| 1 |
package storage |
| 2 |
|
| 3 |
import ( |
| 4 |
"errors" |
| 5 |
"fmt" |
| 6 |
"io" |
| 7 |
"os" |
| 8 |
"path/filepath" |
| 9 |
|
| 10 |
"larc.wejust.rest/larc/internal/hash" |
| 11 |
) |
| 12 |
|
| 13 |
/* BlobStore manages content-addressable blob storage. |
| 14 |
* Files are stored in objects/ directory with 2-char prefix subdirs. |
| 15 |
* All blobs are zstd compressed. |
| 16 |
* |
| 17 |
* Structure: objects/ab/cd1234567890ef (where abcd1234567890ef is the hash) */ |
| 18 |
|
| 19 |
var ( |
| 20 |
ErrBlobNotFound = errors.New("blob not found") |
| 21 |
ErrBlobExists = errors.New("blob already exists") |
| 22 |
) |
| 23 |
|
| 24 |
// BlobStore provides content-addressable storage for file contents |
| 25 |
type BlobStore struct { |
| 26 |
root string // path to objects/ directory |
| 27 |
} |
| 28 |
|
| 29 |
// NewBlobStore creates a new blob store at the given root directory |
| 30 |
func NewBlobStore(root string) (*BlobStore, error) { |
| 31 |
if err := os.MkdirAll(root, 0755); err != nil { |
| 32 |
return nil, fmt.Errorf("create blob store dir: %w", err) |
| 33 |
} |
| 34 |
return &BlobStore{root: root}, nil |
| 35 |
} |
| 36 |
|
| 37 |
// blobPath returns the full path for a blob hash |
| 38 |
func (s *BlobStore) blobPath(h string) string { |
| 39 |
if len(h) < 2 { |
| 40 |
return filepath.Join(s.root, "00", h) |
| 41 |
} |
| 42 |
return filepath.Join(s.root, h[:2], h[2:]) |
| 43 |
} |
| 44 |
|
| 45 |
// Exists checks if a blob exists |
| 46 |
func (s *BlobStore) Exists(h string) bool { |
| 47 |
_, err := os.Stat(s.blobPath(h)) |
| 48 |
return err == nil |
| 49 |
} |
| 50 |
|
| 51 |
// Write stores data and returns its hash |
| 52 |
func (s *BlobStore) Write(data []byte) (string, error) { |
| 53 |
h := hash.Bytes(data) |
| 54 |
|
| 55 |
/* fast path: already exists */ |
| 56 |
if s.Exists(h) { |
| 57 |
return h, nil |
| 58 |
} |
| 59 |
|
| 60 |
/* compress data */ |
| 61 |
compressed, err := Compress(data) |
| 62 |
if err != nil { |
| 63 |
return "", fmt.Errorf("compress blob: %w", err) |
| 64 |
} |
| 65 |
|
| 66 |
/* ensure directory exists */ |
| 67 |
blobPath := s.blobPath(h) |
| 68 |
dir := filepath.Dir(blobPath) |
| 69 |
if err := os.MkdirAll(dir, 0755); err != nil { |
| 70 |
return "", fmt.Errorf("create blob subdir: %w", err) |
| 71 |
} |
| 72 |
|
| 73 |
/* write atomically via temp file */ |
| 74 |
tmpPath := blobPath + ".tmp" |
| 75 |
if err := os.WriteFile(tmpPath, compressed, 0644); err != nil { |
| 76 |
return "", fmt.Errorf("write blob tmp: %w", err) |
| 77 |
} |
| 78 |
|
| 79 |
if err := os.Rename(tmpPath, blobPath); err != nil { |
| 80 |
os.Remove(tmpPath) |
| 81 |
return "", fmt.Errorf("rename blob: %w", err) |
| 82 |
} |
| 83 |
|
| 84 |
return h, nil |
| 85 |
} |
| 86 |
|
| 87 |
// WriteFile stores a file and returns its hash |
| 88 |
func (s *BlobStore) WriteFile(path string) (string, int64, error) { |
| 89 |
data, err := os.ReadFile(path) |
| 90 |
if err != nil { |
| 91 |
return "", 0, fmt.Errorf("read file for blob: %w", err) |
| 92 |
} |
| 93 |
|
| 94 |
h, err := s.Write(data) |
| 95 |
if err != nil { |
| 96 |
return "", 0, err |
| 97 |
} |
| 98 |
|
| 99 |
return h, int64(len(data)), nil |
| 100 |
} |
| 101 |
|
| 102 |
// Read retrieves blob content by hash |
| 103 |
func (s *BlobStore) Read(h string) ([]byte, error) { |
| 104 |
compressed, err := os.ReadFile(s.blobPath(h)) |
| 105 |
if err != nil { |
| 106 |
if os.IsNotExist(err) { |
| 107 |
return nil, ErrBlobNotFound |
| 108 |
} |
| 109 |
return nil, fmt.Errorf("read blob: %w", err) |
| 110 |
} |
| 111 |
|
| 112 |
data, err := Decompress(compressed) |
| 113 |
if err != nil { |
| 114 |
return nil, fmt.Errorf("decompress blob: %w", err) |
| 115 |
} |
| 116 |
|
| 117 |
return data, nil |
| 118 |
} |
| 119 |
|
| 120 |
// ReadTo reads blob content to writer |
| 121 |
func (s *BlobStore) ReadTo(h string, w io.Writer) error { |
| 122 |
data, err := s.Read(h) |
| 123 |
if err != nil { |
| 124 |
return err |
| 125 |
} |
| 126 |
|
| 127 |
if _, err := w.Write(data); err != nil { |
| 128 |
return fmt.Errorf("write blob to dest: %w", err) |
| 129 |
} |
| 130 |
|
| 131 |
return nil |
| 132 |
} |
| 133 |
|
| 134 |
// Size returns the compressed size of a blob |
| 135 |
func (s *BlobStore) Size(h string) (int64, error) { |
| 136 |
info, err := os.Stat(s.blobPath(h)) |
| 137 |
if err != nil { |
| 138 |
if os.IsNotExist(err) { |
| 139 |
return 0, ErrBlobNotFound |
| 140 |
} |
| 141 |
return 0, fmt.Errorf("stat blob: %w", err) |
| 142 |
} |
| 143 |
return info.Size(), nil |
| 144 |
} |
| 145 |
|
| 146 |
// Delete removes a blob (use with caution!) |
| 147 |
func (s *BlobStore) Delete(h string) error { |
| 148 |
if err := os.Remove(s.blobPath(h)); err != nil { |
| 149 |
if os.IsNotExist(err) { |
| 150 |
return ErrBlobNotFound |
| 151 |
} |
| 152 |
return fmt.Errorf("delete blob: %w", err) |
| 153 |
} |
| 154 |
return nil |
| 155 |
} |
| 156 |
|
| 157 |
// Walk iterates over all blobs in the store |
| 158 |
func (s *BlobStore) Walk(fn func(hash string) error) error { |
| 159 |
entries, err := os.ReadDir(s.root) |
| 160 |
if err != nil { |
| 161 |
return fmt.Errorf("read blob root: %w", err) |
| 162 |
} |
| 163 |
|
| 164 |
for _, entry := range entries { |
| 165 |
if !entry.IsDir() || len(entry.Name()) != 2 { |
| 166 |
continue |
| 167 |
} |
| 168 |
|
| 169 |
prefix := entry.Name() |
| 170 |
subdir := filepath.Join(s.root, prefix) |
| 171 |
|
| 172 |
subEntries, err := os.ReadDir(subdir) |
| 173 |
if err != nil { |
| 174 |
continue |
| 175 |
} |
| 176 |
|
| 177 |
for _, subEntry := range subEntries { |
| 178 |
if subEntry.IsDir() { |
| 179 |
continue |
| 180 |
} |
| 181 |
hash := prefix + subEntry.Name() |
| 182 |
if err := fn(hash); err != nil { |
| 183 |
return err |
| 184 |
} |
| 185 |
} |
| 186 |
} |
| 187 |
|
| 188 |
return nil |
| 189 |
} |
| 190 |
|
| 191 |
// Count returns the total number of blobs |
| 192 |
func (s *BlobStore) Count() (int, error) { |
| 193 |
count := 0 |
| 194 |
err := s.Walk(func(_ string) error { |
| 195 |
count++ |
| 196 |
return nil |
| 197 |
}) |
| 198 |
return count, err |
| 199 |
} |
| 200 |
|