Phase 6 Inventory — Read-Only Inspection Pass

Generated: 2026-05-04
Test baseline confirmed: 256 tests collected

Section 1 — Tab and Surface Structure

Three top-level tabs (app.py line 2511)

school_finder_tab, major_explorer_tab, find_your_fit_tab = st.tabs(
    ["School Finder", "Major Explorer", "Find Your Fit"]
)

Shared Sidebar (app.py lines 418–620, render_sidebar())

Renders on every tab. Sub-sections in order:

• Title + caption
• --- divider
• Filters subheader
  • State multiselect
  • School type multiselect
  • Campus setting multiselect
  • School size multiselect
  • Minimum % Pell Grant slider
  • Mission-driven institutions expander (multiselect inside)
  • Schools’ avg SAT range slider
  • Undergraduate enrollment slider
  • Show small schools checkbox
  • Filter by major (CIP) selectbox (Phase 13.2)
• --- divider
• Your test scores (optional) subheader
  • SAT composite number_input
  • ACT composite number_input
  • Family income bracket selectbox
• --- divider
• Priorities subheader
  • Quick start selectbox (preset profiles)
  • 7× scoring dimension weight sliders

School Finder tab (app.py lines 2523–2603)

Primary render file: app.py; helper file: page_modules/school_finder.py (59 lines, cross-tab helpers only)
Approximate lines in app.py owned by this tab’s render logic: ~1,900 (sidebar helpers, PDF/CSV export, pin, results, map, comparison, detail cards)

Sub-sections in render order:

1. Title — “Results” (st.title)
2. Cross-tab CIP banner — shown once after Major Explorer handoff (app.py 2527–2531)
3. Summary banner — school count + active priorities + filter summary (app.py 2534, render_summary_banner())
4. Export row — rows-to-export number_input + CSV download button + PDF download button (app.py 2536, render_export_row())
5. Pin selector — multiselect + Clear pins button (app.py 2538, render_pin_selector())
6. Results table — st.dataframe with 14 columns including ProgressColumn for Match score (app.py 2540, render_results())
7. Map expander — “🗺️ Map of matching schools (N schools)” (app.py 2542, render_map())
8. Comparison expander — “📊 Compare pinned schools (2–4)” containing side-by-side table + grouped bar chart (app.py 2544, render_comparison())
9. Pinned school detail cards — one bordered card per pinned school (app.py 2546, render_pinned_cards())
   • Each card contains: context badges (Carnegie/test policy/MSI/religion), header with acceptance badge, “Why this rank?” explainer, CI bar, sensitivity tier badge, key metrics grid (8 metrics), radar chart, Student Body section (demographics stacked bar + outcomes detail), Earnings Mobility section (grouped bar chart), Program-level outcomes section (4 metric tiles), Programs offered expander (table with Explore buttons)
10. About the Data expander — methodology for BLS/OEWS/EP/geographic/work-environment data (app.py 2548–2601)

Major Explorer tab (page_modules/major_explorer.py — 1,135 lines)

Primary render file: page_modules/major_explorer.py, render() at line 1106

Sub-sections in render order:

1. Header + caption — “Major Explorer” (lines 1108–1111)
2. Alignment context banner — Phase 12.3: tip (quiz not taken), success (quiz taken), or silent (zero-variance vector) (lines 1124–1136)
3. Major picker — st.selectbox “Select a major”, annotated with ⭐ for aligned majors (lines 1161–1171)
4. Major description block — Phase 11.1A: hand-authored overview + “What you’ll learn” + “Typical classes” + related majors, or stub NCES text (lines 1183–1187)
5. “See schools strong in this major →” button — Phase 13.2 cross-tab handoff to School Finder (lines 1190–1200)
6. NY Fed Labor Market Outcomes — Phase 11.2B.3: 5 st.metric tiles (unemployment, underemployment, early wage, mid-career wage, grad degree share) (lines 1205–1207)
7. Scorecard Outcomes panel — 4 st.metric tiles (earnings 1yr/4yr/5yr, debt); “About these numbers” expander (lines 1217–1256)
8. Wages and projections summary — Phase 9.5: 4 st.metric tiles (median wage, 10yr growth, total employment, routing occupations count) (lines 1261–1263)
9. Career paths and occupations — Phase 9.4: st.dataframe of routing occupations with wage/growth/openings/education columns (lines 1267–1269)
10. Industries that hire this major — Phase 11.3.3: horizontal bar chart (top 10 NAICS + optional Other row) (lines 1273–1275)
11. Work environment — Phase 10.5: horizontal bar chart, 13 O*NET work-context elements colored by quartile (lines 1279–1283)
12. Where these jobs are — Phase 10.4: side-by-side USA choropleths (Total Employment + Location Quotient) (lines 1287–1290)
13. Schools strong in this major — Phase 8.4: ranked table of top 25 schools with “📌 Pin” buttons (lines 1294–1350)

Find Your Fit tab (page_modules/find_your_fit.py — 262 lines)

Primary render file: page_modules/find_your_fit.py, render() at line 288

Sub-sections in render order (pre-submission state):

1. Header + intro — “Find Your Fit”, O*NET methodology description (lines 97–112)
2. Questionnaire — 6 st.expander sections (one per RIASEC dimension), each with 10 st.radio items on a 5-point Likert scale (lines 115–145)
3. Progress bar — st.progress(total_answered / 60) (line 146)
4. “See my results” button — disabled until all 60 items answered (lines 149–170)

Post-submission state (fyf_submitted == True):

1. Your Interest Profile — Holland Code st.metric + horizontal bar chart (6 RIASEC dimensions) (lines 178–208)
2. Majors Aligned with Your Interests — top 10 table with correlation scores + “Explore →” buttons (lines 213–281)
3. Shared results banner — shown when ?riasec=... URL param detected (Phase 13.4b) (lines 226–246)
4. “Retake the questionnaire” button (line 303)

Section 2 — Chart Inventory

None of the 11 charts have Streamlit-level alt text, aria-label, or a programmatic descriptive caption rendered as accessible text above the chart. Hover templates exist on all but chart #5, but those are not screen-reader accessible.

Section 3 — Interactive Element Inventory

st.multiselect (6 instances)

• Keys follow consistent filter_* pattern for sidebar filters; pinned_schools is its own category.

st.selectbox (4 instances)

• Quick start selectbox intentionally has no key= (value is derived from weights, not state).

st.slider / st.select_slider (10 instances total)

• Weight slider keys follow consistent weight_{dim_key} pattern.
• SAT range and enrollment range sliders have no key (intentionally excluded from URL state, per comment at app.py 127).

st.checkbox (1 instance)

st.number_input (3 instances)

st.radio (60 instances — FYF questionnaire)

• All 60 items rendered in find_your_fit.py:_render_questionnaire() lines 115–143
• Keys: fyf_item_{item_id} — consistent fyf_item_ prefix
• Labels: verbatim O*NET Interest Profiler Short Form activity text — descriptive by design

st.button / st.form_submit_button

• Dynamic buttons (pin_btn_*, explore_riasec_*, explore_program_*, see_schools_*) encode enough identity in the key to be unique per rerun.
• Labels are descriptive except “Explore →” which lacks context without surrounding row content (open question for accessibility).

st.download_button (2 instances)

st.tabs (1 instance)

• st.tabs(["School Finder", "Major Explorer", "Find Your Fit"]) — app.py 2511
• Cannot be programmatically switched; cross-tab handoffs use banners instead.

st.expander (11 instances)

• RIASEC expander state is persisted in st.session_state["fyf_expander_states"] keyed by dimension letter.

Custom HTML (1 instance)

• Earnings Mobility empty state — st.markdown(..., unsafe_allow_html=True) with inline <div style='background:#f0f0f0;...'> (app.py 1583–1587). No ARIA role.

Section 4 — Data Source Inventory

Open question: data/processed/top_60_cip4.csv exists on disk (scripts/rank_top_cip4.py produces it) but is not consumed by the runtime app. Not counted above.

Section 5 — Color and Palette Usage

Centralized palette

• No centralized palette constant in config.py.
• A named dict _DEMO_COLORS exists in app.py at lines 1408–1421 (module-level, not in config). It is used only for demographics + earnings mobility charts.
• A local palette list ["#0072B2", "#E69F00", "#009E73", "#CC79A7"] is defined inline in _build_comparison_chart() (app.py 2348) — not shared.

Color literals by usage area

Demographics / Earnings Mobility charts — _DEMO_COLORS dict (app.py 1408–1421, Wong 2011 colorblind-safe)

Comparison chart — inline local palette (app.py 2348)

["#0072B2", "#E69F00", "#009E73", "#CC79A7"] — up to 4 schools, Wong 2011

CI bar chart (app.py 1652–1677)

Work environment quartile coloring (major_explorer.py 514–523)

• Colors are hardcoded in _quartile_color() local function; not imported from config.

NAICS distribution chart (major_explorer.py 974–975)

RIASEC interest profile bar (find_your_fit.py 195)

• #1f77b4 — default Plotly blue, NOT Wong palette, inconsistent with comparison/demo charts

Choropleth color scales (major_explorer.py 688, 724)

• Both Total Employment and Location Quotient choropleths use Plotly named scale “Viridis”

Map scatter (app.py 2116)

• Color scale: “RdYlGn” (Plotly named scale, mapped to match score 0–100)
• Marker outline: "black" (named)
• Map land: "lightgray" (named), subunit borders: "white" (named)

Acceptance badges — emoji + PDF colors

Rank emojis in comparison table: 🟢 = best, 🟡 = middle, 🔴 = worst — defined inline at app.py 1244–1249 and 2262–2270.

Miscellaneous

Section 6 — URL State and Shareable Surfaces

All URL state is managed by _parse_url_into_session_state() (app.py 324–383) and _write_session_state_to_url() (app.py 386–412). The canonical parameter map is URL_PARAM_MAP (app.py 132–163). The encoding helpers (encode_tab, decode_tab, encode_riasec, decode_riasec) live in url_state.py.

Complete URL parameter map

Intentionally omitted: SAT range slider, enrollment range slider (tuple values, excluded per comment at app.py 127–130).

Clamping and validation logic

• List params: _clamp_filter_list(key, valid_list) drops any value not in the current dataset (guards against stale URLs from old data versions).
• Income bracket: clamped to config.INCOME_BRACKETS keys; garbage resets to "Don't know / skip" (app.py 380–381).
• Tab: decode_tab() in url_state.py accepts only {"school_finder", "major_explorer", "find_your_fit"}; any other value returns None and is dropped. Default ("school_finder") is also dropped so it doesn’t appear in URLs.
• RIASEC: decode_riasec() validates exact 18-char format (R##I##A##S##E##C##), dimension order, and score range 0–40. Any failure returns None.
• _url_parsed guard: entire parse runs exactly once per browser session to prevent subsequent widget interactions from being overwritten (app.py 330–331).

Section 7 — Empty State and Error Handling

Section 8 — Cross-Tab Navigation

Navigation buttons

Note: No button directly navigates the tab widget. st.tabs() cannot be programmatically switched in Streamlit. All cross-tab transitions require the user to manually click the target tab after seeing a success/info banner.

Session state keys involved in cross-tab handoffs

Focus management / scroll behavior

• None exists. No st.rerun() scroll-to-top behavior, no JavaScript focus management, no st.empty() scroll anchors.
• After clicking a cross-tab button, the user sees a banner on the current tab or must scroll up to see the banner on the target tab after switching.

Section 9 — Test File Inventory

Oversized files — cohesion analysis

tests/test_naics_data_prep.py (26 tests, 205 lines)

Three distinct sub-groupings, each with its own scope="module" fixture:

1. soc_to_naics3 fixture group (lines 1–112, 8 tests): validates data/cleaned/soc_to_naics3.csv schema, shape, share-sum invariant, suppression-marker cleanup, and two spot-checks (Software Devs → NAICS 541, RNs → NAICS 622).
2. cip4_naics3_dist fixture group (lines 113–220, 9 tests): validates data/cleaned/cip4_naics3_distribution.csv schema, share-sum invariant, CIP4/NAICS3 format, spot-checks, routing-dilution observable, and Military Tech empty-state.
3. Loader + utility group (lines 221–273, 7 tests + 2 utility tests): validates naics_distribution.get_naics_distribution() behavior, normalize_cip4() utility with 3 format variants.

Natural split lines: → test_soc_to_naics3.py (group 1), test_cip4_naics3_distribution.py (group 2), test_naics_loader.py (group 3). The 2 normalize_cip4 tests in group 3 could go into test_utils.py.

tests/test_major_explorer_alignment.py (21 tests, 200 lines)

Four distinct sub-groupings, covering four separate Phases and four separate functions/modules:

1. Phase 12.3 — _get_aligned_cip4_set() (lines 36–104, 6 tests): no-quiz-taken gate, shape when quiz taken, zero-variance case, caching, cache invalidation, dot-format contract.
2. Phase 13.1 — set_major_explorer_cip() (lines 111–130, 3 tests): session-state writes, no-dot format normalization, overwrite behavior.
3. Phase 13.2 — set_school_finder_cip(), initial_cip_filter_index(), apply_cip_filter() (lines 133–203, 8 tests): session-state writes, filter restricts to offering schools, stale-value fallback.
4. Phase 13.3 — get_school_all_programs() (lines 207–266, 4 tests): sort order, empty-school case, NaN-last sort, cip4 dot-format in output.

Natural split lines: → test_riasec_alignment.py (group 1), test_major_explorer_xtab.py (groups 2+3), test_programs_offered.py (group 4).