{% extends "global/layout.html" %}
{% block title %}{% trans %}GeoIP File Specification{% endtrans %}{% endblock %}
{% block lastupdated %}{% trans %}May 2013{% endtrans %}{% endblock %}
{% block accuratefor %}0.9.6{% endblock %}
{% block content %}

<h2>{% trans %}Overview{% endtrans %}</h2>
<p>{% trans -%}
This page specifies the format of the various GeoIP files,
used by the router to look up a country for an IP.
{%- endtrans %}</p>


<h2>{% trans %}Country Name (countries.txt) Format{% endtrans %}</h2>
<p>{% trans -%}
This format is easily generated from data files available from many public sources.
For example:
{%- endtrans %}</p>
<pre>
# wget http://geolite.maxmind.com/download/geoip/database/GeoIPCountryCSV.zip
# unzip GeoIPCountryCSV.zip
# cut -d, -f5,6 < GeoIPCountryWhois.csv |sed 's/"//g' | sort | uniq > countries.txt
</pre>
<ul>
<li>Encoding is UTF-8
<li>'#' in column 1 specifies a comment line
<li>Entry lines are CountryCode,CountryName
<li>CountryCode is the ISO two-letter code, upper case
<li>CountryName is in English
</ul>


<h2>{% trans %}IPv4 (geoip.txt) Format{% endtrans %}</h2>
<p>{% trans -%}
This format is borrowed from Tor and is easily generated from data files available from many public sources.
For example:
{%- endtrans %}</p>
<pre>
# wget http://geolite.maxmind.com/download/geoip/database/GeoIPCountryCSV.zip
# unzip GeoIPCountryCSV.zip
# cut -d, -f3-5 < GeoIPCountryWhois.csv|sed 's/"//g' > geoip.txt
# cut -d, -f5,6 < GeoIPCountryWhois.csv |sed 's/"//g' | sort | uniq > countries.txt
</pre>
<ul>
<li>Encoding is ASCII
<li>'#' in column 1 specifies a comment line
<li>Entry lines are FromIP,ToIP,CountryCode
<li>FromIP and ToIP are unsigned integer representations of the 4-byte IP
<li>CountryCode is the ISO two-letter code, upper case
<li>Entry lines must be sorted by numeric FromIP
</ul>


<h2>{% trans %}IPv6 (geoipv6.dat.gz) Format{% endtrans %}</h2>
<p>{% trans -%}
This is a compressed binary format designed for I2P.
The file is gzipped. Ungzipped format:
{%- endtrans %}</p>
<pre>
   Bytes 0-9: Magic number "I2PGeoIPv6"
   Bytes 10-11: Version (0x0001)
   Bytes 12-15 Options (0x00000000) (future use)
   Bytes 16-23: Creation date (Java long)
   Bytes 24-xx: Optional comment (UTF-8)
   Bytes xx-255: null padding
   Bytes 256-: 18 byte records:
       8 byte from (/64)
       8 byte to (/64)
       2 byte ISO country code LOWER case (ASCII)
</pre>
<p>{% trans %}NOTES:{% endtrans %}</p>
<ul>
<li>Data must be sorted (SIGNED long twos complement), no overlap.
So the order is 8000:: ... FFFF:: 0000:: ... 7FFF::
<li>
The GeoIPv6.java class contains a program to generate this format from
public sources such as the Maxmind GeoLite data.
<li>
This specification is preliminary; I2P does not yet support IPv6 GeoIP lookup.
</ul>
{% endblock %}