|
| 1 | +package cache |
| 2 | + |
| 3 | +// This file was initially taken from https://github.com/hauke96/tiny-http-proxy/blob/master/cache.go |
| 4 | +// some extra functionality and fixes have been added: |
| 5 | +// Allow invalidating cahce items |
| 6 | +// Expiring cached items based on a global TTL |
| 7 | + |
| 8 | +import ( |
| 9 | + "bufio" |
| 10 | + "bytes" |
| 11 | + "crypto/sha256" |
| 12 | + "encoding/hex" |
| 13 | + "fmt" |
| 14 | + "hash" |
| 15 | + "io" |
| 16 | + "io/fs" |
| 17 | + "os" |
| 18 | + "path/filepath" |
| 19 | + "sync" |
| 20 | + "time" |
| 21 | + |
| 22 | + log "github.com/sirupsen/logrus" |
| 23 | +) |
| 24 | + |
| 25 | +type Cache struct { |
| 26 | + folder string |
| 27 | + hash hash.Hash |
| 28 | + knownValues map[string][]byte |
| 29 | + busyValues map[string]*sync.Mutex |
| 30 | + mutex *sync.Mutex |
| 31 | + maxSize int64 |
| 32 | + ttl time.Duration |
| 33 | +} |
| 34 | + |
| 35 | +type Options struct { |
| 36 | + Path string |
| 37 | + MaxSize int64 |
| 38 | + TTL time.Duration |
| 39 | +} |
| 40 | + |
| 41 | +func CreateCache(opts Options) (*Cache, error) { |
| 42 | + fileInfos, err := os.ReadDir(opts.Path) |
| 43 | + if err != nil { |
| 44 | + log.Warnf("Cannot open cache folder '%s': %s", opts.Path, err) |
| 45 | + log.Infof("Create cache folder '%s'", opts.Path) |
| 46 | + if err := os.Mkdir(opts.Path, os.ModePerm); err != nil { |
| 47 | + return nil, err |
| 48 | + } |
| 49 | + } |
| 50 | + |
| 51 | + values := make(map[string][]byte, 0) |
| 52 | + busy := make(map[string]*sync.Mutex, 0) |
| 53 | + |
| 54 | + // Go through every file an save its name in the map. The content of the file |
| 55 | + // is loaded when needed. This makes sure that we don't have to read |
| 56 | + // the directory content each time the user wants data that's not yet loaded. |
| 57 | + for _, info := range fileInfos { |
| 58 | + if !info.IsDir() { |
| 59 | + values[info.Name()] = nil |
| 60 | + } |
| 61 | + } |
| 62 | + |
| 63 | + hash := sha256.New() |
| 64 | + |
| 65 | + mutex := &sync.Mutex{} |
| 66 | + |
| 67 | + c := &Cache{ |
| 68 | + folder: opts.Path, |
| 69 | + hash: hash, |
| 70 | + knownValues: values, |
| 71 | + busyValues: busy, |
| 72 | + mutex: mutex, |
| 73 | + maxSize: opts.MaxSize, |
| 74 | + ttl: opts.TTL, |
| 75 | + } |
| 76 | + |
| 77 | + go func() { |
| 78 | + ticker := time.NewTicker(c.ttl) |
| 79 | + defer ticker.Stop() |
| 80 | + for range ticker.C { |
| 81 | + files, err := c.findFilesOlderThanTTL() |
| 82 | + if err != nil { |
| 83 | + continue |
| 84 | + } |
| 85 | + for _, file := range files { |
| 86 | + err := c.deleteFromHash(file.Name()) |
| 87 | + if err != nil { |
| 88 | + continue |
| 89 | + } |
| 90 | + } |
| 91 | + } |
| 92 | + }() |
| 93 | + |
| 94 | + return c, nil |
| 95 | +} |
| 96 | + |
| 97 | +// Returns true if the resource is found, and false otherwise. If the |
| 98 | +// resource is busy, this method will hang until the resource is free. If |
| 99 | +// the resource is not found, a lock indicating that the resource is busy will |
| 100 | +// be returned. Once the resource Has been put into cache the busy lock *must* |
| 101 | +// be unlocked to allow others to access the newly cached resource |
| 102 | +func (c *Cache) Has(key string) (*sync.Mutex, bool) { |
| 103 | + hashValue := calcHash(key) |
| 104 | + |
| 105 | + c.mutex.Lock() |
| 106 | + defer c.mutex.Unlock() |
| 107 | + |
| 108 | + // If the resource is busy, wait for it to be free. This is the case if |
| 109 | + // the resource is currently being cached as a result of another request. |
| 110 | + // Also, release the lock on the cache to allow other readers while waiting |
| 111 | + if lock, busy := c.busyValues[hashValue]; busy { |
| 112 | + c.mutex.Unlock() |
| 113 | + lock.Lock() |
| 114 | + // just waiting in case lock was previously acquired |
| 115 | + lock.Unlock() |
| 116 | + c.mutex.Lock() |
| 117 | + } |
| 118 | + |
| 119 | + // If a resource is in the shared cache, it can't be reserved. One can simply |
| 120 | + // access it directly from the cache |
| 121 | + if _, found := c.knownValues[hashValue]; found { |
| 122 | + return nil, true |
| 123 | + } |
| 124 | + |
| 125 | + // The resource is not in the cache, mark the resource as busy until it has |
| 126 | + // been cached successfully. Unlocking lock is required! |
| 127 | + lock := new(sync.Mutex) |
| 128 | + lock.Lock() |
| 129 | + c.busyValues[hashValue] = lock |
| 130 | + return lock, false |
| 131 | +} |
| 132 | + |
| 133 | +func (c *Cache) Get(key string) (*io.Reader, error) { |
| 134 | + var response io.Reader |
| 135 | + hashValue := calcHash(key) |
| 136 | + |
| 137 | + // Try to get content. Error if not found. |
| 138 | + c.mutex.Lock() |
| 139 | + content, ok := c.knownValues[hashValue] |
| 140 | + c.mutex.Unlock() |
| 141 | + if !ok && len(content) > 0 { |
| 142 | + log.Debugf("Cache doesn't know key '%s'", hashValue) |
| 143 | + return nil, fmt.Errorf("key '%s' is not known to cache", hashValue) |
| 144 | + } |
| 145 | + |
| 146 | + log.Debugf("Cache has key '%s'", hashValue) |
| 147 | + |
| 148 | + // Key is known, but not loaded into RAM |
| 149 | + if content == nil { |
| 150 | + log.Debugf("Cache item '%s' known but is not stored in memory. Using file.", hashValue) |
| 151 | + |
| 152 | + file, err := os.Open(filepath.Join(c.folder, hashValue)) |
| 153 | + if err != nil { |
| 154 | + log.Errorf("Error reading cached file '%s': %s", hashValue, err) |
| 155 | + // forget the cached item |
| 156 | + _ = c.deleteFromHash(hashValue) |
| 157 | + return nil, err |
| 158 | + } |
| 159 | + |
| 160 | + response = file |
| 161 | + |
| 162 | + log.Debugf("Create reader from file %s", hashValue) |
| 163 | + } else { // Key is known and data is already loaded to RAM |
| 164 | + response = bytes.NewReader(content) |
| 165 | + log.Debugf("Create reader from %d byte large cache content", len(content)) |
| 166 | + } |
| 167 | + |
| 168 | + return &response, nil |
| 169 | +} |
| 170 | + |
| 171 | +func (c *Cache) Delete(key string) error { |
| 172 | + return c.deleteFromHash(calcHash(key)) |
| 173 | +} |
| 174 | + |
| 175 | +func (c *Cache) deleteFromHash(hashValue string) error { |
| 176 | + c.mutex.Lock() |
| 177 | + defer c.mutex.Unlock() |
| 178 | + |
| 179 | + // If the resource is busy, wait for it to be free. This is the case if |
| 180 | + // the resource is currently being cached as a result of another request. |
| 181 | + // Also, release the lock on the cache to allow other readers while waiting |
| 182 | + if lock, busy := c.busyValues[hashValue]; busy { |
| 183 | + c.mutex.Unlock() |
| 184 | + lock.Lock() |
| 185 | + // just waiting in case lock was previously acquired |
| 186 | + lock.Unlock() |
| 187 | + c.mutex.Lock() |
| 188 | + } |
| 189 | + |
| 190 | + delete(c.busyValues, hashValue) |
| 191 | + delete(c.knownValues, hashValue) |
| 192 | + |
| 193 | + return os.Remove(filepath.Join(c.folder, hashValue)) |
| 194 | +} |
| 195 | + |
| 196 | +func (c *Cache) findFilesOlderThanTTL() ([]fs.DirEntry, error) { |
| 197 | + var files []fs.DirEntry |
| 198 | + tmpfiles, err := os.ReadDir(c.folder) |
| 199 | + if err != nil { |
| 200 | + return files, err |
| 201 | + } |
| 202 | + |
| 203 | + for _, file := range tmpfiles { |
| 204 | + if file.Type().IsRegular() { |
| 205 | + info, err := file.Info() |
| 206 | + if err != nil { |
| 207 | + return files, err |
| 208 | + } |
| 209 | + if time.Since(info.ModTime()) > c.ttl { |
| 210 | + files = append(files, file) |
| 211 | + } |
| 212 | + } |
| 213 | + } |
| 214 | + return files, err |
| 215 | +} |
| 216 | + |
| 217 | +// release is an internal method which atomically caches an item and unmarks |
| 218 | +// the item as busy, if it was busy before. The busy lock *must* be unlocked |
| 219 | +// elsewhere! |
| 220 | +func (c *Cache) release(hashValue string, content []byte) { |
| 221 | + c.mutex.Lock() |
| 222 | + delete(c.busyValues, hashValue) |
| 223 | + c.knownValues[hashValue] = content |
| 224 | + c.mutex.Unlock() |
| 225 | +} |
| 226 | + |
| 227 | +func (c *Cache) Put(key string, content *io.Reader, contentLength int64) error { |
| 228 | + hashValue := calcHash(key) |
| 229 | + |
| 230 | + // Small enough to put it into the in-memory cache |
| 231 | + if contentLength <= c.maxSize*1024*1024 { |
| 232 | + buffer := &bytes.Buffer{} |
| 233 | + _, err := io.Copy(buffer, *content) |
| 234 | + if err != nil { |
| 235 | + return err |
| 236 | + } |
| 237 | + |
| 238 | + defer c.release(hashValue, buffer.Bytes()) |
| 239 | + log.Debugf("Added %s into in-memory cache", hashValue) |
| 240 | + |
| 241 | + err = os.WriteFile(filepath.Join(c.folder, hashValue), buffer.Bytes(), 0644) |
| 242 | + if err != nil { |
| 243 | + return err |
| 244 | + } |
| 245 | + log.Debugf("Wrote content of entry %s into file", hashValue) |
| 246 | + } else { // Too large for in-memory cache, just write to file |
| 247 | + defer c.release(hashValue, nil) |
| 248 | + log.Debugf("Added nil-entry for %s into in-memory cache", hashValue) |
| 249 | + |
| 250 | + file, err := os.Create(filepath.Join(c.folder, hashValue)) |
| 251 | + if err != nil { |
| 252 | + return err |
| 253 | + } |
| 254 | + |
| 255 | + writer := bufio.NewWriter(file) |
| 256 | + _, err = io.Copy(writer, *content) |
| 257 | + if err != nil { |
| 258 | + return err |
| 259 | + } |
| 260 | + log.Debugf("Wrote content of entry %s into file", hashValue) |
| 261 | + } |
| 262 | + |
| 263 | + log.Debugf("Cache wrote content into '%s'", hashValue) |
| 264 | + |
| 265 | + return nil |
| 266 | +} |
| 267 | + |
| 268 | +func calcHash(data string) string { |
| 269 | + sha := sha256.Sum256([]byte(data)) |
| 270 | + return hex.EncodeToString(sha[:]) |
| 271 | +} |
0 commit comments