SPB Git forge

spb/rareindex

Public
54commits 1branches 0releases
7.1 MBsize
maindefault branch
10 days agolast push
TypeScript 61.9% HTML 37.2% SQL 0.7%
2.6 KB · 37 lines json
Raw Blame History
1{2  "id": "numisbids",3  "displayName": "NumisBids (prices realized across ~370 numismatic auction houses)",4  "sourceId": "numisbids",5  "sourceName": "NumisBids",6  "sourceType": "marketplace",7  "sourceUrl": "https://www.numisbids.com",8  "module": "firecrawl/numisbids",9  "enginePriority": ["firecrawl", "scrapfly"],10  "categories": ["coins", "banknotes", "medals"],11  "regions": ["global", "US", "EU", "GB"],12  "languages": ["en"],13  "currency": ["USD", "EUR", "GBP", "CHF", "AUD", "CAD", "SEK", "NOK", "DKK", "PLN", "CZK", "HKD", "JPY"],14  "supportsListings": false,15  "supportsSold": true,16  "supportsAuctions": false,17  "supportsImages": true,18  "supportsCatalog": false,19  "supportsPopulation": false,20  "supportsLookup": false,21  "refreshFrequencyMinutes": 1440,22  "priority": "high",23  "trustScore": 0.9,24  "attributionRequired": true,25  "termsUrl": "https://www.numisbids.com/terms",26  "accessNotes": "Firecrawl rendering (1 credit per page) because Cloudflare answers plain HTTPS with a 403 challenge; no login, no CAPTCHA solving — the same public pages any visitor sees. robots.txt (User-agent *) only disallows wantlist_add / bidlist / bidsubmit / watchlist paths, none of which are fetched (ClaudeBot and Bytespider are disallowed site-wide; RareIndexBot is not). Pages read: /results (closed sales that publish prices realized → sale id), /sale/<sid> (firm, auction title, closing date, category list) and /sale/<sid>/category/<cid>?pg=N (100 lots per page: lot number, truncated description, estimate, image, 'Price realized: 160 USD'). NumisBids states on every prices-realized page that 'Buyer's premium is not included' → sales are stored as hammer prices (buyer_premium_included=false) in the currency the page prints; lots without a price (unsold, withdrawn) are skipped and 'Large Lots' sections are flagged isBundle. Sale date = the closing date printed in the sale header (first day of multi-day sales). Backfill walks sale ids downwards (≈ 11 000 sales since 2013; ids without published prices cost one page each and are skipped). Not fetched: individual lot pages (only reached via the category listing; the lot URL is kept as sourceUrl), bidding, accounts, want-list searches. Attribution: NumisBids + the consigning auction house (auctionHouse field). 2.5 s politeness, concurrency 1.",27  "enabled": true,28  "schemaVersion": "1.0",29  "historicalDepth": "decades",30  "acquisitionMethod": "Firecrawl rendered HTML (sale category listings)",31  "config": {32    "salesPerRun": 2,33    "pagesPerRun": 30,34    "skipSections": ["Numismatic Literature", "Literature", "Books", "Antiquities"]35  }36}37