使用HealthConnect API时卡路里数据聚合结果存在差异

问题背景 #

通过Kotlin实现校验逻辑，对比过去30天内每日手动汇总的TotalCaloriesBurnedRecord卡路里值，与Health Connect aggregate接口返回的ENERGY_TOTAL聚合值，结果始终存在巨大差异。日志显示手动求和远小于聚合值，使用官方测试工具也复现了该问题。

核心原因 #

代码中仅调用了一次readRecords接口，未处理分页返回逻辑。当每日的卡路里记录数量超过设置的pageSize（2000）时，API会将数据分页返回，仅获取第一页记录会导致手动求和遗漏后续页的内容（比如系统自动生成的基础代谢记录、其他应用写入的批量小记录），而aggregate接口会计算所有记录的总和，因此出现数值差异。

解决方案 #

循环调用readRecords，直到返回的nextPageToken为null，收集所有分页的记录后再进行求和计算。

修正后的代码 #

suspend fun validateCaloriesData(healthConnectClient: HealthConnectClient) {
    // Time zone and date setup
    val zoneId = ZoneId.systemDefault()
    val endTime = Instant.now().atZone(zoneId).truncatedTo(ChronoUnit.DAYS)
    val startTime = endTime.minusDays(30)

    var currentDay = startTime
    while (currentDay.isBefore(endTime)) {
        val startOfDay = currentDay.toInstant()
        val endOfDay = currentDay.plusDays(1).toInstant()

        val dailyRecords = mutableListOf<Double>()

        try {
            var pageToken: String? = null
            // 循环读取所有分页的记录
            do {
                val records = healthConnectClient.readRecords(
                    ReadRecordsRequest(
                        recordType = TotalCaloriesBurnedRecord::class,
                        timeRangeFilter = TimeRangeFilter.between(startOfDay, endOfDay),
                        pageSize = 2000,
                        pageToken = pageToken,
                        ascendingOrder = true
                    )
                )
                records.records.forEach {
                    dailyRecords.add(it.energy.inKilocalories ?: 0.0)
                }
                pageToken = records.nextPageToken
            } while (pageToken != null)

            val sumOfDailyRecords = dailyRecords.sum()
            val response = healthConnectClient.aggregate(
                AggregateRequest(
                    metrics = setOf(TotalCaloriesBurnedRecord.ENERGY_TOTAL),
                    timeRangeFilter = TimeRangeFilter.between(startOfDay, endOfDay)
                )
            )
            val aggregatedCalories = response[TotalCaloriesBurnedRecord.ENERGY_TOTAL]?.inKilocalories ?: 0.0

            // 考虑浮点数精度问题，使用近似对比
            if (kotlin.math.abs(sumOfDailyRecords - aggregatedCalories) > 0.01) {
                Log.e("DailyCaloriesValidation", "Discrepancy found on day $currentDay: Manual sum is $sumOfDailyRecords but aggregated is $aggregatedCalories")
            } else {
                Log.i("DailyCaloriesValidation", "Match on day $currentDay: $sumOfDailyRecords")
            }

        } catch (e: Exception) {
            Log.e("DailyCaloriesValidation", "Error reading or aggregating calories for day $currentDay: ${e.message}")
        }

        currentDay = currentDay.plusDays(1)
    }
}

额外说明 #

1. 浮点数精度处理：由于卡路里数值可能存在小数位，直接用!=对比容易出现精度误差，改用绝对值差小于0.01的近似对比更合理。
2. 记录完整性：确保应用已申请READ_TOTAL_CALORIES_BURNED权限，否则无法读取所有来源的卡路里记录。

内容的提问来源于stack exchange，提问作者Catfoxes