Introduction
The CI Data File Reader API consists of two DLL files that can be integrated with custom software to directly read and extract data from an ATFX file. This article will demonstrate how to read various signals from an ATFX file and convert them into a Pandas Dataframe following a specific format.

For a simple way to import the C# dll files and open an ATFX file in Python & Matlab, please refer to the How to Read CI Data Files in Python and / or How to Read CI Data Files in Matlab article.

For more detailed information on how to implement the following code sections and properties in a class, please refer to the CI Data File Reader manual and the provided C#, Python & Matlab Demo code. The package can be downloaded from our Programming Corner.

The following Python script uses two packages, numpy and pandas, that can be installed using the following commands in the operating system command prompt or integrated development environment software terminal.

pip install numpy

pip install pandas

Reading and Converting to Pandas Dataframe
The following python script can be found in: Convert_to_Pandas_Dataframe.py.

The import modules for the script are below.

#---Pythonnet clr import
from tokenize import Name
import clr
# Change file path here to whereever the DLL files are
parentPath = "C:\\MyStuff\\DevelopmentalVer\\bin\\AnyCPU\\Debug\\Utility\\CIATFXReader\\"

clr.AddReference(parentPath + "CI.ATFX.Reader.dll")
clr.AddReference(parentPath + "Common.dll")
clr.AddReference('System.Linq')
clr.AddReference('System.Collections')

import numpy as np
import pandas as pd

#---C# .NET imports & dll imports
from EDM.Recording import *
from EDM.RecordingInterface import *
from ASAM.ODS.NVH import *
from EDM.Utils import *
from Common import *
from Common import _SpectrumScalingType
from Common.Spider import *
from System import *
from System.Diagnostics import *
from System.Reflection import *
from System.Text import *
from System.IO import *

The next function takes in the ATFX file path, extracts the signals, converts and outputs to a pandas dataframe.

The function initially starts by opening the ATFX file, obtaining the signal list, and setting up some starting variables.

# Function for converting signals from an ATFX file into a pandas dataframe in a specific order below:
# Frequency, profile(f), control(f), HighAbort, HighAlarm, LowAlarm, LowAbort, APS / Spectrum (Table), APS / Spectrum (Ch#)
def ConvertAllSignalsToDataframe(atfxFilePath):
    #OpenRecording(string, out IRecording)
    # openRecordSucceed is required for the OpenRecording as it is the returned boolean
    # Make sure to reference the correct file string
    openRecordSucceed, recording1 = RecordingManager.Manager.OpenRecording(atfxFilePath, None)

# Get a list of signals
signalList = Utility.GetListOfAllSignals(recording1)

PREFIX = 'EoT '
df = pd.DataFrame()

Signal list filtering can filter drive or time domain signals from the list. It can also keep any frequency domain signals while filtering out specified signals, such as H(f).

    # Find the specific signal in the signal list
    for signal in signalList:
        # Skip over any signals with drive or Block in its name
        if 'drive' in signal.Name or 'Block' in signal.Name:
            continue
        # Skip any other signals that doesn't have the below strings in its name, such as H(f)
        if( 'Spectrum' not in signal.Name and 'APS' not in signal.Name and
            'profile' not in signal.Name and 'control' not in signal.Name and
            'HighAbort' not in signal.Name and 'HighAlarm' not in signal.Name and
            'LowAlarm' not in signal.Name `'LowAbort' not in signal.Name):
            continue

Once the signals are filtered, next is obtaining the signal frame data, then converting the System.Double[] object into a numpy array and assigning it to a pandas dataframe.

        # Get the signal frame and if the ATFX file is from a Swept Sine test then use (EU)RMS
        # and get the default engineering unit by passing in a empty string
        if recording1.RecordingProperty.MeasurementType == MeasurementConfigType.VCS_Sine:
            signalFrame = signal.GetFrame(0, _SpectrumScalingType.EURms, '')
        else:
            signalFrame = signal.GetFrame(0, _SpectrumScalingType.EURMS2, '')

        # if the signal is profile(f), get the Frequency column and profile(f) column and assign them at the beginning of the dataframe
        # There are chances where other signals such as control(f) went before profile(f) in the signalList order
        if 'profile' in signal.Name:
            frequencyFrame      = np.fromiter(signalFrame[0], float)
            profileFrameY       = np.fromiter(signalFrame[1], float)
            df.insert(0, PREFIX + "Frequency", frequencyFrame)
            df.insert(1, PREFIX + signal.name, profileFrameY)

        # Every other signal gets appended to the dataframe normally
        signalFrameY = np.fromiter(signalFrame[1], float)
        df[PREFIX + signal.name] = signalFrameY

After the signal list has been filtered and the pandas dataframe contains the necessary data, next is the proper ordering of the columns in the dataframe which is as follows:

Frequency, profile(f), control(f), HighAbort, HighAlarm, LowAlarm, LowAbort, APS / Spectrum (Table), APS / Spectrum (Ch#)

    # Set up the proper order by getting the specific columns of the first dataframe and assigning to a new dataframe
    df2 = df[[PREFIX + 'Frequency', PREFIX + 'profile(f)', PREFIX + 'control(f)',
              PREFIX + 'HighAbort(f)', PREFIX + 'HighAlarm(f)', PREFIX + 'LowAlarm(f)',
              PREFIX + 'LowAbort(f)']]
    # Then drop the above columns to concat the first and second dataframe
    df.drop(columns=[PREFIX + 'Frequency', PREFIX + 'profile(f)', PREFIX + 'control(f)',
              PREFIX + 'HighAbort(f)', PREFIX + 'HighAlarm(f)', PREFIX + 'LowAlarm(f)',
              PREFIX + 'LowAbort(f)'])
    df3 = pd.concat([df2, df])
    df3 = df3.apply(pd.to_numeric, errors='coerce')
    df3 = df3.dropna()
    return df3

The main run block of the script is below with an example screenshot.

# Change file path here to whereever signal or recording files are
recordingPath = "C:\\Users\\KevinCheng\\Documents\\EDM\\importapssignals2\\Sine6\\"
# ATFX file path, change contain the file name and correctly reference it in RecordingManager.Manager.OpenRecording
recordingPathRegular = recordingPath + "Run2 Aug 19, 2022 12-22-48\\SIG0001.atfx"
dataframe = ConvertAllSignalsToDataframe(recordingPathRegular)
print(dataframe.columns)
print(dataframe)

CI Blog