.(com|co\.(jp|uk|za)|com\.(au|be|br|mx|tr)|ae|ca|cn|de|eg|es|fr|ie|in|it|nl|pl|sa|se|sg)$/' => 'amazon',
+        '/^(www\.)?bol\.com$/' => 'bol',
+        '/^(www\.)?coolblue\.be$/' => 'awin,85165',
+    ];
+
+    private $partnerIds = [];
+
+    public function __construct()
+    {
+        // Get all partner ids from ENV.
+        // Multiple ids are supported by space seperating them.
+        foreach ($_ENV as $key => $value) {
+            if (0 === strpos($key, 'PARTNER_')) {
+                $this->partnerIds[strtolower(substr($key, 8))] = array_filter(explode(' ', $value));
+            }
+        }
+    }
+
+    public function extractUrls(string $text): array
+    {
+        $pattern = '#\bhttps?://[^,\s()<>]+(?:\([\w\d]+\)|([^,[:punct:]\s]|/))#';
+        if (preg_match_all($pattern, $text, $matches)) {
+            // remove duplicates
+            $urls = array_unique($matches[0]);
+
+            return $urls;
+        }
+
+        return [];
+    }
+
+    /**
+     * Replace all urls in input array (as key) with its replacements (values).
+     *
+     * Custom function because str_replace could replace occurrences from previous replacements
+     *   or urls that are child/parent urls of other urls that need to be replaced
+     */
+    public function replaceUrls(string $text, array $urls): string
+    {
+        // urls contain the original urls as key, and replacements as value
+        $urlsOnly = array_keys($urls);
+
+        // sort by length longest to shortest
+        usort($urlsOnly, function ($a, $b) {
+            return strlen($b) - strlen($a);
+        });
+
+        // create index of all positions of urls, where a position can only be taken by the longest url (child/parent)
+        $byUrl = [];
+        $byPosition = [];
+        foreach ($urlsOnly as $url) {
+            $byUrl[$url] = [];
+            // get first match
+            $position = strpos($text, $url, 0);
+            while (false !== $position) {
+                if (!isset($byPosition[$position])) {
+                    // position not already matched with longer url
+                    $byUrl[$url][] = $position;
+                    $byPosition[$position] = $url;
+                }
+                // find next occurrence
+                $position = strpos($text, $url, $position + strlen($url));
+            }
+        }
+
+        // start replacements back to front to not mess up earlier positions
+        krsort($byPosition, SORT_NUMERIC);
+        foreach ($byPosition as $position => $url) {
+            $text = substr($text, 0, $position).
+                    $urls[$url].
+                    substr($text, $position + strlen($url))
+            ;
+        }
+
+        return $text;
+    }
+
+    public function transformUrl(string $url): string
+    {
+        // parse URL into parts
+        $urlParts = parse_url($url);
+
+        // find matching hostpattern
+        $matchedFormat = '';
+        foreach ($this->hostFormats as $hostFormat => $key) {
+            if (preg_match($hostFormat, $urlParts['host'])) {
+                $matchedFormat = $key;
+                break;
+            }
+        }
+
+        if ($matchedFormat) {
+            // split by comma. Shift first element off and use as key to identify type of link.
+            $params = explode(',', $matchedFormat);
+            $key = array_shift($params);
+            // have we configured a partner id for this program?
+            if (isset($this->partnerIds[$key][0]) && $this->partnerIds[$key][0]) {
+                if (count($this->partnerIds[$key]) > 1) {
+                    // select random id from array
+                    $partnerId = $this->partnerIds[$key][array_rand($this->partnerIds[$key])];
+                } else {
+                    $partnerId = $this->partnerIds[$key][0];
+                }
+
+                switch ($key) {
+                    case 'amazon':
+                        // append id as tag parameter
+                        if (isset($urlParts['query'])) {
+                            $url .= '&tag='.$partnerId;
+                        } else {
+                            $url .= '?tag='.$partnerId;
+                        }
+                        break;
+
+                    case 'bol':
+                        // generate text link to partner program and append original URL encoded
+                        $url = 'https://partner.bol.com/click/click?p=1&t=url&s='.$partnerId.'&f=TXL&url='.urlencode($url);
+                        break;
+
+                    case 'tradetracker':
+                        // params[0] should contain campaignid, append original URL encoded
+                        $url = 'https://tc.tradetracker.net/?c='.$params[0].'&m=12&a='.$partnerId.'&r=&u='.urlencode($urlParts['path']);
+                        if (isset($urlParts['query'])) {
+                            $url .= urlencode('?'.$urlParts['query']);
+                        }
+                        break;
+
+                    case 'awin':
+                        // params[0] should contain merchantid, append original URL encoded
+                        $url = 'https://www.awin1.com/cread.php?awinmid='.$params[0].'&awinaffid='.$this->partnerIds[$key].'&ued='.urlencode($url);
+                        break;
+
+                    default:
+                        // No matching format
+                }
+            }
+        }
+
+        return $url;
+    }
+}

src/Twig/LinkifyExtension.php

@@ -2,11 +2,16 @@
 
 namespace App\Twig;
 
+use App\Service\UrlTransformerService;
 use Twig\Extension\AbstractExtension;
 use Twig\TwigFilter;
 
 class LinkifyExtension extends AbstractExtension
 {
+    public function __construct(private UrlTransformerService $urlTransformerService)
+    {
+    }
+
     public function getFilters(): array
     {
         return [
@@ -21,19 +26,14 @@ public function linkifyFilter($html)
             return $html;
         }
 
-        // Selects all urls starting with ://
-        $html = preg_replace(
-            '~[[:alpha:]]+://[^<>[:space:]]+[[:alnum:]/=!\?,]~',
-            '<a href="\\0" target="_blank" rel="noopener noreferrer">\\0</a>',
-            $html
-        );
-
-        // Selects all urls starting with www. but do not start with ://
-        $html = preg_replace(
-            '~(?<!://)www.[^<>[:space:]]+[[:alnum:]/=!\?,]~',
-            '<a href="http://\\0" target="_blank" rel="noopener noreferrer">\\0</a>',
-            $html
-        );
+        // extract, transform, create links and replace
+        $urls = $this->urlTransformerService->extractUrls($html);
+        $replacements = [];
+        foreach ($urls as $url) {
+            $replacement = $this->urlTransformerService->transformUrl($url);
+            $replacements[$url] = '<a href="'.$replacement.'" target="_blank" rel="noopener noreferrer">'.$url.'</a>';
+        }
+        $html = $this->urlTransformerService->replaceUrls($html, $replacements);
 
         return $html;
     }

tests/php/unit/Service/UrlTransformerTest.php

@@ -0,0 +1,49 @@
+<?php
+
+namespace App\Tests\Unit\Service;
+
+use App\Service\UrlTransformerService;
+use PHPUnit\Framework\TestCase;
+
+class UrlTransformerTest extends TestCase
+{
+    // Simple party with simple excludes
+    public function testUrlExtraction()
+    {
+        $urlTransformer = new UrlTransformerService();
+
+        $urls = $urlTransformer->extractUrls('https://www.test.com');
+        $this->assertEquals(1, count($urls));
+        $this->assertEquals('https://www.test.com', $urls[0]);
+    }
+
+    public function testAmazonUrlExtraction()
+    {
+        $_ENV['PARTNER_AMAZON'] = 'abc-123';
+        $urlTransformer = new UrlTransformerService();
+
+        $url = $urlTransformer->transformUrl('https://www.amazon.com/Zmart-Funny-Christmas-Coworkers-Secret/dp/B0CC1S12S3');
+        $this->assertEquals('https://www.amazon.com/Zmart-Funny-Christmas-Coworkers-Secret/dp/B0CC1S12S3?tag=abc-123', $url);
+
+        $url = $urlTransformer->transformUrl('https://www.amazon.com/Zmart-Funny-Christmas-Coworkers-Secret/dp/B0CC1S12S3?crid=123456789');
+        $this->assertEquals('https://www.amazon.com/Zmart-Funny-Christmas-Coworkers-Secret/dp/B0CC1S12S3?crid=123456789&tag=abc-123', $url);
+    }
+
+    public function testurlReplacement()
+    {
+        $urlTransformer = new UrlTransformerService();
+
+        $html = 'Multiple links to https://www.tom.be, again https://www.tom.be and a child https://www.tom.be/zeb and a child https://www.tom.be/arne';
+        $replacements = [
+            'https://www.tom.be' => 'https://www.tom.be/zeb',
+            'https://www.tom.be/zeb' => 'https://www.tom.be/Zeb',
+            'https://www.tom.be/arne' => 'https://www.tom.be/Arne',
+        ];
+        // It should not replace the /zeb instances that are the result of the replacements of https://www.tom.be.
+        // It should also not replace https://www.tom.be in any of the other child urls, but instead replace them with the capital names.
+        $expectedHtml = 'Multiple links to https://www.tom.be/zeb, again https://www.tom.be/zeb and a child https://www.tom.be/Zeb and a child https://www.tom.be/Arne';
+
+        $actual = $urlTransformer->replaceUrls($html, $replacements);
+        $this->assertEquals($expectedHtml, $actual);
+    }
+}

tests/php/unit/Twig/LinkifyExtensionTest.php

@@ -2,6 +2,7 @@
 
 namespace App\Tests\Unit\Twig;
 
+use App\Service\UrlTransformerService;
 use App\Twig\LinkifyExtension;
 use PHPUnit\Framework\TestCase;
 
@@ -14,7 +15,7 @@ class LinkifyExtensionTest extends TestCase
      */
     public function itShouldWrapAnHtmlLinkInAnAnchorTag($rawHtml, $anchorHtml)
     {
-        $filter = new LinkifyExtension();
+        $filter = new LinkifyExtension(new UrlTransformerService());
 
         $this->assertEquals($anchorHtml, $filter->linkifyFilter($rawHtml));
     }
@@ -36,7 +37,7 @@ public function htmlProvider()
             ],
             [
                 'www.test.com',
-                '<a href="http://www.test.com" target="_blank" rel="noopener noreferrer">www.test.com</a>',
+                'www.test.com',
             ],
             [
                 'http://test',