SbfParser

A Python module to parse output of Septentrio receiver, including Septentrio Binary Format (SBF) & more !

This project is an update of PySbf made by Jashandeep-Sohi & Nodd and SbfParser made by MJeanneRose.

Install

Using pip :

pip install sbf-parser

From source :

pip install -e .

Release Notes

• 2x faster than SbfParser (12mo/s)
• Parsing based on SBF v4.14.10.1, including more than 35 blocks
• Can encode sbf message
• Lossless decoding, get raw binary of decoded messages

Usage

Parsing input stream

You have many examples in example directory :

• Decoding input streaming : decode.py, decode_with_memory.py
• Save parser state between calls : memory_manipulation.py
• Encoding of sbf blocks : create_complex_block.py, create_simple_block.py And somes tools in utils
• Split sbf files : split_sbf_file.py
• Re-order sbd time of week : replace_header_time.py
• Speed benchmark : benchmark.py
• Compare result of SbfParser with Septentrio sbf2ascii : comapre_with_ascii.py

import serial
from sbf_parser import SbfParser, load

# Open your files 
with open("sbf_files/log_0000.sbf", "rb") as fobj:
    for block_type, infos in load(fobj.read()):
        print(block_type, infos)


# or read directly from serial connection
parser = SbfParser()
ser = serial.Serial(
    port='COM5', baudrate=115200,\
    parity=serial.PARITY_NONE,\
    stopbits=serial.STOPBITS_ONE,\
    bytesize=serial.EIGHTBITS,\
)

while True:
    for binary in ser.read(100):
        for block_type, infos in parser.parse(binary):
            print(block_type, infos)

You can call read(path), load(fobj) or parse(binary), directly or with SbfParser to use memory between calls. infos use this structure for SBF block:

{
    'TOW': 49638143, 
    'WNc': 2368, 
    .... 
    'AGCState': [
        {'FrontendID': 1, 'Gain': 2, 'SampleVar': 4, 'BlankingStat': 8}, 
        {'FrontendID': 16, 'Gain': 32, 'SampleVar': 64, 'BlankingStat': 128}
    ], 
    
    'blockName': 'ReceiverStatus',
    'blockType': 'SBF',
    'payload': bytearray(b'$@.......')
}

block_type can be etheir, the name of the SBF block or "Replie", "Transmission", "Description", "Snmp", "BadSBF".

Encode block

Encode a basic block; for more examples, check example.py (block with sub-block or working with files).

from sbf_parser import encode
    
event_block = {
    'TOW': 123456789,  # Time of Week in milliseconds
    'WNc': 2120,       # Week number (continuous)
    'Source': 1,       # Source identifier
    'Polarity': 0,     # Polarity (0 = rising edge)
    'Offset': 0.005,   # Time offset in seconds
    'RxClkBias': 0.0,  # Receiver clock bias
    'PVTAge': 0,        # Age of PVT in ms

    'BlockType':'SBF',
    'BlockName':'ExtEvent'
}

block_bytes = encode(event_block)

with open('ext_event.sbf', 'wb') as f:
    f.write(block_bytes)

If you don't modify block, payload is the original binary representation of the block from the parser. You can set payload_priority of encode to theses values :

PAYLOAD_PRIORITY_ALWAYS = 0 # Use payload has soon we have one
PAYLOAD_PRIORITY_ONLY_ON_FAIL = -1 # Use payload for not implemented block and when encoding failed
PAYLOAD_PRIORITY_ONLY_NOT_IMPLEMENTED = -2 # Use payload only for not implemented block
PAYLOAD_PRIORITY_NO_PAYLOAD = -3 # Never use payload

Utils

You can also use some script in utils/ to edit your sbf files :

• split_sbf_file : Split Sbf file using two TOW
• replace_herder_time : Change the value of TOW, usefull when your receiver is not connected to satellites during recording
• benchmark : If you want to stress test SbfParser
• compare_with_ascii : Compare Sbf block with ascii given by RxTools

FAQ

Can I call sbf parser from C/C++ project ?

Cython allows you to call cdef functions directly from C. You may need to make minor changes to _parse function from parser.pyx to make it accessible from C, check this tutorial.

Why this code is faster even if it has more feature ?

SbfParser was using fread to parse block from files, whereas this project loads the file directly into memory and casts the blocks in the correct structure rather than reading them.

Can I read a block larger than 1 MB ?

The actual implementation uses a structure memory with a uint_8[1000000] buffer, hence the limits of 1 MB. If needed, you can fork this project and increase this memory; otherwise, larger blocks will be split in smaller unknown blocks.

For your information, large sbf blocks can be up to 4096 bytes.