Log SVSBench version

pyproject.toml

@@ -1,11 +1,11 @@
 [project]
 name = "svsbench"
-version = "0.1.0"
 description = "Benchmarking for Scalable Vector Search"
 readme = "README.md"
 authors = [
     { name = "Mihai Capotă", email = "mihai.capota@intel.com" }
 ]
+dynamic = ["version"]
 requires-python = ">=3.12"
 dependencies = [
     "numpy>=1.10",
@@ -15,7 +15,7 @@ dependencies = [
 ]
 
 [build-system]
-requires = ["hatchling"]
+requires = ["hatchling", "hatch-vcs"]
 build-backend = "hatchling.build"
 
 [dependency-groups]
@@ -25,5 +25,8 @@ dev = [
     "ruff>=0.9.6",
 ]
 
+[tool.hatch.version]
+source = "vcs"
+
 [tool.ruff]
 line-length = 79

src/svsbench/__init__.py

@@ -1,3 +1,9 @@
 # Copyright (C) 2025 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 """Scalable Vector Search Benchmarking."""
+
+from importlib.metadata import version
+
+if __spec__ is None:
+    raise RuntimeError("Running __init__.py directly is not supported.")
+__version__ = version(__spec__.parent)

src/svsbench/build.py

@@ -129,6 +129,7 @@ def main(argv: list[str] | None = None) -> None:
     )
     print("Logging to", log_file, sep="\n")
     logger.info({"argv": argv if argv else sys.argv})
+    utils.check_uncommitted_and_log_version(logger, args.uncommitted)
     args.out_dir.mkdir(exist_ok=True)
     if args.data_matrix_file is not None:
         if args.query_matrix_file is None:

src/svsbench/compress.py

@@ -38,6 +38,7 @@ def main(argv: str | None = None) -> None:
     )
     print("Logging to", log_file, sep="\n")
     logger.info({"argv": argv if argv else sys.argv})
+    utils.check_uncommitted_and_log_version(logger, args.uncommitted)
     compress(
         idx_dir=args.idx_dir,
         svs_type=args.svs_type,

src/svsbench/generate_ground_truth.py

@@ -46,6 +46,7 @@ def main(argv: str | None = None) -> None:
     )
     print("Logging to", log_file, sep="\n")
     logger.info({"argv": argv if argv else sys.argv})
+    utils.check_uncommitted_and_log_version(logger, args.uncommitted)
     generate_ground_truth(
         vecs_path=args.vecs_file,
         query_file=args.query_file,

src/svsbench/search.py

@@ -341,6 +341,7 @@ def main(argv: str | None = None) -> None:
     )
     print("Logging to", log_file, sep="\n")
     logger.info({"argv": sys.argv})
+    utils.check_uncommitted_and_log_version(logger, args.uncommitted)
     search(
         idx_dir=args.idx_dir,
         vecs_path=args.vecs_file,

src/svsbench/utils.py

@@ -5,13 +5,14 @@
 import os
 from datetime import datetime
 from pathlib import Path
-from typing import Any
+from typing import Any, Final
 from warnings import warn
 
 import svs
 
-from . import consts
+from . import consts, __version__
 
+UNCOMMITTED: Final = "--uncommitted"
 
 class JSONFormatter(logging.Formatter):
     """Formatter that dumps msg and created as JSON."""
@@ -209,6 +210,11 @@ def add_common_arguments(parser):
         choices=consts.SVS_TYPES,
         default="float32",
     )
+    parser.add_argument(
+        UNCOMMITTED,
+        action="store_true",
+        help="Run even if there are changes in SVSBench not committed to git",
+    )
 
 
 def ground_truth_path(
@@ -230,3 +236,31 @@ def ground_truth_path(
             f"{seed if seed is not None else False}.ivecs",
         )
     )
+
+def check_uncommitted_and_log_version(
+    logger: logging.Logger, uncommitted: bool
+) -> None:
+    """Check if SVSBench is using uncommitted code and log version.
+
+    The version has a suffix starting with "d2" when there are
+    uncommitted code changes, e.g., "0.1.1.dev4+gcf3cee1.d20250501".
+    Without uncommitted changes, that suffix is missing, e.g.,
+    "0.1.1.dev4+gcf3cee1".
+
+    Args:
+        logger: Logger to log the version.
+        uncommitted: If True, do not raise an error when uncommitted
+            code changes are detected.
+    Raises:
+        RuntimeError: If uncommitted code changes are detected and
+            `uncommitted` is False.
+    """
+    if "." in __version__ and __version__.rsplit(".", 1)[1].startswith("d2"):
+        if not uncommitted:
+            raise RuntimeError(
+                "Uncommitted changes detected. To ignore this error,"
+                f" add {UNCOMMITTED} to the SVSBench call."
+            )
+        else:
+            logger.warning("uncommitted_code_changes")
+    logger.info({"version": __version__})

tests/test_build.py

@@ -66,6 +66,7 @@ def test_main_with_train_query(tmp_path, tmp_vecs, query_path):
             str(query_path),
             "--out_dir",
             str(tmp_path),
+            "--uncommitted",
         ]
     )
 
@@ -100,5 +101,6 @@ def test_main_with_matrices(tmp_path, tmp_vecs, query_path):
             str(query_matrix_path),
             "--out_dir",
             str(tmp_path),
+            "--uncommitted",
         ]
     )