Merge branch 'master' of github.com:lionsoul2016/ip2region into nodejsclient

CHANGES.md

@@ -0,0 +1,7 @@
+### 1.2
+
+1. 新增分列式升级算法，大大的增加了数据的准确率，基本避免之前各大网友反馈的些许ip定位错误！
+2. 数据升级是2016/06/30版！
+3. 优化数据文件生成算法，ip2region.db文件由原来的3.5M降为1.5M，42亿个IP地址，皆大欢喜啊！
+4. C/PHP/JAVA客户端增加纯内存搜索模式（python, php扩展，nodejs会在后续版本加上！）
+说明：因为数据文件只有1.5M，对于PHP,java这类IO优化类语言提升速度不大，C有一个数量级的提升！

README.md

@@ -1,6 +1,6 @@
-ip2region - ip到地区的映射库，妈妈再也不用担心我的ip定位。
+ip2region - 最自由的ip地址查询库，ip到地区的映射库，提供Binary,B树和纯内存三种查询算法，妈妈再也不用担心我的ip地址定位。
 
-**1. 99.9%准确率，不定时更新：**
+**1. 99.9%准确率，定时更新：**
 
 数据聚合了一些知名ip到地名查询提供商的数据，这些是他们官方的的准确率，经测试着实比纯真啥的准确多了。<br />
 每次聚合一下数据需要1-2天，会不定时更新。
@@ -14,20 +14,20 @@ ip2region - ip到地区的映射库，妈妈再也不用担心我的ip定位。
 
 **3. 体积小：**
 
-生成的数据库文件ip2region.db只有3.5M
+生成的数据库文件ip2region.db只有1.5M（1.2版本前是3.5M）
 
 **4. 多查询客户端的支持，0.0x毫秒级别的查询**
 
 已经集成的客户端有：java, php, c, python，php扩展(支持linux, php5, php7版本已支持)。
 
     提供了两种查询算法，响应时间如下：
-    客户端/binary算法/b-tree算法：
+    客户端/binary算法/b-tree算法/Memory算法：
-    java/0.x毫秒/0.x毫秒 (使用RandomAccessFile)
+    java/0.x毫秒/0.x毫秒/0.1x毫秒 (使用RandomAccessFile)
-    php/0.x毫秒/0.1x毫秒
+    php/0.x毫秒/0.1x毫秒/0.1x毫秒
-    c/0.0x毫秒/0.0x毫秒(b-tree算法基本稳定在0.02x毫秒级别)
+    c/0.0x毫秒/0.0x毫秒/0.00x毫秒(b-tree算法基本稳定在0.02x毫秒级别)
-    python/0.x毫秒/0.1x毫秒
+    python/0.x毫秒/0.1x毫秒/未知
 
-任何客户端b-tree都比binary算法快
+任何客户端b-tree都比binary算法快，当然Memory算法固然是最快的！
 
 **5. 测试程序：**
 java: 
@@ -61,10 +61,6 @@ python:
     p2region>> 101.105.35.57
     2163|中国|华南|广东省|深圳市|鹏博士 in 0.02295 millseconds
 
-输入ip地址开始测试，第一次会稍微有点慢，在运行命令后面接入binary来尝试binary算法，建议使用b-tree算法。
+输入ip地址开始测试，第一次会稍微有点慢，在运行命令后面接入binary,memory来尝试其他算法，建议使用b-tree算法，速度和并发需求的可以使用memory算法。
 
 具体集成请参考不同客户端的测试源码。
-
-**6. 联系作者：**
-
-狮子的魂： chenxin619315@gmail.com

binding/c/ip2region.c

@@ -35,12 +35,15 @@ IP2R_API uint_t ip2region_create(ip2region_t ip2rObj, char *dbFile)
     if ( ip2rObj->dbHandler == NULL  ) {
         IP2R_FREE(ip2rObj->HeaderSip);
         IP2R_FREE(ip2rObj->HeaderPtr);
+        //fprintf(stderr, "Fail to open the db file %s\n", ip2rObj>dbFile);
+        //exit(-1);
         return 0;
     }
 
     ip2rObj->firstIndexPtr = 0;
     ip2rObj->lastIndexPtr  = 0;
     ip2rObj->totalBlocks   = 0;
+    ip2rObj->dbBinStr      = NULL;
 
     return 1;
 }
@@ -63,9 +66,94 @@ IP2R_API uint_t ip2region_destroy(ip2region_t ip2rObj)
         ip2rObj->dbHandler = NULL;
     }
 
+    //free the db binary string
+    if ( ip2rObj->dbBinStr != NULL ) {
+        IP2R_FREE(ip2rObj->dbBinStr);
+        ip2rObj->dbBinStr = NULL;
+    }
+
     return 1;
 }
 
+/**
+ * get the region associated with the specified ip address with the memory binary search algorithm
+ *
+ * @param   ip2rObj
+ * @param   ip
+ * @param   datablock
+*/
+IP2R_API uint_t ip2region_memory_search(ip2region_t ip2rObj, uint_t ip, datablock_t datablock) 
+{
+    int l, h, m, p;
+    uint_t sip, eip, dptr;
+    int dataLen, dataptr;
+    long filesize;
+    char *buffer;
+
+    if ( ip2rObj->dbBinStr == NULL ) {
+        //get the size of the file
+        fseek(ip2rObj->dbHandler, 0, SEEK_END);
+        filesize = ftell(ip2rObj->dbHandler);
+        fseek(ip2rObj->dbHandler, 0, SEEK_SET);
+
+        //alloc the buffer size
+        ip2rObj->dbBinStr = IP2R_MALLOC(filesize);
+        if ( ip2rObj->dbBinStr == NULL ) {
+            return 0;
+        }
+
+        //now read the whole file
+        if ( fread(ip2rObj->dbBinStr, filesize, 1, ip2rObj->dbHandler) != 1 ) {
+            return 0;
+        }
+
+        buffer = ip2rObj->dbBinStr;
+        ip2rObj->firstIndexPtr = getUnsignedInt(buffer, 0);
+        ip2rObj->lastIndexPtr  = getUnsignedInt(buffer, 4);
+        ip2rObj->totalBlocks   = (ip2rObj->lastIndexPtr-ip2rObj->firstIndexPtr)/INDEX_BLOCK_LENGTH + 1;
+    }
+
+    l = 0; h = ip2rObj->totalBlocks; dptr = 0;
+    while ( l <= h ) {
+        m = (l + h) >> 1;
+        p = ip2rObj->firstIndexPtr + m * INDEX_BLOCK_LENGTH;
+
+        buffer = ip2rObj->dbBinStr + p;
+        sip    = getUnsignedInt(buffer, 0);
+        if ( ip < sip ) {
+            h = m - 1;
+        } else {
+            eip = getUnsignedInt(buffer, 4);
+            if ( ip > eip ) {
+                l = m + 1;
+            } else {
+                dptr = getUnsignedInt(buffer, 8);
+                break;
+            }
+        }
+    }
+
+    if ( dptr == 0 ) return 0;
+
+    //get the data
+    dataLen = ((dptr >> 24) & 0xFF);
+    dataptr = (dptr & 0x00FFFFFF);
+    buffer  = ip2rObj->dbBinStr + dataptr;
+
+    //fill the data to the datablock
+    datablock->city_id = getUnsignedInt(buffer, 0);
+    dataLen -= 4;    //reduce the length of the city_id
+    memcpy(datablock->region, buffer + 4, dataLen); 
+    datablock->region[dataLen] = '\0';
+
+    return 1;
+}
+
+IP2R_API uint_t ip2region_memory_search_string(ip2region_t ip2rObj, char *ip, datablock_t datablock)
+{
+    return ip2region_memory_search(ip2rObj, ip2long(ip), datablock);
+}
+
 /**
  * get the region associated with the specifield ip address with binary search algorithm
  *

binding/c/ip2region.h

@@ -52,7 +52,9 @@ typedef struct {
 	uint_t *HeaderSip;		//header start ip blocks
 	uint_t *HeaderPtr;		//header ptr blocks
 	uint_t headerLen;		//header block number
+    char *dbFile;           //path of db file
 	FILE *dbHandler;		//file handler
+    char *dbBinStr;         //db binary string for memory search mode
 
 	uint_t firstIndexPtr;	//first index ptr
 	uint_t lastIndexPtr;	//last index ptr
@@ -79,14 +81,26 @@ typedef datablock_entry * datablock_t;
 IP2R_API uint_t ip2region_create(ip2region_t, char *);
 
 /**
- * destroy the specifield ip2region object
+ * destroy the specified ip2region object
  *
  * @param	ip2region_t
 */
 IP2R_API uint_t ip2region_destroy(ip2region_t);
 
 /**
- * get the region associated with the specifield ip address with binary search algorithm
+ * get the region associated with the specified ip address with the memory binary search algorithm
+ *
+ * @param   ip2region_t
+ * @param   uint_t
+ * @param   datablock_t
+ * @date    2016/06/30
+*/
+IP2R_API uint_t ip2region_memory_search(ip2region_t, uint_t, datablock_t);
+IP2R_API uint_t ip2region_memory_search_string(ip2region_t, char *, datablock_t);
+
+
+/**
+ * get the region associated with the specified ip address with binary search algorithm
  *
  * @param	ip2rObj
  * @param	ip
@@ -97,7 +111,7 @@ IP2R_API uint_t ip2region_binary_search(ip2region_t, uint_t, datablock_t);
 IP2R_API uint_t ip2region_binary_search_string(ip2region_t, char *, datablock_t);
 
 /**
- * get the region associated with the specifield ip address with b-tree algorithm
+ * get the region associated with the specified ip address with b-tree algorithm
  *
  * @param	ip2rObj
  * @param	ip
@@ -108,7 +122,7 @@ IP2R_API uint_t ip2region_btree_search(ip2region_t, uint_t, datablock_t);
 IP2R_API uint_t ip2region_btree_search_string(ip2region_t, char *, datablock_t);
 
 /**
- * get a unsinged long(4bytes) from a specifield buffer start from the specifield offset
+ * get a unsinged long(4bytes) from a specified buffer start from the specified offset
  *
  * @param	buffer
  * @param	offset

binding/c/testSearcher.c

@@ -54,22 +54,27 @@ int main( int argc, char **argv )
     memset(&datablock, 0x00, sizeof(datablock_entry));
 
     if ( argc < 2  ) {
-        printf("Usage: a.out [ip2region db file path] [algorithm]");
+        printf("Usage: a.out [ip2region db file path] [algorithm]\n");
         return 0;
     }
 
     dbFile      = argv[1];
     algorithm = "B-tree";
     func_ptr  = ip2region_btree_search_string;
-    if ( argc >= 3 && strcmp(argv[2], "binary") == 0 ) {
+    if ( argc >= 3 ) {
-        algorithm = "Binary";
+        if ( strcmp(argv[2], "binary") == 0 ) {
-        func_ptr = ip2region_binary_search_string;
+            algorithm = "Binary";
+            func_ptr = ip2region_binary_search_string;
+        } else if ( strcmp(argv[2], "memory") == 0 ) {
+            algorithm = "Memory";
+            func_ptr = ip2region_memory_search_string;
+        }
     }
 
     //create a new ip2rObj
     printf("+--initializing %s ... \n", algorithm);
     if ( ip2region_create(&ip2rEntry, dbFile) == 0 ) {
-        println("Error: Fail to create the ip2region object");
+        println("Error: Fail to create the ip2region object\n");
         return 0;
     }
 

binding/java/build.xml

@@ -9,7 +9,7 @@
 		<property name="jars" value="${basedir}"/>
 		<property name="sources" value="${basedir}/src"/>
 		<property name="classes" value="${basedir}/classes"/>
-		<property name="version" value="1.0"/>
+		<property name="version" value="1.2"/>
 		<property name="api" value="${basedir}/doc"/>
 		<mkdir dir="${classes}"/>
 		<mkdir dir="${api}"/>

binding/java/src/org/lionsoul/ip2region/DataBlock.java

@@ -3,54 +3,84 @@ package org.lionsoul.ip2region;
 /**
  * data block class
  * 
- * @author    chenxin<chenxin619315@gmail.com>
+ * @author	chenxin<chenxin619315@gmail.com>
 */
 public class DataBlock 
 {
-    /**
+	/**
-     * city id 
+	 * city id 
-    */
+	*/
-    private int city_id;
+	private int city_id;
-    
+	
-    /**
+	/**
-     * region address
+	 * region address
-    */
+	*/
-    private String region;
+	private String region;
-    
+	
-    public DataBlock( int city_id, String region )
+	/**
-    {
+	 * region ptr in the db file
-        this.city_id = city_id;
+	*/
-        this.region = region;
+	private int dataPtr;
-    }
+	
+	/**
+	 * construct method
+	 * 
+	 * @param  city_id
+	 * @param  region  region string
+	 * @param  ptr data ptr
+	*/
+	public DataBlock( int city_id, String region, int dataPtr )
+	{
+		this.city_id = city_id;
+		this.region  = region;
+		this.dataPtr = dataPtr;
+	}
+	
+	public DataBlock(int city_id, String region)
+	{
+	    this(city_id, region, 0);
+	}
 
-    public int getCityId()
+	public int getCityId()
-    {
+	{
-        return city_id;
+		return city_id;
-    }
+	}
 
-    public DataBlock setCityId(int city_id)
+	public DataBlock setCityId(int city_id)
-    {
+	{
-        this.city_id = city_id;
+		this.city_id = city_id;
-        return this;
+		return this;
-    }
+	}
 
-    public String getRegion()
+	public String getRegion()
-    {
+	{
-        return region;
+		return region;
-    }
+	}
+
+	public DataBlock setRegion(String region)
+	{
+		this.region = region;
+		return this;
+	}
+	
+	public int getDataPtr()
+	{
+	    return dataPtr;
+	}
+	
+	public DataBlock setDataPtr(int dataPtr)
+	{
+	    this.dataPtr = dataPtr;
+	    return this;
+	}
+	
+	@Override
+	public String toString()
+	{
+		StringBuilder sb = new StringBuilder();
+		
+		sb.append(city_id).append('|').append(region).append('|').append(dataPtr);
+		return sb.toString();
+	}
 
-    public DataBlock setRegion(String region)
-    {
-        this.region = region;
-        return this;
-    }
-    
-    @Override
-    public String toString()
-    {
-        StringBuilder sb = new StringBuilder();
-        
-        sb.append(city_id).append('|').append(region);
-        return sb.toString();
-    }
 }

binding/java/src/org/lionsoul/ip2region/DbSearcher.java

@@ -11,51 +11,124 @@ import java.io.RandomAccessFile;
 */
 public class DbSearcher 
 {
-    public static final int BTREE_ALGORITHM = 1;
+    public static final int BTREE_ALGORITHM  = 1;
-    public static final int BIN_ALGORITHM = 2;
+    public static final int BINARY_ALGORITHM = 2;
-    
+    public static final int MEMORY_ALGORITYM = 3;
+
     /**
      * db config
     */
     private DbConfig dbConfig = null;
-    
+
     /**
      * db file access handler
     */
     private RandomAccessFile raf = null;
-    
+
     /**
      * header blocks buffer 
     */
     private long[] HeaderSip = null;
     private int[]  HeaderPtr = null;
     private int headerLength;
-    
+
     /**
      * super blocks info 
     */
     private long firstIndexPtr = 0;
     private long lastIndexPtr = 0;
     private int totalIndexBlocks = 0;
-    
+
+    /**
+     * for memory mode
+     * the original db binary string
+    */
+    private byte[] dbBinStr = null;
+
     /**
      * construct class
      * 
-     * @param    bdConfig
+     * @param   bdConfig
-     * @param    dbFile
+     * @param   dbFile
-     * @throws FileNotFoundException 
+     * @throws  FileNotFoundException 
     */
     public DbSearcher( DbConfig dbConfig, String dbFile ) throws FileNotFoundException
     {
         this.dbConfig = dbConfig;
         raf = new RandomAccessFile(dbFile, "r");
     }
+
+    /**
+     * get the region with a int ip address with memory binary search algorithm
+     *
+     * @param   ip
+     * @throws  IOException
+    */
+    public DataBlock memorySearch(long ip) throws IOException
+    {
+        int blen = IndexBlock.getIndexBlockLength();
+        if ( dbBinStr == null ) {
+            dbBinStr = new byte[(int)raf.length()];
+            raf.seek(0L);
+            raf.readFully(dbBinStr, 0, dbBinStr.length);
+            
+            //initialize the global vars
+            firstIndexPtr    = Util.getIntLong(dbBinStr, 0);
+            lastIndexPtr     = Util.getIntLong(dbBinStr, 4);
+            totalIndexBlocks = (int)((lastIndexPtr - firstIndexPtr)/blen) + 1;
+        }
+        
+        //search the index blocks to define the data
+        int l = 0, h = totalIndexBlocks;
+        long sip, eip, dataptr = 0;
+        while ( l <= h ) {
+            int m = (l + h) >> 1;
+            int p = (int)(firstIndexPtr + m * blen);
+            
+            sip = Util.getIntLong(dbBinStr, p);
+            if ( ip < sip ) {
+                h = m - 1;
+            } else {
+                eip = Util.getIntLong(dbBinStr, p + 4);
+                if ( ip > eip ) {
+                    l = m + 1;
+                } else {
+                    dataptr = Util.getIntLong(dbBinStr, p + 8);
+                    break;
+                }
+            }
+        }
+        
+        //not matched
+        if ( dataptr == 0 ) return null;
+        
+        //get the data
+        int dataLen = (int)((dataptr >> 24) & 0xFF);
+        int dataPtr = (int)((dataptr & 0x00FFFFFF));
+        int city_id = (int)Util.getIntLong(dbBinStr, dataPtr);
+        String region = new String(dbBinStr, dataPtr + 4, dataLen - 4, "UTF-8");
+        
+        return new DataBlock(city_id, region, dataPtr);
+    }
     
+    /**
+     * get the region throught the ip address with memory binary search algorithm
+     * 
+     * @param   ip
+     * @return  DataBlock
+     * @throws  IOException 
+    */
+    public DataBlock memorySearch( String ip ) throws IOException
+    {
+        return memorySearch(Util.ip2long(ip));
+    }
+    
+
     /**
      * get by index ptr
      * 
-     * @param    indexPtr
+     * @param   indexPtr
-     * @throws IOException 
+     * @throws  IOException 
     */
     public DataBlock getByIndexPtr( long ptr ) throws IOException
     {
@@ -76,13 +149,13 @@ public class DbSearcher
         int city_id = (int)Util.getIntLong(data, 0);
         String region = new String(data, 4, data.length - 4, "UTF-8");
         
-        return new DataBlock(city_id, region);
+        return new DataBlock(city_id, region, dataPtr);
     }
-    
+
     /**
      * get the region with a int ip address with b-tree algorithm
      * 
-     * @param    ip
+     * @param   ip
      * @throws  IOException 
     */
     public DataBlock btreeSearch( long ip ) throws IOException
@@ -204,25 +277,25 @@ public class DbSearcher
         int city_id = (int)Util.getIntLong(data, 0);
         String region = new String(data, 4, data.length - 4, "UTF-8");
         
-        return new DataBlock(city_id, region);
+        return new DataBlock(city_id, region, dataPtr);
     }
-    
+
     /**
      * get the region throught the ip address with b-tree search algorithm
      * 
-     * @param    ip
+     * @param   ip
-     * @return    DataBlock
+     * @return  DataBlock
-     * @throws IOException 
+     * @throws  IOException 
     */
     public DataBlock btreeSearch( String ip ) throws IOException
     {
         return btreeSearch(Util.ip2long(ip));
     }
-    
+
     /**
      * get the region with a int ip address with binary search algorithm
      * 
-     * @param    ip
+     * @param   ip
      * @throws  IOException 
     */
     public DataBlock binarySearch( long ip ) throws IOException
@@ -274,15 +347,15 @@ public class DbSearcher
         int city_id = (int)Util.getIntLong(data, 0);
         String region = new String(data, 4, data.length - 4, "UTF-8");
         
-        return new DataBlock(city_id, region);
+        return new DataBlock(city_id, region, dataPtr);
     }
-    
+
     /**
      * get the region throught the ip address with binary search algorithm
      * 
-     * @param    ip
+     * @param   ip
-     * @return    DataBlock
+     * @return  DataBlock
-     * @throws IOException 
+     * @throws  IOException 
     */
     public DataBlock binarySearch( String ip ) throws IOException
     {
@@ -292,13 +365,13 @@ public class DbSearcher
     /**
      * get the db config
      *
-     * @return    DbConfig
+     * @return  DbConfig
     */
     public DbConfig getDbConfig()
     {
         return dbConfig;
     }
-    
+
     /**
      * close the db 
      * 
@@ -308,6 +381,8 @@ public class DbSearcher
     {
         HeaderSip = null;    //let gc do its work
         HeaderPtr = null;
+        dbBinStr  = null;
         raf.close();
     }
+
 }

binding/java/src/org/lionsoul/ip2region/test/TestSearcher.java

@@ -4,6 +4,8 @@ import java.io.BufferedReader;
 import java.io.File;
 import java.io.IOException;
 import java.io.InputStreamReader;
+import java.lang.reflect.InvocationTargetException;
+import java.lang.reflect.Method;
 
 import org.lionsoul.ip2region.DataBlock;
 import org.lionsoul.ip2region.DbConfig;
@@ -25,23 +27,45 @@ public class TestSearcher
             return;
         }
         
-        int algorithm = DbSearcher.BTREE_ALGORITHM;
         File file = new File(argv[0]);
         if ( file.exists() == false ) {
             System.out.println("Error: Invalid ip2region.db file");
             return;
         }
         
+        int algorithm = DbSearcher.BTREE_ALGORITHM;
+        String algoName = "B-tree";
         if ( argv.length > 1 ) {
-            if ( argv[1].equalsIgnoreCase("binary")) algorithm = DbSearcher.BIN_ALGORITHM;
+            if ( argv[1].equalsIgnoreCase("binary")) {
+                algoName  = "Binary"; 
+                algorithm = DbSearcher.BINARY_ALGORITHM;
+            } else if ( argv[1].equalsIgnoreCase("memory") ) {
+                algoName  = "Memory";
+                algorithm = DbSearcher.MEMORY_ALGORITYM;
+            }
         }
         
         try {
-            System.out.println("initializing "+((algorithm==2)?"Binary":"B-tree")+" ... ");
+            System.out.println("initializing "+algoName+" ... ");
             DbConfig config = new DbConfig();
-            DbSearcher seacher = new DbSearcher(config, argv[0]);
+            DbSearcher searcher = new DbSearcher(config, argv[0]);
             BufferedReader reader = new BufferedReader(new InputStreamReader(System.in));
             
+            //define the method
+            Method method = null;
+            switch ( algorithm ) 
+            {
+            case DbSearcher.BTREE_ALGORITHM:
+                method = searcher.getClass().getMethod("btreeSearch", String.class);
+                break;
+            case DbSearcher.BINARY_ALGORITHM:
+                method = searcher.getClass().getMethod("binarySearch", String.class);
+                break;
+            case DbSearcher.MEMORY_ALGORITYM:
+                method = searcher.getClass().getMethod("memorySearch", String.class);
+                break;
+            }
+            
             System.out.println("+----------------------------------+");
             System.out.println("| ip2region test shell             |");
             System.out.println("| Author: chenxin619315@gmail.com  |");
@@ -62,13 +86,13 @@ public class TestSearcher
                 }
                 
                 sTime = System.nanoTime();
-                dataBlock = algorithm==2 ? seacher.binarySearch(line) : seacher.btreeSearch(line);
+                dataBlock = (DataBlock) method.invoke(searcher, line);
                 cTime = (System.nanoTime() - sTime) / 1000000;
                 System.out.printf("%s in %.5f millseconds\n", dataBlock, cTime);
             }
             
             reader.close();
-            seacher.close();
+            searcher.close();
             System.out.println("+--Bye");
         } catch (IOException e) {
             // TODO Auto-generated catch block
@@ -76,6 +100,21 @@ public class TestSearcher
         } catch (DbMakerConfigException e) {
             // TODO Auto-generated catch block
             e.printStackTrace();
+        } catch (NoSuchMethodException e) {
+            // TODO Auto-generated catch block
+            e.printStackTrace();
+        } catch (SecurityException e) {
+            // TODO Auto-generated catch block
+            e.printStackTrace();
+        } catch (IllegalAccessException e) {
+            // TODO Auto-generated catch block
+            e.printStackTrace();
+        } catch (IllegalArgumentException e) {
+            // TODO Auto-generated catch block
+            e.printStackTrace();
+        } catch (InvocationTargetException e) {
+            // TODO Auto-generated catch block
+            e.printStackTrace();
         }
     }
 }

binding/php/Ip2Region.class.php

@@ -2,12 +2,12 @@
 /**
  * ip2region php seacher client class
  *
- * @author    chenxin<chenxin619315@gmail.com>
+ * @author  chenxin<chenxin619315@gmail.com>
  * @date    2015-10-29
 */
 
-defined('INDEX_BLOCK_LENGTH')    or define('INDEX_BLOCK_LENGTH',  12);
+defined('INDEX_BLOCK_LENGTH')   or define('INDEX_BLOCK_LENGTH',  12);
-defined('TOTAL_HEADER_LENGTH')    or define('TOTAL_HEADER_LENGTH', 4096);
+defined('TOTAL_HEADER_LENGTH')  or define('TOTAL_HEADER_LENGTH', 4096);
 
 class Ip2Region 
 {
@@ -29,6 +29,13 @@ class Ip2Region
     private $firstIndexPtr = 0;
     private $lastIndexPtr  = 0;
     private $totalBlocks   = 0;
+
+    /**
+     * for memory mode only
+     *  the original db binary string
+    */
+    private $dbBinStr = NULL;
+    private $dbFile = NULL;
     
     /**
      * construct method
@@ -37,7 +44,65 @@ class Ip2Region
     */
     public function __construct( $ip2regionFile )
     {
-        $this->dbFileHandler = fopen($ip2regionFile, 'r');
+        $this->dbFile = $ip2regionFile;
+    }
+
+    /**
+     * all the db binary string will be loaded into memory
+     * then search the memory only and this will a lot faster than disk base search
+     * @Note: 
+     * invoke it once before put it to public invoke could make it thread safe
+     *
+     * @param   $ip
+    */
+    public function memorySearch($ip)
+    {
+        //check and load the binary string for the first time
+        if ( $this->dbBinStr == NULL ) {
+            $this->dbBinStr = file_get_contents($this->dbFile);
+            if ( $this->dbBinStr == false ) {
+                throw new Exception("Fail to open the db file {$this->dbFile}");
+            }
+
+            $this->firstIndexPtr = self::getLong($this->dbBinStr, 0);
+            $this->lastIndexPtr  = self::getLong($this->dbBinStr, 4);
+            $this->totalBlocks   = ($this->lastIndexPtr-$this->firstIndexPtr)/INDEX_BLOCK_LENGTH + 1;
+        }
+
+        if ( is_string($ip) ) $ip = ip2long($ip);
+
+        //binary search to define the data
+        $l = 0;
+        $h = $this->totalBlocks;
+        $dataPtr = 0;
+        while ( $l <= $h ) {
+            $m = (($l + $h) >> 1);
+            $p = $this->firstIndexPtr + $m * INDEX_BLOCK_LENGTH;
+            $sip = self::getLong($this->dbBinStr, $p);
+            if ( $ip < $sip ) {
+                $h = $m - 1;
+            } else {
+                $eip = self::getLong($this->dbBinStr, $p + 4);
+                if ( $ip > $eip ) {
+                    $l = $m + 1;
+                } else {
+                    $dataPtr = self::getLong($this->dbBinStr, $p + 8);
+                    break;
+                }
+            }
+        }
+
+        //not matched just stop it here
+        if ( $dataPtr == 0 ) return NULL;
+
+        //get the data
+        $dataLen = (($dataPtr >> 24) & 0xFF);
+        $dataPtr = ($dataPtr & 0x00FFFFFF);
+
+        return array(
+            'city_id' => self::getLong($this->dbBinStr, $dataPtr), 
+            'region'  => substr($this->dbBinStr, $dataPtr + 4, $dataLen - 4)
+        );
     }
 
     /**
@@ -51,6 +116,14 @@ class Ip2Region
         //check and conver the ip address
         if ( is_string($ip) ) $ip = ip2long($ip);
         if ( $this->totalBlocks == 0 ) {
+            //check and open the original db file
+            if ( $this->dbFileHandler == NULL ) {
+                $this->dbFileHandler = fopen($this->dbFile, 'r');
+                if ( $this->dbFileHandler == false ) {
+                    throw new Exception("Fail to open the db file {$this->dbFile}");
+                }
+            }
+
             fseek($this->dbFileHandler, 0);
             $superBlock = fread($this->dbFileHandler, 8);
 
@@ -102,9 +175,10 @@ class Ip2Region
 
     /**
      * get the data block associated with the specifield ip with b-tree search algorithm
+     * @Note: not thread safe
      *
-     * @param    ip
+     * @param   ip
-     * @return    Mixed Array for NULL for any error
+     * @return  Mixed Array for NULL for any error
     */
     public function btreeSearch( $ip )
     {
@@ -112,6 +186,14 @@ class Ip2Region
 
         //check and load the header
         if ( $this->HeaderSip == NULL ) {
+            //check and open the original db file
+            if ( $this->dbFileHandler == NULL ) {
+                $this->dbFileHandler = fopen($this->dbFile, 'r');
+                if ( $this->dbFileHandler == false ) {
+                    throw new Exception("Fail to open the db file {$this->dbFile}");
+                }
+            }
+
             fseek($this->dbFileHandler, 8);
             $buffer = fread($this->dbFileHandler, TOTAL_HEADER_LENGTH);
             
@@ -240,7 +322,11 @@ class Ip2Region
     */
     public function __destruct()
     {
-        if ( $this->dbFileHandler != NULL ) fclose($this->dbFileHandler);
+        if ( $this->dbFileHandler != NULL ) {
+            fclose($this->dbFileHandler);
+        }
+
+        $this->dbBinStr  = NULL;
         $this->HeaderSip = NULL;
         $this->HeaderPtr = NULL;
     }

binding/php/testSeacher.php

@@ -15,12 +15,19 @@ EOF;
 
 array_shift($argv);
 $dbFile     = $argv[0];
-$method     = 1;
+$method     = 'btreeSearch';
 $algorithm  = 'B-tree';
-if ( isset($argv[1])
+if ( isset($argv[1]) ) {
-    && strtolower($argv[1]) == 'binary' ) {
+    switch ( strtolower($argv[1]) ) {
-    $method    = 2;
+    case 'binary':
-    $algorithm = 'Binary';
+        $algorithm = 'Binary';
+        $method    = 'binarySearch';
+        break;
+    case 'memory':
+        $algorithm = 'Memory';
+        $method    = 'memorySearch';
+        break;
+    }
 }
 
 require dirname(__FILE__) . '/Ip2Region.class.php';
@@ -47,7 +54,7 @@ while ( true ) {
     }
 
     $s_time = getTime();
-    $data   = $method==2 ? $ip2regionObj->binarySearch($line) : $ip2regionObj->btreeSearch($line);
+    $data   = $ip2regionObj->{$method}($line);
     $c_time = getTime() - $s_time;
     printf("%s|%s in %.5f millseconds\n", $data['city_id'], $data['region'], $c_time);
 }