fix(android-ratings): drop 0 'no data' rating that plots as a spike

analyze_stats.py

@@ -46,6 +46,7 @@ def _load_android_rating():
     df = pd.read_csv("data/android-ratings.csv", index_col="date", parse_dates=True)
     df = df.tz_localize(tz=timezone.utc)
     df.columns = ["Android Play Store rating"]
+    df = df[df.iloc[:, 0] > 0]  # 0 = "no data" placeholder, not a rating
     return df
 
 

android_ratings.py

@@ -42,6 +42,13 @@ def list_rating_files(bucket: str, package: str, token: str) -> list[str]:
     return sorted(n for n in names if n.endswith("_overview.csv"))
 
 
+def _is_valid_rating(value: str) -> bool:
+    try:
+        return 1.0 <= float(value) <= 5.0
+    except ValueError:
+        return False
+
+
 def fetch_rating_series(bucket: str, package: str, credentials) -> dict[str, str]:
     token = _token(credentials)
     series: dict[str, str] = {}
@@ -50,7 +57,9 @@ def fetch_rating_series(bucket: str, package: str, credentials) -> dict[str, str
         for row in csv.DictReader(io.StringIO(text)):
             date = (row.get("Date") or "").strip()
             value = (row.get(METRIC) or "").strip()
-            if date and value:
+            # Play Console emits 0 on days it has no rating data (not a real
+            # rating; valid ratings are 1-5). Skip so it isn't plotted as a spike.
+            if date and value and _is_valid_rating(value):
                 series[date] = value
     return series
 

data/android-ratings.csv

@@ -2751,6 +2751,5 @@ date,rating
 2026-09-21,3.125
 2026-09-22,3.3137
 2026-09-23,3.3999
-2026-09-24,0.0
 2026-09-25,3.2884
 2026-09-26,3.3333

test_android_ratings.py

@@ -0,0 +1,35 @@
+import android_ratings as ar
+from android_ratings import _is_valid_rating, fetch_rating_series
+
+
+def test_zero_and_garbage_ratings_rejected():
+    assert not _is_valid_rating("0.0")
+    assert not _is_valid_rating("0")
+    assert not _is_valid_rating("abc")
+    assert _is_valid_rating("3.2884")
+    assert _is_valid_rating("5.00")
+
+
+def test_fetch_rating_series_skips_placeholder_zero(monkeypatch):
+    """Collection path, not just the validator: 0.0 must not enter the series."""
+    report = (
+        "Date,Package Name,Total Average Rating\n"
+        "2026-09-23,net.activitywatch.android,3.3999\n"
+        "2026-09-24,net.activitywatch.android,0.0\n"
+        "2026-09-25,net.activitywatch.android,3.2884\n"
+        "2026-09-26,net.activitywatch.android,abc\n"
+        "2026-09-27,net.activitywatch.android,5.00\n"
+    ).encode("utf-16")
+
+    monkeypatch.setattr(ar, "_token", lambda creds: "tok")
+    monkeypatch.setattr(
+        ar, "list_rating_files", lambda bucket, package, token: ["overview.csv"]
+    )
+    monkeypatch.setattr(ar, "_download", lambda bucket, name, token: report)
+
+    series = fetch_rating_series("bucket", "net.activitywatch.android", credentials=None)
+    assert series == {
+        "2026-09-23": "3.3999",
+        "2026-09-25": "3.2884",
+        "2026-09-27": "5.00",
+    }

test_vitals.py

@@ -152,7 +152,7 @@ def test_by_version_csv_escapes_comma_in_label(monkeypatch, tmp_path):
         ["by-version", "--as-csv", "--package", "net.activitywatch.android"],
     )
     assert result.exit_code == 0, result.output
-    rows = list(csv.reader(_io.StringIO(result.output)))
+    rows = list(csv.reader(_io.StringIO(result.stdout)))
     # Header + one data row; the label's commas must not create extra columns
     assert rows[0] == ["version", "days", "mean"]
     assert len(rows[1]) == 3