StorageUtils: Add generation-aware rehashing ID file map

Such a file could be used to store mappings of digests from CAS or
Git cache to digests of different hash type that represent same
content.

5 files changed, 39 insertions, 12 deletions

diff --git a/src/buildtool/crypto/hash_function.hpp b/src/buildtool/crypto/hash_function.hpp
index 9e5a9c0b..530e0b2b 100644
--- a/src/buildtool/crypto/hash_function.hpp
+++ b/src/buildtool/crypto/hash_function.hpp
@@ -95,4 +95,15 @@ class HashFunction {
         -> std::optional<std::pair<Hasher::HashDigest, std::uintmax_t>>;
 };
 
+[[nodiscard]] constexpr auto ToString(HashFunction::Type type) noexcept -> const
+    char* {
+    switch (type) {
+        case HashFunction::Type::GitSHA1:
+            return "git-SHA1";
+        case HashFunction::Type::PlainSHA256:
+            return "plain-SHA256";
+    }
+    Ensures(false);  // unreachable
+}
+
 #endif  // INCLUDED_SRC_BUILDTOOL_CRYPTO_HASH_FUNCTION_HPP
diff --git a/src/buildtool/crypto/hash_info.cpp b/src/buildtool/crypto/hash_info.cpp
index 4fb87161..e185d214 100644
--- a/src/buildtool/crypto/hash_info.cpp
+++ b/src/buildtool/crypto/hash_info.cpp
@@ -21,16 +21,6 @@
 #include "src/utils/cpp/hex_string.hpp"
 
 namespace {
-[[nodiscard]] inline auto GetHashTypeName(HashFunction::Type type) noexcept
-    -> std::string {
-    switch (type) {
-        case HashFunction::Type::GitSHA1:
-            return "GitSHA1";
-        case HashFunction::Type::PlainSHA256:
-            return "PlainSHA256";
-    }
-    Ensures(false);
-}
 
 inline constexpr auto kSHA1EmptyGitBlobHash =
     "e69de29bb2d1d6434b8b29ae775ad8c2e48c5391";
@@ -91,7 +81,7 @@ auto HashInfo::ValidateInput(HashFunction::Type type,
             "HashInfo: hash {} is expected to be {}.\nTrees are "
             "not allowed in this mode.",
             hash,
-            GetHashTypeName(type));
+            ToString(type));
     }
 
     if (auto const exp_size = HashFunction{type}.MakeHasher().GetHashLength();
@@ -100,7 +90,7 @@ auto HashInfo::ValidateInput(HashFunction::Type type,
             "HashInfo: hash {} is expected to be {}.\n It must have a length "
             "of {}, but its length is {}.",
             hash,
-            GetHashTypeName(type),
+            ToString(type),
             exp_size,
             hash.size());
     }
diff --git a/src/buildtool/storage/TARGETS b/src/buildtool/storage/TARGETS
index 79af1305..73356708 100644
--- a/src/buildtool/storage/TARGETS
+++ b/src/buildtool/storage/TARGETS
@@ -103,6 +103,7 @@
     [ "config"
     , "storage"
     , ["src/buildtool/common", "user_structs"]
+    , ["src/buildtool/crypto", "hash_function"]
     , ["src/buildtool/file_system/symlinks_map", "pragma_special"]
     ]
   , "stage": ["src", "buildtool", "storage"]
diff --git a/src/buildtool/storage/fs_utils.cpp b/src/buildtool/storage/fs_utils.cpp
index 171a5d1c..05883814 100644
--- a/src/buildtool/storage/fs_utils.cpp
+++ b/src/buildtool/storage/fs_utils.cpp
@@ -104,6 +104,16 @@ auto GetResolvedTreeIDFile(StorageConfig const& storage_config,
            tree_hash;
 }
 
+auto GetRehashIDFile(StorageConfig const& storage_config,
+                     HashFunction::Type target_hash_type,
+                     std::string const& hash,
+                     bool from_git,
+                     std::size_t generation) noexcept -> std::filesystem::path {
+    return storage_config.GenerationCacheRoot(generation) /
+           fmt::format("to-{}", ToString(target_hash_type)) /
+           (from_git ? "from-git" : "from-cas") / hash;
+}
+
 auto WriteTreeIDFile(std::filesystem::path const& tree_id_file,
                      std::string const& tree_id) noexcept -> bool {
     // needs to be done safely, so use the rename trick
diff --git a/src/buildtool/storage/fs_utils.hpp b/src/buildtool/storage/fs_utils.hpp
index 7482457b..25392b91 100644
--- a/src/buildtool/storage/fs_utils.hpp
+++ b/src/buildtool/storage/fs_utils.hpp
@@ -20,6 +20,7 @@
 #include <string>
 
 #include "src/buildtool/common/user_structs.hpp"
+#include "src/buildtool/crypto/hash_function.hpp"
 #include "src/buildtool/file_system/symlinks_map/pragma_special.hpp"
 #include "src/buildtool/storage/config.hpp"
 #include "src/buildtool/storage/storage.hpp"
@@ -73,6 +74,20 @@ namespace StorageUtils {
                                          std::size_t generation = 0) noexcept
     -> std::filesystem::path;
 
+/// \brief Get the path to the file storing the corresponding artifact hashed by
+/// a different hash function.
+/// \param storage_config  Storage under which the file is to be found.
+/// \param target_hash_type  Hash type to identify mapping target.
+/// \param hash  Hash to identify mapping source.
+/// \param from_git  Flag to distinguish further mapping source (CAS / GitCAS)
+/// \param generation  Further specificity in location of the file.
+[[nodiscard]] auto GetRehashIDFile(StorageConfig const& storage_config,
+                                   HashFunction::Type target_hash_type,
+                                   std::string const& hash,
+                                   bool from_git,
+                                   std::size_t generation = 0) noexcept
+    -> std::filesystem::path;
+
 /// \brief Write a tree id to file. The parent folder of the file must exist!
 [[nodiscard]] auto WriteTreeIDFile(std::filesystem::path const& tree_id_file,
                                    std::string const& tree_id) noexcept -> bool;