t.mettler/giesCANne
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

First version of DBC code that functions. There are questions about some functionality but it mostly works.

Browse Source
This commit is contained in:
Collin Kidder
2015-05-31 20:43:34 -04:00
parent 34d68d7e33
commit f272a26dcd
5 changed files with 190 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files
canframemodel.cpp
+12 -2
View File
@@ -26,7 +26,14 @@ void CANFrameModel::setDBCHandler(DBCHandler *handler)
void CANFrameModel::setInterpetMode(bool mode)
{
interpretFrames = mode;
//if the state of interpretFrames changes then we need to reset the model
//so that QT will refresh the view properly
if (interpretFrames != mode)
{
this->beginResetModel();
interpretFrames = mode;
this->endResetModel();
}
}
void CANFrameModel::setOverwriteMode(bool mode)
@@ -70,14 +77,17 @@ QVariant CANFrameModel::data(const QModelIndex &index, int role) const
tempString.append(" ");
}
//now, if we're supposed to interpret the data and the DBC handler is loaded then use it
if (dbcHandler != NULL)
if (dbcHandler != NULL && interpretFrames)
{
DBC_MESSAGE *msg = dbcHandler->findMsgByID(thisFrame.ID);
if (msg != NULL)
{
tempString.append("\r\n");
for (int j = 0; j < msg->msgSignals.length(); j++)
{
tempString.append(dbcHandler->processSignal(thisFrame, msg->msgSignals.at(j)));
tempString.append("\r\n");
}
}
}
dbchandler.cpp
+134 -18
View File
@@ -320,42 +320,158 @@ void DBCHandler::listDebugging()
}
//DBC files use what I'd consider a completely stupid way to count bits. The lowest bit
//in a byte is 7 while the highest is 0. So, the counting for a byte goes like this:
//0 1 2 3 4 5 6 7. That only makes sense if you write it out like that. In reality bits are stored
//7 6 5 4 3 2 1 0. So, plan accordingly. It's confusing when you're used to bit 7 being the highest, not lowest
//But, bytes are still in order. Byte 0 is the first byte, byte 7 would be the last byte in a frame.
//So, the lowest bit of the last byte is 63. The upshot is that, if you took all the bytes in a canbus
//frame and started labeling from left to right you would really number 0 to 63 in complete order. It's
//just that computers don't store data like that. Have I mentioned that already? Screw Vector. Go away on a CANoe.
//Vector uses a special format for bit ordering. It pretends that the bits are numbered
//0 to 63 in ascending order of bits as if a 64 bit integer were stored lowest first
//and highest bit last.
//0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Vector bit ordering
//7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0 Bitwise ordering within bytes
//0 1 2 3 Byte ordering
//7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0 Normal bitwise ordering within bytes
//0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Reversed bit order used by Vector
//0 1 2 3 Byte ordering (same either way)
//A 16 bit integer would be stored Low first high second for intel format and high first, low second for motorola
//For intel format invert the starting bit within a byte.
//Otherwise, iterate over the bytes that it encompasses and
void DBCHandler::processSignal(CANFrame *frame, DBC_SIGNAL *sig)
//Otherwise, iterate over the bytes that it encompasses
//For intel format this works nicely as it means you can just go through the list getting higher and higher
//values for each bit as you go.
//For motorola it is backwards but only partially. For each byte you can go through and it's higher as you go
//but, at each byte boundary the next byte is lower than the multiplier for the last.
QString DBCHandler::processSignal(const CANFrame &frame, const DBC_SIGNAL &sig)
{
int startBit, endBit, startByte, endByte, bitWithinByteStart, bitWithinByteEnd;
int result = 0;
int multiplier;
int bitsToGo;
startBit = sig->startBit;
startBit = sig.startBit;
startByte = startBit / 8;
bitWithinByteStart = startBit % 8;
if (sig->intelByteOrder)
if (sig.intelByteOrder)
{
bitWithinByteStart = 7 - bitWithinByteStart;
startBit = (startByte * 8) + bitWithinByteStart;
}
endBit = startBit + sig->signalSize - 1;
if (sig.valType == STRING)
{
QString buildString;
int bytes = sig.signalSize / 8;
for (int x = 0; x < bytes; x++) buildString.append(frame.data[startByte + x]);
return buildString;
}
endBit = startBit + sig.signalSize - 1;
endByte = endBit / 8;
bitWithinByteEnd = endBit % 8;
bitsToGo = sig.signalSize - 1;
if (sig->intelByteOrder) //little endian - startBit is least sig. bit
multiplier = 1;
if (sig.intelByteOrder)
{
for (int y = startByte; y < endByte; y++) multiplier *= 256;
}
else //motorola / big endian - startBit is most sig. bit
//qDebug() << "Signal Name: " << sig.name;
//qDebug() << "Intel Order: " << sig.intelByteOrder;
//qDebug() << "start byte: " << startByte;
//qDebug() << "End Byte: " << endByte;
int sBit, eBit;
sBit = bitWithinByteStart;
eBit = sBit + bitsToGo;
if (eBit > 7) eBit = 7;
bitsToGo -= (eBit - sBit + 1);
for (int b = startByte; b <= endByte; b++)
{
//qDebug() << "Byte: " << frame.data[b];
//qDebug() << "S: " << sBit;
//qDebug() << "E: " << eBit;
//process this byte
result += processByte(frame.data[b], sBit, eBit) * multiplier;
//add to multiplier
if (!sig.intelByteOrder)
multiplier = multiplier << 8;
else
multiplier = multiplier >> 8;
//Prepare sBit and eBit for next byte
sBit = 0; //fresh byte so we start at the beginning now
eBit = sBit + bitsToGo;
if (eBit > 7) eBit = 7;
bitsToGo -= (eBit - sBit + 1);
}
if (sig.valType == SIGNED_INT)
{
int mask = (1 << (sig.signalSize - 1));
if ((result & mask) == mask) //is the highest bit possible for this signal size set?
{
/*
* if so we need to also set every bit higher in the result int too.
* This leads to the below two lines that are nasty. Here's the theory behind that...
* If the value is signed and the highest bit is set then it is negative. To create
* a negative value out of this even though the variable result is 64 bit we have to
* run 1's all of the way up to bit 63 in result. -1 is all ones for whatever size integer
* you have. So, it's 64 1's in this case.
* signedMask is done this way:
* first you take the signal size and shift 1 up that far. Then subtract one. Lets
* see that for a 16 bit signal:
* (1 << 16) - 1 = the first 16 bits set as 1's. So far so good. We then negate the whole
* thing which flips all bits. Thus signedMask ends up with 1's everwhere that the signal
* doesn't take up in the 64 bit signed integer result. Then, result has an OR operation on
* it with the old value and -1 masked so that the the 1 bits from -1 don't overwrite bits from the
* actual signal. This extends the sign bits out so that the integer result reads as the proper negative
* value. We dont need to do any of this if the sign bit wasn't set.
*/
int signedMask = ~((1 << sig.signalSize) - 1);
result = (-1 & signedMask) | result;
}
}
double endResult = ((double)result * sig.factor) + sig.bias;
result = (int) endResult;
//qDebug() << "Result: " << result;
QString outputString;
outputString = sig.name + ": ";
if (sig.valList.count() > 0) //if this is a value list type then look it up and display the proper string
{
for (int x = 0; x < sig.valList.count(); x++)
{
if (sig.valList.at(x).value == result) outputString += sig.valList.at(x).descript;
}
}
else //otherwise display the actual number and unit (if it exists)
{
outputString += QString::number(endResult) + sig.unitName;
}
return outputString;
}
//given a byte it will reverse the bit order in that byte
unsigned char DBCHandler::reverseBits(unsigned char b) {
b = (b & 0xF0) >> 4 | (b & 0x0F) << 4;
b = (b & 0xCC) >> 2 | (b & 0x33) << 2;
b = (b & 0xAA) >> 1 | (b & 0x55) << 1;
return b;
}
unsigned char DBCHandler::processByte(unsigned char input, int start, int end)
{
unsigned char output = 0, size = end - start + 1;
//first knock it down so that bottom is is start
output = input >> start;
//then mask off all bits above the proper ending
output &= ((1 << size) - 1);
return output;
}
/*
SG_ NLG5_E_B_P : 15|1@0+ (1,0) [0|1] "" Control
SG_ NLG5_S_MC_M_PI : 23|8@0+ (0.1,0) [0|20] "A" Control
SG_ NLG5_S_MC_M_CP : 7|16@0+ (0.1,0) [0|100] "A" Control
*/
dbchandler.h
+3 -1
View File
@@ -17,7 +17,7 @@ public:
explicit DBCHandler(QObject *parent = 0);
void loadDBCFile(QString);
void listDebugging();
void processSignal(CANFrame *frame, DBC_SIGNAL *sig);
QString processSignal(const CANFrame &frame, const DBC_SIGNAL &sig);
DBC_NODE *findNodeByName(QString name);
DBC_MESSAGE *findMsgByID(int id);
DBC_SIGNAL *findSignalByName(DBC_MESSAGE *msg, QString name);
@@ -30,6 +30,8 @@ private:
QList<DBC_NODE> dbc_nodes;
QList<DBC_MESSAGE> dbc_messages;
unsigned char reverseBits(unsigned char);
unsigned char processByte(unsigned char, int, int);
};
#endif // DBCHANDLER_H
mainwindow.cpp
+37 -1
View File
@@ -50,6 +50,8 @@ MainWindow::MainWindow(QWidget *parent) :
ui->canFramesView->setColumnWidth(5, 275);
QHeaderView *HorzHdr = ui->canFramesView->horizontalHeader();
HorzHdr->setStretchLastSection(true); //causes the data column to automatically fill the tableview
//enabling the below line kills performance in every way imaginable. Left here as a warning. Do not do this.
//ui->canFramesView->verticalHeader()->setSectionResizeMode(QHeaderView::ResizeToContents);
ports = QSerialPortInfo::availablePorts();
@@ -106,6 +108,9 @@ MainWindow::MainWindow(QWidget *parent) :
connect(ui->actionLoad_DBC_File, SIGNAL(triggered(bool)), this, SLOT(handleLoadDBC()));
connect(ui->actionEdit_Messages_Signals, SIGNAL(triggered(bool)), this, SLOT(showEditSignalsWindow()));
connect(ui->actionSave_DBC_File, SIGNAL(triggered(bool)), this, SLOT(handleSaveDBC()));
connect(ui->canFramesView, SIGNAL(clicked(QModelIndex)), this, SLOT(gridClicked(QModelIndex)));
connect(ui->cbInterpret, SIGNAL(toggled(bool)), this, SLOT(interpretToggled(bool)));
connect(ui->cbOverwrite, SIGNAL(toggled(bool)), this, SLOT(overwriteToggled(bool)));
lbStatusConnected.setText(tr("Not connected"));
updateBaudLabel(0,0);
@@ -118,6 +123,16 @@ MainWindow::MainWindow(QWidget *parent) :
ui->lbNumFrames->setText("0");
isConnected = false;
//create a temporary frame to be able to capture the correct
//default height of an item in the table. Need to do this in case
//of scaling or font differences between different computers.
CANFrame temp;
model->addFrame(temp, true);
normalRowHeight = ui->canFramesView->rowHeight(0);
qDebug() << "normal row height = " << normalRowHeight;
model->clearFrames();
}
MainWindow::~MainWindow()
@@ -135,6 +150,27 @@ MainWindow::~MainWindow()
delete dbcHandler;
}
void MainWindow::gridClicked(QModelIndex idx)
{
if (ui->canFramesView->rowHeight(idx.row()) > normalRowHeight)
{
ui->canFramesView->setRowHeight(idx.row(), normalRowHeight);
}
else {
ui->canFramesView->resizeRowToContents(idx.row());
}
}
void MainWindow::interpretToggled(bool state)
{
model->setInterpetMode(state);
}
void MainWindow::overwriteToggled(bool state)
{
model->setOverwriteMode(state);
}
void MainWindow::updateBaudLabel(int baud0, int baud1)
{
QString labelText;
@@ -638,7 +674,7 @@ void MainWindow::handleLoadDBC()
//right now there is only one file type that can be loaded here so just do it.
dbcHandler->loadDBCFile(filename);
dbcHandler->listDebugging();
//dbcHandler->listDebugging();
QStringList fileList = filename.split('/');
lbStatusDatabase.setText(fileList[fileList.length() - 1] + tr(" loaded."));
}
mainwindow.h
+4
View File
@@ -44,6 +44,9 @@ private slots:
void connectionFailed();
void gotDeviceInfo(int, int);
void connectionSucceeded(int, int);
void gridClicked(QModelIndex);
void interpretToggled(bool);
void overwriteToggled(bool);
public slots:
void gotFrames(int, int);
@@ -70,6 +73,7 @@ private:
QLabel lbStatusConnected;
QLabel lbStatusBauds;
QLabel lbStatusDatabase;
int normalRowHeight;
bool isConnected;
void loadCRTDFile(QString);
void loadNativeCSVFile(QString);