from flask import Flask, request, render_template_string import json from collections import Counter app = Flask(__name__) # Load the JSON data from the file with open("output.json", "r", encoding="utf-8") as f: data = json.load(f) # Function to analyze the data def analyze_data(results): # Filter out unwanted titles and hostnames filtered_titles = [ entry["title"] for entry in results if entry.get("title") and entry["title"].lower() not in ["unknown", "no title"] ] filtered_hostnames = [ entry["hostname"] for entry in results if entry.get("hostname") and entry["hostname"].lower() != "unknown" ] # Find the most common titles and hostnames most_common_titles = Counter(filtered_titles).most_common(25) most_common_hostnames = Counter(filtered_hostnames).most_common(20) return most_common_titles, most_common_hostnames # HTML Template HTML_TEMPLATE = """ Badger search engine

Badger Search

Results ({{ results_count }} found)

IP: {{ result.ip }}
Hostname: {{ result.hostname }}
Title: {{ result.title }}
Description: {{ result.description }}
No results found.

Statistics

Most Common Titles

{{ title }} ({{ count }})

Most Common Hostnames

{{ hostname }} ({{ count }})

""" @app.route("/", methods=["GET", "POST"]) def index(): results = data["results"] # Access the results key in the data search_query = "" filters = { "has_title": False, "has_description": False, "has_hostname": False } if request.method == "POST": # Get search query search_query = request.form.get("search", "").strip().lower() filters["has_title"] = "has_title" in request.form filters["has_description"] = "has_description" in request.form filters["has_hostname"] = "has_hostname" in request.form # Split the query by space to handle the negative term search positive_terms = [] negative_terms = [] for term in search_query.split(): if term.startswith('-'): negative_terms.append(term[1:]) # Remove the "-" for negative terms else: positive_terms.append(term) # Filter based on the positive terms if positive_terms: results = [ entry for entry in results if any(term in entry["ip"].lower() for term in positive_terms) or any(term in entry["title"].lower() for term in positive_terms) or any(term in entry["description"].lower() for term in positive_terms) or any(term in entry["hostname"].lower() for term in positive_terms) ] # Exclude based on the negative terms if negative_terms: results = [ entry for entry in results if not any(term in entry["ip"].lower() for term in negative_terms) and not any(term in entry["title"].lower() for term in negative_terms) and not any(term in entry["description"].lower() for term in negative_terms) and not any(term in entry["hostname"].lower() for term in negative_terms) ] # Apply the checkbox filters if filters["has_title"]: results = [entry for entry in results if entry["title"] and entry["title"].lower() != "no title"] if filters["has_description"]: results = [entry for entry in results if entry["description"] and entry["description"].lower() != "no description"] if filters["has_hostname"]: results = [entry for entry in results if entry["hostname"] and entry["hostname"].lower() != "unknown"] # Calculate the most common titles and hostnames most_common_titles, most_common_hostnames = analyze_data(data["results"]) # Calculate the number of results results_count = len(results) return render_template_string( HTML_TEMPLATE, results=results, search_query=search_query, filters=filters, results_count=results_count, most_common_titles=most_common_titles, most_common_hostnames=most_common_hostnames ) if __name__ == "__main__": app.run(debug=True)