Cleaning up no_proxy support

docs/request-options.rst

@@ -638,14 +638,24 @@ Pass a string to specify a proxy for all protocols.
     $client->get('/', ['proxy' => 'tcp://localhost:8125']);
 
 Pass an associative array to specify HTTP proxies for specific URI schemes
-(i.e., "http", "https").
+(i.e., "http", "https"). Provide a ``no`` key value pair to provide a list of
+host names that should not be proxied to.
+
+.. note::
+
+    Guzzle will automatically populate this value with your environment's
+    ``NO_PROXY`` environment variable. However, when providing a ``proxy``
+    request option, it is up to your to provide the ``no`` value parsed from
+    the ``NO_PROXY`` environment variable
+    (e.g., ``explode(',', getenv('NO_PROXY'))``).
 
 .. code-block:: php
 
     $client->get('/', [
         'proxy' => [
             'http'  => 'tcp://localhost:8125', // Use this proxy with "http"
-            'https' => 'tcp://localhost:9124'  // Use this proxy with "https"
+            'https' => 'tcp://localhost:9124', // Use this proxy with "https",
+            'no' => ['.mit.edu', 'foo.com']    // Don't use a proxy with these
         ]
     ]);
 

src/Handler/CurlFactory.php

@@ -361,14 +361,16 @@ class CurlFactory implements CurlFactoryInterface
         if (isset($options['proxy'])) {
             if (!is_array($options['proxy'])) {
                 $conf[CURLOPT_PROXY] = $options['proxy'];
-            } elseif ($scheme = $easy->request->getUri()->getScheme()) {
+            } else {
-            	$host = $easy->request->getUri()->getHost();
+                $scheme = $easy->request->getUri()->getScheme();
-            	if (!isset($options['proxy']['no']) 
+                if (isset($options['proxy'][$scheme])) {
-        	       ||!$this->isHostInProxyArea($host, $options['proxy']['no'])) {
+                    $host = $easy->request->getUri()->getHost();
-                    if (isset($options['proxy'][$scheme])) {
+                    if (!isset($options['proxy']['no']) ||
+                        !\GuzzleHttp\is_host_in_noproxy($host, $options['proxy']['no'])
+                    ) {
                         $conf[CURLOPT_PROXY] = $options['proxy'][$scheme];
-                    }        		
+                    }
-            	}
+                }
             }
         }
 
@@ -423,27 +425,6 @@ class CurlFactory implements CurlFactoryInterface
             $conf[CURLOPT_VERBOSE] = true;
         }
     }
-    
-    private function isHostInProxyArea($host, array $areas)
-    {
-        $matches = false;
-        foreach ($areas as $area) {
-            $areaToMatch = $area;
-            // if include all subdomains 
-            if (!empty($area) && substr($area, 0, 1) === '.'){
-                $areaToMatch = "*$area";
-            }
-            // Use wilcards
-            $delimiter = '#';
-            $areaToMatch = preg_quote($areaToMatch, $delimiter);
-            $areaToMatch = str_replace('\*', '.*', $areaToMatch);  
-            if (preg_match("$delimiter^$areaToMatch$$delimiter", $host)) {
-                $matches = true;
-                break;
-            }
-        }
-        return $matches;
-    }
 
     /**
      * This function ensures that a response was set on a transaction. If one

src/Handler/StreamHandler.php

@@ -296,7 +296,14 @@ class StreamHandler
         } else {
             $scheme = $request->getUri()->getScheme();
             if (isset($value[$scheme])) {
-                $options['http']['proxy'] = $value[$scheme];
+                if (!isset($value['no'])
+                    || !\GuzzleHttp\is_host_in_noproxy(
+                        $request->getUri()->getHost(),
+                        $value['no']
+                    )
+                ) {
+                    $options['http']['proxy'] = $value[$scheme];
+                }
             }
         }
     }

src/functions.php

@@ -225,3 +225,56 @@ function normalize_header_keys(array $headers)
 
     return $result;
 }
+
+/**
+ * Returns true if the provided host matches any of the no proxy areas.
+ *
+ * This method will strip a port from the host if it is present. Each pattern
+ * can be matched with an exact match (e.g., "foo.com" == "foo.com") or a
+ * partial match: (e.g., "foo.com" == "baz.foo.com" and ".foo.com" ==
+ * "baz.foo.com", but ".foo.com" != "foo.com").
+ *
+ * Areas are matched in the following cases:
+ * 1. "*" (without quotes) always matches any hosts.
+ * 2. An exact match.
+ * 3. The area starts with "." and the area is the last part of the host. e.g.
+ *    '.mit.edu' will match any host that ends with '.mit.edu'.
+ *
+ * @param string $host         Host to check against the patterns.
+ * @param array  $noProxyArray An array of host patterns.
+ *
+ * @return bool
+ */
+function is_host_in_noproxy($host, array $noProxyArray)
+{
+    if (strlen($host) === 0) {
+        throw new \InvalidArgumentException('Empty host provided');
+    }
+
+    // Strip port if present.
+    if (strpos($host, ':')) {
+        $host = explode($host, ':', 2)[0];
+    }
+
+    foreach ($noProxyArray as $area) {
+        // Always match on wildcards.
+        if ($area === '*') {
+            return true;
+        } elseif (empty($area)) {
+            // Don't match on empty values.
+            continue;
+        } elseif ($area === $host) {
+            // Exact matches.
+            return true;
+        } else {
+            // Special match if the area when prefixed with ".". Remove any
+            // existing leading "." and add a new leading ".".
+            $area = '.' . ltrim($area, '.');
+            if (substr($host, -(strlen($area))) === $area) {
+                return true;
+            }
+        }
+    }
+
+    return false;
+}

tests/Handler/CurlFactoryTest.php

@@ -145,25 +145,28 @@ class CurlFactoryTest extends \PHPUnit_Framework_TestCase
         $this->assertEquals('http://bar.com', $_SERVER['_curl'][CURLOPT_PROXY]);
         $this->checkNoProxyForHost('http://test.test.com', ['test.test.com'], false);
         $this->checkNoProxyForHost('http://test.test.com', ['.test.com'], false);
-        $this->checkNoProxyForHost('http://test.test.com', ['*.test.com'], false);
+        $this->checkNoProxyForHost('http://test.test.com', ['*.test.com'], true);
-        $this->checkNoProxyForHost('http://test.test.com', ['*.test2.com'], true);
+        $this->checkNoProxyForHost('http://test.test.com', ['*'], false);
-        $this->checkNoProxyForHost('http://127.0.0.1', ['127.0.0.*'], false);
+        $this->checkNoProxyForHost('http://127.0.0.1', ['127.0.0.*'], true);
     }
-    
+
-    
     private function checkNoProxyForHost($url, $noProxy, $assertUseProxy)
     {
         $f = new Handler\CurlFactory(3);
         $f->create(new Psr7\Request('GET', $url), [
-            'proxy' => ['http' => 'http://bar.com', 'https' => 'https://t', 'no' => $noProxy],
+            'proxy' => [
+                'http' => 'http://bar.com',
+                'https' => 'https://t',
+                'no' => $noProxy
+            ],
         ]);
-        if($assertUseProxy) {
+        if ($assertUseProxy) {
             $this->assertArrayHasKey(CURLOPT_PROXY, $_SERVER['_curl']);
-        }else{
+        } else {
             $this->assertArrayNotHasKey(CURLOPT_PROXY, $_SERVER['_curl']);
         }
     }
-    
+
 
     /**
      * @expectedException \InvalidArgumentException

tests/Handler/StreamHandlerTest.php

@@ -196,6 +196,17 @@ class StreamHandlerTest extends \PHPUnit_Framework_TestCase
         $this->assertEquals($url, $opts['http']['proxy']);
     }
 
+    public function testAddsProxyButHonorsNoProxy()
+    {
+        $url = str_replace('http', 'tcp', Server::$url);
+        $res = $this->getSendResult(['proxy' => [
+            'http' => $url,
+            'no'   => ['*']
+        ]]);
+        $opts = stream_context_get_options($res->getBody()->detach());
+        $this->assertTrue(empty($opts['http']['proxy']));
+    }
+
     public function testAddsTimeout()
     {
         $res = $this->getSendResult(['stream' => true, 'timeout' => 200]);

tests/functionsTest.php

@@ -70,6 +70,37 @@ class FunctionsTest extends \PHPUnit_Framework_TestCase
     {
         $this->assertFileExists(GuzzleHttp\default_ca_bundle());
     }
+
+    public function noProxyProvider()
+    {
+        return [
+            ['mit.edu', ['.mit.edu'], false],
+            ['foo.mit.edu', ['.mit.edu'], true],
+            ['mit.edu', ['mit.edu'], true],
+            ['mit.edu', ['baz', 'mit.edu'], true],
+            ['mit.edu', ['', '', 'mit.edu'], true],
+            ['mit.edu', ['baz', '*'], true],
+        ];
+    }
+
+    /**
+     * @dataProvider noproxyProvider
+     */
+    public function testChecksNoProxyList($host, $list, $result)
+    {
+        $this->assertSame(
+            $result,
+            \GuzzleHttp\is_host_in_noproxy($host, $list)
+        );
+    }
+
+    /**
+     * @expectedException \InvalidArgumentException
+     */
+    public function testEnsuresNoProxyCheckHostIsSet()
+    {
+        \GuzzleHttp\is_host_in_noproxy('', []);
+    }
 }
 
 final class StrClass