Compare commits
5 Commits
edge
...
1ffea0f61f
| Author | SHA1 | Date | |
|---|---|---|---|
| 1ffea0f61f | |||
| b6cad2a241 | |||
| 6f36809932 | |||
| 0e6df98a50 | |||
| d0bfe15fa4 |
@ -1,20 +1,4 @@
|
|||||||
rank,Latitude,Longitude,Name of City,Country,2021 Population,2020 Population,Growth,Population Difference,Population Change
|
rank,Latitude,Longitude,Name of City,Country,2021 Population,2020 Population,Growth,Population Difference,Population Change
|
||||||
1,35.6828387,139.7594549,Tokyo,Japan,37339804,37393128,-0.0014,53324,declined
|
|
||||||
2,28.6517178,77.2219388,Delhi,India,31181376,30290936,0.0294,890440,grew
|
|
||||||
3,31.2322758,121.4692071,Shanghai,China,27795702,27058480,0.0272,737222,grew
|
|
||||||
4,-23.5506507,-46.6333824,Sao Paulo,Brazil,22237472,22043028,0.0088,194444,grew
|
|
||||||
5,19.4326296,-99.1331785,Mexico City,Mexico,21918936,21782378,0.0063,136558,grew
|
|
||||||
6,23.7861979,90.4026151,Dhaka,Bangladesh,21741090,21005860,0.035,735230,grew
|
|
||||||
7,30.0443879,31.2357257,Cairo,Egypt,21322750,20900604,0.0202,422146,grew
|
|
||||||
8,39.906217,116.3912757,Beijing,China,20896820,20462610,0.0212,434210,grew
|
|
||||||
9,19.0759899,72.8773928,Mumbai,India,20667656,20411274,0.0126,256382,grew
|
|
||||||
10,34.6198813,135.490357,Osaka,Japan,19110616,19165340,-0.0029,54724,declined
|
|
||||||
11,24.8546842,67.0207055,Karachi,Pakistan,16459472,16093786,0.0227,365686,grew
|
|
||||||
12,29.5647398,106.5478767,Chongqing,China,16382376,15872179,0.0321,510197,grew
|
|
||||||
13,41.0096334,28.9651646,Istanbul,Turkey,15415197,15190336,0.0148,224861,grew
|
|
||||||
14,-34.6075682,-58.4370894,Buenos Aires,Argentina,15257673,15153729,0.0069,103944,grew
|
|
||||||
15,22.5414185,88.3576912,Kolkata,India,14974073,14850066,0.0084,124007,grew
|
|
||||||
16,-4.3217055,15.3125974,Kinshasa,DR Congo,14970460,14342439,0.0438,628021,grew
|
|
||||||
17,6.4550575,3.3941795,Lagos,Nigeria,14862111,14368332,0.0344,493779,grew
|
17,6.4550575,3.3941795,Lagos,Nigeria,14862111,14368332,0.0344,493779,grew
|
||||||
18,14.5907332,120.9809674,Manila,Philippines,14158573,13923452,0.0169,235121,grew
|
18,14.5907332,120.9809674,Manila,Philippines,14158573,13923452,0.0169,235121,grew
|
||||||
19,39.0856735,117.1951073,Tianjin,China,13794450,13589078,0.0151,205372,grew
|
19,39.0856735,117.1951073,Tianjin,China,13794450,13589078,0.0151,205372,grew
|
||||||
|
|||||||
|
2098
data/contact_info.csv
Normal file
2098
data/contact_info.csv
Normal file
File diff suppressed because it is too large
Load Diff
39
src/index.ts
39
src/index.ts
@ -9,7 +9,7 @@ import chrome, { ServiceBuilder } from 'selenium-webdriver/chrome';
|
|||||||
import * as fs from 'fs';
|
import * as fs from 'fs';
|
||||||
import * as path from 'path';
|
import * as path from 'path';
|
||||||
import { getCities } from './lib/cities';
|
import { getCities } from './lib/cities';
|
||||||
import { WebDriverUtils, saveContactInfoToCSV } from './lib/utils';
|
import { WebDriverUtils, saveContactInfoToCSV, useExistingChrome, disableCookiesInChrome } from './lib/utils';
|
||||||
import * as UIActions from './lib/UIActions';
|
import * as UIActions from './lib/UIActions';
|
||||||
import { randomUUID } from 'crypto';
|
import { randomUUID } from 'crypto';
|
||||||
|
|
||||||
@ -23,17 +23,8 @@ async function visitCityPages(): Promise<void> {
|
|||||||
|
|
||||||
console.log('Connecting to existing Chrome browser...');
|
console.log('Connecting to existing Chrome browser...');
|
||||||
|
|
||||||
// Connect to an existing Chrome browser running in debug mode on port 9222
|
const driver = await useExistingChrome();
|
||||||
const options = new chrome.Options();
|
if (!driver) return;
|
||||||
|
|
||||||
// Set the debugger address to connect to the existing Chrome instance
|
|
||||||
options.debuggerAddress('localhost:9222');
|
|
||||||
|
|
||||||
// Create WebDriver instance that connects to the existing browser
|
|
||||||
const driver: WebDriver = await new Builder()
|
|
||||||
.forBrowser('chrome')
|
|
||||||
.setChromeOptions(options)
|
|
||||||
.build();
|
|
||||||
|
|
||||||
// Visit each city's TripAdvisor page
|
// Visit each city's TripAdvisor page
|
||||||
for (let i = 0; i < cities.length; i++) {
|
for (let i = 0; i < cities.length; i++) {
|
||||||
@ -50,18 +41,18 @@ async function visitCityPages(): Promise<void> {
|
|||||||
|
|
||||||
console.log("Logo click")
|
console.log("Logo click")
|
||||||
if (!await UIActions.gotoHome(driver)) throw `${city} failed`;
|
if (!await UIActions.gotoHome(driver)) throw `${city} failed`;
|
||||||
await WebDriverUtils.wait(5);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
console.log("Exec Search")
|
console.log("Exec Search")
|
||||||
if (!await UIActions.execSearch(driver, city)) throw `${city} failed`;
|
if (!await UIActions.execSearch(driver, city)) throw `${city} failed`;
|
||||||
await WebDriverUtils.wait(5);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
console.log("Click See all")
|
console.log("Click See all")
|
||||||
if (!await UIActions.clickSeeAll(driver)) {
|
if (!await UIActions.clickSeeAll(driver)) {
|
||||||
if (!await UIActions.clickTourismLink(driver)) throw `${city} failed`;
|
if (!await UIActions.clickTourismLink(driver)) throw `${city} failed`;
|
||||||
if (!await UIActions.clickSeeAll(driver)) throw `${city} failed`;
|
if (!await UIActions.clickSeeAll(driver)) throw `${city} failed`;
|
||||||
}
|
}
|
||||||
await WebDriverUtils.wait(5);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
console.log("Switch tab")
|
console.log("Switch tab")
|
||||||
let windows = await driver.getAllWindowHandles();
|
let windows = await driver.getAllWindowHandles();
|
||||||
@ -75,7 +66,7 @@ async function visitCityPages(): Promise<void> {
|
|||||||
|
|
||||||
console.log("Click See all attractions")
|
console.log("Click See all attractions")
|
||||||
if (!await UIActions.clickSeeAllAttractions(driver)) throw `${city} failed`;
|
if (!await UIActions.clickSeeAllAttractions(driver)) throw `${city} failed`;
|
||||||
await WebDriverUtils.wait(5);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
console.log("Switch tab to Attraction")
|
console.log("Switch tab to Attraction")
|
||||||
windows = await driver.getAllWindowHandles();
|
windows = await driver.getAllWindowHandles();
|
||||||
@ -90,7 +81,7 @@ async function visitCityPages(): Promise<void> {
|
|||||||
// click museum
|
// click museum
|
||||||
console.log("Click Museum link");
|
console.log("Click Museum link");
|
||||||
if (!await UIActions.clickMuseumsLink(driver)) throw `${city} failed`;
|
if (!await UIActions.clickMuseumsLink(driver)) throw `${city} failed`;
|
||||||
await WebDriverUtils.wait(5);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
let page = 1;
|
let page = 1;
|
||||||
while (1) {
|
while (1) {
|
||||||
@ -98,12 +89,12 @@ async function visitCityPages(): Promise<void> {
|
|||||||
// get list of museums
|
// get list of museums
|
||||||
console.log("Get list of museums");
|
console.log("Get list of museums");
|
||||||
const museumElms = await UIActions.getMusiums(driver);
|
const museumElms = await UIActions.getMusiums(driver);
|
||||||
await WebDriverUtils.wait(1);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
for (const listItem of museumElms) {
|
for (const listItem of museumElms) {
|
||||||
|
|
||||||
await listItem.click();
|
await listItem.click();
|
||||||
await WebDriverUtils.wait(3);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
windows = await driver.getAllWindowHandles();
|
windows = await driver.getAllWindowHandles();
|
||||||
for (const handle of windows) {
|
for (const handle of windows) {
|
||||||
@ -120,10 +111,10 @@ async function visitCityPages(): Promise<void> {
|
|||||||
|
|
||||||
museumWindow && await driver.switchTo().window(museumWindow);
|
museumWindow && await driver.switchTo().window(museumWindow);
|
||||||
await driver.close();
|
await driver.close();
|
||||||
await WebDriverUtils.wait(1);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
attactionsWindow && await driver.switchTo().window(attactionsWindow);
|
attactionsWindow && await driver.switchTo().window(attactionsWindow);
|
||||||
await WebDriverUtils.wait(1);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
}
|
}
|
||||||
|
|
||||||
@ -132,7 +123,7 @@ async function visitCityPages(): Promise<void> {
|
|||||||
if (page > 10) break;
|
if (page > 10) break;
|
||||||
|
|
||||||
UIActions.clickPagination(driver, page);
|
UIActions.clickPagination(driver, page);
|
||||||
await WebDriverUtils.wait(5);
|
await WebDriverUtils.wait(driver);
|
||||||
|
|
||||||
}
|
}
|
||||||
|
|
||||||
@ -142,7 +133,7 @@ async function visitCityPages(): Promise<void> {
|
|||||||
|
|
||||||
if (i < cities.length - 1) {
|
if (i < cities.length - 1) {
|
||||||
console.log(`Waiting for 5000 seconds before next city...`);
|
console.log(`Waiting for 5000 seconds before next city...`);
|
||||||
await WebDriverUtils.wait(5); // Wait 5000 seconds before next city
|
await WebDriverUtils.wait(); // Wait 5000 seconds before next city
|
||||||
}
|
}
|
||||||
} catch (error) {
|
} catch (error) {
|
||||||
|
|
||||||
@ -150,7 +141,7 @@ async function visitCityPages(): Promise<void> {
|
|||||||
|
|
||||||
// If the button is not found within the timeout, log and continue to the next city
|
// If the button is not found within the timeout, log and continue to the next city
|
||||||
console.log(`No Museums button found for ${city}. Moving to next city after 5 seconds...`);
|
console.log(`No Museums button found for ${city}. Moving to next city after 5 seconds...`);
|
||||||
await WebDriverUtils.wait(5); // Wait 5 seconds before next city
|
await WebDriverUtils.wait(); // Wait 5 seconds before next city
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|||||||
@ -21,9 +21,9 @@ export async function execSearch(driver: WebDriver, city: string): Promise<boole
|
|||||||
await searchInput.sendKeys(city);
|
await searchInput.sendKeys(city);
|
||||||
|
|
||||||
// Submit the search (press Enter)
|
// Submit the search (press Enter)
|
||||||
await WebDriverUtils.wait(2);
|
await WebDriverUtils.wait(driver);
|
||||||
await searchInput.sendKeys('\uE007'); // Unicode for Enter key
|
await searchInput.sendKeys('\uE007'); // Unicode for Enter key
|
||||||
await WebDriverUtils.wait(5); // Wait 5 seconds before next city
|
await WebDriverUtils.wait(driver); // Wait 5 seconds before next city
|
||||||
|
|
||||||
return true;
|
return true;
|
||||||
} catch (e) {
|
} catch (e) {
|
||||||
|
|||||||
@ -2,7 +2,8 @@
|
|||||||
* Utility class for common WebDriver operations
|
* Utility class for common WebDriver operations
|
||||||
*/
|
*/
|
||||||
|
|
||||||
import { WebDriver, By, until } from 'selenium-webdriver';
|
import { Builder, By, until, WebDriver } from 'selenium-webdriver';
|
||||||
|
import chrome from 'selenium-webdriver/chrome';
|
||||||
import { writeFileSync, existsSync, appendFileSync } from 'fs';
|
import { writeFileSync, existsSync, appendFileSync } from 'fs';
|
||||||
import * as path from 'path';
|
import * as path from 'path';
|
||||||
import { ContactInfo } from './types';
|
import { ContactInfo } from './types';
|
||||||
@ -13,9 +14,36 @@ export class WebDriverUtils {
|
|||||||
* @param seconds Number of seconds to wait
|
* @param seconds Number of seconds to wait
|
||||||
* @returns Promise that resolves after the specified time
|
* @returns Promise that resolves after the specified time
|
||||||
*/
|
*/
|
||||||
static async wait(seconds: number): Promise<void> {
|
static async wait(driver?: WebDriver): Promise<void> {
|
||||||
console.log(`Waiting for ${seconds} seconds...`);
|
const seconds = Math.floor(Math.random() * 1000) % 3 + 3;
|
||||||
return new Promise(resolve => setTimeout(resolve, seconds * 1000));
|
console.log(`Scrolling to bottom for ${seconds} seconds...`);
|
||||||
|
|
||||||
|
const endTime = Date.now() + seconds * 1000;
|
||||||
|
|
||||||
|
let scrollCounter = 0;
|
||||||
|
|
||||||
|
while (Date.now() < endTime) {
|
||||||
|
try {
|
||||||
|
if(driver){
|
||||||
|
if(scrollCounter < 4){
|
||||||
|
await driver.executeScript(`
|
||||||
|
window.scrollBy(0, window.innerHeight);
|
||||||
|
`);
|
||||||
|
}else{
|
||||||
|
await driver.executeScript(`
|
||||||
|
window.scrollTo(0, 0);
|
||||||
|
`);
|
||||||
|
}
|
||||||
|
|
||||||
|
scrollCounter++;
|
||||||
|
}
|
||||||
|
} catch (error) {
|
||||||
|
console.warn('Scroll failed:', error);
|
||||||
|
}
|
||||||
|
|
||||||
|
// Wait a little between scrolls
|
||||||
|
await new Promise(resolve => setTimeout(resolve, 500));
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
@ -43,3 +71,58 @@ export function saveContactInfoToCSV(city: string, contactInfo: ContactInfo, fil
|
|||||||
|
|
||||||
console.log(`Contact info saved to ${filePath}`);
|
console.log(`Contact info saved to ${filePath}`);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
export async function disableCookiesInChrome(): Promise<WebDriver | null> {
|
||||||
|
// Set Chrome options
|
||||||
|
const options = new chrome.Options();
|
||||||
|
|
||||||
|
// 1. Block all cookies
|
||||||
|
//options.setUserPreferences({
|
||||||
|
// 'profile.default_content_setting_values.cookies': 2, // 2 = Block all
|
||||||
|
// 'profile.block_third_party_cookies': true
|
||||||
|
//});
|
||||||
|
|
||||||
|
// 2. Optional: Launch in incognito for extra privacy
|
||||||
|
options.addArguments('--incognito');
|
||||||
|
options.addArguments('--start-maximized');
|
||||||
|
|
||||||
|
let driver: WebDriver | null = null;
|
||||||
|
|
||||||
|
try {
|
||||||
|
driver = await new Builder()
|
||||||
|
.forBrowser('chrome')
|
||||||
|
.setChromeOptions(options)
|
||||||
|
.build();
|
||||||
|
|
||||||
|
await driver.get('https://www.tripadvisor.com/');
|
||||||
|
|
||||||
|
console.log('Chrome launched with cookies disabled.');
|
||||||
|
|
||||||
|
// Optional: Verify cookies are blocked by trying to set/get a cookie
|
||||||
|
await driver.manage().addCookie({ name: 'test', value: '123' });
|
||||||
|
const cookies = await driver.manage().getCookies();
|
||||||
|
console.log('Cookies after trying to add:', cookies); // Should be empty or restricted
|
||||||
|
|
||||||
|
return driver;
|
||||||
|
} catch (error) {
|
||||||
|
console.error('Error:', error);
|
||||||
|
return driver;
|
||||||
|
}
|
||||||
|
|
||||||
|
}
|
||||||
|
|
||||||
|
export async function useExistingChrome(): Promise<WebDriver> {
|
||||||
|
// Connect to an existing Chrome browser running in debug mode on port 9222
|
||||||
|
const options = new chrome.Options();
|
||||||
|
|
||||||
|
// Set the debugger address to connect to the existing Chrome instance
|
||||||
|
options.debuggerAddress('localhost:9222');
|
||||||
|
|
||||||
|
// Create WebDriver instance that connects to the existing browser
|
||||||
|
const driver: WebDriver = await new Builder()
|
||||||
|
.forBrowser('chrome')
|
||||||
|
.setChromeOptions(options)
|
||||||
|
.build();
|
||||||
|
|
||||||
|
return driver;
|
||||||
|
}
|
||||||
Reference in New Issue
Block a user