add path to Distribution namedtuple

ddtrace/internal/packages.py

@@ -1,14 +1,14 @@
 import typing
 
 
-Distribution = typing.NamedTuple("Distribution", [("name", str), ("version", str)])
+Distribution = typing.NamedTuple("Distribution", [("name", str), ("version", str), ("path", str)])
 
 _DISTRIBUTIONS = None  # type: typing.Optional[typing.Set[Distribution]]
 
 
 def get_distributions():
     # type: () -> typing.Set[Distribution]
-    """returns the names, versions and path of all installed distributions"""
+    """returns the name and version of all distributions in a python path"""
     global _DISTRIBUTIONS
     if _DISTRIBUTIONS is not None:
         return _DISTRIBUTIONS
@@ -20,13 +20,15 @@ def get_distributions():
 
     pkgs = set()
     for dist in importlib_metadata.distributions():
-        # accessing dist.metadata reads all lines in the PKG-INFO and/or METADATA files
-        # We should avoid accessing dist.metadata more than once
+        # Get the root path of all files in a distribution
+        path = str(dist.locate_file(""))
+        # PKG-INFO and/or METADATA files are parsed when dist.metadata is accessed
+        # Optimization: we should avoid accessing dist.metadata more than once
         metadata = dist.metadata
         name = metadata["name"]
         version = metadata["version"]
         if name and version:
-            pkgs.add(Distribution(name, version))
+            pkgs.add(Distribution(path=path, name=name, version=version))
 
     _DISTRIBUTIONS = pkgs
     return _DISTRIBUTIONS

ddtrace/internal/telemetry/data.py

@@ -62,8 +62,9 @@ def _get_application(key):
 
 def get_dependencies():
     # type: () -> List[Dict[str, str]]
-    """Creates a list of dictionaries to store the name and version of all installed packages"""
-    return [{"name": dist.name, "version": dist.version} for dist in get_distributions()]
+    """Returns a unique list of the names and versions of all installed packages"""
+    dependencies = {(dist.name, dist.version) for dist in get_distributions()}
+    return [{"name": name, "version": version} for name, version in dependencies]
 
 
 def get_application(service, version, env):

tests/internal/test_packages.py

@@ -1,3 +1,5 @@
+import os
+
 from ddtrace.internal.packages import get_distributions
 
 
@@ -7,13 +9,18 @@ def test_get_distributions():
 
     pkg_resources_ws = {pkg.project_name for pkg in pkg_resources.working_set}
 
-    importlib_pkgs = {pkg.name for pkg in get_distributions()}
-    # The package name in typing_extensions-4.x.x.dist-info/METADATA is set to `typing_extensions`
-    # this is inconsistent with the package name found in pkg_resources. The block below corrects this.
-    # The correct package name is typing-extensions.
-    if "typing_extensions" in importlib_pkgs and "typing-extensions" in pkg_resources_ws:
-        importlib_pkgs.discard("typing_extensions")
-        importlib_pkgs.add("typing-extensions")
+    importlib_pkgs = set()
+    for pkg in get_distributions():
+        assert pkg.name
+        assert pkg.version
+        assert os.path.exists(pkg.path)
+        # The package name in typing_extensions-4.x.x.dist-info/METADATA is set to `typing_extensions`
+        # this is inconsistent with the package name found in pkg_resources. The block below corrects this.
+        # The correct package name is typing-extensions.
+        if pkg.name == "typing_extensions" and "typing-extensions" in pkg_resources_ws:
+            importlib_pkgs.add("typing-extensions")
+        else:
+            importlib_pkgs.add(pkg.name)
 
     # assert that pkg_resources and importlib.metadata return the same packages
     assert pkg_resources_ws == importlib_pkgs

tests/tracer/telemetry/test_data.py

@@ -149,12 +149,6 @@ def test_get_container_id_when_container_does_not_exists():
 
 def test_get_dependencies():
     """asserts that get_dependencies and get_distributions return the same packages"""
-    pkgs_as_dicts = get_dependencies()
-    pkgs_as_distributions = list(get_distributions())
-
-    assert len(pkgs_as_dicts) == len(pkgs_as_distributions)
-    for i in range(len(pkgs_as_dicts)):
-        pkg_dict = pkgs_as_dicts[i]
-        pkg_dist = pkgs_as_distributions[i]
-        assert pkg_dict["name"] == pkg_dist.name
-        assert pkg_dict["version"] == pkg_dist.version
+    pkgs_as_dicts = {(dep["name"], dep["version"]) for dep in get_dependencies()}
+    pkgs_as_distributions = {(dist.name, dist.version) for dist in get_distributions()}
+    assert pkgs_as_dicts == pkgs_as_distributions