geoip 1.4.0 → 1.5.0

checksums.yaml

@@ -1,15 +1,7 @@
 ---
-!binary "U0hBMQ==":
-  metadata.gz: !binary |-
-    YTNjMGMyYjJlNmYzY2JmZTFlZTI2ZDMwYmJiMWZjODVjZjA1ZTI4NQ==
-  data.tar.gz: !binary |-
-    NTIwNDVkMzA2NWI3YjNjODA2ODY5ZDkzZTY3MDI0NDljYjMwNTYxNw==
+SHA1:
+  metadata.gz: 8e46df24a662de22a8500c265c5eaaf2bfb0cb29
+  data.tar.gz: d90310294059804b56d80ac41554fa68826393a2
 SHA512:
-  metadata.gz: !binary |-
-    NThlNDBhZGYwOGFlODY3ODQ1NzBlZDQ1OGI3NzVkYzAzY2ZmOTg5OTEwYTFm
-    NmVkNzY2YmVhOTc1ZTIyNWEyNDlhNjhjYWNhNmZhZTAyYWVhNWIyMWFlYzdm
-    MjAxNjY0NzI4ODkyZmE3MmNiMjIzMWY3YmQwNjc0ZTQxYzYxMTE=
-  data.tar.gz: !binary |-
-    ODBjOWM0ODA1NWUyNDRlYTQ0NmJkY2ViZTI3MGQwNjI5N2JiODRhZTY3YTIx
-    NzdkN2EzZGZkYWZkM2JlZDA0MmFiMTk4MzRmNTU3NTVmMWMxYmYyN2JmNTAx
-    YmUzYWZkMGE4OTE5NmFkOTM1YmVmMjAwMTczODQ2Yzg0ZTJkMzc=
+  metadata.gz: 5ad6a2221b51acf61c658412786484bd5e65ce2b56a4969fdb8954870be6a292360364867584844585c31fb8a04a597c7c30301d9a19d6f0ca125de9f65c00f3
+  data.tar.gz: dba6732195af06951a47b1d36eaadad3d58f5a98fbbd35bae54858fe05a2966ecb1a009c63706873b287b8983d9d5ca38bedce586c0b72ebaeae994c76ea1554

data/bin/geoip

@@ -9,7 +9,38 @@ data = '/usr/share/GeoIP/GeoIP.dat'
 data = ARGV.shift if ARGV[0] =~ /\.dat\Z/
 
 geoip = GeoIP.new data
-req = ([GeoIP::GEOIP_CITY_EDITION_REV1, GeoIP::GEOIP_CITY_EDITION_REV0].include?(geoip.databaseType)) ? :city : :country
+req = case geoip.databaseType
+	when GeoIP::Edition::CITY_REV1,
+	     GeoIP::Edition::CITY_REV0,
+	     GeoIP::Edition::CITY_REV1_V6
+	  :city
+	when GeoIP::Edition::REGION_REV0,
+	     GeoIP::Edition::REGION_REV1
+	  :region
+	when GeoIP::Edition::NETSPEED, GeoIP::Edition::NETSPEED_REV1
+	  :netspeed
+	when GeoIP::Edition::ISP,	# All these looked up the same way
+	     GeoIP::Edition::ORG,
+	     GeoIP::Edition::ISP,
+	     GeoIP::Edition::DOMAIN,
+	     GeoIP::Edition::ACCURACYRADIUS,
+	     GeoIP::Edition::NETSPEED,
+	     GeoIP::Edition::USERTYPE,
+	     GeoIP::Edition::USERTYPE_V6,
+	     GeoIP::Edition::REGISTRAR,
+	     GeoIP::Edition::LOCATIONA,
+	     GeoIP::Edition::CITYCONF,
+	     GeoIP::Edition::COUNTRYCONF,
+	     GeoIP::Edition::REGIONCONF,
+	     GeoIP::Edition::POSTALCONF
+	  :isp
+	when GeoIP::Edition::ASNUM
+	  :asn
+	when GeoIP::Edition::COUNTRY, GeoIP::Edition::PROXY, GeoIP::Edition::COUNTRY_V6
+	  :country
+	else
+	  :asn
+	end
 
 if ARGV.size > 0
   ARGV.each { |a| p geoip.send(req, a) }

data/geoip.gemspec

@@ -2,16 +2,16 @@
 # DO NOT EDIT THIS FILE DIRECTLY
 # Instead, edit Jeweler::Tasks in Rakefile, and run 'rake gemspec'
 # -*- encoding: utf-8 -*-
-# stub: geoip 1.4.0 ruby lib
+# stub: geoip 1.5.0 ruby lib
 
 Gem::Specification.new do |s|
   s.name = "geoip"
-  s.version = "1.4.0"
+  s.version = "1.5.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.require_paths = ["lib"]
   s.authors = ["Clifford Heath", "Roland Moriz"]
-  s.date = "2014-03-26"
+  s.date = "2015-03-10"
   s.description = "GeoIP searches a GeoIP database for a given host or IP address, and\nreturns information about the country where the IP address is allocated,\nand the city, ISP and other information, if you have that database version."
   s.email = ["clifford.heath@gmail.com", "rmoriz@gmail.com"]
   s.executables = ["geoip"]
@@ -36,6 +36,7 @@ Gem::Specification.new do |s|
     "test/csvORG2dat.py",
     "test/organizations.csv",
     "test/organizations.dat",
+    "test/test_file.rb",
     "test/test_geoip.rb",
     "test/test_helper.rb"
   ]

data/lib/geoip.rb

@@ -58,7 +58,7 @@ require 'yaml'
 class GeoIP
 
   # The GeoIP GEM version number
-  VERSION = "1.4.0"
+  VERSION = "1.5.0"
 
   # The +data/+ directory for geoip
   DATA_DIR = File.expand_path(File.join(File.dirname(__FILE__),'..','data','geoip'))
@@ -96,6 +96,55 @@ class GeoIP
   GEOIP_NETSPEED_EDITION = 10
   GEOIP_COUNTRY_EDITION_V6 = 12
   GEOIP_CITY_EDITION_REV1_V6 = 30
+  GEOIP_NETSPEED_EDITION_REV1 = 32
+
+  # Editions list updated from the C API, August 2014:
+  module Edition
+    COUNTRY = 1
+    REGION_REV0 = 7
+    CITY_REV0 = 6
+    ORG = 5
+    ISP = 4
+    CITY_REV1 = 2
+    REGION_REV1 = 3
+    PROXY = 8
+    ASNUM = 9
+    NETSPEED = 10
+    DOMAIN = 11
+    COUNTRY_V6 = 12
+    LOCATIONA = 13
+    ACCURACYRADIUS = 14
+    CITYCONFIDENCE = 15             # unsupported
+    CITYCONFIDENCEDIST = 16         # unsupported
+    LARGE_COUNTRY = 17
+    LARGE_COUNTRY_V6 = 18
+    CITYCONFIDENCEDIST_ISP_ORG = 19 # unused, but gaps are not allowed
+    CCM_COUNTRY = 20                # unused, but gaps are not allowed 
+    ASNUM_V6 = 21
+    ISP_V6 = 22
+    ORG_V6 = 23
+    DOMAIN_V6 = 24
+    LOCATIONA_V6 = 25
+    REGISTRAR = 26
+    REGISTRAR_V6 = 27
+    USERTYPE = 28
+    USERTYPE_V6 = 29
+    CITY_REV1_V6 = 30
+    CITY_REV0_V6 = 31
+    NETSPEED_REV1 = 32
+    NETSPEED_REV1_V6 = 33
+    COUNTRYCONF = 34
+    CITYCONF = 35
+    REGIONCONF = 36
+    POSTALCONF = 37
+    ACCURACYRADIUS_V6 = 38
+  end
+
+  # Numeric codes for NETSPEED (NETSPEED_REV1* is string-based):
+  GEOIP_UNKNOWN_SPEED = 0
+  GEOIP_DIALUP_SPEED = 1
+  GEOIP_CABLEDSL_SPEED = 2
+  GEOIP_CORPORATE_SPEED = 3
 
   COUNTRY_BEGIN = 16776960          #:nodoc:
   STATE_BEGIN_REV0 = 16700000       #:nodoc:
@@ -181,7 +230,7 @@ class GeoIP
     @use_pread = IO.respond_to?(:pread) && !options[:preload]
 
     @options = options
-    @database_type = GEOIP_COUNTRY_EDITION
+    @database_type = Edition::COUNTRY
     @record_length = STANDARD_RECORD_LENGTH
     @file = File.open(filename, 'rb')
 
@@ -210,43 +259,49 @@ class GeoIP
   # * The two-character continent code
   #
   def country(hostname)
-    if (@database_type == GEOIP_CITY_EDITION_REV0 ||
-        @database_type == GEOIP_CITY_EDITION_REV1 ||
-        @database_type == GEOIP_CITY_EDITION_REV1_V6)
-      return city(hostname)
-    end
+    case @database_type
+    when Edition::CITY_REV0, Edition::CITY_REV1, Edition::CITY_REV1_V6
+      city(hostname)
 
-    if (@database_type == GEOIP_REGION_EDITION_REV0 ||
-        @database_type == GEOIP_REGION_EDITION_REV1)
-      return region(hostname)
-    end
+    when Edition::REGION_REV0, Edition::REGION_REV1
+      region(hostname)
 
-    ip = lookup_ip(hostname)
-    if (@database_type == GEOIP_COUNTRY_EDITION ||
-        @database_type == GEOIP_PROXY_EDITION ||
-        @database_type == GEOIP_NETSPEED_EDITION)
-        # Convert numeric IP address to an integer
-        ipnum = iptonum(ip)
-        code = (seek_record(ipnum) - COUNTRY_BEGIN)
-    elsif @database_type == GEOIP_COUNTRY_EDITION_V6
-      ipaddr = IPAddr.new ip
-      code = (seek_record_v6(ipaddr.to_i) - COUNTRY_BEGIN)
+    when Edition::NETSPEED, Edition::NETSPEED_REV1
+      netspeed(hostname)
+
+    when Edition::COUNTRY, Edition::PROXY, Edition::COUNTRY_V6
+      ip = lookup_ip(hostname)
+      if @ip_bits > 32
+	ipaddr = IPAddr.new ip
+	code = (seek_record(ipaddr.to_i) - COUNTRY_BEGIN)
+      else
+	# Convert numeric IP address to an integer
+	ipnum = iptonum(ip)
+	code = (seek_record(ipnum) - @database_segments[0])
+      end
+      read_country(code, hostname, ip)
     else
-      throw "Invalid GeoIP database type, can't look up Country by IP"
+      throw "Invalid GeoIP database type #{@database_type}, can't look up Country by IP"
     end
+  end
 
-    Country.new(
-      hostname,                   # Requested hostname
-      ip,                         # Ip address as dotted quad
-      code,                       # GeoIP's country code
-      CountryCode[code],          # ISO3166-1 alpha-2 code
-      CountryCode3[code],         # ISO3166-2 alpha-3 code
-      CountryName[code],          # Country name, per ISO 3166
-      CountryContinent[code]      # Continent code.
-    )
+  # Search a GeoIP Connection Type (Netspeed) database for the specified host,
+  # returning the speed code.
+  #
+  # +hostname+ is a String holding the host's DNS name or numeric IP address.
+  def netspeed(hostname)
+    unless (@database_type == Edition::NETSPEED ||
+        @database_type == Edition::NETSPEED_REV1)
+      throw "Invalid GeoIP database type #{@database_type}, can't look up Netspeed by IP"
+    end
+    # Convert numeric IP address to an integer
+    ip = lookup_ip(hostname)
+    ipnum = iptonum(ip)
+    offset = (seek_record(ipnum) - @database_segments[0])
+    read_netspeed(offset)
   end
 
-  # Search the GeoIP database for the specified host, retuning region info.
+  # Search the GeoIP database for the specified host, returning region info.
   #
   # +hostname+ is a String holding the hosts's DNS name or numeric IP
   # address.
@@ -262,14 +317,14 @@ class GeoIP
   # * The timezone name, if known
   #
   def region(hostname)
-    if (@database_type == GEOIP_CITY_EDITION_REV0 ||
-        @database_type == GEOIP_CITY_EDITION_REV1 ||
-        @database_type == GEOIP_CITY_EDITION_REV1_V6)
+    if (@database_type == Edition::CITY_REV0 ||
+        @database_type == Edition::CITY_REV1 ||
+        @database_type == Edition::CITY_REV1_V6)
       return city(hostname)
     end
 
-    if (@database_type == GEOIP_REGION_EDITION_REV0 ||
-        @database_type == GEOIP_REGION_EDITION_REV1)
+    if (@database_type == Edition::REGION_REV0 ||
+        @database_type == Edition::REGION_REV1)
       ip = lookup_ip(hostname)
       ipnum = iptonum(ip)
       pos = seek_record(ipnum)
@@ -306,14 +361,14 @@ class GeoIP
   def city(hostname)
     ip = lookup_ip(hostname)
 
-    if (@database_type == GEOIP_CITY_EDITION_REV0 ||
-        @database_type == GEOIP_CITY_EDITION_REV1)
+    if (@database_type == Edition::CITY_REV0 ||
+        @database_type == Edition::CITY_REV1)
       # Convert numeric IP address to an integer
       ipnum = iptonum(ip)
       pos = seek_record(ipnum)
-    elsif (@database_type == GEOIP_CITY_EDITION_REV1_V6)
+    elsif (@database_type == Edition::CITY_REV1_V6)
       ipaddr = IPAddr.new ip
-      pos = seek_record_v6(ipaddr.to_i)
+      pos = seek_record(ipaddr.to_i)
     else
       throw "Invalid GeoIP database type, can't look up City by IP"
     end
@@ -328,7 +383,7 @@ class GeoIP
     # you can test whatever IP range you think is causing problems,
     # as I don't care to undertake an exhaustive search of the 32-bit space.
     unless pos == @database_segments[0]
-      read_city(pos, hostname, ip)
+      read_city(pos-@database_segments[0], hostname, ip)
     end
   end
 
@@ -347,24 +402,34 @@ class GeoIP
     # Convert numeric IP address to an integer
     ipnum = iptonum(ip)
 
-    if (@database_type != GEOIP_ISP_EDITION &&
-        @database_type != GEOIP_ORG_EDITION)
+    case @database_type
+    when Edition::ORG,
+	 Edition::ISP,
+	 Edition::DOMAIN,
+	 Edition::ASNUM,
+	 Edition::ACCURACYRADIUS,
+	 Edition::NETSPEED,
+	 Edition::USERTYPE,
+	 Edition::REGISTRAR,
+	 Edition::LOCATIONA,
+	 Edition::CITYCONF,
+	 Edition::COUNTRYCONF,
+	 Edition::REGIONCONF,
+	 Edition::POSTALCONF
+      pos = seek_record(ipnum)
+      read_isp(pos-@database_segments[0])
+    else
       throw "Invalid GeoIP database type, can't look up Organization/ISP by IP"
     end
-
-    pos = seek_record(ipnum)
-    off = pos + (2*@record_length - 1) * @database_segments[0]
-
-    record = atomic_read(MAX_ORG_RECORD_LENGTH, off)
-    record = record.sub(/\000.*/n, '')
-    record.start_with?('*') ? nil : ISP.new(record)
   end
 
   # Search a ASN GeoIP database for the specified host, returning the AS
   # number and description.
   #
-  # +hostname+ is a String holding the host's DNS name or numeric
-  # IP address.
+  # Many other types of GeoIP database (e.g. userType) mis-identify as ASN type,
+  # and this can read those too.
+  #
+  # +hostname+ is a String holding the host's DNS name or numeric IP address.
   #
   # Returns the AS number and description.
   #
@@ -377,18 +442,12 @@ class GeoIP
     # Convert numeric IP address to an integer
     ipnum = iptonum(ip)
 
-    if (@database_type != GEOIP_ASNUM_EDITION)
-      throw "Invalid GeoIP database type, can't look up ASN by IP"
+    if ![Edition::ASNUM, Edition::ASNUM_V6].include? @database_type
+      throw "Invalid GeoIP database type #{@database_type}, can't look up ASN by IP"
     end
 
     pos = seek_record(ipnum)
-    off = pos + (2*@record_length - 1) * @database_segments[0]
-
-    record = atomic_read(MAX_ASN_RECORD_LENGTH, off)
-    record = record.sub(/\000.*/n, '')
-
-    # AS####, Description
-    ASN.new($1, $2) if record =~ /^(AS\d+)\s(.*)$/
+    read_asn(pos-@database_segments[0])
   end
 
   # Search a ISP GeoIP database for the specified host, returning the
@@ -401,12 +460,12 @@ class GeoIP
   #
   alias_method(:organization, :isp)  # Untested, according to Maxmind docs this should work
 
-  # Iterate through a GeoIP city database
+  # Iterate through a GeoIP city database by 
   def each
     return enum_for unless block_given?
 
-    if (@database_type != GEOIP_CITY_EDITION_REV0 &&
-        @database_type != GEOIP_CITY_EDITION_REV1)
+    if (@database_type != Edition::CITY_REV0 &&
+        @database_type != Edition::CITY_REV1)
       throw "Invalid GeoIP database type, can't iterate thru non-City database"
     end
 
@@ -422,8 +481,65 @@ class GeoIP
     return self
   end
 
+  # Call like this, for example:
+  # GeoIP.new('GeoIPNetSpeedCell.dat').each{|*a| puts("0x%08X\t%d" % a)}
+  # or:
+  # GeoIP.new('GeoIPv6.dat').each{|*a| puts("0x%032X\t%d" % a)}
+  def each_by_ip offset = 0, ipnum = 0, mask = nil, &callback
+    mask ||= 1 << (@ip_bits-1)
+
+    # Read the two pointers and split them:
+    record2 = atomic_read(@record_length*2, @record_length*2*offset)
+    record1 = record2.slice!(0, @record_length)
+
+    # Traverse the left tree
+    off1 = le_to_ui(record1.unpack('C*'))
+    val = off1 - @database_segments[0]
+    if val >= 0
+      yield(ipnum, read_record(ipnum.to_s, ipnum, val)) if val > 0
+    elsif mask != 0
+      each_by_ip(off1, ipnum, mask >> 1, &callback)
+    end
+
+    # Traverse the right tree
+    off2 = le_to_ui(record2.unpack('C*'))
+    val = off2 - @database_segments[0]
+    if val >= 0
+      yield(ipnum|mask, read_record(ipnum.to_s, ipnum, val)) if val > 0
+    elsif mask != 0
+      each_by_ip(off2, ipnum|mask, mask >> 1, &callback)
+    end
+  end
+
   private
 
+  def read_record hostname, ip, offset
+    case @database_type
+    when Edition::CITY_REV0, Edition::CITY_REV1, Edition::CITY_REV1_V6
+      read_city(offset, hostname, ip)
+
+    when Edition::REGION_REV0, Edition::REGION_REV1
+      read_region(offset, hostname, ip)
+
+    when Edition::NETSPEED, Edition::NETSPEED_REV1
+      read_netspeed(offset)
+
+    when Edition::COUNTRY, Edition::PROXY, Edition::COUNTRY_V6
+      read_country(offset, hostname, ip)
+
+    when Edition::ASNUM, Edition::ASNUM_V6
+      read_asn(offset)
+
+    # Add new types here
+    when Edition::ISP, Edition::ORG
+      read_isp offset
+
+    else
+      #raise "Unsupported GeoIP database type #{@database_type}"
+      offset
+    end
+  end
+
   # Loads data into a StringIO which is Copy-on-write friendly
   def preload_data
     @file.seek(0)
@@ -434,57 +550,130 @@ class GeoIP
   # Detects the type of the database.
   def detect_database_type! # :nodoc:
     @file.seek(-3, IO::SEEK_END)
+    @ip_bits = 32
 
     0.upto(STRUCTURE_INFO_MAX_SIZE - 1) do |i|
       if @file.read(3).bytes.all? { |byte| byte == 255 }
-        @database_type = if @file.respond_to?(:getbyte)
-                           @file.getbyte
-                         else
-                           @file.getc
-                         end
+        @database_type =
+	  if @file.respond_to?(:getbyte)
+	    @file.getbyte
+	  else
+	    @file.getc
+	  end
 
         @database_type -= 105 if @database_type >= 106
 
-        if (@database_type == GEOIP_REGION_EDITION_REV0)
+        if (@database_type == Edition::REGION_REV0)
           # Region Edition, pre June 2003
           @database_segments = [STATE_BEGIN_REV0]
-        elsif (@database_type == GEOIP_REGION_EDITION_REV1)
+        elsif (@database_type == Edition::REGION_REV1)
           # Region Edition, post June 2003
           @database_segments = [STATE_BEGIN_REV1]
-        elsif (@database_type == GEOIP_CITY_EDITION_REV0 ||
-               @database_type == GEOIP_CITY_EDITION_REV1 ||
-               @database_type == GEOIP_CITY_EDITION_REV1_V6 ||
-               @database_type == GEOIP_ORG_EDITION ||
-               @database_type == GEOIP_ISP_EDITION ||
-               @database_type == GEOIP_ASNUM_EDITION)
+        elsif @database_type == Edition::CITY_REV0 ||
+               @database_type == Edition::CITY_REV1 ||
+               @database_type == Edition::ORG ||
+               @database_type == Edition::ORG_V6 ||
+               @database_type == Edition::ISP ||
+               @database_type == Edition::ISP_V6 ||
+	       @database_type == Edition::REGISTRAR ||
+	       @database_type == Edition::REGISTRAR_V6 ||
+	       @database_type == Edition::USERTYPE ||	    # Many of these files mis-identify as ASNUM files
+	       @database_type == Edition::USERTYPE_V6 ||
+               @database_type == Edition::DOMAIN ||
+               @database_type == Edition::DOMAIN_V6 ||
+               @database_type == Edition::ASNUM ||
+               @database_type == Edition::ASNUM_V6 ||
+	       @database_type == Edition::NETSPEED_REV1 ||
+	       @database_type == Edition::NETSPEED_REV1_V6 ||
+	       @database_type == Edition::LOCATIONA ||
+	       # @database_type == Edition::LOCATIONA_V6 ||
+	       @database_type == Edition::ACCURACYRADIUS ||
+	       @database_type == Edition::ACCURACYRADIUS_V6 ||
+               @database_type == Edition::CITYCONF ||
+               @database_type == Edition::COUNTRYCONF ||
+               @database_type == Edition::REGIONCONF ||
+               @database_type == Edition::POSTALCONF ||
+               @database_type == Edition::CITY_REV0_V6 ||
+               @database_type == Edition::CITY_REV1_V6
 
           # City/Org Editions have two segments, read offset of second segment
           @database_segments = [0]
           sr = @file.read(3).unpack("C*")
           @database_segments[0] += le_to_ui(sr)
 
-          if (@database_type == GEOIP_ORG_EDITION ||
-              @database_type == GEOIP_ISP_EDITION)
-            @record_length = 4
-          end
         end
 
+	case @database_type
+	when Edition::NETSPEED_REV1
+	when Edition::ASNUM
+	when Edition::CITY_REV0
+	when Edition::CITY_REV1
+	when Edition::REGION_REV0
+	when Edition::REGION_REV1
+	  @ip_bits = 32
+	  @record_length = 3
+
+	when Edition::ORG,
+	    Edition::DOMAIN,
+	    Edition::ISP
+	  @ip_bits = 32
+	  @record_length = 4
+
+	when Edition::ASNUM_V6,
+	    Edition::CITY_REV0_V6,
+	    Edition::CITY_REV1_V6,
+	    Edition::NETSPEED_REV1_V6,
+	    Edition::COUNTRY,
+	    Edition::COUNTRY_V6,
+	    Edition::PROXY
+	  @ip_bits = 128
+	  @record_length = 3
+
+	when Edition::ACCURACYRADIUS_V6,
+	    Edition::DOMAIN_V6,
+	    Edition::ISP_V6,
+	    Edition::LARGE_COUNTRY_V6,
+	    Edition::LOCATIONA_V6,
+	    Edition::ORG_V6,
+	    Edition::REGISTRAR_V6,
+	    Edition::USERTYPE_V6
+	  @ip_bits = 128
+	  @record_length = 4
+
+	else
+	  raise "unimplemented database type"
+	end
+
         break
       else
         @file.seek(-4, IO::SEEK_CUR)
       end
     end
 
-    if (@database_type == GEOIP_COUNTRY_EDITION ||
-        @database_type == GEOIP_PROXY_EDITION ||
-        @database_type == GEOIP_COUNTRY_EDITION_V6 ||
-        @database_type == GEOIP_NETSPEED_EDITION)
+    if (@database_type == Edition::COUNTRY ||
+        @database_type == Edition::PROXY ||
+        @database_type == Edition::COUNTRY_V6 ||
+        @database_type == Edition::NETSPEED)
       @database_segments = [COUNTRY_BEGIN]
     end
+
+    # puts "Detected IPv#{@ip_bits == 32 ? '4' : '6'} database_type #{@database_type} with #{@database_segments[0]} records of length #{@record_length} (data starts at #{@database_segments[0]*@record_length*2})"
+  end
+
+  def read_country code, hostname, ip
+    Country.new(
+      hostname,                   # Requested hostname
+      ip,                         # Ip address as dotted quad
+      code,                       # GeoIP's country code
+      CountryCode[code],          # ISO3166-1 alpha-2 code
+      CountryCode3[code],         # ISO3166-2 alpha-3 code
+      CountryName[code],          # Country name, per ISO 3166
+      CountryContinent[code]      # Continent code.
+    )
   end
 
   def read_region(pos, hostname = '', ip = '') #:nodoc:
-    if (@database_type == GEOIP_REGION_EDITION_REV0)
+    if (@database_type == Edition::REGION_REV0)
       pos -= STATE_BEGIN_REV0
       if (pos >= 1000)
         code = 225
@@ -493,7 +682,7 @@ class GeoIP
         code = pos
         region_code = ''
       end
-    elsif (@database_type == GEOIP_REGION_EDITION_REV1)
+    elsif (@database_type == Edition::REGION_REV1)
       pos -= STATE_BEGIN_REV1
       if (pos < US_OFFSET)
         code = 0
@@ -523,6 +712,40 @@ class GeoIP
     )
   end
 
+  def read_asn off
+    record = atomic_read(MAX_ASN_RECORD_LENGTH, off+index_size)
+    record.slice!(record.index("\0")..-1)
+
+    # AS####, Description
+    # REVISIT: Text is in Latin-1 (ISO8859-1)
+    if record =~ /^(AS\d+)(?:\s(.*))?$/
+      ASN.new($1, $2)
+    else
+      record
+    end
+  end
+
+  def read_netspeed(offset)
+    return offset if @database_type == Edition::NETSPEED  # Numeric value
+
+    record = atomic_read(20, index_size+offset)
+    record.slice!(record.index("\0")..-1)
+    record
+  end
+
+  def read_isp pos
+    off = pos + index_size
+
+    record = atomic_read(MAX_ORG_RECORD_LENGTH, off)
+    record = record.sub(/\000.*/n, '')
+    record.start_with?('*') ? nil : ISP.new(record)
+  end
+
+  # Size of the database index (a binary tree of depth <= @ip_bits)
+  def index_size
+    2 * @record_length * @database_segments[0]
+  end
+
   def lookup_region_name(country_iso2, region_code)
     country_regions = RegionName[country_iso2]
     country_regions && country_regions[region_code]
@@ -545,9 +768,7 @@ class GeoIP
   # * The timezone name, if known
   #
   def read_city(pos, hostname = '', ip = '') #:nodoc:
-    off = pos + (2*@record_length - 1) * @database_segments[0]
-    record = atomic_read(FULL_RECORD_LENGTH, off)
-
+    record = atomic_read(FULL_RECORD_LENGTH, pos+index_size)
     return unless (record && record.size == FULL_RECORD_LENGTH)
 
     # The country code is the first byte:
@@ -565,7 +786,7 @@ class GeoIP
     city = spl[1]
     @iter_pos += (city.size + 1) unless @iter_pos.nil?
     # set the correct encoding in ruby 1.9 compatible environments:
-    city.force_encoding('iso-8859-1') if city.respond_to?(:force_encoding)
+    city = city.force_encoding('iso-8859-1').encode('utf-8') if city.respond_to?(:force_encoding)
 
     # Get the postal code:
     postal_code = spl[2]
@@ -595,7 +816,7 @@ class GeoIP
     # UNTESTED
     if (record &&
         record[0,3] &&
-        @database_type == GEOIP_CITY_EDITION_REV1 &&
+        @database_type == Edition::CITY_REV1 &&
         CountryCode[code] == "US")
 
       dmaarea_combo = le_to_ui(record[0,3].unpack('C*'))
@@ -661,46 +882,18 @@ class GeoIP
     # Binary search in the file.
     # Records are pairs of little-endian integers, each of @record_length.
     offset = 0
-    mask = 0x80000000
-
-    31.downto(0) do |depth|
-      off = (@record_length * 2 * offset)
-      buf = atomic_read(@record_length * 2, off)
-
-      buf.slice!(0...@record_length) if ((ipnum & mask) != 0)
-      offset = le_to_ui(buf[0...@record_length].unpack("C*"))
+    mask = 1 << (@ip_bits-1)
 
-      if (offset >= @database_segments[0])
-        return offset
-      end
+    @ip_bits.downto(1) do |depth|
+      go_right = (ipnum & mask) != 0
+      off = @record_length * (2 * offset + (go_right ? 1 : 0))
+      offset = le_to_ui(r2 = atomic_read(@record_length, off).unpack('C*'))
 
+      return offset if offset >= @database_segments[0]
       mask >>= 1
     end
   end
 
-  def seek_record_v6(ipnum)
-
-    # Binary search in the file.
-    # Records are pairs of little-endian integers, each of @record_length.
-    offset = 0
-    mask = 1 << 127
-
-    127.downto(0) do |depth|
-      off = (@record_length * 2 * offset)
-      buf = atomic_read(@record_length * 2, off)
-
-      buf.slice!(0...@record_length) if ((ipnum & mask) != 0)
-      offset = le_to_ui(buf[0...@record_length].unpack("C*"))
-
-      if (offset >= @database_segments[0])
-        return offset
-      end
-
-      mask >>= 1
-    end
-
-  end
-
   # Convert a big-endian array of numeric bytes to unsigned int.
   #
   # Returns the unsigned Integer.

data/test/test_file.rb

@@ -0,0 +1,17 @@
+#
+# This program walks the specified file and dumps it in IP order
+#
+require 'geoip'
+
+ARGV.each do |file|
+  g = GeoIP.new(file)
+  g.each_by_ip do |ip, val|
+    ip_str =
+      if ip >= (1<<32)
+	(('%032X'%ip).scan(/..../)*':').sub(/\A(0000:)+/, '::')	  # An IPv6 address
+      else
+	'%d.%d.%d.%d' % [ip].pack('N').unpack('C4')
+      end
+    puts "#{ip_str}\t#{val.to_hash.to_a.sort.map{|n,v| "#{n}=#{v.inspect}"}*', '}"
+  end
+end

data/test/test_geoip.rb

@@ -1,11 +1,18 @@
-require File.dirname(__FILE__) + '/test_helper.rb'
-
-class TestGeoip < Test::Unit::TestCase
-
-  def setup
-  end
-  
-  def test_truth
-    assert true
-  end
-end
+require File.dirname(__FILE__) + '/test_helper.rb'
+
+class TestGeoip < Minitest::Test
+
+  def setup
+    data = '/usr/share/GeoIP/GeoIP.dat'
+    begin
+      @g = GeoIP.new(data)
+    rescue Errno::ENOENT => e
+      skip e.message
+    end
+  end
+
+  def test_constructor
+    assert_instance_of GeoIP, @g
+  end
+
+end

data/test/test_helper.rb

@@ -1,3 +1,10 @@
 require 'stringio'
-require 'test/unit'
+require 'minitest/autorun'
 require File.dirname(__FILE__) + '/../lib/geoip'
+
+if Minitest.const_defined?('Test')
+  # We're on Minitest 5+. Nothing to do here.
+else
+  # Minitest 4 doesn't have Minitest::Test yet.
+  Minitest::Test = MiniTest::Unit::TestCase
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: geoip
 version: !ruby/object:Gem::Version
-  version: 1.4.0
+  version: 1.5.0
 platform: ruby
 authors:
 - Clifford Heath
@@ -9,13 +9,12 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-03-26 00:00:00.000000000 Z
+date: 2015-03-10 00:00:00.000000000 Z
 dependencies: []
-description: ! 'GeoIP searches a GeoIP database for a given host or IP address, and
-
+description: |-
+  GeoIP searches a GeoIP database for a given host or IP address, and
   returns information about the country where the IP address is allocated,
-
-  and the city, ISP and other information, if you have that database version.'
+  and the city, ISP and other information, if you have that database version.
 email:
 - clifford.heath@gmail.com
 - rmoriz@gmail.com
@@ -42,6 +41,7 @@ files:
 - test/csvORG2dat.py
 - test/organizations.csv
 - test/organizations.dat
+- test/test_file.rb
 - test/test_geoip.rb
 - test/test_helper.rb
 homepage: http://github.com/cjheath/geoip
@@ -54,12 +54,12 @@ require_paths:
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
-  - - ! '>='
+  - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
-  - - ! '>='
+  - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []