Merge pull request #1234 from virtualtam/lint

[github/shaarli/Shaarli.git] / application / Url.php

diff --git a/application/Url.php b/application/Url.php
index 23356f39d20514ade3f2391458678d6cb7b650c1..3b7f19c207770679d2fff77a709210529ffa2afa 100644 (file)
--- a/application/Url.php
+++ b/application/Url.php
@@ -25,6 +25,68 @@ function unparse_url($parsedUrl)
     return "$scheme$user$pass$host$port$path$query$fragment";
 }
 
+/**
+ * Removes undesired query parameters and fragments
+ *
+ * @param string url Url to be cleaned
+ *
+ * @return string the string representation of this URL after cleanup
+ */
+function cleanup_url($url)
+{
+    $obj_url = new Url($url);
+    return $obj_url->cleanup();
+}
+
+/**
+ * Get URL scheme.
+ *
+ * @param string url Url for which the scheme is requested
+ *
+ * @return mixed the URL scheme or false if none is provided.
+ */
+function get_url_scheme($url)
+{
+    $obj_url = new Url($url);
+    return $obj_url->getScheme();
+}
+
+/**
+ * Adds a trailing slash at the end of URL if necessary.
+ *
+ * @param string $url URL to check/edit.
+ *
+ * @return string $url URL with a end trailing slash.
+ */
+function add_trailing_slash($url)
+{
+    return $url . (!endsWith($url, '/') ? '/' : '');
+}
+
+/**
+ * Replace not whitelisted protocols by 'http://' from given URL.
+ *
+ * @param string $url       URL to clean
+ * @param array  $protocols List of allowed protocols (aside from http(s)).
+ *
+ * @return string URL with allowed protocol
+ */
+function whitelist_protocols($url, $protocols)
+{
+    if (startsWith($url, '?') || startsWith($url, '/')) {
+        return $url;
+    }
+    $protocols = array_merge(['http', 'https'], $protocols);
+    $protocol = preg_match('#^(\w+):/?/?#', $url, $match);
+    // Protocol not allowed: we remove it and replace it with http
+    if ($protocol === 1 && ! in_array($match[1], $protocols)) {
+        $url = str_replace($match[0], 'http://', $url);
+    } elseif ($protocol !== 1) {
+        $url = 'http://' . $url;
+    }
+    return $url;
+}
+
 /**
  * URL representation and cleanup utilities
  *
@@ -47,6 +109,7 @@ class Url
         'action_type_map=',
         'fb_',
         'fb=',
+        'PHPSESSID=',
 
         // Scoop.it
         '__scoop',
@@ -55,7 +118,10 @@ class Url
         'utm_',
 
         // ATInternet
-        'xtor='
+        'xtor=',
+
+        // Other
+        'campaign_'
     );
 
     private static $annoyingFragments = array(
@@ -80,13 +146,47 @@ class Url
      */
     public function __construct($url)
     {
+        $url = self::cleanupUnparsedUrl(trim($url));
         $this->parts = parse_url($url);
+
+        if (!empty($url) && empty($this->parts['scheme'])) {
+            $this->parts['scheme'] = 'http';
+        }
     }
 
+    /**
+     * Clean up URL before it's parsed.
+     * ie. handle urlencode, url prefixes, etc.
+     *
+     * @param string $url URL to clean.
+     *
+     * @return string cleaned URL.
+     */
+    protected static function cleanupUnparsedUrl($url)
+    {
+        return self::removeFirefoxAboutReader($url);
+    }
+
+    /**
+     * Remove Firefox Reader prefix if it's present.
+     *
+     * @param string $input url
+     *
+     * @return string cleaned url
+     */
+    protected static function removeFirefoxAboutReader($input)
+    {
+        $firefoxPrefix = 'about://reader?url=';
+        if (startsWith($input, $firefoxPrefix)) {
+            return urldecode(ltrim($input, $firefoxPrefix));
+        }
+        return $input;
+    }
+    
     /**
      * Returns a string representation of this URL
      */
-    public function __toString()
+    public function toString()
     {
         return unparse_url($this->parts);
     }
@@ -117,7 +217,7 @@ class Url
         }
 
         $this->parts['query'] = implode('&', $queryParams);
-    }    
+    }
 
     /**
      * Removes undesired fragments
@@ -145,6 +245,58 @@ class Url
     {
         $this->cleanupQuery();
         $this->cleanupFragment();
-        return $this->__toString();
+        return $this->toString();
+    }
+
+    /**
+     * Converts an URL with an International Domain Name host to a ASCII one.
+     * This requires PHP-intl. If it's not available, just returns this->cleanup().
+     *
+     * @return string converted cleaned up URL.
+     */
+    public function idnToAscii()
+    {
+        $out = $this->cleanup();
+        if (! function_exists('idn_to_ascii') || ! isset($this->parts['host'])) {
+            return $out;
+        }
+        $asciiHost = idn_to_ascii($this->parts['host'], 0, INTL_IDNA_VARIANT_UTS46);
+        return str_replace($this->parts['host'], $asciiHost, $out);
+    }
+
+    /**
+     * Get URL scheme.
+     *
+     * @return string the URL scheme or false if none is provided.
+     */
+    public function getScheme()
+    {
+        if (!isset($this->parts['scheme'])) {
+            return false;
+        }
+        return $this->parts['scheme'];
+    }
+
+    /**
+     * Get URL host.
+     *
+     * @return string the URL host or false if none is provided.
+     */
+    public function getHost()
+    {
+        if (empty($this->parts['host'])) {
+            return false;
+        }
+        return $this->parts['host'];
+    }
+
+    /**
+     * Test if the Url is an HTTP one.
+     *
+     * @return true is HTTP, false otherwise.
+     */
+    public function isHttp()
+    {
+        return strpos(strtolower($this->parts['scheme']), 'http') !== false;
     }
 }