Fix version endpoint in packaged image

Author: shuchenliu

pyproject.toml

@@ -94,7 +94,7 @@ extend_skip = []
 # pytest configuration
 [tool.pytest.ini_options]
 minversion = "6.2.5"
-pythonpath = ["."]
+pythonpath = ["src", "."]
 
 
 # Options
@@ -122,7 +122,7 @@ norecursedirs = [
     "lib",
     "lib64",
 ]
-testpaths = ["test"]
+testpaths = ["tests"]
 
 
 # Markers

src/nodenorm/handlers/version.py

@@ -1,36 +1,13 @@
-import logging
-import pathlib
-import git
+import asyncio
 
 from biothings.web.handlers import BaseHandler
 
-logger = logging.getLogger(__name__)
+from nodenorm.version import get_version
 
 
 class VersionHandler(BaseHandler):
     name = "version"
 
-    def get_github_commit_hash(self):
-        """Retrieve the current GitHub commit hash using gitpython."""
-        try:
-            # Resolve the absolute path to the current file
-            file_path = pathlib.Path(__file__).resolve()
-
-            # Use git.Repo to find the root of the repository
-            repo = git.Repo(file_path, search_parent_directories=True)
-
-            if repo.bare:
-                # Get the absolute path to the repository root
-                repo_dir = repo.working_tree_dir
-                logger.error(f"Git repository not found in directory: {repo_dir}")
-                return "Unknown"
-
-            commit_hash = repo.head.commit.hexsha  # Get the latest commit hash
-            return commit_hash
-        except Exception as e:
-            logger.error(f"Error getting GitHub commit hash: {e}")
-            return "Unknown"
-
     async def get(self, *args, **kwargs):
-        version = self.get_github_commit_hash()
+        version = await asyncio.to_thread(get_version)
         self.write({"version": version})

src/nodenorm/version.py

@@ -0,0 +1,59 @@
+import logging
+import os
+import pathlib
+from collections.abc import Iterable
+from functools import cache
+
+from git import InvalidGitRepositoryError, NoSuchPathError, Repo
+
+logger = logging.getLogger(__name__)
+
+UNKNOWN_VERSION = "Unknown"
+VERSION_FILE_NAME = "version.txt"
+VERSION_FILE_ENV_VAR = "NODENORM_VERSION_FILE"
+CONTAINER_VERSION_FILE = pathlib.Path("/home/nodenorm/configuration") / VERSION_FILE_NAME
+
+
+def read_version_file(version_file_paths: Iterable[pathlib.Path] | None = None) -> str | None:
+    """Read the build-time version file when the app is running from a packaged image."""
+    candidate_paths = []
+    configured_path = os.getenv(VERSION_FILE_ENV_VAR)
+    if configured_path:
+        candidate_paths.append(pathlib.Path(configured_path))
+
+    if version_file_paths is None:
+        candidate_paths.extend([pathlib.Path.cwd() / VERSION_FILE_NAME, CONTAINER_VERSION_FILE])
+    else:
+        candidate_paths.extend(version_file_paths)
+
+    for version_file_path in candidate_paths:
+        try:
+            version = version_file_path.read_text(encoding="utf-8").strip()
+        except OSError:
+            continue
+
+        if version:
+            return version
+
+    return None
+
+
+@cache
+def get_github_commit_hash(source_path: pathlib.Path | None = None) -> str:
+    """Retrieve the current GitHub commit hash using gitpython."""
+    try:
+        repo_path = source_path or pathlib.Path(__file__).resolve()
+        repo = Repo(repo_path, search_parent_directories=True)
+
+        return repo.head.commit.hexsha
+    except (InvalidGitRepositoryError, NoSuchPathError) as exc:
+        logger.warning("Git repository unavailable for version lookup: %s", exc)
+        return UNKNOWN_VERSION
+    except Exception as exc:
+        logger.exception("Error getting GitHub commit hash: %s", exc)
+        return UNKNOWN_VERSION
+
+
+@cache
+def get_version() -> str:
+    return read_version_file() or get_github_commit_hash()

tests/test_version.py

@@ -0,0 +1,76 @@
+import pytest
+
+from nodenorm import version as nodenorm_version
+
+
+@pytest.fixture(autouse=True)
+def clear_version_caches():
+    nodenorm_version.get_github_commit_hash.cache_clear()
+    nodenorm_version.get_version.cache_clear()
+    yield
+    nodenorm_version.get_github_commit_hash.cache_clear()
+    nodenorm_version.get_version.cache_clear()
+
+
+def test_read_version_file_uses_configured_path(tmp_path, monkeypatch):
+    configured_version_file = tmp_path / "configured-version.txt"
+    configured_version_file.write_text("configured-sha\n", encoding="utf-8")
+    fallback_version_file = tmp_path / "version.txt"
+    fallback_version_file.write_text("fallback-sha\n", encoding="utf-8")
+
+    monkeypatch.setenv(nodenorm_version.VERSION_FILE_ENV_VAR, str(configured_version_file))
+
+    assert nodenorm_version.read_version_file([fallback_version_file]) == "configured-sha"
+
+
+def test_read_version_file_ignores_missing_and_blank_files(tmp_path, monkeypatch):
+    blank_version_file = tmp_path / "blank-version.txt"
+    blank_version_file.write_text("\n", encoding="utf-8")
+    version_file = tmp_path / "version.txt"
+    version_file.write_text("build-sha\n", encoding="utf-8")
+
+    monkeypatch.delenv(nodenorm_version.VERSION_FILE_ENV_VAR, raising=False)
+
+    assert (
+        nodenorm_version.read_version_file([tmp_path / "missing.txt", blank_version_file, version_file])
+        == "build-sha"
+    )
+
+
+def test_read_version_file_allows_empty_fallback_paths(monkeypatch):
+    monkeypatch.delenv(nodenorm_version.VERSION_FILE_ENV_VAR, raising=False)
+
+    assert nodenorm_version.read_version_file([]) is None
+
+
+def test_get_version_prefers_build_version_file(monkeypatch):
+    monkeypatch.setattr(nodenorm_version, "read_version_file", lambda: "build-sha")
+    monkeypatch.setattr(
+        nodenorm_version,
+        "get_github_commit_hash",
+        lambda: pytest.fail("Git fallback should not run when a build version file exists"),
+    )
+
+    assert nodenorm_version.get_version() == "build-sha"
+
+
+def test_get_version_falls_back_to_git(monkeypatch):
+    monkeypatch.setattr(nodenorm_version, "read_version_file", lambda: None)
+    monkeypatch.setattr(nodenorm_version, "get_github_commit_hash", lambda: "git-sha")
+
+    assert nodenorm_version.get_version() == "git-sha"
+
+
+def test_get_version_caches_resolved_version(monkeypatch):
+    calls = 0
+
+    def read_version_file():
+        nonlocal calls
+        calls += 1
+        return "build-sha"
+
+    monkeypatch.setattr(nodenorm_version, "read_version_file", read_version_file)
+
+    assert nodenorm_version.get_version() == "build-sha"
+    assert nodenorm_version.get_version() == "build-sha"
+    assert calls == 1