The live generation test depends on real OpenAI + web scraping.
Adding a second source improves chances of finding articles.
Retries increased from 1 to 2.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Remove content settings from settings table (moved to themes).
Add theme_id to sources and syntheses. Pipeline loads content
settings from the selected theme. Generate endpoint requires theme_id.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The LLM now determines if scraped content is a real article during
classify (zero extra cost). The separate LLM link extraction option
is removed — heuristic extraction is sufficient.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
'AI Weekly' with 7 days was too narrow — most articles filtered as
too old. 'Intelligence Artificielle' with 30 days gives more results.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Replace waitForLoadState('networkidle') with 'domcontentloaded' to
avoid hangs caused by the Cloudflare Turnstile script
- Add { waitUntil: 'domcontentloaded' } to all page.goto() calls
- Rewrite registration test to use the API directly instead of UI form
submission, since the Turnstile script causes continuous DOM mutations
that prevent Playwright from clicking elements
- Fix admin-providers test to select Gemini from the provider dropdown
when multiple providers are enabled
- Fix sources test to clean up leftover sources before asserting counts
- Update generation-live LLM call type assertion from 'rewrite' to
'search' to match the current pipeline (classify_summarize is optional)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add use_brave_search boolean field to all settings structs, DB layer,
SQL queries, frontend types, i18n labels, and test fixtures following
the same pattern as use_llm_for_source_links.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add a user-configurable batch_size setting (default 5, range 1-20)
that controls how many articles are processed in parallel during
Phase 1 scrape+classify. Previously hardcoded to 5.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace the single `models` array in `ProviderConfig` and `AdminProvider`
with separate `models_scraping` / `models_websearch` lists. Rename
`ai_model_writing` → `ai_model_websearch` in `UserSettings` and all
references (Settings page, admin Providers page, E2E test, fixtures,
and unit tests). Update i18n label for the second dropdown to
"Modele d'IA (Recherche Web)".
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- filter_empty_scraped_articles: removes articles with empty scraped content
(too old, soft 404, scrape failure) before the rewrite pass, preventing
empty articles in the final synthesis
- restore_scraped_urls: already existed, now has unit tests
- E2E test: added assertions for no Wikipedia URLs, no empty summaries,
and updated settings payload with new fields (max_articles_per_source,
source_diversity_window)
- 4 new unit tests for filter_empty + restore_scraped_urls
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Use import.meta.url for ESM-compatible __dirname
- Source creation expects 201, not 200
- Clean up existing sources before adding to avoid unique constraint violation
- Fix E2E docker-compose build context to project root
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
oabrivard