wlib · August 6, 2025 01:42 · Aug 6, 2025
diff --git a/scrape-amazon-orders.mjs b/scrape-amazon-orders.mjs
@@ -0,0 +1,198 @@
+const ordersUrl = "https://www.amazon.com/your-orders/orders"
+
+const wait = (ms = 1_000) =>
+  new Promise(resolve => {
+    setTimeout(() =>{
+      resolve()
+    }, ms)
+  })
+
+const fetchDocument = async url => {
+  const html = await fetch(url).then(res => res.ok ? res.text() : undefined)
+  if (!html)
+    return
+
+  return Document.parseHTMLUnsafe(html)
+}
+
+const fetchDocumentViaIframe = async url =>
+  new Promise(resolve => {
+    const iframe = document.createElement("iframe")
+    iframe.addEventListener("load", () => {
+      const html = iframe.contentDocument.documentElement.outerHTML
+      resolve(Document.parseHTMLUnsafe(html))
+    })
+    iframe.src = url
+  })
+
+const ordersPage = await fetchDocument(ordersUrl)
+
+const ordersUrlsByYear =
+  [...ordersPage.querySelector("select[name = timeFilter]").options]
+    .flatMap(option => {
+      if (!option.value.startsWith("year-"))
+        return []
+
+      const url = new URL(ordersUrl)
+      url.searchParams.set("timeFilter", option.value)
+      return [url.href]
+    })
+
+const orderIds = new Set()
+
+const extractOrderIds = (doc, baseUrl) => {
+  for (const a of doc.querySelectorAll("a")) {
+    let url
+    try {
+      url = new URL(a.href, baseUrl)
+    }
+    catch {
+      continue
+    }
+
+    if (!url.href.startsWith("https://www.amazon.com/gp/css/summary/print.html"))
+      continue
+
+    const orderId = url.searchParams.get("orderID")
+    if (!orderId)
+      continue
+
+    orderIds.add(orderId)
+  }
+}
+
+for (const ordersPageUrl of ordersUrlsByYear) {
+  const ordersPage = await fetchDocumentViaIframe(ordersPageUrl)
+  await wait()
+
+  extractOrderIds(ordersPage, ordersPageUrl)
+
+  const maxStartIndex = [...ordersPage.querySelectorAll(".a-pagination a")]
+    .flatMap(a => {
+      let url
+      try {
+        url = new URL(a.href, ordersPageUrl)
+      }
+      catch {
+        return []
+      }
+
+      if (!url.href.startsWith(ordersUrl))
+        return []
+
+      const startIndex = parseInt(url.searchParams.get("startIndex"))
+      if (isNaN(startIndex))
+        return []
+
+      return [startIndex]
+    })
+    .sort((a, b) => b - a)[0]
+  if (!maxStartIndex)
+    continue
+
+  const otherPageUrls = Array.from({ length: maxStartIndex / 10 })
+    .map((_, i) => {
+      const startIndex = (i + 1) * 10
+      const url = new URL(ordersPageUrl)
+      url.searchParams.set("startIndex", startIndex)
+      return url.href
+    })
+
+  for (const otherPageUrl of otherPageUrls) {
+    const otherPage = await fetchDocumentViaIframe(otherPageUrl)
+    await wait()
+
+    extractOrderIds(otherPage, ordersPageUrl)
+  }
+}
+
+const extractModernInvoiceInfo = (invoicePage, baseUrl) => {
+  const orderDate = invoicePage.querySelector("[data-component = orderDate]")?.textContent.trim()
+  const items = [...invoicePage.querySelectorAll("[data-component = purchasedItems] .a-fixed-left-grid")]
+    .map(purchasedItem => {
+      const imageContainer = purchasedItem.querySelector("[data-component = itemImage]")
+      let image
+      try {
+        image = new URL(imageContainer?.querySelector("img")?.getAttribute("src"), baseUrl).href
+      }
+      catch {}
+      let quantity = parseInt(imageContainer?.querySelector(".od-item-view-qty")?.textContent.trim())
+      if (isNaN(quantity))
+        quantity = undefined
+
+      const titleLink = purchasedItem.querySelector("[data-component = itemTitle] a")
+      const title = titleLink?.textContent.trim()
+      let url
+      try {
+        const url_ = new URL(titleLink.href, baseUrl)
+        url_.search = ""
+        url = url_.href
+      }
+      catch {}
+
+      const unitPrice = purchasedItem.querySelector("[data-component = unitPrice] .a-offscreen")?.textContent.trim()
+
+      return { image, title, url, unitPrice, quantity }
+    })
+
+  return { orderDate, items }
+}
+
+const extractLegacyInvoiceInfo = (invoicePage, baseUrl) => {
+  const orderDateCell = [...invoicePage.querySelectorAll("td:not(:has(td))")]
+    .find(td => td.textContent.includes("Order Placed:"))
+  const orderDate = orderDateCell?.textContent.replace("Order Placed:", "").trim()
+
+  const rawItems = [...invoicePage.querySelectorAll("tr")]
+    .flatMap(row => {
+      const italicText = row.querySelector("i")
+      const priceCell = row.querySelector("td[align = right]")
+      if (!(italicText && priceCell && priceCell.textContent.includes("$")))
+        return []
+
+      let quantity = parseInt(italicText.previousSibling?.textContent.match(/(\d+)\s+of:/)?.[1])
+      if (isNaN(quantity))
+        quantity = 1
+
+      const title = italicText.textContent.trim()
+
+      const unitPrice = priceCell.textContent.trim()
+
+      return [{
+        title,
+        unitPrice,
+        quantity
+      }]
+    })
+
+  const itemsMap = new Map()
+
+  for (const item of rawItems) {
+    const key = `${item.title}\0${item.unitPrice}`
+
+    if (itemsMap.has(key))
+      itemsMap.get(key).quantity += item.quantity
+    else
+      itemsMap.set(key, { ...item })
+  }
+
+  const items = [...itemsMap.values()]
+
+  return { orderDate, items }
+}
+
+const orderIdToInfo = {}
+
+for (const orderId of orderIds) {
+  const invoiceUrl = new URL("https://www.amazon.com/gp/css/summary/print.html")
+  invoiceUrl.searchParams.set("orderID", orderId)
+  const invoicePage = await fetchDocument(invoiceUrl)
+  await wait()
+
+  orderIdToInfo[orderId] =
+    invoicePage.querySelector("[data-component = orderDate]")
+      ? extractModernInvoiceInfo(invoicePage, invoiceUrl)
+      : extractLegacyInvoiceInfo(invoicePage, invoiceUrl)
+}
+
+console.log(orderIdToInfo)