Bagazhnikpro Products Without a Photo

Purpose

Enumerate every product in the bagazhnikpro.store catalog whose listing card has no product photo (the card renders the grey "Нет Изображения" / no_image_ru.png placeholder instead of a real image) and export the result as an Excel workbook. For each no-photo product the skill returns its internal id, SKU (артикул), title, category, and canonical product URL. Read-only — it only reads public catalog pages; it never logs in, edits, or touches the cart.

When to Use

• A merchandising / content audit: "which products are missing a photo so we can shoot/upload images?"
• Producing a hand-off spreadsheet (.xlsx) of incomplete product cards for a catalog manager.
• Any periodic data-quality check over the full bagazhnikpro.store catalog (~25.8k products across ~50 leaf categories).
• As a template for the sibling shops on the same engine (e.g. autossport.store), which share the identical "hotengine" CMS markup.

Workflow

bagazhnikpro.store runs the "hotengine" CMS (LiteSpeed/CyberPanel). No anti-bot, no proxy, no stealth needed (the homepage anti-bot probe reported none detected, and ClaudeBot/GPTBot are explicitly Allowed in robots.txt), so the pages are plain static HTML fetchable by any client. Via Browserless, drive each category page with browserless_agent — a goto (waitUntil: "load", no proxy arg) then an evaluate that parses the grid in-page and returns only the no-photo rows (a compact projection — don't ship raw page HTML back; the text return caps ~200k chars). The one non-obvious trick: a bare category URL (/ru/pers_shop/<cat>/) returns a JS-only skeleton with zero products — but appending the pagination query params ?n=<page>&NUM_ONPAGE=<count> makes the server render the full product grid as static HTML in the response. That single insight turns the whole task into deterministic in-page HTML parsing.

1. Collect the leaf category slugs. Fetch the homepage https://bagazhnikpro.store/ru/ (note: https://bagazhnikpro.store/ 301-redirects to /ru/) and harvest every link matching /ru/pers_shop/<slug>/. ~68 such links exist, but many are parent groupings that render a subcategory list and 0 products (e.g. bagajniki, deflektori, kovriki, lampy, farkopi) — that's fine, they simply yield no rows. The leaf categories (e.g. poperechininareylingi, zashitakarteradvigatelya, organayzery, deflektorinaoknavetroviki) are where products live.

2. Page through each category server-side. For each category slug, fetch:

   https://bagazhnikpro.store/ru/pers_shop/<slug>/?n=<N>&NUM_ONPAGE=96

   starting at N=0 and incrementing until a page returns fewer than NUM_ONPAGE product blocks (or zero). Each page is ~430 KB at NUM_ONPAGE=96. Do not crank NUM_ONPAGE arbitrarily high to grab a whole big category in one shot — large categories (e.g. bagajnikinakrishyavtomobilyavsbore has 5,856 products, deflektorinaoknavetroviki 3,203, porogi 2,637) will blow past the fetch body-size limit. 96/page is a safe sweet spot.

3. Parse product cards. Each card is a <div class="hotengine-shop-product-list-block" ...> carrying:

   • data-hotengine-marking-shop_catalog_page_id="<id>" → product id
   • data-hotengine-marking-shop_catalog_page_sku="<sku>" → SKU / артикул
   • title in .hotengine-shop-product-title > h4 > a (also in the anchor title= attr)
   • the listing image <img class="hotengine-shop-products-list-img" ...>

4. Detect "no photo" — the definitive signal. A card with a photo has <img class="hotengine-shop-products-list-img" src="/upload/shop_catalog/s9060/<id>/<id>_0.small_cachedx300.{webp|jpg}">. A card without a photo has an extra class and a fixed placeholder src:

   <img
     class="hotengine-shop-products-list-img hotengine-shop-products-list-img-empty"
     loading="lazy"
     src="/img/shop/no_image_ru.png"
   />

   So a product is "no photo" iff either its listing <img> carries the class hotengine-shop-products-list-img-empty or its src is /img/shop/no_image_ru.png (equivalently: the src does not start with /upload/shop_catalog/). The CSS selector div.hotengine-shop-product-list-block img.hotengine-shop-products-list-img-empty matches exactly the no-photo cards.

5. Collect & de-duplicate by product id (a product can appear under more than one category path), decode HTML entities in the title/SKU (", &,  , «/»), and build the canonical URL https://bagazhnikpro.store/ru/pers_shop/<slug>/<id>.htm.

6. Write the Excel file. Emit one row per no-photo product with columns № | ID товара | Артикул (SKU) | Название | Категория | Ссылка to an .xlsx (e.g. via the xlsx/SheetJS library, or any spreadsheet writer). Over the full catalog this run produced 43 no-photo products out of 25,817 (≈0.17%).

If server-side rendering breaks

If the ?n=&NUM_ONPAGE= server-render ever stops returning the grid in the initial response, let the page hydrate client-side: after the goto, add { "method": "waitForSelector", "params": { "selector": "div.hotengine-shop-product-list-block", "timeout": 10000 } } (or a waitForTimeout), then run the same evaluate — document.querySelectorAll("div.hotengine-shop-product-list-block"), flagging any whose img.hotengine-shop-products-list-img-empty exists (or whose img.src ends in no_image_ru.png). Still no proxy arg; this site needs no stealth. Waiting for hydration is slower per page and only needed as a safety net.

Site-Specific Gotchas

• Bare category pages are empty to a fetch. GET /ru/pers_shop/<slug>/ (no query string) returns the page chrome with 0 product blocks — the grid is hydrated client-side. You must add ?n=0&NUM_ONPAGE=<n> to get server-rendered products. This is the single thing that traps naive scrapers.
• / redirects to /ru/. The root issues a 301 to /ru/. The shop is multilingual (/ru/, /en/, /ua/, /pl/); the placeholder filename is locale-specific — Russian uses /img/shop/no_image_ru.png. On other locales expect no_image_<lang>.png; match on the -empty class (locale-agnostic) rather than the exact filename if you crawl non-RU trees.
• Two reliable no-photo signals, prefer the class. img.hotengine-shop-products-list-img-empty (added only to placeholder images) is the most robust signal; src="/img/shop/no_image_ru.png" is the secondary one. Photo cards carry the class without the -empty suffix and a /upload/shop_catalog/... src. (Caution when regex-parsing: matching the literal class="hotengine-shop-products-list-img" with a closing quote will miss photo cards' true class only if you assume it's unique — the -empty cards have a second class token after a space.)
• Parent vs. leaf categories. Of ~68 /ru/pers_shop/<slug>/ links, roughly a third are parent groupings that render a subcategory list and zero products. They're harmless (0 rows) but don't mistake "0 products" for "0 no-photo products" on those — you simply haven't reached the leaves.
• De-dupe by id. The same product id can surface in multiple category paths; key your result set on data-hotengine-marking-shop_catalog_page_id to avoid double-counting.
• Watch the return-size cap on big categories. Parsing the grid in-page and returning only the no-photo rows keeps output small — but don't return raw page HTML: the browserless_agent/browserless_function text return caps at ~200k chars and a whole big-category page (~430 KB at 96/page) would overflow it. Keep NUM_ONPAGE ≤ ~96 and paginate.
• No anti-bot, no proxy required. Verified: a plain request (no proxy arg, no stealth) returns full server-rendered product HTML; a bare browserless_agent session confirmed the same. Don't waste time/money on stealth or residential proxies here.
• Correlation, not a rule: in this snapshot all 43 no-photo products were also "Нет в наличии" (out of stock), and their embedded schema.org JSON-LD showed availability: OutOfStock. Out-of-stock is not a substitute signal for "no photo" — always test the image, not the stock state.
• robots.txt Crawl-delay: 1 — keep to ≤ 1 request/second to stay polite; the full catalog crawl is a few hundred fetches.

Expected Output

An .xlsx workbook (sheet "Товары без фото") with one row per no-photo product. The underlying data shape:

{
  "site": "bagazhnikpro.store",
  "scanned_products": 25817,
  "no_photo_count": 43,
  "detection_signal": "img.hotengine-shop-products-list-img-empty  /  src=/img/shop/no_image_ru.png",
  "no_photo_products": [
    {
      "id": "326370",
      "sku": "PSV 123323",
      "title": "Набор автомобилиста \"Аварийный\" (огнетушитель, знак, аптечка )",
      "category": "organayzery",
      "url": "https://bagazhnikpro.store/ru/pers_shop/organayzery/326370.htm"
    },
    {
      "id": "99322",
      "sku": "м.00230",
      "title": "Защита MOTODOR радиатора BMW 6 (e63) 2003-2007 куп",
      "category": "zashitakarteradvigatelya",
      "url": "https://bagazhnikpro.store/ru/pers_shop/zashitakarteradvigatelya/99322.htm"
    },
    {
      "id": "327185",
      "sku": "Rola 59504",
      "title": "Багажник-корзина ROLA на поперечены 121,9x95,2x10,2 см",
      "category": "korziniiplatformi",
      "url": "https://bagazhnikpro.store/ru/pers_shop/korziniiplatformi/327185.htm"
    }
  ]
}

Excel column layout (one row per product):

If a run finds no missing photos in the scanned scope, return no_photo_count: 0 with an empty no_photo_products array and still emit an (empty-bodied) workbook with the header row.