-
Notifications
You must be signed in to change notification settings - Fork 0
/
exercise.py
executable file
·146 lines (139 loc) · 7.58 KB
/
exercise.py
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
#!/usr/bin/python3
import sqlite3, sys, json, glob, argparse
from collections import defaultdict
from datetime import datetime
def eval_exercise(c, gold_data, langs, skip_incomplete, date_from, date_to, ex_id = None):
if ex_id:
ex = c.execute("SELECT * FROM exercises WHERE id=?", (ex_id,)).fetchall()
else:
ex = c.execute("SELECT * FROM exercises")
all_results = []
file_results = {}
for e in ex:
if "ALL" not in langs and e["lang"] not in langs:
continue
results = json.loads(e["results"])
if "current" in results:
del results["current"]
if skip_incomplete:
print("WARNING: exercise %s is incomplete, skipping; use -a to include this" % e["id"], file=sys.stderr)
continue
results.pop("previous", None)
ex_sets = results.keys()
correct = 0
skipped = 0
total = len(ex_sets)
mistakes = []
for s in ex_sets:
if "when" in results[s]:
when = datetime.fromisoformat(results[s]["when"])
else:
when = None
if when and ((date_from and when < date_from) or (date_to and when > date_to)):
continue
if s not in file_results:
file_results[s] = {"correct": 0, "total": 0, "skipped": 0, "words": {}}
file_results[s]["total"] += 1
outlier = results[s]["outlier"]
status = "wrong"
if outlier in gold_data[s]["outliers"]:
correct += 1
file_results[s]["correct"] += 1
status = "correct"
else:
outlier = set(results[s]["words"]) & set(gold_data[s]["outliers"])
if len(outlier) > 1:
raise RuntimeError("More than one outlier in the set!")
elif not outlier:
continue
outlier = outlier.pop()
if results[s]["outlier"] == "=SKIP=":
skipped += 1
file_results[s]["skipped"] += 1
status = "skipped"
if outlier not in file_results[s]["words"]:
file_results[s]["words"][outlier] = {"correct": 0, "total": 0, "skipped": 0, "mistakes": defaultdict(int), "ids": []}
file_results[s]["words"][outlier]["correct"] += int(status == "correct")
file_results[s]["words"][outlier]["total"] += 1
if status == "wrong":
file_results[s]["words"][outlier]["mistakes"][results[s]["outlier"]] += 1
mistakes.append((s, results[s]["outlier"], outlier))
file_results[s]["words"][outlier]["skipped"] += int(status == "skipped")
file_results[s]["words"][outlier]["ids"].append(e["id"])
all_results.append({"id": e["id"], "name": e["name"], "lang": e["lang"], "mistakes": mistakes, "correct": correct, "total": total, "skipped": skipped})
return all_results, file_results
def load_exercise_data(datadir):
ex_sets = glob.iglob(datadir + "/*/??-*")
exercises = {}
for ex in ex_sets:
exercises[ex] = load_exercise_file(ex)
return exercises
def load_exercise_file(filename):
ex_file = open(filename).readlines()
ex_data = [l.strip() for l in ex_file if not l.startswith("#")]
if len(ex_data) != 17 or ex_data[8] != "":
raise Exception("Invalid exercise format for file %s" % ex_filename)
return {"words": ex_data[:8], "outliers": ex_data[9:]}
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Evalutes outlier detection annotations.')
parser.add_argument('dbpath', metavar='SQLITE.db', help='path to the SQLITE database with annotations')
parser.add_argument('datadir', metavar='DATADIR', help='path to the data directory with gold dataset')
parser.add_argument('lang', metavar='LANG', help='ALL|CS,EN,IT...')
parser.add_argument('-a', '--all', help='include also incomplete exercises', action='store_false', dest='skip_incomplete')
parser.add_argument('-f', '--from', help='limit evaluation to actions from this datetime in ISO format (e.g. 1970-01-01 23:02:13)', dest='date_from', type=datetime.fromisoformat)
parser.add_argument('-t', '--to', help='limit evaluation to actions to this datetime in ISO format', dest='date_to', type=datetime.fromisoformat)
group = parser.add_mutually_exclusive_group()
group.add_argument('-v', '--verbose', action='count', default=0, help='verbosity, up to -vvv supported')
group.add_argument('-e', '--ex-id', help='limit evaluation to particular exercise ID')
# if len(sys.argv) < 4 or len(sys.argv) > 5:
# print("Usage: %s [-a] SQLITE.db DATADIR LANG=ALL|CS,EN,.. [ VERBOSITY_LEVEL={default=0,1,2,3} | EXERCISE_ID ]", file=sys.stderr)
# print("-a includes also incomplete exercises")
# sys.exit(1)
args = parser.parse_args()
conn = sqlite3.connect(args.dbpath)
conn.row_factory = sqlite3.Row
c = conn.cursor()
langs = args.lang.split(",")
if args.ex_id:
r = eval_exercise (c, load_exercise_data(args.datadir), ["ALL"], args.skip_incomplete, args.date_from, args.date_to, args.ex_id)[0][0]
print("exercise id: %s, name: %s, lang: %s" % (r["id"], r["name"], r["lang"]))
if r["total"] > 0:
print("correct = %d, total = %d, precision = %.2f, skipped = %d" % (r["correct"], r["total"], r["correct"]/r["total"], r["skipped"]))
else:
print("total = 0")
print("mistakes:")
for m in r["mistakes"]:
exset, mistaken_outlier, correct_outlier = m
print("%s instead of %s for %s" % (exset, mistaken_outlier, correct_outlier))
else:
def vp(msg, level):
if level <= args.verbose:
print(msg)
rr, fr = eval_exercise (c, load_exercise_data(args.datadir), langs, args.skip_incomplete, args.date_from, args.date_to)
per_lang = {}
for r in rr:
vp("exercise id: %s, name: %s, lang: %s" % (r["id"], r["name"], r["lang"]), 2)
if not r["total"]:
vp("skipping (total=0)", 2)
continue
vp("correct = %d, total = %d, precision = %.2f, skipped = %d" % (r["correct"], r["total"], r["correct"]/r["total"], r["skipped"]), 2)
if r["lang"] not in per_lang:
per_lang[r["lang"]] = {"correct": 0, "total": 0, "skipped": 0}
per_lang[r["lang"]]["correct"] += r["correct"]
per_lang[r["lang"]]["total"] += r["total"]
per_lang[r["lang"]]["skipped"] += r["skipped"]
vp("PER LANGUAGE:", 0)
sum_total = 0
sum_correct = 0
sum_skipped = 0
for l, v in per_lang.items():
vp("Language %s: correct = %d, total = %d, precision = %.2f, skipped = %d" % (l, v["correct"], v["total"], v["correct"]/v["total"], v["skipped"]), 0)
sum_total += v["total"]
sum_correct += v["correct"]
sum_skipped += v["skipped"]
vp("TOTAL: correct = %d, total = %d, precision = %.2f, skipped = %d" % (sum_correct, sum_total, sum_correct/sum_total, sum_skipped), 0)
vp("\nPER FILE:", 1)
for f, k in fr.items():
vp("\nFilename %s: correct = %d, total = %d, precision = %.2f, skipped = %d" % (f, k["correct"], k["total"], k["correct"]/k["total"], k["skipped"]), 1)
for w, kk in k["words"].items():
vp("Outlier %s: correct = %d, total = %d, precision = %.2f, skipped = %d, mistakes = %s, ids = %s" % (w, kk["correct"], kk["total"], kk["correct"]/kk["total"], kk["skipped"], dict(kk["mistakes"]), args.verbose > 2 and kk["ids"] or "[VERBOSITY3]"), 1)