Merge pull request #757 from kiwix/gzip_compression

src/server/internalServer.cpp

@@ -78,7 +78,6 @@ extern "C" {
 #include "response.h"
 
 #define MAX_SEARCH_LEN 140
-#define KIWIX_MIN_CONTENT_SIZE_TO_DEFLATE 100
 #define DEFAULT_CACHE_SIZE 2
 
 namespace kiwix {

src/server/request_context.cpp

@@ -75,15 +75,15 @@ RequestContext::RequestContext(struct MHD_Connection* connection,
   method(str2RequestMethod(_method)),
   version(version),
   requestIndex(s_requestIndex++),
-  acceptEncodingDeflate(false),
+  acceptEncodingGzip(false),
   byteRange_()
 {
   MHD_get_connection_values(connection, MHD_HEADER_KIND, &RequestContext::fill_header, this);
   MHD_get_connection_values(connection, MHD_GET_ARGUMENT_KIND, &RequestContext::fill_argument, this);
 
   try {
-    acceptEncodingDeflate =
-        (get_header(MHD_HTTP_HEADER_ACCEPT_ENCODING).find("deflate") != std::string::npos);
+    acceptEncodingGzip =
+        (get_header(MHD_HTTP_HEADER_ACCEPT_ENCODING).find("gzip") != std::string::npos);
   } catch (const std::out_of_range&) {}
 
   try {
@@ -127,7 +127,7 @@ void RequestContext::print_debug_info() const {
   printf("Parsed : \n");
   printf("full_url: %s\n", full_url.c_str());
   printf("url   : %s\n", url.c_str());
-  printf("acceptEncodingDeflate : %d\n", acceptEncodingDeflate);
+  printf("acceptEncodingGzip : %d\n", acceptEncodingGzip);
   printf("has_range : %d\n", byteRange_.kind() != ByteRange::NONE);
   printf("is_valid_url : %d\n", is_valid_url());
   printf(".............\n");

src/server/request_context.h

@@ -92,7 +92,7 @@ class RequestContext {
 
     ByteRange get_range() const;
 
-    bool can_compress() const { return acceptEncodingDeflate; }
+    bool can_compress() const { return acceptEncodingGzip; }
 
     std::string get_user_language() const;
 
@@ -103,7 +103,7 @@ class RequestContext {
     std::string version;
     unsigned long long requestIndex;
 
-    bool acceptEncodingDeflate;
+    bool acceptEncodingGzip;
 
     ByteRange byteRange_;
     std::map<std::string, std::string> headers;

src/server/response.cpp

@@ -31,8 +31,17 @@
 #include <mustache.hpp>
 #include <zlib.h>
 
+#include <array>
 
-#define KIWIX_MIN_CONTENT_SIZE_TO_DEFLATE 100
+// This is somehow a magic value.
+// If this value is too small, we will compress (and lost cpu time) too much
+// content.
+// If this value is too big, we will not compress enough content and send too
+// much data.
+// If we assume that MTU is 1500 Bytes it is useless to compress
+// content smaller as the content will be sent in one packet anyway.
+// 1400 Bytes seems to be a common accepted limit.
+#define KIWIX_MIN_CONTENT_SIZE_TO_COMPRESS 1400
 
 namespace kiwix {
 
@@ -58,6 +67,41 @@ bool is_compressible_mime_type(const std::string& mimeType)
       || mimeType.find("application/json") != string::npos;
 }
 
+bool compress(std::string &content) {
+  z_stream strm;
+  strm.zalloc = Z_NULL;
+  strm.zfree = Z_NULL;
+  strm.opaque = Z_NULL;
+
+  auto ret = deflateInit2(&strm, Z_DEFAULT_COMPRESSION, Z_DEFLATED, 31, 8,
+                          Z_DEFAULT_STRATEGY);
+  if (ret != Z_OK) { return false; }
+
+  strm.avail_in = static_cast<decltype(strm.avail_in)>(content.size());
+  strm.next_in =
+      const_cast<Bytef *>(reinterpret_cast<const Bytef *>(content.data()));
+
+  std::string compressed;
+
+  std::array<char, 16384> buff{};
+  do {
+    strm.avail_out = buff.size();
+    strm.next_out = reinterpret_cast<Bytef *>(buff.data());
+    ret = deflate(&strm, Z_FINISH);
+    assert(ret != Z_STREAM_ERROR);
+    compressed.append(buff.data(), buff.size() - strm.avail_out);
+  } while (strm.avail_out == 0);
+
+  assert(ret == Z_STREAM_END);
+  assert(strm.avail_in == 0);
+
+  content.swap(compressed);
+
+  deflateEnd(&strm);
+  return true;
+}
+
+
 
 } // unnamed namespace
 
@@ -331,7 +375,7 @@ ContentResponse::can_compress(const RequestContext& request) const
 {
   return request.can_compress()
       && is_compressible_mime_type(m_mimeType)
-      && (m_content.size() > KIWIX_MIN_CONTENT_SIZE_TO_DEFLATE);
+      && (m_content.size() > KIWIX_MIN_CONTENT_SIZE_TO_COMPRESS);
 }
 
 bool
@@ -365,35 +409,17 @@ ContentResponse::create_mhd_response(const RequestContext& request)
     }
   }
 
-  bool shouldCompress = can_compress(request);
-  if (shouldCompress) {
-    std::vector<Bytef> compr_buffer(compressBound(m_content.size()));
-    uLongf comprLen = compr_buffer.capacity();
-    int err = compress(&compr_buffer[0],
-                       &comprLen,
-                       (const Bytef*)(m_content.data()),
-                       m_content.size());
-    if (err == Z_OK && comprLen > 2 && comprLen < (m_content.size() + 2)) {
-      /* /!\ Internet Explorer has a bug with deflate compression.
-         It can not handle the first two bytes (compression headers)
-         We need to chunk them off (move the content 2bytes)
-         It has no incidence on other browsers
-         See http://www.subbu.org/blog/2008/03/ie7-deflate-or-not and comments */
-      m_content = string((char*)&compr_buffer[2], comprLen - 2);
-      m_etag.set_option(ETag::COMPRESSED_CONTENT);
-    } else {
-      shouldCompress = false;
-    }
-  }
+  const bool isCompressed = can_compress(request) && compress(m_content);
 
   MHD_Response* response = MHD_create_response_from_buffer(
     m_content.size(), const_cast<char*>(m_content.data()), MHD_RESPMEM_MUST_COPY);
 
-  if (shouldCompress) {
+  if (isCompressed) {
+    m_etag.set_option(ETag::COMPRESSED_CONTENT);
     MHD_add_response_header(
         response, MHD_HTTP_HEADER_VARY, "Accept-Encoding");
     MHD_add_response_header(
-        response, MHD_HTTP_HEADER_CONTENT_ENCODING, "deflate");
+        response, MHD_HTTP_HEADER_CONTENT_ENCODING, "gzip");
   }
   return response;
 }

test/server.cpp

@@ -1,4 +1,5 @@
 
+#define CPPHTTPLIB_ZLIB_SUPPORT 1
 #include "./httplib.h"
 #include "gtest/gtest.h"
 
@@ -184,13 +185,10 @@ const ResourceCollection resources200Compressible{
   { WITH_ETAG, "/ROOT/skin/taskbar.css" },
   { WITH_ETAG, "/ROOT/skin/block_external.js" },
 
-  { NO_ETAG,   "/ROOT/catalog/root.xml" },
-  { NO_ETAG,   "/ROOT/catalog/searchdescription.xml" },
   { NO_ETAG,   "/ROOT/catalog/search" },
 
   { NO_ETAG,   "/ROOT/search?content=zimfile&pattern=a" },
 
-  { NO_ETAG,   "/ROOT/suggest?content=zimfile" },
   { NO_ETAG,   "/ROOT/suggest?content=zimfile&term=ray" },
 
   { NO_ETAG,   "/ROOT/catch/external?source=www.example.com" },
@@ -223,7 +221,10 @@ const ResourceCollection resources200Uncompressible{
   { WITH_ETAG, "/ROOT/corner_cases/-/empty.css" },
   { WITH_ETAG, "/ROOT/corner_cases/-/empty.js" },
 
-  // The title and creator are too small to be compressed
+  // The following url's responses are too small to be compressed
+  { NO_ETAG,   "/ROOT/catalog/root.xml" },
+  { NO_ETAG,   "/ROOT/catalog/searchdescription.xml" },
+  { NO_ETAG,   "/ROOT/suggest?content=zimfile" },
   { WITH_ETAG, "/ROOT/raw/zimfile/meta/Creator" },
   { WITH_ETAG, "/ROOT/raw/zimfile/meta/Title" },
 };
@@ -271,9 +272,9 @@ TEST_F(ServerTest, 200)
 TEST_F(ServerTest, CompressibleContentIsCompressedIfAcceptable)
 {
   for ( const Resource& res : resources200Compressible ) {
-    const auto x = zfs1_->GET(res.url, { {"Accept-Encoding", "deflate"} });
+    const auto x = zfs1_->GET(res.url, { {"Accept-Encoding", "gzip"} });
     EXPECT_EQ(200, x->status) << res;
-    EXPECT_EQ("deflate", x->get_header_value("Content-Encoding")) << res;
+    EXPECT_EQ("gzip", x->get_header_value("Content-Encoding")) << res;
     EXPECT_EQ("Accept-Encoding", x->get_header_value("Vary")) << res;
   }
 }
@@ -281,7 +282,7 @@ TEST_F(ServerTest, CompressibleContentIsCompressedIfAcceptable)
 TEST_F(ServerTest, UncompressibleContentIsNotCompressed)
 {
   for ( const Resource& res : resources200Uncompressible ) {
-    const auto x = zfs1_->GET(res.url, { {"Accept-Encoding", "deflate"} });
+    const auto x = zfs1_->GET(res.url, { {"Accept-Encoding", "gzip"} });
     EXPECT_EQ(200, x->status) << res;
     EXPECT_EQ("", x->get_header_value("Content-Encoding")) << res;
   }
@@ -1062,7 +1063,7 @@ TEST_F(ServerTest, CompressionInfluencesETag)
     if ( ! res.etag_expected ) continue;
     const auto g1 = zfs1_->GET(res.url);
     const auto g2 = zfs1_->GET(res.url, { {"Accept-Encoding", ""} } );
-    const auto g3 = zfs1_->GET(res.url, { {"Accept-Encoding", "deflate"} } );
+    const auto g3 = zfs1_->GET(res.url, { {"Accept-Encoding", "gzip"} } );
     const auto etag = g1->get_header_value("ETag");
     EXPECT_EQ(etag, g2->get_header_value("ETag"));
     EXPECT_NE(etag, g3->get_header_value("ETag"));
@@ -1075,7 +1076,7 @@ TEST_F(ServerTest, ETagOfUncompressibleContentIsNotAffectedByAcceptEncoding)
     if ( ! res.etag_expected ) continue;
     const auto g1 = zfs1_->GET(res.url);
     const auto g2 = zfs1_->GET(res.url, { {"Accept-Encoding", ""} } );
-    const auto g3 = zfs1_->GET(res.url, { {"Accept-Encoding", "deflate"} } );
+    const auto g3 = zfs1_->GET(res.url, { {"Accept-Encoding", "gzip"} } );
     const auto etag = g1->get_header_value("ETag");
     EXPECT_EQ(etag, g2->get_header_value("ETag")) << res;
     EXPECT_EQ(etag, g3->get_header_value("ETag")) << res;
@@ -1114,7 +1115,7 @@ std::string make_etag_list(const std::string& etag)
 
 TEST_F(ServerTest, IfNoneMatchRequestsWithMatchingETagResultIn304Responses)
 {
-  const char* const encodings[] = { "", "deflate" };
+  const char* const encodings[] = { "", "gzip" };
   for ( const Resource& res : all200Resources() ) {
     for ( const char* enc: encodings ) {
       if ( ! res.etag_expected ) continue;
@@ -1245,7 +1246,7 @@ TEST_F(ServerTest, RangeHasPrecedenceOverCompression)
 
   const Headers onlyRange{ {"Range", "bytes=123-456"} };
   Headers rangeAndCompression(onlyRange);
-  rangeAndCompression.insert({"Accept-Encoding", "deflate"});
+  rangeAndCompression.insert({"Accept-Encoding", "gzip"});
 
   const auto p1 = zfs1_->GET(url, onlyRange);
   const auto p2 = zfs1_->GET(url, rangeAndCompression);