252 lines
8.0 KiB
Python
252 lines
8.0 KiB
Python
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 = """
|
|
<!DOCTYPE html>
|
|
<html lang="en">
|
|
<head>
|
|
<meta charset="UTF-8">
|
|
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
|
<title>Badger search engine</title>
|
|
<style>
|
|
body {
|
|
font-family: Arial, sans-serif;
|
|
background-color: #20201d;
|
|
color: #a6a28c;
|
|
margin: 20px;
|
|
display: flex;
|
|
flex-direction: row;
|
|
}
|
|
h1 {
|
|
color: #e0e0e0;
|
|
}
|
|
form {
|
|
margin-bottom: 20px;
|
|
}
|
|
input[type="text"] {
|
|
width: 300px;
|
|
padding: 8px;
|
|
margin-right: 10px;
|
|
background-color: #333;
|
|
color: #a6a28c;
|
|
border: 1px solid #555;
|
|
border-radius: 4px;
|
|
}
|
|
label {
|
|
margin-right: 15px;
|
|
color: #e0e0e0;
|
|
}
|
|
button {
|
|
padding: 8px 16px;
|
|
background-color: #444;
|
|
color: #a6a28c;
|
|
border: 1px solid #555;
|
|
cursor: pointer;
|
|
border-radius: 4px;
|
|
}
|
|
button:hover {
|
|
background-color: #555;
|
|
}
|
|
ul {
|
|
list-style-type: none;
|
|
padding: 0;
|
|
}
|
|
li {
|
|
margin-bottom: 10px;
|
|
padding: 10px;
|
|
background: #333;
|
|
color: #a6a28c;
|
|
border: 1px solid #555;
|
|
border-radius: 4px;
|
|
}
|
|
a {
|
|
color: #4e9fff;
|
|
text-decoration: none;
|
|
}
|
|
a:hover {
|
|
color: #80d4ff;
|
|
text-decoration: underline;
|
|
}
|
|
#statistics {
|
|
flex-shrink: 0;
|
|
width: 300px;
|
|
background-color: #2c2c28;
|
|
box-shadow: -4px 0 6px rgba(0, 0, 0, 0.3);
|
|
color: #a6a28c;
|
|
padding: 20px;
|
|
border-left: 1px solid #555;
|
|
height: 100vh;
|
|
overflow-y: auto;
|
|
}
|
|
#statistics h2, #statistics h3 {
|
|
margin-top: 0;
|
|
color: #e0e0e0;
|
|
}
|
|
#main-content {
|
|
flex: 1;
|
|
margin-right: 20px;
|
|
}
|
|
@media (max-width: 600px) {
|
|
#statistics {
|
|
width: 100%;
|
|
height: auto;
|
|
}
|
|
}
|
|
</style>
|
|
</head>
|
|
<body>
|
|
<div id="main-content">
|
|
<h1>Badger Search</h1>
|
|
<form method="POST">
|
|
<input type="text" name="search" placeholder="Search..." value="{{ search_query }}">
|
|
<div>
|
|
<label>
|
|
<input type="checkbox" name="has_title" {% if filters['has_title'] %}checked{% endif %}>
|
|
Has Title
|
|
</label>
|
|
<label>
|
|
<input type="checkbox" name="has_description" {% if filters['has_description'] %}checked{% endif %}>
|
|
Has Description
|
|
</label>
|
|
<label>
|
|
<input type="checkbox" name="has_hostname" {% if filters['has_hostname'] %}checked{% endif %}>
|
|
Has Hostname
|
|
</label>
|
|
</div>
|
|
<button type="submit">Search</button>
|
|
</form>
|
|
<h2>Results ({{ results_count }} found)</h2>
|
|
<ul>
|
|
{% for result in results %}
|
|
<li>
|
|
<strong>IP:</strong> <a href="http://{{ result.ip }}" target="_blank">{{ result.ip }}</a><br>
|
|
<strong>Hostname:</strong> {{ result.hostname }}<br>
|
|
<strong>Title:</strong> {{ result.title }}<br>
|
|
<strong>Description:</strong> {{ result.description }}
|
|
</li>
|
|
{% else %}
|
|
<li>No results found.</li>
|
|
{% endfor %}
|
|
</ul>
|
|
</div>
|
|
<div id="statistics">
|
|
<h2>Statistics</h2>
|
|
<h3>Most Common Titles</h3>
|
|
<ul>
|
|
{% for title, count in most_common_titles %}
|
|
<li>{{ title }} ({{ count }})</li>
|
|
{% endfor %}
|
|
</ul>
|
|
<h3>Most Common Hostnames</h3>
|
|
<ul>
|
|
{% for hostname, count in most_common_hostnames %}
|
|
<li>{{ hostname }} ({{ count }})</li>
|
|
{% endfor %}
|
|
</ul>
|
|
</div>
|
|
</body>
|
|
</html>
|
|
"""
|
|
|
|
@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)
|