import requests from bs4 import BeautifulSoup import cssutils import logging Tắt các…
- User
import requests from bs4 import BeautifulSoup import cssutils import logging
Tắt các thông báo cảnh báo không cần thiết từ cssutils
cssutils.log.setLevel(logging.CRITICAL)
def extract_css(url): # Tải nội dung trang web response = requests.get(url) soup = BeautifulSoup(response.text, 'html.parser')
# Trích xuất tất cả CSS all_css = "" # CSS nội tuyến for style in soup.find_all('style'): all_css += style.string or "" # CSS từ các file external for link in soup.find_all('link', rel='stylesheet'): css_url = link.get('href') if css_url: if not css_url.startswith(('http:', 'https:')): css_url = url + css_url if css_url.startswith('/') else url + '/' + css_url css_response = requests.get(css_url) all_css += css_response.text # Trích xuất các phần CSS liên quan relevant_css = "" # Sử dụng cssutils để phân tích CSS sheet = cssutils.parseString(all_css) relevant_properties = [ 'background-color', 'background', 'color', 'font-family', 'button', 'btn', '.button', '.btn', # Các class và element liên quan đến button 'body' # Element body ] for rule in sheet: if rule.type == rule.STYLE_RULE: if _is_relevant_rule(rule, relevant_properties): relevant_css += rule.cssText + "\n" elif rule.type == rule.MEDIA_RULE: for subrule in rule: if subrule.type == subrule.STYLE_RULE: if _is_relevant_rule(subrule, relevant_properties): relevant_css += rule.cssText + "\n" break return relevant_cssdef _is_relevant_rule(rule, relevant_properties): if any(prop in rule.selectorText.lower() for prop in relevant_properties): return True
style = rule.style for property in style: if property.name in relevant_properties: return True # Kiểm tra các thuộc tính liên quan đến button if 'button' in property.name.lower() or 'btn' in property.name.lower(): return True return FalseSử dụng hàm
url = "https://neuralpit.com/" # Thay thế bằng URL thực tế result = extract_css(url) print(result) \n Cho đoạn code, mục đích từ web url mình lấy các dòng css code liên quan đến: background_color: str = Field(description="the primary background color used on the website. Must be a valid hex color code and not transparent.") button_color: str = Field(description="The primary color used for buttons on the website. Must be a valid hex color code, not transparent.") text_color: str = Field(description="the primary text color used on the website. Must be a valid hex color code and not transparent.") font_style: str = Field(description="the main font style used on the website.") Để sau đó LLMs sẽ dựa vào đó để phân tích, trích xuất ra các properties trên. Làm thế nào lấy không thừa (thừa thì tốn prompt dẫn đến tăng cost), cũng không thiếu (thiếu thì ko đủ thông tin, phân tích sai).
- Assistant
To achieve precise extraction of CSS properties related to
background_color,button_color,text_color, andfont_stylefrom a given URL, you can refine your existing script to focus specifically on these properties. This will help minimize unnecessary CSS code extraction, reducing prompt size and cost when passing the data to an LLM for analysis.Below are the steps and an improved version of your script to accomplish this:
-
Define Target Properties and Selectors:
- Properties:
background-colorcolorfont-family
- Selectors:
- Elements related to buttons such as
button,.button,btn,.btn - The
bodyelement for overall styling
- Elements related to buttons such as
- Properties:
-
Extract Only Relevant CSS Declarations:
- Instead of extracting entire CSS rules, focus on extracting only the declarations that pertain to the target properties.
- This ensures that the extracted CSS is concise and directly useful for the LLM to analyze.
-
Handle External CSS URLs Correctly:
- Ensure that relative URLs are correctly resolved based on the base URL of the website.
-
Optimize the Extraction Function:
- Modify the extraction logic to collect only the necessary declarations.
- Use dictionaries to store the properties for each selector to avoid redundancy and enhance clarity.
Improved Script
import requests from bs4 import BeautifulSoup import cssutils import logging from urllib.parse import urljoin # Suppress cssutils warnings cssutils.log.setLevel(logging.CRITICAL) def extract_relevant_css(url): # Fetch the webpage content response = requests.get(url) soup = BeautifulSoup(response.text, 'html.parser') # Collect all CSS content all_css = "" # Extract inline CSS for style in soup.find_all('style'): if style.string: all_css += style.string + "\n" # Extract external CSS for link in soup.find_all('link', rel='stylesheet'): css_url = link.get('href') if css_url: full_css_url = urljoin(url, css_url) try: css_response = requests.get(full_css_url) if css_response.status_code == 200: all_css += css_response.text + "\n" except requests.exceptions.RequestException as e: print(f"Failed to fetch CSS from {full_css_url}: {e}") # Parse the combined CSS sheet = cssutils.parseString(all_css) # Define target properties and selectors target_properties = { 'background_color': 'background-color', 'button_color': 'background-color', # Assuming button color is set via background-color 'text_color': 'color', 'font_style': 'font-family' } target_selectors = [ 'button', '.button', 'btn', '.btn', 'body' ] # Dictionary to hold extracted properties extracted_properties = { 'background_color': set(), 'button_color': set(), 'text_color': set(), 'font_style': set() } for rule in sheet: if rule.type != rule.STYLE_RULE: continue # Skip non-style rules selectors = [selector.strip().lower() for selector in rule.selectorText.split(',')] # Check if any selector is in target_selectors relevant = any(sel in target_selectors for sel in selectors) if not relevant: continue # Skip rules that don't target the desired selectors for property in rule.style: prop_name = property.name.lower() prop_value = property.value.strip() if prop_name == target_properties['background_color']: if selectors[0] in ['button', '.button', 'btn', '.btn']: extracted_properties['button_color'].add(prop_value) elif selectors[0] == 'body': extracted_properties['background_color'].add(prop_value) elif prop_name == target_properties['text_color']: if selectors[0] == 'body' or selectors[0] in ['button', '.button', 'btn', '.btn']: extracted_properties['text_color'].add(prop_value) elif prop_name == target_properties['font_style']: extracted_properties['font_style'].add(prop_value) # Format the extracted properties as CSS declarations formatted_css = "" for key, values in extracted_properties.items(): for value in values: if key == 'background_color': formatted_css += f"body {{ background-color: {value}; }}\n" elif key == 'button_color': formatted_css += f".btn, button {{ background-color: {value}; }}\n" elif key == 'text_color': formatted_css += f"body {{ color: {value}; }}\n" elif key == 'font_style': formatted_css += f"body {{ font-family: {value}; }}\n" return formatted_css.strip() # Usage if __name__ == "__main__": url = "https://neuralpit.com/" # Replace with the actual URL relevant_css = extract_relevant_css(url) print(relevant_css)Explanation of Improvements
-
Focused Property Extraction:
- The script now specifically looks for the properties
background-color,color, andfont-family. - It differentiates between general background color (
background_color) and button-specific background color (button_color).
- The script now specifically looks for the properties
-
Selector Handling:
- The script targets specific selectors related to buttons (
button,.button,btn,.btn) and thebodyelement. - This ensures that only relevant CSS declarations are extracted.
- The script targets specific selectors related to buttons (
-
Efficient URL Handling:
- Utilizes
urljoinfrom theurllib.parsemodule to correctly handle relative URLs for external CSS files.
- Utilizes
-
Avoiding Redundancy:
- Uses sets to store unique property values, preventing duplicate entries in the final output.
-
Structured Output:
- The extracted CSS is formatted in a structured manner, associating each property with its respective selector.
- This makes it easier for the LLM to parse and understand the context of each property.
-
Error Handling:
- Includes basic error handling for network requests to external CSS files, ensuring the script doesn't fail abruptly.
Final Output Example
When you run the script with the provided URL, the output will look something like this (values will vary based on the actual website's CSS):
body { background-color: #ffffff; } .btn, button { background-color: #ff5733; } body { color: #333333; } body { font-family: 'Arial, sans-serif'; }Passing to the LLM for Analysis
With the extracted and formatted CSS, you can now pass this concise information to an LLM for further analysis or processing. Here's a suggested prompt structure:
Here are the extracted CSS properties from the website:body { background-color: #ffffff; } .btn, button { background-color: #ff5733; } body { color: #333333; } body { font-family: 'Arial, sans-serif'; }
Please analyze and provide the following information: - **Primary Background Color:** - **Button Color:** - **Primary Text Color:** - **Main Font Style:** Ensure that each color is a valid hex code and not transparent.Benefits of This Approach
- Precision: Only relevant CSS declarations are extracted, reducing unnecessary data.
- Clarity: Structured output makes it easier for the LLM to understand and process the information.
- Efficiency: Minimizing the amount of data sent to the LLM reduces prompt size and associated costs.
- Scalability: The script can be easily adapted to include more properties or selectors as needed.
By implementing these improvements, you ensure that the CSS extraction process is both efficient and effective, providing the LLM with exactly the information it needs for accurate analysis.
-