Rotate the random zoom spot over time, not just per-track

Random zoom position was previously picked once per track (stable via
subsonic_id) and held for that track's entire runtime. Changed it to
rotate on a ~10s cadence based on wall-clock time, still offset by a
per-track hash so different tracks don't all land on the same spot in
the same window. This lets a single long track show off different
sections of its own artwork over time, rather than being stuck on one
spot until the next track starts.

subwave_nowplaying.star

@@ -104,13 +104,22 @@ DEFAULT_ZOOM_LEVEL = "moderate"
 # own 36x36 bounds. Kept modest (0.3-0.7, not full 0.0-1.0) and verified
 # safe (no black border, even at the strongest 1.7x zoom) via the
 # calibration tool -- more extreme corners weren't tested and may not be.
-# Picked per-track (stable via subsonic_id, like the old corner-selection
-# logic) rather than per-render, so a still-playing track doesn't jump
-# between spots on every NOWPLAYING_TTL refresh. Only used when the
-# "Zoom position" schema field is set to "random" -- "center" always zooms
-# on the image's own center, exposed as a config choice since some covers
-# (e.g. ones with a centered logo or face) read better with a predictable
-# center zoom than a per-track random one.
+# Only used when the "Zoom position" schema field is set to "random" --
+# "center" always zooms on the image's own center, exposed as a config
+# choice since some covers (e.g. ones with a centered logo or face) read
+# better with a predictable center zoom than a random one.
+#
+# Rotates on a ~10s cadence (matching ART_ZOOM_DURATION below) based on
+# wall-clock time, not just per-track: a time bucket picks the spot, offset
+# by a per-track hash so different tracks don't all land on the same spot
+# at the same moment. This means the spot can change on *every* refresh,
+# including mid-playback of a single still-running track -- not just when
+# the track itself changes -- so a long track gets to show off different
+# sections of its art over time, which is the whole point of "random"
+# rather than "center". One real limitation: a one-shot `pixlet render`
+# CLI call reads the system clock same as the live server would, so this
+# is still deterministic for a given moment -- it just varies *which*
+# moment you rendered at, same as the live progress bar in _track_progress.
 ZOOM_CENTER_SPOT = (0.5, 0.5)
 ZOOM_RANDOM_SPOTS = [
     (0.5, 0.5),
@@ -119,16 +128,19 @@ ZOOM_RANDOM_SPOTS = [
     (0.3, 0.7),
     (0.7, 0.3),
 ]
+ZOOM_SPOT_ROTATE_SECONDS = 10
 DEFAULT_ZOOM_POSITION = "center"
 
-def _zoom_spot_for_track(zoom_position, subsonic_id):
+def _zoom_spot(zoom_position, subsonic_id):
     if zoom_position != "random":
         return ZOOM_CENTER_SPOT
 
-    total = 0
+    track_offset = 0
     for ch in subsonic_id.elems():
-        total += ord(ch)
-    return ZOOM_RANDOM_SPOTS[total % len(ZOOM_RANDOM_SPOTS)]
+        track_offset += ord(ch)
+
+    time_bucket = int(time.now().unix / ZOOM_SPOT_ROTATE_SECONDS)
+    return ZOOM_RANDOM_SPOTS[(time_bucket + track_offset) % len(ZOOM_RANDOM_SPOTS)]
 
 def main(config):
     station_url = config.get("station_url", DEFAULT_STATION_URL)
@@ -266,7 +278,7 @@ def _art_widget(art_bytes, zoom_scale, zoom_position, subsonic_id):
     # toward the bottom-right -- the opposite of the bug it was meant to
     # fix. Since `origin` already does directly and correctly what that
     # Translate was clumsily approximating, this uses `origin` alone.
-    origin_x, origin_y = _zoom_spot_for_track(zoom_position, subsonic_id)
+    origin_x, origin_y = _zoom_spot(zoom_position, subsonic_id)
     top_scale = max(zoom_scale, 1.0)
     zoom = animation.Transformation(
         child = art,