collect blocklists and merge them with weights per source

src/brewserverblocklist/brewserverblocklist.py

@@ -1,4 +1,86 @@
+#!env python3
 # SPDX-FileCopyrightText: 2023 Tobias Diekershoff
 #
-# SPDX-License-Identifier: CC0-1.0
 # SPDX-License-Identifier: GPL-3.0-or-later
+"""
+This script can be used to create server block lists by compining the blocklists
+of any number of other Friendica instances.
+"""
+import argparse
+import configparser
+import sys
+from os.path import exists
+import requests
+
+class BrewBlocklist():
+    """
+    This is the cauldron that is used to
+
+    * collect the blocklists from other servers
+    * rank the blocklists by their trustworthyness
+    * compile a the resulting blocklist
+    """
+    def __init__(self, configfile, outputfile):
+        """
+        Initialise the cauldron with the filenames of the config file
+        and the outputfilename.
+        """
+        self.sources = []
+        config = configparser.RawConfigParser()
+        config.read(configfile)
+        for section in config.sections():
+            section_values = dict(config.items(section))
+            self.sources.append({'url': section, 'trust': int(section_values['trust'])})
+        self.outputfile = outputfile
+        self.blocklist = {}
+        self.reasons = {}
+
+    def collect_ingrediens(self):
+        """
+        Use request.get to collect the blocklists from the Friendica nodes provided
+        in the config file. Collect the reasons why to block the servers (the first
+        mention of the server wins) and sum the trust levels of the blocks.
+        """
+        for source in self.sources:
+            requ = requests.get('https://{}/blocklist/domain/download'.format(source['url']))
+            for line in requ.text.split('\n'):
+                try:
+                    pattern, reason = line.split(',')
+                except ValueError:
+                    break
+                self.blocklist[pattern] = self.blocklist.get(pattern, 0) + source['trust']
+                self.reasons[pattern] = self.reasons.get(pattern, reason)
+
+    def serve_meal(self):
+        """
+        Print the CSV list of the collected blocklist into either STDOUT or
+        the output file that was defined as command line parameter.
+        """
+        print(self.outputfile, not self.outputfile)
+        if self.outputfile:
+            out_file = open(self.outputfile, 'w')
+            orig_stdout = sys.stdout
+            sys.stdout = out_file
+        for key, value in self.blocklist.items():
+            print('{}, {}, {}'.format(key, self.reasons[key], value))
+        if self.outputfile:
+            sys.stdout = orig_stdout
+            out_file.close()
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('-c', '--config',
+                        dest='configfile',
+                        required=True,
+                        help='specify the configuration file')
+    parser.add_argument('-o', '--output',
+                        dest='outputfile',
+                        default=None,
+                        help='specify the output file. STDOUT if none given')
+    args = parser.parse_args()
+    if not exists(args.configfile):
+        print('The config file {} was not found.'.format(args.configfile))
+        sys.exit()
+    brew = BrewBlocklist(args.configfile, args.outputfile)
+    brew.collect_ingrediens()
+    brew.serve_meal()