Thoroughbred source research and acquisition status

Last updated: 2026-08-08

Scope and operating rules

This report records the Thoroughbred fork of the jurisdiction research run based on documents/SITE_LINKS.md. Each supplied base site was opened in the in-app browser before transport and route analysis. Research was bounded; sources that did not expose a reproducible pre-race and post-race path after reasonable checks were classified and left without speculative crawler code. Direct public GET was preferred. Robert's approved Racing and Sports proxy was attempted after a direct transport challenge where eligible. Authentication, CAPTCHA, paywalls, and access controls were not bypassed. Supplementary PDFs and other documents are linked but were not downloaded.

FormCrawler captures raw source responses and only parses enough structure to discover acquisition targets. It does not semantically parse the racing data.

Outcome summary

Jurisdiction / source	Status	Pre-race acquisition	Post-race acquisition	Method and notes
Australia / Racing Australia	Implemented and verified	Meeting program, acceptances, gear and scratchings	Meeting results	HTML discovery and capture through existing adapter au_racing_australia.
New Zealand / LoveRacing	Implemented and verified	Race meeting/card targets	Results targets	Public first-party HTML/JSON routes through nz_loveracing.
United Kingdom / Racing Post	Implemented and verified	Racecards	Results	Public HTML through gb_racing_post; bounded URL and content validation.
Korea / KRA	Implemented and verified	Racecards	Results	Public KRA HTML through kr_kra.
Malaysia / Selangor Turf Club	Implemented and verified	Meeting/racecard pages	Results pages	Public HTML through my_selangor_turf_club.
Qatar / QREC	Implemented and verified	Thoroughbred races from meeting API	Results from the same lifecycle API	First-party API; short-lived public front-end token is generated at request time and is never stored. Arabian rows are excluded.
Bahrain / Bahrain Turf Club	Implemented and verified	Racecards/results archive	Same archive	Supplied BHRA URL was unavailable; replacement first-party Bahrain Turf Club HTML is used by bh_bhra. WAHO rows are excluded.
South Africa / Gold Circle	Document-only	Racecard example	No reproducible machine-readable result archive found	Direct site was Cloudflare-protected; approved proxy returned HTML. Calendar navigation and racecard PDFs were found, but no paired machine archive, so no adapter. Statistics and media are also present.
South Africa / Gallop	Implemented and verified	Fixture, meeting and racecard APIs	Result lifecycle when published	First-party sahorseform.co.za APIs through the approved proxy in za_gallop.
Turkey / TJK	Implemented and verified	Daily/city meeting program HTML	Daily/city results HTML	Public HTML through tr_tjk; foreign and Karma meetings are excluded, and the meeting must identify Thoroughbred racing. Mixed Thoroughbred/Arabian meeting payloads remain raw.
Japan / JBIS	Research-only	No public racecard/calendar route found within the bound	Result example	Excellent result and horse-profile history, but not a paired acquisition source.
Japan / Umanity JRA	Implemented and verified	Date/race racecard HTML	Date/race results HTML	Stable public HTML through jp_umanity.
Japan / Umanity NAR	Implemented and verified	NAR date/race racecard HTML	NAR results HTML	Stable public HTML through jp_umanity_nar.
Japan / NAR Keiba	Research-only: outage	Supplied production, beta and mobile pages returned first-party 404	Same	No adapter while the official routes are unavailable.
Germany / German Racing	Research-only: transport failure	Not reached	Not reached	TLS certificate-authority failure occurred directly and through the proxy. Deutscher Galopp already covers Germany.
Germany / Deutscher Galopp	Implemented and verified	Calendar and meeting discovery	Meeting and race results	Existing de_deutscher_galopp HTML adapter.
Italy / SNAI	Research-only	Live partenti pages exist	Result archive exists	Race and horse URLs carry short-lived signatures and no stable historical pre-race route was verified. Ratings, odds, jockey/trainer/owner statistics, video and PDFs are available.
Belgium / BGalopp	Document-only	Meeting index links conditions/program PDFs	Same index links result PDFs	Paired documents are useful for review, but were not downloaded and no adapter was added.
Argentina / Stud Book	Implemented and verified	Meeting pages	Meeting results	Existing ar_stud_book adapter.
Chile / AMR	Research-only: authentication required	Programs require sign-in	Results require sign-in	Public calendar shell did not expose usable rows. No authentication bypass attempted.
Chile / Hipódromo Chile	Implemented and verified	Existing source-specific meeting acquisition	Per-race result JSON	Existing cl_hipodromo_chile adapter also captures a rolling workout snapshot.
Czechia / Dostihy JC	Implemented and verified	Date/meeting/racecard HTML	Results HTML	Stable date routes through cz_dostihy_jc.
Spain / Jockey Club Español	Implemented and verified	Rolling meeting index and combined partants HTML	Combined results/stewards HTML	Historical day shells expire, so es_jockey_club discovers stable event/data links from the rolling jornadas page.
Uruguay / AMR	Research-only: authentication required	Programs require sign-in	Results require sign-in	Same AMR finding as Chile/Brazil.
Uruguay / Maroñas	Implemented and verified	Meeting calendar	Complete meeting result XML	Existing uy_maronas adapter.
Brazil / AMR	Research-only: authentication required	Programs require sign-in	Results require sign-in	Same AMR finding as Chile/Uruguay.
Canada / Equibase	Implemented and verified	Static racecard mirror	Static summary results	Main Equibase domain returned Imperva both directly and through proxy. The official TVG mirror works through ca_equibase.
USA / Equibase	Implemented and verified	Static racecard mirror	Static summary results	Same adapter family through us_equibase.
Sweden / Swedish Horse Racing	Implemented and verified	Calendar, raceday and race JSON	Race JSON/results state	First-party service with required X-Requested-With header through se_swedish_horse_racing; Swedish non-trot meetings only.
Saudi Arabia / JCSA	Aborted after three validation attempts	Raw runner response exists, but pre validator rejected completed-page jockey/trainer link labels	Results artifact verified	Thoroughbred discovery correctly excluded Arabian races. Direct HTTP was successful. Per the three-attempt ceiling, no fourth crawl was run; see Known limitations.
Norway / RaceBets	Research-only	Current/future sportsbook event browser	No robust historical official results archive found	Public SPA requires preference cookies and dynamic event-browser calls. It is a worldwide betting source rather than an official Norwegian archive; bounded research stopped without an adapter.
Mauritius / MTC Jockey Club	Research-only: crawler transport blocked	Fixtures and results and dated racecard articles are visible in the browser	Same fixture pages expose results and result PDFs	Direct crawler GET returned Cloudflare 403; approved proxy returned an error, so no reproducible crawler transport. Supplementary workouts and trials, championships, stipes reports, replays and horse search are available.
Hungary / Kincsem Park	Implemented and verified	/racecards/gallop/YYYY-MM-DD/	/results/gallop/YYYY-MM-DD	Date-addressable public HTML through hu_kincsem_park. The supplied workbook URL was trotting; the site navigation exposed the separate Gallop route.
Morocco / SOREC Galop	Research-only	Calendar and upcoming meetings	Results search exists	Stateful JSF session/postback workflow did not yield a safe stable date-addressable GET contract inside the bound. Also offers horse/jockey performance, meeting programs, VOD, official bulletins, and technical/regulatory documents.
Netherlands / NDR	Not a Thoroughbred source	None	None	The supplied selectieproeven page is Dutch trotting/selection-trial material. It belongs in the harness fork, so no Thoroughbred adapter.
United Arab Emirates / Emirates Racing	Implemented and verified	Date-addressable declarations in /ajax/races-table?date=YYYY-MM-DD	Results in the same complete meeting table	Public first-party HTML through ae_emirates_racing; the adapter requires a Thoroughbred race and preserves mixed meeting content raw. Previous seasons supplies the date index.
Hong Kong / HKJC	Research-only	Current racecard route exists, but historical racecards expire to “no information”	Stable historical results	Monthly fixture calendar and post-race pages are strong, but the paired historical pre-race requirement could not be met. Trackwork, veterinary records, form lines, incidents, replays and sectionals are available.
France / France Galop	Research-only	Today exposes current meetings	Meeting results exist behind opaque meeting identifiers	Date query attempts still returned the current day; no backward date index that can deterministically produce both phases was verified. Program/agenda documents, ratings, breeders, sires and bulletins are available.
India / IndiaRace	Research-only	Current portal data	Archives require interactive selection	No stable deterministic date acquisition contract was verified within the bound. The site does expose equines, stallions, sire and dam searches, profiles, fixtures and archives.
India / Bangalore Turf Club	Implemented and verified	/racing/racecard?d=YYYY-MM-DD	/racing/results?d=YYYY-MM-DD	Date-addressable public HTML through in_bangalore_races. The approved proxy is selected by its launcher after direct certificate/timeout failure. Fixtures, videos, archives, ratings and horse profiles are also available.
India / Hyderabad Race Club	Implemented and verified	Printable racecard with base64-encoded public date token	Printable results with base64-encoded public date token	Deterministic public HTML through in_hydraces. Also exposes calendar, morning trackwork, entries, handicaps, acceptances, declarations, live results, ratings, swimming reports, veterinary remarks and racing incidents.
India / Racing Pulse	Research-only	Fixtures and venue/season archive selector	Archive selector	ASP.NET postback state did not expose a stable GET date contract within the bound. Horse, trainer, jockey, sire and dam profiles are available.
India / Delhi Race Club	Document-only	Entries, handicaps, acceptances and declarations PDFs	Results, stipes report and rating PDFs	Results/archive page and home-page daily program expose documents. Core material is PDF-only, so links are reported and no documents were downloaded.
India / RCTC	Document/current-page only	Home calendar links a racecard PDF	Current detailed result HTML	Archive uses site state/internal identifiers and the pre-race item is a PDF. No paired date-range adapter. Trackwork, ratings, trainer lots, video, stipes reports and horse profiles are available.

Jurisdictions in the workbook with no supplied Thoroughbred URL remain unstarted by design: Australia (NSW), Australia (QLD), Australia (WA), Austria, Ireland, Peru, Panama, and Mexico.

Breeding and horse-detail examples

“Origin” below means the source displays a country suffix for the horse or parent, not that FormCrawler normalises it.

Source	Sire / dam and origins	Foal year/date	Stable horse detail example
QREC	Yes: sire and dam with country where supplied	Age only on inspected modal; no explicit YOB	No stable URL; runner opens a modal. Example: Ayyal Bjd (QA), sire Heshem (IRE), dam Blue Night (QA).
Bahrain Turf Club	Sire/dam names; parent origins not consistently shown	Age, not explicit YOB	GARSMAN (IRE)
Turkey TJK	Yes, parent origin suffixes where supplied	Exact DOB	Victory Maria — DOB 28 Feb 2024; Tizway / Easy Come Easy Go.
JBIS	Yes, horse/sire/dam origins	Exact DOB	Coconuts Brown (JPN) — DOB 28 Apr 2020; Kitasan Black / Lewers Street.
South Africa Gallop	No breeding fields on inspected profile	Not shown	Example horse search/profile
Czech Dostihy JC	Yes, parent origins	Exact DOB	Example horse 273665
Equibase Canada	Sire/dam; horse jurisdiction shown, parent origin not consistently shown	Exact DOB	Thornfield (ON)
Equibase USA	Sire/dam; horse jurisdiction shown, parent origin not consistently shown	Exact DOB	Warrior of Gold (KY)
Swedish Horse Racing	Not in inspected first-party race/stats modal	Not shown	No stable horse profile URL; API exposes a numeric horse ID only.
Saudi JCSA	Sire/dam names; horse origin suffix; parent origins not consistently shown	Age, not explicit YOB	Application (KSA)
Hungary Kincsem	Yes, horse and parent origins	Year born	First Lady — born 2023; Mokarris (USA) / Falconidae (IRE).
UAE Emirates Racing	Yes, horse and parent origins	Age on profile; no explicit inspected DOB	Nature Watch (USA) — Curlin (USA) / Saucy Dame (USA).
Hong Kong HKJC	Horse origin/age and sire/dam in profile/results	Age/season data; exact foal date not verified on inspected page	Historical horse profile example
Bangalore Turf Club	Racecard shows sire/dam with origin suffixes where supplied	Exact DOB	Infinite Spirit — Phoenix Tower (USA) / Aquila Oculus (IRE), foaled 8 Apr 2017.
Hyderabad Race Club	Racecard shows sire/dam with origin suffixes where supplied	Exact foaling date	Daryoush — Oiseau de Feu (USA) / Ritz, foaled 26 Apr 2023.
RCTC	Yes, parent origin suffixes where supplied	Exact DOB and YOB	Trillionaire — Dali (USA) / Lost In Thought (GB), foaled 5 Feb 2021.
Mauritius MTC	Horse search exists, but breeding/profile fields and stable example URL were not verified	Not verified	Horse search
Morocco SOREC	Horse-performance search exists; breeding availability not verified without stateful interaction	Not verified	No stable example GET verified.
France Galop	Horse, breeder and sire directories exist	Not verified in this run	Horse directory
Racing Pulse	Explicit horse, sire and dam profile searches	Not verified in this run	Archives/profile portal

Sources not listed in this matrix either had their breeding/profile detail already documented by an existing adapter, exposed no stable horse profile in the inspected raw lifecycle response, or were inaccessible/authentication-only. Absence here is recorded as “not verified,” not proof that the source never publishes the field.

New adapters and launchers in this run

• nz_loveracing, gb_racing_post, kr_kra, my_selangor_turf_club, qa_qrec, bh_bhra, za_gallop, tr_tjk, jp_umanity, jp_umanity_nar, cz_dostihy_jc, es_jockey_club, ca_equibase, us_equibase, se_swedish_horse_racing, sa_jcsa, hu_kincsem_park, ae_emirates_racing, in_bangalore_races, and in_hydraces were added or completed during the wider source-research work represented by this report.
• Each adapter uses the shared queue, storage, validation, rate limiter, transport provenance, and parser-handoff workflow.
• Dedicated .bat launchers are under scripts/ for repeatable date-range runs.

Known limitations and stopped work

• Saudi JCSA: stopped after three attempts as requested. Ten Thoroughbred races were correctly discovered for 3 January 2026 and Arabian races 5 and 9 were excluded. Shell, meeting index and Race 1 results artifacts and hashes verified. The pre-race validator looks for visible “jockey” and “trainer” labels, while completed pages expose those identities through /en/jockeys/ and /en/trainers/ links. The raw 84,694-byte response contains the runner data. No fourth attempt was made.
• Hong Kong: historical result routes remain strong, but historical racecard requests now return “no information.” A post-only adapter would not meet this task's paired pre/post requirement.
• Mauritius: browser access succeeded after Cloudflare, but direct crawler and approved-proxy transports did not provide a reproducible response. Browser session state was not converted into a crawler bypass.
• PDF/document sources: Gold Circle, BGalopp, Delhi Race Club and RCTC links are reported for review; documents were not downloaded.
• QDG Knowledge Base: this report and per-jurisdiction status pages were published through the approved versioned KnowledgeBaseService fallback. Every write used optimistic version checks and was read back after publication.

Final live validation evidence

• Hungary/Kincsem, 5 October 2025: attempt 2 passed after the shared range date-key integration fix. Three direct HTTP 200 artifacts (discovery, pre, post), 0 retries/failures; 88,155-byte pre SHA-256 b93465e2dd2d36bd2e29c7ea17212726cf220b6e7c206d1439dd0a1beb4c63c8; 122,893-byte post SHA-256 3344abc7b502db3e56b070713de65f2c78689a1b4ea94c3ce0b422c5f30f4029.
• UAE/Emirates Racing, 28 March 2026: attempt 2 passed after the same integration fix. Three direct HTTP 200 artifacts, 0 retries/failures; combined 42,280-byte lifecycle table SHA-256 30d81a23de69210913f29a207176d82c54995266aa51bb8f35d88c8811d5dc08.
• Bangalore, 19 July 2026: attempt 3 passed through the approved proxy after direct certificate/timeout failure. Three proxy-provenance HTTP 200 artifacts, 0 retries/failures; 206,583-byte pre SHA-256 00ffc4d4c966f109acb7008a3276811bd5ca170769be2ca008584711838d6b49; 338,489-byte post SHA-256 7fbcd2b6f858f4888bc06879cf65d5ed15a7ce3ec556ca494043ebc9415e7cba.
• Hyderabad, 3 August 2026: attempt 1 passed. Three direct HTTP 200 artifacts, 0 retries/failures; 317,832-byte pre SHA-256 b1586f7b0954abc6845ade51feede5b378ebf22ed16ee678ad97f35d005294d8; 127,322-byte post SHA-256 7ea376b858a814a9c7368cb527ff0097b3db6e240455fb51ec05eb05a248bba0.
• For all four passing roots, recomputed content hashes equal the queue ledger and metadata, artifact IDs match, and every FormParser handoff exists with matching artifact/content/metadata URIs.
• Final automated gate: 152 tests passed with ResourceWarning promoted to an error; all Python modules compile and git diff --check passes.