detect.js

// detect.js
// Shared logic: parse an Amazon product page (HTML string) and decide whether the
// item is currently eligible for FREE shipping to the signed-in delivery address.
//
// This runs in the background page (DOMParser available) and in the popup (for
// parseAsin). It intentionally works on raw HTML so the same code path handles
// both a background fetch() and the outerHTML of a rendered fallback tab.

var FreeShipDetect = (function () {
  "use strict";

  // Defaults are tuned for amazon.com with a delivery address set to Israel.
  // Every one is overridable from Settings so the single user can re-tune if
  // Amazon changes its wording, without touching code.
  var DEFAULTS = {
    // A destination-scoped free-shipping line, e.g. "FREE Shipping to Israel",
    // "FREE International Shipping", "FREE delivery".
    freeRegex: "free\\s+(international\\s+)?(shipping|delivery)",
    // A real shipping charge line, e.g. "$12.98 Shipping to Israel".
    // Deliberately NOT matching "Import Fees Deposit" — those can apply even when
    // shipping itself is free, so they must not count as a shipping charge.
    chargeRegex: "\\$\\s?\\d[\\d,]*\\.\\d{2}[^.$|]{0,30}shipping",
    // The item can't be delivered here at all right now.
    unavailableRegex:
      "cannot be shipped to|does(n't| not) (currently )?ship to|item cannot be shipped|this item does not ship"
  };

  // Blocks on the product page that carry delivery / shipping-cost wording.
  var SELECTORS = [
    "#deliveryBlockMessage",
    "#mir-layout-DELIVERY_BLOCK",
    "#amazonGlobal_feature_div",
    "#globalStoreBadgePopoverInsideBuybox",
    "#contextualIngressPtLabel",
    "#unifiedDeliveryMessage",
    "#exportsCosts_feature_div",
    "#deliveryPromiseInsideBuyBox_feature_div",
    "#desktop_buybox",
    "#buybox"
  ];

  var PRICE_SELECTORS = [
    "#corePriceDisplay_desktop_feature_div .a-price .a-offscreen",
    "#corePrice_feature_div .a-price .a-offscreen",
    "#priceblock_ourprice",
    "#priceblock_dealprice",
    "#price_inside_buybox",
    ".a-price .a-offscreen"
  ];

  function firstText(doc, selectors) {
    for (var i = 0; i < selectors.length; i++) {
      var el = doc.querySelector(selectors[i]);
      if (el && el.textContent && el.textContent.trim()) return el.textContent.trim();
    }
    return null;
  }

  function deliveryText(doc) {
    var parts = [];
    for (var i = 0; i < SELECTORS.length; i++) {
      var el = doc.querySelector(SELECTORS[i]);
      if (el) {
        var t = el.textContent.replace(/\s+/g, " ").trim();
        if (t) parts.push(t);
      }
    }
    // Drop generic marketing like "FREE Shipping on orders over $49" so it can't
    // masquerade as a real per-item free-shipping signal.
    return parts
      .join("   |   ")
      .replace(/free\s+(international\s+)?shipping\s+on\s+(all\s+)?orders?[^.|]*/gi, " ");
  }

  // The main product photo. Works on both the static fetched HTML (data-dynamic
  // JSON / og:image) and a rendered tab (resolved src). Skips lazy-load data:
  // URIs so we never store a blank placeholder pixel.
  function extractImage(doc) {
    var el =
      doc.querySelector("#landingImage") ||
      doc.querySelector("#imgTagWrapperId img") ||
      doc.querySelector("#main-image") ||
      doc.querySelector("#imgBlkFront");
    if (el) {
      var hires = el.getAttribute("data-old-hires");
      if (hires && !/^data:/.test(hires)) return hires;
      var dyn = el.getAttribute("data-a-dynamic-image");
      if (dyn) {
        try {
          var k = Object.keys(JSON.parse(dyn))[0];
          if (k && !/^data:/.test(k)) return k;
        } catch (e) {
          /* ignore */
        }
      }
      var src = el.getAttribute("src");
      if (src && !/^data:/.test(src)) return src;
    }
    var og = doc.querySelector('meta[property="og:image"]');
    if (og) {
      var c = og.getAttribute("content");
      if (c && !/^data:/.test(c)) return c;
    }
    return null;
  }

  function isRobotPage(html) {
    return /Robot Check|Enter the characters you see|api-services-support@amazon\.com|To discuss automated access|validateCaptcha/i.test(
      html
    );
  }

  function snippetAround(text, regexes) {
    for (var i = 0; i < regexes.length; i++) {
      var m = regexes[i].exec(text);
      if (m) {
        var start = Math.max(0, m.index - 30);
        var end = Math.min(text.length, m.index + m[0].length + 50);
        return (start > 0 ? "…" : "") + text.slice(start, end).trim() + (end < text.length ? "…" : "");
      }
    }
    return null;
  }

  // Core decision. Returns:
  //   { ok, status, freeShip, title, price, snippet }  on success
  //   { ok:false, reason }                             when the page couldn't be read
  // status ∈ "free" | "paid" | "unavailable" | "unknown"
  function detect(html, opts) {
    opts = opts || {};
    if (!html || html.length < 120) return { ok: false, reason: "empty" };
    if (isRobotPage(html)) return { ok: false, reason: "robot" };

    var free = new RegExp(opts.freeRegex || DEFAULTS.freeRegex, "i");
    var charge = new RegExp(opts.chargeRegex || DEFAULTS.chargeRegex, "i");
    var unavailable = new RegExp(opts.unavailableRegex || DEFAULTS.unavailableRegex, "i");

    var doc = new DOMParser().parseFromString(html, "text/html");
    var title = firstText(doc, ["#productTitle", "#title"]);
    var price = firstText(doc, PRICE_SELECTORS);
    var text = deliveryText(doc);

    if (!title && !text) return { ok: false, reason: "noparse" };

    var status, freeShip = false;
    if (free.test(text)) {
      status = "free";
      freeShip = true;
    } else if (unavailable.test(text)) {
      status = "unavailable";
    } else if (charge.test(text)) {
      status = "paid";
    } else {
      status = "unknown";
    }

    var snippet = snippetAround(text, [free, unavailable, charge]) || text.slice(0, 160).trim();

    return {
      ok: true,
      status: status,
      freeShip: freeShip,
      title: title ? title.replace(/\s+/g, " ").trim() : null,
      price: price || null,
      snippet: snippet || null,
      image: extractImage(doc)
    };
  }

  // Pull a 10-char ASIN out of a URL, a full paste, or a bare ASIN.
  function parseAsin(input) {
    if (!input) return null;
    input = String(input).trim();
    if (/^[A-Z0-9]{10}$/i.test(input)) return input.toUpperCase();
    var m =
      input.match(/\/(?:dp|gp\/product|product|gp\/aw\/d|gp\/offer-listing)\/([A-Z0-9]{10})/i) ||
      input.match(/[?&]asin=([A-Z0-9]{10})/i) ||
      input.match(/\/([A-Z0-9]{10})(?:[/?#]|$)/i);
    return m ? m[1].toUpperCase() : null;
  }

  function canonicalUrl(asin) {
    return "https://www.amazon.com/dp/" + asin + "?psc=1";
  }

  return {
    detect: detect,
    parseAsin: parseAsin,
    canonicalUrl: canonicalUrl,
    DEFAULTS: DEFAULTS
  };
})();

if (typeof module !== "undefined") module.exports = FreeShipDetect;