File indexing completed on 2024-04-06 12:25:45
0001 from __future__ import print_function
0002 import sys
0003
0004 def compare(fn1, fn2):
0005 f1, f2 = open(fn1, "r"), open(fn2, "r")
0006 diff_cols = {}
0007
0008 for l1, l2 in zip(f1.readlines(), f2.readlines()):
0009 if l1 == l2: continue
0010
0011 for i, (k1, k2) in enumerate(zip(l1.strip().split(), l2.strip().split())):
0012 if k1 == k2: continue
0013
0014 if i not in diff_cols:
0015 diff_cols[i] = []
0016
0017 diff = diff_cols[i]
0018
0019 try:
0020 diff_f = abs(float(k2) - float(k1))
0021 diff.append((diff_f, k1, k2, "l1: " + l1, "l2: " + l2, ))
0022
0023 print("diffrence[f%d]: %s -> %s" % (i, k1, k2))
0024 except ValueError:
0025 print("non float-type difference[f%d]: %s -> %s" % (i, k1, k2))
0026
0027 for key, item in sorted(diff_cols.items()):
0028 print("column:", key)
0029 print("\tavg: %f" % (sum(map(lambda x: x[0], item)) / len(item)))
0030
0031 m = max(item)
0032 print("\tmax:", m[:3])
0033 print("\t\t-:", m[3])
0034 print("\t\t+:", m[4])
0035
0036 return diff_cols
0037
0038 if __name__ == "__main__":
0039 if len(sys.argv) != 3:
0040 print("usage %s old.csv new.csv" % sys.argv[0])
0041 sys.exit(1)
0042
0043 diff = compare(sys.argv[1], sys.argv[2])
