Bug/is user code

oracletrace/tracer.py

@@ -1,4 +1,6 @@
 import sys
+import sysconfig
+import site
 import os
 import time
 from collections import defaultdict
@@ -76,11 +78,31 @@ def _is_ignored(self, filename: str) -> bool:
         return False
 
     def _is_user_code(self, filename: str) -> bool:
-        # Filter out files not in the project root
-        if not filename.startswith(str(self._root_path)):
+        # Normalize all paths to resolve symlinks
+        filename = os.path.realpath(filename)
+        root_path = os.path.realpath(self._root_path)
+
+        # Check project root
+        if os.path.commonpath([root_path, filename]) != root_path:
+            return False
+
+        # Exclude Python stdlib
+        stdlib_path = os.path.realpath(sysconfig.get_path("stdlib"))
+        if os.path.commonpath([stdlib_path, filename]) == stdlib_path:
             return False
-        # Filter out third-party libraries
-        if "site-packages" in filename or "dist-packages" in filename:
+
+        # Exclude all site-packages
+        site_paths = site.getsitepackages() + [site.getusersitepackages()]
+        normalized_site_paths = [os.path.realpath(p) for p in site_paths]
+        for sp in normalized_site_paths:
+            if os.path.commonpath([sp, filename]) == sp:
+                return False
+
+        # Exclude venv-like directories that may be inside the project root
+        venv_markers = {"venv", ".venv", "env", ".env", "virtualenv"}
+        path_relative_to_root = os.path.relpath(filename, root_path)
+        path_parts = Path(path_relative_to_root).parts
+        if any(part in venv_markers for part in path_parts):
             return False
         return True
 

tests/test_tracer.py

@@ -0,0 +1,112 @@
+import os
+import sysconfig
+import tempfile
+import pytest
+from pathlib import Path
+
+from oracletrace.tracer import Tracer
+
+
+@pytest.fixture
+def tracer():
+    with tempfile.TemporaryDirectory() as tmpdir:
+        yield Tracer(tmpdir)
+
+
+@pytest.fixture
+def stdlib_path():
+    return os.path.realpath(sysconfig.get_path("stdlib"))
+
+
+@pytest.fixture
+def site_paths():
+    import site
+    paths = site.getsitepackages() + [site.getusersitepackages()]
+    return [os.path.realpath(p) for p in paths]
+
+
+def test_user_code_under_root(tracer):
+    user_file = os.path.join(tracer._root_path, "my_module.py")
+    Path(user_file).touch()
+    assert tracer._is_user_code(user_file)
+
+
+def test_user_code_nested_under_root(tracer):
+    nested_dir = os.path.join(tracer._root_path, "subdir", "deep")
+    os.makedirs(nested_dir, exist_ok=True)
+    user_file = os.path.join(nested_dir, "my_module.py")
+    Path(user_file).touch()
+    assert tracer._is_user_code(user_file)
+
+
+def test_external_code_outside_root(tracer):
+    with tempfile.TemporaryDirectory() as tmpdir:
+        external_file = os.path.join(tmpdir, "external.py")
+        Path(external_file).touch()
+        assert not tracer._is_user_code(external_file)
+
+
+def test_stdlib_excluded(tracer, stdlib_path):
+    stdlib_file = os.path.join(stdlib_path, "os.py")
+    assert not tracer._is_user_code(stdlib_file)
+
+
+def test_site_packages_excluded(tracer, site_paths):
+    for sp in site_paths:
+        if os.path.exists(sp):
+            site_file = os.path.join(sp, "requests", "__init__.py")
+            assert not tracer._is_user_code(site_file)
+
+
+def test_venv_site_packages_excluded(tracer):
+    venv_path = os.path.join(tracer._root_path, "venv", "lib", "python", "site-packages", "numpy", "__init__.py")
+    assert not tracer._is_user_code(venv_path)
+
+
+def test_prefix_false_positive(tracer):
+    similar_path = tracer._root_path + "_backup"
+    os.makedirs(similar_path, exist_ok=True)
+    similar_file = os.path.join(similar_path, "module.py")
+    Path(similar_file).touch()
+    assert not tracer._is_user_code(similar_file)
+
+
+def test_symlinked_user_code(tracer):
+    real_file = os.path.join(tracer._root_path, "real_module.py")
+    Path(real_file).touch()
+
+    symlink_dir = tempfile.mkdtemp()
+    symlink_file = os.path.join(symlink_dir, "linked_module.py")
+    os.symlink(real_file, symlink_file)
+
+    try:
+        assert tracer._is_user_code(symlink_file)
+    finally:
+        os.unlink(symlink_file)
+        os.rmdir(symlink_dir)
+
+
+def test_dist_packages_excluded(tracer):
+    dist_pkg_path = "/usr/lib/python3/dist-packages/numpy/__init__.py"
+    assert not tracer._is_user_code(dist_pkg_path)
+
+
+def test_dot_venv_excluded(tracer):
+    venv_path = os.path.join(tracer._root_path, ".venv", "lib", "python", "site-packages", "flask", "__init__.py")
+    assert not tracer._is_user_code(venv_path)
+
+
+def test_env_excluded(tracer):
+    env_path = os.path.join(tracer._root_path, "env", "lib", "python", "site-packages", "django", "__init__.py")
+    assert not tracer._is_user_code(env_path)
+
+
+def test_dot_env_excluded(tracer):
+    env_path = os.path.join(tracer._root_path, ".env", "lib", "python", "site-packages", "pandas", "__init__.py")
+    assert not tracer._is_user_code(env_path)
+
+
+def test_root_path_with_trailing_slash(tracer):
+    user_file = os.path.join(tracer._root_path, "module.py")
+    Path(user_file).touch()
+    assert tracer._is_user_code(user_file)