findRegressions-simple.py
3.95 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
#!/usr/bin/env python
from __future__ import print_function
import re, string, sys, os, time, math
DEBUG = 0
(tp, exp) = ('compile', 'exec')
def parse(file):
f = open(file, 'r')
d = f.read()
# Cleanup weird stuff
d = re.sub(r',\d+:\d', '', d)
r = re.findall(r'TEST-(PASS|FAIL|RESULT.*?):\s+(.*?)\s+(.*?)\r*\n', d)
test = {}
fname = ''
for t in r:
if DEBUG:
print(t)
if t[0] == 'PASS' or t[0] == 'FAIL' :
tmp = t[2].split('llvm-test/')
if DEBUG:
print(tmp)
if len(tmp) == 2:
fname = tmp[1].strip('\r\n')
else:
fname = tmp[0].strip('\r\n')
if fname not in test:
test[fname] = {}
test[fname][t[1] + ' state'] = t[0]
test[fname][t[1] + ' time'] = float('nan')
else :
try:
n = t[0].split('RESULT-')[1]
if DEBUG:
print("n == ", n);
if n == 'compile-success':
test[fname]['compile time'] = float(t[2].split('program')[1].strip('\r\n'))
elif n == 'exec-success':
test[fname]['exec time'] = float(t[2].split('program')[1].strip('\r\n'))
if DEBUG:
print(test[fname][string.replace(n, '-success', '')])
else :
# print "ERROR!"
sys.exit(1)
except:
continue
return test
# Diff results and look for regressions.
def diffResults(d_old, d_new):
regressions = {}
passes = {}
removed = ''
for x in ['compile state', 'compile time', 'exec state', 'exec time']:
regressions[x] = ''
passes[x] = ''
for t in sorted(d_old.keys()) :
if t in d_new:
# Check if the test passed or failed.
for x in ['compile state', 'compile time', 'exec state', 'exec time']:
if x not in d_old[t] and x not in d_new[t]:
continue
if x in d_old[t]:
if x in d_new[t]:
if d_old[t][x] == 'PASS':
if d_new[t][x] != 'PASS':
regressions[x] += t + "\n"
else:
if d_new[t][x] == 'PASS':
passes[x] += t + "\n"
else :
regressions[x] += t + "\n"
if x == 'compile state' or x == 'exec state':
continue
# For execution time, if there is no result it's a fail.
if x not in d_old[t] and x not in d_new[t]:
continue
elif x not in d_new[t]:
regressions[x] += t + "\n"
elif x not in d_old[t]:
passes[x] += t + "\n"
if math.isnan(d_old[t][x]) and math.isnan(d_new[t][x]):
continue
elif math.isnan(d_old[t][x]) and not math.isnan(d_new[t][x]):
passes[x] += t + "\n"
elif not math.isnan(d_old[t][x]) and math.isnan(d_new[t][x]):
regressions[x] += t + ": NaN%\n"
if d_new[t][x] > d_old[t][x] and d_old[t][x] > 0.0 and \
(d_new[t][x] - d_old[t][x]) / d_old[t][x] > .05:
regressions[x] += t + ": " + "{0:.1f}".format(100 * (d_new[t][x] - d_old[t][x]) / d_old[t][x]) + "%\n"
else :
removed += t + "\n"
if len(regressions['compile state']) != 0:
print('REGRESSION: Compilation Failed')
print(regressions['compile state'])
if len(regressions['exec state']) != 0:
print('REGRESSION: Execution Failed')
print(regressions['exec state'])
if len(regressions['compile time']) != 0:
print('REGRESSION: Compilation Time')
print(regressions['compile time'])
if len(regressions['exec time']) != 0:
print('REGRESSION: Execution Time')
print(regressions['exec time'])
if len(passes['compile state']) != 0:
print('NEW PASSES: Compilation')
print(passes['compile state'])
if len(passes['exec state']) != 0:
print('NEW PASSES: Execution')
print(passes['exec state'])
if len(removed) != 0:
print('REMOVED TESTS')
print(removed)
# Main
if len(sys.argv) < 3 :
print('Usage:', sys.argv[0], '<old log> <new log>')
sys.exit(-1)
d_old = parse(sys.argv[1])
d_new = parse(sys.argv[2])
diffResults(d_old, d_new)