From fb1b182fbf0ee5afed586f77eec84d7a906831ef Mon Sep 17 00:00:00 2001
From: VirtualTam <virtualtam@flibidi.net>
Date: Mon, 3 Dec 2018 00:23:35 +0100
Subject: namespacing: \Shaarli\Http\Url

Signed-off-by: VirtualTam <virtualtam@flibidi.net>
---
 application/http/Url.php | 217 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 217 insertions(+)
 create mode 100644 application/http/Url.php

(limited to 'application/http/Url.php')

diff --git a/application/http/Url.php b/application/http/Url.php
new file mode 100644
index 00000000..260231c6
--- /dev/null
+++ b/application/http/Url.php
@@ -0,0 +1,217 @@
+<?php
+
+namespace Shaarli\Http;
+
+/**
+ * URL representation and cleanup utilities
+ *
+ * Form
+ *   scheme://[username:password@]host[:port][/path][?query][#fragment]
+ *
+ * Examples
+ *   http://username:password@hostname:9090/path?arg1=value1&arg2=value2#anchor
+ *   https://host.name.tld
+ *   https://h2.g2/faq/?vendor=hitchhiker&item=guide&dest=galaxy#answer
+ *
+ * @see http://www.faqs.org/rfcs/rfc3986.html
+ */
+class Url
+{
+    private static $annoyingQueryParams = array(
+        // Facebook
+        'action_object_map=',
+        'action_ref_map=',
+        'action_type_map=',
+        'fb_',
+        'fb=',
+        'PHPSESSID=',
+
+        // Scoop.it
+        '__scoop',
+
+        // Google Analytics & FeedProxy
+        'utm_',
+
+        // ATInternet
+        'xtor=',
+
+        // Other
+        'campaign_'
+    );
+
+    private static $annoyingFragments = array(
+        // ATInternet
+        'xtor=RSS-',
+
+        // Misc.
+        'tk.rss_all'
+    );
+
+    /*
+     * URL parts represented as an array
+     *
+     * @see http://php.net/parse_url
+     */
+    protected $parts;
+
+    /**
+     * Parses a string containing a URL
+     *
+     * @param string $url a string containing a URL
+     */
+    public function __construct($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 $input;
+    }
+
+    /**
+     * Returns a string representation of this URL
+     */
+    public function toString()
+    {
+        return unparse_url($this->parts);
+    }
+
+    /**
+     * Removes undesired query parameters
+     */
+    protected function cleanupQuery()
+    {
+        if (!isset($this->parts['query'])) {
+            return;
+        }
+
+        $queryParams = explode('&', $this->parts['query']);
+
+        foreach (self::$annoyingQueryParams as $annoying) {
+            foreach ($queryParams as $param) {
+                if (startsWith($param, $annoying)) {
+                    $queryParams = array_diff($queryParams, array($param));
+                    continue;
+                }
+            }
+        }
+
+        if (count($queryParams) == 0) {
+            unset($this->parts['query']);
+            return;
+        }
+
+        $this->parts['query'] = implode('&', $queryParams);
+    }
+
+    /**
+     * Removes undesired fragments
+     */
+    protected function cleanupFragment()
+    {
+        if (!isset($this->parts['fragment'])) {
+            return;
+        }
+
+        foreach (self::$annoyingFragments as $annoying) {
+            if (startsWith($this->parts['fragment'], $annoying)) {
+                unset($this->parts['fragment']);
+                break;
+            }
+        }
+    }
+
+    /**
+     * Removes undesired query parameters and fragments
+     *
+     * @return string the string representation of this URL after cleanup
+     */
+    public function cleanup()
+    {
+        $this->cleanupQuery();
+        $this->cleanupFragment();
+        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;
+    }
+}
-- 
cgit v1.2.3


From 51753e403fa69c0ce124ede27d300477e3e799ca Mon Sep 17 00:00:00 2001
From: VirtualTam <virtualtam@flibidi.net>
Date: Mon, 3 Dec 2018 00:34:53 +0100
Subject: namespacing: move HTTP utilities along \Shaarli\Http\ classes

Signed-off-by: VirtualTam <virtualtam@flibidi.net>
---
 application/http/Url.php | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'application/http/Url.php')

diff --git a/application/http/Url.php b/application/http/Url.php
index 260231c6..90444a2f 100644
--- a/application/http/Url.php
+++ b/application/http/Url.php
@@ -206,7 +206,7 @@ class Url
     }
 
     /**
-     * Test if the Url is an HTTP one.
+     * Test if the UrlUtils is an HTTP one.
      *
      * @return true is HTTP, false otherwise.
      */
-- 
cgit v1.2.3