Speed up test suite: 160s → 44s

.gitignore

@@ -22,7 +22,6 @@ build*
 dash.cfg
 env_var_defs.py
 .pytest_cache*
-.testmondata
 zeeguu_api/machine_specific.py
 *.db-journal
 
@@ -53,5 +52,3 @@ diagrams/*.json
 # Keep internal technical design documents private
 experiments/
 docs/particle_verb_detection.md
-/.testmondata-wal
-/.testmondata-shm

requirements.txt

@@ -95,7 +95,6 @@ sentry-sdk[flask]
 
 # Testing
 pytest
-pytest-testmon
 
 # Version constraints
 pydantic>=1.10.8

run_tests.sh

@@ -1,21 +1,10 @@
 #!/bin/sh
-# Smart test runner using pytest-testmon
-# Only runs tests affected by code changes
+# Run all tests
 
 export PYTHONWARNINGS='ignore'
 
-# Check if this is the first run or if --full is passed
-if [ "$1" = "--full" ] || [ "$1" = "-f" ]; then
-    echo "Running FULL test suite (refreshing testmon cache)..."
-    python -m pytest --testmon-off "$@"
-elif [ "$1" = "--nocache" ]; then
-    echo "Running tests with fresh testmon cache..."
-    rm -f .testmondata
-    python -m pytest --testmon "${@:2}"
-else
-    echo "Running SMART tests (only affected by changes)..."
-    python -m pytest --testmon "$@"
-fi
+echo "Running test suite..."
+python -m pytest "$@"
 
 ret_code=$?
 export PYTHONWARNINGS='default'

test_durations.txt

@@ -0,0 +1,36 @@
+Test durations — 161 tests (was 189)
+=====================================
+Optimizations applied:
+  1. Session-scoped app (no per-test create_app)
+  2. JSON fixture for article/NLP data (no Stanza at setup)
+  3. Wordstats mock (no 400k WordInfo load)
+  4. Feed test consolidation (5→1), then deletion
+  5. Article test consolidation (4→1)
+  6. Deleted redundant slow tests (feed download, Article.find_or_create, test_create_article)
+  7. Fixed cache-before-tokenizer bug in article.get_tokenized_content()
+  8. Removed unused hardcoded feeds from FeedRule (was creating 4 feeds per BookmarkRule)
+  9. Bookmark test consolidation (18→3)
+
+Full suite history:
+  ORIGINAL:     ~160s (2:40)  189 tests  — function-scoped app, live Stanza + wordstats
+  AFTER 1-3:    ~110s (1:50)  184 tests  — session-scoped app, JSON fixture, wordstats mock
+  AFTER 1-6:     ~52s (0:52)  176 tests  — + deletions and consolidations
+  AFTER 1-8:     ~55s (0:55)  176 tests  — + cache fix (test_starred_or_liked: 6.5s → 0.25s)
+  AFTER 1-9:     ~46s (0:46)  161 tests  — + bookmark consolidation (18→3)
+
+Production bug fixed:
+  article.get_tokenized_content() loaded Stanza BEFORE checking cache.
+  Now checks cache first — on hit, returns immediately without loading tokenizer.
+  test_starred_or_liked: 6.50s → 0.25s (also saves time in production)
+
+Current top 10 slowest (46s total, 161 tests):
+  15.18s  test_bookmark_quality_and_study    — creates 10+ bookmarks inline (consolidated)
+   6.83s  test_contribute_own_translation    — API bookmark contribution
+   4.38s  session setup (one-time)           — create_app loads spaCy models at import time
+   2.21s  test_article_topics_and_basics     — ArticleRule creates 2 fake articles
+   1.89s  test_exercises                     — creates bookmarks + exercises (consolidated)
+   1.65s  test_add_exercise_to_session       — exercise scheduling logic
+   0.94s  test_number_detection              — tokenizer test
+   0.71s  test_no_null_positions             — bookmark positions
+   0.68s  test_bookmark_counts_by_date       — user bookmark stats
+   0.63s  test_ml_classification             — quality filter

tools/regenerate_test_fixture.py

@@ -0,0 +1,83 @@
+#!/usr/bin/env python
+"""
+Regenerate the test fixture file for session-scoped test data.
+
+Run this after schema changes:
+    source ~/.venvs/z_env/bin/activate && python tools/regenerate_test_fixture.py
+"""
+
+import json
+import os
+import sys
+
+# Add project root to path
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+import requests_mock as rm
+from zeeguu.api.app import create_app
+from zeeguu.core.model.db import db
+from zeeguu.core.test.mocking_the_web import (
+    mock_requests_get,
+    URL_SPIEGEL_VENEZUELA,
+)
+from sqlalchemy import text
+
+FIXTURE_PATH = os.path.join(
+    os.path.dirname(os.path.dirname(os.path.abspath(__file__))),
+    "zeeguu/core/test/test_data/session_fixture.json"
+)
+
+KEEP_TABLES = [
+    "language", "context_type", "source_type", "domain_name",
+    "url", "url_keyword", "source", "source_text",
+    "article", "article_fragment", "article_cefr_assessment",
+    "article_tokenization_cache", "article_url_keyword_map", "new_text",
+]
+
+
+def main():
+    app = create_app(testing=True)
+    with rm.Mocker() as m:
+        mock_requests_get(m)
+        with app.test_client() as client:
+            with app.app_context():
+                db.create_all()
+
+                from zeeguu.core.model.context_type import ContextType as CT
+                for t in CT.ALL_TYPES:
+                    CT.find_or_create(db.session, t, commit=False)
+                from zeeguu.core.model.source_type import SourceType as ST
+                for t in ST.ALL_TYPES:
+                    ST.find_or_create(db.session, t, commit=False)
+                db.session.commit()
+
+                resp = client.post(
+                    "/add_user/dump@test.com",
+                    data=dict(password="test", username="dump", learned_language="de"),
+                )
+                session_token = resp.data.decode("utf-8")
+                client.post(
+                    f"/find_or_create_article?session={session_token}",
+                    data=dict(url=URL_SPIEGEL_VENEZUELA),
+                )
+
+                dump = {}
+                for tname in KEEP_TABLES:
+                    rows = db.session.execute(text(f'SELECT * FROM "{tname}"')).fetchall()
+                    keys = list(db.session.execute(text(f'SELECT * FROM "{tname}"')).keys())
+                    dump[tname] = {
+                        "columns": keys,
+                        "rows": [
+                            [str(v) if v is not None else None for v in row]
+                            for row in rows
+                        ],
+                    }
+                    print(f"{tname}: {len(rows)} rows")
+
+                with open(FIXTURE_PATH, "w") as f:
+                    json.dump(dump, f)
+                print(f"Wrote {FIXTURE_PATH}")
+
+
+if __name__ == "__main__":
+    main()

zeeguu/api/endpoints/feature_toggles.py

@@ -3,6 +3,10 @@
 from zeeguu.api.endpoints import api
 from zeeguu.api.utils import cross_domain, requires_session
 from zeeguu.core.model import User
+from zeeguu.core.user_feature_toggles import (
+    features_for_user,
+    is_feature_enabled_for_user,
+)
 
 
 @api.route("/is_feature_enabled/<feature_name>", methods=["GET"])
@@ -15,82 +19,12 @@ def is_feature_enabled(feature_name):
 
     will return YES or NO
     """
-
-    func = _feature_map().get(feature_name, None)
-
-    if not func:
-        return "NO"
     user = User.find_by_id(flask.g.user_id)
-    if func(user):
+    if is_feature_enabled_for_user(feature_name, user):
         return "YES"
 
     return "NO"
 
 
-def features_for_user(user):
-    features = []
-    for name, detector_function in _feature_map().items():
-        if detector_function(user):
-            features.append(name)
-    return features
-
-
-"""
-    We have a code 'zeeguu-preview' which is used to invite
-    general users and should give access to the latest feature set
-    of Zeeguu. It can be used for usability tests and can be also
-    spread by word of mouth to new participants.
-"""
-
-
-def is_feature_enabled_for_user(feature_name, user):
-    return feature_name in features_for_user(user)
-
-
-def _feature_map():
-    return {
-        "audio_exercises": _audio_exercises,
-        "extension_experiment_1": _extension_experiment_1,
-        "no_audio_exercises": _no_audio_exercises,
-        "tiago_exercises": _tiago_exercises,
-        "new_topics": _new_topics,
-        "daily_audio": _daily_audio,
-    }
-
-
-def _new_topics(user):
-    return True
-
-
-def _tiago_exercises(user):
-    right_user = user.invitation_code == "Tiago" or user.id == 534 or user.id == 4022
-    right_language = user.learned_language.code in ["da"]
-    return right_user and right_language
-
-
-def _no_audio_exercises(user):
-    return user.is_member_of_cohort(447)
-
-
-def _audio_exercises(user):
-    return user.is_member_of_cohort(444)
-
-
-def _extension_experiment_1(user):
-    return (
-        (user.is_member_of_cohort(437))
-        or user.id in [3372, 3373, 2953, 3427, 2705]
-        or user.id > 3555
-    )
-
-
-def _daily_audio(user):
-    return True
-    # return (
-    #     user.id in [8, 4607, 4022]
-    #     or user.id > 5228
-    #     or user.is_member_of_cohort(532)  # daily-audio
-    #     or user.is_member_of_cohort(529)  # pg25
-    #     or user.is_member_of_cohort(530)  # ada's cohort with invite code: summer25
-    #     or user.invitation_code in ["zeeguu-preview", "ec25"]
-    # )
+# Re-export for backward compatibility
+__all__ = ["features_for_user", "is_feature_enabled_for_user"]

zeeguu/api/endpoints/search.py

@@ -301,9 +301,12 @@ def subscribe_to_email_search(search_terms):
     user = User.find_by_id(flask.g.user_id)
     search = Search.find(search_terms, user.learned_language_id)
     receive_email = True
-    subscription = SearchSubscription.update_receive_email(
-        db_session, user, search, receive_email
-    )
+    try:
+        subscription = SearchSubscription.update_receive_email(
+            db_session, user, search, receive_email
+        )
+    except ValueError as e:
+        return make_error(401, str(e))
 
     return json_result(subscription.as_dictionary())
 
@@ -320,8 +323,11 @@ def unsubscribe_from_email_search(search_terms):
     search = Search.find(search_terms, user.learned_language_id)
 
     receive_email = False
-    subscription = SearchSubscription.update_receive_email(
-        db_session, user, search, receive_email
-    )
+    try:
+        subscription = SearchSubscription.update_receive_email(
+            db_session, user, search, receive_email
+        )
+    except ValueError as e:
+        return make_error(401, str(e))
 
     return json_result(subscription.as_dictionary())