feat: Spline_interpolate changed to interpolate and filter either in rows or cols

	Keyword parameter 'rowwise' added. It can take value True or False

processing.py

@@ -190,8 +190,8 @@ def holey_periodogram(data, fs=1.0, window='boxcar', FFTlength=None, detrend='co
             return ( np.linspace(0, 0.5*fs, num= N2, endpoint=False),   np.square(np.abs(F[:,N2:])/N))
         else :
             return ( np.linspace((-0.5+1/N)*fs, 0.5*fs, num= N2, endpoint=True),   np.square(np.abs(F)/N))
-
-def spline_interpolate(data, window=('tukey', 0.2) ):
+'''
+def spline_interpolateA(data, window=('tukey', 0.2) ):
     windata=data*scipy.signal.get_window(window, data.shape[1])
     xbase=np.arange(0., data.shape[1],dtype=float)
     outrow=0
@@ -229,6 +229,64 @@ def spline_interpolate(data, window=('tukey', 0.2) ):
     print("output shape", outdata.shape)
     return outdata
     
+'''
+
+def spline_interpolate(data, rowwise=True, window=('tukey', 0.2) ):
+    if rowwise:
+        segment_size=data.shape[1]
+        segment_num=data.shape[0]
+    else: 
+        segment_size=data.shape[0]
+        segment_num=data.shape[1]
+
+    windata=data*scipy.signal.get_window(window, segment_size)
+    xbase=np.arange(0., segment_size,dtype=float)
+    index_out=0
+    invalids=0
+    outdata=np.ma.empty(dtype=float, shape=data.shape)
+    for index_in in range(segment_num):
+        
+        segment=np.ma.array(windata[index_in, :], mask=windata[index_in, :].mask) if rowwise else \
+                ma.array(windata[:,index_in], mask=windata[:,index_in].mask)
+        if segment.mask[0]:
+            segment[0]=0
+            segment.mask[0]=False
+        if segment.mask[-1]:
+            segment[-1]=0
+            segment.mask[-1]=False
+                
+        num_invalid=segment[segment.mask].size
+        if num_invalid :
+            print (index_in, ":" , num_invalid, "invalid data points")
+            if num_invalid > 0.2 * segment_size :
+                print( "   too many datapoints, row/col will be skipped from computation")
+                invalids+=segment_size
+                continue
+                
+            invalids+=num_invalid
+            
+            x=xbase[~ segment.mask]
+            y=segment[~ segment.mask]
+            # defining 1st derivative null at ends since Tuckey window was applied (periodic is not working)
+            spline3=scipy.interpolate.make_interp_spline(x, y, bc_type=([(1, 0.0)], [(1, 0.0)]))
+            if rowwise:
+                outdata[index_out,:]=spline3(xbase, extrapolate='periodic')
+            else:
+                outdata[:,index_out]=spline3(xbase, extrapolate='periodic')
+        else:
+            if  rowwise:
+                outdata[index_out,:]=windata[index_in,:]
+            else:
+                outdata[:,index_out]=windata[:,index_in]
+        index_out+=1
+    if  rowwise:
+        np.ma.resize(outdata, (index_out,))
+    else:
+        np.ma.resize(outdata, (data.shape[0],index_out))
+    print ("percent of invalid data=", 100*invalids/(data.shape[0]*data.shape[1]) )
+    print("output shape", outdata.shape)
+    return outdata
+    
 def makePowerList(nx,ny, N):
     """! Create a power list suitable for use with fit2D_poly where
     @param nx is the maximum degree in X