fix: Guard article.inner_text() against stale element errors in BigDeal scraper

[Copilot]

A detached/stale Playwright element during article.inner_text() would abort the entire scrape attempt with no retry, since the outer wrapper only retries on PlaywrightTimeoutError.

Changes

• bigdeal.py: Wrap article.inner_text() in try/except PlaywrightError + except Exception with continue, matching Leon's scraper pattern — a bad element skips to the next post instead of aborting

try:
    text_content = article.inner_text()
except PlaywrightError as e:
    self.logger.debug(f"Failed to extract text from post {i}: {e}")
    continue
except Exception as e:
    self.logger.debug(f"Unexpected error extracting text from post {i}: {e}")
    continue

• tests/test_bigdeal_scraper.py: Fixed 5 broken tests where mock articles lacked evaluate.return_value = False — the comment-filtering step calls article.evaluate(...), and an unset Mock() returns a truthy value, causing all articles to be silently dropped as nested comments

• tests/test_bigdeal_scraper.py: Added test_scrape_facebook_inner_text_error_continues_to_next_post — verifies that a PlaywrightError on one article's inner_text() skips that element and returns the flavor from the next valid post

---

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.

[cloudflare-workers-and-pages]

Deploying daily-custard with    Cloudflare Pages

Latest commit:	0d1dbe6
Status:	✅  Deploy successful!
Preview URL:	https://470bccd1.daily-custard.pages.dev
Branch Preview URL:	https://copilot-sub-pr-30-again.daily-custard.pages.dev

View logs

[Copilot]

Just as a heads up, I was blocked by some firewall rules while working on your feedback. Expand below for details.

[Copilot]

Pull request overview

This PR improves the resilience of the Big Deal Facebook Playwright scraper by preventing a single stale/detached post element from aborting an entire scrape, and updates unit tests to reflect the scraper’s nested-comment filtering behavior.

Changes:

• Guard article.inner_text() in the BigDeal scraper so Playwright DOM errors skip the bad post and continue scanning.
• Fix existing BigDeal scraper tests by setting article.evaluate.return_value = False so mocked posts aren’t filtered out as nested comments.
• Add a regression test ensuring an inner_text() Playwright error on one post still allows a later valid post to be returned.

Reviewed changes

Copilot reviewed 1 out of 1 changed files in this pull request and generated 1 comment.

File	Description
app/scrapers/bigdeal.py	Handles inner_text() Playwright errors per-article to avoid aborting the scrape attempt on a single bad element.
tests/test_bigdeal_scraper.py	Updates mocks to pass nested-article filtering and adds a test covering inner_text() PlaywrightError continuation behavior.

[Copilot]

In this test, mock_page.query_selector_all.return_value is used for all selectors, but the scraper calls query_selector_all('text="See more"') before query_selector_all('[role="article"]'). With the current setup the article mocks can be treated as "See more" buttons (calling is_visible()/click()), which makes the test less representative and could hide regressions. Consider routing query_selector_all via side_effect (as done in test_scrape_facebook_page_expands_see_more_buttons) to return [] for the "See more" selector and the articles list for [role="article"].

Suggested change

	mock_page.query_selector_all.return_value = [mock_article1, mock_article2]
	def query_selector_all_side_effect(selector: str):
	if selector == 'text="See more"':
	# No "See more" buttons in this scenario
	return []
	if selector == '[role="article"]':
	# Return the mocked posts
	return [mock_article1, mock_article2]
	return []
	mock_page.query_selector_all.side_effect = query_selector_all_side_effect