Build GeoIPISP legacy database

[malakudi]

Trying to build a GeoIPISP legacy database. I copied code from the ASN related functions (ASNRadixTree etc) but resulting file is not valid. Field of input csv called isp and parsed correctly but output file is wrong. What am I missing? Can you help?

Here is my patch so far

--- ../100/geolite2legacy/pygeoip_const.py	2024-10-05 10:05:47.870249802 +0300
+++ ../geolite2legacy/pygeoip_const.py	2024-10-05 10:22:07.172732546 +0300
@@ -405,6 +405,7 @@
 CITY_EDITION_REV1_V6 = 30
 ORG_EDITION = 5
 ISP_EDITION = 4
+ISP_EDITION_V6 = 22
 ASNUM_EDITION = 9
 ASNUM_EDITION_V6 = 21
 NETSPEED_EDITION = 10
@@ -416,7 +417,7 @@
 
 
 # Collection of databases
-IPV6_EDITIONS = (COUNTRY_EDITION_V6, ASNUM_EDITION_V6, CITY_EDITION_REV1_V6)
+IPV6_EDITIONS = (COUNTRY_EDITION_V6, ASNUM_EDITION_V6, ISP_EDITION_V6, CITY_EDITION_REV1_V6)
 CITY_EDITIONS = (CITY_EDITION_REV0, CITY_EDITION_REV1, CITY_EDITION_REV1_V6)
 REGION_EDITIONS = (REGION_EDITION_REV0, REGION_EDITION_REV1)
 REGION_CITY_EDITIONS = REGION_EDITIONS + CITY_EDITIONS

--- ../100/geolite2legacy/geolite2legacy.py	2024-10-05 10:05:47.854249430 +0300
+++ ../geolite2legacy/geolite2legacy.py	2024-10-05 11:01:57.014387404 +0300
@@ -241,6 +241,29 @@
     reclen = STANDARD_RECORD_LENGTH
     segreclen = SEGMENT_RECORD_LENGTH
 
+class ISPRadixTree(RadixTree):
+    seek_depth = 31
+    edition = ISP_EDITION
+    reclen = STANDARD_RECORD_LENGTH
+    segreclen = SEGMENT_RECORD_LENGTH
+
+    def gen_nets(self, locations, infile):
+        for row in csv.DictReader(infile):
+            nets = [IPNetwork(row['network'])]
+            isp = decode_text(row['isp'])
+            entry = u'{}'.format(isp)
+            yield nets, (serialize_text(entry),)
+
+    def encode(self, data):
+        return data + b'\0\0\0'
+
+
+class ISPv6RadixTree(ISPRadixTree):
+    seek_depth = 127
+    edition = ISP_EDITION_V6
+    reclen = STANDARD_RECORD_LENGTH
+    segreclen = SEGMENT_RECORD_LENGTH
+
 
 class CityRev1RadixTree(RadixTree):
     seek_depth = 31
@@ -366,13 +389,15 @@
 RTree = {
     'Country': {'IPv4': CountryRadixTree, 'IPv6': Countryv6RadixTree},
     'City': {'IPv4': CityRev1RadixTree, 'IPv6': CityRev1v6RadixTree},
-    'ASN': {'IPv4': ASNRadixTree, 'IPv6': ASNv6RadixTree}
+    'ASN': {'IPv4': ASNRadixTree, 'IPv6': ASNv6RadixTree},
+    'ISP': {'IPv4': ISPRadixTree, 'IPv6': ISPv6RadixTree}
 }
 
 Filenames = {
     'Country': {'IPv4': "GeoIP.dat", 'IPv6': "GeoIPv6.dat"},
     'City': {'IPv4': "GeoIPCity.dat", 'IPv6': "GeoIPCityv6.dat"},
-    'ASN': {'IPv4': "GeoIPASNum.dat", 'IPv6': "GeoIPASNumv6.dat"}
+    'ASN': {'IPv4': "GeoIPASNum.dat", 'IPv6': "GeoIPASNumv6.dat"},
+    'ISP': {'IPv4': "GeoIPISP.dat", 'IPv6': "GeoIPISPv6.dat"},
 }
 
 
@@ -423,8 +448,9 @@
     # noinspection PyUnboundLocalVariable
     datfilecomment = '{} converted to legacy MaxMind DB with geolite2legacy'.format(os.path.dirname(entry.filename))
     dbtype, entries = entries.popitem()
+    print(dbtype)
 
-    if dbtype == 'ASN':
+    if dbtype == 'ISP' or dbtype == 'ASN':
         locs = None
     else:
         if not {'Locations', 'Blocks'} <= set(entries.keys()):

[hege-li]

@malakudi Atleast you need to change STANDARD_RECORD_LENGTH -> ORG_RECORD_LENGTH (edit: sorry meant ORG)

And possibly this (not sure if \0\0\0 -> \0 change was required for isp/org but try it)
def encode(self, data):
return data + b'\0'

[malakudi]

@malakudi Atleast you need to change STANDARD_RECORD_LENGTH -> GEO_RECORD_LENGTH

And possibly this (not sure if \0\0\0 -> \0 change was required for isp/org but try it) def encode(self, data): return data + b'\0'

GEO_RECORD_LENGTH is not defined in GeoIP.h and of course at pygeoip_const.py. Maybe you mean ORG_RECORD_LENGTH?

[malakudi]

It worked with reclen = ORG_RECORD_LENGTH
Thank you very much!
I will create a pull request later

[malakudi]

@malakudi Atleast you need to change STANDARD_RECORD_LENGTH -> ORG_RECORD_LENGTH (edit: sorry meant ORG)

And possibly this (not sure if \0\0\0 -> \0 change was required for isp/org but try it) def encode(self, data): return data + b'\0'

Database seems correct with both \0\0\0 and \0, at least when checked with geoiplookup utility. Same for ASNum db. Why \0\0\0 was needed?

[malakudi]

OK, I reply to myself, when you do not use \0\0\0 , the version string (shown in geoiplookup with -v) is empty.