oshi-core/src/main/java/oshi/hardware/platform/windows/WindowsPowerSource.java (2)

75-77: ⚠️ Potential issue | 🟠 Major

Don't turn a single System Battery into a per-pack percentage.

getPowerSources() still returns one System Battery, but Line 271 stops after the first enumerated battery and Line 200 now exposes that pack's current/max ratio as the public remaining-capacity percent. If Windows exposes more than one battery, callers will get a per-pack number instead of the system-wide percentage. Please accumulate current/max across all batteries before setting psRemainingCapacityPercent.

Also applies to: 200-201, 270-271

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@oshi-core/src/main/java/oshi/hardware/platform/windows/WindowsPowerSource.java`
around lines 75 - 77, getPowerSources() currently returns a single "System
Battery" and the code that computes psRemainingCapacityPercent (in
getPowerSource/getPowerSources logic) stops after the first enumerated pack,
causing per-pack percentages to be exposed; change the logic to enumerate all
batteries, return a List of PowerSource objects (one system-level entry if
that's the intended public API) and—critically—when computing
psRemainingCapacityPercent (in the method(s) that read battery entries, e.g.,
getPowerSource and its battery enumeration loop) sum the available/current
capacities and the full/max capacities across all detected battery packs and
compute remainingPercent = totalCurrent / totalMax instead of using the first
pack only; also remove any early break that stops after the first battery so all
packs are included in the aggregation.

---

251-269: ⚠️ Potential issue | 🟠 Major

The new discharging estimate can publish invalid fallback values.

If the BatteryEstimatedTime query fails, psTimeRemainingInstant is still 0d, and Line 268 exposes that as the estimated value for a discharging battery. Even when Lines 260-265 run, the fallback uses psMaxCapacity - psCurrentCapacity, which is time-to-full, not time-to-empty, so the new discharging path can report the wrong number. That violates oshi-core/src/main/java/oshi/hardware/PowerSource.java:51-69 and the existing assertion in oshi-core/src/test/java/oshi/hardware/PowerSourceTest.java:27-40.

💡 Suggested adjustment

-                                                        // Fallback
-                                                        if (psTimeRemainingInstant < 0 && psPowerUsageRate != 0) {
-                                                            psTimeRemainingInstant = (psMaxCapacity - psCurrentCapacity)
-                                                                    * 3600d / psPowerUsageRate;
-                                                            if (psTimeRemainingInstant < 0) {
-                                                                psTimeRemainingInstant *= -1;
-                                                            }
-                                                        }
-                                                        if (psDischarging) {
+                                                        // Fallback
+                                                        if (psTimeRemainingInstant <= 0 && psPowerUsageRate != 0) {
+                                                            double capacityDelta = psDischarging ? psCurrentCapacity
+                                                                    : psMaxCapacity - psCurrentCapacity;
+                                                            psTimeRemainingInstant = capacityDelta * 3600d
+                                                                    / Math.abs(psPowerUsageRate);
+                                                        }
+                                                        if (psDischarging && psTimeRemainingInstant > 0d) {
                                                             psTimeRemainingEstimated = psTimeRemainingInstant;
                                                         }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@oshi-core/src/main/java/oshi/hardware/platform/windows/WindowsPowerSource.java`
around lines 251 - 269, The fallback logic in WindowsPowerSource that computes
psTimeRemainingInstant can produce wrong (time-to-full) values when reporting a
discharging battery and can publish zero as a valid estimate; update the
fallback so it only runs when the BatteryEstimatedTime query truly failed
(psTimeRemainingInstant not set/negative) and psPowerUsageRate != 0, and compute
time-to-empty when psDischarging is true (use psCurrentCapacity * 3600d /
psPowerUsageRate) and time-to-full when not discharging (use (psMaxCapacity -
psCurrentCapacity) * 3600d / psPowerUsageRate); also ensure you only assign
psTimeRemainingEstimated when the computed psTimeRemainingInstant is a
positive/valid value. Reference: WindowsPowerSource class and the variables
psTimeRemainingInstant, psDischarging, psPowerUsageRate, psMaxCapacity,
psCurrentCapacity, and psTimeRemainingEstimated.