1
0
mirror of https://codeberg.org/Freeyourgadget/Gadgetbridge synced 2024-11-28 21:06:50 +01:00

Experimental code for an activity sync problem. Commented out for now. #45

@daniele do you have an idea what this byte equence on the control point
characteristic could mean (and why the transfer stops then)?
This commit is contained in:
cpfeiffer 2015-07-12 00:02:51 +02:00
parent b516ceda01
commit 6553558947

View File

@ -59,7 +59,7 @@ public class MiBandSupport extends AbstractBTLEDeviceSupport {
private static final Logger LOG = LoggerFactory.getLogger(MiBandSupport.class);
//temporary buffer, size is a multiple of 60 because we want to store complete minutes (1 minute = 3 bytes)
private static final int activityDataHolderSize = 60 * 24; // 8h
private static final int activityDataHolderSize = 3 * 60 * 4; // 8h
private byte[] activityDataHolder = new byte[activityDataHolderSize];
//index of the buffer above
private int activityDataHolderProgress = 0;
@ -588,12 +588,7 @@ public class MiBandSupport extends AbstractBTLEDeviceSupport {
// byte 0 is the data type: 1 means that each minute is represented by a triplet of bytes
int dataType = value[0];
// byte 1 to 6 represent a timestamp
GregorianCalendar timestamp = new GregorianCalendar(value[1] + 2000,
value[2],
value[3],
value[4],
value[5],
value[6]);
GregorianCalendar timestamp = parseTimestamp(value, 1);
// counter of all data held by the band
int totalDataToRead = (value[7] & 0xff) | ((value[8] & 0xff) << 8);
@ -625,6 +620,17 @@ public class MiBandSupport extends AbstractBTLEDeviceSupport {
}
}
private GregorianCalendar parseTimestamp(byte[] value, int offset) {
GregorianCalendar timestamp = new GregorianCalendar(
value[offset] + 2000,
value[offset + 1],
value[offset + 2],
value[offset + 3],
value[offset + 4],
value[offset + 5]);
return timestamp;
}
private void bufferActivityData(byte[] value) {
if (this.activityDataRemainingBytes >= value.length) {
@ -644,6 +650,8 @@ public class MiBandSupport extends AbstractBTLEDeviceSupport {
LOG.warn("DATA: " + String.format("0x%8x", b));
}
}
} else {
LOG.error("error buffering activity data: remaining bytes: " + activityDataRemainingBytes + ", received: " + value.length);
}
}
@ -666,10 +674,10 @@ public class MiBandSupport extends AbstractBTLEDeviceSupport {
category);
timestamp.add(Calendar.MINUTE, 1);
}
} finally {
this.activityDataHolderProgress = 0;
this.activityDataTimestampProgress = timestamp;
}
this.activityDataHolderProgress = 0;
this.activityDataTimestampProgress = timestamp;
}
private void handleControlPointResult(byte[] value, int status) {
@ -681,6 +689,10 @@ public class MiBandSupport extends AbstractBTLEDeviceSupport {
if (getDevice().isBusy()) {
if (isActivityDataSyncFinished(value)) {
unsetBusy();
} else {
if (value != null && value.length == 9 && value[0] == 0xa) {
handleActivityCheckpoint(value);
}
}
}
if (value != null) {
@ -692,6 +704,22 @@ public class MiBandSupport extends AbstractBTLEDeviceSupport {
}
}
// I have no idea if this is anything we could or should do.
// For some reason, I got these bytes on the control point characteristic
// while in the middle of a activity data transfer:
// { 0xa, 0xf, 0x6, 0x5, 0x14, 0xe, 0x15, 0x70, 0x8 }
// 'a' = activity, then timestamp, then bytes to read
// After this, the communication stops. And this happened again and again
// so that I couldn't transfer the remaning data anymore.
// My idea was that this might be some kind of checkpointing, asking for an ACK
// of the transfer till now.
// Commented out for now until it can be reproduced and tested again.
private void handleActivityCheckpoint(byte[] value) {
GregorianCalendar timestamp = parseTimestamp(value, 1);
sendAckDataTransfer(timestamp, activityDataHolderProgress);
flushActivityDataHolder();
}
private boolean isActivityDataSyncFinished(byte[] value) {
// byte 0 is the kind of message
// byte 1 to 6 represent a timestamp