[k2] fix support http content-encoding on ob_gzhandler (#1523)

Author: astrophysik

runtime-light/server/http/http-server-state.h

@@ -59,6 +59,7 @@ struct HttpServerInstanceState final : private vk::not_copyable {
 
   std::optional<string> opt_raw_post_data;
 
+  bool auto_encoding_enabled{};
   uint32_t encoding{};
   uint64_t status_code{kphp::http::status::NO_STATUS};
   kphp::http::method http_method{kphp::http::method::other};

runtime-light/server/http/init-functions.cpp

@@ -212,10 +212,11 @@ string get_http_response_body(HttpServerInstanceState& http_server_instance_st)
     const auto appender{[&body](const auto& buffer) noexcept { body.append(buffer.buffer(), buffer.size()); }};
     std::ranges::for_each(user_buffers | std::views::filter([](const auto& buffer) noexcept { return buffer.size() > 0; }), appender);
 
+    const bool auto_encoding_enabled{http_server_instance_st.auto_encoding_enabled};
     const bool gzip_encoded{static_cast<bool>(http_server_instance_st.encoding & HttpServerInstanceState::ENCODING_GZIP)};
     const bool deflate_encoded{static_cast<bool>(http_server_instance_st.encoding & HttpServerInstanceState::ENCODING_DEFLATE)};
     // compress body if needed
-    if (gzip_encoded || deflate_encoded) {
+    if (auto_encoding_enabled && (gzip_encoded || deflate_encoded)) {
       auto encoded_body{kphp::zlib::encode({body.c_str(), static_cast<size_t>(body.size())}, kphp::zlib::DEFAULT_COMPRESSION_LEVEL,
                                            gzip_encoded ? kphp::zlib::ENCODING_GZIP : kphp::zlib::ENCODING_DEFLATE)};
       if (encoded_body.has_value()) [[likely]] {
@@ -389,9 +390,10 @@ kphp::coro::task<> finalize_server() noexcept {
     }
     [[fallthrough]];
   case kphp::http::response_state::sending_headers: {
+    const bool auto_encoding_enabled{http_server_instance_st.auto_encoding_enabled};
     const bool gzip_encoded{static_cast<bool>(http_server_instance_st.encoding & HttpServerInstanceState::ENCODING_GZIP)};
     const bool deflate_encoded{static_cast<bool>(http_server_instance_st.encoding & HttpServerInstanceState::ENCODING_DEFLATE)};
-    if (gzip_encoded || deflate_encoded) {
+    if (auto_encoding_enabled && (gzip_encoded || deflate_encoded)) {
       auto& static_SB{RuntimeContext::get().static_SB};
       static_SB.clean() << kphp::http::headers::CONTENT_ENCODING.data() << ": " << (gzip_encoded ? ENCODING_GZIP.data() : ENCODING_DEFLATE.data());
       kphp::http::header({static_SB.c_str(), static_SB.size()}, true, kphp::http::status::NO_STATUS);

runtime-light/stdlib/output/output-buffer.cpp

@@ -24,7 +24,7 @@ void f$ob_start(const string& callback) noexcept {
   if (!callback.empty()) {
     if (current_buffering_level == 0 && std::string_view{callback.c_str(), callback.size()} == ob_gzhandler_name) {
       auto& http_server_instance_st{HttpServerInstanceState::get()};
-      http_server_instance_st.encoding |= HttpServerInstanceState::ENCODING_GZIP;
+      http_server_instance_st.auto_encoding_enabled = true;
     } else {
       kphp::log::error("unsupported callback {} at buffering level {}", callback.c_str(), output_instance_st.output_buffers.user_level());
     }

runtime-light/stdlib/output/output-buffer.h

@@ -42,7 +42,9 @@ inline bool f$ob_end_clean() noexcept {
 
   output_instance_st.output_buffers.prev_user_buffer();
   auto& http_server_instance_st{HttpServerInstanceState::get()};
-  http_server_instance_st.encoding &= ~HttpServerInstanceState::ENCODING_GZIP;
+  if (const auto current_buffering_level{output_instance_st.output_buffers.user_level()}; current_buffering_level == 0) {
+    http_server_instance_st.auto_encoding_enabled = false;
+  }
   return true;
 }
 
@@ -56,7 +58,10 @@ inline Optional<string> f$ob_get_clean() noexcept {
   string result{(*opt_user_buffer).get().str()};
   output_instance_st.output_buffers.prev_user_buffer();
   auto& http_server_instance_st{HttpServerInstanceState::get()};
-  http_server_instance_st.encoding &= ~HttpServerInstanceState::ENCODING_GZIP;
+  if (const auto current_buffering_level{output_instance_st.output_buffers.user_level()}; current_buffering_level == 0) {
+    http_server_instance_st.auto_encoding_enabled = false;
+  }
+
   return result;
 }
 

tests/python/tests/http_server/php/index.php

@@ -190,6 +190,23 @@ public function work(string $output) {
         echo 'OK';
         ob_end_flush();
         break;
+      case "gzhandler-after-reset":
+        ob_start("ob_gzhandler");
+        echo 'OK';
+        ob_end_flush();
+
+        ob_start("ob_gzhandler");
+        break;
+      case "gzhandler-with-nested-buffer":
+        ob_start("ob_gzhandler");
+        echo 'OK';
+        ob_start();
+        ob_end_clean();
+
+        break;
+      case "gzhandler-absent":
+        echo 'OK';
+        break;
       case "ignore-second-handler":
         header('Transfer-Encoding: chunked');
         $chunk = "OK";

tests/python/tests/http_server/test_gzip_header_reset.py

@@ -13,12 +13,36 @@ def test_single_gzip_buffer_closed(self):
         with self.assertRaises(KeyError):
             _ = response.headers["Content-Encoding"]
 
+    def test_gzip_handler_after_reset(self):
+        response = self.gzip_request("gzhandler-after-reset")
+        self.assertEqual(response.headers["Content-Encoding"], "gzip")
+
+    def test_deflate_handler_after_reset(self):
+        response = self.deflate_request("gzhandler-after-reset")
+        self.assertEqual(response.headers["Content-Encoding"], "deflate")
+
+    def test_gzip_handler_with_nested_buffer(self):
+        response = self.gzip_request("gzhandler-with-nested-buffer")
+        self.assertEqual(response.headers["Content-Encoding"], "gzip")
+
+    def test_gzip_without_handler(self):
+        response = self.gzip_request("gzhandler-absent")
+        with self.assertRaises(KeyError):
+            _ = response.headers["Content-Encoding"]
+
     @pytest.mark.k2_skip
     def test_ignore_second_gzip_handler(self):
         response = self.gzip_request("ignore-second-handler")
         with self.assertRaises(KeyError):
             _ = response.headers["Content-Encoding"]
 
+    def deflate_request(self, type):
+        url = "/test_script_gzip_header?type=" + type
+        return self.web_server.http_get(url, headers={
+            "Host": "localhost",
+            "Accept-Encoding": "deflate"
+        })
+
     def gzip_request(self, type):
         url = "/test_script_gzip_header?type=" + type
         return self.web_server.http_get(url, headers={