Any python coders about?

[philthy]

I've been building a raspberry pi based tracker.
Followed a tutorial online, and it's up and running, but not as well as it could.
The GPS data is parsed by pynmea, then uploaded to my live database.The problem is, loss of GPS signal. The code defaults back to a lat & lng of 0.0 0.0
Frustrating, as while watching it live, it suddenly leaps off to the west of Africa, then returns, leaving a trace line.
I've tried tweaking it, but I'm not familiar with python (or any other code for that matter), but can't crack it.
My thinking is, if lat & lng = 0.0 don't update database, just leave the old coordinates, until GPS signal returns.
Here's the code:

firebase=pyrebase.initialize_app(firebaseConfig)
db=firebase.database()

while True:
port="/dev/ttyAMA0"
ser=serial.Serial(port, baudrate=9600, timeout=0.5)
dataout = pynmea2.NMEAStreamReader()
newdata=ser.readline()
n_data = newdata.decode('latin-1')
if n_data[0:6] == '$GPRMC':
newmsg=pynmea2.parse(n_data)
lat=newmsg.latitude
lng=newmsg.longitude
gps = "Latitude=" + str(lat) + " and Longitude=" + str(lng)
print(gps)
data = {"LAT": lat, "LNG": lng}
db.update(data)
print("Data sent")

Thanks in advance for any help.

[sbridgey]

I’d probably check the gps_qual property to ensure that the data is trustworthy?

[ATG]

If you can test for quality, that'd be good. If not, your original suggestion would work too and you could write it like this:

if lat == 0 and lng == 0:
continue

I imagine PH will screw up the formatting, but the "continue" needs to be indented by however many spaces you used to indent the block after "while True"

"Continue" means "start the next iteration of the current loop immediately"

Given you were testing for zero, there are some other options, but what I've written is the most readable way of expressing your logic.

Edited by ATG on Thursday 27th March 13:21

[ATG]

Re. "quality" it's been a long, long time since I played with NMEA . If memory serves, it is signal compatible with our old friend RS233. Seeing 9600 baud jogged my memory as that seemed tragically slow even in the 1990s. Anyway, I'm pretty sure it has an "estimated position error" value as part of each fix, so you could test that that is sufficiently small before you record a fix.

[philthy]

Thanks guys, some food for thought there.
It's the formatting that's throwing me I think?
Lat & lng are uploaded as 0.0 values.

[fat80b]

ChatGPT says :

You're on the right track! The issue is that when your GPS signal is lost, pynmea2 likely returns 0.0 for latitude and longitude, and your code blindly updates the database with these values.

A simple fix is to check if the new coordinates are (0.0, 0.0), and only update the database when valid data is received. Here's the corrected version:

python
Copy
Edit
import serial
import pynmea2
import pyrebase

1. Initialize Firebase

firebase = pyrebase.initialize_app(firebaseConfig)
db = firebase.database()

1. Store last known location

last_lat = None
last_lng = None

1. Serial port setup

port = "/dev/ttyAMA0"
ser = serial.Serial(port, baudrate=9600, timeout=0.5)
dataout = pynmea2.NMEAStreamReader()

while True:
newdata = ser.readline()
try:
n_data = newdata.decode('latin-1').strip()
if n_data.startswith('$GPRMC'):
newmsg = pynmea2.parse(n_data)
lat = newmsg.latitude
lng = newmsg.longitude

# Check if lat & lng are valid
if lat != 0.0 and lng != 0.0:
last_lat, last_lng = lat, lng # Update last known good location
data = {"LAT": lat, "LNG": lng}
db.update(data)
print(f"Data sent: Latitude={lat}, Longitude={lng}")
else:
print("Invalid GPS data received, keeping last known location.")
except (UnicodeDecodeError, pynmea2.ParseError):
print("Error parsing GPS data, skipping.")



What this does:
Stores the last known location (last_lat, last_lng).

Checks if the new GPS data is valid (i.e., not (0.0, 0.0)). If valid, it updates Firebase.

If invalid GPS data is received, it simply ignores it and keeps the last known good location.

Handles errors like decoding or parsing issues to prevent crashes.

Now, when the signal drops, your tracker won’t jump to the west of Africa! 🚀 Let me know if you need any tweaks.

[NWMark]

For a quick basic fix to exclude any 0.0 returned data you can add this line into your code, will only work if lat and lng are actually strings of data of '0.0' if they are '0.000', change it to that, if the are numbers (probably floats) remove the '' so its just 0.0

if lat !='0.0' and lng !='0.0':

if n_data[0:6] == '$GPRMC':
newmsg=pynmea2.parse(n_data)
lat=newmsg.latitude
lng=newmsg.longitude
if lat !='0.0' and lng !='0.0':
gps = "Latitude=" + str(lat) + " and Longitude=" + str(lng)
print(gps)
data = {"LAT": lat, "LNG": lng}
db.update(data)
print("Data sent")

Ideally you'd want to do the check on the if n_data line so your not parsing data unnecessarily, but id need to know what a whole line looks like to advise on that.
Guessing it may be something like

if n_data[0:6] == '$GPRMC' and n_data[0:14] != '$GPRMC 0.0 0.0':


EDIT: beaten to it by ChatGPT

Edited by NWMark on Thursday 27th March 14:49

[philthy]

NWMark said:

For a quick basic fix to exclude any 0.0 returned data you can add this line into your code, will only work if lat and lng are actually strings of data of '0.0' if they are '0.000', change it to that, if the are numbers (probably floats) remove the '' so its just 0.0

if lat !='0.0' and lng !='0.0':

if n_data[0:6] == '$GPRMC':
newmsg=pynmea2.parse(n_data)
lat=newmsg.latitude
lng=newmsg.longitude
if lat !='0.0' and lng !='0.0':
gps = "Latitude=" + str(lat) + " and Longitude=" + str(lng)
print(gps)
data = {"LAT": lat, "LNG": lng}
db.update(data)
print("Data sent")

Ideally you'd want to do the check on the if n_data line so your not parsing data unnecessarily, but id need to know what a whole line looks like to advise on that.
Guessing it may be something like

if n_data[0:6] == '$GPRMC' and n_data[0:14] != '$GPRMC 0.0 0.0':


EDIT: beaten to it by ChatGPT

Edited by NWMark on Thursday 27th March 14:49

Throws this error:


pi@raspi:~ $ python3 gps_send.py
Traceback (most recent call last):
File "/home/pi/gps_send.py", line 28, in <module>
if lat !='0.0' and lng !='0.0':
NameError: name 'lat' is not defined

[philthy]

fat80b said:

ChatGPT says :

You're on the right track! The issue is that when your GPS signal is lost, pynmea2 likely returns 0.0 for latitude and longitude, and your code blindly updates the database with these values.

A simple fix is to check if the new coordinates are (0.0, 0.0), and only update the database when valid data is received. Here's the corrected version:

python
Copy
Edit
import serial
import pynmea2
import pyrebase

1. Initialize Firebase

firebase = pyrebase.initialize_app(firebaseConfig)
db = firebase.database()

1. Store last known location

last_lat = None
last_lng = None

1. Serial port setup

port = "/dev/ttyAMA0"
ser = serial.Serial(port, baudrate=9600, timeout=0.5)
dataout = pynmea2.NMEAStreamReader()

while True:
newdata = ser.readline()
try:
n_data = newdata.decode('latin-1').strip()
if n_data.startswith('$GPRMC'):
newmsg = pynmea2.parse(n_data)
lat = newmsg.latitude
lng = newmsg.longitude

# Check if lat & lng are valid
if lat != 0.0 and lng != 0.0:
last_lat, last_lng = lat, lng # Update last known good location
data = {"LAT": lat, "LNG": lng}
db.update(data)
print(f"Data sent: Latitude={lat}, Longitude={lng}")
else:
print("Invalid GPS data received, keeping last known location.")
except (UnicodeDecodeError, pynmea2.ParseError):
print("Error parsing GPS data, skipping.")



What this does:
Stores the last known location (last_lat, last_lng).

Checks if the new GPS data is valid (i.e., not (0.0, 0.0)). If valid, it updates Firebase.

If invalid GPS data is received, it simply ignores it and keeps the last known good location.

Handles errors like decoding or parsing issues to prevent crashes.

Now, when the signal drops, your tracker won’t jump to the west of Africa! ?? Let me know if you need any tweaks.

I'm having problems with indents when I copy and paste, but worked through it. It then returned :

pi@raspi:~ $ python3 gps_send.py
File "/home/pi/gps_send.py", line 41

^
SyntaxError: unexpected EOF while parsing

[philthy]

sbridgey said:

I’d probably check the gps_qual property to ensure that the data is trustworthy?

Having a look into this now.
gps_qual returns 0 if not locked.
Now how to squeeze it in the code......

[philthy]

fat80b said:

Stuff .

A virtual pint to you good man.
I was ballsing up the indents and formatting when copying and pasting.

[ATG]

Yeah, you can't copy and paste that stuff raw from Pistonheads because the code has lost all its indentation. Indentation is part of Python's core syntax (which is unusual amongst programming languages, but rather a good idea). Everyone who is new to programming struggles with syntax errors. Every dot, comma, bracket, etc has to be in just the right place or nothing works at all. Once you've built up a little bit of experience syntax errors will look as obvious and wrong to you as they do to the computer.

There are a couple of things that could be going wrong.

Programming languages treat text and numbers as being two completely separate things. So if I've set the variable "f" equal to 3 like this:

f = "3"

then f is a string ... its a piece of text, and I can't do things like divide it by a number or subtract something from it.

However, if instead I wrote:

f = 3

then f is a number, and I can divide it by another number, or subtract something from it or whatever.

Crucially, if I wrote this (and I'm using dots here in place of spaces so that you can see where the indentation needs to go):

f = 3
g = "3"

if f == g:
......print("u wot??")
else:
......print("obviously")

The code will print "obviously" because a bit of text is never equal to a number. f and g are different types of things.

Why am I banging on about this? Because you need to know if pynmea.parse() returns the lat and long as numbers or strings. I'd expect it to be as numbers as latitude and longitude are conceptually numbers ... but ... it might just be returning them as strings.

And it matters because "0.0" is never equal to 0.

My code example assumed lat and lng were numeric, so I compared them to the number 0. NWMark's code is treating them as strings .. i.e. he's writing '0.0'


And ChatGPT's suggested code is horrible. It might work if indentation was sorted out, but initialising lat and lng with None, the place where it is catching the exception and its logical expression lat != 0.0 and lng != 0.0 are gopping and a good example of how it is risky to use ChatGPT to learn how to do things.

[ATG]

philthy said:

Having a look into this now.
gps_qual returns 0 if not locked.
Now how to squeeze it in the code......

Probably as simple as this:

while True:

........<your code that pulls data and parses it>


........if newmsg.gps_qual == 0:
............continue

........<your code that writes lat and lng to your database>

[jeremyc]

Top tip: use braces to preserve formatting in your code when posting on PH.

f = 3
g = "3"

if f == g:
      print("u wot??")
else:
      print("obviously")

[ATG]

jeremyc said:

Top tip: use braces to preserve formatting in your code when posting on PH.

f = 3
g = "3"

if f == g:
      print("u wot??")
else:
      print("obviously")

Cool, thanks!

[philthy]

ATG said:

Probably as simple as this:

while True:

........<your code that pulls data and parses it>


........if newmsg.gps_qual == 0:
............continue

........<your code that writes lat and lng to your database>

Thanks for that. It is working, but as you say, using gps_qual strikes me as more elegant.
I'll have a tweak of the code another time, I think my brain melted earlier?

[ATG]

philthy said:

ATG said:

Probably as simple as this:

while True:

........<your code that pulls data and parses it>


........if newmsg.gps_qual == 0:
............continue

........<your code that writes lat and lng to your database>

Thanks for that. It is working, but as you say, using gps_qual strikes me as more elegant.
I'll have a tweak of the code another time, I think my brain melted earlier?

If you're new to this sort of stuff and you've managed to get _anything_ working, then you're doing really well. "Elegant" is exactly the right way to think about this stuff, and elegant is better because it is easier to understand. But "working" is a lot more important than "elegant".

[philthy]

ATG said:

If you're new to this sort of stuff and you've managed to get _anything_ working, then you're doing really well. "Elegant" is exactly the right way to think about this stuff, and elegant is better because it is easier to understand. But "working" is a lot more important than "elegant".

Amen to that!
I will twiddle with it though.

Next thing is to get my webpage authentication to firebase sorted. Currently I have it as unrestricted, which obviously is far from ideal. Again, the code looks logical, but it's syntax that's confusing me.

[droopsnoot]

philthy said:

Thanks for that. It is working, but as you say, using gps_qual strikes me as more elegant.

As well as allowing it to record position when either of the values is legitimately zero.

[davek_964]

ATG said:

....If memory serves, it is signal compatible with our old friend RS233.

It's a very long time since I messed about with UARTs - but have you applied inflation there?