from bs4 import BeautifulSoup
import json
def generate_form_names(html_content):
"""
Generates unique form names for text and image elements in an HTML page, modifies the HTML, and creates a corresponding JSON file.
1. Parses the HTML to locate text (h1, h2, h3, p) and image (img) elements.
2. For each element found:
- Generates a unique form name using the format `{page-class-name}_{parent-class-name}_{element-type}_{index}`.
- Assigns the form name as an `id` attribute and replaces text content or image `src` with Django's template syntax.
3. Builds a JSON file containing form names, element types, and metadata like `sub_type` and `tags`.
4. Returns the modified HTML with form names and a JSON structure for further use.
This automates form name generation and content replacement, simplifying integration with CMS app.
"""
# Parse the HTML content
soup = BeautifulSoup(html_content, 'html.parser')
json_output = {
"About": []
}
text_counter = 0 #FIX TEXT COUNTER TO BE FOR EACH ELEMNT
img_counter = 0
# Find all divs with the 'page' class
for page_div in soup.find_all('div', class_='page'):
page_class = page_div.get('class')[0] # Get the first class as the page name
# For each text element (h1, h2, h3, p)
for el in page_div.find_all(['h1', 'h2', 'h3', 'p']):
parent_class = el.find_parent().get('class')[0] # Get the parent's class name
tag_name = el.name.lower() # Element type (e.g., h1, h3, etc.)
unique_form_name = f"{page_class}_{parent_class}_{tag_name}_{text_counter}"
text_counter += 1
# Set the ID in the HTML
el['id'] = unique_form_name
# Update the content with Django template tag
el.string = f"{{{{content.{unique_form_name}}}}}"
# Add to the JSON output
json_output["About"].append({
"title": "Lorem Ipsum",
"type": "text",
"max_length": 255,
"form_name": unique_form_name,
"sub_type": "regular_field",
"tags": ""
})
# For each image element
for img in page_div.find_all('img'):
parent_class = img.find_parent().get('class')[0] # Get the parent's class name
tag_name = img.name.lower() # Should always be 'img'
unique_form_name = f"{page_class}_{parent_class}_{tag_name}_{img_counter}"
img_counter += 1
# Set the ID in the HTML
img['id'] = unique_form_name
# Update the 'src' with Django template tag
img['src'] = f"{{% static 'img/' %}}{{{{content.{unique_form_name}}}}}"
# Add to the JSON output
json_output["About"].append({
"title": "Lorem Ipsum",
"type": "image",
"form_name": unique_form_name,
"sub_type": "drop_down",
"tags": "img"
})
# Convert modified soup object back to a string
modified_html = str(soup)
# Convert json_output to a JSON string
json_string = json.dumps(json_output, indent=4)
# Return the modified HTML and JSON
return modified_html, json_string
# Example HTML input
html_content = """
{% load static %}
{% load widget_tweaks %}
CMS - Demo
Company Profile Template Demo
Hi thereeeee
Company Profile Template Demo
Hi thereeeee
"""
# Call the function to generate form names and modify HTML
modified_html, json_output = generate_form_names(html_content)
# Save the modified HTML to a file
with open('modified_about.html', 'w') as html_file:
html_file.write(modified_html)
# Save the JSON output to a file
with open('about_content_index.json', 'w') as json_file:
json_file.write(json_output)
# Output to verify
print("Modified HTML:")
print(modified_html)
print("\nGenerated JSON:")
print(json_output)
