api/test_durations.txt at master · zeeguu/api · GitHub

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
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