Skip to content

Place detail panel for map POI search and add modal (#1430)#1451

Open
maxjb-xyz wants to merge 234 commits into
liketrek:devfrom
maxjb-xyz:feature/poi-detail-panel
Open

Place detail panel for map POI search and add modal (#1430)#1451
maxjb-xyz wants to merge 234 commits into
liketrek:devfrom
maxjb-xyz:feature/poi-detail-panel

Conversation

@maxjb-xyz

Copy link
Copy Markdown

Summary

Resolves the request in discussion #1430 — see a place's info (photo, rating, hours, reviews) before adding it, instead of a name-only guess.

  • Map POI explore: clicking a POI icon on the map now opens a Google-Maps-style detail panel (styled like PlaceInspector) instead of jumping straight to the Add Place form. An Add to trip button opens the pre-filled form, now carrying the resolved Google place ID so the saved place links to Google.
  • Add Place modal: selecting a Google search result / autocomplete suggestion still fills the form instantly, and now also shows a details card docked to the right (stacked below on mobile) as a verification aid.

How it works

Map POIs are OSM-only (no Google place ID), so a new server endpoint resolves them:

  • GET /api/maps/poi-detailsgetPoiDetails() runs a Google Text Search biased to the POI's coordinates, accepts a match by name + within ~150 m, and returns the existing expanded details (rating, reviews, hours, editorial summary, photo). It falls back to OSM-only data when there's no Google key, no confident match, or the places_details_enabled kill-switch is off. Results (including negative matches) are cached in place_details_cache under poi:<osm_id> with a 7-day TTL, so each POI costs Google quota at most once.
  • Client fetches are cached per-POI (module map + sessionStorage). Every failure path degrades gracefully — adding a place is never blocked.
  • A shared PlaceDetailsCard renders the details; PoiDetailPanel (map) and the modal dock wrap it.

Photos reuse the existing single-photo pipeline (no carousel). Admin kill-switches (places_details_enabled, places_photos_enabled) are respected.

i18n

New poi.details.* strings added to en and translated across all non-en locales (parity --strict passes).

Testing

  • New unit tests: getPoiDetails matcher (match / distance-reject / name-reject / no-key / cache-hit), the poi-details endpoint, PlaceDetailsCard, PoiDetailPanel (Add forwards resolved place ID; close), and the modal dock (shown for Google results, absent for OSM-only).
  • Full client suite green; server + client typecheck clean; i18n parity strict OK.

🤖 Generated with Claude Code

mauriceboe and others added 30 commits June 28, 2026 10:30
…set-admin

The first-run seeder only applies ADMIN_EMAIL/ADMIN_PASSWORD on an empty
database and then silently ignores them. People add the vars after the first
boot, or pull a fresh image without clearing ./data, restart, and cannot log
in with no hint why (liketrek#1339). The default is a generated password (not the
.env.example placeholder), printed once in the first-run box. Now: warn loudly
when the vars are set but a user already exists, and warn on a partial
(one-of-two) config instead of quietly falling back.

Also ship the reset-admin recovery script in the image -- it was never COPYed in
despite the wiki referencing it. node server/reset-admin.js resets/creates
admin@trek.local with a generated password (RESET_ADMIN_EMAIL/RESET_ADMIN_PASSWORD
overridable), picks a free username so it cannot trip UNIQUE(username), and sets
must_change_password.
Booking import only fell back to the LLM when each user flipped an 'always retry with AI' toggle, so by default files kitinerary returned nothing for just failed. Run the fallback automatically whenever the AI Parsing addon is on (fallback-on-empty); drop the now-redundant per-user toggle and its setting.
client: the import call inherited the global 8s axios timeout and aborted long LLM extractions even though the server finished it; remove the timeout. server: raise the OpenAI-compatible LLM timeout 60s->180s (a cold Ollama model can take ~45s to first token). server: cap extracted text to 8000 chars before the LLM - multi-page T&C tails (30k+ chars) overflowed the context window, truncating the relevant head and making CPU inference crawl; booking details sit at the top.
…gn days

- rental car: request+map dropoffLocation, emit pickup->return from/to endpoints, set a location string (G1/G2/G3). - geocode endpoints (stations/stops/terminals/rental desks) on confirm via Nominatim; mapper now emits coordless named endpoints and confirm persists only the geocoded ones (G6). - assign every dated booking to the nearest trip day so it still shows when slightly out of range, and keep hotel accommodation from vanishing when a check date misses (G5/G10). - fix bus mislabelled as train + add bus_number metadata (G7/G8), flag malformed boats (G9), accept root start/end time for events (G11). - raise the local-LLM timeout to 300s for CPU-only Ollama.
On a GPU-less host the model's prompt-eval time scales with input length and dominates total latency. Booking details sit at the top of a confirmation, so capping the extracted text at 4000 chars (was 8000) roughly halves extraction time (~50s warm for a capable local 7B model) with no loss of fields on real hotel/rental confirmations. Tunable if a long multi-segment itinerary needs more.
…tact

Request and map root-level seat/class/platform and a total price/currency into reservation metadata (shown on the card; price reuses the existing label). Read both the root and reservationFor and tolerate common field-name aliases (priceAmount, priceCurrencyISO4217Code, fareClass, ...) since models name these inconsistently. Also capture event/attraction venue telephone + url onto the auto-created place, matching lodging/restaurant.
When a confirmation carries a total price, record it as a real expense
linked to the reservation (in the matching Costs category) instead of
leaving the amount in metadata only. Gated on the Costs addon.
A freshly imported hotel links to an accommodation that lives outside the
trip store, so loadTrip alone left the reservation edit modal with blank
place/date fields. Reload the accommodations list once the import finishes.
NuExtract isn't an instruct model — fed a plain chat prompt it just echoes the
schema back. Detect a NuExtract model by id and talk to it the way the model
cards document: the JSON template inlined in a single user message, no system
prompt, no json_schema, temperature 0. Its flat result is mapped back to the
same KiReservation shape the rest of the pipeline already uses, so nothing
downstream changes; every other model keeps the generic prompt.

Money is taken as a verbatim string and parsed locally (German "1.580,22 €"
otherwise comes back as 1.49772), a rental car's pickup/return ride the from/to
fields so a stray form label doesn't become the location, and a lodging with no
name falls back to its address instead of being dropped.
The provider picker is the shared CustomSelect now and the form is split into
clear sections rather than a flat stack of inputs. NuExtract 2.0 2B is the
recommended default — fastest on a CPU-only host and MIT licensed; the 4B
carries a non-commercial licence, so it's no longer flagged as recommended.
Instead of writing parsed items straight to the trip, the import opens the
normal edit modal pre-filled for each one, so you can check and fix it before
saving — useful when a model guesses a wrong date or address. Hotels gained an
editable address field; on save an existing place is matched by name, otherwise
the reviewed address is geocoded and a new place is created.
The single-shot prompt was unreliable on multi-leg flights and longer
documents, and slow on a CPU host. For the local provider, run a small
router instead:

- deterministic vendor templates first, with no model call at all
- exactly one grammar-enforced call per document via Ollama's native
  `format` (flights as a flat array of legs, everything else as one flat
  reservation, the type picked from keywords or a union schema)
- booking-wide fields (booking reference, total price, the overnight
  arrival day) filled deterministically from the text afterwards, and
  dates coerced to ISO so a natural-language date can't slip through

Recommend qwen2.5 in the AI-parsing settings instead of NuExtract.
Parsing a booking can take a while on a CPU host, so don't hold the
upload modal open for it. The async import endpoint returns a job id
right away; the parse runs server-side (one at a time per user) and
pushes progress over the user's WebSocket, and a small widget in the
bottom corner tracks it while the user keeps navigating and editing.
A finished job opens the per-item review from the widget.
…kings

Reviewing an imported booking saves it through the normal reservation
form, which dropped the parsed price (so no linked cost was created) and
only created the accommodation when both nights matched a trip day.
Carry the parsed price into a linked cost on save, and create the
accommodation from whichever day the check-in/out dates resolve to.
Pull the hotel/rental fields these vendors print in a stable text layout (name, address, stay/pickup dates, price, reference) deterministically, so the import stops depending on the local model for them. Handles German long/abbreviated months and English dates incl. 12-hour and comma forms.
Apply the deterministic confirmation-code and total fill to vendor-template results too (not just model output), and require the captured reference to contain a digit so a bare 'Confirmation'/'Reference' label no longer grabs the next prose word.
Persist the background-import tasks (id/trip/status only) and re-fetch each job's status on mount, so a parse still running when the page reloads keeps its widget instead of vanishing; expired jobs (404) are dropped and a restored 'done' task re-fetches its items.
…save

reservation_endpoints.lat/lng are NOT NULL, so saving a reviewed transport whose pick-up/return couldn't be geocoded threw a 500 and lost the whole booking (dates, linked cost). Skip those rows; the dates still persist on reservation_time/reservation_end_time.
A reviewed transport (e.g. a rental car) arrived with only its parsed pick-up/return dates and no day_id, so the modal kept just the time and saved a bare "HH:MM" with no date. Resolve start/end day from the parsed dates (exact match, else nearest trip day) so the booking lands on the right days.
…s appear without a reload

Imported bookings auto-create their linked budget items server-side, but the saving client suppresses its own budget:created echo, so the Costs list stayed stale until a manual reload. Reload the budget items when the review session ends.
… support JPY vouchers

Code-audit clean-ups: share one normCurrency between the router and the templates, lift the duplicated nearest-day resolver into formatters.resolveDayId, drop two needless as-unknown-as casts at the fillBookingWideFields call sites, restore routeExtraction's doc comment, and give the broker template readable names. Plus recognise ¥/JPY and fall back to a standalone symbol amount, so a Klook-style voucher whose price sits far from any label still yields a cost.
Keep the uploaded files on the background task and hand them to the review flow, so each reviewed booking pre-fills its Files with the document it was parsed from (uploaded with the booking on save). The two modals also adopt the shared resolveDayId helper.
Hybrid/reasoning models (Qwen3 and similar) default to emitting reasoning tokens, which collide with Ollama's format-grammar constraint — on CPU this produced null/unparseable output and blew the latency budget (qwen3:8b: null or 300s timeouts vs ~20s with thinking off). Send think:false on the /api/chat call; Ollama ignores it for non-thinking models (verified on qwen2.5:7b), so it's safe and unlocks the stronger Qwen3 family.
A/B against the prior default (qwen2.5:7b) on CPU showed Qwen3-8B is both faster and more accurate on tricky/multilingual booking docs (correct Airbnb year+price, correct DisneySea admission date), once thinking is disabled — which the router now does. Feature it as the recommended pull, keep qwen2.5:7b as the fallback.
…terministic backfill

Now that a capable instruct model (Qwen3-8B, thinking off) reads name/address/dates/legs reliably across formats, the per-vendor template short-circuit distorted more than it fixed: brittle on layout variations and overriding the better model output. Remove the template layer; the model extracts the structure and Schicht 2 backfills the confirmation/total and takes the currency from the document's own symbol (correcting model misreads like ¥→$). Per-type prompts now also ask for address and price/currency.
After dropping the vendor templates, the model skipped the (often unlabeled) Expedia-style hotel address — making address a required schema field forces it to emit the street-address line, restoring the booking's location/place. Also hint the rental company so a car booking gets a real title instead of the generic fallback.
…aved

The saving client gets no budget:created echo (X-Socket-Id) and the create response omits the linked budget item, so the booking's Costs section and the Costs tab stayed stale until a manual reload. Reload the budget items right after a create that carried a budget entry.
mauriceboe and others added 13 commits July 5, 2026 00:23
…etrek#1411)

The great-circle sampler normalizes longitudes to [-180,180], so a
transpacific leg jumped +-360 between neighbours and got split into two
polylines pinned to opposite map edges. Unwrap the longitudes instead
(shared flightGeodesy module for both renderers): Leaflet additionally
draws a +-360-shifted copy so both halves show in the standard view, GL
maps repeat world copies themselves.
…trek#1404)

Clicking an off-center place recenters the map under a stationary
cursor, so mouseout/mouseleave never fires and the hover card sticks.
Clear it on marker click and on movestart, and suppress re-shows while
the camera is animating (marker rebuilds re-fire mouseenter mid-pan).

feat(map): long-press + plain right-click add-place on GL maps (liketrek#1398)

The GL providers only bound middle-click, so mobile had no way to add a
place at a position (and Macs have no middle button). Add a 600ms touch
long-press with move tolerance and the map contextmenu event - both GL
libs suppress it while the right-button rotate/pitch drag is active, so
the gesture keeps winning.
…related saves (liketrek#1414)

Three separate causes for the reconnect-per-tool-call pain:
- no keep-alive on the standalone GET stream, so reverse proxies with
  idle timeouts (nginx default 60s) killed it between calls - send an
  SSE comment ping every 25s (MCP_SSE_KEEPALIVE, 0 = off) and count an
  open stream as session activity
- the session TTL was hard-coded - MCP_SESSION_TTL (seconds, clamped to
  24h) now works as the issue expected
- every addon save invalidated ALL sessions: config-only saves, photo
  provider toggles and addons with no MCP surface included. Only a real
  enabled-flip of an MCP-relevant addon (or an actual collab-feature
  change) tears sessions down now.
…BLED (liketrek#1412)

Swagger UI + raw spec (/api/docs-json, -yaml) over all controllers, with
a bearer button that works with a plain session JWT. Off by default -
the spec enumerates the whole surface incl. admin routes, so exposing
it is an explicit self-hoster decision (same kill-switch pattern as
TREK_PLUGINS_ENABLED).

Request bodies come from the Zod schemas the routes already validate
with: an enricher walks every controller, finds whole-body
ZodValidationPipe params and lifts their schema into the document via
zod v4's native z.toJSONSchema - nothing is annotated twice, and any
route that gains a Zod pipe is documented automatically.
- mapbox-gl (unlike maplibre) still emits the map contextmenu after a
  right-button rotate/pitch drag on Windows - guard it with the pressed
  position so ending a rotate can't open the Add-Place form (liketrek#1398)
- a long-press whose fire was deduped (or that never yields a click) no
  longer leaves suppressNextClick armed to swallow a later real tap
- MCP_SSE_KEEPALIVE=0 keeps the open-stream-counts-as-activity
  guarantee: the touch interval survives, only the pings stop (liketrek#1414)
- swagger-ui-dist ships @scarf/scarf install-time analytics - disabled
  via scarfSettings in the root package.json, TREK sends no telemetry
- Budget wiki currency list: 47 incl. KGS (liketrek#1400)
… of straight lines

Road-based transport bookings drew an as-the-crow-flies line; only
transit journeys (Transitous) showed the real path. A shared
useTransportRoutes hook now fetches the OSRM road geometry (driving for
car/bus/taxi, cycling for bicycle) — reusing the day-route router and
its cache — and both renderers draw it in place of the straight arc,
falling back to the straight line until it loads or if routing fails.
Trains/other keep their straight line (not road-routable); a 2000 km
sanity cap avoids hammering the public router on cross-continent quirks.
Long train trips are usually several trains under one booking. Trains
now get the same multi-leg editor flights have: an ordered chain of
stations (station search instead of the airport picker) with a per-leg
train number + platform, saved as from/stop/to endpoints + metadata.legs
— mirroring the flight leg contract, so the map draws the whole chain
and the day plan splits it into one row per leg (drag/reorder/position
persistence come for free from the shared __leg machinery). A single-leg
train saves exactly as before (flat metadata, no legs), and the flat
train-fields block is gone in favour of the per-leg inputs. Day sidebar,
shared trip view and the PDF render each train leg like a flight leg.
Each list can now define its own labels (e.g. Berlin, Hamburg, Ostsee in a
"Germany 2026" list) and organise its places by them:

- manage labels (create / rename / recolor / delete) from a label manager
- assign labels to a place from its detail sheet, or to many places at once
  from the selection toolbar
- filter the place list AND the map by label (multi-select, any-match)

Labels are scoped to a collection and shared by all its members. Managing and
assigning labels needs edit rights; filtering is available to everyone. Moving
a place to another list drops its labels, since they belong to the source list.
# Conflicts:
#	client/src/pages/SharedTripPage.tsx
#	client/src/pages/tripPlanner/useTripPlanner.ts
…tail test

The per-collection labels feature (a5522e9) made `labels` a required prop
and renders `labels.filter(...)`, but the test's props cast to
Omit<DetailProps,'t'> hid the missing prop, so `labels` was undefined at
runtime and crashed the whole suite (Cannot read properties of undefined
reading 'filter'). Pass labels: [] like categories.
…te overlays

- Collections: add a Custom labels section (manage / assign / filter), note the
  label filter + bulk assign, and the view-vs-edit permission split
- Transport: rewrite the train fields as the multi-leg route editor, correct the
  transport type list (nine types) and the map/day-plan behaviour
- Map Features: car/bus/taxi/bicycle overlays follow real roads; multi-leg trains
  draw their full station chain; date-line routes render as one continuous arc
'Tàn tật' means physically handicapped/disabled-person, not the
off/disabled state of a toggle. Replace with 'Tắt' (off), matching
the existing 'admin.plugins.stateOff' translation.

Affects admin.notifications.none and admin.addons.disabled.
@github-actions github-actions Bot added the wrong-base-branch PR is targeting the wrong base branch label Jul 5, 2026
@github-actions

github-actions Bot commented Jul 5, 2026

Copy link
Copy Markdown
Contributor

Wrong target branch

This PR targets main, but contributions must go through dev first.

To fix this, click Edit next to the PR title and change the base branch to dev.

This PR will be automatically closed in 24 hours if the base branch has not been updated.

If you need to merge directly to main, contact a maintainer.

@maxjb-xyz
maxjb-xyz changed the base branch from main to dev July 5, 2026 19:18
@github-actions github-actions Bot removed the wrong-base-branch PR is targeting the wrong base branch label Jul 5, 2026
MadMaximusJB and others added 10 commits July 5, 2026 15:19
Adds getPoiDetails() to mapsService with haversineMeters/poiNamesMatch/osmFallbackDetails
helpers, DB caching under poi:<osm_id> (7-day TTL), and 5 TDD tests (POI-001…005).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…esults

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.