Introduce MD5 from libmd, use it for hashing queuedLines for park path

4 files changed, 32 insertions, 2 deletions

diff --git a/Jamroot.jam b/Jamroot.jam
index 2f38c94..37e4a0e 100644
--- a/Jamroot.jam
+++ b/Jamroot.jam
@@ -13,6 +13,7 @@ lib adhocutil : : <link>shared : : <include>/usr/include/adhocutil <library>glib
 lib dbppcore : : <link>shared : : <include>/usr/include/dbpp <library>adhocutil ;
 lib dbpp-postgresql : : <link>shared : : <include>/usr/include/dbpp-postgresql <library>dbppcore ;
 lib z : : <link>shared ;
+lib md : : <link>shared ;
 
 project webstat : requirements
 	<cxxstd>26
diff --git a/src/Jamfile.jam b/src/Jamfile.jam
index 9459dd8..eca0c24 100644
--- a/src/Jamfile.jam
+++ b/src/Jamfile.jam
@@ -4,6 +4,7 @@ lib webstat : [ glob *.cpp : *_main.cpp ] :
 	<library>..//dbppcore
 	<library>../thirdparty//scn
 	<library>..//z
+	<library>..//md
 	<library>..//curl
 	: :
 	<include>.
diff --git a/src/ingestor.cpp b/src/ingestor.cpp
index 0659f1f..5ae2487 100644
--- a/src/ingestor.cpp
+++ b/src/ingestor.cpp
@@ -24,6 +24,32 @@ namespace DB {
 
 namespace WebStat {
 	namespace {
+		using ByteArrayView = std::span<const uint8_t>;
+
+		auto
+		bytesToHexRange(const ByteArrayView bytes)
+		{
+			constexpr auto HEXN = 16ZU;
+			return bytes | std::views::transform([](auto byte) {
+				return std::array {byte / HEXN, byte % HEXN};
+			}) | std::views::join
+					| std::views::transform([](auto nibble) {
+						  return "0123456789abcdef"[nibble];
+					  });
+		}
+
+		EntityHash
+		makeHash(const std::string_view value)
+		{
+			MD5_CTX ctx {};
+			MD5Init(&ctx);
+			// NOLINTNEXTLINE(cppcoreguidelines-pro-type-reinterpret-cast) - correct for md5ing raw bytes
+			MD5Update(&ctx, reinterpret_cast<const uint8_t *>(value.data()), value.length());
+			EntityHash hash {};
+			MD5Final(hash.data(), &ctx);
+			return hash;
+		}
+
 		Crc32Value
 		crc32(const std::string_view value)
 		{
@@ -281,8 +307,8 @@ namespace WebStat {
 		if (queuedLines.empty()) {
 			return std::unexpected(0);
 		}
-		const std::filesystem::path path {
-				settings.fallbackDir / std::format("parked-{}.short", crc32(queuedLines.front()))};
+		const std::filesystem::path path {settings.fallbackDir
+				/ std::format("parked-{:s}.short", bytesToHexRange(makeHash(queuedLines.front())))};
 		if (auto parked = FilePtr(fopen(path.c_str(), "w"))) {
 			fprintf(parked.get(), "%zu\n", queuedLines.size());
 			for (const auto & line : queuedLines) {
diff --git a/src/logTypes.hpp b/src/logTypes.hpp
index 71393b2..6556d5c 100644
--- a/src/logTypes.hpp
+++ b/src/logTypes.hpp
@@ -1,5 +1,6 @@
 #pragma once
 
+#include <md5.h>
 #include <optional>
 #include <scn/scan.h>
 #include <string>
@@ -35,6 +36,7 @@ namespace WebStat {
 	};
 
 	using Crc32Value = uint32_t;
+	using EntityHash = std::array<uint8_t, MD5_DIGEST_LENGTH>;
 	using Entity = std::tuple<Crc32Value, EntityType, std::string_view>;
 }