Frason's picture
updated
d785968 verified
import time
import logging
import argparse
import os
import json
import random
import re
import uuid
from collections import defaultdict
from datetime import datetime
from typing import List, Dict, Any, Optional, Union, Tuple
from selenium import webdriver
from selenium.webdriver.chrome.options import Options
from selenium.webdriver.chrome.service import Service
from selenium.webdriver.common.by import By
from selenium.webdriver.common.keys import Keys
from selenium.webdriver.common.action_chains import ActionChains
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC
from selenium.common.exceptions import (
TimeoutException, NoSuchElementException, WebDriverException
)
import gradio as gr
import pandas as pd
# Setup logging
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
datefmt='%Y-%m-%d %H:%M:%S'
)
logger = logging.getLogger(__name__)
# Predefined advertisers list
ADVERTISERS = [
{"id": "AR10051102910143528961", "name": "Theory Sabers"},
{"id": "AR12645693856247971841", "name": "Artsabers"},
{"id": "AR07257050693515608065", "name": "bmlightsabers"},
{"id": "AR01506694249926623233", "name": "Padawan Outpost Ltd"},
{"id": "AR10584025853845307393", "name": "GalaxySabers"},
{"id": "AR16067963414479110145", "name": "nsabers"},
{"id": "AR12875519274243850241", "name": "es-sabers"},
{"id": "AR05144647067079016449", "name": "Ultra Sabers"},
{"id": "AR15581800501283389441", "name": "SuperNeox"},
{"id": "AR06148907109187584001", "name": "Sabertrio"}
]
#####################################
### FACEBOOK SCRAPER SECTION #######
#####################################
# Constants for Facebook Scraper
FB_DEFAULT_TIMEOUT = 60 # seconds
FB_MIN_WAIT_TIME = 1 # minimum seconds for random waits
FB_MAX_WAIT_TIME = 3 # maximum seconds for random waits
FB_MAX_SCROLL_ATTEMPTS = 5 # maximum number of scroll attempts
FB_SELECTOR_HISTORY_FILE = "fb_selector_stats.json" # File to store selector success stats
# User agents for rotation
USER_AGENTS = [
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/121.0.0.0 Safari/537.36",
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.0 Safari/605.1.15",
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/122.0.0.0 Safari/537.36 Edg/122.0.0.0",
"Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/118.0"
]
# Viewport sizes for randomization
VIEWPORT_SIZES = [
(1366, 768),
(1920, 1080),
(1536, 864),
(1440, 900)
]
class SelectorStats:
"""Class to track and optimize selector performance"""
def __init__(self, file_path=FB_SELECTOR_HISTORY_FILE):
self.file_path = file_path
self.stats = self._load_stats()
def _load_stats(self) -> Dict:
"""Load stats from file or initialize if not exists"""
if os.path.exists(self.file_path):
try:
with open(self.file_path, 'r') as f:
return json.load(f)
except (json.JSONDecodeError, IOError) as e:
logger.warning(f"Error loading selector stats: {e}, initializing new stats")
# Initialize structure for platform stats
return {
"facebook": {"selectors": {}, "last_updated": datetime.now().isoformat()}
}
def update_selector_success(self, selector: str, count: int = 1) -> None:
"""Record successful use of a selector"""
platform = "facebook" # Only using Facebook for this version
if platform not in self.stats:
self.stats[platform] = {"selectors": {}, "last_updated": datetime.now().isoformat()}
if selector not in self.stats[platform]["selectors"]:
self.stats[platform]["selectors"][selector] = {"successes": 0, "attempts": 0}
self.stats[platform]["selectors"][selector]["successes"] += count
self.stats[platform]["selectors"][selector]["attempts"] += 1
self.stats[platform]["last_updated"] = datetime.now().isoformat()
# Save after each update
self._save_stats()
def update_selector_attempt(self, selector: str) -> None:
"""Record attempt to use a selector regardless of success"""
platform = "facebook" # Only using Facebook for this version
if platform not in self.stats:
self.stats[platform] = {"selectors": {}, "last_updated": datetime.now().isoformat()}
if selector not in self.stats[platform]["selectors"]:
self.stats[platform]["selectors"][selector] = {"successes": 0, "attempts": 0}
self.stats[platform]["selectors"][selector]["attempts"] += 1
self.stats[platform]["last_updated"] = datetime.now().isoformat()
# Don't save on every attempt to reduce disk I/O
def get_best_selectors(self, min_attempts: int = 3, max_count: int = 10) -> List[str]:
"""Get the best performing selectors for Facebook"""
platform = "facebook" # Only using Facebook for this version
if platform not in self.stats:
return []
selectors = []
for selector, data in self.stats[platform]["selectors"].items():
if data["attempts"] >= min_attempts:
success_rate = data["successes"] / data["attempts"] if data["attempts"] > 0 else 0
selectors.append((selector, success_rate))
# Sort by success rate (descending)
selectors.sort(key=lambda x: x[1], reverse=True)
# Return top N selectors
return [s[0] for s in selectors[:max_count]]
def _save_stats(self) -> None:
"""Save stats to file"""
try:
with open(self.file_path, 'w') as f:
json.dump(self.stats, f, indent=2)
except IOError as e:
logger.error(f"Error saving selector stats: {e}")
class FacebookAdsScraper:
def __init__(self, headless=True, debug_mode=False):
"""Initialize the ads scraper with browser configuration"""
self.debug_mode = debug_mode
self.headless = headless
self.driver = self._setup_driver(headless)
# Initialize selector stats tracker
self.selector_stats = SelectorStats()
# Track navigation history for smart retry
self.navigation_history = []
# Track success/failure for self-healing
self.success_rate = defaultdict(lambda: {"success": 0, "failure": 0})
# Generate a session ID for this scraping session
self.session_id = str(uuid.uuid4())[:8]
def _setup_driver(self, headless):
"""Set up and configure the Chrome WebDriver with anti-detection measures"""
chrome_options = Options()
if headless:
chrome_options.add_argument("--headless")
# Select a random user agent
user_agent = random.choice(USER_AGENTS)
chrome_options.add_argument(f"--user-agent={user_agent}")
logger.info(f"Using user agent: {user_agent}")
# Select a random viewport size
viewport_width, viewport_height = random.choice(VIEWPORT_SIZES)
chrome_options.add_argument(f"--window-size={viewport_width},{viewport_height}")
logger.info(f"Using viewport size: {viewport_width}x{viewport_height}")
# Add common options to avoid detection
chrome_options.add_argument("--disable-blink-features=AutomationControlled")
chrome_options.add_argument("--no-sandbox")
chrome_options.add_argument("--disable-dev-shm-usage")
chrome_options.add_argument("--disable-gpu")
chrome_options.add_argument("--start-maximized")
chrome_options.add_argument("--enable-unsafe-swiftshader")
# Performance improvements
chrome_options.add_argument("--disable-extensions")
chrome_options.add_argument("--disable-notifications")
chrome_options.add_argument("--blink-settings=imagesEnabled=true")
# Add experimental options to avoid detection
chrome_options.add_experimental_option("excludeSwitches", ["enable-automation"])
chrome_options.add_experimental_option("useAutomationExtension", False)
# Additional preferences to improve performance
chrome_options.add_experimental_option("prefs", {
"profile.default_content_setting_values.notifications": 2,
"profile.managed_default_content_settings.images": 1,
"profile.managed_default_content_settings.cookies": 1,
# Add some randomness to the profile
"profile.default_content_setting_values.plugins": random.randint(1, 3),
"profile.default_content_setting_values.popups": random.randint(1, 2)
})
try:
# Try to create driver with service for newer Selenium versions
service = Service()
driver = webdriver.Chrome(service=service, options=chrome_options)
# Execute CDP commands to avoid detection (works in newer Chrome versions)
driver.execute_cdp_cmd("Page.addScriptToEvaluateOnNewDocument", {
"source": """
Object.defineProperty(navigator, 'webdriver', {
get: () => undefined
});
// Overwrite the languages with random order
Object.defineProperty(navigator, 'languages', {
get: () => ['en-US', 'en', 'de'].sort(() => 0.5 - Math.random())
});
// Modify plugins length
Object.defineProperty(navigator, 'plugins', {
get: () => {
// Randomize plugins length between 3 and 7
const len = Math.floor(Math.random() * 5) + 3;
const plugins = { length: len };
for (let i = 0; i < len; i++) {
plugins[i] = {
name: ['Flash', 'Chrome PDF Plugin', 'Native Client', 'Chrome PDF Viewer'][Math.floor(Math.random() * 4)],
filename: ['internal-pdf-viewer', 'mhjfbmdgcfjbbpaeojofohoefgiehjai', 'internal-nacl-plugin'][Math.floor(Math.random() * 3)]
};
}
return plugins;
}
});
"""
})
except TypeError:
# Fallback for older Selenium versions
driver = webdriver.Chrome(options=chrome_options)
except Exception as e:
# If there's an issue with CDP, continue anyway
logger.warning(f"CDP command failed, continuing: {e}")
driver = webdriver.Chrome(options=chrome_options)
# Set default timeout
driver.set_page_load_timeout(FB_DEFAULT_TIMEOUT)
return driver
def random_wait(self, min_time=None, max_time=None):
"""Wait for a random amount of time to simulate human behavior"""
min_time = min_time or FB_MIN_WAIT_TIME
max_time = max_time or FB_MAX_WAIT_TIME
wait_time = random.uniform(min_time, max_time)
time.sleep(wait_time)
return wait_time
def human_like_scroll(self, scroll_attempts=None):
"""Scroll down the page in a human-like way"""
attempts = scroll_attempts or random.randint(3, FB_MAX_SCROLL_ATTEMPTS)
# Get page height before scrolling
initial_height = self.driver.execute_script("return document.body.scrollHeight")
for i in range(attempts):
# Calculate a random scroll amount (25-90% of viewport)
scroll_percent = random.uniform(0.25, 0.9)
viewport_height = self.driver.execute_script("return window.innerHeight")
scroll_amount = int(viewport_height * scroll_percent)
# Scroll with a random speed
scroll_steps = random.randint(5, 15)
current_position = self.driver.execute_script("return window.pageYOffset")
target_position = current_position + scroll_amount
for step in range(scroll_steps):
# Calculate next position with easing
t = (step + 1) / scroll_steps
# Ease in-out function
factor = t * t * (3.0 - 2.0 * t)
next_position = current_position + (target_position - current_position) * factor
self.driver.execute_script(f"window.scrollTo(0, {next_position})")
time.sleep(random.uniform(0.01, 0.05))
# Occasionally pause longer as if reading content
if random.random() < 0.3: # 30% chance to pause
self.random_wait(1.5, 3.5)
else:
self.random_wait(0.5, 1.5)
# Log progress
logger.info(f"Human-like scroll {i + 1}/{attempts} completed")
# Check if we've reached the bottom of the page
new_height = self.driver.execute_script("return document.body.scrollHeight")
if new_height == initial_height and i > 1:
# We haven't loaded new content after a couple of scrolls
# Do one big scroll to the bottom to trigger any lazy loading
self.driver.execute_script("window.scrollTo(0, document.body.scrollHeight)")
self.random_wait()
initial_height = new_height
def simulate_human_behavior(self):
"""Simulate random human-like interactions with the page"""
# Random chance to move the mouse around
if random.random() < 0.7: # 70% chance
try:
# Find a random element to hover over
elements = self.driver.find_elements(By.CSS_SELECTOR, "a, button, input, div")
if elements:
element = random.choice(elements)
ActionChains(self.driver).move_to_element(element).perform()
self.random_wait(0.2, 1.0)
except:
# Ignore any errors, this is just for randomness
pass
# Random chance to click somewhere non-interactive
if random.random() < 0.2: # 20% chance
try:
# Find a safe area to click (like a paragraph or heading)
safe_elements = self.driver.find_elements(By.CSS_SELECTOR, "p, h1, h2, h3, h4, span")
if safe_elements:
safe_element = random.choice(safe_elements)
ActionChains(self.driver).move_to_element(safe_element).click().perform()
self.random_wait(0.2, 1.0)
except:
# Ignore any errors, this is just for randomness
pass
def check_headless_visibility(self):
"""
Check if elements are visible in headless mode
Returns True if everything is working properly
"""
if not self.headless:
# If not in headless mode, no need to check
return True
logger.info("Performing headless visibility check...")
# Use a simpler page for testing interactivity
test_url = "https://www.example.com"
try:
self.driver.get(test_url)
# Just check if the page loads at all - don't try to interact with elements
WebDriverWait(self.driver, 10).until(
EC.presence_of_element_located((By.TAG_NAME, "body"))
)
logger.info("Headless check passed: Page loaded successfully")
return True
except Exception as e:
logger.error(f"Headless check failed: {e}")
# Try switching to non-headless mode
logger.info("Switching to non-headless mode...")
self.driver.quit()
self.headless = False
self.driver = self._setup_driver(headless=False)
return True # Continue without rechecking
def fetch_facebook_ads(self, query):
"""Fetch ads from Facebook's Ad Library with anti-detection measures"""
ads_data = []
base_url = "https://www.facebook.com/ads/library/"
logger.info(f"Fetching Facebook ads for {query}")
try:
# Add some randomness to URL parameters
params = {
"active_status": "all",
"ad_type": "all",
"country": "ALL",
"q": query,
# Random parameters to avoid fingerprinting
"_": int(time.time() * 1000),
"session_id": self.session_id
}
# Construct URL with parameters
url = base_url + "?" + "&".join(f"{k}={v}" for k, v in params.items())
logger.info(f"Navigating to Facebook URL: {url}")
# Navigate to the URL
self.driver.get(url)
# Wait for page to initially load
try:
WebDriverWait(self.driver, FB_DEFAULT_TIMEOUT).until(
EC.any_of(
EC.presence_of_element_located((By.CSS_SELECTOR, "div[role='main']")),
EC.presence_of_element_located((By.TAG_NAME, "body"))
)
)
except TimeoutException:
logger.warning("Timeout waiting for Facebook page to load initially, continuing anyway")
# Human-like scrolling to trigger lazy loading
self.human_like_scroll()
# Simulate human behavior
self.simulate_human_behavior()
# Save debug data at this point
if self.debug_mode:
self._save_debug_data("facebook_after_scroll", query)
# Find ad elements using self-healing selectors
ad_elements = self._find_facebook_ad_elements()
if not ad_elements:
logger.info("No Facebook ads found")
if self.debug_mode:
self._save_debug_data("facebook_no_ads", query)
# Return placeholder data as fallback
return self._generate_placeholder_facebook_data(query)
# Process the found ad elements
for i, ad in enumerate(ad_elements[:10]): # Limit to 10 ads for performance
try:
ad_data = {
"platform": "Facebook",
"query": query,
"timestamp": datetime.now().isoformat(),
"index": i + 1,
"session_id": self.session_id
}
# Extract data using smarter methods
full_text = ad.text.strip()
# Log first ad text for debugging
if i == 0:
logger.info(f"First Facebook ad full text (first 150 chars): {full_text[:150]}...")
# Smart data extraction
extracted_data = self._extract_facebook_ad_data(ad, full_text)
# Merge extracted data
ad_data.update(extracted_data)
# Add fallback values if needed
if "advertiser" not in ad_data or not ad_data["advertiser"]:
ad_data["advertiser"] = "Unknown Advertiser"
if "text" not in ad_data or not ad_data["text"]:
ad_data["text"] = "Ad content not available"
ads_data.append(ad_data)
except Exception as e:
logger.warning(f"Error processing Facebook ad {i + 1}: {e}")
return ads_data if ads_data else self._generate_placeholder_facebook_data(query)
except Exception as e:
logger.error(f"Error fetching Facebook ads: {e}")
if self.debug_mode:
self._save_debug_data("facebook_error", query)
return self._generate_placeholder_facebook_data(query)
def _find_facebook_ad_elements(self):
"""Find Facebook ad elements using a self-healing selector strategy"""
# Historical best performers
historical_best = self.selector_stats.get_best_selectors()
# Base selectors
base_selectors = [
"div[class*='_7jvw']",
"div[data-testid='ad_library_card']",
"div[class*='AdLibraryCard']",
"div.AdLibraryCard",
"div[class*='adCard']",
"div[class*='ad_card']"
]
# Combine selectors, prioritizing historical best
combined_selectors = historical_best + [s for s in base_selectors if s not in historical_best]
# Try each selector
for selector in combined_selectors:
try:
# Record attempt
self.selector_stats.update_selector_attempt(selector)
elements = self.driver.find_elements(By.CSS_SELECTOR, selector)
if elements:
logger.info(f"Found {len(elements)} Facebook ads using selector: {selector}")
# Record success
self.selector_stats.update_selector_success(selector, len(elements))
return elements
except Exception as e:
logger.debug(f"Facebook selector {selector} failed: {e}")
# No elements found with standard selectors, try a more aggressive approach
try:
# Look for text patterns that typically appear in ads
patterns = [
"//div[contains(., 'Library ID:')]",
"//div[contains(., 'Sponsored')]",
"//div[contains(., 'Active')][contains(., 'Library ID')]",
"//div[contains(., 'Inactive')][contains(., 'Library ID')]"
]
for pattern in patterns:
elements = self.driver.find_elements(By.XPATH, pattern)
if elements:
ad_containers = []
for element in elements:
try:
# Try to find containing card by navigating up
container = element
for _ in range(5): # Try up to 5 levels up
if container.get_attribute("class") and "card" in container.get_attribute(
"class").lower():
ad_containers.append(container)
break
container = container.find_element(By.XPATH, "..")
except:
continue
if ad_containers:
logger.info(f"Found {len(ad_containers)} Facebook ads using text pattern approach")
# Record this special method
self.selector_stats.update_selector_success("text_pattern_method", len(ad_containers))
return ad_containers
except Exception as e:
logger.debug(f"Facebook text pattern approach failed: {e}")
return []
def _extract_facebook_ad_data(self, ad_element, full_text):
"""Extract data from Facebook ad using multiple intelligent methods"""
extracted_data = {}
# Process text content if available
if full_text:
# Split into lines
lines = full_text.split('\n')
# Check for status (Active/Inactive)
if lines and lines[0] in ["Active", "Inactive"]:
extracted_data["status"] = lines[0]
# Look for advertiser - typically after "See ad details"
for i, line in enumerate(lines):
if "See ad details" in line or "See summary details" in line:
if i + 1 < len(lines):
extracted_data["advertiser"] = lines[i + 1].strip()
break
else:
# First line is likely the advertiser
if lines:
extracted_data["advertiser"] = lines[0].strip()
# Extract ad content
# Look for patterns to determine content boundaries
content_start_idx = -1
content_end_idx = len(lines)
# Find where "Sponsored" appears
for i, line in enumerate(lines):
if "Sponsored" in line:
content_start_idx = i + 1
break
# If no "Sponsored" found, look for advertiser + status
if content_start_idx == -1:
# Skip metadata lines
metadata_patterns = [
"Library ID:",
"Started running on",
"Platforms",
"Open Drop-down",
"See ad details",
"See summary details",
"This ad has multiple versions"
]
for i, line in enumerate(lines):
if any(pattern in line for pattern in metadata_patterns):
continue
if i > 0: # Skip first line (advertiser)
content_start_idx = i
break
# Find where UI elements start
ui_elements = [
"Like", "Comment", "Share", "Learn More", "Shop Now",
"Sign Up", "Visit Instagram profile", "See More"
]
for i, line in enumerate(lines):
# Skip lines before content start
if i <= content_start_idx:
continue
if any(ui in line for ui in ui_elements):
content_end_idx = i
break
# Extract content between boundaries
if content_start_idx != -1 and content_start_idx < content_end_idx:
content_lines = lines[content_start_idx:content_end_idx]
extracted_data["text"] = "\n".join(content_lines).strip()
# If text extraction failed, try element-based approaches
if "text" not in extracted_data or not extracted_data["text"]:
facebook_text_selectors = [
"div[data-ad-preview='message']", # Direct message container
"div[class*='_7jy6']", # Known ad text container
"div[data-testid='ad-creative-text']", # Test ID for ad text
"div[class*='_38ki']", # Another text container
"span[class*='_7oe']", # Text span
"div.text_exposed_root" # Exposed text root
]
for selector in facebook_text_selectors:
try:
elements = ad_element.find_elements(By.CSS_SELECTOR, selector)
text_content = " ".join([e.text.strip() for e in elements if e.text.strip()])
if text_content:
extracted_data["text"] = text_content
break
except:
pass
# If advertiser extraction failed, try element-based approaches
if "advertiser" not in extracted_data or not extracted_data["advertiser"]:
facebook_advertiser_selectors = [
"span[class*='fsl']", # Facebook specific large text class
"a[aria-label*='profile']", # Profile links often contain advertiser name
"h4", # Often contains advertiser name
"div[class*='_8jh5']", # Known advertiser class
"a[role='link']", # Links are often advertiser names
"div[class*='_3qn7']", # Another known advertiser container
"div[class*='_7jvw'] a", # Links within the ad card
]
for selector in facebook_advertiser_selectors:
try:
elements = ad_element.find_elements(By.CSS_SELECTOR, selector)
for element in elements:
text = element.text.strip()
if text and len(text) < 50: # Advertiser names are usually short
extracted_data["advertiser"] = text
break
if "advertiser" in extracted_data and extracted_data["advertiser"]:
break
except:
pass
return extracted_data
def _generate_placeholder_facebook_data(self, query):
"""Generate placeholder Facebook ad data when real ads cannot be scraped"""
logger.info(f"Returning placeholder Facebook ad data for query: {query}")
return [
{
"platform": "Facebook",
"query": query,
"advertiser": "Placeholder Advertiser 1",
"text": f"This is a placeholder ad for {query} since no actual ads could be scraped.",
"timestamp": datetime.now().isoformat(),
"index": 1,
"is_placeholder": True,
"session_id": self.session_id
},
{
"platform": "Facebook",
"query": query,
"advertiser": "Placeholder Advertiser 2",
"text": f"Another placeholder ad for {query}. Please check your scraping settings.",
"timestamp": datetime.now().isoformat(),
"index": 2,
"is_placeholder": True,
"session_id": self.session_id
}
]
def _save_debug_data(self, prefix, query):
"""Save debugging data for investigation"""
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
debug_dir = "debug_data"
if not os.path.exists(debug_dir):
os.makedirs(debug_dir)
# Save screenshot
screenshot_path = f"{debug_dir}/{prefix}_{query}_{timestamp}.png"
self.driver.save_screenshot(screenshot_path)
logger.info(f"Saved debug screenshot to {screenshot_path}")
# Save HTML
html_path = f"{debug_dir}/{prefix}_{query}_{timestamp}.html"
with open(html_path, "w", encoding="utf-8") as f:
f.write(self.driver.page_source)
logger.info(f"Saved debug HTML to {html_path}")
# Save sample of first ad structure if available
try:
ad_elements = self.driver.find_elements(By.CSS_SELECTOR, "div[class*='_7jvw']")
if ad_elements:
first_ad = ad_elements[0]
# Get sample HTML structure
first_ad_html = first_ad.get_attribute('outerHTML')
# Save first ad HTML
sample_path = f"{debug_dir}/{prefix}_sample_ad_{timestamp}.html"
with open(sample_path, "w", encoding="utf-8") as f:
f.write(first_ad_html)
logger.info(f"Saved sample ad HTML to {sample_path}")
# Log the text structure
logger.info(f"Sample ad text structure: {first_ad.text[:300]}...")
except Exception as e:
logger.error(f"Error saving ad sample: {e}")
def close(self):
"""Close the WebDriver and save stats"""
if self.driver:
self.driver.quit()
# Save selector stats one last time
self.selector_stats._save_stats()
# Facebook Gradio Interface Function
def fetch_facebook_ads(query):
"""Fetch Facebook ads only for Gradio interface"""
logger.info(f"Processing Facebook ad search for: {query}")
scraper = FacebookAdsScraper(headless=True, debug_mode=True)
# Perform headless check first
visibility_ok = scraper.check_headless_visibility()
if not visibility_ok:
logger.warning("Headless visibility check failed, results may be affected")
# Fetch ads from Facebook
facebook_ads = scraper.fetch_facebook_ads(query)
# Format for display
formatted_results = []
for ad in facebook_ads:
formatted_ad = f"Platform: {ad['platform']}\n"
# Include status if available
if 'status' in ad:
formatted_ad += f"Status: {ad['status']}\n"
formatted_ad += f"Advertiser: {ad['advertiser']}\n"
# Format ad text with word wrapping
text_lines = []
if ad['text'] and ad['text'] != "Ad content not available":
# Split long text into readable chunks
words = ad['text'].split()
current_line = ""
for word in words:
if len(current_line) + len(word) + 1 <= 80: # 80 chars per line
current_line += (" " + word if current_line else word)
else:
text_lines.append(current_line)
current_line = word
if current_line:
text_lines.append(current_line)
formatted_text = "\n".join(text_lines)
else:
formatted_text = ad['text']
formatted_ad += f"Ad Text: {formatted_text}\n"
formatted_ad += f"Timestamp: {ad['timestamp']}\n"
if ad.get('is_placeholder', False):
formatted_ad += "[THIS IS PLACEHOLDER DATA]\n"
formatted_ad += "-" * 50
formatted_results.append(formatted_ad)
scraper.close()
return "\n\n".join(formatted_results) if formatted_results else "No Facebook ads found for your query."
# Create a function to save ads to JSON
def save_ads_to_json(ads, query):
"""Save ads to a JSON file"""
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
filename = f"facebook_ads_{query.replace(' ', '_')}_{timestamp}.json"
try:
with open(filename, 'w', encoding='utf-8') as f:
json.dump(ads, f, indent=2, ensure_ascii=False)
logger.info(f"Saved ads to {filename}")
return filename
except Exception as e:
logger.error(f"Error saving ads to JSON: {e}")
return None
#####################################
### GOOGLE ADS SCRAPER SECTION #####
#####################################
# Constants for Google Ads Scraper
MAX_ADS_DEFAULT = 5
# Import the actual GoogleAds class and regions
try:
from GoogleAds.main import GoogleAds, show_regions_list
from GoogleAds.regions import Regions
USING_ACTUAL_GOOGLE_ADS = True
logger.info("Successfully imported GoogleAds module")
except ImportError as e:
# Fallback to mock implementation if module is missing
logger.warning(f"GoogleAds module not found: {e}. Using mock implementation.")
USING_ACTUAL_GOOGLE_ADS = False
# Mock Regions dictionary - only used if real module fails to import
Regions = {
"GB": {"Region": "United Kingdom"}
}
def show_regions_list():
"""Mock function - only used if real module fails to import"""
return [("GB", "United Kingdom"), ("US", "United States")]
# Mock GoogleAds class - only used if real module fails to import
class GoogleAds:
def __init__(self, region="GB"):
self.region = region
logger.warning(f"Using MOCK GoogleAds implementation with region: {region}")
logger.warning("Please install the GoogleAds module for actual data")
def creative_search_by_advertiser_id(self, advertiser_id, count=5):
# Mock implementation - only used if real module fails to import
logger.warning(f"MOCK: Searching for creatives from advertiser {advertiser_id}")
return [f"creative_{i}_{advertiser_id}" for i in range(min(count, 3))]
def get_detailed_ad(self, advertiser_id, creative_id):
# Mock implementation - only used if real module fails to import
logger.warning(f"MOCK: Getting details for creative {creative_id}")
# Find advertiser name
advertiser_name = "Unknown"
for adv in ADVERTISERS:
if adv["id"] == advertiser_id:
advertiser_name = adv["name"]
break
# Return mock ad details
return {
"Ad Format": "Text",
"Advertiser": advertiser_name,
"Advertiser Name": advertiser_name,
"Ad Title": f"MOCK DATA - INSTALL GOOGLE ADS MODULE",
"Ad Body": f"This is MOCK data because the GoogleAds module is not installed. Please install the proper module.",
"Last Shown": datetime.now().strftime("%Y-%m-%d"),
"Creative Id": creative_id,
"Ad Link": "#"
}
def clean_ad_text(text):
"""Clean ad text by removing special characters and formatting issues."""
if text is None or not isinstance(text, str):
return ""
# Remove Unicode special characters often found in Google ads data
cleaned = text.replace('â¦', '') # Opening symbol
cleaned = cleaned.replace('â©', '') # Closing symbol
cleaned = cleaned.replace('<dynamically generated based on landing page content>', '[Dynamic Content]')
# Remove any other strange characters that might appear
cleaned = re.sub(r'[^\x00-\x7F]+', '', cleaned)
return cleaned.strip()
def get_regions_list():
"""Get a limited list of regions - only GB and anywhere."""
regions = [
("anywhere", "Global (anywhere)"),
("GB", f"{Regions['GB']['Region']} (GB)")
]
return regions
def search_by_advertiser_id(advertiser_id: str, max_ads=MAX_ADS_DEFAULT, region="GB", progress=gr.Progress(),
provided_name=None) -> Tuple[
str, Optional[pd.DataFrame], Optional[Dict]]:
try:
progress(0, desc="Initializing scraper...")
# Fix for region handling
region_val = region
if isinstance(region, tuple) and len(region) > 0:
region_val = region[0]
# Ensure 'anywhere' is handled correctly
if region_val == "Global (anywhere)" or "anywhere" in str(region_val).lower():
region_val = "anywhere"
# Initialize the Google Ads scraper
scraper = GoogleAds(region=region_val)
progress(0.2, desc=f"Fetching ads for advertiser ID: {advertiser_id}")
# Get creative IDs for this advertiser
creative_ids = scraper.creative_search_by_advertiser_id(advertiser_id, count=max_ads)
if not creative_ids:
return f"No ads found for advertiser ID: {advertiser_id}", None, None
progress(0.3, desc=f"Found {len(creative_ids)} ads. Fetching details...")
# Fetch detailed information for each ad
ads_data = []
ad_formats = {}
for i, creative_id in enumerate(creative_ids):
progress_val = 0.3 + (0.7 * (i / len(creative_ids)))
progress(progress_val, desc=f"Processing ad {i + 1}/{len(creative_ids)}")
try:
ad_details = scraper.get_detailed_ad(advertiser_id, creative_id)
# Fix encoding issues for Ad Title and Ad Body fields
if 'Ad Title' in ad_details:
ad_details['Ad Title'] = clean_ad_text(ad_details['Ad Title'])
if 'Ad Body' in ad_details:
ad_details['Ad Body'] = clean_ad_text(ad_details['Ad Body'])
ads_data.append(ad_details)
# Count ad formats
ad_format = ad_details.get("Ad Format", "Unknown")
ad_formats[ad_format] = ad_formats.get(ad_format, 0) + 1
# Brief pause to avoid overwhelming the server
time.sleep(0.2)
except Exception as e:
print(f"Error fetching details for ad {creative_id}: {e}")
if not ads_data:
return f"Retrieved creative IDs but couldn't fetch ad details for advertiser ID: {advertiser_id}", None, None
# Create a DataFrame for display
df = pd.DataFrame(ads_data)
# Generate summary info
timestamp = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
# Use provided name if available, otherwise try to determine from predefined list or ad data
advertiser_name = "Unknown"
# First, use the provided name if it exists
if provided_name:
advertiser_name = provided_name
else:
# Check our predefined list
for adv in ADVERTISERS:
if adv["id"] == advertiser_id:
advertiser_name = adv["name"]
break
# If still unknown, try to get from the ad data
if advertiser_name == "Unknown" and ads_data and len(ads_data) > 0:
# The field might be "Advertiser" or "Advertiser Name" depending on the version
for field in ["Advertiser", "Advertiser Name", "advertiser_name"]:
if field in ads_data[0]:
advertiser_name = ads_data[0][field]
break
summary = {
'advertiser_id': advertiser_id,
'advertiser_name': advertiser_name,
'ads_count': len(ads_data),
'timestamp': timestamp,
'region': region_val,
'ad_formats': ad_formats
}
# Find the earliest and latest ad
dates = []
for ad in ads_data:
# The field might be "Last Shown" or "last_shown_date" depending on the version
for field in ["Last Shown", "last_shown_date"]:
if field in ad and ad[field]:
dates.append(ad[field])
break
if dates:
summary['earliest_ad'] = min(dates)
summary['latest_ad'] = max(dates)
# Don't save the data, just prepare the summary info
summary = {
'advertiser_id': advertiser_id,
'advertiser_name': advertiser_name,
'ads_count': len(ads_data),
'timestamp': timestamp,
'region': region_val,
'ad_formats': ad_formats
}
# Find the earliest and latest ad
dates = []
for ad in ads_data:
# The field might be "Last Shown" or "last_shown_date" depending on the version
for field in ["Last Shown", "last_shown_date"]:
if field in ad and ad[field]:
dates.append(ad[field])
break
if dates:
summary['earliest_ad'] = min(dates)
summary['latest_ad'] = max(dates)
success_message = (
f"Found {len(ads_data)} ads for advertiser '{advertiser_name}' (ID: {advertiser_id})."
)
progress(1.0, desc="Complete!")
return success_message, df, summary
except Exception as e:
error_message = f"Error searching for advertiser ID: {str(e)}"
return error_message, None, None
def process_advertiser_search(advertiser_selection, region, max_ads, progress=gr.Progress()):
"""Handle the advertiser selection form submission and update the UI."""
# Extract advertiser ID and name from the selection format "ID: Name"
if not advertiser_selection:
return "Please select an advertiser to search", None, None, None
# Split the selection string to get the ID and name
parts = advertiser_selection.split(":", 1)
advertiser_id = parts[0].strip()
advertiser_name = parts[1].strip() if len(parts) > 1 else "Unknown"
# Perform the search
result_message, ads_df, summary_info = search_by_advertiser_id(
advertiser_id, max_ads, region, progress, advertiser_name
)
# Generate analysis if data is available
analysis_html = analyze_ads(ads_df, summary_info) if ads_df is not None and not ads_df.empty else None
return result_message, ads_df, analysis_html, summary_info
def analyze_ads(df: pd.DataFrame, summary: Dict) -> str:
"""
Analyze ads data and generate insights.
Args:
df: DataFrame containing ad data
summary: Dictionary with summary information
Returns:
HTML string with analysis results
"""
if df is None or df.empty or summary is None:
return "<h3>No data available for analysis</h3>"
try:
# Create a simple HTML report with the analysis
html = f"""
<div style="font-family: Arial, sans-serif;">
<h2>{summary.get('advertiser_name', 'Unknown Advertiser')} - Ad Analysis</h2>
<div style="background-color: #f5f5f5; padding: 15px; border-radius: 5px; margin-bottom: 20px;">
<h3>Overview</h3>
<p><b>Advertiser ID:</b> {summary.get('advertiser_id', 'Unknown')}</p>
<p><b>Total Ads Found:</b> {summary['ads_count']}</p>
<p><b>Region:</b> {summary['region']}</p>
<p><b>Data Collected:</b> {summary['timestamp'].replace('_', ' ').replace('-', '/')}</p>
{f"<p><b>Ad Date Range:</b> {summary.get('earliest_ad')} to {summary.get('latest_ad')}</p>" if 'earliest_ad' in summary else ""}
</div>
<div style="display: flex; margin-bottom: 20px;">
<div style="flex: 1; background-color: #f5f5f5; padding: 15px; border-radius: 5px; margin-right: 10px;">
<h3>Ad Format Distribution</h3>
<table style="width: 100%; border-collapse: collapse;">
<tr style="background-color: #eaeaea;">
<th style="text-align: left; padding: 8px; border-bottom: 1px solid #ddd;">Format</th>
<th style="text-align: center; padding: 8px; border-bottom: 1px solid #ddd;">Count</th>
<th style="text-align: center; padding: 8px; border-bottom: 1px solid #ddd;">Percentage</th>
</tr>
"""
total = sum(summary['ad_formats'].values())
for format_name, count in summary['ad_formats'].items():
percentage = (count / total) * 100
html += f"""
<tr>
<td style="padding: 8px; border-bottom: 1px solid #ddd;">{format_name}</td>
<td style="text-align: center; padding: 8px; border-bottom: 1px solid #ddd;">{count}</td>
<td style="text-align: center; padding: 8px; border-bottom: 1px solid #ddd;">{percentage:.1f}%</td>
</tr>
"""
html += """
</table>
</div>
"""
# Common words in ad titles
if 'Ad Title' in df.columns and not df['Ad Title'].isna().all():
from collections import Counter
import re
# Extract words from titles
all_titles = ' '.join(df['Ad Title'].dropna().astype(str).tolist())
words = re.findall(r'\b\w+\b', all_titles.lower())
# Remove common stop words
stop_words = {'the', 'a', 'an', 'and', 'or', 'but', 'in', 'on', 'at', 'to', 'for', 'with', 'by', 'of', 'is',
'are'}
filtered_words = [word for word in words if word not in stop_words and len(word) > 2]
# Count word frequencies
word_counts = Counter(filtered_words).most_common(10)
if word_counts:
html += """
<div style="flex: 1; background-color: #f5f5f5; padding: 15px; border-radius: 5px;">
<h3>Most Common Words in Ad Titles</h3>
<table style="width: 100%; border-collapse: collapse;">
<tr style="background-color: #eaeaea;">
<th style="text-align: left; padding: 8px; border-bottom: 1px solid #ddd;">Word</th>
<th style="text-align: center; padding: 8px; border-bottom: 1px solid #ddd;">Frequency</th>
</tr>
"""
for word, count in word_counts:
html += f"""
<tr>
<td style="padding: 8px; border-bottom: 1px solid #ddd;">{word}</td>
<td style="text-align: center; padding: 8px; border-bottom: 1px solid #ddd;">{count}</td>
</tr>
"""
html += """
</table>
</div>
"""
html += """
</div>
<h3>SEO & Marketing Insights</h3>
<div style="background-color: #f5f5f5; padding: 15px; border-radius: 5px; margin-bottom: 20px;">
"""
# Add general insights
html += f"""
<h4>Competitive Intelligence</h4>
<ul>
<li>The advertiser has been active in advertising until {summary.get('latest_ad', 'recently')}</li>
<li>Their ad strategy focuses primarily on {max(summary['ad_formats'].items(), key=lambda x: x[1])[0]} ads</li>
<li>Consider monitoring changes in their ad frequency and creative strategy over time</li>
</ul>
<h4>UK Market Insights</h4>
<ul>
<li>The ads were collected for the {summary['region']} market</li>
<li>Regular monitoring can reveal seasonal UK advertising patterns</li>
<li>Compare with other regions to identify UK-specific marketing approaches</li>
</ul>
"""
html += """
</div>
<h3>All Ad Examples</h3>
"""
# Add example ads (all of them, not just the most recent)
if not df.empty:
# Sort by Last Shown date if available
if 'Last Shown' in df.columns:
df = df.sort_values(by='Last Shown', ascending=False)
# Get all ads, not just the top 3
for i, (_, ad) in enumerate(df.iterrows()):
html += f"""
<div style="background-color: #f5f5f5; padding: 15px; border-radius: 5px; margin-bottom: 15px;">
<h4>Ad {i + 1}: {ad.get('Creative Id', '')}</h4>
<p><b>Format:</b> {ad.get('Ad Format', 'Unknown')}</p>
<p><b>Last Shown:</b> {ad.get('Last Shown', 'Unknown')}</p>
"""
# Display title and body if available
if 'Ad Title' in ad and pd.notna(ad['Ad Title']) and ad['Ad Title']:
html += f"<p><b>Title:</b> {ad['Ad Title']}</p>"
if 'Ad Body' in ad and pd.notna(ad['Ad Body']) and ad['Ad Body']:
body = ad['Ad Body']
if len(body) > 150:
body = body[:150] + "..."
html += f"<p><b>Body:</b> {body}</p>"
# Display image or video links if available
if 'Image URL' in ad and pd.notna(ad['Image URL']) and ad['Image URL']:
html += f"""<p><img src="{ad['Image URL']}" style="max-width: 300px; max-height: 200px;" /></p>"""
if 'Ad Link' in ad and pd.notna(ad['Ad Link']) and ad['Ad Link'] and ad.get('Ad Format') != 'Text':
html += f"""<p><b>Ad Link:</b> <a href="{ad['Ad Link']}" target="_blank">View Ad</a></p>"""
html += "</div>"
html += """
</div>
"""
return html
except Exception as e:
return f"<h3>Error analyzing data: {str(e)}</h3>"
#####################################
### COMBINED INTERFACE SECTION #####
#####################################
def create_combined_app():
"""Create the combined Gradio interface with Facebook and Google Ads scrapers"""
# Create dropdown choices for advertiser selection
advertiser_choices = [f"{adv['id']}: {adv['name']}" for adv in ADVERTISERS]
with gr.Blocks(title="Combined Ads Transparency Scraper") as app:
gr.Markdown("# Combined Ads Transparency Scraper")
gr.Markdown("## Search for ads from Facebook and Google Ads transparency tools")
# Create tabs for the two different scrapers
with gr.Tabs() as tabs:
# Tab 1: Facebook Ad Library Scraper
with gr.TabItem("Facebook Ad Library"):
gr.Markdown("### Facebook Ad Library Search")
gr.Markdown("Search for ads by brand, domain, or keyword")
with gr.Row():
fb_query_input = gr.Textbox(
label="Search Query",
placeholder="Enter brand, domain or product name",
value=""
)
fb_search_button = gr.Button("Find Facebook Ads", variant="primary")
fb_results_output = gr.Textbox(label="Search Results", lines=20)
fb_save_button = gr.Button("Save Results to JSON")
fb_save_status = gr.Textbox(label="Save Status", lines=1)
# Define the save function for Facebook
def save_fb_results(query, results_text):
if not results_text or "No Facebook ads found" in results_text:
return "No ads to save"
# Get the scraper to fetch fresh ads for JSON format
scraper = FacebookAdsScraper(headless=True, debug_mode=False)
ads = scraper.fetch_facebook_ads(query)
scraper.close()
# Save to JSON
filename = save_ads_to_json(ads, query)
if filename:
return f"Saved {len(ads)} ads to {filename}"
else:
return "Error saving ads to JSON"
# Connect Facebook interface components
fb_search_button.click(
fn=fetch_facebook_ads,
inputs=[fb_query_input],
outputs=[fb_results_output]
)
fb_save_button.click(
fn=save_fb_results,
inputs=[fb_query_input, fb_results_output],
outputs=[fb_save_status]
)
# Tab 2: Lightsaber Companies Google Ads Scraper
with gr.TabItem("Google Ads (Lightsaber Companies)"):
gr.Markdown("### Lightsaber Companies Ads Transparency Scraper")
gr.Markdown("View Google Ads data for popular lightsaber companies")
with gr.Row():
with gr.Column(scale=3):
advertiser_dropdown = gr.Dropdown(
choices=advertiser_choices,
label="Select Lightsaber Company",
info="Choose a company to view their Google Ads data"
)
with gr.Row():
region_dropdown = gr.Dropdown(
choices=get_regions_list(),
value="GB", # UK is the default
label="Region",
info="Choose between Global or UK"
)
max_ads_slider = gr.Slider(
minimum=1,
maximum=10,
value=5,
step=1,
label="Max Ads to Retrieve"
)
search_button = gr.Button("Search Ads", variant="primary")
with gr.Column(scale=2):
result_message = gr.Markdown(label="Search Result")
# Tabs for displaying Google Ads search results
with gr.Tabs() as google_result_tabs:
with gr.Tab("Analysis"):
analysis_html = gr.HTML()
with gr.Tab("Raw Data"):
ads_table = gr.DataFrame()
# State for storing summary info
summary_info = gr.State()
# Connect the Google Ads inputs to the output function
search_button.click(
fn=process_advertiser_search,
inputs=[advertiser_dropdown, region_dropdown, max_ads_slider],
outputs=[result_message, ads_table, analysis_html, summary_info]
)
# About section for the combined app
with gr.Accordion("About This Tool", open=False):
gr.Markdown("""
## About Combined Ads Transparency Scraper
This tool combines two different ad transparency scrapers:
1. **Facebook Ad Library Scraper**: Search for any advertiser's ads on Facebook.
2. **Google Ads Transparency Scraper**: View ads for popular lightsaber companies.
### Technical Details
- The Facebook scraper uses Selenium WebDriver with anti-detection techniques.
- The Google Ads scraper leverages the Google Ad Transparency API.
- Both scrapers include adaptive error handling and fallback mechanisms.
### Usage Notes
- Facebook scraping may take 30-60 seconds to complete
- Search results are not stored permanently
- Use the "Save Results" button to save data for later analysis
**Note**: This tool is intended for research and educational purposes only.
""")
return app
# Main execution
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="Combined Ads Transparency Scraper")
parser.add_argument("--headless", action="store_true", default=True, help="Run in headless mode")
parser.add_argument("--debug", action="store_true", help="Enable debug mode with extra logging")
parser.add_argument("--fb-query", type=str, help="Facebook search query to run directly without Gradio")
parser.add_argument("--google-advertiser", type=str, help="Google Ads advertiser ID to run directly without Gradio")
parser.add_argument("--save", action="store_true", help="Save results to JSON file when using direct query")
args = parser.parse_args()
if args.fb_query:
# Run direct query mode for Facebook
scraper = FacebookAdsScraper(headless=args.headless, debug_mode=args.debug)
scraper.check_headless_visibility()
facebook_ads = scraper.fetch_facebook_ads(args.fb_query)
# Display results
print(f"\nFound {len(facebook_ads)} Facebook ads for '{args.fb_query}'")
if facebook_ads:
for i, ad in enumerate(facebook_ads):
print(f"\n--- Ad {i + 1} ---")
print(f"Platform: {ad['platform']}")
if 'status' in ad:
print(f"Status: {ad['status']}")
print(f"Advertiser: {ad['advertiser']}")
print(f"Text: {ad['text']}")
if ad.get('is_placeholder', False):
print("[THIS IS PLACEHOLDER DATA]")
# Save to JSON if requested
if args.save:
filename = save_ads_to_json(facebook_ads, args.fb_query)
if filename:
print(f"\nSaved {len(facebook_ads)} ads to {filename}")
else:
print("No Facebook ads found.")
scraper.close()
elif args.google_advertiser:
# Run direct query mode for Google Ads
advertiser_id = args.google_advertiser
# Find advertiser name if it's in our list
advertiser_name = "Unknown"
for adv in ADVERTISERS:
if adv["id"] == advertiser_id:
advertiser_name = adv["name"]
break
print(f"\nSearching for Google Ads from advertiser '{advertiser_name}' (ID: {advertiser_id})")
# Use a dummy progress object for CLI
class DummyProgress:
def __call__(self, value, desc=None):
if desc:
print(f"{desc} ({value * 100:.0f}%)")
result_message, ads_df, summary_info = search_by_advertiser_id(
advertiser_id,
max_ads=5,
region="GB",
progress=DummyProgress(),
provided_name=advertiser_name
)
print(f"\n{result_message}")
if ads_df is not None and not ads_df.empty:
print("\nFound ads:")
for i, (_, ad) in enumerate(ads_df.iterrows()):
print(f"\n--- Ad {i + 1} ---")
print(f"Format: {ad.get('Ad Format', 'Unknown')}")
print(f"Title: {ad.get('Ad Title', 'Unknown')}")
body_text = ad.get('Ad Body', 'Unknown')
if len(body_text) > 100:
body_text = body_text[:100] + "..."
print(f"Body: {body_text}")
print(f"Last Shown: {ad.get('Last Shown', 'Unknown')}")
print(f"Creative ID: {ad.get('Creative Id', 'Unknown')}")
else:
print("No Google ads found or error occurred.")
else:
# Run Gradio interface
app = create_combined_app()
print("Starting Combined Ads Transparency Scraper")
print("Facebook: Search for any brand or company")
print("Google Ads: Available lightsaber companies:")
for adv in ADVERTISERS:
print(f" - {adv['name']}")
app.launch()