CMSIS_5/CMSIS/DSP/Testing/PatternGeneration/Transform.py

import os.path
import numpy as np
import itertools
import Tools
import scipy.fftpack

# Those patterns are used for tests and benchmarks.
# For tests, there is the need to add tests for saturation


FFTSIZES=[16,32,64,128,256,512,1024,2048,4096]
SINES=[0.25,0.5,0.9]
NOISES=[0.1,0.4]


def asReal(a):
    #return(a.view(dtype=np.float64))
    return(a.reshape(np.size(a)).view(dtype=np.float64))

def noiseSignal(nb):
    return(np.random.randn(nb))

def sineSignal(freqRatio,nb):
    fc = nb / 2.0
    f = freqRatio*fc 
    time = np.arange(0,nb)
    return(np.sin(2 * np.pi * f *  time/nb))

def noisySineSignal(noiseAmp,r,nb):
    return(noiseAmp*noiseSignal(nb) + r*sineSignal(0.25,nb))

def stepSignal(r,nb):
   n = int(nb/2)
   return(np.concatenate((np.zeros(n), r*np.ones(n))))

def writeFFTForSignal(config,sig,scaling,i,j,nb,signame):
    fft=scipy.fftpack.fft(sig)
    ifft = np.copy(fft)
    if scaling:
        fft = np.array([x/2**scaling[j] for x in fft])
    config.writeInput(i, asReal(sig),"ComplexInputSamples_%s_%d_" % (signame,nb))
    config.writeInput(i, asReal(fft),"ComplexFFTSamples_%s_%d_" % (signame,nb))
    config.writeInput(i, asReal(fft),"ComplexInputIFFTSamples_%s_%d_" % (signame,nb))

def writeRFFTForSignal(config,sig,scaling,i,j,nb,signame):
    rfft=scipy.fftpack.rfft(sig)
    # Changed for f32 and f64 to reproduce CMSIS behavior.
    if not scaling:
       rfft=np.insert(rfft, 1, rfft[-1])
       rfft[-1]=0.0
       
    rifft = np.copy(rfft)
    if scaling:
        rfft = np.array([x/2**scaling[j] for x in rfft])
    config.writeInput(i, (sig),"RealInputSamples_%s_%d_" % (signame,nb))
    config.writeInput(i, (rfft),"RealFFTSamples_%s_%d_" % (signame,nb))
    config.writeInput(i, (rfft),"RealInputIFFTSamples_%s_%d_" % (signame,nb))


def writeTests(configs):
    i = 1

    # Write FFT tests for sinusoid
    j = 0
    for nb in FFTSIZES:
        sig = noisySineSignal(0.05,0.7,nb)
        sigc = np.array([complex(x) for x in sig])
        for config,scaling in configs:
            writeFFTForSignal(config,sigc,scaling,i,j,nb,"Noisy")
            writeRFFTForSignal(config,sig,scaling,i,j,nb,"Noisy")
        i = i + 1
        j = j + 1

    # Write FFT tests for step
    j = 0
    for nb in FFTSIZES:
        sig = stepSignal(0.9,nb)
        sigc = np.array([complex(x) for x in sig])
        for config,scaling in configs:
            writeFFTForSignal(config,sigc,scaling,i,j,nb,"Step")
            writeRFFTForSignal(config,sig,scaling,i,j,nb,"Step")
        i = i + 1
        j = j + 1

    # Used for benchmarks
    data1=np.random.randn(512)
    data1 = Tools.normalize(data1)
    for config,scaling in configs:
        config.writeInput(i, data1,"RealInputSamples" )

   
def generatePatterns():
    PATTERNDIR = os.path.join("Patterns","DSP","Transform","Transform")
    PARAMDIR = os.path.join("Parameters","DSP","Transform","Transform")
    
    configf64=Tools.Config(PATTERNDIR,PARAMDIR,"f64")
    configf32=Tools.Config(PATTERNDIR,PARAMDIR,"f32")
    configq31=Tools.Config(PATTERNDIR,PARAMDIR,"q31")
    configq15=Tools.Config(PATTERNDIR,PARAMDIR,"q15")
    
    
    scalings = [4,5,6,7,8,9,10,11,12]
    writeTests([(configf64,None),
        (configf32,None)
        ,(configq31,scalings)
        ,(configq15,scalings)])



if __name__ == '__main__':
  generatePatterns()