"""SeeSway browser preview 0.2.0. New definitions, not legacy-equivalence certified.""" import csv,io,json,math import numpy as np VERSION='0.2.0' def analyse(payload): text=payload.get('csv','').lstrip('\ufeff') if len(text)>5_000_000: raise ValueError('File exceeds the 5 MB analysis limit.') fs=float(payload.get('sample_rate',100)); unit=payload.get('unit','mm'); method=payload.get('filter','none') if not math.isfinite(fs) or fs<=0 or fs>10000: raise ValueError('Sample rate must be greater than 0 and no more than 10000 Hz.') if unit not in ('mm','cm','m'): raise ValueError('Select mm, cm or m.') delimiter=payload.get('delimiter',',') if delimiter not in (',',';','\t'): raise ValueError('Choose comma, semicolon or tab.') columns=payload.get('columns',[0,1]);header_mode=payload.get('header','auto') if not isinstance(columns,list) or len(columns)!=2 or any(not isinstance(c,int) or isinstance(c,bool) or c<0 or c>99 for c in columns) or columns[0]==columns[1]: raise ValueError('Choose two different column numbers from 1 to 100.') if header_mode not in ('auto','yes','no'): raise ValueError('Choose the header setting.') rows=[]; header=None; expected=None for line,row in enumerate(csv.reader(io.StringIO(text),delimiter=delimiter),1): if not row or all(not c.strip() for c in row):continue if expected is None:expected=len(row) if len(row)!=expected or len(row)<=max(columns):raise ValueError(f'Row {line}: expected a consistent column count, with at least ML and AP.') if not rows and header is None: names=[row[c].strip().lower() for c in columns] if header_mode=='yes' or (header_mode=='auto' and names[0] in ('ml','medial-lateral','ml_mm') and names[1] in ('ap','anterior-posterior','ap_mm')): header=row;continue try: values=[float(v.strip().replace(',','.')) if delimiter!=',' else float(v.strip()) for v in [row[c] for c in columns]] except ValueError:raise ValueError(f'Row {line}: ML and AP must be numbers. Optional header: ML,AP.') if not all(math.isfinite(v) for v in values):raise ValueError(f'Row {line}: missing or non-finite values are not accepted.') rows.append(values) if len(rows)>100000:raise ValueError('Maximum 100000 samples per analysis.') if len(rows)<16:raise ValueError('At least 16 valid samples are required.') x=np.asarray(rows,dtype=float)*{'mm':1.,'cm':10.,'m':1000.}[unit] y=x.copy();settings={'method':method,'sample_rate_hz':fs,'input_unit':unit,'output_unit':'mm','duration_definition':'(N - 1) / sampling rate','columns_zero_based':columns,'column_names':[header[c] for c in columns] if header else None,'header_mode':header_mode,'unused_columns_zero_based':[c for c in range(expected) if c not in columns],'extra_columns':'Unused columns ignored; no vertical-force measures'} if method=='moving': raw_width=float(payload.get('window',5)) if not math.isfinite(raw_width) or not raw_width.is_integer():raise ValueError('Moving-average window must be a whole number.') width=int(raw_width) if width<3 or width>101 or width%2!=1 or width>=len(x):raise ValueError('Moving-average window must be odd, 3–101, and shorter than the trace.') k=width//2;y=np.column_stack([np.convolve(np.pad(x[:,i],(k,k),mode='reflect'),np.ones(width)/width,mode='valid') for i in range(2)]) settings.update(window_samples=width,padding='reflect, endpoints excluded') elif method=='butterworth': from scipy.signal import butter,sosfiltfilt cutoff=float(payload.get('cutoff',10));order=int(payload.get('order',6)) if not math.isfinite(cutoff) or cutoff<=0 or cutoff>=fs/2:raise ValueError('Cutoff must be positive and below half the sample rate (Nyquist).') if order<1 or order>10:raise ValueError('Butterworth order must be from 1 to 10.') sos=butter(order,cutoff,fs=fs,output='sos') try:y=sosfiltfilt(sos,x,axis=0,padtype='odd') except ValueError:raise ValueError('Trace is too short for this filter order. Use more samples or a lower order.') settings.update(cutoff_hz=cutoff,design_order=order,passes='forward/backward',padding='SciPy odd extension; default SOS pad length') elif method=='wavelet': import pywt level=3 margin=(pywt.Wavelet('sym8').dec_len-1)*(2**level-1) if len(x)<2*margin+1: raise ValueError('Symlet-8 filtering requires at least 211 samples.') extra=(-(len(x)+2*margin))%(2**level) padded=np.pad(x,((margin,margin+extra),(0,0)),mode='reflect') coeffs=pywt.swt(padded,'sym8',level=level,axis=0) approximation=[(a,np.zeros_like(d)) for a,d in coeffs] y=pywt.iswt(approximation,'sym8',axis=0)[margin:margin+len(x)] settings.update(wavelet='sym8',wavelet_level=level,transform='stationary undecimated; inverse reconstruction with all detail coefficients zero',nominal_upper_band_hz=fs/16,padding='reflect, endpoints excluded; 105 samples each end plus right padding to multiple of 8',padding_right_extra=extra) elif method!='none':raise ValueError('Unknown filter.') if not np.isfinite(y).all():raise ValueError('Analysis produced non-finite values; check input scale and settings.') duration=(len(y)-1)/fs;d=np.diff(y,axis=0);centre=y-y.mean(axis=0);path=float(np.linalg.norm(d,axis=1).sum()) metrics={'samples':len(y),'duration_s':duration,'resultant_path_mm':path,'mean_resultant_speed_mm_s':path/duration} for i,axis in enumerate(('ML','AP')): metrics.update({f'{axis}_path_mm':float(np.abs(d[:,i]).sum()),f'{axis}_range_mm':float(np.ptp(y[:,i])),f'{axis}_rms_about_mean_mm':float(np.sqrt(np.mean(centre[:,i]**2))),f'{axis}_rms_about_zero_mm':float(np.sqrt(np.mean(y[:,i]**2))),f'{axis}_sd_sample_mm':float(np.std(y[:,i],ddof=1))}) raw_path=float(np.linalg.norm(np.diff(x,axis=0),axis=1).sum()) baseline_metrics={'resultant_path_mm':raw_path,'mean_resultant_speed_mm_s':raw_path/duration} bands=wavelet_bands(x,fs) if payload.get('wavelet_bands',True) else None dfa=None if payload.get('dfa',False): if payload.get('dfa') is not True: raise ValueError('DFA selection must be a boolean.') dfa={'ML':dfa_analysis(x[:,0],fs,payload), 'AP':dfa_analysis(x[:,1],fs,payload)} settings['dfa']={'input':'unfiltered displacement in mm', 'order':1, 'integration':'cumulative sum after subtracting mean', 'segments':'non-overlapping, forward and backward; duplicate segments omitted when exactly divisible', 'scale_spacing':'up to 20 unique rounded logarithmic integer scales', 'fit':'unweighted least-squares log10 F versus log10 window size', 'min_samples':dfa['ML']['scales_samples'][0], 'max_samples':dfa['ML']['scales_samples'][-1]} step=max(1,math.ceil(len(y)/2000));idx=list(range(0,len(y),step)) if idx[-1]!=len(y)-1:idx.append(len(y)-1) return {'version':VERSION,'status':'research_preview_not_legacy_equivalent','settings':settings,'metrics':metrics,'baseline_metrics':baseline_metrics,'dfa':dfa,'wavelet_bands':bands,'plot':{'time':[i/fs for i in idx],'raw':x[idx].tolist(),'filtered':y[idx].tolist(),'display_decimation':step},'filtered':y.tolist(),'warnings':['New implementation: equivalence to 2018 SeeSway has not been established.','Plots may be decimated for display; metrics use every sample.','Only ML/AP displacement is analysed. Extra columns are ignored.']} def dfa_analysis(signal,fs,payload): """DFA-1, explicit new method; not an implementation specified in the ACLR paper.""" n=len(signal) if n<128: raise ValueError('DFA requires at least 128 samples.') lo=float(payload.get('dfa_min',16));hi=float(payload.get('dfa_max',n//4)) if not all(math.isfinite(v) and v.is_integer() for v in (lo,hi)): raise ValueError('DFA window limits must be whole sample counts.') lo,hi=int(lo),int(hi) if lo<4 or hi>n//4 or hi