X-Git-Url: http://www.git.stargrave.org/?p=glocate.git;a=blobdiff_plain;f=diff.go;fp=diff.go;h=61ba1ac9f83e9f80cb3ac2a7b81e88079e1e9a2d3a8e8afabd34f0485b739d9f;hp=0000000000000000000000000000000000000000000000000000000000000000;hb=d5b8c235a1f3088c6c1e7261df3d1b565d042db2ba2ad1bbd1018782b9178e1f;hpb=411a031ec7cc707b8269acc3dfe28bc8db1bab5a9a91781c26809ae9853c6f6a

diff --git a/diff.go b/diff.go
new file mode 100644
index 0000000..61ba1ac
--- /dev/null
+++ b/diff.go
@@ -0,0 +1,254 @@
+package main
+
+import (
+	"bufio"
+	"io"
+	"log"
+	"os"
+	"sort"
+	"strings"
+)
+
+type Ren struct {
+	src []string
+	dst []string
+}
+
+type BySrc []Ren
+
+func (a BySrc) Len() int {
+	return len(a)
+}
+
+func (a BySrc) Swap(i, j int) {
+	a[i], a[j] = a[j], a[i]
+}
+
+func (a BySrc) Less(i, j int) bool {
+	return namesCmp(a[i].src, a[j].src) < 0
+}
+
+type EntByName []*Ent
+
+func (a EntByName) Len() int {
+	return len(a)
+}
+
+func (a EntByName) Swap(i, j int) {
+	a[i], a[j] = a[j], a[i]
+}
+
+func (a EntByName) Less(i, j int) bool {
+	return namesCmp(a[i].name, a[j].name) < 0
+}
+
+func updateWithDiff(dbPath, strip string) *os.File {
+	scanner := bufio.NewScanner(os.Stdin)
+	var t string
+	var delsNames []string
+	var addsNames []string
+	var modsNames []string
+	var rens []Ren
+	var isDir bool
+	for scanner.Scan() {
+		t = scanner.Text()
+		if len(t) == 0 {
+			continue
+		}
+		cols := strings.Split(t, "\t")
+		if len(cols) < 3 {
+			log.Fatalln("bad zfs-diff format")
+		}
+		isDir = cols[1] == "/"
+		name := deoctalize(strings.TrimPrefix(cols[2], strip))
+		if name == "" {
+			continue
+		}
+		name = "./" + name
+		if isDir {
+			name += "/"
+		}
+		switch cols[0] {
+		case "-":
+			delsNames = append(delsNames, name)
+		case "+":
+			addsNames = append(addsNames, name)
+		case "M":
+			modsNames = append(modsNames, name)
+		case "R":
+			if len(cols) != 4 {
+				log.Fatalln("bad zfs-diff format for R")
+			}
+			dst := "./" + deoctalize(strings.TrimPrefix(cols[3], strip))
+			if isDir {
+				dst += "/"
+				rens = append(rens, Ren{
+					src: nameSplit(name),
+					dst: nameSplit(dst),
+				})
+			} else {
+				delsNames = append(delsNames, name)
+				addsNames = append(addsNames, dst)
+			}
+		default:
+			log.Fatalln("bad zfs-diff format")
+		}
+	}
+
+	entsReader := make(chan Ent, 1<<10)
+	db, err := os.Open(dbPath)
+	if err != nil {
+		log.Fatalln(err)
+	}
+	dels := make([][]string, 0, len(delsNames)+len(rens))
+	adds := make([][]string, 0, len(addsNames)+len(rens))
+	mods := make([]*Ent, 0, len(modsNames)+len(rens))
+	if len(rens) > 0 {
+		sort.Sort(BySrc(rens))
+		go reader(bufio.NewReaderSize(db, 1<<17), entsReader)
+		var ent Ent
+		var ok, met bool
+		for {
+			ent, ok = <-entsReader
+			if !ok {
+				break
+			}
+		Retry:
+			if len(rens) > 0 {
+				if hasPrefix(ent.name, rens[0].src) {
+					dels = append(dels, ent.name)
+					dst := append(
+						append([]string{}, rens[0].dst...),
+						ent.name[len(rens[0].src):]...,
+					)
+					adds = append(adds, dst)
+					mods = append(mods, &Ent{name: dst})
+					if !met {
+						// strip "/" from prefix directory
+						dst := rens[0].dst
+						last := dst[len(dst)-1]
+						dst[len(dst)-1] = last[:len(last)-1]
+						met = true
+					}
+				} else if met {
+					met = false
+					rens = rens[1:]
+					goto Retry
+				}
+			}
+		}
+		rens = nil
+	}
+
+	for _, name := range delsNames {
+		dels = append(dels, nameSplit(name))
+	}
+	delsNames = nil
+	sort.Sort(ByName(dels))
+
+	for _, name := range addsNames {
+		adds = append(adds, nameSplit(name))
+		modsNames = append(modsNames, name)
+	}
+	addsNames = nil
+	sort.Sort(ByName(adds))
+
+	for _, name := range modsNames {
+		mods = append(mods, &Ent{name: nameSplit(name)})
+	}
+	modsNames = nil
+	sort.Sort(EntByName(mods))
+	var info os.FileInfo
+	for _, ent := range mods {
+		info, err = os.Stat(nameJoin(ent.name))
+		if err != nil {
+			log.Println("can not stat:", nameJoin(ent.name), ":", err)
+			continue
+		}
+		if info.Mode().IsRegular() {
+			ent.size = info.Size()
+		}
+		ent.mtime = info.ModTime().Unix()
+	}
+
+	_, err = db.Seek(0, io.SeekStart)
+	if err != nil {
+		log.Fatalln(err)
+	}
+	tmp0, err := os.CreateTemp("", "glocate-idx")
+	if err != nil {
+		log.Fatalln(err)
+	}
+	defer os.Remove(tmp0.Name())
+	entsReader = make(chan Ent, 1<<10)
+	entsDirSizer := make(chan Ent, 1<<10)
+	entsWriter := make(chan Ent, 1<<10)
+	go reader(bufio.NewReaderSize(db, 1<<17), entsReader)
+
+	dirSizerJob := make(chan struct{})
+	var dirSizes []int64
+	sinkBack := make(chan Ent, 1)
+	go func() {
+		dirSizer(&dirSizes, 1, sinkBack, entsDirSizer, entsWriter)
+		close(dirSizerJob)
+	}()
+
+	bw := bufio.NewWriterSize(tmp0, 1<<17)
+	writerJob := make(chan struct{})
+	go func() {
+		writer(bw, entsWriter)
+		close(writerJob)
+	}()
+
+	for ent := range entsReader {
+		if len(dels) > 0 && namesCmp(ent.name, dels[0]) == 0 {
+			dels = dels[1:]
+			continue
+		}
+		for len(adds) > 0 && namesCmp(adds[0], ent.name) < 0 {
+			if namesCmp(mods[0].name, adds[0]) != 0 {
+				panic("+ and M lists are out of sync")
+			}
+			newEnt := Ent{
+				name:  adds[0],
+				mtime: mods[0].mtime,
+				size:  mods[0].size,
+			}
+			entsDirSizer <- newEnt
+			adds = adds[1:]
+			mods = mods[1:]
+		}
+		if len(mods) > 0 && namesCmp(ent.name, mods[0].name) == 0 {
+			ent.mtime = mods[0].mtime
+			ent.size = mods[0].size
+			mods = mods[1:]
+		}
+		entsDirSizer <- ent
+	}
+	for len(adds) > 0 {
+		if namesCmp(mods[0].name, adds[0]) != 0 {
+			panic("+ and M lists are out of sync")
+		}
+		newEnt := Ent{
+			name:  adds[0],
+			mtime: mods[0].mtime,
+			size:  mods[0].size,
+		}
+		entsDirSizer <- newEnt
+		adds = adds[1:]
+		mods = mods[1:]
+	}
+
+	close(entsDirSizer)
+	<-dirSizerJob
+	close(entsWriter)
+	<-writerJob
+	if err = bw.Flush(); err != nil {
+		log.Fatalln(err)
+	}
+
+	tmp1 := applyDirSizes(tmp0, dirSizes)
+	tmp0.Close()
+	os.Remove(tmp0.Name())
+	return tmp1
+}