Malin/krawl.es
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

modified default analyzer values

Browse Source
This commit is contained in:
Leonardo Bambini
2026-01-05 10:01:51 +01:00
parent ff98a77e1a
commit bf02fdb351
3 changed files with 11 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
config.yaml
View File

@@ -38,9 +38,9 @@ behavior:
probability_error_codes: 0 # 0-100 percentage probability_error_codes: 0 # 0-100 percentage
analyzer: analyzer:
http_risky_methods_threshold: 0.1 # http_risky_methods_threshold: 0.1
violated_robots_threshold: 0.1 # violated_robots_threshold: 0.1
uneven_request_timing_threshold: 5 # uneven_request_timing_threshold: 5
uneven_request_timing_time_window_seconds: 300 # uneven_request_timing_time_window_seconds: 300
user_agents_used_threshold: 1 # user_agents_used_threshold: 2
attack_urls_threshold: 1 # attack_urls_threshold: 1

10
src/analyzer.py
View File

@@ -111,9 +111,7 @@ class Analyzer:
delete_accesses_count = len([item for item in accesses if item["method"] == "DELETE"]) delete_accesses_count = len([item for item in accesses if item["method"] == "DELETE"])
head_accesses_count = len([item for item in accesses if item["method"] == "HEAD"]) head_accesses_count = len([item for item in accesses if item["method"] == "HEAD"])
options_accesses_count = len([item for item in accesses if item["method"] == "OPTIONS"]) options_accesses_count = len([item for item in accesses if item["method"] == "OPTIONS"])
patch_accesses_count = len([item for item in accesses if item["method"] == "PATCH"]) patch_accesses_count = len([item for item in accesses if item["method"] == "PATCH"])
#print(f"TOTAL: {total_accesses_count} - GET: {get_accesses_count} - POST: {post_accesses_count}")
if total_accesses_count > http_risky_methods_threshold: if total_accesses_count > http_risky_methods_threshold:
http_method_attacker_score = (post_accesses_count + put_accesses_count + delete_accesses_count + options_accesses_count + patch_accesses_count) / total_accesses_count http_method_attacker_score = (post_accesses_count + put_accesses_count + delete_accesses_count + options_accesses_count + patch_accesses_count) / total_accesses_count
@@ -131,10 +129,6 @@ class Analyzer:
score["good_crawler"]["risky_http_methods"] = False score["good_crawler"]["risky_http_methods"] = False
score["bad_crawler"]["risky_http_methods"] = False score["bad_crawler"]["risky_http_methods"] = False
score["regular_user"]["risky_http_methods"] = False score["regular_user"]["risky_http_methods"] = False
#print(f"Updated score: {score}")
#--------------------- Robots Violations --------------------- #--------------------- Robots Violations ---------------------
#respect robots.txt and login/config pages access frequency #respect robots.txt and login/config pages access frequency
@@ -248,6 +242,8 @@ class Analyzer:
#--------------------- Calculate score --------------------- #--------------------- Calculate score ---------------------
attacker_score = good_crawler_score = bad_crawler_score = regular_user_score = 0
attacker_score = score["attacker"]["risky_http_methods"] * weights["attacker"]["risky_http_methods"] attacker_score = score["attacker"]["risky_http_methods"] * weights["attacker"]["risky_http_methods"]
attacker_score = attacker_score + score["attacker"]["robots_violations"] * weights["attacker"]["robots_violations"] attacker_score = attacker_score + score["attacker"]["robots_violations"] * weights["attacker"]["robots_violations"]
attacker_score = attacker_score + score["attacker"]["uneven_request_timing"] * weights["attacker"]["uneven_request_timing"] attacker_score = attacker_score + score["attacker"]["uneven_request_timing"] * weights["attacker"]["uneven_request_timing"]

4
src/config.py
View File

@@ -103,7 +103,7 @@ class Config:
api = data.get('api', {}) api = data.get('api', {})
database = data.get('database', {}) database = data.get('database', {})
behavior = data.get('behavior', {}) behavior = data.get('behavior', {})
analyzer = data.get('analyzer', {}) analyzer = data.get('analyzer') or {}
# Handle dashboard_secret_path - auto-generate if null/not set # Handle dashboard_secret_path - auto-generate if null/not set
dashboard_path = dashboard.get('secret_path') dashboard_path = dashboard.get('secret_path')
@@ -142,7 +142,7 @@ class Config:
violated_robots_threshold=analyzer.get('violated_robots_threshold', 0.1), violated_robots_threshold=analyzer.get('violated_robots_threshold', 0.1),
uneven_request_timing_threshold=analyzer.get('uneven_request_timing_threshold', 5), uneven_request_timing_threshold=analyzer.get('uneven_request_timing_threshold', 5),
uneven_request_timing_time_window_seconds=analyzer.get('uneven_request_timing_time_window_seconds', 300), uneven_request_timing_time_window_seconds=analyzer.get('uneven_request_timing_time_window_seconds', 300),
user_agents_used_threshold=analyzer.get('user_agents_used_threshold', 1), user_agents_used_threshold=analyzer.get('user_agents_used_threshold', 2),
attack_urls_threshold=analyzer.get('attack_urls_threshold', 1) attack_urls_threshold=analyzer.get('attack_urls_threshold', 1)
) )