Add UA rotation (8 real Chrome UAs), random viewport, en-GB locale,
Europe/London timezone, stealth browser args to suppress automation
signals, and a per-context init script masking navigator.webdriver,
plugins, and chrome runtime.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The previous approach walked siblings from [id^="room_type_id_"] anchors
and only processed rows with js-rt-block-row class, causing the first rate
row of some room types to be silently skipped when that class was absent.
Now iterates all #available_rooms tbody tr rows with data-block-id (same
strategy as the original scrapy spider), using the room_type_id_ element
presence to identify room names only on the first row, with fallback to the
stored name for subsequent rows of the same room type.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Include tier=own in get_active_hotels() so the own hotel Booking.com
listing is scraped alongside competitors. Change matrix max_persons
filter from hard WHERE to ORDER BY priority so hotels with only
1-person rates (e.g. Old Stocks) still show up rather than being
silently excluded.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
_proxy_enabled() was checking cfg.get('server') but load_config() returns
{host, port, username, ...} — no 'server' key. Server URL is built by
proxy_util.playwright_proxy(). This meant proxy was silently disabled
and _proxy_kwargs() would also KeyError if called.
- _proxy_enabled() now delegates to proxy_util.is_enabled() (checks host+username)
- _proxy_kwargs() now calls proxy_util.playwright_proxy() with credentials
embedded in the URL (avoids 14s DataImpulse 407 round-trip)
- _get_context() generates a sticky session_id per context so each worker
gets a distinct residential IP lane
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The proxy-off → serial guard was correct for search-results (all workers
hit the same URL), but hotel-page workers each scrape a different hotel's
property page — parallel requests look like multi-tab browsing, not a
hammered aggregation endpoint. Added require_proxy param to
_effective_concurrency; hotel-page path passes require_proxy=False.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Scraping a past date via the hotel-page backend wouldn't return anything
useful (Booking.com shows no availability for dates gone by). The ↻
button is now only rendered for today and future dates.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Scraper: restrict hotel-page scraper to 'competitor' tier only (was own+competitor); own hotel rates come from Newbook API, not Booking.com
- Backend: POST /competitors/discover — runs search-results scrape for one date to find market hotels regardless of current backend setting
- Backend: POST /competitors/hotels — add hotel manually from Booking.com URL + name + tier; upserts on slug conflict
- Frontend (Settings tab): Discover Market Hotels button added below manual date-range scrape
- Frontend (Hotels tab): Add Hotel form at top — paste URL (auto-derives name from slug), choose tier, submit
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Market-tier hotels were auto-discovered from search results and don't
need room-level rate tracking — scraping all 25+ of them per date was
unnecessary. Only 'own' and 'competitor' hotels are now scraped.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Sold out + last price: show SOLD label above strikethrough price; remove price-index badge (rate no longer bookable, delta misleading)
- Sold out, no history: unchanged — shows 'Sold'
- Past days: price-index badge now grey/faded (#94a3b8 on #e2e8f0) instead of coloured — historical context only
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Scraper: extract max_persons from block_id (3rd segment) instead of cell text; cell[0] was picking up the price value (e.g. 189/166) causing the max_persons=2 filter to exclude all scraped rows and show "No rates available" in snapshot
- MarketView: add namelength: -1 to Plotly legend to prevent label truncation; increase bottom margin from 50→65 and legend y from -0.25→-0.3
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- MarketView: split tickformat '£,.0f' into tickprefix+'£' + tickformat ',.0f' (Plotly d3 format doesn't accept £ prefix inline)
- Settings System tab: add editable dropdown for booking_scraper_backend (hotel page vs search results) with description of the tradeoff; backend was previously read-only in the table
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The 'We have X left' scarcity indicator only appears when availability is
low (typically ≤5). The quantity <select> dropdown in the booking form
always shows 0..N where N = actual available qty (capped at 10).
Extract max option value from the <select> in the last table cell per rate
plan row, falling back to regex parsing of the cell text if the select
isn't rendered, then to the scarcity text as a last resort.
Stored in both rooms_left and available_qty on RateData.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Modal now shows two sections:
- Current availability: all room types with rooms-left count, then each rate
plan (meal plan × cancel policy × price) from the latest scrape batch
- Rate history chart: one line per room type, cheapest 2-adult rate per
scrape run (max_persons filter added to exclude 1-adult variants)
New endpoint: GET /competitors/hotels/{id}/rate-snapshot/{date}
Returns all rate plan variants grouped by room type from the latest batch.
Handles legacy single-row data (pre hotel-page scraper) gracefully.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Switches booking.com scraping from search-results page (Cloudflare-targeted)
to individual hotel property pages (not CF-protected). Each page load returns
all room types, all rate plan variants (room-only/B&B × refundable/non-ref ×
1-2 adults), and availability counts.
Key changes:
- New PlaywrightHotelPageBackend: proxy reuse until block, rotate on CF/WAF
- booking_scraper.py: _run_hotel_page_scrape(), scrape_hotel_date(),
_scrape_hotels_concurrent() — sharded by hotel so one proxy session covers
all dates for one hotel (looks human)
- schema.sql: ADD COLUMN rate_plan_id, max_persons on booking_com_rates
- competitors API: filter max_persons=2, order by rate_gross ASC as tiebreaker
so DISTINCT ON returns cheapest 2-adult rate from latest batch
Enable via Settings → Scraper Backend → playwright_hotel_page
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When page.goto() times out while Cloudflare's JS challenge redirect is in
progress, page.content() silently blocks until that navigation completes
before raising — adding ~1m46s of hidden delay per blocked-page attempt.
Fix: skip page.content() entirely when page_loaded=False (goto already
timed out), setting content="" so block detection treats it as no signal.
Also break out of the page loop early when page 1 failed to load with 0
hotels — no point spending another 60s on the offset-URL page 2 when the
browser is in a Cloudflare challenge state.
Per-blocked-attempt time: ~4 min → ~1 min (30s goto + 30s selectors).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Timeouts were set for the old proxy 407 round-trip world (14.5s latency
per challenge). Now that credentials are embedded in the proxy URL, pages
load in ~7s — the 90s goto timeout was forcing 5+ minutes of wasted wait
on Cloudflare-blocked dates before rotating sessions.
homepage warmup goto: 90s → 20s
page 1 search goto: 90s → 30s (4× normal load time headroom)
property-card waits: 30s → 15s (cards appear in <3s on clean pages)
Default concurrency 3 → 6: each worker uses its own residential proxy IP
so parallelism is safe. ~0.4 GB per Chromium; 6 workers fits in 4 GB LXC.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Two rapid POST /scrape requests could both pass the lock check before
either background task acquired SCRAPE_LOCK, queuing two sequential
scrapes for the same date range. The second would hit Cloudflare after
the first already succeeded, causing repeated retries.
_SCRAPE_PENDING is set on submission and cleared when the task starts,
closing the gap between the 409 check and lock acquisition.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When the JWT expired, AuthGate redirected window.location to /auth/login
which loaded inside the portal iframe. The portal detected window.self !==
window.top and showed the EmbeddedFallback ("This app isn't available yet")
instead of the login page.
Redirect window.top instead so the portal itself navigates to /login.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
DataImpulse takes ~14s to issue a 407 challenge from this network.
Passing credentials as separate Playwright proxy fields caused Chromium
to wait for that challenge before sending auth on every CONNECT request,
making scrapes consistently time out at 90s.
Embedding credentials directly in the proxy server URL makes Chromium
send Proxy-Authorization on the first CONNECT, bypassing the round-trip.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
page.content() throws when page is still navigating after a goto timeout;
catch it and continue with empty content. Also broaden the soft-block check
to trigger IP rotation whenever no hotels are found (not just on explicit
blocks), so proxy rotation fires even when the error path is hit.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When all 3 proxy rotation attempts are soft-blocked (bad IP pool), fall
back to the server's own IP for a final attempt. Keeps proxy as primary
for IP diversity / rate-limit protection while ensuring a clean fallback
when the assigned residential pool is Cloudflare-challenged.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
A direct hit to searchresults.en-gb.html with no Referer causes Cloudflare
to hold the response body open (never sending HTML), so domcontentloaded
never fires. Setting referer=HOMEPAGE makes the request look like the user
searched from the homepage, which resolves the stall.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Headless mode gets blocked by Cloudflare on the search endpoint. Headed
mode with Xvfb works on dev (Unraid). The missing piece on Proxmox LXC was
insufficient /dev/shm (Docker default 64 MB); setting shm_size: 256m gives
Xvfb enough shared memory to start.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
DataImpulse residential proxy takes ~29s to route through Booking.com's
Cloudflare edge — the previous 30s timeout was too short and the page
never had a chance to deliver its body. Also raised wait_for_selector
from 15s to 30s so property cards have time to render post-load.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Xvfb exits immediately on the production LXC (no framebuffer device
support), so headed mode was never working. Switch to headless=True
and remove Xvfb from the Dockerfile entirely.
Stealth coverage is unchanged: playwright-stealth patches webdriver,
plugins and chrome.runtime; the residential proxy + sticky session
provide the GB residential fingerprint; homepage warmup carries real
session cookies into the search. --enable-unsafe-swiftshader and
--disable-blink-features=AutomationControlled are kept.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Root cause: threading.Lock held indefinitely when Playwright browser
hangs inside run_in_executor (finally never fires from the async side).
Fixes:
- _acquire_scrape_lock/_release_scrape_lock track monotonic timestamp
- POST /competitors/scrape/reset force-releases the lock and marks any
running batch as interrupted (queue rows stay intact for retry)
- GET /competitors/status now includes lock_held_seconds
- APScheduler watchdog job every 30 min auto-releases if held >3h
- Settings → Scraper Proxy tab shows live lock status (green/amber)
with a Force Reset button requiring confirmation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Sold-out cells: amber (was red), show last known available rate with
strikethrough. Past dates: grey, same strikethrough treatment, no eye
icon (Booking.com won't serve past availability), history chart still
accessible. Date column headers also hide the eye link for past dates.
Backend adds a second query returning last_available_rate (most recent
available + non-null rate_gross) per hotel+date for the full range.
Icons are now 12px and stacked below the rate text in a flex-column
cell layout. Price index badge sits between rate and icons.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace full-cell link with two small icon buttons in each matrix cell:
- Eye icon links to the hotel's Booking.com page for that check-in date
- LineChart icon opens a history modal (Plotly multi-line chart)
The history chart plots best available rate over scrape runs, with one
line per room type so that changes in which room is cheapest show as
separate traces rather than a single jumpy line.
Backend: GET /competitors/hotels/{id}/rate-history/{stay_date} groups
by scrape batch, takes MIN(rate_gross) per room type per run, returns
series [{room_type, points: [{t, rate}]}].
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
New GET /competitors/own-direct-rates: cheapest bookable non-dinner
Newbook tariff per date (same selection rules as the parity check).
'Show direct rates' now renders a 'Direct (Newbook)' sub-row under the
own-hotel row alongside the competitors' scraped direct rows; cell
tooltip names the tariff.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
parity_match_mode config ('best_available' default | 'match_terms').
Best available = cheapest bookable non-dinner direct tariff vs the BC
lead-in (already BC's best available) — simplest like-for-like. Term
matching kept as an option for days where BC's cheapest basis differs
from direct's. Settings gains a comparison-basis select.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The check compared Newbook's FIRST tariff (typically flexi B&B) against
Booking.com's lead-in card (often prepaid) — apples vs oranges, e.g.
21 Jul flagged B&B FLEX £289 vs a prepaid BC rate whose true comparable
was B&B PPAY £279.
- classify Newbook tariffs from names/descriptions (PPAY/prepay/advance/
ADV/NRF/saver = prepaid; DBB/dinner/half board = dinner-inclusive)
- use the scraped BC rate's flags (free_cancellation/no_prepayment/
breakfast_included) to pick the cheapest COMPARABLE tariff per date,
excluding tariffs not bookable for that date (success=false, min-stay
>1, unmet advance-purchase windows)
- tiered fallback (matched -> any non-dinner -> any -> legacy headline),
recorded per alert in room_category as 'TARIFF vs BC basis'
- shared gather_comparisons() now drives both the daily job and
GET /competitors/parity (issues gain newbook_tariff, booking_basis,
match_quality, expected_rate)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- config: parity_markup_value/_unit + parity_tolerance_value/_unit
(pct|gbp), legacy *_pct keys still read as fallback
- expected rate = newbook + £X or newbook × (1 + X%); breach test uses
the tolerance in its own unit
- Settings parity tab: unit selects + live worked example line
- Parity Alerts tab: unit-aware description, badge now shows £ deviation
alongside %
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Parity was read-only — the alerts table had no producer, so the Market
View badge could never fire. Now:
- jobs/check_rate_parity.py: daily 06:45 job comparing own Booking.com
lead-in rate vs cheapest Newbook rate per date, measured against an
EXPECTED markup (we deliberately price Booking.com higher to cover
commission): alert when deviation from newbook*(1+markup%) exceeds the
tolerance. Creates/updates active alerts, auto-resolves dates back in
line, leaves acknowledged dates alone.
- config keys: parity_check_enabled, parity_expected_markup_pct,
parity_tolerance_pct (system_config)
- POST /competitors/parity/check manual trigger; GET /parity now uses the
same markup/tolerance and cheapest-across-categories Newbook rate
- Settings -> Rate Parity tab: markup %, tolerance %, enable toggle,
run-now with result summary
- Market View -> Parity Alerts tab: status-filtered list w/ acknowledge
- Market View -> Hotels: direct-link dropdown per competitor (new PUT
/competitors/hotels/{id}/direct-link) — closes the never-written
direct_hotel_id gap so the matrix direct-rates sub-row can populate
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- new GET /direct/hotels/{id}/config returns full hotel config plus
distinct room/rate IDs from scraped data (and any known via labels)
- Configure panel: benchmark room/rate selects, tier base room, per-room
£ offsets (base locked to 0), friendly room/rate names, room display
order (up/down) — saved via the existing PUT endpoint
- enables apples-to-apples estimated benchmark when only e.g. a suite
is left: bench = room_price - room_offset + bench_offset
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Direct-rates sub-row in Market View was dead: the frontend filters
competitors on direct_hotel_id but /matrix and /hotels never returned it.
Add direct_hotel_id to both queries (+ HotelResponse), and only render the
"Direct" sub-row when a hotel actually has a non-null direct rate (was
rendering all-dashes on an empty {} object).
Remove the pause-on-block flow entirely — dormant since rotate-on-block
replaced it (nothing set booking_scraper_paused=true after set_scraper_paused
was dropped): is_scraper_paused, /config/unpause, the /scrape paused guard,
ScraperStatusResponse.paused/pause_until, and the frontend Paused badge +
Unpause button. Trim now-unused datetime import.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Proxy config, DataImpulse sticky-session username syntax, and Playwright/
httpx proxy builders now live in one place (services/proxy.py) instead of
being duplicated across the Booking.com backend and the /config/proxy test
endpoint. Both scrapers consume it.
- services/proxy.py: load_config/normalize (DB-authoritative, env fallback),
new_session_id, username, playwright_proxy, httpx_proxy_url
- PlaywrightLocalBackend delegates proxy building to the module
- get_scraper_backend factory uses proxy.load_config (one resolution path)
- test_proxy_config endpoint uses the shared URL builder; httpx proxies= ->
proxy= (forward-compatible, 0.28-safe)
- Direct booking-engine scraper (httpx) can now route through the same proxy,
gated by the direct_scraper_use_proxy flag (default off, plumbing ready)
Dead code removed: set_scraper_paused (never called — rotate-on-block
replaced pause-on-block), get_competitor_matrix / get_hotels_list /
update_hotel_tier (endpoints have their own SQL), unused PROXY_KEYS tuple.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fan a scrape run out across N workers, each with its own DB session and
its own scraper backend — and since every backend picks a random sticky
session id, each worker scrapes from a distinct residential IP. Dates are
interleaved across workers so each covers a spread of the range. Cuts a
150-date batch from ~30-45 min to ~12-15 min at 3 workers.
- booking_scraper_concurrency config key (default 3); Settings → Scraper
Proxy has a "Parallel workers" field
- Forced to 1 when the proxy is off (N workers would share one IP and
hammer it) or when there's a single date
- Both manual and queue paths routed through _scrape_dates_concurrent;
workers mark their own queue items
- Per-worker rotate-on-block replaces the old global pause-on-block
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The location_search_url column existed but was dead — the scraper always
rebuilt the URL from the free-text location name. Now the Location
Configuration form takes a "Booking.com search URL" field: paste the
address-bar URL from a real search and the scraper lifts ss/dest_id/
dest_type from it (the most reliable destination pin). Falls back to
dest_id, then plain name. Server also extracts dest_id from the URL for
the column and derives a display name from ss when none is typed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Free-text ss= destination resolution is non-deterministic: tonight's
30-day run resolved "Stow on the Wold" to St. Wolfgang, Austria for 8 of
30 dates, saving Salzkammergut hotel rates into the matrix. dest_id +
dest_type=city in the search URL pins the destination.
- booking_scrape_config gains a dest_id column (idempotent ALTER)
- scrape_location_search/_build_search_url thread dest_id through
- /config/location accepts dest_id
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- New "Scraper Proxy" tab: enable toggle, host/port/username/password/
country, Save, and a Test Connection button that reports the live exit
IP + country through the proxy
- Backend proxy config now lives in system_config (DB authoritative when
booking_proxy_enabled is set; BOOKING_PROXY_* env vars are the fallback)
- Dedicated /config/proxy GET/POST/test endpoints; password is write-only
(never returned, blank keeps the stored value) and masked in /config/system
- Surface proxy status keys in the read-only System tab
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Route through DataImpulse residential proxy when BOOKING_PROXY_* env
vars are set (empty = direct connection, unchanged behaviour)
- Sticky one IP per session via DataImpulse sessid; rotate_session()
burns the IP for a fresh one
- Rotate-on-block: retry a date on a new IP when the page is challenged
or page 1 renders zero hotels (up to 3 IPs; single attempt sans proxy)
- Persistent, pre-warmed context across dates so cache stays hot
(~5 MB first page, ~0.3 MB per date after) instead of per-date cold loads
- Block images/media/fonts and third-party ad/consent/analytics hosts to
cut bandwidth; never touch the anti-bot challenge (awswaf)
- Faster inter-page pacing now that a burned IP is cheap
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
xvfb-run as container PID 1 hangs before exec'ing uvicorn — its Xvfb
readiness handshake (SIGUSR1) is never delivered to PID 1, so the backend
sat 'Up (unhealthy)' with no logs and the frontend never started.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Tier 1+2 anti-detection to stop Booking.com throttling page 2:
- Run Chromium headful under xvfb (Dockerfile) — headless leaks SwiftShader
WebGL, empty plugins, missing chrome.runtime
- playwright-stealth patches navigator.webdriver/plugins/WebGL vendor
- Single coherent Chrome-121 identity: UA + matching sec-ch-ua client hints +
platform (dropped the Firefox/Safari UA strings — a mismatched UA on a
Chromium engine is a stronger tell than no rotation)
- Homepage warm-up so search requests carry real session cookies + cookie
consent dismiss
- Paginate by clicking next (offset= deep-link was the page-2 tell), offset
URL kept as fallback
- Viewport jitter, mouse movement, longer scroll, selector retry on lazy load
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Concurrent manual scrapes were interleaving (two Chromium sessions on one
LXC) causing the page timeouts behind partial results. SCRAPE_LOCK guards
run_manual_scrape and process_queue; the trigger endpoint returns 409 when
busy, and the frontend keeps the job queued and retries after 30s.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- matrix response gains last_scraped per date (max scraped_at across all
hotels), so column headers show the newest scrape touching the date even
when the visible hotels were on a failed page
- cells >1h older than the column's latest scrape render italic with *
- every cell tooltip now includes the datestamp the price was scraped
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A page-load timeout was logged and skipped, so a scrape could 'succeed'
with only page 1 of results — and the not_listed flagging then marked
every page-2 hotel absent, suppressing their last known rates.
- ScraperResult now tracks pages_requested/pages_ok
- scrape_date skips not_listed flagging when pages failed or when the
scrape saw <60% of the date's 7-day coverage baseline; scraped rates
are still saved, unseen hotels keep last known rate + scrape time
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
