feature: cache downloaded wheel information (#2276)

Co-authored-by: Bernát Gábor <[email protected]>

docs/changelog/2268.feature.rst

@@ -0,0 +1,2 @@
+Add downloaded wheel information in the relevant JSON embed file to
+prevent additional downloads of the same wheel. - by :user:`mayeut`.

src/virtualenv/app_data/via_disk_folder.py

@@ -14,7 +14,7 @@
 │       │           └── <install class> -> CopyPipInstall / SymlinkPipInstall
 │       │               └── <wheel name> -> pip-20.1.1-py2.py3-none-any
 │       └── embed
-│           └── 2 -> json format versioning
+│           └── 3 -> json format versioning
 │               └── *.json -> for every distribution contains data about newer embed versions and releases
 └─── unzip <in zip app we cannot refer to some internal files, so first extract them>
      └── <virtualenv version>
@@ -101,7 +101,7 @@ def py_info_clear(self):
                             filename.unlink()
 
     def embed_update_log(self, distribution, for_py_version):
-        return EmbedDistributionUpdateStoreDisk(self.lock / "wheel" / for_py_version / "embed" / "2", distribution)
+        return EmbedDistributionUpdateStoreDisk(self.lock / "wheel" / for_py_version / "embed" / "3", distribution)
 
     @property
     def house(self):

src/virtualenv/seed/wheels/acquire.py

@@ -10,6 +10,7 @@
 from virtualenv.util.subprocess import Popen, subprocess
 
 from .bundle import from_bundle
+from .periodic_update import add_wheel_to_update_log
 from .util import Version, Wheel, discover_wheels
 
 
@@ -35,6 +36,8 @@ def get_wheel(distribution, version, for_py_version, search_dirs, download, app_
             to_folder=app_data.house,
             env=env,
         )
+        if wheel is not None and app_data.can_update:
+            add_wheel_to_update_log(wheel, for_py_version, app_data)
 
     return wheel
 

src/virtualenv/seed/wheels/periodic_update.py

@@ -82,6 +82,19 @@ def handle_auto_update(distribution, for_py_version, wheel, search_dirs, app_dat
         trigger_update(distribution, for_py_version, wheel, search_dirs, app_data, periodic=True, env=env)
 
 
+def add_wheel_to_update_log(wheel, for_py_version, app_data):
+    embed_update_log = app_data.embed_update_log(wheel.distribution, for_py_version)
+    logging.debug("adding %s information to %s", wheel.name, embed_update_log.file)
+    u_log = UpdateLog.from_dict(embed_update_log.read())
+    if any(version.filename == wheel.name for version in u_log.versions):
+        logging.warning("%s already present in %s", wheel.name, embed_update_log.file)
+        return
+    # we don't need a release date for sources other than "periodic"
+    version = NewVersion(wheel.name, datetime.now(), None, "download")
+    u_log.versions.append(version)  # always write at the end for proper updates
+    embed_update_log.write(u_log.to_dict())
+
+
 DATETIME_FMT = "%Y-%m-%dT%H:%M:%S.%fZ"
 
 
@@ -248,23 +261,27 @@ def _run_do_update(app_data, distribution, embed_filename, for_py_version, perio
     embed_update_log = app_data.embed_update_log(distribution, for_py_version)
     u_log = UpdateLog.from_dict(embed_update_log.read())
     now = datetime.now()
+
+    update_versions, other_versions = [], []
+    for version in u_log.versions:
+        if version.source in {"periodic", "manual"}:
+            update_versions.append(version)
+        else:
+            other_versions.append(version)
+
     if periodic:
         source = "periodic"
-        # mark everything not updated manually as source "periodic"
-        for version in u_log.versions:
-            if version.source != "manual":
-                version.source = source
     else:
         source = "manual"
-        # mark everything as source "manual"
-        for version in u_log.versions:
-            version.source = source
+        # mark the most recent one as source "manual"
+        if update_versions:
+            update_versions[0].source = source
 
     if wheel_filename is not None:
         dest = wheelhouse / wheel_filename.name
         if not dest.exists():
             copy2(str(wheel_filename), str(wheelhouse))
-    last, last_version, versions = None, None, []
+    last, last_version, versions, filenames = None, None, [], set()
     while last is None or not last.use(now, ignore_grace_period_ci=True):
         download_time = datetime.now()
         dest = acquire.download_wheel(
@@ -276,21 +293,24 @@ def _run_do_update(app_data, distribution, embed_filename, for_py_version, perio
             to_folder=wheelhouse,
             env=os.environ,
         )
-        if dest is None or (u_log.versions and u_log.versions[0].filename == dest.name):
+        if dest is None or (update_versions and update_versions[0].filename == dest.name):
             break
         release_date = release_date_for_wheel_path(dest.path)
         last = NewVersion(filename=dest.path.name, release_date=release_date, found_date=download_time, source=source)
         logging.info("detected %s in %s", last, datetime.now() - download_time)
         versions.append(last)
-        last_wheel = Wheel(Path(last.filename))
+        filenames.add(last.filename)
+        last_wheel = last.wheel
         last_version = last_wheel.version
         if embed_version is not None:
             if embed_version >= last_wheel.version_tuple:  # stop download if we reach the embed version
                 break
     u_log.periodic = periodic
     if not u_log.periodic:
         u_log.started = now
-    u_log.versions = versions + u_log.versions
+    # update other_versions by removing version we just found
+    other_versions = [version for version in other_versions if version.filename not in filenames]
+    u_log.versions = versions + update_versions + other_versions
     u_log.completed = datetime.now()
     embed_update_log.write(u_log.to_dict())
     return versions
@@ -395,6 +415,7 @@ def _run_manual_upgrade(app_data, distribution, for_py_version, env):
 
 
 __all__ = (
+    "add_wheel_to_update_log",
     "periodic_update",
     "do_update",
     "manual_upgrade",

tests/unit/seed/wheels/test_acquire.py

@@ -2,16 +2,25 @@
 
 import os
 import sys
+from datetime import datetime
 from subprocess import CalledProcessError
 
 import pytest
 
+from virtualenv.app_data import AppDataDiskFolder
+from virtualenv.info import IS_PYPY, PY2
 from virtualenv.seed.wheels.acquire import download_wheel, get_wheel, pip_wheel_env_run
 from virtualenv.seed.wheels.embed import BUNDLE_FOLDER, get_embed_wheel
+from virtualenv.seed.wheels.periodic_update import dump_datetime
 from virtualenv.seed.wheels.util import Wheel, discover_wheels
 from virtualenv.util.path import Path
 
 
+@pytest.fixture(autouse=True)
+def fake_release_date(mocker):
+    mocker.patch("virtualenv.seed.wheels.periodic_update.release_date_for_wheel_path", return_value=None)
+
+
 def test_pip_wheel_env_run_could_not_find(session_app_data, mocker):
     mocker.patch("virtualenv.seed.wheels.acquire.from_bundle", return_value=None)
     with pytest.raises(RuntimeError, match="could not find the embedded pip"):
@@ -74,24 +83,64 @@ def test_download_fails(mocker, for_py_version, session_app_data):
 @pytest.fixture
 def downloaded_wheel(mocker):
     wheel = Wheel.from_path(Path("setuptools-0.0.0-py2.py3-none-any.whl"))
-    mocker.patch("virtualenv.seed.wheels.acquire.download_wheel", return_value=wheel)
-    yield wheel
+    yield wheel, mocker.patch("virtualenv.seed.wheels.acquire.download_wheel", return_value=wheel)
 
 
 @pytest.mark.parametrize("version", ["bundle", "0.0.0"])
-def test_get_wheel_download_called(for_py_version, session_app_data, downloaded_wheel, version):
+def test_get_wheel_download_called(mocker, for_py_version, session_app_data, downloaded_wheel, version):
     distribution = "setuptools"
+    write = mocker.patch("virtualenv.app_data.via_disk_folder.JSONStoreDisk.write")
     wheel = get_wheel(distribution, version, for_py_version, [], True, session_app_data, False, os.environ)
     assert wheel is not None
-    assert wheel.name == downloaded_wheel.name
+    assert wheel.name == downloaded_wheel[0].name
+    assert downloaded_wheel[1].call_count == 1
+    assert write.call_count == 1
 
 
 @pytest.mark.parametrize("version", ["embed", "pinned"])
-def test_get_wheel_download_not_called(for_py_version, session_app_data, downloaded_wheel, version):
+def test_get_wheel_download_not_called(mocker, for_py_version, session_app_data, downloaded_wheel, version):
     distribution = "setuptools"
     expected = get_embed_wheel(distribution, for_py_version)
     if version == "pinned":
         version = expected.version
+    write = mocker.patch("virtualenv.app_data.via_disk_folder.JSONStoreDisk.write")
     wheel = get_wheel(distribution, version, for_py_version, [], True, session_app_data, False, os.environ)
     assert wheel is not None
     assert wheel.name == expected.name
+    assert downloaded_wheel[1].call_count == 0
+    assert write.call_count == 0
+
+
+@pytest.mark.skipif(IS_PYPY and PY2, reason="mocker.spy failing on PyPy 2.x")
+def test_get_wheel_download_cached(tmp_path, freezer, mocker, for_py_version, downloaded_wheel):
+    from virtualenv.app_data.via_disk_folder import JSONStoreDisk
+
+    app_data = AppDataDiskFolder(folder=str(tmp_path))
+    expected = downloaded_wheel[0]
+    write = mocker.spy(JSONStoreDisk, "write")
+    # 1st call, not cached, download is called
+    wheel = get_wheel(expected.distribution, expected.version, for_py_version, [], True, app_data, False, os.environ)
+    assert wheel is not None
+    assert wheel.name == expected.name
+    assert downloaded_wheel[1].call_count == 1
+    assert write.call_count == 1
+    # 2nd call, cached, download is not called
+    wheel = get_wheel(expected.distribution, expected.version, for_py_version, [], True, app_data, False, os.environ)
+    assert wheel is not None
+    assert wheel.name == expected.name
+    assert downloaded_wheel[1].call_count == 1
+    assert write.call_count == 1
+    wrote_json = write.call_args[0][1]
+    assert wrote_json == {
+        "completed": None,
+        "periodic": None,
+        "started": None,
+        "versions": [
+            {
+                "filename": expected.name,
+                "release_date": None,
+                "found_date": dump_datetime(datetime.now()),
+                "source": "download",
+            },
+        ],
+    }