wip

data/cities.csv

@@ -1,4 +1,20 @@
 rank,Latitude,Longitude,Name of City,Country,2021 Population,2020 Population,Growth,Population Difference,Population Change
+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
+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
+20,23.1301964,113.2592945,Guangzhou,China,13635397,13301532,0.0251,333865,grew
+21,-22.9110137,-43.2093727,Rio de Janeiro,Brazil,13544462,13458075,0.0064,86387,grew
+22,31.5656822,74.3141829,Lahore,Pakistan,13095166,12642423,0.0358,452743,grew
+23,12.9767936,77.590082,Bangalore,India,12764935,12326532,0.0356,438403,grew
 24,55.7504461,37.6174943,Moscow,Russia,12593252,12537954,0.0044,55298,grew
 25,22.555454,114.0543297,Shenzhen,China,12591696,12356820,0.019,234876,grew
 26,13.0836939,80.270186,Chennai,India,11235018,10971108,0.0241,263910,grew

src/index.ts

@@ -23,115 +23,39 @@ async function visitCityPages(): Promise<void> {
 
     console.log('Connecting to existing Chrome browser...');
 
-    const driver = await useChrome();
+    const driver = await useExistingChrome();
     if (!driver) return;
 
     // Visit each city's TripAdvisor page
     for (let i = 0; i < cities.length; i++) {
+
         const city = cities[i];
         console.log(`[${i + 1}/${cities.length}] Visiting TripAdvisor page for ${city}...`);
-        let originalWindow;
-        let cityTopWindow;
-        let attactionsWindow;
-        let museumWindow;
 
         try {
 
             const originalWindow = await driver.getWindowHandle();
+            await driver.get("https://www.tripadvisor.com/");
+            await WebDriverUtils.wait(driver);
 
             console.log("Logo click")
             if (!await UIActions.gotoHome(driver)) throw `${city} failed`;
             await WebDriverUtils.wait(driver);
 
             console.log("Exec Search")
-            if (!await UIActions.execSearch(driver, city)) throw `${city} failed`;
-            await WebDriverUtils.wait(driver);
-
-            console.log("Click See all")
-            let seeAllUrl = await UIActions.getSeeAllUrl(driver);
-
-            if (seeAllUrl.length == 0) {
-                if (!await UIActions.clickTourismLink(driver)) throw `${city} failed`;
-                seeAllUrl = await UIActions.getSeeAllUrl(driver);
-            }
-
-            if (seeAllUrl.length == 0) throw `${city} failed`;
-            await WebDriverUtils.wait();
-
-
-            // open new incognito window
-            const driver2 = await useChrome();
-            if (!driver2) throw `${city} failed`;
-
-            await WebDriverUtils.wait();
-            await driver2.get(seeAllUrl);
-            await WebDriverUtils.wait();
-
-
-
-            console.log("Switch tab")
-            let windows = await driver.getAllWindowHandles();
-            // Switch to the newly opened window/tab
-            for (const handle of windows) {
-                if (handle !== originalWindow) {
-                    cityTopWindow = handle;
-                    await driver.switchTo().window(handle);
-                }
-            }
-
-            console.log("Click See all attractions")
-            if (!await UIActions.getSeeAllAttractionsUrl(driver)) throw `${city} failed`;
-            await WebDriverUtils.wait();
-
-            console.log("Switch tab to Attraction")
-            windows = await driver.getAllWindowHandles();
-            // Switch to the newly opened window/tab
-            for (const handle of windows) {
-                if (handle !== originalWindow && handle !== cityTopWindow) {
-                    attactionsWindow = handle;
-                    await driver.switchTo().window(attactionsWindow);
-                }
-            }
-
-            // click museum
-            console.log("Click Museum link");
-            if (!await UIActions.clickMuseumsLink(driver)) throw `${city} failed`;
+            if (!await UIActions.execSearch(driver, `"${city}" museums`)) throw `${city} failed`;
             await WebDriverUtils.wait(driver);
 
             let page = 1;
+            let museumNames: string[] = [];
+
             while (1) {
 
                 // get list of museums
                 console.log("Get list of museums");
-                const museumElms = await UIActions.getMusiums(driver);
+                const tmpMuseumNames = await UIActions.getAttractionNames(driver);
                 await WebDriverUtils.wait(driver);
-
-                for (const listItem of museumElms) {
-
-                    await listItem.click();
-                    await WebDriverUtils.wait(driver);
-
-                    windows = await driver.getAllWindowHandles();
-                    for (const handle of windows) {
-                        if (handle !== originalWindow && handle !== cityTopWindow && handle !== attactionsWindow) {
-                            museumWindow = handle;
-                            await driver.switchTo().window(museumWindow);
-                        }
-                    }
-
-                    const { websiteUrl, email } = await UIActions.getWebsiteAndEmail(driver);
-
-                    console.log(`${websiteUrl} / ${email}`);
-                    saveContactInfoToCSV(city, { websiteUrl: websiteUrl, email: email }, path.join(__dirname, '../data/contact_info.csv'));
-
-                    museumWindow && await driver.switchTo().window(museumWindow);
-                    await driver.close();
-                    await WebDriverUtils.wait(driver);
-
-                    attactionsWindow && await driver.switchTo().window(attactionsWindow);
-                    await WebDriverUtils.wait(driver);
-
-                }
+                museumNames = [...museumNames, ...tmpMuseumNames]
 
                 page++;
 
@@ -142,17 +66,41 @@ async function visitCityPages(): Promise<void> {
 
             }
 
+            for (const museumIndex in museumNames) {
 
-            await UIActions.closeAllTabsExceptFirst(driver);
+                const museumName = museumNames[museumIndex];
 
+                try{
+                    console.log(`reading museum ${museumIndex}/${museumNames.length}`);
+    
+                    if (!await UIActions.gotoHome(driver)) throw `${city} failed`;
+                    await WebDriverUtils.wait(driver);
+    
+                    if (!await UIActions.typeSearch(driver, `${city} ${museumName}`)) throw `${city} failed`;
+                    await WebDriverUtils.wait(driver);
+    
+                    await UIActions.clickFirstAttractionLinkInForm(driver)
+                    await WebDriverUtils.wait(driver);
+    
+                    const { websiteUrl, email } = await UIActions.getWebsiteAndEmail(driver);
+    
+                    console.log(`${websiteUrl} / ${email}`);
+                    saveContactInfoToCSV(city, { name: museumName, websiteUrl: websiteUrl, email: email }, path.join(__dirname, '../data/contact_info.csv'));
+    
+                    await UIActions.simulateClickAt(driver,100,100);
+                    await WebDriverUtils.wait(driver);
+                }catch(e){
+                    console.error(`failed ${museumName}`)
+                }
+               
+            }
 
             if (i < cities.length - 1) {
                 console.log(`Waiting for 5000 seconds before next city...`);
                 await WebDriverUtils.wait(); // Wait 5000 seconds before next city
             }
-        } catch (error) {
 
-            await UIActions.closeAllTabsExceptFirst(driver);
+        } catch (error) {
 
             // 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...`);

src/lib/UIActions.ts

@@ -1,4 +1,4 @@
-import { Builder, By, until, WebDriver, WebElement } from 'selenium-webdriver';
+import { Builder, By, until, WebDriver, WebElement, Actions } from 'selenium-webdriver';
 import * as chromedriver from 'chromedriver';
 import chrome, { ServiceBuilder } from 'selenium-webdriver/chrome';
 import * as fs from 'fs';
@@ -8,7 +8,7 @@ import { ContactInfo } from './types';
 
 
 
-export async function execSearch(driver: WebDriver, city: string): Promise<boolean> {
+export async function execSearch(driver: WebDriver, searchTerm: string): Promise<boolean> {
     try {
         // Find the search input field
         const searchSelector = 'input[name="q"][placeholder="Places to go, things to do, hotels..."]';
@@ -18,7 +18,7 @@ export async function execSearch(driver: WebDriver, city: string): Promise<boole
 
         // Clear any existing text and enter the city name
         await searchInput.clear();
-        await searchInput.sendKeys(city);
+        await searchInput.sendKeys(searchTerm);
 
         // Submit the search (press Enter)
         await WebDriverUtils.wait(driver);
@@ -31,6 +31,25 @@ export async function execSearch(driver: WebDriver, city: string): Promise<boole
     }
 }
 
+export async function typeSearch(driver: WebDriver, searchTerm: string): Promise<boolean> {
+    try {
+        // Find the search input field
+        const searchSelector = 'input[name="q"][placeholder="Places to go, things to do, hotels..."]';
+        await WebDriverUtils.waitForElement(driver, searchSelector, 10000);
+        console.log("Search box found");
+        const searchInput = await driver.findElement(By.css(searchSelector));
+
+        // Clear any existing text and enter the city name
+        await searchInput.clear();
+        await searchInput.sendKeys(searchTerm);
+
+        return true;
+    } catch (e) {
+        return false;
+    }
+}
+
+
 export async function clickSeeAll(driver: WebDriver): Promise<boolean> {
     try {
         const seeAllElement = await driver.wait(
@@ -80,7 +99,7 @@ export async function gotoHome(driver: WebDriver): Promise<boolean> {
 }
 
 export async function clickSeeAllAttractions(driver: WebDriver): Promise<boolean> {
-    const xpath = `//h2[starts-with(normalize-space(.), 'Top Attractions in')]/parent::*[1]//a[starts-with(@href, '/Attractions') and .//span[normalize-space(.)='See all']]`;
+    const xpath = `//h3[starts-with(normalize-space(.), 'Things to do')]/parent::*[1]//a[starts-with(@href, '/Attractions') and .//span[normalize-space(.)='See all']]`;
 
     try {
         const anchorElement = await driver.wait(until.elementLocated(By.xpath(xpath)), 5000);
@@ -98,7 +117,7 @@ export async function clickSeeAllAttractions(driver: WebDriver): Promise<boolean
 }
 
 export async function getSeeAllAttractionsUrl(driver: WebDriver): Promise<string | null> {
-    const xpath = `//h2[starts-with(normalize-space(.), 'Top Attractions in')]/parent::*[1]//a[starts-with(@href, '/Attractions') and .//span[normalize-space(.)='See all']]`;
+    const xpath = `//h3[starts-with(normalize-space(.), 'Top Attractions in')]/parent::*[1]//a[starts-with(@href, '/Attractions') and .//span[normalize-space(.)='See all']]`;
 
     try {
         const anchorElement = await driver.wait(until.elementLocated(By.xpath(xpath)), 5000);
@@ -161,8 +180,51 @@ export async function getMusiums(driver: WebDriver): Promise<WebElement[]> {
     }
 }
 
+export async function getMuseumsLinks(driver: WebDriver): Promise<string[]> {
+    const xpath = `//div//section[.//a[starts-with(@href, '/Attraction')] and .//h3]//a[starts-with(@href, '/Attraction') and .//img]`;
+
+    try {
+        const links = await driver.findElements(By.xpath(xpath));
+
+        const urls: string[] = [];
+        for (const link of links) {
+            const href = await link.getAttribute('href');
+            if (href) {
+                urls.push(href);
+            }
+        }
+
+        return urls;
+    } catch (error) {
+        console.warn('Error getting attraction URLs:', error);
+        return [];
+    }
+}
+
+export async function getAttractionNames(driver: WebDriver): Promise<string[]> {
+    // XPath to find <h3> inside <a> whose href starts with /Attraction_Review
+    const xpath = `//a[starts-with(@href, '/Attraction_Review')]/h3`;
+
+    try {
+        const h3Elements = await driver.findElements(By.xpath(xpath));
+
+        const names: string[] = [];
+        for (const h3 of h3Elements) {
+            const text = await h3.getText();
+            if (text) {
+                names.push(text.trim().replace(/^\d+\.\s*/, ''));
+            }
+        }
+
+        return names;
+    } catch (error) {
+        console.warn('Error getting attraction names:', error);
+        return [];
+    }
+}
+
 export async function getWebsiteAndEmail(driver: WebDriver): Promise<ContactInfo> {
-    const result: ContactInfo = { websiteUrl: null, email: null };
+    const result: ContactInfo = { name: null, websiteUrl: null, email: null };
 
     // XPath to find URL (starting with 'http' but not containing 'tripadvisor')
     const urlXPath = `//a[starts-with(@href, 'http') and not(contains(@href, 'tripadvisor'))]`;
@@ -224,4 +286,42 @@ export async function closeAllTabsExceptFirst(driver: WebDriver): Promise<void>
 
     await driver.switchTo().window(originalHandle);
     console.log(`Switched back to original tab: ${originalHandle}`);
-}
+}
+
+export async function clickFirstAttractionLinkInForm(driver: WebDriver): Promise<boolean> {
+    const xpath = `//form//a[starts-with(@href, '/Attraction')]`;
+
+    try {
+        // Wait for the link to appear inside a form
+        const link = await driver.wait(until.elementLocated(By.xpath(xpath)), 5000);
+
+        await driver.wait(until.elementIsVisible(link), 5000);
+        await driver.wait(until.elementIsEnabled(link), 5000);
+
+        await link.click();
+        console.log('Clicked the first attraction link in the form.');
+
+        return true;
+    } catch (error) {
+        console.warn('Attraction link not found or not clickable.', error);
+        return false;
+    }
+}
+
+export async function simulateClickAt(driver: WebDriver, x: number, y: number): Promise<void> {
+    try {
+
+        const bodyElement = await driver.findElement(By.css('body'));
+
+        await driver
+            .actions({ bridge: true }) // optional; required in some environments
+            .move({ x, y, origin: bodyElement }) // move relative to the viewport
+            .click()
+            .perform();
+
+        console.log(`Clicked at (${x}, ${y})`);
+    } catch (error) {
+        console.error('An error occurred:', error);
+    }
+}
+

src/lib/types.ts

@@ -1,4 +1,5 @@
 export interface ContactInfo {
+    name: string | null,
     websiteUrl: string | null;
     email: string | null;
 }

src/lib/utils.ts

@@ -15,32 +15,29 @@ export class WebDriverUtils {
      * @returns Promise that resolves after the specified time
      */
     static async wait(driver?: WebDriver): Promise<void> {
-        const seconds = Math.floor(Math.random() * 1000) % 3 + 3;
+        const seconds = Math.floor(Math.random() * 1000) % 3 + 2;
         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);
-                        `); 
-                    }
+                if (driver) {
+                    await driver.executeScript(`
+                        window.scrollBy(0, 10);
+                    `);
 
                     scrollCounter++;
                 }
             } catch (error) {
                 console.warn('Scroll failed:', error);
             }
-    
+            */
+
             // Wait a little between scrolls
             await new Promise(resolve => setTimeout(resolve, 500));
         }
@@ -61,7 +58,7 @@ export class WebDriverUtils {
 
 export function saveContactInfoToCSV(city: string, contactInfo: ContactInfo, filePath: string): void {
     const headers = 'City,Website URL,Email\n';
-    const line = `"${city},"${contactInfo.websiteUrl}","${contactInfo.email}"\n`;
+    const line = `"${city}","${contactInfo.name}","${contactInfo.websiteUrl}","${contactInfo.email}"\n`;
 
     if (!existsSync(filePath)) {
         writeFileSync(filePath, headers + line);
@@ -111,7 +108,7 @@ export async function disableCookiesInChrome(): Promise<WebDriver | null> {
 
 }
 
-export async function useChrome(): Promise<WebDriver | null> {
+export async function useChrome(initialUrl: string = "https://www.tripadvisor.com/"): Promise<WebDriver | null> {
     // Set Chrome options
     const options = new chrome.Options();
 
@@ -133,7 +130,7 @@ export async function useChrome(): Promise<WebDriver | null> {
             .setChromeOptions(options)
             .build();
 
-        await driver.get('https://www.tripadvisor.com/');
+        await driver.get(initialUrl);
 
         console.log('Chrome launched with cookies disabled.');