Add BDD-style Playwright test suite for website verification (#91)

62 tests across 4 test files:
- test_navigation.py: sidebar sections, navbar, wireframe alignment (21 tests)
- test_requirements_page.py: accordion behavior, 18 callouts (7 tests)
- test_key_pages.py: homepage, about, how-to-use, architecture, pubs, data endpoint (22 tests)
- test_explorer.py: search, filters, facet counts, sample card, view modes (12 tests)

Run with: pytest tests/ -p no:base-url
Against local: ISAMPLES_BASE_URL=http://localhost:5860 pytest tests/

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

Author: rdhyee

tests/conftest.py

@@ -0,0 +1,42 @@
+"""
+Shared fixtures for iSamples website BDD tests.
+
+Usage:
+    # Against live site (default):
+    pytest tests/
+
+    # Against local Quarto preview:
+    ISAMPLES_BASE_URL=http://localhost:5860 pytest tests/
+
+    # With visible browser:
+    pytest tests/ --headed
+"""
+import os
+import pytest
+from playwright.sync_api import sync_playwright
+
+
+SITE_URL = os.environ.get("ISAMPLES_BASE_URL", "https://isamples.org")
+
+
+@pytest.fixture(scope="session")
+def browser():
+    with sync_playwright() as p:
+        browser = p.chromium.launch(
+            headless="--headed" not in " ".join(os.sys.argv),
+        )
+        yield browser
+        browser.close()
+
+
+@pytest.fixture
+def page(browser):
+    context = browser.new_context(viewport={"width": 1280, "height": 900})
+    page = context.new_page()
+    yield page
+    context.close()
+
+
+@pytest.fixture(scope="session")
+def site_url():
+    return SITE_URL

tests/test_explorer.py

@@ -0,0 +1,83 @@
+"""
+Interactive Explorer tests — verify the search/filter/globe experience.
+
+These tests hit the live Explorer page and wait for DuckDB-WASM to initialize.
+They are slower (~30s+) due to remote parquet loading.
+"""
+import pytest
+from conftest import SITE_URL
+
+EXPLORER_URL = f"{SITE_URL}/tutorials/isamples_explorer.html"
+
+
+@pytest.fixture
+def explorer_page(page):
+    """Navigate to Explorer and wait for initial load."""
+    page.goto(EXPLORER_URL, wait_until="domcontentloaded", timeout=60000)
+    return page
+
+
+class TestExplorerLoads:
+    """Explorer page should load and initialize DuckDB-WASM."""
+
+    def test_page_loads(self, explorer_page):
+        assert "Explorer" in explorer_page.title()
+
+    def test_has_search_input(self, explorer_page):
+        # Observable inputs render after JS executes — wait for them
+        search = explorer_page.locator("input[type='text']")
+        search.first.wait_for(state="visible", timeout=15000)
+        assert search.count() > 0
+
+    def test_has_source_filter_section(self, explorer_page):
+        assert explorer_page.get_by_text("Source", exact=True).count() > 0
+
+    def test_has_material_filter_section(self, explorer_page):
+        assert explorer_page.get_by_text("Material", exact=True).count() > 0
+
+    def test_has_sampled_feature_filter(self, explorer_page):
+        assert explorer_page.get_by_text("Sampled Feature").count() > 0
+
+    def test_has_specimen_type_filter(self, explorer_page):
+        assert explorer_page.get_by_text("Specimen Type").count() > 0
+
+    def test_has_max_samples_slider(self, explorer_page):
+        # Observable range input renders after JS — wait for it
+        slider = explorer_page.locator("input[type='range']")
+        slider.first.wait_for(state="attached", timeout=15000)
+        assert slider.count() > 0
+
+    def test_has_view_mode_selector(self, explorer_page):
+        assert explorer_page.get_by_text("Globe").count() > 0
+        assert explorer_page.get_by_text("List").count() > 0
+        assert explorer_page.get_by_text("Table").count() > 0
+
+
+class TestExplorerFacetCounts:
+    """Facet counts should appear from pre-computed summaries."""
+
+    def test_source_checkboxes_have_counts(self, explorer_page):
+        """Source checkboxes should show sample counts (loaded from 2KB summary)."""
+        # Wait for facet summaries to load (they're tiny, should be fast)
+        explorer_page.wait_for_timeout(5000)
+        # Check that at least one source has a count in parentheses
+        sesar = explorer_page.get_by_text("SESAR")
+        assert sesar.count() > 0
+
+    def test_four_sources_present(self, explorer_page):
+        """All 4 data sources should appear as filter options."""
+        explorer_page.wait_for_timeout(5000)
+        for source in ["SESAR", "OPENCONTEXT", "GEOME", "SMITHSONIAN"]:
+            assert explorer_page.get_by_text(source).count() > 0, f"Missing source: {source}"
+
+
+class TestExplorerSampleCard:
+    """Sample Card section should exist."""
+
+    def test_has_sample_card_section(self, explorer_page):
+        assert explorer_page.get_by_text("Sample Card").count() > 0
+
+    def test_sample_card_shows_click_prompt(self, explorer_page):
+        """Before clicking a point, card should show instructions."""
+        explorer_page.wait_for_timeout(3000)
+        assert explorer_page.get_by_text("Click a point").count() > 0

tests/test_key_pages.py

@@ -0,0 +1,128 @@
+"""
+Key page smoke tests — verify all important pages load and have expected content.
+
+These catch broken pages, missing assets, and regressions in page structure.
+"""
+import pytest
+from conftest import SITE_URL
+
+
+class TestHomepage:
+    """Homepage should load with hero, globe animation, and showcase."""
+
+    def test_homepage_loads(self, page):
+        response = page.goto(SITE_URL, wait_until="domcontentloaded")
+        assert response.status == 200
+
+    def test_homepage_has_title(self, page):
+        page.goto(SITE_URL, wait_until="domcontentloaded")
+        assert "iSamples" in page.title()
+
+    def test_homepage_has_hero_text(self, page):
+        page.goto(SITE_URL, wait_until="domcontentloaded")
+        assert page.get_by_text("Internet of Samples").count() > 0
+
+    def test_homepage_has_globe_animation(self, page):
+        page.goto(SITE_URL, wait_until="domcontentloaded")
+        globe = page.locator("img[src*='isamples_globe']")
+        assert globe.count() > 0
+
+
+class TestAboutPage:
+    """About page should have all 4 wireframe sections."""
+
+    def test_about_loads(self, page):
+        response = page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        assert response.status == 200
+
+    def test_has_objectives(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        assert page.locator("h2:has-text('Objectives')").count() > 0
+
+    def test_has_team(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        assert page.locator("h2:has-text('Team')").count() > 0
+
+    def test_has_photo_gallery(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        assert page.locator("h2:has-text('Photo Gallery')").count() > 0
+
+    def test_has_background_history(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        assert page.locator("h2:has-text('Background')").count() > 0
+
+    def test_has_pi_names(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        for name in ["Kerstin Lehnert", "Andrea Thomer", "Neil Davies", "David Vieglais"]:
+            assert page.get_by_text(name).count() > 0, f"Missing PI: {name}"
+
+
+class TestHowToUsePage:
+    """How to Use page should have quick start and data tables."""
+
+    def test_how_to_use_loads(self, page):
+        response = page.goto(f"{SITE_URL}/how-to-use.html", wait_until="domcontentloaded")
+        assert response.status == 200
+
+    def test_has_quick_start(self, page):
+        page.goto(f"{SITE_URL}/how-to-use.html", wait_until="domcontentloaded")
+        assert page.get_by_text("Quick Start").count() > 0
+
+    def test_has_data_sources_table(self, page):
+        page.goto(f"{SITE_URL}/how-to-use.html", wait_until="domcontentloaded")
+        assert page.get_by_text("SESAR").count() > 0
+        assert page.get_by_text("OpenContext").count() > 0
+
+
+class TestArchitecturePage:
+    """Architecture overview should have structured sections."""
+
+    def test_architecture_loads(self, page):
+        response = page.goto(f"{SITE_URL}/design/index.html", wait_until="domcontentloaded")
+        assert response.status == 200
+
+    def test_has_core_principles(self, page):
+        page.goto(f"{SITE_URL}/design/index.html", wait_until="domcontentloaded")
+        assert page.get_by_text("Core Principles").count() > 0
+
+    def test_has_link_to_requirements(self, page):
+        page.goto(f"{SITE_URL}/design/index.html", wait_until="domcontentloaded")
+        assert page.locator("a[href*='requirements']").count() > 0
+
+    def test_has_link_to_metadata_model(self, page):
+        page.goto(f"{SITE_URL}/design/index.html", wait_until="domcontentloaded")
+        assert page.locator("a[href*='metadata']").count() > 0
+
+
+class TestPublicationsPage:
+    """Publications page should have presentations and bibliography."""
+
+    def test_publications_loads(self, page):
+        response = page.goto(f"{SITE_URL}/pubs.html", wait_until="domcontentloaded")
+        assert response.status == 200
+
+    def test_has_presentations_section(self, page):
+        page.goto(f"{SITE_URL}/pubs.html", wait_until="domcontentloaded")
+        assert page.get_by_text("Presentations").count() > 0
+
+    def test_has_spnhc_talk_link(self, page):
+        page.goto(f"{SITE_URL}/pubs.html", wait_until="domcontentloaded")
+        assert page.locator("a[href*='youtu']").count() > 0
+
+
+class TestDataEndpoint:
+    """data.isamples.org should serve parquet files with range requests."""
+
+    def test_facet_summaries_accessible(self, page):
+        # Use Playwright's API request context (not page.goto which triggers download)
+        response = page.request.head(
+            "https://data.isamples.org/isamples_202601_facet_summaries.parquet"
+        )
+        assert response.status in (200, 206)
+
+    def test_wide_parquet_supports_range_requests(self, page):
+        response = page.request.head(
+            "https://data.isamples.org/isamples_202601_wide.parquet"
+        )
+        assert response.status in (200, 206)
+        assert "bytes" in response.headers.get("accept-ranges", "")

tests/test_navigation.py

@@ -0,0 +1,137 @@
+"""
+Navigation structure tests — verify sidebar and navbar match the April 2026 wireframe.
+
+These tests verify the structural changes from PRs #89 (nav restructure)
+and #90 (accordion menus).
+"""
+import pytest
+from conftest import SITE_URL
+
+
+class TestSidebarSections:
+    """Sidebar should show the wireframe's section names."""
+
+    def test_sidebar_shows_architecture_and_vocabularies(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Architecture and Vocabularies").count() > 0
+
+    def test_sidebar_does_not_show_old_information_architecture(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Information Architecture").count() == 0
+
+    def test_sidebar_shows_research_and_resources(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Research & Resources").count() > 0
+
+    def test_sidebar_does_not_show_separate_published_research(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Published Research", exact=True).count() == 0
+
+    def test_sidebar_does_not_show_separate_resources(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        # "Resources" alone shouldn't appear as a section header
+        # (it's OK inside "Research & Resources")
+        sections = sidebar.locator(".sidebar-section .sidebar-section-header")
+        texts = [s.text_content().strip() for s in sections.all()]
+        assert "Resources" not in texts
+
+
+class TestSidebarHowToUse:
+    """How to Use section should have 5 sub-items matching wireframe."""
+
+    def test_how_to_use_has_overview(self, page):
+        page.goto(f"{SITE_URL}/how-to-use.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Overview", exact=True).count() > 0
+
+    def test_how_to_use_has_deep_dive(self, page):
+        page.goto(f"{SITE_URL}/how-to-use.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Deep-Dive Analysis").count() > 0
+
+    def test_how_to_use_has_globe_viz(self, page):
+        page.goto(f"{SITE_URL}/how-to-use.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("3D Globe Visualization").count() > 0
+
+    def test_how_to_use_has_search_explorer(self, page):
+        page.goto(f"{SITE_URL}/how-to-use.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Search Explorer").count() > 0
+
+    def test_how_to_use_has_narrow_vs_wide(self, page):
+        page.goto(f"{SITE_URL}/how-to-use.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Technical: Narrow vs Wide").count() > 0
+
+
+class TestSidebarAbout:
+    """About section should have 4 items matching wireframe."""
+
+    def test_about_has_objectives(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Objectives", exact=True).count() > 0
+
+    def test_about_has_pis_and_contributors(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("PIs and Contributors").count() > 0
+
+    def test_about_has_photo_gallery(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Photo Gallery").count() > 0
+
+    def test_about_has_background_history(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Background & History").count() > 0
+
+
+class TestSidebarResearchResources:
+    """Research & Resources should have 3 items matching wireframe."""
+
+    def test_has_publications_and_conferences(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Publications & Conferences").count() > 0
+
+    def test_has_zenodo_community(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Zenodo Community").count() > 0
+
+    def test_has_github_repositories(self, page):
+        page.goto(f"{SITE_URL}/about.html", wait_until="domcontentloaded")
+        sidebar = page.locator(".sidebar-navigation")
+        assert sidebar.get_by_text("Github Repositories").count() > 0
+
+
+class TestNavbar:
+    """Top navbar should have the 4 main items."""
+
+    def test_navbar_has_home(self, page):
+        page.goto(SITE_URL, wait_until="domcontentloaded")
+        navbar = page.locator(".navbar")
+        assert navbar.get_by_text("Home", exact=True).count() > 0
+
+    def test_navbar_has_interactive_explorer(self, page):
+        page.goto(SITE_URL, wait_until="domcontentloaded")
+        navbar = page.locator(".navbar")
+        assert navbar.get_by_text("Interactive Explorer").count() > 0
+
+    def test_navbar_has_how_to_use(self, page):
+        page.goto(SITE_URL, wait_until="domcontentloaded")
+        navbar = page.locator(".navbar")
+        assert navbar.get_by_text("How to Use").count() > 0
+
+    def test_navbar_has_about(self, page):
+        page.goto(SITE_URL, wait_until="domcontentloaded")
+        navbar = page.locator(".navbar")
+        assert navbar.get_by_text("About", exact=True).count() > 0