func main() {
var kind string = "files"
var ndirs, nfiles int
var paths []string
flag.Var(&_fthreshold, "ft", "file sizes <= threshhold will not be considered")
flag.Var(&_dthreshold, "dt", "directory sizes <= threshhold will not be considered")
//fmt.Printf("dedup\n")
hf = aeshash.NewAES(0)
flag.Parse()
if *pat != "" {
re, err := regexp.Compile(*pat)
if err != nil {
return
}
patre = re
}
if *dd != "" {
re, err := regexp.Compile(*dd)
if err != nil {
return
}
ddre = re
}
fthreshold = int64(_fthreshold.V)
dthreshold = int64(_dthreshold.V)
//fmt.Printf("fthreshold=%d\n", fthreshold)
//fmt.Printf("dthreshold=%d\n", dthreshold)
if *dirf {
kind = "dirs"
}
if len(flag.Args()) != 0 {
for _, path := range flag.Args() {
fi, err := os.Stat(path)
if err != nil || fi == nil {
fmt.Printf("fi=%#v, err=%v\n", fi, err)
panic("bad")
}
if fi.Mode()&os.ModeDir == os.ModeDir {
ndirs++
} else {
nfiles++
}
paths = append(paths, path)
}
}
scan(paths, ndirs)
if *s {
asort()
check2(kind, ndirs)
} else {
check(kind, ndirs)
}
}
func descend(path string, fis []os.FileInfo,
ffp func(path string, fis os.FileInfo, hash uint64, size int64),
dfp func(path string, fis os.FileInfo, hash uint64, size int64)) (uint64, int64) {
var des func(path string, fis []os.FileInfo) (uint64, int64)
des = func(path string, fis []os.FileInfo) (uint64, int64) {
var hash uint64
var size, sizes int64
var gh = aeshash.NewAES(0)
outer:
for _, fi := range fis {
//fmt.Printf("descend: enter fi.Name=%q\n", fi.Name())
switch {
case fi.Mode()&os.ModeDir == os.ModeDir:
//fmt.Printf("descend: dir: path=%q, fi.Name()=%q\n", path, fi.Name())
stats.scannedDirs++
if *dd != "" {
b := ddre.MatchString(fi.Name())
if b {
//fmt.Printf("descend: skip dir=%q\n", fi.Name())
continue
}
}
for _, name := range ignoreList {
if fi.Name() == name {
fmt.Printf("descend: skip dir=%q\n", fi.Name())
continue outer
}
}
p := fullName(path, fi)
//fmt.Printf("descend: dir=%q\n", p)
d, err := os.Open(p)
if err != nil {
continue
}
fis, err := d.Readdir(-1)
if err != nil || fis == nil {
fmt.Printf("descend: can't read %q\n", fullName(path, fi))
continue
}
d.Close()
h, size := des(p, fis)
hash = h
gh.Write64(hash)
sizes += size
//fmt.Printf("descend: dir: path=%q, fi.Name()=%q, sizes=%d\n", path, fi.Name(), sizes)
stats.matchedDirs++
if dfp != nil {
//fmt.Printf("descend: dfp: path=%q, fi.Name()=%q, hash=0x%016x, size=%d\n", path, fi.Name(), hash, size)
dfp(path, fi, hash, size)
}
case fi.Mode()&os.ModeType == 0:
stats.scannedFiles++
if fi.Size() > fthreshold && (*pat == "" || (*pat != "" && patre.MatchString(fi.Name()))) {
if *fr {
hash = readFullHash(path, fi)
} else {
hash = readPartialHash(path, fi)
}
gh.Write64(hash)
sizes += fi.Size()
stats.matchedFiles++
//fmt.Printf("descend: file: path=%q, fi.Name()=%q, hash=0x%016x, size=%d, sizes=%d\n", path, fi.Name(), hash, size, sizes)
if ffp != nil {
ffp(path, fi, hash, size)
}
}
default:
continue
}
}
hashes := gh.Sum64()
//fmt.Printf("descend: return dir=%q, hashes=0x%016x, sizes=%d\n", path, hashes, sizes)
return hashes, sizes
}
//fmt.Printf("descend: path=%q\n", path)
return des(path, fis)
}