A confidence score reports how many rows were READ, never what they say:
the `all-positive` fixture scores a perfect 100 % while every credit is
imported as an expense, because each of those rows is perfectly readable.
Nothing between that score and the database looked at the signs — the
preview was an optional modal of 20 rows with no totals, and the final
confirmation listed the delimiter and the date format but neither the
amount mode nor the sign convention.
The `file-preview` step had been declared in `ImportWizardStep` since the
beginning and no dispatch ever aimed at it. It is a real step now,
traversed at every import and gated by nothing — in particular not by the
detection score, which is sign-blind by construction.
- `useImportWizard`: `parseAndPreview` parses and stops at the preview,
replacing `parsePreview` and the `parseAndCheckDuplicates` that jumped
straight to the duplicates ("skips preview step"); `checkDuplicates`,
dead code until now, is the preview's next button, so the rows the user
validated are the rows that get checked.
- `summarizeParsedRows`: the recap, pure and tested — outflows and their
total, inflows and theirs, rows in error. Totals stay SIGNED, since
magnitudes would hide the one thing the recap exists to expose. Computed
over the whole file, never over the twenty rows displayed.
- `flipSignFormat` + "Inverser les signes": the correction lands on the
CONFIGURATION, so it is persisted with the source and the next file from
that bank reads right on its own. In debit/credit mode it swaps the two
column indices rather than toggling a convention `mapRow` ignores there,
where a toggle would have been inert.
- `ImportConfirmation` states the amount mode, the sign convention (in the
mode that applies it) and the column mapping, named by header.
- `FilePreviewModal` removed: it was the redundant surface, and editing
the table alone would have mutated a still-live copy of it.
The `all-positive` KNOWN DEFECT marker is dropped rather than deleted. Its
three original expectations still hold — an unsigned file carries no
direction and detection cannot invent one — and two cases were added: the
recap tell (six outflows, zero inflows) and the honest limit, that
flipping this particular file only produces its mirror image.
1088 vitest (1054 before), tsc and vite build clean. No DB migration.
Resolves#329
Detection handed back a configuration it had never tested, and only ever
ran behind the magic-wand button. A source opened for the first time
therefore started on `defaultConfig` — `;`, `DD/MM/YYYY`, columns 0/1/2 —
plausible enough to import a whole file wrong rather than fail visibly.
`detectImportFormat` now REPLAYS what it just decided over every data row
of the file and returns the rate as a `DetectionScore`. The replay runs
`mapRow`, the same pure function `parseFilesInternal` runs at import time,
under the same column-level decimal arbitration: two mappers would be the
exact divergence this chantier removes — a score reading 100 % while the
import wrote different amounts. A test asserts the two agree row for row
on every fixture of the corpus.
The threshold is 90 %, and it only colours the banner. Below it the panel
warns and prints the detailed count ("132 of 150 rows read"); at or above
it the banner is neutral. Nothing is blocked, because a perfect score says
nothing about the SIGN of what was read — `all-positive` scores 100 %
while every credit imports as an expense — and the preview step (#329) is
the real net, traversed at every import.
Detection now also fires on its own, guarded on `!existing`: a source that
has never been configured. A source that HAS one is never re-detected, the
stored format wins. The condition is deliberately not `!restored` — a
stored format that fails to decode already reports its own error, and
detecting over it would replace that message with a silent guess.
The button and the automatic run share one `detectFormatForFile`, so the
button replays detection instead of running a second, drifting variant of
it. The score is cleared when its source changes, when the format is
edited by hand (compared through the codec, so a rename keeps it) and when
a template overwrites the format — a banner vouching for a configuration
nobody measured is the misinformation it exists to remove.
Finally, the sign-convention selector is hidden in debit/credit mode.
`mapRow` computes `credit - debit` on magnitudes there and never reads
`signConvention`, so the control changed nothing. Hidden, not reset: the
stored value is left untouched.
Tests: 1054 vitest (+20), build and cargo check green. No DB migration.
CHANGELOG and docs are centralized in link 10 of the stack per the plan.
Resolves#328
Detection reasoned on the shape of the data alone, so it could not tell a debit
column from a credit one: a file laid out `Date;Description;Credit;Debit` was
mapped by position and every sign of the import came out inverted — silently,
since the total is merely negated and no aggregate check notices.
A new `headerDictionary.ts` carries the FR/EN transaction dictionary (date,
description, amount, debit, credit, balance) plus the two matching helpers,
moved out of `csvAutoDetect.ts` whose holdings tables stay untouched: `montant`
is an exclusion token there and the primary amount keyword here, so the two
tables cannot be merged. Moving the generic helpers rather than exporting them
keeps the dependency one-way.
`csvAutoDetect` puts that layer in front of the shape heuristics. Labels resolve
the debit/credit order, the date, description and single-amount columns, and
give `detectHeader` a second signal for a header row carrying a bare number.
Every hint is a preference the data can veto — a labelled date column must still
parse, a labelled balance column is never excluded if it would leave nothing to
map — and a mute file (no header row, unknown labels) falls back to the shape
heuristics unchanged.
Files pairing unsigned amounts with an adjacent D/C indicator column are now
detected and REFUSED with a dedicated message, instead of being configured as
`positive_expense` and importing every deposit as an expense. Detection reports
that reason through `detectImportFormat`; `autoDetectConfig` keeps its previous
shape for the callers that only need the configuration.
Resolves#327
Three ways an imported amount could be silently wrong, all of them passing
validation, all of them fixed here.
The two-column rule was `isNaN(credit) ? -debit : credit`, so the credit always
won. Many banks write `0,00` in the unused column rather than leaving it empty,
and `isNaN(0)` is false -- every debit of such a file imported as 0,00 and the
expense simply vanished, with no error anywhere. The rule is `credit - debit` on
magnitudes now, which needs no special case for a `0,00` cell (zero is the
identity of the subtraction) and implements the documented convention even when
an export negates its debits. A row unreadable in BOTH columns is an error
instead of a free 0,00 transaction.
`parseFrenchAmount` ended on `parseFloat`, which returns the longest valid
PREFIX instead of rejecting. Measured before the fix: `"50,00-"` -> 5000,
`"1 234,56 CR"` -> 123456, `"100,00 CAD"` -> 10000. A factor-100 error, and it
passes `isNaN`, so those rows counted as VALID everywhere downstream -- which
would have defeated the signed preview (#329), the safety net of the whole
chantier. Validation is anchored over the whole normalized string now and any
residual character yields NaN.
NaN, not a rescued magnitude, for a trailing `CR`/`DB` or currency code. Two
reasons: `CR`/`DB` carry a DIRECTION, so returning a magnitude for both would
trade a loud failure for a silent SIGN error (the D/C-indicator shape is refused
upstream by design, #328); and `"100,00 CAD"` is structurally identical to
`"2025 Montant"`, so whitelisting a trailing word to rescue the first re-blinds
`detectHeader` on the second. Two accounting forms ARE legitimate and supported:
parentheses `(50,00)` and a trailing sign `50,00-`.
The `?? 0` fallbacks read column 0 -- usually the date -- when the mapping was
incomplete. An unmapped amount column is an explicit row error now, reported
ahead of any per-row problem since it is a format error affecting every row.
`1.234` is 1234 in a French column and 1.234 in an English one, and no rule
applied to that cell ALONE can tell. `detectDecimalSeparator` arbitrates from
the decisive siblings of the column and `parseFrenchAmount` takes the verdict as
an option. Detection deliberately stays out of it: it runs before a column is
known to be an amount column at all, so the verdict is applied where the value
actually becomes a transaction.
The rule itself moves out of the hook as a pure `mapRow(raw, format)` in
`importFormat.ts`. That is what lets the corpus tests run the REAL rule -- the
hand-written mirror in `csvAutoDetect.test.ts` and the static guard pinning five
`parseFilesInternal` expressions are both deleted -- and what stops the
detection score (#328) and the signed preview (#329) each re-implementing it.
Hardening is global: the parser is shared by 11 call sites, 8 in
`csvAutoDetect.ts` and 3 in the holdings CSV import (#245), where a price cell
`150,25 CAD` used to store 15025. It is refused now and `buildDetailedLines`
raises on the empty price. An unreadable QUANTITY was worse -- coerced to 0, so a
zero-value position saved in silence; the draft keeps the offending text instead
and the existing `snapshot_priced_quantity_required` fires.
Row errors become i18n keys (`import.rowErrors.*`) rather than the raw English
literals rendered straight into the preview table, since this adds a
user-visible string. The report table also carries raw exception messages, so
both render sites resolve through `isRowErrorKey` and never feed `t()` anything
that is not ours.
Test churn, per link 1's handoff (update the expectation, drop the marker, never
delete the test): the three `#325` KNOWN DEFECT blocks in `amountParser.test.ts`
flip, plus `unused-column-zero` in `csvAutoDetect.test.ts`. One block tagged
`#328` flips too -- `header-numeric-label`, whose own comment reads "#328 adds a
lexical signal to detectHeader, and #325 anchors the parser [...] either fix
closes this". The anchored parser landed first. The other `#328` blocks
(`debit-credit-reversed`, `absolute-indicator`) and the `#329` block are
verified unchanged. CHANGELOG and docs stay centralized in the last link of the
stack, as the plan specifies.
989 vitest (963 before), tsc + vite build clean, cargo check clean. No DB
migration.
Resolves#325
The root bug of this chantier. `import_sources` carried no `amount_mode` and no
`sign_convention` until v17, so restoring a configured source re-inferred the
mode from `mapping.debitAmount !== undefined` and wrote
`signConvention: "negative_expense"` outright (useImportWizard.ts:321-323). A
credit-card statement configured for positive expenses came back on the default
convention at its second import and `parseFilesInternal` negated every amount:
expenses landed as income, with no error shown anywhere. The format is a read
value now, not a guessed one.
Two types and a codec, not one composed type. The four carriers are
structurally incompatible -- `ImportSource.has_header` is declared boolean,
`ImportConfigTemplate.has_header` is a number, `SourceConfig` is camelCase on a
parsed mapping -- so the guarantee cannot come from a shared shape. It comes
from `src/utils/importFormat.ts` being the single conversion point between
`ImportFormatRow` (persisted: snake_case, mapping as JSON, `has_header`
normalized to 0/1) and `ImportFormat` (domain), and from its completeness test.
That test is enforced on two levels, and both were mutation-checked:
`FORMAT_FIELD_PAIRS` is typed `Record<keyof ImportFormat, keyof
ImportFormatRow>`, so a field added to the format fails to BUILD until it is
listed; the test then compares each codec's real output keys against that table,
so a field listed but not wired fails the TEST. Dropping `sign_convention` from
`formatToRow` -- the shape of the original bug -- fails 14 tests.
`formatFromRow` validates rather than falls back. The v17 CHECK admits
`absolute_indicator` so the third amount mode ships without another migration,
but the app cannot map one: falling through to the `single` branch would read
the wrong column for every row, and anything other than `positive_expense`
would silently mean `negative_expense`. It raises an `ImportFormatError`
carrying an i18n key, and the wizard opens on a fresh configuration so
"reconfigure this source" stays an action the user can actually take.
Also here:
- The config write moves from `checkDuplicatesInternal` to `executeImport`, so
an import abandoned at the duplicate step leaves no configuration behind. It
is the only write point in the hook and a guard test holds that.
- Switching amount mode prunes the abandoned mode's columns, so the mode owns
the mapping rather than the reverse. The column the `<select>` merely displays
is deliberately not materialized -- #325 turns an unmapped amount column into
an explicit row error, and writing a 0 here would make it unreachable. The
mode and the pruned mapping land in ONE state update: the panel's handlers
each spread the same `config` prop, so two calls would see the same stale
value.
- `template_id` is recorded and restored as provenance only, never re-read as
format. `selectedTemplateId` is no longer blanked on every source selection.
An acceptance test rewrites a template end to end and asserts the linked
source reads identically, plus a non-vacuity check that the template really
changed.
- Both template writers go through the codec too, so a new format field cannot
reach one table and miss the other.
`parseFilesInternal` is untouched: link 1's static guard on its five pinned
expressions still passes. 39 new tests (963 vitest total, was 924), build clean,
`cargo check` clean, no migration.
Resolves#324
First link of the import-format stack. `autoDetectConfig`, `detectAmountMode`,
`detectSingleAmount`, `preprocessQuotedCSV` and `parseFrenchAmount` had no test
at all on a suite of 871, while carrying every imported amount. This adds the
reference the rewrite (#323-#332) measures itself against, before any of it
moves.
Corpus — 11 synthetic files under `src/__fixtures__/csv/`, no real statement
data, covering shapes a single real statement never contains at once: signed
amount, debit/credit, debit/credit in reversed column order, unused column
filled with `0,00`, preamble before the header, header carrying a number,
header cell starting with digits, no header row, all-positive amounts,
whole-line-quoted (Desjardins style), absolute amount + D/C indicator.
Every expectation was derived by running the code, not by reading it. Four
cases are frozen as DEFECTIVE, each named `KNOWN DEFECT` with the right answer
in a comment and the issue that owes the fix:
- reversed debit/credit — the pair is assigned by column position, never by
label, so every sign is inverted while the total merely negates (#328)
- unused column at `0,00` — the rule branches on `isNaN(credit)` and `"0,00"`
parses to 0, so every debit imports as zero (#325)
- header cell starting with digits — `parseFloat` returns the numeric prefix,
`detectHeader` reads the header as data (#328/#325)
- absolute amount + D/C indicator — the indicator column is ignored entirely
and every credit imports as an expense (#328)
`parseFrenchAmount` is pinned form by form, including the prefix-scan defect
the spec flagged: `"100,00 CAD"` yields 10000 and passes `isNaN`. Per the
/review-spec revision, the assertions reach the holdings call sites too, which
surfaced the same x100 leak in the #245 holdings import — a price cell of
`"150,25 CAD"` stores the position at 15025.
The end-to-end tests replay the wizard's row-mapping rule from a mirror, since
it still lives inside a `useCallback` and the repo has no jsdom. A guard test
asserts the production expressions are still literally present, so the mirror
cannot drift; #325 extracts `mapRow` and must then delete it.
No production file is modified. 924 vitest green (was 871), build clean.
Resolves#326
"tout la Gratuite/Base" -> "tout de la Gratuite/Base", flagged as the
one user-facing correction in the /pr-review pass on PR #308.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Document the edition-gating work (#297-#301):
- ADR 0017 (accepted): tier->features matrix, signed features[] override
fail-closed in Free (CWE-863), UI-only enforcement as an assumed GPL
soft-paywall (server-enforced price fetching stays the only hard gate),
non-destructive downgrade, dev-override behind an explicit Cargo
feature (CWE-489), rejected alternatives.
- architecture.md: new 'Gating par edition' section (entitlements matrix,
LicenseContext, useEntitlement, RequireFeature/UpsellGate, NavLock,
profileGate, Rust side), rewritten entitlements.rs section (auto-update
now Base+, stale 'open to free' note removed), gated routes listed in
the routing section, hooks table updated (useLicense removed in #297 ->
useEntitlement/useIsPremium), ADR index + header refreshed.
- guide-utilisateur.md + docs.editions.* i18n keys (FR/EN) wired into
DocsContent: new 'Editions' section with the Free/Base/Premium table,
unlock flow and non-destructive locking tips.
- CHANGELOG.md + CHANGELOG.fr.md: one global [Unreleased] entry listing
the modules now gated Base (Budget, Adjustments, advanced reports,
multi-profile, auto-update) and Premium (Balance), the visible-but-
locked upsell with disabled 'coming soon' purchase CTA, and the
data-preserving behaviour.
Resolves#302
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A Free user keeps full access to their active profile; profiles beyond
it show a lock in ProfileSwitcher and open an upsell dialog instead of
switching. Creating a profile beyond the first is locked at the single
creation point, ProfileFormModal (reached from both ProfileSwitcher and
ProfileSelectionPage), with a race guard in handleSave covering the
license boot window. Both creation entries stay visible with a lock
(locked-not-hidden). Nothing is ever removed from profiles.json — an
upgrade to Base/Premium makes every profile reappear untouched.
- New pure predicates in src/shared/profileGate.ts
(isProfileSwitchLocked, isProfileCreationLocked) + 10 vitest
- ProfileFormModal upsell panel reuses upsell.* keys WITHOUT UpsellGate:
the modal also opens from ProfileSelectionPage, which renders outside
BrowserRouter, where UpsellGate's useNavigate would throw
- UpsellGate gains an optional onNavigate callback so the
ProfileSwitcher upsell dialog can close itself after navigation
- No lock while the license is loading (anti-flash, ready guard);
zero new i18n keys; no DB migration
Resolves#300
Generated autonomously by /autopilot run of 2026-07-20
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Apply the tier gating to routes and navigation on top of the #298 UI
guard:
- App.tsx: pathless RequireFeature layout-routes grouping /balance,
/balance/accounts, /balance/snapshot under "balance"; /reports/
highlights|compare|category|cartes under "reports-advanced"; /budget
under "budget"; /adjustments under "adjustments". The /reports hub and
/reports/trends stay Free and ungated.
- NavItem gains an optional `feature?: FeatureKey`; set in NAV_ITEMS on
budget, adjustments and balance only — NOT on reports (Free hub).
- Sidebar: local NavLock child component (hook at component top level)
renders a lock badge only when the license is ready AND the feature is
not allowed — no locked flash at boot; items stay clickable and lead
to the upsell via the gated route. Tooltip/aria reuse nav.locked.
- ReportsPage hub: single useEntitlement("reports-advanced") call
drives a `locked` badge on the 4 advanced tiles via a new additive
HubReportNavCard `locked?` prop; the Trends tile is never locked.
- Pure contract test on NAV_ITEMS (gated trio present, reports/Free
items ungated, exactly 3 of 9 gated).
No new i18n strings (nav.locked shipped with #298), no DB migration.
Changelog centralized in #302.
Resolves#299
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Add the reusable gating guard components on top of the #297 foundation:
- UpsellGate: full locked screen (lock icon, tier title, per-feature
description). Two CTAs: "Get <tier>" rendered VISIBLE but DISABLED with
an "online purchase coming soon" note (per planning decision — #270 will
activate it), and "I already have a key" navigating to /settings/users
(LicenseCard).
- RequireFeature: renders a neutral loader while the license is not ready
(no upsell flash at boot), then children or UpsellGate. Renders <Outlet/>
when children are omitted so it also works as a layout route grouping
all routes of one feature.
- requiredTierFor() pure helper in shared/entitlements.ts derives the
minimum unlocking tier from matrix membership (not array order).
- i18n: upsell.* (title, per-feature descriptions, CTAs) + nav.locked in
BOTH locales; tier labels reuse the existing license.editions.* keys.
- Tests: requiredTierFor mapping/minimality + upsell i18n coverage for
every FeatureKey in fr and en (the components themselves are not
testable without jsdom).
Resolves#298
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A rejected submitKey dispatched the same ERROR action as a failed boot
load, so the CWE-703 retry backoff armed on it and the auto refresh
(LOAD_START) cleared the "invalid key" message ~1s after submit —
LicenseCard has no local error state, the context is the only source.
Split the state: load lifecycle (status/error, retried) vs validation
(validating/validationError, never retried). VALIDATE_ERROR leaves
status untouched, so a ready license stays ready on a typo'd key (no
`ready` regression for gating consumers) and a boot-error retry loop
keeps running through a failed validation. Reducer + initial state
exported for tests, covered by LicenseContext.test.ts.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Custom categories with no standard match were shown as read-only text in
the migration wizard and force-parented under a catch-all bucket. They now
get the same inline target picker as seeded rows: picking a standard leaf
merges the custom category — its transactions, budgets, keywords and
suppliers are reassigned to the leaf — and deactivates it. Leaving a custom
unmapped keeps the previous behaviour and never blocks the wizard.
- Reducer: RESOLVE_ROW resolves rows in both plan.rows and plan.preserved;
the Next-button guard still counts seeded rows only.
- Writer: the rewrite mapping now includes resolved preserved rows; the
catch-all parent is created only when a custom is left unmerged; merged
customs are deactivated instead of re-parented (shared isResolvedTarget
helper across the three sites).
- UI: the preserved block renders MappingRow instead of plain text.
- i18n (FR/EN) + CHANGELOG (FR/EN).
Tests: reducer (resolve a preserved custom, guard unchanged, GO_NEXT still
proceeds) + writer (reassign to the chosen leaf, soft-delete, no empty
parent when all merged, orphan-free merge of a custom parent with an
unresolved child). 836 vitest green, tsc + vite build clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Document the final multi-level category collapse behavior (shipped in
#288/#289/#290) and the structural decision to persist per-profile UI
state in user_preferences rather than localStorage.
- ADR 0016 (accepted): profile-specific UI state (category collapse) lives
in the profile's own SQLite user_preferences table, not localStorage.
deleteProfile drops the .db but purges no localStorage, so a per-profile
localStorage key would be a residue surviving profile deletion, leaking
which categories a (possibly PIN-protected) profile explored. States the
boundary: profile-specific -> user_preferences; machine-global (theme,
subtotals position, Cartes period mode) -> localStorage.
- guide-utilisateur.md sections 8 (Budget) + 9 (Reports): multi-level
collapse, collapsed-by-default, "Expand all / Collapse all" button,
memory per profile.
- docs.* i18n keys (fr + en): mirror the guide additions in the in-app help
page (docs.budget + docs.reports features/tips).
- architecture.md: user_preferences line now names the 4 collapse keys,
useCollapsibleGroups cross-cutting hook note, ADR index row.
No DB migration, no behavior change (docs only). Build clean, 828 vitest.
Resolves#291
Generated autonomously by /autopilot run of 2026-07-15
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the two hand-rolled Set-of-ids collapse state machines in the category
trees with the shared useCollapsibleGroups hook (a strict superset after #288),
keeping each tree's distinct recursive render and CategoryTree's drag-and-drop
untouched.
- CategoryTree (Categories page): storageKey null + defaultExpanded true, so
every parent opens with no seeding; drops the local Set + collectExpandable.
- CategoryTaxonomyTree + guide page: storageKey null + defaultExpanded false
(collapsed by default); exports a shared TAXONOMY_COLLAPSE_ACCESSORS.
- Fix the guide's button bug: allExpanded = expanded.size > 0 flipped to
"Collapse all" after opening a single node; now uses the hook's correct
allExpanded (every group must be open).
- Also migrate StepDiscover (4th consumer of CategoryTaxonomyTree, same button
bug) onto the hook for a green build and consistency.
Both trees pass a flattened node list to the bulk ops; behaviour preserved:
Categories opens expanded, the guide/wizard open collapsed.
Resolves#290
Generated autonomously by /autopilot run of 2026-07-15
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Reverse #278's deliberate no-collapse decision for the budget grid: it now
folds/unfolds at every category level like the hierarchical reports, opening
fully collapsed with a one-click "Expand all".
- BudgetTable: wire useCollapsibleGroups (defaultExpanded: false), inline
BUDGET_COLLAPSE_ACCESSORS + BUDGET_EXPANDED_KEY (mirrors ComparePeriodTable /
BudgetVsActualTable — no budgetTableModel.ts). Chevron + aria-expanded +
aria-level on every parent row; groups.visible(group) before reorderRows.
- BudgetTable: section subtotal now uses the tested sumLeavesForType on the RAW
group (drop-in for the hand-rolled loop) so folding stays purely visual.
- BudgetTable: rename STORAGE_KEY to "budget-subtotals-position", decoupling the
subtotals-position preference from BudgetVsActualTable (they collided).
- useBudget: extract the pure buildBudgetYearRows(); the grid's rows are
level-order (BFS), not DFS — document the invariant and pin it in a test, since
the #288 ancestor-walk collapse is order-independent (the v1 plan assumed DFS
and would have broken here).
- Tests: useBudget.test.ts locks the level-order emission, the DFS-killer, the
end-to-end multi-level collapse on real builder output, and that subtotals sum
raw rows regardless of collapse. 828 vitest pass.
Resolves#289
Generated autonomously by /autopilot run of 2026-07-15
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Generalize the report category collapse from level-1-only to every hierarchy
level, on the three hierarchical report tables (real-vs-real Compare,
real-vs-budget Compare, Trends by category). Visibility is now decided by an
ancestor walk, not by row adjacency, so it is independent of row order (the
level-ordered budget grid emits a non-DFS order).
- collapsibleRows: rewrite visibleRows as an ancestor walk (a row is hidden iff
any ancestor is collapsed); add parentKeyOf + injective `p:` keys;
collapsibleKeys returns all parents (any depth); extract the pure, tested
isCollapsedFor polarity helper; MAX_TREE_DEPTH cycle guard.
- useCollapsibleGroups: persist in user_preferences (per-profile, destroyed with
the profile) instead of localStorage; storageKey nullable (no persistence);
options.defaultExpanded; async hydration (no flash); collapseAll(rows).
- 3 tables: fix BOTH gates (collapsed flag + button) isTopParent -> isParent, add
parentKeyOf accessors, aria-level on parent rows.
- Delete dead CategoryTable.tsx (0 imports).
- Tests: rewrite collapsibleRows.test.ts (BFS==DFS masking, cycle guard,
cross-section ancestor, "(direct)" leaf, polarity); extend overTimeTableModel
fixture to 3 levels with cascade assertions.
Collapse stays purely visual: subtotals and result lines are computed from raw
rows, never from visible rows.
Resolves#288
Generated autonomously by /autopilot run of 2026-07-15
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
getCartesSnapshot forwarded the account filter only to the top-movers and
budget sub-reports, leaving fetchMonthlyFlows (KPIs, sparklines, 12-month
overlay) and fetchSeasonality unfiltered. The Dashboard is the first page to
expose the account filter over these series, so its KPI cards showed
unfiltered totals while the category bars / trend / top-movers respected the
filter — figures that did not reconcile on the same screen.
Thread an optional accountIds through both fetchers (parameterized
`source_id IN (...)` via inPlaceholders) and pass it from getCartesSnapshot,
so the whole dashboard honours the filter — which is what the #279 CHANGELOG
entry already promises. No CHANGELOG change: the code now matches the note.
Resolves#279
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
DashboardPage now reuses the Cartes report's presentational widgets
(KpiCard with MoM/YoY deltas, top movers, budget adherence) sourced from
getCartesSnapshot against a Dashboard-owned reference month (defaults to
the last complete month). The expense-only pie chart is replaced by a
ranked bar chart of top expense categories (resurrecting the previously
unused CategoryBarChart), and the account (import-source) filter
introduced on Trends/Compare/Budget now also applies to the Dashboard's
own transactional widgets via a Dashboard-local accountIds state (kept
separate from useReportsPeriod, since the Dashboard owns its own two
temporal axes).
A new net-worth tile surfaces the Balance sheet's latest total
(getSnapshotTotalsByDate) — a distinct metric from every transactional
card here, so it stays hidden (never a misleading "$0") until at least
one balance account has a recorded snapshot, reusing deriveLandingState
rather than re-inferring emptiness from nulls. It is not scoped by the
account filter (balance_accounts is a disjoint concept from
import_sources) and is fetched independently on mount.
The category-over-time trend chart now passes typeFilter "expense",
fixing a latent mismatch where a revenue category could silently show
up in a chart titled "expenses over time".
Resolves#279
Align BudgetVsActualTable on the income-statement gold standard already
shipped for the real-vs-real compare report (#253/#256): flip TYPE_ORDER
to income-first (revenue -> expense -> transfer) and replace the flat
"Total" row with interleaved Result before transfers / Net result lines.
Roll-up logic extracted into a new pure, tested module
(budgetVsActualResults.ts) mirroring compareResults.ts, with one sign
difference: BVA amounts already arrive signed the accounting way
(expense actual/budget are negative), so the operating result is a
straight add of income + expense rather than a subtraction.
Resolves#277
Generated autonomously by /autopilot run of 2026-07-11
Align BudgetTable on the income-statement standard shipped for the
compare/trend reports: sections now read Income -> Expense -> Transfer
(useBudget's TYPE_ORDER and BudgetTable's typeOrder both flipped), and
the previously unlabeled grand-total row is replaced by two interleaved
result rows (Result before transfers / Net result), computed by a new
pure, tested module (budgetTableResults.ts) mirroring compareResults.ts's
shape. Roll-up covers the previous-year-actual, budgeted-annual, and
budgeted-monthly columns alike.
All categories remain displayed (the grid stays a full edit surface) --
no collapse, no empty-row toggle, matching this issue's frozen decisions.
Resolves#278
Mounts <FilterPanel> on ReportsComparePage (next to the existing
CompareReferenceMonthPicker) and BudgetPage (next to the existing
YearNavigator), keeping each page's temporal control unchanged. Threads
accountIds from useReportsPeriod through useCompare into
getCompareMonthOverMonth/getCompareYearOverYear, and through
CompareBudgetView into getBudgetVsActualData — closing the gap where
that sub-tab silently ignored the filter while the rest of the Compare
page respected it.
Budget: the issue named getBudgetVsActualData as the target for
useBudget, but that hook never calls it (it's exclusive to
CompareBudgetView/Dashboard/Cartes) — its only real actuals fetch is
getActualTotalsForYear(year - 1), the previous-year reference column.
Extended that function with the same optional accountIds pass-through
established by #273, rather than wiring in an unrelated call shape.
Both hooks fetch their account checkbox list via getAllImportSources,
mirroring the useTrends/#275 pattern. Empty selection = no filter,
byte-identical to pre-#276 output (regression-tested).
Resolves#276
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Mounts <FilterPanel> (from #274) on ReportsTrendsPage next to the period
selector, and threads accountIds from useReportsPeriod through useTrends
into getMonthlyTrends/getCategoryOverTime (both already accountIds-aware
since #273). The account checkbox list reuses the same getAllImportSources
query as the Transactions single-source filter. Empty selection = no
filter, byte-identical to pre-#275 output.
Resolves#275
Adds src/components/reports/FilterPanel.tsx, the shared filter bar for the
report pages. It renders the page's own temporal control (PeriodSelector /
CompareReferenceMonthPicker / YearNavigator / …) as-is via a `temporalControl`
ReactNode prop, plus a checkbox multi-select over `accounts: ImportSource[]`.
No `temporalMode` enum — that design was rejected to avoid mixing state
sources; the panel owns no temporal state at all.
Empty `accountIds` means no filter (all sources). Selection toggling is a
pure, exported `toggleAccountId` helper. Copy says "sources"/"import
sources" everywhere, never "comptes"/"accounts", so the filter reads as
distinct from the Bilan module's own "Compte" vocabulary.
This issue creates the standalone component only; wiring it into report
pages is left to follow-up issues (#275/#276).
Resolves#274
The constant became the compareDeltaSql() function in the previous
commit (Issue #273); three comments still named the old constant.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Give every report service an optional accountIds?: number[] filter that
narrows results to a subset of import sources (transactions.source_id),
matched via a parameterized IN (...) clause — one bound placeholder per
id, never a joined string (CWE-89). New shared inPlaceholders() helper
(src/utils/sqlFilters.ts) generates the placeholder list so the pattern
lives in exactly one place.
New accountIds param: getCompareMonthOverMonth, getCompareYearOverYear,
getBudgetVsActualData, getCartesSnapshot (forwarded to its
getCompareMonthOverMonth + getBudgetVsActualData sub-calls, so the
Cartes dashboard's top-movers and budget-adherence cards respect an
active filter instead of silently ignoring it).
Signature change scalar -> plural: getMonthlyTrends, getCategoryOverTime,
getExpensesByCategory (sourceId?: number -> accountIds?: number[]). No
scalar production caller of these three passed sourceId today, so only
the test call-sites needed updating to the array shape.
Omitted/empty accountIds adds no clause at all, so every service stays
byte-identical to its pre-#273 SQL/results — pinned by a regression test
per service (new budgetService.test.ts / dashboardService.test.ts files,
extended reportService.test.ts / reportService.cartes.test.ts).
This is backend plumbing only: no report page exposes an account filter
control yet (follow-up issues #274-#276 add the shared <FilterPanel> and
wire it into Trends/Compare/Budget).
Resolves#273
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add accountIds: number[] to the shared useReportsPeriod hook, URL-backed
via a dedicated `sources` query param (comma-separated, bookmarkable like
the existing period/from/to). Purely additive: the hook's existing shape
is unchanged, so all 6 consumers (useTrends, useCompare, useCategoryZoom,
ReportsCategoryPage, ReportsPage, ReportsComparePage) keep compiling as-is.
Also adds the shared ReportFilters { period, accountIds } type for
follow-up issues to consume, and exports pure parseAccountIds/
serializeAccountIds helpers (same hookless-testability pattern as
resolveReportsPeriod). URL parsing validates each token as a finite
integer via a strict regex + Number.isSafeInteger, dropping invalid
tokens individually rather than discarding the whole selection.
This is foundation only: no UI control exposes the filter yet, and no
service reads accountIds yet (both land in follow-up issues of the
"rapports uniformes" epic).
Resolves#272
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Renders the trends-by-category table from the #264 id-keyed tree as an indented
parent/child hierarchy (parity with ComparePeriodTable): top-level parents carry
a chevron and fold their subtree down to a subtotal, with an "expand all /
collapse all" button. Collapsed by default; expansion state is persisted under a
trends-specific localStorage key (reports-trends-expanded), distinct from the
comparable tables. The "result before transfers" line is interleaved before the
transfers section instead of sitting at the bottom.
Collapse is purely visual: section subtotals, before/net results and totals come
from computeOverTimeResults over the raw tree, never the visible rows.
New pure module overTimeTableModel.ts (section grouping + collapse accessors)
keeps the component thin and unit-testable without a React render harness;
overTimeResults.ts is left untouched. 8 new tests; build + 728 vitest green.
Resolves#265
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds an id-keyed hierarchical `tree: OverTimeRow[]` sidecar to
getCategoryOverTime, built via the generic buildLeafDrivenTree (#263) as
buildOverTimeTree — leaf-driven, income-first, no top-N / no "Other" bucket.
The name-keyed pivot (data/categories/colors/types/categoryIds) is left
BYTE-IDENTICAL, so the Trends chart and the dashboard render unchanged.
computeOverTimeResults now consumes the id-keyed tree LEAVES (never the
`is_parent` subtotals) grouped by each leaf's category_id-resolved type, so
the section subtotals and the Result-before/net lines are exact: two homonym
categories of different types no longer collide, and a non-top-N income or
transfer category is no longer lumped into "Other" as an expense.
CategoryOverTimeTable renders the tree leaves (every category, id-safe)
instead of the top-N pivot; the chart keeps reading the top-N-capped pivot so
its stacked series stay bounded. OverTimeRow mirrors CategoryDelta's snake_case
hierarchy block (parent_id/is_parent/depth/category_type) so it composes with
collapsibleRows / useCollapsibleGroups for the #265 hierarchy work.
Tests: rewrote overTimeResults.test.ts onto the tree (homonym regression +
leaves-only-not-parents); added buildOverTimeTree suite (nesting, income-first,
grand-total invariance, orphan, A->B->A cycle depth guard) and a
getCategoryOverTime tree-wiring test proving the pivot stays top-N+Other while
the tree carries every category. 720 vitest green, build + tsc green.
Resolves#264
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Factor the hand-rolled hierarchy walk out of buildCompareTree into a
generic buildLeafDrivenTree<T>(leaves, opts): the shared skeleton (leaf/
orphan split, ancestor "relevant" set, DB-order adjacency, buildNode
recursion with the MAX_TREE_DEPTH guard + loop-break, magnitude sibling
ordering, orphan append, contiguous income->expense->transfer section
sort) now lives in one place, parameterised by injection points
(categoryIdOf, makeLeaf, makeSubtotal, decorateDirectLeaf, makeOrphan,
sortKey, isSubtotal, sectionOf, sectionOrder, maxDepth).
buildCompareTree becomes a thin specialisation supplying only the delta
arithmetic; output is byte-identical (compare tests unchanged as the
non-regression guard). Exported (with TreeCatMeta / TreeSectionType /
LeafDrivenTreeOptions) so the trends tree reuses the exact same skeleton.
Behavior-preserving refactor: no user-facing change, no CHANGELOG entry,
no DB migration.
Resolves#263
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The Trends report previously opened on the global monthly chart, which
has no income section — so a user's revenue (e.g. their pay) was hidden
until they manually switched to the "By category" table view.
Open the report on byCategory + table by default so the income section
(and therefore payroll) is visible immediately. Two scoped call-site
changes:
- useTrends initial subView: "global" -> "byCategory"
- ReportsTrendsPage viewMode fallback: readViewMode(key, "table")
Both defaults are scoped to Trends. The real-vs-real compare report
(distinct "reports-viewmode-compare" storage key) is unaffected, and a
user's own persisted view choice still wins over the new fallback.
Resolves#262
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The two hierarchical comparable reports — real-vs-real Compare and
real-vs-budget — now let the user collapse or expand each top-level
category's sub-categories. Groups start collapsed by default (#260): only
the parent's subtotal row shows until the user expands it, and each expanded
group is remembered per report.
- New pure module utils/collapsibleRows.ts (visibility filter, group-key
extraction, expanded-set (de)serialization) + useCollapsibleGroups hook
wrapping the localStorage-persisted expanded set. Persisting the *expanded*
set (not the collapsed one) makes "all collapsed" the zero/default state.
- A chevron toggles each top-level parent; an "Expand all / Collapse all"
button toggles them together. Section subtotals and result lines keep
summing every leaf, so collapsing never changes any total.
- Accessors mirror each table's own depth/parent logic so hidden rows are
exactly a group's indented descendants.
- i18n keys reports.collapse.{expandAll,collapseAll} (FR/EN); CHANGELOG.
CategoryOverTimeTable (Trends -> by category) is intentionally left out: its
rows are a flat top-N category list with no parent/child hierarchy to fold.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
getCategoryOverTime's monthly pivot selected COALESCE(c.type,'expense') AS
category_type but nothing read it — the `types` map is built from the top-N
query. Remove the unused projection and its row-type field. No behaviour change.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Apply the compare report's "result analysis" presentation to Trends -> By
category (table view). Categories are grouped into Income -> Expenses ->
Transfers sections with per-month and total subtotals; the old mixed "Total"
row (which summed absolute magnitudes into a meaningless figure) is replaced by
a Net result row per month (income - expenses + transfers) plus a Result before
transfers row shown only when transfers exist, both coloured by sign. Category
cells stay neutral levels.
- reportService.getCategoryOverTime: project COALESCE(c.type,'expense') in both
SELECTs and return an additive `types` map (built from the top-N rows, like
`colors`). Dashboard widget and CategoryOverTimeChart are untouched.
- shared types: add `types` to CategoryOverTimeData.
- new pure module overTimeResults.ts (computeOverTimeResults) with unit tests -
the per-month result reducer, mirroring the compare's pure Totals helper.
- i18n: add reports.compare.resultNet / resultBeforeTransfers (FR+EN); reuse the
existing sections.* and total* keys.
- .gitignore: anchor the autopilot `reports/` rule to `/reports/` so it no
longer swallows new files under src/components/reports/.
Section order is income-first (Revenus -> Depenses -> Transferts), per the
issue; the existing COMPARE_TYPE_ORDER is expense-first, so a local
OVER_TIME_TYPE_ORDER is used rather than reusing it.
Resolves#256
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
/pr-review on #255 caught a cross-consumer regression: broadening
COMPARE_DELTA_SQL to surface income (this PR) also feeds getCartesSnapshot,
whose "top movers" card is a spending view (up = red). A salary rise would land
under "biggest increases" coloured red — inverted meaning.
Filter significantMovers to expense leaves ((category_type ?? "expense") ===
"expense"), mirroring ComparePeriodChart. The surviving expense output is
byte-identical to pre-#253. Add a Cartes regression test (an income category
with the biggest delta must not top the movers list) — the existing test mocked
the compare SQL without category types, so it stayed silently green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Addresses /pr-review REQUEST_CHANGES on #255.
- useCompare: the "skip first sync" boolean was not StrictMode-safe — the dev
double-invoke of effects flipped the flag on setup #1, so setup #2 re-synced
the reference month to the civil-year December, re-introducing the very bug
Changement 2 fixes (dev only; prod has no double-invoke). Replace it with a
value-change guard: a ref seeded with the initial `to` plus a pure
syncReferenceOnPeriodChange() that only dispatches when `to` actually changes.
Idempotent across the double-invoke, and now unit-tested (5 cases) since the
decision is a pure function (the project has no renderHook harness).
- Remove the now-orphaned reports.compare.totalRow i18n key (both locales) — the
flat grand total it labelled was replaced by the result lines.
- ComparePeriodTable: gate the "before transfers" line on results.hasTransfers
(previously computed/tested but unused).
- ComparePeriodChart: show the no-data empty state when the expense filter
leaves nothing (a pure income/transfer period) instead of bare axes.
Build + 690 vitest green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The "actual vs actual" compare was expenses-only (WHERE amount < 0), so revenue
categories were invisible — you could not tell a surplus from a deficit. Make it
an income statement (the product's namesake "résultat"):
- COMPARE_DELTA_SQL: broaden the WHERE and per-type CASE to include `income` as
a signed SUM (like `transfer`), so revenue credits survive the outflow filter.
Expenses stay ABS-of-outflows, byte-identical. Transfer netting (#243) intact.
- Section order is now income -> expense -> transfer (COMPARE_TYPE_ORDER).
- ComparePeriodTable: replace the now-meaningless flat grand total with two
result lines — "result before transfers" (revenues - expenses) and the net
"result" (after transfers) — extracted into a pure, tested compareResults
module. Delta colours are direction-aware (income/result up = green, spending
up = red); result amounts are coloured by sign (surplus/deficit).
- ComparePeriodChart stays a spending view: filter to expense leaves so revenue
bars don't mix into the same axis.
- useCompare: skip the initial period-sync so the compare opens on the previous
(last complete) month instead of the civil-year December that useReportsPeriod
yields by default.
Also anchor the `reports/` gitignore to `/reports/` — the unanchored rule was
silently ignoring new files under src/components/reports/.
Tests: new compareResults.test.ts (result roll-up, unbalanced transfer, deficit,
subtotal exclusion); reportService.test.ts updated for the broadened SQL, the
income bucket rule, and income-first section order. Build + 685 vitest green.
Resolves#253
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Address the two non-blocking review notes on PR #252:
- The re-injected non-leaf parent (e.g. #1710) is technically clickable, not
'never offered'; correct the MappingRow comment to state selecting it is a
no-op (only leaves resolve).
- Extract the per-row option-building glue into a pure, unit-tested helper
comboboxCategoriesForTarget (4 cases: null/leaf pass-through by reference,
non-leaf append, unresolvable stale id).
The v1 target combobox in the migration preview now offers only leaf
categories (the inputable end-categories), not intermediate parents, so a
transaction is never mapped to a grouping bucket. Leaves render as a flat,
un-indented list in taxonomy order.
A low-confidence default can still point a v2 category at a non-leaf parent
(e.g. 'Divertissement' #1710); MappingRow re-injects that current target for
its single row via findTaxonomyCategory so the input stays populated, while
new picks remain leaves-only. Adds a findTaxonomyCategory pure helper + tests
(adapter now returns 112 leaves, not the 150-node full tree).
Make the v1 target editable on EVERY row of the categories-migration
preview (StepSimulate), not only the "needs review" ones. Each row now
renders a reused CategoryCombobox (accent-insensitive type-ahead,
keyboard, hierarchical) in place of the read-only text / flat <select>.
A pure adapter (taxonomyToComboboxCategories) flattens the full v1
taxonomy into Category-shaped rows once in StepSimulate and passes it as
a prop to the rows. The full tree (not just leaves) is fed so mid-tree
default targets such as "Divertissement" (id 1710, the default for
"Jeux, Films & Livres") both display and can be re-picked. The target
cell stops click/keydown propagation so picker interaction never toggles
the row preview. Reducer/guard behavior unchanged: resolving a "none"
row still bumps confidence to medium, and the unresolved-count guard
stays functional.
Resolves#246
A detailed (by-security) account can now import its positions from a CSV
instead of adding each security one by one. An "Import CSV" button next to
"Add a title" opens a native picker; the delimiter, encoding and columns
(symbol, quantity, optional price + book_cost) are auto-detected via new
csvAutoDetect helpers and adjustable in a small mapping editor. Rows sharing
a symbol are merged (SUM qty + book_cost, first price) and the batch is merged
into the basket by normalized symbol, so no UNIQUE(snapshot_line_id,
security_id) violation can occur at save. A CSV without a price column imports
quantities with an empty unit_price (fetch/type later) — the existing atomic
save path is unchanged (no SQL/Rust change).
- csvAutoDetect: autoDetectHoldingColumns + analyzeHoldingsCsv (+ tests)
- useSnapshotEditor: holdingsFromCsvRows + IMPORT_HOLDINGS action + importHoldings (+ tests)
- HoldingsCsvImportModal: file pick -> parse -> mapping editor + preview
- i18n FR/EN under balance.snapshot.detailed.importCsv.*
- CHANGELOG (Added / Ajouté)
Resolves#245
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Decouple the /balance empty-state guard into three explicit states via a
new pure helper `deriveLandingState` (empty / accounts-no-snapshot /
populated). Having accounts but no snapshot no longer strands the user on a
snapshot-only onboarding card: each state now renders navigation tiles
(HubReportNavCard, modeled on the Reports hub), and "Manage accounts"
(/balance/accounts) is reachable in every state, including the populated
dashboard.
- Remove BalanceOnboardingCard (component + test) and the dead
balance.onboarding.* i18n keys; add balanceLanding.ts + test.
- Add balance.hub.* and balance.landing.* keys in FR and EN.
- CHANGELOG (EN + FR) under [Unreleased].
Resolves#244
Render the real-vs-real comparable report as a parent/child category tree
with subtotal rows, mirroring the budget report — rows grouped into
expense/income/transfer sections with per-section and grand totals and a
subtotals-on-top/bottom toggle. The compare service now builds the tree on
top of the flat per-category deltas, so leaf-category figures are unchanged
and the #243 transfer netting is preserved (a balanced transfer group
subtotals to ~0, i.e. the group's net).
- reportService: buildCompareTree() synthesizes subtotal rows from the flat
leaves + category metadata; getCompareMonthOverMonth/YoY return the tree
(COMPARE_DELTA_SQL and rowsToDeltas untouched).
- CategoryDelta gains optional parent_id/is_parent/depth/category_type.
- ComparePeriodTable: sections, depth indentation, reorderRows toggle,
section/grand net totals.
- Cartes top-movers and ComparePeriodChart filter to leaf rows only.
- reorderRows constraint loosened to the two fields it reads.
- i18n (FR+EN) section labels; CHANGELOG entries.
Resolves#247
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Reports > Compare (real vs real) summed only outflows (SUM(ABS(amount))
filtered on amount < 0), so a `transfer` category like "Paiement CC" -
whose debit and matching credit should cancel - always showed the sum of
its debits instead of ~0.
Net transfer-type categories via a signed SUM(t.amount) while keeping the
expense behavior (ABS + amount < 0) for every other type, and broaden the
WHERE so transfer credits survive the expense filter. Both MoM and YoY now
share a single COMPARE_DELTA_SQL constant so they stay in sync. Real-vs-
budget already nets (signed SUM, no filter) and is unchanged.
Resolves#243
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Database lock (the reported 0.10.0 regression):
- tauri-plugin-sql loads SQLite through a default multi-connection sqlx pool
(Pool::connect => max_connections = 10) and exposes no JS transaction
primitive. BEGIN and COMMIT issued as separate db.execute calls could land on
different pooled connections, stranding an open write transaction on an idle
connection — a permanent write lock until app restart. Triggered when the
Bilan page's concurrent reads interleaved with an in-flight snapshot save.
- Fix: funnel every db operation through one FIFO lock in db.ts; withTransaction
holds it across the whole BEGIN..COMMIT so a transaction's statements never
span connections (and the saves become genuinely atomic). A reentrancy guard
lets nested getDb() calls run directly instead of deadlocking. Applied to all
5 transaction sites (saveSnapshotAtomic, upsertSnapshotLines,
proposeStarterAccounts, applyKeywordWithReassignment, applyMigration) via
in-place helper extraction. New db.test.ts covers serialization,
cross-transaction non-interleaving, reentrancy, and lock release on error.
Log console (Settings -> Systems -> Journaux):
- getLogs returned the same mutated array reference, so useSyncExternalStore
(identity comparison) never re-rendered on a new log -> the console never
updated live. getLogs now returns an immutable snapshot rebuilt on each
mutation (stable between mutations, new identity after each).
- Add logInfo/logWarn/logError app-logging API and instrument the snapshot save
(info on success, error on failure) so DB issues surface in the console.
tsc + 635 vitest + vite build green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cross-cutting Étape 2 coverage proving the per-security detail feature does not
regress aggregations, returns, date-move, or deletion.
Rust (lib.rs, +3 real-SQLite tests):
- regression_detailed_account_totals_equal_simple_account_totals: a detailed
account whose holdings sum to V yields byte-identical date/category/vehicle
SUM() totals to a simple account worth V (frozen golden numbers 1300/300/...).
Proven where SUM() GROUP BY actually runs, unlike the TS mock harness.
- migration_v14_to_v16_on_populated_db_preserves_integrity_and_totals: realistic
multi-type DB (simple + convertible priced w/ 2-snapshot history + non-
convertible priced) — totals byte-identical before/after v16, values
preserved, qty/price NULLed only on converted lines, one shared security,
non-convertible line fully intact.
- regression_snapshot_delete_cascades_to_holdings: two-hop CASCADE
(snapshot -> line -> holdings); security survives (RESTRICT).
TS (balance-flow.test.ts, +6 integration tests):
- detailed snapshot save end-to-end (aggregated line + securities + holdings in
one BEGIN/COMMIT); aggregated line value = rounded-cent SUM(holdings).
- rollback on injected holding INSERT failure (no partial line/holdings).
- snapshot date-move with a detailed account: line + holdings move together;
collision rolls both back (#200).
- golden-value invariant: detailed line stores the same value a simple account
would, feeding getSnapshotTotalsByDate identically.
- deleteSnapshot emits exactly one parent DELETE (FK cascades the rest).
No production code changed — pure test PR. Builds on the v16 tests (#211) and
the detailed-save/securities unit tests (#212) without duplicating them.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds a light confirmation modal (DetailAccountWizard) that flips a simple
balance account to detailed entry mode: sets kind='detailed' and
detailed_since = today (local civil day, YYYY-MM-DD) via updateBalanceAccount.
Toggle-only — no title capture; per-security holdings are entered at the next
normal snapshot, where validateDetailedSnapshot requires them from the pivot on.
Entry point: a 'Détailler en titres' action in the per-row actions menu of
BalanceAccountsTable, shown only for kind==='simple' rows (replaces the disabled
'Détail / coming soon' placeholder). Past aggregated history stays frozen
read-only. The flip is one-way: the #212 service backstop rejects
detailed -> simple once holdings exist, and the UI exposes no inverse action.
Exports buildDetailToggleInput() as a pure helper for a focused unit test
(project has no jsdom harness). FR/EN i18n under balance.detailWizard.*; removed
the now-dead balance.overview.detailAction / detailComingSoon keys.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Surface unrealized (latent) gain on the existing balance surfaces, no new
visualization (decision 2026-06-04).
Service:
- AccountLatestSnapshot gains account `kind` + `latest_snapshot_line_id`, so
the table knows which rows are detailed and where their holdings live.
- getAccountLatentGainByLine(lineId) folds a line's holdings through the
existing computeUnrealizedGain (shares the book_cost=0 / NULL -> N/A guard).
- rollupLatentGain(accounts): pure aggregation by asset class, by envelope
(vehicle_type, 'none' bucket), and grand total; per-bucket % denominator is
the known book_cost only (null when none), mirroring the aggregate guard.
Hook (useBalanceOverview):
- Prefetches per-detailed-account latent gain in parallel (failures isolated),
exposes latentGainByAccount + latentGainRollup. Simple accounts skipped.
UI:
- BalanceAccountsTable: expandable detailed rows -> per-security value + latent
gain %; a latent-gain column (shown only when a detailed account has one);
a summary block aggregating latent gain by asset class / envelope.
- BalanceOverviewCard: total latent gain line (hidden without detailed accounts).
- N/A rendered (never divide-by-zero) when book_cost is NULL or 0; partial-%
flag when some positions lack a cost basis.
Native non-CAD currency display de-scoped this round (untestable while all
securities are CAD). Modified Dietz return columns (#204) unchanged.
i18n: balance.latentGain.* in FR + EN. Focused unit tests for
getAccountLatentGainByLine and rollupLatentGain (grouping, 'none' envelope,
null-% / unknown book_cost, empty).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Turn the detailed-account snapshot variant into the real per-title entry
surface (building on the minimal sub-rows from #213):
- New SecurityPicker (src/components/balance/SecurityPicker.tsx): an
autocomplete combobox over the existing balance_securities catalogue
(loaded via listSecurities()) with inline creation. Accepts any
normalized symbol (UPPER/TRIM) with NO live ticker validation — the
price fetch is best-effort and separate. On pick/create it emits a
SecurityPick {symbol, asset_type, name, isNew}; a stock/crypto toggle
lets the user set the asset class when creating a new symbol (default
'stock'). Built on the CategoryCombobox UI idiom (ARIA listbox,
keyboard nav, click-outside). Pure helpers filterSecurities /
decideCreateOption are exported and unit-tested (no jsdom harness).
- SnapshotLineRow detailed sub-rows: labeled columns
[title (SecurityPicker), quantity, price (+ existing PriceFetchControl),
value (qty x price, read-only), book_cost, live unrealized gain].
Account value = displayed SUM of positions. Simple accounts unchanged.
- useSnapshotEditor: new SET_HOLDING_SECURITY action + setHoldingSecurity
callback (atomically sets symbol + asset_type + name and drops the
stale fetched-price attribution since the symbol changed). The
securities catalogue is loaded in loadForDate and exposed as
state.securities, so it refreshes after a save that creates a security.
- i18n: extended balance.snapshot.detailed.* (col.*, picker.*, book cost,
unrealized gain) in FR + EN — no hardcoded UI text.
- CHANGELOG (EN + FR) under [Unreleased]: first user-visible surface of
the per-title detail chain (#210-#213 were schema/service/reducer).
Build (tsc + vite) green; npm test green (613 tests, +10 SecurityPicker).
Generated autonomously by /autopilot run of 2026-06-06
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Structural rewrite of useSnapshotEditor for N holdings per detailed account,
and switch ALL simple/detailed dispatch from category_kind to the account's
own kind. This is the state/plumbing layer; the full multi-security entry UI
(SecurityPicker, rich sub-rows) lands in #214. Simple accounts behave
identically.
Reducer state shape:
- values: Record<accountId, string> (simple accounts, scalar)
- holdings: Record<accountId, HoldingDraft[]> (detailed accounts, one per title)
HoldingDraft is a string-typed, editable mirror of SnapshotHoldingInput with a
stable client-side rowId for React keys. Actions: ADD_HOLDING / REMOVE_HOLDING /
SET_HOLDING_FIELD (plus existing SET_VALUE/PREFILL/RESET). The legacy priced
scalar path (SET_PRICED_FIELD / pricedValues) is removed: after migration v16
(#211) every former-priced account is kind='detailed' with one holding, so those
accounts flow through the holdings path.
Dispatch:
- LOADED hydrates detailed baskets via listHoldingsBySnapshotLine (edit) keeping
the saved price, or getHoldingsForLatestSnapshot (new) dropping the price
(qty-0 excluded server-side). Simple accounts keep the scalar value path.
- SnapshotLineRow / SnapshotEditor / AccountForm now gate on account.kind, not
category_kind. category.kind survives ONLY as the suggested seed default for a
NEW account in AccountForm.
Save: detailed accounts pass their holdings array into SnapshotLineInput.holdings
(presence marks the line detailed; value = rounded-cent SUM); simple accounts
pass a scalar value with no holdings. Blank holding rows are skipped; a partial
row throws a typed error before any DB mutation.
AccountForm: adds an entry-mode selector (defaulting to the category-mapped
kind). New accounts persist as 'simple' (CreateBalanceAccountInput carries no
kind, and the service is out of this issue's scope); converting a fresh account
to detailed + pivot date is #215. Editing locks the selector for an already-
detailed account (the detailed->simple downgrade is service-guarded).
Tests: 19 new reducer/helper unit tests (pure exports; the project has no
renderHook harness) covering ADD/REMOVE/SET_HOLDING_FIELD, LOADED-vs-PREFILL
hydration (price drop, book_cost), qty-0 already excluded upstream, the
build*Lines save builders, and the dispatch-on-account.kind regression
(detailed account under a 'simple' category).
Resolves#213.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Service layer for detailed (per-security) balance accounts:
- findOrCreateSecurity (UPSERT on normalized UPPER(TRIM) symbol, callable
in-txn via an executor), listSecurities, getSecurity, updateSecurity.
- saveSnapshotAtomic / upsertSnapshotLines: a detailed account (line carrying
a holdings array) writes its aggregated line (value = rounded-cent SUM,
qty/price NULL) AND its holdings in the SAME BEGIN/COMMIT; the line id is
captured, existing holdings DELETEd, each security find-or-created and each
holding INSERTed. A holding-insert failure rolls the whole save back. Simple
/ legacy-priced scalar path is unchanged. upsertSnapshotLines is now wrapped
in an explicit transaction for the same atomicity.
- validateDetailedSnapshot: detailed+holdings => line qty/price NULL and
value === rounded-cent SUM(holdings) compared EXACTLY (no float tolerance);
detailed without holdings => pre-pivot aggregated tolerated.
validateLineKindInvariants stays byte-for-byte for the scalar path.
- roundToCent helper; detailed path uses per-holding cent rounding then exact
comparison to avoid N-holding rounding accumulation (decision 2026-06-03).
- Service backstop in updateBalanceAccount: detailed->simple rejected with a
typed error (account_kind_detailed_has_holdings) when holdings exist; adds
kind/detailed_since to the account input + SELECT.
- getHoldingsForLatestSnapshot (prefill; excludes quantity-0 positions),
listHoldingsBySnapshotLine (drill-down).
- computeUnrealizedGain: per-holding and aggregated value - book_cost and %;
book_cost = 0 OR NULL => gain % null (no divide-by-zero); NULL book_cost
excluded from the aggregate and flagged.
Existing aggregators (getSnapshotTotalsBy*) and computeAccountReturn untouched.
Unit tests for every new function incl. casing dedup, N>=20 holdings rounding,
book_cost=0/NULL, detailed->simple guard, atomic save + rollback. Existing
upsertSnapshotLines/updateBalanceAccount tests updated for the BEGIN/COMMIT
wrapping and the kind/detailed_since columns.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Document Étape 1 of the balance audit (vehicle_type axis), already shipped
in #202/#203/#204:
- ADR 0014 (Accepted): fiscal envelope is an account attribute, the category
is a pure asset class; Étape 2 (per-security detail) explicitly out of scope.
- ADR 0012 marked Rejected (never accepted, not Superseded) + pointer to 0014.
- User guide (markdown + in-app docs.balance i18n FR/EN): optional fiscal
envelope, the two chart axes, type renaming, and the historical-reclass note.
- CHANGELOG.md + CHANGELOG.fr.md [Unreleased]: Added (envelope field, envelope
axis, collapsible returns) + Changed (asset-class category, CELI/REER reclass,
rename no longer alters translation, historical-reclass note).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Issue 3 of overnight-2026-06-01-bilan-axe-vehicule. Builds the tracking UI
on top of the merged data layer (#202) and input UI (#203).
- Service: getSnapshotTotalsByVehicleAndDate(range) mirrors the by-category
aggregation with GROUP BY COALESCE(a.vehicle_type, 'none') so null-envelope
accounts land in a single 'none' bucket (never a SQL NULL key). Add
vehicle_type to getAccountsLatestSnapshot SELECT + type.
- useBalanceOverview: new groupAxis ('class'|'vehicle') state ORTHOGONAL to
chartMode; loads byVehicle alongside byCategory. Default groupAxis='class'.
- BalanceEvolutionChart + BalancePage: stacked-mode sub-toggle 'Par classe
d'actif' (default) / 'Par enveloppe'. Vehicle legend reuses the #203
vehicleType.* labels; the 'none' bucket uses balance.vehicle.none.
- BalanceAccountsTable: 4 return columns collapsed by default with a toggle,
persisted across sessions via userPreferenceService key balance_show_returns.
- i18n FR/EN: balance.chart.axis.{byAssetClass,byVehicle}, balance.vehicle.none,
balance.accountsTable.toggleReturns.{show,hide} (+ axisLegend aria label).
Tests: npm run build green (0 type errors); vitest 3314 passed. Added 5
service tests for the 'none' bucket + mixed envelopes + date range.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The internal Step component received `t: TFunction` as a prop while every
other component in the codebase calls useTranslation() directly at the
top of the function. Aligns with the majority pattern.
Suggestion S5 from PR #184 review (#187).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Before this commit, /balance rendered the BalanceOnboardingCard plus the
period selector + evolution chart + accounts table whenever the user had
no accounts or no snapshot. The lower three components surfaced their
own empty states, producing 3 stacked "no data" messages under the
onboarding card.
Lifts the (accountsCount, hasAnySnapshot) computation out of the inline
IIFE and uses a single isEmpty branch: empty profiles see only the
BalanceOnboardingCard; populated profiles see the full overview.
No logic change — only JSX restructuring. Tests covering useBalanceOverview
and BalanceOnboardingCard remain green (61 tests passing).
Suggestion S2 from PR #184 review (#187).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Before this commit, a brand-new profile briefly showed the
StarterAccountsModal even though the 4 starter accounts were already
seeded — the modal rendered 4 collision rows with no actionable choice
before being dismissed. Pre-seeding the pref in consolidated_schema.sql
suppresses the modal on first /balance visit for new profiles entirely.
Existing profiles already running the app are unaffected: they handle
the modal once on their first /balance visit (the pref-write happens on
dismiss). No migration is needed for them.
Suggestion S1 from PR #185 review (#187).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds defense-in-depth: each iteration runs a SELECT COUNT(*) WHERE name=?
AND balance_category_id=? AND archived_at IS NULL inside the BEGIN/COMMIT
block, immediately before its INSERT. On a hit, the iteration skips the
INSERT silently and the returned ids array excludes the skipped starter.
Rationale: balance_accounts has no UNIQUE constraint on (name, category)
and the upstream pre-filter (getStarterCollisions) is best-effort. If a
race or a bypass slips a duplicate through, the in-txn check catches it
without surfacing a confusing error to the user.
Existing two tests in StarterAccountsModal.test.tsx updated to mock the
new SELECT call sequence; new test "skips silently when in-txn collision
check finds an existing account" added.
Suggestion S3 from PR #185 review (#187).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
getStarterCollisions now filters `archived_at IS NULL` so a starter
account the user voluntarily archived no longer blocks re-creation
through the StarterAccountsModal. Matches the rest-of-codebase
convention (active = is_active=1 AND archived_at IS NULL).
Suggestion S4 from PR #185 review (#187).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extends PR #189's fix (one input on /balance/snapshot) to the 7 remaining
native <input type="date"> fields across 4 components:
- transactions/TransactionFilterBar.tsx (dateFrom + dateTo)
- adjustments/AdjustmentForm.tsx (form.date)
- balance/LinkTransfersModal.tsx (from + to)
- dashboard/PeriodSelector.tsx (localFrom + localTo)
Each onChange handler now calls e.currentTarget.blur() after the state
update to dismiss the native date popup on Linux Tauri WebView. The call
is a no-op on Windows WebView2 / macOS WKWebView, where the picker
already auto-closes.
No automated test added: this is a WebKitGTK-specific WebView quirk that
cannot be reproduced in jsdom/vitest. Manual smoke test on Linux Tauri
dev was the validation, mirroring PR #189's approach.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The single 12-card SettingsPage is replaced by a hub at /settings linking
to three thematic sub-pages mounted via a shared SettingsLayout (Outlet):
/settings SettingsHomePage (3 cards-cluster + PageHelp)
/settings/users UsersSettingsPage (Account, License, DocsContent)
/settings/data DataSettingsPage (Categories, DataManagement,
PriceFetchConsentToggle)
/settings/systems SystemsSettingsPage (Version, UpdateCard,
ChangelogContent, LogViewer)
DocsPage and ChangelogPage are extracted into reusable DocsContent /
ChangelogContent components and the standalone /docs and /changelog
routes become Navigate redirects to preserve external bookmarks and
release-note links. UpdateCard is extracted from the inline updater
block for symmetry and testability.
TokenStoreFallbackBanner is mounted once in SettingsLayout, surfacing
the OS-keychain-fallback warning across the four main routes only.
The two existing /settings/categories/{standard,migrate} sub-routes
stay flat (siblings of SettingsLayout) to keep their focused flows
free of the banner — their internal back-links now point to
/settings/data.
i18n FR/EN gain settings.{home, users, data, systems, backToHome};
docs/architecture.md and CHANGELOG{,.fr}.md updated. Pure refactor of
presentation: no new business logic, no Tauri commands, no SQL
migrations.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
WebKitGTK (Linux Tauri WebView) does not auto-dismiss the native
<input type="date"> popup after a value commit — the user has to
press Esc. Force-blur on change is a no-op on Edge Chromium WebView2
(Windows) and WKWebView (macOS), where the popup already closes.
Scope narrowed to /balance/snapshot per issue body. Six other date
inputs across the app share the same WebKitGTK bug; tracked in a
follow-up issue rather than bundled here.
Diagnostic: Ubuntu 24.04 + libwebkit2gtk-4.1-0 2.50.4-0ubuntu0.24.04.1.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Part 1 — New profiles: seed 4 starter accounts in
consolidated_schema.sql (Compte chèque/CELI/REER/Compte
non-enregistré, currency CAD, is_active=1) right after the
balance_categories seeds. Categories resolved via SELECT subquery
on the seeded `key` values for robustness.
Part 2 — Existing profiles: StarterAccountsModal proposes the same
4 starters at first /balance visit. Default-checked checkboxes,
collision rule (case-insensitive trim name + matching category)
disables matches with a "Déjà présent" tooltip. The atomic helper
`proposeStarterAccounts` wraps the inserts in BEGIN/COMMIT (rolls
back on error). user_preferences.balance_starter_proposed records
{shown_at, accepted} so the modal never reappears, dismissed or
confirmed.
Part 3 — docs/adr/0012-balance-two-level-model.md (Proposed):
captures the future vehicles × compositions model for reflection,
no code change. Numbered 0012 because 0011 was already taken by
the providers-best-effort-yahoo ADR. Linked from architecture.md
ADR table and Bilan section.
Tests: StarterAccountsModal.test.tsx covers STARTER_ACCOUNTS shape,
getStarterCollisions (case-insensitive trim, category-scoped) and
proposeStarterAccounts (insert order, COMMIT, ROLLBACK on failure).
No render tests — mirrors the BalanceOnboardingCard pattern (no
jsdom configured).
Resolves#179
Replace empty BalanceOverviewCard with BalanceOnboardingCard showing
two steps:
1. Create an account
2. Enter a snapshot
Step 2 is grayed out until at least one account exists; the entire
card is replaced by BalanceOverviewCard once a snapshot is recorded.
Hide "+ New snapshot" button when 0 accounts (it lives inside the
overview card, which is now hidden in that state).
Improve SnapshotEditPage noAccounts copy to clarify account vs
snapshot semantics.
Resolves#178
useSnapshotEditor.save now validates all simple/priced lines in-memory
before any DB write, then delegates to a new saveSnapshotAtomic helper
that wraps INSERT snapshot + INSERT lines in an explicit BEGIN/COMMIT
transaction (ROLLBACK on catch). Pattern matches categorizationService.
Migration v11 cleans existing orphan snapshots in profiles that hit the
old race; new orphans are no longer possible thanks to the transaction.
Resolves#176
SQLite raised "misuse of aggregate function MIN()" because MIN was used
in the WHERE clause of a scalar subquery. Replace with ROW_NUMBER()
OVER (PARTITION BY account_id ORDER BY snapshot_date ASC) filtered on
rn = 1.
Adds vitest coverage and a regression test for /balance load.
Resolves#175
Priced balance categories now carry an explicit `asset_type`
('stock' | 'crypto') so PriceFetchControl can route to the right
provider without symbol heuristics. ETH = Ethan Allen NYSE AND
Ethereum crypto are no longer ambiguous.
Migration v10 adds a nullable column and backfills the two seeded
priced categories (key='stock','crypto'). Legacy custom priced rows
stay NULL until the user edits the category — SnapshotLineRow hides
the price-fetch button when asset_type is NULL on a priced row, so
manual entry remains available.
Service-side validation rejects priced creation without asset_type
('asset_type_required') and rejects values outside ('stock','crypto')
('asset_type_invalid'). Simple kind coerces asset_type to NULL.
The CategoryVariant of AccountForm shows the selector only when
kind=priced, requires it on submit, and resets it on kind switch.
i18n keys added under balance.category.assetType.* (FR + EN).
Tests:
- 4 new Rust migration tests in lib.rs (column add, seed backfill,
legacy row stays NULL, CHECK rejects 'gold')
- 6 new vitest cases on createBalanceCategory + listBalanceAccounts
asserts c.asset_type AS category_asset_type in the join
- balance-flow integration test updated to pass asset_type='stock'
No new test for SnapshotLineRow render guard — project lacks
@testing-library/react + jsdom; the guard is one boolean expression
covered by manual QA per autopilot decisions in PR #167.
Fixes#169
- Adds PriceFetchConsentToggle to SettingsPage Privacy section
- Reads/writes user_preferences.price_fetching_consent for active profile
- Confirmation dialog before revoke (DELETE the key entirely so next click re-opens consent modal)
- Disabled (with notPremium tooltip) when license is not premium
- Adds deletePreference() to userPreferenceService
- Adds settings.privacy.title i18n key (FR + EN)
- 10 vitest tests covering all paths
Closes#159
- New component renders button + consent modal + spinner + attribution
- Best-effort warning shown once per session for stock categories
- Hidden if not premium or category kind != 'priced'
- Consent persisted per-profile in user_preferences.price_fetching_consent
- Manual unit_price input remains active in all paths
- 17 vitest tests (no RTL/jsdom — logged MEDIUM in decisions-log.md)
- Wired into SnapshotLineRow/SnapshotEditor/SnapshotEditPage
- asset_type hardcoded to 'stock' pending category schema extension (MEDIUM)
Closes#158
- prices.fetchPrice wraps invoke('fetch_price', ...) with local rate-limit (1/2s), in-flight dedup, exp backoff on 5xx (2/4/8s, max 3 retries), no retry on 4xx/429, hard 100/session cap
- 9 vitest tests with vi.useFakeTimers() (happy, 401/403/404, 429 no-retry, 5xx retries, dedup, pacing, session cap)
- Annexe B i18n mapping wired (PriceError → balance.priceFetching.errors.* keys)
- Session cap checked before rate-limit/dedup; failures do not consume budget (MEDIUM decision)
Closes#156
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add docs.balance.* keys (FR + EN) for the new Balance Sheet section
consumed by DocsPage (title / overview / features / steps / tips).
- Wire the new section into DocsPage.tsx SECTIONS array with a Wallet
icon, slotted between reports and settings.
- Add bilingual [Unreleased] CHANGELOG entry for #145 covering the
architecture.md updates, the 3 ADRs, the new guide section and the
i18n keys.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Source-level structural test on `TransactionTable.tsx` to lock down the
inlined transfer icon contract introduced in #142. Without RTL or jsdom
in the dev-deps, the test reads the component source and asserts:
- the icon is gated by `linkedTransfersByTxId?.has(row.id)`,
- optional-chaining short-circuits cleanly when the prop is omitted
(zero-impact on pre-#142 callers),
- the prop is declared OPTIONAL on the component interface,
- the `Link2` glyph comes from lucide-react,
- tooltip + aria-label go through `transactions.transferIcon.*` i18n
keys,
- the row's description cell layout (truncate span + title) stays
shared between linked and non-linked rows.
Catches the specific regression vectors: someone removing the gate,
renaming the prop, or breaking the optional-chaining pattern that
guarantees the page renders identically when no transfers are linked.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
End-to-end happy path through the full Bilan stack: account → priced
category → priced snapshot → linked transfer → return. Drives every
service against the existing in-memory FakeDb harness used by
category-migration tests so SQL shape (table names + parameters) can be
asserted alongside service outputs.
Currency lock: USD / EUR / GBP / JPY / AUD all rejected up-front by the
service with a typed `currency_unsupported` code, no DB hit. The CAD
default is verified to land in the INSERT params explicitly.
Priced-kind safety: a snapshot save with one out-of-tolerance line must
NOT clear pre-existing lines (the DELETE is gated behind the validation
loop). A drift just within ε is accepted unchanged.
computeAccountReturn wiring: malformed dates are rejected client-side
without invoking the Rust command; missing active profile yields a typed
`transfer_active_profile_unknown`; partial-period payloads are forwarded
unchanged (null fields preserved).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Issue #142 / Bilan #4 — translations and changelog entries.
i18n (FR + EN):
- `balance.returns.partialTooltip`, `balance.returns.noTransfersWarning`
- `balance.accountsTable.return3m/return1y/sinceCreation/unadjusted`
(label + tooltip variants)
- `balance.transfers.linkAction` + `balance.transfers.direction.{in,out}`
- `balance.transfers.modal.*` (every modal label, including the
partial-failure summary and the per-row direction toggle)
- `balance.transfers.errors.*` (5 new typed error codes)
- `balance.evolution.transferIn/transferOut` (chart label)
- `transactions.transferIcon.tooltip/ariaLabel`
CHANGELOG (English source + French translation):
- New entry under `[Unreleased]` summarising the Modified Dietz
formula, the per-account return columns (3M / 1A / since-inception
+ unadjusted), the link-transfers modal, the transactions-page
inline icon, the typed FK error on bulk-delete paths, and the
vertical reference markers on the evolution chart. References #142.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Issue #142 / Bilan #4 — vertical reference lines for tagged transfers.
`BalanceEvolutionChart.tsx` accepts a new optional prop
`transferMarkers?: BalanceAccountTransferWithTransaction[]`. For every
marker whose `transaction_date` matches a date already on the X axis,
the chart renders a `<ReferenceLine>` (Recharts) — green for `in`
(capital added), red for `out` (capital removed). The marker is drawn
in both `line` and `stacked` modes; in line mode an inline label
("In" / "Out") sits at the top-right of the marker so the user can
identify the direction without hovering.
Markers whose date is between two snapshot ticks are filtered out
(Recharts categorical axis silently drops unknown ticks; preferred
over an off-axis bug). A future improvement is to switch the X axis
to a numeric/time scale so markers can land anywhere — out of scope
here per the autopilot prompt's "least invasive" guideline.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Issue #142 / Bilan #4 — non-regressive transfer awareness in the
transactions table + clean error mapping on bulk delete.
- `TransactionTable.tsx`: optional new prop
`linkedTransfersByTxId?: Map<txId, links[]>`. When supplied, a small
`<Link2>` icon appears next to the description for every linked
transaction; tooltip lists the account name(s) and direction(s).
Without the prop, the table renders byte-for-byte identical to
before — preserves the spec's non-regression invariant.
- `TransactionsPage.tsx`: loads the linked-transfers map once on mount
via `listAllLinkedTransfersForTooltip()` (one batch SELECT) and
threads it through to the table. Failure to load the map degrades
gracefully to an empty map (icon simply doesn't appear).
- `transactionService.ts`: new `deleteTransaction(id)` helper +
`TransactionLinkedToBalanceError` (typed FK guard). Pre-checks
`balance_account_transfers` before attempting the DELETE so the
error carries the offending account names; falls back to the FK
pattern matcher if a race linked the transaction between the
SELECT and the DELETE.
- `importedFileService.ts`: both bulk delete paths
(`deleteImportWithTransactions`, `deleteAllImportsWithTransactions`)
now pre-check for any linked transfer and surface the same typed
error before they would explode on FK RESTRICT. The pre-check has
a `LIMIT 50` safety cap on the global path.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Issue #142 / Bilan #4 — UI for transfer linking + per-account returns.
- New `LinkTransfersModal.tsx`: portal modal with date-range / category /
free-text filters, multi-select with auto-proposed direction (`in` for
negative bank amounts, `out` for positive — flippable per row).
Submits via sequential `linkTransfer` calls; reports per-row failures
inline (most common case: `transfer_already_linked` on a re-submit).
- `BalanceAccountsTable.tsx`: 4 new columns rendered side-by-side —
3M / 1A / Since-inception (Modified Dietz via `compute_account_return`)
+ Unadjusted (`(V_end - V_start) / V_start`). Returns load lazily
after mount via `Promise.all` over (account × horizon); per-cell
failure leaves the slot at "—" without blocking the rest of the
table. The actions menu gains a *Link transfers* item that bubbles
the request up to the parent page. New props:
`sinceCreationDate` (anchors the since-inception horizon) and
`onLinkTransfers` (modal opener).
- `BalancePage.tsx`: hosts the new modal, loads the categories list
once on mount for the filter dropdown, fetches the union of
`listAccountTransfers` per account so the chart can render markers,
and threads the earliest snapshot date down to the table. Reload
is triggered after the modal reports at least one successful link.
- `balance.service.ts`: dropped the unused `BalanceAccountTransfer`
import to satisfy `tsc --noUnusedLocals`.
`npm run build` clean. `npm test` → 429 passed. Manual sanity check:
the table renders "…" placeholders during the per-row return load,
then resolves to either a percentage or a "—" with the partial
tooltip when the underlying snapshot endpoint is missing.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Issue #142 / Bilan #4 — TS bridge for the Modified Dietz command + plain
CRUD for transfer linking.
Types (`src/shared/types/index.ts`):
- `BalanceTransferDirection` ('in' | 'out')
- `BalanceAccountTransfer` (raw row) +
`BalanceAccountTransferWithTransaction` (joined view)
- `AccountReturn` (mirrors the Rust struct, ready to receive the invoke
payload as-is)
Service (`src/services/balance.service.ts`):
- `computeAccountReturn(accountId, periodStart, periodEnd)`: resolves the
active profile's `db_filename` from `loadProfiles()` and calls the
`compute_account_return` Tauri command.
- `linkTransfer(accountId, transactionId, direction, notes?)`: INSERT
with duplicate guard (typed `transfer_already_linked` error instead of
raw SQL UNIQUE failure).
- `unlinkTransfer(accountId, transactionId)`: DELETE with
`transfer_not_linked` guard for stale-UI calls.
- `listAccountTransfers(accountId, dateRange?)`: joined SELECT for
modal/list rendering.
- `listLinkedTransactionIds()`: returns a `Set<number>` for the
transaction icon (one query, in-memory `.has()` lookups thereafter).
- `listAllLinkedTransfersForTooltip()`: returns
`Map<transactionId, links[]>` for tooltip rendering.
- `suggestTransferDirection(amount)`: pure helper for the modal — maps
negative bank amounts to 'in', positive to 'out'.
- `isLinkedTransactionFkError(error)`: detects the canonical SQLite "FK
constraint failed" text so `transactionService.deleteTransaction` can
surface a clear i18n message.
- 5 new error codes added to `BalanceErrorCode`.
Tests (`balance.service.test.ts`): 22 new vitest cases bringing the file
to 85 passed. Mocks `@tauri-apps/api/core` `invoke` and
`./profileService` `loadProfiles`.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Insert a new "Bilan" / "Balance sheet" entry in NAV_ITEMS pointing
at /balance with the Wallet lucide-react icon. Position: between
Reports and Settings, matching the autopilot prompt instruction
and the spec-plan-bilan v2 ordering.
Sidebar.tsx imports Wallet and registers it in iconMap.
Refs: #141
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three new components composed under a new BalancePage at /balance:
- BalanceOverviewCard — latest aggregate net worth, Δ% vs the
previous chronological snapshot (rendered as "—" when only
one snapshot exists), 60-day staleness warning, and a
"+ Nouveau snapshot" CTA pointing at /balance/snapshot.
- BalanceEvolutionChart — Recharts-based line / stacked-area
toggle. Line mode plots SUM(value) per snapshot_date with a
single primary-coloured stroke. Stacked mode transposes the
byCategory series into one Area per category_key with a
fixed 10-color palette indexed deterministically. Tooltip
formats CAD via Intl.NumberFormat.
- BalanceAccountsTable — one row per active account with name,
category label, latest value, and Δ% over the active period
(latest_value vs the period anchor). Returns columns
(3M / 1Y / since-creation / unadjusted) reserved for #142
with a TODO marker. Action menu includes a disabled "Detail"
placeholder + functional "Archive" wired through reload().
BalancePage composes the three with an inline period selector
(3M / 6M / 1A / 3A / Tout) and chart-mode toggle, both styled
as segmented controls. State flows through useBalanceOverview.
Route /balance registered before /balance/accounts in App.tsx.
Refs: #141
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Scoped useReducer hook backing BalancePage. Tracks:
- period (3M / 6M / 1A / 3A / all) — defaults to 1A
- chartMode (line / stacked) — defaults to line
- evolutionTotals + evolutionByCategory + accountsLatest +
accountsPeriodAnchor (parallel-fetched on mount and on every
period change via Promise.all)
- isLoading + error
Exposes computeBalanceDateRange(period, today) as a pure helper
so the date math is unit-testable without mocking time. Anchors
on `today` rather than the latest snapshot — keeps the chart's
right edge stable as the user enters new snapshots.
Refs: #141
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add four service helpers used by the upcoming `/balance` overview:
- getSnapshotTotalsByDate(range?) — SUM(value) GROUP BY snapshot_date
with an optional inclusive [from, to] range. LEFT JOIN preserves
empty snapshots as zero rows so the chart shows continuity.
- getSnapshotTotalsByCategoryAndDate(range?) — same aggregation broken
down by balance_categories.key, returned as one row per snapshot
date with a `byCategory` map. Powers the stacked-area variant.
- getAccountsLatestSnapshot() — one row per active account with the
value of its most-recent snapshot line (NULL when none exists).
Filters archived accounts via WHERE is_active = 1 AND archived_at
IS NULL, matches the listBalanceAccounts default.
- getAccountsPeriodAnchor(range) — earliest snapshot_date >= from
per account, with the value at that date — the anchor used to
compute the per-account Δ% column on the accounts table.
Tests cover empty DB, single/multi snapshot, archived exclusion via
SQL inspection, date-range params (from-only, both bounds, open).
Refs: #141
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
