[Maintenance] Move DieLibrary git tag handling to a separate file and expose it to Python

.dielib_commit

@@ -0,0 +1 @@
+09df9ccafe48a0531987ad1e605402ed79d4c3f6

cmake/FindDieLibrary.cmake

@@ -65,10 +65,12 @@ list(INSERT CMAKE_MODULE_PATH 0
 
 find_package(Qt6 REQUIRED COMPONENTS Core Qml Concurrent)
 
+file(STRINGS "${ROOT_DIR}/.dielib_commit" DIE_LIBRARY_GIT_TAG)
+message(STATUS "Using tag ${DIE_LIBRARY_GIT_TAG} for DieLibrary")
 FetchContent_Declare(
   DieLibrary
   GIT_REPOSITORY "https://github.com/horsicq/die_library"
-  GIT_TAG 09df9ccafe48a0531987ad1e605402ed79d4c3f6
+  GIT_TAG "${DIE_LIBRARY_GIT_TAG}"
 )
 
 set(DIE_BUILD_AS_STATIC ON CACHE INTERNAL "")

python/CMakeLists.txt

@@ -60,6 +60,7 @@ target_compile_definitions(_die
     PRIVATE
         DIE_VERSION="${DIE_VERSION}"
         DIELIB_VERSION="${DIELIB_VERSION}"
+        DIELIB_TAG="${DIE_LIBRARY_GIT_TAG}"
 )
 
 target_link_libraries(_die PRIVATE $<TARGET_FILE:die> $<TARGET_PROPERTY:die,LINK_LIBRARIES>)

python/die/__init__.py

@@ -2,7 +2,7 @@
 import pathlib
 import warnings
 
-from typing import Generator, Optional, Union
+from typing import Generator
 
 from ._die import __version__  # ty:ignore[unresolved-import]
 from ._die import DieFlags as _DieFlags  # ty:ignore[unresolved-import]
@@ -13,9 +13,9 @@
     ScanMemoryExA as _ScanMemoryExA,
     LoadDatabaseA as _LoadDatabaseA,
 )
-from ._die import die_version, dielib_version  # ty:ignore[unresolved-import]
+from ._die import die_version, dielib_version, dielib_tag  # ty:ignore[unresolved-import]
 
-__all__ = ["die_version", "dielib_version"]
+__all__ = ["die_version", "dielib_version", "dielib_tag"]
 version_major, version_minor, version_patch = map(int, __version__.split("."))
 
 
@@ -108,7 +108,7 @@ def iterdir(self):
 
 
 # Initialize database path with smart handling
-database_path = _DatabasePath(__path__[0]) / "db"
+database_path: pathlib.Path = _DatabasePath(__path__[0]) / "db"
 """Path to the DIE signature database
 
 This path automatically points to the correct database location,
@@ -138,8 +138,8 @@ class ScanFlags(enum.IntFlag):
 
 
 def scan_file(
-    filepath: Union[pathlib.Path, str], flags: ScanFlags, database: Optional[str] = None
-) -> Optional[str]:
+    filepath: pathlib.Path | str, flags: ScanFlags, database: str | None = None
+) -> str | None:
     """
     Scan the given file against the signature database, if specified
 
@@ -192,8 +192,8 @@ def __enum_db(root: pathlib.Path) -> Generator[pathlib.Path, None, None]:
 
 
 def scan_memory(
-    memory: Union[bytes, bytearray], flags: ScanFlags, database: Optional[str] = None
-) -> Optional[str]:
+    memory: bytes | bytearray, flags: ScanFlags, database: str | None = None
+) -> str | None:
     """
     Scan the given sequence of bytes against the signature database, if specified
 
@@ -220,11 +220,14 @@ def scan_memory(
     return res.strip()
 
 
-def load_database(database: str) -> int:
+def load_database(database: str | pathlib.Path) -> int:
     """
     Load a database
     """
-    if not isinstance(database, str):
+    if isinstance(database, pathlib.Path):
+        database = str(database)
+
+    elif not isinstance(database, str):
         raise TypeError
 
     return _LoadDatabaseA(database)

python/inc/die.hpp

@@ -14,6 +14,10 @@
 #define DIE_VERSION ""
 #endif // DIE_VERSION
 
+#ifndef DIELIB_TAG
+#define DIELIB_TAG ""
+#endif // DIELIB_TAG
+
 #ifdef __cplusplus
 namespace DIE
 {

python/src/die.cpp

@@ -126,6 +126,7 @@ NB_MODULE(_die, m)
     m.attr("__version__")    = "0.5.0";
     m.attr("die_version")    = DIE_VERSION;
     m.attr("dielib_version") = DIELIB_VERSION;
+    m.attr("dielib_tag")     = DIELIB_TAG;
 
     m.def("ScanFileA", DIE::ScanFileA, "filename"_a, "flags"_a, "database"_a, "Scan a file against known signatures");
 

python/tests/test_die.py

@@ -17,6 +17,8 @@ def test_constants():
     assert die.die_version
     assert isinstance(die.dielib_version, str)
     assert die.dielib_version
+    assert isinstance(die.dielib_tag, str)
+    assert die.dielib_tag
 
     # validate die database
     assert isinstance(die.database_path, die._DatabasePath)
@@ -117,9 +119,11 @@ def test_scan_export_format_xml(target_binary: pathlib.Path) -> None:
     assert xml.Result
     if platform.system() == "Windows":
         assert hasattr(xml.Result, "PE64")
+        assert xml.Result.PE64
         assert xml.Result.PE64["filetype"] == "PE64"
     elif platform.system() == "Linux":
         assert hasattr(xml.Result, "ELF64")
+        assert xml.Result.ELF64
         assert xml.Result.ELF64["filetype"] == "ELF64"
 
 
@@ -170,16 +174,18 @@ def test_database_path_backward_compatibility():
     # Test 2: database_path should resolve to a valid location with PE/ directory
     db_path = pathlib.Path(path_new)
     assert db_path.exists(), f"Database path does not exist: {db_path}"
-    assert (db_path / 'PE').exists(), f"PE directory not found at {db_path}"
+    assert (db_path / "PE").exists(), f"PE directory not found at {db_path}"
 
     # Test 3: Old usage with /'db' should work through smart path resolution
     # The smart path should detect the version and handle accordingly
     with warnings.catch_warnings(record=True) as w:
         warnings.simplefilter("always")
-        path_old = str(die.database_path / 'db')
+        path_old = str(die.database_path / "db")
 
         # The path should exist in both old and new versions
-        assert pathlib.Path(path_old).exists(), f"Old usage path doesn't exist: {path_old}"
+        assert pathlib.Path(path_old).exists(), (
+            f"Old usage path doesn't exist: {path_old}"
+        )
 
         if len(w) > 0:
             # New fixed version: got deprecation warning
@@ -200,13 +206,14 @@ def test_database_path_resolves_correctly():
     db_path = pathlib.Path(str(die.database_path))
 
     # Check for PE directory (main signature database)
-    assert (db_path / 'PE').exists(), f"PE directory not found at {db_path}"
+    assert (db_path / "PE").exists(), f"PE directory not found at {db_path}"
 
     # Check for other expected directories
-    expected_dirs = ['PE', 'ELF', 'MACH']
+    expected_dirs = ["PE", "ELF", "MACH"]
     for dir_name in expected_dirs:
-        assert (db_path / dir_name).exists(), \
+        assert (db_path / dir_name).exists(), (
             f"Expected directory {dir_name} not found at {db_path}"
+        )
 
 
 def test_scan_with_explicit_database_path(target_binary: pathlib.Path):
@@ -230,7 +237,7 @@ def test_scan_with_explicit_database_path(target_binary: pathlib.Path):
         res = die.scan_file(
             target_binary,
             die.ScanFlags.DEEP_SCAN,
-            database=str(die.database_path / 'db'),
+            database=str(die.database_path / "db"),
         )
         assert res
         assert isinstance(res, str)