|
| 1 | +// Package dbfs implement the fs.FS over a sqlite3 database backend. |
| 2 | +package dbfs |
| 3 | + |
| 4 | +import ( |
| 5 | + "database/sql" |
| 6 | + "errors" |
| 7 | + "fmt" |
| 8 | + "io" |
| 9 | + "io/fs" |
| 10 | + "path" |
| 11 | + "strings" |
| 12 | + |
| 13 | + "github.com/hashicorp/go-multierror" |
| 14 | + "github.com/jmoiron/sqlx" |
| 15 | + _ "github.com/mattn/go-sqlite3" |
| 16 | +) |
| 17 | + |
| 18 | +type FS struct { |
| 19 | + db *sqlx.DB |
| 20 | + rootInode int |
| 21 | +} |
| 22 | + |
| 23 | +var ( |
| 24 | + RelativePathErr = fmt.Errorf("relative path are not supported") |
| 25 | + InodeNotFoundErr = fmt.Errorf("cannot find inode") |
| 26 | + IncorrectTypeErr = fmt.Errorf("incorrect file type") |
| 27 | +) |
| 28 | + |
| 29 | +const ( |
| 30 | + DirectoryType = "d" |
| 31 | + RegularFileType = "f" |
| 32 | +) |
| 33 | + |
| 34 | +func NewSqliteFS(dbName string) (*FS, error) { |
| 35 | + db, err := sqlx.Open("sqlite3", dbName) |
| 36 | + if err != nil { |
| 37 | + return nil, fmt.Errorf("canot open the database: %w", err) |
| 38 | + } |
| 39 | + err = runMigrations(db) |
| 40 | + if err != nil { |
| 41 | + return nil, err |
| 42 | + } |
| 43 | + if _, err := db.Exec("PRAGMA foreign_key = ON"); err != nil { |
| 44 | + return nil, fmt.Errorf("cannot activate foreign keys check: %w", err) |
| 45 | + } |
| 46 | + |
| 47 | + fs := &FS{db: db} |
| 48 | + row := db.QueryRow(` |
| 49 | + SELECT inode |
| 50 | + FROM github_dgsb_dbfs_files |
| 51 | + WHERE fname = '/' AND parent IS NULL`) |
| 52 | + if err := row.Scan(&fs.rootInode); err != nil { |
| 53 | + return nil, fmt.Errorf("no root inode: %w %w", InodeNotFoundErr, err) |
| 54 | + } |
| 55 | + |
| 56 | + return fs, nil |
| 57 | +} |
| 58 | + |
| 59 | +func (fs *FS) Close() error { |
| 60 | + return fs.db.Close() |
| 61 | +} |
| 62 | + |
| 63 | +func (fs *FS) addRegularFileNode(tx *sqlx.Tx, fname string) (int, error) { |
| 64 | + if !path.IsAbs(fname) { |
| 65 | + return 0, fmt.Errorf("%w: %s", RelativePathErr, fname) |
| 66 | + } |
| 67 | + |
| 68 | + components := strings.Split(fname, "/")[1:] |
| 69 | + var parentInode = fs.rootInode |
| 70 | + for i, searchMode := 0, true; i < len(components); i++ { |
| 71 | + if searchMode { |
| 72 | + var ( |
| 73 | + inode int |
| 74 | + ftype string |
| 75 | + ) |
| 76 | + row := tx.QueryRowx( |
| 77 | + "SELECT inode, type FROM github_dgsb_dbfs_files WHERE fname = ? AND parent = ?", |
| 78 | + components[i], parentInode) |
| 79 | + err := row.Scan(&inode, &ftype) |
| 80 | + if err == nil { |
| 81 | + parentInode = inode |
| 82 | + if (i < len(components)-1 && ftype != DirectoryType) || |
| 83 | + (i == len(components)-1 && ftype != RegularFileType) { |
| 84 | + return 0, fmt.Errorf( |
| 85 | + "%w: %s %s", IncorrectTypeErr, "/"+strings.Join(components[:i+1], "/"), ftype) |
| 86 | + } |
| 87 | + continue |
| 88 | + } |
| 89 | + if !errors.Is(err, sql.ErrNoRows) { |
| 90 | + return 0, fmt.Errorf("cannot query files table: %w", err) |
| 91 | + } |
| 92 | + searchMode = false |
| 93 | + } |
| 94 | + |
| 95 | + componentType := func() string { |
| 96 | + if i < len(components)-1 { |
| 97 | + return DirectoryType |
| 98 | + } |
| 99 | + return RegularFileType |
| 100 | + }() |
| 101 | + row := tx.QueryRow(` |
| 102 | + INSERT INTO github_dgsb_dbfs_files (fname, parent, type) |
| 103 | + VALUES (?, ?, ?) |
| 104 | + RETURNING inode`, components[i], parentInode, componentType) |
| 105 | + if err := row.Scan(&parentInode); err != nil { |
| 106 | + return 0, fmt.Errorf( |
| 107 | + "cannot insert node %s as child of %d: %w", components[i], parentInode, err) |
| 108 | + } |
| 109 | + } |
| 110 | + |
| 111 | + return parentInode, nil |
| 112 | +} |
| 113 | + |
| 114 | +func (fs *FS) UpsertFile(fname string, chunkSize int, data []byte) (ret error) { |
| 115 | + if !path.IsAbs(fname) { |
| 116 | + return fmt.Errorf("%w: %s", RelativePathErr, fname) |
| 117 | + } |
| 118 | + fname = path.Clean(fname) |
| 119 | + |
| 120 | + tx, err := fs.db.Beginx() |
| 121 | + if err != nil { |
| 122 | + return fmt.Errorf("cannot start transaction: %w", err) |
| 123 | + } |
| 124 | + defer func() { |
| 125 | + if err != nil { |
| 126 | + ret = multierror.Append(ret, tx.Rollback()) |
| 127 | + } else { |
| 128 | + ret = tx.Commit() |
| 129 | + } |
| 130 | + }() |
| 131 | + |
| 132 | + inode, err := fs.addRegularFileNode(tx, fname) |
| 133 | + if err != nil { |
| 134 | + return fmt.Errorf("cannot insert file node: %w", err) |
| 135 | + } |
| 136 | + |
| 137 | + if _, err := tx.Exec(`DELETE FROM github_dgsb_dbfs_chunks WHERE inode = ?`, inode); err != nil { |
| 138 | + return fmt.Errorf("cannot delete previous chunks of the same file %s: %w", fname, err) |
| 139 | + } |
| 140 | + |
| 141 | + for i, position := 0, 0; i < len(data); i, position = i+chunkSize, position+1 { |
| 142 | + toWrite := func() int { |
| 143 | + remaining := len(data) - i |
| 144 | + if remaining < chunkSize { |
| 145 | + return remaining |
| 146 | + } |
| 147 | + return chunkSize |
| 148 | + }() |
| 149 | + _, err := tx.Exec(` |
| 150 | + INSERT INTO github_dgsb_dbfs_chunks (inode, position, data, size) |
| 151 | + VALUES (?, ?, ?, ?)`, inode, position, data[i:i+toWrite], toWrite) |
| 152 | + if err != nil { |
| 153 | + return fmt.Errorf("cannot insert file chunk in database: %w", err) |
| 154 | + } |
| 155 | + } |
| 156 | + return nil |
| 157 | +} |
| 158 | + |
| 159 | +func (fs *FS) namei(fname string) (int, error) { |
| 160 | + if !path.IsAbs(fname) { |
| 161 | + return 0, fmt.Errorf("%w: %s", RelativePathErr, fname) |
| 162 | + } |
| 163 | + components := strings.Split(fname, "/")[1:] |
| 164 | + if len(components) == 0 { |
| 165 | + return fs.rootInode, nil |
| 166 | + } |
| 167 | + |
| 168 | + var inode int |
| 169 | + for i, parentInode := 0, fs.rootInode; i < len(components); i, parentInode = i+1, inode { |
| 170 | + row := fs.db.QueryRow( |
| 171 | + "SELECT inode FROM github_dgsb_dbfs_files WHERE parent = ? AND fname = ?", |
| 172 | + parentInode, components[i]) |
| 173 | + if err := row.Scan(&inode); errors.Is(err, sql.ErrNoRows) { |
| 174 | + return 0, fmt.Errorf( |
| 175 | + "%w: parent inode %d, fname %s", InodeNotFoundErr, parentInode, components[i]) |
| 176 | + } else if err != nil { |
| 177 | + return 0, fmt.Errorf( |
| 178 | + "querying file table: inode %d, fname %s, %w", parentInode, components[i], err) |
| 179 | + } |
| 180 | + } |
| 181 | + return inode, nil |
| 182 | +} |
| 183 | + |
| 184 | +func (fs *FS) Open(fname string) (fs.File, error) { |
| 185 | + fmt.Println("calling open") |
| 186 | + defer fmt.Println("calling open: return") |
| 187 | + if !path.IsAbs(fname) { |
| 188 | + return nil, fmt.Errorf("relative path are not supported") |
| 189 | + } |
| 190 | + |
| 191 | + f := &File{db: fs.db} |
| 192 | + inode, err := fs.namei(fname) |
| 193 | + if err != nil { |
| 194 | + return nil, fmt.Errorf("namei on %s: %w", fname, err) |
| 195 | + } |
| 196 | + f.inode = inode |
| 197 | + |
| 198 | + row := f.db.QueryRowx( |
| 199 | + "SELECT COALESCE(sum(size), 0) FROM github_dgsb_dbfs_chunks WHERE inode = ?", f.inode) |
| 200 | + if err := row.Scan(&f.size); err != nil { |
| 201 | + return nil, fmt.Errorf("file chunks not found: %d, %w", inode, err) |
| 202 | + } |
| 203 | + |
| 204 | + return f, nil |
| 205 | +} |
| 206 | + |
| 207 | +type File struct { |
| 208 | + db *sqlx.DB |
| 209 | + inode int |
| 210 | + offset int64 |
| 211 | + size int64 |
| 212 | + closed bool |
| 213 | +} |
| 214 | + |
| 215 | +func (f *File) Read(out []byte) (int, error) { |
| 216 | + fmt.Println("calling read") |
| 217 | + defer fmt.Println("calling read: return") |
| 218 | + if f.closed { |
| 219 | + return 0, fmt.Errorf("file closed") |
| 220 | + } |
| 221 | + if f.offset >= f.size { |
| 222 | + return 0, io.EOF |
| 223 | + } |
| 224 | + toRead := func(a, b int64) int64 { |
| 225 | + if a < b { |
| 226 | + return a |
| 227 | + } |
| 228 | + return b |
| 229 | + }(f.size-f.offset, int64(len(out))) |
| 230 | + |
| 231 | + // XXX the query is incorrect |
| 232 | + rows, err := f.db.NamedQuery(` |
| 233 | + WITH offsets AS ( |
| 234 | + SELECT |
| 235 | + COALESCE( |
| 236 | + SUM(size) OVER ( |
| 237 | + ORDER BY POSITION ROWS BETWEEN UNBOUNDED PRECEDING AND 1 PRECEDING |
| 238 | + ), |
| 239 | + 0 |
| 240 | + ) AS start, |
| 241 | + position |
| 242 | + FROM github_dgsb_dbfs_chunks |
| 243 | + WHERE inode = :inode |
| 244 | + ) |
| 245 | + SELECT |
| 246 | + github_dgsb_dbfs_chunks.position, |
| 247 | + data, |
| 248 | + size, |
| 249 | + start |
| 250 | + FROM github_dgsb_dbfs_chunks JOIN offsets |
| 251 | + WHERE inode = :inode |
| 252 | + AND :offset < start + size |
| 253 | + AND :offset + :size >= start |
| 254 | + ORDER BY github_dgsb_dbfs_chunks.position |
| 255 | + `, map[string]interface{}{"inode": f.inode, "offset": f.offset, "size": toRead}) |
| 256 | + if err != nil { |
| 257 | + return 0, fmt.Errorf("cannot query the database: %w", err) |
| 258 | + } |
| 259 | + defer rows.Close() |
| 260 | + |
| 261 | + copied := int64(0) |
| 262 | + for rows.Next() { |
| 263 | + var ( |
| 264 | + position int |
| 265 | + buf []byte |
| 266 | + size int64 |
| 267 | + offset int64 |
| 268 | + ) |
| 269 | + if err := rows.Scan(&position, &buf, &size, &offset); err != nil { |
| 270 | + return 0, fmt.Errorf("cannot retrieve file chunk: %w", err) |
| 271 | + } |
| 272 | + |
| 273 | + numByte := int64(copy(out[copied:], buf[f.offset-offset:])) |
| 274 | + copied += numByte |
| 275 | + f.offset += numByte |
| 276 | + if copied >= toRead { |
| 277 | + break |
| 278 | + } |
| 279 | + } |
| 280 | + if err := rows.Err(); err != nil { |
| 281 | + return 0, fmt.Errorf("cannot iterate over file chunks: %w", err) |
| 282 | + } |
| 283 | + |
| 284 | + return int(toRead), nil |
| 285 | +} |
| 286 | + |
| 287 | +func (f *File) Close() error { |
| 288 | + fmt.Println("calling close") |
| 289 | + defer fmt.Println("calling close: return") |
| 290 | + f.db = nil |
| 291 | + f.closed = true |
| 292 | + return nil |
| 293 | +} |
| 294 | + |
| 295 | +func (f *File) Stat() (fs.FileInfo, error) { |
| 296 | + fmt.Println("calling stat") |
| 297 | + defer fmt.Println("calling stat: return") |
| 298 | + return nil, fmt.Errorf("not yet implemented") |
| 299 | +} |
0 commit comments