v0.2.0
A deep data-model audit: richer product and card records, canonical image URLs, and fuller chart paging.
A pass over every parsed surface to capture the fields Amazon shows but earlier versions dropped, plus one canonical image URL everywhere.
Richer records
- Products gained
savingsandsavings_pctoff the list price,coupon,bought_past_month, anin_stockflag derived from availability, inlinevideos,ships_from, and arankslist that captures every Best Sellers Rank, overall and per subcategory, instead of only the overall one. Availability and image also read from the JSON-LD block now. - Search cards gained
list_price,badge(Amazon's Choice, Best Seller),prime, andbought_past_month.
The data model documents the full shape of every record.
Canonical images
Amazon serves the same photo at dozens of sizes and from several CDN hosts.
amz now strips the size modifier from every image URL and pins one canonical
host, so images holds one full-resolution URL per distinct photo, with
thumbnails, tracking pixels, and sprites removed. The same canonicalization runs
on every surface that carries an image: products, search cards, related cards,
reviews, brand logos and banners, and author photos.
Fuller paging
Charts now page until a page comes back empty rather than stopping at two, and ranks are offset by page so the numbering continues across pages even when Amazon drops the rank badges.
Output
The high-value new columns are surfaced in the table and CSV views; JSON and JSONL already carried the complete record.