diff --git a/binding/c/ip2region.c b/binding/c/ip2region.c index 67931b3..c87ecc3 100644 --- a/binding/c/ip2region.c +++ b/binding/c/ip2region.c @@ -1,9 +1,15 @@ /** * default ip2region implementation * +<<<<<<< HEAD * @see #ip2region.h * @author chenxin * @date 2015-10-30 +======= + * @see #ip2region.h + * @author chenxin + * @date 2015-10-30 +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 */ #include "ip2region.h" @@ -13,6 +19,7 @@ /** * create a new ip2region object * +<<<<<<< HEAD * @param dbFile path */ IP2R_API uint_t ip2region_create(ip2region_t ip2rObj, char *dbFile) @@ -46,11 +53,47 @@ IP2R_API uint_t ip2region_create(ip2region_t ip2rObj, char *dbFile) ip2rObj->totalBlocks = 0; return 1; +======= + * @param dbFile path +*/ +IP2R_API uint_t ip2region_create(ip2region_t ip2rObj, char *dbFile) +{ + memset(ip2rObj, 0x00, sizeof(ip2region_entry)); + ip2rObj->headerLen = 0; + ip2rObj->HeaderSip = (uint_t *) IP2R_MALLOC(TOTAL_HEADER_LENGTH); + if ( ip2rObj->HeaderSip == NULL ) + { + return 0; + } + + ip2rObj->HeaderPtr = (uint_t *) IP2R_MALLOC(TOTAL_HEADER_LENGTH); + if ( ip2rObj->HeaderPtr == NULL ) + { + IP2R_FREE(ip2rObj->HeaderSip); + return 0; + } + + //open the db file + ip2rObj->dbHandler = fopen(dbFile, "rb"); + if ( ip2rObj->dbHandler == NULL ) + { + IP2R_FREE(ip2rObj->HeaderSip); + IP2R_FREE(ip2rObj->HeaderPtr); + return 0; + } + + ip2rObj->firstIndexPtr = 0; + ip2rObj->lastIndexPtr = 0; + ip2rObj->totalBlocks = 0; + + return 1; +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } /** * destroy the specifield ip2region object * +<<<<<<< HEAD * @param ip2region_t */ IP2R_API uint_t ip2region_destroy(ip2region_t ip2rObj) @@ -68,11 +111,31 @@ IP2R_API uint_t ip2region_destroy(ip2region_t ip2rObj) } return 1; +======= + * @param ip2region_t +*/ +IP2R_API uint_t ip2region_destroy(ip2region_t ip2rObj) +{ + IP2R_FREE(ip2rObj->HeaderSip); + ip2rObj->HeaderSip = NULL; + IP2R_FREE(ip2rObj->HeaderPtr); + ip2rObj->HeaderPtr = NULL; + + //close the db file resource + if ( ip2rObj->dbHandler != NULL ) + { + fclose(ip2rObj->dbHandler); + ip2rObj->dbHandler = NULL; + } + + return 1; +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } /** * get the region associated with the specifield ip address with binary search algorithm * +<<<<<<< HEAD * @param ip2rObj * @param ip * @param datablock @@ -140,16 +203,90 @@ IP2R_API uint_t ip2region_binary_search(ip2region_t ip2rObj, uint_t ip, databloc datablock->region[dataLen] = '\0'; return 1; +======= + * @param ip2rObj + * @param ip + * @param datablock + * @return uint_t +*/ +IP2R_API uint_t ip2region_binary_search(ip2region_t ip2rObj, uint_t ip, datablock_t datablock) +{ + int l, h, m, p; + uint_t sip, eip, dptr; + char buffer[256]; + int dataLen, dataptr; + + if ( ip2rObj->totalBlocks == 0 ) + { + fseek(ip2rObj->dbHandler, 0, 0); + if ( fread(buffer, 8, 1, ip2rObj->dbHandler) != 1 ) { + return 0; + } + + ip2rObj->firstIndexPtr = getUnsignedInt(buffer, 0); + ip2rObj->lastIndexPtr = getUnsignedInt(buffer, 4); + ip2rObj->totalBlocks = (ip2rObj->lastIndexPtr-ip2rObj->firstIndexPtr)/INDEX_BLOCK_LENGTH + 1; + } + + //binary search the index blocks to define the data block + l = 0; h = ip2rObj->totalBlocks; dptr = 0; + while ( l <= h ) { + m = (l + h) >> 1; + p = ip2rObj->firstIndexPtr + m * INDEX_BLOCK_LENGTH; + + fseek(ip2rObj->dbHandler, p, 0); + if ( fread(buffer, INDEX_BLOCK_LENGTH, 1, ip2rObj->dbHandler) != 1 ) { + return 0; + } + 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); + + //memset(data, 0x00, sizeof(data)); + fseek(ip2rObj->dbHandler, dataptr, 0); + if ( fread(buffer, dataLen, 1, ip2rObj->dbHandler) != 1 ) { + return 0; + } + + //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; +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } IP2R_API uint_t ip2region_binary_search_string(ip2region_t ip2rObj, char *ip, datablock_t datablock) { +<<<<<<< HEAD return ip2region_binary_search(ip2rObj, ip2long(ip), datablock); +======= + return ip2region_binary_search(ip2rObj, ip2long(ip), datablock); +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } /** * get the region associated with the specifield ip address with b-tree algorithm * +<<<<<<< HEAD * @param ip2rObj * @param ip * @param datablock @@ -271,16 +408,144 @@ IP2R_API uint_t ip2region_btree_search(ip2region_t ip2rObj, uint_t ip, datablock datablock->region[dataLen] = '\0'; return 1; +======= + * @param ip2rObj + * @param ip + * @param datablock + * @return uint_t +*/ +IP2R_API uint_t ip2region_btree_search(ip2region_t ip2rObj, uint_t ip, datablock_t datablock) +{ + int i, idx; + int l, m, h, p, sptr, eptr, indexBlockLen, dataLen, dataptr; + uint_t sip, eip, idxptr, dptr; + char buffer[TOTAL_HEADER_LENGTH]; + + if ( ip2rObj->headerLen == 0 ) + { + idx = 0; + fseek(ip2rObj->dbHandler, 8, 0); //pass the super block + if ( fread(buffer, TOTAL_HEADER_LENGTH, 1, ip2rObj->dbHandler) != 1 ) { + return 0; + } + + for ( i = 0; i < TOTAL_HEADER_LENGTH; i += 8 ) + { + sip = getUnsignedInt(buffer, i); + idxptr = getUnsignedInt(buffer, i + 4); + if ( idxptr == 0 ) break; + + ip2rObj->HeaderSip[idx] = sip; + ip2rObj->HeaderPtr[idx] = idxptr; + idx++; + } + + ip2rObj->headerLen = idx; + } + + //search the header block to define the index block + l = 0; h = ip2rObj->headerLen; sptr = 0; eptr = 0; + while ( l <= h ) { + m = ((l + h) >> 1); + + //perfetc matched, just return it + if ( ip == ip2rObj->HeaderSip[m] ) { + if ( m > 0 ) { + sptr = ip2rObj->HeaderPtr[m-1]; + eptr = ip2rObj->HeaderPtr[m ]; + } else { + sptr = ip2rObj->HeaderPtr[m ]; + eptr = ip2rObj->HeaderPtr[m+1]; + } + + break; + } + + //less then the middle value + if ( ip < ip2rObj->HeaderSip[m] ) { + if ( m == 0 ) { + sptr = ip2rObj->HeaderPtr[m ]; + eptr = ip2rObj->HeaderPtr[m+1]; + break; + } else if ( ip > ip2rObj->HeaderSip[m-1] ) { + sptr = ip2rObj->HeaderPtr[m-1]; + eptr = ip2rObj->HeaderPtr[m ]; + break; + } + h = m - 1; + } else { + if ( m == ip2rObj->headerLen - 1 ) { + sptr = ip2rObj->HeaderPtr[m-1]; + eptr = ip2rObj->HeaderPtr[m ]; + break; + } else if ( ip <= ip2rObj->HeaderSip[m+1] ) { + sptr = ip2rObj->HeaderPtr[m ]; + eptr = ip2rObj->HeaderPtr[m+1]; + break; + } + l = m + 1; + } + } + + //not matched just stop it + if ( sptr == 0 ) return 0; + + indexBlockLen = eptr - sptr; + fseek(ip2rObj->dbHandler, sptr, 0); + if ( fread(buffer, indexBlockLen + INDEX_BLOCK_LENGTH, 1, ip2rObj->dbHandler) != 1 ) { + return 0; + } + + dptr = 0; l = 0; h = indexBlockLen / INDEX_BLOCK_LENGTH; + while ( l <= h ) { + m = ((l + h) >> 1); + p = m * INDEX_BLOCK_LENGTH; + sip = getUnsignedInt(buffer, p); + if ( ip < sip ) { + h = m - 1; + } else { + eip = getUnsignedInt(buffer, p + 4); + if ( ip > eip ) { + l = m + 1; + } else { + dptr = getUnsignedInt(buffer, p + 8); + break; + } + } + } + + if ( dptr == 0 ) return 0; + + dataLen = ((dptr >> 24) & 0xFF); + dataptr = (dptr & 0x00FFFFFF); + + fseek(ip2rObj->dbHandler, dataptr, 0); + if ( fread(buffer, dataLen, 1, ip2rObj->dbHandler) != 1 ) { + return 0; + } + + datablock->city_id = getUnsignedInt(buffer, 0); + dataLen -= 4; + memcpy(datablock->region, buffer + 4, dataLen); + datablock->region[dataLen] = '\0'; + + return 1; +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } IP2R_API uint_t ip2region_btree_search_string(ip2region_t ip2rObj, char *ip, datablock_t datablock) { +<<<<<<< HEAD return ip2region_btree_search(ip2rObj, ip2long(ip), datablock); +======= + return ip2region_btree_search(ip2rObj, ip2long(ip), datablock); +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } /** * get a unsinged long(4bytes) from a specifield buffer start from the specifield offset * +<<<<<<< HEAD * @param buffer * @param offset * @return uint_t @@ -293,11 +558,26 @@ IP2R_API uint_t getUnsignedInt(char *buffer, int offset) ((buffer[offset+2] << 16) & 0x00FF0000) | ((buffer[offset+3] << 24) & 0xFF000000) ); +======= + * @param buffer + * @param offset + * @return uint_t +*/ +IP2R_API uint_t getUnsignedInt(char *buffer, int offset) +{ + return ( + ((buffer[offset ]) & 0x000000FF) | + ((buffer[offset+1] << 8) & 0x0000FF00) | + ((buffer[offset+2] << 16) & 0x00FF0000) | + ((buffer[offset+3] << 24) & 0xFF000000) + ); +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } /** * string ip to long * +<<<<<<< HEAD * @param ip * @return uint_t */ @@ -334,11 +614,50 @@ IP2R_API uint_t ip2long(char *ip) ipval |= atoi(buffer); return ipval; +======= + * @param ip + * @return uint_t +*/ +IP2R_API uint_t ip2long(char *ip) +{ + int i = 0, p = 24; + char buffer[4], *cs = ip; + uint_t ipval = 0; + + while ( *cs != '\0' ) + { + if ( *cs == '.' ) { + //single part length limit + if ( i > 3 ) { + ipval = 0; + break; + } + + if ( p < 0 ) break; + buffer[i] = '\0'; + ipval |= (atoi(buffer) << p); + p -= 8; + i = 0; + } else { + buffer[i++] = *cs; + } + + cs++; + } + + //append the rest parts + if ( i > 3 ) return 0; + buffer[i] = '\0'; + ipval |= atoi(buffer); + + return ipval; +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } /** * long to string ip * +<<<<<<< HEAD * @param ip * @param buffer * @return uint_t(1 for success and 0 for failed) @@ -346,4 +665,13 @@ IP2R_API uint_t ip2long(char *ip) IP2R_API uint_t long2ip(uint_t ip, char *buffer) { return 0; +======= + * @param ip + * @param buffer + * @return uint_t(1 for success and 0 for failed) +*/ +IP2R_API uint_t long2ip(uint_t ip, char *buffer) +{ + return 0; +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/c/testSearcher.c b/binding/c/testSearcher.c index f1646a2..7b9f69d 100644 --- a/binding/c/testSearcher.c +++ b/binding/c/testSearcher.c @@ -1,8 +1,13 @@ /** * test ip2region searcher program * +<<<<<<< HEAD * @author chenxin * @date 2015-10-30 +======= + * @author chenxin + * @date 2015-10-30 +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 */ #include "ip2region.h" @@ -12,15 +17,23 @@ #include #define __PRINT_ABOUT__ \ +<<<<<<< HEAD println("+-------------------------------------+"); \ println("| ip2region test program |"); \ println("| Author: chenxin619315@gmail.com. |"); \ println("| Type 'quit' to exit the program. |"); \ +======= + println("+-------------------------------------+"); \ + println("| ip2region test program |"); \ + println("| Author: chenxin619315@gmail.com. |"); \ + println("| Type 'quit' to exit the program. |"); \ +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 println("+-------------------------------------+"); //read a line from a command line. static char *getLine( FILE *fp, char *__dst ) { +<<<<<<< HEAD register int c; register char *cs; @@ -32,19 +45,41 @@ static char *getLine( FILE *fp, char *__dst ) *cs = '\0'; return ( c == EOF && cs == __dst ) ? NULL : __dst; +======= + register int c; + register char *cs; + + cs = __dst; + while ( ( c = getc( fp ) ) != EOF ) { + if ( c == '\n' ) break; + *cs++ = c; + } + *cs = '\0'; + + return ( c == EOF && cs == __dst ) ? NULL : __dst; +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } static double getTime() { +<<<<<<< HEAD struct timeval tv; struct timezone tz; gettimeofday(&tv, &tz); return (tv.tv_sec * 1000 + ((double)tv.tv_usec)/1000); +======= + struct timeval tv; + struct timezone tz; + gettimeofday(&tv, &tz); + + return (tv.tv_sec * 1000 + ((double)tv.tv_usec)/1000); +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } int main( int argc, char **argv ) { +<<<<<<< HEAD ip2region_entry ip2rEntry; datablock_entry datablock; char *dbFile = NULL, *algorithm = NULL; @@ -96,4 +131,57 @@ int main( int argc, char **argv ) ip2region_destroy(&ip2rEntry); return 0; +======= + ip2region_entry ip2rEntry; + datablock_entry datablock; + char *dbFile = NULL, *algorithm = NULL; + char line[256]; + uint_t (*func_ptr)(ip2region_t, char *, datablock_t); + double s_time, c_time; + memset(&datablock, 0x00, sizeof(datablock_entry)); + + if ( argc < 2 ) { + printf("Usage: a.out [ip2region db file path] [algorithm]"); + return 0; + } + + dbFile = argv[1]; + algorithm = "B-tree"; + func_ptr = ip2region_btree_search_string; + if ( argc >= 3 && strcmp(argv[2], "binary") == 0 ) { + algorithm = "Binary"; + func_ptr = ip2region_binary_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"); + return 0; + } + + __PRINT_ABOUT__; + + while ( 1 ) + { + print("ip2region>> "); + getLine( stdin, line ); + if ( strlen(line) < 2 ) continue; + if ( strcasecmp( line, "quit" ) == 0 ) { + println("+--Bye!"); + break; + } + + s_time = getTime(); + func_ptr(&ip2rEntry, line, &datablock); + c_time = getTime() - s_time; + printf("%d|%s in %.5f millseconds\n", datablock.city_id, datablock.region, c_time); + } + + //destory the ip2rObj + ip2region_destroy(&ip2rEntry); + + return 0; +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/DataBlock.java b/binding/java/src/org/lionsoul/ip2region/DataBlock.java index e29fbe4..a7a2185 100644 --- a/binding/java/src/org/lionsoul/ip2region/DataBlock.java +++ b/binding/java/src/org/lionsoul/ip2region/DataBlock.java @@ -3,6 +3,7 @@ package org.lionsoul.ip2region; /** * data block class * +<<<<<<< HEAD * @author chenxin */ public class DataBlock @@ -49,4 +50,52 @@ public class DataBlock sb.append(city_id).append('|').append(region); return sb.toString(); } +======= + * @author chenxin +*/ +public class DataBlock +{ + /** + * city id + */ + private int city_id; + + /** + * region address + */ + private String region; + + public DataBlock( int city_id, String region ) + { + this.city_id = city_id; + this.region = region; + } + + public int getCityId() { + return city_id; + } + + public DataBlock setCityId(int city_id) { + this.city_id = city_id; + return this; + } + + public String getRegion() { + return region; + } + + 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(); + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/DbConfig.java b/binding/java/src/org/lionsoul/ip2region/DbConfig.java index 5df40c1..cc4c1af 100644 --- a/binding/java/src/org/lionsoul/ip2region/DbConfig.java +++ b/binding/java/src/org/lionsoul/ip2region/DbConfig.java @@ -7,6 +7,7 @@ package org.lionsoul.ip2region; */ public class DbConfig { +<<<<<<< HEAD /** * total header data block size */ @@ -56,4 +57,55 @@ public class DbConfig this.indexBlockSize = dataBlockSize; return this; } +======= + /** + * total header data block size + */ + private int totalHeaderSize; + + /** + * max index data block size + * u should always choice the fastest read block size + */ + private int indexBlockSize; + + /** + * construct method + * + * @param totalHeaderSize + * @param dataBlockSize + * @throws DbMakerConfigException + */ + public DbConfig( int totalHeaderSize ) throws DbMakerConfigException { + if ( (totalHeaderSize % 8) != 0 ) + { + throw new DbMakerConfigException("totalHeaderSize must be times of 8"); + } + + this.totalHeaderSize = totalHeaderSize; + this.indexBlockSize = 4096; //4 * 1024 + } + + public DbConfig() throws DbMakerConfigException { + this(8192); + } + + public int getTotalHeaderSize() { + return totalHeaderSize; + } + + public DbConfig setTotalHeaderSize(int totalHeaderSize) { + this.totalHeaderSize = totalHeaderSize; + return this; + } + + public int getIndexBlockSize() { + return indexBlockSize; + } + + public DbConfig setIndexBlockSize(int dataBlockSize) { + this.indexBlockSize = dataBlockSize; + return this; + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/DbMakerConfigException.java b/binding/java/src/org/lionsoul/ip2region/DbMakerConfigException.java index b84334f..5fdedd2 100644 --- a/binding/java/src/org/lionsoul/ip2region/DbMakerConfigException.java +++ b/binding/java/src/org/lionsoul/ip2region/DbMakerConfigException.java @@ -7,6 +7,7 @@ package org.lionsoul.ip2region; */ public class DbMakerConfigException extends Exception { +<<<<<<< HEAD private static final long serialVersionUID = 4495714680349884838L; public DbMakerConfigException( String info ) { @@ -20,4 +21,19 @@ public class DbMakerConfigException extends Exception public DbMakerConfigException( String info, Throwable res ) { super(info, res); } +======= + private static final long serialVersionUID = 4495714680349884838L; + + public DbMakerConfigException( String info ) { + super(info); + } + + public DbMakerConfigException( Throwable res ) { + super(res); + } + + public DbMakerConfigException( String info, Throwable res ) { + super(info, res); + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/DbSearcher.java b/binding/java/src/org/lionsoul/ip2region/DbSearcher.java index 01f69cb..17fce7f 100644 --- a/binding/java/src/org/lionsoul/ip2region/DbSearcher.java +++ b/binding/java/src/org/lionsoul/ip2region/DbSearcher.java @@ -11,6 +11,7 @@ import java.io.RandomAccessFile; */ public class DbSearcher { +<<<<<<< HEAD public static final int BTREE_ALGORITHM = 1; public static final int BIN_ALGORITHM = 2; @@ -313,4 +314,308 @@ public class DbSearcher HeaderPtr = null; raf.close(); } +======= + public static final int BTREE_ALGORITHM = 1; + public static final int BIN_ALGORITHM = 2; + + /** + * 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; + + /** + * construct class + * + * @param bdConfig + * @param dbFile + * @throws FileNotFoundException + */ + public DbSearcher( DbConfig dbConfig, String dbFile ) throws FileNotFoundException + { + this.dbConfig = dbConfig; + raf = new RandomAccessFile(dbFile, "r"); + } + + /** + * get by index ptr + * + * @param indexPtr + * @throws IOException + */ + public DataBlock getByIndexPtr( long ptr ) throws IOException + { + raf.seek(ptr); + byte[] buffer = new byte[12]; + raf.readFully(buffer, 0, buffer.length); + //long startIp = Util.getIntLong(buffer, 0); + //long endIp = Util.getIntLong(buffer, 4); + long extra = Util.getIntLong(buffer, 8); + + int dataLen = (int)((extra >> 24) & 0xFF); + int dataPtr = (int)((extra & 0x00FFFFFF)); + + raf.seek(dataPtr); + byte[] data = new byte[dataLen]; + raf.readFully(data, 0, data.length); + + 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); + } + + /** + * get the region with a int ip address with b-tree algorithm + * + * @param ip + * @throws IOException + */ + public DataBlock btreeSearch( long ip ) throws IOException + { + //check and load the header + if ( HeaderSip == null ) + { + raf.seek(8L); //pass the super block + //byte[] b = new byte[dbConfig.getTotalHeaderSize()]; + byte[] b = new byte[4096]; + raf.readFully(b, 0, b.length); + + //fill the header + int len = b.length >> 3, idx = 0; //b.lenght / 8 + HeaderSip = new long[len]; + HeaderPtr = new int [len]; + long startIp, dataPtr; + for ( int i = 0; i < b.length; i += 8 ) { + startIp = Util.getIntLong(b, i); + dataPtr = Util.getIntLong(b, i + 4); + if ( dataPtr == 0 ) break; + + HeaderSip[idx] = startIp; + HeaderPtr[idx] = (int)dataPtr; + idx++; + } + + headerLength = idx; + } + + //1. define the index block with the binary search + if ( ip == HeaderSip[0] ) { + return getByIndexPtr(HeaderPtr[0]); + } else if ( ip == HeaderSip[headerLength-1] ) { + return getByIndexPtr(HeaderPtr[headerLength-1]); + } + + int l = 0, h = headerLength, sptr = 0, eptr = 0; + while ( l <= h ) + { + int m = (l + h) >> 1; + + //perfetc matched, just return it + if ( ip == HeaderSip[m] ) { + if ( m > 0 ) { + sptr = HeaderPtr[m-1]; + eptr = HeaderPtr[m ]; + } else { + sptr = HeaderPtr[m ]; + eptr = HeaderPtr[m+1]; + } + + break; + } + + //less then the middle value + if ( ip < HeaderSip[m] ) { + if ( m == 0 ) { + sptr = HeaderPtr[m ]; + eptr = HeaderPtr[m+1]; + break; + } else if ( ip > HeaderSip[m-1] ) { + sptr = HeaderPtr[m-1]; + eptr = HeaderPtr[m ]; + break; + } + h = m - 1; + } else { + if ( m == headerLength - 1 ) { + sptr = HeaderPtr[m-1]; + eptr = HeaderPtr[m ]; + break; + } else if ( ip <= HeaderSip[m+1] ) { + sptr = HeaderPtr[m ]; + eptr = HeaderPtr[m+1]; + break; + } + l = m + 1; + } + } + + //match nothing just stop it + if ( sptr == 0 ) return null; + + //2. search the index blocks to define the data + int blockLen = eptr - sptr, blen = IndexBlock.getIndexBlockLength(); + byte[] iBuffer = new byte[blockLen + blen]; //include the right border block + raf.seek(sptr); + raf.readFully(iBuffer, 0, iBuffer.length); + + l = 0; h = blockLen / blen; + long sip, eip, dataptr = 0; + while ( l <= h ) { + int m = (l + h) >> 1; + int p = m * blen; + sip = Util.getIntLong(iBuffer, p); + if ( ip < sip ) { + h = m - 1; + } else { + eip = Util.getIntLong(iBuffer, p + 4); + if ( ip > eip ) { + l = m + 1; + } else { + dataptr = Util.getIntLong(iBuffer, p + 8); + break; + } + } + } + + //not matched + if ( dataptr == 0 ) return null; + + //3. get the data + int dataLen = (int)((dataptr >> 24) & 0xFF); + int dataPtr = (int)((dataptr & 0x00FFFFFF)); + + raf.seek(dataPtr); + byte[] data = new byte[dataLen]; + raf.readFully(data, 0, data.length); + + 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); + } + + /** + * get the region throught the ip address with b-tree search algorithm + * + * @param ip + * @return DataBlock + * @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 + * @throws IOException + */ + public DataBlock binarySearch( long ip ) throws IOException + { + int blen = IndexBlock.getIndexBlockLength(); + if ( totalIndexBlocks == 0 ) + { + raf.seek(0L); + byte[] superBytes = new byte[8]; + raf.readFully(superBytes, 0, superBytes.length); + //initialize the global vars + firstIndexPtr = Util.getIntLong(superBytes, 0); + lastIndexPtr = Util.getIntLong(superBytes, 4); + totalIndexBlocks = (int)((lastIndexPtr - firstIndexPtr)/blen) + 1; + } + + //search the index blocks to define the data + int l = 0, h = totalIndexBlocks; + byte[] buffer = new byte[blen]; + long sip, eip, dataptr = 0; + while ( l <= h ) { + int m = (l + h) >> 1; + raf.seek(firstIndexPtr + m * blen); //set the file pointer + raf.readFully(buffer, 0, buffer.length); + sip = Util.getIntLong(buffer, 0); + if ( ip < sip ) { + h = m - 1; + } else { + eip = Util.getIntLong(buffer, 4); + if ( ip > eip ) { + l = m + 1; + } else { + dataptr = Util.getIntLong(buffer, 8); + break; + } + } + } + + //not matched + if ( dataptr == 0 ) return null; + + //get the data + int dataLen = (int)((dataptr >> 24) & 0xFF); + int dataPtr = (int)((dataptr & 0x00FFFFFF)); + + raf.seek(dataPtr); + byte[] data = new byte[dataLen]; + raf.readFully(data, 0, data.length); + + 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); + } + + /** + * get the region throught the ip address with binary search algorithm + * + * @param ip + * @return DataBlock + * @throws IOException + */ + public DataBlock binarySearch( String ip ) throws IOException + { + return binarySearch(Util.ip2long(ip)); + } + + /** + * get the db config + * + * @return DbConfig + */ + public DbConfig getDbConfig() + { + return dbConfig; + } + + /** + * close the db + * + * @throws IOException + */ + public void close() throws IOException + { + HeaderSip = null; //let gc do its work + HeaderPtr = null; + raf.close(); + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/HeaderBlock.java b/binding/java/src/org/lionsoul/ip2region/HeaderBlock.java index c772602..35de6d3 100644 --- a/binding/java/src/org/lionsoul/ip2region/HeaderBlock.java +++ b/binding/java/src/org/lionsoul/ip2region/HeaderBlock.java @@ -7,6 +7,7 @@ package org.lionsoul.ip2region; */ public class HeaderBlock { +<<<<<<< HEAD /** * index block start ip address */ @@ -61,4 +62,60 @@ public class HeaderBlock return b; } +======= + /** + * index block start ip address + */ + private long indexStartIp; + + /** + * ip address + */ + private int indexPtr; + + public HeaderBlock( long indexStartIp, int indexPtr ) + { + this.indexStartIp = indexStartIp; + this.indexPtr = indexPtr; + } + + public long getIndexStartIp() { + return indexStartIp; + } + + public HeaderBlock setIndexStartIp(long indexStartIp) { + this.indexStartIp = indexStartIp; + return this; + } + + public int getIndexPtr() { + return indexPtr; + } + + public HeaderBlock setIndexPtr(int indexPtr) { + this.indexPtr = indexPtr; + return this; + } + + /** + * get the bytes for db storage + * + * @return byte[] + */ + public byte[] getBytes() + { + /* + * +------------+-----------+ + * | 4bytes | 4bytes | + * +------------+-----------+ + * start ip index ptr + */ + byte[] b = new byte[8]; + + Util.writeIntLong(b, 0, indexStartIp); + Util.writeIntLong(b, 4, indexPtr); + + return b; + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/IndexBlock.java b/binding/java/src/org/lionsoul/ip2region/IndexBlock.java index ae9ef74..eb5bffb 100644 --- a/binding/java/src/org/lionsoul/ip2region/IndexBlock.java +++ b/binding/java/src/org/lionsoul/ip2region/IndexBlock.java @@ -7,6 +7,7 @@ package org.lionsoul.ip2region; */ public class IndexBlock { +<<<<<<< HEAD private static int LENGTH = 12; /** @@ -101,4 +102,100 @@ public class IndexBlock return b; } +======= + private static int LENGTH = 12; + + /** + * start ip address + */ + private long startIp; + + /** + * end ip address + */ + private long endIp; + + /** + * data ptr and data length + */ + private int dataPtr; + + /** + * data length + */ + private int dataLen; + + public IndexBlock(long startIp, long endIp, int dataPtr, int dataLen) + { + this.startIp = startIp; + this.endIp = endIp; + this.dataPtr = dataPtr; + this.dataLen = dataLen; + } + + public long getStartIp() { + return startIp; + } + + public IndexBlock setStartIp(long startIp) { + this.startIp = startIp; + return this; + } + + public long getEndIp() { + return endIp; + } + + public IndexBlock setEndIp(long endIp) { + this.endIp = endIp; + return this; + } + + public int getDataPtr() { + return dataPtr; + } + + public IndexBlock setDataPtr(int dataPtr) { + this.dataPtr = dataPtr; + return this; + } + + public int getDataLen() { + return dataLen; + } + + public IndexBlock setDataLen(int dataLen) { + this.dataLen = dataLen; + return this; + } + + public static int getIndexBlockLength() { + return LENGTH; + } + + /** + * get the bytes for storage + * + * @return byte[] + */ + public byte[] getBytes() + { + /* + * +------------+-----------+-----------+ + * | 4bytes | 4bytes | 4bytes | + * +------------+-----------+-----------+ + * start ip end ip data ptr + len + */ + byte[] b = new byte[12]; + + Util.writeIntLong(b, 0, startIp); //start ip + Util.writeIntLong(b, 4, endIp); //end ip + + //write the data ptr and the length + long mix = dataPtr | ((dataLen << 24) & 0xFF000000L); + Util.writeIntLong(b, 8, mix); + + return b; + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/Util.java b/binding/java/src/org/lionsoul/ip2region/Util.java index 0f9d40e..b25de2c 100644 --- a/binding/java/src/org/lionsoul/ip2region/Util.java +++ b/binding/java/src/org/lionsoul/ip2region/Util.java @@ -7,6 +7,7 @@ package org.lionsoul.ip2region; */ public class Util { +<<<<<<< HEAD /** * write specfield bytes to a byte array start from offset * @@ -142,4 +143,141 @@ public class Util return true; } +======= + /** + * write specfield bytes to a byte array start from offset + * + * @param b + * @param offset + * @param v + * @param bytes + */ + public static void write( byte[] b, int offset, long v, int bytes) + { + for ( int i = 0; i < bytes; i++ ) + { + b[offset++] = (byte)((v >>> (8 * i)) & 0xFF); + } + } + + /** + * write a int to a byte array + * + * @param b + * @param offet + * @param v + */ + public static void writeIntLong( byte[] b, int offset, long v ) + { + b[offset++] = (byte)((v >> 0) & 0xFF); + b[offset++] = (byte)((v >> 8) & 0xFF); + b[offset++] = (byte)((v >> 16) & 0xFF); + b[offset ] = (byte)((v >> 24) & 0xFF); + } + + /** + * get a int from a byte array start from the specifiled offset + * + * @param b + * @param offset + */ + public static long getIntLong( byte[] b, int offset ) + { + return ( + ((b[offset++] & 0x000000FFL)) | + ((b[offset++] << 8) & 0x0000FF00L) | + ((b[offset++] << 16) & 0x00FF0000L) | + ((b[offset ] << 24) & 0xFF000000L) + ); + } + + /** + * get a int from a byte array start from the specifield offset + * + * @param b + * @param offset + */ + public static int getInt3( byte[] b, int offset ) + { + return ( + (b[offset++] & 0x000000FF) | + (b[offset++] & 0x0000FF00) | + (b[offset ] & 0x00FF0000) + ); + } + + public static int getInt2( byte[] b, int offset ) + { + return ( + (b[offset++] & 0x000000FF) | + (b[offset ] & 0x0000FF00) + ); + } + + public static int getInt1( byte[] b, int offset ) + { + return ( + (b[offset] & 0x000000FF) + ); + } + + /** + * string ip to long ip + * + * @param ip + * @return long + */ + public static long ip2long( String ip ) + { + String[] p = ip.split("\\."); + if ( p.length != 4 ) return 0; + + int p1 = ((Integer.valueOf(p[0]) << 24) & 0xFF000000); + int p2 = ((Integer.valueOf(p[1]) << 16) & 0x00FF0000); + int p3 = ((Integer.valueOf(p[2]) << 8) & 0x0000FF00); + int p4 = ((Integer.valueOf(p[3]) << 0) & 0x000000FF); + + return ((p1 | p2 | p3 | p4) & 0xFFFFFFFFL); + } + + /** + * int to ip string + * + * @param ip + * @return string + */ + public static String long2ip( long ip ) + { + StringBuilder sb = new StringBuilder(); + + sb + .append((ip >> 24) & 0xFF).append('.') + .append((ip >> 16) & 0xFF).append('.') + .append((ip >> 8) & 0xFF).append('.') + .append((ip >> 0) & 0xFF); + + return sb.toString(); + } + + /** + * check the validate of the specifeld ip address + * + * @param ip + * @return boolean + */ + public static boolean isIpAddress( String ip ) + { + String[] p = ip.split("\\."); + if ( p.length != 4 ) return false; + + for ( String pp : p ) + { + if ( pp.length() > 3 ) return false; + int val = Integer.valueOf(pp); + if ( val > 255 ) return false; + } + + return true; + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/test/TestSearcher.java b/binding/java/src/org/lionsoul/ip2region/test/TestSearcher.java index 619623b..8abce99 100644 --- a/binding/java/src/org/lionsoul/ip2region/test/TestSearcher.java +++ b/binding/java/src/org/lionsoul/ip2region/test/TestSearcher.java @@ -18,6 +18,7 @@ import org.lionsoul.ip2region.Util; */ public class TestSearcher { +<<<<<<< HEAD public static void main(String[] argv) { if ( argv.length == 0 ) { @@ -79,4 +80,67 @@ public class TestSearcher e.printStackTrace(); } } +======= + public static void main(String[] argv) + { + if ( argv.length == 0 ) { + System.out.println("| Usage: java -jar ip2region-{version}.jar [ip2region db file]"); + 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; + } + + if ( argv.length > 1 ) { + if ( argv[1].equalsIgnoreCase("binary")) algorithm = DbSearcher.BIN_ALGORITHM; + } + + try { + System.out.println("initializing "+((algorithm==2)?"Binary":"B-tree")+" ... "); + DbConfig config = new DbConfig(); + DbSearcher seacher = new DbSearcher(config, argv[0]); + BufferedReader reader = new BufferedReader(new InputStreamReader(System.in)); + + System.out.println("+----------------------------------+"); + System.out.println("| ip2region test shell |"); + System.out.println("| Author: chenxin619315@gmail.com |"); + System.out.println("| Type 'quit' to exit program |"); + System.out.println("+----------------------------------+"); + + double sTime = 0, cTime = 0; + String line = null; + DataBlock dataBlock = null; + while ( true ) + { + System.out.print("ip2region>> "); + line = reader.readLine().trim(); + if ( line.length() < 2 ) continue; + if ( line.equalsIgnoreCase("quit") ) break; + if ( Util.isIpAddress(line) == false ) { + System.out.println("Error: Invalid ip address"); + continue; + } + + sTime = System.nanoTime(); + dataBlock = algorithm==2 ? seacher.binarySearch(line) : seacher.btreeSearch(line); + cTime = (System.nanoTime() - sTime) / 1000000; + System.out.printf("%s in %.5f millseconds\n", dataBlock, cTime); + } + + reader.close(); + seacher.close(); + System.out.println("+--Bye"); + } catch (IOException e) { + // TODO Auto-generated catch block + e.printStackTrace(); + } catch (DbMakerConfigException e) { + // TODO Auto-generated catch block + e.printStackTrace(); + } + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/test/TestUnit.java b/binding/java/src/org/lionsoul/ip2region/test/TestUnit.java index 79572a3..a555997 100644 --- a/binding/java/src/org/lionsoul/ip2region/test/TestUnit.java +++ b/binding/java/src/org/lionsoul/ip2region/test/TestUnit.java @@ -19,6 +19,7 @@ import org.lionsoul.ip2region.DbSearcher; public class TestUnit { +<<<<<<< HEAD public static void main(String[] args) { try { DbSearcher _searcher = new DbSearcher(new DbConfig(), "./data/ip2region.db"); @@ -92,5 +93,80 @@ public class TestUnit { e.printStackTrace(); } } +======= + public static void main(String[] args) { + try { + DbSearcher _searcher = new DbSearcher(new DbConfig(), "./data/ip2region.db"); + BufferedReader bfr = new BufferedReader(new FileReader("./data/ip.merge.txt")); + BufferedWriter bwr = new BufferedWriter(new FileWriter("./data/error_log.txt", true)); + int errCount = 0; + int lineCount = 0; + String str = null; + + while ( (str = bfr.readLine()) != null ) { + StringBuffer line = new StringBuffer(str); + //get first ip + int first_idx = line.indexOf("|"); + String first_ip = line.substring(0, first_idx); + + line = new StringBuffer( line.substring(first_idx + 1) ); + + //get second ip + int second_idx = line.indexOf("|"); + String second_ip = line.substring(0, second_idx); + + //get addr + String source_region = line.substring(second_idx + 1); + + //search from DbSearcher + System.out.println("+---Start, start to search"); + System.out.println("+---[Info]: Source region = "+source_region); + + System.out.println("+---[Info]: Step1, search for first IP: "+first_ip); + DataBlock fdata = _searcher.binarySearch(first_ip); + if ( ! fdata.getRegion().equalsIgnoreCase( source_region ) ) { + System.out.println("[Error]: Search first IP failed, DB region = "+fdata.getRegion()); + bwr.write("[Source]: Region: "+fdata.getRegion()); + bwr.newLine(); + bwr.write("[Source]: First Ip: "+first_ip); + bwr.newLine(); + bwr.write("[DB]: Region: "+fdata.getRegion()); + bwr.newLine(); + bwr.flush(); + errCount++; + } + + System.out.println("+---[Info]: Step2, search for second IP: "+second_ip); + DataBlock sdata = _searcher.btreeSearch(second_ip); + if ( ! sdata.getRegion().equalsIgnoreCase( source_region ) ) { + System.out.println("[Error]: Search second IP failed, DB region = "+sdata.getRegion()); + bwr.write("[Source]: Region: "+sdata.getRegion()); + bwr.newLine(); + bwr.write("[Source]: First Ip: "+second_ip); + bwr.newLine(); + bwr.write("[DB]: Region: "+sdata.getRegion()); + bwr.newLine(); + bwr.flush(); + errCount++; + } + + lineCount++; + } + + bwr.close(); + bfr.close(); + System.out.println("+---Done, search complished"); + System.out.println("+---Statistics, Error count = "+errCount + +", Total line = "+lineCount + +", Fail ratio = "+((float)(errCount/lineCount))*100+"%"); + } catch (FileNotFoundException e) { + e.printStackTrace(); + } catch (DbMakerConfigException e) { + e.printStackTrace(); + } catch ( Exception e ) { + e.printStackTrace(); + } + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/java/src/org/lionsoul/ip2region/test/TestUtil.java b/binding/java/src/org/lionsoul/ip2region/test/TestUtil.java index c10cc53..1d32342 100644 --- a/binding/java/src/org/lionsoul/ip2region/test/TestUtil.java +++ b/binding/java/src/org/lionsoul/ip2region/test/TestUtil.java @@ -9,6 +9,7 @@ import org.lionsoul.ip2region.Util; */ public class TestUtil { +<<<<<<< HEAD public static void main(String[] argv) { /* //1. test the ip2long @@ -54,4 +55,51 @@ public class TestUtil System.out.println(Util.ip2long("255.255.255.0")); } +======= + public static void main(String[] argv) + { +/* //1. test the ip2long + String[] ipSet = new String[]{ + "120.24.78.68", + "120.24.229.68", + "120.24.87.145", + "218.17.162.99" + }; + + for ( String ip : ipSet ) + { + int ipInt = Util.ip2Int(ip); + System.out.println("src ip: " + ip + ", ip2Int: " + ipInt + ", int2IP: " + Util.int2IP(ipInt)); + }*/ + +/* int[] arr = new int[]{12344, -1234, 2146789, 0, -1024}; + byte[] b = new byte[arr.length * 4]; + + //write the int + System.out.println("+--Testing writeInt ... "); + int i, idx = 0; + for ( i = 0; i < b.length; i += 4 ) + { + System.out.println("offset: " + i); + Util.writeInt(b, i, arr[idx++]); + } + System.out.println("|----[Ok]"); + + //read the int + System.out.println("+--Testing getInt ... "); + idx = 0; + for ( i = 0; i < b.length; i += 4 ) + { + System.out.println(arr[idx++]+", " + Util.getInt(b, i)); + } + System.out.println("|----[Ok]");*/ + +/* HeaderBlock headerBlock = new HeaderBlock(241658345, 2134785); + byte[] b = headerBlock.getBytes(); + System.out.println(headerBlock.getIndexStartIp() + ", " + headerBlock.getIndexPtr()); + System.out.println(Util.getInt(b, 0) + ", " + Util.getInt(b, 4));*/ + + System.out.println(Util.ip2long("255.255.255.0")); + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } diff --git a/binding/php/Ip2Region.class.php b/binding/php/Ip2Region.class.php index 8bfe64e..4c46c09 100644 --- a/binding/php/Ip2Region.class.php +++ b/binding/php/Ip2Region.class.php @@ -2,6 +2,7 @@ /** * ip2region php seacher client class * +<<<<<<< HEAD * @author chenxin * @date 2015-10-29 */ @@ -249,5 +250,254 @@ class Ip2Region $this->HeaderSip = NULL; $this->HeaderPtr = NULL; } +======= + * @author chenxin + * @date 2015-10-29 +*/ + +defined('INDEX_BLOCK_LENGTH') or define('INDEX_BLOCK_LENGTH', 12); +defined('TOTAL_HEADER_LENGTH') or define('TOTAL_HEADER_LENGTH', 4096); + +class Ip2Region +{ + /** + * db file handler + */ + private $dbFileHandler = NULL; + + /** + * header block info + */ + private $HeaderSip = NULL; + private $HeaderPtr = NULL; + private $headerLen = 0; + + /** + * super block index info + */ + private $firstIndexPtr = 0; + private $lastIndexPtr = 0; + private $totalBlocks = 0; + + /** + * construct method + * + * @param ip2regionFile + */ + public function __construct( $ip2regionFile ) + { + $this->dbFileHandler = fopen($ip2regionFile, 'r'); + } + + /** + * get the data block throught the specifield ip address or long ip numeric with binary search algorithm + * + * @param ip + * @return mixed Array or NULL for any error + */ + public function binarySearch( $ip ) + { + //check and conver the ip address + if ( is_string($ip) ) $ip = ip2long($ip); + if ( $this->totalBlocks == 0 ) + { + fseek($this->dbFileHandler, 0); + $superBlock = fread($this->dbFileHandler, 8); + + $this->firstIndexPtr = self::getLong($superBlock, 0); + $this->lastIndexPtr = self::getLong($superBlock, 4); + $this->totalBlocks = ($this->lastIndexPtr-$this->firstIndexPtr)/INDEX_BLOCK_LENGTH + 1; + } + + //binary search to define the data + $l = 0; + $h = $this->totalBlocks; + $dataPtr = 0; + while ( $l <= $h ) + { + $m = (($l + $h) >> 1); + $p = $m * INDEX_BLOCK_LENGTH; + + fseek($this->dbFileHandler, $this->firstIndexPtr + $p); + $buffer = fread($this->dbFileHandler, INDEX_BLOCK_LENGTH); + $sip = self::getLong($buffer, 0); + if ( $ip < $sip ) { + $h = $m - 1; + } else { + $eip = self::getLong($buffer, 4); + if ( $ip > $eip ) { + $l = $m + 1; + } else { + $dataPtr = self::getLong($buffer, 8); + break; + } + } + } + + //not matched just stop it here + if ( $dataPtr == 0 ) return NULL; + + + //get the data + $dataLen = (($dataPtr >> 24) & 0xFF); + $dataPtr = ($dataPtr & 0x00FFFFFF); + + fseek($this->dbFileHandler, $dataPtr); + $data = fread($this->dbFileHandler, $dataLen); + + return array( + 'city_id' => self::getLong($data, 0), + 'region' => substr($data, 4) + ); + } + + /** + * get the data block associated with the specifield ip with b-tree search algorithm + * + * @param ip + * @return Mixed Array for NULL for any error + */ + public function btreeSearch( $ip ) + { + if ( is_string($ip) ) $ip = ip2long($ip); + + //check and load the header + if ( $this->HeaderSip == NULL ) + { + fseek($this->dbFileHandler, 8); + $buffer = fread($this->dbFileHandler, TOTAL_HEADER_LENGTH); + + //fill the header + $idx = 0; + $this->HeaderSip = array(); + $this->HeaderPtr = array(); + for ( $i = 0; $i < TOTAL_HEADER_LENGTH; $i += 8 ) + { + $startIp = self::getLong($buffer, $i); + $dataPtr = self::getLong($buffer, $i + 4); + if ( $dataPtr == 0 ) break; + + $this->HeaderSip[] = $startIp; + $this->HeaderPtr[] = $dataPtr; + $idx++; + } + + $this->headerLen = $idx; + } + + //1. define the index block with the binary search + $l = 0; $h = $this->headerLen; $sptr = 0; $eptr = 0; + while ( $l <= $h ) + { + $m = (($l + $h) >> 1); + + //perfetc matched, just return it + if ( $ip == $this->HeaderSip[$m] ) { + if ( $m > 0 ) { + $sptr = $this->HeaderPtr[$m-1]; + $eptr = $this->HeaderPtr[$m ]; + } else { + $sptr = $this->HeaderPtr[$m ]; + $eptr = $this->HeaderPtr[$m+1]; + } + + break; + } + + //less then the middle value + if ( $ip < $this->HeaderSip[$m] ) { + if ( $m == 0 ) { + $sptr = $this->HeaderPtr[$m ]; + $eptr = $this->HeaderPtr[$m+1]; + break; + } else if ( $ip > $this->HeaderSip[$m-1] ) { + $sptr = $this->HeaderPtr[$m-1]; + $eptr = $this->HeaderPtr[$m ]; + break; + } + $h = $m - 1; + } else { + if ( $m == $this->headerLen - 1 ) { + $sptr = $this->HeaderPtr[$m-1]; + $eptr = $this->HeaderPtr[$m ]; + break; + } else if ( $ip <= $this->HeaderSip[$m+1] ) { + $sptr = $this->HeaderPtr[$m ]; + $eptr = $this->HeaderPtr[$m+1]; + break; + } + $l = $m + 1; + } + } + + //match nothing just stop it + if ( $sptr == 0 ) return NULL; + + //2. search the index blocks to define the data + $blockLen = $eptr - $sptr; + fseek($this->dbFileHandler, $sptr); + $index = fread($this->dbFileHandler, $blockLen + INDEX_BLOCK_LENGTH); + + $dataptr = 0; + $l = 0; $h = $blockLen / INDEX_BLOCK_LENGTH; + while ( $l <= $h ) { + $m = (($l + $h) >> 1); + $p = (int)($m * INDEX_BLOCK_LENGTH); + $sip = self::getLong($index, $p); + if ( $ip < $sip ) { + $h = $m - 1; + } else { + $eip = self::getLong($index, $p + 4); + if ( $ip > $eip ) { + $l = $m + 1; + } else { + $dataptr = self::getLong($index, $p + 8); + break; + } + } + } + + //not matched + if ( $dataptr == 0 ) return NULL; + + //3. get the data + $dataLen = (($dataptr >> 24) & 0xFF); + $dataPtr = ($dataptr & 0x00FFFFFF); + + fseek($this->dbFileHandler, $dataPtr); + $data = fread($this->dbFileHandler, $dataLen); + + return array( + 'city_id' => self::getLong($data, 0), + 'region' => substr($data, 4) + ); + } + + /** + * read a long from a byte buffer + * + * @param b + * @param offset + */ + public static function getLong( $b, $offset ) + { + return ( + (ord($b[$offset++])) | + (ord($b[$offset++]) << 8) | + (ord($b[$offset++]) << 16) | + (ord($b[$offset ]) << 24) + ); + } + + /** + * destruct method, resource destroy + */ + public function __destruct() + { + if ( $this->dbFileHandler != NULL ) fclose($this->dbFileHandler); + $this->HeaderSip = NULL; + $this->HeaderPtr = NULL; + } +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } ?> diff --git a/binding/php/testSeacher.php b/binding/php/testSeacher.php index d3c991f..48271f7 100644 --- a/binding/php/testSeacher.php +++ b/binding/php/testSeacher.php @@ -2,11 +2,16 @@ /** * Ip2Region php client test script * +<<<<<<< HEAD * @author chenxin +======= + * @author chenxin +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 */ if ( $argc < 2 ) { +<<<<<<< HEAD $usage = <<>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } require dirname(__FILE__) . '/Ip2Region.class.php'; @@ -39,6 +61,7 @@ echo $initStr, "\n"; while ( true ) { +<<<<<<< HEAD echo "ip2region>> "; $line = trim(fgets(STDIN)); if ( strlen($line) < 2 ) continue; @@ -52,10 +75,29 @@ while ( true ) $data = $method==2 ? $ip2regionObj->binarySearch($line) : $ip2regionObj->btreeSearch($line); $c_time = getTime() - $s_time; printf("%s|%s in %.5f millseconds\n", $data['city_id'], $data['region'], $c_time); +======= + echo "ip2region>> "; + $line = trim(fgets(STDIN)); + if ( strlen($line) < 2 ) continue; + if ( $line == 'quit' ) break; + if ( ip2long($line) == NULL ) { + echo "Error: invalid ip address\n"; + continue; + } + + $s_time = getTime(); + $data = $method==2 ? $ip2regionObj->binarySearch($line) : $ip2regionObj->btreeSearch($line); + $c_time = getTime() - $s_time; + printf("%s|%s in %.5f millseconds\n", $data['city_id'], $data['region'], $c_time); +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } function getTime() { +<<<<<<< HEAD return (microtime(true) * 1000); +======= + return (microtime(true) * 1000); +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 } ?> diff --git a/binding/python/ip2Region.py b/binding/python/ip2Region.py index 087ae3c..12d73d0 100644 --- a/binding/python/ip2Region.py +++ b/binding/python/ip2Region.py @@ -8,6 +8,7 @@ import struct, io, socket, sys class Ip2Region(object): +<<<<<<< HEAD __headerSip = [] __headerPtr = [] __f = None @@ -190,3 +191,187 @@ class Ip2Region(object): self.__headerPtr = None self.__f.close() self.__f = None +======= + __headerSip = [] + __headerPtr = [] + __f = None + + def __init__(self, dbfile): + self.initDatabase(dbfile) + + def binarySearch(self, ip): + """ + " binary search method + " param: ip + """ + if not ip.isdigit(): ip = self.ip2long(ip) + + self.__f.seek(0) + b = self.__f.read(8) + startPtr = self.getLong(b, 0) + endPtr = self.getLong(b, 4) + + indexLen = endPtr - startPtr + self.__f.seek(startPtr) + b = self.__f.read(indexLen+12) + + l, h, mixPtr = (0, int(indexLen/12), 0) + while l <= h: + m = int((l+h)/2) + ptr = startPtr + m*12 + self.__f.seek(ptr) + + b = self.__f.read(12) + sip = self.getLong(b, 0) + eip = self.getLong(b, 4) + + if ip > sip: + if ip > eip: + l = m + 1 + else: + mixPtr = self.getLong(b, 8) + break; + else: + h = m - 1 + + if mixPtr == 0: return "N2" + + dataPtr = mixPtr & 0x00FFFFFFL + dataLen = (mixPtr >> 24) & 0xFF + + self.__f.seek(dataPtr) + data = self.__f.read(dataLen) + return { + "city_id": self.getLong(data, 0), + "region" : data[4:] + } + + def btreeSearch(self, ip): + """ + " b-tree search method + " param: ip + """ + if not ip.isdigit(): ip = self.ip2long(ip) + + headerLen = len(self.__headerSip) - 1 + l, h, sptr, eptr = (0, headerLen, 0, 0) + while l <= h: + m = int((l+h)/2) + + if ip == self.__headerSip[m]: + if m > 0: + sptr = self.__headerPtr[m-1] + eptr = self.__headerPtr[m] + break; + else: + sptr = self.__headerPtr[m] + eptr = self.__headerPtr[m+1] + break; + + if ip > self.__headerSip[m]: + if m == headerLen: + sptr = self.__headerPtr[m-1] + eptr = self.__headerPtr[m] + break; + elif ip < self.__headerSip[m+1]: + sptr = self.__headerPtr[m] + eptr = self.__headerPtr[m+1] + break; + + l = m + 1 + else: + if m == 0: + sptr = self.__headerPtr[m] + eptr = self.__headerPtr[m+1] + break; + elif ip > self.__headerSip[m-1]: + sptr = self.__headerPtr[m-1] + eptr = self.__headerPtr[m] + break; + + h = m - 1 + + if sptr == 0: return "N1" + + indexLen = eptr - sptr + self.__f.seek(sptr) + b = self.__f.read(indexLen + 12) + + l, h, mixPtr = (0, int(indexLen/12), 0) + while l <= h: + m = int((l+h)/2) + offset = m * 12 + + if ip > self.getLong(b, offset): + if ip > self.getLong(b, offset+4): + l = m + 1 + else: + mixPtr = self.getLong(b, offset+8) + break; + else: + h = m - 1 + + if mixPtr == 0: return "N2" + + dataPtr = mixPtr & 0x00FFFFFFL + dataLen = (mixPtr >> 24) & 0xFF + + self.__f.seek(dataPtr) + data = self.__f.read(dataLen) + return { + "city_id": self.getLong(data, 0), + "region" : data[4:] + } + + def initDatabase(self, dbfile): + """ + " initialize the database for search + " param: dbFile + """ + try: + self.__f = io.open(dbfile, "rb") + #pass the super block + self.__f.seek(8) + #read the header block + b = self.__f.read(4086) + #parse the header block + sip = None + ptr = None + for i in range(0, len(b)-1, 8): + sip = self.getLong(b, i) + ptr = self.getLong(b, i+4) + if ptr == 0: + break + self.__headerSip.append(sip) + self.__headerPtr.append(ptr) + + except IOError, e: + print "[Error]: ", e + sys.exit() + + def ip2long(self, ip): + _ip = socket.inet_aton(ip) + return struct.unpack("!L", _ip)[0] + + def isip(self, ip): + p = ip.split(".") + + if len(p) != 4 : return False + for pp in p: + if not pp.isdigit(): return False + if len(pp) > 3 : return False + if int(pp) > 255 : return False + + return True + + def getLong(self, b, offset): + if len( b[offset:offset+4] ) == 4: + return struct.unpack('I', b[offset:offset+4])[0] + return 0 + + def close(self): + self.__headerSip = None + self.__headerPtr = None + self.__f.close() + self.__f = None +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56 diff --git a/binding/python/testSearcher.py b/binding/python/testSearcher.py index af02480..1560ad5 100644 --- a/binding/python/testSearcher.py +++ b/binding/python/testSearcher.py @@ -9,6 +9,7 @@ import struct, sys, os, time from ip2Region import Ip2Region def testSearch(): +<<<<<<< HEAD """ " ip2region test function """ @@ -72,4 +73,70 @@ def testSearch(): searcher.close() if __name__ == "__main__": - testSearch() \ No newline at end of file + testSearch() +======= + """ + " ip2region test function + """ + llen = len(sys.argv) + + if llen < 2: + print "Usage: python ip2Region.py [ip2region db file] [alrogrithm]" + print "Algorithm: binary or b-tree" + return 0 + + dbFile = sys.argv[1] + method = 1 + algorithm = "b-tree" + + if (not os.path.isfile(dbFile)) or (not os.path.exists(dbFile)): + print "[Error]: Specified db file is not exists." + return 0 + + if llen > 2: + algorithm = sys.argv[2] + if algorithm == "binary": + method = 2 + + print "initializing %s..." % (algorithm) + print "+----------------------------------+" + print "| ip2region test script |" + print "| Author: komazhang@foxmail.com |" + print "| Type 'quit' to exit program |" + print "+----------------------------------+" + + searcher = Ip2Region(dbFile); + + while True: + line = raw_input("ip2region>> ") + line = line.strip() + + if line == "": + print "[Error]: Invalid ip address." + continue + + if line == "quit": + print "[Info]: Thanks for your use, Bye." + break + + if not searcher.isip(line): + print "[Error]: Invalid ip address." + continue + + sTime = time.time() * 1000 + if method == 1: + data = searcher.btreeSearch(line) + else: + data = searcher.binarySearch(line) + eTime = time.time() * 1000 + + if isinstance(data, dict): + print "[Return]: %s|%s in %f millseconds" % (data["city_id"], data["region"], eTime-sTime) + else: + print "[Error]: ", data + + searcher.close() + +if __name__ == "__main__": + testSearch() +>>>>>>> 7e51a4909fdb01014f948c2dbc5cfb5fbed9ce56