chg: [domains search] search domains by name

bin/lib/Domain.py

@@ -10,9 +10,10 @@ The ``Domain``
 import os
 import sys
 import itertools
-import time
+import re
 import redis
 import random
+import time
 
 sys.path.append(os.path.join(os.environ['AIL_BIN'], 'packages/'))
 import Cryptocurrency
@@ -241,6 +242,52 @@ def create_domains_metadata_list(list_domains, domain_type):
                             ports=True, tags=True, languages=True, screenshot=True, tags_safe=True))
     return l_domains
 
+def sanithyse_domain_name_to_search(name_to_search, domain_type):
+    if domain_type == 'onion':
+        r_name = r'[a-z0-9\.]+'
+    else:
+        r_name = r'[a-zA-Z0-9\.-_]+'
+    # invalid domain name
+    if not re.fullmatch(r_name, name_to_search):
+        return None
+    return name_to_search.replace('.', '\.')
+
+
+def search_domains_by_name(name_to_search, domain_types, r_pos=False):
+    domains_dict = {}
+    for domain_type in domain_types:
+        r_name = sanithyse_domain_name_to_search(name_to_search, domain_type)
+        if not name_to_search:
+            break
+        r_name = re.compile(r_name)
+        for domain in get_all_domains_up(domain_type):
+            res = re.search(r_name, domain)
+            if res:
+                domains_dict[domain] = {}
+                if r_pos:
+                    domains_dict[domain]['hl-start'] = res.start()
+                    domains_dict[domain]['hl-end'] = res.end()
+    return domains_dict
+
+def api_search_domains_by_name(name_to_search, domains_types, domains_metadata=False, page=1):
+    domains_types = sanitize_domain_types(domains_types)
+    domains_dict = search_domains_by_name(name_to_search, domains_types, r_pos=True)
+    l_domains = sorted(domains_dict.keys())
+    l_domains = paginate_iterator(l_domains, nb_obj=28, page=page)
+    if not domains_metadata:
+        return l_domains
+    else:
+        l_dict_domains = []
+        for domain in l_domains['list_elem']:
+            dict_domain = get_domain_metadata(domain, get_domain_type(domain), first_seen=True, last_ckeck=True,
+                                                        status=True, ports=True, tags=True, tags_safe=True,
+                                                        languages=True, screenshot=True)
+            dict_domain = {**domains_dict[domain], **dict_domain}
+            l_dict_domains.append(dict_domain)
+        l_domains['list_elem'] = l_dict_domains
+        l_domains['search'] = name_to_search
+        return l_domains
+
 
 ######## LANGUAGES ########
 def get_all_domains_languages():
@@ -940,3 +987,6 @@ class Domain(object):
         '''
         port = sanathyse_port(port, self.domain, self.type, strict=True, current_port=self.current_port)
         return get_domain_items_crawled(self.domain, self.type, port, epoch=epoch, items_link=items_link, item_screenshot=item_screenshot, item_tag=item_tag)
+
+if __name__ == '__main__':
+    search_domains_by_name('c', 'onion')

bin/lib/crawlers.py

@@ -47,6 +47,20 @@ faup = Faup()
 def generate_uuid():
     return str(uuid.uuid4()).replace('-', '')
 
+def is_valid_onion_domain(domain):
+    if not domain.endswith('.onion'):
+        return False
+    domain = domain.replace('.onion', '', 1)
+    if len(domain) == 16: # v2 address
+        r_onion = r'[a-z0-9]{16}'
+        if re.match(r_onion, domain):
+            return True
+    elif len(domain) == 56: # v3 address
+        r_onion = r'[a-z0-9]{56}'
+        if re.fullmatch(r_onion, domain):
+            return True
+    return False
+
 ################################################################################
 
 # # TODO: handle prefix cookies

var/www/blueprints/crawler_splash.py

@@ -232,6 +232,25 @@ def domains_search_languages_get():
                                 l_dict_domains=l_dict_domains, bootstrap_label=bootstrap_label,
                                 current_languages=languages, domains_types=domains_types)
 
+@crawler_splash.route('/domains/name/search', methods=['GET'])
+@login_required
+@login_analyst
+def domains_search_name():
+    name = request.args.get('name')
+    page = request.args.get('page')
+    try:
+        page = int(page)
+    except:
+        page = 1
+    domains_types = request.args.getlist('domain_types')
+    if domains_types:
+        domains_types = domains_types[0].split(',')
+
+    l_dict_domains = Domain.api_search_domains_by_name(name, domains_types, domains_metadata=True, page=page)
+    return render_template("domains/domains_result_list.html", template_folder='../../',
+                                l_dict_domains=l_dict_domains, bootstrap_label=bootstrap_label,
+                                domains_types=domains_types)
+
 ##--  --##
 
 

var/www/modules/hiddenServices/templates/Crawler_dashboard.html

@@ -105,23 +105,7 @@
 					</div>
 				</div>
 
-			<div class="d-flex justify-content-center my-4">
+			{% include 'domains/block_domains_name_search.html' %}
-				<div class="card border-secondary" style="max-width: 40rem;">
-					<div class="card-body text-dark">
-						<h5 class="card-title">Show Domain:</h5>
-						<form class="" action="{{url_for('crawler_splash.showDomain')}}" method="post">
-							<div class="input-group mb-3">
-							  <input type="text" class="form-control" style="min-width: 30rem;" placeholder="Domain name" aria-label="Domain name" aria-describedby="btn_show_domain" id="in_show_domain" , name="in_show_domain">
-							  <div class="input-group-append">
-							    <button class="btn btn-info" type="submit" id="btn_show_domain">
-										<i class="fas fa-search"></i>
-									</button>
-							  </div>
-							</div>
-						</form>
-					</div>
-				</div>
-			</div>
 
 
 			<hr>

var/www/static/css/ail-project.css

@@ -0,0 +1,7 @@
+.hg-text{
+    padding-top: 0.2em;
+    padding-bottom: 0.2em;
+    padding-right: 0.15em;
+    padding-left: 0.15em;
+    background-color: #2e5;
+}

var/www/templates/domains/block_domains_name_search.html

@@ -0,0 +1,50 @@
+<div class="d-flex justify-content-center my-4">
+	<div class="card border-secondary" style="max-width: 40rem;">
+		<div class="card-body text-dark">
+			<h5 class="card-title">Search Domain by name:</h5>
+			<div class="input-group mb-3">
+				<input type="text" class="form-control" id="in_search_name" value="{{search}}" style="min-width: 30rem;" placeholder="Domain name" aria-label="Domain name" aria-describedby="btn_show_domain">
+				<div class="input-group-append">
+					<button class="btn btn-info" type="button" id="btn_search_name" onclick="searchDomainName()">
+						<i class="fas fa-search"></i>
+					</button>
+				</div>
+			</div>
+
+			<div class="mb-3">
+				<div class="custom-control custom-switch">
+					<input class="custom-control-input" type="checkbox" name="domain_onion_switch" value="" id="domain_onion_switch" {%if 'onion' in domains_types or not domains_types%}checked{%endif%}>
+					<label class="custom-control-label" for="domain_onion_switch">
+						<span class="badge badge-danger"><i class="fas fa-user-secret"></i> Onion Domains</span>
+					</label>
+				</div>
+				<div class="custom-control custom-switch">
+					<input class="custom-control-input" type="checkbox" name="domain_regular_switch" value="True" id="domain_regular_switch"{%if 'regular' in domains_types%}checked{%endif%}>
+					<label class="custom-control-label" for="domain_regular_switch">
+						<span class="badge badge-warning"><i class="fab fa-html5"></i> Web Domains</span>
+					</label>
+				</div>
+			</div>
+
+		</div>
+	</div>
+</div>
+
+<script>
+  function searchDomainName() {
+		var all_domain_types = ['onion', 'regular'] // TODO: load from flask
+		var l_domains_types = [];
+
+		console.log(document.getElementById('in_search_name'));
+
+  	var data = document.getElementById('in_search_name').value;
+		for (var i = 0; i < all_domain_types.length; i++) {
+			if (document.getElementById('domain_'+ all_domain_types[i] +'_switch').checked) {
+				l_domains_types.push(all_domain_types[i])
+			}
+		}
+
+		var parameter = "?name=" + data + "&domain_types=" + l_domains_types  +"{%if page%}&page={{ page }}{%endif%}";
+		window.location.href = "{{ url_for('crawler_splash.domains_search_name') }}" + parameter;
+  }
+</script>

var/www/templates/domains/card_img_domain.html

@@ -12,7 +12,7 @@
           <h5 class="card-title">
             <a target="_blank" href="{{ url_for('crawler_splash.showDomain') }}?domain={{dict_domain["id"]}}">
               {% if 'hl-start' in  dict_domain %}
-                {{dict_domain["id"][:dict_domain['hl-start']]}}<mark>{{dict_domain["id"][dict_domain['hl-start']:dict_domain['hl-end']]}}</mark>{{dict_domain["id"][dict_domain['hl-end']:]}}
+                {{dict_domain["id"][:dict_domain['hl-start']]}}<span class="hg-text">{{dict_domain["id"][dict_domain['hl-start']:dict_domain['hl-end']]}}</span>{{dict_domain["id"][dict_domain['hl-end']:]}}
               {% else %}
                 {{dict_domain["id"]}}
               {% endif %}

var/www/templates/domains/domains_result_list.html

@@ -0,0 +1,195 @@
+<!DOCTYPE html>
+<html>
+<head>
+  <title>Domain Search - AIL</title>
+	<link rel="icon" href="{{ url_for('static', filename='image/ail-icon.png') }}">
+	<!-- Core CSS -->
+	<link href="{{ url_for('static', filename='css/bootstrap4.min.css') }}" rel="stylesheet">
+	<link href="{{ url_for('static', filename='css/font-awesome.min.css') }}" rel="stylesheet">
+  <link href="{{ url_for('static', filename='css/ail-project.css') }}" rel="stylesheet">
+	<!-- JS -->
+	<script src="{{ url_for('static', filename='js/jquery.js')}}"></script>
+  <script src="{{ url_for('static', filename='js/popper.min.js')}}"></script>
+  <script src="{{ url_for('static', filename='js/bootstrap4.min.js') }}"></script>
+
+  <style>
+    .card-columns {
+      column-count: 4;
+    }
+  </style>
+
+</head>
+<body>
+
+	{% include 'nav_bar.html' %}
+
+	<div class="container-fluid">
+		<div class="row">
+
+			{% include 'crawler/menu_sidebar.html' %}
+
+			<div class="col-12 col-lg-10" id="core_content">
+
+        <div class="row">
+          	<div class="col-12 col-lg-6">
+
+              {% with page=l_dict_domains['page'], search=l_dict_domains['search'] %}
+                {% include 'domains/block_domains_name_search.html' %}
+              {% endwith %}
+
+
+            </div>
+            <div class="col-12 col-xl-6">
+
+              <div class="card my-2 border-secondary" >
+      					<div class="card-body py-2">
+      						<div class="row">
+                    <div class="col-md-3 text-center">
+      								<button class="btn btn-primary" onclick="blocks.value=0;pixelate_all();">
+      										<i class="fas fa-eye-slash"></i>
+      										<span class="label-icon">Hide</span>
+      								</button>
+      							</div>
+      							<div class="col-md-6">
+      								<input class="custom-range mt-2" id="blocks" type="range" min="1" max="50" value="5">
+      							</div>
+      							<div class="col-md-3 text-center">
+      								<button class="btn btn-primary" onclick="blocks.value=50;pixelate_all();">
+                          <i class="fas fa-plus-square"></i>
+      										<span class="label-icon">Full resolution</span>
+      								</button>
+      							</div>
+      						</div>
+      					</div>
+      				</div>
+
+            </div>
+        </div>
+
+
+        {% with l_dict_domains=l_dict_domains['list_elem'], bootstrap_label=bootstrap_label %}
+          {% include 'domains/card_img_domain.html' %}
+        {% endwith %}
+
+        <br>
+        <br>
+
+        {%if l_dict_domains['list_elem']%}
+          {% with page=l_dict_domains['page'], nb_page_max=l_dict_domains['nb_pages'],  nb_first_elem=l_dict_domains['nb_first_elem'], nb_last_elem=l_dict_domains['nb_last_elem'], nb_all_elem=l_dict_domains['nb_all_elem'] %}
+            {% set target_url=url_for('crawler_splash.domains_search_name') + "?name=" + l_dict_domains['search']%}
+            {%if domains_types %}
+              {% set target_url = target_url + '&domain_types=' + ','.join(domains_types)%}
+            {%endif%}
+            {% include 'pagination.html' %}
+  				{% endwith %}
+        {%endif%}
+
+
+			</div>
+		</div>
+	</div>
+</body>
+
+
+<script>
+  $(document).ready(function(){
+    $('#nav_title_domains_explorer').removeClass("text-muted");
+
+  });
+
+function toggle_sidebar(){
+	if($('#nav_menu').is(':visible')){
+		$('#nav_menu').hide();
+		$('#side_menu').removeClass('border-right')
+		$('#side_menu').removeClass('col-lg-2')
+		$('#core_content').removeClass('col-lg-10')
+	}else{
+		$('#nav_menu').show();
+		$('#side_menu').addClass('border-right')
+		$('#side_menu').addClass('col-lg-2')
+		$('#core_content').addClass('col-lg-10')
+	}
+}
+</script>
+
+
+<script>
+
+// img_url
+// ctx
+// canevas_id
+
+var dict_canevas_blurr_img = {}
+
+function init_canevas_blurr_img(canevas_id, img_url){
+
+  // ctx, turn off image smoothin
+  dict_canevas_blurr_img[canevas_id] = {}
+  var canvas_container = document.getElementById(canevas_id);
+  var ctx = canvas_container.getContext('2d');
+  ctx.webkitImageSmoothingEnabled = false;
+  ctx.imageSmoothingEnabled = false;
+  dict_canevas_blurr_img[canevas_id]["ctx"] = ctx;
+
+  // img
+  dict_canevas_blurr_img[canevas_id]["img"] = new Image();
+  dict_canevas_blurr_img[canevas_id]["img"].onload = function() {pixelate_img(canevas_id);};
+  dict_canevas_blurr_img[canevas_id]["img"].addEventListener("error", function() {img_error(canevas_id);});
+  dict_canevas_blurr_img[canevas_id]["img"].src = img_url;
+}
+
+function pixelate_all(){
+  Object.entries(dict_canevas_blurr_img).forEach(([key, value]) => {
+   pixelate_img(key);
+  });
+}
+
+function pixelate_img(canevas_id) {
+
+  if (typeof canevas_id !== 'undefined') {
+    var canevas_to_blurr = document.getElementById(canevas_id);
+
+    /// use slider value
+    if( blocks.value == 50 ){
+        size = 1;
+    } else {
+        var size = (blocks.value) * 0.01;
+    }
+
+    canevas_to_blurr.width = dict_canevas_blurr_img[canevas_id]["img"].width;
+    canevas_to_blurr.height = dict_canevas_blurr_img[canevas_id]["img"].height;
+
+    /// cache scaled width and height
+    w = canevas_to_blurr.width * size;
+    h = canevas_to_blurr.height * size;
+
+    /// draw original image to the scaled size
+    dict_canevas_blurr_img[canevas_id]["ctx"].drawImage(dict_canevas_blurr_img[canevas_id]["img"], 0, 0, w, h);
+
+    /// pixelated
+    dict_canevas_blurr_img[canevas_id]["ctx"].drawImage(canevas_to_blurr, 0, 0, w, h, 0, 0, canevas_to_blurr.width, canevas_to_blurr.height);
+  }
+
+}
+
+function img_error(canevas_id) {
+  dict_canevas_blurr_img[canevas_id]["img"].onerror=null;
+  dict_canevas_blurr_img[canevas_id]["img"].src="{{ url_for('static', filename='image/AIL.png') }}";
+}
+
+blocks.addEventListener('change', pixelate_all, false);
+
+{% for dict_domain in l_dict_domains['list_elem'] %}
+  {% if 'screenshot' in dict_domain %}
+    {% if dict_domain['is_tags_safe'] %}
+      var screenshot_url = "{{ url_for('showsavedpastes.screenshot', filename="") }}{{dict_domain['screenshot']}}";
+    {% else %}
+    var screenshot_url = "{{ url_for('static', filename='image/AIL.png') }}";
+    {% endif %}
+    init_canevas_blurr_img("canvas_{{loop.index0}}", screenshot_url);
+  {% endif %}
+{% endfor %}
+
+</script>
+
+</html>