changed import, added referer cleanup

default.conf

@@ -8,29 +8,45 @@ db = /home/my_user/analytics/my_website.db
 # these changes will only apply to newly collected data/creation of new database
 # path to the nginx access log to parse.
 access_log = /home/my_user/analytics/access.log
+
 # nginx locations and their root directory: location:directory,location:directory,...
 locs_and_dirs = /:/www/my_website,/error:/www/error
 # filetypes that should be grouped (comma separated)
 auto_group_filetypes = png,jpg,jpeg,gif,svg,css,ico,pdf,txt
-# wether a request with 30x http status counts as success
-status_300_is_success = False
-# wether a user needs to make at least 1 successful request to be a human
-humans_need_success = True
-# filegroups, eg group index.html and home.html
+# group certain files
 filegroups = home:index.html,home.html;images:image1.png,image2.png
 # filegroups = 
 
+# wether a request with 30x http status counts as success
+status_300_is_success = False
+# if False, unique user is (ip-address - user agent) pair, if True only ip addess
+unique_user_is_ip_address = False
+# wether a user needs to make at least 1 successful request to be a human
+humans_need_success = True
+
+# dont collect requests to locations matched by this
+request_location_regex_blacklist = /analytics.*
+
 # VISUALIZATION
 # separate users into all and humans 
 get_human_percentage = True
 # regex expression as whitelist for file ranking
-# file_ranking_regex_whitelist = .*\.((txt)|(html)|(css)|(php)|(png)|(jpeg)|(jpg)|(svg)|(gif))
-file_ranking_regex_whitelist =
+
+# ignore the protocol in referers, so https://url.com = http://url.com -> url.com
+referer_ranking_ignore_protocol = True
+# ignore the subdomains in referers, so foo.url.com = bar.url.com -> url.com
+referer_ranking_ignore_subdomain = False
+# ignore the location in referers, so url.com/foo = url.com/bar -> url.com
+referer_ranking_ignore_location = True
 # regex expression as whitelist for referer ranking, minus means empty
 # eg: exclude empty referers
 referer_ranking_regex_whitelist = ^[^\-].*
+
 # regex expression as whitelist for user agent ranking
 user_agent_ranking_regex_whitelist = 
+
+# file_ranking_regex_whitelist = .*\.((txt)|(html)|(css)|(php)|(png)|(jpeg)|(jpg)|(svg)|(gif))
+file_ranking_regex_whitelist =
 # maximum number of file(group)s on the file ranking
 file_ranking_plot_max_files = 20
 # wether to ignore non existing files in the ranking
@@ -38,6 +54,7 @@ file_ranking_ignore_error_files = True
 
 # "plot_figsize" = (60 40),
 plot_dpi = 300
+
 # output directory for the generated plots
 img_dir = /www/analytics/images
 # nginx location for the generated images, its root must be img_dir

regina/db_operation/collect.py

@@ -2,10 +2,10 @@ import sqlite3 as sql
 from re import match
 from time import mktime
 from datetime import datetime as dt
-from db_operation.database import t_request, t_user, t_file, t_filegroup, database_tables, get_filegroup
-from utility.sql_util import sanitize, sql_select, sql_exists, sql_insert, sql_tablesize
-from utility.utility import pdebug, warning
-from utility.globals import user_agent_operating_systems, user_agent_browsers, settings
+from regina.db_operation.database import t_request, t_user, t_file, t_filegroup, database_tables, get_filegroup
+from regina.utility.sql_util import sanitize, sql_select, sql_exists, sql_insert, sql_tablesize
+from regina.utility.utility import pdebug, warning
+from regina.utility.globals import user_agent_operating_systems, user_agent_browsers, settings
 
 """
 collect information from the access log and put it into the database
@@ -73,13 +73,23 @@ def parse_log(logfile:str) -> list[Request]:
                                 status=g[4], bytes_sent=g[5], referer=g[6], user_agent=g[7]))
     return requests
 
+
+def user_exists(cursor, request) -> bool:
+    if settings["unique_user_is_ip_address"]:
+        return sql_exists(cursor, t_user, [("ip_address", request.ip_address)])
+    else:
+        return sql_exists(cursor, t_user, [("ip_address", request.ip_address), ("user_agent", request.user_agent)])
+
 def get_user_id(request: Request, cursor: sql.Cursor) -> int:
     """
     get the user_id. Adds the user if not already existing
     """
     # if user exists
-    if sql_exists(cursor, t_user, [("ip_address", request.ip_address), ("user_agent", request.user_agent)]):
-        user_id = sql_select(cursor, t_user, [("ip_address", request.ip_address), ("user_agent", request.user_agent)])[0][0]
+    if user_exists(cursor, request):
+        if settings["unique_user_is_ip_address"]:
+            user_id = sql_select(cursor, t_user, [("ip_address", request.ip_address)])[0][0]
+        else:
+            user_id = sql_select(cursor, t_user, [("ip_address", request.ip_address), ("user_agent", request.user_agent)])[0][0]
     else:  # new user 
         # new user_id is number of elements
         user_id: int = sql_tablesize(cursor, t_user)
@@ -139,8 +149,12 @@ def add_requests_to_db(requests: list[Request], db_name: str):
     cursor = conn.cursor()
     # check the new users later
     max_user_id = sql_tablesize(cursor, t_user)
+    request_blacklist = settings["request_location_regex_blacklist"]
     for i in range(len(requests)):
         request = requests[i]
+        # skip requests to blacklisted locations
+        if request_blacklist:
+            if match(request_blacklist, request.request_file): continue
         # pdebug("add_requests_to_db:", i, "request:", request)
         user_id = get_user_id(request, cursor)
         conn.commit()

regina/db_operation/database.py

@@ -3,8 +3,8 @@
 import sqlite3 as sql
 from os import path, listdir
 # local
-from utility.sql_util import sanitize, sql_select, sql_exists, sql_insert, sql_tablesize
-from utility.utility import pdebug
+from regina.utility.sql_util import sanitize, sql_select, sql_exists, sql_insert, sql_tablesize
+from regina.utility.utility import pdebug
 
 """
 create reginas database as shown in the uml diagram database.uxf

regina/db_operation/visualize.py

@@ -9,17 +9,14 @@ from datetime import datetime as dt
 
 from numpy import empty
 # local
-from db_operation.database import t_request, t_user, t_file, t_filegroup
-from utility.sql_util import sanitize, sql_select, sql_exists, sql_insert, sql_tablesize, sql_get_count_where
-from utility.utility import pdebug, warning, missing_arg
-from utility.globals import settings
+from regina.db_operation.database import t_request, t_user, t_file, t_filegroup
+from regina.utility.sql_util import sanitize, sql_select, sql_exists, sql_insert, sql_tablesize, sql_get_count_where
+from regina.utility.utility import pdebug, warning, missing_arg
+from regina.utility.globals import settings
 
 
 """
 visualize information from the databse
-TODO:
-- bei referrers ähnliche zusammenlegen, z.b. www.google.de und https://google.com
-- ignore 404
 """
 
 palette = {
@@ -141,12 +138,18 @@ def get_where_date_str(at_date=None, min_date=None, max_date=None):
 def get_earliest_date(cur: sql.Cursor) -> int:
     """return the earliest time as unixepoch"""
     cur.execute(f"SELECT MIN(date) FROM {t_request}")
-    return cur.fetchone()[0]
+    date = cur.fetchone()[0]
+    if not isinstance(date, int): return 0
+    else: return date
+
 # get the latest date
 def get_latest_date(cur: sql.Cursor) -> int:
     """return the latest time as unixepoch"""
     cur.execute(f"SELECT MAX(date) FROM {t_request}")
-    return cur.fetchone()[0]
+    date = cur.fetchone()[0]
+    if not isinstance(date, int): return 0
+    else: return date
+
 # get all dates
 # the date:str parameter in all these function must be a sqlite constraint
 def get_days(cur: sql.Cursor, date:str) -> list[str]:
@@ -296,6 +299,48 @@ def get_ranking(field_name: str, table: str, whitelist_regex: str, cur: sql.Curs
     # print(ranking)
     return ranking
 
+re_uri_protocol = f"(https?)://"
+re_uri_ipv4 = r"(?:(?:(?:\d{1,3}\.?){4})(?::\d+)?)"
+# re_uri_ipv6 = ""
+re_uri_domain = r"(?:([^/]+\.)*[^/]+\.[a-zA-Z]{2,})"
+re_uri_location = r"(?:/(.*))?"
+re_uri_full = f"{re_uri_protocol}({re_uri_domain}|{re_uri_ipv4})({re_uri_location})"
+
+def cleanup_referer(referer: str) -> str:
+    """
+    split the referer uri into its parts and reassemeble them depending on settings
+    """
+    m = fullmatch(re_uri_full, referer)
+    if not m:
+        pdebug(f"cleanup_referer: Could not match referer '{referer}'")
+        return referer
+    # pdebug(f"cleanup_referer: {referer} - {m.groups()}")
+    protocol = m.groups()[0]
+    subdomains = m.groups()[2]
+    if not subdomains: subdomains = ""
+    domain = m.groups()[1].replace(subdomains, "")
+    location = m.groups()[3]
+
+    referer = domain
+    if not settings["referer_ranking_ignore_subdomain"]: referer = subdomains + referer
+    if not settings["referer_ranking_ignore_protocol"]: referer = protocol + "://" + referer
+    if not settings["referer_ranking_ignore_location"]: referer += location
+    # pdebug(f"cleanup_referer: cleaned up: {referer}")
+    return referer
+
+def cleanup_referer_ranking(referer_ranking: list[tuple[int, str]]):
+    unique_referers = dict()
+    for count, referer in referer_ranking:
+        referer = cleanup_referer(referer)
+        if referer in unique_referers:
+            unique_referers[referer] += count
+        else:
+            unique_referers[referer] = count
+    referer_ranking.clear()
+    for referer, count in unique_referers.items():
+        referer_ranking.append((count, referer))
+    referer_ranking.sort()
+
 
 #
 # PLOTTING
@@ -463,7 +508,8 @@ def visualize(loaded_settings: dict):
     get_humans = settings["get_human_percentage"]
     # pdebug(f"visualize: settings {settings}")
     # DATE STRINGS
-    names["earliest_date"] = dt.fromtimestamp(get_earliest_date(cur)).strftime("%Y-%m-%d")
+    earliest_date = get_earliest_date(cur)
+    names["earliest_date"] = dt.fromtimestamp(earliest_date).strftime("%Y-%m-%d")
     names["generation_date"] = dt.now().strftime("%Y-%m-%d %H:%M:%S")
     # LAST_X_DAYS
     # last_x_days_min_date: latest_date - last_x_days
@@ -506,6 +552,7 @@ def visualize(loaded_settings: dict):
 
         # REFERER
         referer_ranking = get_ranking("referer", t_request, settings["referer_ranking_regex_whitelist"], cur, date_str)
+        cleanup_referer_ranking(referer_ranking)
         if gen_img:
             fig_referer_ranking = plot_ranking(referer_ranking, xlabel="HTTP Referer", ylabel="Number of requests", color_settings=color_settings_alternate)
             fig_referer_ranking.savefig(f"{img_dir}/{names[f'img_referer_ranking{suffix}']}")

regina/main.py

@@ -3,14 +3,23 @@
 # __package__="."
 from sys import argv, exit
 from os.path import isfile
-from db_operation.collect import parse_log, add_requests_to_db
-from db_operation.database import create_db
-from db_operation.visualize import visualize
-from utility.settings_manager import read_settings_file
-from utility.globals import settings, version
+from regina.db_operation.collect import parse_log, add_requests_to_db
+from regina.db_operation.database import create_db
+from regina.db_operation.visualize import visualize
+from regina.utility.settings_manager import read_settings_file
+from regina.utility.globals import settings, version
 
 """
 start regina, launch either collect or visualize
+TODO:
+- bei referrers ähnliche zusammenlegen, z.b. www.google.de und https://google.com
+- optionen:
+    - unique user = ip address
+    - max requests/time
+
+- wenn datenbankgröße zum problem wird:
+    - referrer table die die schon zusammengelegten referrer enthält, request verlinkt nur mit id
+    - selbes für platforms und browsers
 """
 
 
@@ -71,9 +80,9 @@ def main():
         error(f"Not a file: '{config_file}'")
     read_settings_file(config_file, settings)
     settings["version"] = version
-    if log_file: settings["access-log"] = log_file
+    if log_file: settings["access_log"] = log_file
 
-    print(f"regina version {version} with server-name '{settings['server_name']}' and database '{settings['db']}'")
+    print(f"regina version {version} with server-name '{settings['server_name']}', database '{settings['db']}' and logfile '{settings['access_log']}'")
 
     if not settings["server_name"]: missing_arg("server-name")
     if not settings["access_log"]: missing_arg("log")

regina/utility/globals.py

@@ -12,6 +12,8 @@ settings = {
     "locs_and_dirs": [],
     "auto_group_filetypes": [],
     "filegroups": "",
+    "request_location_regex_blacklist": "",
+    "unique_user_is_ip_address": False,
 
     # VISUALIZATION
     "get_human_percentage": False,
@@ -20,6 +22,9 @@ settings = {
     # "file_ranking_regex_whitelist": r".*\.((txt)|(html)|(css)|(php)|(png)|(jpeg)|(jpg)|(svg)|(gif))",
     "file_ranking_regex_whitelist": r".*\.(html)",
     "file_ranking_ignore_error_files": False,  # skip files that only had unsuccessful requests (status < 300)
+    "referer_ranking_ignore_protocol": True,
+    "referer_ranking_ignore_subdomain": False,
+    "referer_ranking_ignore_location": True,
     "referer_ranking_regex_whitelist": r"^[^\-].*",  # minus means empty
     "user_agent_ranking_regex_whitelist": r"",
     "file_ranking_plot_max_files": 15,

regina/utility/utility.py

@@ -6,7 +6,7 @@ from sys import exit
 Various utitity
 """
 
-DEBUG = False
+DEBUG = True
 def pdebug(*args):
     if DEBUG: print(*args)