feat: add extra chips in chips challenges

This adds extra chips in chips challenges that will not be used in
the correct solution.

apps/librelingo_json_export/librelingo_json_export/__init__.py

@@ -4,4 +4,4 @@
 
 import librelingo_json_export.export
 
-__version__ = '0.6.3'
+__version__ = '0.7.0'

apps/librelingo_json_export/librelingo_json_export/challenge_types.py

@@ -1,4 +1,6 @@
-from librelingo_utils import get_dumb_opaque_id, audio_id, clean_word
+import editdistance
+
+from librelingo_utils import get_dumb_opaque_id, audio_id, clean_word, iterate_phrases
 from .dictionary import _define_words_in_sentence
 
 
@@ -56,10 +58,42 @@ def get_options_challenge(phrase, _):
     }]
 
 
-def get_chips(phrase, course):
+def get_chips_from_string(phrase):
     return list(map(clean_word, phrase.split()))
 
 
+def get_chips_from_phrase(get_input_texts, phrase, course):
+    extra_chips = []
+    solution_chips = get_chips_from_string(get_input_texts(phrase)[0])
+
+    for phrase in iterate_phrases(course):
+        for variant in get_input_texts(phrase):
+            for chip in get_chips_from_string(variant):
+                if chip not in solution_chips:
+                    extra_chips.append(chip)
+
+    chips_already_added = set()
+    deduplicated_chips = []
+
+    for chip in extra_chips:
+        if chip.lower() not in chips_already_added:
+            deduplicated_chips.append(chip)
+            chips_already_added.add(chip.lower())
+
+    extra_chips = sorted(
+        deduplicated_chips,
+        key=lambda chip: sum(
+            editdistance.eval(other_chip, chip) for other_chip in solution_chips
+        )
+    )
+
+    return solution_chips + extra_chips[0:max(len(solution_chips) - 1, 2)]
+
+
+def get_solutions_from_phrase(get_input_texts, phrase):
+    return [get_chips_from_string(x) for x in get_input_texts(phrase)]
+
+
 def create_chips_challenge_generator(reverse):
     def get_input_texts(phrase):
         return phrase.in_source_language if reverse else phrase.in_target_language
@@ -88,8 +122,8 @@ def get_chips_challenge(phrase, course):
             "type": "chips",
             "translatesToSourceLanguage": reverse,
             "phrase": _define_words_in_sentence(course, get_phrase_text(phrase), reverse),
-            "chips": get_chips(get_input_text(phrase), course),
-            "solutions": [get_chips(x, course) for x in get_input_texts(phrase)],
+            "chips": get_chips_from_phrase(get_input_texts, phrase, course),
+            "solutions": get_solutions_from_phrase(get_input_texts, phrase),
             "formattedSolution": get_input_text(phrase),
             "id": get_dumb_opaque_id("Chips", phrase, "reverse chips" if reverse else "chips"),
             "priority": 2,

apps/librelingo_json_export/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "librelingo_json_export"
-version = "0.6.3"
+version = "0.7.0"
 description = "Export LibreLingo courses in the JSON format used by the web app"
 authors = ["Dániel Kántor <[email protected]>"]
 license = "AGPLv3"
@@ -12,6 +12,7 @@ librelingo-types = "^3.0.0"
 librelingo-utils = "^2.3.0"
 python-slugify = "^4.0.1"
 librelingo-yaml-loader = "^1.0.0"
+editdistance = "^0.5.3"
 
 [tool.poetry.dev-dependencies]
 pytest = "^6.2.2"

apps/librelingo_json_export/tests/snapshots/snap_test_integration.py

@@ -32,7 +32,10 @@
                     'La',
                     'femme',
                     'dit',
-                    'bonjour'
+                    'bonjour',
+                    'la',
+                    'salut',
+                    "L'homme"
                 ],
                 'formattedSolution': 'La femme dit bonjour',
                 'group': '921dd5d24167',
@@ -80,7 +83,9 @@
                     'The',
                     'woman',
                     'says',
-                    'hello'
+                    'hello',
+                    'man',
+                    'hi'
                 ],
                 'formattedSolution': 'The woman says hello',
                 'group': '921dd5d24167',
@@ -143,7 +148,9 @@
                 'chips': [
                     "L'homme",
                     'dit',
-                    'bonjour'
+                    'bonjour',
+                    'La',
+                    'femme'
                 ],
                 'formattedSolution': "L'homme dit bonjour",
                 'group': '0f389ca8a0c4',
@@ -189,7 +196,9 @@
                     'The',
                     'man',
                     'says',
-                    'hello'
+                    'hello',
+                    'hi',
+                    'woman'
                 ],
                 'formattedSolution': 'The man says hello',
                 'group': '0f389ca8a0c4',

apps/librelingo_json_export/tests/test_challenges.py

@@ -1,3 +1,5 @@
+import collections
+
 from unittest.mock import patch
 from unittest import TestCase
 from librelingo_json_export.challenges import _get_challenges_data
@@ -8,7 +10,7 @@
 from librelingo_json_export.challenge_types import get_listening_challenge
 from librelingo_json_export.challenge_types import get_chips_challenge
 from librelingo_json_export.challenge_types import get_options_challenge
-from librelingo_json_export.challenge_types import get_chips
+from librelingo_json_export.challenge_types import get_chips_from_phrase
 from librelingo_types import Phrase
 from librelingo_types import Settings
 from librelingo_fakes import fakes
@@ -234,14 +236,15 @@ def test_returns_correct_value1(self):
             'id': '9f9b09771a07',
             'group': '930c4c4e7552',
             "priority": 2,
-            "chips": ["foous", "barus"],
+            "chips": ["foous", "barus", "lorem", "ipsum"],
             "solutions": [["foous", "barus"]],
             "formattedSolution": "foous barus",
         }
 
-    @patch('librelingo_json_export.challenge_types.get_chips')
-    def test_returns_correct_value2(self, get_chips):
-        get_chips.return_value = fakes.fake_value()
+    @patch('librelingo_json_export.challenge_types.get_solutions_from_phrase')
+    @patch('librelingo_json_export.challenge_types.get_chips_from_phrase')
+    def test_returns_correct_value2(self, get_chips_from_phrase, get_solutions_from_phrase):
+        get_chips_from_phrase.return_value = fakes.fake_value()
         challenge = get_chips_challenge(
             fakes.phrase_with_alternatives, fakes.course1)[0]
         assert challenge == {
@@ -256,36 +259,73 @@ def test_returns_correct_value2(self, get_chips):
             'id': '4b0e9208ce1b',
             'group': '66a39e74a2c8',
             "priority": 2,
-            "solutions": [get_chips.return_value, get_chips.return_value],
-            "chips": get_chips.return_value,
+            "solutions": get_solutions_from_phrase.return_value,
+            "chips": get_chips_from_phrase.return_value,
             "formattedSolution": "foous barus foous barus ",
         }
 
-    @patch('librelingo_json_export.challenge_types.get_chips')
-    def test_calls_get_chips_with_correct_value(self, get_chips):
-        get_chips_challenge(fakes.phrase1, fakes.course1)
-        get_chips.assert_called_with(fakes.phrase1.in_target_language[0], fakes.course1)
-
 
 class GetChipsTest(TestCase):
     def test_empty_string(self):
-        assert get_chips('', fakes.course1) == []
+        assert get_chips_from_phrase(lambda _: [''], None, fakes.course1) == []
 
     @patch('librelingo_json_export.challenge_types.clean_word')
     def test_empty_string_doesnt_call_clean_word(self, clean_word):
-        get_chips('', fakes.course1)
+        get_chips_from_phrase(lambda _: [''], None, fakes.course1)
         assert not clean_word.called
 
     @patch('librelingo_json_export.challenge_types.clean_word')
     def test_calls_clean_word_with_correct_argument(self, clean_word):
-        get_chips('foo', fakes.course2)
+        get_chips_from_phrase(lambda _: ['foo'], None, fakes.course2)
         clean_word.assert_called_with('foo')
 
     @patch('librelingo_json_export.challenge_types.clean_word')
     def test_returns_correct_value(self, clean_word):
         clean_word.return_value = fakes.fake_value()
-        assert get_chips('foo', fakes.course1) == [clean_word.return_value]
+        assert get_chips_from_phrase(lambda _: 'foo', None, fakes.course1) == [clean_word.return_value]
 
     @patch('librelingo_json_export.challenge_types.clean_word')
     def test_returns_correct_number_of_words(self, clean_word):
-        assert len(get_chips('foo bar', fakes.course1)) == 2
+        # Includes extra words
+        clean_word.side_effect = lambda x: x
+        assert len(get_chips_from_phrase(lambda x: ['Sleep well!'] if x is None else x[0], None, fakes.course1)) == 4
+
+    def test_includes_words_from_other_phrases_sorted_by_edit_distance(self):
+        extra_chips = list(get_chips_from_phrase(lambda x: ['something lipsum'] if x is None else x[0], None, fakes.course1))
+        while "something" in extra_chips:
+            extra_chips.remove("something")
+        while "lipsum" in extra_chips:
+            extra_chips.remove("lipsum")
+        assert extra_chips == ['ipsum', 'lorem']
+
+    def test_doesnt_include_duplicates_from_another_phrase(self):
+        my_fake_module = fakes.customize(
+            fakes.course1.modules[0],
+            skills=[
+                fakes.customize(
+                    fakes.course1.modules[0].skills[0],
+                    phrases=[
+                        fakes.customize(
+                            fakes.course1.modules[0].skills[0].phrases[0],
+                            in_target_language=["Ipsum"]
+                        )
+                    ]
+                )
+            ]
+        )
+        fake_course_with_duplicated_chip = fakes.customize(
+            fakes.course1,
+            modules=[fakes.course1.modules[0], fakes.course1.modules[0], my_fake_module]
+        )
+        chips_count = collections.Counter(get_chips_from_phrase(lambda x: ['something lipsum'] if x is None else x[0], None, fake_course_with_duplicated_chip))
+
+        assert chips_count["ipsum"] + chips_count["Ipsum"] == 1
+
+    def test_returns_correct_number_of_words_2(self):
+        fake_course_with_duplicated_chip = fakes.customize(
+            fakes.course1,
+            modules=[fakes.course1.modules[0], fakes.course1.modules[0]]
+        )
+        chips = list(get_chips_from_phrase(lambda x: ['something lipsum dolor'] if x is None else x[0], None, fake_course_with_duplicated_chip))
+
+        assert len(chips) == 5

apps/web/cypress/integration/chips.feature

@@ -36,6 +36,7 @@ Feature: Chips challenge
     And I click "Como"
     And I click "estás"
     And I click "hoy"
+    Then I have unused chips
     Given I click "Submit"
     Then I see the challenge panel with no skip button
     And I read "Correct solution"

apps/web/cypress/integration/chips/index.js

@@ -13,3 +13,7 @@ Then("I see the correct chips", () => {
     cy.get(".chip").contains("estás").should("be.visible")
     cy.get(".chip").contains("hoy").should("be.visible")
 })
+
+Then("I have unused chips", () => {
+    cy.get("#chips .chip").should("be.visible")
+})