2
2
mirror of https://github.com/octoleo/restic.git synced 2024-11-23 05:12:10 +00:00
restic/repository/index.go
Alexander Neumann 5cdcc99eba Use array instead of hash for backend.ID
Since backend.ID is always a slice of constant length, use an array
instead of a slice. Mostly, arrays behave as slices, except that an
array cannot be nil, so use `*backend.ID` insteaf of `backend.ID` in
places where the absence of an ID is possible (e.g. for the Subtree of a
Node, which may not present when the node is a file node).

This change allows to directly use backend.ID as the the key for a map,
so that arbitrary data structures (e.g. a Set implemented as a
map[backend.ID]struct{}) can easily be formed.
2015-07-25 18:01:57 +02:00

325 lines
7.4 KiB
Go

package repository
import (
"encoding/json"
"fmt"
"io"
"sync"
"github.com/restic/restic/backend"
"github.com/restic/restic/crypto"
"github.com/restic/restic/debug"
"github.com/restic/restic/pack"
)
// Index holds a lookup table for id -> pack.
type Index struct {
m sync.Mutex
pack map[string]indexEntry
}
type indexEntry struct {
tpe pack.BlobType
packID *backend.ID
offset uint
length uint
old bool
}
// NewIndex returns a new index.
func NewIndex() *Index {
return &Index{
pack: make(map[string]indexEntry),
}
}
func (idx *Index) store(t pack.BlobType, id backend.ID, pack *backend.ID, offset, length uint, old bool) {
idx.pack[id.String()] = indexEntry{
tpe: t,
packID: pack,
offset: offset,
length: length,
old: old,
}
}
// Store remembers the id and pack in the index.
func (idx *Index) Store(t pack.BlobType, id backend.ID, pack *backend.ID, offset, length uint) {
idx.m.Lock()
defer idx.m.Unlock()
debug.Log("Index.Store", "pack %v contains id %v (%v), offset %v, length %v",
pack.Str(), id.Str(), t, offset, length)
idx.store(t, id, pack, offset, length, false)
}
// Remove removes the pack ID from the index.
func (idx *Index) Remove(packID backend.ID) {
idx.m.Lock()
defer idx.m.Unlock()
debug.Log("Index.Remove", "id %v removed", packID.Str())
s := packID.String()
if _, ok := idx.pack[s]; ok {
delete(idx.pack, s)
}
}
// Lookup returns the pack for the id.
func (idx *Index) Lookup(id backend.ID) (packID *backend.ID, tpe pack.BlobType, offset, length uint, err error) {
idx.m.Lock()
defer idx.m.Unlock()
if p, ok := idx.pack[id.String()]; ok {
debug.Log("Index.Lookup", "id %v found in pack %v at %d, length %d",
id.Str(), p.packID.Str(), p.offset, p.length)
return p.packID, p.tpe, p.offset, p.length, nil
}
debug.Log("Index.Lookup", "id %v not found", id.Str())
return nil, pack.Data, 0, 0, fmt.Errorf("id %v not found in index", id)
}
// Has returns true iff the id is listed in the index.
func (idx *Index) Has(id backend.ID) bool {
_, _, _, _, err := idx.Lookup(id)
if err == nil {
return true
}
return false
}
// LookupSize returns the length of the cleartext content behind the
// given id
func (idx *Index) LookupSize(id backend.ID) (cleartextLength uint, err error) {
_, _, _, encryptedLength, err := idx.Lookup(id)
if err != nil {
return 0, err
}
return encryptedLength - crypto.Extension, nil
}
// Merge loads all items from other into idx.
func (idx *Index) Merge(other *Index) {
debug.Log("Index.Merge", "Merge index with %p", other)
idx.m.Lock()
defer idx.m.Unlock()
for k, v := range other.pack {
if _, ok := idx.pack[k]; ok {
debug.Log("Index.Merge", "index already has key %v, updating", k[:8])
}
idx.pack[k] = v
}
debug.Log("Index.Merge", "done merging index")
}
// PackedBlob is a blob already saved within a pack.
type PackedBlob struct {
pack.Blob
PackID backend.ID
}
// Each returns a channel that yields all blobs known to the index. If done is
// closed, the background goroutine terminates. This blocks any modification of
// the index.
func (idx *Index) Each(done chan struct{}) <-chan PackedBlob {
idx.m.Lock()
ch := make(chan PackedBlob)
go func() {
defer idx.m.Unlock()
defer func() {
close(ch)
}()
for ids, blob := range idx.pack {
id, err := backend.ParseID(ids)
if err != nil {
// ignore invalid IDs
continue
}
select {
case <-done:
return
case ch <- PackedBlob{
Blob: pack.Blob{
ID: id,
Offset: blob.offset,
Type: blob.tpe,
Length: uint32(blob.length),
},
PackID: *blob.packID,
}:
}
}
}()
return ch
}
// Count returns the number of blobs of type t in the index.
func (idx *Index) Count(t pack.BlobType) (n uint) {
debug.Log("Index.Count", "counting blobs of type %v", t)
idx.m.Lock()
defer idx.m.Unlock()
for id, blob := range idx.pack {
if blob.tpe == t {
n++
debug.Log("Index.Count", " blob %v counted: %v", id[:8], blob)
}
}
return
}
type packJSON struct {
ID string `json:"id"`
Blobs []blobJSON `json:"blobs"`
}
type blobJSON struct {
ID string `json:"id"`
Type pack.BlobType `json:"type"`
Offset uint `json:"offset"`
Length uint `json:"length"`
}
// generatePackList returns a list of packs containing only the index entries
// that selsectFn returned true for. If selectFn is nil, the list contains all
// blobs in the index.
func (idx *Index) generatePackList(selectFn func(indexEntry) bool) ([]*packJSON, error) {
list := []*packJSON{}
packs := make(map[string]*packJSON)
for id, blob := range idx.pack {
if selectFn != nil && !selectFn(blob) {
continue
}
debug.Log("Index.generatePackList", "handle blob %q", id[:8])
if blob.packID.IsNull() {
debug.Log("Index.generatePackList", "blob %q has no packID! (type %v, offset %v, length %v)",
id[:8], blob.tpe, blob.offset, blob.length)
return nil, fmt.Errorf("unable to serialize index: pack for blob %v hasn't been written yet", id)
}
// see if pack is already in map
p, ok := packs[blob.packID.String()]
if !ok {
// else create new pack
p = &packJSON{ID: blob.packID.String()}
// and append it to the list and map
list = append(list, p)
packs[p.ID] = p
}
// add blob
p.Blobs = append(p.Blobs, blobJSON{
ID: id,
Type: blob.tpe,
Offset: blob.offset,
Length: blob.length,
})
}
debug.Log("Index.generatePackList", "done")
return list, nil
}
// encode writes the JSON serialization of the index filtered by selectFn to enc.
func (idx *Index) encode(w io.Writer, selectFn func(indexEntry) bool) error {
list, err := idx.generatePackList(func(entry indexEntry) bool {
return !entry.old
})
if err != nil {
return err
}
debug.Log("Index.Encode", "done")
enc := json.NewEncoder(w)
return enc.Encode(list)
}
// Encode writes the JSON serialization of the index to the writer w. This
// serialization only contains new blobs added via idx.Store(), not old ones
// introduced via DecodeIndex().
func (idx *Index) Encode(w io.Writer) error {
debug.Log("Index.Encode", "encoding index")
idx.m.Lock()
defer idx.m.Unlock()
return idx.encode(w, func(e indexEntry) bool { return !e.old })
}
// Dump writes the pretty-printed JSON representation of the index to w.
func (idx *Index) Dump(w io.Writer) error {
debug.Log("Index.Dump", "dumping index")
idx.m.Lock()
defer idx.m.Unlock()
list, err := idx.generatePackList(nil)
if err != nil {
return err
}
buf, err := json.MarshalIndent(list, "", " ")
if err != nil {
return err
}
_, err = w.Write(append(buf, '\n'))
if err != nil {
return err
}
debug.Log("Index.Dump", "done")
return nil
}
// DecodeIndex loads and unserializes an index from rd.
func DecodeIndex(rd io.Reader) (*Index, error) {
debug.Log("Index.DecodeIndex", "Start decoding index")
list := []*packJSON{}
dec := json.NewDecoder(rd)
err := dec.Decode(&list)
if err != nil {
return nil, err
}
idx := NewIndex()
for _, pack := range list {
packID, err := backend.ParseID(pack.ID)
if err != nil {
debug.Log("Index.DecodeIndex", "error parsing pack ID %q: %v", pack.ID, err)
return nil, err
}
for _, blob := range pack.Blobs {
blobID, err := backend.ParseID(blob.ID)
if err != nil {
debug.Log("Index.DecodeIndex", "error parsing blob ID %q: %v", blob.ID, err)
return nil, err
}
idx.store(blob.Type, blobID, &packID, blob.Offset, blob.Length, true)
}
}
debug.Log("Index.DecodeIndex", "done")
return idx, err
}