Support IDNA - internationalized domain names

[phancap.git] / src / phancap / Options.php

diff --git a/src/phancap/Options.php b/src/phancap/Options.php
index d39af73218fafc31009f53db328a9f865ec33622..2cfec7e23cd276021befc035dd08e8fa2f4e4a2e 100644 (file)
--- a/src/phancap/Options.php
+++ b/src/phancap/Options.php
@@ -116,6 +116,8 @@ class Options
     public $values = array();
 
     /**
+     * Configuration object
+     *
      * @var Config
      */
     protected $config;
@@ -152,11 +154,12 @@ class Options
                 $this->values[$name] = $this->validateUrl($arValues[$name]);
             } else if ($arOption['type'] == 'int') {
                 $this->values[$name] = $this->validateInt(
-                    $arValues[$name], $arOption['min'], $arOption['max']
+                    $name, $arValues[$name],
+                    $arOption['min'], $arOption['max']
                 );
             } else if (gettype($arOption['type']) == 'array') {
                 $this->values[$name] = $this->validateArray(
-                    $arValues[$name], $arOption['type']
+                    $name, $arValues[$name], $arOption['type']
                 );
             } else if ($arOption['type'] == 'age') {
                 $this->values[$name] = $this->clamp(
@@ -283,17 +286,18 @@ class Options
     /**
      * Check that a given value exists in an array
      *
+     * @param string $name    Variable name
      * @param string $value   Value to check
      * @param array  $options Array of allowed values
      *
      * @return string Value
      * @throws \InvalidArgumentException If the value does not exist in $options
      */
-    protected function validateArray($value, $options)
+    protected function validateArray($name, $value, $options)
     {
         if (array_search($value, $options) === false) {
             throw new \InvalidArgumentException(
-                'Invalid value ' . $value . '.'
+                'Invalid value ' . $value . ' for ' . $name . '.'
                 . ' Allowed: ' . implode(', ', $options)
             );
         }
@@ -303,6 +307,7 @@ class Options
     /**
      * Validate that a value is numeric and between $min and $max (inclusive)
      *
+     * @param string  $name  Variable name
      * @param string  $value Value to check
      * @param integer $min   Minimum allowed value
      * @param integer $max   Maximum allowed value
@@ -310,11 +315,11 @@ class Options
      * @return integer Value as integer
      * @throws \InvalidArgumentException When outside range or not numeric
      */
-    protected function validateInt($value, $min, $max)
+    protected function validateInt($name, $value, $min, $max)
     {
         if (!is_numeric($value)) {
             throw new \InvalidArgumentException(
-                'Value must be a number'
+                $name . ' value must be a number'
             );
         }
         $value = (int) $value;
@@ -331,6 +336,9 @@ class Options
      */
     protected function validateUrl($url)
     {
+        if ($url === '') {
+            throw new \InvalidArgumentException('URL is empty');
+        }
         $parts = parse_url($url);
         if ($parts === false) {
             throw new \InvalidArgumentException('Invalid URL');
@@ -344,6 +352,33 @@ class Options
         if (!isset($parts['host'])) {
             throw new \InvalidArgumentException('URL host missing');
         }
+
+        $rebuild = false;
+        if (strlen(preg_replace('#[[:ascii:]]#', '', $parts['host']))) {
+            //non-ascii characters in the host name
+            $host = idn_to_ascii($parts['host']);
+            if ($host === false) {
+                //incoming URL was not UTF-8 but some ISO dialect
+                $host = idn_to_ascii(utf8_encode($parts['host']));
+                if ($host === false) {
+                    throw new \InvalidArgumentException(
+                        'Strange characters in host name'
+                    );
+                }
+            }
+            $parts['host'] = $host;
+            $rebuild = true;
+        }
+        if (strlen(preg_replace('#[[:ascii:]]#', '', $parts['path']))) {
+            //non-ascii characters in the path
+            $parts['path'] = str_replace('%2F', '/', urlencode($parts['path']));
+            $rebuild = true;
+        }
+
+        if ($rebuild) {
+            $url = static::http_build_url($parts);
+        }
+
         return $url;
     }
 
@@ -360,5 +395,36 @@ class Options
         $this->options['smaxage']['default'] = $this->config->screenshotMaxAge;
         $this->options['smaxage']['min']     = $this->config->screenshotMinAge;
     }
+
+    /**
+     * Re-build an URL parts array generated by parse_url()
+     *
+     * @param string $parts Array of URL parts
+     *
+     * @return string URL
+     */
+    protected static function http_build_url($parts)
+    {
+        $scheme   = isset($parts['scheme'])
+            ? $parts['scheme'] . '://' : '';
+        $host     = isset($parts['host'])
+            ? $parts['host'] : '';
+        $port     = isset($parts['port'])
+            ? ':' . $parts['port'] : '';
+        $user     = isset($parts['user'])
+            ? $parts['user'] : '';
+        $pass     = isset($parts['pass'])
+
+            ? ':' . $parts['pass']  : '';
+        $pass     = ($user || $pass)
+            ? "$pass@" : '';
+        $path     = isset($parts['path'])
+            ? $parts['path'] : '';
+        $query    = isset($parts['query'])
+            ? '?' . $parts['query'] : '';
+        $fragment = isset($parts['fragment'])
+            ? '#' . $parts['fragment'] : '';
+        return "$scheme$user$pass$host$port$path$query$fragment";
+    }
 }
 ?>