Merge pull request #478 from lionsoul2014/golang_maker_region_mgr

Golang maker region struct and cache impl
This commit is contained in:
Leon / 狮子的魂 2026-04-29 17:41:08 +08:00 committed by GitHub
commit 593e028677
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
8 changed files with 182 additions and 33 deletions

View File

@ -90,12 +90,16 @@ func Bench(sCmd string) {
fmt.Printf("failed to open source text file: %s\n", err)
return
}
defer handle.Close()
var rgCache = xdb.NewRegionCache()
defer rgCache.Clean()
var count, errCount, tStart = 0, 0, time.Now()
slog.Info("Bench start", "xdbPath", dbFile, "srcPath", srcFile)
_, _, iErr := xdb.IterateSegments(handle, false, nil, nil, func(seg *xdb.Segment) error {
_, _, iErr := xdb.IterateSegments(handle, false, func(l string) {
// do thing here
}, nil, rgCache.Region, func(seg *xdb.Segment) error {
var l = fmt.Sprintf("%d|%d|%s", seg.StartIP, seg.EndIP, seg.Region)
slog.Debug("try to bench", "segment", l)
// mip := xdb.IPMiddle(seg.StartIP, seg.EndIP)
@ -109,7 +113,7 @@ func Bench(sCmd string) {
// check the region info
count++
if r != seg.Region {
if r != seg.Region.Str {
errCount++
slog.Error(" --[Failed] region not match", "src", r, "dst", seg.Region)
if !ignoreError {

View File

@ -24,6 +24,9 @@ type Editor struct {
// segments list
segments *list.List
// region cache
rgCache *RegionCache
}
func NewEditor(version *Version, srcFile string) (*Editor, error) {
@ -44,6 +47,7 @@ func NewEditor(version *Version, srcFile string) (*Editor, error) {
srcHandle: srcHandle,
toSave: false,
segments: list.New(),
rgCache: NewRegionCache(),
}
// load the segments
@ -54,6 +58,10 @@ func NewEditor(version *Version, srcFile string) (*Editor, error) {
return e, nil
}
func (e *Editor) Region(str string) *Region {
return e.rgCache.Region(str)
}
// Load all the segments from the source file
func (e *Editor) loadSegments() error {
var last *Segment = nil
@ -62,7 +70,7 @@ func (e *Editor) loadSegments() error {
_, _, iErr := IterateSegments(e.srcHandle, true, func(l string) {
// do nothing here
}, nil, func(seg *Segment) error {
}, nil, e.Region, func(seg *Segment) error {
// version check
if len(seg.StartIP) != e.verison.Bytes {
return fmt.Errorf("invalid ip segment(%s expected)", e.verison.Name)
@ -100,15 +108,12 @@ func (e *Editor) loadSegments() error {
// to Keep the entire data continuous.
last = nil
for _, seg := range segments {
if err := seg.After(last); err != nil {
}
if last == nil {
if IPCompare(seg.StartIP, e.verison.Min) > 0 {
e.segments.PushBack(&Segment{
StartIP: e.verison.Min,
EndIP: IPSubOne(seg.StartIP),
Region: "",
Region: EmptyRegion,
})
}
} else if err := seg.RightBehind(last); err == nil {
@ -121,7 +126,7 @@ func (e *Editor) loadSegments() error {
e.segments.PushBack(&Segment{
StartIP: IPAddOne(last.EndIP),
EndIP: IPSubOne(seg.StartIP),
Region: "",
Region: EmptyRegion,
})
}
@ -138,7 +143,7 @@ func (e *Editor) loadSegments() error {
e.segments.PushBack(&Segment{
StartIP: IPAddOne(back.Value.(*Segment).EndIP),
EndIP: e.verison.Max,
Region: "",
Region: EmptyRegion,
})
}
}
@ -182,7 +187,7 @@ func (e *Editor) Slice(offset int, size int) []*Segment {
}
func (e *Editor) Put(ip string, cb func(newSeg *Segment, oldList []*Segment) []*Segment) (int, int, error) {
seg, err := SegmentFrom(ip)
seg, err := SegmentFrom(ip, e.rgCache.Region)
if err != nil {
return 0, 0, err
}
@ -350,7 +355,7 @@ func (e *Editor) PutFile(src string, cb func(newSeg *Segment, oldList []*Segment
var oldRows, newRows = 0, 0
_, _, iErr := IterateSegments(handle, true, func(l string) {
// do nothing here
}, nil, func(seg *Segment) error {
}, nil, NewRegion, func(seg *Segment) error {
o, n, err := e.PutSegment(seg, cb)
if err == nil {
oldRows += o
@ -395,7 +400,7 @@ func (e *Editor) SaveToFile(dstFile string) error {
}
// ignore the padded or empty segment
if s.Region == "" {
if s.Region.IsEmpty() {
continue
}
@ -415,4 +420,5 @@ func (e *Editor) SaveToFile(dstFile string) error {
func (e *Editor) Close() {
_ = e.srcHandle.Close()
e.rgCache.Clean()
}

View File

@ -84,6 +84,7 @@ type Maker struct {
indexPolicy IndexPolicy
segments []*Segment
regionPool map[string]uint32
regionCache *RegionCache
vectorIndex []byte
}
@ -118,6 +119,7 @@ func NewMaker(version *Version, policy IndexPolicy, srcFile string, dstFile stri
indexPolicy: policy,
segments: []*Segment{},
regionPool: map[string]uint32{},
regionCache: NewRegionCache(),
vectorIndex: make([]byte, VectorIndexLength),
}, nil
}
@ -173,7 +175,7 @@ func (m *Maker) loadSegments() error {
}, func(region string) (string, error) {
// apply the field filter
return RegionFiltering(region, m.fields)
}, func(seg *Segment) error {
}, m.regionCache.Region, func(seg *Segment) error {
// ip version check
if len(seg.StartIP) != m.version.Bytes {
return fmt.Errorf("invalid ip segment(%s expected)", m.version.Name)
@ -276,13 +278,13 @@ func (m *Maker) Start() error {
slog.Info("try to write the data block ... ")
for _, seg := range m.segments {
slog.Debug("try to write", "region", seg.Region)
ptr, has := m.regionPool[seg.Region]
ptr, has := m.regionPool[seg.Region.Str]
if has {
slog.Debug(" --[Cached]", "ptr=", ptr)
continue
}
var region = []byte(seg.Region)
var region = []byte(seg.Region.Str)
if len(region) > 0xFFFF {
return fmt.Errorf("too long region info `%s`: should be less than %d bytes", seg.Region, 0xFFFF)
}
@ -303,7 +305,7 @@ func (m *Maker) Start() error {
return fmt.Errorf("write region '%s': %w", seg.Region, err)
}
m.regionPool[seg.Region] = uint32(pos)
m.regionPool[seg.Region.Str] = uint32(pos)
slog.Debug(" --[Added] with", "ptr", pos)
}
@ -312,14 +314,14 @@ func (m *Maker) Start() error {
var indexBuff = make([]byte, m.version.SegmentIndexSize)
var counter, startIndexPtr, endIndexPtr = 0, int64(-1), int64(-1)
for _, seg := range m.segments {
dataPtr, has := m.regionPool[seg.Region]
dataPtr, has := m.regionPool[seg.Region.Str]
if !has {
return fmt.Errorf("missing ptr cache for region `%s`", seg.Region)
}
// @Note: data length should be the length of bytes.
// this works fine because of the string feature (byte sequence) of golang.
var dataLen = len(seg.Region)
var dataLen = len(seg.Region.Str)
if dataLen < 1 {
// @TODO: could this even be a case ?
// return fmt.Errorf("empty region info for segment '%s'", seg)
@ -412,5 +414,7 @@ func (m *Maker) End() error {
return err
}
m.regionCache.Clean()
return nil
}

View File

@ -26,6 +26,9 @@ type Processor struct {
fields []int
segments []*Segment
// region cache
rgCache *RegionCache
}
func NewProcessor(srcFile string, dstFile string, fields []int,
@ -55,6 +58,7 @@ func NewProcessor(srcFile string, dstFile string, fields []int,
fields: fields,
segments: []*Segment{},
rgCache: NewRegionCache(),
}, nil
}
@ -94,7 +98,7 @@ func (p *Processor) loadSegments() error {
}
return RegionFiltering(region, p.fields)
}, func(seg *Segment) error {
}, p.rgCache.Region, func(seg *Segment) error {
// check the continuity of the data segment
// if err := seg.AfterCheck(last); err != nil {
// return err
@ -152,5 +156,7 @@ func (p *Processor) End() error {
return err
}
p.rgCache.Clean()
return nil
}

129
maker/golang/xdb/region.go Normal file
View File

@ -0,0 +1,129 @@
package xdb
import (
"fmt"
"strings"
"sync"
)
// region manager with:
// 1, content cache.
// 2, util functions
// --- region
type Region struct {
Str string // region string
fields []string // region fields
}
var EmptyRegion = NewRegion("")
// Create a new region without checking cache info
func NewRegion(str string) *Region {
return &Region{
Str: str,
fields: nil,
}
}
func (r *Region) Fields() []string {
if r.fields == nil {
r.fields = strings.Split(r.Str, "|")
}
return r.fields
}
func (r *Region) Join(sep string) string {
if sep == "|" {
return r.Str
}
return strings.Join(r.Fields(), sep)
}
func (r *Region) Filtering(fields []int) (*Region, error) {
if len(fields) == 0 {
return r, nil
}
fs := r.Fields()
var sb []string
for _, idx := range fields {
if idx < 0 {
return r, fmt.Errorf("negative filter index %d", idx)
}
if idx >= len(fs) {
return r, fmt.Errorf("field index %d exceeded the max length of %d", idx, len(fs))
}
sb = append(sb, fs[idx])
}
return &Region{
Str: strings.Join(sb, "|"),
fields: sb,
}, nil
}
// Equal check ptr (share the same region cache) or the Str is the same.
func (r *Region) Equal(dst *Region) bool {
return (r == dst || r.Str == dst.Str)
}
func (r *Region) IsEmpty() bool {
return r.Str == ""
}
func (r *Region) String() string {
return r.Str
}
// ---
// --- region cache
type RegionCache struct {
lock sync.Mutex
cache map[string]*Region
}
func NewRegionCache() *RegionCache {
return &RegionCache{
lock: sync.Mutex{},
cache: make(map[string]*Region),
}
}
func (rc *RegionCache) Region(str string) *Region {
rc.lock.Lock()
defer rc.lock.Unlock()
region, ok := rc.cache[str]
if ok {
return region
}
// cache the new region
region = &Region{
Str: str,
fields: nil,
}
rc.cache[str] = region
return region
}
func (rc *RegionCache) Swap(r *Region) *Region {
return rc.Region(r.Str)
}
func (rc *RegionCache) Clean() {
rc.lock.Lock()
defer rc.lock.Unlock()
for k := range rc.cache {
delete(rc.cache, k)
}
}

View File

@ -12,10 +12,10 @@ import (
type Segment struct {
StartIP []byte
EndIP []byte
Region string
Region *Region
}
func SegmentFrom(seg string) (*Segment, error) {
func SegmentFrom(seg string, cRegion func(string) *Region) (*Segment, error) {
var ps = strings.SplitN(strings.TrimSpace(seg), "|", 3)
if len(ps) != 3 {
return nil, fmt.Errorf("invalid ip segment `%s`", seg)
@ -38,7 +38,7 @@ func SegmentFrom(seg string) (*Segment, error) {
return &Segment{
StartIP: sip,
EndIP: eip,
Region: ps[2],
Region: cRegion(ps[2]),
}, nil
}

View File

@ -140,7 +140,7 @@ func IPMiddle(sip, eip []byte) ([]byte, error) {
return IPHalf(buf), nil
}
func IterateSegments(handle *os.File, autoMerge bool, before func(l string), filter func(region string) (string, error), done func(seg *Segment) error) (int, int, error) {
func IterateSegments(handle *os.File, autoMerge bool, before func(l string), filter func(region string) (string, error), cRegion func(string) *Region, done func(seg *Segment) error) (int, int, error) {
var last *Segment = nil
var totalCount, mergeCount = 0, 0
var scanner = bufio.NewScanner(handle)
@ -193,14 +193,14 @@ func IterateSegments(handle *os.File, autoMerge bool, before func(l string), fil
if filter != nil {
region, err = filter(ps[2])
if err != nil {
return totalCount, mergeCount, fmt.Errorf("failed to filter region `%s`: %s", ps[2], err)
return totalCount, mergeCount, fmt.Errorf("failed to filter region `%s`: %s", region, err)
}
}
var seg = &Segment{
StartIP: sip,
EndIP: eip,
Region: region,
Region: cRegion(region),
}
// check and automatic merging the Consecutive Segments, which means:
@ -209,7 +209,7 @@ func IterateSegments(handle *os.File, autoMerge bool, before func(l string), fil
if last == nil {
last = seg
continue
} else if autoMerge && last.Region == seg.Region {
} else if autoMerge && last.Region.Equal(seg.Region) {
if err = seg.RightBehind(last); err == nil {
mergeCount++
last.EndIP = seg.EndIP
@ -266,7 +266,7 @@ func MergeSegments(segList []*Segment) []*Segment {
if last == nil {
last = seg
continue
} else if last.Region == seg.Region {
} else if last.Region.Equal(seg.Region) {
if err = seg.RightBehind(last); err == nil {
last.EndIP = seg.EndIP
continue

View File

@ -201,7 +201,7 @@ func TestSplitSegmentV4(t *testing.T) {
// var str = "0.0.0.0|1.255.225.254|0|0|0|内网IP|内网IP"
// var str = "29.0.0.0|29.34.191.255|美国|0|0|0|0"
var str = "28.201.224.0|29.34.191.255|美国|0|0|0|0"
seg, err := SegmentFrom(str)
seg, err := SegmentFrom(str, NewRegion)
if err != nil {
t.Fatalf("failed to parser segment '%s': %s", str, err)
}
@ -220,12 +220,12 @@ func TestSplitSegmentV4(t *testing.T) {
func TestRegionFiltering(t *testing.T) {
var line = "2001:1203:31:8000::|2001:1203:31:bfff:ffff:ffff:ffff:ffff||墨西哥|瓜纳华托州||||专线用户|"
seg, err := SegmentFrom(line)
seg, err := SegmentFrom(line, NewRegion)
if err != nil {
t.Fatalf("failed to parse segment '%s': %s", line, err)
}
fReg, err := RegionFiltering(seg.Region, []int{1, 2, 4, 6})
fReg, err := seg.Region.Filtering([]int{1, 2, 4, 6})
if err != nil {
t.Fatalf("failed to filter region '%s': %s", seg.Region, err)
}
@ -235,7 +235,7 @@ func TestRegionFiltering(t *testing.T) {
func TestSplitSegmentV6(t *testing.T) {
var str = "fec0::|ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff||瑞士|弗里堡州||||专线用户|IANA"
seg, err := SegmentFrom(str)
seg, err := SegmentFrom(str, NewRegion)
if err != nil {
t.Fatalf("failed to parser segment '%s': %s", str, err)
}
@ -260,7 +260,7 @@ func TestIterateSegments(t *testing.T) {
_, _, _ = IterateSegments(handle, true, func(l string) {
// fmt.Printf("load segment: `%s`\n", l)
}, nil, func(seg *Segment) error {
}, nil, NewRegion, func(seg *Segment) error {
fmt.Printf("get segment: `%s`\n", seg)
return nil
})