读取纯真IP数据库代码

参考文档:http://lumaqq.linuxsir.org/article/qqwry_format_detail.html
作者: gounliey
Email:4473117@qq.com
第一次写东西出来大家分享，还有很多不足之处，请大家指点。

'辅助类，用于保存IP索引信息
Public Class CZ_INDEX_INFO

Public IpSet As UInt32
Public IpEnd As UInt32
Public Offset As UInt32

Sub New()
IpSet = 0
IpEnd = 0
Offset = 0
End Sub

End Class

'读取纯真IP数据库类
Public Class PHCZIP

Protected bFilePathInitialized As Boolean
Protected FilePath As String
Protected FileStrm As FileStream

Protected Index_Set As UInt32
Protected Index_End As UInt32
Protected Index_Count As UInt32

Protected Search_Index_Set As UInt32
Protected Search_Index_End As UInt32
Protected Search_Set As CZ_INDEX_INFO
Protected Search_Mid As CZ_INDEX_INFO
Protected Search_End As CZ_INDEX_INFO

Public Sub New()
bFilePathInitialized = False
End Sub

Public Sub New(ByVal dbFilePath As String)
bFilePathInitialized = False
SetDbFilePath(dbFilePath)
End Sub

'使用二分法查找索引区，初始化查找区间
Protected Sub Initialize()
Search_Index_Set = 0
Search_Index_End = Index_Count - 1
End Sub

'关闭文件
Public Sub Dispose()
If (bFilePathInitialized) Then
bFilePathInitialized = False
FileStrm.Close()
FileStrm.Dispose()
End If
End Sub

'设置纯真IP数据库的文件路径
Public Function SetDbFilePath(ByVal dbFilePath As String) As Boolean

If (dbFilePath = "") Then
Return False
End If

'Try
'打开数据文件

FileStrm = New FileStream(dbFilePath, FileMode.Open, FileAccess.Read)
'Catch
'Return False
'End Try

'检查文件长度
If (FileStrm.Length < 8) Then
FileStrm.Close()
FileStrm.Dispose()
Return False
End If

'得到第一条索引的绝对偏移和最后一条索引的绝对偏移
FileStrm.Seek(0, SeekOrigin.Begin)

Index_Set = GetUInt32()
Index_End = GetUInt32()

'得到总索引条数
Index_Count = (Index_End - Index_Set) / 7 + 1
bFilePathInitialized = True

Return True

End Function

'主接口函数，根据传入的IP返回该IP的地址信息
Public Function GetAddressWithIP(ByVal IPvalue As String) As String

If Not bFilePathInitialized Then
Return ""
End If

Initialize()

'将IP转化为数字
Dim ip As UInt32 = IPToUInt32(IPvalue)

While (True)

'首先初始化本轮查找的区间

'区间头
Search_Set = IndexInfoAtPos(Search_Index_Set)
'区间尾
Search_End = IndexInfoAtPos(Search_Index_End)

'判断IP是否在区间头内
If (ip >= Search_Set.IpSet And ip <= Search_Set.IpEnd) Then
Return ReadAddressInfoAtOffset(Search_Set.Offset)
End If

'判断IP是否在区间尾内
If (ip >= Search_End.IpSet And ip <= Search_End.IpEnd) Then
Return ReadAddressInfoAtOffset(Search_End.Offset)
End If

'计算出区间中点
Search_Mid = IndexInfoAtPos((Search_Index_End + Search_Index_Set) / 2)

'判断IP是否在中点
If (ip >= Search_Mid.IpSet And ip <= Search_Mid.IpEnd) Then
Return ReadAddressInfoAtOffset(Search_Mid.Offset)
End If

'本轮没有找到，准备下一轮
If (ip < Search_Mid.IpSet) Then
'IP比区间中点要小，将区间尾设为现在的中点，将区间缩小1倍。
Search_Index_End = (Search_Index_End + Search_Index_Set) / 2
Else
'IP比区间中点要大，将区间头设为现在的中点，将区间缩小1倍。
Search_Index_Set = (Search_Index_End + Search_Index_Set) / 2
End If

End While

Return ""

End Function

'读取指定文件偏移位置的国家和地区信息
Protected Function ReadAddressInfoAtOffset(ByVal Offset As UInt32) As String

Dim country As String = ""
Dim area As String = ""
Dim country_Offset As UInt32 = 0
Dim Tag As Byte = 0
'跳过4字节，因这4个字节是该索引的IP区间上限。
FileStrm.Seek(Offset + 4, SeekOrigin.Begin)

'读取一个字节，得到描述国家信息的“寻址方式”
Tag = GetTag()

If (Tag = &H1) Then

'模式0x01，表示接下来的3个字节是表示偏移位置
FileStrm.Seek(GetOffset(), SeekOrigin.Begin)

'继续检查“寻址方式”
Tag = GetTag()
If (Tag = &H2) Then
'模式0x02，表示接下来的3个字节代表国家信息的偏移位置
'先将这个偏移位置保存起来，因为我们要读取它后面的地区信息。
country_Offset = GetOffset()
'读取地区信息（注：按照Luma的说明，好像没有这么多种可能性，但在测试过程中好像有些情况没有考虑到，
'所以写了个ReadArea()来读取。
area = ReadArea()
'读取国家信息
FileStrm.Seek(country_Offset, SeekOrigin.Begin)
country = ReadString()

Else
'这种模式说明接下来就是保存的国家和地区信息了，以'/0'代表结束。
FileStrm.Seek(-1, SeekOrigin.Current)
country = ReadString()
area = ReadArea()

End If

ElseIf (Tag = &H2) Then
'模式0x02，说明国家信息是一个偏移位置
country_Offset = GetOffset()
'先读取地区信息
area = ReadArea()
'读取国家信息
FileStrm.Seek(country_Offset, SeekOrigin.Begin)
country = ReadString()

Else
'这种模式最简单了，直接读取国家和地区就OK了
FileStrm.Seek(-1, SeekOrigin.Current)
country = ReadString()
area = ReadArea()

End If

Return country + " " + area

End Function

'从当前位置读取一个字符串（国家或者地区信息）
Protected Function ReadString() As String

Dim Offset As UInt32 = 0
Dim TempByteArray(256) As Byte
TempByteArray(Offset) = FileStrm.ReadByte()
While (TempByteArray(Offset) <> &H0)
Offset += 1
TempByteArray(Offset) = FileStrm.ReadByte()
End While

Return System.Text.Encoding.Default.GetString(TempByteArray)

End Function

'从当前文件位置读取地区信息
Protected Function ReadArea() As String

Dim Tag As Byte = GetTag()

If (Tag = &H1 Or Tag = &H2) Then
FileStrm.Seek(GetOffset(), SeekOrigin.Begin)
Return ReadString()
Else
FileStrm.Seek(-1, SeekOrigin.Current)
Return ReadString()
End If

End Function

'从当前文件位置读取一个字节的标识
Protected Function GetTag() As Byte
Return FileStrm.ReadByte()
End Function

'得到指定索引位置的信息（IP起始范围）
Protected Function IndexInfoAtPos(ByVal Index_Pos As Int32) As CZ_INDEX_INFO

Dim Index_Info As New CZ_INDEX_INFO
'根据索引编号计算出在文件中在偏移位置
FileStrm.Seek(Index_Set + 7 * Index_Pos, SeekOrigin.Begin)
Index_Info.IpSet = GetUInt32()
Index_Info.Offset = GetOffset()
FileStrm.Seek(Index_Info.Offset, SeekOrigin.Begin)
Index_Info.IpEnd = GetUInt32()

Return Index_Info

End Function

'读取3个字节，得到偏移位置
Protected Function GetOffset() As UInt32

Dim TempByte4(4) As Byte
TempByte4(0) = FileStrm.ReadByte()
TempByte4(1) = FileStrm.ReadByte()
TempByte4(2) = FileStrm.ReadByte()
TempByte4(3) = 0
Return BitConverter.ToUInt32(TempByte4, 0)

End Function

Protected Function GetUInt32() As UInt32

Dim TempByte4(4) As Byte
FileStrm.Read(TempByte4, 0, 4)
Return BitConverter.ToUInt32(TempByte4, 0)

End Function

'将字符串的IP转换成Int32
Public Shared Function IPToUInt32(ByVal Ipvalue As String) As UInt32

Dim IpByte() As String = Ipvalue.Split(".")
Dim nUpperBound As Int32 = IpByte.GetUpperBound(0)
If nUpperBound <> 3 Then
ReDim Preserve IpByte(4)
For i As Int32 = 1 To 3 - nUpperBound
IpByte(nUpperBound + i) = "0"
Next
End If

Dim TempByte4(4) As Byte
For i As Int32 = 0 To 3
'如果是.Net 2.0可以支持TryParse。
'If Not (Byte.TryParse(IpByte(i), TempByte4(3 - i))) Then
'TempByte4(3 - i) = &H0
'End If
If (IsNumeric(IpByte(i))) Then
TempByte4(3 - i) = CInt(IpByte(i)) And &HFF
End If
Next

Return BitConverter.ToUInt32(TempByte4, 0)
'System.Net.IPAddress.NetworkToHostOrder

End Function