add script for splitting into different acc, ecg, ppg125, ppg512 csv file

Author: rbtchc

convert.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+import argparse
+import numpy as np
+import os
+
+def parse_args():
+    p = argparse.ArgumentParser()
+    p.add_argument('raw_data_file', nargs=1, help='Specify the raw data file')
+    return vars(p.parse_args())
+
+MSEC_PER_SEC = 1000
+
+def get_interpolated_ts(ts, ratio):
+    new_ts = None
+    vals, idx = np.unique(ts, return_index=True)
+    idx = np.hstack((idx,ts.shape[0])) # so we can do idx[i+1] - idx[i]
+    vals = np.hstack((vals,vals[-1] + MSEC_PER_SEC))
+    for i in xrange(vals.size-1):
+        num = idx[i+1] - idx[i]
+        # we have to expand time intervals according to the ratio for
+        # we might have multiple data in one row
+        tmp = np.linspace(vals[i], vals[i+1], num * ratio, False)
+        new_ts = np.append(new_ts, tmp) if new_ts is not None else tmp
+    return new_ts
+
+def convert_ecg_to_mv(raw, ratio=1000./6/2097152):
+    idx = np.where(raw >= 4194304)
+    raw[idx] -= 8388608
+    return raw * ratio
+
+def convert_ppg_to_mv(raw):
+    return convert_ecg_to_mv(raw, 3.2*1000/65536)
+
+if __name__ == "__main__":
+    args = parse_args()
+    input_file = os.path.basename(args["raw_data_file"][0])
+
+    # 0: ACC, 5: ECG, 9: PPG 125 Hz, 12: PPG 512 Hz
+    basename = os.path.splitext(input_file)[0]
+    params = {
+             #type: [filename, wanted indexes, data per row, data per item, convert function (if any)]
+              0:  [ basename + "_acc.csv", [2,3,4,6,7,8,10,11,12], 3, 3, None],
+              5:  [ basename + "_ecg.csv", range(2,13), 11, 1, convert_ecg_to_mv],
+              9:  [ basename + "_ppg125.csv", range(2,13,2), 6, 1, convert_ppg_to_mv],
+              12: [ basename + "_ppg512.csv", range(2,14), 12, 1, convert_ppg_to_mv],
+             }
+
+    data = np.genfromtxt(input_file, delimiter=",")
+    for data_type in params.keys():
+        fname, wanted, data_per_row, data_per_item, fn_convert = params[data_type]
+
+        # check if data_type exists in the input file
+        idx = np.where(data[:,0] == data_type)[0]
+        if len(idx) <= 0:
+            continue
+
+        # filter out specific type
+        raw = data[idx]
+        # filter out columns and reshape into given row, column
+        raw = raw[:,wanted].reshape(raw.shape[0] * data_per_row, data_per_item)
+        # if further convert() is required, call it
+        if fn_convert:
+            raw = fn_convert(raw)
+        # get the interpolated timestamp
+        ts = get_interpolated_ts(data[idx,-1] * MSEC_PER_SEC, data_per_row)
+        # match time stamps w/ the parsed data
+        output = np.column_stack((ts, raw))
+        # dump into output file
+        np.savetxt(fname, output, delimiter=',')
+