Example using Pandas and csv file

[jharming]

Hi,

I am using windows with anaconda.

I tried to import pandas and then put a csv file into a subfolder called FMU_files.

It does not work like this, so is there an example I can build on ? :)

from pythonfmu.fmi2slave import Fmi2Slave, Fmi2Causality, Fmi2Variability, Real
import pandas as pd
try:
    from sympy import symbols, exp
except ImportError:  # Trick to be able to generate the FMU without sympy installed
    symbols, exp = None, None

df_ = pd.read_csv('FMU_files\data.csv', sep=';')

class Hydraulic(Fmi2Slave):
"""This class is an example to demonstrate installing new Python dependencies.
The code is not efficient.
"""

def __init__(self, **kwargs):
    super().__init__(**kwargs)
    # self.df = df_
    # self.df['rpm'] = self.df['n']
    self.Q = 1.07
    self.rpm = 1475
    self.H = 32.87
    self.torque = 165.245

    self.register_variable(Real("Q", causality=Fmi2Causality.input))
    self.register_variable(Real("rpm", causality=Fmi2Causality.input))
    self.register_variable(Real("H", causality=Fmi2Causality.output))
    self.register_variable(Real("torque", causality=Fmi2Causality.output))

def do_step(self, current_time, step_size):
    
    # Q, rpm, H, Torque = symbols("Q, rpm, H, tau")
    #expr = i * a * (1 - exp(-1 * t / tau))
    # self.H = 500. #realOut = expr.evalf(subs={i: self.realIn, a: self.a, t: current_time + step_size, tau: self.tau})
    # self.torque = 300.
    return True 

[markaren]

Did you embed the FMU_files with the FMU?

pythonfmu build -f hydraulic.py FMU_files

[jharming]

yes, I did do that. Now I tried to take your example pythonslave.py.
After each step I build using pythonfmu build -f pythonslave.py FMU_files I used FMPy GUI to test the FMU
If I only import pandas as pd then I can still build it and run it. When I try to read the csv file I can still build it but not run it. I think it must be the path I am using is wrong. I read using df = pd.read_csv('FMU_files\hyd_data.csv, sep=';')

[jharming]

I only read the csv file I dont use it for anything at the moment

[markaren]

You probably need a custom getter if you need to fetch data straight from the csv:
self.register_variable(String("file_content", getter=lambda: data, causality=Fmi2Causality.output))
where data in this case is:

with (open('FMU_files/hello.txt', 'r')) as file:
    data = file.read()

I don't see why df = pd.read_csv('FMU_files\hyd_data.csv, sep=';') should fail. If you look inside the resources folder of the FMU, the file structure should be preserved.

[markaren]

Could it be that you are using \ not / ?

[markaren]

Actually, forget what I said about the custom getter. The natural approach of reading an csv is to cache data into instance variables.

[markaren]

You might also want to look at the CSV builder that ships with pythonfmu. https://github.com/NTNU-IHB/PythonFMU/blob/master/pythonfmu/csvbuilder.py

[jharming]

Thanks for the fast replies 👍
I think I start taking a step back and then I need to understand the syntax a bit better. By the way I took the SympySlave.fmu and checked it at https://fmu-check.herokuapp.com/ here it comes up with an error. I also tried to run it in Simulink but simulink crashed. I can run it just fin using FMPy.

[markaren]

here it comes up with an error

Technically an error with the modelDescription.xml. In practice you probably will not notice. Depends on the importing tool ofcourse.

I also tried to run it in Simulink but simulink crashed.

Simulink might load some version of Python, which is not compatible. E.g. old version or 32 bit. In which case it will crash. Not much to to about it.

[markaren]

So, I think the correct way to read files is to append the value of self.resources.
This means that you need to make the call in the context of the slave instance.

class PythonSlaveReadFile(Fmi2Slave):

    def __init__(self, **kwargs):
        super().__init__(**kwargs)

        with (open(f'{self.resources}/hello.txt', 'r')) as file:
            data = file.read()