提交 1d4d3f92 编写于 作者: L Lion

add vector index and xdb content load util func

上级 79b3dd2a
......@@ -54,52 +54,6 @@ func (s *Searcher) Close() {
}
}
// LoadVectorIndex load and cache the vector index for search speedup.
// this will take up VectorIndexRows x VectorIndexCols x VectorIndexSize bytes memory.
func (s *Searcher) LoadVectorIndex() error {
// loaded already
if s.vectorIndex != nil {
return nil
}
// load all the vector index block
_, err := s.handle.Seek(HeaderInfoLength, 0)
if err != nil {
return fmt.Errorf("seek to vector index: %w", err)
}
var buff = make([]byte, VectorIndexRows*VectorIndexCols*VectorIndexSize)
rLen, err := s.handle.Read(buff)
if err != nil {
return err
}
if rLen != len(buff) {
return fmt.Errorf("incomplete read: readed bytes should be %d", len(buff))
}
// decode the vector index blocks
var vectorIndex = make([][]*VectorIndexBlock, VectorIndexRows)
for r := 0; r < VectorIndexRows; r++ {
vectorIndex[r] = make([]*VectorIndexBlock, VectorIndexCols)
for c := 0; c < VectorIndexCols; c++ {
offset := r*VectorIndexCols*VectorIndexSize + c*VectorIndexSize
vectorIndex[r][c], err = VectorIndexBlockDecode(buff[offset:])
if err != nil {
return fmt.Errorf("decode vector index at [%d][%d]: %w", r, c, err)
}
}
}
s.vectorIndex = vectorIndex
return nil
}
// ClearVectorIndex clear preloaded vector index cache
func (s *Searcher) ClearVectorIndex() {
s.vectorIndex = nil
}
// SearchByStr find the region for the specified ip string
func (s *Searcher) SearchByStr(str string) (string, error) {
ip, err := CheckIP(str)
......
......@@ -3,6 +3,7 @@ package xdb
import (
"encoding/binary"
"fmt"
"os"
"strconv"
"strings"
)
......@@ -43,3 +44,86 @@ func Long2IP(ip uint32) string {
func MidIP(sip uint32, eip uint32) uint32 {
return uint32((uint64(sip) + uint64(eip)) >> 1)
}
// LoadVectorIndex util function to load the vector index from the specified file handle
func LoadVectorIndex(handle *os.File) ([][]*VectorIndexBlock, error) {
// load all the vector index block
_, err := handle.Seek(HeaderInfoLength, 0)
if err != nil {
return nil, fmt.Errorf("seek to vector index: %w", err)
}
var buff = make([]byte, VectorIndexRows*VectorIndexCols*VectorIndexSize)
rLen, err := handle.Read(buff)
if err != nil {
return nil, err
}
if rLen != len(buff) {
return nil, fmt.Errorf("incomplete read: readed bytes should be %d", len(buff))
}
// decode the vector index blocks
var vectorIndex = make([][]*VectorIndexBlock, VectorIndexRows)
for r := 0; r < VectorIndexRows; r++ {
vectorIndex[r] = make([]*VectorIndexBlock, VectorIndexCols)
for c := 0; c < VectorIndexCols; c++ {
offset := r*VectorIndexCols*VectorIndexSize + c*VectorIndexSize
vectorIndex[r][c], err = VectorIndexBlockDecode(buff[offset:])
if err != nil {
return nil, fmt.Errorf("decode vector index at [%d][%d]: %w", r, c, err)
}
}
}
return vectorIndex, nil
}
// LoadVectorIndexFromFile load vector index from a specified file path
func LoadVectorIndexFromFile(dbFile string) ([][]*VectorIndexBlock, error) {
handle, err := os.OpenFile(dbFile, os.O_RDONLY, 0600)
if err != nil {
return nil, fmt.Errorf("open xdb file `%s`: %w", dbFile, err)
}
return LoadVectorIndex(handle)
}
// LoadContent load the whole xdb content from the specified file handle
func LoadContent(handle *os.File) ([]byte, error) {
// get file size
fi, err := handle.Stat()
if err != nil {
return nil, fmt.Errorf("stat: %w", err)
}
size := fi.Size()
// seek to the head of the file
_, err = handle.Seek(0, 0)
if err != nil {
return nil, fmt.Errorf("seek to get xdb file length: %w", err)
}
var buff = make([]byte, size)
rLen, err := handle.Read(buff)
if err != nil {
return nil, err
}
if rLen != len(buff) {
return nil, fmt.Errorf("incomplete read: readed bytes should be %d", len(buff))
}
return buff, nil
}
// LoadContentFromFile load the whole xdb content from the specified db file path
func LoadContentFromFile(dbFile string) ([]byte, error) {
handle, err := os.OpenFile(dbFile, os.O_RDONLY, 0600)
if err != nil {
return nil, fmt.Errorf("open xdb file `%s`: %w", dbFile, err)
}
return LoadContent(handle)
}
// Copyright 2022 The Ip2Region Authors. All rights reserved.
// Use of this source code is governed by a Apache2.0-style
// license that can be found in the LICENSE file.
package xdb
import (
"fmt"
"testing"
)
func TestLoadVectorIndex(t *testing.T) {
vIndex, err := LoadVectorIndexFromFile("../../../data/ip2region.xdb")
if err != nil {
fmt.Printf("failed to load vector index: %s\n", err)
return
}
fmt.Printf("vIndex length: %d\n", len(vIndex))
}
func TestLoadContent(t *testing.T) {
buff, err := LoadContentFromFile("../../../data/ip2region.xdb")
if err != nil {
fmt.Printf("failed to load xdb content: %s\n", err)
}
fmt.Printf("buff length: %d\n", len(buff))
}
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册