template-fingerprinting

Template Fingerprinting Tool

Automatically identify and classify pages by their template type using HTML structure analysis and machine learning clustering.

Features

• Analyzes HTML structure (tags, classes, IDs, meta tags)
• Uses TF-IDF vectorization for feature extraction
• K-Means clustering to identify template patterns
• Identifies common structural patterns per cluster
• Exports results with page type classifications

Use Cases

• Identify different page templates on a website (PDP, PLP, blog, etc.)
• Audit template usage across large sites
• Find pages with unusual/broken templates
• Group pages for template-specific SEO recommendations

Requirements

pip install -r requirements.txt

Usage

1. Export URLs from Screaming Frog (or create a CSV with an Address column)
2. Update configuration variables in the script:
   • INPUT_FILE: Path to your CSV file
   • OUTPUT_FILE: Where to save results
   • N_CLUSTERS: Number of template types to identify
3. Run the script:

python template_fingerprinting.py

Configuration

Variable	Default	Description
INPUT_FILE	./urls.csv	CSV file with 'Address' column
OUTPUT_FILE	./classified_urls.csv	Output file path
N_CLUSTERS	5	Number of template types to detect
TIMEOUT	10	HTTP request timeout in seconds

Output

The script generates a CSV file with:

• Original URL data
• Cluster: Numeric cluster ID
• Page Type: Human-readable type label (Type 0, Type 1, etc.)

Console output includes top features for each cluster to help identify what each template type represents.

How It Works

1. Feature Extraction: For each URL, the script fetches HTML and extracts:

   • Tag counts (e.g., div:15, article:1)
   • CSS class names
   • ID attributes
   • Meta tag properties

2. Vectorization: Features are converted to TF-IDF vectors

3. Clustering: K-Means groups similar page structures

4. Analysis: Top features per cluster help identify template types

Author

Lee Foot - eCommerce SEO Consultant