whyrating-engine-legacy/diagnose_reviews_panel.py

#!/usr/bin/env python3
"""
Better diagnostic: Actually wait for reviews panel to load and find correct selector.
"""

import time
from seleniumbase import Driver

url = "https://www.google.com/maps/search/?api=1&query=panevezio%20respubliikine%20ligonine"

print("Opening browser...")
driver = Driver(uc=True, headless=False)

try:
    # Add English locale
    if '?' in url:
        url += '&hl=en'
    else:
        url += '?hl=en'

    driver.get(url)
    print(f"Loaded: {url}")
    time.sleep(5)

    # Handle GDPR
    try:
        form_btns = driver.find_elements('css selector', 'form button')
        for btn in form_btns:
            btn_text = (btn.text or '').lower()
            if 'accept all' in btn_text:
                print(f"Clicking GDPR: {btn.text}")
                btn.click()
                time.sleep(2)
                break
    except:
        pass

    # Click reviews tab and WAIT for panel to load
    print("\nClicking reviews tab...")
    time.sleep(2)
    tabs = driver.find_elements('css selector', 'button[role="tab"]')
    for tab in tabs:
        text = (tab.text or '').lower()
        aria = (tab.get_attribute('aria-label') or '').lower()
        if 'review' in text or 'review' in aria:
            print(f"Found reviews tab: {tab.text or aria[:50]}")
            driver.execute_script("arguments[0].click();", tab)
            print("Clicked! Waiting for reviews panel to load...")
            time.sleep(5)  # Wait longer for reviews to actually load
            break

    # Try scrolling the reviews pane to load more
    print("\nTrying to find and scroll reviews pane...")
    pane_selectors = [
        'div.m6QErb.WNBkOb.XiKgde',
        'div.m6QErb',
        'div[role="main"]'
    ]

    for selector in pane_selectors:
        try:
            pane = driver.find_element('css selector', selector)
            print(f"Found pane: {selector}")
            driver.execute_script("arguments[0].scrollBy(0, 500);", pane)
            time.sleep(2)
            driver.execute_script("arguments[0].scrollBy(0, 500);", pane)
            time.sleep(2)
            break
        except:
            continue

    # NOW check for review selectors
    print("\n" + "="*80)
    print("CHECKING REVIEW SELECTORS AFTER PANEL LOADED:")
    print("="*80)

    selectors_to_try = [
        ('div.jftiEf.fontBodyMedium', 'Standard Google Maps reviews'),
        ('div.jftiEf', 'Just jftiEf class'),
        ('div.fontBodyMedium', 'Just fontBodyMedium'),
        ('div[data-review-id]', 'data-review-id attribute'),
        ('div[jsaction*="review"]', 'jsaction with review'),
        ('[data-review]', 'data-review attribute'),
        ('div[class*="review" i]', 'Class containing review'),
        ('[role="article"]', 'role=article'),
        ('div[jslog]', 'Elements with jslog (Google tracking)'),
    ]

    for selector, description in selectors_to_try:
        count = driver.execute_script(
            f"return document.querySelectorAll('{selector}').length;"
        )
        print(f"{description:35} | {selector:40} | Found: {count}")

    # Get detailed info about most promising selector
    print("\n" + "="*80)
    print("ANALYZING MOST PROMISING SELECTOR:")
    print("="*80)

    analysis = driver.execute_script("""
        // Try selectors in order of likelihood
        const selectors = [
            'div.jftiEf.fontBodyMedium',
            'div.jftiEf',
            'div.fontBodyMedium',
            'div[jslog*="impression"]',
            '[role="article"]'
        ];

        for (let selector of selectors) {
            const elements = document.querySelectorAll(selector);
            if (elements.length > 5) {  // Need at least a few to be reviews
                // Analyze first element
                const first = elements[0];
                const analysis = {
                    selector: selector,
                    total_found: elements.length,
                    first_element: {
                        tag: first.tagName,
                        classes: first.className,
                        has_rating: !!first.querySelector('[aria-label*="star" i]'),
                        has_author: !!first.querySelector('button, a, div[aria-label]'),
                        has_avatar: !!first.querySelector('img'),
                        has_date: !!first.textContent.match(/\\d+\\s*(day|week|month|year|hour|minute)/i),
                        text_length: first.textContent.length,
                        sample_text: first.textContent.substring(0, 100)
                    }
                };

                // Check if multiple elements have review characteristics
                let reviewLikeCount = 0;
                for (let i = 0; i < Math.min(10, elements.length); i++) {
                    const elem = elements[i];
                    const hasRating = !!elem.querySelector('[aria-label*="star" i]');
                    const hasText = elem.textContent.length > 30;
                    if (hasRating && hasText) reviewLikeCount++;
                }
                analysis.review_like_count_in_first_10 = reviewLikeCount;

                return analysis;
            }
        }

        return {error: 'No selector found with >5 elements'};
    """)

    if 'error' in analysis:
        print(f"ERROR: {analysis['error']}")
    else:
        print(f"Best selector: {analysis['selector']}")
        print(f"Total found: {analysis['total_found']}")
        print(f"Review-like in first 10: {analysis['review_like_count_in_first_10']}")
        print(f"\nFirst element analysis:")
        for key, value in analysis['first_element'].items():
            print(f"  {key}: {value}")

    print("\n" + "="*80)
    print("Keeping browser open for 120 seconds for manual inspection...")
    print("="*80)
    time.sleep(120)

finally:
    driver.quit()