Move testing code from monkey.py to test.py

Drop monkey/db.py and use Django models instead.

3 files changed, 171 insertions, 213 deletions

diff --git a/monkey/db.py b/monkey/db.py
deleted file mode 100644
index 0634098..0000000
--- a/monkey/db.py
+++ /dev/null
@@ -1,44 +0,0 @@
-#!/usr/bin/python3
-
-import sqlite3
-
-db = sqlite3.connect('misc/solutions.db')
-db.row_factory = sqlite3.Row
-db.text_factory = bytes
-cursor = db.cursor()
-
-def b_to_utf8(bytestring):
-    return str(bytestring, encoding='utf-8')
-
-def get_problem_ids():
-    cursor.execute('SELECT id FROM problems ORDER BY id ASC')
-    return [row['id'] for row in cursor]
-
-def get_problem(pid):
-    cursor.execute('SELECT name, solution, library FROM problems WHERE id=?', (pid,))
-    row = cursor.fetchone()
-    name = b_to_utf8(row['name'])
-    solution = b_to_utf8(row['solution']).replace('\r', '')
-    lib_id = row['library'] if row['library'] else None
-    return name, solution, lib_id
-
-def get_depends(pid):
-    cursor.execute('SELECT dependency FROM depends WHERE problem=?', (pid,))
-    return [r['dependency'] for r in cursor.fetchall()]
-
-def get_library(lid):
-    cursor.execute('SELECT facts FROM libraries WHERE id=?', (lid,))
-    row = cursor.fetchone()
-    return b_to_utf8(row['facts']).replace('\r', '') if row else None
-
-def get_tests(pid):
-    cursor.execute('SELECT query FROM tests WHERE problem=?', (pid,))
-    return [b_to_utf8(row['query']) for row in cursor]
-
-def get_traces(pid):
-    cursor.execute('SELECT * FROM attempts WHERE problem=? AND done=1 ORDER BY id ASC', (pid,))
-    return {(pid, attempt['user']): attempt['log'] for attempt in cursor}
-
-def get_solved(uid):
-    cursor.execute('SELECT problem FROM attempts WHERE user=? AND done=1 ORDER BY problem ASC', (uid,))
-    return [row['problem'] for row in cursor.fetchall()]
diff --git a/monkey/monkey.py b/monkey/monkey.py
index 8e805f5..693aa67 100755
--- a/monkey/monkey.py
+++ b/monkey/monkey.py
@@ -8,11 +8,10 @@ import time
 
 from termcolor import colored
 
-from . import db
 from .action import parse
 from .edits import classify_edits, clean_graph, edit_graph, get_edits_from_traces
 from .graph import Node, graphviz
-from .prolog.engine import PrologEngine
+from .prolog.engine import test
 from .prolog.util import compose, decompose, map_vars, rename_vars, stringify
 from .util import PQueue, Token, indent
 
@@ -144,170 +143,3 @@ def fix(name, code, edits, timeout=30, debug=False):
                 add_task(new_lines, new_rules, prev=task, step=new_step, cost=cost)
 
     return '', [], total_time, n_tested
-
-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 line, (before, after) in steps:
-            print('  {}:\t{} → {}'.format(line, 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'))
-
-# Find official solutions to all problems.
-def init_problems():
-    names = {}
-    codes = {}
-    libraries = {}
-
-    pids = db.get_problem_ids()
-    for pid in pids:
-        names[pid], codes[pid], libraries[pid] = db.get_problem(pid)
-
-    return names, codes, libraries
-
-# Submit code to Prolog server for testing.
-def test(name, code):
-    # TODO also load fact library and solved predicates
-    engine = PrologEngine(code=code)
-    result = engine.ask("run_tests({}, '{}')".format(name, engine.id))
-    engine.destroy()
-    return result['event'] == 'success'
-
-if __name__ == '__main__':
-    # Get problem id from commandline.
-    if len(sys.argv) < 2:
-        print('usage: ' + sys.argv[0] + ' <pid>')
-        sys.exit(1)
-    pid = int(sys.argv[1])
-
-    names, codes, libraries = init_problems()
-
-    # Analyze traces for this problem to get edits, submissions and queries.
-    traces = db.get_traces(pid)
-    edits, lines, submissions, queries = get_edits_from_traces(traces.values())
-
-    # Find incorrect submissions.
-    incorrect = []
-    for submission, count in sorted(submissions.items()):
-        if not test(names[pid], submission):
-            # This incorrect submission appeared in [count] attempts.
-            incorrect += [submission]*count
-
-    # XXX only for testing
-    try:
-        done = pickle.load(open('status-'+str(pid)+'.pickle', 'rb'))
-    except:
-        done = []
-
-    # 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(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(names[pid], program, edits, timeout=timeout)
-            if solution:
-                done.append(program)
-            print_hint(solution, steps, fix_time, n_tested)
-            print()
-
-            pickle.dump(done, open('status-'+str(pid)+'.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(
-                pid, colored(names[pid], 'yellow'),
-                colored(str(len(edits)), 'yellow'), colored(str(len(traces)), 'yellow'),
-                colored(str(len([p for p in incorrect if p in done])), 'yellow'),
-                colored(str(len(incorrect)), 'yellow'),
-                colored(str(len(set(done))), 'yellow'),
-                colored(str(len(set(incorrect))), 'yellow')))
-        else:
-            if sys.argv[3] == 'users':
-                print(' '.join([str(uid) for (pid, uid) in sorted(traces.keys())]))
-            # 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(' {:.2f}\t{}'.format(cost, stringify(after)))
-                print('Removes')
-                for before, cost in sorted(removes.items(), key=lambda x: x[1]):
-                    print(' {:.2f}\t{}'.format(cost, stringify(before)))
-                print('Changes')
-                for (before, after), cost in sorted(changes.items(), key=lambda x: x[1]):
-                    print(' {:.2f}\t{} → {}'.format(cost,
-                                                   stringify(before if before else [('INVALID', 'ε')]),
-                                                   stringify(after if after else [('INVALID', 'ε')])))
-            # print all student submissions not (yet) corrected
-            elif sys.argv[3] == 'unsolved':
-                for p in sorted(set(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])
-        actions = parse(traces[(pid, uid)])
-
-        nodes, submissions, queries = edit_graph(actions)
-
-        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(names[pid], code, edits, debug=True)
-            print_hint(solution, steps, fix_time, n_tested)
diff --git a/monkey/test.py b/monkey/test.py
new file mode 100755
index 0000000..3143274
--- /dev/null
+++ b/monkey/test.py
@@ -0,0 +1,170 @@
+#!/usr/bin/python3
+
+import os
+import pickle
+import sys
+
+import django
+from termcolor import colored
+
+from .action import parse
+from .edits import classify_edits, edit_graph, get_edits_from_traces
+from .graph import graphviz
+from .monkey import fix
+from .prolog.engine import test
+from .prolog.util import compose, decompose, stringify
+from .util import indent
+
+# Load django models.
+os.environ['DJANGO_SETTINGS_MODULE'] = 'webmonkey.settings'
+django.setup()
+from django.contrib.auth.models import User
+from tutor.models import Attempt, Problem
+
+# Get problem id from commandline.
+if len(sys.argv) < 2:
+    print('usage: ' + sys.argv[0] + ' <pid>')
+    sys.exit(1)
+pid = int(sys.argv[1])
+
+# Analyze traces for this problem to get edits, submissions and queries.
+problem = Problem.objects.get(pk=pid)
+attempts = Attempt.objects.filter(problem=problem)
+
+traces = [a.trace for a in attempts]
+edits, lines, submissions, queries = get_edits_from_traces(traces)
+
+# Find incorrect submissions.
+incorrect = []
+for submission, count in sorted(submissions.items()):
+    if not test(problem.name, submission):
+        # This incorrect submission appeared in [count] attempts.
+        incorrect += [submission]*count
+
+# 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 line, (before, after) in steps:
+            print('  {}:\t{} → {}'.format(line, 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(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, 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(traces)), 'yellow'),
+            colored(str(len([p for p in incorrect if p in done])), 'yellow'),
+            colored(str(len(incorrect)), '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(' {:.2f}\t{}'.format(cost, stringify(after)))
+            print('Removes')
+            for before, cost in sorted(removes.items(), key=lambda x: x[1]):
+                print(' {:.2f}\t{}'.format(cost, stringify(before)))
+            print('Changes')
+            for (before, after), cost in sorted(changes.items(), key=lambda x: x[1]):
+                print(' {:.2f}\t{} → {}'.format(cost,
+                                               stringify(before if before else [('INVALID', 'ε')]),
+                                               stringify(after if after else [('INVALID', 'ε')])))
+        # Print all student submissions not (yet) corrected.
+        elif sys.argv[3] == 'unsolved':
+            for p in sorted(set(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)
+    actions = parse(Attempt.objects.get(problem=problem, user=user).trace)
+
+    nodes, submissions, queries = edit_graph(actions)
+
+    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, debug=True)
+        print_hint(solution, steps, fix_time, n_tested)