add auto update ozon_wb_dpr

scripts/update.js

@@ -2,6 +2,7 @@ import fs from "node:fs/promises";
 import { asyncMap } from "modern-async";
 import { JSDOM } from "jsdom";
 import { getFinalURL } from "./utils.js";
+import ozonWbDpr from "./update/ozon-wb-dpr.js";
 
 async function woyag() {
   const apiResponse = await fetch("https://login.woyag.ru/ajax/pvz-list");
@@ -84,6 +85,7 @@ async function sevenDostavka() {
 async function main() {
   await woyag();
   await sevenDostavka();
+  await ozonWbDpr();
 }
 
 main();

scripts/update/ozon-wb-dpr.js

@@ -0,0 +1,124 @@
+import { distance } from "fastest-levenshtein";
+import { getFinalURL, getTelegramMessage } from "../utils.js";
+import { JSDOM } from "jsdom";
+import { asyncMap } from "modern-async";
+import fs from "node:fs/promises";
+
+const MAIN_URL = "https://t.me/ozon_wb_dpr/627";
+
+const QR_FOR_PVZ_STRING = "QR для ПВЗ";
+
+const closestObj = (str, arr, key) => {
+  let min_distance = Infinity;
+  let min_index = 0;
+  for (let i = 0; i < arr.length; i++) {
+    const dist = distance(str, arr[i][key]);
+    if (dist < min_distance) {
+      min_distance = dist;
+      min_index = i;
+    }
+  }
+  return arr[min_index];
+};
+
+async function getPVZFromPost(post) {
+  const message = await getTelegramMessage(post);
+  const dom = new JSDOM(message.text);
+  const document = dom.window.document;
+
+  const linkElement = document.querySelector(
+    'a[href*="vk.cc"],a[href*="ozon.ru"]'
+  );
+
+  const link = await getFinalURL(linkElement["href"]).then((u) => {
+    const final = new URL(u);
+    final.search = "";
+    return final.toString();
+  });
+
+  return link;
+}
+
+async function getFromTelegram() {
+  const message = await getTelegramMessage(MAIN_URL);
+
+  const dom = new JSDOM(message.text);
+  const document = dom.window.document;
+
+  const links = document.querySelectorAll("a");
+
+  return (
+    await asyncMap(links, async (link) => {
+      const textContent = link.textContent;
+
+      if (textContent.startsWith(QR_FOR_PVZ_STRING)) {
+        const address = textContent.replace(QR_FOR_PVZ_STRING, "").trim();
+
+        let postLink = link["href"];
+        // TEMP FIX
+        if (address === "г.Донецк, ул.Университетская, 76") {
+          postLink = "https://t.me/ozon_wb_dpr/774";
+        }
+
+        return {
+          name: `ПВЗ ${address}`,
+          address,
+          link: await getPVZFromPost(postLink),
+          operationTime: "пн-вс с 9:00 до 17:45",
+        };
+      }
+    })
+  ).filter(Boolean);
+}
+
+async function getFromSite() {
+  const res = await fetch("https://ozon-wb-dpr.ru/");
+  const html = await res.text();
+
+  const dom = new JSDOM(html);
+  const document = dom.window.document;
+
+  const scripts = document.querySelectorAll("script");
+
+  for (let script of scripts) {
+    if (script.textContent.includes('descr: "Бесплатный озон')) {
+      let x = script.textContent.split("[\n{\n").pop().split("\n},\n]")[0];
+
+      if (x) {
+        const points = new Function(`return [{${x}}]`)();
+        return points;
+      }
+    }
+  }
+}
+
+async function ozonWbDpr() {
+  const telegramPoints = await getFromTelegram();
+  let points = (await getFromSite()).map((point) => ({
+    coordinates: [parseFloat(point["lng"]), parseFloat(point["lat"])],
+    name: point.title,
+  }));
+
+  points = points.map((point) => {
+    const obj = closestObj(point.name, telegramPoints, "name");
+    return {
+      coordinates: point.coordinates,
+      ...obj,
+    };
+  });
+
+  await fs.writeFile(
+    "data/03_ozon-wb-dpr.json",
+    JSON.stringify(
+      {
+        name: "ПВЗ ДНР",
+        source: MAIN_URL,
+        points,
+      },
+      undefined,
+      4
+    )
+  );
+}
+
+export default ozonWbDpr;

scripts/utils.js

@@ -1,23 +1,65 @@
+import { JSDOM } from "jsdom";
+
+async function resolveVKcc(url) {
+  const response = await fetch(url);
+  const text = await response.text();
+
+  const match = new RegExp(/value=["']([^"']+)["']/).exec(text);
+
+  if (match && match.length > 1) {
+    return match[1];
+  } else {
+    return;
+  }
+}
+
 export async function getFinalURL(url) {
-    let response = await fetch(url, {
+  if (url?.startsWith('https://vk.cc')) {
+    url = await resolveVKcc(url);
+  }
+
+  if (!url) {
+    return;
+  }
+
+  let response = await fetch(url, {
+    method: "HEAD",
+    redirect: "manual",
+  });
+  while (
+    response.status >= 300 &&
+    response.status < 400 &&
+    response.headers.get("location") &&
+    !url.startsWith('https://ozon.ru/point/')
+  ) {
+    url = response.headers.get("location");
+    if (!url.startsWith("http")) {
+      const baseUrl = new URL(response.url);
+      url = `${baseUrl.protocol}//${baseUrl.host}${url}`;
+    }
+    response = await fetch(url, {
       method: "HEAD",
       redirect: "manual",
     });
-    while (
-      response.status >= 300 &&
-      response.status < 400 &&
-      response.headers.get("location") &&
-      new URL(url).hostname !== 'ozon.ru'
-    ) {
-      url = response.headers.get("location");
-      if (!url.startsWith("http")) {
-        const baseUrl = new URL(response.url);
-        url = `${baseUrl.protocol}//${baseUrl.host}${url}`;
-      }
-      response = await fetch(url, {
-        method: "HEAD",
-        redirect: "manual",
-      });
-    }
-    return url;
-  }
+  }
+  return url;
+}
+
+export async function getTelegramMessage(input) {
+  const url = new URL(input);
+
+  url.search = "?embed=1&mode=tme";
+
+  const res = await fetch(url);
+  const htmlText = await res.text();
+
+  const dom = new JSDOM(htmlText);
+
+  const document = dom.window.document;
+  const message = document.querySelector(".tgme_widget_message");
+
+  return {
+    text: message.querySelector("div.tgme_widget_message_text.js-message_text")
+      .innerHTML,
+  };
+}