func new_docsummary_internal(n *html.Node, f hash.Hash64) *DocumentSummary {
rtn := &DocumentSummary{}
if n == nil {
return rtn
}
foreach_child(n, func(child *html.Node) {
switch {
case child.Type == html.CommentNode:
case child.Type == html.DoctypeNode:
case child.Type == html.TextNode:
c, _ := f.Write([]byte(child.Data))
rtn.WordCount += c
if node_is_in_a(child) {
rtn.LinkWordCount += c
}
rtn.Text += child.Data
case child.Data == "img":
rtn.Images = append(rtn.Images, make_mediasummary(child))
case node_is_media(child):
rtn.Medias = append(rtn.Medias, make_mediasummary(child))
case child.Data == "a":
rtn.LinkCount++
ac := new_docsummary_internal(child, f)
rtn.Images = append(rtn.Images, ac.Images...)
rtn.Medias = append(rtn.Medias, ac.Medias...)
default:
sc := new_docsummary_internal(child, f)
rtn.add(sc)
}
})
return rtn
}
func TestRef(t *testing.T) {
for _, elem := range data {
var h32 hash.Hash32 = New32()
h32.Write([]byte(elem.s))
if v := h32.Sum32(); v != elem.h32 {
t.Errorf("'%s': 0x%x (want 0x%x)", elem.s, v, elem.h32)
}
if v := Sum32([]byte(elem.s)); v != elem.h32 {
t.Errorf("'%s': 0x%x (want 0x%x)", elem.s, v, elem.h32)
}
var h64 hash.Hash64 = New64()
h64.Write([]byte(elem.s))
if v := h64.Sum64(); v != elem.h64_1 {
t.Errorf("'%s': 0x%x (want 0x%x)", elem.s, v, elem.h64_1)
}
var h128 Hash128 = New128()
h128.Write([]byte(elem.s))
if v1, v2 := h128.Sum128(); v1 != elem.h64_1 || v2 != elem.h64_2 {
t.Errorf("'%s': 0x%x-0x%x (want 0x%x-0x%x)", elem.s, v1, v2, elem.h64_1, elem.h64_2)
}
if v1, v2 := Sum128([]byte(elem.s)); v1 != elem.h64_1 || v2 != elem.h64_2 {
t.Errorf("'%s': 0x%x-0x%x (want 0x%x-0x%x)", elem.s, v1, v2, elem.h64_1, elem.h64_2)
}
}
}
func makeHashes(d []byte, h hash.Hash64) ([]uint32, error) {
_, err := h.Write(d)
if err != nil {
return nil, err
}
hashed := h.Sum64()
lower := uint32(hashed)
upper := uint32(hashed >> 32)
return []uint32{lower, upper}, nil
}
// ID with which you can identify a daemon connection to the same SMTP server
// independent of the scope ID.
func (dm *Daemon) ID() uint64 {
var h hash.Hash64
h = fnv.New64()
data := []byte(dm.getHost() + strconv.Itoa(dm.getPort()) + dm.getUsername())
if _, err := h.Write(data); err != nil {
log.Error("mail.daemon.ID", "err", err, "hashWrite", string(data))
return 0
}
return h.Sum64()
}
func TestRef(t *testing.T) {
for _, elem := range data {
var h32 hash.Hash32 = New32()
h32.Write([]byte(elem.s))
if v := h32.Sum32(); v != elem.h32 {
t.Errorf("'%s': 0x%x (want 0x%x)", elem.s, v, elem.h32)
}
var h32_byte hash.Hash32 = New32()
h32_byte.Write([]byte(elem.s))
target := fmt.Sprintf("%08x", elem.h32)
if p := fmt.Sprintf("%x", h32_byte.Sum(nil)); p != target {
t.Errorf("'%s': %s (want %s)", elem.s, p, target)
}
if v := Sum32([]byte(elem.s)); v != elem.h32 {
t.Errorf("'%s': 0x%x (want 0x%x)", elem.s, v, elem.h32)
}
var h64 hash.Hash64 = New64()
h64.Write([]byte(elem.s))
if v := h64.Sum64(); v != elem.h64_1 {
t.Errorf("'%s': 0x%x (want 0x%x)", elem.s, v, elem.h64_1)
}
var h64_byte hash.Hash64 = New64()
h64_byte.Write([]byte(elem.s))
target = fmt.Sprintf("%016x", elem.h64_1)
if p := fmt.Sprintf("%x", h64_byte.Sum(nil)); p != target {
t.Errorf("Sum64: '%s': %s (want %s)", elem.s, p, target)
}
if v := Sum64([]byte(elem.s)); v != elem.h64_1 {
t.Errorf("Sum64: '%s': 0x%x (want 0x%x)", elem.s, v, elem.h64_1)
}
var h128 Hash128 = New128()
h128.Write([]byte(elem.s))
if v1, v2 := h128.Sum128(); v1 != elem.h64_1 || v2 != elem.h64_2 {
t.Errorf("New128: '%s': 0x%x-0x%x (want 0x%x-0x%x)", elem.s, v1, v2, elem.h64_1, elem.h64_2)
}
var h128_byte Hash128 = New128()
h128_byte.Write([]byte(elem.s))
target = fmt.Sprintf("%016x%016x", elem.h64_1, elem.h64_2)
if p := fmt.Sprintf("%x", h128_byte.Sum(nil)); p != target {
t.Errorf("New128: '%s': %s (want %s)", elem.s, p, target)
}
if v1, v2 := Sum128([]byte(elem.s)); v1 != elem.h64_1 || v2 != elem.h64_2 {
t.Errorf("Sum128: '%s': 0x%x-0x%x (want 0x%x-0x%x)", elem.s, v1, v2, elem.h64_1, elem.h64_2)
}
}
}
func benchmarkHash(b *testing.B, h hash.Hash64) {
uids := getUids(b.N)
var s uint64
b.ResetTimer()
for i := 0; i < b.N; i++ {
h.Reset()
io.WriteString(h, uids[i])
s = h.Sum64()
}
result = s
}
func hashUpdateOrdered(h hash.Hash64, a, b uint64) uint64 {
// For ordered updates, use a real hash function
h.Reset()
// We just panic if the binary writes fail because we are writing
// an int64 which should never be fail-able.
e1 := binary.Write(h, binary.LittleEndian, a)
e2 := binary.Write(h, binary.LittleEndian, b)
if e1 != nil {
panic(e1)
}
if e2 != nil {
panic(e2)
}
return h.Sum64()
}
// ID with which you can identify a daemon connection to the same SMTP server
// independent of the scope ID.
func (u *uniqueID) Get() (id uint64, hasChanged bool) {
var h hash.Hash64
h = fnv.New64()
data := []byte(u.getHost() + strconv.Itoa(u.getPort()) + u.getUsername())
if _, err := h.Write(data); err != nil {
log.Error("mail.daemon.ID", "err", err, "hashWrite", string(data))
return
}
if u.lastID != h.Sum64() {
u.lastID = h.Sum64()
return u.lastID, true // ID has changed, means some one updated the configuration.
}
return h.Sum64(), false // has not changed
}
func testCollissions(t *testing.T, h hash.Hash64) {
uids := getUids(uidSize)
results := make(map[uint64]bool)
cols := 0
for i := 0; i < uidSize; i++ {
h.Reset()
io.WriteString(h, uids[i])
s := h.Sum64()
if _, col := results[s]; col {
cols += 1
} else {
results[s] = true
}
}
if cols > 0 {
t.Errorf("Found %v collissions for uidSize %v\n", cols, uidSize)
}
}
func writeChecksumBlock(hash hash.Hash64, output io.Writer) error {
// file path length... zero
err := binary.Write(output, binary.BigEndian, uint16(0))
if err == nil {
blockType := []byte{byte(blockTypeChecksum)}
_, err = output.Write(blockType)
}
if err == nil {
err = binary.Write(output, binary.BigEndian, hash.Sum64())
}
return err
}
// Hash64 is a convenience method for hashing a string against a hash.Hash64
func Hash64(s string, h hash.Hash64) uint64 {
h.Reset()
h.Write([]byte(s))
return h.Sum64()
}
func (ds *Directory) Get(key string, opts store.GetOptions) ([]byte, store.Stat, error) {
ds.mu.RLock()
defer ds.mu.RUnlock()
select {
case <-opts.Cancel:
return nil, store.Stat{}, store.ErrCancelled
default:
}
fh, path, err := ds.findAndOpen(key)
if err != nil {
return nil, store.Stat{}, err
}
if fh == nil {
return nil, store.Stat{}, store.ErrNotFound
}
defer fh.Close()
var expectedFNV [8]byte
_, err = io.ReadFull(fh, expectedFNV[:])
if err != nil {
return nil, store.Stat{}, err
}
var rdr io.Reader
var fnver hash.Hash64
if opts.NoVerify {
rdr = fh
} else {
fnver = fnv.New64a()
rdr = io.TeeReader(fh, fnver)
}
var expectedSHA256 [32]byte
_, err = io.ReadFull(rdr, expectedSHA256[:])
if err != nil {
return nil, store.Stat{}, err
}
fi, err := fh.Stat()
if err != nil {
return nil, store.Stat{}, err
}
size := fi.Size() - 40
if int64(int(size)) != size {
return nil, store.Stat{}, errors.New("file is too big")
}
writeTime := fi.ModTime().Unix()
data := make([]byte, int(size))
at := 0
for at < len(data) {
readInto := data[at:]
if len(readInto) > 1024*1024 {
readInto = readInto[:1024*1024]
}
n, err := rdr.Read(readInto)
at += n
if err != nil {
if err == io.EOF {
break
}
return nil, store.Stat{}, err
}
select {
case <-opts.Cancel:
return nil, store.Stat{}, store.ErrCancelled
default:
}
}
if at != len(data) {
return nil, store.Stat{}, errors.New("file was shortened during read")
}
if !opts.NoVerify {
actualFNV := fnver.Sum(nil)
if !bytes.Equal(actualFNV, expectedFNV[:]) {
fh.Close()
// TODO: this relocking is f****d and racy
ds.mu.RUnlock()
ds.mu.Lock()
ds.quarantine(key, path)
ds.mu.Unlock()
ds.mu.RLock()
return nil, store.Stat{}, ErrCorruptObject
}
}
return data, store.Stat{
SHA256: expectedSHA256,
Size: size,
WriteTime: writeTime,
}, nil
}
// hashKernel returns the upper and lower base hash values from which the k
// hashes are derived.
func hashKernel(data []byte, hash hash.Hash64) (uint32, uint32) {
hash.Write(data)
sum := hash.Sum(nil)
hash.Reset()
return binary.BigEndian.Uint32(sum[4:8]), binary.BigEndian.Uint32(sum[0:4])
}