Avoid reallocations in base64 encoding (#6951)

Author: SiarheiFedartsou

CHANGELOG.md

@@ -24,6 +24,7 @@
     - NodeJS:
       - CHANGED: Use node-api instead of NAN. [#6452](https://github.com/Project-OSRM/osrm-backend/pull/6452)
     - Misc:
+      - CHANGED: Avoid reallocations in base64 encoding. [#6951](https://github.com/Project-OSRM/osrm-backend/pull/6951)
       - CHANGED: Get rid of unused Boost dependencies. [#6960](https://github.com/Project-OSRM/osrm-backend/pull/6960)
       - CHANGED: Apply micro-optimisation for Table & Trip APIs. [#6949](https://github.com/Project-OSRM/osrm-backend/pull/6949)
       - CHANGED: Apply micro-optimisation for Route API. [#6948](https://github.com/Project-OSRM/osrm-backend/pull/6948)

include/engine/base64.hpp

@@ -47,24 +47,29 @@ namespace engine
 // Encodes a chunk of memory to Base64.
 inline std::string encodeBase64(const unsigned char *first, std::size_t size)
 {
-    std::vector<unsigned char> bytes{first, first + size};
-    BOOST_ASSERT(!bytes.empty());
+    BOOST_ASSERT(size > 0);
 
-    std::size_t bytes_to_pad{0};
+    std::string encoded;
+    encoded.reserve(((size + 2) / 3) * 4);
 
-    while (bytes.size() % 3 != 0)
+    auto padding = (3 - size % 3) % 3;
+
+    BOOST_ASSERT(padding == 0 || padding == 1 || padding == 2);
+
+    for (auto itr = detail::Base64FromBinary(first); itr != detail::Base64FromBinary(first + size);
+         ++itr)
     {
-        bytes_to_pad += 1;
-        bytes.push_back(0);
+        encoded.push_back(*itr);
     }
 
-    BOOST_ASSERT(bytes_to_pad == 0 || bytes_to_pad == 1 || bytes_to_pad == 2);
-    BOOST_ASSERT_MSG(0 == bytes.size() % 3, "base64 input data size is not a multiple of 3");
+    for (size_t index = 0; index < padding; ++index)
+    {
+        encoded.push_back('=');
+    }
 
-    std::string encoded{detail::Base64FromBinary{bytes.data()},
-                        detail::Base64FromBinary{bytes.data() + (bytes.size() - bytes_to_pad)}};
+    BOOST_ASSERT(encoded.size() == (size + 2) / 3 * 4);
 
-    return encoded.append(bytes_to_pad, '=');
+    return encoded;
 }
 
 // C++11 standard 3.9.1/1: Plain char, signed char, and unsigned char are three distinct types

unit_tests/engine/base64.cpp

@@ -74,4 +74,98 @@ BOOST_AUTO_TEST_CASE(hint_encoding_decoding_roundtrip_bytewise)
                            reinterpret_cast<const unsigned char *>(&decoded)));
 }
 
+BOOST_AUTO_TEST_CASE(long_string_encoding)
+{
+    using namespace osrm::engine;
+    std::string long_string(1000, 'A'); // String of 1000 'A's
+    std::string encoded = encodeBase64(long_string);
+    BOOST_CHECK_EQUAL(decodeBase64(encoded), long_string);
+}
+
+BOOST_AUTO_TEST_CASE(invalid_base64_decoding)
+{
+    using namespace osrm::engine;
+    BOOST_CHECK_THROW(decodeBase64("Invalid!"), std::exception);
+}
+
+BOOST_AUTO_TEST_CASE(hint_serialization_size)
+{
+    using namespace osrm::engine;
+    using namespace osrm::util;
+
+    const Coordinate coordinate;
+    const PhantomNode phantom;
+    const osrm::test::MockDataFacade<osrm::engine::routing_algorithms::ch::Algorithm> facade{};
+
+    const SegmentHint hint{phantom, facade.GetCheckSum()};
+    const auto base64 = hint.ToBase64();
+
+    BOOST_CHECK_EQUAL(base64.size(), 112);
+}
+
+BOOST_AUTO_TEST_CASE(extended_roundtrip_tests)
+{
+    using namespace osrm::engine;
+
+    std::vector<std::string> test_strings = {
+        "Hello, World!",                    // Simple ASCII string
+        "1234567890",                       // Numeric string
+        "!@#$%^&*()_+",                     // Special characters
+        std::string(1000, 'A'),             // Long repeating string
+        "¡Hola, mundo!",                    // Non-ASCII characters
+        "こんにちは、世界！",               // Unicode characters
+        std::string("\x00\x01\x02\x03", 4), // Binary data
+        "a",                                // Single character
+        "ab",                               // Two characters
+        "abc",                              // Three characters (no padding in Base64)
+        std::string(190, 'x')               // String that doesn't align with Base64 padding
+    };
+
+    for (const auto &test_str : test_strings)
+    {
+        std::string encoded = encodeBase64(test_str);
+        std::string decoded = decodeBase64(encoded);
+        BOOST_CHECK_EQUAL(decoded, test_str);
+
+        // Additional checks
+        BOOST_CHECK(encoded.find_first_not_of(
+                        "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=") ==
+                    std::string::npos);
+        if (test_str.length() % 3 != 0)
+        {
+            BOOST_CHECK(encoded.back() == '=');
+        }
+    }
+}
+
+BOOST_AUTO_TEST_CASE(roundtrip_with_url_safe_chars)
+{
+    using namespace osrm::engine;
+
+    std::string original = "Hello+World/Nothing?Is:Impossible";
+    std::string encoded = encodeBase64(original);
+
+    // Replace '+' with '-' and '/' with '_'
+    std::replace(encoded.begin(), encoded.end(), '+', '-');
+    std::replace(encoded.begin(), encoded.end(), '/', '_');
+
+    std::string decoded = decodeBase64(encoded);
+    BOOST_CHECK_EQUAL(decoded, original);
+}
+
+BOOST_AUTO_TEST_CASE(roundtrip_stress_test)
+{
+    using namespace osrm::engine;
+
+    std::string test_str;
+    for (int i = 0; i < 1000; ++i)
+    {
+        test_str += static_cast<char>(i % 256);
+    }
+
+    std::string encoded = encodeBase64(test_str);
+    std::string decoded = decodeBase64(encoded);
+    BOOST_CHECK_EQUAL(decoded, test_str);
+}
+
 BOOST_AUTO_TEST_SUITE_END()