From 09d51fdf2fe64d8e020f348b87b4a2af3424b5bc Mon Sep 17 00:00:00 2001
From: server <server@jakubkadlec.dev>
Date: Wed, 15 Apr 2026 15:43:25 +0200
Subject: [PATCH] Optimize archive compression selection

---
 docs/client-integration.md |  5 +++++
 docs/format.md             |  7 ++++---
 src/archive.cpp            | 29 +++++++++++++++++++++++++++--
 3 files changed, 36 insertions(+), 5 deletions(-)

diff --git a/docs/client-integration.md b/docs/client-integration.md
index 59292dc..ba56103 100644
--- a/docs/client-integration.md
+++ b/docs/client-integration.md
@@ -54,6 +54,7 @@ Minimum validation:
 - reject duplicate paths
 - reject path traversal
 - verify AEAD tag before decompression
+- verify plaintext hash only in debug or explicit strict mode
 
 ## Launcher key delivery
 
@@ -73,6 +74,7 @@ Supported inputs, highest priority last:
 --m2pack-pubkey-hex <64-hex-public-key>
 --m2pack-key-map <mapping-name>
 --m2pack-key-id <integer>
+--m2pack-strict-hash [0|1]
 ```
 
 ### Environment
@@ -82,6 +84,7 @@ M2PACK_MASTER_KEY_HEX
 M2PACK_SIGN_PUBKEY_HEX
 M2PACK_KEY_MAP
 M2PACK_KEY_ID
+M2PACK_STRICT_HASH
 ```
 
 ### Shared memory
@@ -119,6 +122,8 @@ Recommended production path:
 - Keep a path-to-entry map in lowercase normalized form.
 - Decrypt per request, not by unpacking the full archive.
 - Keep a small decompression scratch buffer pool if the client reads in parallel.
+- Skip plaintext hash verification in release hot paths unless `M2PACK_STRICT_HASH`
+  or `--m2pack-strict-hash` explicitly enables it.
 
 ## Release model
 
diff --git a/docs/format.md b/docs/format.md
index 52028cd..8e410fa 100644
--- a/docs/format.md
+++ b/docs/format.md
@@ -43,9 +43,10 @@ this order:
 
 Each file payload is:
 
-1. compressed with `zstd`
-2. encrypted with `XChaCha20-Poly1305`
-3. stored in-place in the archive
+1. compressed with `zstd` when it produces a meaningful size win
+2. otherwise stored as plaintext bytes
+3. encrypted with `XChaCha20-Poly1305`
+4. stored in-place in the archive
 
 The file path is used as associated data so path tampering invalidates the
 payload authentication tag.
diff --git a/src/archive.cpp b/src/archive.cpp
index b76843f..b27b340 100644
--- a/src/archive.cpp
+++ b/src/archive.cpp
@@ -16,12 +16,36 @@ namespace
 
 constexpr char kArchiveMagic[kMagicSize] = {'M', '2', 'P', 'A', 'C', 'K', '2', '\0'};
 constexpr std::uint32_t kArchiveVersion = 1;
+constexpr std::size_t kCompressionMinSavingsBytes = 64;
+constexpr std::size_t kCompressionMinSavingsPercent = 5;
 
 std::array<std::uint8_t, kAeadKeySize> require_master_key(const KeyMaterial& keys)
 {
     return keys.master_key;
 }
 
+Compression select_compression_mode(
+    const std::vector<std::uint8_t>& plain,
+    const std::vector<std::uint8_t>& compressed)
+{
+    if (compressed.size() >= plain.size())
+    {
+        return Compression::None;
+    }
+
+    const auto savings = plain.size() - compressed.size();
+    const bool meaningful_size_win = savings >= kCompressionMinSavingsBytes;
+    const bool meaningful_ratio_win = !plain.empty() &&
+        (savings * 100 >= plain.size() * kCompressionMinSavingsPercent);
+
+    if (!meaningful_size_win && !meaningful_ratio_win)
+    {
+        return Compression::None;
+    }
+
+    return Compression::Zstd;
+}
+
 }  // namespace
 
 std::string normalize_path(const std::filesystem::path& root, const std::filesystem::path& file)
@@ -143,13 +167,14 @@ BuildResult build_archive(
         const auto compressed = compress_zstd(plain);
         ManifestEntry entry;
         entry.path = normalize_path(input_dir, path);
-        entry.compression = Compression::Zstd;
+        entry.compression = select_compression_mode(plain, compressed);
         entry.data_offset = payload_bytes.size();
         entry.original_size = plain.size();
         entry.nonce = random_nonce();
         entry.plaintext_hash = hash_bytes(plain);
 
-        const auto encrypted = encrypt_payload(compressed, require_master_key(keys), entry.nonce, entry.path);
+        const auto& payload = entry.compression == Compression::Zstd ? compressed : plain;
+        const auto encrypted = encrypt_payload(payload, require_master_key(keys), entry.nonce, entry.path);
         entry.stored_size = encrypted.size();
 
         payload_bytes.insert(payload_bytes.end(), encrypted.begin(), encrypted.end());