# Retail platforms. # # kind marketplace | national_chain | tn_regional # region national | TN (TN = a Tamil Nadu retail chain) # policy serp_only -> NEVER fetched directly; everything comes from web # search results (titles, snippets, image results) # probe -> fetched only if the site probe grades it A or B # (robots.txt allows, HTTP 200, no CAPTCHA); otherwise # it falls back to search results like serp_only # product_url regex a URL must match to count as a single product page. # Group 1, when present, is the site's own product id. # pincode_param optional query parameter the site accepts for a delivery # pincode. Only sites that actually honour it get # pincode_applied=true on their prices. # # Brand official sites are generated from brands.yaml (kind brand_official). sites: - domain: amazon.in name: Amazon.in kind: marketplace region: national policy: serp_only product_url: '/(?:dp|gp/product)/([A-Z0-9]{10})' - domain: flipkart.com name: Flipkart kind: marketplace region: national policy: serp_only product_url: '/p/(itm[0-9a-z]+)' - domain: croma.com name: Croma kind: national_chain region: national policy: probe product_url: '/p/(\d{5,})' - domain: reliancedigital.in name: Reliance Digital kind: national_chain region: national policy: probe product_url: '(?:/p/|/product/[^?#]*?-)(\d{6,})' - domain: vijaysales.com name: Vijay Sales kind: national_chain region: national policy: probe product_url: '/p/(?:P?)(\d{3,})/' - domain: tatacliq.com name: Tata CLiQ kind: marketplace region: national policy: probe product_url: '/p-(mp\d+)' - domain: poorvika.com name: Poorvika kind: tn_regional region: TN policy: probe product_url: '/([a-z0-9-]{8,})/p/?$' - domain: sangeethamobiles.com name: Sangeetha Mobiles kind: tn_regional region: TN policy: probe product_url: '(?i)/product-?details/(?:[^/?#]+/)?(\d+)' - domain: vasanthandco.in name: Vasanth & Co kind: tn_regional region: TN policy: probe product_url: '/(?:product|products)/([a-z0-9-]{8,})' - domain: viveks.com name: Viveks kind: tn_regional region: TN policy: probe product_url: '/([a-z0-9-]{8,})\.html$'