analytics: send remote types and git-remote's path hashes (#10816)

tests/func/test_analytics.py

@@ -2,6 +2,7 @@ import os
 
 import pytest
 
+from dvc import __version__, env
 from dvc.analytics import _scm_in_use, collect_and_send_report
 from dvc.cli import main
 from dvc.repo import Repo
@@ -33,7 +34,8 @@ def mock_daemon(mocker):
     return mocker.patch("dvc.daemon.daemon", mocker.MagicMock(side_effect=func))
 
 
-def test_collect_and_send_report(mocker, dvc, mock_daemon):
+def test_collect_and_send_report(monkeypatch, mocker, dvc, mock_daemon):
+    monkeypatch.delenv(env.DVC_ANALYTICS_ENDPOINT, raising=False)
     mock_post = mocker.patch("requests.post")
     collect_and_send_report()
 
@@ -41,19 +43,28 @@ def test_collect_and_send_report(mocker, dvc, mock_daemon):
     assert mock_post.call_count == 1
     assert mock_post.call_args == mocker.call(
         "https://analytics.dvc.org",
-        json=ANY(dict),
+        json={
+            "dvc_version": __version__,
+            "scm_class": type(dvc.scm).__name__,
+            "is_binary": False,
+            "system_info": ANY(dict),
+            "user_id": ANY(str),
+            "group_id": mocker.ANY,
+            "remotes": ANY(list),
+            "git_remote_hash": None,
+        },
         headers={"content-type": "application/json"},
         timeout=5,
     )
 
 
 def test_scm_dvc_only(tmp_dir, dvc):
-    scm = _scm_in_use()
+    scm = _scm_in_use(dvc.scm)
     assert scm == "NoSCM"
 
 
 def test_scm_git(tmp_dir, scm, dvc):
-    scm = _scm_in_use()
+    scm = _scm_in_use(scm)
     assert scm == "Git"
 
 
@@ -62,7 +73,7 @@ def test_scm_subrepo(tmp_dir, scm):
     subdir.mkdir()
 
     with subdir.chdir():
-        Repo.init(subdir=True)
-        scm = _scm_in_use()
+        repo = Repo.init(subdir=True)
+        scm = _scm_in_use(repo.scm)
 
     assert scm == "Git"

tests/unit/test_analytics.py

@@ -1,10 +1,11 @@
+import hashlib
 import json
 import platform
 
 import pytest
 from voluptuous import Any, Schema
 
-from dvc import analytics
+from dvc import analytics, env
 from dvc.cli import parse_args
 
 
@@ -50,6 +51,8 @@ def test_runtime_info(tmp_global_dir):
             "user_id": str,
             "system_info": dict,
             "group_id": Any(str, None),
+            "remotes": Any(list, None),
+            "git_remote_hash": Any(str, None),
         },
         required=True,
     )
@@ -57,7 +60,8 @@ def test_runtime_info(tmp_global_dir):
     assert schema(analytics._runtime_info())
 
 
-def test_send(mocker, tmp_path):
+def test_send(monkeypatch, mocker, tmp_path):
+    monkeypatch.delenv(env.DVC_ANALYTICS_ENDPOINT, raising=False)
     mock_post = mocker.patch("requests.post")
 
     import requests
@@ -158,3 +162,38 @@ def test_system_info():
         )
 
     assert schema(analytics._system_info())
+
+
+@pytest.mark.parametrize(
+    "git_remote",
+    [
+        "git://github.com/iterative/dvc.git",
+        "[email protected]:iterative/dvc.git",
+        "http://github.com/iterative/dvc.git",
+        "https://github.com/iterative/dvc.git",
+        "ssh://[email protected]/iterative/dvc.git",
+    ],
+)
+def test_git_remote_hash(mocker, git_remote):
+    m = mocker.patch("dvc.analytics._git_remote_url", return_value=git_remote)
+    expected = hashlib.md5(b"iterative/dvc.git").hexdigest()
+
+    assert analytics._git_remote_path_hash(None) == expected
+    m.assert_called_once_with(None)
+
+
+@pytest.mark.parametrize(
+    "git_remote",
+    [
+        "C:\\Users\\user\\dvc.git",
+        "/home/user/dvc.git",
+        "file:///home/user/dvc.git",
+        "./dvc.git",
+    ],
+)
+def test_git_remote_hash_local(mocker, git_remote):
+    m = mocker.patch("dvc.analytics._git_remote_url", return_value=git_remote)
+
+    expected = hashlib.md5(git_remote.encode("utf-8")).hexdigest()
+    assert analytics._git_remote_path_hash(None) == expected
+    m.assert_called_once_with(None)