pyserial data gaps - python-2.7

Question

I have python code that acquires serial data from 2 devices and writes to a .txt file. Every 4-15 minutes there is approx 30-45 seconds of data missing in the .txt file and this is not acceptable for our use case. I've spent hours googling and searching SO about multiprocessing and serial port data acquisition and haven't come up with a solution.
Here is my code
gpsser = input(("Enter GPS comport as 'COM_': "))
ser = serial.Serial(port=gpsser,
baudrate=38400,
timeout=2,
parity=serial.PARITY_NONE,
stopbits=serial.STOPBITS_ONE,
bytesize=serial.EIGHTBITS)
root = Tk()
root.title("DualEM DAQ")
path = filedialog.asksaveasfilename() + ".txt"
file = glob.glob(path)
filename = path
with open(filename, 'wb') as f:
w = csv.writer(f, dialect='excel')
w.writerow(['header'])
def sensor():
while True:
try:
NMEA1 = dser.readline().decode("ascii")
while dser.inWaiting() == 0:
pass
NMEA1_array = NMEA1.split(',')
NMEA2_array = NMEA2.split(',')
NMEA3_array = NMEA3.split(',')
NMEA4_array = NMEA4.split(',')
if NMEA1_array[0] == '$PDLGH':
value1 = NMEA1_array[2]
value2 = NMEA1_array[4]
if NMEA1_array[0] == '$PDLG1':
value3 = NMEA1_array[2]
value4 = NMEA1_array[4]
if NMEA1_array[0] == '$PDLG2':
value5 = NMEA1_array[2]
value6 = NMEA1_array[4]
return (float(value1), float(value2), float(value3),
float(value4), float(value5), float(value6),
except (IndexError, NameError, ValueError, UnicodeDecodeError):
pass
def gps():
while True:
try:
global Status, Latitude, Longitude, Speed, Truecourse, Date
global GPSQuality, Satellites, HDOP, Elevation, Time
while ser.inWaiting() == 0:
pass
msg = ser.readline()
pNMEA = pynmea2.parse(msg)
if isinstance(pNMEA, pynmea2.types.talker.RMC):
Latitude = pynmea2.dm_to_sd(pNMEA.lat)
Longitude = -(pynmea2.dm_to_sd(pNMEA.lon))
Date = pNMEA.datestamp
Time = datetime.datetime.now().time()
if () is not None:
return (Longitude, Latitude, Date, Time)
except (ValueError, UnboundLocalError, NameError):
pass
while True:
try:
with open(filename, "ab") as f:
data = [(gps() + sensor())]
writer = csv.writer(f, delimiter=",", dialect='excel')
writer.writerows(data)
f.flush()
print(data)
except (AttributeError, TypeError) as e:
pass
The program is writing to the file but I need help understanding why I'm losing 30-45 seconds of data every so often. Where is my bottle neck that is causing this to happen?
Here is an example of where the breaks are, note the breaks are approx 50 seconds in this case.
Breaks in writing data to csv
DB

Answer 1

Back when I used PySerial, I did this:
nbytes = ser.inWaiting()
if nbytes > 0:
indata = ser.read(nbytes)
#now parse bytes in indata to look for delimiter, \n in your case
#and if found process the input line(s) until delimiter not found
else:
#no input yet, do other processing or allow other things to run
#by using time.sleep()
Also note that new versions (3.0+) of PySerial have .in_waiting as a property not a method, so no (), it used to be .inWaiting().

Answer 2

You should not flush the serial port input. Data is arriving on its own timing into a buffer in the driver, not when your read happens, so you are throwing away data with your flush. You may need to add code to synchronize with the input stream.

Answer 3

I used threading with a queue and changed my mainloop to look like this.
while True:
try:
with open(filename, "ab") as f:
writer = csv.writer(f, delimiter=",", dialect='excel')
data = []
data.extend(gpsdata())
data.extend(dualemdata())
writer.writerows([data])
f.flush()
f.close()
dser.flushInput()
ser.flushInput()
print(data)
sleep(0.05)
except (AttributeError, TypeError) as e:
pass
I had to flush the serial port input data before the looping back to the read functions so it was reading new realtime data(this eliminated any lag of the incoming data stream). I've ran a 30 minute test and the time gaps appear to have go away. Thankyou to Cmaster for giving me some diagnostic ideas.