summaryrefslogtreecommitdiff
path: root/monkey/test.py
blob: 4b0d192e2fe2a7169a468a897ee6d2742de799b8 (plain)
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
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
#!/usr/bin/python3

import os
import pickle
import sys

import django
from termcolor import colored

from .action import parse
from .edits import classify_edits, trace_graph, get_edits_from_traces
from .graph import graphviz
from .monkey import fix
from prolog.engine import test
from prolog.util import Token, compose, decompose, stringify
from .util import indent

# Load django models.
os.environ['DJANGO_SETTINGS_MODULE'] = 'webmonkey.settings'
django.setup()
from django.apps import apps as tutor_apps
from django.contrib.auth.models import User
from tutor.models import Attempt, Problem, get_aux_code

# Get problem id from commandline.
if len(sys.argv) < 2:
    print('usage: ' + sys.argv[0] + ' <pid>')
    sys.exit(1)
pid = int(sys.argv[1])

problem = Problem.objects.get(pk=pid)
aux_code = get_aux_code(user=User.objects.get(pk=1), problem=problem)

attempts = Attempt.objects.filter(problem=problem)

# Load hint database stored in edits.pickle.
edits = tutor_apps.get_app_config('tutor').edits[problem.pk]
lines = tutor_apps.get_app_config('tutor').lines[problem.pk]
submissions = tutor_apps.get_app_config('tutor').submissions[problem.pk]
queries = tutor_apps.get_app_config('tutor').queries[problem.pk]

# Find incorrect submissions.
incorrect_all = []
for submission, count in sorted(submissions.items()):
    if not test(problem.name, submission + '\n' + aux_code):
        # This incorrect submission appeared in [count] attempts.
        incorrect_all += [submission]*count
incorrect = set(incorrect_all)

# Load current status (programs for which a hint was found).
try:
    done = pickle.load(open('status-'+str(problem.pk)+'.pickle', 'rb'))
except:
    done = []

def print_hint(solution, steps, fix_time, n_tested):
    if solution:
        print(colored('Hint found! Tested {} programs in {:.1f} s.'.format(n_tested, fix_time), 'green'))
        print(colored(' Edits', 'blue'))
        for step_type, line, (before, after) in steps:
            print('  {}: {} {} → {}'.format(line, step_type, stringify(before), stringify(after)))
        print(colored(' Final version', 'blue'))
        print(indent(compose(*decompose(solution)), 2))
    else:
        print(colored('Hint not found! Tested {} programs in {:.1f} s.'.format(n_tested, fix_time), 'red'))

# Test fix() on incorrect student submissions.
if len(sys.argv) >= 3 and sys.argv[2] == 'test':
    timeout = int(sys.argv[3]) if len(sys.argv) >= 4 else 10

    print('Fixing {}/{} programs (timeout={})…'.format(
        len([p for p in incorrect if p not in done]), len(incorrect), timeout))

    for i, program in enumerate(sorted(incorrect)):
        if program in done:
            continue
        print(colored('Analyzing program {0}/{1}…'.format(i+1, len(incorrect)), 'yellow'))
        print(indent(compose(*decompose(program)), 2))

        solution, steps, fix_time, n_tested = fix(problem.name, program, edits, lines, aux_code=aux_code, timeout=timeout)
        if solution:
            done.append(program)
        print_hint(solution, steps, fix_time, n_tested)
        print()

        pickle.dump(done, open('status-'+str(problem.pk)+'.pickle', 'wb'))

    print('Found hints for ' + str(len(done)) + ' of ' + str(len(incorrect)) + ' incorrect programs')

# Print info for this problem.
elif len(sys.argv) >= 3 and sys.argv[2] == 'info':
    # With no additional arguments, print some stats.
    if len(sys.argv) == 3:
        print('Problem {} ({}): {} edits in {} traces, fixed {}/{} ({}/{} unique)'.format(
            problem.pk, colored(problem.name, 'yellow'),
            colored(str(len(edits)), 'yellow'), colored(str(len([a.trace for a in attempts])), 'yellow'),
            colored(str(len([p for p in incorrect_all if p in done])), 'yellow'),
            colored(str(len(incorrect_all)), 'yellow'),
            colored(str(len(set(done))), 'yellow'),
            colored(str(len(set(incorrect))), 'yellow')))
    else:
        if sys.argv[3] == 'users':
            print(' '.join([str(a.user.pk) for a in attempts]))
        # Print all observed edits and their costs.
        elif sys.argv[3] == 'edits':
            inserts, removes, changes = classify_edits(edits)
            print('Inserts')
            for after, cost in sorted(inserts.items(), key=lambda x: x[1]):
                print(' {:.4f}\t{}'.format(cost, stringify(after)))
            print('Removes')
            for before, cost in sorted(removes.items(), key=lambda x: x[1]):
                print(' {:.4f}\t{}'.format(cost, stringify(before)))
            print('Changes')
            for (before, after), cost in sorted(changes.items(), key=lambda x: x[1]):
                print(' {:.4f}\t{} → {}'.format(cost,
                                               stringify(before if before else [Token('INVALID', 'ε')]),
                                               stringify(after if after else [Token('INVALID', 'ε')])))
        # Print all student submissions not (yet) corrected.
        elif sys.argv[3] == 'unsolved':
            for p in sorted(incorrect):
                if p in done:
                    continue
                print(indent(compose(*decompose(p)), 2))
                print()
        # Print all student queries and their counts.
        elif sys.argv[3] == 'queries':
            for query, count in queries.most_common():
                print('  ' + str(count) + '\t' + query)

# Print the edit graph in graphviz dot syntax.
elif len(sys.argv) == 4 and sys.argv[2] == 'graph':
    uid = int(sys.argv[3])
    user = User.objects.get(pk=uid)
    attempt = Attempt.objects.get(problem=problem, user=user)

    nodes, submissions, queries = trace_graph(attempt.trace)

    def position(node):
        return (node.data[1]*150, node.data[0]*-60)

    def label(node):
        return stringify(node.data[2])

    def node_attr(node):
        if node.ein and node.data[2] == node.ein[0].data[2]:
            return 'color="gray", shape="point"'
        return ''

    def edge_attr(a, b):
        if a.data[2] == b.data[2]:
            return 'arrowhead="none"'
        return ''

    graphviz_str = graphviz(nodes, pos=position, label=label,
                            node_attr=node_attr, edge_attr=edge_attr)
    print(graphviz_str)

# Run interactive loop.
else:
    while True:
        # Read the program from stdin.
        print('Enter program, end with empty line:')
        code = ''
        try:
            while True:
                line = input()
                if not line:
                    break
                code += line + '\n'
        except EOFError:
            break

        # Try finding a fix.
        print(colored('Analyzing program…', 'yellow'))
        solution, steps, fix_time, n_tested = fix(problem.name, code, edits, aux_code=aux_code, debug=True)
        print_hint(solution, steps, fix_time, n_tested)