Updates for Mercurial's HTTP protocol

Summary:
While testing https://secure.phabricator.com/D21864 I ran into some issues getting mercurial HTTP access working. Using wireshark I confirmed that my local mercurial 6.4 was not including command arguments as HTTP headers but in the querystring.

I didn't dig too deep into understanding when/why this started happening. The protocol documents this in [[ https://repo.mercurial-scm.org/hg/file/tip/mercurial/helptext/internals/wireprotocol.txt | wireprotocol.txt ]].

>Command arguments can be sent multiple ways. The simplest is part of the URL query string using ``x-www-form-urlencoded`` encoding (see Python's ``urllib.urlencode()``. However, many servers impose length limitations on the URL. So this mechanism is typically only used if the server doesn't support other mechanisms.

Based on that either the mercurial on the server is really old (it's 6.1.1 tho) or maybe some other parsing/info passing in Phab's handling of the wire protocol is causing the client to downgrade the wire protocol support.

Test Plan: Host mercurial repo using HTTP, test push/pull.

Reviewers: #blessed_reviewers, epriestley

Reviewed By: #blessed_reviewers, epriestley

Subscribers: Korvin, epriestley

Differential Revision: https://secure.phabricator.com/D21867

Author: neandrake

src/applications/diffusion/controller/DiffusionServeController.php

@@ -886,10 +886,29 @@ private function getMercurialArguments() {
       }
       $args_raw[] = $_SERVER[$header];
     }
-    $args_raw = implode('', $args_raw);
 
-    return id(new PhutilQueryStringParser())
-      ->parseQueryString($args_raw);
+    if ($args_raw) {
+      $args_raw = implode('', $args_raw);
+      return id(new PhutilQueryStringParser())
+        ->parseQueryString($args_raw);
+    }
+
+    // Sometimes arguments come in via the query string. Note that this will
+    // not handle multi-value entries e.g. "a[]=1,a[]=2" however it's unclear
+    // whether or how the mercurial protocol should handle this.
+    $query = idx($_SERVER, 'QUERY_STRING', '');
+    $query_pairs = id(new PhutilQueryStringParser())
+      ->parseQueryString($query);
+    foreach ($query_pairs as $key => $value) {
+      // Filter out private/internal keys as well as the command itself.
+      if (strncmp($key, '__', 2) && $key != 'cmd') {
+        $args_raw[$key] = $value;
+      }
+    }
+
+    // TODO: Arguments can also come in via request body for POST requests. The
+    // body would be all arguments, url-encoded.
+    return $args_raw;
   }
 
   private function formatMercurialArguments($command, array $arguments) {