From 36b85a1017b967a1cbb1bcce4de9f207a6888b11 Mon Sep 17 00:00:00 2001
From: Gregor Jasny <gjasny@googlemail.com>
Date: Mon, 18 Dec 2023 23:34:56 +0100
Subject: [PATCH] fix(util): avoid sign extension in base64 encodesr

---
 util/include/prometheus/detail/base64.h | 14 +++++++-------
 util/tests/unit/base64_test.cc          | 16 ++++++++++++++++
 2 files changed, 23 insertions(+), 7 deletions(-)

diff --git a/util/include/prometheus/detail/base64.h b/util/include/prometheus/detail/base64.h
index 4adf6100..21c3e31d 100644
--- a/util/include/prometheus/detail/base64.h
+++ b/util/include/prometheus/detail/base64.h
@@ -37,9 +37,9 @@ inline std::string base64_encode(const std::string& input) {
   auto it = input.begin();
 
   for (std::size_t i = 0; i < input.size() / 3; ++i) {
-    temp = (*it++) << 16;
-    temp += (*it++) << 8;
-    temp += (*it++);
+    temp = static_cast<std::uint8_t>(*it++) << 16;
+    temp += static_cast<std::uint8_t>(*it++) << 8;
+    temp += static_cast<std::uint8_t>(*it++);
     encoded.append(1, kEncodeLookup[(temp & 0x00FC0000) >> 18]);
     encoded.append(1, kEncodeLookup[(temp & 0x0003F000) >> 12]);
     encoded.append(1, kEncodeLookup[(temp & 0x00000FC0) >> 6]);
@@ -48,14 +48,14 @@ inline std::string base64_encode(const std::string& input) {
 
   switch (input.size() % 3) {
     case 1:
-      temp = (*it++) << 16;
+      temp = static_cast<std::uint8_t>(*it++) << 16;
       encoded.append(1, kEncodeLookup[(temp & 0x00FC0000) >> 18]);
       encoded.append(1, kEncodeLookup[(temp & 0x0003F000) >> 12]);
       encoded.append(2, kPadCharacter);
       break;
     case 2:
-      temp = (*it++) << 16;
-      temp += (*it++) << 8;
+      temp = static_cast<std::uint8_t>(*it++) << 16;
+      temp += static_cast<std::uint8_t>(*it++) << 8;
       encoded.append(1, kEncodeLookup[(temp & 0x00FC0000) >> 18]);
       encoded.append(1, kEncodeLookup[(temp & 0x0003F000) >> 12]);
       encoded.append(1, kEncodeLookup[(temp & 0x00000FC0) >> 6]);
@@ -118,7 +118,7 @@ inline std::string base64_decode(const std::string& input) {
 
     decoded.push_back((temp >> 16) & 0x000000FF);
     decoded.push_back((temp >> 8) & 0x000000FF);
-    decoded.push_back((temp)&0x000000FF);
+    decoded.push_back((temp) & 0x000000FF);
   }
 
   return decoded;
diff --git a/util/tests/unit/base64_test.cc b/util/tests/unit/base64_test.cc
index 1f05410a..7efba320 100644
--- a/util/tests/unit/base64_test.cc
+++ b/util/tests/unit/base64_test.cc
@@ -13,6 +13,12 @@ struct TestVector {
 };
 
 const TestVector testVector[] = {
+    // RFC 3548 examples
+    {"\x14\xfb\x9c\x03\xd9\x7e", "FPucA9l+"},
+    {"\x14\xfb\x9c\x03\xd9", "FPucA9k="},
+    {"\x14\xfb\x9c\x03", "FPucAw=="},
+
+    // RFC 4648 examples
     {"", ""},
     {"f", "Zg=="},
     {"fo", "Zm8="},
@@ -20,6 +26,16 @@ const TestVector testVector[] = {
     {"foob", "Zm9vYg=="},
     {"fooba", "Zm9vYmE="},
     {"foobar", "Zm9vYmFy"},
+
+    // Wikipedia examples
+    {"sure.", "c3VyZS4="},
+    {"sure", "c3VyZQ=="},
+    {"sur", "c3Vy"},
+    {"su", "c3U="},
+    {"leasure.", "bGVhc3VyZS4="},
+    {"easure.", "ZWFzdXJlLg=="},
+    {"asure.", "YXN1cmUu"},
+    {"sure.", "c3VyZS4="},
 };
 
 using namespace testing;