new: Add ASN descriptions

bgpranking/asn_descriptions.py

@@ -0,0 +1,66 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+import logging
+from pathlib import Path
+import requests
+from redis import StrictRedis
+from dateutil.parser import parse
+import re
+
+from .libs.helpers import set_running, unset_running, get_socket_path, safe_create_dir
+
+
+class ASNDescriptions():
+
+    def __init__(self, storage_directory: Path, loglevel: int=logging.DEBUG):
+        self.__init_logger(loglevel)
+        self.asn_meta = StrictRedis(unix_socket_path=get_socket_path('storage'), db=2, decode_responses=True)
+        self.logger.debug('Starting ASN History')
+        self.directory = storage_directory / 'ans_descriptions'
+        safe_create_dir(self.directory)
+        self.archives = self.directory / 'archive'
+        safe_create_dir(self.archives)
+        self.url = 'http://www.cidr-report.org/as2.0/autnums.html'
+
+    def __init_logger(self, loglevel):
+        self.logger = logging.getLogger(f'{self.__class__.__name__}')
+        self.logger.setLevel(loglevel)
+
+    def __update_available(self):
+        r = requests.head(self.url)
+        current_last_modified = parse(r.headers['Last-Modified'])
+        if not self.asn_meta.exists('ans_description_last_update'):
+            return True
+        last_update = parse(self.asn_meta.get('ans_description_last_update'))
+        if last_update < current_last_modified:
+            return True
+        return False
+
+    def load_descriptions(self):
+        if not self.__update_available():
+            self.logger.debug('No new file to import.')
+            return
+        set_running(self.__class__.__name__)
+        self.logger.info('Importing new ASN descriptions.')
+        r = requests.get(self.url)
+        last_modified = parse(r.headers['Last-Modified']).isoformat()
+        p = self.asn_meta.pipeline()
+        new_asn = 0
+        new_description = 0
+        for asn, descr in re.findall('as=AS(.*)&.*</a> (.*)\n', r.text):
+            existing_descriptions = self.asn_meta.hgetall(f'{asn}|descriptions')
+            if not existing_descriptions:
+                self.logger.debug(f'New ASN: {asn} - {descr}')
+                p.hset(f'{asn}|descriptions', last_modified, descr)
+                new_asn += 1
+            else:
+                last_descr = sorted(existing_descriptions.keys(), reverse=True)[0]
+                if descr != existing_descriptions[last_descr]:
+                    self.logger.debug(f'New description for {asn}: {existing_descriptions[last_descr]} -> {descr}')
+                    p.hset(f'{asn}|descriptions', last_modified, descr)
+                    new_description += 1
+        p.set('ans_description_last_update', last_modified)
+        p.execute()
+        self.logger.info(f'Done with import. New ASNs: {new_asn}, new descriptions: {new_description}')
+        unset_running(self.__class__.__name__)

bgpranking/parser.py

@@ -57,19 +57,23 @@ class RawFilesParser():
 
     def parse_raw_files(self):
         set_running(f'{self.__class__.__name__}-{self.source}')
-        for filepath in self.files_to_parse:
-            self.logger.debug('Parsing {}, {} to go.'.format(filepath, len(self.files_to_parse) - 1))
-            with open(filepath, 'rb') as f:
-                to_parse = BytesIO(f.read())
-            p = self.redis_intake.pipeline()
-            for ip in self.parse_raw_file(to_parse):
-                uuid = uuid4()
-                p.hmset(uuid, {'ip': ip, 'source': self.source,
-                               'datetime': self.datetime.isoformat()})
-                p.sadd('intake', uuid)
-            p.execute()
-            self._archive(filepath)
-        unset_running(f'{self.__class__.__name__}-{self.source}')
+        try:
+            for filepath in self.files_to_parse:
+                self.logger.debug('Parsing {}, {} to go.'.format(filepath, len(self.files_to_parse) - 1))
+                with open(filepath, 'rb') as f:
+                    to_parse = BytesIO(f.read())
+                p = self.redis_intake.pipeline()
+                for ip in self.parse_raw_file(to_parse):
+                    uuid = uuid4()
+                    p.hmset(uuid, {'ip': ip, 'source': self.source,
+                                   'datetime': self.datetime.isoformat()})
+                    p.sadd('intake', uuid)
+                p.execute()
+                self._archive(filepath)
+        except Exception as e:
+            self.logger.exception("That didn't go well")
+        finally:
+            unset_running(f'{self.__class__.__name__}-{self.source}')
 
     def _archive(self, filepath: Path):
         '''After processing, move file to the archive directory'''

bgpranking/querying.py

@@ -69,3 +69,9 @@ class Querying():
         d = self.__normalize_date(date)
         key = f'{d}|sources'
         return self.storage.smembers(key)
+
+    def get_asn_descriptions(self, asn: int, all_descriptions=False):
+        descriptions = self.asn_meta.hgetall(f'{asn}|descriptions')
+        if all_descriptions or not descriptions:
+            return descriptions
+        return descriptions[sorted(descriptions.keys(), reverse=True)[0]]

bin/asn_descriptions.py

@@ -0,0 +1,29 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+import logging
+from pathlib import Path
+
+from bgpranking.abstractmanager import AbstractManager
+from bgpranking.asn_descriptions import ASNDescriptions
+from bgpranking.libs.helpers import get_homedir
+
+logging.basicConfig(format='%(asctime)s %(name)s %(levelname)s:%(message)s',
+                    level=logging.INFO, datefmt='%I:%M:%S')
+
+
+class ASNDescriptionsManager(AbstractManager):
+
+    def __init__(self, storage_directory: Path=None, loglevel: int=logging.DEBUG):
+        super().__init__(loglevel)
+        if not storage_directory:
+            storage_directory = get_homedir() / 'rawdata'
+        self.asn_descr = ASNDescriptions(storage_directory, loglevel)
+
+    def _to_run_forever(self):
+        self.asn_descr.load_descriptions()
+
+
+if __name__ == '__main__':
+    asnd_manager = ASNDescriptionsManager()
+    asnd_manager.run(sleep_in_sec=3600)

setup.py

@@ -14,7 +14,7 @@ setup(
     packages=['bgpranking'],
     scripts=['bin/archiver.py', 'bin/dbinsert.py', 'bin/fetcher.py', 'bin/parser.py',
              'bin/loadprefixes.py', 'bin/rislookup.py', 'bin/sanitizer.py', 'bin/run_backend.py',
-             'bin/monitor.py', 'bin/ranking.py', 'bin/start.py', 'bin/stop.py', 'bin/shutdown.py'],
+             'bin/monitor.py', 'bin/ranking.py', 'bin/asn_descriptions.py', 'bin/start.py', 'bin/stop.py', 'bin/shutdown.py'],
     classifiers=[
         'License :: OSI Approved :: GNU Affero General Public License v3 or later (AGPLv3+)',
         'Development Status :: 3 - Alpha',

website/web/__init__.py

@@ -49,7 +49,9 @@ def index():
     sources = q.get_sources(date=session['date'])
     session.pop('asn', None)
     ranks = q.asns_global_ranking(limit=-1, **session)
-    return render_template('index.html', ranks=ranks, sources=sources, **session)
+    descriptions = [q.get_asn_descriptions(int(asn)) for asn, rank in ranks]
+    r = zip(ranks, descriptions)
+    return render_template('index.html', ranks=r, sources=sources, **session)
 
 
 @app.route('/asn', methods=['GET', 'POST'])

website/web/templates/index.html

@@ -39,11 +39,13 @@
       <tr>
           <th>ASN</th>
           <th>Rank</th>
+          <th>Description</th>
       </tr>
-      {% for asn, rank in ranks %}
+      {% for (asn, rank), description in ranks %}
         <tr>
             <td><a href="{{ url_for('asn_details', asn=asn) }}">{{ asn }}</a></td>
             <td>{{ rank }}</td>
+            <td>{{ description }}</td>
         </tr>
       {% endfor %}
   </table>