Task 1 set context['project_team_pairs_json'] = json.dumps(pairs), then the
template rendered it with |json_script — which also calls json.dumps on the
value. Result was a JSON-encoded string-of-a-string in the <script
id="projectTeamPairs"> tag, so JSON.parse() returned a string (not a list)
and the pill-popover IIFE died on pairs.forEach(...). Symptom: all three
filter pills clickable but unresponsive.
Fix: pass the raw Python list; let |json_script own the serialisation (the
established pattern for team_workers_map_json and the other *_json keys on
the payroll dashboard).
Tests updated to read the raw list from resp.context. Added an end-to-end
regression test that extracts the rendered <script id="projectTeamPairs">
payload and asserts JSON.parse() would return a list (not a string) —
catches any future regression of this class even if the test suite and the
view drift apart.
Verified in the browser: all three pill popovers now open on click and
Choices.js lazy-initialises correctly for projects/teams.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Main interactive layer for the inline-filters feature. Appends two
blocks to report.html (inside {% block content %}, before the final
{% endblock %}):
1. Choices.js CDN <link> + <script> (admin-only gated, SRI-hashed) —
moved here because Task 5 will delete _report_config_modal.html,
which previously loaded the CDN. Keeping this on the report page
directly means the pills stay functional after modal retirement.
2. A scoped IIFE that wires up the three filter pills into an
interactive, state-managed UI:
- Click pill -> open popover (lazy-inits Choices.js on first open)
- Click outside / Esc / other pill -> close
- OK commits popover's local edits into pending state; dirty pills
get the orange outline + pulsing dot; Apply button slides in
- Cross-filter: picking projects auto-removes now-invalid teams
with toast notice ("Removed Team X — no logs on selected
projects"), and vice versa. Scope = entire history.
- Apply -> rebuilds querystring from pending state + navigates
(full page reload, same URL scheme as the retired modal)
- Reset -> reverts all pills to URL-current values
XSS-safe throughout: textContent and createElement; no innerHTML with
user data. Matches the pattern in base.html's work-log-payroll modal
from the Work-Log Payroll feature.
Graceful fallback: if Choices.js CDN fails to load, the module bails
early with a console warning; native <select multiple> still works
inside the popovers.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces the three static filter pills with clickable buttons and
inline popover shells below each one. Popovers remain hidden by
default (hidden attribute) — the JS module in Task 4 will wire up
open/close, dirty state, and Apply behaviour.
Structure per pill:
- .filter-pill-wrap (position-relative container)
- <button class="filter-pill filter-pill--editable" data-filter="...">
with chevron indicating clickability
- <a class="filter-pill__x"> (existing × clear-filter link, preserved)
- .filter-popover (the editable widget — date picker for the Date
pill, Choices.js multi-select for Projects/Teams pills)
Apply + Reset buttons sit in .apply-filters-group at the right end,
initially hidden. A <div id="filter-toast-container"> is pre-placed
for the cross-filter auto-removal notices.
Three json_script blocks embed the data the JS needs:
- projectTeamPairs: (project_id, team_id) pairs for cross-filter
- urlSelectedProjectIds / urlSelectedTeamIds: current URL state for
dirty diffing + reset
No visible behaviour change yet (no CSS, no JS). Page renders same
as before until Tasks 3-4 light it up.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Serialises distinct (project_id, team_id) pairs from WorkLog as JSON on
the generate_report context. The upcoming pill-popover JS (Task 4 of the
inline-filters plan) uses this to hide teams that haven't worked on a
selected project (and vice versa) without any extra HTTP round-trips.
Scope: entire history (not the report date range) — cross-filter is about
data possibility, not data shown in this period. Filters out NULL
project or team (can't cross-filter on NULL).
2 tests cover: key is populated with correct pairs; NULL-team logs don't
leak into the pairs list.
No visible behaviour change — template doesn't consume the new key yet.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Task-by-task plan for the design at 30d0991. 6 tasks, 1 checkpoint
after Task 4 (all pill interactions demoable, modal still as fallback).
Tasks:
1. Backend: project_team_pairs_json context + 2 tests
2. Template: popover shells + Apply button + json_script embeds
3. CSS: pill-editable, pill-dirty, popover, apply-group, toast (~150 lines)
4. JS: pill-popover interactive module (~300 lines, scoped IIFE)
--- CHECKPOINT 1 ---
5. Retire modal: delete _report_config_modal.html, update index + report
templates, keep backend context keys (still used by pill markup)
6. QA + shipped note
Scope: ~480 LOC net added (not the ~330 estimated — JS came out larger
once written with proper state management + cross-filter). Tests grow
42 -> 44. One new CDN-loaded library? No — Choices.js already loaded
from Executive Report v2. Zero model changes, zero migrations.
Noted trade-off in Task 5: selected_project_ids / selected_team_ids
context keys were KEPT despite design doc suggesting removal — the
pill popovers still use them for pre-selection + URL-diff init. Only
the modal markup was retired.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Second brainstorm output of the day. New tab alongside Pending /
History / Loans & Advances (the URL pattern already established at
?status=pending|paid|loans — this slots in at ?status=adjustments).
Key decisions:
- Semantic badge palette: 5 colour categories mapped across 7 types.
Loan/advance repayments get +15% saturation — same family, hotter
signal for "money coming back" vs "money going out".
- Three multi-select filters (Type, Workers, Teams) via Choices.js.
Teams cross-filter Workers using JSON pair map (mirrors Feature 1's
project<->team pattern). Auto-remove invalid selections with toast.
- Single-date default with optional range toggle; presets for
Today / This week / This month.
- Sticky filter bar; sortable columns (Date / Worker / Amount / Status).
- Group-by toggle: Flat / By Type / By Worker. Collapsible group
headers show count + net sum per group (+R additive / -R deductive).
- Bulk action bar (floating) for multi-row delete on unpaid rows only.
New endpoint POST /payroll/adjustments/bulk-delete/ filters
payroll_record__isnull=True for safety.
- Inline row actions reuse existing modals: Preview (unpaid) /
View Payslip (paid) / Edit + Delete (unpaid). Zero new modal code.
- Empty state, keyboard Esc, URL state for everything (bookmark-safe).
Scope: ~960 lines, ~12 tasks, 2 checkpoints. Uses existing
#addAdjustmentModal / #editAdjustmentModal / #payslipPreviewModal
already on payroll_dashboard.html — zero duplication.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Brainstorm output — Konrad's Checkpoint-3 UX request, now spec'd.
Key decisions:
- Pill-as-dropdown: existing filter pills become clickable popovers
- Explicit Apply button; hidden when no pending changes
- Modal retired; dashboard 'Generate Report' becomes a plain link
- Bidirectional cross-filter: selecting a project hides teams that
haven't worked on it (and vice versa). Strict behaviour with
auto-removal of now-invalid selections + toast notice.
- URL contract unchanged; PDF download unchanged (still uses
current querystring).
One new context key (project_team_pairs_json) serialises distinct
(project_id, team_id) pairs from WorkLog for client-side cross-filter.
~80 CSS lines for popover + dirty state + toast; ~150 JS lines for
one scoped module (createElement + textContent, XSS-safe).
Scope: 5-6 focused tasks, 1 checkpoint.
Next step: Feature 2 brainstorm (Payroll Adjustments Browser) before
handing both to writing-plans.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
QA summary:
- 42/42 tests pass
- manage.py check clean
- No pending migrations
- Route sanity: /report/, /report/?project=1&project=2, /report/pdf/ all
resolve (302 as anon, 200 as admin)
- PDF generation verified for populated and empty date ranges
Appends a "Shipped" block to the design doc that captures the final
QA state, the deferred items, and the notable design decisions made
during implementation. Konrad's inline-filter UX improvement (raised
during Checkpoint 3) is explicitly flagged for a future brainstorm.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
PDF template updated to match the new HTML structure: cover block
with static filter labels, hero KPI band (4 stacked 2x2), Chapter I
lifetime (Projects + Teams full-width, Projects now with Start /
Working Days / Avg-R-per-Working-Day columns), Chapter II selected
period (existing Total Paid Out hero + Loans/Advances pairs +
Labour Cost + Payments/Adjustments), Chapter III worker breakdown
(heading renamed), Chapter IV team x project pivot (new).
THIS YEAR section dropped per design doc section 3 (redundant with
All Time + Selected Period).
Same _build_report_context helper so HTML and PDF cannot drift in
data. All numbers identical. WeasyPrint-friendly: absolute units,
single-column body, no Font Awesome.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Final chapter of the executive redesign. Renders the
team_project_activity context as a pivot: rows=teams, columns=projects,
cell=COUNT(DISTINCT work-log dates). Zero cells show em-dashes in muted
grey (not '0') so non-zero cells stand out. Row totals, column totals,
and grand total on the bottom row.
Adds a tiny dictlookup template filter (format_tags.py) — Django
templates can't index a dict by a dynamic variable key, and the pivot
cell lookup is cells_by_project_id[col.id]. Defensive None + TypeError
guards so a malformed context can't 500 the page.
.table-total-row CSS: 2px top border + inset background for the footer
row so totals visually separate from the data rows.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds the numbered 'III' chapter heading above the existing Worker
Breakdown card (the widest table in the report). Promotes the table
to .report-numeric (tabular-nums) for perfect column alignment
across dynamic adjustment columns — Inter's tabular-nums variant
keeps the rand amounts pixel-aligned.
No data or structural changes to the breakdown itself.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces the old narrow four-card All-Time/YTD row (dropped in Task 9)
with two wider cards under a numbered 'Chapter I - Lifetime Context'
heading. Projects card gains Start, Working Days, Total Cost, and
Avg R / Working Day columns per the design. Teams card keeps name +
total.
Adds .chapter-heading and .chapter-num CSS for the orange numbered
markers (I, II, III, IV) and .report-numeric class that applies
tabular-nums across the number columns of every report table.
Renames the existing 'Selected Period' heading to Chapter II.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Chapter 0 of the executive redesign: four large cards at the top
showing Paid This Period, Outstanding Now (live, stamped with the
generation time), FoxFitt Avg/Day, and FoxFitt Avg/Month.
Drops the old four-small-cards All-Time/YTD row (YTD specifically
documented as redundant per design doc section 3). All-Time detail
moves into Chapter I in the next task.
New .stat-card--hero variant uses Poppins 1.85rem for the number,
uppercase tracked labels, subtle tertiary sub-lines. tabular-nums
keeps the R-amounts pixel-aligned across cards.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The b0d3829 overrides lost the cascade battle: the Choices.js CDN CSS
loads AFTER custom.css (inside the modal partial near </body>), so the
CDN's same-specificity rules won by load order. Dropdown still showed
white background + light-grey text in dark mode.
Fix: chain the root `.choices` class to every override (specificity
0,2,0 → 0,3,0) and add !important to color + background + border
properties that Choices.js hardcodes most aggressively. Now the
theme tokens always win regardless of load order.
Visual effect: dropdown option hover state now matches the selected
"Month(s)" button aesthetic (--bg-card-hover subtle lift with
--text-primary text) per Konrad's feedback.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Code review of Task 7 showed Choices.js shipping with white-bg +
light-grey-text defaults, which were unreadable in dark mode and
clashed with the app's premium aesthetic. The design doc section 10
scoped these overrides into the CSS work but they hadn't been written.
Adds ~70 lines to custom.css re-themeing every .choices__ selector to
use existing design tokens (--bg-card, --bg-inset, --text-primary,
--accent, --border-default, etc.). No hardcoded colours; both :root
and :root.light themes work automatically.
Key visual changes:
- Dropdown popup: dark-card background with shadow
- Options: primary text colour, accent hover highlight
- Selected chips: orange accent pill with white text
- Focus ring: 0.15rem accent glow on the input
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three pills under the header: date range, project(s), team(s). Shows
comma-joined names when multi-valued (project_name in context is
already a comma-joined string from Task 6). × buttons on the project
and team pills remove just that filter via a rebuilt querystring;
the calendar pill has no × (date range is required).
Helper context keys query_string_without_project / _without_team do
the rebuild in the view via QueryDict.setlist so multi-value keys
are properly stripped (pop() only removes the first occurrence).
Pill CSS uses existing design tokens (--bg-inset, --accent,
--text-primary, --border-default, --text-tertiary, --bg-card-hover)
so dark and light themes work without overrides.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Code review (on 748c7c7) flagged that Bootstrap CDN tags in base.html
use integrity=sha384-... + crossorigin=anonymous, but the Choices.js
tags added in Task 7 did not. Since both are admin-only privileged
contexts and Bootstrap sets the precedent, Choices.js should match.
Hashes computed from cdn.jsdelivr.net/npm/choices.js@10.2.0 via
curl ... | openssl dgst -sha384 -binary | openssl base64
No behavior change when the CDN is healthy; defense against a
compromised CDN serving altered bytes.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces the two single <select> elements in the report config modal
with <select multiple> enhanced by Choices.js (CDN 10.2.0, admin-only
gated, graceful fallback to native on CDN failure).
Removes the 'All Projects' / 'All Teams' placeholder option rows —
empty selection = all, matching Choices.js convention.
Persists selected values across submissions via two new context keys
(selected_project_ids, selected_team_ids) threaded through index() and
generate_report().
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
project_id/team_id become project_ids/team_ids (list[int] or None).
Every internal filter uses the __in lookup; M2M filters use the
id__in subquery pattern documented in CLAUDE.md's Django ORM gotcha.
generate_report and generate_report_pdf switch to request.GET.getlist.
Old URL ?project=1 still works - getlist returns a single-element list.
Return dict gains six hero-KPI keys: current_outstanding, current_as_of,
company_avg_daily, company_avg_monthly, company_working_days,
team_project_activity - ready for the template restructure in Tasks 9-12.
Tests: 3 new multi-filter tests; existing inflation tests updated to the
new kwarg names. 42 total, all pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Chapter I of the executive report needs per-project working-day count and
avg rand per working day. Instead of modifying the shared _get_labour_costs
helper (used by other sections with different column sets), enrich the
output INSIDE _build_report_context: wrap the raw result and add
working_days (distinct work-log dates per project) and avg_per_working_day
(total_cost / working_days, null-safe).
Also attaches start_date from the Project model (may be None if not set).
1 test added.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Chapter IV pivot backend: for each (team, project) pair in the given
work-logs queryset, counts distinct work-log dates. Returns columns
(projects), rows (teams with cell dict), column totals, and grand total
ready for direct template rendering.
Logs with NULL team or NULL project are excluded (can't pivot on NULL).
Teams/projects with zero activity don't appear as rows/columns — keeps
the pivot tight.
Tests cover shape, cell counts, row+column+grand totals, and
zero-activity team omission.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two tweaks from code review on 82594fa:
1. The sort `key=lambda r: -r['amount']` placed NEGATIVE amounts
(rare but possible: a project with only a deductive adjustment)
AHEAD of larger positive exposures. Swapped to
`key=lambda r: r['amount'], reverse=True` — same runtime, clearer
intent, correct for negatives.
2. test_team_filter_scopes_total only asserted the net total. A
partial scoping regression where the adjustment leaked but netted
to zero would have silently passed. Added two assertions that
by_project has exactly the expected 2 entries and R 500 never
appears in the amount list.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Hero KPI card 2 needs 'Outstanding NOW' scoped to the report's selected
projects/teams. This helper wraps _compute_outstanding, reshapes the
by_project dict into a sorted list, and exposes the net total for direct
rendering.
Tests cover unfiltered total, project-scoped total, and team-scoped
total (including the worker__teams subquery path for adjustments).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Computes company-wide avg daily and monthly labour cost for the
executive report's hero KPI band (cards 3 and 4). Denominator is
working days (distinct work-log dates), not calendar days — true
cost-of-a-productive-day metric per design section 2.
Monthly = daily * 30.44 (the 365.25/12 month-length approximation,
which keeps annualised totals correct on average).
Tests cover: empty DB returns zero, known values with assertAlmostEqual
for the 30.44 multiplication, and that multiple workers on one date
count as 1 working day (not N).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pure refactor: the ~45 lines of outstanding-payment math inside index()
(computing unpaid_wages + pending_adj_add - pending_adj_sub, with a
per-project breakdown) move into a standalone _compute_outstanding()
helper. index() now calls it with no arguments for unchanged behaviour.
The helper accepts optional project_ids / team_ids for Task 3.
No tests changed; 28/28 still pass. Dashboard Outstanding Payments
card shows the same value before and after.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Task-by-task plan for the design committed at 27cdb46. 14 tasks with
4 hard-pause checkpoints at natural demo points:
- After Task 6 (backend helpers done)
- After Task 8 (multi-select modal + filter pills)
- After Task 12 (full HTML layout — all 4 chapters)
- After Task 14 (PDF mirrored + QA + shipped note)
Task 1 is a pure refactor (extract _compute_outstanding from index())
so later tasks can reuse the dashboard math with filters. Tasks 2-5
add the new helpers alongside existing code with failing-test-first
discipline. Task 6 switches the main helper to multi-value filters
(project_ids/team_ids) — existing behaviour preserved via backward-
compatible getlist. Tasks 7-12 restructure the HTML template into
Hero + 4 chapters. Task 13 mirrors in the PDF. Task 14 QAs and ships.
~11 new tests across 4 test classes; total grows from 28 to ~39.
One new dependency: Choices.js 10.2.0 via CDN, admin-only gated,
graceful fallback to native multi-select on CDN failure.
Follows the CLAUDE.md conventions: # === SECTION === comments,
plain-English docstrings, subquery-filter pattern for M2M filters,
single-batched push at the end, Co-Authored-By trailer on every
commit, never amend.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Brainstorm output for rebuilding /report/ as an executive-grade dashboard.
Key decisions captured:
- Multi-select filters (Choices.js) with empty=all semantics
- Hero KPI band: Paid / Outstanding NOW / Avg R/day / Avg R/month
- Chapter I: Lifetime context with working-day denominator for avg cost
- Chapter II: Selected period (existing content, restructured)
- Chapter III: Worker breakdown (existing, restyled)
- Chapter IV: NEW team × project activity pivot
Current Outstanding reuses dashboard math (live, stamped with generation
time). Company cost velocity = lifetime cost / distinct work-log dates;
monthly = daily × 30.44.
No model changes. One new CDN dep (Choices.js). Target: ~650 LOC
including ~120 new tests. Four checkpoint pauses proposed for the
subsequent implementation plan.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Five focused updates from the Apr 22-23 bug-fix + gitignore session:
1. Fix stale supervisor-picker queryset doc: it was showing the pre-fix
Q(is_staff)|Q(is_superuser)|Q(groups__name='Work Logger') filter.
Since commit 0ceceeb the queryset is just User.objects.filter(is_active=True).
2. Update "How to add a new supervisor" step 2: Work Logger group
membership is no longer required for picker visibility — optional now.
3. Add "Schema name-drifts to remember" block near Key Models. Three
recurring gotchas that burned four subagent tasks across two sessions:
- PayrollAdjustment.description (not reason)
- log.adjustments_by_work_log (not payrolladjustment_set)
- log.overtime_amount (not log.overtime)
4. Add canonical test-command one-liner to the Commands section:
USE_SQLITE=true DJANGO_DEBUG=true python manage.py test core.tests -v 2
5. Add "Django ORM gotcha" subsection documenting the M2M filter +
values().annotate(Sum()) inflation bug and the id__in subquery fix
pattern (refs commit f1e246c, ReportContextFilterInflationTests).
No code changes; no test impact.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Problem: every time collectstatic ran on the VM, Flatlogic's web UI
detected the modified files in staticfiles/ and auto-committed them
with a generic "Ver XX.YY" message (e.g. "Ver 30.04 Fix reports and
add Supervisor"), pushing the result to gitea but not GitHub. Every
push of CSS/JS changes triggered a reconciliation dance. See the
"Ver 30.04" divergence resolved by commit e0d2c74 for the most recent
example — that was the 3rd or 4th recurrence of this exact pattern.
Fix:
1. Add staticfiles/ to .gitignore
2. Untrack all 627 currently-tracked files via `git rm -r --cached`
3. Document the change in CLAUDE.md (Project Structure, Static Assets,
and a new "NOT tracked in git" subsection)
Deploy consequence: the NEXT pull on the VM will delete
staticfiles/ from the working tree (because git sees those files
removed from the tree). Gemini MUST run `collectstatic --noinput`
IMMEDIATELY after `git pull` to repopulate from source, then
restart the service. Brief window of 404s on static assets is
acceptable at this scale (seconds).
After this change: collectstatic output lives on the VM's filesystem
but outside git's view, so Flatlogic's UI has nothing to auto-commit.
The recurring divergence pattern is permanently eliminated.
No runtime code changes — all 28 tests still pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Restores the .work-log-row hover rule into the collected CSS.
Replaces the Flatlogic-auto-noise commit (683e2b0) which had the misleading message 'Ver 30.04 Fix reports and add Supervisor' but only contained this same collectstatic output.
Reported: when creating a new team or project from the friendly UI
(/teams/new/ or /projects/new/), the Supervisor dropdown only lists
is_staff / is_superuser accounts. Users who should be eligible to
supervise (e.g. eendman, supervisor_smoke) are invisible in the
picker even though they are active.
Root cause:
core.forms._supervisor_user_queryset filtered to
is_active=True AND (is_staff OR is_superuser OR groups__name='Work Logger')
That was strictly more restrictive than the app's own permission
helper is_supervisor(user) in views.py, which grants supervisor
powers to ANYONE assigned to a team/project (via the team.supervisor
FK or project.supervisors M2M), regardless of group membership.
On Konrad's dev DB that excluded 2 of 6 active users from the picker
(one in a custom group, one in no group) even though both were valid
supervisor candidates by the permission model.
Fix:
Queryset now returns every active user. The act of assigning a user
to a team/project is what confers supervisor-ness downstream, so
the picker no longer needs a pre-registered allow-list. Inactive
users (is_active=False) remain excluded — the one hard guardrail.
Docstring rewritten to explain the new behavior and why. Stale comment
in TeamForm.__init__ updated to match (the old comment still described
the pre-fix Work-Logger-group requirement).
Tests: 4 new regression tests in SupervisorPickerQuerysetTests:
- regular active user is selectable (the core bug)
- user in an unrelated group is selectable
- inactive user is still excluded (guardrail)
- admin is still selectable (no regression for prior use case)
All 28 tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reported: when the generate-report page is filtered by BOTH project and
team, every amount in the "Worker Breakdown" and "Payments by Date"
tables blew up by ~100x. Example: Billy Baloyi R 5,400 (correct)
became R 604,800 (wrong, 112x) after selecting Wilkot + Civils One.
Root cause:
_build_report_context chained `records.filter(work_logs__project_id=X)
.distinct().filter(work_logs__team_id=Y).distinct()`. In Django's ORM
each chained M2M filter creates a SEPARATE JOIN alias on
core_payrollrecord_work_logs, so the SQL produces the cartesian product
of (matching-logs-for-project) x (matching-logs-for-team) rows per
PayrollRecord. A downstream `.values().annotate(Sum('amount_paid'))`
then summed across those duplicated rows - inflating every total by
N * M where N and M are the log counts per record.
Why total_paid_out looked correct: `.aggregate(Sum(...))` wraps the
query in a subquery when distinct() is in play, so it dedupes before
summing. `.values().annotate(Sum(...))` uses GROUP BY on the raw
joined rows and doesn't get that help.
Fix:
Replace chained M2M filters with id__in subquery filters:
records.filter(id__in=PayrollRecord.objects.filter(
work_logs__project_id=X).values('id'))
This keeps the outer queryset JOIN-free, so values().annotate(Sum())
aggregates over distinct records. Same pattern applied to the
adjustments team-filter (worker__teams M2M) for the adjustment
summary.
Tests: 5 new regression tests in ReportContextFilterInflationTests
covering project-only, team-only, both-filters, total_paid_out
invariant, and the adjustment summary path. All 24 tests pass
(19 existing + 5 new).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a consolidated regression test to WorkLogPayrollAjaxTests that
exercises: paid worker serialization shape, null team branch, OT flag
in JSON, full_page_url value, and adjustment payslip-link serialization.
Closes the 'Important' coverage gap flagged in Task 3's quality review.
Also appends a 'Shipped' block to the design doc summarising QA
status and capturing all five deferred nits (admin-gate consistency,
template branch tests, |default:0 redundancy, admin-gate expression
readability, background vs background-color) so they survive the
merge into project history.
All 19 tests pass. manage.py check clean. No migrations needed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Subtle background tint on hover to cue that the row is clickable.
Applied via .work-log-row class which Tasks 6-8 added to admin-only
rows in work_history.html, teams/detail.html, and projects/detail.html.
Supervisors never get the class, so hover never applies for them.
Deployment_timestamp cache-bust in base.html will beat Cloudflare's
edge cache (per CLAUDE.md Static Assets section).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Admins see cursor:pointer + data-log-id on each row. Click opens the
shared modal from base.html. Supervisors unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Admins see cursor:pointer + data-log-id on each row. Click opens the
shared modal from base.html. Supervisors unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Admin users get cursor:pointer + data-log-id on each row. Click
opens the shared modal from base.html. Supervisors unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two issues caught by code quality review on commit 2e60124:
1. C1 (critical): the <script> at line ~398 runs during HTML parsing,
BEFORE the modal markup at line ~627 has been parsed. getElementById
returned null, the `if (!modalEl) return;` guard silently exited the
IIFE, and the delegated click listener was never attached — so the
modal was completely dormant. Wrapped the IIFE body in a
DOMContentLoaded handler so the DOM is fully parsed before lookups.
2. I1 (a11y): added aria-labelledby on the modal root + a matching id on
the modal-title h5 so screen readers announce the title correctly
(Bootstrap 5 a11y convention).
No behavioural changes to the JS logic itself — only the wrapping and
two aria attributes on the markup.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Modal shell + JS click handler live in base.html so any page opts in
by adding data-log-id to a row. JS uses createElement + textContent
(matches worker_lookup_ajax pattern) to build the modal body from
JSON — no innerHTML. Supervisors never receive the markup.
Footer 'Open full page' links to /history/<id>/.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Active breadcrumb item now has aria-current="page" so screen
readers correctly announce the current page (Bootstrap 5 convention).
- Template section comments changed from {# --- #} to {# === #} to
match the CLAUDE.md Python convention used elsewhere in the project.
No logic or rendering changes.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
work_log_payroll_ajax serializes the helper's output to JSON with
floats (not Decimals), ISO dates, and payroll_record/worker IDs for
client-side link construction. Admin-only; supervisor = 403, anon =
302, unknown log = 404. Matches the worker_lookup_ajax pattern.
Added 4 view-level tests (total 16 passing).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The helper used log.overtime (which doesn't exist on WorkLog); the
correct field is overtime_amount. Combined with a defensive
`getattr(..., None) or 0`, the bug made the flag permanently False,
which would have silently hidden the 'Price now' banner in Tasks 3
and 4. Now reads overtime_amount directly (it's non-nullable with a
0.00 default, so no defensive shim is needed).
Adds 4 regression tests:
- test_overtime_needs_pricing_flag: the bug that just got fixed
- test_query_count_is_bounded: N+1 guard (4 queries regardless of worker count)
- test_empty_log_returns_zero_totals: log with no workers attached
- test_log_without_team_has_no_pay_period: log whose team became NULL
Also removes unused `reverse` import from tests.py.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pure-function helper that classifies each worker on a work log as
Paid / Priced-not-paid / Unpaid, collects log-linked adjustments,
and computes totals + pay-period context. Used by both the AJAX
endpoint and the full-page view so they can't drift.
Bootstraps core/tests.py (was empty); 8 tests cover the three
statuses, totals, log-linked adjustments, and the pay-period branch.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Routes /history/<id>/ and /history/<id>/payroll/ajax/ to stub views.
Both admin-gated; no data yet. Sets up the surface for Tasks 2-4.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Task-by-task plan for implementing the modal + /history/<id>/ page
designed in the companion design doc. 10 tasks, 4 hard-pause review
checkpoints (after tasks 2, 4, 6, 10). TDD for the pure helper
function (bootstraps the currently-empty core/tests.py), view-level
tests for the AJAX + detail endpoints, manual smoke tests for the
template/JS work.
Uses the existing worker_lookup_ajax JSON+DOM pattern for the modal
(createElement + textContent, not innerHTML) to match the codebase's
XSS-safe convention. Full page is server-side rendered via a Django
template.
No model changes. No migrations. Admin-only.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Brainstorm output for the next UI refinement. Adds a click-through from
any historic work log (Work History, team detail Recent Work Logs, project
detail Recent Work Logs) to a compact modal showing paid/unpaid status per
worker, with links out to /workers/<id>/ and /payroll/payslip/<pk>/. The
modal has a "Open full page" button that navigates to a new
/history/<log_id>/ route for bookmark-able detail + pay-period context
(via get_pay_period). Admin-only; supervisors unchanged.
Read-only pass; no model changes, no migrations. Uses existing data:
PayrollRecord.work_logs (M2M) and PayrollAdjustment.work_log (FK).
Also fixes local dev: run_dev.bat now sets DJANGO_DEBUG=true so runserver
auto-serves /static/ (prior behaviour: CSS 404 on localhost because
Django's dev server only serves static files when DEBUG=True; production
keeps DEBUG=false and is served by Apache, so unaffected).
Design doc: docs/plans/2026-04-22-work-log-payroll-crosslink-design.md
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Documents three things that came out of today's Phase 2 deploy session
and weren't previously written down:
1. Static Assets & Cache-Busting (new section): explains that production
traffic goes through Cloudflare with 4h edge cache; the
`deployment_timestamp` template variable is what breaks stale caches;
and why `request.timestamp` must never be used (the silent-default-to-1.0
bug that ate a couple of hours).
2. Environment Variables: inline notes for each var. Most important new
fact is that DEFAULT_FROM_EMAIL is now optional — falls back to
EMAIL_HOST_USER if unset (prevents the "Invalid address ''" failure
mode on outbound mail). Also documents that .env lives at BASE_DIR.parent
on Flatlogic and can only be edited via Gemini/shell.
3. Flatlogic Deployment: collectstatic isn't auto-run, django-dev.service
runs manage.py runserver (dev server in prod — known but works at this
scale), Cloudflare sits in front, VM has two git remotes (github +
gitea) that must stay in sync, VM-local safety branches for rollback,
and the "pick one write path" workflow rule to avoid divergence.
No code changes — documentation only.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When DEFAULT_FROM_EMAIL env var isn't set, it defaulted to an empty
string, causing every outbound email (receipts, payslips) to fail
with: Invalid address "".
Phase 1 removed the hardcoded Gmail fallback for security. The
cleanest restore — without reintroducing a secret default — is to
fall back to EMAIL_HOST_USER, which is already the authenticated
Gmail address we send AS. That address is always valid when SMTP
auth works, and it's already set on the VM (otherwise sending
would fail with an auth error instead).
Now:
DEFAULT_FROM_EMAIL = os.getenv("DEFAULT_FROM_EMAIL", "") or EMAIL_HOST_USER
Verified locally: when DEFAULT_FROM_EMAIL is unset and EMAIL_HOST_USER
is 'test@example.com', DEFAULT_FROM_EMAIL resolves to the same address.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
