1 files changed, 86 insertions, 0 deletions
diff --git a/python/problems/dictionaries/text/common.py b/python/problems/dictionaries/text/common.py
new file mode 100644
index 0000000..134f69b
--- /dev/null
+++ b/python/problems/dictionaries/text/common.py
@@ -0,0 +1,86 @@
+import re
+from python.util import has_token_sequence, string_almost_equal, \
+    string_contains_number, get_tokens, get_numbers, get_exception_desc
+from server.hints import Hint
+
+id = 20612
+number = 13
+visible = True
+
+solution = '''\
+import collections
+
+def following_words(txt):
+    words = txt.split()
+    freq = collections.defaultdict(list)
+    for word, next_word in zip(words, words[1:]):
+        freq[word].append(next_word)
+    return freq
+
+def freq_following_word(txt):
+    following = following_words(txt)
+    for f in following:
+        vals = collections.Counter(following[f])
+        s = sorted(vals.most_common(), key = lambda x: (-x[1], x[0]))
+        following[f] = s[0][0]
+    return following
+
+def text(word, full_text, num):
+    freq = freq_following_word(full_text)
+    words = []
+    for i in range(num):
+        words.append(word)
+        word = freq[word]
+    return ' '.join(words)
+'''
+
+hint_type = {
+    'final_hint': Hint('final_hint')
+}
+
+def test(python, code):
+    func_name = 'text'
+    tokens = get_tokens(code)
+    if not has_token_sequence(tokens, ['def', func_name]):
+        return False, [{'id' : 'no_func_name', 'args' : {'func_name' : func_name}}]
+
+    in_out = [
+        (('in', 'in in in in', 5), 'in in in in in'),
+        (('in', 'in to in ono in to smo mi', 5), 'in to in to in'),
+        (('danes', 'danes je lep dan danes sije sonce', 5),
+         'danes je lep dan danes'),
+        (('danes', 'danes je lep dan danes sije sonce danes sije dan ki je sonce', 5),
+         'danes sije dan danes sije'),
+    ]
+
+    test_in = [(func_name+'%s'%str(l[0]), None) for l in in_out]
+    test_out = [l[1] for l in in_out]
+
+    answers = python(code=code, inputs=test_in, timeout=1.0)
+    n_correct = 0
+    tin, tout = None, None
+    for i, (ans, to) in enumerate(zip(answers, test_out)):
+        corr = ans[0] == to
+        n_correct += corr
+        if not corr:
+            tin = test_in[i][0]
+            tout = to
+
+    passed = n_correct == len(test_in)
+    hints = [{'id': 'test_results', 'args': {'passed': n_correct, 'total': len(test_in)}}]
+    if tin:
+        hints.append({'id': 'problematic_test_case', 'args': {'testin': str(tin), 'testout': str(tout)}})
+    if passed:
+        hints.append({'id': 'final_hint'})
+    return passed, hints
+
+
+def hint(python, code):
+    tokens = get_tokens(code)
+
+    # run one test first to see if there are any exceptions
+    answer = python(code=code, inputs=[(None, None)], timeout=1.0)
+    exc = get_exception_desc(answer[0][3])
+    if exc: return exc
+
+    return None