An Arduino library to parse NMEA sentences.

NMEA is a communication standard in the marine equipment industry: GPS, anemometers,... The NMEAParser library allows you to analyze NMEA sentences and associate handlers to each of those that need to be recognized. The library provides the means to associate a handler to each identifier and also provides functions to retrieve the arguments of the sentence in different data formats: character, integer, float, character string.

• 1.1 : Added joker in type. Remove binaries from extra
• 1.0 : Initial release

On an Arduino Uno, an instance of a NMEAParser requires 97 bytes with only one handler. 8 bytes per additional handler are required.

As usual, the library should be included at the beginning of the sketch.

#include <NMEAParser.h>

Then, you must instantiate a parser as follows:

NMEAParser<4> parser;

The 4 is the maximum number of handlers desired, parser is the name of the object that will allow the analysis of NMEA sentences. you can of course use the name you want.

In setup you configure your parser as you wish using the following functions.

where <type> is a character string and the type of sentence to recongnize, and <handler> the function to call when a sentence is recognize. <type> can be a string stored un RAM or a string stored in flash : F("ASTRI"). If <type> has more than 5 characters, it is trucated.

For example, suppose you want to recognize the following sounder sentences which give the water depth below keel (DBK) and below surface (DBS):

$SDDBK,...

and

$SDDBS,...

You will install two handlers as follows (assuming your parsing object is named parser):

parser.addHandler("SDDBK", handleDepthBelowKeel);
parser.addHanlder("SDDBS", handleDepthBelowSurface);

handleDepthBelowKeel and handleDepthBelowSurface are functions that will be called when sentences are recognized.

With version 1.1, <type> may include hyphens. An hyphens matches any character. For instance if you want the handler to match all sentences coming from the sounder, you would write:

parser.addHandler("SD---", handleSounder);

handleSounder function would be called for any sentence with the type beginning with SD.

When a sentence is succefully parsed but none of the handler correspond to it, <handler> is called. It is a function corresponding to a void f(void) prototype. By default, no function is called.

When a sentence is malformed : too long, wrong characters, etc, <handler> is called. It is a function corresponding to a void f(void) prototype. By default, no function is called.

Specifies whether the CRC is checked or not. By default, the CRC is checked. If you do not want CRC verification, pass false to setHandleCRC.

In the handlers, you will get the arguments of the sentence, the sentence type or the error if any by using the following functions:

is used to get the arguments of the sentence. <argnum> is the number of the argument, starting at 0 for the argument that follows the identifier. <argval is a variable where the argument value will be stored if successful. getArg returns a boolean which is true if successfull, false if not.

Continuing with the example, both sentences have the same 6 arguments

• Argument 0 is a float number giving the depth in feet.
• Argument 1 is a f for feet.
• Argument 2 is a float number giving the depth in meters.
• Argument 3 is a M for Meters.
• Argument 4 is a float number giving the depth in fathoms.
• At last Argument 5 is a F for Fathoms.

Suppose you are interested by the depths in meters and you have two float variables to store theses data:

float depthBelowKeel;
float depthBelowSurface;

You would implement handleDepthBelowKeel and handleDepthBelowSurface as follow:

void handleDepthBelowKeel(void)
{
  float depth;
  if (parser.getArg(2, depth))
  {
    depthBelowKeel = depth;
  }
}

void handleDepthBelowSurface(void)
{
  float depth;
  if (parser.getArg(2, depth))
  {
    depthBelowSurface = depth;
  }
}

3 versions of getType exist. The first one puts the type of the sentence in <type> which is a char *. The second one does the same but <type> is a String. Return true if a type has been parsed, false otherwise. The third one puts a character of the type at position <num>

Return the number of arguments.

Return the error. The returned code can be:

• NMEA::NO_ERROR: no error;
• NMEA::UNEXPECTED_CHAR: a char which is not expected in the sentence has been encountered;
• NMEA::BUFFER_FULL: the sentence is too long to fit in the buffer;
• NMEA::TYPE_TOO_LONG: the sentence type has more than 5 characters;
• NMEA::CRC_ERROR: the CRC is wrong;
• NMEA::INTERNAL_ERROR: the internal state of the parser is wrong, should not happen by the way.

Characters are fed to the parser in loop, assuming we get the characters from Serial, the following way:

while (Serial.available()) {
  parser << Serial.read();
}

This can also be done in serialEvent. while could be replaced by if.

Let's say you want to turn the Arduino's LED on and off. We define a NMEA sentence taking a single argument: 1 to turn on the LED and 0 to turn it off. The sentence can therefore be:

$ARLED,1*43

to turn the LED on and

$ARLED,0*42

to turn the LED off. We define a single handler to retrieve the argument and control the LED accordingly:

void ledHandler()
{
  Serial.print("Got ARLED with ");
  Serial.print(parser.argCount());
  Serial.println(" arguments");
  int ledState;
  if (parser.getArg(0,ledState)) {
    digitalWrite(LED_BUILTIN, ledState == 0 ? LOW : HIGH);
  }
}

We define 2 other handlers for anything else than ARLED and for errors

void errorHandler()
{
  Serial.print("*** Error : ");
  Serial.println(parser.error());
}

void unknownCommand()
{
  Serial.print("*** Unkown command : ");
  char buf[6];
  parser.getType(buf);
  Serial.println(buf);
}

In setup, the handlers are installed:

void setup() {
  Serial.begin(115200);
  parser.setErrorHandler(errorHandler);
  parser.addHandler("ARLED", ledHandler);
  parser.setDefaultHandler(unknownCommand);
  pinMode(LED_BUILTIN, OUTPUT);
}

At last in loop, we feed the parser with the chars coming from Serial.

void loop() {
  if (Serial.available()) {
    parser << Serial.read();
  }
}

Additional software can be found in the extra directory.

The gen subdirectory contains nmeagen, a NMEA sentence generator program. This program generates well formed sentences with good or bad CRC. It can be used to test the parser. To build nmeagen, run the build.sh script. nmeagen takes 1 or 2 arguments. The first argument is the number of sentences to generate. The second optional one is the number of sentences with bad CRC.

The test subdirectory contains a test program that compile on Linux or Mac OS X. It takes sentences from the standard input, parse them and print out the type, the arguments and if an error occured.

• NMEA CRC calculator
• NMEA 0183 on Wikipedia
• NMEA revealed

That's all folks !